Beginning_Python______Programmer_to_Programmer_

Beginning_Python______Programmer_to_Programmer_

1 Pages 1-10

▲back to top

1.1 Page 1

▲back to top

1.2 Page 2

▲back to top
Beginning Python
Peter Norton, Alex Samuel, David Aitel, Eric Foster-Johnson,
Leonard Richardson, Jason Diamond,
Aleatha Parker, Michael Roberts

1.3 Page 3

▲back to top

1.4 Page 4

▲back to top
Beginning Python

1.5 Page 5

▲back to top

1.6 Page 6

▲back to top
Beginning Python
Peter Norton, Alex Samuel, David Aitel, Eric Foster-Johnson,
Leonard Richardson, Jason Diamond,
Aleatha Parker, Michael Roberts

1.7 Page 7

▲back to top
Beginning Python
Published by
Wiley Publishing, Inc.
10475 Crosspoint Boulevard
Indianapolis, IN 46256
www.wiley.com
Copyright © 2005 by Wiley Publishing, Inc., Indianapolis, Indiana
Published simultaneously in Canada
ISBN-10: 0-7645-9654-3
ISBN-13: 978-0-7645-9654-4
Manufactured in the United States of America
10 9 8 7 6 5 4 3 2 1
1B/SQ/QX/QV/IN
Library of Congress Cataloging-in-Publication Data:
Beginning Python / Peter Norton.
p. cm.
Includes bibliographical references and index.
ISBN-13: 978-0-7645-9654-4 (paper/website)
ISBN-10: 0-7645-9654-3 (paper/website)
1. Python (Computer program language) I. Norton, Peter, 1974-
QA76.73.P98B45 2005
005.13’3--dc22
2005013968
No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means,
electronic, mechanical, photocopying, recording, scanning or otherwise, except as permitted under Sections 107 or 108 of
the 1976 United States Copyright Act, without either the prior written permission of the Publisher, or authorization
through payment of the appropriate per-copy fee to the Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA
01923, (978) 750-8400, fax (978) 646-8600. Requests to the Publisher for permission should be addressed to the Legal
Department, Wiley Publishing, Inc., 10475 Crosspoint Blvd., Indianapolis, IN 46256, (317) 572-3447, fax (317) 572-4355, or
online at http://www.wiley.com/go/permissions.
LIMIT OF LIABILITY/DISCLAIMER OF WARRANTY: THE PUBLISHER AND THE AUTHOR MAKE NO REPRESEN-
TATIONS OR WARRANTIES WITH RESPECT TO THE ACCURACY OR COMPLETENESS OF THE CONTENTS OF
THIS WORK AND SPECIFICALLY DISCLAIM ALL WARRANTIES, INCLUDING WITHOUT LIMITATION WAR-
RANTIES OF FITNESS FOR A PARTICULAR PURPOSE. NO WARRANTY MAY BE CREATED OR EXTENDED BY
SALES OR PROMOTIONAL MATERIALS. THE ADVICE AND STRATEGIES CONTAINED HEREIN MAY NOT BE SUIT-
ABLE FOR EVERY SITUATION. THIS WORK IS SOLD WITH THE UNDERSTANDING THAT THE PUBLISHER IS NOT
ENGAGED IN RENDERING LEGAL, ACCOUNTING, OR OTHER PROFESSIONAL SERVICES. IF PROFESSIONAL
ASSISTANCE IS REQUIRED, THE SERVICES OF A COMPETENT PROFESSIONAL PERSON SHOULD BE SOUGHT.
NEITHER THE PUBLISHER NOR THE AUTHOR SHALL BE LIABLE FOR DAMAGES ARISING HEREFROM. THE
FACT THAT AN ORGANIZATION OR WEBSITE IS REFERRED TO IN THIS WORK AS A CITATION AND/OR A
POTENTIAL SOURCE OF FURTHER INFORMATION DOES NOT MEAN THAT THE AUTHOR OR THE PUBLISHER
ENDORSES THE INFORMATION THE ORGANIZATION OR WEBSITE MAY PROVIDE OR RECOMMENDATIONS IT
MAY MAKE. FURTHER, READERS SHOULD BE AWARE THAT INTERNET WEBSITES LISTED IN THIS WORK MAY
HAVE CHANGED OR DISAPPEARED BETWEEN THEN THIS WORK WAS WRITTEN AND WHEN IT IS READ.
For general information on our other products and services please contact our Customer Care Department within the
United States at (800) 762-2974, outside the United States at (317) 572-3993 or fax (317) 572-4002.
Trademarks: Wiley, the Wiley logo, Wrox, the Wrox logo, Programmer to Programmer, and related trade dress are trade-
marks or registered trademarks of John Wiley & Sons, Inc. and/or its affiliates, in the United States and other countries,
and may not be used without written permission. All other trademarks are the property of their respective owners. Wiley
Publishing, Inc., is not associated with any product or vendor mentioned in this book.
Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available
in electronic books.

1.8 Page 8

▲back to top
About the Authors
Peter Norton (NY, NY) has been working with Unix and Linux for over a decade at companies large and
small solving problems with Linux. An officer of the NY Linux Users Group, he can be found on the
nylug-talk mailing list. Peter coauthored Professional RHEL3. He works for a very large financial com-
pany in NYC, plying his Python and open-source skills.
Alex Samuel (San Diego, CA) has developed software for biology researchers and now studies high-
energy physics at Caltech. Alex has worked on many GNU/Linux development tools, including GCC,
and co-founded CodeSourcery LLC, a consulting firm specializing in GNU/Linux development tools.
David Aitel (NY, NY) is the CEO of Immunity and a coauthor of Shellcoder’s Handbook.
Eric Foster-Johnson (Minneapolis, MN) uses Python extensively with Java, and is a veteran author,
most recently completing Beginning Shell Scripting.
Leonard Richardson (San Francisco, CA) writes useful Python packages with silly names.
Jason Diamond (CA) Jason Diamond is a software development instructor for DevelopMentor and a
consultant specializing in C++, .NET, Python, and XML. He spends most of his spare time contributing
to open-source projects using his favorite language, Python.
Aleathea Parker (San Francisco CA) is a programmer working as a publication engineer for a major
software company, coding primarily in Python and XSLT. She has a background in web applications and
content management.
Michael Roberts (Puerto Rico) has been programming professionally in C, Perl, and Python for long
enough that Python didn’t actually exist when he started. He is the chief perpetrator of the wftk
open-source workflow toolkit, and he swears that it will someday be finished, for certain values of
“finished”.

1.9 Page 9

▲back to top

1.10 Page 10

▲back to top
Credits
Acquisitions Editor
Debra Williams Cauley
Development Editor
Kelly D. Henthorne
Production Editor
William A. Barton
Copy Editor
Luann Rouff
Production Manager
Tim Tate
Editorial Manager
Mary Beth Wakefield
Graphics and Production Specialists
Sean Decker
Carrie Foster
Lauren Goddard
Denny Hager
Jennifer Heleine
Amanda Spagnuolo
Quality Control Technicians
Leann Harney
Joe Niesen
Carl William Pierce
Media Development Specialists
Angela Denny
Kit Malone
Travis Silvers
Vice President & Executive Group Publisher
Richard Swadley
Proofreading and Indexing
TECHBOOKS Production Services
Vice President and Publisher
Joseph B. Wikert
Project Coordinator
Kristie Rees

2 Pages 11-20

▲back to top

2.1 Page 11

▲back to top

2.2 Page 12

▲back to top
To my Claudia, for keeping me thinking straight through a crazy time.
To my mom, Eunice, for bringing me food and asking if I was okay throughout.
To Debra, for roping me into this. And to all of the authors,
I want to thank you for making it to the finish line.
Whoa! I didn’t know what I was getting you all into! —P. N.
To my dad, Clarence A. Johnson, 1922–2005. —E. F-J.
For my mother. —L. R.
For Jilly: 1 = 2. —J. D.
To Aaron, for putting up with me. —A. P.
To my wife, Agnes, in revenge for her doctoral thesis. —M. R.

2.3 Page 13

▲back to top

2.4 Page 14

▲back to top
Contents
Acknowledgments
Introduction
Chapter 1: Programming Basics and Strings
How Programming Is Different from Using a Computer
Programming Is Consistency
Programming Is Control
Programming Copes with Change
What All That Means Together
The First Steps
Starting codeEditor
Using codeEditor’s Python Shell
Try It Out: Starting the Python Shell
Beginning to Use Python — Strings
What Is a String?
Why the Quotes?
Try It Out: Entering Strings with Different Quotes
Understanding Different Quotes
Putting Two Strings Together
Try It Out: Using + to Combine Strings
Putting Strings Together in Different Ways
Try It Out: Using a Format Specifier to Populate a String
Try It Out: More String Formatting
Displaying Strings with Print
Try It Out: Printing Text with Print
Summary
Exercises
Chapter 2: Numbers and Operators
Different Kinds of Numbers
Numbers in Python
Try It Out: Using Type with Different Numbers
Try It Out: Creating an Imaginary Number
xxix
xxxi
1
1
2
2
2
3
3
3
4
4
5
5
6
6
6
8
8
9
9
9
10
10
10
11
13
13
14
14
15

2.5 Page 15

▲back to top
Contents
Program Files
15
Try It Out: Using the Shell with the Editor
16
Using the Different Types
17
Try It Out Including Different Numbers in Strings
18
Try It Out: Escaping the % Sign in Strings
18
Basic Math
19
Try It Out Doing Basic Math
19
Try It Out: Using the Modulus Operation
20
Some Surprises
20
Try It Out: Printing the Results
21
Using Numbers
21
Order of Evaluation
21
Try It Out: Using Math Operations
21
Number Formats
22
Try It Out: Using Number Formats
22
Mistakes Will Happen
23
Try It Out: Making Mistakes
23
Some Unusual Cases
24
Try It Out: Formatting Numbers as Octal and Hexadecimal
24
Summary
24
Exercises
25
Chapter 3: Variables — Names for Values
27
Referring to Data – Using Names for Data
27
Try It Out: Assigning Values to Names
28
Changing Data Through Names
28
Try It Out: Altering Named Values
29
Copying Data
29
Names You Can’t Use and Some Rules
29
Using More Built-in Types
30
Tuples — Unchanging Sequences of Data
30
Try It Out: Creating and Using a Tuple
30
Try It Out: Accessing a Tuple Through Another Tuple
31
Lists — Changeable Sequences of Data
33
Try It Out Viewing the Elements of a List
33
Dictionaries — Groupings of Data Indexed by Name
34
Try It Out: Making a Dictionary
34
Try It Out: Getting the Keys from a Dictionary
35
Treating a String Like a List
36
Special Types
38
xii

2.6 Page 16

▲back to top
Other Common Sequence Properties
Referencing the Last Elements
Ranges of Sequences
Try It Out: Slicing Sequences
Growing Lists by Appending Sequences
Using Lists to Temporarily Store Data
Try It Out: Popping Elements from a List
Summary
Exercises
Chapter 4: Making Decisions
Comparing Values — Are They the Same?
Try It Out: Comparing Values for Sameness
Doing the Opposite — Not Equal
Try It Out: Comparing Values for Difference
Comparing Values — Which One Is More?
Try It Out: Comparing Greater Than and Less Than
More Than or Equal, Less Than or Equal
Reversing True and False
Try It Out: Reversing the Outcome of a Test
Looking for the Results of More Than One Comparison
How to Get Decisions Made
Try It Out: Placing Tests within Tests
Repetition
How to Do Something — Again and Again
Try It Out: Using a while Loop
Stopping the Repetition
Try It Out: Using else While Repeating
Try It Out: Using continue to Keep Repeating
Handling Errors
Trying Things Out
Try It Out: Creating an Exception with Its Explanation
Summary
Exercises
Chapter 5: Functions
Putting Your Program into Its Own File
Try It Out: Run a Program with Python -i
Contents
38
38
39
39
40
40
40
41
42
43
43
43
45
45
45
45
47
47
47
48
48
49
51
51
51
52
54
54
55
55
56
57
58
59
59
61
xiii

2.7 Page 17

▲back to top
Contents
Functions: Grouping Code under a Name
61
Try It Out: Defining a Function
61
Choosing a Name
62
Describing a Function in the Function
63
Try It Out: Displaying __doc__
63
The Same Name in Two Different Places
64
Making Notes to Yourself
65
Try It Out: Experimenting with Comments
65
Asking a Function to Use a Value You Provide
66
Try It Out Invoking a Function with Parameters
67
Checking Your Parameters
68
Try It Out: Determining More Types with the type Function
69
Try It Out: Using Strings to Compare Types
69
Setting a Default Value for a Parameter — Just in Case
70
Try It Out: Setting a Default Parameter
70
Calling Functions from within Other Functions
71
Try It Out: Invoking the Completed Function
72
Functions Inside of Functions
72
Flagging an Error on Your Own Terms
73
Layers of Functions
74
How to Read Deeper Errors
74
Summary
75
Exercises
76
Chapter 6: Classes and Objects
79
Thinking About Programming
79
Objects You Already Know
79
Looking Ahead: How You Want to Use Objects
81
Defining a Class
81
How Code Can Be Made into an Object
81
Try It Out: Defining a Class
82
Try It Out: Creating an Object from Your Class
82
Try It Out: Writing an Internal Method
84
Try It Out: Writing Interface Methods
85
Try It Out: Using More Methods
87
Objects and Their Scope
89
Try It Out: Creating Another Class
89
Summary
92
Exercises
93
xiv

2.8 Page 18

▲back to top
Chapter 7: Organizing Programs
Modules
Importing a Module So That You Can Use It
Making a Module from Pre-existing Code
Try It Out: Creating a Module
Try It Out: Exploring Your New Module
Using Modules — Starting With the Command Line
Try It Out: Printing sys.argv
Changing How Import Works — Bringing in More
Packages
Try It Out: Making the Files in the Kitchen Class
Modules and Packages
Bringing Everything into the Current Scope
Try It Out: Exporting Modules from a Package
Re-importing Modules and Packages
Try It Out: Examining sys.modules
Basics of Testing Your Modules and Packages
Summary
Exercises
Chapter 8: Files and Directories
File Objects
Writing Text Files
Reading Text Files
Try It Out: Printing the Lengths of Lines in the Sample File
File Exceptions
Paths and Directories
Paths
Directory Contents
Try It Out: Getting the Contents of a Directory
Try It Out: Listing the Contents of Your Desktop or Home Directory
Obtaining Information about Files
Recursive Directory Listings
Renaming, Moving, Copying, and Removing Files
Example: Rotating Files
Creating and Removing Directories
Globbing
Contents
95
96
96
97
97
98
99
100
101
101
102
103
103
104
104
105
106
106
107
109
109
110
111
112
113
113
114
116
116
118
118
118
119
120
121
122
xv

2.9 Page 19

▲back to top
Contents
Pickles
Try It Out: Creating a Pickle File
Pickling Tips
Efficient Pickling
Summary
Exercises
Chapter 9: Other Features of the Language
Lambda and Filter: Short Anonymous Functions
Reduce
Try It Out: Working with Reduce
Map: Short-Circuiting Loops
Try It Out: Use Map
Decisions within Lists — List Comprehension
Generating Lists for Loops
Try It Out: Examining an xrange Object
Special String Substitution Using Dictionaries
Try It Out: String Formatting with Dictionaries
Featured Modules
Getopt — Getting Options from the Command Line
Using More Than One Process
Threads — Doing Many Things in the Same Process
Storing Passwords
Summary
Exercises
Chapter 10: Building a Module
Exploring Modules
Importing Modules
Finding Modules
Digging through Modules
Creating Modules and Packages
Try It Out: Creating a Module with Functions
Working with Classes
Defining Object-Oriented Programming
Creating Classes
Try It Out: Creating a Meal Class
Extending Existing Classes
xvi
123
123
124
125
125
125
127
127
128
128
129
129
130
131
132
133
133
134
134
137
139
140
141
142
143
143
145
145
146
150
150
151
151
151
152
153

2.10 Page 20

▲back to top
Finishing Your Modules
Defining Module-Specific Errors
Choosing What to Export
Documenting Your Modules
Try It Out: Viewing Module Documentation
Testing Your Module
Running a Module as a Program
Try It Out: Running a Module
Creating a Whole Module
Try It Out: Finishing a Module
Try It Out: Smashing Imports
Installing Your Modules
Try It Out: Creating an Installable Package
Summary
Exercises
Chapter 11: Text Processing
Why Text Processing Is So Useful
Searching for Files
Clipping Logs
Sifting through Mail
Navigating the File System with the os Module
Try It Out: Listing Files and Playing with Paths
Try It Out: Searching for Files of a Particular Type
Try It Out: Refining a Search
Working with Regular Expressions and the re Module
Try It Out: Fun with Regular Expressions
Try It Out: Adding Tests
Summary
Exercises
Chapter 12: Testing
Assertions
Try It Out: Using Assert
Test Cases and Test Suites
Try It Out: Testing Addition
Try It Out: Testing Faulty Addition
Test Fixtures
Try It Out: Working with Test Fixtures
Contents
154
154
155
156
157
162
164
164
165
165
169
170
171
174
174
175
175
176
177
178
178
180
181
183
184
186
187
189
189
191
191
192
193
194
195
196
197
xvii

3 Pages 21-30

▲back to top

3.1 Page 21

▲back to top
Contents
Putting It All Together with Extreme Programming
Implementing a Search Utility in Python
Try It Out: Writing a Test Suite First
Try It Out: A General-Purpose Search Framework
A More Powerful Python Search
Try It Out: Extending the Search Framework
Formal Testing in the Software Life Cycle
Summary
Chapter 13: Writing a GUI with Python
GUI Programming Toolkits for Python
PyGTK Introduction
pyGTK Resources
Creating GUI Widgets with pyGTK
Try It Out: Writing a Simple pyGTK Program
GUI Signals
GUI Helper Threads and the GUI Event Queue
Try It Out: Writing a Multithreaded pyGTK App
Widget Packing
Glade: a GUI Builder for pyGTK
GUI Builders for Other GUI Frameworks
Using libGlade with Python
A Glade Walkthrough
Starting Glade
Creating a Project
Using the Palette to Create a Window
Putting Widgets into the Window
Glade Creates an XML Representation of the GUI
Try It Out: Building a GUI from a Glade File
Creating a Real Glade Application
Advanced Widgets
Further Enhancing PyRAP
Summary
Exercises
Chapter 14: Accessing Databases
Working with DBM Persistent Dictionaries
Choosing a DBM Module
Creating Persistent Dictionaries
Try It Out: Creating a Persistent Dictionary
xviii
199
200
201
203
205
206
207
208
209
209
210
211
213
213
214
216
219
222
223
224
225
225
226
227
227
228
230
231
231
238
241
248
248
249
250
250
251
251

3.2 Page 22

▲back to top
Accessing Persistent Dictionaries
Try It Out: Accessing Persistent Dictionaries
Deciding When to Use DBM and When to Use a Relational Database
Working with Relational Databases
Writing SQL Statements
Defining Tables
Setting Up a Database
Try It Out: Creating a Gadfly Database
Using the Python Database APIs
Downloading Modules
Creating Connections
Working with Cursors
Try It Out: Inserting Records
Try It Out: Writing a Simple Query
Try It Out: Writing a Complex Join
Try It Out: Updating an Employee’s Manager
Try It Out: Removing Employees
Working with Transactions and Committing the Results
Examining Module Capabilities and Metadata
Handling Errors
Summary
Exercises
Chapter 15: Using Python for XML
What Is XML?
A Hierarchical Markup Language
A Family of Standards
What Is a Schema/DTD?
What Are Document Models For?
Do You Need One?
Document Type Definitions
An Example DTD
DTDs Aren’t Exactly XML
Limitations of DTDs
Schemas
An Example Schema
Schemas Are Pure XML
Schemas Are Hierarchical
Other Advantages of Schemas
Schemas Are Less Widely Supported
Contents
252
253
255
255
257
259
260
261
262
263
263
264
264
266
267
269
270
271
272
272
273
274
275
275
275
277
278
278
278
278
278
280
280
280
280
281
281
281
281
xix

3.3 Page 23

▲back to top
Contents
XPath
282
HTML as a Subset of XML
282
The HTML DTDs
283
HTMLParser
283
Try It Out: Using HTMLParser
283
htmllib
284
Try It Out: Using htmllib
284
XML Libraries Available for Python
285
Validating XML Using Python
285
What Is Validation?
286
Well-Formedness versus Validation
286
Available Tools
286
Try It Out: Validation Using xmlproc
What Is SAX?
286
287
Stream-based
288
Event-driven
288
What Is DOM?
288
In-memory Access
288
Why Use SAX or DOM
289
Capability Trade-Offs
289
Memory Considerations
289
Speed Considerations
289
SAX and DOM Parsers Available for Python
289
PyXML
290
xml.sax
290
xml.dom.minidom
290
Try It Out: Working with XML Using DOM
290
Try It Out: Working with XML Using SAX
292
Intro to XSLT
293
XSLT Is XML
293
Transformation and Formatting Language
293
Functional, Template-Driven
293
Using Python to Transform XML Using XSLT
294
Try It Out: Transforming XML with XSLT
294
Putting It All Together: Working with RSS
296
RSS Overview and Vocabulary
296
Making Sense of It All
296
RSS Vocabulary
297
An RSS DTD
297
xx

3.4 Page 24

▲back to top
A Real-World Problem
Try It Out: Creating an RSS Feed
Creating the Document
Checking It Against the DTD
Another Real-World Problem
Try It Out: Creating An Aggregator
Summary
Exercises
Chapter 16: Network Programming
Try It Out: Sending Some E-mail
Understanding Protocols
Comparing Protocols and Programming Languages
The Internet Protocol Stack
A Little Bit About the Internet Protocol
Internet Addresses
Internet Ports
Sending Internet E-mail
The E-mail File Format
MIME Messages
MIME Encodings: Quoted-printable and Base64
MIME Content Types
Try It Out: Creating a MIME Message with an Attachment
MIME Multipart Messages
Try It Out: Building E-mail Messages with SmartMessage
Sending Mail with SMTP and smtplib
Try It Out: Sending Mail with MailServer
Retrieving Internet E-mail
Parsing a Local Mail Spool with mailbox
Try It Out: Printing a Summary of Your Mailbox
Fetching Mail from a POP3 Server with poplib
Try It Out: Printing a Summary of Your POP3 Mailbox
Fetching Mail from an IMAP Server with imaplib
Try It Out: Printing a Summary of Your IMAP Mailbox
IMAP’s Unique Message IDs
Try It Out: Fetching a Message by Unique ID
Secure POP3 and IMAP
Webmail Applications Are Not E-mail Applications
Socket Programming
Introduction to Sockets
Try It Out: Connecting to the SuperSimpleSocketServer with Telnet
Contents
297
298
300
301
301
301
303
303
305
305
307
307
308
309
309
310
311
311
313
313
314
315
316
320
321
323
323
323
324
325
327
327
329
330
330
331
331
331
332
333
xxi

3.5 Page 25

▲back to top
Contents
Binding to an External Hostname
The Mirror Server
Try It Out: Mirroring Text with the MirrorServer
The Mirror Client
SocketServer
Multithreaded Servers
The Python Chat Server
Design of the Python Chat Server
The Python Chat Server Protocol
Our Hypothetical Protocol in Action
Initial Connection
Chat Text
Server Commands
General Guidelines
The Python Chat Client
Single-Threaded Multitasking with select
Other Topics
Miscellaneous Considerations for Protocol Design
Trusted Servers
Terse Protocols
The Twisted Framework
Deferred Objects
The Peer-to-Peer Architecture
Summary
Exercises
Chapter 17: Extension Programming with C
Extension Module Outline
Building and Installing Extension Modules
Passing Parameters from Python to C
Returning Values from C to Python
The LAME Project
The LAME Extension Module
Using Python Objects from C Code
Summary
Exercises
Chapter 18: Writing Shareware and Commercial Programs
A Case Study: Background
How Much Python Should You Use?
xxii
334
335
336
336
337
339
340
340
341
341
342
342
342
343
346
348
350
350
350
350
351
351
354
354
354
355
356
358
360
363
364
368
380
383
383
385
385
386

3.6 Page 26

▲back to top
Pure Python Licensing
Web Services Are Your Friend
Pricing Strategies
Watermarking
Other Models
Selling as a Platform, Rather Than a Product
Your Development Environment
Finding Python Programmers
Training non-Python Programmers
Python Employment Resources
Python Problems
Porting to Other Versions of Python
Porting to Other Operating Systems
Debugging Threads
Common Gotchas
Portable Distribution
Essential Libraries
Timeoutsocket
PyGTK
GEOip
Summary
Chapter 19: Numerical Programming
Numbers in Python
Integers
Long Integers
Floating-point Numbers
Formatting Numbers
Characters as Numbers
Mathematics
Arithmetic
Built-in Math Functions
The math Module
Complex Numbers
Arrays
The array Module
The numarray Package
Using Arrays
Computing the Standard Deviation
Summary
Exercises
Contents
387
388
389
390
394
395
395
396
397
397
397
397
398
399
399
400
401
401
402
402
403
405
405
406
406
407
408
410
412
412
414
415
416
418
420
422
422
423
424
425
xxiii

3.7 Page 27

▲back to top
Contents
Chapter 20: Python in the Enterprise
Enterprise Applications
Document Management
The Evolution of Document Management Systems
What You Want in a Document Management System
People in Directories
Taking Action with Workflow
Auditing, Sarbanes-Oxley, and What You Need to Know
Auditing and Document Management
Working with Actual Enterprise Systems
Introducing the wftk Workflow Toolkit
Try It Out: Very Simple Record Retrieval
Try It Out: Very Simple Record Storage
Try It Out: Data Storage in MySQL
Try It Out: Storing and Retrieving Documents
Try It Out: A Document Retention Framework
The python-ldap Module
Try It Out: Using Basic OpenLDAP Tools
Try It Out: Simple LDAP Search
More LDAP
Back to the wftk
Try It Out: Simple Workflow Trigger
Try It Out: Action Queue Handler
Summary
Exercises
Chapter 21: Web Applications and Web Services
REST: The Architecture of the Web
Characteristics of REST
A Distributed Network of Interlinked Documents
A Client-Server Architecture
Servers Are Stateless
Resources
Representations
REST Operations
HTTP: Real-World REST
Try It Out: Python’s Three-Line Web Server
The Visible Web Server
Try It Out: Seeing an HTTP Request and Response
The HTTP Request
The HTTP Response
xxiv
427
428
428
429
430
431
432
433
434
435
435
436
438
439
441
446
448
449
451
453
453
454
456
458
458
459
460
460
461
461
461
461
462
462
463
463
464
465
466
467

3.8 Page 28

▲back to top
CGI: Turning Scripts into Web Applications
Try It Out: Running a CGI Script
The Web Server Makes a Deal with the CGI Script
CGI’s Special Environment Variables
Accepting User Input through HTML Forms
The cgi Module: Parsing HTML Forms
Try It Out: Printing Any HTML Form Submission
Building a Wiki
The BittyWiki Core Library
Back-end Storage
WikiWords
Writing the BittyWiki Core
Try It Out: Creating Wiki Pages from an Interactive Python Session
The BittyWiki Web Interface
Resources
Request Structure
But Wait — There’s More (Resources)
Wiki Markup
Web Services
How Web Services Work
REST Web Services
REST Quick Start: Finding Bargains on Amazon.com
Try It Out: Peeking at an Amazon Web Services Response
Introducing WishListBargainFinder
Giving BittyWiki a REST API
Wiki Search-and-Replace Using the REST Web Service
Try It Out: Wiki Searching and Replacing
XML-RPC
XML-RPC Quick Start: Get Tech News from Meerkat
The XML-RPC Request
Representation of Data in XML-RPC
The XML-RPC Response
If Something Goes Wrong
Exposing the BittyWiki API through XML-RPC
Try It Out: Manipulating BittyWiki through XML-RPC
Wiki Search-and-Replace Using the XML-RPC Web Service
SOAP
SOAP Quick Start: Surfing the Google API
The SOAP Request
The SOAP Response
If Something Goes Wrong
Contents
468
469
470
471
473
474
478
480
481
481
481
481
483
484
484
484
485
486
493
494
494
495
496
497
500
503
507
508
509
511
512
513
513
514
517
518
520
520
522
524
524
xxv

3.9 Page 29

▲back to top
Contents
Exposing a SOAP Interface to BittyWiki
Try It Out: Manipulating BittyWiki through SOAP
Wiki Search-and-Replace Using the SOAP Web Service
Documenting Your Web Service API
Human-Readable API Documentation
The BittyWiki REST API Document
The BittyWiki XML-RPC API Document
The BittyWiki SOAP API Document
The XML-RPC Introspection API
Try It Out: Using the XML-RPC Introspection API
WSDL
Try It Out: Manipulating BittyWiki through a WSDL Proxy
Choosing a Web Service Standard
Web Service Etiquette
For Consumers of Web Services
For Producers of Web Services
Using Web Applications as Web Services
A Sampling of Publicly Available
Web Services
Summary
Exercises
Chapter 22: Integrating Java with Python
Scripting within Java Applications
Comparing Python Implementations
Installing Jython
Running Jython
Running Jython Interactively
Try It Out: Running the Jython Interpreter
Running Jython Scripts
Try It Out Running a Python Script
Controlling the jython Script
Making Executable Commands
Try It Out: Making an Executable Script
Running Jython on Your Own
Packaging Jython-Based Applications
Integrating Java and Jython
Using Java Classes in Jython
Try It Out: Calling on Java Classes
Try It Out: Creating a User Interface from Jython
Accessing Databases from Jython
Working with the Python DB API
xxvi
525
526
527
529
529
529
529
530
530
530
531
533
534
535
535
535
536
536
538
538
539
540
541
541
542
542
542
543
543
544
545
546
546
547
547
548
548
550
552
553

3.10 Page 30

▲back to top
Setting Up a Database
Try It Out: Create Tables
Writing J2EE Servlets in Jython
Setting Up an Application Server
Adding the PyServlet to an Application Server
Extending HttpServlet
Try It Out: Writing a Python Servlet
Choosing Tools for Jython
Testing from Jython
Try It Out: Exploring Your Environment with Jython
Embedding the Jython Interpreter
Calling Jython Scripts from Java
Try It Out: Embedding Jython
Compiling Python Code to Java
Handling Differences between C Python and Jython
Summary
Exercises
Appendix A: Answers to Exercises
Appendix B: Online Resources
Appendix C: What’s New in Python 2.4
Glossary
Index
Contents
554
555
558
559
560
561
562
564
565
565
566
566
567
568
569
570
571
573
605
609
613
623
xxvii

4 Pages 31-40

▲back to top

4.1 Page 31

▲back to top

4.2 Page 32

▲back to top
Acknowledgments
I would like to acknowledge Macktez at www.macktez.com. They’re a great group, and not just for loan-
ing me the Mac. Most of all, thank you to the folks who hack on Python and other great projects so that
everyone else can learn. —Peter Norton
I would like to acknowledge the pyGTK team, the hardworking individuals at Immunity, Inc., and my
wife, Justine. —Dave Aitel
I would like to acknowledge Sumana Harihareswara, Mark A. Hershberger, Danny O’Brien, and Tony
Steidler-Dennison. —Leonard Richardson
The publisher would like to thank the early efforts of Oleg Kolesnikov and Levi Cook, who had to bow
out of the writing process due to personal matters, but who contributed to the overall structure and con-
tent outline.

4.3 Page 33

▲back to top

4.4 Page 34

▲back to top
Introduction
What Is Python?
Python is a programming language that embodies many of the best practices and intentions of other pro-
gramming languages that have been created and refined during the last two decades.
❑ Python is interpreted, which makes it easy to test programs and to use interactively.
❑ Python is object-oriented, which enables you to leverage the work of other programmers and
share your own work with the community at large.
❑ Python is open-source software, so you may use it and deploy it without being concerned about
a single company that owns it one day going out of business or otherwise making the language
unavailable to you.
❑ Python is small; it is a simple language that can be explained to a programmer in a very short
time and can be taught to a new programmer without having to worry about the innumerable
special cases that occur with more complex and involved languages.
❑ Python is extensible; it comes with a set of modules that act as tools that build on the language.
Python can also be extended with libraries that come from other languages such as C. Python
makes this integration simple.
❑ Python is familiar; it incorporates many language features from other languages in a way that
makes it friendly to use if you are already familiar with the languages to which it is similar, such
as Java, C, or Perl. At the same time, these concepts that Python has in common with other lan-
guages makes it possible to move to these other languages with confidence.
The Benefits of Python
Python’s benefits are at least twofold: to the person programming in Python and to the user of programs
written in Python.
Some of the benefits to the programmer are as follows:
❑ Consistency — Python is maintained by a loose organization of people, but the language itself is
changed slowly and with the goal of not breaking anything within the same major version
series (that is, 2.3 to 2.4 will have a minimal impact on existing programs).
❑ Scalability — Projects written in Python have been proven to scale to very large projects as both
the primary language and an auxiliary language.

4.5 Page 35

▲back to top
Introduction
❑ Flexibility — Python can be used to work with libraries and programs written in other pro-
gramming languages so that its interpreter can be used with other languages in a manner that
almost always provides more convenience than the language in which the library or program
was written.
❑ Multiple cooperating implementations — There is more than one Python. Different implemen-
tations enable Python to be specialized for special situations or problems in ways that a one-
size-fits-all approach doesn’t allow.
❑ Interactive interpreter — Python provides an environment that can be used to interact with a
program while it is running, which enables programs to be tested, and their behavior changed,
without having to have an extended cycle in which the application is rebuilt to test theories.
The advantages for people who use code written in Python come naturally from the features that make it
so useful to programmers:
❑ Maintainability — Python’s syntax and the structure that the language provides makes it easier
to maintain and, thus, easier for users to make changes and for application programmers to
respond to the needs of their clients.
❑ Cross-platform — Python exists on many hardware and software platforms, making applica-
tions and tools written in Python more durable to changes in a client environment.
❑ Easy to implement — When a product is written in Python, the entire product, including the
Python interpreter, can be included in a single area of the system. This allows the release to be
better controlled and the stability of the product to rely less on the system it’s installed.
Who This Book Is For
This book is written for users who are computer literate and have reached a point where their desire to
control their computing environment has naturally led to wanting to write their own programs. This
population includes the following groups:
❑ Enterprise users who have found that the increasing trend toward the acceptance of open-source
software has led to the ability to write company-sanctioned programs instead of having to wait
for or hire outside parties for simple tasks or to evaluate the feasibility of solving a problem.
❑ Users who have discovered that Linux, FreeBSD, or other Unix or Unix-like operating systems
provide them with programming tools that they can use, but do not necessarily provide the guid-
ance needed to give a new programmer the tools needed to solve their technological problems.
❑ Students who are learning to program and who need a language that is clear and well defined,
and useful in the sense of being able to create real, usable programs at every level.
The first section of this book covers topics that will be useful for both beginning programmers and inter-
mediate practitioners of the discipline. The second section provides useful code and concepts to give the
reader a starting point in understanding how to solve problems by adding to the information and pro-
grams provided to solve their specific problem.
xxxii

4.6 Page 36

▲back to top
Introduction
What This Book Covers
This book begins with the most basic concepts of the Python programming languages and proceeds to
introduce the core concepts of the language. The main body of the book assumes that the reader already
has an installed version of Python 2.4 (or later). To help with this, Appendix B contains references, and
the web site contains download and installation instructions for Python for Linux, Windows, and
MacOS X.
The initial chapters provide the material discussion of all of the concepts needed to write simple pro-
grams in Python, as well as the understanding to peruse and understand programs written by more
experienced programmers, because it is very important to read other people’s programs when learning
how to program.
The early chapters explain important concepts such as the basic types in Python — strings, lists, tuples,
and dictionaries — and build up to the use of classes and objects, and the structure for larger projects in
modules and packages. In addition, these chapters touch on some of the features and changes in recent
versions of Python.
The latter portion of the book covers a variety of topics that are useful to a broad cross-section of the
computer-using population. The intention is that readers who have covered the basic topics will have
the knowledge to use some of the material in the latter half, while other chapters will be more useful for
the novice who has progressed to an intermediate skill level and for programmers who already under-
stand programming or who already have prior knowledge of the problem being solved and want to
understand how to work on that problem in Python.
How This Book Is Structured
Chapters 1–10 walk the reader through programming in Python using examples in the text and explana-
tions of the concepts being introduced, starting with the assumption of a minimal knowledge of pro-
gramming or Python and proceeding from there, building upon concepts as they are introduced.
Examples are based on the theme of food and serving food, because it is a concept with which everyone
should be familiar. Although these examples are simple, they can be used to understand how to write
programs that accomplish other tasks as well.
❑ Chapter 1 covers using the Python interpreter interactively to become familiar with strings, a
basic data type that almost every program will use heavily to provide information to users and
to create any text that will ever be viewed by a user.
❑ Chapter 2 introduces numbers and the basic numeric types involved in arithmetic and the kind
of simple math that is used in day-to-day life, as well as the mathematical operations that you
can perform with them in Python, such as addition, subtraction, multiplication, and division.
❑ Chapter 3 introduces the concept of variables, which are the names that you use to refer to and
store and manipulate data in your programs.
❑ Chapter 4 covers making decisions based on the conditions of the data in the program. Data ref-
erenced through variables can determine the decisions made at different points in time during
the program, and leveraging that control enables programs to be designed to do their work.
xxxiii

4.7 Page 37

▲back to top
Introduction
❑ Chapter 5 covers functions. Functions are a way of taking groups of code that achieve one or a
few goals and putting them together so that they can be invoked by a name that makes obvious
sense when it is viewed in a program.
❑ Chapter 6 adds the concept of the class and the object to programming with Python. Python is
built around the idea of objects, and is usually thought of as an object-oriented language. A class
is a set of data and functions (although functions, when used specifically as part of a class, are
called methods) that create the definition of an object. An object is the result of a class being
invoked and bound to a name that is, from then on, an independent entity that contains data
and methods that can be used in a program independently of other objects of the same type. The
data and methods of an object are not available in general — they are confined to the object in its
scope.
❑ Chapter 7 introduces modules. Modules are a scope whose functions, classes, and data can func-
tion independently of other code, so they can be used separately from any other code in your
program. Modules are the building blocks of Python, as Python’s behavior is extended beyond
the base language by modules that add features. Chapter 7 also introduces Packages, which pro-
vide a structure by associating the files on disk to a deepening scope, using the same form as
that of modules.
❑ Chapter 8 explains files and input/output (I/O). Input/output is the general term for anything
that involves reading data or writing data outside of the program itself. Examples of I/O
include saving and reading files, asking a user a question in a screen, input from a mouse or a
keyboard, and any action that takes place over a network.
❑ Chapter 9 offers examples of many common features of Python that are useful to programmers
and that can be so convenient, but are easily overlooked.
❑ Chapter 10 is an example of a complete Package that combines most of the concepts covered
previously. The module is contrived but provides a complete example of the things that should
be kept in mind when writing modules and packages.
Chapters 11–22 cover more advanced topics. After the language is understood, these examples can be
directly used to build programs that can solve problems in the real world.
❑ Chapter 11 covers dealing with text and regular expressions. Regular expressions are a concept
describing the capability to create rules that can match and change text more flexibly than a lit-
eral letter-by-letter match. Regular expressions provide a syntax to do things such as find all
words that begin with a capital “R” or every file that has a line that begins with a “g” and ends
with an exclamation point.
❑ Chapter 12 delves more deeply into the topic of how to write tests for your modules and pro-
grams so that you can automatically confirm that your program continues to work as you make
changes to it, and so that your clients and users can confirm it works as well.
❑ Chapter 13 introduces programming with a graphical user interface in Python, using GTK+, a
toolkit that is available on Linux, Windows, and MacOS X. A graphical user interface, or GUI, is
any environment that provides windows, buttons, slide-bars and other features that are com-
monly available on modern computers and through which users expect programs to communi-
cate, with the mouse, keyboard, and so on.
xxxiv

4.8 Page 38

▲back to top
Introduction
❑ Chapter 14 shows how to use a database for storing data. The database support available
through Python modules is very powerful, and these modules are available for multiple rela-
tional databases, such as MySQL, Oracle, DB2, and Postgresql.
❑ Chapter 15 focuses on XML, HTML, XSL, and related technologies. XML is a foundational tech-
nology that offers a structured way to access and manipulate text and data. HTML is very simi-
lar to XML, except that it is much simpler and less flexible, and it presents almost all data on the
web one way or the other. XSL is a way to define, in XML, how to transform text, almost like a
language of its own for dealing with documents. Python’s modules to handle these types of
documents very powerful and widely used.
❑ Chapter 16 covers simple networks and programming and e-mail. Most serious programs
require the capability to communicate over the network using network modules. In addition,
e-mail is an extremely common way of using the pervasive network infrastructure provided by
the Internet to send messages to people. Many small and large applications will have some level
of e-mail integration, and Python has the facilities to use e-mail.
❑ Chapter 17 shows how to extend Python for intermediate programmers who have libraries that
they have used or have written in other languages, such as C.
❑ Chapter 18 offers advice, almost a case study, on how to assess the issues involved with packag-
ing shareware or commercial software that has been written in Python.
❑ Chapter 19 covers numerical programming techniques that would often be used in scientific or
other disciplines where advanced math is used.
❑ Chapter 20 provides information for the more advanced user who needs to solve enterprise-
class problems such as workflow and document management.
❑ Chapter 21 covers web applications and APIs that enable applications to access the rich features
available via the Internet from sites such as Google and Yahoo!.
❑ Chapter 22 explores Jython, a version of the Python language that is written in pure Java, and
which enables you to use the Python language while also giving you access to all of the mod-
ules and extensions available to the Java community.
Appendixes
❑ Appendix A provides answers to the exercise questions at the end of each chapter in the book.
❑ Appendix B provides the web sites from which software can be downloaded, and describes the
steps required to install the required software.
❑ Appendix C provides references to more information for people who want to find out more
about Python or the examples in specific chapters.
❑ The Glossary provides definitions of terms that are used in the book.
What You Need to Use This Book
There are some minimal requirements to use the material in this book. The following are recommenda-
tions, as Python itself runs on many different platforms. However, the first chapters assume that you
xxxv

4.9 Page 39

▲back to top
Introduction
have access to a GUI such as is available in Windows, Mac OS X, or the X Window system on Unix and
Linux. Naturally, some chapters, such as the GUI chapter, require the GUI as well, and chapters involv-
ing networking will make much more sense if a network connection is in place.
Following are the suggested minimum requirements:
❑ A PC running Linux, a BSD Unix, or Windows running at 500MHz or faster, or a G3 or later
Macintosh running Mac OS X version 10.2 or later
❑ 256MB of memory (at a minimum)
❑ A graphical user interface — whichever is native to the platform you are on
❑ Administrative access to the computer you are on so that you may install required software
❑ Network access to a TCP/IP network such as the Internet or a campus network
❑ Internet access to download required software
❑ For the enterprise software chapter, an LDAP server needs to be available for testing
Conventions
To help you get the most from the text and keep track of what’s happening, we’ve used a number of con-
ventions throughout the book.
Try It Out
The Try It Out is an exercise you should work through, following the text in the book.
1. They usually consist of a set of steps.
2. Each step has a number.
3. Follow the steps through with your copy of the database.
How It Works
After each Try It Out, the code you’ve typed will be explained in detail.
Boxes like this one hold important, not-to-be forgotten information that is directly
relevant to the surrounding text.
Tips, hints, tricks, and asides to the current discussion are offset and placed in italics like this.
As for styles in the text:
❑ We highlight new terms and important words when we introduce them.
❑ We show keyboard strokes like this: Ctrl+A.
xxxvi

4.10 Page 40

▲back to top
Introduction
❑ We show file names, URLs, and code within the text like so: persistence.properties.
❑ We present code in two different ways:
In code examples we highlight new and important code with a gray background.
The gray highlighting is not used for code that’s less important in the present
context, or has been shown before.
On the Web Site
As you work through the examples in this book, you may choose to either type in all the code manually
or use the source code files that accompany the book. All the source code used in this book is available
for download at www.wrox.com. Once at the site, simply locate the book’s title by using either the Search
box or one of the topic lists and click the Download Code link. You can then choose what you want to
download from a list of files zipped by chapter.
Because many books have similar titles, you may find it easiest to search by ISBN; for this book the
ISBN is 0-7645-9654-3 (changing to 978-0-7645-9654-4 as the new industry wide 13-digit ISBN
numbering system is phased in by January 2007).
After you download the code, just decompress it with your favorite compression tool. Mac OS X 10.3
and later supports uncompressing zip files directly in the Finder. Alternatively, you can go to the main
Wrox code download page at www.wrox.com/dynamic/books/download.aspx to see the code avail-
able for this book and all other Wrox books.
Besides source code, you will find some additional resource material on the wrox.com site. When you get
to the download page, as described in the preceding text, look for a file called Additional Resources.zip. In
this file are two documents: Developer Resources and Developer Tools Roadmap. Take a look at these files
when you have a chance. You might find that they offer some valuable information to help you during the
development process.
Errata
We make every effort to ensure that there are no errors in the text or in the code. However, no one is per-
fect, and mistakes do occur. If you find an error in one of our books, such as a spelling mistake or faulty
piece of code, we would be very grateful for your feedback. By sending in errata, you may save another
reader hours of frustration; at the same time, you are helping us provide higher quality information.
To find the errata page for this book, go to www.wrox.com and locate the title using the Search box or
one of the topic lists. When the information page for the book appears, click the Errata link in the
middle of the page. On this page, you can view all errata that has been submitted for this book and
posted by Wrox editors. A complete book list, including links to each book’s errata, is also available at
www.wrox.com/misc-pages/booklist.shtml.
xxxvii

5 Pages 41-50

▲back to top

5.1 Page 41

▲back to top
Introduction
If you don’t spot “your” error on the book’s Errata page, click the Errata Form link in the paragraph just
above the actual listing of errata or go to www.wrox.com/contact/techsupport.shtml. Complete the
form to send us the error you have found. We’ll check the information and, if appropriate, post a mes-
sage to the book’s errata page and fix the problem in subsequent editions of the book.
p2p.wrox.com
For author and peer discussion, join the P2P forums at p2p.wrox.com. The forums are a web-based sys-
tem for you to post messages relating to Wrox books and related technologies and to interact with other
readers and technology users. The forums offer a subscription feature to e-mail you topics of interest of
your choosing when new posts are made to the forums. Wrox authors, editors, other industry experts,
and your fellow readers are present on these forums.
At http://p2p.wrox.com, you will find a number of different forums that will help you not only as
you read this book, but also as you develop your own applications. To join the forums, just follow these
steps:
1. Go to p2p.wrox.com and click the Register link.
2. Read the terms of use and click Agree.
3. Complete the required information to join as well as any optional information you want to pro-
vide and click Submit.
4. You will receive an e-mail with information describing how to verify your account and com-
plete the joining process.
You can read messages in the forums without joining P2P but in order to post your own messages, you
must join.
After you join, you can post new messages and respond to messages other users post. You can read mes-
sages at any time on the web. If you would like to have new messages from a particular forum e-mailed
to you, click the Subscribe to this Forum icon by the forum name in the forum listing.
For more information about how to use the Wrox P2P, be sure to read the P2P FAQs for answers to ques-
tions about how the forum software works as well as many common questions specific to P2P and Wrox
books. To read the FAQs, click the FAQ link on any P2P page.
xxxviii

5.2 Page 42

▲back to top
1
Programming Basics
and Strings
This chapter is a gentle introduction to the practice of programming in Python. Python is a very
rich language with many features, so it is important to learn to walk before you learn to run.
Chapters 1 through 3 provide a basic introduction to common programming ideas, explained in
easily digestible paragraphs with simple examples.
If you are already an experienced programmer interested in Python, you may want to read this
chapter quickly and take note of the examples, but until Chapter 3 you will be reading material
with which you’ve probably already gained some familiarity in another language.
If you are a novice programmer, by the end of this chapter you will have learned some guiding
principles for programming, as well as directions for your first interactions with a programming
language — Python. The exercises at the end of the chapter provide hands-on experience with the
basic information that you’ll have learned.
How Programming Is Different
from Using a Computer
The first thing you need to understand about computers when you’re programming is that you
control the computer. Sometimes the computer doesn’t do what you expect, but even when it
doesn’t do what you want the first time, it should do the same thing the second and third time —
until you take charge and change the program.
The trend in personal computers has been away from reliability and toward software being built
on top of other, unreliable, software. The results that you live with might have you believing that
computers are malicious and arbitrary beasts, existing to taunt you with unbearable amounts of
extra work and various harassments while you’re already trying to accomplish something. If you
do feel this way, you already know that you’re not alone. However, after you’ve learned how to
program, you gain an understanding of how this situation has come to pass, and perhaps you’ll
find that you can do better than some of the programmers whose software you’ve used.

5.3 Page 43

▲back to top
Chapter 1
Note that programming in a language like Python, an interpreted language, means that you are not
going to need to know a whole lot about computer hardware, memory, or long sequences of 0s and 1s.
You are going to write in text form like you are used to reading and writing but in a different and sim-
pler language. Python is the language, and like English or any other language(s) you speak, it makes
sense to other people who already speak the language. Learning a programming language can be even
easier, however, because programming languages aren’t intended for discussions, debates, phone calls,
plays, movies, or any kind of casual interaction. They’re intended for giving instructions and ensuring
that those instructions are followed. Computers have been fashioned into incredibly flexible tools that
have found a use in almost every business and task that people have found themselves doing, but they
are still built from fundamentally understandable and controllable pieces.
Programming Is Consistency
In spite of the complexity involved in covering all of the disciplines into which computers have crept,
the basic computer is still relatively simple in principle. The internal mechanisms that define how a com-
puter works haven’t changed a lot since the 1950s when transistors were first used in computers.
In all that time, this core simplicity has meant that computers can, and should, be held to a high stan-
dard of consistency. What this means to you, as the programmer, is that anytime you tell a computer to
metaphorically jump, you must tell it how high and where to land, and it will perform that jump — over
and over again for as long as you specify. The program should not arbitrarily stop working or change
how it works without you facilitating the change.
Programming Is Control
Programming a computer is very different from creating a program, as the word applies to people in real
life. In real life, we ask people to do things, and sometimes we have to struggle mightily to ensure that
our wishes are carried out — for example, if we plan a party for 30 people and assign two of them to
bring the chips and dip and two of them to bring the drinks.
With computers that problem doesn’t exist. The computer does exactly what you tell it to do. As you can
imagine, this means that you must pay some attention to detail to ensure that the computer does just
what you want it to do.
One of the goals of Python is to program in blocks that enable you to think about larger and larger pro-
jects by building each project as pieces that behave in well-understood ways. This is a key goal of a pro-
gramming style known as object-oriented programming. The guiding principle of this style is that you
can create reliable pieces that still work when you piece them together, that are understandable, and that
are useful. This gives you, the programmer, control over how the parts of your programs run, while
enabling you to extend your program as the problems you’re solving evolve.
Programming Copes with Change
Programs are run on computers that handle real-world problems; and in the real world, plans and cir-
cumstances frequently change. Because of these shifting circumstances, programmers rarely get the
opportunity to create perfectly crafted, useful, and flexible programs. Usually, you can achieve only two
of these goals. The changes that you will have to deal with should give you some perspective and lead
you to program cautiously. With sufficient caution, you can create programs that know when they’re
2

5.4 Page 44

▲back to top
Programming Basics and Strings
being asked to exceed their capabilities, and they can fail gracefully by notifying their users that they’ve
stopped. In the best cases, you can create programs that explain what failed and why. Python offers
especially useful features that enable you to describe what conditions may have occurred that prevented
your program from working.
What All That Means Together
Taken together, these beginning principles mean that you’re going to be introduced to programming as a
way of telling a computer what tasks you want it to do, in an environment where you are in control. You
will be aware that sometimes accidents can happen and that these mistakes can be accommodated
through mechanisms that offer you some discretion regarding how these conditions will be handled,
including recovering from problems and continuing to work.
The First Steps
First, you should go online to the web site for the book, following the procedure in the Introduction, and
follow the instructions there for downloading PythonCard. PythonCard is a set of utilities that provides
an environment for programming in Python. PythonCard is a product that’s free to use and distribute
and is tailor-made for writing in Python. It contains an editor, called codeEditor, that you will be using
for the first part of this book. It has a lot in common with the editor that comes with Python, called idle,
but in the opinion of the authors, codeEditor works better as a teaching tool because it was written with
a focus on users who may be working on simpler projects. In addition, codeEditor is a program written
in Python.
Programs are written in a form called source code. Source code contains the instruc-
tions that the language follows, and when the source code is read and processed, the
instructions that you’ve put in there become the actions that the computer takes.
Just as authors and editors have specialized tools for writing for magazines, books, or online publica-
tions, programmers also need specialized tools. As a starting Python programmer, the right tool for the
job is codeEditor.
Starting codeEditor
Depending on your operating system, you will start codeEditor in different ways.
Once it is installed on your system with PythonCard, on Linux or Unix-based systems, you can just type
codeEditor in a terminal or shell window and it will start.
On Windows, codeEditor should be in your Start menu under Programs ➪ PythonCard. Simply launch-
ing the program will get you started.
When you start codeEditor for the first time, it doesn’t display an open file to work with, so it gives you
the simplest possible starting point, a window with very little in it. Along the left side, you’ll see line
numbers. Programmers are often given information by their programs about where there was a problem,
3

5.5 Page 45

▲back to top
Chapter 1
or where something happened, based on the line number in the file. This is one of the features of a good
programming editor, and it makes it much easier to work with programs.
Using codeEditor’s Python Shell
Before starting to write programs, you’re going to learn how to experiment with the Python shell. For
now, you can think of the Python shell as a way to peer within running Python code. It places you inside
of a running instance of Python, into which you can feed programming code; at the same time, Python
will do what you have asked it to do and will show you a little bit about how it responds to its environ-
ment. Because running programs often have a context — things that you as the programmer have tai-
lored to your needs — it is an advantage to have the shell because it lets you experiment with the context
you have created. Sometimes the context that you’re operating in is called your environment.
Try It Out Starting the Python Shell
To start the Python shell from codeEditor, pull down the Shell menu in the codeEditor’s menu bar and
select Shell window. This will open a window with the Python shell in it (no surprises here) that just has
simple text, with line numbers along the left side (see Figure 1-1). You can get a similar interface without
using PythonCard by starting the regular Python interpreter, without PythonCard’s additions, by just
typing python on a Unix system or by invoking Python from the Start menu on a Windows system.
Figure 1-1
4

5.6 Page 46

▲back to top
Programming Basics and Strings
After you’ve started the shell, you’ll be presented with some information that you don’t have to be con-
cerned about now (from, import, pcapp, and so on), followed by the sign that the interpreter is ready
and waiting for you to work with it: >>>.
>>> import wx
>>> from PythonCard import dialog, util
>>> bg = pcapp.getCurrentBackground()
>>> self = bg
>>> comp = bg.components
>>>
How It Works
The codeEditor is a program written in Python, and the Python shell within it is actually a special pro-
gramming environment that is enhanced with features that you will use later in the book to help you
explore Python. The import, from, and other statements are covered in Chapter 7 in depth, but for now
they’re not important.
Beginning to Use Python — Strings
At this point, you should feel free to experiment with using the shell’s basic behavior. Type some text, in
quotes; for starters, you could type the following:
>>> “This text really won’t do anything”
“This text really won’t do anything”
>>>
You should notice one thing immediately: After you entered a quote (“), codeEditor’s Python shell changed
the color of everything up to the quote that completed the sentence. Of course, the preceding text is abso-
lutely true. It did nothing: It didn’t change your Python environment; it was merely evaluated by the run-
ning Python instance, in case it did determine that in fact you’d told it to do something. In this case, you’ve
asked it only to read the text you wrote, but doing this doesn’t constitute a change to the environment.
However, you can see that Python indicated that it saw what you entered. It showed you the text you
entered, and it displayed it in the manner it will always display a string — in quotes. As you learn about
other data types, you’ll find that Python has a way of displaying each one differently.
What Is a String?
The string is the first data type that you’re being introduced to within Python. Computers in general,
and programming languages specifically, segregate everything they deal with into types. Types are cate-
gories for things within a program with which the program will work. After a thing has a type, the pro-
gram (and the programmer) knows what to do with that thing. This is a fundamental aspect of how
computers work, because without a named type for the abstract ideas that they work with, the computer
won’t know how to do basic things like combine two different values. However, if you have two things,
and they’re of the same type, you can define easy rules for combining them. Therefore, when the type of
a thing has been confirmed, Python knows what its options are, and you as the programmer know more
about what to do with it.
5

5.7 Page 47

▲back to top
Chapter 1
Why the Quotes?
Now, back to strings in particular. Strings are the basic unit of text in Python. Unlike some other pro-
gramming languages, a single letter is represented as a one-letter string. Instead of trying to explain
strings in terms of other concepts in a vacuum, let’s create some examples of strings using the Python
shell and build from there.
Try It Out Entering Strings with Different Quotes
Enter the following strings, keeping in mind the type of quotes (single or double) and the ends of lines
(use the Enter key when you see that the end of a line has been reached):
>>> “This is another string”
‘This is another string’
>>> ‘This is also a string’
‘This is also a string’
>>> “””This is a third string that is some
...
how different”””
‘This is a third string that is some\\n how different’
How It Works
If you use different quotes, they may look different to you; to the Python interpreter; however all of
them can be used in the same situations and are very similar. For more information, read on.
These examples raise a few questions. In your first text example, you saw that the text was enclosed
in double quotes, and when python saw two quotes it repeated those double quotes on the next line.
However, in the preceding example, double quotes are used for “This is another string”, but below it
single quotes are used. Then, in the third example, three double quotes in a row are used, and after the
word “some” we used the Enter key, which caused a new line to appear. The following section explains
these seemingly arbitrary conventions.
Understanding Different Quotes
Three different types of quotes are used in Python. First, there are the single and double quotes, which
you can look at in two ways. In one way, they are identical. They work the same way and they do the
same things. Why have both? Well, there are a couple of reasons. First, strings play a huge part in almost
any program that you’re going to write, and quotes define strings. One challenge when you first use
them is that quotes aren’t special characters that appear only in computer programs. They are a part of
any normal English text to indicate that someone has spoken. In addition, they are used for emphasis or
to indicate that something is literally what was seen or experienced.
The dilemma for a programming language is that when you’re programming, you can only use charac-
ters that are already on a keyboard. However, the keys on a keyboard can be entered by the average user,
so obviously people normally use those keys for tasks other than programming! Therefore, how do you
make it a special character? How do you indicate to the language that you, the programmer, mean some-
thing different when you type a set of quotes to pass a string to your program, versus when you, as the
programmer, enter quotes to explain something to the person using your program?
One solution to this dilemma is a technique that’s called escaping. In most programming languages, at
least one character, called an escape character, is designated; and it has the power to remove the special
6

5.8 Page 48

▲back to top
Programming Basics and Strings
significance from other special characters, such as quotes. This character in Python is the backslash ( \\).
Therefore, if you have to quote some text within a string and it uses the same style of quote in which you
enclosed the entire string, you need to escape the quote that encloses the string to prevent Python from
thinking that it has prematurely reached the end of a string. If that sounds confusing, it looks like this:
>>> ‘And he said \\’this string has escaped quotes\\’’
“And he said ‘this string has escaped quotes’”
Returning to those three examples, normally a running Python shell will show you a string that it has
evaluated in single quotes. However, if you use a single quote within a string that begins and ends with
double quotes, Python will display that string with double quotes around it to make it obvious to you
where the string starts and where it ends:
>>> ‘Ben said “How\\’re we supposed to know that?”’
‘Ben said “How\\’re we supposed to know that?”’
>>>
This shows you that there is no difference between single and double quoted strings. The only thing to
be aware of is that when you start a string with a double quote, it can’t be ended by a single quote, and
vice versa. Therefore, if you have a string that contains single quotes, you can make your life easier by
enclosing the string in double quotes, and vice versa if you’ve got strings with quotes that have been
enclosed in single quotes. SQL, the language that is used to obtain data from databases, will often have
single quoted strings inside of them that have nothing to do with Python. You can learn more about this
when you reach Chapter 14. One more important rule to know is that by themselves, quotes will not let
you create a newline in a string. The newline is the character that Python uses internally to mark the end
of a line. It’s how computers know that it’s time to start a new line.
Within strings, Python has a way of representing special characters that you
normally don’t see — in fact, that may indicate an action, such as a newline, by
using sequences of characters starting with a backslash (\\). (Remember that it’s
already special because it’s the escape character and now it’s even more special.)
The newline is \\n, and it is likely the most common special character you will
encounter.
Until you see how to print your strings, you’ll still see the escaped characters look-
ing as you entered them, as \\n, instead of, say, an actual line ending, with any more
tests starting on the next line.
Python has one more special way of constructing strings, one that will almost always avoid the entire
issue of requiring an escape character and will let you put in new lines as well: the triple quote. If you
ever use a string enclosed in three quotes in a row — either single or double quotes, but all three have to
be the same kind — then you do not have to worry about escaping any single instance of a single or dou-
ble quote. Until Python sees three of the same quotes in a row, it won’t consider the string ended, and it
can save you the need to use escape characters in some situations:
>>> “””This is kind of a special string, because it violates some
...
rules that we haven’t talked about yet”””
“This is kind of a special string, because it violates some\\n rules that we
haven’t talked about yet”
7

5.9 Page 49

▲back to top
Chapter 1
As you can see here, Python enables you to do what you want in triple-quoted strings. However, it does
raise one more question: What’s that \\n doing there? In the text, you created a new line by pressing the
Enter key, so why didn’t it just print the rest of the sentence on another line? Well, Python will provide an
interpretation to you in the interest of accuracy. The reason why \\n may be more accurate than showing
you the next character on a new line is twofold: First, that’s one way for you to tell Python that you’re
interested in printing a new line, so it’s not a one-way street. Second, when displaying this kind of data,
it can be confusing to actually be presented with a new line. Without the \\n, you may not know whether
something is on a new line because you’ve got a newline character or because there are spaces that lead
up to the end of the line, and the display you’re using has wrapped around past the end of the current
line and is continued on the next line. By printing \\n, Python shows you exactly what is happening.
Putting Two Strings Together
Something that you are probably going to encounter more than a few times in your programming
adventures is multiple strings that you want to print at once. A simple example is when you have sepa-
rate records of a person’s first name and last name, or their address, and you want to put them together.
In Python, each one of these items can be treated separately, as shown here:
>>> “John”
‘John’
>>> “Q.”
‘Q.’
>>> “Public”
‘Public’
>>>
Try It Out Using + to Combine Strings
To put each of these distinct strings together, you have a couple of options. One, you can use Python’s
own idea of how strings act when they’re added together:
>>> “John” + “Q.” + “Public”
‘JohnQ.Public’
How It Works
This does put your strings together, but notice how this doesn’t insert spaces the way you would expect
to read a person’s name; it’s not readable, because using the plus sign doesn’t take into account any con-
cepts of how you want your string to be presented.
You can easily insert spaces between them, however. Like newlines, spaces are characters that are treated
just like any other character, such as A, s, d, or 5. Spaces are not removed from strings, even though they
can’t be seen:
>>> “John” + “ “ + “Q.” + “ “ + “Public”
‘John Q. Public’
After you determine how flexible you need to be, you have a lot of control and can make decisions about
the format of your strings.
8

5.10 Page 50

▲back to top
Programming Basics and Strings
Putting Strings Together in Different Ways
Another way to specify strings is to use a format specifier. It works by putting in a special sequence of
characters that Python will interpret as a placeholder for a value that will be provided by you. This may
initially seem like it’s too complex to be useful, but format specifiers also enable you to control what the
displayed information looks like, as well as a number of other useful tricks.
Try It Out Using a Format Specifier to Populate a String
In the simplest case, you can do the same thing with your friend, John Q.:
>>> “John Q. %s” % (“Public”)
‘John Q. Public’
How It Works
That %s is the format specifier for a string. Several other specifiers will be described as their respective
types are introduced. Each specifier acts as a placeholder for that type in the string; and after the string,
the % sign outside of the string indicates that after it, all of the values to be inserted into the format speci-
fier will be presented there to be used in the string.
You may be wondering why the parentheses are there. The parentheses indicate to the string that it
should expect to see a sequence that contains the values to be used by the string to populate its format
specifiers.
Sequences are a very important part of programming in Python, and they are covered in some detail
later. For now, we are just going to use them. What is important to know at this point is that every for-
mat specification in a string has to have an element that matches it in the sequence that’s provided to it.
The items we are putting in the sequence are strings that are separated by commas (if there is more than
one). If there is only one, as in the preceding example, the sequence isn’t needed, but it can be used.
The reason why this special escape sequence is called a format specifier is because you can do some
other special things with it — that is, rather than just insert values, you can provide some specifications
about how the values will be presented, how they’ll look.
Try It Out More String Formatting
You can do a couple of useful things when formatting a simple string:
>>> “%s %s %10s” % (“John”, “Q.”, “Public”)
‘John Q.
Public’
>>> “%-10s %s %10s” % (“John”, “Q.”, “Public”)
‘John
Q.
Public’
How It Works
In the first string, the reason why Public is so alone along the right side is because the third format
specifier in the main string, on the left side, has been told to make room for something that has 10 char-
acters. That’s what the %10s means. However, because the word Public only has 6 characters, Python
padded the string with space for the remaining four characters that it had reserved.
9

6 Pages 51-60

▲back to top

6.1 Page 51

▲back to top
Chapter 1
In the second string, the Q. is stranded in the middle, with John and Public far to either side. The
behavior on its right-hand side has just been explained. The behavior on its left happens for very sim-
ilar reasons. An area with 10 spaces has been created in the string, but this string was specified with a
%-10s. The - in that specifier means that the item should be pushed to the left, instead of to the right,
as it would normally.
Displaying Strings with Print
Up until now, you have seen how Python represents the strings you type, but only how it represents
them internally. However, you haven’t actually done anything that your program would show to a user.
The point of the vast majority of programs is to present users with information — programs produce
everything from sports statistics to train schedules to web pages to automated telephone voice response
units. The key point is that they all have to make sense to a person eventually.
Try It Out Printing Text with Print
For displaying text, a special feature is built into useful languages, one that helps the programmer dis-
play information to users. The basic way to do this in Python is by using the print function:
>>> print “%s %s %10s” % (“John”, “Q.”, “Public”)
John Q.
Public
>>>
You’ll notice that there are no longer any quotes surrounding the first, middle, and last name. In this
case, it’s significant — this is the first thing that you’ve done that would actually be seen by someone
using a program that you’ve written!
How It Works
print is a function — a special name that you can put in your programs that will perform one or more
tasks behind the scenes. Normally, you don’t have to worry about how it happens. (When you start writ-
ing your own functions in Chapter 5, you’ll naturally start to think more about how this works.)
In this case, the print function is an example of a built-in function, which is a function included as a
part of Python, as opposed to a function that you or another programmer has written. The print func-
tion performs output — that is, it presents something to the user using a mechanism that they can see,
such as a terminal, a window, a printer, or perhaps another device (such as a scrolling LED display).
Related routines perform input, such as getting information from the user, from a file, from the network,
and so on. Python considers these input/output (I/O) routines. I/O commonly refers to anything in a
program that prints, saves, goes to or from a disk, or connects to a network. You will learn more about
I/O in Chapter 8.
Summary
In this chapter, you’ve begun to learn how to use the programming editor codeEditor, which is a pro-
gram written in Python for the purpose of editing Python programs. In addition to editing files,
10

6.2 Page 52

▲back to top
Programming Basics and Strings
codeEditor can run a Python shell, where you can experiment with simple Python programming lan-
guage statements.
Within the shell, you have learned the basics of how to handle strings, including adding strings together
to create longer strings as well as using format specifiers to insert one or more strings into another string
that has format specifiers. The format specifier %s is used for strings, and it can be combined with num-
bers, such as %8s, to specify that you want space for eight characters — no more and no less. In later
chapters, you will learn about other format specifiers that work with other types.
You also learned how to print strings that you have created. Printing is a type of input/output operation
(input/output is covered in more detail in Chapter 8). Using the print function, you can present users
of your program with strings that you have created.
In the next chapter, you will learn about dealing with simple numbers and the operations that you can
perform on them, as well as how to combine numbers and strings so that print can render numbers dis-
playable. This technique of using format specifiers will enable you to display other types of data as well.
Exercises
1. In the Python shell, type the string, "Rock a by baby,\\n\\ton the tree top,\\t\\twhen the wind
blows\\n\\t\\t\\t the cradle will drop." Experiment with the number and placement of the \\t and
\\n escape sequences and see how this affects what you see. What do you think will happen?
2. In the Python shell, use the same string indicated in the prior exercise, but display the string
using the print function. Again, play with the number and location of the \\n and \\t escape
sequences. What do you think will happen?
11

6.3 Page 53

▲back to top

6.4 Page 54

▲back to top
2
Numbers and Operators
When you think of numbers, you can probably invoke pleasant memories like Sesame Street and
its counting routine or more serious memories like math lessons. Either way, you are familiar with
numbers. Indeed, numbers are such a familiar concept that you probably don’t notice the many
different ways in which you use them depending on their context.
In this chapter, you will be re-introduced to numbers and some of the ways in which Python
works with them, including basic arithmetic and special string format specifiers for its different
types of numbers. When you have finished the chapter, you will be familiar with the different
basic categories of numbers that Python uses and with the methods for using them, including
displaying and mixing the various number types.
Different Kinds of Numbers
If you have ever used a spreadsheet, you’ve noticed that the spreadsheet doesn’t just look at num-
bers as numbers but as different kinds of numbers. Depending on how you’ve formatted a cell, the
spreadsheet will have different ways of displaying the numbers. For instance, when you deal with
money, your spreadsheet will show one dollar as 1.00. However, if you’re keeping track of the
miles you’ve traveled in your car, you’d probably only record the miles you’ve traveled in tenths
of a mile, such as 10.2. When you name a price you’re willing to pay for a new house, you proba-
bly only think to the nearest thousand dollars. At the large end of numbers, your electricity bills
are sent to you with meter readings that come in at kilowatt hours, which are each one thousand
watts per hour.
What this means in terms of Python is that, when you want to use numbers, you sometimes need
to be aware that not all numbers relate to each other (as you’ll see with imaginary numbers in this
chapter), and sometimes you’ll have to be careful about what kind of number you have and what
you’re trying to do with it. However, in general, you will use numbers in two ways: The first way
will be to tell Python to repeat a certain action, while the second way will be to represent things
that exist in the real world (that is, in your program, which is trying to model something in the real
world). You will rarely have to think of numbers as anything besides simple numbers when you
are counting things inside of Python. However, when you move on to trying to solve problems

6.5 Page 55

▲back to top
Chapter 2
that exist in the real world — things that deal with money, science, cars, electricity, or anything else,
you’ll find yourself more aware about how you use numbers.
Numbers in Python
Python offers four different kinds of numbers with which you can work: integers, long numbers
(or longs), floating-point numbers (or floats), and imaginary numbers.
Integers and longs are very closely related and can be mixed freely. Each one is a whole number, positive
or negative, but plain integers only run between –2,147,483,648 and +2,147,483,647. That’s pretty big —
big enough for a lot of tasks. However, if you find that you need more than that, Python will notice this
and automatically promote your number from a plain integer to a long number.
To determine the type of a number, you can use a special function that is built into Python, called type.
When you use type, Python will tell you what kind of data you’re looking at. Let’s try this with a few
examples.
Try It Out Using Type with Different Numbers
In the codeEditor’s Python shell, you can enter different numbers and see what type tells you about
how Python sees them:
>>> type(1)
<type ‘int’>
>>> type(2000)
<type ‘int’>
>>> type(999999999999)
<type ‘long’>
>>> type(1.0)
<type ‘float’>
How It Works
Although in everyday life 1.0 is the same number as 1, Python will automatically perceive 1.0 as being a
float; without the .0, the number 1 would be dealt with as the integer number one (which you probably
learned as a whole number in grade school), which is a different kind of number.
In essence, the special distinction between a float and an integer or a long integer is that a float has a
component that is a fraction of 1. Numbers such as 1.01, 2.34, 0.02324, and any other number that con-
tains a fractional component is treated as a floating-point number (except for imaginary numbers, which
have rules of their own). This is the type that you would want to use for dealing with money or with
things dealt with in partial quantities, like gasoline or pairs of socks. (There’s always a stray single sock
in the drawers, right?)
The last type of number that Python offers is oriented toward engineers and mathematicians. It’s the
imaginary number, and you may remember it from school; it’s defined as the square root of –1. Despite
being named imaginary, it does have a lot of practical uses in modeling real-world engineering situa-
tions, as well as in other disciplines like physics and pure math. The imaginary number is built into
Python so that it’s easily usable by user communities who frequently need to solve their problems with
computers. Having this built-in type enables Python to help them do that. If you happen to be one of
those people, you will be happy to learn that you’re not alone, and Python is there for you.
14

6.6 Page 56

▲back to top
Numbers and Operators
A word to the wise: Numbers can be tricky
Experts in engineering, financial, and other fields who deal with very large and very
small numbers (small with a lot of decimal places) need even more accuracy and con-
sistency than what built-in types like floats offer. If you’re going to explore these disci-
plines within programming, you should use the available modules, a concept
introduced in Chapter 7, which are written to handle the types of issues pertinent to
the field in which you’re interested. Or at least using modules that are written to han-
dle high-precision floating-point values in a manner that is specifically different than
the default behavior is worth investigating if you’ve got the need for them.
Try It Out Creating an Imaginary Number
The imaginary number behaves very much like a float, except that it cannot be mixed with a float. When
you see an imaginary number, it will have the letter j trailing it:
>>> 12j
12j
How It Works
When you use the letter j next to a number and outside the context of a string (that is, not enclosed in
quotes), Python knows that you’ve asked it to treat the number you’ve just entered as an imaginary
number. When any letter appears outside of a string, it has to have a special meaning, such as this modi-
fier, which specifies the type of number, or a named variables (which you’ll see in Chapter 3), or another
special name. Otherwise, the appearance of a letter by itself will cause an error!
You can combine imaginary and nonimaginary numbers to create complex numbers:
>>> 12j + 1
(1+12j)
>>> 12j + 1.01
(1.01+12j)
>>> type (12j + 1)
<type ‘complex’>
You can see that when you try to mix imaginary numbers and other numbers, they are not added
(or subtracted, multiplied, or divided); they’re kept separate, in a way that creates a complex number.
Complex numbers have a real part and an imaginary part, but an explanation of how they are used
is beyond the scope of this chapter, although if you’re someone who needs to use them, the complex
number module (that word again!) is something that you can explore once you’ve gotten through
Chapter 6. The module’s name is cmath, for complex math. Complex numbers are discussed further
in Chapter 19.
Program Files
At this point, it’s worth looking at the codeEditor environment again. You have already used the
codeEditor’s Python shell to enter examples, and you have looked at the examples and questions in
15

6.7 Page 57

▲back to top
Chapter 2
Chapter 1. The Python shell, however, is not the main part of codeEditor. Its main use is what its name
suggests, a programmer’s editor.
For the remainder of this chapter, you will be encouraged to use the Python shell along with its editor
functionality to save your work as you go along.
Try It Out Using the Shell with the Editor
Enter the following in the Python shell window:
>>> print “This is another string”
This is another string
>>> print “Joining two strings with “ + “the plus operation”
Joining two strings with the plus operation
>>> print “This is an example of including %s %s” % (“two strings”, “together”)
This is an example of including two strings together
Select the entire area from the first print statement to the final print statement. Now, in the editor win-
dow, select Save Shell Selection from the Shell menu (see Figure 2-1).
Figure 2-1
After you’ve selected a filename and saved the file, you can reopen it. You need to remove the output —
the printed lines. After you’ve done this, you can rerun the remaining lines any time by selecting Run
from the File menu.
16

6.8 Page 58

▲back to top
Numbers and Operators
You will see that everything in quotes has been colored; this is codeEditor’s way of indicating that this is
a string, including where it begins and where it ends (see Figure 2-2).
Figure 2-2
Do this a few more times with different strings, saving them in different files. Each one of these sessions
is now available for you, and you can refer to them later.
Using the Different Types
Except for the basic integer, the other number types can grow to an unwieldy number of digits to look at
and make sense of. Therefore, very often when these numbers are generated, you will see them in a format
that is similar to scientific notation. Python will let you input numbers in this format as well, so it’s a two-
way street. There are many snags to using very large long integers and floats. The topic is quite detailed
and not necessarily pertinent to learning Python. If you want to know more about floating points numbers
in general, and what they really mean to a computer, the paper at http://docs.sun.com/source/
806-3568/ncg_goldberg.html is a very good reference, although the explanation will only make sense
to someone with prior experience with computers and numbers. Don’t let that stop you from looking,
though. It may be something you want to know about at some point in the future.
More commonly, you will be using integers and floats. It wouldn’t be unusual to acquire a number from
somewhere such as the date, the time, or information about someone’s age or the time of day. After that
data, in the form of a number, is acquired, you’ll have to display it.
The usual method of doing this is to incorporate numbers into strings. You can use the format specifier
method that was used in Chapter 1. It may make intuitive sense to you that you should also be able to
use the + method for including a number in a string, but in fact this does not work, because deep down
they are different types, and the + operator is intended for use only with two things of the same type:
two strings, two numbers, or two other objects and types that you will encounter later. The definite
exceptions are that floats, longs, and integers can be added together. Otherwise, you should expect that
different types won’t be combined with the + operation.
You are likely wondering why a string format specifier can be used to include a number, when a + can’t.
The reason is that the + operation relies on information contained in the actual items being added.
Almost everything you use in Python can be thought of as an object with properties, and all of the
17

6.9 Page 59

▲back to top
Chapter 2
properties combined define the object. One important property of every object is its type, and for now
the important thing to understand about a type is that certain naturally understood things like the +
operation work only when you perform them with two objects of compatible types. In most cases,
besides numbers, compatible types should be thought of as the same type.
If you do want to use the + operation with numbers and strings (and doing this is usually a matter of
style that you can decide for yourself), you can use a built-in function called str that will transform, if
possible, numbers into a string. It enables you to do things such as add strings and numbers into a sin-
gle string. You can use str with most objects because most objects have a way of displaying themselves
as strings. However, for the sake of consistency, you’ll use string format specifiers for now.
Try It Out Including Different Numbers in Strings
When you combined two strings in the first chapter by using a format specifier, you used the format
specifier %s, which means “a string.” Because numbers and strings have different types, you will use a
different specifier that will enable your numbers to be included in a string:
>>> “Including an integer works with %%d like this: %d” % 10
‘Including an integer works with %d like this: 10’
>>> “An integer converted to a float with %%f: %f” % 5
‘An integer converted to a float with %f: 5.000000’
>>> “A normal float with %%f: %f” % 1.2345
‘A normal float with %f: 1.234500’
>>> “A really large number with %%E: %E” % 6.789E10
‘A really large number with %E: 6.789000E+10’
>>> “Controlling the number of decimal places shown: %.02f” % 25.101010101
‘Controlling the number of decimal places shown: 25.10’
If you’re wondering where you can use format specifiers, note that the last example looks very similar to
the way we print monetary values, and, in fact, any program that deals with dollars and cents will need
to have at least this much capability to deal with numbers and strings.
How It Works
Anytime you are providing a format specifier to a string, there may be options that you can use to con-
trol how that specifier displays the value associated with it. You’ve already seen this with the %s speci-
fier in Chapter 1, where you could control how many characters were displayed. With numeric specifiers
are also conventions regarding how the numbers of a particular type should be displayed. These con-
ventions result in what you see when you use any of the numeric format specifiers.
Try It Out Escaping the % Sign in Strings
One other trick was shown before. In case you wanted to print the literal string %d in your program, you
achieve that in Python strings by using two % signs together. This is needed only when you also have
valid format specifiers that you want Python to substitute for you in the same string:
print “The %% behaves differently when combined with other letters, like this: %%d
%%s %%f %d” % 10
The % behaves differently when combined with other letters, like this: %d %s %f 10
18

6.10 Page 60

▲back to top
Numbers and Operators
How It Works
Note that Python pays attention to the combinations of letters and will behave correctly in a string that
has both format specifiers as well as a double percent sign.
Basic Math
It’s more common than not that you’ll have to use the numbers in your program in basic arithmetic.
Addition, subtraction, division, and multiplication are all built in. Addition and subtraction are per-
formed by the + and – symbols.
Try It Out Doing Basic Math
You can enter basic arithmetic at the Python shell prompt and use it like a calculator. Like a calculator,
Python will accept a set of operations, and when you hit the Enter key, it will evaluate everything you’ve
typed and give you your answer:
>>> 5 + 300
305
>>> 399 + 3020 + 1 + 3456
6876
>>> 300 - 59994 + 20
-59674
>>> 4023 - 22.46
4000.54
How It Works
Simple math looks about how you’d expect it to look. In addition to + and –, multiplication is performed
by the asterisk, *, and division is performed by the forward slash, /. Multiplication and division may not
be as straightforward as you’d expect in Python, because of the distinction between floating point num-
bers and whole numbers.
Also, you can see below that numbers will be automatically promoted as they become larger and
larger — for instance, from integer to as long as needed:
>>> 2000403030 * 392381727
784921595607432810L
>>> 2000403030 * 3923817273929
7849215963933911604870L
>>> 2e304 * 3923817273929
inf
>>> 2e34 * 3923817273929
7.8476345478579995e+46
Note that while Python can deal with some very large numbers, the results of some operations will
exceed what Python can accommodate. The shorthand for infinity, inf, is what Python will return when
a result is larger than what it can handle.
Division is also interesting. Without help, Python won’t coax one kind of number into another through
division. Only when you have at least one number that has a floating-point component — that is, a
19

7 Pages 61-70

▲back to top

7.1 Page 61

▲back to top
Chapter 2
period followed by a number — will floating-point answers be displayed. If two numbers that are nor-
mal integers or longs (in either case, lacking a component that specifies a value less than one, even if that
is .0) are divided, the remainder will be discarded:
>>> 44 / 11
4
>>> 324 / 101
3
>>> 324.0 / 101.0
3.2079207920792081
>>> 324.0 / 101
3.2079207920792081
Try It Out Using the Modulus Operation
There is one other basic operation of Python that you should be aware of: the remainder, or modulus
operation. When you try to do division, like the 324/101 in the preceding example, Python will return
only the whole number portion of the result: the 3. To find out the rest of the answer, you have to use the
modulus operator, which is the %. Don’t let this confuse you! The % means modulus only when it is used
on numbers. When you are using strings, it retains its meaning as the format specifier. When something
has different meanings in different contexts, it is called overloading, and it is very useful; but don’t get
caught by surprise when something behaves differently by design.
>>> 5 / 3
1
>>> 5 % 3
2
How It Works
The preceding code indicates that 5 divided by 3 is 1 and ⁄23. One very useful task the modulus operator
is used for is to discover whether one thing can be evenly divided by another, such as determining
whether the items in one sequence will fit into another evenly (you will learn more about sequences in
Chapter 3). Here are some more examples that you can try out:
>>> 123 % 44
35
>>> 334 % 13
9
>>> 652 % 4
0
Some Surprises
You need to be careful when you are dealing with common floating-point values, such as money.
Some things in Python are puzzling. For one thing, if you manipulate certain numbers with seemingly
straightforward math, you may still receive answers that have extra values trailing them, such as the
following:
>>> 4023 - 22.4
4000.5999999999999
20

7.2 Page 62

▲back to top
Numbers and Operators
The trailing nines could worry you, but they merely reflect the very high precision that Python offers.
However, when you print or perform math, this special feature actually results in precise answers.
Try It Out Printing the Results
Try actually printing the results, so that the preceding math with the unusual-looking results has its
results displayed to a user, as it would from inside of a program:
>>> print “%f” % (4023 - 22.4)
4000.600000
How It Works
Floating-point numbers can be confusing. A complete discussion of floating-point numbers is beyond
the scope of this book, but if you are experienced with computers and numbers and want to know more
about floating-point numbers, read the paper at http://docs.sun.com/source/806-3568/ncg_
goldberg.html. The explanation offered there should help round out this discussion.
Using Numbers
You can display numbers with the print function by including the numbers into strings, for instance by
using a format specifier. The important point is that you must determine how to display your numbers
so that they mean what you intend them to mean, and that depends on knowing your application.
Order of Evaluation
When doing math, you may find yourself looking at an expression like 4*3+1/4–12. The puzzle you’re
confronted with is determining how you’re going to evaluate this sort of expression and whether the way
you would evaluate it is the same way that Python would evaluate it. The safest way to do this is to
always enclose your mathematical expressions in parentheses, which will make it clear which math
operations will be evaluated first.
Python evaluates these basic arithmetic operations as follows: Multiplication and division operations
happen before addition and subtraction, but even this can become confusing.
Try It Out Using Math Operations
When you’re thinking about a particular set of mathematical operations, it can seem straightforward
when you write it down (or type it in). When you look at it later, however, it can become confusing. Try
these examples, and imagine them without the parentheses:
>>> (24 * 8)
192
>>> (24 * (8 + 3))
264
>>> (24 * (8 + 3 + 7.0))
432.0
>>> (24 * (8 + 3 + 7.0 + 9))
21

7.3 Page 63

▲back to top
Chapter 2
648.0
>>> (24 * (8 + 3 + 7.0 + 9))/19
34.10526315789474
>>> (24 * (8 + 3 + 7 + 9))/19
34
>>> (24 * (8 + 3 + 7 + 9))%19
2
Notice in the examples here how the presence of any floating-point numbers changes the entire equation
to using floating-point numbers, and how removing any floating-point numbers will cause Python to
evaluate everything as integers (or longs for larger numbers).
How It Works
The examples are grouped in something that resembles the normal order of evaluation, but the paren-
theses ensure that you can be certain which groups of arithmetic operations will be evaluated first. The
innermost (the most contained) are evaluated first, and the outermost last. Within a set of parentheses,
the normal order takes place.
Number Formats
When you prepare strings to contain a number, you have a lot of flexibility. In the following Try It Out,
you’ll see some examples.
For displaying money, use a format specifier indicating that you want to limit the number of decimal
places to two.
Try It Out Using Number Formats
Try this, for example. Here, you print a number as though you were printing a dollar amount:
>>> print “$%.02f” % 30.0
$30.00
You can use a similar format to express values less than a cent, such as when small items are listed for
sale individually. When you have more digits than you will print, notice what Python does:
>>> print “$%.03f” % 30.00123
$30.001
>>> print “$%.03f” % 30.00163
$30.002
>>> print “%.03f” % 30.1777
30.178
print “%.03f” % 30.1113
>>> 30.111
How It Works
As you can see, when you specify a format requiring more accuracy than you have asked Python to dis-
play, it will not just cut off the number. It will do the mathematically proper rounding for you as well.
22

7.4 Page 64

▲back to top
Numbers and Operators
Mistakes Will Happen
While you are entering these examples, you may make a mistake. Obviously, there is nothing that Python
can do to help you if you enter a different number; you will get a different answer than the one in this
book. However, for mistakes such as entering a letter as a format specifier that doesn’t mean anything to
Python or not providing enough numbers in a sequence you’re providing to a string’s format specifiers,
Python tries to give you as much information as possible to indicate what’s happened so that you can
fix it.
Try It Out Making Mistakes
To understand what’s going on when a mistake happens, here are some examples you can try. Their full
meanings are covered later, starting in Chapter 4, but in the meantime, you should know this.
>>> print “%.03f” % (30.1113, 12)
Traceback (most recent call last):
File “<input>”, line 1, in ?
TypeError: not all arguments converted during string formatting
How It Works
In the preceding code, there are more elements in the sequence (three in all) than there are format speci-
fiers in the string (just two), so Python helps you out with a message. What’s less than helpful is that this
mistake would cause a running program to stop running, so this is normally an error condition, or an
exception. The term arguments here refers to the format specifiers but is generally used to mean param-
eters that are required in order for some object to work. When you call a function that expects a certain
number of values to be specified, each one of those anticipated values is called an argument.
This is something that programmers take for granted; this specialized technical language may not make
sense immediately, but it will begin to feel right when you get used to it. Through the first ten chapters
of this book, arguments will be referred to as parameters to make them less puzzling, since no one is
arguing, just setting the conditions that are being used at a particular point in time. When you are pro-
gramming, though, the terms are interchangeable.
Here is another potential mistake:
>>> print “%.03f, %f %d” % (30.1113, 12)
Traceback (most recent call last):
File “<input>”, line 1, in ?
TypeError: not enough arguments for format string
Now that you know what Python means by an argument, it makes sense. You have a format specifier,
and you don’t have a value in the accompanying sequence that matches it; thus, there aren’t enough
parameters.
If you try to perform addition with a string and a number, you will also get an error:
“This is a string” + 4
Traceback (most recent call last):
File “<input>”, line 1, in ?
TypeError: cannot concatenate ‘str’ and ‘int’ objects
23

7.5 Page 65

▲back to top
Chapter 2
This should make sense because you’ve already read about how you can and can’t do this. However,
here is definite proof: Python is telling you clearly that it can’t do what it has been asked to do, so now
it’s up to you to resolve the situation. (Hint: You can use the str function.)
Some Unusual Cases
There is one other feature that Python offers with its numbers that is worth knowing about so that you
understand it when you encounter it. The normal counting system that we use is called base 10, or radix
10. It included numbers from 0 to 9. Numbers above that just involve combining 0 through 9. However,
computers commonly represent the binary numbers they actually deal with in base 8, called octal, and
base 16, also called hexadecimal. These systems are often used to give programmers an easier way to
understand bytes of data, which often come in one and two chunks of 8 bits.
In addition, neither octal nor hexadecimal can be displayed as negative numbers. Numbers described in
this way are said to be unsigned, as opposed to being signed. The sign that is different is the + or – sign.
Normally, numbers are assumed to be positive, but if a number is a signed type, it can be negative as
well. If a number is unsigned, it has to be positive; and if you ask for the display of a negative number
but in a signed format string, you’ll get unusual answers.
Try It Out Formatting Numbers as Octal and Hexadecimal
>>> print ‘Octal uses the letter “o” lowercase. %d %o’ % (10, 10)
Octal uses the letter “o” lowercase. 10 12
It may seem like a mistake that the second number printed is 12 when you’ve provided the string with a
10. However, octal only has 8 numbers (0 to 7), so from 0 to 10 in octal is 0, 1, 2, 3, 4, 5, 6, 7, 8, 10, 11.
print ‘Hex uses the letter “x” or “X”. %d %x %X’ % (10, 10, 10)
Hex uses the letter “x” or “X”. 10 a A
Here is another case that needs explaining. Hexadecimal uses numbers from 0 to 15, but because you
run out of numbers at 9, hex utilizes the letters a–f; and the letters are lowercase if you used the format
specifier %x and is capitalized if you used %X. Therefore, the numbers 0 to 20 in decimal are as follows in
hex: 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, a, b, c, d, e, f, 10, 11, 12, 13.
Summary
This chapter introduced you to numbers in Python, although it doesn’t cover everything available.
You’ve seen and used the four kinds of built-in numbers that Python knows about: integers, longs,
floats, and imaginary numbers. You have learned how to use string format specifiers to allow you to
include numbers in your strings, and you’ve formatted those numbers in strings with different styles.
The format, or how the number is displayed in a string, doesn’t change the value of the number. Floats
remain floats even when they are printed as integers, and longs remain longs even when they are
printed as floats.
You’ve performed the major built-in arithmetical operations: addition, subtraction, multiplication,
division, and modulus. If integers and longs are mixed in an arithmetic calculation, the result is a long.
24

7.6 Page 66

▲back to top
Numbers and Operators
If integers and longs are mixed with a float, the result is a float. If arithmetic is done with an integer,
long, or a float combined with an imaginary number, the result will be a complex number that separates
the real component and the imaginary component. You’ve also learned about the type function, which
enables you to determine what type of number you actually have.
Lastly, we generally use numbers in base 10, or radix 10. Computers in general, and Python in particular,
can easily translate numbers to base 8, or octal, and base 16, or hexadecimal.
Exercises
Do the following four exercises in codeEditor and save the results in a file called ch2_exercises.py.
You can run it from within codeEditor by selecting and dragging everything into the Python shell.
1. In the Python shell, multiply 5 and 10. Try this with other numbers as well.
2. In the Python shell, divide 122 by 23. Now divide 122.0 by 23. Try multiplying floats and inte-
gers using other numbers.
3. Print every number from 6 through 14 in base 8.
4. Print every number from 9 through 19 in base 16.
5. Try to elicit other errors from the Python interpreter — for instance, by deliberately misspelling
print as pinrt. Notice how as you work on a file in codeEditor, it will display print differ-
ently than it does pinrt.
6. Create a string with the format specifier %u and pass a negative number to it. When Python
evaluates it, consider the answer it returns, which you may find surprising.
25

7.7 Page 67

▲back to top

7.8 Page 68

▲back to top
3
Variables — Names
for Values
In the previous two chapters, you learned how Python views strings, integers, longs, floats, and
imaginary numbers and how they can be created and displayed. This chapter presents more exam-
ples that will demonstrate how these data types can be used.
In this chapter, you will learn how to use names to store the types you already know as well as
other basic types to which you will be introduced. The same facility will enable you to work with
different types of objects that you haven’t learned about yet, too. By the end of this chapter, you
should be familiar with variables and new, different types — specifically, you will become better
acquainted with lists, tuples, and dictionaries. You will know what a reference is and have some
experience in using references.
To get the most out of this chapter, you should type the examples yourself and alter them to see
what happens.
Referring to Data — Using
Names for Data
It’s difficult to always write strings and numbers explicitly throughout a program because it forces
you to remember everything. The exacting memory that computers have enable them to remember
far more details than people can, and taking advantage of that capability is a huge part of pro-
gramming. However, to make using data more flexible and easy, you want to give the data names
that can be used to refer to them.

7.9 Page 69

▲back to top
Chapter 3
Try It Out Assigning Values to Names
These names are commonly called variables, which indicates that the data to which they refer can vary
(it can be changed), while the name remains the same. You’ll see them referred to as names, as well,
because that is what you are presented with by Python.
>>> first_string = “This is a string”
>>> second_string = “This is another string”
>>> first_number = 4
>>> second_number = 5
>>> print “The first variables are %s, %s, %d, %d” % (first_string, second_string,
first_number, second_number)
The first variables are This is a string, This is another string, 4, 5
How It Works
You can see that you can associate a name with a value — either a string or an integer — by using the
equals (=) sign. The name that you use doesn’t relate to the data to which it points in any direct sense
(that is, if you name it “number,” that doesn’t actually have to mean that it holds a number).
>>> first_string = 245
>>> second_number = “This isn’t a number”
>>> first_string
245
>>> second_number
“This isn’t a number”
The benefit of being able to name your data is that you can decide to give it a name that means some-
thing. It is always worthwhile to give your data a name that reminds you of what it contains or how you
will use it in your program. If you were to inventory the lightbulbs in your home, you might want a
piece of your program to contain a count of the lightbulbs in your closets and another piece to contain a
count of those actually in use:
>>> lightbulbs_in_closet = 10
>>> lightbulbs_in_lamps = 12
As lightbulbs are used, they can be moved from the closet into the lamps, and a name can be given to the
number of lightbulbs that have been thrown out this year, so that at the end of the year you have an idea
of what you’ve bought, what you have, and what you’ve used; and when you want to know what you
still have, you have only to refer to lightbulbs_in_closet or lightbulbs_in_lamps.
When you have names that relate to the value stored in them, you’ve created an informal index that
enables you to look up and remember where you put the information that you want so that it can be eas-
ily used in your program.
Changing Data Through Names
If your data is a number or a string, you can change it by using the operations you already know you
can do with them.
28

7.10 Page 70

▲back to top
Variables — Names for Values
Try It Out Altering Named Values
Every operation you’ve learned for numbers and strings can be used with a variable name so that you
can treat them exactly as if they were the numbers they referenced:
>>> proverb = “A penny saved”
>>> proverb = proverb + “ is a penny earned”
>>> print proverb
A penny saved is a penny earned
>>> pennies_saved = 0
>>> pennies_saved = pennies_saved + 1
>>> pennies_saved
1
How It Works
Whenever you combine named values on the right-hand side of an equals sign, the names will be oper-
ated on as though you had presented Python with the values referenced by the names, even if the same
name is on the left-hand side of the equals sign. When Python encounters a situation like that, it will first
evaluate and find the result of the operations on the right side and then assign the result to the name on
the left side. That way, there’s no confusion about how the name can exist on both sides — Python will
do the right thing.
Copying Data
The name that you give data is only a name. It’s how you refer to the data that you’re trying to access.
This means that more than one name can refer to the same data:
>>> pennies_earned = pennies_saved
>>> pennies_earned
1
When you use the = sign again, you are referring your name to a new value that you’ve created, and the
old value will still be pointed to by the other name:
>>> pennies_saved = pennies_saved + 1
>>> pennies_saved
2
>>> pennies_earned
1
Names You Can’t Use and Some Rules
Python uses a few names as special built-in words that it reserves for special use to prevent ambiguity.
The following words are reserved by Python and can’t be used as the names for data:
and, assert, break, class, continue, def, del, elif, else, except, exec, finally,
for, from, global, if, import, in, is, lambda, not, or, pass, print, raise, return,
try, while, yield
29

8 Pages 71-80

▲back to top

8.1 Page 71

▲back to top
Chapter 3
In addition, the names for data cannot begin with numbers or most non-alphabet characters (such as
commas, plus or minus signs, slashes, and so on), with the exception of the underscore character. The
underscore is allowed and even has a special meaning in some cases (specifically with classes and mod-
ules, which you’ll see in Chapter 6 and later).
You will see a number of these special reserved words in the remaining discussion in this chapter.
They’re important when you are using Python to do various tasks.
Using More Built-in Types
Beside strings and numbers, Python provides three other important basic types: tuples, lists, and diction-
aries. These three types have a lot in common because they all allow you to group more than one item of
data together under one name. Each one also gives you the capability to search through them because of
that grouping. These groupings are indicated by the presence of enclosing parentheses “()”, square brackets
“[]”, and curly braces “{}”.
When you write a program, or read someone else’s program, it is important to pay
attention to the type of enclosing braces when you see groupings of elements. The
differences among {}, [], and () are important.
Tuples — Unchanging Sequences of Data
In Chapters 1 and 2, you saw tuples (rhymes with supple) being used when you wanted to assign values
to match more than one format specifier in a string. Tuples are a sequence of values, each one accessible
individually, and a tuple is a basic type in Python. You can recognize tuples when they are created
because they’re surrounded by parentheses:
>>> print “A %s %s %s %s” % (“string”, “filled”, “by a”, “tuple”)
A string filled by a tuple
Try It Out Creating and Using a Tuple
Tuples contain references to data such as strings and numbers. However, even though they refer to data,
they can be given names just like any other kind of data.
>>> filler = (“string”, “filled”, “by a”, “tuple”)
>>> print “A %s %s %s %s” % filler
A string filled by a tuple
How It Works
You can see in the example that filler is treated exactly as though its data — the tuple with strings —
were present and being used by the string to fill in its format specifiers because the tuple was treated
exactly as though you had typed in a sequence to satisfy the format specification.
You can access a single value inside of a tuple. The value referred to by each element can be accessed
directly by using the dereference feature of the language. With tuples, you dereference the value by
30

8.2 Page 72

▲back to top
Variables — Names for Values
placing square brackets after the name of the tuple, counting from zero to the element that you’re
accessing. Therefore, the first element is 0, and the second element is 1, the third element is 2, and so on
until you reach the last element in the tuple:
>>> a = (“first”, “second”, “third”)
>>> print “The first element of the tuple is %s” % a[0]
The first element of the tuple is first
>>> print “The second element of the tuple is %s” % a[1]
The second element of the tuple is second
>>> print “The third element of the tuple is %s” % a[2]
The third element of the tuple is third
A tuple keeps track of how many elements it contains, and it can tell you when you ask it by using the
built-in function len:
>>> print “%d” % len(a)
3
This returns 3, so you need to remember that the len function starts counting at 1, but when you access
your tuple, because tuples are counted starting from zero, you must stop accessing at one less than the
number returned by len:
>>> print a[len(a) - 1]
3
You can also have one element of a tuple refer to an entirely different tuple. In other words, you can cre-
ate layers of tuples:
>>> b = (a, “b’s second element”)
Now you can access the elements of the tuple a by adding another set of brackets after the first one,
and the method for accessing the second element is no different from accessing the first — you just add
another set of square brackets.
Try It Out Accessing a Tuple Through Another Tuple
Recreate the a and b tuples so that you can look at how this works. When you have these layers of
sequences, they are sometimes referred to as multidimensional because there are two layers that can be
visualized as going down and across, like a two-dimensional grid for graph paper or a spreadsheet.
Adding another one can be thought of as being three-dimensional, like a stack of blocks. Beyond that,
though, visualizing this can give you a headache, and it’s better to look at it as layers of data.
>>> a = (“first”, “second”, “third”)
>>> b = (a, “b’s second element”)
>>> print “%s” % b[1]
b’s second element
>>> print “%s” % b[0][0]
first
>>> print “%s” % b[0][1]
second
>>> print “%s” % b[0][2]
3
31

8.3 Page 73

▲back to top
Chapter 3
How It Works
In each case, the code works exactly as though you had followed the reference in the first element of the
tuple named b and then followed the references for each value in the second layer tuple (what originally
came from the tuple a). It’s as though you had done the following:
>>> a = (“first”, “second”, “third”)
>>> b = (a, “b’s second element”)
>>> layer2 = b[0]
>>> layer2[0]
‘first’
>>> layer2[1]
‘second’
>>> layer2[2]
‘third’
Note that tuples have one oddity when they are created: To create a tuple with one element, you abso-
lutely have to follow that one element with a comma:
>>> single_element_tuple = (“the sole element”,)
Doing otherwise will result in the creation of a string, and that could be confusing when you try to
access it later.
A tuple can have any kind of data in it, but after you’ve created one it can’t be changed. It is immutable,
and in Python this is true for a few types (for instance, strings are immutable after they are created; and
operations on them that look like they change them actually create new strings).
Tuples are immutable because they are supposed to be used for ordered groups of things that will not
be changed while you’re using them. Trying to change anything in them will cause Python to complain
with an error, similar to the errors you were shown at the end of Chapter 2:
>>> a[1] = 3
Traceback (most recent call last):
File “<stdin>”, line 1, in ?
TypeError: object does not support item assignment
>>> print “%s” % a[1]
second
You can see that the error Python returns when you try to assign a value to an element in the tuple is a
TypeError, which means that this type doesn’t support the operation you asked it to do (that’s what the
equals sign does — it asks the tuple to perform an action). In this case, you were trying to get the second
element in a to refer to an integer, the number 3, but that’s not going to happen. Instead, a remains
unchanged.
An unrelated error will happen if you try to refer to an element in a tuple that doesn’t exist. If you try to
refer to the fourth element in a, you will get an error (remember that because tuples start counting their
elements at zero, the fourth element would be referenced using the number three):
>>> a[3]
Traceback (most recent call last):
File “<stdin>”, line 1, in ?
IndexError: tuple index out of range
32

8.4 Page 74

▲back to top
Variables — Names for Values
Note that this is an IndexError and that the explanation of the error is provided (although it doesn’t tell
you what the index value that was out of range was, you do know that you tried to access an element
using an index value that doesn’t exist in the tuple). To fix this in a program, you would have to find out
what value you were trying to access and how many elements are in the tuple. Python makes finding
these errors relatively simple compared to many other languages that will fail silently.
Lists — Changeable Sequences of Data
Lists, like tuples, are sequences that contain elements referenced starting at zero. Lists are created by
using square brackets:
>>> breakfast = [ “coffee”, “tea”, “toast”, “egg” ]
Try It Out Viewing the Elements of a List
The individual elements of a list can be accessed in the same way as tuples. Like tuples, the elements in a
list are referenced starting at 0 and are accessed in the same order from zero until the end.
>>> count = 0
>>> print “Todays breakfast is %s” % breakfast[count]
Todays breakfast is coffee
>>> count = 1
>>> print “Todays breakfast is %s” % breakfast[count]
Todays breakfast is tea
>>> count = 2
>>> print “Todays breakfast is %s” % breakfast[count]
Todays breakfast is toast
>>> count = 3
>>> print “Todays breakfast is %s” % breakfast[count]
Todays breakfast is egg
How It Works
When you are accessing more than one element of a list, one after the other, it is essential to use a name
to hold the value of the numbered position where you are in the list. In simple examples like this, you
should do it to get used to the practice, but in practice, you will always do this. Most often, this is done
in a loop to view every element in a sequence (see Chapter 4 for more about loops).
Here, you’re manually doing the work of increasing the value referred to by count to go through each
element in the breakfast list to pull out the special for four days of the week. Because you’re increasing
the count, whatever number is referred to by count is the element number in the breakfast list that is
accessed.
The primary difference in using a list versus using a tuple is that a list can be modified after it has been
created. The list can be changed at any time:
>>> breakfast[count] = “sausages”
>>> print “Todays breakfast is %s” % breakfast[count]
Todays breakfast is sausages
33

8.5 Page 75

▲back to top
Chapter 3
You don’t just have to change elements that already exist in the list, you can also add elements to the list
as you need them. You can add elements at the end by using the append method that is built in to the
list type. Using append enables you to append exactly one item to the end of a list:
>>> breakfast.append(“waffle”)
>>> count = 4
>>> print “Todays breakfast is %s” % breakfast[count]
Todays breakfast is waffle
If you wanted to add more than one item to the end of a list — for instance, the contents of a tuple or of
another list — you can use the extend method to append the contents of a list all at once. The list isn’t
included as one item in one slot; each element is copied from one list to the other:
>>> breakfast.extend([“juice”, “decaf”, “oatmeal”])
>>> breakfast
[‘coffee’, ‘tea’, ‘toast’, ‘egg’, ‘waffle’, ‘juice’, ‘decaf’, ‘oatmeal’]
As with tuples, you can’t ask for an element beyond the end of a list, but the error message is slightly
different from a tuple because the error will tell you that it’s a list index that’s out of range, instead of a
tuple index that’s out of range:
>>> count = 5
>>> print “Todays breakfast is %s” % breakfast[count]
Traceback (most recent call last):
File “<stdin>”, line 1, in ?
IndexError: list index out of range
The length of an array can also be determined by using the len function. Just like tuples, lengths start at
one, whereas the first element of a list starts at zero. It’s important to always remember this.
Dictionaries — Groupings of Data Indexed by Name
A dictionary is similar to lists and tuples. It’s another type of container for a group of data. However,
whereas tuples and lists are indexed by their numeric order, dictionaries are indexed by names that you
choose. These names can be letters, numbers strings, or symbols — whatever suits you.
Try It Out Making a Dictionary
Dictionaries are created using the curly braces. To start with, you can create the simplest dictionary,
which is an empty dictionary, and populate it using names and values that you specify one per line:
>>> menus_specials = {}
>>> menus_specials[“breakfast”] = “canadian ham”
>>> menus_specials[“lunch”] = “tuna surprise”
>>> menus_specials[“dinner”] = “Cheeseburger Deluxe”
How It Works
When you first assign to menus_specials, you’re creating an empty dictionary with the curly braces.
Once the dictionary is defined and referenced by the name, you may start to use this style of specifying the
34

8.6 Page 76

▲back to top
Variables — Names for Values
name that you want to be the index as the value inside of the square brackets, and the values that will be
referenced through that index are on the right side of the equals sign. Because they’re indexed by names
that you choose, you can use this form to assign indexes and values to the contents of any dictionary that’s
already been defined.
When you’re using dictionaries, there are special names for the indexes and values. Index names in dic-
tionaries are called keys, and the values are called, well, values. To create a fully specified (or you can
think of it as a completely formed) dictionary — one with keys and values assigned at the outset — you
have to specify each key and its corresponding value, separated by a colon, between the curly braces. For
example, a different day’s specials could be defined all at once:
>>> menu_specials = {“breakfast” : “sausage and eggs”,
...
“lunch” : “split pea soup and garlic bread”,
...
“dinner”: “2 hot dogs and onion rings”}
To access any of the values, you use square brackets with the name of the key enclosed in the brackets. If
the key is a string, the key has to be enclosed in quotes. If the key is a number (you can use numbers,
too, making a dictionary look a lot like a list or a tuple), you need only the bare number.
>>> print “%s” % menu_specials[“breakfast”]
sausage and eggs
>>> print “%s” % menu_specials[“lunch”]
split pea soup and garlic bread
>>> print “%s” % menu_specials[“dinner”]
2 hot dogs and onion rings
If a key that is a string is accidentally not enclosed in quotes when you try to use it within square brack-
ets, Python will try to treat it as a name that should be dereferenced to find the key. In most cases, this
will raise an exception — a NameError — unless it happens to find a name that is the same as the string,
in which case you will probably get an IndexError from the dictionary instead!
Try It Out Getting the Keys from a Dictionary
Dictionaries can tell you what all of their keys are, or what all of their values are, if you know how to ask
them. The keys method will ask the dictionary to return all of its keys to you as a list so that you can
examine them for the key (or keys) you are looking for, and the values method will return all of the val-
ues as a list.
>>> menu_specials.keys()
[‘lunch’, ‘breakfast’, ‘dinner’]
>>> menu_specials.values()
[‘split pea soup and garlic bread’, ‘sausage and eggs’, ‘2 hot dogs and onion
rings’]
How It Works
Both the keys and values methods return lists, which you can assign and use like any normal list.
When you have the items in a list from the keys method, you can use the items in the list, which are
keys, to get their matching values from that dictionary. Note that while a particular key will lead you
to a value, you cannot start with a value and reliably find the key associated with it. You try to find the
key when you know only a value; you need to exhaustively test all the possible keys to find a matching
value, and even then, two different keys can have the same values associated with them.
35

8.7 Page 77

▲back to top
Chapter 3
In addition, the way that dictionaries work is that each key is different (you can’t have two keys that are
exactly the same), but you can have multiple duplicate values:
>>> count = 0
>>> specials_keys = menu_specials.keys()
>>> print “%s is the key, and %s is the value” % (specials_keys[count],
menu_specials[specials_keys[count]])
lunch is the key, and split pea soup and garlic bread is the value
>>> count = 1
>>> print “%s is the key, and %s is the value” % (specials_keys[count],
menu_specials[specials_keys[count]])
breakfast is the key, and sausage and eggs is the value
>>> count = 2
>>> print “%s is the key, and %s is the value” % (specials_keys[count],
menu_specials[specials_keys[count]])
dinner is the key, and 2 hot dogs and onion rings is the value
One other thing to note about a dictionary is that you can ask the list whether a particular key already
exists. You can use a special built-in method called __contains__, which will return True or False.
When you invoke a method like __contains__, you are asking a question of it, and it will answer with
its version of yes, which is True or no, which is False.
>>> menu_specials.__contains__(“test”)
False
>>> menu_specials.__contains__(“Brunch”)
False
>>> menu_specials.__contains__(“midnight specials”)
False
>>> menu_specials.__contains__(“Lunch”)
False
>>> menu_specials.__contains__(“lunch”)
True
True and False are special values that are actually 1 and 0 (1 is True, and 0 is
False), but True and False make more sense, because you are asking a question
(1 or 0 is more ambiguous than True or False). True and False are talked about a
bit later in this chapter.
Note that in versions of Python before 2.4, you’ll see 0 instead of False and 1 instead
of True in many cases, such as the previous one.
Treating a String Like a List
Python offers an interesting feature of strings. Sometimes, it is useful to be able to treat a string as
though it were a list of individual characters. It’s not uncommon to have extraneous characters at the
end of a string. People may not recognize these, but computers will get hung up on them. It’s also com-
mon to only need to look at the first character of a string to know what you want to do with it. For
instance, if you had a list of last names and first names, you could view the first letter of each by using
the same syntax that you would for a list. This method of looking at strings is called slicing and is one
of the fun things about Python:
36

8.8 Page 78

▲back to top
Variables — Names for Values
>>> last_names = [ “Douglass”, “Jefferson”, “Williams”, “Frank”, “Thomas” ]
>>> print “%s” % last_names[0]
Douglass
>>> print “%s” % last_names[0][0]
D
>>> print “%s” % last_names[1]
Jefferson
>>> print “%s” % last_names[1][0]
J
>>> print “%s” % last_names[2]
Williams
>>> print “%s” % last_names[2][0]
W
>>> print “%s” % last_names[3]
Frank
>>> print “%s” % last_names[3][0]
F
>>> print “%s” % last_names[4]
Thomas
>>> print “%s” % last_names[4][0]
T
For example, you can use the letter positioning of strings to arrange them into groups in a dictionary
based on the first letter of the last name. You don’t need to do anything complicated; you can just check
to see which letter the string containing the name starts with and file it under that:
>>> by_letter = {}
>>> by_letter[last_names[0][0]] = last_names[0]
>>> by_letter[last_names[1][0]] = last_names[1]
>>> by_letter[last_names[2][0]] = last_names[2]
>>> by_letter[last_names[3][0]] = last_names[3]
>>> by_letter[last_names[4][0]] = last_names[4]
>>> by_letter[last_names[5][0]] = last_names[5]
The by_letter dictionary will, thanks to string slicing, only contain the first letter from each of the last
names. Therefore, by_letter is a dictionary indexed by the first letter of each last name. You could also
make each key in by_letter reference a list instead and use the append method of that list to create a
list of names beginning with that letter (if, of course, you wanted to have a dictionary that indexed a
larger group of names, where each one did not begin with a different letter).
Remember that, like tuples, strings are immutable. When you are slicing strings, you are actually creating
new strings that are copies of sections of the original string.
String slicing is very useful
If you’re new to programming, string slicing may seem like an unusual feature at first.
Programmers who have used a lower-level language like C or C++ would have learned
how to program viewing strings as special lists (and in Python you can also slice lists,
as you’ll be shown), so for them this is natural. For you, it will be a very convenient
tool once you’ve learned how to control repetition over lists in Chapter 4.
37

8.9 Page 79

▲back to top
Chapter 3
Special Types
There are a handful of special types in Python. You’ve seen them all, but they bear mentioning on their
own: None, True, and False are all special built-in values that are useful at different times.
None is special because there is only one None. It’s a name that no matter how many times you use it, it
doesn’t match any other object, just itself. When you use functions that don’t have anything to return
to you — that is, when the function doesn’t have anything to respond with — it will return None.
True and False are special representations of the numbers 1 and 0. This prevents a lot of the confusion
that is common in other programming languages where the truth value of a statement is arbitrary. For
instance, in a Unix shell (shell is both how you interact with the system, as well as a programming lan-
guage), 0 is true and anything else is false. With C and Perl, 0 is false and anything else is true. However,
in all of these cases, there are no built-in names to distinguish these values. Python makes this easier by
explicitly naming the values. The names True and False can be used in elementary comparisons, which
you’ll see a lot; and in Chapter 4, you will learn how these comparisons can dramatically affect your
programs — in fact, they enable you to make decisions within your program.
>>> True
True
>>> False
False
>>> True == 1
True
>>> True == 0
False
>>> False == 1
False
>>> False == 0
True
Other Common Sequence Properties
The two types of sequences are tuples and lists; and as you’ve seen, in some cases strings can be accessed
as though they were sequences as well. Strings make sense because you can view the letters in a string as
a sequence.
Even though dictionaries represent a group of data, they are not sequences, because they do not have a
specific ordering from beginning to end, which is a feature of sequences.
Referencing the Last Elements
All of the sequence types provide you with some shortcuts to make their use more convenient. You often
need to know the contents of the final element of a sequence, and you can get that information in two
ways. One way is to get the number of elements in the list and then use that number to directly access
the value there:
38

8.10 Page 80

▲back to top
Variables — Names for Values
>>> last_names = [ “Douglass”, “Jefferson”, “Williams”, “Frank”, “Thomas” ]
>>> len(last_names)
5
>>> last_element = len(last_names) - 1
>>> print “%s” % last_names[last_element]
Thomas
However, that method takes two steps; and as a programmer, typing it repeatedly in a program can be
time-consuming. Fortunately, Python provides a shortcut that enables you to access the last element of a
sequence by using the number –1, and the next-to-last element with –2, letting you reverse the order of
the list by using negative numbers from –1 to the number that is the negative length of the list (–5 in the
case of the last_names list).
>>> print “%s” % last_names[-1]
Thomas
>>> print “%s” % last_names[-2]
Frank
>>> print “%s” % last_names[-3]
Williams
Ranges of Sequences
You can take sections of a sequence and extract a piece from it, making a copy that you can use sepa-
rately. The term for creating these groupings is called slicing (the same term used for this practice when
you did it with strings). Whenever a slice is created from a list or a tuple, the resulting slice is the same
type as the type from which it was created, and you’ve already seen this with strings. For example, a
slice that you make from a list is a list, a slice you make from a tuple is a tuple, and the slice from a string
is a string.
Try It Out Slicing Sequences
You’ve already sliced strings, so try using the same idea to slice tuples, lists, and strings and see what
the results are side-by-side:
>>> slice_this_tuple = (“The”, “next”, “time”, “we”, “meet”, “drinks”, “are”, “on”,
“me”)
>>> sliced_tuple = slice_this_tuple[5:9]
>>> sliced_tuple
(‘drinks’, ‘are’, ‘on’, ‘me’)
>>> slice_this_list = [“The”, “next”, “time”, “we”, “meet”, “drinks”, “are”, “on”,
“me”]
>>> sliced_list = slice_this_list[5:9]
>>> sliced_list
[‘drinks’, ‘are’, ‘on’, ‘me’]
>>> slice_this_string = “The next time we meet, drinks are on me”
>>> sliced_string = slice_this_string[5:9]
>>> sliced_string
‘ext ‘
39

9 Pages 81-90

▲back to top

9.1 Page 81

▲back to top
Chapter 3
How It Works
In each case, using the colon to specify a slice of the sequence instructs Python to create a new sequence
that contains just those elements.
Growing Lists by Appending Sequences
Suppose you have two lists that you want to join together. You haven’t been shown a purposely built
way to do that yet. You can’t use append to take one sequence and add it to another. Instead, you will
find that you have layered a sequence into your list:
>>> living_room = (“rug”, “table”, “chair”, “TV”, “dustbin”, “shelf”)
>>> apartment = []
>>> apartment.append(living_room)
>>> apartment
[(‘rug’, ‘table’, ‘chair’, ‘TV’, ‘dustbin’, ‘shelf’)]
This is probably not what you want if you were intending to create a list from the contents of the tuple
living_room that could be used to create a list of all the items in the apartment.
To copy all of the elements of a sequence, instead of using append, you can use the extend method of
lists and tuples, which takes each element of the sequence you give it and inserts those elements into the
list from which it is called:
>>> apartment = []
>>> apartment.extend(living_room)
>>> apartment
[‘rug’, ‘table’, ‘chair’, ‘TV’, ‘dustbin’, ‘shelf’]
Using Lists to Temporarily Store Data
You’ll often want to acquire data from another source, such as a user entering data or another computer
whose information you need. To do that, it is best to put this data in a list so that it can be processed later
in the same order in which it arrived.
However, after you’ve processed the data, you no longer need it to be in the list, because you won’t need
it again. Temporal (time-specific) information such as stock tickers, weather reports, or news headlines
would be in this category.
To keep your lists from becoming unwieldy, a method called pop enables you to remove a specific refer-
ence to data from the list after you’re done with it. When you’ve removed the reference, the position it
occupied will be filled with whatever the next element was, and the list will be reduced by as many ele-
ments as you’ve popped.
Try It Out Popping Elements from a List
You need to tell pop which element it is acting on. If you tell it to work on element 0, it will pop the first
item in its list, while passing pop a parameter of 1 will tell it to use the item at position 1 (the second ele-
ment in the list), and so on. The element pop acts on is the same number that you’d use to access the
list’s elements using square brackets:
40

9.2 Page 82

▲back to top
Variables — Names for Values
>>> todays_temperatures = [23, 32, 33, 31]
>>> todays_temperatures.append(29)
>>> todays_temperatures
[23, 32, 33, 31, 29]
>>> morning = todays_temperatures.pop(0)
>>> print “This mornings temperature was %.02f” % morning
This mornings temperature was 23.00
>>> late_morning = todays_temperatures.pop(0)
>>> print “Todays late morning temperature was %.02f” % late_morning
Todays late morning temperature was 32.00
>>> noon = todays_temperatures.pop(0)
>>> print “Todays noon temperature was %.02f” % noon
Todays noon temperature was 33.00
>>> todays_temperatures
[31, 29]
How It Works
When a value is popped, if the action is on the right-hand side of an equals sign, you can assign the ele-
ment that was removed to a value on the left-hand side, or just use that value in cases where it would be
appropriate. If you don’t assign the popped value or otherwise use it, it will be discarded from the list.
You can also avoid the use of an intermediate name, by just using pop to populate, say, a string format,
because pop will return the specified element in the list, which can be used just as though you’d speci-
fied a number or a name that referenced a number:
>>> print “Afternoon temperature was %.02f” % todays_temperatures.pop(0)
Afternoon temperature was 31.00
>>> todays_temperatures
[29]
If you don’t tell pop to use a specific element (0 in the examples) from the list it’s invoked from, it will
remove the last element of the list, not the first as shown here.
Summary
In this chapter, you learned how to manipulate many core types that Python offers. These types are
tuples, lists, dictionaries, and three special types: None, True, and False. You’ve also learned a special
way that strings can be treated like a sequence. The other sequence types are tuples and lists.
A tuple is a sequence of data that’s indexed in a fixed numeric order, starting at zero. The references in
the tuple can’t be changed after the tuple is created. Nor can it have elements added or deleted. However,
if a tuple contains a data type that has changeable elements, such as a list, the elements of that data
type are not prevented from changing. Tuples are useful when the data in the sequence is better off not
changing, such as when you want to explicitly prevent data from being accidentally changed.
A list is another type of sequence, which is similar to a tuple except that its elements can be modified.
The length of the list can be modified to accommodate elements being added using the append method,
and the length can be reduced by using the pop method. If you have a sequence whose data you want to
append to a list, you can append it all at once with the extend method of a list.
41

9.3 Page 83

▲back to top
Chapter 3
Dictionaries are yet another kind of indexed grouping of data. However, whereas lists and tuples are
indexed by numbers, dictionaries are indexed by values that you choose. To explore the indexes, which
are called keys, you can invoke the keys method. To explore the data that is referred to, called the
values, you can use the values method. Both of these methods return lists.
Other data types are True, False, and None. True and False are a special way of looking at 1 and 0,
but when you want to test whether something is true or false, explicitly using the names True and
False is always the right thing to do. None is a special value that is built into Python that only equals
itself, and it is what you receive from functions that otherwise would not return any value (such as
True, False, a string, or other values).
Exercises
Perform all of the following in the codeEditor Python shell:
1. Create a list called dairy_section with four elements from the dairy section of a supermarket.
2. Print a string with the first and last elements of the dairy_section list.
3. Create a tuple called milk_expiration with three elements: the month, day, and year of the
expiration date on the nearest carton of milk.
4. Print the values in the milk_expiration tuple in a string that reads “This milk carton will
expire on 12/10/2005”.
5. Create an empty dictionary called milk_carton. Add the following key/value pairs. You can
make up the values or use a real milk carton:
expiration_date — Set it to the milk_expiration tuple.
fl_oz — Set it to the size of the milk carton on which you are basing this.
Cost — Set this to the cost of the carton of milk.
brand_name — Set this to the name of the brand of milk you’re using.
6. Print out the values of all of the elements of the milk_carton using the values in the dictionary,
and not, for instance, using the data in the milk_expiration tuple.
7. Show how to calculate the cost of six cartons of milk based on the cost of milk_carton.
8. Create a list called cheeses. List all of the cheeses you can think of. Append this list to the
dairy_section list, and look at the contents of dairy_section. Then remove the list of
cheeses from the array.
9. How do you count the number of cheeses in the cheese list?
10. Print out the first five letters of the name of your first cheese.
42

9.4 Page 84

▲back to top
4
Making Decisions
So far, you have only seen how to manipulate data directly or through names to which the data is
bound. Now that you have the basic understanding of how those data types can be manipulated
manually, you can begin to exercise your knowledge of data types and use your data to make
decisions.
In this chapter, you’ll learn about how Python makes decisions using True and False and how to
make more complex decisions based on whether a condition is True or False.
You will learn how to create situations in which you can repeat the same actions using loops
that give you the capability to automate stepping through lists, tuples, and dictionaries. You’ll
also learn how to use lists or tuples with dictionaries cooperatively to explore the contents of a
dictionary.
You will also be introduced to exception handling, which enables you to write your programs to
cope with problematic situations that you can handle within the program.
Comparing Values — Are They the Same?
You saw True and False in Chapter 3, but you weren’t introduced to how they can be used. True
and False are the results of comparing values, asking questions, and performing other actions.
However, anything that can be given a value and a name can be compared with the set of compari-
son operations that return True and False.
Try It Out Comparing Values for Sameness
Testing for equality is done with two equal signs — remember that the single equal sign will bind
data to a name, which is different from what you want to do here, which is elicit a True or False:
>>> 1 == 1
True
>>> 1 == 2
False

9.5 Page 85

▲back to top
Chapter 4
How It Works
When you use the equality comparison, Python will compare the values on both sides. If the numbers
are different, False will be the result. If the numbers are the same, then True will be the result.
If you have different types of numbers, Python will still be able to compare them and give you the
correct answer:
>>> 1.23 == 1
False
>>> 1.0 == 1
True
You can also use the double equals to test whether strings have the same contents, and you can even
restrict this test to ranges within the strings (remember from the last chapter that slices create copies of
the part of the strings they reference, so you’re really comparing two strings that represent just the range
that a slice covers):
>>> a = “Mackintosh apples”
>>> b = “Black Berries”
>>> c = “Golden Delicious apples”
>>> a == b
False
>>> b == c
False
>>> a[-len(“apples”):-1] == c[-len(“apples”):-1]
True
Sequences can be compared in Python with the double equals as well. Python considers two sequences
to be equal when every element in the same position is the same in each list. Therefore, if you have three
items each in two sequences and they contain the same data but in a different order, they are not equal:
>>> apples = [“Mackintosh”, “Golden Delicious”, “Fuji”, “Mitsu”]
>>> apple_trees = [“Golden Delicious”, “Fuji”, “Mitsu”, “Mackintosh”]
>>> apples == apple_trees
False
>>> apple_trees = [“Mackintosh”, “Golden Delicious”, “Fuji”, “Mitsu”]
>>> apples == apple_trees
True
In addition, dictionaries can be compared. Like lists, every key and value (paired, together) in one dictionary
has to have a key and value in the other dictionary in which the key in the first is equal to the key in the sec-
ond, and the value in the first is equal to the value in the second:
>>> tuesday_breakfast_sold = {“pancakes”:10, “french toast”: 4, “bagels”:32,
“omelets”:12, “eggs and sausages”:13}
>>> wednesday_breakfast_sold = {“pancakes”:8, “french toast”: 5, “bagels”:22,
“omelets”:16, “eggs and sausages”:22}
>>> tuesday_breakfast_sold == wednesday_breakfast_sold
False
>>> thursday_breakfast_sold = {“pancakes”:10, “french toast”: 4, “bagels”:32,
“omelets”:12, “eggs and sausages”:13}
>>> tuesday_breakfast_sold == thursday_breakfast_sold
True
44

9.6 Page 86

▲back to top
Making Decisions
Doing the Opposite — Not Equal
There is an opposite operation to the equality comparison. If you use the exclamation and equals
together, you are asking Python for a comparison between any two values that are not equal (by the
same set of rules of equality that you saw for the double equal signs) to result in a True value.
Try It Out Comparing Values for Difference
>>> 3 == 3
True
>>> 3 != 3
False
>>> 5 != 4
True
How It Works
Every pair of numbers that would generate a True result when they’re compared using the == will now
generate a False, while any two numbers that would have generated a False when compared using ==
will now result in True.
These rules hold true for all of the more complex types, like sequences and dictionaries:
>>> tuesday_breakfast_sold != wednesday_breakfast_sold
True
>>> tuesday_breakfast_sold != thursday_breakfast_sold
False
Like numbers, any situation that would be true with == will be False with != with these types.
Comparing Values — Which One Is More?
Equality isn’t the only way to find out what you want to know. Sometimes you will want to know
whether a quantity of something is greater than that of another, or whether a value is less than some
other value. Python has greater than and less than operations that can be invoked with the > and < charac-
ters, respectively. These are the same symbols you are familiar with from math books, and the question
is always asking whether the value on the left is greater than (>) or less than (<) the value on the right.
Try It Out Comparing Greater Than and Less Than
>>> 5 < 3
False
>>> 10 > 2
True
How It Works
The number on the left is compared to the number on the right. You can compare letters, too. There are a
few conditions where this might not do what you expect, such as trying to compare letters to numbers.
(The question just doesn’t come up in many cases, so what you expect and what Python expects is
45

9.7 Page 87

▲back to top
Chapter 4
probably not the same.) The values of the letters in the alphabet run roughly this way: A capital “A” is
the lowest letter. “B” is the next, followed by “C”, and so on until “Z.” This is followed by the lowercase
letters, with “a” being the lowest lowercase letter and “z” the highest. However, “a” is higher than “Z”:
>>> “a” > “b”
False
>>> “A” > “b”
False
>>> “A” > “a”
False
>>> “b” > “A”
True
>>> “Z” > “a”
False
If you wanted to compare two strings that were longer than a single character, Python will look at each
letter until it finds one that’s different. When that happens, the outcome will depend on that one differ-
ence. If the strings are completely different, the first letter will decide the outcome:
>>> “Zebra” > “aardvark”
False
>>> “Zebra” > “Zebrb”
False
>>> “Zebra” < “Zebrb”
True
You can avoid the problem of trying to compare two words that are similar but have differences in capi-
talization by using a special method of strings called lower, which acts on its string and return a new
string with all lowercase letters. There is also a corresponding upper method. These are available for
every string in Python:
>>> “Pumpkin” == “pumpkin”
False
>>> “Pumpkin”.lower() == “pumpkin”.lower()
True
>>> “Pumpkin”.lower()
‘pumpkin’
>>> “Pumpkin”.upper() == “pumpkin”.upper()
True
>>> “pumpkin”.upper()
‘PUMPKIN’
When you have a string referenced by a name, you can still access all of the methods that strings nor-
mally have:
>>> gourd = “Calabash”
>>> gourd
‘Calabash’
>>> gourd.lower()
‘calabash’
>>> gourd.upper()
‘CALABASH’
46

9.8 Page 88

▲back to top
Making Decisions
More Than or Equal, Less Than or Equal
There is a useful variation on greater than and less than. It’s common to think of things in terms of greater
than or equal to or less than or equal to. You can use a simple shorthand to do that: Join the two symbols in
a way that makes sense when you look at it:
>>> 1 > 1
False
>>> 1 >= 2
False
>>> 10 < 10
False
>>> 10 <= 10
True
Reversing True and False
When you are creating situations where you’re comparing their outcomes, sometimes you want to know
whether something is true, and sometimes you want to know whether something is not true. Sensibly
enough, Python has an operation to create the opposite situation — the word not provides the opposite
of the truth value that follows it.
Try It Out Reversing the Outcome of a Test
>>> not True
False
>>> not 5
False
>>> not 0
True
How It Works
The not operation applies to any test that results in a True or False. However, remember from Chapter 3
that anything that’s not zero will be seen as True, so you can use not in many situations where you
wouldn’t expect it or where it doesn’t necessarily make sense:
>>> not 5 > 2
False
>>> not “A” < 3
True
>>> not “A” < “z”
False
47

9.9 Page 89

▲back to top
Chapter 4
Looking for the Results of More
Than One Comparison
You can also combine the results of more than one operation, which enables your programs to make
more complex decisions by evaluating the truth values of more than one operation.
One kind of combination is the and operation, which says “if the operation, value, or object on my left
evaluates to being True, move to my right and evaluate that. If it doesn’t evaluate to True, just stop and
say False — don’t do any more.”
>>> True and True
True
>>> False and True
False
>>> True and False
False
>>> False and False
False
The other kind of combining operation is the or compound. Using the or tells Python to evaluate the
expression on the left, and if it is False, Python will evaluate the expression on the right. If it is True,
Python will stop evaluation of any more expressions:
>>> True or True
True
>>> True or False
True
>>> False or True
True
>>> False or False
False
You may also want to place sequences of these together based on actions you want to happen. In these
cases, evaluation starts with the leftmost and or or and continues following the rules above — in other
words, until a False value is evaluated for and, or until a True value is evaluated for or.
How to Get Decisions Made
Python has a very simple way of letting you make decisions. The reserved word for decision making is
if, and it is followed by a test for the truth of a condition, and the test is ended with a colon, so you’ll
see it referred to here as if ... :. It can be used with anything that evaluates to True or False to say “if
something is true, do what follows”:
>>> if 1 > 2:
...
print ‘No, its not’
...
>>> if 2 > 1:
...
print ‘Yes, it is’
...
Yes, it is
48

9.10 Page 90

▲back to top
Making Decisions
You have just seen one of the most distinctive visual aspects of Python and the one
that most people remark on when they encounter Python.
When you see the colon in Python programs, it’s an indication that Python is enter-
ing a part of its program that is partially isolated from the rest of the program. At
this point, indentation becomes important. The indentation is how Python knows
that a particular block of code is separate from the code around it. The number of
spaces used is important, and a Python-oriented programming editor will always
carefully help you maintain the proper indentation for the code that is being writ-
ten. The number of spaces is relevant, so it is important to use the editor to deter-
mine your indentation and not change the number of spaces manually.
You will see more keywords paired with the colon; and in all cases, you need to pay
attention to the indentation. Python will warn you with an error if your program has
changes in indentation that it doesn’t understand.
Only when the statements to be evaluated between the if and the colon evaluate to True will the
indented statements below be visited by Python to be evaluated. The indentation indicates that the code
that follows it is a part of the program but is only executed only if the right conditions occur. For the
if ... : statement, the proper condition is when the comparison being made evaluates to True.
You can place if ... : statements within the indentation of other if ... : statements to perform more
complex decisions than what can be achieved with and and or because using if ... : enables you to
perform any series of statements that you may need before evaluating the indented if ... : statement.
Try It Out Placing Tests within Tests
Try the following example, where one if ...: appears within another:
>>> omelet_ingredients = {“egg”:2, “mushroom”:5, “pepper”:1, “cheese”:1, “milk”:1}
>>> fridge_contents = {“egg”:10, “mushroom”:20, “pepper”:3, “cheese”:2, “tomato”:4,
“milk”:15}
>>> have_ingredients = [True]
>>> if fridge_contents[“egg”] > omelet_ingredients[“egg”]:
...
have_ingredients[0] = False
...
have_ingredients.append(“egg”)
...
>>> if fridge_contents[“mushroom”] > omelet_ingredients[“mushroom”]:
...
if have_ingredients[0] == True:
...
have_ingredients[0] = False
...
have_ingredients.append(“mushroom”)
...
How It Works
After a condition is tested with an if ...: and there is an additional level of indentation, Python will
continue to evaluate the rest of the code that you’ve placed in the indentation. If the first if ...: isn’t
true, then none of the code below it will be evaluated — it would be skipped entirely.
However, if the first if ...: statement is true, the second one at the same level will be evaluated. The
outcome of a comparison only determines whether the indented code beneath it will be run. Code at the
49

10 Pages 91-100

▲back to top

10.1 Page 91

▲back to top
Chapter 4
same level, or above, won’t be stopped without something special happening, like an error or another
condition that would prevent the program from continuing to run.
To complete the example, you could enter the rest of this (if you want to make a computer representation
of an omelet):
>>> if fridge_contents[“pepper”] > omelet_ingredients[“pepper”]:
...
if have_ingredients[0] == True:
...
have_ingredients[0] = False
...
have_ingredients.append(“pepper”)
...
>>> if fridge_contents[“cheese”] > omelet_ingredients[“cheese”]:
...
if have_ingredients[0] == True:
...
have_ingredients[0] = False
...
have_ingredients.append(“cheese”)
...
>>> if fridge_contents[“milk”] > omelet_ingredients[“milk”]:
...
if have_ingredients[0] == True:
...
have_ingredients[0] = False
...
have_ingredients.append(“milk”)
...
>>> if have_ingredients[0] == True :
...
print “I can make an omelet now”
...
I can make an omelet now
You can create a chain of tests beginning with if ... : using elif ... :. elif ... : enables a variety
of conditions to be tested for but only if a prior condition wasn’t met. If you use a series of if ... :
statements they will all be executed. If you use an if ... : followed by an elif ... :, the elif ... :
will be evaluated only if the if ... : results in a False value:
>>> milk_price = 1.50
>>> if milk_price < 1.25:
...
print “Buy two cartons of milk, they’re on sale”
... elif milk_price < 2.00:
...
print “Buy one carton of milk, prices are normal”
... elif milk_price > 2.00:
...
print “Go somewhere else! Milk costs too much here”
...
Buy one carton of milk, prices are normal
There is also a fall-through statement that you can insert to handle those cases where none of the prior
tests resulted in a True value: the else: statement. If none of the if ... : or elif ... : statements
have test conditions that evaluate to True, the else: clause is invoked:
>>> OJ_price = 2.50
>>> if OJ_price < 1.25:
...
print “Get one, I’m thirsty”
... elif OJ_price <= 2.00:
...
print “Ummm... sure, but I’ll drink it slowly”
... else:
...
print “I don’t have enough money. Never mind”
...
I don’t have enough money. Never mind
50

10.2 Page 92

▲back to top
Making Decisions
Repetition
You have seen how many times every element in a sequence, or every element in a dictionary, needs to
be examined and compared. Doing this manually is impossibly boring and error prone for a person,
even a fast touch-typist. In addition, if you enter these things in manually, you’ll be caught off-guard
when the inevitable typo happens, or when something that you’re evaluating is changed elsewhere,
and your manually entered code can’t easily accommodate that change.
To perform repetitive tasks, Python offers two kinds of repetition operations. Both are similar — in fact,
they’re almost identical — but each one lets you think about what you’re doing differently so each one
should have its place in your skill set.
How to Do Something — Again and Again
The two operations that enable you to initiate and control repetitive tasks are the while and for opera-
tions. The while operation tests for one truth condition, so it will be referred to as while ... :. The for
operation uses each value from within a list, so it will be referred to as for ... in ... :.
The while ... : operation will first check for its test condition (the ... between the while and the :)
and if it’s True, it will evaluate the statements in its indented block a first time. After it reaches the end of
its indented block, which can include other indented blocks that it may contain, it will once again evalu-
ate its test condition to see whether it is still True. If it is, it will repeat its actions again; however, if it is
False, Python leaves the indented section and continues to evaluate the rest of the program after the
while ...: section. When names are used in the test condition, then between the first repetition and the
next (and the next, and so on), the value referred to by the name could have changed and on and on
until there is some reason to stop.
Try It Out Using a while Loop
>>> ingredients = omelet_ingredients.keys()
>>> ingredients
[‘cheese’, ‘pepper’, ‘egg’, ‘milk’, ‘mushroom’]
>>> while len(ingredients) > 0:
...
current_ingredient = ingredients.pop()
...
print “Adding %d %s to the mix” % (omelet_ingredients[current_ingredient],
current_ingredient)
...
Adding 5 mushroom to the mix
Adding 1 milk to the mix
Adding 2 egg to the mix
Adding 1 pepper to the mix
Adding 1 cheese to the mix
How It Works
In making this omelet, first you have taken the list of ingredients from the omelet_ingredients
dictionary. The dictionary contains both the ingredients and the corresponding quantities that are
needed for an omelet. The ingredients list only has the names of the ingredients.
The repetition using the while ... : operation will ensure that at least one element is left in the
ingredients list. For as long as there are elements in ingredients, the looping repetition will use
51

10.3 Page 93

▲back to top
Chapter 4
pop to remove the last element from the ingredients list and reference its value with the name
current_ingredient. Then, in the print statement, the current_ingredient is always going to be
the name of an ingredient from the keys of the omelet_ingredients dictionary because that’s where
the list ingredients came from.
Doing this the other way, with the for ... in ... : form of repetition, is, as shown before, very similar
to the while ... : form, but it saves you a couple of steps. In the first part, the for ..., you provide a
name that you will use inside of the indented code. In the second part, the in ... : part, you provide a
sequence, such as a list or a tuple, which takes each element and assigns the value of the element to the
name you provided in the first part. This saves you some of the effort that went in to using the while
loop in showing the omelet ingredients being put together:
>>> for ingredient in omelet_ingredients.keys():
...
print “adding %d %s to the mix” % (omelet_ingredients[ingredient],
ingredient)
...
adding 1 cheese to the mix
adding 1 pepper to the mix
adding 2 egg to the mix
adding 1 milk to the mix
adding 5 mushroom to the mix
You can see that this method is performed in the opposite order of the while ... : example. This is
because the first example used the pop method to remove elements from the end of the list, but the sec-
ond example with for ... in ... : uses each element in the keys of the omelet_ingredients in order
from first to last.
Stopping the Repetition
The common term infinite loop refers to a sequence of code that will repeat forever. A simple example
just sets up a while ... : statement that tests against something that is always going to result in True.
For instance, just using True will always work. You should not type in the following code, because it’s
the kind of thing that’s better to see than to have to do yourself:
>>> while True:
...
print “You’re going to get bored with this quickly”
...
You’re going to get bored with this quickly
You’re going to get bored with this quickly
You’re going to get bored with this quickly
You’re going to get bored with this quickly
You’re going to get bored with this quickly
The preceding code continues forever or until you break out of it. Inconvenient as it seems at first glance
to have something that repeats forever, there are times you may want this — for instance, in a program
that waits for the user to type something in, and when the user is done, returns to waiting.
However, sometimes you will want to know that if certain conditions are met, such as the right time of
day, when the water has run out, when there are no more eggs to be made into omelets, and so on, that
52

10.4 Page 94

▲back to top
Making Decisions
the repetition can be broken out of even when there is no explicit test in the top of the while ... : or
when the list that’s being used in the for ... in ... : doesn’t have an end.
Infinite loops can be exited by using the break statement. Some of the lines in the example here continue
for a long time. When you try this out, if you see a line that doesn’t begin with a >>> or a . . ., then it’s
actually part of the prior line, so type the entire line. In addition, make sure your indentation matches
what’s on the page:
>>> omlettes_ordered = 5 >>> omlettes_delivered = 0
>>> fridge_contents = {“egg”:8, “mushroom”:20, “pepper”:3, “cheese”:2, “tomato”:4,
“milk”:13}
>>>
>>> while omelets_delivered < omelets_ordered:
...
break_out = False
...
for ingredient in omelet_ingredients.keys():
...
ingredients_needed = omelet_ingredients[ingredient]
...
print “adding %d %s to the mix” % (omelet_ingredients[ingredient],
ingredient)
...
fridge_contents[ingredient] = fridge_contents[ingredient] -
ingredients_needed
...
if fridge_contents[ingredient] < ingredients_needed:
...
print “There isn’t enough %s for another omelet!” % ingredient
...
break_out = True
...
omelets_delivered = omelets_delivered + 1
...
print “One more omelet made! %d more to go” % (omelets_ordered -
omelets_delivered)
...
if break_out == True:
...
print “Out of ingredients, go shopping if you want to make more
omelets!”
...
break
...
adding 1 cheese to the mix
adding 1 pepper to the mix
adding 2 egg to the mix
adding 1 milk to the mix
adding 5 mushroom to the mix
One more omelet made! 4 more to go
adding 1 cheese to the mix
There isn’t enough cheese for another omelet!
adding 1 pepper to the mix
adding 2 egg to the mix
adding 1 milk to the mix
adding 5 mushroom to the mix
One more omelet made! 3 more to go
Out of ingredients, go shopping if you want to make more omelets!
If you use break, it will only take you out of the most recent loop — if you have a while ... : loop that
contains a for ... in ... : loop indented within it, a break within the for ... in ... : will not break
out of the while ... :.
53

10.5 Page 95

▲back to top
Chapter 4
Both while ... : and for ... in ... : loops can have an else: statement at the end of the loop, but
it will be run only if the loop doesn’t end due to a break statement. In this case, else: could be better
named something like done or on_completion, but else: is a convenient name because you’ve
already seen it, and it’s not hard to remember.
Try It Out Using else While Repeating
>>> for food in (“pate”, “cheese”, “crackers”, “yogurt”):
...
if food == “yogurt”:
...
break
... else:
...
print “There’s no yogurt!”
...
>>> for food in (“pate”, “cheese”, “crackers”):
...
if food == “yogurt”:
...
break
... else:
...
print “There’s no yogurt!”
...
There’s no yogurt!
How It Works
In each example, there is a test to determine whether there is any yogurt. If there is, the while ...: is
terminated by using a break. However, in the second loop, there is no yogurt in the list, so when the
loop terminates after reaching the end of the list, the else: condition is invoked.
There is one other commonly used feature for loops: the continue statement. When continue is used,
you’re telling Python that you do not want the loop to be terminated, but that you want to skip the rest
of the current repetition of the loop, and if you’re in a for ... in ...: loop, re-evaluate the conditions
and the list for the next round.
Try It Out Using continue to Keep Repeating
>>> for food in (“pate”, “cheese”, “rotten apples”, “crackers”, “whip cream”,
“tomato soup”):
...
if food[0:6] == “rotten”:
...
continue
...
print “Hey, you can eat %s” % food
...
Hey, you can eat pate
Hey, you can eat cheese
Hey, you can eat crackers
Hey, you can eat whip cream
Hey, you can eat tomato soup
How It Works
Because you’ve used an if ...: test to determine whether the first part of each item in the food list
contains the string “rotten”, the “rotten apples” element will be skipped by the continue, whereas
everything else is printed as safe to eat.
54

10.6 Page 96

▲back to top
Making Decisions
Handling Errors
You have seen examples of how Python reports errors in Chapter 2 and Chapter 3. Those errors usually
contain a lot of information pertaining to what failed and how:
>>> fridge_contents = {“egg”:8, “mushroom”:20, “pepper”:3, “cheese”:2, “tomato”:4,
“milk”:13}
>>> if fridge_contents[“orange juice”] > 3:
...
print “Sure, let’s have some juice”
...
Traceback (most recent call last):
File “<stdin>”, line 1, in ?
KeyError: ‘orange juice’
Oops. There is no orange juice in the fridge right now, but it would be nice to be able to learn this with-
out having to crash out of the program.
You have already learned one way to find out about the keys that are present in a dictionary, by using
the keys method of the dictionary and then searching through the list of keys to determine whether the
key you want is present. However, there’s no reason not to take a shortcut. The last line of the error
shown in the preceding code is:
KeyError: ‘orange juice’
This says that the error Python encountered was an error with the key in the fridge_contents dictionary.
You can use the error that Python has told you about to brace the program against that particular class of
error. You do this with the special word try:, telling Python to prepare for an error.
Trying Things Out
A try: statement sets up a situation in which an except: statement can follow it. Each except: state-
ment handles the error, which is formally named an exception, that was just raised when Python evalu-
ated the code within the try: statement instead of failing. To start with, use except: to handle one type
of error — for instance, the KeyError that you get when trying to check the fridge.
You have only one line in which to handle the error, which may seem restrictive, but in Chapter 5 you’ll
learn how to write your own functions so that you can handle errors with more flexibility:
>>> fridge_contents = {“egg”:8, “mushroom”:20, “pepper”:3, “cheese”:2, “tomato”:4,
“milk”:13}
>>> try:
...
if fridge_contents[“orange juice”] > 3:
...
print “Sure, let’s have some juice”
... except KeyError:
...
print “Aww, there’s no juice. Lets go shopping”
...
Aww, there’s no juice. Lets go shopping
You may find that you need to print more information about the error itself, and this is the information
that you have access to.
55

10.7 Page 97

▲back to top
Chapter 4
There are multiple kinds of exceptions, and each one’s name reflects the problem
that’s occurred and, when possible, the condition under which it can happen.
Because dictionaries have keys and values, the KeyError indicates that the key that
was requested from a dictionary isn’t present. Similarly, a TypeError indicates that
while Python was expecting one type of data (such as a string or an integer), another
type was provided that can’t do what’s needed.
In addition, when an exception occurs, the message that you would have otherwise
seen when the program stops (when you run interactively) can be accessed.
When you’ve learned more, you’ll be able to define your own types of exceptions for
conditions that require it.
Try It Out Creating an Exception with Its Explanation
>>> fridge_contents = {“egg”:8, “mushroom”:20, “pepper”:3, “cheese”:2, “tomato”:4,
“milk”:13}
>>> try:
...
if fridge_contents[“orange juice”] > 3:
...
print “Sure, let’s have some juice”
... except KeyError, error:
...
print “Woah! There is no %s” % error
...
Woah! There is no ‘orange juice’
How It Works
Because there is no key in fridge_contents dictionary for “orange juice”, a KeyError is raised by
Python to let you know that no such key is available. In addition, you specified the name error, which
Python will use to reference a string that contains any information about the error that Python can offer.
In this case, the string relates to the key that was requested but not present in the fridge_contents
dictionary (which is, again, “orange juice”).
There may be times when you handle more than one type of error in exactly the same way; and in those
cases, you can use a tuple with all of those exception types described:
>>> fridge_contents = {“egg”:8, “mushroom”:20, “pepper”:3, “cheese”:2, “tomato”:4,
“milk”:13}
>>> try: ...
if fridge_contents[“orange juice”] > 3:
...
print “Sure, let’s have some juice”
... except (KeyError, TypeError), error:
...
print “Woah! There is no %s” % error
...
Woah! There is no ‘orange juice’
If you have an exception that you need to handle, but you want to handle it by not doing anything
(for cases in which failure isn’t actually a big deal), Python will let you skip that case by using the
special word pass:
56

10.8 Page 98

▲back to top
Making Decisions
>>> fridge_contents = {“egg”:8, “mushroom”:20, “pepper”:3, “cheese”:2, “tomato”:4,
“milk”:13}
>>> try:
...
if fridge_contents[“orange juice”] > 3:
...
print “Sure, let’s have some juice”
... except KeyError, error:
...
print “Woah! There is no %s” % error
... except TypeError:
...
pass
...
Woah! There is no ‘orange juice’
There is also an else: clause that can be put at the end of the try: block of code. This will only be run
when there are no errors to be caught. Like before, else may not be the obvious choice for a name that
could better be described as “in case it all works” or “all_clear” or something like that. By now, however,
you can see how else: has become a flexible catch-all that means “in case something happens”
although it’s not consistent. In any case, it’s there for you to use.
Summary
In this chapter, you’ve learned about the methods for making decisions that Python offers. Any opera-
tion that results in True or False can be used by if ...: statements to determine whether a program will
evaluate an indented block of code.
You have seen for the first time the important role that indentation plays in Python programs. Even in
the interactive Python shell the number of spaces in the indentation matters.
You now have the knowledge to use sequence and dictionary elements in repetition loops. By using rep-
etitions, you can perform operations on every element in a list and make decisions about the values of
each list element.
The two types of repeating loops that Python offers you are the while ... : loop and the for ... in
... : loop. They perform similar jobs, continuing until a condition causes them to finish. The difference
between the two lies in the conditions that will permit them to evaluate their indented block of code.
The while ... : loop only tests for true or false in its test case, while the for ... in ... : loop will
take a sequence you provide in the in ... : section, and each element from first to last in the sequence
will be assigned to the value provided in the for ... section.
Both types of repeating loops can be exited before their test conditions are met by using the break
operation. The break operation will cause the loop that is being evaluated to stop without further evalu-
ations of any more code in the loop’s code block. However, if a break operation is performed, then
the optional else: condition for a loop will not be run. In addition to break is the continue operation,
which will skip the rest of the current loop but return to the top of the loop and evaluate the next
test case.
57

10.9 Page 99

▲back to top
Chapter 4
You also learned about one other kind of decision-making, which is handling the exceptions that Python
uses to report errors. These exceptions are how any error is reported. If they are not accommodated,
these errors will result in your program stopping at the point at which the error occurred. However, if
you enclose code that may cause an error in a code block indented beneath a try: you can specify
how to prevent the program from exiting, even going so far as handling the error and continuing with
the program. The errors that you anticipate encountering will be specified in the except ... : clause,
where the first value provided defines the type of the error (or types if a tuple of error types is provided);
and, optionally, a comma followed by a name used to refer to data containing information about the
error, can be provided.
Exercises
Perform all of the following in the codeEditor Python Shell:
1. Using a series of if ... : statements, evaluate whether the numbers from 0 through 4 are True
or False by creating five separate tests.
2. Create a test using a single if ... : statement that will tell you whether a value is between 0
and 9 inclusively (that is, the number can be 0 or 9 as well as all of the numbers in between, not
just 1-8) and print a message if it’s a success. Test it.
3. Using if ... :, elseif ...: and else:, create a test for whether a value referred to by a name
is in the first two elements of a sequence. Use the if ... : to test for the first element of the list;
use elif ... : to test the second value referenced in the sequence; and use the else: clause to
print a message indicating whether the element being searched for is not in the list.
4. Create a dictionary containing foods in an imaginary refrigerator, using the name fridge. The
name of the food will be the key, and the corresponding value of each food item should be a
string that describes the food. Then create a name that refers to a string containing the name of
a food. Call the name food_sought. Modify the test from question 2 to be a simple if ... :
test (no elif ... : or else: will be needed here) for each key and value in the refrigerator using
a for ... in ... : loop to test every key contained in the fridge. If a match is found, print a
message that contains the key and the value and then use break to leave the loop. Use an else
... : statement at the end of the for loop to print a message for cases in which the element
wasn’t found.
5. Modify question 3 to use a while ... : loop by creating a separate list called fridge_list
that will contain the values given by fridge.keys. As well, use a variable name, current_key
that will refer to the value of the current element in the loop that will be obtained by the method
fridge_list.pop. Remember to place fridge_list.pop as the last line of the while ... :
loop so that the repetition will end normally. Use the same else: statement at the end of the
while loop as the one used at the end of question 3.
6. Query the fridge dictionary created in question 3 for a key that is not present, and elicit an error.
In cases like this, the KeyError can be used as a shortcut to determining whether or not the
value you want is in the list. Modify the solution to question 3 so that instead of using a for
... in ... : a try: block is used.
58

10.10 Page 100

▲back to top
5
Functions
Up until this point, any time you wanted to accomplish a task, you have needed to type out entire
programs to do the job. If you needed to do the same work again, you could type the entire pro-
gram again or place it in a loop. However, loops are most useful when you are repeating the same
thing, but writing the same loop repeatedly in different parts of your program with slightly modi-
fied values in each one is not a sane way to live your life.
Python has functions that enable you to gather sections of code into more convenient groupings
that can be called on when you have a need for them.
In this chapter, you will learn how to create and use your own functions. You will be given guide-
lines to help facilitate your thinking about how to create and structure your programs to use func-
tions. You will also learn to write your functions so that you can later interrogate them for
information about how they behave and what you intend for them to do.
Putting Your Program into Its Own File
As the examples in this book get longer, typing the entire code block begins to be a burden. A sin-
gle mistake causes you to retype in the entire block of code you are working on. Long before
you’ve gotten to the point where you’ve got more than, say, 40 lines of code to type, you are
unlikely to want to have to do it more than once.
You are probably already aware that programmers write programs that are saved as source code
into files that can be opened, edited, and run without a great deal of work.
To reach this far more convenient state of affairs, from here on out you should type the programs
you are using into the main codeEditor window, and save the examples from the book into a sin-
gle folder from which you can reference them and run them. One suggestion for naming the folder
could be “Learning Python,” and then you could name the programs according to the chapters in
which they appear.

11 Pages 101-110

▲back to top

11.1 Page 101

▲back to top
Chapter 5
You can do two things to make your programs easy to run. The first line of all of your Python files
should look like this:
#!/usr/bin/env python
This enables Unix and Linux systems to run the script if you follow the instructions in the appendix at
the end of the book. A second important thing to do is to name all of your Python files with names that
end in .py. On Windows systems, this will provide the operating system with the information that it
needs to launch the file as a Python file and to not just treat it as a text file. For instance, if you put all of
the examples from the chapters you’ve read so far into their own files, you may have a folder with the
following files:
chapter_1.py
chapter_2.py
chapter_3.py
chapter_4.py
chapter_5.py
After you save your first program into a file, you’ll notice that codeEditor has begun to emphasize cer-
tain parts of the file by displaying them in a few different colors and styles. You’ll notice a pattern —
some of the built-in functions and reserved words are treated one way, while strings get a different treat-
ment and a few keywords are treated yet another way. However, most of the text in your files will still
be plain black and white, as shown in Figure 5-1.
Figure 5-1
60

11.2 Page 102

▲back to top
Functions
Using these files enables you to type any example only once. After an example has been typed in and
saved, you can run it with python -i <filename>. The -i tells python to read your program file, and then
lets you continue to interact with Python, instead of exiting immediately, which is what it normally would
do. Within codeEditor, you can do this automatically by selecting Run with Interpreter from the File menu.
Try It Out Run a Program with Python -i
To show how you can take advantage of running python -i or Run with Interpreter, enter the following
code into a file called ch5-demo.py:
a = 10
b = 20
print “A added to B is %d” % (a + b)
Now when you invoke Python with the -i option, you will be in a Python interactive session that looks
like the following:
A added to B is 30
>>>
How It Works
The code you entered into your ch5-demo.py file has all been evaluated now, and you can continue
to interact with the values of a and b, as well as expand upon it, just as though you’d entered them by
hand. This will save you time as the examples get longer. Now that you know all of this, some things
will be demonstrated in the shell first, but that you can save yourself to be run later. Other things will
be shown as code within a file that needs to be saved and run. You’ll be seeing programs in files because
either the material being covered doesn’t demonstrate an idea that is best shown off by forcing you to do
the extra work of typing in the same thing over and over, or of having you interact with it. Or it’s simply
too long to subject you to entering over and over each time you want to test it.
Functions: Grouping Code under a Name
Most modern programming languages provide you with the capability to group code together under a
name; and whenever you use that name, all of the code that was grouped together is invoked and evalu-
ated without having to be retyped every time.
To create a named function that will contain your code, you use the word def, which you can think of as
defining a functional block of code.
Try It Out Defining a Function
Try saving the following in your file for Chapter 5, ch5.py.def in_fridge:
try:
count = fridge[wanted_food]
except KeyError:
count = 0
return count
61

11.3 Page 103

▲back to top
Chapter 5
How It Works
When you invoke ch5.py with just the in_fridge function defined, you won’t see any output.
However, the function will be defined, and it can be invoked from the interactive Python session that
you’ve created.
To take advantage of the in_fridge function, though, you have to ensure that there is a dictionary
called fridge with food names in it. In addition, you have to have a string in the name wanted_food. This
string is how you can ask, using in_fridge, whether that food is available. Therefore, from the interac-
tive session, you can do this to use the function:
>>> fridge = {‘apples’:10, ‘oranges’:3, ‘milk’:2}
>>> wanted_food = ‘apples’
>>> in_fridge()
10
>>> wanted_food = ‘oranges’
>>> in_fridge()
3
>>> wanted_food = ‘milk’
>>> in_fridge()
2
This is more than just useful — it makes sense and it saves you work. This grouping of blocks of code
under the cover of a single name means that you can now simplify your code, which in turn enables you
to get more done more quickly. You can type less and worry less about making a mistake as well.
Functions are a core part of any modern programming language, and they are a key part of getting prob-
lems solved using Python.
Functions can be thought of as a question and answer process when you write them. When they are
invoked, a question is often being asked of them: “how many,” “what time,” “does this exist?” “can this
be changed?” and more. In response, functions will often return an answer — a value that will contain
an answer, such as True, a sequence, a dictionary, or another type of data. In the absence of any of these,
the answer returned is the special value None.
Even when a function is mainly being asked to just get something simple done, there is usually an
implied question that you should know to look for. When a function has completed its task, the ques-
tions “Did it work?” or “How did it work out?” are usually part of how you invoke the function.
Choosing a Name
One of the first guidelines to writing functions well is that you should name your functions to reflect
their purpose. They should indicate what you want them to do. Examples of this that come with Python
that you have seen are print, type, and len.
When you decide on a name, you should think about how it will be invoked in the program. It is
always good to name a function so that when it’s called, it will be read naturally by yourself and others
later. It is very common to forget the specifics of what you put into a function within a couple of weeks,
so the name becomes the touchstone that you use to recall what it’s doing when you return to use it
again later.
62

11.4 Page 104

▲back to top
Functions
Describing a Function in the Function
After you’ve chosen a name for your function, you should also add a description of the function. Python
enables you to do this in a way that is simple and makes sense.
If you place a string as the first thing in a function, without referencing a name to the string, Python will
store it in the function so you can reference it later. This is commonly called a docstring, which is short for
documentation string.
Documentation in the context of a function is anything written that describes the part of the program
(the function, in this case) that you’re looking at. It’s famously rare to find computer software that is well
documented. However, the simplicity of the docstring feature in Python makes it so that, generally, much
more information is available inside Python programs than in programs written in other languages that
lack this friendly and helpful convention.
The text inside the docstring doesn’t necessarily have to obey the indentation rules that the rest of the
source code does, because it’s only a string. Even though it may visually interrupt the indentation, it’s
important to remember that, when you’ve finished typing in your docstring, the remainder of your func-
tions must still be correctly indented.
def in_fridge ():
“””This is a function to see if the fridge has a food.
fridge has to be a dictionary defined outside of the function.
the food to be searched for is in the string wanted_food”””
try:
count = fridge[wanted_food]
except KeyError:
count = 0
return count
The docstring is referenced through a name that is part of the function, almost as though the function
were a dictionary. This name is __doc__, and it’s found by following the function name with a period
and the name __doc__.
Try It Out Displaying __doc__
You should now exit the interactive session that you entered in the last example and re-invoke ch5.py,
since it now has the docstring added to in_fridge. After you’ve done that, you can do the following:
>>> print “%s” % in_fridge.__doc__
This is a function to see if the fridge has a food.
fridge has to be a dictionary defined outside of the function.
the food to be searched for is in the string wanted_food
How It Works
Functions, like other types you’ve seen, have properties that can be used by following the name of the
function with a period and the name of the property. __doc__ is a string like any other and can be easily
printed for your reference while you’re in an interactive session.
The function has other information too (a set of information that it maintains that can be viewed with
the built-in function dir).
63

11.5 Page 105

▲back to top
Chapter 5
dir shows you all of the properties of the object in which you’re interested, such as a function, including
things that Python uses internally:
>>> dir(in_fridge)
[‘__call__’, ‘__class__’, ‘__delattr__’, ‘__dict__’, ‘__doc__’, ‘__get__’,
‘__getattribute__’, ‘__hash__’, ‘__init__’, ‘__module__’, ‘__name__’, ‘__new__’,
‘__reduce__’, ‘__reduce_ex__’, ‘__repr__’, ‘__setattr__’, ‘__str__’,
‘func_closure’, ‘func_code’, ‘func_defaults’, ‘func_dict’, ‘func_doc’,
‘func_globals’, ‘func_name’]
Any of these properties can be accessed using the same notation that you used for getting the data refer-
enced by in_fridge.__doc__, but normally you don’t need to use most of these attributes directly,
although it is a good exercise to explore these elements with the type built-in function to see how
Python describes them.
The Same Name in Two Different Places
One special property of a function is that it’s the first example you’ve seen of how the names that refer to
values can be compartmentalized. What this means is that if you have a name outside of a function, that
name refers to a particular value — whether it’s a string, a number, a dictionary, a sequence, or a function.
All of these share the same space.
For example, if you create a name for a string and then on the next line create a dictionary and reference
it to the same name, the string would no longer be referenced by that name, only the dictionary:
>>> fridge = “Chilly Ice Makers”
>>> fridge = {‘apples’:10, ‘oranges’:3, ‘milk’:2}
>>> print “%s” % fridge
{‘apples’: 10, ‘oranges’: 3, ‘milk’: 2}
This makes sense; however, this changes within a function when it’s being used. The function creates a
new space in which names can be reused and re-created without affecting the same names if they exist
in other spaces in your program. This enables you to write functions without worrying about having to
micromanage whether somewhere, in another function, a name that you are using is already being used.
Therefore, when you are writing a function, your function has its names, and another function has its
own names, and they are separate. Even when a name in both functions contains all of the same letters,
because they’re each in separate functions they are completely separate entities that will reference sepa-
rate values.
At the same time, if a function is going to be used in a known situation, where you have ensured that a
name it needs to use will be defined and have the right data already referenced, it is able to access this
global data by using that already-defined name. Python’s ability to do this comes from separating the
visibility of a name into separate conceptual areas. Each one of these areas is called a scope.
Scope defines how available any name is to another part of the program. The scope of a name that’s used
inside of a function can be thought of as being on a vertical scale. The names that are visible everywhere
are at the top level and they are referred to in python as being global. Names in any particular function
are a level below that — a scope that is local to each function. Functions do not share these with other
functions at the same level; they each have their own scope.
64

11.6 Page 106

▲back to top
Functions
Any name in the top-level scope can be reused in a lower-level scope without affecting the data referred
to by the top-level name:
>>> special_sauce = [‘ketchup’, ‘mayonnaise’, ‘french dressing’]
>>> def make_new_sauce():
...
“””This function makes a new special sauce all its own”””
...
special_sauce = [“mustard”, “yogurt”]
...
return special_sauce
...
At this point, there is a special sauce in the top-level scope, and another that is used in the function
make_new_sauce. When they are run, you can see that the name in the global scope is not changed:
>>> print “%s” % special_sauce
[‘ketchup’, ‘mayonnaise’, ‘french dressing’]
>>> new_sauce = make_new_sauce()
>>> print special_sauce
[‘ketchup’, ‘mayonnaise’, ‘french dressing’]
>>> print new_sauce
[‘mustard’, ‘yogurt’]
Remember that different functions can easily use the same name for a variable defined inside the
function — a name that will make sense in both functions, but reference different values, without con-
flicting with each other.
Making Notes to Yourself
Python has an additional feature of the language to help you to keep track of your program. Everything
that you type into a program, even if it doesn’t change how the program behaves (like docstrings) up to
this point, has been processed by Python. Even unused strings will cause Python to create the string just
in case you were going to use it.
In addition to unneeded strings, every programming language gives you the capability to place com-
ments within your code that don’t have any affect whatsoever on the program. They are not there for
Python to read but for you to read.
If at any point a line has the # character and it’s not in a string, Python will ignore everything that fol-
lows it. It will only begin to evaluate statements by continuing on the next line and reading the remain-
der of the program from there.
Try It Out Experimenting with Comments
If you test out comments interactively you can see how they’re different from strings when Python
reads them:
>>> “This is a string”
‘This is a string’
>>> # This is a comment
>>>
>>> “This is a string” # with a comment at the end
‘This is a string’
65

11.7 Page 107

▲back to top
Chapter 5
How It Works
When a comment appears by itself, Python ignores it and returns with the prompt asking for your next
request, trying to prompt you to enter a statement that it can evaluate. When a comment appears on a
line with something that can be evaluated, even just a string, Python knows that you have already given
your instructions to it.
Normally, comments will appear in program files. It’s unlikely you’ll ever bother entering comments as
annotations in your interactive sessions, but that’s how you’ll want to use them in your program files.
In addition, when you want to test changes in a program, it’s very useful to use comments to disable a
line (or more than one line) of code that is causing problems by placing a comment in front of it. Be care-
ful, though. A comment does affect the indentation that Python pays strict attention to. You need to be
careful to place comments that are within functions at the proper indentation level, because if you don’t,
Python will treat the comment as though it has closed out that function, if ...: block, or other cause of
indentation, and that’s almost certainly not what you want!
Keeping comments at the same indentation level also makes reading the comment much easier because
it is obvious to which part of the code the comment applies.
Asking a Function to Use a Value You Provide
In the in_fridge example, the values used by the function were in the global scope. The function
in_fridge only operated on already defined values whose names were already visible to the whole
program. This works only when you have a very small program.
When you move to larger programs consisting of hundreds, thousands, or more lines of code (the length
of a program is often measured in terms of the numbers of lines it contains), you usually can’t count on
the global availability of a particular name — it may be changed, based on decisions made by other peo-
ple and without your involvement! Instead, you can specify that a function will, every time it is invoked,
require that it be given the values that you want it to work with.
With many of the examples in the book, those that progress by offering different
and improved versions of themselves can be added to the same file unless you are
instructed to explicitly change the function you are working on.
You don’t always need to remove the prior revision of a function, since the next ver-
sion will simply “bump” it. This gives you the opportunity to look at the changes
that are being made to the function by comparing the old to the new.
As long as the most recent version is at the bottom of the file when you load it, that
version will be used.
This can be a useful practice when you’re writing your own programs as well.
There’s little as painful as fiddling with a piece of code that was working and then
not remembering how to return it to a working state.
66

11.8 Page 108

▲back to top
Functions
These values are the specifications or parameters that the function will use to do its job. When the function is
invoked, these parameters can be names that reference data, or they can be static data such as a number like
5 or a string. In all cases, the actual data will enter the scope of the called function instead of being global.
Notice that, in the following code, def — the definition of the function — has now changed so that it
specifies that the function will expect two parameters by naming them in the tuple that follows the func-
tion name. Those parameters will enter and remain in the scope of the in_fridge function, and they’ll
be seen as the names some_fridge and desired_item.
def in_fridge(some_fridge, desired_item):
“””This is a function to see if the fridge has a food.
fridge has to be a dictionary defined outside of the function.
the food to be searched for is in the string wanted_food”””
try:
count = some_fridge[desired_item]
except KeyError:
count = 0
return count
When you invoke a function with parameters, you specify the values for the parameters by placing the
values or the names you want to use between the parentheses in the invocation of the in_fridge func-
tion, separated by commas. You’ve already done this with functions like len.
Try It Out Invoking a Function with Parameters
Once again, you should re-invoke an interactive Python session by running python -i ch5.py or use
Run with Interpreter so that you will have an interactive session with the new in_fridge function
defined:
>>> fridge = {‘apples’:10, ‘oranges’:3, ‘milk’:2}
>>> wanted_food = “oranges”
>>> in_fridge(fridge, wanted_food)
3
How It Works
The fridge dictionary and the wanted_food string are given as parameters to the new in_fridge
function. After the scope of the function is entered, the dictionary referenced by fridge is now refer-
enced by the name some_fridge. At the same time, the string “oranges”, referenced by wanted_food,
is associated with the name desired_item upon entering the scope of the in_fridge function. After
this set-up is done, the function has the information it needs to do its job.
To further demonstrate how this works, you can use un-named values — data that isn’t referenced from
names:
>>> in_fridge({‘cookies’:10, ‘broccoli’:3, ‘milk’:2}, “cookies”)
10
These values are brought into the scope of the in_fridge function and assigned by the definition of the
function to the names that are used inside of the functions. The proof of this is that there is no longer a
global top-level name to be referenced from within the function.
67

11.9 Page 109

▲back to top
Chapter 5
Checking Your Parameters
The parameters that you intend to be used could be expecting different types than what they are given
when the function is called. For example, you could write a function that expects to be given a dictionary
but by accident is instead given a list, and your function will run until an operation unique to a dictio-
nary is accessed. Then the program will exit because an exception will be generated. This is different
from some other languages, which try to ensure that the type of each parameter is known, and can be
checked to be correct.
Python does not check to see what kind of data it’s associating to the names in a function. In most cases
this isn’t a problem because an operation on the provided data will be specific to a type, and then fail to
work properly if the type of data that the name references is not correct.
For instance, if in_fridge is given a number instead of a dictionary, Python, when trying to access the
number as though it were a dictionary, will raise an error that the except: will not catch.. A TypeError
will be generated indicating that the type Python tried to operate on isn’t capable of doing what Python
expected:
>>> in_fridge(4, “cookies”)
Traceback (most recent call last):
File “<stdin>”, line 1, in ?
File “<stdin>”, line 7, in in_fridge
TypeError: unsubscriptable object
In this case, you’ve been shown a number being given to a function where you know that the function
expects to operate on a dictionary. No matter what, a number does not have a property where a name
can be used as the key to find a value. A number doesn’t have keys and it doesn’t have values. The idea
is that in any context, finding 4(“cookies”) can’t be done in Python, and so an exception is raised.
The term unsubscriptable is how Python indicates that it can’t find a way to follow a key to a value the
way it needs to with a dictionary. Subscripting is the term for describing when you access an element in
a list or a tuple as well as a dictionary, so you can encounter this error in any of those contexts.
This behavior — not requiring you to specifically define what type you expect, and allowing you to flexi-
bly decide how you want to treat it — can be used to your advantage. It enables you to write a single
function that handles any kind of input that you want. You can write a single function that can take
more than one type as its parameter and then decide how the function should behave based on the type
it is given. Which approach you take depends on what you need to do in your own program.
To determine the type of some data, remember that you can use the type built-in function, which was
introduced in Chapter 2. Using the output of this, you can verify the type of variable in the beginning of
your functions:
def make_omelet(omelet_type):
“””This will make an omelet. You can either pass in a dictionary
that contains all of the ingredients for your omelet, or provide
a string to select a type of omelet this function already knows
about”””
if type(omelet_type) == type({}):
print “omelet_type is a dictionary with ingredients”
return make_food(omelet_type, “omelet”)
68

11.10 Page 110

▲back to top
Functions
elif type(omelet_type) == type(“”):
omelet_ingredients = get_omelet_ingredients(omelet_type)
return make_food(omelet_ingredients, omelet_type)
else:
print “I don’t think I can make this kind of omelet: %s” % omelet_type
By itself, this definition of make_omelet won’t work because it relies on a few functions that you
haven’t written yet. You will sometimes do this as you program — create names for functions that need
to be written later. You’ll see these functions later in this chapter, at which point this code will become
fully useable.
Try It Out Determining More Types with the type Function
The following should be entered after loading your ch5.py file with python -i or the Run with
Interpreter command:
>>> fridge = {‘apples’:10, ‘oranges’:3, ‘milk’:2}
>>> type(fridge)
<type ‘dict’>
>>> type({})
<type ‘dict’>
>>> type(“Omelet”)
<type ‘str’>
>>> type(“”)
<type ‘str’>
How It Works
The first thing to note here is that the type function returns a type object. You can use this type object in
tests — it can be compared to another type object.
Try It Out Using Strings to Compare Types
There is one other feature you can use here. You have seen that for the print function, many objects in
Python can be represented as strings. This is because many objects have a built-in capability to convert
themselves into strings for the times when that’s needed.
For example, an alternative way of writing the preceding comparison could be as follows:
>>> fridge = {‘apples’:10, ‘oranges’:3, ‘milk’:2}
>>> str(type(fridge))
“<type ‘dict’>”
>>> if str(type(fridge)) == “<type ‘dict’>”:
...
print “They match!”
...
They match!
How It Works
Because you can find out ahead of time what the string representation of a type object looks like, you can
use that string to compare to a type object that has been rendered into a string by the str function.
69

12 Pages 111-120

▲back to top

12.1 Page 111

▲back to top
Chapter 5
Setting a Default Value for a Parameter — Just in Case
There is one more trick available to you to ensure that your functions will be easier to use. Every param-
eter to a function needs to have a value. If values aren’t assigned to the names of all of the required
parameters, a function will raise an error — or worse, it could somehow return data that is wrong.
To avoid this condition, Python enables you to create functions with default values that will be assigned
to the parameter’s name if the function is invoked without that parameter being explicitly provided in
its invocation. You’ve already seen this behavior — for instance, with the pop method of lists, which can
either be told to work on a particular element in a list, or if no value is given, will automatically use the
last element.
You can do this in your own functions by using the assignment operator (the = sign) in the parameter list
when you define them. For instance, if you wanted a variation on make_omelet that will make a cheese
omelet by default, you have only to change its definition and nothing else.
Try It Out Setting a Default Parameter
Cut and paste the entire make_omelet function. Then, by changing only the definition in your new copy
of the function to the following, you’ll get the behavior of having a cheese omelet by default:
def make_omelet2(omelet_type = “cheese”):
How It Works
This definition doesn’t change the way that any of the remaining code in the function behaves. It sets up
omelet_type only if it hasn’t been defined when the make_omelet2 function is invoked.
This still enables you to specify an omelet by using a dictionary or a different kind of omelet! However,
if make_omelet is defined this way, you can call it without any particular kind of omelet being speci-
fied; and instead of bailing out on you, the function will make you a cheese omelet.
Doing this same thing to make_omelet is the first step toward writing a make_omelet function that will
be able to behave in a friendly and obvious way. Remember, though, that you still need to write other
functions! The goal is to have output like the following:
>>> make_omelet()
Adding 2 of eggs to make a cheese
Adding 2 of cheddar to make a cheese
Adding 1 of milk to make a cheese
Made cheese
‘cheese’
>>> make_omelet(“western”)
Adding 1 of pepper to make a western
Adding 1 of ham to make a western
Adding 1 of onion to make a western
Adding 2 of eggs to make a western
Adding 2 of jack_cheese to make a western
Adding 1 of milk to make a western
Made western
‘western’
70

12.2 Page 112

▲back to top
Functions
If you write a function with more than one parameter and you want to have both required and optional
parameters, you have to place the optionals at the end of your list of parameters. This is because once
you’ve specified that a parameter is optional, it may or may not be there. From the first optional param-
eter, Python can’t guarantee the presence of the remaining parameters — those to the right of your
optional parameters. In other words, every parameter after the first default parameter becomes optional.
This happens automatically, so be careful and be aware of this when you use this feature.
Calling Functions from within Other Functions
Functions declared within the top level, or global scope, can be used from within other functions and
from within the functions inside of other functions. The names in the global scope can be used from
everywhere, as the most useful functions need to be available for use within other functions.
In order to have a make_omelet function work the way you saw above, it should rely on other functions
to be available, so they can be used by make_omelet.
This is how it should work: First, a function acts like sort of a cookbook. It will be given a string that names
a type of omelet and return a dictionary that contains all of the ingredients and their quantities. This
function will be called get_omelet_ingredients, and it needs one parameter — the name of the omelet:
def get_omelet_ingredients(omelet_name):
“””This contains a dictionary of omelet names that can be produced,
and their ingredients”””
# All of our omelets need eggs and milk
ingredients = {“eggs”:2, “milk”:1}
if omelet_name == “cheese”:
ingredients[“cheddar”] = 2
elif omelet_name == “western”:
ingredients[“jack_cheese”] = 2
ingredients[“ham”]
=1
ingredients[“pepper”]
=1
ingredients[“onion”]
=1
elif omelet_Name == “greek”:
ingredients[“feta_cheese”] = 2
ingredients[“spinach”]
=2
else:
print “That’s not on the menu, sorry!”
return None
return ingredients
The second function you need to make omelets is a function called make_food that takes two param-
eters. The first is a list of ingredients needed — exactly what came from the get_omelet_ingredients
function. The second is the name of the food, which should be the type of omelet:
def make_food(ingredients_needed, food_name):
“””make_food(ingredients_needed, food_name)
Takes the ingredients from ingredients_needed and makes food_name”””
for ingredient in ingredients_needed.keys():
print “Adding %d of %s to make a %s” % (ingredients_needed[ingredient],
ingredient, food_name)
print “Made %s” % food_name
return food_name
71

12.3 Page 113

▲back to top
Chapter 5
At this point, all of the pieces are in place to use the make_omelet function. It needs to call on the
get_omelet_ingredients and the make_food functions to do its job. Each function provides some
part of the process of making an omelet. The get_omelet_ingredients provides the specific instruc-
tions for specific kinds of omelets, while the make_food function provides the information needed to
know that any kind of food can, if you look at it one way (a very simplistic way for the sake of demon-
stration!), be represented as the result of just mixing the right quantities of a number of ingredients.
Try It Out Invoking the Completed Function
Now that you have all of the functions in place for make_omelet to work, invoke your ch5.py file with
python -i or the Run with Interpreter command, and then try out the following code in the shell:
>>> omelet_type = make_omelet(“cheese”)
Adding 2 of eggs to make a cheese
Adding 2 of cheddar to make a cheese
Adding 1 of milk to make a cheese
Made cheese
>>> print omelet_type
cheese
>>> omelet_type = make_omelet({“eggs”:2, “jack_cheese”:2, “milk”:1, “mushrooms”:2})
omelet_type is a dictionary with ingredients
Adding 2 of jack_cheese to make a omelet
Adding 2 of mushrooms to make a omelet
Adding 2 of eggs to make a omelet
Adding 1 of milk to make a omelet
Made omelet
>>> print omelet_type
omelet
How It Works
Now that all of the functions are in place and can be called, one from another, make_omelet can be used
by only specifying the name of the omelet that you want to make.
Functions Inside of Functions
While it’s unlikely that you’ll be modeling any omelet-making in your professional or amateur career,
the same process of designing partial simulations of real-world situations is likely, so this section will
provide some ideas about how you could refine the solution you already have.
You may decide that a particular function’s work is too much to define in one place and want to break it
down into smaller, distinct pieces. To do this, you can place functions inside of other functions and have
them invoked from within that function. This allows for more sense to be made of the complex function.
For instance, get_omelet_ingredients could be contained entirely inside the make_omelet function
and not be available to the rest of the program.
Limiting the visibility of this function would make sense, as the usefulness of the function is limited to
making omelets. If you were writing a program that had instructions for making other kinds food as
well, the ingredients for omelets wouldn’t be of any use for making these other types of food, even simi-
lar foods like scrambled eggs or soufflés. Each new food would need its own functions to do the same
72

12.4 Page 114

▲back to top
Functions
thing, with one function for each type of food. However, the make_food function would still make sense
on its own and could be used for any kind of food.
Defining a function within another function looks exactly like defining it at the top level. The only differ-
ence is that it is indented at the same level as the other code in the function in which it’s contained. In
this case, all of the code looks exactly the same:
def make_omelet(omelet_type):
“””This will make an omelet. You can either pass in a dictionary
that contains all of the ingredients for your omelet, or provide
a string to select a type of omelet this function already knows
about”””
def get_omelet_ingredients(omelet_name):
“””This contains a dictionary of omelet names that can be produced,
and their ingredients”””
ingredients = {“eggs”:2, “milk”:1}
if omelet_name == “cheese”:
ingredients[“cheddar”] = 2
elif omelet_name == “western”:
ingredients[“jack_cheese”] = 2
# You need to copy in the remainder of the original
# get_omelet_ingredients function here. They are not being
# included here for brevity’s sake
if type(omelet_type) == type({}):
print “omelet_type is a dictionary with ingredients”
return make_food(omelet_type, “omelet”)
elif type(omelet_type) == type(“”):
omelet_ingredients = get_omelet_ingredients(omelet_type)
return make_food(omelet_ingredients, omelet_type)
else:
print “I don’t think I can make this kind of omelet: %s” % omelet_type
It is important to define a function before it is used. If an attempt is made to invoke a function before
it’s defined, Python won’t be aware of its existence at the point in the program where you’re trying to
invoke it, and so it can’t be used! Of course, this will result in an error and an exception being raised. So,
define your functions at the beginning of your files so you can use them toward the end.
Flagging an Error on Your Own Terms
If you need to indicate that a particular error has occurred, you may want to use one of the errors you’ve
already encountered to indicate, through the function that’s being called, what has gone wrong.
There is a counterpart to the try: and except: special words: the raise ... command. A good time to
use the raise ... command might be when you’ve written a function that expects multiple parameters
but one is of the wrong type.
You can check the parameters that are passed in and use raise ... to indicate that the wrong type was
given. When you use raise ..., you provide a message that an except ... : clause can capture for
display — an explanation of the error.
73

12.5 Page 115

▲back to top
Chapter 5
The following code changes the end of the make_omelet function by replacing a printed error, which is
suitable for being read by a person running the program, with a raise ... statement that makes it pos-
sible for a problem to be either handled by functions or printed so that a user can read it:
if type(omelet_type) == type({}):
print “omelet_type is a dictionary with ingredients”
return make_food(omelet_type, “omelet”)
elif type(omelet_type) == type(“”):
omelet_ingredients = get_omelet_ingredients(omelet_type)
return make_food(omelet_ingredients, omelet_type)
else:
raise TypeError, “No such omelet type: %s” % omelet_type
After making this change, make_omelet can give you precise information about this kind of error when
it’s encountered, and it still provides information for a user.
Layers of Functions
Now that you’ve got an idea of what functions are and how they work, it’s useful to think about them in
terms of how they are called and how Python keeps track of these layers of invocations.
When your program calls a function, or a function calls a function, Python creates a list inside of itself
that is called the stack or sometimes the call stack. When you invoke a function (or call on, which is why it
can be called a call stack), Python will stop for a moment, take note of where it is when the function was
called and then stash that information in its internal list. It’ll then enter the function and execute it, as
you’ve seen. For example, the following code illustrates how Python keeps track of how it enters and
leaves functions:
[{‘top_level’: ‘line 1’}, {‘make_omelet’: ‘line 64’}, {‘make food’: ‘line 120’}]
At the top, Python keeps track starting at line 1. Then, as the function make_omelet is called at line
sixty-four, it keeps track of that. Then, from inside of make_omelet, make_food is called. When the
make_food function finishes, Python determines that it was on line 64, and it returns to line 64 to con-
tinue. The line numbers in the example are made up, but you get the idea.
The list is called a stack because of the way in which a function is entered. You can think of a function as
being on the top of a stack until it is exited, when it is taken off, and the stack is shortened by one.
How to Read Deeper Errors
When an error does happen in a program and an uncaught error is raised, you might find yourself look-
ing at a more complex error than what you’ve seen before. For example, imagine that you’ve passed a
dictionary that contains a list instead of a number. This will cause an error that looks like the following:
>>> make_omelet({“a”:1, “b”:2, “j”:[“c”, “d”, “e”]})
omelet_type is a dictionary with ingredients
Adding 1 of a to make a omelet
Adding 2 of b to make a omelet
74

12.6 Page 116

▲back to top
Functions
Traceback (most recent call last):
File “<stdin>”, line 1, in ?
File “ch5.py”, line 96, in make_omelet
return make_food(omelet_type, “omelet”)
File “ch5.py”, line 45, in make_food
print “Adding %d of %s to make a %s” % (ingredients_needed[ingredient],
ingredient, food_name)
TypeError: int argument required
After you’ve entered a function from a file, Python will do its best to show you where in the stack you
are (which means how many layers there are when the error occurs and at what line in the file each layer
in the stack was called from) so that you can open the problem file to determine what happened.
As you create deeper stacks (which you can think of as longer lists) by calling more functions or using
functions that call other functions, you gain experience in using the stack trace. (This is the common
name for the output that Python gives you when you raise an error or when an exception is otherwise
raised.)
With the preceding stack trace, which is three levels deep, you can see that in line 45, when make_food
is called, there was a problem with the type of an argument. You could now go back and fix this.
If you thought that this problem would happen a lot, you could compensate for it by enclosing calls to
make_food in a try ...: block so that TypeErrors can always be prevented from stopping the program.
However, it’s even better if you handle them in the function where they will occur.
In the case of something like a blatantly incorrect type or member of a dictionary, it’s usually not neces-
sary to do any more than what Python does on its own, which is to raise a TypeError. How you want to
handle any specific situation is up to you, however.
The stack trace is the readable form of the stack, which you can examine to see where the problem hap-
pened. It shows everything that is known at the point in time when a problem occurred, and it is pro-
duced by Python whenever an exception has been raised.
Summary
This chapter introduced you to functions. Functions are a way of grouping a number of statements in
Python into a single name that can be invoked any time that it’s needed. When a function is defined, it
can be created so that when it’s invoked it will be given parameters to specify the values on which it
should operate.
The names of the parameters for a function are defined along with the function by enclosing them in
parentheses after the function is named. When no parameters are used, the parentheses are still present,
but they will be empty.
As functions are invoked, they each create a scope of their own whereby they have access to all of the
names that are present in the global scope of the program, as well as names that have been assigned and
created inside of the function. If a name that is present in the global scope is assigned in the scope of a
75

12.7 Page 117

▲back to top
Chapter 5
particular function, it will not change value when referenced by the global name but will instead only
be changed within the function.
If a function is defined within another function, then it can access all of the names of the function in
which it was defined, as well as names that are in the global scope. Remember that this visibility
depends on where the function is defined and not where it was called.
Functions can be called from within other functions. Doing this can make understanding programs
easier. Functions enable you to reduce repetitive typing by making common tasks achievable with a
brief name.
Functions that are defined with parameters are invoked with values — each value provided will be
assigned, in the function, to the name inside the function’s parameter list. The first parameter passed to
a function will be assigned to the first name, the second to the second, and so on. When functions are
passed parameters, each one can be either mandatory or optional. Optional parameters must be placed
after mandatory parameters when the function is defined, and they can be given a default value.
You can use the raise . . . : feature to signal errors that can be received and handled by except . . . :.
This enables you to provide feedback from your functions by providing both the type of error and a
string that describes the error so it can be handled.
You have also learned about the stack. When an error condition is raised with raise . . . :, or by
another error in the program, the location of the error is described not just by naming the function where
the error occurred, but also by naming any and all intervening functions that were invoked and specify-
ing on what line in which file that invocation happened. Therefore, if the same function is useful enough
that you use it in different places and it only has problems in one of them, you can narrow the source of
the problem by following the stack trace that is produced.
Exercises
1. Write a function called do_plus that accepts two parameters and adds them together with the
“+” operation.
2. Add type checking to confirm that the type of the parameters is either an integer or a string. If the
parameters aren’t good, raise a TypeError.
3. This one is a lot of work, so feel free to take it in pieces. In Chapter 4, a loop was written to make
an omelet. It did everything from looking up ingredients to removing them from the fridge and
making the omelet. Using this loop as a model, alter the make_omelet function by making a
function called make_omelet_q3. It should change make_omelet in the following ways to get it
to more closely resemble a real kitchen:
a. The fridge should be passed into the new make_omelet as its first parameter.
The fridge’s type should be checked to ensure it is a dictionary.
b. Add a function to check the fridge and subtract the ingredients to be used. Call
this function remove_from_fridge. This function should first check to see if
enough ingredients are in the fridge to make the omelet, and only after it has
checked that should it remove those items to make the omelet. Use the error type
LookupError as the type of error to raise.
76

12.8 Page 118

▲back to top
Functions
c. The items removed from the fridge should be placed into a dictionary and
returned by the remove_from_fridge function to be assigned to a name that
will be passed to make_food. After all, you don’t want to remove food if it’s not
going to be used.
d. Rather than a cheese omelet, choose a different default omelet to make. Add the
ingredients for this omelet to the get_omelet_ingredients function.
4. Alter make_omelet to raise a TypeError error in the get_omelet_ingredients function if
a salmonella omelet is ordered. Try ordering a salmonella omelet and follow the resulting
stack trace.
77

12.9 Page 119

▲back to top

12.10 Page 120

▲back to top
6
Classes and Objects
So far, you have been introduced to most of the building blocks of programming. You have used
data; you have referenced that data to names (the names are more commonly called variables
when programmers talk); and you have used that data in loops and functions. The use of these
three elements are the foundation of programming and problem-solving with computers. Named
variables enable you to store values, reference them, and manipulate them. Repeating loops enable
you to evaluate every possible element in a list, or every other element, or ever third element, and
so on. Finally, functions enable you to combine bunches of code into a name that you can invoke
whenever and wherever you need it.
In this chapter, you will see how Python provides a way to combine functions and data so that
they are accessed using a single object’s name. You’ll also gain some knowledge about how and
why classes and objects are used and how they make programs easier to write and use in a variety
of circumstances.
Thinking About Programming
At this point, you’ve only been given a rudimentary introduction to Python. To create a descrip-
tion of an object in Python right now, you have just enough knowledge to achieve two views. One
is of the data, which comes and goes as needed, except for parts that live in the top level, or global
scope. The other view is of functions, which have no persistent data of their own. They interact
only with data that you give them.
Objects You Already Know
The next tool you will be given will enable you to think of entire objects that contain both data and
functions. You’ve already seen these when you used strings. A string is not just the text that it con-
tains. As you’ve learned, methods are associated with strings, which enable them to be more than
just the text, offering such features as allowing you to make the entire string upper or lowercase.
To recap what you’ve already learned, a string is mainly the text that you’ve input:
>>> omelet_type = “Cheese”

13 Pages 121-130

▲back to top

13.1 Page 121

▲back to top
Chapter 6
In addition to the data that you’ve worked with the most, the text “Cheese,” the string is an object that
has methods, or behaviors that are well known. Examples of methods that every string has are lower,
which will return the string it contains as all lowercase, and upper, which will return the string as an
entirely uppercase string:
>>> omelet_type.lower()
‘cheese’
>>> omelet_type.upper()
‘CHEESE’
Also available are methods built into tuple, list, and dictionary objects, like the keys method of dictionar-
ies, which you’ve already used:
>>> fridge = {“cheese”:1, “tomato”:2, “milk”:4}
>>> fridge.keys()
[‘tomato’, ‘cheese’, ‘milk’]
When you want to find out more about what is available in an object, Python exposes everything that
exists in an object when you use the dir function:
dir(omelet_type)
[‘__add__’, ‘__class__’, ‘__contains__’, ‘__delattr__’, ‘__doc__’, ‘__eq__’,
‘__ge__’, ‘__getattribute__’, ‘__getitem__’, ‘__getnewargs__’, ‘__getslice__’,
‘__gt__’, ‘__hash__’, ‘__init__’, ‘__le__’, ‘__len__’, ‘__lt__’, ‘__mod__’,
‘__mul__’, ‘__ne__’, ‘__new__’, ‘__reduce__’, ‘__reduce_ex__’, ‘__repr__’,
‘__rmod__’, ‘__rmul__’, ‘__setattr__’, ‘__str__’, ‘capitalize’, ‘center’, ‘count’,
‘decode’, ‘encode’, ‘endswith’, ‘expandtabs’, ‘find’, ‘index’, ‘isalnum’,
‘isalpha’, ‘isdigit’, ‘islower’, ‘isspace’, ‘istitle’, ‘isupper’, ‘join’, ‘ljust’,
‘lower’, ‘lstrip’, ‘replace’, ‘rfind’, ‘rindex’, ‘rjust’, ‘rsplit’, ‘rstrip’,
‘split’, ‘splitlines’, ‘startswith’, ‘strip’, ‘swapcase’, ‘title’, ‘translate’,
‘upper’, ‘zfill’]
Every bit of data, every method, and, in short, every name in a string or any other object in Python can be
exposed with the dir function. dir lists all of the available names in the object it is examining in alpha-
betical order, which tends to group those names beginning with underscores first. By convention, these
names refer to items considered to be internal pieces of the object and should be treated as though they
are invisible. In other words, you shouldn’t use them, but Python leaves that decision up to you —
there’s no reason not to look at these items interactively to learn from them:
>>> type(omelet_type.__len__)
<type ‘method-wrapper’>
This is interesting. Because this is a method, it can be invoked to see what it does:
>>> omelet_type.__len__()
6
This returns the same value as the len built-in function. When a function is built in to an object, it’s
called a method of that object.
80

13.2 Page 122

▲back to top
Classes and Objects
In fact, the method __len__ is how the len function works: It asks an object how long it is by asking
this built-in method. This enables the designer of an object to define how the length is determined and to
have the built-in function len behave correctly for any object that defines a __len__ method.
The other names beginning with an underscore also have special meanings. You can explore these in the
Python shell. Pythoncard’s Python Shell will help you explore the normal methods of a string object, or
any other method, by displaying possible names within the object that you are trying to call on, but it
will not display internal names that begin with an underscore. You can determine those with the dir
function yourself if you decide to do this.
Looking Ahead: How You Want to Use Objects
When you have an object, you want to be able to use it naturally. For instance, once you’ve defined it,
the Omelet class could produce objects that behave in a way that would feel natural when you read the
source code. You’re going to try to make something that can do this (you’ll see how to do this in the next
section):
>>> o1 = Omelet()
>>> o1.show_kind()
‘cheese’
You’d also want to have a refrigerator that can be used as an object instead of just as a dictionary. It may
be nice for you to be able to do things like be able to think of using it like a real fridge, whereby you can
add food, remove food, check for foods, add or remove more than one thing at a time, and so on.
In other words, when you create an object that models something from the real world, you can form
your program’s objects and classes so they help the pieces of the program work in a way that someone
familiar with the real life object will recognize and be able to understand.
Defining a Class
When you are considering how even small programs of a few hundred lines of Python code is working,
you will often realize that the program is keeping track of data in groups — when one thing is accessed,
it affects other things that need to go along with it. Almost as often, you’ll realize that you’ve got whole
lists of this interdependent data — lists in which the first element in list1 is matched to the first element
in list2 and list3, and so on. Sometimes this can and should be solved by combining the lists creatively.
Python employs the concept of creating an entire class of code that acts as a placeholder. When a class is
invoked, it creates an object bound to a name.
How Code Can Be Made into an Object
After you have an object bound to a name, using that name provides you with access to all of the data
and functions you’ve defined. When you are writing code for a class, you start by declaring that class.
This is done with the class keyword.
81

13.3 Page 123

▲back to top
Chapter 6
Try It Out Defining a Class
The definition of a class is simple and mainly involves the use of the special word class along with a
name. The style is similar to the definition of a function, except that you do not follow a simple class def-
inition with a tuple containing terms. (Doing that defines a class to inherit from, which you will see in
Chapter 10.)
class Fridge:
“””This class implements a fridge where ingredients can be
added and removed individually, or in groups.”””
How It Works
From here on out, everything indented will be available through the objects created inside of this class.
You’ve already seen this with functions in Chapter 5, and similar rules apply to classes. Note that you
have the option for the built-in docstring with classes, as you do with functions. They behave the same
way and are very useful for providing an easy way to get information about the class.
You should try creating the Fridge class as shown in the preceding example. Note that a capital “F” was
used for this. It’s a common convention for Python programmers to begin their class names with a capi-
tal letter; and when a class name has more than one word, it’s also common convention to run the words
together, but to have each word begin with a capital letter to make it easier to read. For instance, a class
that is modeling a fridge and a freezer together could be called FridgeAndFreezer.
Try It Out Creating an Object from Your Class
Try typing the Fridge class into your ch6.py file (or a similar file for the examples here) and then
invoke that file with python -i or the Run with Interpreter command, as you did in Chapter 5.
You can create a single object that is a Fridge by invoking it with the open and close parentheses:
>>> f = Fridge()
How It Works
At this point, you don’t have anything complicated defined yet. Fridge is basically empty, so this is
your starting point. However, even without anything else, you should notice that you created an empty
class that is usable. It does almost nothing, but there are situations in which you need very little. For
instance, you can now treat this nearly empty object you’ve created like a special kind of dictionary. You
can do this by adding names to your class interactively while you’re testing. This can help you develop
an idea how you’d like it to work:
>>> f.items = {}
>>> f.items[“mystery meat”] = 1
In addition, as you’ll see demonstrated in Chapter 10, exceptions are actually classes, and sometimes all
you need is an empty class to make an effective exception. You should only use this sort of direct access
to a class when you have a simple, undefined class like this. When you have a more developed class,
accessing the names inside of its scope can interfere with how the class was written, so it can cause a lot
of trouble.
82

13.4 Page 124

▲back to top
Classes and Objects
The best way to start writing a class is to decide what you want it to do. For this, a Python-based model
of refrigerator behaviors, Fridge, is the first thing, and it should be basic. While you’re thinking about
it, focus on what you will need a particular Fridge object to do for your own purposes. You want
enough behaviors available that this object can be used to make food, yet you don’t want to worry about
aspects of real-life refrigerators that won’t be included in a simplified example, such as temperature, the
freezer, defrosting, and electricity — all of these are unnecessary details that would only complicate our
purpose here. For now, let’s just add to the docstring for the Fridge class to define the behaviors that
you will be building soon.
First, you will want to have a way of stocking your Fridge. There are a couple of ways you’re going to
do this: adding one type of a single item at a time and adding an entire dictionary at the same time so
that it’s easy to initialize. Or simulating occasions when a refrigerator is filled, such as after you’ve come
back from a shopping trip.
Second, you’ll want to have a way to take things out of the Fridge. You want to have the capability to
do all of the same things when removing items as you do when you add: get a single item or get a whole
bunch of things out of the Fridge.
You’ll want to write a couple of other things into this object to make this selective model of a Fridge: a
function that will determine whether a particular item is available in the Fridge and another one that
will check an entire dictionary worth of ingredients. These enable you to prepare to begin cooking.
These are all of the things that you would need to have in order to use a Fridge to store ingredients and
to get them out when you want to cook but only for this limited purpose of modeling, of course. In other
words, these will work as a model of this specific situation, while glossing over every possible scenario.
The methods that an object makes available for use are called its interface because these methods are
how the program outside of the object makes use of the object. They’re what make the object useable.
The interface is everything you make available from the object. With Python, this usually means all of
the methods and any other names that don’t begin with one or more underscores are your interfaces;
however, it’s a good practice to distinguish which functions you expect to have called by explicitly stat-
ing what methods can be used, and how they’re used, in the class’s docstring:
class Fridge:
“””This class implements a fridge where ingredients can be
added and removed individually, or in groups.
The fridge will retain a count of every ingredient added or removed,
and will raise an error if a sufficient quantity of an ingredient
isn’t present.
Methods:
has(food_name [, quantity]) - checks if the string food_name is in the fridge.
Quantity will be set to 1 if you don’t specify a number.
has_various(foods) - checks if enough of every food in the dictionary is in the
fridge
add_one(food_name) - adds a single food_name to the fridge
add_many(food_dict) - adds a whole dictionary filled with food
get_one(food_name) - takes out a single food_name from the fridge
get_many(food_dict) - takes out a whole dictionary worth of food.
get_ingredients(food) - If passed an object that has the __ingredients__
method, get_many will invoke this to get the list of ingredients.
“””
83

13.5 Page 125

▲back to top
Chapter 6
def __init__(self, items={}):
“””Optionally pass in an initial dictionary of items”””
if type(items) != type({}):
raise TypeError, “Fridge requires a dictionary but was given %s” %
type(items)
self.items = items
return
In addition, documenting the methods you expect to be used is a good practice when you sit down to
write a class — in effect, it is your outline for what you need to do to consider the class complete, and
this can go hand-in-hand with testing your program as you write it. (See Chapter 12 for more about how
to do this.)
When you write your interface methods, you’ll notice that, a lot of the time, simpler methods will share
a lot of common features, like “get one thing” or “get two things” or “get some large number of things,”
but to make them simple to call, you’ll want to keep all of these variations. At first, this will look seem as
though it means that you need to duplicate a lot of the source code for each of these functions. However,
instead of retyping the common components of your interface methods, you can save a lot of work by
writing methods that are for internal use only.
These private methods can perform actions common to some or all of your interface methods. You’d
want to do this when the private methods are more complex, or contain details that a user may not need
to know in order to use them. By doing this, you can prevent confusion when your class is called, while
making it easier for you to write. At its best, this is a clear win-win situation.
For the Fridge class, and in many classes you’ll write, it’s common to have a method that can operate
on a group of data, and another method that works with just a single element. Whenever you have this
situation, you can save your effort by making the method that works on a single item simply invoke
the method that works on any number of items. In fact, sometimes it’s useful to have this method be
considered private, or not a part of the interface. This way it can be used or not used and changed with-
out affecting how the class is used, because any changes you make will not be seen outside an object,
only inside.
For your Fridge class, you can minimize your work by creating an internal method called __add_multi
that will take two parameters — the name of the item and the quantity of items — and have it add those
to the items dictionary that each object has.
Try It Out Writing an Internal Method
When you add this to your file for this chapter, remember to ensure that you have the right indentation
for this to appear under your Fridge class, not alone at the top level. The class declaration is shown
here to make this clear:
class Fridge:
# the docstring and intervening portions of the class would be here, and
# __add_multi should go afterwards.
def __add_multi(self, food_name, quantity):
“””
__add_multi(food_name, quantity) - adds more than one of a
food item. Returns the number of items added
84

13.6 Page 126

▲back to top
Classes and Objects
This should only be used internally, after the type checking has been
done
“””
if not self.items.has_key(food_name):
self.items[food_name] = 0
self.items[food_name] = self.items[food_name] + quantity
How It Works
Now you have a way of adding any number of single food items to a Fridge object. However, this is
an internal method that doesn’t confirm whether the type that it is being given — either for food_name
or quantity — is valid. You should use your interface functions to do this checking because, being a
conscientious programmer, you will always ensure that you only pass the right values into your private
methods. OK, just kidding. It’s always a good idea to check everywhere you can. For this example, you’re
not going to check here, though, because you’re only going to use __add_multi in a foolproof way.
Now that you have the generally useful method __add_multi for your Fridge class, the add_one and
the add_many methods can both be written to use it instead of your having to write similar functions
two times. This will save you work.
Try It Out Writing Interface Methods
To make this faster, you can avoid typing in the docstrings for now. They are here so that you under-
stand better what the actual code is doing in case you have any questions.
Like before, these need to be indented beneath the Fridge class definition. Anything that seems to begin
at the start of a line is actually a continuation from the line before and should all be entered on one line:
def add_one(self, food_name):
“””
add_one(food_name) - adds a single food_name to the fridge
returns True
Raises a TypeError if food_name is not a string.
“””
if type(food_name) != type(“”):
raise TypeError, “add_one requires a string, given a %s” %
type(food_name)
else:
self.__add_multi(food_name, 1)
return True
def add_many(self, food_dict):
“””
add_many(food_dict) - adds a whole dictionary filled with food as keys and
quantities as values.
returns a dictionary with the removed food.
raises a TypeError if food_dict is not a dictionary
returns False if there is not enough food in the fridge.
“””
85

13.7 Page 127

▲back to top
Chapter 6
if type(food_dict) != type({}):
raise TypeError, “add_many requires a dictionary, got a %s” % food_dict
for item in food_dict.keys():
self.__add_multi(item, food_dict[item])
return
How It Works
add_one and add_many each serve similar purposes, and each one has the code to ensure that it is being
used appropriately. At the same time, they both use __add_multi_to actually do the heavy lifting. Now
if anything changes regarding how your class works inside of __add_multi, you will save time because
it will change how both of these methods behave.
Now that you’ve written all of this, you have enough code written to put items into a Fridge object, but
no way of taking items out. You can just directly access the object.items dictionary, but that is never a
good idea except when testing. Of course, you’re testing now, so why not do that?
>>> f = Fridge({“eggs”:6, “milk”:4, “cheese”:3})
>>> f.items
{‘cheese’: 3, ‘eggs’: 6, ‘milk’: 4}
>>> f.add_one(“grape”)
True
>>> f.items
{‘cheese’: 3, ‘eggs’: 6, ‘grape’: 1, ‘milk’: 4}
>>> f.add_many({“mushroom”:5, “tomato”:3})
>>> f.items
{‘tomato’: 3, ‘cheese’: 3, ‘grape’: 1, ‘mushroom’: 5, ‘eggs’: 6, ‘milk’: 4}
>>>
So far, everything works! This is the simple part. The second thing you’ll want to add are the methods
that enable you to determine whether something is in the Fridge.
It is important to write code that gives you a way to confirm that something is present because it can be
used by the methods that remove items, get_one and get_many and get_ingredients, so that they
ensure that they can check if enough of the items wanted are present. That’s exactly what the has and
has_various methods are for:
def has(self, food_name, quantity=1):
“””
has(food_name, [quantity]) - checks if the string food_name is in the
fridge. Quantity defaults to 1
Returns True if there is enough, False otherwise.
“””
return self.has_various({food_name:quantity})
def has_various(self, foods):
“””
has_various(foods) determines if the dictionary food_name
has enough of every element to satisfy a request.
returns True if there’s enough, False if there’s not or if an element does
86

13.8 Page 128

▲back to top
Classes and Objects
not exist.
“””
try:
for food in foods.keys():
if self.items[food] < foods[food]:
return False
return True
except KeyError:
return False
After has and has_various are in place, you can use a Fridge object in tests, and when you read the
code, it will almost make sense when you read your code out loud.
Try It Out Using More Methods
You can now invoke your ch6.py file with python -i or the Run with Interpreter command so that you
can use everything you’ve added to the Fridge class. If you get errors instead of the >>> prompt, pay
attention to the exception raised and try to fix any indentation, spelling, or other basic errors identified.
The class should be usable like this now:
>>> f = Fridge({“eggs”:6, “milk”:4, “cheese”:3})
>>> if f.has(“cheese”, 2):
...
print “Its time to make an omelet!”
...
Its time to make an omelet!
How It Works
Now that you’ve defined new methods, the f object can use them. When you re-created f with the eggs,
milk, and cheese you made the object out of the new Fridge class, so it has the new methods you’ve
added available to it.
Finally, it’s time for the methods to get items from the Fridge. Here you can do the same thing you did
for the methods to add to the Fridge, focusing on a single method that will take care of the hard stuff
and letting the interface methods rely on this hard-working guy:
def __get_multi(self, food_name, quantity):
“””
_get_multi(food_name, quantity) - removes more than one of a
food item. Returns the number of items removed
returns False if there isn’t enough food_name in the fridge.
This should only be used internally, after the type checking has been
done
“””
try:
if not self.has(food_name, quantity):
return False
self.items[food_name] = self.items[food_name] - quantity
87

13.9 Page 129

▲back to top
Chapter 6
except KeyError:
return False
return quantity
After this has been defined, you can create the remaining methods that the Fridge class’s docstring has
specified. They each use __get_multi so that they can remove items from the Fridge with a minimal
amount of extra coding on your part:
def get_one(self, food_name):
“””
get_one(food_name) - takes out a single food_name from the fridge
returns a dictionary with the food:1 as a result, or False if there wasn’t
enough in the fridge.
“””
if type(food_name) != type(“”):
raise TypeError, “get_one requires a string, given a %s” %
type(food_name)
else:
result = self.__get_multi(food_name, 1)
return result
def get_many(self, food_dict):
“””
get_many(food_dict) - takes out a whole dictionary worth of food.
returns a dictionary with all of the ingredients
returns False if there are not enough ingredients or if a dictionary
isn’t provided.
“””
if self.has_various(food_dict):
foods_removed = {}
for item in food_dict.keys():
foods_removed[item] = self.__get_multi(item, food_dict[item])
return foods_removed
def get_ingredients(self, food):
“””
get_ingredients(food) - If passed an object that has the __ingredients__
method, get_many will invoke this to get the list of ingredients.
“””
try:
ingredients = self.get_many(food.__ingredients__())
except AttributeError:
return False
if ingredients != False:
return ingredients
You’ve now written a completely usable class for a refrigerator. Remember that there are many direc-
tions in which you can take this. Although you may be making omelets that use the Fridge class now,
you can also use it for other projects — to model the product flow of a business, for example, such as a
deli that has ten refrigerators with different products in each one.
88

13.10 Page 130

▲back to top
Classes and Objects
When you do find an opportunity to repurpose a class that you’ve written (or a class that you’ve used),
you can take advantage of the opportunity that is presented by adding features to support new needs
without sacrificing what it already does.
For instance, an application that needs to take into account several refrigerators may result in a need for
each Fridge object to have extra attributes, such as a name for it (like “dairy fridge”), its position in the
store, its preferred temperature setting, and its dimensions. You can add these to the class, along with
methods to get and set these values, while still keeping it completely usable for the omelet examples in
this book. This is how interfaces help you. As long as the interfaces to the Fridge class you’ve already
written here aren’t changed, or at least as long as they behave the same, you can otherwise modify any-
thing. This capability to keep interfaces behaving the same is what is called their stability.
Objects and Their Scope
As you saw in Chapter 5, functions create their own space, a scope, for the names that they use. While
the function is being invoked, the name and value are present, and any changes made to the name per-
sist for as long as the function is in use. However, after the function has finished running and is invoked
again, any work that was done in any prior invocations is lost, and the function has to start again.
With objects, the values inside of them can be stored and attached to self on the inside of the object (self
in this case is a name that refers to the object itself, and it’s also the same as what is referenced by a name
on the outside of the object, such as f). As long as the object is referenced by a name that is still active,
all of the values contained in it will be available as well. If an object is created in a function and isn’t
returned by that function to be referenced to a name in a longer-lived scope, it will be available for as
long as the single invocation of the function in which it was called, in the same way as any other data
in the function.
Multiple objects are often created in tandem so that they can be used together. For instance, now that
you’ve implemented all of the features you need to have a workable Fridge in your program, you need
to have an Omelet object that works with it.
Try It Out Creating Another Class
You’ve already created a class — a Fridge. Using the same format, create an Omelet class that you
can use:
class Omelet:
“””This class creates an omelet object. An omelet can be in one of
two states: ingredients, or cooked.
An omelet object has the following interfaces:
get_kind() - returns a string with the type of omelet
set_kind(kind) - sets the omelet to be the type named
set_new_kind(kind, ingredients) - lets you create an omelet
mix() - gets called after all the ingredients are gathered from the fridge
cook() - cooks the omelet
“””
def __init__(self, kind=”cheese”):
“””__init__(self, kind=”cheese”)
This initializes the Omelet class to default to a cheese omelet.
Other methods
89

14 Pages 131-140

▲back to top

14.1 Page 131

▲back to top
Chapter 6
“””
self.set_kind(kind)
return
How It Works
You’ve now got a class whose intent is clearly spelled out. You’ve seen most of these behaviors in func-
tions that you saw in Chapter 5, but now you have a structure within which you can combine all of these
behaviors.
This class will have interface methods that enable the omelet to use a Fridge object cooperatively, and
it will still offer the capability to create customized omelets as it could in Chapter 5.
Remember that all of the following code has to be indented one level beneath the Omelet class to be used:
def __ingredients__(self):
“””Internal method to be called on by a fridge or other objects
that need to act on ingredients.
“””
return self.needed_ingredients
def get_kind(self):
return self.kind
def set_kind(self, kind):
possible_ingredients = self.__known_kinds(kind)
if possible_ingredients == False:
return False
else:
self.kind = kind
self.needed_ingredients = possible_ingredients
def set_new_kind(self, name, ingredients):
self.kind = name
self.needed_ingredients = ingredients
return
def __known_kinds(self, kind):
if kind == “cheese”:
return {“eggs”:2, “milk”:1, “cheese”:1}
elif kind == “mushroom”:
return {“eggs”:2, “milk”:1, “cheese”:1, “mushroom”:2}
elif kind == “onion”:
return {“eggs”:2, “milk”:1, “cheese”:1, “onion”:1}
else:
return False
def get_ingredients(self, fridge):
self.from_fridge = fridge.get_ingredients(self)
def mix(self):
for ingredient in self.from_fridge.keys():
90

14.2 Page 132

▲back to top
Classes and Objects
print “Mixing %d %s for the %s omelet” % (self.from_fridge[ingredient],
ingredient, self.kind)
self.mixed = True
def make(self):
if self.mixed == True:
print “Cooking the %s omelet!” % self.kind
self.cooked = True
Now you have an Omelet class that can create Omelet objects. The Omelet class has the same features
as the process for making omelets in Chapters 4 and 5, but using it is much easier because everything is
combined and the presentation of the Omelet is confined to a few purposefully simpler interfaces.
Now that you have your two classes, you can make an omelet after loading everything with python -i
or the Run with Interpreter command.
>>> o = Omelet(“cheese”)
>>> f = Fridge({“cheese”:5, “milk”:4, “eggs”:12})
>>> o.get_ingredients(f)
>>> o.mix()
Mixing 1 cheese for the cheese omelet
Mixing 2 eggs for the cheese omelet
Mixing 1 milk for the cheese omelet
>>> o.make()
Cooking the cheese omelet!
This isn’t any easier or harder to use than making a single omelet in Chapter 5 was. However, the benefit
of using objects becomes obvious when you have many things to work with at the same time — for
instance, many omelets being made at the same time:
>>> f = Fridge({“cheese”:5, “milk”:4, “eggs”:12, “mushroom”:6, “onion”:6})
>>> o = Omelet(“cheese”)
>>> m = Omelet(“mushroom”)
>>> c = Omelet(“onion”)
>>> o.get_ingredients(f)
>>> o.mix()
Mixing 1 cheese for the cheese omelet
Mixing 2 eggs for the cheese omelet
Mixing 1 milk for the cheese omelet
>>> m.get_ingredients(f)
>>> m.mix()
Mixing 1 cheese for the mushroom omelet
Mixing 2 eggs for the mushroom omelet
Mixing 1 milk for the mushroom omelet
Mixing 2 mushroom for the mushroom omelet
>>> c.get_ingredients(f)
>>> c.mix()
Mixing 1 cheese for the onion omelet
Mixing 2 eggs for the onion omelet
Mixing 1 milk for the onion omelet
Mixing 1 onion for the onion omelet
>>> o.make()
Cooking the cheese omelet!
91

14.3 Page 133

▲back to top
Chapter 6
>>> m.make()
Cooking the mushroom omelet!
>>> c.make()
Cooking the onion omelet!
Take a moment to compare this to how you’d do the same thing using the functions from Chapter 5, and
you’ll realize why so much programming is done in this style — and why this kind of programming,
called object-oriented programming, is used to make larger systems.
As long as the Fridge has the ingredients needed, making different kinds of omelets is very, very easy
now — it involves only invoking the class to create a new object and then just calling three methods for
each Omelet object. Of course, you could reduce it to one. That will be an exercise question.
Summary
In this chapter, you’ve been introduced to how Python provides you with the tools to program with
classes and objects. These are the basic concepts behind what is called object-oriented programming.
When they are used inside a class, functions are referred to as methods because now every one has a
special name called self that, when that method is invoked as part of an object, contains all of the data
and methods of the object.
A class is invoked to create an object by using the class’s name followed by parentheses, (). Initial
parameters can be given at this time and whether or not parameters are given, the newly created object
will invoke the method __init__. Like normal functions, methods in classes (including __init__) can
accept parameters, including optional and default parameters.
The process of creating a class includes deciding what methods should be created to provide all of the
functionality that you want in your class. Two general kinds of methods were described: public inter-
faces that should be invoked on the outside of the objects and private methods that should be called only
by methods inside of the object. The interfaces should be made to change as little as possible, whereas the
internal methods may change without affecting how the class can be used. This is especially important to
remember when using a class written by someone else. Python expects any name within the scope of an
object beginning with two underscores to be private, so this convention should be used by you as well.
Other names are generally considered public.
To specify how you expect the class to be used you should create a docstring for the class by entering a
string on the first line after the class’s definition. In that docstring, it is best to always provide the names
of the methods that you expect to be used, and their purpose. It’s not a bad idea to include an explana-
tion of the class as a whole, too.
All of the names that are defined in a class (both data and methods) are distinct in each object that is cre-
ated. When a method is invoked in one object and that changes data in that object, other types of the
same object are not affected. Examples of this that are built in to Python are strings, which are objects
that include special methods that help with common tasks when you are using text.
To make objects easier to use, it’s common to provide multiple interfaces that behave similarly. This can
save you a lot of work, by finding ways for these interfaces to call a single internal method that is more
92

14.4 Page 134

▲back to top
Classes and Objects
complex or accepts more parameters than the interfaces. This gives you two distinct advantages. First, it
makes the code that calls on these methods easier to read because the names of the parameters don’t
need to be remembered by the programmer — the name of the method provides needed information to
the programmer. Second, if you need to change the internal method that its related interfaces call on,
you can change how all of them behave by just changing the internal method. This is especially useful
when fixing problems because a single fix will correct how all of the interfaces work as well. In addition,
the method that provides this support to other methods can itself be a public interface. There’s no strict
rule about whether a hard-working method like this should be private and internal or not. It’s really up
to you.
One goal of writing objects is to duplicate as little code as possible, while providing as many features
as possible. Creating a class that can use objects can save a lot of code writing because they are usually
manipulated more conveniently than when functions and data are kept separated because methods
within the same class can count on the methods and data that they use being present. Groups of classes
can be written so that they have interdependent behaviors, enabling you to model groups of things
that work together. You will learn how to structure these interdependent and cooperative classes in
Chapter 7.
Last, you’ve seen how codeEditor’s Python shell helps you explore your objects by showing you all
of the interface names once you type a period. This is much easier than typing dir to get the same infor-
mation because of the more convenient and easier to use manner in which codeEditor displays the
information.
Exercises
Each of the following exercises builds on the exercises that preceded it:
1. Add an option to the Omelet class’s mix method to turn off the creation messages by adding a
parameter that defaults to True, indicating that the “mixing ...” messages should be printed.
2. Create a method in class Omelet that uses the new mix method from exercise 1. Called
quick_cook, it should take three parameters: the kind of omelet, the quantity wanted, and the
Fridge that they’ll come from. The quick_cook method should do everything required instead
of requiring three method calls, but it should use all of the existing methods to accomplish this,
including the modified mix method with the mix messages turned off.
3. For each of the methods in the Omelet class that do not have a docstring, create one. In each
docstring, make sure you include the name of the method, the parameters that the method
takes, what the method does, and what value or values it returns upon success, and what it
returns when it encounters an error (or what exceptions it raises, if any).
4. View the docstrings that you’ve created by creating an Omelet object.
5. Create a Recipe class that can be called by the Omelet class to get ingredients. The Recipe
class should have the ingredient lists of the same omelets that are already included in the
Omelet class. You can include other foods if you like. The Recipe class should include methods
to retrieve a recipe, get(recipe_name), a method to add a recipe as well as name it, and
create (recipe_name, ingredients), where the ingredients are a dictionary with the same
format as the one already used in the Fridge and Omelet classes.
93

14.5 Page 135

▲back to top
Chapter 6
6. Alter the __init__ method of Omelet so that it accepts a Recipe class. To do this, you can do
the following:
a. Create a name, self.recipe, that each Omelet object will have.
b. The only part of the Omelet class that stores recipes is the internal method
__known_kinds. Alter __known_kinds to use the recipes by calling
self.recipe.get() with the kind of omelet that’s desired.
c. Alter the set_new_kind method so that it places the new recipe into
self.recipe and then calls set_kind to set the current omelet to the kind just
added to the recipe.
d. In addition, modify __known_kinds to use the recipe method’s get method to
find out the ingredients of an omelet.
7. Try using all of the new classes and methods to determine whether you understand them.
94

14.6 Page 136

▲back to top
7
Organizing Programs
In Chapter 6, you began using Python’s features to create separate classes that can be used to cre-
ate entirely self-contained objects. Classes and the objects that are created from them are tools that
enable you to gather data and functions into a contained space so that they can be viewed as part
of a larger entity.
So far, the definitions of classes have all been in a single file and were not run in the way you nor-
mally think of programs being run. Instead, they were invoked interactively so that you could use
them as you would from within another program. However, if you wanted to use the classes
you’ve written with what you know so far, you would make the same file that defined the classes
the program. That means putting all of the classes at the beginning of the file, and the important
decision making code at the end. The end is where it takes the most time to find the code that
you’re going to want to find the most often.
Another cautionary note needs to be sounded. Classes are very useful, but not all problems should
be solved by creating a class. Sometimes the work of designing them is overkill, and other times
what you really need are functions that don’t require the long life span that data and methods can
have in objects.
To make Python more useful, therefore, it offers you the great feature of enabling you to create
modules that create a named scope for functions and data, but which are simpler than classes and
objects. Modules give you a tool to separate your program into distinctly named pieces, without
using classes to do it. In fact, classes can be defined within a module.
As an extension of this, you can also divide these modules into different files; Python calls this fea-
ture a package. Packages enable you to divide your programs among several files and even into
separate directories to help you organize your programs.
So far, you have only been introduced to intrinsic pieces of the Python language — things that deal
with how Python itself works. Python is also very flexible, and though it comes with a small core
set of features, these are expanded in a variety of modules. To extend Python to use features pro-
vided by the operating system, there is a module called os. To extend Python to have networking
features, Python provides modules that offer both low-level networking (such as sockets) and

14.7 Page 137

▲back to top
Chapter 7
higher-level protocols (such as http, ftp, and so on). Many modules come with Python, but because it is
very easy to write modules, a variety of additional modules are available from third parties, both com-
mercial and free.
By the end of this chapter, you will have learned how to write simple modules for your own use or to
share. You’ll also be introduced to some of the bundled Python modules. You will be familiar with the
concept of importing modules, and you will be able to use packages to contain useful functions and
names, separately from the global scope. You will also find out more about how scope can be used to
your advantage for tasks such as testing your packages.
Modules
Modules present a whole group of functions, methods, or data that should relate to a common theme.
Such a theme might be networking components (see Chapter 16), performing more complicated work
with strings and text (see Chapter 12), dealing with graphical user interfaces (see Chapter 13), and other
services.
After you’ve learned how to program in a language, you often find that you need to work with compo-
nents that the language doesn’t initially bundle. Python, by itself, is no different. At its core, it is a very
small and simple language that doesn’t offer many special features. However, because of its simplicity,
it is easy to use as a platform that can be extended with additional functions and objects that can be
used by anyone.
Importing a Module So That You Can Use It
To make a module usable, two things need to be available. First, the module itself has to be installed on
the system. For the most part, you’ll find that a lot of the basic things you want to do, such as reading
and writing files (more on this in Chapter 8) and other fundamental important things that differ between
platforms, are available as bundled modules with Python — that is, they are free and universally avail-
able with the language.
The simplest way to begin using a module is with the import keyword:
import sys
This will import the module named sys that contains services Python offers that mostly involve
system-specific items. This means that it relates to things that involve how the system works, how a
particular installation of Python is installed, or how the program you’ve written was invoked from the
command line.
To start looking at modules, you’re also going to begin to write in a style that facilitates running the file
you’re working on by itself, as a standalone program. To that end, create a file called ch7.py and type
the following:
#!/usr/bin/env python2.4
# Chapter 7 module demonstration
import sys
96

14.8 Page 138

▲back to top
Organizing Programs
The first line is for users of Linux and other Unix systems (or Python under a Unix based environment
like Cygwin). This is a way to get the python2.4 binary run in case other Python interpreters are on
the system. See the web site for this book for more information on running Python. For Window and
Macintosh systems, the file extension should provide information that the operating system needs to
launch the Python interpreter, whether it’s python, python2.4, or some other name when it’s installed
on your system (although some configuration may be needed). See the web site for more information on
this, too.
Making a Module from Pre-existing Code
To create a module, all you need to do is choose a name for your module and open a file with that name
and the extension .py in your editor. For example, to create a Foods module, you only have to create a
file called Foods.py. When that’s finished, you can import it using the name “Foods” without the .py at
the end. That’s it! You’ve imported a simple module.
Try It Out Creating a Module
Take your file with all of the source code from Chapter 6 and copy it to a file called Foods.py. When
you’ve done this, open the Python shell so you can import the Foods module:
>>> import Foods
>>> dir(Foods)
[‘Fridge’, ‘Omelet’, ‘Recipe’, ‘__builtins__’, ‘__doc__’, ‘__file__’, ‘__name__’]
>>>
How It Works
You now have access to the Fridge class, the Omelet class and, from the previous exercises, the Recipe
class. Together, you have a file that is a module that contains all of these classes, and they’ll be able to
work together. However, you’ll now access them through the name Foods.Fridge, Foods.Omelet, and
Foods.Recipe, and they remain fully usable, albeit with some new rules.
Be aware that this is the first time you’re getting the examples in the book to be run directly with your
computer! By default, Python keeps a list of directories in which it will look for modules to load. This list
contains several directories, though the exact locations of all of them will depend on how your running
Python interpreter was installed. Therefore, if you’re trying to import the Foods module but the shell
has started itself in a directory other than the one in which you’ve saved the Foods.py file, you’re going
to receive an error (but you can fix this by changing to the right directory).
This path, or list of directories that Python should search through, is stored in the sys module, in a vari-
able named path. To access this name, you will need to import the sys module. Until you do that, the
sys.path won’t be available to you:
>>> import sys
>>> print sys.path
[‘’, ‘D:\\\\Python24\\\\Lib\\\\site-packages\\\\PythonCard\\\\tools\\\\codeEditor’,
‘C:\\\\WINNT\\\\system32\\\\python24.zip’, ‘C:\\\\Documents and Settings\\\\J Jones’,
‘D:\\\\Python24\\\\DLLs’, ‘D:\\\\Python24\\\\lib’, ‘D:\\\\Python24\\\\lib\\\\plat-win’,
‘D:\\\\Python24\\\\lib\\\\lib-tk’, ‘D:\\\\Python24’, ‘D:\\\\Python24\\\\lib\\\\site-packages’,
‘D:\\\\Python24\\\\lib\\\\site-packages\\\\wx-2.5.3-msw-ansi’]
97

14.9 Page 139

▲back to top
Chapter 7
You can see that sys.path is a normal list, and if you want to add directories that will be checked for
your modules, because you want them somewhere that isn’t already in sys.path, you can alter it by
using the usual methods — either the append method to add one directory, or the extend method to
add any number of directories.
When you’ve imported the Foods module, as above, you can use codeEditor’s feature of interactively
helping you by popping up a list of all of the names in the scope of the module while you’re typing in
a name. Every time you come to a period, if the name you’ve just typed in has names associated with
you, codeEditor will allow you to select from the interfaces that the name provides. This will help you
explore the module you’ve just created but is even more useful with larger, more complex modules!
You can now run through examples from the prior chapters, but now you access your classes through
the Foods module. For instance, you can invoke Foods.Fridge, but not just Fridge by itself. If you
wanted to access Fridge alone, you’ll see how to do this soon.
Try It Out Exploring Your New Module
codeEditor provides you with a special feature in the Python shell that will interact with you as you
type. You may have noticed already that when you finish typing the name of something such as a class
or a module, when you type a period at the end of the name, within the shell a menu of names that exist
in the scope of the module or object is shown to you. Figure 7-1 shows what this looks like, so you can
do the same for yourself.
Figure 7-1
How It Works
As you type in codeEditor’s Python shell, it evaluates what you are typing as you type When it notices
that you’ve typed certain characters, it takes actions on them. You notice this when strings take on a
98

14.10 Page 140

▲back to top
Organizing Programs
different color once you type in any kind of quote, or when words that are special to Python are given
colors. Whenever the shell sees that you’re typing a period, it knows that what you’re typing will be
looking inside a module or an object, so it queries that object behind the scenes and shows you the
results so you can work with it.
Using Modules — Starting With the Command Line
So far, you’ve started by using import with a module name by itself. When a module is imported this
way, all of the names it contains are put into a scope that is named for the module — that is, the name
that was used in the import statement.
For example, in the case of sys, everything available is referred to by using the name sys, followed by
a period, and then the name inside of sys, such as sys.path or sys.copyright, which, as it suggests,
specifies the copyright on Python (Programmers love to be clever like that). Now that you know how
modules are structured, you can interactively explore the sys module with the codeEditor Python shell,
or with the dir function, as you saw in Chapter 6. (dir will show you even more than the helpful dialog
box in the codeEditor shell, as it shows private names that aren’t part of the interface of the module.
These concepts, which you’ve seen in classes and objects, still apply to modules!) You can also explore
the docstrings that are present in the module and in the functions and classes provided by it.
On Unix and Unix-like environments, it’s common to ask users to provide command-line parameters
that will determine how a program as a whole will behave. This is conceptually very similar to how
functions use parameters in Python. These command-line parameters show up in Python programs as
a special name inside the sys module. That name is argv. This name may not make much sense at first,
but it’s an important term to know because it is common across most languages and platforms.
argv is an abbreviation for the term argument vector. In computer programming lingo, argument is
another word for what you’ve seen called a parameter. This term is used with functions and when you
run a program with parameters on the command line (another word for parameters and arguments on
the command line are flags). A vector is another word for a list of options. In some languages, it has a
very specific and different meaning, but Python doesn’t make the same distinction, so you don’t have to
worry about it.
If you translate argv back through those definitions, you’ll see that it simply means the parameters that
were on the command line, accessible as a list (see Figure 7-2)! It’s hard to convert that information into
a short and comprehensible word that makes sense in English (or any other nonprogramming language
that the author has heard of), so the term argv persists.
To print out the parameters from the command line, you just have to use sys.argv as you would with
any other list:
print “This was given the command line parameters: %s” % sys.argv
To make running this the same procedure on any platform, you can launch this from codeEditor. Select
File ➪ Run Options and then put anything you want in the Other argv field. You’ve used this facility
before, starting in Chapter 5, but taking advantage of the Run Options dialog box’s capability to let you
set the command line that your program will start with is something new.
99

15 Pages 141-150

▲back to top

15.1 Page 141

▲back to top
Chapter 7
For testing programs that are changing and that aren’t meant to be used interactively, you are generally
better off using python -i or Run with Interpreter; this way, you can try running your program repeat-
edly, starting it program from the beginning each time.
Figure 7-2
Try It Out Printing sys.argv
Now, anytime you run this program using the Run with Interpreter option from your File menu, you
will get a printed representation of the list that becomes the sys.argv. For example, if the command-
line arguments provided in the Other args field were “test 123 test”, your program will print something
like the following (which was run on Windows, while a Unix shell would have a very different looking
sys.path):
This was given the command line parameters: [‘D:\\\\Documents\\\\Chapter7.py’, ‘test’,
‘123’, ‘test’]
How It Works
The first element of the sys.argv list will always be the name of the program, and anything else will
become the elements of the sys.argv list, starting at the number one in the list.
Classes that live within a module are accessed in the same way as any other name. For modules that pro-
vide classes that you use, the invocation is what you’d expect — just the addition of the parentheses to
the fully spelled out path, such as calling Foods.Recipe().
100

15.2 Page 142

▲back to top
Organizing Programs
Changing How Import Works — Bringing in More
Import can be used alone; when it’s used that way, it creates a named scope from which everything in
the module can be referenced. Sometimes it can be useful to have specific parts of the module brought
into your program’s top-level global scope, though. Eliminating the need to type the name of the mod-
ule before the function or class you have to access reduces a lot of typing and makes your code a lot
more straightforward. With your Foods module, you have to do the following to get an onion Omelet:
import Foods
r = Foods.Recipe()
onion_ingredients = Foods.Omelet(r, “onion”)
You can see by this example that when you want to invoke or access something inside of a module, it
means spelling out the entire path. You can quickly tire of doing this. However, you can change this
behavior by bringing the names you want closer in to your code, by using the from modifier to the
import command:
from Foods import Omelet
from Foods import Recipe
r = Recipe()
onion_ingredients = Omelet(r, “onion”)
If you have to descend more levels, such as to (the made-up food) Foods.Recipes.Breads.Muffins
.Bran and you want to bring the names from Bran into the current scope, you’d write something simi-
lar. It would look like you’d expect:
from Foods.Recipes.Breads.Muffins import Bran
Packages
After you’ve gotten a module built and in its own file, it’s not uncommon to find that a single file runs
headlong into organizational issues. Mainly, the issue is that an individual class becomes more useful on
its own and may gain far more code than all of the rest of the classes in the module. This would be a
good reason to move it to its own file, but that would break code that already uses the module!
However, there is a solution.
To provide a structure for doing this, Python provides the organizational idea of packages. Packages use
the structure of the directories (another name for folders) that every operating system uses to give you a
methodology for making many files in the same directory look like a single module when they’re used
together.
You can start by simply making the directory. Let’s break up the Foods module. First, you need to use a
new name — Foods.py already exists, and it would be confusing to keep working with the module by
calling it “Foods”. Therefore, to work around that, let’s start working on a new package, and call this
new one the Kitchen package (this name is also general enough to leave you a lot of room for your
imagination to work with later if you’d like to).
Simply enough, create a Kitchen directory. Then create a file in Kitchen called __init__.py (this
name has to be the same name as the method in a class that you’ve seen already, and note that it has two
101

15.3 Page 143

▲back to top
Chapter 7
underscores before and after the name). This file is the hint that tells Python that this is a package direc-
tory, and not just a directory with Python files in it. This is important because it ensures that you know
you’re responsible for maintaining this and controlling its behavior. This file has a lot of control over
how the package is going to be used, because unlike a module, when a package is imported, every file
in the directory isn’t immediately imported and evaluated. Instead, the __init__.py file is evaluated,
and here you can specify which files are used and how they’re used!
Try It Out Making the Files in the Kitchen Class
To make your three already written classes a part of the Kitchen package, create four files underneath
the Kitchen directory and place the appropriate classes into each of the files named after a class
name. Remember that under all versions of Windows, anywhere you see a forward slash (/) you should
use a backslash (\\) because that’s what Windows uses to separate directories. In other words, create the
Kitchen/Fridge.py file inside the Kitchen directory, and you’ll put only the Fridge class in it.
Make one file for each of the classes, as well as making for the __init_ _.py file:
❑ Kitchen/Fridge.py — All of the code and comments for the Fridge class should go in here,
starting from where your ch6.py says class Fridge:.
❑ Kitchen/Omelet.py — All of the code and comments for the Omelet class should go here. Use
the revision of the Omelet class that you have as the solution to the Exercises from Chapter 6.
❑ Kitchen/Recipe.py — All of the code and comments for the Recipe class should go here.
❑ Kitchen/__init__.py (remember to use two underscores before and after the filename) —
Nothing has to go in this file.
How It Works
You have a class in each file and __init__.py created, so you can now import the Kitchen package.
However, when you import Kitchen, Python evaluates only __init__.py. This is a very important
detail, because without putting some further code in __init__.py, you’ll never get to see your code.
Currently, nothing is actually imported if you do what you’d assume you should do by default, which is
import Kitchen!
To make all of your classes available when you’ve imported Kitchen, you need to put explicit import
statements in __init__.py:
from Fridge import Fridge
from Recipe import Recipe
from Omelet import Omelet
After you’ve added these lines to __init__.py, you have all of these classes available when you’ve
imported the Kitchen package:
>>> import Kitchen
>>> r = Kitchen.Recipe()
>>> r.recipes
{‘cheese’: {‘cheese’: 1, ‘eggs’: 2, ‘milk’: 1}, ‘onion’: {‘cheese’: 1, ‘eggs’: 2,
‘milk’: 1, ‘onion’: 1}, ‘mushroom’: {‘cheese’: 1, ‘eggs’: 2, ‘milk’: 1, ‘mushroom’:
2}}
102

15.4 Page 144

▲back to top
Organizing Programs
By itself, this doesn’t buy you much yet because this is only a very small project, but for any project that
begins to grow, this facility is very important and can make development among multiple developers far
easier by letting the natural assignment of functions and classes be divided into files, enabling each pro-
grammer to work on his or her own group of files in the package.
Modules and Packages
Now that modules and packages have been defined, you will continue to see how to use them — mostly
interchangeably. You’ll generally have your attention drawn to where packages behave differently
from a single module. Because the module has been named Foods and the package has been named
Kitchen, you won’t be confused when you’re shown something that deals with a package instead of a
module. Just remember: Kitchen references are highlighting packages; Foods references are highlight-
ing modules.
Bringing Everything into the Current Scope
Note a special feature of modules: Sometimes you may want to have the entire contents of a module
available without having to specify each name that is available from it explicitly. To do this, Python pro-
vides a special character, the asterisk, which can be used with the from . . . import . . . statement.
It’s important to understand that you can only import using the * when you are importing into the
global scope:
from Foods import *
This would bring Omelet into your current scope, as well as everything else at the top of the recipe
module. In other words, now you no longer have type Foods.Omelet(), just Omelet(), and you need
to do this only once, instead of one time for each name you want to make local.
Packages can be made to work in a similar fashion, but underneath, they actually work differently. For
packages, you need to specify the names you want to be provided when from . . . import *, and these
need to be stated explicity. You can make the three modules in the Kitchen package available by using
the __all__ list in __init__.py. Any names that appear in the __all__ list will be exported by the *
but only those names.
The elements that are present in the __all__ list are the names of functions, classes, or data that will be
automatically imported into the global scope of the program that is asked to import *.
You can expect users of modules and packages you write to automatically use the from . . . import *
syntax within their programs. To work with packages, you must specify a list of names that will be
exported! However, if you have a large module, you can also create an __all__ list at the top of your
module file, and it will also have the effect of restricting the names in the module in the same way as it
would in a package.
103

15.5 Page 145

▲back to top
Chapter 7
Try It Out Exporting Modules from a Package
The __all__ list exists because using from . . . import * is common. You will use (at first) and write
(later) packages that have many layers, functions, data names, and individual modules that a user
shouldn’t see — they’re not part of your public interface. Because you need to be careful about over-
whelming a user with a lot of things they don’t need, the __all__ list enforces your interface decisions.
__all__ = [‘Fridge’, ‘Recipe’, ‘Omelet’]
How It Works
Now these names will come into the global space of your program when you invoke them with from
Kitchen import *. It’s important to know that if your __init__.py looked like this:
from Fridge import Fridge
from Omelet import Omelet
__all__ = [‘Omelet’, ‘Recipe’, ‘Fridge’]
With the from Recipe import Recipe statement eliminated, you would have to invoke
Recipe.Recipe() to create a new recipe object after calling from Kitchen import *.
Re-importing Modules and Packages
Programming involves a lot of trial and error. You will often realize that you’ve made a mistake in the
work you’ve done in your module while you’re in the shell interactively. Because you may have done
a lot of typing to get your shell set up perfectly for your test before your problem module was loaded,
you’d like to be able to fix your module and have Python re-load it so that you can save yourself the
work of having to set up your session again. So far, you haven’t been shown how to do this, but you can.
The first thing you need to know to do this is that it’s normal for a common module to be required by
multiple other modules and effectively be called up multiple times in the same program. When this
happens, instead of going through the extra time it would take to re-load, re-evaluate, and re-compile
the module each time (see the sidebar “Compiling and .pyc Files”), Python stashes away the name of
the module, and where it came from, in a special dictionary of all the modules that have been imported
so far, called sys.modules. In fact, when you use the Python shell from within codeEditor, its already
loaded sys and many other modules for you, so any time you’ve called it in your own, you’ve had
this happen!
Compiling and .pyc Files
If you’ve looked at your ch5.py, ch6.py, or any other Python files that you’ve
worked on so far, you’ll notice that after you run them, a file with almost the same
name appears — the difference is that it ends in .pyc. This is a special file that Python
writes out that contains a form of your program that can be loaded and run faster
than the plaintext source code. If you make changes to the .py file, the next time it is
invoked (that is, by double-clicking it, running python -i, or using the Run or Run
with Interpreter menu options in codeEditor), Python will re-create the .pyc file from
the newer, changed source code that you’ve updated.
104

15.6 Page 146

▲back to top
Organizing Programs
Try It Out Examining sys.modules
If you look at the list returned by sys.modules.keys, you’ll see the name of every module that’s
loaded. Even if you start a Python shell outside of codeEditor, you’ll find that after you’ve imported sys
and can look at sys.modules, many modules are loaded by the system without your knowledge. Each
operating system and installation will have slight variations on the exact contents of the dictionary, but it
will usually look something like this:
>>> sys.modules.keys()
[‘copy_reg’, ‘__main__’, ‘site’, ‘__builtin__’, ‘Kitchen.Omelet’, ‘encodings’,
‘posixpath’, ‘encodings.codecs’, ‘os.path’, ‘_codecs’, ‘encodings.exceptions’,
‘stat’, ‘zipimport’, ‘warnings’, ‘encodings.types’, ‘UserDict’, ‘encodings.ascii’,
‘’sys’,’codecs’, ‘readline’, ‘types’, ‘signal’, ‘linecache’, ‘posix’,
‘encodings.aliases’, ‘exceptions’]
How It Works
Depending on the operating system and when you call it, the sys.modules dictionary shows you all of
the modules that have been called. For modules that you haven’t explicitly imported, you can assume
that they are automatically called in by Python to handle things like the operating system or other mech-
anisms that Python doesn’t force you to deal with directly. The preceding sample is from a Linux system,
and certain things are obviously OS-related — posix and posixpath, for example, if you have worked
with Unix — while some other things are not.
You can take this opportunity to look at the values associated with any keys that interest you. You’ll see
that some modules are listed as built-in and some are listed as being from a file, and when this is the
case, the entire path to the module file is listed in the information that the module provides to you. Don’t
worry if the list of modules that comes up in your Python shell looks very different from the preceding
example. After you’ve loaded the Foods module, it will be present in the sys.modules dictionary, and
when it’s there, Python will not re-evaluate the Foods.py module, even if you’ve changed it! To fix this
in an interactive session, you can simply remove the record of the Foods module from the sys.modules
dictionary and then import the module again. Because Python no longer has a record in sys.modules it
will do as you ask instead of trying to save effort as it did before you removed the reference:
>>> import Kitchen
>>> sys.modules.has_key(‘Kitchen’)
True
>>> sys.modules[‘Kitchen’]
<module ‘Kitchen’ from ‘Kitchen\\__init__.py’>
>>> sys.modules.pop(‘Kitchen’)
<module ‘Kitchen’ from ‘Kitchen\\__init__.py’>
>>> sys.modules[‘Kitchen’]
Traceback (most recent call last):
File “<input>”, line 1, in ?
KeyError: ‘Kitchen’
However, now that you know how this works under the hood, you also need to know that you have a
simplified way of doing the same thing. Python provides a built-in function called reload that reloads
the module you specified as though you’d done the manual labor you’ve just seen:
105

15.7 Page 147

▲back to top
Chapter 7
import Kitchen
reload(Kitchen)
<module ‘Kitchen’ from ‘Kitchen\\__init__.pyc’>
Note that this doesn’t change any objects that already exist. They’re still potentially tied to the old defini-
tion, which you could have changed in the module you’ve just reloaded! If you altered the Recipe and
the Omelet classes, you’d need to re-invoke the classes and use them to re-create new versions of all
objects of these types, but you already know how to initialize objects:
>>> r = Omelet.Recipe()
>>> o = Omelet.Omelet(r, ‘onion’)
Basics of Testing Your Modules
and Packages
There is a very interesting side effect of the scope that is created for modules. Within your program is
always a special name, __name__, that tells you what the scope you’re running in is called. For instance,
if the value of __name__ were checked from within the Foods module, it would return the string ‘Foods’.
One special reserved name, the name of the top-level global scope, is __main__. If you have a module
that’s normally never used directly, you can stick some code at the end that has one purpose in life —
verifying that your module works! This is a great opportunity to make your testing easy.
You’ll have many occasions when you see a module with the following code at the end:
if __name__ == ‘__main__’:
You can use this statement at the end of your modules; and from this point on, you can have tests that
will ensure that classes are made, that functions will return the values that you expect, or any other tests
you can think of. It’s very common as you program to have situations in which something that once
worked suddenly breaks. It’s always a great idea to place tests for these situations in your packages so
that you never forget that they can happen, and you can be ahead of the game! There is a lot more infor-
mation about testing in Chapter 12.
Summary
In the previous chapters, you learned how to write code at the interactive Python shell, as well as put
code into individual files that can be run. In this chapter, you’ve been shown ways of organizing your
programs into modules and packages.
Modules are distinct names that Python uses to keep a scope for local names. Within a module, a name
can be used directly; however, from outside of a particular module (for instance, in the global top-level
scope whose name is actually __main__), the names within a module can be accessed by first specifying
the name of the module where the name you want to use is defined, followed by a period, followed by
the name you’re looking for. An example of this is sys.path. This enables you to use the same name in
different modules for different purposes, without being confusing.
106

15.8 Page 148

▲back to top
Organizing Programs
To use a module, it must be brought into your program with the import statement. Import will find a
file with the name of the module you want to use, with the extension .py, and make it available. It does
this by examining each of the directories in the list sys.path until it finds the file.
You will often want specific parts of a module to be available with less typing than the entire specifi-
cation would require — the long form would be the name of the module, any intermediate modules
(separated with periods), and then the name you actually want. In such cases, you can use the construct
from . . . import . . . to just import names that you will be frequently using. When a module is
imported, it is evaluated, and any code that is not inside of a function or a class will be evaluated.
When you have a lot of code to write, you can use a package to group your code into a structure that is
provided by the underlying file system of your operation system. This structure begins with a directory
(the same thing as a folder), which will be the name of the package when it is imported into your pro-
gram. What makes a directory into a package is the presence of a file called __init__.py. This file will
be read and parsed, and it can contain any code that could be useful to the entire package, such as data
that should be available to all parts of the package, such as version information, locations of important
files, and so on, as well as import statements that could be required to bring in modules that will be
needed in order for other parts of the package to work correctly.
When you have a package, the files in that package will not be automatically exported when a program-
mer requests it by using from . . . import *, even if those files are modules that have been imported
inside of __init__.py. With a package, the names that will be exported by default to this request have
to be specified in a list called __all__.
Exercises
Moving code to modules and packages is straightforward and doesn’t necessarily require any changes to
the code to work, which is part of the ease of using Python.
In these exercises, the focus is on testing your modules, as testing is essentially writing small programs
for an automated task.
1. Write a test for the Foods.Recipe module that creates a recipe object with a list of foods, and
then verifies that the keys and values provided are all present and match up. Write the test so
that it is run only when Recipe.py is called directly, and not when it is imported.
2. Write a test for the Foods.Fridge module that will add items to the Fridge, and exercise all
of its interfaces except get_ingredients, which requires an Omelet object.
3. Experiment with these tests. Run them directly from the command line. If you’ve typed them
correctly, no errors should come up. Try introducing errors to elicit error messages from
your tests.
107

15.9 Page 149

▲back to top

15.10 Page 150

▲back to top
8
Files and Directories
In this chapter, you’ll get to know some of the types and functions that Python provides for writ-
ing and reading files and accessing the contents of directories. These functions are important,
because almost all nontrivial programs use files to read input or store output.
Python provides a rich collection of input/output functions; this chapter covers those that are
most widely used. First, you’ll use file objects, the most basic implementation of input/output
in Python. Then you’ll learn about functions for manipulating paths, retrieving information about
files, and accessing directory contents. Even if you are not interested in these, make sure that you
glance at the last section on pickling, which is an extremely handy tool for storing and retrieving
Python objects.
File Objects
The simplest way to read and write files in Python is with a file object. It represents a connection
to a file on your disk. Because file is a built-in type, there is no need to import any module before
you use it.
In this chapter, most of the examples use Windows path names. If you are working on a different
platform, replace the example paths with paths appropriate for your system.
If you do use Windows, however, remember that a backslash is a special character in a Python
string, so you must escape (that is, double up) any backslash in a path. For instance, the path
C:\\Windows\\Temp is represented by the Python string “C:\\\\Windows\\\\Temp”. If you prefer, you
can instead disable special treatment of backslashes in a string by placing an r before the opening
quotes, so this same path may be written r”C:\\Windows\\Temp”.
We’ll use a string object to hold the path name for a sample file we’ll create and access. If you’re
using Windows, enter the following (you can choose another path if you want):
>>> path = “C:\\\\sample.txt”

16 Pages 151-160

▲back to top

16.1 Page 151

▲back to top
Chapter 8
Other Uses of file Objects
A file object is actually more general than a connection to a disk file. It can represent a
network connection, a connection to a hardware device such as a modem, or a connec-
tion to another running program. If you understand how to use file objects, you are
one step closer to understanding network programming and other advanced topics.
If you’re using Linux, enter the following (or choose a path of your own):
>>> path = “/tmp/sample.txt”
Writing Text Files
Let’s start by creating a file with some simple text. To create a new file on your system, create a file
object, and tell Python you want to write to it. A file object represents a connection to a file, not the file
itself, but if you open a file for writing that doesn’t exist, Python creates the file automatically. Enter the
following:
>>> sample_file = file(path, “w”)
The first argument is the path where Python creates the file. The “w” argument tells Python that you
intend to write to the file; without it, Python would assume you intend to read from the file and would
raise an exception when it found that the file didn’t exist.
When opening a file, and with all the other file-manipulation functions discussed in this chapter, you
can specify either a relative path (a path relative to the current directory, the directory in which your
program or Python was run) or an absolute path (a path starting at the root of the drive or file system).
For example, /tmp/sample.txt is an absolute path, while just sample.txt, without the specification
of what directory is above it, is a relative path.
Using the file object’s write method, you can write text to the file:
>>> sample_file.write(“About Pythons\\n”)
Because write doesn’t add line breaks automatically, you must add one yourself with the escape
sequence \\n wherever you want a line break in the file.
If you use write again, the text is appended to what you wrote before. If the string you pass is more
than one line long, more than one line is added to the file:
>>> sample_file.write(“””
... Pythons are snakes. They eat small mammals, killing
... them by squeezing them to death.
... “””)
We’ve used a multi-line triple-quoted string here. Until you close the triple quotes, Python prompts you
to continue the string with “...”. In a multi-line string, Python adds line breaks between lines.
110

16.2 Page 152

▲back to top
Files and Directories
If you prefer the print statement, you may use it to write to a file, like this:
>>> print >> sample_file, “The end.”
Be careful here with the punctuation: Python prints the first >>> as its prompt, while you type the >>
after print to specify that the output should be added to your file. Unlike write, the print statement
adds a line break after the text you specify; to suppress it, end the print statement with a comma.
When you’re done writing text, you must close the file. The text you wrote may not actually be written
to disk until you do so. To close the file, you can simply delete the file object. This doesn’t delete the
file. It only deletes the Python file object, which represents a connection to the file on disk and thus
closes the file. You’ll learn later in the chapter how to delete the actual file.
>>> del sample_file
If you had created sample_file inside a function, Python would have deleted it automatically upon
returning from the function, but it’s a good idea to delete the file object explicitly to remind yourself
that the file is being closed.
Reading Text Files
Reading from a file is similar. First, open the file by creating a file object. This time, use “r” to tell
Python you intend to read from the file. It’s the default, so you can omit the second argument altogether
if you want.
>>> input = file(path, “r”)
Make sure you use the path to the file you created earlier, or use the path to some other file you want to
read. If the file doesn’t exist, Python will raise an exception.
You can read a line from the file using the readline method. The first time you call this method on a
file object, it will return the first line of text in the file:
>>> input.readline()
‘About Pythons\\n’
Notice that readline includes the newline character at the end of the string it returns. To read the con-
tents of the file one line at a time, call readline repeatedly.
You can also read the rest of the file all at once, with the read method. This method returns any text in
the file that you haven’t read yet. (If you call read as soon as you open a file, it will return the entire con-
tents of the file, as one long string.)
>>> text = input.read()
>>> print text
Pythons are snakes. They eat small mammals, killing
them by squeezing them to death.
The end.
111

16.3 Page 153

▲back to top
Chapter 8
Because you’ve used print to print the text, Python shows newline characters as actual line breaks,
instead of as \\n.
When you’re done reading the file, close the file by deleting the file object:
>>> del input
It’s convenient to have Python break a text file into lines, but it’s nice to be able to get all the lines at one
time — for instance, to use in a loop. The readlines method does exactly that: It returns the remaining
lines in the file as a list of strings. Suppose, for instance, that you want to print out the length of each line
in a file. This function will do that:
def print_line_lengths(path):
input = file(path)
for line in input.readlines():
print len(line)
Try It Out Printing the Lengths of Lines in the Sample File
Using the function print_line_lengths, you can examine the file you just created, displaying the
length of each line:
>>> print_line_lengths(“C:\\\\sample.txt”)
14
1
53
33
9
How It Works
Each line is read as a string. Each line, as it’s read, has its length displayed by using the string as an
argument to the len function. Remember that the newline character is included in each line, so what
looks like an empty line has a length of one.
Looping over the lines in a text file is such a common operation that Python lets you use the file object
itself as if it were the lines in the file. Therefore, if you’re in a rush, you can get the same effect as the pre-
ceding function with the following:
>>> for line in file(path):
...
print len(line)
...
14
1
53
33
9
You may sometimes see programs use the open function instead of calling file to create file objects.
The two are equivalent, but older versions of Python only provided open for this purpose. Calling file
to create file objects is more consistent with Python’s type system, in which you call a type to create an
instance of it, so you should use file instead of open in your programs, unless you intend to support
older versions of Python.
112

16.4 Page 154

▲back to top
Files and Directories
File Exceptions
Because your Python program does not have exclusive control of the computer’s file system, it must be
prepared to handle unexpected errors when accessing files. When Python encounters a problem per-
forming a file operation, it raises an IOError exception. (Exceptions are described in Chapter 4.) The
string representation of the exception will describe the problem.
There are many circumstances in which you can get an IOError, including the following:
❑ If you attempt to open for reading a file that does not exist
❑ If you attempt to create a file in a directory that does not exist
❑ If you attempt to open a file for which you do not have read access
❑ If you attempt to create a file in a directory for which you do not have write access
❑ If your computer encounters a disk error (or network error, if you are accessing a file on a net-
work disk)
If you want your program to react gracefully when errors occur, you must handle these exceptions.
What to do when you receive an exception depends on what your program does. In some cases, you
may want to try a different file, perhaps after printing a warning message. In other cases, you may have
to ask the user what to do next or simply exit if recovery is not possible. Make sure that you provide the
user with a clear description of what went wrong.
The following code fragment shows how you might handle the case in which an input file is not avail-
able, if your program is able to continue successfully without the contents of the file:
try:
input_file = file(path)
except IOError, error:
print “problem while reading ‘%s’: %s” % (path, error)
input_text = “”
else:
input_text = input_file.read()
Paths and Directories
The file systems on Windows, Linux, Unix, and Mac OS/X have a lot in common but differ in some of
their rules, conventions, and capabilities. For example, Windows uses a backslash to separate directory
names in a path, whereas Linux and Unix (and Mac OS/X is a type of Unix) use a forward slash. In addi-
tion, Windows uses drive letters, whereas the others don’t. These differences can be a major irritation if
you are writing a program that will run on different platforms. Python makes your life easier by hiding
some of the annoying details of path and directory manipulation in the os module. Using os will not
solve all of your portability problems, however; some functions in os are not available on all platforms.
This section describes only those functions that are.
Even if you intend to use your programs only on a single platform and anticipate being able to avoid
most of these issues, if your program is useful you never know if someone will try to run it on another
platform someday. So it’s better to tap the os module, because it provides many useful services. Don’t
forget to import os first so you can use it.
113

16.5 Page 155

▲back to top
Chapter 8
Exceptions in os
The functions in the os module raise OSError exceptions on failure. If you want your
program to behave nicely when things go wrong, you must handle this exception. As
with IOError, the string representation of the exception will provide a description of
the problem.
Paths
The os module contains another module, os.path, which provides functions for manipulating paths.
Because paths are strings, you could use ordinary string manipulation to assemble and disassemble file
paths. Your code would not be as easily portable, however, and would probably not handle special cases
that os.path knows about. Use os.path to manipulate paths, and your programs will be better for it.
To assemble directory names into a path, use os.path.join. Python uses the path separator appropri-
ate for your operating system. Don’t forget to import the os.path module before you use it. For exam-
ple, on Windows, enter the following:
>>> import os.path
>>> os.path.join(“snakes”, “Python”)
‘snakes\\\\Python’
On Linux, however, using the same parameters to os.path.join gives you the following, different,
result:
>>> import os.path
>>> os.path.join(“snakes”, “Python”)
‘snakes/Python’
You can specify more than two components as well.
The inverse function is os.path.split, which splits off the last component of a path. It returns a tuple
of two items: the path of the parent directory and the last path component. Here’s an example:
>>> os.path.split(“C:\\\\Program Files\\\\Python24\\\\Lib”)
(‘C:\\\\Program Files\\\\Python24’, ‘Lib’)
On Unix or Linux, it would look like this:
>>> os.path.split(“/usr/bin/python”)
(‘/usr/bin’, ‘python’)
Automatic unpacking of sequences comes in handy here. What happens is that when os.path.split
returns a tuple, the tuple can be broken up into the elements on the left-hand side of the equals sign:
>>> parent_path, name = os.path.split(“C:\\\\Program Files\\\\Python24\\\\Lib”)
>>> print parent_path
C:\\Program Files\\Python24
>>> print name
Lib
114

16.6 Page 156

▲back to top
Files and Directories
Although os.path.split only splits off the last path component, sometimes you might want to split a
path completely into directory names. Writing a function to do this is not difficult; what you want to do
is call os.path.split on the path, and then call os.path.split on the parent directory path, and so
forth, until you get all the way to the root directory. An elegant way to do this is with a recursive func-
tion, which is a function that calls itself. It might look like this:
def split_fully(path):
parent_path, name = os.path.split(path)
if name == “”:
return (parent_path, )
else:
return split_fully(parent_path) + (name, )
The key line is the last line, where the function calls itself to split the parent path into components. The
last component of the path, name, is then attached to the end of the fully split parent path. The lines in
the middle of split_fully prevent the function from calling itself infinitely. When os.path.split
can’t split a path any further, it returns an empty string for the second component; split_fully notices
this and returns the parent path without calling itself again.
A function can call itself safely, as Python keeps track of the arguments and local variables in each run-
ning instance of the function, even if one is called from another. In this case, when split_fully calls
itself, the outer (first) instance doesn’t lose its value of name even though the inner (second) instance
assigns a different value to it, because each has its own copy of the variable name. When the inner
instance returns, the outer instance continues with the same variable values it had when it made the
recursive call.
When you write a recursive function, make sure that it never calls itself infinitely, which would be bad
because it would never return. (Actually, Python would run out of space in which to keep track of all
the calls, and would raise an exception.) The function split_fully won’t call itself infinitely, because
eventually path is short enough that name is an empty string, and the function returns without calling
itself again.
Notice in this function the two uses of single-element tuples, which must include a comma in the paren-
theses. Without the comma, Python would interpret the parentheses as ordinary grouping parentheses,
as in a mathematical expression: (name, ) is a tuple with one element; (name) is the same as name.
Let’s see the function in action:
>>> split_fully(“C:\\\\Program Files\\\\Python24\\\\Lib”)
(‘C:\\\\’, ‘Program Files’, ‘Python24’, ‘Lib’)
After you have the name of a file, you can split off its extension with os.path.splitext:
>>> os.path.splitext(“image.jpg”)
(‘image’, ‘.jpg’)
The call to splitext returns a two-element tuple, so you can extract just the extension as shown here:
>>> parts = os.path.splitext(path)
>>> extension = parts[1]
115

16.7 Page 157

▲back to top
Chapter 8
You don’t actually need the variable parts at all. You can extract the second component, the
extension, directly from the return value of splitext:
>>> extension = os.path.splitext(path)[1]
Also handy is os.path.normpath, which normalizes or “cleans up” a path:
>>> print os.path.normpath(r”C:\\\\Program Files\\Perl\\..\\Python24”)
C:\\Program Files\\Python24
Notice how the “..” was eliminated by backing up one directory component, and the double separator
was fixed. Similar to this is os.path.abspath, which converts a relative path (a path relative to the cur-
rent directory) to an absolute path (a path starting at the root of the drive or file system):
>>> print os.path.abspath(“other_stuff”)
C:\\Program Files\\Python24\\other_stuff
Your output will depend on your current directory when you call abspath. As you may have noticed,
this works even though you don’t have an actual file or directory named other_stuff in your Python
directory. None of the path manipulation functions in os.path check whether the path you are manipu-
lating actually exists.
If you want to know whether a path actually does exist, use os.path.exists. It simply returns True
or False:
>>> os.path.exists(“C:\\\\Windows”)
True
>>> os.path.exists(“C:\\\\Windows\\\\reptiles”)
False
Of course, if you’re not using Windows, or your Windows is installed in another directory (like
C:\\WinNT), both of these will return False!
Directory Contents
Now you know how to construct arbitrary paths and take them apart. But how can you find out what’s
actually on your disk? The os.listdir module tells you, by returning a list of the names entries in a
directory — the files, subdirectories, and so on that it contains.
Try It Out Getting the Contents of a Directory
The following code gets a list of entries in a directory. In Windows, you can list the contents of your
Python installation directory:
>>> os.listdir(“C:\\\\Program Files\\\\Python24”)
[‘DLLs’, ‘Doc’, ‘include’, ‘Lib’, ‘libs’, ‘LICENSE.txt’, ‘NEWS.txt’, ‘py.ico’,
‘pyc.ico’, ‘python.exe’, ‘pythonw.exe’, ‘pywin32-wininst.log’, ‘README.txt’,
‘Removepywin32.exe’, ‘Scripts’, ‘tcl’, ‘Tools’, ‘w9xpopen.exe’]
116

16.8 Page 158

▲back to top
Files and Directories
In other operating systems, or if you installed Python in a different directory, substitute some other path.
You can use “.” to list your current directory. Of course, you will get back a different list of names if you
list a different directory.
In any case, you should note a few important things here. First, the results are names of directory entries,
not full paths. If you need the full path to an entry, you must construct it yourself, with os.path.join.
Second, names of files and directories are mixed together, and there is no way to distinguish the two
from the result of os.listdir. Finally, notice that the results do not include ‘.’ and ‘..’, the two spe-
cial directory names that represent the same directory and its parent.
Let’s write a function that lists the contents of a directory but prints full paths instead of just file and
directory names, and prints only one entry per line:
def print_dir(dir_path):
for name in os.listdir(dir_path):
print os.path.join(dir_path, name)
This function loops over the list returned by os.listdir and calls os.path.join on each entry to con-
struct the full path before printing it. Try it like this:
>>> print_dir(“C:\\\\Program Files\\\\Python24”)
C:\\Program Files\\Python24\\DLLs
C:\\Program Files\\Python24\\Doc
C:\\Program Files\\Python24\\include
...
There is no guarantee that the list of entries returned by os.listdir will be sorted in any particular
way: The order can be anything. You may prefer to have the entries in some specific order to suit your
application. Because it’s just a list of strings, you can sort it yourself using the sorted function (which is
new in Python version 2.4). By default, this produces a case-sensitive alphabetical sort:
>>> sorted(os.listdir(“C:\\\\Program Files\\\\Python24”))
[‘DLLs’, ‘Doc’, ‘LICENSE.txt’, ‘Lib’, ‘NEWS.txt’, ‘README.txt’,
‘Removepywin32.exe’, ‘Scripts’, ‘Tools’, ‘include’, ‘libs’, ‘py.ico’, ‘pyc.ico’,
‘python.exe’, ‘pythonw.exe’, ‘pywin32-wininst.log’, ‘tcl’, ‘w9xpopen.exe’]
Let’s try something more complicated: Suppose that you want to list directory contents, but sorted by
file extension. For this, you need a comparison function like cmp that compares only the extensions of
two filenames. Remember that os.path.splitext splits a filename into the name and extension.
The comparison function looks like this:
def cmp_extension(path0, path1):
return cmp(os.path.splitext(path0)[1], os.path.splitext(path1)[1])
Using this function, you can augment the directory listing function to sort by extension:
def print_dir_by_ext(dir_path):
for name in sorted(os.listdir(dir_path), cmp_extension):
print os.path.join(dir_path, name)
117

16.9 Page 159

▲back to top
Chapter 8
Try It Out Listing the Contents of Your Desktop or Home Directory
Use print_dir_by_ext to list the contents of your desktop or home directory. On Windows, your desk-
top is a folder, whose path is typically C:\\\\Documents and Settings\\\\username\\\\Desktop, where
username is your account name. On GNU/Linux or Unix, your home directory’s path is typically
/home/username. Is the output what you expected?
Obtaining Information about Files
You can easily determine whether a path refers to a file or to a directory. If it’s a file, os.path.isfile
will return True; if it’s a directory, os.path.isdir will return True. Both return False if the path does
not exist at all:
>>> os.path.isfile(“C:\\\\Windows”)
False
>>> os.path.isdir(“C:\\\\Windows”)
True
Recursive Directory Listings
You can combine os.path.isdir with os.listdir to do something very useful: process subdirectories
recursively. For instance, you can list the contents of a directory, its subdirectories, their subdirectories,
and so on. To do this, it’s again useful to write a recursive function. This time, when the function finds a
subdirectory, it calls itself to list the contents of that subdirectory:
def print_tree(dir_path):
for name in os.listdir(dir_path):
full_path = os.path.join(dir_path, name)
print full_path
if os.path.isdir(full_path):
print_tree(full_path)
You’ll notice the similarity to the function print_dir you wrote previously. This function, however,
constructs the full path to each entry as full_path, because it’s needed both for printing out and for
consideration as a subdirectory. The last two lines check whether it is a subdirectory, and if so, the func-
tion calls itself to list the subdirectory’s contents before continuing. If you try this function, make sure
that you don’t call it for a large directory tree; otherwise, you’ll have to wait a while as it prints out the
full path of every single subdirectory and file in the tree.
Other functions in os.path provide information about a file. For instance, os.path.getsize returns
the size, in bytes, of a file without having to open and scan it. Use os.path.getmtime to obtain the
time when the file was last modified. The return value is the number of seconds between the start of the
year 1970 and when the file was last modified — not a format users prefer for dates! You’ll have to call
another function, time.ctime, to convert the result to an easily understood format (don’t forget to
import the time module first). Here’s an example that outputs when your Python installation directory
was last modified, which is probably the date and time you installed Python on your computer:
>>> import time
>>> mod_time = os.path.getmtime(“C:\\\\Program Files\\\\Python24”)
>>> print time.ctime(mod_time)
Tue Dec 07 02:25:01 2004
118

16.10 Page 160

▲back to top
Files and Directories
Other Types of Directory Entries
On some platforms, a directory may contain additional types of entries, such as sym-
bolic links, sockets, and devices. The semantics of these are specific to the platform and
too complicated to cover here. Nonetheless, the os module provides some support for
examining these; consult the module documentation for details for your platform.
Now you know how to modify print_dir to print the contents of a directory, including the size and
modification time of each file. In the interest of brevity, the version that follows prints only the names of
entries, not their full paths:
def print_dir_info(dir_path):
for name in os.listdir(dir_path):
full_path = os.path.join(dir_path, name)
file_size = os.path.getsize(full_path)
mod_time = time.ctime(os.path.getmtime(full_path))
print “%-32s: %8d bytes, modified %s” % (name, file_size, mod_time)
The last statement uses Python’s built-in string formatting that you saw in Chapters 1 and 2 to produce
neatly aligned output. If there’s other file information you would like to print, browse the documenta-
tion for the os.path module to learn how to obtain it.
Renaming, Moving, Copying, and Removing Files
The shutil module contains functions for operating on files. You can use the function shutil.move to
rename a file:
>>> import shutil
>>> shutil.move(“server.log”, “server.log.backup”)
Alternately, you can use it to move a file to another directory:
>>> shutil.move(“old mail.txt”, “C:\\\\data\\\\archive\\\\”)
You might have noticed that os also contains a function for renaming or moving files, os.rename.
You should generally use shutil.move instead, because with os.rename, you may not specify a direc-
tory name as the destination and on some systems os.rename cannot move a file to another disk or
file system.
The shutil module also provides the copy function to copy a file to a new name or directory. You can
simply use the following:
>>> shutil.copy(“important.dat”, “C:\\\\backups”)
Deleting a file is easiest of all. Just call os.remove:
>>> os.remove(“junk.dat”)
119

17 Pages 161-170

▲back to top

17.1 Page 161

▲back to top
Chapter 8
File Permissions
File permissions work differently on different platforms, and explaining them is
beyond the scope of this book. However, if you need to change the permissions of a file
or directory, you can use the os.chmod function. It works in the same way as the Unix
or Linux chmod system call. See the documentation for the os module for details.
If you’re an old-school Unix hacker (or want to pass yourself off as one), you may prefer os.unlink,
which does the same thing.
Example: Rotating Files
Let’s now tackle a more difficult real-world file management task. Suppose that you need to keep old
versions of a file around. For instance, system administrators will keep old versions of system log files.
Often, older versions of a file are named with a numerical suffix — for instance, web.log.1, web.log.2,
and so on — in which a larger number indicates an older version. To make room for a new version of the
file, the old versions are rotated: The current version of web.log becomes version web.log.1,
web.log.1 becomes web.log.2, and so on.
This is clearly tedious to do by hand, but Python can make quick work of it. There are a few tricky points
to consider, however. First, the current version of the file is named differently than old versions; whereas
old versions have a numerical suffix, the current version does not. One way to get around this is to treat
the current version as version zero. A short function, make_version_path, constructs the right path for
both current and old versions.
The other subtle point is that you must make sure to rename the oldest version first. For instance, if you
rename web.log.1 to web.log.2 before renaming web.log.2, the latter will be overwritten and its
contents lost before you get to it, which isn’t what you want. Once again, a recursive function will save
you. The function can call itself to rotate the next-older version of the log file before it gets overwritten:
import os
import shutil
def make_version_path(path, version):
if version == 0:
# No suffix for version 0, the current version.
return path
else:
# Append a suffix to indicate the older version.
return path + “.” + str(version)
def rotate(path, version=0):
# Construct the name of the version we’re rotating.
old_path = make_version_path(path, version)
if not os.path.exists(old_path):
# It doesn’t exist, so complain.
raise IOError, “‘%s’ doesn’t exist” % path
# Construct the new version name for this file.
new_path = make_version_path(path, version + 1)
120

17.2 Page 162

▲back to top
Files and Directories
# Is there already a version with this name?
if os.path.exists(new_path):
# Yes. Rotate it out of the way first!
rotate(path, version + 1)
# Now we can rename the version safely.
shutil.move(old_path, new_path)
Take a few minutes to study this code and the comments. The rotate function uses a technique com-
mon in recursive functions: a second argument for handing recursive cases — in this case, the version
number of the file being rotated. The argument has a default value, zero, which indicates the current ver-
sion of the file. When you call the function (as opposed to when the function is calling itself), you don’t
specify a value for this argument. For example, you can just call rotate(“web.log”).
You may have noticed that the function checks to make sure that the file being rotated actually exists and
raises an exception if it doesn’t. But suppose you want to rotate a system log file that may or may not
exist. One way to handle this is to create an empty log file whenever it’s missing. Remember that when
you open a file that doesn’t exist for writing, Python creates the file automatically. If you don’t actually
write anything to the new file, it will be empty. Here’s a function that rotates a log file that may or may
not exist, creating it first if it doesn’t. It uses the rotate function you wrote previously.
def rotate_log_file(path):
if not os.path.exists(path):
# The file is missing, so create it.
new_file = file(path, “w”)
# Close the new file immediately, which leaves it empty.
del new_file
# Now rotate it.
rotate(path)
Creating and Removing Directories
Creating an empty directory is even easier than creating a file. Just call os.mkdir. The parent directory
must exist, however. The following will raise an exception if the parent directory C:\\photos\\zoo does
not exist:
>>> os.mkdir(“C:\\\\photos\\\\zoo\\\\snakes”)
You can create the parent directory itself using os.mkdir, but the easy way out is instead to use
os.makedirs, which creates missing parent directories. For example, the following will create C:\\
photos and C:\\photos\\zoo, if necessary:
>>> os.makedirs(“C:\\\\photos\\\\zoo\\\\snakes”)
Remove a directory with os.rmdir. This works only for empty directories; if the directory is not empty,
you’ll have to remove its contents first:
>>> os.rmdir(“C:\\\\photos\\\\zoo\\\\snakes”)
This removes only the snakes subdirectory.
121

17.3 Page 163

▲back to top
Chapter 8
There is a way to remove a directory even when it contains other files and subdirectories. The function
shutil.rmtree does this. Be careful, however; if you make a programming or typing mistake and pass
the wrong path to this function, you could delete a whole bunch of files before you even know what’s
going on! For instance, this will delete your entire photo collection — zoo, snakes, and all:
>>> shutil.rmtree(“C:\\\\photos”)
Globbing
If you have used the command prompt on Windows, or a shell command line on GNU/Linux, Unix,
or Mac OS X, you probably have encountered wildcard patterns before. These are the special characters,
such as * and ?, which you use to match many files with similar names. For example, you may have
used the pattern P* to match all files that start with P, or *.txt to match all files with the extension .txt.
Globbing is hackers’ jargon for expanding wildcards in filename patterns. Python provides a
function glob, in the module also named glob, which implements globbing of directory contents. The
glob.glob function takes a glob pattern and returns a list of matching filenames or paths, similar to
os.listdir.
For example, try the following command to list entries in your C:\\Program Files directory that start
with M:
>>> import glob
>>> glob.glob(“C:\\\\Program Files\\\\M*”)
[‘C:\\\\Program Files\\\\Messenger’, ‘C:\\\\Program Files\\\\Microsoft Office’,
‘C:\\\\Program Files\\\\Mozilla Firefox’]
Your computer’s output will vary depending on what software you have installed. Observe that
glob.glob returns paths containing drive letters and directory names if the pattern includes them,
unlike os.listdir, which only returns the names in the specified directory.
The following table lists the wildcards you can use in glob patterns. These wildcards are not necessarily
the same as those available in the command shell of your operating system, but Python’s glob module
uses the same syntax on all platforms. Note that the syntax for glob patterns resembles but is not the
same as the syntax for regular expressions.
Wildcard
*
?
[...]
[!...]
Matches
Any zero or more characters
Any one character
Any one character listed in
the brackets
Any one character not listed
in the brackets
Example
*.m* matches names whose extensions
begin with m.
??? matches names exactly three char-
acters long.
[AEIOU]* matches names that begin
with capital vowels.
*[!s] matches names that don’t end
with an s.
122

17.4 Page 164

▲back to top
Files and Directories
Globbing and Case-sensitivity
On Windows, the pattern M* matches filenames that begin with both M and m, as file-
names and, therefore, filename globbing, are case-insensitive. On most other operating
systems, globbing is case-sensitive.
You can also use a range of characters in square brackets. For example, [m-p] matches any one of the let-
ters m, n, o, or p, and [!0-9] matches any character other than a digit.
Globbing is a handy way of selecting a group of similar files for a file operation. For instance, deleting all
backup files with the extension .bak in the directory C:\\source\\ is as easy as these two lines:
>>> for path in glob.glob(“C:\\\\source\\\\*.bak”):
...
os.remove(path)
Globbing is considerably more powerful than os.listdir, because you can specify wildcards in direc-
tory and subdirectory names. For patterns like this, glob.glob can return paths in more than one direc-
tory. For instance, the following code returns all files with the extension .txt in subdirectories of the
current directory:
>>> glob.glob(“*\\\\*.txt”)
Pickles
Pickles are one of the crunchiest, tastiest, and most useful features of Python. A pickle is a representation
of a Python object as a string of bytes. You can save these bytes to file, for later use or for transfer to
another computer; store them in a database; or transfer them over a network. Then you can unpickle
the string of bytes to reconstitute the original Python object, even in another instance of the Python
interpreter or on a different computer.
The most common use of pickling is to write pickle files. Typically, a pickle file contains the pickled rep-
resentation of a Python object. The fact that it contains only a single object isn’t a limitation, because the
single object may be a tuple or other collection of many other objects. Use a pickle file as an easy way to
store temporary results, store results to be used as input for another Python program, write backups,
and many other purposes.
The pickle module contains the functions you need: the dump function pickles an object to a file, and
the load function unpickles a pickle file and restores the Python object.
Try It Out Creating a Pickle File
Create a pickle file from an object; in this case, a tuple containing a string, an integer, and a floating-
point number:
>>> import pickle
>>> important_data = (“hello world”, 10, 16.5)
123

17.5 Page 165

▲back to top
Chapter 8
>>> pickle_file = file(“test.pickle”, “w”)
>>> pickle.dump(important_data, pickle_file)
>>> del pickle_file
The preceding code passed to the dump function two arguments: the object to pickle — in this case,
important_data — and a file object to which to write the pickle. Remember that pickle_file isn’t
written and closed until the file object is deleted.
You can now restore the pickled data. If you like, close your Python interpreter and open a new instance,
to convince yourself that the data is actually loaded from the pickle file. You can even copy test.pickle
to another computer and try unpickling it there:
>>> import pickle
>>> pickle_file = file(“test.pickle”)
>>> important_data = pickle.load(pickle_file)
>>> print important_data
(‘hello world’, 10, 16.5)
You don’t have to write pickles to or load pickles from a file, since you can also deal with them as strings.
The function dumps returns a pickled object as a character string, and loads restores the object from a
character string. You don’t usually want to print the string out, however, as it’s not particularly readable:
>>> pickle.dumps(important_data)
“(S’hello world’\\np0\\nI10\\nF16.5\\ntp1\\n.”
Pickling Tips
Keep in mind these features and gotchas when you use pickles:
❑ Most, but not all, Python objects can be pickled. The basic Python data types can all be pickled:
None, numbers, strings, lists, tuples, and dictionaries.
❑ You can pickle a class instance, but the class itself must be available when you unpickle the
object. This isn’t a problem for instances of classes in standard Python modules, but if you
pickle an instance of a class you wrote, make sure that the module containing that class is avail-
able when you unpickle. The class itself isn’t pickled. You don’t have to import the module con-
taining the class; Python does this for you, as long as the module is available.
❑ Other types of objects may or may not be pickleable. An instance of an extension type (see
Chapter 17) generally cannot be pickled, unless specialized pickling functions are available for
that extension type. This includes some types in the standard Python modules.
❑ You can pickle compound objects, including containers such as lists and tuples. The contents of
containers are included in the pickle. Similarly, if you pickle an object that has another object as
an attribute, both objects are included in the pickle.
❑ Pickles are portable between operating systems and architectures. For example, you can create a
pickle on a Windows or GNU/Linux PC and unpickle it on a Mac, or even a Sun workstation.
This enables you to move pickle files and transfer pickles over a network between different
types of computers.
❑ Pickles are Python-specific. There’s no easy way to access the contents of a pickle with programs
written in other languages.
124

17.6 Page 166

▲back to top
Files and Directories
Efficient Pickling
If your program performs a lot of pickling and unpickling, and/or uses very large pickles of large or
complicated data structures, you might want to consider the following two techniques. The first uses the
cPickle module for faster pickling and unpickling. The second uses an alternate binary pickle protocol
to write more compact pickles.
In addition to the pickle module, Python provides a second implementation of pickling in the cPickle
module. Both modules contain the same functions for pickling and unpickling, and their pickles are
compatible. The difference is that pickle is itself written in Python, whereas cPickle is an extension
module written in C, and therefore runs much faster. You can use them interchangeably:
>>> import cPickle
>>> print cPickle.load(file(“test.pickle”))
(‘hello world’, 10, 16.5)
In addition, both pickle and cPickle support an additional format for pickles. The default format uses
ordinary (albeit unintelligible) text to represent objects; the alternate binary pickle protocol uses a more
compact (and even less intelligible) binary representation. You can specify the protocol version with an
extra argument to dump or dumps:
>>> pickle.dump(important_data, file(“test.pickle”, “w”), 2)
Here, you’ve specified protocol version 2, which as of Python 2.4 is the newest binary pickle protocol
available. There’s no need to specify the protocol version when unpickling; Python figures it out
automatically.
Summary
In this chapter, you learned how to write data to and read data from files on your disk. Using a file
object, you can now write strings to a file, and read back the contents of a file, line-by-line or all at once.
You can use these techniques to read input into your program, to generate output files, or to store inter-
mediate results.
You also learned about paths, which specify the location of a file on your disk, and how to manipulate
them. Using os.listdir or glob, you can find out what’s on your disk.
Finally, you learned about pickles. Pickles enable you to store many kinds of Python objects, not just
strings, in a file and restore them later.
Exercises
1. Create another version of the (nonrecursive) print_dir function that lists all subdirectory
names first, followed by names of files in the directory. Names of subdirectories should be
alphabetized, as should filenames. (For extra credit, write your function in such a way that it
calls os.listdir only one time. Python can manipulate strings faster than it can execute
os.listdir.)
125

17.7 Page 167

▲back to top
Chapter 8
2. Modify the rotate function to keep only a fixed number of old versions of the file. The number
of versions should be specified in an additional parameter. Excess old versions above this num-
ber should be deleted.
3. Write a program to maintain a simple diary (like a blog, but not on the web). Put the diary
entries into a list and store the list in a pickle file. Every time the program is run, it should read
in the diary data from the pickle file, ask the user for a new entry, append this entry to the diary
data, and write back the diary data to the pickle file. Use the built-in raw_entry function to
prompt for the new entry.
a. For extra credit, call time.ctime(time.time()) to obtain the current date and
time and store this with the diary entry.
b. Finally, write a program to print out the diary. Print it in reverse order — that is,
the most recent entry first.
126

17.8 Page 168

▲back to top
9
Other Features of
the Language
In this chapter, you’ll be introduced to some other aspects of Python that are less frequently used,
as well as modules that are very commonly used. Each section describes at least one way that the
feature is typically used and then offers example code.
Lambda and Filter: Short
Anonymous Functions
Sometimes you need a very simple function invocation — something that is not generally useful or
that is so specific that its use is going to need to be completely different if it is invoked in another
location in your code. For these occasions, there is a special operation: lamba. Lambda is not a
function itself but a special word that tells Python to create a function and use it in place, rather
than reference it from a name.
To demonstrate lambda being used, filter will be used, which is a function that can use lambda
effectively. It enables you to take a list and remove elements based on criteria you define within
a function you write. Normal functions can be used, but in simple cases, such as where you want
only odd numbers (or odd-numbered elements, or strings beginning with something, and so on),
a fully defined function could be overkill.
# use lambda with filter
filter_me = [1, 2, 3, 4, 6,7 ,8, 11, 12, 14, 15, 19, 22]
# This will only return true for even numbers (because x%2 is 0, or False,
# for odd numbers)
result = filter(lambda x: x%2 == 0, filter_me)
print result

17.9 Page 169

▲back to top
Chapter 9
The functions that lambda creates are called anonymous functions because of their lack of a name.
However, you can use the result of the lambda statement to bind the name to a function yourself. That
name will be available only in the scope in which the name was created, like any other name:
# use lambda with filter, but bind it to a name
filter_me = [1, 2, 3, 4, 6,7 ,8, 11, 12, 14, 15, 19, 22]
# This will only return true for even numbers (because x%2 is 0, or False,
# for odd numbers)
func = lambda x: x%2 == 0
result = filter(func, filter_me)
print result
Lambda can only be a simple function, and it can’t contain statements, such as creating a name for a
variable. Inside a lambda, you can only perform a limited set of operations, such as testing for equality,
multiplying numbers, or using other already existing functions in a specific manner. You can’t do things
like use if ... : elsif ... : else: constructs or even create new names for variables! You can only
use the parameters passed into the lambda function. You can, however, do slightly more than perform
simple declarative statements by using the and and or operations. However, you should still keep in
mind that lambda is for very limited uses.
The main use for lambda is with the built-in functions map, reduce, and filter. Used with lambda,
these functions provide compact ways to perform some great operations while avoiding the need for
loops. You’ve already seen filter in action, which could be a difficult loop to write.
Reduce
Reduce is a way to take the elements of a list or a tuple and run a function on the first two elements.
Then it uses the result of that operation and runs the result and the next element in the list through the
same operation. It is a lot like a distillation process that gradually gets to a processed summary of the
contents of a sequence. For simple operations like adding or multiplying all of the elements, using an
anonymous function is quite convenient:
# Use reduce with a lambda function to make small numbers into a very big number
reduce_me = [ 2, 4, 4, 2, 6 ]
result = reduce(lambda first, second: first**second, reduce_me)
print “The result of reduce is: %d” % result
This produces a very large number when you run it:
The result of reduce is: 6277101735386680763835789423207666416102355444464034512896
To see how this works, let’s build up to the same answer with smaller lists.
Try It Out Working with Reduce
In the shell, set up a reduce invocation that just contains the first two elements from reduce_me, and
run the code again:
>>> reduce(lambda first, second: first**second, [2, 4])
16
128

17.10 Page 170

▲back to top
Other Features of the Language
Now let’s use more elements from reduce_me and build up to the answer:
>>> reduce(lambda first, second: first**second, [2, 4, 4])
65536
>>> reduce(lambda first, second: first**second, [2, 4, 4, 2])
4294967296L
>>> reduce(lambda first, second: first**second, [2, 4, 4, 2, 6])
6277101735386680763835789423207666416102355444464034512896L
How It Works
You can see that when a list containing only 2, 4 is passed to result, the result is 16, which is the value
of two to the fourth power, or 2**4, or the first list element to the power of the second.
When the list is expanded to 2, 4, 4, the value leaps to 65536. This is the result of 16, the result of 2**4,
being raised to the fourth power, or 16**4.
Reduce continues to do this to every member of the list.
Map: Short-Circuiting Loops
One common place to use anonymous functions is when the map function is called. Map is a special
function for cases when you need to do a specific action on every element of a list. It enables you to
accomplish this without having to write the loop.
Try It Out Use Map
Try this basic test:
# Now map gets to be run in the simple case
map_me = [ ‘a’, ‘b’, ‘c’, ‘d’, ‘e’, ‘f’, ‘g’ ]
result = map(lambda x: “The letter is %s” % x, map_me)
print result
How It Works
Just like being in a loop, every element in the list will be visited, and a list with the new values is
returned. This is how it will look:
>>> print result
[‘The letter is a’, ‘The letter is b’, ‘The letter is c’, ‘The letter is d’, ‘The
letter is e’, ‘The letter is f’, ‘The letter is g’]
There are some special things worth knowing about map. If you pass in a list of lists (or tuples — any
kind of sequence can be given to map), then your function needs to expect that list. Each sequence in the
main list should have the same number of elements:
# use map with a list of lists, to re-order the output.
map_me_again = [ [1, 2, 3], [4, 5, 6], [7, 8, 9]]
result = map(lambda list: [ list[1], list[0], list[2]], map_me_again)
print result
129

18 Pages 171-180

▲back to top

18.1 Page 171

▲back to top
Chapter 9
This results in a list of lists, where everything has been shuffled around:
>>> print result
[[2, 1, 3], [5, 4, 6], [8, 7, 9]]
You can see that map always returns a list. Map is not usable when you need to print output or do any-
thing besides get a resulting list.
Map can be given the name of a non-anonymous function if you like, and it operates in the same way.
Map has one other interesting feature. If you have multiple lists passed into your function, the first ele-
ment of each list will be supplied as the parameters to your function, and then each second element, and
then each of the third elements, and so on. If any elements aren’t present in the lists that are provided,
the special value None will be inserted where there is a missing element:
result = map(lambda x,y,z: “%s” % str(x) + str(y) + str(z), [1, 2, 3], [4, 5, 6],
[7])
print result
The output from this is especially interesting — you should evaluate this yourself to determine why the
output looks like this:
[‘147’, ‘25None’, ‘36None’]
When it is given many lists, map becomes a way to go through some number of horizontal lists of data
by acting on all of the elements vertically.
Decisions within Lists —
List Comprehension
The oddly named list comprehension feature entered the language in Python 2.0. It enables you to write
miniature loops and decisions within the list dereferencing operators (the square brackets) to define
parameters that will be used to restrict the range of elements being accessed.
For instance, to create a list that just prints the positive numbers in a list, you can use list comprehension:
# First, just print even numbers
everything = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12 ]
print [ x for x in everything if x%2 == 0 ]
This can be a nice and compact way of providing a portion of a list to a loop — however, with only the per-
tinent parts of the list, based what you want in your program at the moment, being presented to your loop.
List comprehension provides you with the same functionality as filter or map combined with lambda,
but it is a form that gives you more decision-making power because it can include loops and condition-
als, whereas lambda only enables you to perform one simple expression.
In most cases, list comprehension will also run faster than the alternative.
130

18.2 Page 172

▲back to top
Other Features of the Language
Generating Lists for Loops
Python has a special feature that enables you to create lists: the range function:
list = range (10, 20)
print list
This code produces an obvious-looking result:
>>> print list
[10, 11, 12, 13, 14, 15, 16, 17, 18, 19]
By itself, this doesn’t seem profound, but it is essential for situations when you need to use a for loop
that will continue for a specific number of iterations, and that isn’t based on an existing list; and this
number may not be determined at the time when the program was written, but it becomes known only
when the program is already running.
If range is only given a single number, it will count from zero to that number. The number can be posi-
tive or negative:
for number in range(10):
print “Number is now %d” % number
This produces the obvious output, which is what you want:
Number is now 0
Number is now 1
Number is now 2
Number is now 3
Number is now 4
Number is now 5
Number is now 6
Number is now 7
Number is now 8
Number is now 9
In addition, if you only want, for example, every other number, or every third number, you can use an
even more optional third parameter, called the step, that describes what the interval will be between
each number that range creates:
for number in range(5, 55, 4):
print “Number from 5 to 55, by fours: %d” % number
This results in the selective list of numbers that you specified:
Number from 5 to 55, by fours: 5
Number from 5 to 55, by fours: 9
Number from 5 to 55, by fours: 13
Number from 5 to 55, by fours: 17
Number from 5 to 55, by fours: 21
Number from 5 to 55, by fours: 25
131

18.3 Page 173

▲back to top
Chapter 9
Number from 5 to 55, by fours: 29
Number from 5 to 55, by fours: 33
Number from 5 to 55, by fours: 37
Number from 5 to 55, by fours: 41
Number from 5 to 55, by fours: 45
Number from 5 to 55, by fours: 49
Number from 5 to 55, by fours: 53
In some situations, a program could be handling huge numbers of elements — perhaps hundreds of
thousands, or millions. In this case, range does one thing wrong: It will create an array with every ele-
ment that you’ve asked for — for example, from zero to the number of all the possible systems on the
Internet. When this many things need to be examined, each element uses a bit of computer memory,
which can eventually take up all of the memory on a system. To avoid any problems with this sort of
really large list, a special built-in class called xrange creates fewer elements in memory, so it is perfect
for really large sets of numbers. It behaves the same as range does, except instead of returning a list it
returns an xrange object.
The following code will produce the same output as the range function, so the output is omitted:
# xrange provides a special case useful for large sets. This is unnecessary.
for r in xrange(0, 10):
print r
Try It Out Examining an xrange Object
Interestingly, note that xrange returns an object that behaves like a list. Note that this object has no pub-
lic interfaces — just private methods that look like a subset of what most lists and tuples have:
>>> xr = xrange(0,10)
>>> dir(xr)
[‘__class__’, ‘__delattr__’, ‘__doc__’, ‘__getattribute__’, ‘__getitem__’,
‘__hash__’, ‘__init__’, ‘__iter__’, ‘__len__’, ‘__new__’, ‘__reduce__’,
‘__reduce_ex__’, ‘__repr__’, ‘__reversed__’, ‘__setattr__’, ‘__str__’]
Tying to call it directly doesn’t result in a list; it results in a representation of how it was called:
>>> xr
xrange(10)
You can, however, still access it by using the same dereferencing operation (the square brackets) that you
can with lists, sequences, and dictionaries.
>>> xr[0]
0
>>> xr[1]
1
How It Works
Xrange produces an object that doesn’t have any public methods. The only methods it has are built-in
methods that enable it to act as a very simple sequence. Internally, when you use the square brackets to
132

18.4 Page 174

▲back to top
Other Features of the Language
access a list, tuple, or a dictionary, you are telling python to invoke the __getitem__ method of that list,
tuple, or dictionary. An xrange object has this private method, so it can act as a sequence and be derefer-
enced this way.
When you call an xrange object, it doesn’t produce a list — instead, it tells you how it was created so
you know what the parameters were, in case you wanted to know about the numbers it is generating.
The point is that even though it behaves like a sequence, it is different; and that’s kind of cool.
Special String Substitution Using
Dictionaries
One syntax you haven’t been shown yet is a special syntax for using dictionaries to populate string sub-
stitutions. This can come up when you want a configurable way to print out strings — such as a format-
ted report or something similar.
Try It Out String Formatting with Dictionaries
When you are doing this, you want to take individual named elements from a known set of elements,
such as what you have in a dictionary, and print them out in the order that you have specified, which
can be defined outside of the program itself:
person = {“name”: “James”, “camera”: “nikon”, “handedness”: “lefty”,
“baseball_team”: “angels”, “instrument”: “guitar”}
print “%(name)s, %(camera)s, %(baseball_team)s” % person
The output of this code looks like this:
>>> print “%(name)s, %(camera)s, %(baseball_team)s” % person
James, nikon, angels
How It Works
Note that the information in the parentheses is the name of the key whose value will be substituted from
the dictionary into the string. However, to use this properly, you still need to specify the type of the data
being inserted after the closing parenthesis so that the string substitution knows what to do. Here, all the
types were strings, but you could use the i for int, j for imaginary, l for long, and all the other format
specifiers you’ve learned. To see different formats being used with this new format, try the following
example. Notice that person should appear on the same line as the print statement — it’s not on the
next line; it’s just the end of a long line:
person[“height”] = 1.6
person[“weight”] = 80
print “%(name)s, %(camera)s, %(baseball_team)s, %(height)2.2f, %(weight)2.2f” %
person
133

18.5 Page 175

▲back to top
Chapter 9
This gives you the following terse output:
>>> print “%(name)s, %(camera)s, %(baseball_team)s, %(height)2.2f, %(weight)2.2f” %
person
James, nikon, angels, 1.60, 80.00
These examples work with almost the same syntax that you learned in the first three chapters.
Python 2.4 has added another form of string substitution within the String module, with a new syntax
for a substitution grammar. This form has been created to enable you to give users — for example, of a
program you’ve written — a format that may make more sense to them at first glance:
import string
person = {“name”: “James”, “camera”: “nikon”, “handedness”: “lefty”,
“baseball_team”: “angels”, “instrument”: “guitar”}
person[“height”] = 1.6
person[“weight”] = 80
t = string.Template(“$name is $height m high and $weight kilos”)
print t.substitute(person)
This produces output that’s no better or worse than the first way, except that you can’t control the for-
mat information anymore:
print t.substitute(person)
James is 1.6 m high and 80 kilos
Think about using this feature when you are asking users to describe what information they want from
a set of data. This can be used as an easily supported way for someone else to specify the data they want
without saddling you with the need to rewrite your program. You just need to ask them to specify the
template, and you can supply the string they’ve given you to the string.Template class to create a
template object that will perform the desired substitution.
Featured Modules
Starting in Chapter 7, you’ve seen modules used to add functionality to Python. In Chapter 8, you
learned how interaction with the operating system and its files is achieved through modules that pro-
vide interfaces to how the system works with the os module.
In this section, you’ll see examples of some other common modules that will help you to start building
your own programs.
Getopt — Getting Options from the Command Line
On Unix systems, the most common way to specify the behavior of a program when it runs is to add
parameters to the command line of a program. Even when a program is not run from the command
line but is instead run using fork and exec (more on this later in this chapter), a command line
is constructed when it is invoked. This makes it a universal way of controlling the behavior of your
programs.
134

18.6 Page 176

▲back to top
Other Features of the Language
You may have seen, for instance, that many programs can be run so that they provide you with some
basic information about how they should be run. Python enables you to do this with -h:
$ python –h
usage: python2.4 [option] ... [-c cmd | -m mod | file | -] [arg] ...
Options and arguments (and corresponding environment variables):
-c cmd : program passed in as string (terminates option list)
-d
: debug output from parser (also PYTHONDEBUG=x)
-E
: ignore environment variables (such as PYTHONPATH)
[ etc. ]
In the past, different conventions were available on different Unix platforms to specify these options, but
this has largely resulted in two forms of options being used by most projects: the short form, such as the
help-message producing option to Python, and a long form, such as --help for help.
To accept these sorts of options makes sense. Ideally, you’d like to offer a short and a long form of com-
mands that are common, and allow each one to optionally take a specification. So if you wanted to write
a program that had a configuration file that the user could specify, you may want one option like -c
short for experienced users, but provide a longer option too, like --config-file. In either case, you’d
want them to be the same function in your program to save you time, but you’d like to give users the
freedom to use these options however they want to use them.
The getopt module provides two functions to make this standard convention easy to use:
getopt.getopt and getopt.gnu_getopt. They are both basically the same. The basic getopt only
works until the first non-option is encountered — nothing else is checked.
For getopt to be useful, you have to know what options you want to be useful. Normally, it’s consid-
ered the least you can do for your users to write programs that provide them with information about
how to run the program, such as how Python prints information with the -h option.
In addition, it’s often very useful to have a configuration file. Using these ideas as a starting point, you
could start your new programs so that -h and --help both produce a minimal message about how your
program is used, and using -c or --config-file=file would enable you to specify a configuration
file that is different from the default configuration:
import sys
import getopt
# Remember, the first thing in the sys.argv list is the name of the command
# You don’t need that.
cmdline_params = sys.argv[1:]
opts, args = getopt.getopt(cmdline_params, ‘hc:’, [‘help’, ‘config=’])
print opts, args
for option, parameter in opts:
if option == ‘-h’ or option == ‘--help’:
print “This program can be run with either -h or --help for this message,”
print “or with -c or --config=<file> to specify a different configuration
file”
print
if option in (‘-c’, ‘--config’): # this means the same as the above
print “Using configuration file %s” % parameter
135

18.7 Page 177

▲back to top
Chapter 9
When long options are used and require a parameter (like --config in the preceding example), the
equal sign must connect the option and the value of the parameter. However, when short options are
used, one or more space or tab characters can separate the option from its corresponding value. This dis-
tinction is to duplicate the behavior of the options on older Unix machines that persist to the modern
day. They persist because so many people expect that behavior. What can you do?
The preceding code snippet, if run in a program with the parameters -c test -h --config=
secondtest, produces the following output:
[(‘-c’, ‘test’), (‘-h’, ‘’), (‘--config’, ‘secondtest’)] []
Using configuration file test
This program can be run with either -h or --help for this message,
or with -c or --config=<file> to specify a different configuration file
Using configuration file secondtest
Note how the second instance of the configuration file is accepted silently; and when it is reached, the
same code that sets the config file is revisited so that the second instance is used.
The second list, the args data, is an empty list because all of the options provided to the program on the
command line were valid options, or valid parameters to options. If you inserted other strings in the
middle of your options, the normal getopt would behave differently. If the parameters used were
instead -c test useless_information_here -h --config=secondtest, the output would say a lot
less, and the args array would have a lot more in it.
[(‘-c’, ‘test’)] [‘useless_information_here’, ‘-h’, ‘--config=secondtest’]
Using configuration file test
The gnu_getopt lets you mix and match on the command line so that nonoptions can appear anywhere
in the midst of the options, with more options parsed afterward instead of stopping there:
opts, args = getopt.gnu_getopt(cmdline_params, ‘hc:’, [‘help’, ‘config=’])
print opts, args
for option, parameter in opts:
if option == ‘-h’ or option == ‘--help’:
print “This program can be run with either -h or --help for this message,”
print “or with -c or --config=<file> to specify a different configuration
file”
print
if option in (‘-c’, ‘--config’): # this means the same as the above
print “Using configuration file %s” % parameter
The important point to note is that if you use something that doesn’t meet the criteria for an option (by
beginning with a – or a +, or following an option that takes a parameter), the two behave differently.
Using the options -c test useless_information_here -h --config=secondtest, the gnu_getopt
function provides the following output, with the odd duck being the only part of the command line left
in the args array:
[(‘-c’, ‘test’), (‘-h’, ‘’), (‘--config’, ‘secondtest’)]
[‘useless_information_here’]
136

18.8 Page 178

▲back to top
Other Features of the Language
Using configuration file test
This program can be run with either -h or --help for this message,
or with -c or --config=<file> to specify a different configuration file
Using configuration file secondtest
Using More Than One Process
In Unix and Unix-like operating systems, the main way of performing certain kinds of subtasks is to
create a new process running a new program. On Unix systems, this is done using a system call that is
available in Python by using os.fork. This actually tells the computer to copy everything about the cur-
rently running program into a newly created program that is separate, but almost entirely identical. The
only difference is that the return value for os.fork is zero in the newly created process (the child), and
is the process ID (PID) of the newly created process in the original process (the parent). This can be diffi-
cult to understand, and the only way to really get it is to use it a few times and to read some other mate-
rial on fork and exec that’s available on-line. Or talk to your nearest Unix guru.
Based on the one critical difference, a parent and child can perform different functions. The parent can
wait for an event while the child processes, or vice versa. The code to do this is simple, and common, but
it works only on Unix and Unix-like systems:
import os
pid = os.fork()
if pid == 0: # This is the child
print “this is the child”
else:
print “the child is pid %d” % pid
One of the most common things to do after an os.fork call is to call os.execl immediately afterward
to run another program. os.execl is an instruction to replace the running program with a new pro-
gram, so the calling program goes away, and a new program appears in its place (in case you didn’t
already know this, Unix systems use the fork and exec method to run all programs):
import os
pid = os.fork()
# fork and exec together
print “second test”
if pid == 0: # This is the child
print “this is the child”
print “I’m going to exec another program now”
os.execl(‘/bin/cat’, ‘cat’, ‘/etc/motd’)
else:
print “the child is pid %d” % pid
os.wait()
The os.wait function instructs Python that you want the parent to not do anything until the child pro-
cess returns. It is very useful to know how this works because it works well only under Unix and Unix-
like platforms such as Linux. Windows also has a mechanism for starting up new processes.
To make the common task of starting a new program easier, Python offers a single family of functions
that combines os.fork and os.exec on Unix-like systems, and enables you to do something similar
137

18.9 Page 179

▲back to top
Chapter 9
on Windows platforms. When you want to just start up a new program, you can use the os.spawn fam-
ily of functions. They are a family because they are named similarly, but each one has slightly different
behaviors.
On Unix-like systems, the os.spawn family contains spawnl, spawnle, spawnlp, spawnlpe, spawnv,
spawnve, spawnvp, and spawnvpe. On Windows systems, the spawn family only contains spawnl,
spawnle, spawnv, and spawnve.
In each case, the letters after the word spawn mean something specific. The v means that a list (a vector
is what the v actually stands for) will be passed in as the parameters. This allows a command to be run
with very different commands from one instance to the next without needing to alter the program at all.
The l variations just requires a simple list of parameters.
The e occurrences require that a dictionary containing names and values that will be used as the environ-
ment for the newly created program will be passed in instead of using the current environment.
The p occurrence uses the value of the PATH key in the environment dictionary to find the program. The
p variants are available only on Unix-like platforms. The least of what this means is that on Windows
your programs must have a completely qualified path to be usable by the os.spawn calls, or you have to
search the path yourself:
import os, sys
if sys.platform == ‘win32’:
print “Running on a windows platform”
command = “C:\\\\winnt\\\\system32\\\\cmd.exe”
params = []
if sys.platform == ‘linux2’:
print “Running on a Linux system, identified by %s” % sys.platform
command = ‘/bin/uname’
params = [‘uname’, ‘-a’]
print “Running %s” % command
os.spawnv(os.P_WAIT, command, params)
Of course, this example will only work on a limited range of systems. You can use the contents of
sys.platform on your own computer and for something besides linux2 in case you are on another
Unix system such as Solaris, Mac OS X, AIX, or others.
When you do this, you can either wait for the process to return (that is, until it finishes and exits) or you
can tell Python that you’d prefer to allow the program to run on its own, and that you will confirm that
it completed successfully later. This is done with the os.P_ family of values. Depending on which one
you set, you will be given a different behavior when an os.spawn function returns.
If you need only the most basic invocation of a new command, sometimes the easiest way to do this is to
use the os.system function. If you are running a program and just want to wait for it to finish, you can
use this function very simply:
# Now system
if sys.platform == ‘win32’:
print “Running on a windows platform”
command = “cmd.exe”
138

18.10 Page 180

▲back to top
Other Features of the Language
if sys.platform == ‘linux2’:
print “Running Linux”
command = “uname -a”
os.system(command)
This can be much simpler because it uses the facilities that the operating system provides, and that users
expect normally, to search for the program you want to run, and it defaults to waiting for the child pro-
cess to finish.
Threads — Doing Many Things in the Same Process
Creating a new process using fork or spawn can sometimes be too much effort and not provide enough
benefit. Specifically, regarding the too much effort, when a program grows to be large, fork has to copy
everything in the program to the new program and the system must have enough resources to handle
that. Another downside for fork is that sometimes when you need your program to do many things at
the same time, some things may need to wait while others need to proceed. When this happens, you
want to have all of the different components communicating their needs to other parts of the program.
Using multiple processes, this becomes very difficult. These processes share many things because
the child was originally created using the data in the parent. However, they are separate entities —
completely separate. Because of this, it can be very tricky to make two processes work together
cooperatively.
So, to make some complex situations where subprocesses are not appropriate workable, the concept of
threads is available.
Many cooperative threads of program execution are able to exist at the same time in the same program.
Each one has potentially different objects, with different state, but they can all communicate, while also
being able to run semi-independently of one another.
This means that in many situations, using threads is much more convenient than using a separate pro-
cess. Note that the following example uses subclassing, which is covered in Chapter 10. To see how this
works, try running it with a fairly large parameter, say two million (2000000):
import math
from threading import Thread
import time
class SquareRootCalculator:
“””This class spawns a separate thread to calculate a bunch of square
roots, and checks in it once a second until it finishes.”””
def __init__(self, target):
“””Turn on the calculator thread and, while waiting for it to
finish, periodically monitor its progress.”””
self.results = []
counter = self.CalculatorThread(self, target)
print “Turning on the calculator thread...”
counter.start()
139

19 Pages 181-190

▲back to top

19.1 Page 181

▲back to top
Chapter 9
while len(self.results) < target:
print “%d square roots calculated so far.” % len(self.results)
time.sleep(1)
print “Calculated %s square root(s); the last one is sqrt(%d)=%f” % \\
(target, len(self.results), self.results[-1])
class CalculatorThread(Thread):
“””A separate thread which actually does the calculations.”””
def __init__(self, controller, target):
“””Set up this thread, including making it a daemon thread
so that the script can end without waiting for this thread to
finish.”””
Thread.__init__(self)
self.controller = controller
self.target = target
self.setDaemon(True)
def run(self):
“””Calculate square roots for all numbers between 1 and the target,
inclusive.”””
for i in range(1, self.target+1):
self.controller.results.append(math.sqrt(i))
if __name__ == ‘__main__’:
import sys
limit = None
if len(sys.argv) > 1:
limit = sys.argv[1]
try:
limit = int(limit)
except ValueError:
print “Usage: %s [number of square roots to calculate]” \\
% sys.argv[0]
SquareRootCalculator(limit)
For many situations, such as network servers (see Chapter 16) or graphical user interfaces (see Chapter
13), threads make much more sense because they require less work from you as the programmer, and
less resources from the system.
Note how separate threads can access each other’s names and data easily. This makes it very easy to
keep track of what different threads are doing, an important convenience.
Subprocesses are really available only on Unix and Unix-like platforms on which they are well supported.
A self-contained program using threads can be much more easily ported across different platforms.
Storing Passwords
You will frequently need to store passwords in an encrypted form. Most operating systems have their
own way of doing this. On Unix, the standard encryption is a protocol called DES, though newer sys-
tems also allow a type of hashing called md5, and on some sha-1 is available. Windows systems usually
keep passwords in an entirely different format in the registry.
140

19.2 Page 182

▲back to top
Other Features of the Language
This profusion of standards isn’t necessarily a bad thing — as computers get faster and older password
systems become easier to crack, systems should evolve.
Python provides two reasonably secure built-in algorithms that you can use for password generation
in your applications if you need them. These are passwords that can’t be reversed; they’re useful for
authenticating users to an application that could contain sensitive information:
import sha
import random
import base64
def _gen_salt():
salt = [chr(random.randint(0,255)) for i in range(4) ]
return ‘’.join(salt)
def make_pass(cleartext):
salt = _gen_salt()
text = salt + cleartext
hash = sha.new(text).digest()
data = salt + hash
return base64.encodestring(data)
def check_pass(cipher, cleartext):
cipher = base64.decodestring(cipher)
salt, hash = cipher[:4], cipher[4:]
hash2 = sha.new(salt + cleartext).digest()
return hash2 == hash
if __name__ == ‘__main__’:
cipher = make_pass(‘TEST’)
for word in ‘spam’, ‘TEST’, ‘Test’, ‘omelette’:
passwd = check_pass(cipher, word)
print ‘%s: %d’ % (word, passwd)
The same code could be used with md5 as the core encryption mechanism, although sha is usually con-
sidered stronger.
The base64 module is used to turn what is often binary data into text data that can be easily accessed by
common tools like text editors. Although passwords are things that only computers should be dealing
with, it’s often necessary to manually view the files in which they reside.
Summary
In this chapter, you’ve been introduced to some of the many available functions and modules that
Python offers. These features build on the material you’ve already learned and most of them will be
expanded on in the remaining chapters in the book.
You learned how to use some basic features that enable what is usually called a functional style of
programming, which in Python is offered through the functions lambda, map, and reduce. Lambda
enables you to write a simple function without having to declare it elsewhere. These functions are called
141

19.3 Page 183

▲back to top
Chapter 9
anonymous because they can be written and run without ever having to be bound to a name. Map oper-
ates on lists, and when used on a simple list will run a function on each element from beginning to end.
It has some more complex behaviors, too, which occur when lists within lists, or more than one list, is
provided to map. The last function of these three, reduce, offers you the capability to run the same func-
tion, one that accepts two parameters, on all of the elements of a list, starting with the first and second,
and then using the result of the first and second, and using it with the third element in the list, and then
using that result, and so on.
List comprehension is the capability to run a limited amount of code — a simple loop, for instance —
within the square brackets that dereference a sequence, so that only those elements that meet the criteria
within the brackets will be returned. This enables you easily and quickly to access specific members of a
sequence.
The range and xrange operations enable you to generate lists, and are commonly used in for loops
because they can provide you with numeric lists starting at any number, and ending at any number.
Range simply creates a list, while xrange should be used when you are creating large lists, because it
creates an xrange object that behaves like a list, but for many elements it will use less memory and can
even go faster in these cases.
In addition to simple string substitution, you can provide a string with format specifiers that reference
the name of keys in dictionaries by using a special syntax. This form enables you to continue to use the
format specifier options, such as how many spaces you want reserved for the substitution or how many
decimal points should be used.
An alternative form for simple key-name based string formatting is provided in the string.Template
module that has been added to Python 2.4. It provides a slightly simpler format that is more appropriate
(or at least easier to explain) when you allow your users to specify templates. Generating form letters is
one example of how this could be used.
Getopt enables you to specify options on the command line that lets you offer your users options that
determine the behavior of your programs when they’re run.
You now know how to create more processes when needed, and how to create threads for use in more
complex programs that need to do many things in parallel. You will get a chance to learn more about
using threads in Chapters 13 and 16.
Finally, you learned how to create a password hash that can be used to authenticate users in your
programs.
The features and modules presented here give you an idea of the different directions in which Python
can be extended and used, and how easy it is to use these extensions. In Chapter 10, you’ll see most of
the concepts you’ve used already tied into an example working program.
Exercises
Chapter 9 is a grab-bag of different features. At this point, the best exercise is to test all of the sample
code, looking at the output produced and trying to picture how the various ideas introduced here could
be used to solve problems that you’d like to solve or would have liked to solve in the past.
142

19.4 Page 184

▲back to top
10
Building a Module
As you saw in Chapter 7, modules provide a convenient way to share Python code between appli-
cations. A module is a very simple construct. In Python, a module is merely a file of Python state-
ments. The module might define functions and classes. It can contain simple executable code that’s
not inside a function or class. And, best yet, a module might contain documentation about how to
use the code in the module.
Python comes with a library of hundreds of modules that you can call in your scripts. You can also
create your own modules to share code among your scripts. This chapter shows you how to create
a module, step by step. This includes the following:
❑ Exploring the internals of modules
❑ Creating a module that contains only functions
❑ Defining classes in a module
❑ Extending classes with subclasses
❑ Defining exceptions to report error conditions
❑ Documenting your modules
❑ Testing your modules
❑ Running modules as programs
❑ Installing modules
The first step is to examine what modules really are and how they work.
Exploring Modules
A module is just a Python source file. The module can contain variables, classes, functions, and
any other element available in your Python scripts.

19.5 Page 185

▲back to top
Chapter 10
You can get a better understanding of modules by using the dir function. Pass the name of some Python
element, such as a module, and dir will tell you all of the attributes of that element. For example, to
see the attributes of __builtins__, which contain built-in functions, classes, and variables, use the
following:
dir(_ _builtins_ _)
For example:
>>> dir(__builtins__)
[‘ArithmeticError’, ‘AssertionError’, ‘AttributeError’, ‘DeprecationWarning’,
‘EOFError’, ‘Ellipsis’, ‘EnvironmentError’, ‘Exception’, ‘False’,
‘FloatingPointError’, ‘FutureWarning’, ‘IOError’, ‘ImportError’,
‘IndentationError’, ‘IndexError’, ‘KeyError’, ‘KeyboardInterrupt’,
‘LookupError’, ‘MemoryError’, ‘NameError’, ‘None’, ‘NotImplemented’,
‘NotImplementedError’, ‘OSError’, ‘OverflowError’, ‘OverflowWarning’,
‘PendingDeprecationWarning’, ‘ReferenceError’, ‘RuntimeError’, ‘RuntimeWarning’,
‘StandardError’, ‘StopIteration’, ‘SyntaxError’, ‘SyntaxWarning’,
‘SystemError’, ‘SystemExit’, ‘TabError’, ‘True’, ‘TypeError’,
‘UnboundLocalError’, ‘UnicodeDecodeError’, ‘UnicodeEncodeError’,
‘UnicodeError’, ‘UnicodeTranslateError’, ‘UserWarning’, ‘ValueError’,
‘Warning’, ‘ZeroDivisionError’, ‘__debug__’, ‘__doc__’, ‘__import__’,
‘__name__’, ‘abs’, ‘apply’, ‘basestring’, ‘bool’, ‘buffer’, ‘callable’, ‘chr’,
‘classmethod’, ‘cmp’, ‘coerce’, ‘compile’, ‘complex’, ‘copyright’, ‘credits’,
‘delattr’, ‘dict’, ‘dir’, ‘divmod’, ‘enumerate’, ‘eval’, ‘execfile’, ‘exit’,
‘file’, ‘filter’, ‘float’, ‘getattr’, ‘globals’, ‘hasattr’, ‘hash’, ‘help’,
‘hex’, ‘id’, ‘input’, ‘int’, ‘intern’, ‘isinstance’, ‘issubclass’, ‘iter’,
‘len’, ‘license’, ‘list’, ‘locals’, ‘long’, ‘map’, ‘max’, ‘min’, ‘object’,
‘oct’, ‘open’, ‘ord’, ‘pow’, ‘property’, ‘quit’, ‘range’, ‘raw_input’,
‘reduce’, ‘reload’, ‘repr’, ‘round’, ‘setattr’, ‘slice’, ‘staticmethod’,
‘str’, ‘sum’, ‘super’, ‘tuple’, ‘type’, ‘unichr’, ‘unicode’, ‘vars’,
‘xrange’, ‘zip’]
The example shown here uses Python 2.3, but the techniques apply to Python 2.4 as well.
For a language with as many features as Python, there are surprisingly few built-in elements. You can
run the dir function on modules you import as well. For example:
>>> import sys
>>> dir(sys)
[‘__displayhook__’, ‘__doc__’, ‘__excepthook__’, ‘__name__’,
‘__stderr__’, ‘__stdin__’, ‘__stdout__’, ‘_getframe’, ‘api_version’,
‘argv’, ‘builtin_module_names’, ‘byteorder’, ‘call_tracing’, ‘callstats’,
‘copyright’, ‘displayhook’, ‘exc_clear’, ‘exc_info’, ‘exc_type’, ‘excepthook’,
‘exec_prefix’, ‘executable’, ‘exit’, ‘getcheckinterval’, ‘getdefaultencoding’,
‘getdlopenflags’, ‘getfilesystemencoding’, ‘getrecursionlimit’, ‘getrefcount’,
‘hexversion’, ‘last_traceback’, ‘last_type’, ‘last_value’, ‘maxint’,
‘maxunicode’, ‘meta_path’, ‘modules’, ‘path’, ‘path_hooks’,
‘path_importer_cache’, ‘platform’, ‘prefix’, ‘ps1’, ‘ps2’, ‘setcheckinterval’,
‘setdlopenflags’, ‘setprofile’, ‘setrecursionlimit’, ‘settrace’, ‘stderr’,
‘stdin’, ‘stdout’, ‘version’, ‘version_info’, ‘warnoptions’]
Use dir to help examine modules, including the modules you create.
144

19.6 Page 186

▲back to top
Building a Module
Importing Modules
Before using a module, you need to import it. The standard syntax for importing follows:
import module
You can use this syntax with modules that come with Python or with modules you create. You can also
use the following alternative syntax:
from module import item
The alternative syntax enables you to specifically import just a class or function if that is all you need.
If a module has changed, you can reload the new definition of the module using the reload function.
The syntax follows:
reload(module)
Replace module with the module you want to reload.
With reload, always use parentheses. With import, do not use parentheses.
Finding Modules
To import a module, the Python interpreter needs to find the module. With a module, the Python inter-
preter first looks for a file named module.py, where module is the name of the module you pass to the
import statement. On finding a module, the Python interpreter will compile the module into a .pyc file.
When you next import the module, the Python interpreter can load the pre-compiled module, speeding
your Python scripts.
When you place an import statement in your scripts, the Python interpreter has to be able to find the
module. The key point is that the Python interpreter only looks in a certain number of directories for
your module. If you enter a name the Python interpreter cannot find, it will display an error, as shown in
the following example:
>>> import foo
Traceback (most recent call last):
File “<stdin>”, line 1, in ?
ImportError: No module named foo
The Python interpreter looks in the directories that are part of the module search path. These directories
are listed in the sys.path variable from the sys module:
To list where the Python interpreter looks for modules, print out the value of the
sys.path variable in the Python interpreter. For example
>>> import sys
>>> print sys.path
[‘’, ‘/System/Library/Frameworks/Python.framework/Versions/2.3/lib/python23.zip’,
‘/System/Library/Frameworks/Python.framework/Versions/2.3/lib/python2.3’,
‘/System/Library/Frameworks/Python.framework/Versions/2.3/lib/python2.3/
145

19.7 Page 187

▲back to top
Chapter 10
plat-darwin’,
‘/System/Library/Frameworks/Python.framework/Versions/2.3/lib/python2.3/plat-mac’,
‘/System/Library/Frameworks/Python.framework/Versions/2.3/lib/python2.3/plat-
mac/lib-scriptpackages’,
‘/System/Library/Frameworks/Python.framework/Versions/2.3/lib/python2.3/lib-tk’,
‘/System/Library/Frameworks/Python.framework/Versions/2.3/lib/python2.3/lib-
dynload’,
‘/System/Library/Frameworks/Python.framework/Versions/2.3/lib/python2.3/site-
packages’]
Note that one of the directory entries is empty, signifying the current directory.
Digging through Modules
Because Python is an open-source package, you can get the source code to the Python interpreter as well
as all modules. In fact, even with a binary distribution of Python, you’ll find the source code for modules
written in Python.
Start by looking in all the directories listed in the sys.path variable for files with names ending in .py.
These are Python modules. Some modules contain functions, and others contain classes and functions.
For example, the following module, MimeWriter, defines a class in the Python 2.3 distribution:
“””Generic MIME writer.
This module defines the class MimeWriter. The MimeWriter class implements
a basic formatter for creating MIME multi-part files. It doesn’t seek around
the output file nor does it use large amounts of buffer space. You must write
the parts out in the order that they should occur in the final file.
MimeWriter does buffer the headers you add, allowing you to rearrange their
order.
“””
import mimetools
__all__ = [“MimeWriter”]
class MimeWriter:
“””Generic MIME writer.
Methods:
__init__()
addheader()
flushheaders()
startbody()
startmultipartbody()
nextpart()
lastpart()
146

19.8 Page 188

▲back to top
Building a Module
A MIME writer is much more primitive than a MIME parser. It
doesn’t seek around on the output file, and it doesn’t use large
amounts of buffer space, so you have to write the parts in the
order they should occur on the output file. It does buffer the
headers you add, allowing you to rearrange their order.
General usage is:
f = <open the output file>
w = MimeWriter(f)
...call w.addheader(key, value) 0 or more times...
followed by either:
f = w.startbody(content_type)
...call f.write(data) for body data...
or:
w.startmultipartbody(subtype)
for each part:
subwriter = w.nextpart()
...use the subwriter’s methods to create the subpart...
w.lastpart()
The subwriter is another MimeWriter instance, and should be
treated in the same way as the toplevel MimeWriter. This way,
writing recursive body parts is easy.
Warning: don’t forget to call lastpart()!
XXX There should be more state so calls made in the wrong order
are detected.
Some special cases:
- startbody() just returns the file passed to the constructor;
but don’t use this knowledge, as it may be changed.
- startmultipartbody() actually returns a file as well;
this can be used to write the initial ‘if you can read this your
mailer is not MIME-aware’ message.
- If you call flushheaders(), the headers accumulated so far are
written out (and forgotten); this is useful if you don’t need a
body part at all, e.g. for a subpart of type message/rfc822
that’s (mis)used to store some header-like information.
- Passing a keyword argument ‘prefix=<flag>’ to addheader(),
start*body() affects where the header is inserted; 0 means
append at the end, 1 means insert at the start; default is
append for addheader(), but insert for start*body(), which use
it to determine where the Content-Type header goes.
147

19.9 Page 189

▲back to top
Chapter 10
“””
def __init__(self, fp):
self._fp = fp
self._headers = []
def addheader(self, key, value, prefix=0):
“””Add a header line to the MIME message.
The key is the name of the header, where the value obviously provides
the value of the header. The optional argument prefix determines
where the header is inserted; 0 means append at the end, 1 means
insert at the start. The default is to append.
“””
lines = value.split(“\\n”)
while lines and not lines[-1]: del lines[-1]
while lines and not lines[0]: del lines[0]
for i in range(1, len(lines)):
lines[i] = “ “ + lines[i].strip()
value = “\\n”.join(lines) + “\\n”
line = key + “: “ + value
if prefix:
self._headers.insert(0, line)
else:
self._headers.append(line)
def flushheaders(self):
“””Writes out and forgets all headers accumulated so far.
This is useful if you don’t need a body part at all; for example,
for a subpart of type message/rfc822 that’s (mis)used to store some
header-like information.
“””
self._fp.writelines(self._headers)
self._headers = []
def startbody(self, ctype, plist=[], prefix=1):
“””Returns a file-like object for writing the body of the message.
The content-type is set to the provided ctype, and the optional
parameter, plist, provides additional parameters for the
content-type declaration. The optional argument prefix determines
where the header is inserted; 0 means append at the end, 1 means
insert at the start. The default is to insert at the start.
“””
for name, value in plist:
ctype = ctype + ‘;\\n %s=\\”%s\\”’ % (name, value)
self.addheader(“Content-Type”, ctype, prefix=prefix)
self.flushheaders()
148

19.10 Page 190

▲back to top
Building a Module
self._fp.write(“\\n”)
return self._fp
def startmultipartbody(self, subtype, boundary=None, plist=[], prefix=1):
“””Returns a file-like object for writing the body of the message.
Additionally, this method initializes the multi-part code, where the
subtype parameter provides the multipart subtype, the boundary
parameter may provide a user-defined boundary specification, and the
plist parameter provides optional parameters for the subtype. The
optional argument, prefix, determines where the header is inserted;
0 means append at the end, 1 means insert at the start. The default
is to insert at the start. Subparts should be created using the
nextpart() method.
“””
self._boundary = boundary or mimetools.choose_boundary()
return self.startbody(“multipart/” + subtype,
[(“boundary”, self._boundary)] + plist,
prefix=prefix)
def nextpart(self):
“””Returns a new instance of MimeWriter which represents an
individual part in a multipart message.
This may be used to write the part as well as used for creating
recursively complex multipart messages. The message must first be
initialized with the startmultipartbody() method before using the
nextpart() method.
“””
self._fp.write(“\\n--” + self._boundary + “\\n”)
return self.__class__(self._fp)
def lastpart(self):
“””This is used to designate the last part of a multipart message.
It should always be used when writing multipart messages.
“””
self._fp.write(“\\n--” + self._boundary + “--\\n”)
if __name__ == ‘__main__’:
import test.test_MimeWriter
The majority of this small module is made up of documentation that instructs users how to use the mod-
ule. Documentation is important.
When you look through the standard Python modules, you can get a feel for how modules are put
together. It also helps when you want to create your own modules.
149

20 Pages 191-200

▲back to top

20.1 Page 191

▲back to top
Chapter 10
Creating Modules and Packages
Creating modules is easier than you might think. A module is merely a Python source file. In fact, any
time you’ve created a Python file, you have already been creating modules without even knowing it.
Use the following example to help you get started creating modules.
Try It Out Creating a Module with Functions
Enter the following Python code and name the file food.py:
def favoriteFood():
print ‘The only food worth eating is an omelet.’
This is your module. You then can import the module using the Python interpreter. For example:
>>> import food
>>> dir(food)
[‘__builtins__’, ‘__doc__’, ‘__file__’, ‘__name__’, ‘favoriteFood’]
How It Works
Python uses a very simple definition for a module. You can use any Python source file as a module, as
shown in this short example. The dir function lists the items defined in the module, including the func-
tion favoriteFood.
Once imported, you can execute the code in the module with a command like the following:
>>> food.favoriteFood()
The only food worth eating is an omelet.
If you don’t use the module name prefix, food in this case, you will get an error, as shown in the follow-
ing example:
>>> favoriteFood()
Traceback (most recent call last):
File “<stdin>”, line 1, in ?
NameError: name ‘favoriteFood’ is not defined
Using the alternative syntax for imports can eliminate this problem:
>>> from food import favoriteFood
>>> favoriteFood()
The only food worth eating is an omelet.
>>>
Congratulations! You are now a certified module creator.
150

20.2 Page 192

▲back to top
Building a Module
Working with Classes
Most modules define a set of related functions or classes. A class, as introduced in Chapter 6, holds data
as well as the methods that operate on that data. Python is a little looser than most programming lan-
guages, such as Java, C++, or C#, in that Python lets you break rules enforced in other languages. For
example, Python, by default, lets you access data inside a class. This does violate some of the concepts of
object-oriented programming but with good reason: Python aims first and foremost to be practical.
Defining Object-Oriented Programming
Computer geeks argue endlessly over what is truly object-oriented programming (OOP). Most experts,
however, agree on the following three concepts:
❑ Encapsulation
❑ Inheritance
❑ Polymorphism
Encapsulation is the idea that a class can hide the internal details and data necessary to perform a cer-
tain task. A class holds the necessary data, and you are not supposed to see that data under normal cir-
cumstances. Furthermore, a class provides a number of methods to operate on that data. These methods
can hide the internal details, such as network protocols, disk access, and so on. Encapsulation is a tech-
nique to simplify your programs. At each step in creating your program, you can write code that concen-
trates on a single task. Encapsulation hides the complexity.
Inheritance means that a class can inherit, or gain access to, data and methods defined in a parent class.
This just follows common sense in classifying a problem domain. For example, a rectangle and a circle
are both shapes. In this case, the base class would be Shapes. The Rectangle class would then inherit
from Shapes, as would the Circle class. Inheritance enables you to treat objects of both the Rectangle
and Circle classes as children and members of the Shape class, meaning you can write more generic
code in the base class, and become more specific in the children. (The terms children and child class, and
membership in a class, are similar and can be used interchangeably here.) For the most part, the base class
should be general and the subclasses specialized. Inheritance is often called specialization.
Polymorphism means that subclasses can override methods for more specialized behavior. For example,
a rectangle and a circle are both shapes. You may define a set of common operations such as move and
draw, that should apply to all shapes. However, the draw method for a Circle will obviously be differ-
ent than the draw method for a Rectangle. Polymorphism enables you to name both methods draw
and then call these methods as if the Circle and the Rectangle were both Shapes, which they are, at
least in this example.
Creating Classes
As described in Chapter 6, creating classes is easy. (In fact, most things in Python are pleasantly easy.)
The following example shows a simple class that represents a meal.
151

20.3 Page 193

▲back to top
Chapter 10
Try It Out Creating a Meal Class
The following code defines the Meal class. The full source file appears in the section on “Creating a
Whole Module.”
class Meal:
‘’’Holds the food and drink used in a meal.
In true object-oriented tradition, this class
includes setter methods for the food and drink.
Call printIt to pretty-print the values.
‘’’
def __init__(self, food=’omelet’, drink=’coffee’):
‘’’Initialize to default values.’’’
self.name = ‘generic meal’
self.food = food
self.drink = drink
def printIt(self, prefix=’’):
‘’’Print the data nicely.’’’
print prefix,’A fine’,self.name,’with’,self.food,’and’,self.drink
# Setter for the food.
def setFood(self, food=’omelet’):
self.food = food
# Setter for the drink.
def setDrink(self, drink=’coffee’):
self.drink = drink
# Setter for the name.
def setName(self, name=’’):
self.name = name
How It Works
Each instance of the Meal class holds three data values: the name of the meal, the food, and the drink.
By default, the Meal class sets the name to generic meal, the drink to coffee, and the food to an
omelet.
As with gin and tonics, omelets are not just for breakfast anymore.
The __init__ method initializes the data for the Meal. The printIt method prints out the internal
data in a friendly manner. Finally, to support developers used to stricter programming languages,
the Meal class defines a set of methods called setters. These setter methods, such as setFood and
setDrink, set data into the class.
These methods are not necessary in Python, as you can set the data directly.
See Chapter 6 for more information about classes.
152

20.4 Page 194

▲back to top
Building a Module
Extending Existing Classes
After you have defined a class, you can extend the class by defining subclasses. For example, you can
create a Breakfast class that represents the first meal of the day:
class Breakfast(Meal):
‘’’Holds the food and drink for breakfast.’’’
def __init__(self):
‘’’Initialize with an omelet and coffee.’’’
Meal.__init__(self, ‘omelet’, ‘coffee’)
self.setName(‘breakfast’)
The Breakfast class extends the Meal class as shown by the class definition:
class Breakfast(Meal):
Another subclass would naturally be Lunch:
class Lunch(Meal):
‘’’Holds the food and drink for lunch.’’’
def __init__(self):
‘’’Initialize with a sandwich and a gin and tonic.’’’
Meal.__init__(self, ‘sandwich’, ‘gin and tonic’)
self.setName(‘midday meal’)
# Override setFood().
def setFood(self, food=’sandwich’):
if food != ‘sandwich’ and food != ‘omelet’:
raise AngryChefException
Meal.setFood(self, food)
With the Lunch class, you can see some use for the setter methods. In the Lunch class, the setFood
method allows only two values for the food: a sandwich and an omelet. Nothing else is allowed or you
will make the chef angry.
The Dinner class also overrides a method — in this case, the printIt method:
class Dinner(Meal):
‘’’Holds the food and drink for dinner.’’’
def __init__(self):
‘’’Initialize with steak and merlot.’’’
Meal.__init__(self, ‘steak’, ‘merlot’)
self.setName(‘dinner’)
def printIt(self, prefix=’’):
‘’’Print even more nicely.’’’
print prefix,’A gourmet’,self.name,’with’,self.food,’and’,self.drink
Normally, you would place all these classes into a module. See the section “Creating a Whole Module”
for an example of a complete module.
153

20.5 Page 195

▲back to top
Chapter 10
Finishing Your Modules
After defining the classes and functions that you want for your module, the next step is to finish the
module to make it better fit into the conventions expected by Python users and the Python interpreter.
Finishing your module can include a lot of things, but at the very least you need to do the following:
❑ Define the errors and exceptions that apply to your module.
❑ Define which items in the module you want to export. This defines the public API for the
module.
❑ Document your module.
❑ Test your module.
❑ Provide a fallback function in case your module is executed as a program.
The following sections describe how to finish up your modules.
Defining Module-Specific Errors
Python defines a few standard exception classes, such as IOError and NotImplementedError. If those
classes apply, then by all means use those. Otherwise, you may need to define exceptions for specific
issues that may arise when using your module. For example, a networking module may need to define
a set of exceptions relating to network errors.
For the food-related theme used in the example module, you can define an AngryChefException.
To make this more generic, and perhaps allow reuse in other modules, the AngryChefException is
defined as a subclass of the more general SensitiveArtistException, representing issues raised by
touchy artsy types.
In most cases, your exception classes will not need to define any methods or initialize any data. The base
Exception class provides enough. For most exceptions, the mere presence of the exception indicates
the problem.
This is not always true. For example, an XML-parsing exception should probably contain the line num-
ber where the error occurred, as well as a description of the problem.
You can define the exceptions for the meal module as follows:
class SensitiveArtistException(Exception):
pass
class AngryChefException(SensitiveArtistException):
pass
This is just an example, of course. In your modules, define exception classes as needed. In addition to
exceptions, you should carefully decide what to export from your module.
154

20.6 Page 196

▲back to top
Building a Module
Choosing What to Export
When you use the from form of importing a module, you can specify which items in the module to
import. For example, the following statement imports the AngryChefException from the module meal:
from meal import AngryChefException
To import all public items from a module, you can use the following format:
from module_name import *
For example:
from meal import *
The asterisk, or star (*), tells the Python interpreter to import all public items from the module. What
exactly, is public? You, as the module designer, can choose to define whichever items you want to be
exported as public.
The Python interpreter uses two methods to determine what should be considered public:
❑ If you have defined the variable __all__ in your module, the interpreter uses __all__ to
determine what should be public.
❑ If you have not defined the variable __all__, the interpreter imports everything except items
with names that begin with an underscore, _, so printIt would be considered public, but
_printIt would not.
See Chapter 7 for more information about modules and the import statement.
As a best practice, always define __all__ in your modules. This provides you with explicit control over
what other Python scripts can import. To do this, simply create a sequence of text strings with the names
of each item you want to export from your module. For example, in the meal module, you can define
__all__ in the following manner:
__all__ = [‘Meal’, ‘AngryChefException’, ‘makeBreakfast’,
‘makeLunch’, ‘makeDinner’, ‘Breakfast’, ‘Lunch’, ‘Dinner’]
Each name in this sequence names a class or function to export from the module.
Choosing what to export is important. When you create a module, you are creating an API to perform
some presumably useful function. The API you export from a module then defines what users of your
module can do. You want to export enough for users of the module to get their work done, but you
don’t have to export everything. You may want to exclude items for a number of reasons, including the
following:
❑ Items you are likely to change should remain private until you have settled on the API for those
items. This gives you the freedom to make changes inside the module without impacting users
of the module.
155

20.7 Page 197

▲back to top
Chapter 10
❑ Modules can oftentimes hide, in purpose, complicated code. For example, an e-mail module can
hide the gory details of SMTP, POP3, and IMAP network e-mail protocols. Your e-mail module
could present an API that enables users to send messages, see which messages are available,
download messages, and so on.
Hiding the gory details of how your code is implemented is called encapsulation. Impress your friends
with lines like “making the change you are asking for would violate the rules of encapsulation . . .”
Always define, explicitly, what you want to export from a module. You should also always document
your modules.
Documenting Your Modules
It is vitally important that you document your modules. If not, no one, not even you, will know what
your modules do. Think ahead six months. Will you remember everything that went into your modules?
Probably not. The solution is simple: document your modules.
Python defines a few easy conventions for documenting your modules. Follow these conventions and
your modules will enable users to view the documentation in the standard way. At its most basic, for
each item you want to document, write a text string that describes the item. Enclose this text string in
three quotes, and place it immediately inside the item.
For example, to document a method or function, use the following code as a guide:
def makeLunch():
‘’’ Creates a Breakfast object.’’’
return Lunch()
The line in bold shows the documentation. The documentation that appears right after the function is
defined with the def statement.
Document a class similarly:
class Meal:
‘’’Holds the food and drink used in a meal.
In true object-oriented tradition, this class
includes setter methods for the food and drink.
Call printIt to pretty-print the values.
‘’’
Place the documentation on the line after the class statement.
Exceptions are classes, too. Document them as well:
class SensitiveArtistException(Exception):
‘’’Exception raised by an overly-sensitive artist.
Base class for artistic types.’’’
pass
156

20.8 Page 198

▲back to top
Building a Module
Note that even though this class adds no new functionality, you should describe the purpose of each
exception or class.
In addition, document the module itself. Start your module with the special three-quoted text string, as
shown here:
“””
Module for making meals in Python.
Import this module and then call
makeBreakfast(), makeDinner() or makeLunch().
“””
Place this documentation on the first line of the text file that contains the module. For modules, start
with one line that summarizes the purpose of the module. Separate this line from the remaining lines of
the documentation, using a blank line as shown previously. The Python help function will extract the
one-line summary and treat it specially. (See the Try It Out example, following, for more details about
how to call the help function.)
Usually, one or two lines per class, method, or function should suffice. In general, your documentation
should tell the user the following:
❑ How to call the function or method, including what parameters are necessary and what type of
data will be returned. Describe default values for parameters.
❑ What a given class was designed for, what is its purpose. Include how to use objects of the class.
❑ Any conditions that must exist prior to calling a function or method
❑ Any side effects or other parts of the system that will change as a result of the class. For exam-
ple, a method to erase all of the files on a disk should be documented as to what it does.
❑ Exceptions that may be raised and under what reasons these exceptions will be raised
Note that some people go way overboard in writing documentation. Too much documentation doesn’t
help, but don’t use this as an excuse to do nothing. Too much documentation is far better than none
at all.
A good rule of thumb comes from enlightened self-interest. Ask yourself what you would like to see in
someone else’s module and document to that standard.
You can view the documentation you write using the help function, as shown in the following example:
Try It Out Viewing Module Documentation
Launch the Python interpreter in interactive mode and then run the import and help commands as
shown in the following:
>>> import meal
>>> help(meal)
Help on module meal:
157

20.9 Page 199

▲back to top
Chapter 10
NAME
meal - Module for making meals in Python.
FILE
/Users/ericfj/writing/python/meal.py
DESCRIPTION
Import this module and then call
makeBreakfast(), makeDinner() or makeLunch().
CLASSES
exceptions.Exception
AngryChefException
SensitiveArtistException
Meal
Breakfast
Dinner
Lunch
class AngryChefException(exceptions.Exception)
| Exception that indicates the chef is unhappy.
|
| Methods inherited from exceptions.Exception:
|
| __getitem__(...)
|
| __init__(...)
|
| __str__(...)
class Breakfast(Meal)
| Holds the food and drink for breakfast.
|
| Methods defined here:
|
| __init__(self)
|
Initialize with an omelet and coffee.
|
| ----------------------------------------------------------------------
| Methods inherited from Meal:
|
| printIt(self, prefix=’’)
|
Print the data nicely.
|
| setDrink(self, drink=’coffee’)
|
# Setter for the name.
|
| setFood(self, food=’omelet’)
|
# Setter for the drink.
|
| setName(self, name=’’)
|
# Setter for the name.
158

20.10 Page 200

▲back to top
Building a Module
class Dinner(Meal)
| Holds the food and drink for dinner.
|
| Methods defined here:
|
| __init__(self)
|
Initialize with steak and merlot.
|
| printIt(self, prefix=’’)
|
Print even more nicely.
|
| ----------------------------------------------------------------------
| Methods inherited from Meal:
|
| setDrink(self, drink=’coffee’)
|
# Setter for the name.
|
| setFood(self, food=’omelet’)
|
# Setter for the drink.
|
| setName(self, name=’’)
|
# Setter for the name.
class Lunch(Meal)
| Holds the food and drink for lunch.
|
| Methods defined here:
|
| __init__(self)
|
Initialize with a sandwich and a gin and tonic.
|
| setFood(self, food=’sandwich’)
|
# Override setFood().
|
| ----------------------------------------------------------------------
| Methods inherited from Meal:
|
| printIt(self, prefix=’’)
|
Print the data nicely.
|
| setDrink(self, drink=’coffee’)
|
# Setter for the name.
|
| setName(self, name=’’)
|
# Setter for the name.
class Meal
| Holds the food and drink used in a meal.
| In true object-oriented tradition, this class
| includes setter methods for the food and drink.
|
| Call printIt to pretty-print the values.
159

21 Pages 201-210

▲back to top

21.1 Page 201

▲back to top
Chapter 10
|
| Methods defined here:
|
| __init__(self, food=’omelet’, drink=’coffee’)
|
Initialize to default values.
|
| printIt(self, prefix=’’)
|
Print the data nicely.
|
| setDrink(self, drink=’coffee’)
|
# Setter for the name.
|
| setFood(self, food=’omelet’)
|
# Setter for the drink.
|
| setName(self, name=’’)
|
# Setter for the name.
class SensitiveArtistException(exceptions.Exception)
| Exception raised by an overly-sensitive artist.
|
| Base class for artistic types.
|
| Methods inherited from exceptions.Exception:
|
| __getitem__(...)
|
| __init__(...)
|
| __str__(...)
FUNCTIONS
makeBreakfast()
Creates a Breakfast object.
makeDinner()
Creates a Breakfast object.
makeLunch()
Creates a Breakfast object.
test()
Test function.
DATA
__all__ = [‘Meal’, ‘AngryChefException’, ‘makeBreakfast’, ‘makeLunch’,...
(END)
Press q to quit the listing.
How It Works
The help function is your friend. It can show you the documentation for your modules, as well as the
documentation on any Python module.
160

21.2 Page 202

▲back to top
Building a Module
You must import a module prior to calling the help function to read the modules documentation.
The help function first prints the documentation for the module:
Help on module meal:
NAME
meal - Module for making meals in Python.
FILE
/Users/ericfj/writing/python/meal.py
DESCRIPTION
Import this module and then call
makeBreakfast(), makeDinner() or makeLunch().
Note how the help function separates the first summary line of the module documentation from the rest
of the documentation. The following shows the original string that documents this module:
“””
Module for making meals in Python.
Import this module and then call
makeBreakfast(), makeDinner() or makeLunch().
“””
The help function pulls out the first line for the NAME section of the documentation and the rest for the
DESCRIPTION section.
The help function summarizes the classes next and then shows the documentation for each class:
CLASSES
exceptions.Exception
AngryChefException
SensitiveArtistException
Meal
Breakfast
Dinner
Lunch
Each class is shown indented based on inheritance. In this example, the summary shows that the
Breakfast class inherits from the Meal class.
For each function and method, the help function prints out the documentation:
| printIt(self, prefix=’’)
|
Print the data nicely.
However, if you just have comments near a function or method definition, the help function will try to
associate a comment with the function or method. This doesn’t always work, however, as the help func-
tion alphabetizes the methods and functions. For example:
161

21.3 Page 203

▲back to top
Chapter 10
|
| setDrink(self, drink=’coffee’)
|
# Setter for the name.
|
| setFood(self, food=’omelet’)
|
# Setter for the drink.
|
| setName(self, name=’’)
|
# Setter for the name.
Note how the comments are associated with the wrong methods. Here is the original code:
# Setter for the food.
def setFood(self, food=’omelet’):
self.food = food
# Setter for the drink.
def setDrink(self, drink=’coffee’):
self.drink = drink
# Setter for the name.
def setName(self, name=’’):
self.name = name
The lesson here is to follow the Python conventions for documenting methods. To fix this error, change
the comments that appear above each method into a Python documentation string. Move the Python
documentation string down to the line immediately following the corresponding def command.
As you develop your module, you can call the help function repeatedly to see how changes in the code
change the documentation. If you have changed the Python source file for your module, however, you
need to reload the module prior to calling help. The reload function takes a module, as does help. The
syntax follows:
reload(module)
help(module)
For example, to reload the module meal, use the following code:
>>> reload(meal)
<module ‘meal’ from ‘meal.py’>
Just as documentation is important, so is testing. The more you can test your modules, the better your
modules will fit into Python applications. You’ll know that the functionality of the modules works prior
to using those modules in a program.
Testing Your Module
Testing is hard. Testing is yucky. That’s why testing is often skipped. Even so, testing your module can
verify that it works. More important, creating tests enables you to make changes to your module and
then verify that the functionality still works.
162

21.4 Page 204

▲back to top
Building a Module
Any self-respecting module should include a test function that exercises the functionality in the module.
Your tests should create instances of the classes defined in the module, and call methods on those
instances.
For example, the following method provides a test of the meal module:
def test():
‘’’Test function.’’’
print ‘Module meal test.’
# Generic no arguments.
print ‘Testing Meal class.’
m = Meal()
m.printIt(“\\t”)
m = Meal(‘green eggs and ham’, ‘tea’)
m.printIt(“\\t”)
# Test breakfast
print ‘Testing Breakfast class.’
b = Breakfast()
b.printIt(“\\t”)
b.setName(‘breaking of the fast’)
b.printIt(“\\t”)
# Test dinner
print ‘Testing Dinner class.’
d = Dinner()
d.printIt(“\\t”)
# Test lunch
print ‘Testing Lunch class.’
l = Lunch()
l.printIt(“\\t”)
print ‘Calling Lunch.setFood().’
try:
l.setFood(‘hotdog’)
except AngryChefException:
print “\\t”,’The chef is angry. Pick an omelet.’
Make your test functions part of your modules, so the tests are always available. You’ll learn more about
testing in Python in Chapter 12.
Testing is never done. You can always add more tests. Just do what you can.
163

21.5 Page 205

▲back to top
Chapter 10
Running a Module as a Program
Normally, modules aren’t intended to be run on their own. Instead, other Python scripts import items
from a module and then use those items. However, because a module can be any file of Python code,
you can indeed run a module.
Because modules aren’t meant to be run on their own, Python defines a convention for modules. When a
module is run on its own, it should execute the module tests. This provides a simple means to test your
modules: Just run the module as a Python script.
To help with this convention, Python provides a handy idiom to detect whether your module is run as a
program. Using the test function shown previously, you can use the following code to execute your
module tests:
if __name__ == ‘__main__’:
test()
If you look at the source code for the standard Python modules, you’ll find this idiom used repeatedly.
The next example runs the meal module, created in the section “Creating a Whole Module.”
Try It Out Running a Module
You can run a module, such as the meal module, as a program by using a command like the following:
$ python meal.py
Module meal test.
Testing Meal class.
A fine generic meal with omelet and coffee
A fine generic meal with green eggs and ham and tea
Testing Breakfast class.
A fine breakfast with omelet and coffee
A fine breaking of the fast with omelet and coffee
Testing Dinner class.
A gourmet dinner with steak and merlot
Testing Lunch class.
A fine midday meal with sandwich and gin and tonic
Calling Lunch.setFood().
The chef is angry. Pick an omelet.
How It Works
This example runs a module as a Python program. Using the idiom to detect this situation, the module
merely runs the test function. The output you see is the output of the tests.
Note how the output runs an instance of each class defined in the module, as well as tests the raising of
the AngryChefException.
If you follow all of the guidelines in this section, your modules will meet the expectations of other
Python developers. Moreover, your modules will work better in your scripts. You can see all of this in
action in the next section, which shows a complete Python module.
164

21.6 Page 206

▲back to top
Building a Module
Creating a Whole Module
The sections in this chapter so far show the elements you need to include in the modules you create. The
following example shows a complete module using the techniques described so far.
The meal module doesn’t do much. It supposedly modules a domain that includes food and drink over
three daily meals.
Obviously, this module doesn’t support Hobbits who require more than three meals a day.
The code in this module is purposely short. The intent is not to perform a useful task but instead to
show how to put together a module.
Try It Out Finishing a Module
Enter the following code and name the file meal.py:
“””
Module for making meals in Python.
Import this module and then call
makeBreakfast(), makeDinner() or makeLunch().
“””
__all__ = [‘Meal’,’AngryChefException’, ‘makeBreakfast’,
‘makeLunch’, ‘makeDinner’, ‘Breakfast’, ‘Lunch’, ‘Dinner’]
# Helper functions.
def makeBreakfast():
‘’’ Creates a Breakfast object.’’’
return Breakfast()
def makeLunch():
‘’’ Creates a Breakfast object.’’’
return Lunch()
def makeDinner():
‘’’ Creates a Breakfast object.’’’
return Dinner()
# Exception classes.
class SensitiveArtistException(Exception):
‘’’Exception raised by an overly-sensitive artist.
Base class for artistic types.’’’
pass
165

21.7 Page 207

▲back to top
Chapter 10
class AngryChefException(SensitiveArtistException):
‘’’Exception that indicates the chef is unhappy.’’’
pass
class Meal:
‘’’Holds the food and drink used in a meal.
In true object-oriented tradition, this class
includes setter methods for the food and drink.
Call printIt to pretty-print the values.
‘’’
def __init__(self, food=’omelet’, drink=’coffee’):
‘’’Initialize to default values.’’’
self.name = ‘generic meal’
self.food = food
self.drink = drink
def printIt(self, prefix=’’):
‘’’Print the data nicely.’’’
print prefix,’A fine’,self.name,’with’,self.food,’and’,self.drink
# Setter for the food.
def setFood(self, food=’omelet’):
self.food = food
# Setter for the drink.
def setDrink(self, drink=’coffee’):
self.drink = drink
# Setter for the name.
def setName(self, name=’’):
self.name = name
class Breakfast(Meal):
‘’’Holds the food and drink for breakfast.’’’
def __init__(self):
‘’’Initialize with an omelet and coffee.’’’
Meal.__init__(self, ‘omelet’, ‘coffee’)
self.setName(‘breakfast’)
class Lunch(Meal):
‘’’Holds the food and drink for lunch.’’’
def __init__(self):
‘’’Initialize with a sandwich and a gin and tonic.’’’
Meal.__init__(self, ‘sandwich’, ‘gin and tonic’)
self.setName(‘midday meal’)
166

21.8 Page 208

▲back to top
Building a Module
# Override setFood().
def setFood(self, food=’sandwich’):
if food != ‘sandwich’ and food != ‘omelet’:
raise AngryChefException
Meal.setFood(self, food)
class Dinner(Meal):
‘’’Holds the food and drink for dinner.’’’
def __init__(self):
‘’’Initialize with steak and merlot.’’’
Meal.__init__(self, ‘steak’, ‘merlot’)
self.setName(‘dinner’)
def printIt(self, prefix=’’):
‘’’Print even more nicely.’’’
print prefix,’A gourmet’,self.name,’with’,self.food,’and’,self.drink
def test():
‘’’Test function.’’’
print ‘Module meal test.’
# Generic no arguments.
print ‘Testing Meal class.’
m = Meal()
m.printIt(“\\t”)
m = Meal(‘green eggs and ham’, ‘tea’)
m.printIt(“\\t”)
# Test breakfast
print ‘Testing Breakfast class.’
b = Breakfast()
b.printIt(“\\t”)
b.setName(‘breaking of the fast’)
b.printIt(“\\t”)
# Test dinner
print ‘Testing Dinner class.’
d = Dinner()
d.printIt(“\\t”)
# Test lunch
print ‘Testing Lunch class.’
l = Lunch()
l.printIt(“\\t”)
167

21.9 Page 209

▲back to top
Chapter 10
print ‘Calling Lunch.setFood().’
try:
l.setFood(‘hotdog’)
except AngryChefException:
print “\\t”,’The chef is angry. Pick an omelet.’
# Run test if this module is run as a program.
if __name__ == ‘__main__’:
test()
How It Works
The meal module follows the techniques shown in this chapter for creating a complete module, with
testing, documentation, exceptions, classes, and functions. Note how the tests are about as long as the
rest of the code. You’ll commonly find this to be the case.
After you’ve built a module, you can import the module into other Python scripts. For example, the fol-
lowing script calls on classes and functions in the meal module:
import meal
print ‘Making a Breakfast’
breakfast = meal.makeBreakfast()
breakfast.printIt(“\\t”)
print ‘Making a Lunch’
lunch = meal.makeLunch()
try:
lunch.setFood(‘pancakes’)
except meal.AngryChefException:
print “\\t”,’Cannot make a lunch of pancakes.’
print “\\t”,’The chef is angry. Pick an omelet.’
This example uses the normal form for importing a module:
import meal
When you run this script, you’ll see output like the following:
$ python mealtest.py
Making a Breakfast
A fine breakfast with omelet and coffee
Making a Lunch
Cannot make a lunch of pancakes.
The chef is angry. Pick an omelet.
The next script shows an alternate means to import the module:
from meal import *
168

21.10 Page 210

▲back to top
Building a Module
The full script follows:
from meal import *
print ‘Making a Breakfast’
breakfast = makeBreakfast()
breakfast.printIt(“\\t”)
print ‘Making a Lunch’
lunch = makeLunch()
try:
lunch.setFood(‘pancakes’)
except AngryChefException:
print “\\t”,’Cannot make a lunch of pancakes.’
print “\\t”,’The chef is angry. Pick an omelet.’
Note how with this import form, you can call the makeLunch and makeBreakfast functions without
using the module name, meal, as a prefix on the call.
The output of this script should look familiar.
$ python mealtest2.py
Making a Breakfast
A fine breakfast with omelet and coffee
Making a Lunch
Cannot make a lunch of pancakes.
The chef is angry. Pick an omelet.
Be very careful with the names you use for variables. The example module has a name of meal. This
means you don’t want to use that name in any other context, such as for a variable. If you do, you will
effectively overwrite the definition of meal as a module. The following example shows the pitfall to this
approach.
Try It Out Smashing Imports
Enter the following script and name the file mealproblem.py:
import meal
print ‘Making a Breakfast’
meal = meal.makeBreakfast()
meal.printIt(“\\t”)
print ‘Making a Lunch’
lunch = meal.makeLunch()
169

22 Pages 211-220

▲back to top

22.1 Page 211

▲back to top
Chapter 10
try:
lunch.setFood(‘pancakes’)
except meal.AngryChefException:
print “\\t”,’Cannot make a lunch of pancakes.’
print “\\t”,’The chef is angry. Pick an omelet.’
When you run this script, you’ll see the following error:
$ python mealproblem.py
Making a Breakfast
A fine breakfast with omelet and coffee
Making a Lunch
Traceback (most recent call last):
File “mealproblem.py”, line 10, in ?
lunch = meal.makeLunch()
AttributeError: Breakfast instance has no attribute ‘makeLunch’
How It Works
This script uses meal as a module as well as meal as an instance of the class Breakfast. The following
lines are the culprit:
import meal
meal = meal.makeBreakfast()
When you run this code, the name meal is now a variable, an instance of the class Breakfast. This
changes the interpretation of the following line:
lunch = meal.makeLunch()
The intent of this line is to call the function makeLunch in the module meal. However, because meal is
now an object, the Python interpreter tries to call the makeLunch method on the object, an instance of the
Breakfast class. Because the Breakfast class has no method named makeLunch, the Python inter-
preter raises an error.
The syntax for using modules and calling functions in modules looks very much like the syntax for call-
ing methods on an object. Be careful.
After building your module and testing it, the next step is to install it.
Installing Your Modules
The Python interpreter looks for modules in the directories listed in the sys.path variable. The
sys.path variable includes the current directory, so you can always use modules available locally. If
you want to use a module you’ve written in multiple scripts, or on multiple systems, however, you need
to install it into one of the directories listed in the sys.path variable.
170

22.2 Page 212

▲back to top
Building a Module
In most cases, you’ll want to place your Python modules in the site-packages directory. Look in the
sys.path listing and find a directory name ending in site-packages. This is a directory for packages
installed at a site that are not part of the Python standard library of packages.
In addition to modules, you can create packages of modules, a set of related modules that install into the
same directory structure. See the Python documentation at http://docs.python.org for more on
this subject.
You can install your modules using one of three mechanisms:
❑ You can do everything by hand and manually create an installation script or program.
❑ You can create an installer specific to your operating system, such as MSI files on Windows, an
RPM file on Linux, or a DMG file on Mac OS X.
❑ You can use the handy Python distutils package, short for distribution utilities, to create a
Python-based installer.
To use the Python distutils, you need to create a setup script, named setup.py. A minimal setup
script can include the following:
from distutils.core import setup
setup(name=’NameOfModule’,
version=’1.0’,
py_modules=[‘NameOfModule’],
)
You need to include the name of the module twice. Replace NameOfModule with the name of your mod-
ule, such as meal in the examples in this chapter.
Name the script setup.py.
After you have created the setup.py script, you can create a distribution of your module using the fol-
lowing command:
python setup.py sdist
The argument sdist is short for software distribution. You can try this out with the following example.
Try It Out Creating an Installable Package
Enter the following script and name the file setup.py:
from distutils.core import setup
setup(name=’meal’,
version=’1.0’,
py_modules=[‘meal’],
)
171

22.3 Page 213

▲back to top
Chapter 10
Run the following command to create a Python module distribution:
$ python setup.py sdist
running sdist
warning: sdist: missing required meta-data: url
warning: sdist: missing meta-data: either (author and author_email) or (maintainer
and maintainer_email) must be supplied
warning: sdist: manifest template ‘MANIFEST.in’ does not exist (using default file
list)
warning: sdist: standard file not found: should have one of README, README.txt
writing manifest file ‘MANIFEST’
creating meal-1.0
making hard links in meal-1.0...
hard linking meal.py -> meal-1.0
hard linking setup.py -> meal-1.0
creating dist
tar -cf dist/meal-1.0.tar meal-1.0
gzip -f9 dist/meal-1.0.tar
removing ‘meal-1.0’ (and everything under it)
How It Works
Notice all the complaints. The setup.py script was clearly not complete. It included enough to create
the distribution, but not enough to satisfy the Python conventions. When the setup.py script com-
pletes, you should see the following files in the current directory:
$ ls
MANIFEST
dist/
meal.py
setup.py
The setup.py script created the dist directory and the MANIFEST file. The dist directory contains one
file, a compressed version of our module:
$ ls dist
meal-1.0.tar.gz
You now have a one-file distribution of your module, which is kind of silly because the module itself
was just one file. The advantage of distutils is that your module will be properly installed.
You can then take the meal-1.0.tar.gz file to another system and install the module. First, uncom-
press and expand the bundle. On Linux, Unix, and Mac OS X, use the following commands:
$ gunzip meal-1.0.tar.gz
$ tar xvf meal-1.0.tar
meal-1.0/
meal-1.0/meal.py
meal-1.0/PKG-INFO
meal-1.0/setup.py
On Windows, use a compression program such as WinZip, which can handle the .tar.gz files.
You can install the module after it is expanded with the following command:
python setup.py install
172

22.4 Page 214

▲back to top
Building a Module
For example:
$ python setup.py install
running install
running build
running build_py
creating build
creating build/lib
copying meal.py -> build/lib
running install_lib
copying build/lib/meal.py -> /System/Library/Frameworks/Python.framework/
Versions/2.3/lib/python2.3/site-packages
byte-compiling /System/Library/Frameworks/Python.framework/Versions/2.3/lib/
python2.3/site-packages/meal.py to meal.pyc
The neat thing about the distutils is that it works for just about any Python module. The installation
command is the same, so you just need to know one command to install Python modules on any system.
Another neat thing is that the installation creates documentation on your module that is viewable with
the pydoc command. For example, the following shows the first page of documentation on the meal
module:
$ pydoc meal
Help on module meal:
NAME
meal - Module for making meals in Python.
FILE
/Users/ericfj/writing/python/inst2/meal-1.0/meal.py
DESCRIPTION
Import this module and then call
makeBreakfast(), makeDinner() or makeLunch().
CLASSES
exceptions.Exception
SensitiveArtistException
AngryChefException
Meal
Breakfast
Dinner
Lunch
class AngryChefException(SensitiveArtistException)
| Exception that indicates the chef is unhappy.
:
See the Python documentation at www.python.org/doc/2.4/dist/dist.html for more on writ-
ing distutils setup scripts.
173

22.5 Page 215

▲back to top
Chapter 10
Summary
This chapter pulls together concepts from the earlier chapters to delve into how to create modules by
example. If you follow the techniques described in this chapter, your modules will fit in with other mod-
ules and follow the import Python conventions.
A module is simply a Python source file that you choose to treat as a module. Simple as that sounds, you
need to follow a few conventions when creating a module:
❑ Document the module and all classes, methods, and functions in the module.
❑ Test the module and include at least one test function.
❑ Define which items in the module to export — which classes, functions, and so on.
❑ Create any exception classes you need for the issues that can arise when using the module.
❑ Handle the situation in which the module itself is executed as a Python script.
Inside your modules, you’ll likely define classes, which Python makes exceedingly easy.
While developing your module, you can use the help and reload functions to display documentation
on your module (or any other module for that matter) and reload the changed module, respectively.
After you have created a module, you can create a distributable bundle of the module using the
distutils. To do this, you need to create a setup.py script.
Chapter 11 describes regular expressions, an important concept used for finding relevant information
in a sea of data.
Exercises
1. How can you get access to the functionality provided by a module?
2. How can you control which items from your modules are considered public? (Public items are
available to other Python scripts.)
3. How can you view documentation on a module?
4. How can you find out what modules are installed on a system?
5. What kind of Python commands can you place in a module?
174

22.6 Page 216

▲back to top
11
Text Processing
There is a whole range of applications for which scripting languages like Python are perfectly
suited; and in fact scripting languages were arguably invented specifically for these applications,
which involve the simple search and processing of various files in the directory tree. Taken
together, these applications are often called text processing. Python is a great scripting tool for
both writing quick text processing scripts and then scaling them up into more generally useful
code later, using its clean object-oriented coding style. This chapter will show you the following:
❑ Some of the typical reasons you need text processing scripts
❑ A few simple scripts for quick system administration tasks
❑ How to navigate around in the directory structure in a platform-independent way, so
your scripts will work fine on Linux, Windows, or even the Mac
❑ How to create regular expressions to compare the files found by the os and os.path
modules.
❑ How to use successive refinement to keep enhancing your Python scripts to winnow
through the data found.
Text processing scripts are one of the most useful tools in the toolbox of anybody who seriously
works with computer systems, and Python is a great way to do text processing. You’re going to
like this chapter.
Why Text Processing Is So Useful
In general, the whole idea behind text processing is simply finding things. There are, of course,
situations in which data is organized in a structured way; these are called databases and that’s
not what this chapter is about. Databases carefully index and store data in such a way that if you
know what you’re looking for, you can retrieve it quickly. However, in some data sources, the
information is not at all orderly and neat, such as directory structures with hundreds or thousands
of files, or logs of events from system processes consisting of thousands or hundreds of thousands
of lines, or even e-mail archives with months of exchanges between people.

22.7 Page 217

▲back to top
Chapter 11
When data of that nature needs to be searched for something, or processed in some way, then text pro-
cessing is in its element. Of course, there’s no reason not to combine text processing with other data-
access methods; you might find yourself writing scripts rather often that run through thousands of lines
of log output and do occasional RDBMS lookups (Relational DataBase Management Systems — you’ll
learn about these in Chapter 14) on some of the data they run across. This is a natural way to work.
Ultimately, this kind of script can very often get used for years as part of a back-end data processing sys-
tem. If the script is written in a language like Perl, it can sometimes be quite opaque when some poor
soul is assigned five years later to “fix it.” Fortunately, this is a book about Python programming, and so
the scripts written here can easily be turned into reusable object classes — later, you’ll look at an illustra-
tive example.
The two main tools in your text processing belt are directory navigation, and an arcane technology
called regular expressions. Directory navigation is one area in which different operating systems can
really wreak havoc on simple programs, because the three major operating system families (Unix,
Windows, and the Mac) all organize their directories differently; and, most painfully, they use different
characters to separate subdirectory names. Python is ready for this, though — a series of cross-platform
tools are available for the manipulation of directories and paths that, when used consistently, can elimi-
nate this hassle entirely. You saw these in Chapter 8, and you’ll see more uses of these tools here.
A regular expression is a way of specifying a very simple text parser, which then can be applied rela-
tively inexpensively (which means that it will be fast) to any number of lines of text. Regular expressions
crop up in a lot of places, and you’ve likely seen them before. If this is your first exposure to them, how-
ever, you’ll be pretty pleased with what they can do. In the scope of this chapter, you’re just going to
scratch the surface of full-scale regular expression power, but even this will give your scripts a lot of
functionality.
You’ll first look at some of the reasons you might want to write text processing scripts, and then you’ll
do some experimentation with your new knowledge. The most common reasons to use regular expres-
sions include the following:
❑ Searching for files
❑ Extracting useful data from program logs, such as a web server log
❑ Searching through your e-mail
The following sections introduce these uses.
Searching for Files
Searching for files, or doing something with some files, is a mainstay of text processing. For example,
suppose that you spent a few months ripping your entire CD collection to MP3 files, without really pay-
ing attention to how you were organizing the hundreds of files you were tossing into some arbitrarily
made-up set of directories. This wouldn’t be a problem if you didn’t wait a couple of months before
thinking about organizing your files into directories according to artist — and only then realized that the
directory structure you ended up with was hopelessly confused.
176

22.8 Page 218

▲back to top
Text Processing
Text processing to the rescue! Write a Python script that scans the hopelessly nonsensical directory struc-
ture and then divide each filename into parts that might be an artist’s name. Then take that potential
name and try to look it up in a music database. The result is that you could rearrange hundreds of files
into directories by, if not the name of the artist, certainly some pretty good guesses which will get you
close to having a sensible structure. From there, you would be able to explore manually and end up
actually having an organized music library.
This is a one-time use of a text processing script, but you can easily imagine other scenarios in which
you might use a similarly useful script on a regular basis, as when you are handling data from a client or
from a data source that you don’t control. Of course, if you need to do this kind of sorting often, you can
easily use Python to come up with some organized tool classes that perform these tasks to avoid having
to duplicate your effort each time.
Whenever you face a task like this, a task that requires a lot of manual work manipulating data on your
computer, think Python. Writing a script or two could save you hours and hours of tedious work.
A second but similar situation results as a fallout of today’s large hard disks. Many users store files
willy-nilly on their hard disk, but never seem to have the time to organize their files. A worse situation
occurs when you face a hard disk full of files and you need to extract some information you know is
there on your computer, but you’re not sure where exactly. You are not alone. Apple, Google, Microsoft
and others are all working on desktop search techniques that help you search through the data in the
files you have collected to help you to extract useful information.
Think of Python as a desktop search on steroids, because you can create scripts with a much finer control
over the search, as well as perform operations on the files found.
Clipping Logs
Another common text-processing task that comes up in system administration is the need to sift
through log files for various information. Scripts that filter logs can be spur-of-the-moment affairs meant
to answer specific questions (such as “When did that e-mail get sent?” or “When was the last time my
program log one specific message?”), or they might be permanent parts of a data processing system
that evolves over time to manage ongoing tasks. These could be a part of a system administration and
performance-monitoring system, for instance. Scripts that regularly filter logs for particular subsets of
the information are often said to be clipping logs — the idea being that, just as you clip polygons to fit
on the screen, you can also clip logs to fit into whatever view of the system you need.
However you decide to use them, after you gain some basic familiarity with the techniques used, these
scripts become almost second nature. This is an application where regular expressions are used a lot,
for two reasons: First, it’s very common to use a Unix shell command like grep to do first-level log clip-
ping; second, if you do it in Python, you’ll probably be using regular expressions to split the line into
usable fields before doing more work with it. In any one clipping task, you may very well be using both
techniques.
After a short introduction to traversing the file system and creating regular expressions, you’ll look at a
couple of scripts for text processing in the following sections.
177

22.9 Page 219

▲back to top
Chapter 11
Sifting through Mail
The final text processing task is one that you’ve probably found useful (or if you haven’t, you’ve badly
wanted it): the processing of mailbox files to find something that can’t be found by your normal Inbox
search feature. The most common reason you need something more powerful for this is that the mailbox
file is either archived, so that you can access the file, but not read it with your mail reader easily, or it has
been saved on a server where you’ve got no working mail client installed. Rather than go through the
hassle of moving it into your Inbox tree and treating it like an active folder, you might find it simpler just
to write a script to scan it for whatever you need.
However, you can also easily imagine a situation in which your search script might want to get data
from an outside source, such as a web page or perhaps some other data source, like a database (see
Chapter 14 for more about databases), to cross-reference your data, or do some other task during the
search that can’t be done with a plain vanilla mail client. In that case, text processing combined with any
other technique can be an incredibly useful way to find information that may not be easy to find any
other way.
Navigating the File System
with the os Module
The os module and its submodule os.path are one of the most helpful things about using Python for a
lot of day-to-day tasks that you have to perform on a lot of different systems. If you often need to write
scripts and programs on either Windows or Unix that would still work on the other operating system,
you know from Chapter 8 that Python takes care of much of the work of hiding the differences between
how things work on Windows and Unix.
In this chapter, we’re going to completely ignore a lot of what the os module can do (ranging from pro-
cess control to getting system information) and just focus on some of the functions useful for working
with files and directories. Some things you’ve been introduced to already, while others are new.
One of the difficult and annoying points about writing cross-platform scripts is the fact that directory
names are separated by backslashes (\\) under Windows, but forward slashes (/) under Unix. Even
breaking a full path down into its components is irritatingly complicated if you want your code to work
under both operating systems.
Furthermore, Python, like many other programming languages, makes special use of the backslash char-
acter to indicate special text, such as \\n for a newline. This complicates your scripts that create file paths
on Windows.
With Python’s os.path module, however, you get some handy functions that will split and join path
names for you automatically with the right characters, and they’ll work correctly on any OS that Python
is running on (including the Mac.) You can call a single function to iterate through the directory struc-
ture and call another function of your choosing on each file it finds in the hierarchy. You’ll be seeing a lot
of that function in the examples that follow, but first let’s look at an overview of some of the useful func-
tions in the os and os.path modules that you’ll be using.
178

22.10 Page 220

▲back to top
Function Name, as Called
os.getcwd()
os.listdir(directory)
os.stat(path)
os.path.split(path)
os.path.join(components)
Text Processing
Description
Returns the current directory. You can think of this function
as the basic coordinate of directory functions in whatever
language.
Returns a list of the names of files and subdirectories stored
in the named directory. You can then run os.stat() on the
individual files — for example, to determine which are files
and which are subdirectories.
Returns a tuple of numbers, which give you everything you
could possibly need to know about a file (or directory). These
numbers are taken from the structure returned by the ANSI C
function of the same name, and they have the following mean-
ings (some are dummy values under Windows, but they’re in
the same places!):
st_mode:
permissions on the file
st_ino:
inode number (Unix)
st_dev:
device number
st_nlink:
link number (Unix)
st_uid:
userid of owner
st_gid:
groupid of owner
st_size:
size of the file
st_atime:
time of last access
st_mtime:
time of last modification
st_ctime:
time of creation
Splits the path into its component names appropriately for the
current operating system. Returns a tuple, not a list. This
always surprises me.
Joins name components into a path appropriate to the current
operating system
Table continued on following page
179

23 Pages 221-230

▲back to top

23.1 Page 221

▲back to top
Chapter 11
Function Name, as Called
os.path.normcase(path)
os.path.walk(start, function, arg)
Description
Normalizes the case of a path. Under Unix, this has no effect
because filenames are case-sensitive; but under Windows,
where the OS will silently ignore case when comparing file-
names, it’s useful to run normcase on a path before comparing
it to another path so that if one has capital letters, but the other
doesn’t, Python will be able to compare the two the same way
that the operation system would — that is, they’d be the same
regardless of capitalizations in the path names, as long as that’s
the only difference. Under Windows, the function returns a
path in all lowercase and converts any forward slashes into
backslashes.
This is a brilliant function that iterates down through a direc-
tory tree starting at start. For each directory, it calls the function
function like this: function(arg, dir, files), where the arg is any
arbitrary argument (usually something that is modified, like a
dictionary), dir is the name of the current directory, and files is
a list containing the names of all the files and subdirectories in
that directory. If you modify the files list in place by removing
some subdirectories, you can prevent os.path.walk() from
iterating into those subdirectories.
There are more functions where those came from, but these are the ones used in the example code that
follows. You will likely use these functions far more than any others in these modules. Many other use-
ful functions can be found in the Python module documentation for os and os.path.
Try It Out Listing Files and Playing with Paths
The best way to get to know functions in Python is to try them out in the interpreter. Try some of the pre-
ceding functions to see what the responses will look like.
1. From the Python interpreter, import the os and os.path modules:
>>> import os, os.path
2. First, see where you are in the file system. This example is being done under Windows, so your
mileage will vary:
>>> os.getcwd()
‘C:\\\\Documents and Settings\\\\michael’
3. If you want to do something with this programmatically, you’ll probably want to break it down
into the directory path, as a tuple (use join to put the pieces back together):
>>> os.path.split (os.getcwd())
(‘C:\\\\Documents and Settings’, ‘michael’)
180

23.2 Page 222

▲back to top
Text Processing
4. To find out some interesting things about the directory, or any file, use os.stat:
>>> os.stat(‘.’)
(16895, 0, 2, 1, 0, 0, 0, 1112654581, 1097009078, 1019063193)
Note that the directory named ‘.’ is shorthand for the current directory.
5. If you actually want to list the files in the directory, do this:
>>> os.listdir(‘.’)
[‘.javaws’, ‘.limewire’, ‘Application Data’, ‘Cookies’, ‘Desktop’, ‘Favorites’,
‘gsview32.ini’, ‘Local Settings’, ‘My Documents’, ‘myfile.txt’, ‘NetHood’,
‘NTUSER.DAT’, ‘ntuser.dat.LOG’, ‘ntuser.ini’, ‘PrintHood’, ‘PUTTY.RND’, ‘Recent’,
‘SendTo’, ‘Start Menu’, ‘Templates’, ‘UserData’, ‘WINDOWS’]
How It Works
Most of that was perfectly straightforward and easy to understand, but let’s look at a couple of points
before going on and writing a complete script or two.
First, you can easily see how you might construct an iterating script using listdir, split, and stat —
but you don’t have to, because os.path provides the walk function to do just that, as you’ll see later.
The walk function not only saves you the time and effort of writing and debugging an iterative algo-
rithm where you search everything in your own way, but it also runs a bit faster because it’s a built-in to
Python, but written in C, which can make things in cases like this. You probably will seldom want to
write iterators in Python when you’ve already got something built-in that does the same job.
Second, note that the output of the stat call, which comes from a system call, is pretty opaque. The
tuple it returns corresponds to the structure returned from the POSIX C library function of the same
name, and its component values are described in the preceding table; and, of course, in the Python docu-
mentation. The stat function really does tell you nearly anything you might want to know about a file
or directory, so it’s a valuable function to understand for when you’ll need it, even though it’s a bit
daunting at first glance.
Try It Out Searching for Files of a Particular Type
If you have worked with any other programming languages, you’ll like how easy searching for files is
with Python. Whether or not you’ve done this before in another language, you’ll notice how the example
script is extremely short for this type of work. The following example uses the os and os.path modules
to search for PDF files in the directory — which means the current directory — wherever you are when
you call the function. On a Unix or Linux system, you could use the command line and, for example,
the Unix find command. However, if you don’t do this too often that would mean that each time you
wanted to look for files, you’d need to figure out the command-line syntax for find yet again. (Because
of how much find does, that can be difficult — and that difficulty is compounded by how it expects you
to be familiar already with how it works!) Also, another advantage to doing this in Python is that by
using Python to search for files you can refine your script to do special things based on what you find,
and as you discover new uses for your program, you can add new features to it to find files in ways that
you find you need. For instance, as you search for files you may see far too many results to look at. You
can refine your Python script to further winnow the results to find just what you need.
181

23.3 Page 223

▲back to top
Chapter 11
This is a great opportunity to show off the nifty os.path.walk function, so that’s the basis of this script.
This function is great because it will do all the heavy lifting of file system iteration for you, leaving you
to write a simple function to do something with whatever it finds along the way:
1. Using your favorite text editor, open a script called scan_pdf.py in the directory you want to
scan for PDFs and enter the following code:
import os, os.path
import re
def print_pdf (arg, dir, files):
for file in files:
path = os.path.join (dir, file)
path = os.path.normcase (path)
if re.search (r”.*\\.pdf”, path):
print path
os.path.walk (‘.’, print_pdf, 0)
2. Run it. Obviously, the following output will not match yours. For the best results, add a bunch
of files that end in .pdf to this directory!
$ python scan_pdf.py
.\\95-04.pdf
.\\non-disclosure agreement 051702.pdf
.\\word pro - dokument in lotus word pro 9 dokument45.pdf
.\\101translations\\2003121803\\2003121803.pdf
.\\101translations\\2004101810\\scan.pdf
.\\bluemangos\\purchase order - michael roberts smb-pt134.pdf
.\\bluemangos\\smb_pt134.pdf
.\\businessteam.hu\\aok.pdf
.\\businessteam.hu\\chn14300-2.pdf
.\\businessteam.hu\\diplom_bardelmeier.pdf
.\\businessteam.hu\\doktor_bardelmeier.pdf
.\\businessteam.hu\\finanzamt_1.pdf
.\\businessteam.hu\\zollbescheinigung.pdf
.\\businessteam.hu\\monday\\s3.pdf
.\\businessteam.hu\\monday\\s4.pdf
.\\businessteam.hu\\monday\\s5.pdf
.\\gerard\\done\\tg82-20nc-md-04.07.pdf
.\\gerard\\polytronic\\iau-reglement_2005.pdf
.\\gerard\\polytronic\\tg82-20bes user manual\\tg82-20bes-md-27.05.pdf
.\\glossa\\neumag\\de_993_ba_s5.pdf
.\\glossa\\pepperl+fuchs\\5626eng3con\\vocab - 3522a_recom_flsd.pdf
.\\glossa\\pepperl+fuchs\\5769eng4\\5769eng4 - td4726_8400 d-e - 16.02.04.pdf
How It Works
This is a nice little script, isn’t it? Python does all the work, and you get a list of the PDFs in your directo-
ries, including their location and their full names — even with spaces, which can be difficult to deal with
under Unix and Linux.
182

23.4 Page 224

▲back to top
Text Processing
A little extra work with the paths has been done so that it’s easier to see what’s where: a call to
os.path.join builds the full (relative) pathname of each PDF from the starting directory and a call
to os.path.normcase makes sure that all the filenames are lowercase under Windows. Under Unix,
normcase would have no effect, because case is significant under Unix, so you don’t want to change the
capitalization (and it doesn’t change it), but under Windows, it makes it easier to see whether the file-
name ends in .pdf if you have them all appear in lowercase.
Note the use of a very simple regular expression to check the ending of the filename. You could also
have used os.path.splitext to get a tuple with the file’s base name and its extension, and compared
that to pdf, which arguably would have been cleaner. However, because this script is effectively laid out
as a filter, starting it out with a regular expression, also called regexp, comparison from the beginning
makes sense. Doing it this way means that if you decide later to restrict the output in some way, like
adding more filters based on needs you find you have, you can just add more regexp comparisons and
have nice, easy-to-understand code in the text expression. This is more a question of taste than anything
else. (It was also a good excuse to work in a first look at regular expressions and to demonstrate that
they’re really not too hard to understand.)
If you haven’t seen it before, the form r”<string constant>” simply tells Python that the string con-
stant should suppress all special processing for backslash values. Thus, while “\\n” is a string one char-
acter in length containing a newline, r”\\n” is a string two characters in length, containing a backslash
character followed by the letter ‘n’. Because regular expressions tend to contain a lot of backslashes, it’s
very convenient to be able to suppress their special meaning with this switch.
Try It Out Refining a Search
As it turned out, there were few enough PDF files (about 100) in the example search results that you
should be able to find the files you were looking for simply by looking through the list; but very often
when doing a search of this kind you first look at the results you get on the first pass and then use that
knowledge to zero in on what you ultimately need. The process of zeroing in involves trying out the
script, and then as you see that it could be returning better results, making successive changes to your
scripts to better find the information you want.
To get a flavor of that kind of successive or iterative programming, assume that instead of just showing
all the PDFs, you also want to exclude all PDFs with a space in the name. For example, because the
files you were looking for were downloaded from web sites, they in fact wouldn’t have spaces, whereas
many of the files you received in e-mail messages were attachments from someone’s file system and
therefore often did. Therefore, this refinement is a very likely one that you’ll have an opportunity to use:
1. Using your favorite text editor again, open scan_pdf.py and change it to look like the follow-
ing (the changed portions are in italics; or, if you skipped the last example, just enter the entire
code as follows):
import os, os.path
import re
def print_pdf (arg, dir, files):
for file in files:
183

23.5 Page 225

▲back to top
Chapter 11
path = os.path.join (dir, file)
path = os.path.normcase (path)
if not re.search (r”.*\\.pdf”, path): continue
if re.search (r” “, path): continue
print path
os.path.walk (‘.’, print_pdf, 0)
2. Now run the modified script — and again, this output will not match yours:
$ python scan_pdf.py
.\\95-04.pdf
.\\101translations\\2003121803\\2003121803.pdf
.\\101translations\\2004101810\\scan.pdf
.\\bluemangos\\smb_pt134.pdf
.\\businessteam.hu\\aok.pdf
.\\businessteam.hu\\chn14300-2.pdf
.\\businessteam.hu\\diplom_bardelmeier.pdf
.\\businessteam.hu\\doktor_bardelmeier.pdf
.\\businessteam.hu\\finanzamt_1.pdf
.\\businessteam.hu\\zollbescheinigung.pdf
.\\businessteam.hu\\monday\\s3.pdf
.\\businessteam.hu\\monday\\s4.pdf
.\\businessteam.hu\\monday\\s5.pdf
.\\gerard\\done\\tg82-20nc-md-04.07.pdf
.\\gerard\\polytronic\\iau-reglement_2005.pdf
.\\glossa\\neumag\\de_993_ba_s5.pdf
How It Works
There’s a stylistic change in this code — one that works well when doing these quick text-processing-
oriented filter scripts. Look at the print_pdf function in the code — first build and normalize the path-
name and then run tests on it to ensure that it’s the one you want. After a test fails, it will use continue
to skip to the next file in the list. This technique enables a whole series of tests to be performed one after
another, while keeping the code easy to read.
Working with Regular Expressions
and the re Module
Perhaps the most powerful tool in the text processing toolbox is the regular expression. While matching
on simple strings or substrings is useful, they’re limited. Regular expressions pack a lot of punch into a
few characters, but they’re so powerful that it really pays to get to know them. The basic regular expres-
sion syntax is used identically in several programming languages, and you can find at least one book
written solely on their use and thousands of pages in other books (like this one).
184

23.6 Page 226

▲back to top
Text Processing
As mentioned previously, a regular expression defines a simple parser that matches strings within a
text. Regular expressions work essentially in the same way as wildcards when you use them to specify
multiple files on a command line, in that the wildcard enables you to define a string that matches many
different possible filenames. In case you didn’t know what they were, characters like * and ? are wild-
cards that, when you use them with commands such as dir on Windows or ls on Unix, will let you
select more than one file, but possiblly fewer files than every file (as does dir win*, which will print
only files in your directory on Windows that start with the letters w, i, and n and are followed by
anything — that’s why the * is called a wildcard). There are two major differences between a regular
expression and a simple wildcard:
❑ A regular expression can match multiple times anywhere in a longer string.
❑ Regular expressions are much, much more complicated and much richer than simple wildcards,
as you will see.
The main thing to note when starting to learn about regular expressions is this: A string always matches
itself. Therefore, for instance, the pattern ‘xxx’ will always match itself in ‘abcxxxabc’. Everything
else is just icing on the cake; the core of what we’re doing is just finding strings in other strings.
You can add special characters to make the patterns match more interesting things. The most commonly
used one is the general wildcard ‘.’ (a period, or dot). The dot matches any one character in a string;
so, for instance, ‘x.x’ will match the strings ‘xxx’ or ‘xyx’ or even ‘x.x’.
The last example raises a fundamental point in dealing with regular expressions. What if you really only
want to find something with a dot in it, like ‘x.x’? Actually, specifying ‘x.x’ as a pattern won’t work;
it will also match ‘x!x’ and ‘xqx’. Instead, regular expressions enable you to escape special characters
by adding a backslash in front of them. Therefore, to match ‘x.x’ and only ‘x.x’, you would use the
pattern ‘x\\.x’, which takes away the special meaning of the period as with an escaped character.
However, here you run into a problem with Python’s normal processing of strings. Python also uses the
backslash for escape sequences, because ‘\\n’ specifies a carriage return and ‘\\t’ is a tab character. To
avoid running afoul of this normal processing, regular expressions are usually specified as raw strings,
which as you’ve seen is a fancy way of saying that you tack an ‘r’ onto the front of the string constant,
and then Python treats them specially.
So after all that verbiage, how do you really match ‘x.x’? Simple: You specify the pattern r”x\\.x”.
Fortunately, if you’ve gotten this far, you’ve already made it through the hardest part of coming to grips
with regular expressions in Python. The rest is easy.
Before you get too far into specifying the many special characters used by regular expressions, first look
at the function used to match strings, and then do some learning by example, by typing a few regular
expressions right into the interpreter.
185

23.7 Page 227

▲back to top
Chapter 11
Try It Out Fun with Regular Expressions
This exercise uses some functional programming tools that you may have seen before but perhaps not
had an opportunity to use yet. The idea is to be able to apply a regular expression to a bunch of different
strings to determine which ones it matches and which ones it doesn’t. To do this in one line of typing,
you can use the filter function, but because filter applies a function of one argument to each mem-
ber of its input list, and re.match and re.search take two arguments, you’re forced to use either a
function definition or an anonymous lambda form (as in this example). Don’t think too hard about it
(you can return to Chapter 9 to see how this works again), as it will be obvious what it’s doing:
1. Start the Python interpreter and import the re module:
$ python
>>> import re
2. Now define a list of interesting-looking strings to filter with various regular expressions:
>>> s = (‘xxx’, ‘abcxxxabc’, ‘xyx’, ‘abc’, ‘x.x’, ‘axa’, ‘axxxxa’, ‘axxya’)
3. Do the simplest of all regular expressions first:
>>> filter ((lambda s: re.match(r”xxx”, s)), s)
(‘xxx’,)
4. Hey, wait! Why didn’t that find ‘axxxxa’, too? Even though you normally talk about matches
inside the string, in Python the re.match function looks for matches only at the start of its
input. To find strings anywhere in the input, use re.search (which spells the word research,
so it’s cooler and easy to remember anyway):
>>> filter ((lambda s: re.search(r”xxx”, s)), s)
(‘xxx’, ‘abcxxxabc’, ‘axxxxa’)
5. OK, look for that period:
>>> filter ((lambda s: re.search(r”x.x”, s)), s)
(‘xxx’, ‘abcxxxabc’, ‘xyx’, ‘x.x’, ‘axxxxa’)
6. Here’s how you match only the period (by escaping the special character):
>>> filter ((lambda s: re.search(r”x\\.x”, s)), s)
(‘x.x’,)
7. You also can search for any number of x’s by using the asterisk, which can match a series of
whatever character is in front of it:
>>> filter ((lambda s: re.search(r”x.*x”, s)), s)
(‘xxx’, ‘abcxxxabc’, ‘xyx’, ‘x.x’, ‘axxxxa’, ‘axxya’)
186

23.8 Page 228

▲back to top
Text Processing
8. Wait a minute! How did ‘x.*x’ match ‘axxya’ if there was nothing between the two x’s? The
secret is that the asterisk is tricky — it matches zero or more occurrences of a character between
two x’s. If you really want to make sure something is between the x’s, use a plus instead, which
matches one or more characters:
>>> filter ((lambda s: re.search(r”x.+x”, s)), s)
(‘xxx’, ‘abcxxxabc’, ‘xyx’, ‘x.x’, ‘axxxxa’)
9. Now you know how to match anything with, say, an ‘c’ in it:
>>> filter ((lambda s: re.search(r”c+”, s)), s)
(‘abcxxxabc’, ‘abc’)
10. Here’s where things get really interesting: How would you match anything without an ‘c’?
Regular expressions use square brackets to denote special sets of characters to match, and if
there’s a caret at the beginning of the list, it means all characters that don’t appear in the set, so
your first idea might be to try this:
>>> filter ((lambda s: re.search(r”[^c]*”, s)), s)
(‘xxx’, ‘abcxxxabc’, ‘xyx’, ‘abc’, ‘x.x’, ‘axa’, ‘axxxxa’, ‘axxya’)
11. That matched the whole list. Why? Because it matches anything that has a character that isn’t an
‘c’, you negated the wrong thing. To make this clearer, you can filter a list with more c’s in it:
>>> filter ((lambda s: re.search(r”[^c]*”, s)), (‘c’, ‘cc’, ‘ccx’))
(‘c’, ‘cc’, ‘ccx’)
Note that older versions of Python may return a different tuple, (‘ccx’,).
12. To really match anything without an ‘c’ in it, you have to use the ^ and $ special characters to
refer to the beginning and end of the string and then tell re that you want strings composed
only of non-c characters from beginning to end:
>>> filter ((lambda s: re.search(r”^[^c]*$”, s)), s)
(‘xxx’, ‘xyx’, ‘x.x’, ‘axa’, ‘axxxxa’, ‘axxya’)
As you can see from the last example, getting re to understand what you mean can sometimes require
a little effort. It’s often best to try out new regular expressions on a bunch of data you understand and
then check the results carefully to ensure that you’re getting what you intended; otherwise, you can get
some real surprises later!
Use the techniques shown here in the following example. You can usually run the Python interpreter in
interactive mode, and test your regular expression with sample data until it matches what you want.
Try It Out Adding Tests
The example scan_pdf.py scripts shown so far provide a nicely formatted framework for testing files.
As mentioned previously, the os.path.walk function provides the heavy lifting. The print_pdf func-
tion you write performs the tests — in this case, looking for PDF files.
187

23.9 Page 229

▲back to top
Chapter 11
Clocking in at less than 20 lines of code, these examples show the true power of Python. Following the
structure of the print_pdf function, you can easily add tests to refine the search, as shown in the fol-
lowing example:
1. Using your favorite text editor again, open scan_pdf.py and change it to look like the follow-
ing. The changed portions are in italics (or, if you skipped the last example, just enter the entire
code that follows):
import os, os.path
import re
def print_pdf (arg, dir, files):
for file in files:
path = os.path.join (dir, file)
path = os.path.normcase (path)
if not re.search (r”.*\\.pdf”, path): continue
if re.search (r”.\\.hu”, path): continue
print path
os.path.walk (‘.’, print_pdf, 0)
2. Now run the modified script — and again, this output will not match yours:
C:\\projects\\translation>python scan_pdf.py
.\\businessteam.hu\\aok.pdf
.\\businessteam.hu\\chn14300-2.pdf
.\\businessteam.hu\\diplom_bardelmeier.pdf
.\\businessteam.hu\\doktor_bardelmeier.pdf
.\\businessteam.hu\\finanzamt_1.pdf
.\\businessteam.hu\\zollbescheinigung.pdf
.\\businessteam.hu\\monday\\s3.pdf
.\\businessteam.hu\\monday\\s4.pdf
.\\businessteam.hu\\monday\\s5.pdf
...
How It Works
This example follows the structure set up in the previous examples and adds another test. You can add
test after test to create the script that best meets your needs.
In this example, the test looks only for filenames (which include the full paths) with a .hu in the name.
The assumption here is that files with a .hu in the name (or in a directory with .hu in the name) are trans-
lations from Hungarian (hu is the two-letter country code for Hungary). Therefore, this example shows
how to narrow the search to files translated from Hungarian. (In real life, you will obviously require dif-
ferent search criteria. Just add the tests you need.)
You can continue refining your script to create a generalized search utility in Python. Chapter 12 goes
into this in more depth.
188

23.10 Page 230

▲back to top
Text Processing
Summary
Text processing scripts are generally short, useful, reusable programs, which are either written for one-
time and occasional use, or used as components of a larger data-processing system. The chief tools for
the text processing programmer are directory structure navigation and regular expressions, both of
which were examined in brief in this chapter.
Python is handy for this style of programming because it offers a balance where it is easy to use for
simple, one-time tasks, and it’s also structured enough to ease the maintenance of code that gets reused
over time.
The specific techniques shown in this chapter include the following:
❑ Use the os.path.walk function to traverse the file system.
❑ Place the search criteria in the function you write and pass it to the os.path.walk function.
❑ Regular expressions work well to perform the tests on each file found by the os.path.walk
function.
❑ Try out regular expressions in the Python interpreter interactively to ensure they work.
Chapter 12 covers an important concept: testing. Testing enables you not only to ensure that your scripts
work but that the scripts still work when you make a change.
Exercises
1. Modify the scan_pdf.py script to start at the root, or topmost, directory. On Windows,
this should be the topmost directory of the current disk (C:, D:, and so on). Doing this on a
network share can be slow, so don’t be surprised if your G: drive takes a lot more time when
it comes from a file server). On Unix and Linux, this should be the topmost directory (the
root directory, /).
2. Modify the scan_pdy.py script to only match PDF files with the text boobah in the filename.
3. Modify the scan_pdf.py script to exclude all files with the text boobah in the filename.
189

24 Pages 231-240

▲back to top

24.1 Page 231

▲back to top

24.2 Page 232

▲back to top
12
Testing
Like visits to the dentist, thorough testing of any program is something that you should be doing
if you want to avoid the pain of having to trace a problem that you thought you’d taken care of.
This lesson is one that normally takes a programmer many years to learn, and to be honest, you’re
still going to be working on it for many years. However, the one thing that is of the utmost impor-
tance is that testing must be organized; and to be the most effective, you must start writing your
programs knowing that it will be tested as you go along, and plan around having the time to write
and confirm your test cases.
Fortunately, Python offers an excellent facility for organizing your testing called PyUnit. It is
a Python port of the Java JUnit package, so if you’ve worked with JUnit, you’re already on
firm ground when testing in Python — but if not, don’t worry. This chapter will show you the
following:
❑ The concept and use of assertions
❑ The basic concepts of unit testing and test suites
❑ A few simple example tests to show you how to organize a test suite
❑ Thorough testing of the search utility from Chapter 11
The beauty of PyUnit is that you can set up testing early in the software development life cycle,
and you can run it as often as needed while you’re working. By doing this, you can catch errors
early on, before they’re painful to rework — let alone before anybody else sees them. You can also
set up test cases before you write code, so that as you write, you can be sure that your results
match what you expect! Define your test cases before you even start coding, and you’ll never find
yourself fixing a bug only to discover that your changes have spiraled out of control and cost you
days of work.
Assertions
An assertion, in Python, is in practice similar to an assertion in day-to-day language. When you
speak and you make an assertion, you have said something that isn’t necessarily proven but that

24.3 Page 233

▲back to top
Chapter 12
you believe to be true. Of course, if you are trying to make a point, and the assertion you made is incor-
rect, then your entire argument falls apart.
In Python, an assertion is a similar concept. Assertions are statements that can be made within the code
while you are developing it that you can use to test the validity of your code, but if the statement doesn’t
turn out to be true, an AssertionError is raised, and the program will be stopped if the error isn’t caught
(in general, they shouldn’t be caught, as AssertionErrors should be taken as a warning that you didn’t
think something through correctly!)
Assertions enable you to think of your code in a series of testable cases. That way, you can make sure
that while you develop, you can make tests along the lines of “this value is not None” or “this object is
a String” or “this number is greater than zero.” All of these statements are useful while developing to
catch errors in terms of how you think about the program.
Try It Out Using Assert
Creating a set of simple cases, you can see how the assert language feature works:
# Demonstrate the use of assert()
large = 1000
string = “This is a string”
float = 1.0
broken_int = “This should have been an int”
assert large > 500
assert type(string) == type(“”)
assert type(float) != type(1)
assert type(broken_int) == type(4)
Try running the preceding with python -i.
How It Works
The output from this simple test case looks like this:
Traceback (most recent call last):
File “D:\\Documents\\ch12\\try_assert.py”, line 13, in ?
assert type(broken_int) == type(4)
AssertionError
You can see from this stack trace that this simply raises the error. assert is implemented very simply. If
a special internal variable called __debug__ is True, assertions are checked; and if any assertion doesn’t
succeed, an AssertionError is raised. Because assert is actually a combination of an if statement that,
when there’s a problem, will raise an exception, you are allowed to specify a custom message, just as you
would with raise, by adding a comma and the message that you’d want to see when you see the error in
a try ... : and except ...: block. You should experiment by replacing the last assertion with this
code and running it:
try:
assert type(broken_int) == type(4), “broken_int is broken”
except AssertionError, message:
print “Handle the error here. The message is: %s” % message
192

24.4 Page 234

▲back to top
Testing
The variable __debug__, which activates assert, is special; it’s immutable after Python has started up,
so in order to turn it off you need to specify the -O (a dash, followed by the capital letter O) parameter to
Python. -O tells Python to optimize code, which among other things for Python means that it removes
assert tests, because it knows that they’ll cause the program to slow down (not a lot, but optimization
like this is concerned with getting every little bit of performance). -O is intended to be used when a pro-
gram is deployed, so it removes assertions that are considered to be development-time features.
As you can see, assertions are useful. If you even think that you may have made a mistake and want to
catch it later in your development cycle, you can put in an assertion to catch yourself, and move on and
get other work done until that code is tested. When your code is tested, it can tell you what’s going
wrong if an assertion fails instead of leaving you to wonder what happened. Moreover, when you
deploy and use the -O flag, your assertion won’t slow down the program.
Assert lacks a couple of things by itself. First, assert doesn’t provide you with a structure in which to
run your tests. You have to create a structure, and that means that until you learn what you want from
tests, you’re liable to make tests that do more to get in your way than confirm that your code is correct.
Second, assertions just stop the program and they provide only an exception. It would be more useful to
have a system that would give you summaries, so you can name your tests, add tests, remove tests, and
compile many tests into a package that let you summarize whether your program tests out or not. These
ideas and more make up the concepts of unit tests and test suites.
Test Cases and Test Suites
Unit testing revolves around the test case, which is the smallest building block of testable code for any
circumstances that you’re testing. When you’re using PyUnit, a test case is a simple object with at least
one test method that runs code; and when it’s done, it then compares the results of the test against vari-
ous assertions that you’ve made about the results.
PyUnit is the name of the package as named by its authors, but the module you
import is called the more generic-sounding name unittest.
Each test case is subclassed from the TestCase class, which is a good, memorable name for it. The sim-
plest test cases you can write just override the runTest method of TestCase and enable you to define a
basic test, but you can also define several different test methods within a single test case class, which can
enable you to define things that are common to a number of tests, such as setup and cleanup procedures.
A series of test cases run together for a particular project is called a test suite. You can find some simple
tools for organizing test suites, but they all share the concept of running a bunch of test cases together
and recording what passed, what failed, and how, so you can know where you stand.
Because the simplest possible test suite consists of exactly one test case, and you’ve already had the sim-
plest possible test case described to you, let’s write a quick testing example so you can see how all this
fits together. In addition, just so you really don’t have anything to distract you, let’s test arithmetic,
which has no external requirements on the system, the file system, or, really, anything.
193

24.5 Page 235

▲back to top
Chapter 12
Try It Out Testing Addition
1. Use your favorite editor to create a file named test1.py in a directory named ch12. Using your
programming editor, edit your file to have the following code:
import unittest
class ArithTest (unittest.TestCase):
def runTest (self):
“”” Test addition and succeed. “””
self.failUnless (1+1==2, ‘one plus one fails!’)
self.failIf (1+1 != 2, ‘one plus one fails again!’)
self.failUnlessEqual (1+1, 2, ‘more trouble with one plus one!’)
def suite():
suite = unittest.TestSuite()
suite.addTest (ArithTest())
return suite
if __name__ == ‘__main__’:
runner = unittest.TextTestRunner()
test_suite = suite()
runner.run (test_suite)
2. Now run the code using python:
.
----------------------------------------------------------------------
Ran 1 tests in 0.000s
OK
How It Works
In step 1, after you’ve imported unittest (the module that contains the PyUnit framework), you define
the class ArithTest, which is a subclass of the class from unittest, TestCase. ArithTest has only
defined the runTest method, which performs the actual testing. Note how the runTest method has its
docstring defined. It is at least as important to document your tests as it is to document your code.
Lastly, a series of three assertions takes place in runTest.
TestCase classes beginning with fail, such as failUnless, failIf, and failUnlessEqual, come
in additional varieties to simplify setting up the conditions for your tests. When you’re programming,
you’ll likely find yourself resistant to writing tests (they can be very distracting; sometimes they are bor-
ing; and they are rarely something other people notice, which makes it harder to motivate yourself to
write them). PyUnit tries to make things as easy as possible for you.
After the unit test is defined in ArithTest, you may like to define the suite itself in a callable function,
as recommended by the PyUnit developer, Steve Purcell, in the modules documentation. This enables
you to simply define what you’re doing (testing) and where (in the function you name). Therefore, after
the definition of ArithTest, you have crated the suite function, which simply instantiates a vanilla,
unmodified test suite. It adds your single unit test to it and returns it. Keep in mind that the suite func-
tion only invokes the TestCase class in order to make an object that can be returned. The actual test is
performed by the returned TestCase object.
194

24.6 Page 236

▲back to top
Testing
As you learned in Chapter 6, only when this is being run as the main program will Python invoke the
TextTestRunner class to create the runner object. The runner object has a method called run that
expects to have an object of the unittests.TestSuite class. The suite function creates one such
object, so test_suite is assigned a reference to the TestSuite object. When that’s finished, the
runner.run method is called, which uses the suite in test_suite to test the unit tests defined in
test_suite.
The actual output in this case is dull, but in that good way you’ll learn to appreciate because it means
everything has succeeded. The single period tells you that it has successfully run one unit test. If, instead
of the period, you see an F, it means that a test has failed. In either case, PyUnit finishes off a run with a
report. Note that arithmetic is run very, very fast.
Now, let’s see what failure looks like.
Try It Out Testing Faulty Addition
1. Use your favorite text editor to add a second set of tests to test1.py. These will be based on
the first example. Add the following to your file:
class ArithTestFail (unittest.TestCase):
def runTest (self):
“”” Test addition and fail. “””
self.failUnless (1+1==2, ‘one plus one fails!’)
self.failIf (1+1 != 2, ‘one plus one fails again!’)
self.failUnlessEqual (1+1, 2, ‘more trouble with one plus one!’)
self.failIfEqual (1+1, 2, ‘expected failure here’)
self.failIfEqual (1+1, 2, ‘second failure’)
def suite_2():
suite = unittest.TestSuite()
suite.addTest (ArithTest())
suite.addTest (ArithTestFail())
return suite
You also need to change the if statement that sets off the tests, and you need to make sure that it
appears at the end of your file so that it can see both classes:
if __name__ == ‘__main__’:
runner = unittest.TextTestRunner()
test_suite = suite_2()
runner.run (test_suite)
2. Now run the newly modified file (after you’ve saved it). You’ll get a very different result with
the second set of tests. In fact, it’ll be very different from the prior test:
.F
======================================================================
FAIL: Test addition and fail.
----------------------------------------------------------------------
Traceback (most recent call last):
File “D:\\Documents\\ch12\\test1.py”, line 27, in runTest
self.failIfEqual (1+1, 2, ‘expected failure here’)
AssertionError: expected failure here
195

24.7 Page 237

▲back to top
Chapter 12
----------------------------------------------------------------------
Ran 2 tests in 0.000s
FAILED (failures=1)
>>>
How It Works
Here, you’ve kept your successful test from the first example and added a second test that you know
will fail. The result is that you now have a period from the first test, followed by an ‘F’ for ‘Failed’ from
the second test, all in the first line of output from the test run.
After the tests are run, the results report is printed out so you can examine exactly what happened. The
successful test still produces no output at all in the report, which makes sense: Imagine you have a hun-
dred tests but only two fail — you would have to slog through a lot more output to find the failures than
you do this way. It may seem like looking on the negative side of things, but you’ll get used to it.
Because there was a failed test, the stack trace from the failed test is displayed. In addition, a couple of
different messages result from the runTest method. The first thing you should look at is the FAIL mes-
sage. It actually uses the docstring from your runTest method and prints it at the top, so you can refer-
ence the test that failed. Therefore, the first lesson to take away from this is that you should document
your tests in the docstring! Second, you’ll notice that the message you specified in the runTest for the
specific test that failed is displayed along with the exception that PyUnit generated.
The report wraps up by listing the number of test cases actually run and a count of the failed test cases.
Test Fixtures
Well, this is all well and good, but real-world tests usually involve some work to set up your tests before
they’re run (creating files, creating an appropriate directory structure, generally making sure everything
is in shape, and other things that may need to be done to ensure that the right things are being tested). In
addition, cleanup also often needs to be done at the end of your tests.
In PyUnit, the environment in which a test case runs is called the test fixture, and the base TestCase
class defines two methods: setUp, which is called before a test is run, and tearDown, which is called
after the test case has completed. These are present to deal with anything involved in creating or clean-
ing up the test fixture.
You should know that if setUp fails, tearDown isn’t called. However, tearDown is
called even if the test case itself fails.
Remember that when you set up tests, the initial state of each test shouldn’t rely on a prior test having
succeeded or failed. Each test case should create a pristine test fixture for itself. If you don’t ensure this,
you’re going to get inconsistent test results that will only make your life more difficult.
To save time when you run similar tests repeatedly on an identically configured test fixture, subclass the
TestCase class to define the setup and cleanup methods. This will give you a single class that you can
196

24.8 Page 238

▲back to top
Testing
use as a starting point. Once you’ve done that, subclass your class to define each test case. You can alter-
natively define several test case methods within your unit case class, and then instantiate test case objects
for each method. Both of these are demonstrated in the next example.
Try It Out Working with Test Fixtures
1. Use your favorite text editor to add a new file test2.py. Make it look like the following exam-
ple. Note that this example builds on the previous examples.
import unittest
class ArithTestSuper (unittest.TestCase):
def setUp (self):
print “Setting up ArithTest cases”
def tearDown (self):
print “Cleaning up ArithTest cases”
class ArithTest (ArithTestSuper):
def runTest (self):
“”” Test addition and succeed. “””
print “Running ArithTest”
self.failUnless (1+1==2, ‘one plus one fails!’)
self.failIf (1+1 != 2, ‘one plus one fails again!’)
self.failUnlessEqual (1+1, 2, ‘more trouble with one plus one!’)
class ArithTestFail (ArithTestSuper):
def runTest (self):
“”” Test addition and fail. “””
print “Running ArithTestFail”
self.failUnless (1+1==2, ‘one plus one fails!’)
self.failIf (1+1 != 2, ‘one plus one fails again!’)
self.failUnlessEqual (1+1, 2, ‘more trouble with one plus one!’)
self.failIfEqual (1+1, 2, ‘expected failure here’)
self.failIfEqual (1+1, 2, ‘second failure’)
class ArithTest2 (unittest.TestCase):
def setUp (self):
print “Setting up ArithTest2 cases”
def tearDown (self):
print “Cleaning up ArithTest2 cases”
def runArithTest (self):
“”” Test addition and succeed, in one class. “””
print “Running ArithTest in ArithTest2”
self.failUnless (1+1==2, ‘one plus one fails!’)
self.failIf (1+1 != 2, ‘one plus one fails again!’)
self.failUnlessEqual (1+1, 2, ‘more trouble with one plus one!’)
def runArithTestFail (self):
“”” Test addition and fail, in one class. “””
print “Running ArithTestFail in ArithTest2”
self.failUnless (1+1==2, ‘one plus one fails!’)
197

24.9 Page 239

▲back to top
Chapter 12
self.failIf (1+1 != 2, ‘one plus one fails again!’)
self.failUnlessEqual (1+1, 2, ‘more trouble with one plus one!’)
self.failIfEqual (1+1, 2, ‘expected failure here’)
self.failIfEqual (1+1, 2, ‘second failure’)
def suite():
suite = unittest.TestSuite()
# First style:
suite.addTest (ArithTest())
suite.addTest (ArithTestFail())
# Second style:
suite.addTest (ArithTest2(“runArithTest”))
suite.addTest (ArithTest2(“runArithTestFail”))
return suite
if __name__ == ‘__main__’:
runner = unittest.TextTestRunner()
test_suite = suite()
runner.run (test_suite)
2. Run the code:
Setting up ArithTest cases
Running ArithTest
Cleaning up ArithTest cases
.Setting up ArithTest cases
Running ArithTestFail
FCleaning up ArithTest cases
Setting up ArithTest2 cases
Running ArithTest in ArithTest2
Cleaning up ArithTest2 cases
.Setting up ArithTest2 cases
Running ArithTestFail in ArithTest2
FCleaning up ArithTest2 cases
======================================================================
FAIL: Test addition and fail.
----------------------------------------------------------------------
Traceback (most recent call last):
File “D:\\Documents\\ch12\\test2.py”, line 25, in runTest
self.failIfEqual (1+1, 2, ‘expected failure here’)
AssertionError: expected failure here
======================================================================
FAIL: Test addition and fail, in one class.
----------------------------------------------------------------------
Traceback (most recent call last):
File “D:\\Documents\\ch12\\test2.py”, line 48, in runArithTestFail
self.failIfEqual (1+1, 2, ‘expected failure here’)
AssertionError: expected failure here
198

24.10 Page 240

▲back to top
Testing
----------------------------------------------------------------------
Ran 4 tests in 0.000s
FAILED (failures=2)
>>>
How It Works
Take a look at this code before moving along. The first thing to note about this is that you’re doing the
same tests as before. One test is made to succeed and the other one is made to fail, but you’re doing two
sets, each of which implements multiple unit test cases with a test fixture, but in two different styles.
Which style you use is completely up to you; it really depends on what you consider readable and
maintainable.
The first set of classes in the code (ArithTestSuper, ArithTest, and ArithTestFail) are essentially
the same tests as shown in the second set of examples in test1.py, but this time a class has been created
called ArithTestSuper. ArithTestSuper implements a setUp and tearDown method. They don’t do
much but they do demonstrate where you’d put in your own conditions. Each of the unit test classes are
subclassed from your new ArithTestSuper class, so now they will perform the same setup of the test
fixture. If you needed to make a change to the test fixture, you can now modify it in ArithTestSuper’s
classes, and have it take effect in all of its subclasses.
The actual test cases, ArithTest and ArithTestFail, are the same as in the previous example, except
that you’ve added print calls to them as well.
The final test case class, ArithTest2, does exactly the same thing as the prior three classes that you’ve
already defined. The only difference is that it combines the test fixture methods with the test case meth-
ods, and it doesn’t override runTest. Instead ArithTest2 defines two test case methods: runArithTest
and runArithTestFail. These are then invoked explicitly when you created test case instances during
the test run, as you can see from the changed definition of suite.
Once this is actually run, you can see one change immediately: Because our setup, test, and cleanup
functions all write to stdout, you can see the order in which everything is called. Note that the cleanup
functions are indeed called even after a failed test. Finally, note that the tracebacks for the failed tests
have been gathered up and displayed together at the end of the report.
Putting It All Together with
Extreme Programming
A good way to see how all of this fits together is to use a test suite during the development of an
extended coding project. This strategy underlies the XP (Extreme Programming) methodology, which is
a popular trend in programming: First, you plan the code; then you write the test cases as a framework;
and only then do you write the actual code. Whenever you finish a coding task, you rerun the test suite
to see how closely you approach the design goals as embodied in the test suite. (Of course, you are also
debugging the test suite at the same time, and that’s fine!) This technique is a great way to find your pro-
gramming errors early in the process, so that bugs in low-level code can be fixed and the code made sta-
ble before you even start on higher-level work, and it’s extremely easy to set up in Python using PyUnit,
as you will see in the next example.
199

25 Pages 241-250

▲back to top

25.1 Page 241

▲back to top
Chapter 12
This example includes a realistic use of text fixtures as well, creating a test directory with a few files in it
and then cleaning up the test directory after the test case is finished. It also demonstrates the convention
of naming all test case methods with test followed by the name, such as testMyFunction, to enable
the unittest.main procedure to recognize and run them automatically.
Implementing a Search Utility in Python
The first step in this programming methodology, as with any, is to define your objectives — in this case,
a general-purpose, reusable search function that you can use in your own work. Obviously, it would be
a waste of time to anticipate all possible text-processing functionality in a single search utility program,
but certain search tasks tend to recur a lot. Therefore, if you wanted to implement a general-purpose
search utility, how would you go about it? The Unix find command is a good place to look for useful
functionality — it enables you not only to iterate through the directory tree and perform actions on each
file found but also to specify certain directories to skip, to specify rather complex logic combinations on
the command line, and a number of other things, such as searching by file modification date and size.
On the other hand, the find command doesn’t include any searching on the content of files (the standard
way to do this under Unix is to call grep from within find) and it has a lot of features involving the invo-
cation of post-processing programs that we don’t really need for a general-purpose Python search utility.
What you might need when searching for files in Python could include the following:
❑ Return values you can use easily in Python: A tuple including the full path, the filename, the
extension, and the size of the file is a good start.
❑ Specification of a regular expression for the filename to search for and a regular expression for
the content (if no content search is specified, then the files shouldn’t be opened, to save overhead).
❑ Optional specifications of additional search terms: The size of the file, its age, last modification,
and so on are all useful.
A truly general search utility might include a function to be called with the parameters of the file, so that
more advanced logic can be specified. The Unix find command enables very general logic combinations
on the command line, but frankly, let’s face it — complex logic on the command line is hard to under-
stand. This is the kind of thing that really works better in a real programming language like Python, so
you could include an optional logic function for narrowing searches as well.
In general, it’s a good idea to approach this kind of task by focusing first on the core functionality,
adding more capability after the initial code is already in good shape. That’s how the following example
is structured — first you start with a basic search framework that encapsulates the functionality you cov-
ered in the examples for the os and re modules, and then you add more functionality once that first part
is complete. This kind of incremental approach to software development can help keep you from getting
bogged down in details before you have anything at all to work with, and the functionality of something
like this general-purpose utility is complicated enough that it would be easy to lose the thread.
Because this is an illustration of the XP methodology as well, you’ll follow that methodology and first
write the code to call the find utility, build that code into a test suite, and only then will you write the
find utility. Here, of course, you’re cheating a little. Ordinarily, you would be changing the test suite as
you go, but in this case, the test suite is already guaranteed to work with the final version of the tested
code. Nonetheless, you can use this example for yourself.
200

25.2 Page 242

▲back to top
Testing
Try It Out Writing a Test Suite First
1. Use your favorite text editor to create the file test_find.py. Enter the following code:
import unittest
import find
import os, os.path
def filename(ret):
return ret[1]
class FindTest (unittest.TestCase):
def setUp (self):
os.mkdir (“_test”)
os.mkdir (os.path.join(“_test”, “subdir”))
f = open (os.path.join(“_test”, “file1.txt”), “w”)
f.write (“””first line
second line
third line
fourth line”””)
f.close()
f = open (os.path.join(“_test”, “file2.py”), “w”)
f.write (“””This is a test file.
It has many words in it.
This is the final line.”””)
f.close()
def tearDown (self):
os.unlink (os.path.join (“_test”, “file1.txt”))
os.unlink (os.path.join (“_test”, “file2.py”))
os.rmdir (os.path.join (“_test”, “subdir”))
os.rmdir (“_test”)
def test_01_SearchAll (self):
“”” 1: Test searching for all files. “””
res = find.find (r”.*”, start=”_test”)
self.failUnless (map(filename,res) == [‘file1.txt’, ‘file2.py’],
‘wrong results’)
def test_02_SearchFileName (self):
“”” 2: Test searching for specific file by regexp. “””
res = find.find (r”file”, start=”_test”)
self.failUnless (map(filename,res) == [‘file1.txt’, ‘file2.py’],
‘wrong results’)
res = find.find (r”py$”, start=”_test”)
self.failUnless (map(filename,res) == [‘file2.py’],
‘Python file search incorrect’)
def test_03_SearchByContent (self):
“”” 3: Test searching by content. “””
res = find.find (start=”_test”, content=”first”)
self.failUnless (map(filename,res) == [‘file1.txt’],
“didn’t find file1.txt”)
201

25.3 Page 243

▲back to top
Chapter 12
res = find.find (where=”py$”, start=”_test”, content=”line”)
self.failUnless (map(filename,res) == [‘file2.py’],
“didn’t find file2.py”)
res = find.find (where=”py$”, start=”_test”, content=”second”)
self.failUnless (len(res) == 0,
“found something that didn’t exist”)
def test_04_SearchByExtension (self):
“”” 4: Test searching by file extension. “””
res = find.find (start=”_test”, ext=’py’)
self.failUnless (map(filename,res) == [‘file2.py’],
“didn’t find file2.py”)
res = find.find (start=”_test”, ext=’txt’)
self.failUnless (map(filename,res) == [‘file1.txt’],
“didn’t find file1.txt”)
def test_05_SearchByLogic (self):
“”” 5: Test searching using a logical combination callback. “””
res = find.find (start=”_test”, logic=lambda (x): (x[‘size’] < 50))
self.failUnless (map(filename,res) == [‘file1.txt’],
“failed to find by size”)
if __name__ == ‘__main__’:
unittest.main()
2. Now create another code file named find.py — note that this is only the skeleton of the actual
find utility and will fail miserably. That’s okay; in testing and in extreme programming, failure
is good because it tells you what you still need to do:
import os, os.path
import re
from stat import *
def find (where=’.*’, content=None, start=’.’, ext=None, logic=None):
return ([])
3. Run the test_find.py test suite from the command line. An excerpt is shown here:
C:\\projects\\articles\\python_book\\ch12_testing>python test_find.py
FFFFF
======================================================================
FAIL: 1: Test searching for all files.
----------------------------------------------------------------------
[a lot more information]
Ran 5 tests in 0.421s
FAILED (failures=5)
How It Works
The first three lines of the testing suite import the PyUnit module, the find module to be tested (which
hasn’t actually been written yet), and the os and os.path modules for file and directory manipulation
202

25.4 Page 244

▲back to top
Testing
when setting up and tearing down the test fixtures. Following this, there’s a simple helper function to
extract the filename from the search results, to make it simpler to check the results for correctness.
After that, the test suite itself starts. All test cases in this example are instances of the base class FindTest.
The FindTest class starts out with setUp and tearDown methods to define the test fixtures used in the
test cases, followed by five test cases.
The test fixture in all test cases consists of a testing directory; a subdirectory under that main directory
to ensure that subdirectories aren’t treated as files when scanning; and two test files with .txt and .py
extensions. The contents of the test files are pretty arbitrary, but they contain different words so that the
test suite can include tests to distinguish between them using a content search.
The test cases themselves are named with both a sequential number and a descriptive name, and each
starts with the characters “test”. This allows the unittest.main function to autodetect them when run-
ning the test suite. The sequential numbers ensure that the tests will be run in the proper order defined,
as a simple character sort is used to order them when testing. Each docstring then cites the test number,
followed by a simple description of the type of test. All of this enables the results of failed tests to be
understood quickly and easily, so that you can trace exactly where the error occurred.
Finally, after the test cases are defined, there are exactly two lines of code to detect that the script is being
run directly instead of being called as a module, and if it is being run, to create a default test runner
using unittest.main in that case. The unittest.main call then finds all of the test cases, sorts them
by the sequential number, and runs them in order.
The second file is the skeleton of the find utility itself. Beyond determining what it has to do and how
it’s called, you haven’t done anything at all yet to write the code itself, so that’s your next task.
Try It Out A General-Purpose Search Framework
1. Using your favorite text editor, open find.py and change it to look like this:
import os, os.path
import re
from stat import *
def find (where=’.*’, content=None, start=’.’, ext=None, logic=None):
context = {}
context[‘where’] = where
context[‘content’] = content
context[‘return’] = []
os.path.walk (start, find_file, context)
return context[‘return’]
def find_file (context, dir, files):
for file in files:
# Find out things about this file.
path = os.path.join (dir, file)
path = os.path.normcase (path)
try:
ext = os.path.splitext (file)[1][1:]
203

25.5 Page 245

▲back to top
Chapter 12
except:
ext = ‘’
stat = os.stat(path)
size = stat[ST_SIZE]
# Don’t treat directories like files
if S_ISDIR(stat[ST_MODE]): continue
# Do filtration based on the original parameters of find()
if not re.search (context[‘where’], file): continue
# Do content filtration last, to avoid it as much as possible
if context[‘content’]:
f = open (path, ‘r’)
match = 0
for l in f.readlines():
if re.search(context[‘content’], l):
match = 1
break
f.close()
if not match: continue
# Build the return value for any files that passed the filtration tests.
file_return = (path, file, ext, size)
context[‘return’].append (file_return)
2. Now, for example, to find Python files containing “find,” you can start Python and do the
following:
>>> import find
>>> find.find(r”py$”, content=’find’)
[(‘.\\\\find.py’, ‘find.py’, ‘py’, 1297), (‘.\\\\test_find.py’, ‘test_find.py’, ‘py’,
1696)]
How It Works
This example is really doing the same thing as the first example in the last chapter on text processing,
except that instead of a task-specific print_pdf function, there is a more general find_file function
to scan the files in each directory. Because this code is more complex than the other example scripts, you
can see that having a testing framework available in advance will help you immensely in debugging the
initial versions. This first version satisfies the first three test cases of the test suite.
Because the find_file function is doing most of the filtration work, it obviously needs access to the
search parameters. In addition, because it also needs a place to keep the list of hits it is building during
the search, a dictionary structure is a good choice for its argument, as a dictionary is mutable and can
contain any number of named values. Therefore, the first thing the main find function does is to build
that dictionary and put the search parameters into it. It then calls os.path.walk to do the work of iter-
ating through the directory structure, just as in the PDF search code example at the beginning of this
chapter. Once the walk is done, it returns the return value (the list of files found and information about
them), which was built during the search.
During the search, os.path.walk calls find_file on each directory it finds, passing the dictionary
argument built at the start of the search, the name of the current directory, and a list of all the files in the
204

25.6 Page 246

▲back to top
Testing
directory. The first thing the find_file function does, then, is to scan that list of files and determine
some basic information for each one by running os.stat on it. If the “file” is actually a subdirectory, the
function moves on; because all of the search parameters apply to filenames, not to points in the directory
tree (and because the content search will result in an error unless a file is being opened!), the function
skips the subdirectories using the information gleaned from the os.stat call.
When that’s finished, the function applies the search parameters stored in the dictionary argument to
eliminate whatever files it can. If a content parameter is specified, it opens and reads each file, but other-
wise no manipulation of the file itself is done.
If a file has passed all the search parameter tests (there are only two in this initial version), an entry is built
for it and appended to the hit list; this entry consists of the full pathname of the file relative to the starting
point of the search, the filename itself, its extension, and its size. Naturally, you could return any set of
values for files you find useful, but these are a good basic set that you could use to build a directory-like
listing of hits, or use to perform some sort of task on the files.
A More Powerful Python Search
Remember that this is an illustration of an incremental programming approach, so the first example was
a good place to stop and give an explanation, but there are plenty of other search parameters it would be
nice to include in this general search utility, and of course there are still two unit cases to go in the test
suite you wrote at the outset. Because Python gives you a keyword parameter mechanism, it’s very sim-
ple to add new named parameters to your function definition and toss them into the search context dic-
tionary, and then use them in find_file as needed, without making individual calls to the find
function unwieldy.
The next example shows you how easy it is to add a search parameter for the file’s extension, and
throws in a logic combination callback just for good measure. You can add more search parameters at
your leisure; the following code just shows you how to get started on your own extensions (one of the
exercises for the chapter asks you to add search parameters for the date on which the file was last modi-
fied, for instance).
While the file extension parameter, as a single simple value, is easy to conceive and implement — it’s
really just a matter of adding the parameter to the search context and adding a filter test in find_file —
planning a logic combination callback parameter requires a little thought. The usual strategy for specifi-
cation of a callback is to define a set of parameters — say, the filename, size, and modification date — and
then pass those values in on each call to the callback. If you add a new search parameter, you’re faced with
a choice — you can arbitrarily specify that the new parameter can’t be included in logical combinations,
you can change the callback specification and invalidate all existing callbacks for use with the new code,
or you can define multiple categories of logic callbacks, each with a different set of parameters. None of
these alternatives is terribly satisfying, and yet they’re decisions that have to be made all the time.
In Python, however, the dictionary structure provides you with a convenient way to circumvent this
problem. If you define a dictionary parameter that passes named values for use in logic combinations,
then unused parameters are simply ignored. Thus, older callbacks can still be used with newer code that
defines more search parameters, without any changes to code you’ve already got being necessary. In the
updated search code below, the callback function is defined to be a function that takes a dictionary and
returns a flag — a true filter function. You can see how it’s used in the example section and in the next
chapter, in test case 5 in the search test suite.
205

25.7 Page 247

▲back to top
Chapter 12
Adding a logical combination callback also makes it simple to work with numerical parameters such
as the file size or the modification date. It’s unlikely that a caller will search on the exact size of a file;
instead, one usually searches for files larger or smaller than a given value, or in a given size range — in
other words, most searches on numerical values are already logical combinations. Therefore, the logical
combination callback should also get the size and dates for the file, so that a filter function can already
be written to search on them. Fortunately, this is simple — the results of os.stat are already available
to copy into the dictionary.
Try It Out Extending the Search Framework
1. Again using your favorite text editor, open the file find.py from the last example and add the
lines in italics:
import os, os.path
import re
from stat import *
def find (where=’.*’, content=None, start=’.’, ext=None, logic=None):
context = {}
context[‘where’] = where
context[‘content’] = content
context[‘return’] = []
context[‘ext’] = ext
context[‘logic’] = logic
os.path.walk (start, find_file, context)
return context[‘return’]
def find_file (context, dir, files):
for file in files:
# Find out things about this file.
path = os.path.join (dir, file)
path = os.path.normcase (path)
try:
ext = os.path.splitext (file)[1][1:]
except:
ext = ‘’
stat = os.stat(path)
size = stat[ST_SIZE]
# Don’t treat directories like files
if S_ISDIR(stat[ST_MODE]): continue
# Do filtration based on the original parameters of find()
if not re.search (context[‘where’], file): continue
if context[‘ext’]:
if ext != context[‘ext’]: continue
if context[‘logic’]:
arg = {}
arg[‘path’] = path
arg[‘ext’] = ext
arg[‘stat’] = stat
arg[‘size’] = size
arg[‘mod’] = stat[ST_MTIME]
206

25.8 Page 248

▲back to top
Testing
if not context[‘logic’](arg): continue
# Do content filtration last, to avoid it as much as possible
if context[‘content’]:
f = open (path, ‘r’)
match = 0
for l in f.readlines():
if re.search(context[‘content’], l):
match = 1
break
f.close()
if not match: continue
# Build the return value for any files that passed the filtration tests.
file_return = (path, file, ext, size)
context[‘return’].append (file_return)
2. Now to find files larger than 1,000 bytes and older than yesterday:
>>> import find
>>> find.find(r”py$”, content=’find’)
[(‘.\\\\find.py’, ‘find.py’, ‘py’, 1297), (‘.\\\\test_find.py’, ‘test_find.py’, ‘py’,
1696)]
3. You can also run the test_find.py test suite from the command line:
C:\\projects\\python_book\\ch11_regexp>python test_find.py
.....
----------------------------------------------------------------------
Ran 5 tests in 0.370s
OK
(During development, this run was not quite so smooth!)
Formal Testing in the Software Life Cycle
The result of the test suite shown above is clean and stable code in a somewhat involved programming
example, and well-defined test cases that are documented as working correctly. This is a quick and easy
process in the case of a software “product” that is some 30 lines long, although it can be astounding how
many programming errors can be made in only 30 lines!
In a real-life software life cycle, of course, you will have thousands of lines of code. In projects of realistic
magnitude like this, nobody can hope to define all possible test cases before releasing the code. It’s true
that formal testing during the development phase will dramatically improve both your code and your
confidence in it, but there will still be errors in it when it goes out the door.
During the maintenance phase of the software life cycle, bug reports are filed after the target code is placed
in production. If you’re taking an integrated testing approach to your development process, then you can
see that it’s logical to think of bug reports as highlighting errors in your test cases as well as errors in the code
itself. Therefore, the first thing you should do with a bug report is to use it to modify an existing test case,
or to define a new test case from scratch, and only then should you start to modify the target code itself.
207

25.9 Page 249

▲back to top
Chapter 12
By doing this, you accomplish several things. First, you’re giving the reported bugs a formal definition.
This enables you to agree with other people regarding what bugs are actually being fixed, and it enables
further discussion to take place as to whether the bugs have really been understood correctly. Second,
by defining test fixtures and test cases, you are ensuring that the bugs can be duplicated at will. As I’m
sure you know if you’ve ever need to reproduce elusive bugs, this alone can save you a lot of lost sleep.
Finally, the third result of this approach might be the most significant: If you never make a change to
code that isn’t covered by a test case, you will always know that later changes aren’t going to break fixes
already in place. The result is happier users and a more relaxed you. And you’ll owe it all to unit testing.
Summary
Testing is a discipline best addressed at the very outset of the development life cycle. In general, you will
know that you’ve got a firm grip on the problem you’re solving when you understand it enough to write
tests for it.
The most basic kind of test is an assertion. Assertions are conditions that you’ve placed inside of your
program confirming that conditions that should exist do in fact exist. They are for use while you’re
developing a program to ensure that conditions you expect are met.
Assertions will be turned off if Python is run with the -O option. The -O indicates that you want Python
to run in a higher performance mode, which would usually also be the normal way to run a program in
production. This means that using assert is not something that you should rely on to catch errors in a
running system.
PyUnit is the default way of doing comprehensive testing in Python, and it makes it very easy to man-
age the testing process. PyUnit is implemented in the unittest module.
When you use PyUnit to create your own tests, PyUnit provides you with functions and methods to test
for specific conditions based on questions such as “is value A greater than value B,” giving you a num-
ber of methods in the TestCase class that fail when the conditions reflected by their names fail. The
names of these methods all begin with “fail” and can be used to set up most of the conditions for which
you will ever need to test.
The TestCase class should be subclassed — it’s the run method that is called on to run the tests, and
this method needs to be customized to your tests. In addition, the test fixture, or the environment in
which the tests should be run, can be set up before each test if the TestCase’s setUp and tearDown
methods are overridden, and code is specified for them.
You’ve seen two approaches to setting up a test framework for yourself. One subclasses a customized
class, and another uses separate functions to implement the same features but without the need to sub-
class. You should use both and find out which ones work for your way of doing things. These tests do
not have to live in the same file as your modules or programs; they should be kept separate so they don’t
bloat your code.
As you go through the remainder of this book, try to think about writing tests for the functions and
classes that you see, and perhaps write tests as you go along. It’s good exercise; better than having exer-
cises here.
208

25.10 Page 250

▲back to top
13
Writing a GUI with Python
Python plays a large role behind the scenes in some of the world’s largest and most important
server-side applications, but Python has also made a big impact on end-user applications. Writing
a GUI is an expensive and painful project in C, C++, or even Java or C#, but it can be done quickly
and easily in Python. Even if you only write simple Python scripts, being able to whip up a GUI
can be a force multiplier that makes your script usable by less technical people, compounding its
value. Python, being cross-platform and truly object oriented, has advantages that Visual Basic
programmers would love to have in their rapid application development toolbox.
Python enables you to lay out GUIs one component at a time, like other programming languages.
However, these days, no real programmer is writing GUI code by hand. If that’s what you’re used
to, get ready to embrace all the rapid application development magic of Delphi with the power of
a real language in Python. Of course, this kind of power is also available in other stacks, such as
C#; and Microsoft’s next-generation Avalon programming toolkit draws heavily on these concepts
(although they’d never admit it).
GUI Programming Toolkits for Python
There is wide support for writing GUIs with Python with many different toolkits: You can find a
dozen options at www.python.org/moin/GuiProgramming to try out. These toolkits, binary
modules for Python that interface with native GUI code written in C/C++, all have different API’s
and offer different feature sets. Only one comes with Python by default, the venerable TK GUI
toolkit. TK, while always available, offers only a basic set of features, and is fairly useless in any
real sense. It’s always possible that if you’re just using Windows, you’ll install win32all and use
the Win32 API directly. The truly brave will write their entire GUI in pyGame and add sound to
every slider.
The real options are wxPython, pyQT, and pyGTK. These differ in many ways, but one important
way is the license. The pyQT web page shows this problem of how it could restrict the decisions
you can make if you are trying to create certain classes of applications or libraries. You can see this
in the following paragraph:

26 Pages 251-260

▲back to top

26.1 Page 251

▲back to top
Chapter 13
“PyQt is licensed under the GNU GPL (for UNIX/Linux and MacOS/X), under the Qt Non-commercial
License (for use with the Qt v2.3.0 non-commercial version for windows), under the Qt Educational
License (for use with the educational edition of Qt for Windows), and under a commercial license
(for Windows, UNIX/Linux and MacOS/X). . . .”
They go on to state:
“When deploying commercial PyQt applications it is necessary to discourage users from accessing the
underlying PyQt modules for themselves. A user that used the modules shipped with your application
to develop new applications would themselves be considered a developer and would need their own com-
mercial Qt and PyQt licenses.”
“One solution to this problem is the VendorID (www.riverbankcomputing.co.uk/vendorid/)
package. This enables you to build Python extension modules that can only be imported by a digitally
signed custom interpreter. The package enables you to create such an interpreter with your application
embedded within it. The result is an interpreter that can only run your application, and PyQt modules
that can only be imported by that interpreter. You can use the package to similarly restrict access to any
extension module.”
As you can see, unless there is a very good reason, you’ll probably want to skip the whole QT toolset for
this section of the license alone. No one in their right mind wants to deal with that kind of confusing
licensing landscape. The QT people would claim that the advantages of their toolkit overwhelm the cost
of licensing for the few people who use Windows. If you agree, tread warily into their licensing mine-
field. Most people simply discount it.
One open-source option is wxPython. WxPython is based on wxWidgets, a portable (Windows, Linux,
Mac OS X) graphics toolkit with a long history and a tradition of looking and running just like native
code. You can find the best information on wxPython on the really nice wiki at
http://wiki.wxpython.org/index.cgi/FrontPage.
Beginners to GUI creation may feel overwhelmed by wxPython. Although there is good user support in
mailing lists and professional organizations, the wxPython library is intimidating. Nevertheless, it’s a
good option for people willing to climb the learning curve.
For the rest of us, there’s pyGTK. Based on the same core libraries the Gnome wizards put together to
develop their desktop (and the Graphic design program “The Gimp”), pyGTK is licensed under the
LGPL for all of the platforms it supports. Currently, it supports Windows, Linux, and Mac OS X (under
X11). The core feature pyGTK offers over its competition is the integration of Glade and libglade into
the GUI design process. Glade is a RAD tool that enables users to quickly create a GUI design. This
design is then saved as an XML document, which is loaded by the application at runtime using libglade.
PyGTK fully supports this method of operation, and even improves on the C implementation of it by
enabling you to use introspection and exceptions to their full extent. That said, pyGTK does have some
limitations, and users of pyGTK often find that keeping up with the development pace of GTK and
pyGTK can be dizzying.
PyGTK Introduction
GUIs are not as simple as they look. Once you’ve understood the basic concepts, however, you’ll find
them understandable, and proper program design will help you navigate around the major roadblocks.
210

26.2 Page 252

▲back to top
Writing a GUI with Python
The author’s experience with GUI toolkits, and with pyGTK specifically, stems from developing
Immunity CANVAS, a cross-platform commercial product written completely in Python. Note that the
same techniques described here are the basis for the new large projects being written by the Ximian team
(now part of Novell) as they build the next-generation SuSe desktop application suite.
Of course, not all pyGTK applications have to be complex. Your application may be a simple dialog box
that you’ve written to automate a business process you often do. The same things that made large appli-
cations like CANVAS, Dashboard, and PythonCAD quick and easy to write make simple applications
nearly trivial.
pyGTK Resources
You’ll first need to make sure you have pyGTK installed. If you did a complete install on a modern
Linux distribution, you’ll have pyGTK 2.0 installed already. If you’re running Windows, you can install
the latest pyGTK with two clicks.
The latest Win32 installations of pyGTK are available at www.pcpm.ucl.ac.be/~gustin/win32_
ports/pygtk.html.
If you don’t have pyGTK installed on your Linux system, you’ll likely find that the platform-specific
packaging commands will quickly produce them for you. For gentoo, use “emerge pyGTK”. On debian
or Red Hat installations with apt, invoking “apt-get pygtk-devel” will remedy the situation. Even if you
do have it installed, it doesn’t hurt to make sure that it’s the latest pyGTK package your distribution
offers. See Appendix B and the web site for more information on installing pyGTK.
After you have pyGTK installed, you can make sure it works by importing the pygtk module:
>>> import pygtk
>>> pygtk.require(“2.0”)
>>> import gtk
A more reliable method of importing pyGTK follows. This code is more complex but also more portable
across the different versions of pyGTK that exist in the wild. Put it into a file called findgtk.py and you
can just import findgtk to ensure that Python loads the right version of pyGTK, and that import gtk
will work anytime afterwards. findgtk.py is used by all the examples in this chapter.
#!/usr/bin/env python
“””
findgtk.py - Find the pyGTK libraries, wherever they are.
“””
import os
import sys
sys.path.append(“/usr/local/lib/python2.3/site-packages/”)
def try_import():
import sys
“””tries to import gtk and if successful, returns 1”””
#print “Attempting to load gtk...Path=%s”%sys.path
# To require 2.0
211

26.3 Page 253

▲back to top
Chapter 13
try:
import pygtk
pygtk.require(“2.0”)
except:
print “pyGTK not found. You need GTK 2 to run this.”
print “Did you \\”export PYTHONPATH=/usr/local/lib/python2.2/site-
packages/\\” first?”
print “Perhaps you have GTK2 but not pyGTK, so I will continue to try
loading.”
try:
import gtk,gtk.glade
import atk,pango #for py2exe
import gobject
except:
import traceback,sys
traceback.print_exc(file=sys.stdout)
print “I’m sorry, you apparently do not have GTK2 installed - I tried”
print “to import gtk, gtk.glade, and gobject, and I failed.”
return 0
return 1
if not try_import():
site_packages=0
#for k in sys.path:
# if k.count(“site-packages”):
#
print “existing site-packages path %s found\\n”%k
#
site_packages=1
if site_packages == 0:
from stat import *
#print “no site-packages path set, checking.\\n”
check_lib = [ “/usr/lib/python2.2/site-packages/”,
“/usr/local/lib/python2.2/site-packages/”,
“/usr/local/lib/python2.3/site-packages/” ]
for k in check_lib:
try:
path=os.path.join(k,”pygtk.py”)
#print “Path=%s”%path
if open(path)!=None:
#print “appending”, k
sys.path=[k]+sys.path
if try_import():
break
except:
pass
if not try_import():
sys.exit(0)
212

26.4 Page 254

▲back to top
Writing a GUI with Python
pyGTK Resources
The pyGTK FAQ is really more of a Wiki. This has everything you need to know and is
actively maintained. Often, when people post questions to the pyGTK mailing list, the
maintainers simply reply with a FAQ number and URL:
www.async.com.br/faq/pygtk/index.py?req=index
The pyGTK mailing list is actively used. You’ll find the authors of both pyGTK and this
chapter on pyGTK on this list, actively helping newcomers:
www.daa.com.au/mailman/listinfo/pygtk
This list of tutorials can be handy for beginners. Some are unfinished, but they all pre-
sent useful information:
www.pygtk.org/articles.html
Creating GUI Widgets with pyGTK
The first thing to understand is that most GUI frameworks, including pyGTK, are based on a widget
model. A widget is a component of a GUI — buttons, labels, and text boxes are all widgets. Most widgets
have graphical representations on screen, but some widgets, such as tables and boxes, exist only to con-
tain other widgets and arrange them on the screen. A GUI is constructed out of an arrangement of wid-
gets. In the following section, you’ll create a simple GUI by defining some widgets and placing them
inside each other.
Try It Out Writing a Simple pyGTK Program
With pyGTK in place, you’re ready to write a real GUI application. This script, SingleButtonGUI, creates
a GUI of two widgets: a window, which contains a button. The label of the button displays a message:
#!/usr/bin/env python
import findgtk
import gtk
class SingleButtonGUI:
def __init__(self, msg=”Hello World”):
“Set up the window and the button within.”
self.window=gtk.Window()
self.button=gtk.Button(msg)
self.window.add(self.button)
#Show the GUI
self.button.show()
self.window.show()
if __name__ == ‘__main__’:
SingleButtonGUI()
gtk.main()
Run this program and you’ll see the Hello World button in the window, as shown in Figure 13-1.
213

26.5 Page 255

▲back to top
Chapter 13
Figure 13-1
If you’re running Windows, you can use Cygwin’s bash to execute this script, but don’t use Cygwin’s
Python; it doesn’t come linked with pyGTK. Try this instead:
$ /cygdrive/c/Python24/python.exe SingleButtonGUI.py
How It Works
The first thing to do is to create pyGTK objects for each widget. Then, the child widget (the button) is
associated with its parent (the window). Finally, both widgets are displayed. It’s important to call the
show method on every widget in your GUI. In this example, if you call show on the button but not the
window, the window will show up but nothing will be inside it. If you call show on the window but
not the button, nothing will show up on the screen at all.
One problem with this script is that you can’t kill this window by clicking the Close window in the GUI.
You’ll need to press Ctrl+C (that is, the control key and the c key, together) in the script terminal to close
it, or otherwise kill the Python process. Another problem with this script is that unlike most buttons in
GUI applications, the button here doesn’t actually do anything when you click it. Both problems share a
cause: The script as it is doesn’t handle any GUI events.
GUI Signals
GUI programs aren’t just about putting widgets up on the screen. You also need to be able to respond
to the user’s actions. GUIs generally handle this with the notion of events, or (in pyGTK terminology)
signals.
Each GUI widget can generate a number of different signals in response to user actions: For instance, a
button may be clicked, or a window destroyed. In pyGTK, these would correspond to signals named
clicked and destroy. The other half of GUI programming is setting up handlers for GUI signals:
pieces of code that are triggered each time a corresponding signal is sent by the framework.
If no piece of code is listening for a signal, nothing happens when the user triggers the signal. That’s
why in the previous example you couldn’t close the window through the GUI, and why nothing hap-
pened when you clicked the button. Signals could have been spawned, but they wouldn’t have gone
anywhere.
In pyGTK, you register a function with a signal handler by calling the connect method on the widget
whose signals you want to capture. Pass in the name of the signal you want to receive and the function
you want to be called every time that widget emits that signal.
The following script, ClickCountGUI.py, presents a similar interface to the previous example. The dif-
ference is that this GUI application responds to some signals. You can see ClickCountGUI.py working
in Figure 13-2.
214

26.6 Page 256

▲back to top
Writing a GUI with Python
#!/usr/bin/env python
import findgtk
import gtk
class ClickCountGUI:
“When you click, it increments the label.”
CLICK_COUNT = ‘Click count: %d’
def __init__(self):
“Set up the window and the button within.”
self.window=gtk.Window()
self.button=gtk.Button(self.CLICK_COUNT % 0)
self.button.timesClicked = 0
self.window.add(self.button)
#Call the buttonClicked method when the button is clicked.
self.button.connect(“clicked”, self.buttonClicked)
#Quit the program when the window is destroyed.
self.window.connect(“destroy”, self.destroy)
#Show the GUI
self.button.show()
self.window.show()
def buttonClicked(self, button):
“This button was clicked; increment the message on its label.”
button.timesClicked += 1
button.set_label(self.CLICK_COUNT % button.timesClicked)
def destroy(self, window):
“Remove the window and quit the program.”
window.hide()
gtk.main_quit()
if __name__ == ‘__main__’:
ClickCountGUI()
gtk.main()
Figure 13-2
This GUI responds to the destroy signal of the window object, which means you can close the window
through the GUI. It also responds to the clicked signal of the button object, so the button can change to
display the number of times you’ve clicked it.
215

26.7 Page 257

▲back to top
Chapter 13
GUI Helper Threads and the GUI Event Queue
One common problem GUIs must deal with is handling long-running events, such as data reads from
the network. It doesn’t take much time to change the label on a button, so our click-counting program is
safe. However, what if clicking a button started a process that took a minute to finish? A script like the
one shown in the previous example would freeze the GUI until the process finished. There would be no
processor time allocated to sending out the GUI signals triggered by the user. To the end user, it would
look like your application had frozen.
Even worse, what if clicking a button started a process that would stop only in response to another GUI
action? For example, consider a stopwatch-like application in which clicking a button starts a counter,
and clicking the button again stops it. It wouldn’t do to write code that started counting after receiving
the first signal and stopped counting after receiving a second signal. Once you clicked the button, you’d
never be able to click it again; the program would be busy doing the count, not listening for signals. Any
GUI program that performs a potentially long-running task needs to delegate that task to a separate
thread for the duration. A GUI is always doing two things: It’s doing whatever job is specified for that
particular program, and it’s constantly gathering signals from the user.
With pyGTK, you can run code in other threads without disrupting the GUI, so long as each thread calls
the gtk module’s threads_enter function before calling any pyGTK code, and calls threads_leave
afterwards. Make one mistake, though, and your application will truly freeze. That’s why it’s better to
keep all the pyGTK code in the main thread, and have other threads request changes to the GUI by
putting them into a GUI event queue.
Note that pyGTK under Linux is pretty forgiving of threading mistakes. Nonetheless, having to debug a
random freeze in your application that happens only after running it for several hours can make for a
frustrating week. Getting threading right is difficult in any GUI framework, and the concepts listed
below are applicable to C programming as well as Python programming.
Let’s start with some basics. The problem of cross-platform threading under pyGTK is complicated by
some architectural difficulties on Windows. But if you keep to the strict design decisions outlined below,
you’ll have no problems on any platform. A bonus payoff is that your program will become more orga-
nized in general, and you won’t have to learn all the intricacies of managing threads yourself.
1. Your *GUI.py is the only Python file allowed to call GTK functions.
2. Only one thread is allowed to run in *GUI.py.
3. The thread in *GUI.py will read from and clear a GUI queue object; other threads will add
actions to the queue object.
4. For any operation that might take a long time to complete, your GUI will start another worker
thread. This especially includes network calls.
The term *GUI.py means that once you’ve decided on a name for your program, you’ll create
nameGUI.py so that you know it will be the file that follows these rules.
216

26.8 Page 258

▲back to top
Writing a GUI with Python
This simple design will prevent you from eons of nearly impossible debugging problems as your project
gets more complicated. The following library module (placed in gui_queue.py) will accomplish this for
you. There are several ways to do this sort of queue, but this is the only way that I can absolutely guar-
antee works:
This module requires the timeoutsocket module: www.steffensiebert.de/soft/python/
timeoutsocket.py. See Appendix B for details.
#!/usr/bin/env python
“””
gui_queue.py
This Python modules does what we need to do to avoid threading issues on both Linux
and Windows.
Your other modules can include this file and use it without knowing anything about
gtk.
“””
#Python License for Beginner’s Python book
import findgtk
import gtk
import random
import socket
import time
from threading import RLock
import timeoutsocket #used for set_timeout()
class gui_queue:
“””wakes up the gui thread which then clears our queue”””
def __init__(self,gui,listenport=0):
“””If listenport is 0, we create a random port to listen on”””
self.mylock=RLock()
self.myqueue=[]
if listenport==0:
self.listenport=random.randint(1025,10000)
else:
self.listenport=listenport
print “Local GUI Queue listening on port %s”%self.listenport
s=socket.socket(socket.AF_INET, socket.SOCK_STREAM)
s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
s.bind((“”, self.listenport))
self.listensocket=s
self.listensocket.listen(300) #listen for activity.
#time.sleep(15)
self.gui=gui
return
217

26.9 Page 259

▲back to top
Chapter 13
Above, we use initialize the self.mylock with the Rlock function which we will use to create a
“mutex” to ensure that certain parts of the code are only run by one thread at a time. (This is what mutex
means: mutually exclusive. If one thread is holding on to the mutex, that excludes the other threads from
doing the same action). The code listens for GUI events on a network socket (see Chapter 16 for more
information on sockets and ports). If no listening port is specified, this code will choose a random high
port on which to listen. Other threads will add an item to the GUI queue by connecting to that socket
over the operating system’s local network interface:
def append(self,command,args):
“””
Append can be called by any thread
“””
#print “about to acquire...”
self.mylock.acquire()
self.myqueue.append((command,args))
#this won’t work on a host with a ZoneAlarm firewall
#or no internet connectivity...
s=socket.socket(socket.AF_INET, socket.SOCK_STREAM)
#small timeout will wake up the gui thread, but not
#cause painful pauses if we are already in the gui thread.
#important to note that we use timeoutsocket and it
#is already loaded.
s.set_timeout(0.01)
#wakey wakey!
#print “Connecting to port %d”%self.listenport
try:
s=s.connect((“localhost”,self.listenport))
except:
#ignore timeouts
pass
#print “About to release”
self.mylock.release()
return
def clearqueue(self, socket, x):
“””
Clearqueue is only called by the main GUI thread
Don’t forget to return 1
“””
#print “Clearing queue”
#clear this...TODO: add select call here.
newconn,addr=self.listensocket.accept()
for i in self.myqueue:
(command,args)=i
self.gui.handle_gui_queue(command,args)
self.myqueue=[]
return 1
The preceding code’s clearqueue function will be called periodically by the main GUI thread, which
will then get each of the gui_queue’s new commands sent to the GUI’s handle_gui_queue function
in turn.
218

26.10 Page 260

▲back to top
Writing a GUI with Python
Your GUI application will need to set up a GUI queue, and have its signal hook methods append items
to the GUI queue instead of handling the signals directly. Here’s a class you can subclass that sets up a
GUI queue and provides a method for appending to it, and handling what comes out of it. Note that the
code to connect the queue to the network differs between versions of pyGTK.
class Queued:
def __init__(self):
self.gui_queue=gui_queue(self) #our new gui queue
#for older pyGTK:
#gtk.input_add(self.gui_queue.listensocket,
#
gtk.gdk.INPUT_READ, self.gui_queue.clearqueue)
#
#for newer pyGTK (2.6):
import gobject
gobject.io_add_watch(self.gui_queue.listensocket, gobject.IO_IN,
self.gui_queue.clearqueue)
def handle_gui_queue(self, command, args):
“””
Callback the gui_queue uses whenever it receives a command for us.
command is a string
args is a list of arguments for the command
“””
gtk.threads_enter()
#print “handle_gui_queue”
method = getattr(self, command, None)
if method:
apply(method, args)
else:
print “Did not recognize action to take %s: %s”%(command,args)
#print “Done handling gui queue”
gtk.threads_leave()
return 1
def gui_queue_append(self,command,args):
self.gui_queue.append(command,args)
return 1
Try It Out Writing a Multithreaded pyGTK App
Here’s an application, CountUpGUI.py, that implements the stopwatch idea mentioned earlier. It uses
a separate thread to count off the seconds, a thread that modifies the GUI by putting items on the
gui_queue for the main thread to process:
#!/usr/bin/env python
import time
from threading import Thread
import findgtk
import gtk
from gui_queue import Queued
219

27 Pages 261-270

▲back to top

27.1 Page 261

▲back to top
Chapter 13
class CountUpGUI(Queued):
“””Does counting in a separate thread. To be safe, the other
thread puts calls to threads_enter() and threads_leave() around
all GTK code.”””
START = “Click me to start counting up.”
STOP = “I’ve counted to %s (click me to stop).”
def __init__(self):
Queued.__init__(self)
self.window=gtk.Window()
self.button=gtk.Button(self.START)
self.button.timesClicked = 0
self.window.add(self.button)
self.thread = None
#Call the toggleCount method when the button is clicked.
self.button.connect(“clicked”, self.toggleCount)
#Quit the program when the window is destroyed.
self.window.connect(“destroy”, self.destroy)
#Show the GUI
self.button.show()
self.window.show()
def destroy(self, window):
“Remove the window and quit the program.”
window.hide()
gtk.main_quit()
def toggleCount(self, button):
if self.thread and self.thread.doCount:
#Stop counting.
self.thread.doCount = False
else:
#Start counting.
self.thread = self.CountingThread(self, self.button)
self.thread.start()
def incrementCount(self, button, count):
button.set_label(self.STOP % count)
def resetCount(self, button):
button.set_label(self.START)
class CountingThread(Thread):
“””Increments a counter once per second and updates the button
label accordingly. Updates the button label by putting an
event on the GUI queue, rather than manipulating the GUI
directly.”””
def __init__(self, gui, button):
self.gui = gui
220

27.2 Page 262

▲back to top
Writing a GUI with Python
Thread.__init__(self)
self.button = button
self.doCount = False
self.count = 0
self.setDaemon(True)
def run(self):
self.doCount = True
while self.doCount:
self.gui.gui_queue_append(“incrementCount”,
[self.button, self.count])
self.count += 1
time.sleep(1)
self.gui.gui_queue_append(“resetCount”, [self.button])
self.count = 0
if __name__ == ‘__main__’:
CountUpGUI()
try:
gtk.threads_init()
except:
print “No threading was enabled when you compiled pyGTK!”
import sys
sys.exit(1)
gtk.threads_enter()
gtk.main()
gtk.threads_leave()
You can see how this looks in Figure 13-3.
Figure 13-3
How It Works
When you click the button the first time, it initializes a CountingThread object. This thread object
spends most of its time sleeping, but it wakes up every second to update the label with a new number.
If it were updating the label directly, then to avoid freezing the program, it would have to know to call
gtk.threads_enter before calling incrementCount, and to call gtk.threads_leave afterward.
Instead, it puts an incrementCount command onto the gui_queue object. The main thread (which
called gtk.threads_enter before entering the main body of the code) retrieves this command from the
queue and executes it. The other thread can change the GUI without having to know any of the details of
pyGTK or its thread handling.
221

27.3 Page 263

▲back to top
Chapter 13
Widget Packing
So far, all of our examples have explored GUI concepts with a GUI consisting only of a single button.
Needless to say, most real GUI applications are more complicated. You might be tempted to create a GUI
with multiple widgets by simply creating the widgets and attaching them to the window:
#This is bad code! Don’t actually try it!
button1=gtk.Button(“Button 1”)
window.add(button1)
button2=gtk.Button(“Button 2”)
window.add(button2)
If you try this code, you’ll notice that only the first button shows up. This is because a window can only
contain one widget. Once you associate the first button with the window, you can’t associate anything
else with it. How, then, are complex GUIs possible? The answer is a technique called widget packing.
Widget packing makes use of boxes and tables, virtual widgets that don’t necessarily show up on the
screen the way a button does. A window can only have one child widget, but if that widget happens to
be a box or table, it can contain a number of child widgets, and display them all, either beside or on top
of each other. As you’ll see, you can put boxes inside boxes to get the exact layout you need.
Here’s TwoButtonsGUI.py, a script that is like our original “Hello World” application
SingleButtonGUI.py (in which clicking the button does nothing), but this time there are two buttons
instead of one:
#!/usr/bin/env python
import findgtk
import gtk
class TwoButtonsGUI:
def __init__(self, msg1=”Hello World”, msg2=”Hello Again”):
#Set up the window and the button within.
self.window=gtk.Window()
self.box = gtk.VBox()
self.window.add(self.box)
The window widget only has space for one child widget: If we put one of our buttons directly in the
window, there wouldn’t be anywhere to put the other one. Instead, we put a box widget in the window.
This widget doesn’t show anything onscreen, but it can contain more than one child widget. We use a
VBox, which means that widgets will be packed vertically into the box, one on top of the other. The alter-
native is an HBox, which packs widgets next to each other horizontally:
self.button1 = gtk.Button(msg1)
self.button2 = gtk.Button(msg2)
self.box.pack_start(self.button1)
self.box.pack_start(self.button2)
222

27.4 Page 264

▲back to top
Writing a GUI with Python
We create our two buttons and put each one in the box. Next we must show all four widgets: the two
buttons, the box, and the window. Remember that if you don’t show a widget, neither it nor any of its
children appear on the screen:
#Show the GUI
self.button1.show()
self.button2.show()
self.box.show()
self.window.show()
if __name__ == ‘__main__’:
TwoButtonsGUI()
gtk.main()
See Figure 13-4 for an example of what this code will look like when its run.
Figure 13-4
As you can see, adding just one more GUI widget greatly increased the amount of design and program-
ming work we had to do. For complex layouts, it gets even worse. Writing GUIs by hand in this day and
age is insane. Fortunately, we don’t have to: The best way to lay out a GUI is graphically. This is as far as
we’re going to go with hand-coded GUIs. From this point on, we’ll use a GUI builder tool called Glade.
Glade: a GUI Builder for pyGTK
The great thing about pyGTK is that you almost never have to write the GUI by hand. That’s what Glade
is for. Glade is a GUI construction kit: It provides a GUI you can use to design your own GUI. Once
you’re done, it writes a description of a GUI layout to an XML file (see Chapter 15 for more information
about XML). A library called libglade can then read that XML file and render the corresponding GUI.
Instead of instantiating a bunch of Python objects and calling show on all of them, you just feed a file
describing your GUI into libglade.
To give you some idea of the sorts of complex applications you can build with Glade, Figure 13-5 shows
a screenshot of CANVAS 5.4 running on Windows XP. It runs identically on Linux — and, of course, inte-
grates with the standard GTK themes.
Here you can see many of the widgets you’ll shortly know how to use: a notebook container for the log
and debug information panes, several list and tree views, a horizontal scale for the covertness bar, text
entries, a menu bar, and a button with an icon in it.
Given a little practice, you too could lay out this complex application in Glade in just a few moments. Of
course, the real payoff is when you want to move parts of your application around. Rather than regener-
ate code, you simply change the pieces you want changed in Glade, and click Save. No code changes
need to be made at all.
223

27.5 Page 265

▲back to top
Chapter 13
Figure 13-5
GUI Builders for Other GUI Frameworks
Writing GUI code is painful — no matter which toolkit you use, you’re going to want to use some sort of
generator for as much of it as possible. All the non-TK toolkits include a construction kit. If you’re using
wxPython to run your GUI, you should consider using Boa Constructor, wxDesigner, or wxGlade to
build it. QT has the KDE builder tools to work with, and of course pyGTK has Glade.
You’ll want to avoid any GUI builder that actually generates GUI code (for instance, anything that gen-
erates all those Python objects and calls to show). A good code generator will go to an intermediate lan-
guage, such as XML, which your graphics toolkit will load and parse at runtime. By this token, wxGlade
would be preferable to Boa Constructor. wxGlade generates XRC: the wxWindows XML language.
224

27.6 Page 266

▲back to top
Writing a GUI with Python
For programmers used to HyperCard, wxPython also offers PythonCard. You can obtain more informa-
tion from the developer’s SourceForge page at http://pythoncard.sourceforge.net/.
Using libGlade with Python
Libglade is a library that reads in an XML file and makes the corresponding calls to GTK to create a GUI.
Glade (actually Glade-2.exe or glade2) will present you with a GUI for creating these XML files.
There are many advantages for doing your GUI in this way:
❑ You can change your GUI quickly and easily.
❑ You can have multiple GUIs that drive the same underlying application.
❑ You spend your time designing the GUI, and not debugging the GUI creation code.
As Immunity developed CANVAS, we also strove as much as possible to isolate the code from the GUI
altogether. Although we liked the pyGTK model, there was a distinct possibility that someday we would
want to port to a platform that GTK did not support, such as the Sharp Zaurus. Good application design
specifies that you actually build another layer in between your GUI and your application code, such that
you have three layers:
❑ The XML file that describes your GUI.
❑ An application file (called something like mygui.py), which loads the GUI and application, and
other application files as needed for major GUI components to allow for testing them indepen-
dently of the rest of the application. These are the only files that use GTK functionality directly.
❑ The application logic itself, which should never import GTK or any GUI toolkit library. All calls
to the GUI should be through the code in mygui.py.
Using this design will save you years of time later trying to debug some small threading error. All of
Immunity CANVAS (not a trivial application) was ported from GTK version 1 (and pyGTK version 1) to
GTK version 2 (and corresponding pyGTK) within one day. Because GTK development is proceeding
quite quickly, this sort of capability is going to be key to maintaining compatibility with the library itself.
A Glade Walkthrough
If you’ve never used a GUI builder, then you can’t fully appreciate how easy Glade has made the whole
process. If you have, you’ll find that Glade offers the same kind of features you’re used to.
Installing Glade
If you’re using Windows, you can download a port of Glade from http://gladewin32
.sourceforge.net/. The package systems of Unix-like operating systems (including
Mac OS X) usually make a glade, glade2, or python-glade2 package available. See the
web site for the book for more information.
225

27.7 Page 267

▲back to top
Chapter 13
Starting Glade
Start glade by running glade-2.exe, or on Unix-like systems (including Mac OS X), a simple glade or
glade-2 will do. Glade starts up with three windows: a project window as in Figure 13-6, a palette of
widgets you can use to build your GUI as in Figure 13-7, and a property sheet displaying information on
the currently selected GUI widget as in Figure 13-8. Because you have no GUI widgets yet, the property
sheet is blank.
Figure 13-6
226
Figure 13-7

27.8 Page 268

▲back to top
Writing a GUI with Python
Figure 13-8
Creating a Project
First, start a new Glade project called GladeTwoButtonsGUI.
Glade might ask you some questions at this point, but your answers don’t matter much. Glade might
offer you options for two types of projects: GTK or Gnome projects. A Gnome project would use features
unique to the Gnome desktop environment, which is usually available on Linux, but is not a good idea
for cross-platform projects. You want something portable, so choose a GTK project instead. If Glade asks
you to pick a language (for instance, C, C++, or Ada), choose any of them; it doesn’t matter. You’re not
generating the GUI code from Glade, You’re going to be using only the XML file that Glade generates to
describe the GUI layout.
Save your project now, and you can start creating a GUI with the palette.
Using the Palette to Create a Window
The Glade widget palette is one of the most important tools you’ll be using. You can see an image of the
palette in Figure 13-9. Each icon on the palette corresponds to a type of widget or widget container. To
create a widget, you click its icon and then the container in which you want to place that widget.
Of course, you don’t start out with anywhere to put any of the widgets. Let’s change that by creating a
new window. Click the top-left icon on the palette (it’s called Window and it looks like a little empty
window) to create a root window for your application.
227

27.9 Page 269

▲back to top
Chapter 13
Figure 13-9
The Window and Dialog widgets are top-level widgets: They have their own GUI windows and they
don’t need to go into a widget container. Every other widget in your application will have a Window
widget or a Dialog widget as its ultimate parent.
This window starts out with the name of “window1,” but you can change this from its property sheet.
You’ll find that Glade selects names for each widget based on a simple incrementing number plan. The
first text view widget you create is called “textview1”, the first window is “window1”, and so on. If you
hover over an icon on the palette, or click to select it, Glade will tell you what kind of widget that icon
represents.
Putting Widgets into the Window
The window you just created provides a container in which you can place a widget (or another con-
tainer). Let’s use this to recreate the two-button GUI from the earlier example.
Recall that a window can only contain one child widget. Before you can place any buttons, you need to
fill the window with a box: a virtual widget that can contain multiple child widgets. Click the Vertical Box
icon on the palette and then click on the window to place a vertical box in the window. You’ll be asked
how many rows you want in the box. Because you’re going to place two buttons, enter 2. Now you have
a window that contains a vertical box (see Figure 13-10), which itself can contain up to two widgets.
The presence of the vertical box is denoted graphically by a white line partitioning the window into
two parts. When the GUI is actually run, though, all you’ll see are the widgets inside the vertical box.
Remember that virtual widgets such as boxes don’t show up in the GUI; they just determine how the
other widgets appear.
Let’s put buttons in the box. Click the Button icon on the palette and then click on the top partition of the
vertical box. Repeat this to put another button in the bottom partition of the box. Resize the window if
necessary, and you should have something that looks almost like our other two-button example (see
Figure 13-11).
228

27.10 Page 270

▲back to top
Writing a GUI with Python
Figure 13-10
Figure 13-11
Use the properties sheet for each button to change its label, and the illusion will be complete (see
Figure 13-12). If you can’t find the window with the properties sheet, select View ➪ Show Property
Editor in the main Glade window to make it show up.
Figure 13-12
229

28 Pages 271-280

▲back to top

28.1 Page 271

▲back to top
Chapter 13
By setting these properties, you’ll get a window with two buttons that have changed according to the
text you’ve entered (see Figure 13-13).
Figure 13-13
Glade Creates an XML Representation of the GUI
It should already be clear how powerful Glade is. With it, you can construct a GUI visually instead
of by writing code. But how do you get this GUI into a representation that Python can understand?
Save your Glade project, and then look in the project’s directory. You should find a
GladeTwoButtonsGUI.glade file that contains an XML representation of the GUI you just created.
That XML representation will look something like this (although a lot of it has been edited out for
clarity):
<?xml version=”1.0” standalone=”no”?>
<!DOCTYPE glade-interface SYSTEM “http://glade.gnome.org/glade-2.0.dtd”>
<glade-interface>
<widget class=”GtkWindow” id=”window1”>
<child>
<widget class=”GtkVBox” id=”vbox1”>
<child>
<widget class=”GtkButton” id=”button1”>
<property name=”label” translatable=”yes”>Hello World</property>
</widget>
</child>
<child>
<widget class=”GtkButton” id=”button2”>
<property name=”label” translatable=”yes”>Hello Again</property>
</widget>
</child>
</widget>
</child>
</widget>
</glade-interface>
If this looks like gibberish to you, consult Chapter 15 for more information on XML. If you can read
XML, notice that this data structure defines a tree of tags that corresponds to the tree structure of
the GUI. The interface as a whole contains a window (GtkWindow), which contains a vertical box
(GtkVBox), which contains two buttons (GtkButton). The buttons have customized label properties,
just as you defined them in Glade.
230

28.2 Page 272

▲back to top
Writing a GUI with Python
In short, this XML file contains the same information as the GUI we defined visually, and the same infor-
mation as the several lines of Python we used to define the same GUI in TwoButtonsGUI.py. If there
were a way to get Python to parse this file and create a GUI out of it, we could save a significant amount
of code. This is where libglade comes in.
Try It Out Building a GUI from a Glade File
libglade parses the XML file and makes GTK widgets corresponding to the widgets described in the
XML file. Here’s GladeTwoButtonsGUI.py, a version of TwoButtonsGUI.py that loads its GUI from
the XML file instead of using a series of Python statements:
#!/usr/bin/env python
import findgtk
import gtk.glade
class TwoButtonsGUI:
def __init__(self):
self.window = gtk.glade.XML(‘GladeTwoButtonsGUI.glade’, ‘window1’)
if __name__ == ‘__main__’:
TwoButtonsGUI()
gtk.main()
How It Works
This program uses libglade to load a set of GUI widgets from the GladeTwoButtonsGUI.glade
file we created with Glade. The GUI looks just the same as the Glade mock-up, and the same as we
created with pyGTK calls in the TwoButtonsGUI.py program. The advantage over the original
TwoButtonsGUI.py is that we had to write a lot less code to get the same GUI.
Glade greatly simplifies the layout of even small GUIs. As you’ll see, it also provides a framework for
designating which events a GUI is expected to handle.
You may need to install the Python libglade bindings separately from Glade. If all else fails, you can
download the bindings as part of the pygtk package, at http://ftp.gnome.org/pub/GNOME/
sources/pygtk/. See Appendix B for more information.
Creating a Real Glade Application
Of course, the Glade version of our two-button application doesn’t do anything, any more than the ver-
sion that just used Python code did. In this section, we’ll create a complex GUI, with some signal han-
dlers, for an application called pyRAP. This is a chat-themed GUI that could be used as a client for the
Python Chat Server described in Chapter 16.
Create a new Glade project called PyRAP, and create a new window as shown previously. To create a
basic GUI, start with a Vertical Box widget, also shown previously. Click the Vertical Box in the widget
palette, and then click the crosshatch marks in the new window to place it there. When Glade asks you
231

28.3 Page 273

▲back to top
Chapter 13
how many rows you want in your Vertical Box, enter 3 (as opposed to the two-row box created in the
previous example).
Put a Menu Bar widget in the top row, and a Status Bar widget in the bottom row. You should now have
a GUI that looks a lot like the application interface most people have come to expect, with an empty
space in the middle (see Figure 13-14).
Figure 13-14
That empty middle container (note the cross-hatching) is where we’ll put the guts of our application. For
starters, we’ll just have pyRAP take the contents of an Input widget and write it to another widget. To do
this, we’ll split our central container into two portions with a two-column Horizontal Box, as shown in
Figure 13-15.
232
Figure 13-15

28.4 Page 274

▲back to top
Writing a GUI with Python
Now we’ve got a window that is divided into three portions by a vertical box. The middle portion of
the Vertical Box is itself divided in two by a Horizontal Box. Let’s go one step further and use another
Vertical Box to divide the left portion of the Horizontal Box into three sections, as shown in Figure 13-16.
Figure 13-16
That’s enough layout widgets. Now it’s time to place some real widgets. In the Vertical Box you just cre-
ated, put a Label in the top slot, a Text Entry in the middle slot, and a Button in the bottom slot. Your
GUI should now look like the window shown in Figure 13-17.
Figure 13-17
Note that the label and button widgets appear with some rather bland default text. In a little bit, you’ll
change that text using the properties sheets for those widgets. Right now, though, let’s fill up the only
remaining slot in your GUI with a Text View widget, as shown in Figure 13-18.
233

28.5 Page 275

▲back to top
Chapter 13
Figure 13-18
Let’s change that default text. Select the label and the button in turn, and use the property sheet to set
their Label properties (see Figure 13-19). As you change the default text, you’ll see the text in the mock-
up change as well.
Figure 13-19
Now your mock-up should look like the GUI for a real application (see Figure 13-20).
What you’re seeing is a Label, a Text Entry, and a Button on the left side, and a Text View on the right
side. GTK supports most of the widgets you can expect from any windowing interface — Combo-Boxes,
Spin Buttons for numeric input, and so on. The only difficult part of GTK is understanding and using
Trees and Lists and properly designing your application to handle threads. Now you’ve reached the fun
part: deciding what to do with this application.
234

28.6 Page 276

▲back to top
Writing a GUI with Python
Figure 13-20
Now it’s time to learn how to connect the application to some Python code. Save your Glade project and
let’s start writing PyRAP.py, the code for the application that uses it:
#!/usr/bin/env python
import time
import findgtk
import gtk.glade
class PyRAPGUI:
def __init__(self):
self.wTree = gtk.glade.XML(“PyRAP.glade”, “window1”)
if __name__ == ‘__main__’:
PyRAPGUI()
try:
gtk.threads_init()
except:
print “No threading was enabled when you compiled pyGTK!”
import sys
sys.exit(1)
gtk.threads_enter()
gtk.main()
gtk.threads_leave()
This code is just a skeleton, and it has the same problem as the earlier Glade example. You can enter text
into the Text Enter widget, but clicking the button doesn’t do anything. You need to set up a signal so the
program does something when the button is clicked.
Go back into Glade, and select the Send button in your mock-up. Select the Properties View. Add a sig-
nal that’s activated when the Send button is clicked by clicking first on the Signals tab, and then on the
ellipses (...) button next to the Signal: label (see Figure 13-21). Select the clicked signal and then click
Add. When the GUI gets a click on the button, it’ll generate a signal on_button1_clicked for pyGTK
to process.
235

28.7 Page 277

▲back to top
Chapter 13
Figure 13-21
Click the window1 object in the main screen of Glade to bring focus on the main widget. Next, go to the
window’s properties sheet. Carry out the same process as before to add an on_window1_destroy signal
for the window widget.
Now let’s redo PyRAP.py to respond to those signals. When you kill the window, the program will exit,
as in the previous examples. When you click the Send button, PyRAP will copy to the Text View widget
on the right anything you typed into the Text Entry widget on the left:
#!/usr/bin/env python
import findgtk
import gtk
import time
class PyRAPGUI:
def __init__(self):
self.wTree = gtk.glade.XML (“PyRAP.glade”, “window1”)
dic={ “on_window1_destroy” : self.quit,
“on_button1_clicked” : self.send,
}
self.wTree.signal_autoconnect (dic)
self.username=”Bob”
#setup the text view to act as a log window
self.logwindowview=self.wTree.get_widget(“textview1”)
self.logwindow=gtk.TextBuffer(None)
self.logwindowview.set_buffer(self.logwindow)
return
#Handlers for the GUI signals
def quit(self,obj):
“Handles the ‘destroy’ signal of the window.”
gtk.main_quit()
sys.exit(1)
236

28.8 Page 278

▲back to top
Writing a GUI with Python
def send(self,obj):
“Handles the ‘clicked’ signal of the button.”
message=self.wTree.get_widget(“entry1”).get_text()
print “Message=%s” % message
self.log(self.username + “: “ + message, “black”)
def log(self,message,color,enter=”\\n”):
“””
A helper method for the “send” GUI signal handler:
logs a message to the log window and scrolls the window to the bottom
“””
message=message+enter
buffer = self.logwindow
iter = buffer.get_end_iter()
#gtk versioning avoidance
if color != “black”:
tag = buffer.create_tag()
tag.set_property(“foreground”, color)
self.logwindow.insert_with_tags(buffer.get_end_iter(), message, tag)
else:
self.logwindow.insert(iter, message)
#gtk.FALSE and gtk.TRUE on older pyGTK
mark = buffer.create_mark(“end”, buffer.get_end_iter(), False)
self.logwindowview.scroll_to_mark(mark,0.05,True,0.0,1.0)
#print “Exited log function”
if __name__ == ‘__main__’:
PyRAPGUI()
try:
gtk.threads_init()
except:
print “No threading was enabled when you compiled pyGTK!”
import sys
sys.exit(1)
gtk.threads_enter()
gtk.main()
gtk.threads_leave()
First, we initialize the Text View widget to contain a text buffer. Then we must handle writing into the
text buffer and scrolling it down so it always displays the latest message. As a bonus, we also put in
some code to display the text in different colors, if desired. We’ll probably use that later. As you can see,
with a few widgets and two signals, we’ve created the bare bones of a working GUI for a text messenger
system (see Figure 13-22).
Figure 13-22
237

28.9 Page 279

▲back to top
Chapter 13
The exciting thing about Glade is that you can go from concept to working demo in a couple of hours.
As you revise your program, the GUI can morph completely, without ever affecting your code. Of
course, pyRAP is lacking networking code in this example, but that could be fleshed out with either
socket calls (to connect to IRC, an instant messaging system, or Chapter 16’s Python Chat server) or a
nice XML-RPC client (see Chapter 21).
Advanced Widgets
Not all widgets are as easy to use as simple Entry Boxes or Spin Buttons. As you’ve seen with the text
view in the previous demo, some widgets require initialization to use. A large part of this initialization
requirement is that the widgets themselves are portals into the data set, not the data set itself. For exam-
ple, you might have multiple text views, each representing different parts of the same text buffer.
Classical GUI design text refers to this as the model, view, controller design. It is hoped that you’ll need
to know as little about that as possible.
Now you’re going to add a tree view to the left side of your application, which will contain a server list.
The easiest way to do this is to use the Widget Tree (another window you can activate from Glade’s View
menu). Select the first object under the hbox1, as shown in Figure 13-23, and insert a new container
before it. This will add another column to the horizontal box, to the left of the label and button.
Figure 13-23
The Widget Tree is extremely useful for advanced Glade users. As you modify your application, you can
cut and paste various widgets from one container to another, or simply add new containers where you
want to, without having to dramatically redo the GUI mock-up. Unfortunately, there is currently no
Undo feature in Glade so save when you’re happy with the current state, and then experiment.
Add a Tree View widget to the newly created slot to the left of the other widgets. Your Glade mock-up
will now look like the window shown in Figure 13-24.
238

28.10 Page 280

▲back to top
Writing a GUI with Python
Figure 13-24
The new GUI you can see in figure 13-25 will work just as well with your existing PyRAP.py code, but
the Tree View widget won’t do anything, because there’s no initialization code for it and we haven’t set
up any signals for it.
Figure 13-25
Tree Views display data in a columnar format, but as you can see, no columns will show up in your
application yet; you need to manually set the column headers. In addition, the figure shows that the
author has changed the main window’s title to be “pyRAP,” in anticipation of adding network support
and some application logic to enable two applications to communicate with a central server and have a
“rap battle” with each other.
To fill in your Tree View, you’ll have to initialize it, much as you did with your text view earlier.
Normally, it is wise to split this process off into another file, but in this case, you’ll keep it all together.
The following code first creates a model variable that contains a TreeStore model. The model variable
knows it is going to take two strings as columns. The insert_row function (further below) is a wrapper
to the model.insert_after and model.set_value functions. This code is largely cut-and-paste when
designing your own projects.
An important concept in this code, and in the gtk API in general (and many other APIs), is the concept of
an iterator. An iterator is simply an object that holds a position within a list or other data structure. In
this case, the insert_row function is returning an iterator that holds the position within the tree model
into which a row was inserted. Later, we can pass that iterator back into insert_row to insert a row
239

29 Pages 281-290

▲back to top

29.1 Page 281

▲back to top
Chapter 13
under the “host” line. The following code fragment also sets the TreeView widget to use the new model
we created with the API call set_model. Also notice that we’re grabbing the treeview1 widget from
wherever it happens to be in the Glade-created GUI. If we move treeview1, this code does not have to
change.
Put this code at the end of the __init__ method of your PyRAP class:
#initialize our host tree
self.hosttree=self.wTree.get_widget(“treeview1”)
model=gtk.TreeStore(gobject.TYPE_STRING, gobject.TYPE_STRING)
self.hostsmodel=model
host_inter=insert_row(model,None,’www.immunitysec.com’, ‘Default Main
Server’)
self.hosttree.set_model(model)
In the following code you define a renderer, which you will use to display a line of text inside one of
the columns of the TreeView. You then append that column to the TreeView widget. The text=0 is not,
as it appears, a Boolean value but rather the index into the model from which the text of the column
should come. In this case, insert_row is going to put the hostname (in this case ‘www.immunitysec.
com’) as the first value in a row in the model:
renderer=gtk.CellRendererText()
column=gtk.TreeViewColumn(“Host/Channel”,renderer, text=0)
column.set_resizable(True)
self.hosttree.append_column(column)
You do this again for a second column, giving it an index into the model of 1 (the second value in the
model’s row):
renderer=gtk.CellRendererText()
column=gtk.TreeViewColumn(“Users”,renderer, text=1)
column.set_resizable(True)
self.hosttree.append_column(column)
And, of course, you’ll need to add the insert_row method:
def insert_row(self,model,parent,firstcolumn,secondcolumn, thirdcolumn=None):
myiter=model.insert_after(parent,None)
model.set_value(myiter,0,firstcolumn)
model.set_value(myiter,1,secondcolumn)
if thirdcolumn != None:
model.set_value(myiter,2,thirdcolumn)
return myiter
When all of this is inserted into the __init__ function of your PyRAP class, and you run your applica-
tion, you should see your column headers and some initial information in your tree view, as shown in
Figure 13-26.
240

29.2 Page 282

▲back to top
Writing a GUI with Python
Figure 13-26
Currently, there will only be one pyRAP server, so you can leave that tree view at that. If you want, you
can add signal handlers that respond to button presses to generate drop-down menus or perform other
actions based on which line in the host tree the user has selected. The following sample code is included
to demonstrate how this can be done.
In this sample code, you add a button_press signal to the GUI in Glade and in your code’s signal con-
nection dictionary:
#add this code to your signal dictionary in PyRAP:__init__ to
#capture treeview1’s button presses
“on_treeview1_button_press_event”: self.moduletree_press,
#The handler for the button press looks like this.
#You can place this directly in your PyRAP class as a method
def treeview1_press(self,obj,event):
“””
Handle people double clicking on our tree view
“””
#print “Clicked”
if event.type==gtk.gdk._2BUTTON_PRESS:
#print “Double Click”
model,iter=self.treeview1.get_selection().get_selected()
if iter==None:
print “weird - nothing was selected, yet we got a double-click”
return
nodetext=model.get_value(iter,0)
#now do something based on nodetext.
#...
Further Enhancing PyRAP
Tree Views can quickly become quite tricky. Not every column needs to be visible to the user. In some
columns, you may want to store references to the actual object being displayed. For example, if you had
a Server class, you could specify a Server object as an invisible object on the line, and if the user double-
clicks on that line, we can pass that information on to the Server object itself.
The trick to doing this is to set_value with a number and then never use that number in insert_row
as text=number. This enables you to have columns in your model that are never referenced in the
TreeView. In fact, you can have different TreeViews, each of which displays different columns in your
241

29.3 Page 283

▲back to top
Chapter 13
model. Hence, the “model, view, controller” name of this way of doing things. (We haven’t gone over the
“controller” part here.)
The following code demonstrates several techniques you will find useful as you code more intensive
applications in pyGTK. First you’ll notice it has little icons embedded in it as XPM files. It also can react
dynamically to requests to add or remove lines from the TreeView. It stores objects in an invisible col-
umn for later use, and has a few other code snippets you might feel like copying at some point as you
become more advanced with your pyGTK work. For what it’s worth, in modern pyGTK, Lists are essen-
tially the same as Trees, so once you become comfortable with Tree’s, you’ll find Lists quite simple.
You’ll walk through the code, and it will be explained it as you go.
First, you have some global variables to declare. The first one is a “XPM” picture of a big capital L. XPM
is a very basic image format. The colors come first with a space being color #000000, a . being color
#ffff04, and the X being the color #b2c0dc, and then these defined characters represent the individual
pixels in the image. It’s a quick and easy way to add icons to your program:
#### START CODE
localNodeXPM = [
“12 12 3 1”,
“ c #000000”,
“. c #ffff04”,
“X c #b2c0dc”,
“X
X”,
“X ..
X”,
“X ..
X”,
“X ..
X”,
“X ..
X”,
“X ..
X”,
“X ..
X”,
“X ..
X”,
“X ..
X”,
“X .........X”,
“X .........X”,
“X
X”
]
You then need to convert that XPM picture into a PixBuf:
localNodePB = gtk.gdk.pixbuf_new_from_xpm_data(localNodeXPM)
We store a reference to this new PixBuf in a dictionary:
text_to_PB={}
text_to_PB[“”]=None
text_to_PB[“LocalNode”]=localNodePB
The next code fragment shows how to expand a TreeView as if someone had clicked on it to expand it. It
takes in a path, which is a numeric description of a row and column in the treeview. It does some basic
error-checking to ensure that path is not None:
242

29.4 Page 284

▲back to top
Writing a GUI with Python
def treeview_expand_to_path(treeview, path):
“””Expand row at path, expanding any ancestors as needed.
This function is provided by gtk+ >=2.2, but it is not yet wrapped
by pygtk 2.0.0.”””
if path==None:
return
for i in range(len(path)):
treeview.expand_row(path[:i+1], open_all=False)
You also have a function to find objects in the TreeModel, starting from an iterator within that model.
This function is recursive (which means that it calls itself), as most tree iteration algorithms are. Of
course, this is not ideal for extremely large data sets, as Python has a somewhat limited stack space com-
pared to languages that are built to use recursive functions more extensively. Still, if your data set is not
over a couple of hundred rows deep, you’ll find this works for your needs. It returns an iterator:
def findobj(model,searchobj,current):
myiter=current
row=model.get_value(myiter,0)
#print “row[0]=%s searchobj=%s”%(row,searchobj)
if row==searchobj:
#print “Found! - returning %s”%(myiter)
return myiter
else:
if model.iter_has_child(myiter):
childiter=model.iter_children(myiter)
while childiter!=None:
myiter=findobj(model,searchobj,childiter)
if myiter!=None:
return myiter
childiter=model.iter_next(childiter)
#print “Not found!”
return None
Now start your nodegui class. You can ignore the engine referenced throughout thanks to proper object
isolation. You’ll find that a lot of your code uses “engines” of various sorts as middleware within your
own application.
The constructor here defers initialization to the init_app function:
class nodegui:
def __init__(self,nodetree,local,engine):
self.engine=engine
self.init_app(nodetree,local)
If you refer back to Figure 13-8 (the image of CANVAS), you’ll see what this code has to manage. It was
originally responsible for adding new nodes to the treeview on the right-hand side (the “Node Tree”).
243

29.5 Page 285

▲back to top
Chapter 13
The addNode function takes a high-level object, the Node, and adds it; then it adds all its displayable
sub-objects to treeview. In this case, we first add the interfaces, followed by the hosts that the node
knows about, and then we finally add any other Nodes that are under this Node, making it a recursive
function.
We then expand the treeview to show this new Node using the expand_to_path function detailed
earlier:
def addNode(self,node):
#print “nodegui::addNode called”
#recursively go through and set up the node tree from the start node
p=self.addLine(node)
self.addLine(node.interfaces)
for interface in node.interfaces.get_children():
self.addLine(interface)
for listeners in interface.get_children():
self.addLine(listeners)
self.addLine(node.hostsknowledge)
for host in node.hostsknowledge.get_children():
self.addLine(host)
for c in host.get_children():
self.addLine(c)
self.addLine(node.connected_nodes)
for n in node.connected_nodes.get_children():
self.addNode(n)
#print “nodegui::addNode leaving”
#self.nodetree.set_cursor(p)
treeview_expand_to_path(self.nodetree, p)
return
The addLine function takes an object and adds that object to the tree model (and hence to the tree view).
Each line has two columns: a potentially empty pixbuf that represents the class of the line and a text
field. The model itself has another column, never displayed by the treeview, which is the line object
itself. This way, the tree model is connected to the objects it represents.
Of course, addLine also checks to ensure that no duplicate objects are in the tree. Each line object has a
parent attribute that is set to None if it is the root object. Otherwise, the line object is added under its
parent object:
def addLine(self,lineobj):
#no duplicates
start=self.model.get_iter_first()
if start!=None and findobj(self.model,lineobj,start):
return
lineobj.set_engine(self.engine)
#print “\\naddLine(%s)”%lineobj
if lineobj.parent==None:
myiter=self.model.insert_after(None,None)
244

29.6 Page 286

▲back to top
Writing a GUI with Python
else:
#somehow find the parent node in the tree
parentobj=lineobj.parent
#for line in tree, if line[0]==parentobj, return line
#http://www.pygtk.org/pygtk2`tutorial/sec-TreeModelInterface.html
start=self.model.get_iter_first()
myiter=findobj(self.model,parentobj,start)
myiter=self.model.insert_after(myiter,None)
lineobj.gui=self
pix=lineobj.get_pix()
#print “Pix=%s”%pix
if pix!=None:
pix=text_to_PB[pix]
if pix==””:
pix=None
#NOT A VISIBLE COLUMN (since text=0 has never been set)
self.model.set_value(myiter,0,lineobj)
self.model.set_value(myiter,1,pix) #Set the icon in the first column
self.model.set_value(myiter,2,lineobj.get_text()) #set the text in the
first column
return self.model.get_path(myiter)
#return
This function deletes a row and all of its children from the TreeView by deleting them from the model.
Iterators are again used to traverse the tree downwards:
def delete(self, line):
treestore=self.model
start=treestore.get_iter_first()
from_parent=findobj(treestore,line,start)
iter = treestore.iter_children(from_parent)
while iter:
treestore.remove(iter)
iter = treestore.iter_children(from_parent)
treestore.remove(from_parent)
return
If an object changes, it may have changed how it wants to be represented by the TreeView. The
update_object method enables it to tell the TreeView that it’s time to refresh its pixbuf or its textual
description:
def update_object(self,object):
start=self.model.get_iter_first()
myiter=findobj(self.model,object,start)
if myiter==None:
#error!
return
pix=object.get_pix()
#print “Pix=%s”%pix
if pix!=None:
pix=text_to_PB[pix]
245

29.7 Page 287

▲back to top
Chapter 13
if pix==””:
pix=None
self.model.set_value(myiter,0,object) #NOT A VISIBLE COLUMN (since text=0
has never been set)
self.model.set_value(myiter,1,pix) #Set the icon in the first column
self.model.set_value(myiter,2,object.get_text()) #set the text in the first
column
self.model.row_changed(self.model.get_path(myiter),myiter)
#TODO: we need to force an expose event to the treeview now, somehow!
return
Next is the deferred initialization procedure. This configures a TreeView and a model for use. You can
see that instead of a CellRendererText, we use a CellRendererPixbuf to create the pretty pictures:
def init_app (self,nodetree,local):
“Initialize the application.”
self.nodetree=nodetree
#set up columns
column
column
#this “text=X” is the column number
cellpb = gtk.CellRendererPixbuf()
column=gtk.TreeViewColumn(“Node Tree”, cellpb, pixbuf=1)
cell = gtk.CellRendererText()
column.pack_start(cell, False) #here we pack a text “column” into the same
column.add_attribute(cell, ‘text’, 2) #column 2 is in “Name” but is a text
#to right align it - we don’t like that very much
#cell.set_property(‘xalign’, 1.0)
self.nodetree.append_column(column)
model=gtk.TreeStore(gobject.TYPE_PYOBJECT,gtk.gdk.Pixbuf,gobject.TYPE_STRING)
self.nodetree.set_model(model)
self.model=model
self.addNode(local)
return
The final method handles interaction with the user. This shows one of the rare times you’ll find yourself
constructing widgets by hand — when doing pop-up menus. Of course, here the pop-up menu is con-
structed out of a list of strings automatically pulled from the line object. This is one of the reasons why
we have a reference to the line object in the model.
All line objects have a menu_response method (not shown here) that will react to being clicked by
the user:
246

29.8 Page 288

▲back to top
Writing a GUI with Python
def line_press(self, obj, event):
#print “Line Press called”
if event.button == 3:
model,iter=self.nodetree.get_selection().get_selected()
if iter==None:
#print “weird - nothing was selected, yet we got a right-click”
return
x=int(event.x)
y=int(event.y)
try:
path, col, colx, celly= obj.get_path_at_pos(x,y)
except TypeError:
return
obj.grab_focus()
obj.set_cursor(path, col, 0)
nodetext=model.get_value(iter,2)
lineobj=model.get_value(iter,0)
menulines=lineobj.get_menu()
if menulines==[]:
#print “Nothing in menu...returning”
return
else:
#print “Something in menu of %s: %s”%(nodetext,menulines)
pass
mymenu=gtk.Menu()
for l in menulines:
mline=gtk.MenuItem(l)
mline.connect(“activate”, lineobj.menu_response, l)
mline.show()
mymenu.append(mline)
#print nodetext, str(event)
mymenu.show()
mymenu.popup(None,None, None,event.button, event.time)
You also have a global quit function, of course:
def quit(args):
gtk.mainquit()
return
Our main function initializes everything and starts the main loop. The trick here is that we use this
module from the main GUI module (and it runs as the main GUI thread), but it can also be tested
independently:
if __name__ == ‘__main__’:
local=localNode()
#do splashscreen here maybe
gladefile=”newgui.glade”
window=”window1”
wTree = gtk.glade.XML (gladefile, window)
247

29.9 Page 289

▲back to top
Chapter 13
nodetree=wTree.get_widget(“nodeview”)
mygui=nodegui(nodetree,local,None)
#window1 must be the main app window!!!
dic = {“on_quit_button_clicked”
: quit,
“on_window1_destroy” : (quit),
“on_nodeview_button_press_event”:mygui.line_press,
}
window=wTree.get_widget(“window1”) # sure there must be another way
wTree.signal_autoconnect (dic)
#hmmm
try:
gtk.threads_init()
except:
print “No threading was enabled when you compiled pyGTK!”
sys.exit(1)
gtk.threads_enter()
gtk.mainloop ()
gtk.threads_leave()
Summary
There’s no limit to the things you can do with your GUI using pyGTK. You can take screenshots, display
graphics, handle complex information sets in large windows, draw on a blank canvas, or simply pop up
quick GUIs for custom command-line utilities, exposing them to less technically oriented users.
There are, of course, personal styles to every programming project. Many people have developed tools
that enable automated application development. Python’s bevy of introspection and OO features enables
you to dynamically handle all sorts of changes in your GUI. As you become more familiar with pyGTK,
you’ll find these sorts of techniques to be extremely natural.
Even if you don’t use pyGTK, understanding how pyGTK works will be a valuable asset in your pro-
gramming toolbox. Furthermore, there’s always the possibility that you have a spare 15 minutes and
want to write a custom GUI chat client for your friends.
Exercises
1. Write a Glade interface and a pyGTK class that runs a command and puts the results into a
TextView.
2. Modify exercise 1 to put the results into the TextView as they come back from the command.
(Hint: You’ll need to use threading to do this).
248

29.10 Page 290

▲back to top
14
Accessing Databases
Just about every large enterprise system uses a database for storing data. For example, amazon.com,
the online retailer, needs a database to store information on each product for sale. For Python to
prove capable of handling these types of enterprise applications, the language must be able to
access databases.
Luckily, Python provides a database API (Application Programming Interface — how you program
for the database), which enables you to access most databases using an API that is very similar in
all of the databases that the API works with, in spite of the databases’ different native APIs. The
database, or DB, API doesn’t define all aspects of working with databases, so there are some minor
differences. For the most part, though, you can access databases such as Oracle or MySQL from
your Python scripts without worrying too much about the details of the specific databases.
Having a generic database API proves very useful, as you may need to switch databases or have
your application work with multiple databases, and you won’t want to recode major parts of your
program to allow this. Normally, you can do all of this in Python without a lot of programming
changes being needed.
Even if you aren’t writing the next amazon.com online site, databases provide a convenient means
to persist data for longer than the program is running (so that you don’t lose the data that a user
has entered if you want to restart your program), query for items, and modify your data in a safe
manner.
This chapter covers the two main database systems supported by Python, DBM persistent diction-
aries, and relational databases with the DB API. In addition, this chapter describes how to set up a
database, in case you don’t have a database handy.
Specific topics include the following:
❑ Using the DBM libraries to create persistent dictionaries
❑ Learning about relational databases
❑ Setting up the Gadfly database

30 Pages 291-300

▲back to top

30.1 Page 291

▲back to top
Chapter 14
❑ Setting up the MySQL database
❑ Working with the Python DB API
❑ Creating connections
❑ Accessing data with cursors
❑ Connecting to databases
❑ Querying and modifying data
❑ Working with transactions
❑ Handling errors
❑ Using other database tools
In many cases, you don’t require a full-blown relational database. In such cases, creating a persistent dic-
tionary using dbm files is enough.
Working with DBM Persistent Dictionaries
A persistent dictionary acts exactly like you’d expect. You can store name/value pairs in the dictionary,
which are saved to a disk, and so their data will endure between various times that your program is run.
So if you save data to a dictionary that’s backed by a dbm, the next time that you start your program,
you can read the value stored under a given key again, once you’ve loaded the dbm file. These dictionar-
ies work like normal Python dictionaries, which are covered in Chapter 3. The main difference is that the
data is written to and read from disk.
An additional difference is that the keys and the values must both be strings.
DBM, short for database manager, acts as a generic name for a number of C language libraries originally
created on Unix systems. These libraries sport names such as dbm, gdbm, ndbm, sdbm, and so on. These
names correspond closely to the available modules in Python that provide the requisite functionality.
Choosing a DBM Module
Python supports a number of DBM modules. Each DBM module supports a similar interface and uses a
particular C library to store the data to disk. The main difference lies in the underlying binary format of
the data files on disk. Each DBM module, unfortunately, creates incompatible files. That is, if you create
a DBM persistent dictionary with one DBM module, you must use the same module to read the data.
None of the other modules will work with that data file.
The following table lists the DBM modules.
anydbm
dbhash
dbm
Chooses best DBM module
Uses the Berkeley Unix DB library
Uses the Unix DBM library
250

30.2 Page 292

▲back to top
Accessing Databases
dumbdbm
gdbm
whichdb
Uses a simple, but portable, implementation of the DBM library
Uses the GNU DBM library
Guesses which DBM module to use to open a file
All of these libraries exist because of the history of the DBM library. Originally, this library was only
available on commercial versions of Unix. Free versions of Unix, and later Linux, Windows, and so on,
could not use the DBM library. This lead to the creation of alternative libraries, such as the Berkeley Unix
library and the GNU gdbm library.
With all the incompatible file formats, this plethora of libraries can be a real pain. The anydbm module,
though, offers a handy alternative to choosing a specific DBM module. With the anydbm module, you
can let it choose for you. In general, the anydbm module will choose the best implementation available
on your system when creating a new persistent dictionary. When reading a file, the anydbm module
uses the whichdb module to make an informed guess as to which library created the data file.
Unless you need a specific advanced feature of one of the DBM libraries, use the anydbm module.
Creating Persistent Dictionaries
All of the DBM modules support an open function to create a new dbm object. Once opened, you can
store data in the dictionary, read data, close the dbm object (and the associated data file or files), remove
items, and test for the existence of a key in the dictionary.
To open a DBM persistent dictionary, use the open function on the module you choose. For example,
you can create a persistent dictionary with the anydbm module.
Try It Out Creating a Persistent Dictionary
Enter the following code and name your file dbmcreate.py:
import anydbm
db = anydbm.open(‘websites’, ‘c’)
# Add an item.
db[‘www.python.org’] = ‘Python home page’
print db[‘www.python.org’]
# Close and save to disk.
db.close()
When you run this script, you’ll see output like the following:
$ python dbmcreate.py
Python home page
251

30.3 Page 293

▲back to top
Chapter 14
How It Works
This example uses the recommended anydbm module.
The open function requires the name of the dictionary to create. This name gets translated into the name
of the data file or files that may already be on the disk. (The DBM module may — though not always —
create more than one file, usually a file for the data and one for the index of the keys.) The name of the
dictionary is treated as a base filename, including the path. Usually, the underlying DBM library will
append a suffix such as .dat for data. You can find the file yourself by looking for the file named
websites, most likely in your current working directory.
You should also pass the optional flag. This flag is not optional for the dbhash module. The following
table lists the available flags.
Flag
Usage
c
Opens the data file for reading and writing, creating the file if needed.
n
Opens the file for reading and writing, but always creates a new empty file.
If one already exists, it will be overwritten and its contents lost.
w
Opens the file for reading and writing, but if the file doesn’t exist it will not
be created.
You can also pass another optional parameter, the mode. The mode holds a set of Unix file permissions.
See Chapter 8 for more on opening files.
The open method of the dbm modules returns a new dbm object, which you can
then use to store and retrieve data.
After you open a persistent dictionary, you can write values as you normally would with Python dictio-
naries, as shown in the following example:
db[‘www.python.org’] = ‘Python home page’
Both the key and the value must be strings and can’t be other objects, like numbers or python objects.
Remember, however, that if you want to save an object, you can serialize it using the pickle module, as
you saw in Chapter 8.
The close method closes the file or files and saves the data to disk.
Accessing Persistent Dictionaries
With the DBM modules, you can treat the object you get back from the open function as a dictionary
object. Get and set values using code like the following:
db[‘key’] = ‘value’
value = db[‘key’]
Remember that the key and the value must both be text strings.
252

30.4 Page 294

▲back to top
Accessing Databases
You can delete a value in the dictionary using del:
del db[‘key’]
You can determine whether a particular key is stored in the dictionary using if:
if db[‘key’] != None:
print ‘Key exists in dictionary’
If you use the dbhash module, you can use the following syntax as an alternate means to determine
whether a particular key is stored in the dictionary:
if (‘key’ in db):
print ‘Key exists in dictionary’
This syntax works with the dbhash type of DBM module. It does not work with all other DBM modules.
The keys method returns a list of all the keys, in the same way it would with a normal dictionary:
for key in db.keys():
# do something...
The keys method may take a long time to execute if there are a huge number of keys in the file. In addi-
tion, this method may require a lot of memory to store the potentially large list that it would create with
a large file.
You can use the following script as a guide for how to program with DBM persistent dictionaries.
Try It Out Accessing Persistent Dictionaries
Enter the following script and name the file dbmaccess.py:
import anydbm
import whichdb
# Check the type.
print “Type of DBM file =”, whichdb.whichdb(‘websites’)
# Open existing file.
db = anydbm.open(‘websites’, ‘w’)
# Add another item.
db[‘www.wrox.com’] = ‘Wrox home page’
# Verify the previous item remains.
if db[‘www.python.org’] != None:
print ‘Found www.python.org’
else:
print ‘Error: Missing item’
# Iterate over the keys. May be slow.
253

30.5 Page 295

▲back to top
Chapter 14
# May use a lot of memory.
for key in db.keys():
print “Key =”,key,” value =”,db[key]
del db[‘www.wrox.com’]
print “After deleting www.wrox.com, we have:”
for key in db.keys():
print “Key =”,key,” value =”,db[key]
# Close and save to disk.
db.close()
When you run this script, you’ll see output similar to the following:
$ python dbmaccess.py
Type of DBM file = dbhash
Found www.python.org
Key = www.wrox.com value = Wrox home page
Key = www.python.org value = Python home page
After deleting www.wrox.com, we have:
Key = www.python.org value = Python home page
How It Works
This script works with a small database of web site URLs and descriptions. You need to first run the
dbmcreate.py example, shown previously. That example creates the DBM file and stores data in the
file. The dbmaccess.py script then opens the pre-existing DBM file. The dbmaccess.py script starts out
using the whichdb.whichdb function to determine the type of DBM file created by the previous exam-
ple, dbmcreate.py, for the DBM persistent dictionary websites. In the example here, it’s correctly
determined that the type is dbhash.
The dbmaccess.py script then opens the persistent dictionary websites in read/write mode. The call
to the open function will generate an error if the necessary data file or files do not exist on disk in the
current directory.
From the previous example, dbmcreate.py, there should be one value in the dictionary, under the key
www.python.org. This example adds the Wrox web site, www.wrox.com, as another key.
The script verifies that the www.python.org key exists in the dictionary, using the following code:
if db[‘www.python.org’] != None:
print ‘Found www.python.org’
else:
print ‘Error: Missing item’
Next, the script prints out all of the keys and values in the dictionary:
for key in db.keys():
print “Key =”,key,” value =”,db[key]
Note that there should be only these two entries.
254

30.6 Page 296

▲back to top
Accessing Databases
After printing out all of the entries, the script removes one using del:
del db[‘www.wrox.com’]
The script then prints all of the keys and values again, which should result in just one entry, as shown in
the output.
Finally, the close method closes the dictionary, which involves saving all the changes to disk, so the
next time the file is opened it will be in the state you left it.
As you can see, the API for working with persistent dictionaries is incredibly simple because it works
like files and like dictionaries, which you’re already familiar with.
Deciding When to Use DBM and When
to Use a Relational Database
The DBM modules work when your data needs can be stored as key/value pairs. You can store more
complicated data within key/value pairs with some imagination — for instance, by creating formatted
strings that use a comma or some other character to delimit items in the strings, both on the key and the
value part of the dictionary. This can be useful, but it can also be very difficult to maintain, and it can
restrict you because your data is stored in an inflexible manner. Another way that you can be limited is
technical: Note that some DBM libraries limit the amount of space you can use for the values (sometimes
to a maximum of 1024 bytes, which is very, very little).
You can use the following guidelines to help determine which of these two types of data storage is
appropriate for your needs:
❑ If your data needs are simple, use a DBM persistent dictionary.
❑ If you only plan to store a small amount of data, use a DBM persistent dictionary.
❑ If you require support for transactions, use a relational database. (Transactions are when more
than one thing happens at once — they let you keep your data from getting changed in one
place but not in another; you get to define what happens concurrently with transactions.)
❑ If you require complex data structures or multiple tables of linked data, use a relational
database.
❑ If you need to interface to an existing system, use that system, obviously. Chances are good this
type of system will be a relational database.
Unlike the simple DBM modules, relational databases provide a far richer and more complex API.
Working with Relational Databases
Relational databases have been around for decades so they are a mature and well-known technology.
People who work with relational databases know what they are supposed to do, and how they are sup-
posed to work, so relational databases are the technology of choice for complex data storage.
255

30.7 Page 297

▲back to top
Chapter 14
In a relational database, data is stored in tables that can be viewed as two-dimensional data structures.
The columns, or vertical part of the two-dimensional matrix, are all of the same type of data; like strings,
numbers, dates, and so on. Each horizontal component of the table is made up of rows, also called records.
Each row in turn is made up of columns. Typically, each record holds the information pertaining to one
item, such as an audio CD, a person, a purchase order, an automobile, and so on.
For example, the following table shows a simple employee table.
empid
105
201
firstname
Peter
Bob
lastname
Tosh
Marley
department
2
1
manager
45
36
phone
555-5555
555-5551
This table holds six columns:
❑ empid holds the employee ID number. Relational databases make extensive use of ID numbers
where the database manages the assignment of unique numbers so that each row can be refer-
enced with these numbers to make each row unique (even if they have identical data). We can
then refer to each employee by the ID number. The ID alone provides enough information to
look up the employee.
❑ firstname holds the person’s first name.
❑ lastname holds the person’s last name.
❑ department holds the ID of the department in which the employee works. This would likely be
a numeric ID of the department, where departments are defined in a separate table that has a
unique ID for each department.
❑ manager holds the employee ID of the manager of the given employee. This is sort of self-
referential, because in this example, a manager is actually an employee.
❑ phone holds the office phone number.
In real life, a company would likely store a lot more information about an employee, such as a taxation
authority identification number (social security number in the U.S.), home address, and more, but not
anything that’s really different in principle to what you’ve already seen.
In this example, the column empid, the employee ID, would be used as the primary key. A primary key
is a unique index for a table, where each element has to be unique because the database will use that
element as the key to the given row and as the way to refer to the data in that row, in a manner similar
to dictionary keys and values in Python. So, each employee needs to have a unique ID number, and
once you have an ID number, you can look up any employee. So, the empid will act as the key into this
table’s contents.
The department column holds an ID of a department — that is, an ID of a row in another table. This ID
could be considered a foreign key, as the ID acts as a key into another table. (In databases, a foreign key
has a much more strict definition, so it’s okay to think of it this way.)
For example, the following table shows a possible layout for the department table.
256

30.8 Page 298

▲back to top
Accessing Databases
departmentid
1
2
name
development
qa
Manager
47
32
In these examples, the employee Peter Tosh works for department 2, the qa, or quality assurance, depart-
ment in a dynamic world-class high-quality software development firm. Bob Marley works for depart-
ment 1, the development department.
In a large enterprise, there may be hundreds of tables in the database, with thousands or even millions
of records in some tables.
Writing SQL Statements
The Structured Query Language, or SQL, defines a standard language for querying and modifying
databases.
You can pronounce SQL as “sequel” or “s-q-l.”
SQL supports the basic operations listed in the following table.
Operation
Select
Update
Insert
Delete
Usage
Perform a query to search the database for specific data.
Modify a row or rows, usually based on a certain condition.
Create new rows in the database.
Remove a row or rows from the database.
In general, these basic operations are called QUID, short for Query, Update, Insert, and Delete, or CRUD,
short for Create, Read, Update, and Delete. SQL offers more than these basic operations, but for the most
part, these are the majority of what you’re going to use to write applications.
If you are not familiar with SQL, look at a SQL book or search on the Internet. You will find a huge
amount of tutorial material. You may also look at the web site for this book for more references to SQL
resources.
SQL is important because when you access databases with the Python DB API, you must first create SQL
statements and then execute these statements by having the database evaluate them. You then retrieving
the results and use them. Thus, you will find yourself in the awkward position of using one language,
Python, to create commands in another language, SQL.
257

30.9 Page 299

▲back to top
Chapter 14
The basic SQL syntax for the CRUD operations follows:
SELECT columns FROM tables WHERE condition ORDER BY columns ascending_or_descending
UPDATE table SET new values WHERE condition
INSERT INTO table (columns) VALUES (values)
DELETE FROM table WHERE condition
In addition to this basic look at the available syntax, there are many more parameters and specifiers for
each operation that are optional. You can still use them with Python’s DB API if you’re familiar with SQL.
To insert a new row in the employee table, using the previous employee example, you can use a SQL
query like the following (even though it’s adding data and not getting data, the convention is that all
SQL commands or statements can also be called queries):
insert into employee (empid, firstname, lastname, manager, dept, phone)
values (3, ‘Bunny’, ‘Wailer’, 2, 2, ‘555-5553’)
In this example, the first tuple (it’s useful to think of these in Python terms, even though SQL will give
these different names) holds the names of the columns in the order you are using for inserting your data.
The second tuple, after the keyword values, holds the data items in the same order. Notice how SQL
uses single quotes to delimit strings, and no quotes around numbers. (The phone number is different —
it’s actually a string because it has to be able to contain nonnumbers, like dashes, periods, and plus
signs, depending on how the data is entered.)
With queries, you can use shortcuts such as * to say that you want an operation to be performed using
all of the columns in a table. For example, to query all of the rows in the department table, showing all
of the columns for each row, you can use a query like the following:
select * from department
Note that SQL is not case-sensitive for its keywords, such as SELECT and FROM. But, some databases
require table and column names to be all uppercase. It is common, therefore, to see people use SELECT
and FROM and other operations in all capital letters to make them easily distinguished from other parts
of the query.
This SQL statement omits the names of the columns to read and any conditions that would otherwise
narrow down the data that would be returned. Thus the query will return all of the columns (from the *)
and all of the rows (due to there being no where clause).
You can perform a join with the select command, to query data from more than one table, but present it
all in a single response. It’s called a join because the data from both tables will be returned as though it was
queried from a single table. For example, to extract the department name with each employee, you could
perform a query like the following (all of which would need to be in one string to be a single query):
select employee.firstname, employee.lastname, department.name
from employee, department
where employee.dept = department.departmentid
order by lastname desc
258

30.10 Page 300

▲back to top
Accessing Databases
In this example, the select statement requests two columns from the employee table (the firstname
and the lastname, but these are specified as coming from employee by the convention of specifying the
table name and the column name in the table) and one from the department table (department.name).
The order by section of the statement tells the database to order the results by the value in the lastname
column, in descending order.
To simplify these queries, you can use aliases for the table names, which make them easier to type and
to read (but don’t change the logic or the syntax of your queries). For example, to use the alias e with the
employee table, you can start a query as follows:
select e.firstname, e.lastname
from employee e
...
In this case, you must place the alias, e, after the table name in the from clause. You can also use the fol-
lowing format with the optional key word as, which could be easier for you to read:
select e.firstname, e.lastname
from employee as e
...
To modify (or update) a row, use a SQL statement like the following:
update employee set manager=55 where empid=3
This example modifies the employee with an ID of 3 by setting that employee’s manager to the
employee with an ID of 55. As with other queries, numbers don’t need to have quotes around them;
however, strings would need to be quoted with single quotes.
To delete a row, use a SQL statement like the following:
delete employee where empid=42
This example deletes the employee with an ID of 42 but doesn’t affect anything else in the database.
Defining Tables
When you first set up a database, you need to define the tables and the relations between them. To do
this, you use the part of the SQL language called the DDL, or Data-Definition Language. (It defines the
structure of your tables — get it?) DDL basics are pretty simple, where you use one operation to create
tables, and another one to remove them:
CREATE TABLE tablename (column, type column type, . . . )
DROP TABLE tablename
There is also an ALTER TABLE command to modify an existing table, but you won’t need to do that
for now. When you want to use this, a dedicated SQL book or web page will have more about this
command.
Unfortunately, SQL is not an entirely standard language, and there are parts of it that each database
doesn’t do the same. The DDL remains a part of SQL that has not been standardized. Thus, when
259

31 Pages 301-310

▲back to top

31.1 Page 301

▲back to top
Chapter 14
defining tables you will find differences between the SQL dialects supported by the different databases,
though the basics concepts are the same.
Setting Up a Database
In most cases when you’re the programmer, you will already have a database that’s up and running,
perhaps even a database chosen by some other organization that you’re going to have to use. For exam-
ple, if you host your web site with a web site hosting company that provides bells and whistles, like a
database, your hosting package may include access to the MySQL database. If you work for a large orga-
nization, your IT department may have already standardized on a particular database such as Oracle,
DB/2, Sybase, or Informix. These latter packages are likely present in your workplace if you create enter-
prise applications with Python.
If you have no database at all, yet still want to work on the examples in this chapter, then a good starting
database is Gadfly. The main virtues of Gadfly include the fact that the database is written in Python, so
Gadfly can run on any platform where Python runs. In addition, Gadfly is simple and small, but func-
tional. This makes it a great candidate for your experimentation while you’re learning, even if you’ve
got another database available to you. Just keep in mind that each database has its own quirks.
The examples in this chapter were written to work with Gadfly so that you can follow them without any
external infrastructure being needed. You can easily modify these examples, though, to work with a dif-
ferent database. That’s one of the great aspects of the Python DB API.
Download the ZIP file for the latest Gadfly release from http://gadfly.sourceforge.net/. As with
other Python modules, you can install Gadfly with the following steps:
1. Unpack the file. (You can use Unzip on Unix, Winzip or something similar on Windows. Make
sure to use the options that will create the directory structure that’s embedded in the zip file.)
2. Change to the gadflyZip directory.
3. Run the command python setup.py install.
For example on a Linux or Unix platform (such as Mac OS/X):
$ python setup.py install
When you run this command, you may need administrator or root permissions to install the Gadfly
scripts in the system-wide location alongside the Python installation.
Once you have installed the Gadfly modules, you need to create a database. This part of working with a
database is not standardized as part of the DB API, so you need to write some Python code that is spe-
cific to the Gadfly database to handle this.
If you are working with another database, such as SQL Server, chances are good that a database has
already been created. If not, follow the instructions from your database vendor. (A lot of the time, you
can get help on tasks like this from your Database Administrator, or DBA, who would really rather have
you working on a test database instead of on a production database.)
With Gadfly, creating a database is rather easy.
260

31.2 Page 302

▲back to top
Accessing Databases
Try It Out Creating a Gadfly Database
Enter the following script and name the file createdb.py:
import os
import gadfly
connection = gadfly.gadfly()
os.mkdir(‘db’)
connection.startup(‘pydb’, ‘db’)
cursor = connection.cursor()
# Create tables.
cursor.execute(“””
create table employee
(empid integer,
firstname varchar,
lastname varchar,
dept integer,
manager integer,
phone varchar)
“””)
cursor.execute(“””
create table department
(departmentid integer,
name varchar,
manager integer)
“””)
cursor.execute(“””
create table user
(userid integer,
username varchar,
employeeid integer)
“””)
# Create indices.
cursor.execute(“””create index userid on user (userid)”””)
cursor.execute(“””create index empid on employee (empid)”””)
cursor.execute(“””create index deptid on department (departmentid)”””)
cursor.execute(“””create index deptfk on employee (dept)”””)
cursor.execute(“””create index mgr on employee (manager)”””)
cursor.execute(“””create index emplid on user (employeeid)”””)
cursor.execute(“””create index deptmgr on department (manager)”””)
connection.commit()
cursor.close()
connection.close()
261

31.3 Page 303

▲back to top
Chapter 14
When you run this script, you should see no output unless the script raised an error:
$ python createdb.py
$
How It Works
Gadfly has its own API along with the standard Python DB API. This script uses the Gadfly API, but
you’ll notice that this API is very similar to the DB API covered in the following section, “Using the
Python Database APIs.” This section briefly describes the Gadfly-specific code in the creatdb.py script.
Among the Gadfly-specific code, you need to create a Connection object using the gadfly function on
the gadfly module. For example:
connection = gadfly.gadfly()
connection.startup(‘pydb’, ‘db’)
Note that the gadfly module has the Gadfly-specific API. You need to use the gadfly.dbapi20
module to work with the DB API 2.0.
Once you get a Connection object, you need to start up the database. Pass the name of the database,
pydb here, and the path to the directory to use, db in this example. (This script creates the db directory
using the standard Python os module.)
From there, the script gets a Cursor object, covered in the section “Working with Cursors.” The Cursor
object is used to create three tables and define indexes on these tables.
The script calls the commit method on the Connection to save all the changes to disk.
Gadfly stores all of its data in a directory you define, db in this case. After running the createdb.py
script, you should see the following in the db directory, where each .grl file is a gadfly table:
$ ls db
DEPARTMENT.grl EMPLOYEE.grl pydb.gfd USER.grl
You are now ready to start working with the Python database APIs.
Using the Python Database APIs
First, some history about Python and relational databases. Python’s support for relational databases
started out with ad hoc solutions, with one solution written to interface with each particular database,
such as Oracle. Each database module created its own API, which was highly specific to that database
because each database vendor evolved its own API based on its own needs. This is hard to support,
because coding for one database and trying to move it to the other gives a programmer severe heart-
burn, as everything needs to be completely rewritten and retested.
Over the years, though, Python has matured to support a common database, or DB, API, that’s called the
DB API. Specific modules enable your Python scripts to communicate with different databases, such as
262

31.4 Page 304

▲back to top
Accessing Databases
DB/2, PostgreSQL, and so on. All of these modules, however, support the common API, making your
job a lot easier when you write scripts to access databases. This section covers this common DB API.
The DB API provides a minimal standard for working with databases, using Python structures and syn-
tax wherever possible. This API includes the following:
❑ Connections, which cover guidelines for how to connect to databases
❑ Executing statements and stored procedures to query, update, insert, and delete data with
cursors
❑ Transactions, with support for committing or rolling back a transaction
❑ Examining metadata on the database module as well as on database and table structure
❑ Defining the types of errors
The following sections take you step by step through the Python database APIs.
Downloading Modules
You must download a separate DB API module for each database you need to access. For example, if
you need to access an Oracle database as well as a MySQL database, you must download both the
Oracle and the MySQL database modules.
See www.python.org/topics/database/modules.html for a listing of database modules.
Modules exist for most major databases with the notable exception of Microsoft’s SQL Server. You can
access MQL Server using an ODBC module, though. In fact, the mxODBC module can communicate
with most databases using ODBC on Windows or an ODBC bridge on Unix (including Mac OS X) or
Linux. If you need to do this, you can search for more information on these terms online to find out how
other people are doing it.
Download the modules you need. Follow the instructions that come with the modules to install them.
You may need a C compiler and build environment to install some of the database modules. If you do,
this will be described in the module’s own documentation, which you’ll need to read.
For some databases, such as Oracle, you can choose among a number of different modules that are
slightly different. You should choose the module that seems to best fit your needs or go to the web site
for this book and ask the authors for any recommendations if you’re not sure.
Once you have verified that the necessary modules are installed, you can start working with
Connections.
Creating Connections
A Connection object provides the means to communicate from your script to a database program. Note
the major assumption here that the database is running in a separate process (or processes). The Python
database modules connect to the database. They do not include the database application itself.
263

31.5 Page 305

▲back to top
Chapter 14
Each database module needs to provide a connect function that returns a Connection object. The param-
eters that are passed to connect vary by the module and what is required to communicate with the
database. The following table lists the most common parameters.
Parameter
dsn
host
database
user
password
Usage
Data source name, from ODBC terminology. This usually includes
the name of your database and the server where it’s running.
Host, or network system name, on which the database runs
Name of the database
User name for connecting to the database
Password for the given user name
For example, you can use the following code as a guide:
connection = dbmodule.connect(dsn=’localhost:MYDB’,user=’tiger’,password=’scott’)
Use your database module documentation to determine which parameters are needed.
With a Connection object, you can work with transactions, covered later in this chapter; close the connec-
tion to free system resources, especially on the database; and get a cursor.
Working with Cursors
A Cursor is a Python object that enables you to work with the database. In database terms, the cursor is
positioned at a particular location within a table or tables in the database, sort of like the cursor on your
screen when you’re editing a document, which is positioned at a pixel location.
To get a Cursor, you need to call the cursor method on the Connection object:
cursor = connection.cursor()
Once you have a cursor, you can perform operations on the database, such as inserting records.
Try It Out Inserting Records
Enter the following script and name the file insertdata.py:
import gadfly.dbapi20
connection = gadfly.dbapi20.connect(‘pydb’, ‘db’)
cursor = connection.cursor()
# Create employees.
cursor.execute(“””
insert into employee (empid,firstname,lastname,manager,dept,phone)
values (1,’Eric’,’Foster-Johnson’,1,1,’555-5555’)”””)
264

31.6 Page 306

▲back to top
Accessing Databases
cursor.execute(“””
insert into employee (empid,firstname,lastname,manager,dept,phone)
values (2,’Peter’,’Tosh’,2,3,’555-5554’)”””)
cursor.execute(“””
insert into employee (empid,firstname,lastname,manager,dept,phone)
values (3,’Bunny’,’Wailer’,2,2,’555-5553’)”””)
# Create departments.
cursor.execute(“””
insert into department (departmentid,name,manager)
values (1,’development’,1)”””)
cursor.execute(“””
insert into department (departmentid,name,manager)
values (2,’qa’,2)”””)
cursor.execute(“””
insert into department (departmentid,name,manager)
values (3,’operations’,2)”””)
# Create users.
cursor.execute(“””
insert into user (userid,username,employeeid)
values (1,’ericfj’,1)”””)
cursor.execute(“””
insert into user (userid,username,employeeid)
values (2,’tosh’,2)”””)
cursor.execute(“””
insert into user (userid,username,employeeid)
values (3,’bunny’,3)”””)
connection.commit()
cursor.close()
connection.close()
When you run this script, you will see no output unless the script raises an error:
$ python insertdata.py
How It Works
The first few lines of this script set up the database connection and create a cursor object:
import gadfly.dbapi20
connection = gadfly.dbapi20.connect(‘pydb’, ‘db’)
cursor = connection.cursor()
265

31.7 Page 307

▲back to top
Chapter 14
Note the use of the gadfly.dbapi20 module, which connects to a Gadfly database. To connect to a dif-
ferent database, replace this with your database-specific module, and modify the call to use the connect
function from that database module, as needed.
The next several lines execute a number of SQL statements to insert rows into the three tables set up
earlier: employee, department, and user. The execute method on the cursor object executes the SQL
statement:
cursor.execute(“””
insert into employee (empid,firstname,lastname,manager,dept,phone)
values (2,’Peter’,’Tosh’,2,3,’555-5554’)”””)
This example uses a triple-quoted string to cross a number of lines as needed. You’ll find that SQL com-
mands, especially those embedded within Python scripts, are easier to understand if you can format the
commands over a number of lines. This becomes more important with complex queries covered in
examples later in this chapter.
To save your changes to the database, you must commit the transaction:
connection.commit()
Note that this method is called on the connection, not the cursor.
When you are done with the script, close the Cursor and then the Connection to free up resources. In
short scripts like this, it may not seem important, but this helps the database program free its resources,
as well as your Python script:
cursor.close()
connection.close()
You now have a very small amount of sample data to work with using other parts of the DB API, such as
querying for data.
Try It Out Writing a Simple Query
The following script implements a simple query that performs a join on the employee and department
tables:
import gadfly.dbapi20
connection = gadfly.dbapi20.connect(‘pydb’, ‘db’)
cursor = connection.cursor()
cursor.execute(“””
select employee.firstname, employee.lastname, department.name
from employee, department
where employee.dept = department.departmentid
order by employee.lastname desc
“””)
266

31.8 Page 308

▲back to top
Accessing Databases
for row in cursor.fetchall():
print row
cursor.close()
connection.close()
Save this script under the name simplequery.py.
When you run this script, you will see output like the following:
$ python simplequery.py
(‘Bunny’, ‘Wailer’, ‘qa’)
(‘Peter’, ‘Tosh’, ‘operations’)
(‘Eric’, ‘Foster-Johnson’, ‘development’)
How It Works
This script initializes the connection and cursor in the same manner as the previous script. This
script, though, passes a simple join query to the cursor execute method. This query selects two
columns from the employee table and one from the department table.
This is truly a simple query, but, even so, you’ll want to format your queries so they are readable, simi-
lar to what is shown here.
When working with user interfaces, you will often need to expand IDs stored in the database to human-
readable values. In this case, for example, the query expands the department ID, querying for the
department name. You simply cannot expect people to remember the meaning of strange numeric IDs.
The query also orders the results by the employees’ last names, in descending order. (This means that it
starts at the beginning of the alphabet, which is what you’d normally expect. However, you can reverse
this and have them sorted in ascending order.)
After calling the execute method, the data, if any was found, is stored in the cursor object. You can use
the fetchall method to extract the data.
You can also use the fetchone method to fetch one row at a time from the results.
Note how the data appears as Python tuples:
(‘Bunny’, ‘Wailer’, ‘qa’)
(‘Peter’, ‘Tosh’, ‘operations’)
(‘Eric’, ‘Foster-Johnson’, ‘development’)
You can use this example as a template to create other queries, such as the more complex join shown in
the following Try It Out.
Try It Out Writing a Complex Join
Enter this script and name the file finduser.py:
267

31.9 Page 309

▲back to top
Chapter 14
import sys
import gadfly.dbapi20
connection = gadfly.dbapi20.connect(‘pydb’, ‘db’)
cursor = connection.cursor()
username = sys.argv[1]
query = “””
select u.username,e.firstname,e.lastname,m.firstname,m.lastname, d.name
from user u, employee e, employee m, department d where username=?
and u.employeeid = e.empid
and e.manager = m.empid
and e.dept = d.departmentid
“””
cursor.execute(query, (username,))
for row in cursor.fetchall():
(username,firstname,lastname,mgr_firstname,mgr_lastname,dept) = row
name=firstname + “ “ + lastname
manager=mgr_firstname + “ “ + mgr_lastname
print username,”:”,name,”managed by”,manager,”in”,dept
cursor.close()
connection.close()
When you run this script, you will see results like the following:
$ python finduser.py bunny
bunny : Bunny Wailer managed by Peter Tosh in qa
You need to pass the user name of a person to query from the database. This must be a valid user name
of a person in the database. In this example, bunny is a user name previously inserted into the database.
How It Works
This script performs a join on all three example tables, using table-name aliases to create a shorter query.
The purpose is to find a given user in the database by searching for that user name. This script also
shows an example of expanding both the manager’s ID to the manager’s name and the department’s ID
to the department’s name. All of this makes for more readable output.
This example also shows how you can extract data from each row into Python variables. For example:
(username,firstname,lastname,mgr_firstname,mgr_lastname,dept) = row
Note that this is really nothing new. See Chapter 3 for more on Python tuples, which is all row is.
An important new feature of this script, though, is the use of a question mark to enable you to build a
query using dynamic data. When you call the execute method on the Cursor, you can pass a tuple of
dynamic data, which the execute method will fill in for the question marks in the SQL statement. (This
example uses a tuple of one element.) Each element in the tuple is used, in order, to replace the question
268

31.10 Page 310

▲back to top
Accessing Databases
marks. Thus, it is very important to have as many dynamic values as you do question marks in the SQL
statement, as shown in the following example:
query = “””
select u.username,e.firstname,e.lastname,m.firstname,m.lastname, d.name
from user u, employee e, employee m, department d where username=?
and u.employeeid = e.empid
and e.manager = m.empid
and e.dept = d.departmentid
“””
cursor.execute(query, (username,))
The query used in this example is very helpful when you want to start updating rows in the tables.
That’s because users will want to enter meaningful values. It is up to you, with your SQL statements, to
translate the user input into the necessary IDs.
For example, the following script enables you to change the manager for an employee:
Personally, I’d like to make myself my own manager.
Try It Out Updating an Employee’s Manager
Enter the following script and name the file updatemgr.py:
import sys
import gadfly.dbapi20
connection = gadfly.dbapi20.connect(‘pydb’, ‘db’)
cursor = connection.cursor()
newmgr = sys.argv[2]
employee = sys.argv[1]
# Query to find the employee ID.
query = “””
select e.empid
from user u, employee e
where username=? and u.employeeid = e.empid
“””
cursor.execute(query,(newmgr,));
for row in cursor.fetchone():
if (row != None):
mgrid = row
# Note how we use the same query, but with a different name.
cursor.execute(query,(employee,));
for row in cursor.fetchone():
if (row != None):
empid = row
269

32 Pages 311-320

▲back to top

32.1 Page 311

▲back to top
Chapter 14
# Now, modify the employee.
cursor.execute(“update employee set manager=? where empid=?”, (mgrid,empid))
connection.commit()
cursor.close()
connection.close()
When you run this script, you need to pass the name of the user to update, as well as the name of the
manager. Both names are user names from the user table. For example:
$ python finduser.py bunny
bunny : Bunny Wailer managed by Peter Tosh in qa
$ python updatemgr.py bunny ericfj
$ python finduser.py bunny
bunny : Bunny Wailer managed by Eric Foster-Johnson in qa
How It Works
The example output shows the before and after picture of the employee row, verifying that the
updatemgr.py script worked.
The updatemgr.py script expects two values from the user: the user name of the employee to update
and the user name of the new manager. Both of these names must be user names stored in the database.
Both names are converted into IDs using a simple query. This is not very efficient, as it involves two
extra round-trips to the database. A more efficient means would be to perform an inner select state-
ment on the update statement. For simplicity, though, the separate queries are far easier to understand.
This example also shows the use of the fetchone method on the Cursor. The final SQL statement then
updates the employee row for the given user to have a new manager.
The next example uses a similar technique to terminate an employee. You can really have fun with this
one (terminate your friends, your enemies, and so on).
Try It Out Removing Employees
Enter the following script and name the file terminate.py:
import sys
import gadfly.dbapi20
connection = gadfly.dbapi20.connect(‘pydb’, ‘db’)
cursor = connection.cursor()
employee = sys.argv[1]
# Query to find the employee ID.
query = “””
select e.empid
from user u, employee e
where username=? and u.employeeid = e.empid
“””
270

32.2 Page 312

▲back to top
Accessing Databases
cursor.execute(query,(employee,));
for row in cursor.fetchone():
if (row != None):
empid = row
# Now, modify the employee.
cursor.execute(“delete from employee where empid=?”, (empid,))
connection.commit()
cursor.close()
connection.close()
When you run this script, you need to pass the user name of the person to terminate. You should see no
output unless the script raises an error:
$ python finduser.py bunny
bunny : Bunny Wailer managed by Eric Foster-Johnson in qa
$ python terminate.py bunny
$ python finduser.py bunny
How It Works
This script uses the same techniques as the updatemgr.py script by performing an initial query to get
the employee ID for the given user name and then using this ID in a later SQL statement. With the final
SQL statement, the script deletes the employee from the employee table.
Note that this script leaves the record in the user table. Question 3 of the exercises at the end of this
chapter addresses this.
Working with Transactions and Committing the Results
Each connection, while it is engaged in an action, manages a transaction. With SQL, data is not modified
unless you commit a transaction. The database then guarantees that it will perform all of the modifications
in the transaction or none. Thus, you will not leave your database in an uncertain and potentially erro-
neous state.
To commit a transaction, call the commit method of a connection:
connection.commit()
Note that the transaction methods are part of the Connection class, not the Cursor class.
If something goes wrong, like an exception is thrown that you can handle, you should call the rollback
method to undo the effects of the incomplete transaction; this will restore the database to the state it was
in before you started the transaction, guaranteed:
connection.rollback()
The capability to roll back a transaction is very important, as you can handle errors by ensuring that the
database does not get changed. In addition, rollbacks are very useful for testing. You can insert, modify,
and delete a number of rows as part of a unit test and then roll back the transaction to undo the effects of
271

32.3 Page 313

▲back to top
Chapter 14
all the changes. This enables your unit tests to run without making any permanent changes to the
database. It also enables your unit tests to be run repeatedly, because each run resets the data.
See Chapter 12 for more on testing.
Examining Module Capabilities and Metadata
The DB API defines several globals that need to be defined at the module level. You can use these glob-
als to determine information about the database module and the features it supports. The following
table lists these globals.
Global
apilevel
paramstyle
Holds
Should hold ‘2.0’ for the DB API 2.0, or ‘1.0’ for the 1.0 API.
Defines how you can indicate the placeholders for dynamic data in your SQL
statements. The values include the following:
‘qmark’ — Use question marks, as shown in the examples in this chapter.
‘numeric’ — Use a positional number style, with ‘:1’, ‘:2’, and so on.
‘named’ — Use a colon and a name for each parameter, such as :name.
‘format’ — Use the ANSI C sprintf format codes, such as %s for a string and
%d for an integer.
‘pyformat’ — Use the Python extended format codes, such as %(name)s.
In addition, remember that pydoc is your friend. You can use pydoc to display information on modules,
such as the database modules.
With a Cursor object, you can check the definition attribute to see information about the data returned.
This information should be a set of seven-element sequences, one for each column of result data. These
sequences include the following items:
(name, type_code, display_size, internal_size, precision, scale, null_ok)
None can be used for all but the first two items. The Gadfly database, though, does not fill in the type
code, as shown in this example:
((‘FIRSTNAME’, None, None, None, None, None, None),
(‘LASTNAME’, None, None, None, None, None, None),
(‘NAME’, None, None, None, None, None, None))
Handling Errors
Errors happen. With databases, errors happen a lot. The DB API defines a number of errors that must
exist in each database module. The following table lists these exceptions.
272

32.4 Page 314

▲back to top
Accessing Databases
Exception
Warning
Error
InterfaceError
DatabaseError
DataError
OperationalError
IntegrityError
InternalError
ProgrammingError
NotSupportedError
Usage
Used for non-fatal issues. Must subclass StandardError.
Base class for errors. Must subclass StandardError.
Used for errors in the database module, not the database itself. Must
subclass Error.
Used for errors in the database. Must subclass Error.
Subclass of DatabaseError that refers to errors in the data.
Subclass of DatabaseError that refers to errors such as the loss of a con-
nection to the database. These errors are generally outside of the control
of the Python scripter.
Subclass of DatabaseError for situations that would damage the rela-
tional integrity, such as uniqueness constraints or foreign keys.
Subclass of DatabaseError that refers to errors internal to the database
module, such as a cursor no longer being active.
Subclass of DatabaseError that refers to errors such as a bad table name
and other things that can safely be blamed on you.
Subclass of DatabaseError that refers to trying to call unsupported
functionality.
Your Python scripts should handle these errors. You can get more information about them by reading
the DB API specification. See www.python.org/topics/database/ and http://www.python.org/
peps/pep-0249.html for more information.
Summary
Databases provide a handy means for storing data. You can write Python scripts that can access all the
popular databases using add-on modules. This chapter provided a whirlwind tour of SQL, the
Structured Query Language, and covered Python’s database APIs.
You also learned about the DBM modules that enable you to persist a dictionary using a variety of DBM
libraries. These modules enable you to use dictionaries and transparently persist the data.
In addition, this chapter covered the Python database APIs, which define a standard set of methods and
functions that you should expect from all database modules. This includes the following:
❑ A Connection object encapsulates a connection to the database. Use the connect function on
the database module to get a new Connection. The parameters you pass to the connect func-
tion may differ for each module.
❑ A Cursor provides the main object for interacting with a database. Use the Connection object
to get a Cursor. The Cursor enables you to execute SQL statements.
273

32.5 Page 315

▲back to top
Chapter 14
❑ You can pass dynamic data as a tuple of values to the Cursor execute method. These values
will get filling into your SQL statements, enabling you to create reusable SQL statements.
❑ After performing a query operation, the Cursor object holds the data. Use the fetchone or
fetchall methods to extract the data.
❑ After modifying the database, call commit on the Connection to commit the transaction and
save the changes. Use the rollback method to undo the changes.
❑ Call close on each Cursor when done. Call close on the Connection when done.
❑ The DB APIs include a defined set of exceptions. Your Python scripts should check for these
exceptions to handle the variety of problems that may arise.
Chapter 15 covers XML, HTML and XSL style sheets, technologies frequently used for web development.
Exercises
1. Suppose you need to write a Python script to store the pizza preferences for the workers in your
department. You need to store each person’s name along with that person’s favorite pizza top-
pings. Which technologies are most appropriate to implement this script?
a. Set up a relational database such as MySQL or Gadfly.
b. Use a DBM module such as anydbm.
c. Implement a web-service-backed rich Web application to create a buzzword-compliant
application.
2. Rewrite the following example query using table name aliases:
select employee.firstname, employee.lastname, department.name
from employee, department
where employee.dept = department.departmentid
order by employee.lastname desc
3. The terminate.py script, shown previously, removes an employee row from the employee
table; but this script is not complete. There remains a row in the user table for the same person.
Modify the terminate.py script to delete both the employee and the user table rows for
that user.
274

32.6 Page 316

▲back to top
15
Using Python for XML
XML has exploded in popularity over the past few years as a medium for storing and transmitting
structured data. Python supports the wealth of standards that have sprung up around XML, either
through standard libraries or a number of third-party libraries.
This chapter explains how to use Python to create, manipulate, and validate XML. It also covers
the standard libraries bundled with Python, as well as the popular PyXML library.
What Is XML?
The term XML is bantered around in corporate boardrooms and meetings around the world. Its
flexibility and extensibility have encouraged people to think big, advocating XML for everything
from a new, formatting-independent semantic code storage mechanism to a replacement for object
serialization. But beyond the buzzwords and hype, what is it, really? Is it a panacea for the world’s
woes? Probably not. But it is a powerful, flexible, open-standards-based method of data storage.
Its vocabulary is infinitely customizable to fit whatever kind of data you want to store. Its format
makes it human readable, while remaining easy to parse for programs. It encourages semantic
markup, rather than formatting-based markup, separating content and presentation from each
other, so that a single piece of data can be repurposed many times and displayed in many ways.
A Hierarchical Markup Language
At the core of XML is a simple hierarchical markup language. Tags are used to mark off sections
of content with different semantic meanings, and attributes are used to add metadata about the
content.

32.7 Page 317

▲back to top
Chapter 15
Following is an example of a simple XML document that could be used to describe a library:
<?xml version=”1.0”?>
<library owner=”John Q. Reader”>
<book>
<title>Sandman Volume 1: Preludes and Nocturnes</title>
<author>Neil Gaiman</author>
</book>
<book>
<title>Good Omens</title>
<author>Neil Gamain</author>
<author>Terry Pratchett</author>
</book>
<book>
<title>”Repent, Harlequin!” Said the Tick-Tock Man</title>
<author>Harlan Ellison</author>
</book>
</library>
Notice that every piece of data is wrapped in a tag and that tags are nested in a hierarchy that contains
further information about the data it wraps. Based on the previous document, you can surmise that
<author> is a child piece of information for <book>, as is <title>, and that a library has an attribute
called owner.
Unlike semantic markup languages like LaTeX, every piece of data in XML must be enclosed in tags. The
top-level tag is known as the document root, which encloses everything in the document. An XML docu-
ment can have only one document root.
Just before the document root is the XML declaration: <?xml version=”1.0”?>. This mandatory ele-
ment lets the processor know that this is an XML document. As of the writing of this book, 1.0 is the only
version of XML, so every document will use that version, and this element can just be ignored. If later
versions of XML are released, you may need to parse this element to handle the document correctly.
One problem with semantic markup is the possibility for confusion as data changes contexts. For instance,
you might want to ship a list of book titles off to a database about authors. However, without a human
to look at it, the database has no way of knowing that <title> means a book title, as opposed to an edi-
tor’s business title or an author’s honorific. This is where namespaces come in. A namespace is used
to provide a frame of reference for tags and is given a unique ID in the form of a URL, plus a prefix to
apply to tags from that namespace. For example, you might create a library namespace, with an identi-
fier of http://server.domain.tld/NameSpaces/Library and with a prefix of lib: and use that to
provide a frame of reference for the tags. With a namespace, the document would look like this:
<?xml version=”1.0”?>
<lib:library owner=”John Q. Reader”
xmlns:lib=”http://server.domain.tld/NameSpaces/Library”>
<lib:book>
<lib:title>Sandman Volume 1: Preludes and Nocturnes</lib:title>
<lib:author>Neil Gaiman</lib:author>
</lib:book>
<lib:book>
<lib:title>Good Omens</lib:title>
276

32.8 Page 318

▲back to top
Using Python for XML
<lib:author>Neil Gamain</lib:author>
<lib:author>Terry Pratchett</lib:author>
</lib:book>
<lib:book>
<lib:title>”Repent, Harlequin!” Said the Tick-Tock Man</lib:title>
<lib:author>Harlan Ellison</lib:author>
</lib:book>
</lib:library>
It’s now explicit that the title element comes from a set of elements defined by a library namespace, and
can be treated accordingly.
A namespace declaration can be added to any node in a document, and that namespace will be available
to every descendant node of that node. In most documents, all namespace declarations are applied to the
root element of the document, even if the namespace isn’t used until deeper in the document. In this
case, the namespace is applied to every tag in the document, so the namespace declaration must be on
the root element.
A document can have and use multiple namespaces. For instance, the preceding example library might
use one namespace for library information and a second one to add publisher information.
Notice the xmlns: prefix for the namespace declaration. Certain namespace prefixes are reserved for use
by XML and its associated languages, such as xml:, xsl:, and xmlns:. A namespace declaration can be
added to any node in a document, and that namespace will be available to every descendant node of
that node.
This is a fairly simple document. A more complex document might contain CDATA sections for
storing unprocessed data, comments, and processing instructions for storing information specific to
a single XML processor. For more thorough coverage of the subject, you may want to visit http://
w3cschools.org or pick up Wrox Press’s Beginning XML, 3rd Edition (0764570773) by David Hunter
et al.
A Family of Standards
XML is more than just a way to store hierarchical data. If that were all there were to it, XML would
quickly fall to more lightweight data storage methods that already exist. XML’s big strength lies in its
extensibility, and its companion standards, XSLT, XPath, Schema, and DTD languages, and a host of
other standards for querying, linking, describing, displaying, and manipulating data. Schemas and
DTDs provide a way for describing XML vocabularies and a way to validate documents. XSLT provides
a powerful transformation engine to turn one XML vocabulary into another, or into HTML, plaintext,
PDF, or a host of other formats. XPath is a query language for describing XML node sets. XSL-FO pro-
vides a way to create XML that describes the format and layout of a document for transformation to PDF
or other visual formats.
Another good thing about XML is that most of the tools for working with XML are also written in XML,
and can be manipulated using the same tools. XSLTs are written in XML, as are schemas. What this
means in practical terms is that it’s easy to use an XSLT to write another XSLT or a schema or to validate
XSLTs or schemas using schemas.
277

32.9 Page 319

▲back to top
Chapter 15
What Is a Schema/DTD?
Schemas and DTDs (Document Type Definitions) are both ways of implementing document models. A
document model is a way of describing the vocabulary and structure of a document. It’s somewhat akin
to what a DBA does when creating a database. You define the data elements that will be present in your
document, what relationship they have to one another, and how many of them you expect. In plain
English, a document model for the previous XML example might read as follows: “A library is a collec-
tion of books with a single owner. Each book has a title and at least one author.”
DTDs and schemas have different ways of expressing this document model, but they both describe the
same basic formula for the document. There are subtle differences between the two, as you shall see
later, but they have roughly the same capabilities.
What Are Document Models For?
Document models are used when you want to be able to validate content against a standard before
manipulating or processing it. They are useful whenever you will be interchanging data with an applica-
tion that may change data models unexpectedly, or when you want to constrain what a user can enter, as
in an XML-based documentation system where you will be working with hand-created XML rather than
with something from an application.
Do You Need One?
In some applications, a document model might not be needed. If you control both ends of the data
exchange and can predict what elements you are going to be receiving, a document model would be
redundant.
Document Type Definitions
A DTD is a Document Type Definition. These were the original method of expressing a document
model and are ubiquitous throughout the Internet. DTDs were originally created for describing SGML,
and the syntax has barely changed since that time, so DTDs have had quite a while to proliferate. The
W3C (the World Wide Web Consortium, or one of the groups that brings standards to the Internet) con-
tinues to express document types using DTDs, so there are DTDs for each of the HTML standards, for
Scalable Vector Graphics (SVG), MathML, and for many other useful XML vocabularies.
An Example DTD
If you were to translate the English description of the example library XML document into a DTD, it
might look something like the following:
<?xml version=”1.0”?>
<!ELEMENT library (book+)>
<!ATTLIST library
owner CDATA #REQUIRED
>
278

32.10 Page 320

▲back to top
Using Python for XML
<!ELEMENT book (title, author+)>
<!ELEMENT title (#PCDATA)>
<!ELEMENT author (#PCDATA)>
To add a reference to this DTD in the library file discussed before, you would insert a line at the top of
the file after the XML declaration that read <!DOCTYPE config SYSTEM “library.dtd”>, where
library.dtd was the path to the DTD on your system.
Let’s break this down, one step at a time. The first line, <?xml version=”1.0”?>, tells you that this is
going to be an XML document. Technically, this line is optional; DTDs don’t behave like other XML doc-
uments, but we’ll get to that later. The next line, <!ELEMENT library (book+)>, tells you that there is
an element known as library, which can have one or more child elements of the book type. The syntax
for element frequencies and grouping in DTDs is terse, but similar to that of regular expressions. The fol-
lowing table lists element frequency and element grouping operators in DTDs.
Operator
?
+
,
(list)
|
*
Definition
Specifies zero or one of the preceding elements. For instance, editor? would
mean that a book could have an optional editor element.
Specifies one or more of the preceding element. As in the previous example,
author+ means that a book has one or more authors.
Specifies a sequence of elements that must occur in that order. (title,
author+) means that the book must have a title, followed by one or more
authors, in that order.
Groups elements together. An operator applied after parentheses applies to all
elements in the group. For instance, (author, editor)+ would mean that a
document could have one or more authors and one or more editors.
Or operator. This operator permits a choice between alternatives. As an exam-
ple, (author | editor) would permit a book to have an author or an editor,
but not both.
Specifies that zero or more of the preceding element or group can appear.
(book, CD)* would permit the library to have any number of books and CDs
in it, or none at all.
The next bit is a little more complex:
<!ATTLIST library
owner CDATA #REQUIRED
>
The first line specifies that the library element has a list of attributes. Notice that the attribute list is sepa-
rate from the library element declaration itself and linked to it by the element name. If the element name
changes, the attribute list must be updated to point to the new element name. Next is a list of attributes
for the element. In this case, library has only one attribute, but the list can contain an unbounded
number of attributes. The attribute declaration has three mandatory elements: an attribute name, an
279

33 Pages 321-330

▲back to top

33.1 Page 321

▲back to top
Chapter 15
attribute type, and an attribute description. An attribute type can either be a data type, as specified by
the DTD specification, or a list of allowed values. The attribute description is used to specify the behav-
ior of the attribute. A default value can be described here, and whether the attribute is optional or
required.
DTDs Aren’t Exactly XML
As a holdover from SGML, DTDs are technically not exactly XML. Unlike schemas, they are difficult to
manipulate and validate using the same tools as XML. If you apply a document type declaration at the
beginning of a DTD, your parser will either ignore it or, more likely, generate a syntax error. Although
there is a specification for creating DTDs, there is no document model in the form of a DTD for validat-
ing the structure of a DTD. There are tools for validating DTDs, but they are distinct from the tools used
to validate XML. On the other hand, there is a document model in the form of a schema against which
schemas can be validated using standard XML tools.
Limitations of DTDs
DTDs have a number of limitations. Although it is possible to express complex structures in DTDs, it
becomes very difficult to maintain. DTDs have difficulty cleanly expressing numeric bounds on a docu-
ment model. If you wanted to specify that a library could contain no more than 100 books, you could
write <!ELEMENT library (book, book, book, book etc etc)>, but that quickly becomes an unread-
able morass of code. DTDs also make it hard to permit a number of elements in any order. If you have
three elements that you could receive in any order, you have to write <!ELEMENT book ( ( (author,
((title, publisher) | (publisher, title))) | (title, ((author, publisher) | (publisher,
author))) | (publisher, ((author, title) | (title, publisher)))))>, which is beginning to
look more like LISP (which is a language with a lot of parentheses) than XML and is far more compli-
cated than it really should be. Finally, DTDs don’t permit you to specify a pattern for data, so you can’t
express constructs such as “A telephone number should be composed of digits, dashes, and plus signs.”
Thankfully, the W3C has published a specification for a slightly more sophisticated language for describ-
ing documents, known as Schema.
Schemas
Schema was designed to address some of the limitations of DTDs and provide a more sophisticated
XML-based language for describing document models. It enables you to cleanly specify numeric models
for content, describe character data patterns using regular expressions, and express content models such
as sequences, choices, and unrestricted models.
An Example Schema
If you wanted to translate the hypothetical library model into a schema with the same information con-
tained in the DTD, you would wind up with something like the following:
<?xml version=”1.0”?>
<xs:schema xmlns:xs=”http://www.w3.org/2001/XMLSchema”>
<xs:element name=”library”>
280

33.2 Page 322

▲back to top
Using Python for XML
<xs:complexType>
<xs:sequence>
<xs:element name=”book” maxOccurs=”unbounded”>
<xs:complexType>
<xs:sequence>
<xs:element name=”title” type=”xs:string”/>
<xs:element name=”author” type=”xs:string” maxOccurs=”unbounded”/>
</xs:sequence>
</xs:complexType>
</xs:element>
</xs:sequence>
<xs:attribute name=”owner” type=”xs:string” use=”required”/>
</xs:complexType>
</xs:element>
</xs:schema>
This expresses exactly the same data model as the DTD, but some differences are immediately apparent.
Schemas Are Pure XML
To begin with, this document’s top-level node contains a namespace declaration, specifying that all
tags starting with xs: belong to the namespace identified by the URI “http://www.w3.org/2001/
XMLSchema”. What this means for practical purposes is that you now have a document model that
you can validate your schema against, using the same tools you would use to validate any other XML
document.
Schemas Are Hierarchical
Next, notice that the preceding document has a hierarchy very similar to the document it is describing.
Rather than create individual elements and link them together using references, the document model
mimics the structure of the document as closely as possible. You can also create global elements and then
reference them in a structure, but you are not required to use references; they are optional. This creates a
more intuitive structure for visualizing the form of possible documents that can be created from this
model.
Other Advantages of Schemas
Finally, schemas support attributes such as maxOccurs, which will take either a numeric value from 1
to infinity or the value unbounded, which expresses that any number of that element or grouping may
occur. Although this schema doesn’t illustrate it, schemas can express that an element matches a specific
regular expression, using the pattern attribute, and schemas can express more flexible content models
by mixing the choice and sequence content models.
Schemas Are Less Widely Supported
One of the downsides of schemas is that they haven’t been around as a standard for very long. If you
are using commercial processors and XML editors, they are more likely to support DTDs than schemas.
Schemas are slowly gaining popularity in the marketplace, but DTDs are still the language of choice, and
if you want to include other vocabularies into yours, especially from the W3C, odds are good that it’ll be
281

33.3 Page 323

▲back to top
Chapter 15
a DTD, not a schema. RSS (Rich Site Summary, which you’ll learn more about in this chapter), is speci-
fied using a DTD.
XPath
XPath is a language for describing locations and node sets within an XML document. Entire books have
been written on it. However, the basics are fairly simple. An XPath expression contains a description of a
pattern that a node must match. If the node matches, it is selected; otherwise, it is ignored. Patterns are
composed of a series of steps, either relative to a context node or absolutely defined from the document
root. An absolute path begins with a slash, a relative one does not, and each step is separated by a slash.
A step contains three parts: an axis that describes the direction to travel, a node test to select nodes along
that axis, and optional predicates, which are Boolean (true or false) tests that a node must meet. An
example step might be ancestor-or-self::book[1], where ancestor-or-self is the axis to move
along, book is the node test, and [1] is a predicate specifying to select the first node that meets all the
other conditions. If the axis is omitted, it is assumed to refer to the child axis for the current node, so
library/book[1]/author[1] would select the first author of the first book in the library.
A node test can be a function as well as a node name. For instance, book/node() will return all nodes
below the selected book node, regardless of whether they are text or elements.
The following table describes a handful of shortcuts for axes.
Shortcut
@
*
//
Meaning
Specifies the attribute axis. This is an abbreviation for attribute::
Specifies all children of the current node
Specifies any descendant of the current node. This is an abbreviation for
descendant-or-self::*//. If used at the beginning of an XPath, matches ele-
ments anywhere in the document.
For a more thorough coverage of the subject, you may want to visit http://w3schools.org or pick up
a book on XPath.
HTML as a Subset of XML
XML bears a striking resemblance to HTML. This isn’t entirely by accident. XML and HTML both sprang
from SGML and share a number of syntactic features. Earlier versions of HTML aren’t directly compati-
ble with XML, because XML requires that every tag be closed, and certain HTML tags don’t require a
closing tag, such as <br> and <img>. However, the W3C has declared the XHTML schema in an attempt
to bring the two standards in line with each other. XHTML can be manipulated using the same sets of
tools as pure XML. However, Python also comes with specialized libraries designed specifically for deal-
ing with HTML.
282

33.4 Page 324

▲back to top
Using Python for XML
The HTML DTDs
The current version of HTML is 4.01, which includes 4.01 Transitional, 4.01 Strict, and 4.01 Frameset,
specifically for dealing with frames. However, many people still use HTML 3.2, so it’s useful to be able
to parse documents from earlier DTDs.
HTMLParser
The HTMLParser class, unlike the htmllib class, is not based on an SGML parser and can be used for
both XHTML and earlier versions of HTML.
Try It Out Using HTMLParser
1. Create a sample HTML file named headings.html that contains at least one h1 tag.
2. Cut and paste the following code from the wrox.com web site into a file:
from HTMLParser import HTMLParser
class HeadingParser(HTMLParser):
inHeading = False
def handle_starttag(self, tag, attrs):
if tag == “h1”:
self.inHeading = True
print “Found a Heading 1”
def handle_data(self, data):
if self.inHeading:
print data
def handle_endtag(self, tag):
if tag ==”h1”:
self.inHeading = False
hParser = HeadingParser()
file = open(“headings.html”, “r”)
html = file.read()
file.close()
hParser.feed(html)
3. Run the code.
How It Works
The HTMLParser class defines methods, which are called when the parser finds certain types of content,
such as a beginning tag, an end tag, or a processing instruction. By default, these methods do nothing.
To parse an HTML document, a class that inherits from HTMLparser and implements the necessary
methods must be created. After a parse class has been created and instantiated, the parser is fed data
using the feed method. Data can be fed to it one line at a time or all at once.
283

33.5 Page 325

▲back to top
Chapter 15
This example class only handles tags of type <h1>. When an HTMLParser encounters a tag, the
handle_starttag method is called, and the tag name and any attached attributes are passed to it. This
handle_starttag method determines whether the tag is an <h1>. If so, it prints a message saying it
has encountered an h1 and sets a flag indicating that it is currently in an <h1>.
If text data is found, the handle_data function is called, which determines whether it is in an h1, based
on the flag. If the flag is true, the method prints the text data.
If a closing tag is encountered, the handle_endtag method is called, which determines whether the tag
that was just closed was an <h1>. If so, it prints a message, and then sets the flag to false.
htmllib
htmllib is a parser based on the sgmllib SGML parser. It defines an HTMLParser class that extends the
SGML parser class, and in turn, expects to be extended as a subclass to implement its handler methods.
It must be provided with input in string form via a method, and makes calls to methods of a formatter
object in order to produce output and it does not work with XHTML. It comes with predefined methods
for all HTML 2.0 elements and a number of 3.0 and 3.2 elements.
To parse an HTML document, the parser must override the handler methods for each HTML element.
Handler methods for tags that don’t have closing tags, such as <br>, take the form do_<tagname>.
Tags that have both a closing and opening tag have handler methods of the form start_<tagname>
and end_<tagname>.
Try It Out Using htmllib
To see how the htmllib can be used, try the following example:
from formatter import AbstractFormatter , DumbWriter
from htmllib import HTMLParser
class HeadingParser(HTMLParser):
def start_h1(self, tag):
print “found H1”
writer = DumbWriter()
formatter = AbstractFormatter (writer)
parser=HeadingParser(formatter)
parser.feed(open(‘headings.html’).read())
parser.close()
print “Finished parsing”
How It Works
The HeadingParser class implements the HTMLParser interface. As an example, it implements a han-
dler method for the h1 element. The HTMLParser interface expects a formatter object to handle format-
ted output. The formatter, in turn, expects a writer object. Fortunately, the formatter module contains
some simple default implementations of these interfaces called AbstractFormatter and DumbWriter.
When the formatter for the HeadingParser has been set, the feed method is used to feed data into the
284

33.6 Page 326

▲back to top
Using Python for XML
parser, either all at once, as this example shows, or one line at a time. Because the parser is event-driven,
either way of feeding data will have the same result. When the parser is done, it should be closed to
release any open handles.
XML Libraries Available for Python
Python comes standard with a number of libraries designed to help you work with XML. You have your
choice of several DOM (Document Object Model) implementations, an interface to the nonvalidating
Expat XML parser, and several libraries for using SAX (the Simple API for XML).
The available DOM implementations are as follows:
❑ xml.dom: A fully compliant DOM processor
❑ Xml.dom.minidom: A lightweight and much faster but not fully compliant implementation of
the DOM specification
The PyXML package is a freely available open-source collection of third-party libraries to process
XML with Python. Documentation and downloads are available from Sourceforge at http://pyxml
.sourceforge.net/. It contains a number of useful utility libraries for dealing with XML, such as a
pretty printer for outputting easy-to-read XML, as well as some additional parsers. The full list includes
the following:
❑ xmlproc: A validating XML parser
❑ Expat: A fast nonvalidating parser
❑ sgmlop: A C helper module that can speed up xmllib.py and sgmllib.py by a factor of 5
❑ PySAX: SAX1 and SAX2 libraries with drivers for most of the parsers
❑ 4DOM: A fully compliant DOM Level 2 implementation
❑ javadom: An adapter from Java DOM implementations to the standard Python DOM binding
❑ pulldom: A DOM implementation that supports lazy instantiation of nodes
❑ marshall: Enables Python objects to be serialized to XML
If you don’t already have PyXML installed in your system, please install it now. You will need it to com-
plete examples later in this chapter. Detailed installation instructions are available with the download.
Validating XML Using Python
Document models are wonderful things for describing the kind of data that’s expected, but they aren’t
very useful if the document isn’t verified against it. Surprisingly, many XML processors don’t do this
automatically; you are expected to supply your own code for verifying the XML. Luckily, there are
libraries that do just that.
285

33.7 Page 327

▲back to top
Chapter 15
What Is Validation?
Validation is the process of verifying that a document matches the document model that has been speci-
fied for it. It verifies that tag names match the vocabulary specified, that attributes match the enumera-
tion or pattern that has been specified for them, and so on.
Well-Formedness versus Validation
All of the XML parsers available will check documents for well formedness. This guarantees that any
documents being processed are complete, that every tag opened has been closed, that all tags are well
formed (that is, those that need to have matching opening and closing tags have these matching sets),
and so on.
If these properties are all satisfied, then the document is well-formed. But validation involves more
than that.
Available Tools
Only one of the parsers available for Python today actually validates against a document model, and
that is xmlproc. Xmlproc is available as part of the PyXML package; it is not a part of the core Python
libraries. To continue with the XML examples in this chapter, you will need to download and install
the pyxml package.
Try It Out Validation Using xmlproc
1. Change the line reading <library owner=”John Q. Reader”> to the following line in your
example XML library and save it to a file called library.xml:
<library owner=”John Q. Reader” xmlns:xsi=”http://www.w3.org/2001/XMLSchema-
instance” xsi:noNameSpaceSchemaLocation=”library.xsd”>
2. Save the example schema from earlier in the chapter to a file called library.xsd.
3. Download and install PyXML on your system if you haven’t already. The following code has
been tested using PyXML 0.8.4
4. Place the following code into a file called validator.py:
#!/usr/bin/python
from xml.parsers.xmlproc import xmlval
class docErrorHandler(xmlval.ErrorHandler):
def warning(self, message):
print message
def error(self, message):
print message
def fatal(self, message):
print message
286

33.8 Page 328

▲back to top
Using Python for XML
parser = xmlval.XMLValidator()
parser.set_error_handler(docErrorHandler(parser))
parser.parse_resource(“library.xml”)
5. From the command line, run python validator.py.
How It Works
Including the line <library owner=”John Q. Reader” xmlns:xsi=”http://www.w3.org/2001/
XMLSchema-instance” xsi:noNameSpaceSchemaLocation=”library.xsd”> in the file registers the
prefix xsi to point to the namespace and then uses the noNameSpaceSchemaLocation attribute from
that namespace to specify that this document uses the library.xsd schema as a content model.
The xmlval module from xmlproc is a module for doing document validation. XMLValidator is a vali-
dating parser. It can also use an external parser such as Expat and validate after the external parser has
parsed the document.
The XMLValidator class creates four classes: Application, ErrorHandler, PubIdResolver, and
InputSourceFactory. An Application object handles document events, and an ErrorHandler han-
dles document parse errors. In a full-fledged XML application, you would implement the Application
interface as described later in the section on SAX, but for pure validation, only the ErrorHandler inter-
face needs to be implemented, so that any validation errors that might occur can be printed.
The ErrorHandler has three methods that will need to be implemented: the warning, error, and
fatal methods. As the names might indicate, warning handles all warnings, error handles nonfatal
errors, and fatal handles fatal errors. For a simple validator, it is only necessary to print any warnings,
errors, or fatal errors that may occur, so each of these simply prints the error message.
After the ErrorHandler interface has been implemented, the validating parser needs to be instanti-
ated, and the ErrorHandler needs to be registered with it, using parser.set_error_handler
(docErrorHandler(parser)). The __init__ method for an ErrorHandler requires a locator
parameter to locate error events, which needs to be of the Parser type.
When everything has been configured, the parse method takes a filename as an argument and parses it,
using the ErrorHandler as a callback interface when parsing and validation errors are found.
What Is SAX?
When parsing XML, you have your choice of two different types of parsers: SAX and DOM. SAX stands
for the Simple API for XML. Originally only implemented for Java, it was added to Python as of version
2.0. It is a stream-based, event-driven parser. The events are known as document events, and a docu-
ment event might be the start of an element, the end of an element, encountering a text node, or encoun-
tering a comment. For example, the following simple document:
<?xml version=”1.0”?>
<author>
<name>Ursula K. LeGuin</name>
</author>
287

33.9 Page 329

▲back to top
Chapter 15
might fire the following events:
start document
start element: author
start element: name
characters: Ursula K. LeGuin
end element: name
end element: author
end document
Whenever a document event occurs, the parser fires an event for the calling application to handle. More
precisely, it fires an event for the calling application’s Content Handler object to handle. Content
Handlers are objects that implement a known interface specified by the SAX API from which the parser
can call methods. In the preceding example, the parser would call the startDocument method of the
content handler, followed by two calls to the startElement method, and so on.
Stream-based
When parsing a document with SAX, the document is read and parsed in the order in which it appears.
The parser opens the file or other datasource (such as a URL) as a stream of data (which means that it
doesn’t have to have it all at once) and then fires events whenever an element is encountered.
Because the parser does not wait for the whole document to load before beginning parsing, SAX can
parse documents very soon after it starts reading the document. However, because SAX does not read
the whole document, it may process a partial document before discovering that the document is badly
formed. SAX-based applications should implement error-checking for such conditions.
Event-driven
When working with SAX, document events are handled by event handlers, similar to a GUI. You declare
callback functions for specific types of document events, which are then passed to the parser and called
when a document event occurs that matches the callback function.
What Is DOM?
At the heart of DOM lies the document object. This is a tree-based representation of the XML document.
Tree-based models are a natural fit for XML’s hierarchical structure, making this a very intuitive way of
working with XML. Each element in the tree is called a Node object, and it may have attributes, child
nodes, text, and so on, all of which are also objects stored in the tree. DOM objects have a number of
methods for creating and adding nodes, for finding nodes of a specific type or name, and for reordering
or deleting nodes.
In-memory Access
The major difference between SAX and DOM is the latter’s ability to store the entire document in mem-
ory and manipulate and search it as a tree, rather than force you to parse the document repeatedly, or
force you to build your own in-memory representation of the document. The document is parsed once,
288

33.10 Page 330

▲back to top
Using Python for XML
and then nodes can be added, removed, or changed in memory and then written back out to a file when
the program is finished.
Why Use SAX or DOM
Although either SAX or DOM can do almost anything you might want to do with XML, there are rea-
sons why you might want to use one over the other for a given task. For instance, if you are working on
an application in which you will be modifying an XML document repeatedly based on user input, you
might want the convenient random access capabilities of DOM. On the other hand, if you’re building an
application that needs to process a stream of XML quickly with minimal overhead, SAX might be a bet-
ter choice for you. Following are some of the advantages and disadvantages you might want to be aware
of when architecting your application to use XML.
Capability Trade-Offs
DOM is architected with random access in mind. It provides a tree that can be manipulated at runtime
and needs to be loaded into memory only once. SAX is stream-based so data comes in as a stream one
character after the next, but the document isn’t seen in it’s entirety before it starts getting processed; there-
fore, if you want to randomly access data, you have to either build a partial tree of the document in mem-
ory based on document events, or reparse the document every time you want a different piece of data.
Most people find the object-oriented behavior of DOM very intuitive and easy to learn. The event-driven
model of SAX is more similar to functional programming and can be more challenging to get up to
speed on.
Memory Considerations
If you are working in a memory-limited environment, DOM is probably not the right choice. Even on a
fairly high-end system, constructing a DOM tree for a 2 or 3 MB XML document can bring the computer
grinding to a halt while it processes. Because SAX treats the document as a stream, it never loads the
whole document into memory, so it is preferable if you are memory constrained or working with very
large documents.
Speed Considerations
Using DOM requires a great deal of up-front processing time while the document tree is being built, but
once the tree is built DOM allows for much faster searching and manipulation of nodes because the
entire document is in memory. SAX is somewhat fast for searching documents, but not as efficient for
their manipulation. However, for document transformations, SAX is considered to be the parser of
choice because the event-driven model is fast and very compatible with how XSLT works.
SAX and DOM Parsers Available for Python
The following Python SAX and DOM parsers are available: PyXML, xml.sax, and xml.dom.minidom.
They each behave a bit differently, so here is an overview of each of them.
289

34 Pages 331-340

▲back to top

34.1 Page 331

▲back to top
Chapter 15
PyXML
PyXML contains the following parsers:
Name
xmlproc
Expat
PySAX
4DOM
javadom
pulldom
Description
A validating XML parser
A fast nonvalidating parser
SAX1 and SAX2 libraries with drivers for most of the parsers
A fully compliant DOM Level 2 implementation
An adapter from Java DOM implementations to the standard Python DOM
binding
A DOM implementation that supports lazy instantiation of nodes
xml.sax
xml.sax is the built-in SAX package that comes with Python. It uses the Expat nonvalidating parser by
default but can be passed a list of parser instances that can change this behavior.
xml.dom.minidom
xml.dom.minidom is a lightweight DOM implementation, designed to be simpler and smaller than a
full DOM implementation.
Try It Out Working with XML Using DOM
1. If you haven’t already, save the example XML file from the beginning of this chapter in a file
called library.xml.
2. Either type in or get the following code from this book’s web site, and save it to a file called
xml_minidom.py:
from xml.dom.minidom import parse
import xml.dom.minidom
def printLibrary(library):
books = myLibrary.getElementsByTagName(“book”)
for book in books:
print “*****Book*****”
print “Title: %s” % book.getElementsByTagName(“title”)[0].childNodes[0].data
for author in book.getElementsByTagName(“author”):
print “Author: %s” % author.childNodes[0].data
# open an XML file and parse it into a DOM
myDoc = parse(‘library.xml’)
myLibrary = myDoc.getElementsByTagName(“library”)[0]
290

34.2 Page 332

▲back to top
Using Python for XML
#Get all the book elements in the library
books = myLibrary.getElementsByTagName(“book”)
#Print each book’s title and author(s)
printLibrary(myLibrary)
#Insert a new book in the library
newBook = myDoc.createElement(“book”)
newBookTitle = myDoc.createElement(“title”)
titleText = myDoc.createTextNode(“Beginning Python”)
newBookTitle.appendChild(titleText)
newBook.appendChild(newBookTitle)
newBookAuthor = myDoc.createElement(“author”)
authorName = myDoc.createTextNode(“Peter Norton, et al”)
newBookAuthor.appendChild(authorName)
newBook.appendChild(newBookAuthor)
myLibrary.appendChild(newBook)
print “Added a new book!”
printLibrary(myLibrary)
#Remove a book from the library
#Find ellison book
for book in myLibrary.getElementsByTagName(“book”):
for author in book.getElementsByTagName(“author”):
if author.childNodes[0].data.find(“Ellison”) != -1:
removedBook= myLibrary.removeChild(book)
removedBook.unlink()
print “Removed a book.”
printLibrary(myLibrary)
#Write back to the library file
lib = open(“library.xml”, ‘w’)
lib.write(myDoc.toprettyxml(“ “))
lib.close()
3. Run the file with python xml_minidom.py.
How It Works
To create a DOM, the document needs to be parsed into a document tree. This is accomplished
by calling the parse method from xml.dom.minidom. This method returns a Document object,
which contains methods for querying for child nodes, getting all nodes in the document of a certain
name, and creating new nodes, among other things. The getElementsByTagName method returns
a list of Node objects whose names match the argument, which is used to extract the root node of
the document, the <library> node. The print method uses getElementsByTagName again, and
then for each book node, prints the title and author. Nodes with text that follows them are con-
sidered to have a single child node, and the text is stored in the data attribute of that node, so
book.getElementsByTagName(“title”)[0].childNodes[0].data simply retrieves the text
node below the <title> element and returns its data as a string.
291

34.3 Page 333

▲back to top
Chapter 15
Constructing a new node in DOM requires creating a new node as a piece of the Document object,
adding all necessary attributes and child nodes, and then attaching it to the correct node in the docu-
ment tree. The createElement(tagName) method of the Document object creates a new node with a
tag name set to whatever argument has been passed in. Adding text nodes is accomplished almost the
same way, with a call to createTextNode(string). When all the nodes have been created, the struc-
ture is created by calling the appendChild method of the node to which the newly created node will be
attached. Node also has a method called insertBefore(newChild, refChild) for inserting nodes in
an arbitrary location in the list of child nodes, and replaceChild(newChild, oldChild) to replace
one node with another.
Removing nodes requires first getting a reference to the node being removed and then a call to
removeChild(childNode). After the child has been removed, it’s advisable to call unlink() on it to
force garbage collection for that node and any children that may still be attached. This method is specific
to the minidom implementation and is not available in xml.dom.
Finally, having made all these changes to the document, it would be useful to be able to write the
DOM back to the file from which it came. A utility method is included with xml.dom.minidom called
toprettyxml, which takes two optional arguments: an indentation string and a newline character. If not
specified, these default to a tabulator and \\n, respectively. This utility prints a DOM as nicely indented
XML and is just the thing for printing back to the file.
Try It Out Working with XML Using SAX
This example will show you how you can explore a document with SAX.
#!/usr/bin/python
from xml.sax
import make_parser
from xml.sax.handler import ContentHandler
#begin bookHandler
class bookHandler(ContentHandler):
inAuthor = False
inTitle = False
def startElement(self, name, attributes):
if name == “book”:
print “*****Book*****”
if name == “title”:
self.inTitle = True
print “Title: “,
if name == “author”:
self.inAuthor = True
print “Author: “,
def endElement(self, name):
if name == “title”:
self.inTitle = False
if name == “author”:
self.inAuthor = False
292

34.4 Page 334

▲back to top
Using Python for XML
def characters(self, content):
if self.inTitle or self.inAuthor:
print content
#end bookHandler
parser = make_parser()
parser.setContentHandler(bookHandler())
parser.parse(“library.xml”)
How It Works
The xml.sax parser uses Handler objects to deal with events that occur during the parsing of a docu-
ment. A handler may be a ContentHandler, a DTDHandler, an EntityResolver for handling entity
references, or an ErrorHandler. A SAX application must implement handler classes, which conform to
these interfaces and then set the handlers for the parser.
The ContentHandler interface contains methods that are triggered by document events, such as the
start and end of elements and character data. When parsing character data, the parser has the option of
returning it in one large block or several smaller whitespace-separated blocks, so the characters
method may be called repeatedly for a single block of text.
The make_parser method creates a new parser object and returns it. The parser object created will be
of the first parser type the system finds. The make_parser method can also take an optional argument
consisting of a list of parsers to use, which must all implement the make_parser method. If a list is sup-
plied, those parsers will be tried before the default list of parsers.
Intro to XSLT
XSLT stands for Extensible Stylesheet Language Transformations. Used for transforming XML into out-
put formats such as HTML, it is a procedural, template-driven language.
XSLT Is XML
Like a Schema, XSLT is defined in terms of XML, and it’s being used to supplement the capabilities of
XML. The XSLT namespace is “http://www.w3.org/1999/XSL/Transform”, which specifies the
structure and syntax of the language. XSLT can be validated, like all other XML.
Transformation and Formatting Language
XSLT is used to transform one XML syntax into another or into any other text-based format. It is often
used to transform XML into HTML in preparation for web presentation or a custom document model
into XSL-FO for conversion into PDF.
Functional, Template-Driven
XSLT is a functional language, much like LISP. The XSLT programmer declares a series of templates,
which are functions triggered when a node in the document matches an XPath expression. The
293

34.5 Page 335

▲back to top
Chapter 15
programmer cannot guarantee the order of execution, so each function must stand on its own and make
no assumptions about the results of other functions.
Using Python to Transform XML Using XSLT
Python doesn’t directly supply a way to create an XSLT, unfortunately. To transform XML documents,
an XSLT must be created, and then it can be applied via Python to the XML.
In addition, Python’s core libraries don’t supply a method for transforming XML via XSLT, but a couple
of different options are available from other libraries. Fourthought, Inc., offers an XSLT engine as part
of its freely available 4Suite package. There are also Python bindings for the widely popular libxslt
C library.
The following example uses the latest version of the 4Suite library, which, as of this writing, is 1.0a4.
If you don’t have the 4Suite library installed, please download it from http://4suite.org/index.
xhtml. You will need it to complete the following exercises.
Try It Out Transforming XML with XSLT
1. If you haven’t already, save the example XML file from the beginning of this chapter to a file
called library.xml.
2. Cut and paste the following XSL from the wrox.com web site into a file called
HTMLLibrary.xsl:
<?xml version=”1.0”?>
<xsl:stylesheet version=”1.0”
xmlns:xsl=”http://www.w3.org/1999/XSL/Transform”>
<xsl:template match=”/library”>
<html>
<head>
<xsl:value-of select=”@owner”/>’s Library
</head>
<body>
<h1><xsl:value-of select=”@owner”/>’s Library</h1>
<xsl:apply-templates/>
</body>
</html>
</xsl:template>
<xsl:template match=”book”>
<xsl:apply-templates/>
<br/>
</xsl:template>
<xsl:template match=”title”>
<b><xsl:value-of select=”.”/></b>
</xsl:template>
<xsl:template match=”author[1]”>
by <xsl:value-of select=”.”/>
</xsl:template>
294

34.6 Page 336

▲back to top
Using Python for XML
<xsl:template match=”author”>
, <xsl:value-of select=”.”/>
</xsl:template>
</xsl:stylesheet>
3. Either type this or download it from the web site for this book and save it to a file called
transformLibrary.py:
#!/usr/bin/python
from Ft.Xml import InputSource
from Ft.Xml.Xslt.Processor import Processor
#Open the XML and stylesheet as streams
xml = open(‘library.xml’)
xsl = open(‘HTMLLibrary.xsl’)
#Parse the streams and build input sources from them
parsedxml = InputSource.DefaultFactory.fromStream(xml , “library.xml”)
parsedxsl = InputSource.DefaultFactory.fromStream(xsl, “HTMLLibrary.xsl”)
#Create a new processor and attach stylesheet, then transform XML
processor = Processor()
processor.appendStylesheet(parsedxsl)
HTML = processor.run(parsedxml)
#Write HTML out to a file
output = open(“library.html”, ‘w’)
output.write(HTML)
output.close
4. Run python transformLibrary.py from the command line. This will create library.html.
5. Open library.html in a browser or text editor and look at the resulting web page.
How It Works
The first line of the stylesheet, <xsl:stylesheet version=”1.0” xmlns:xsl=”http://www.w3.org/
1999/XSL/Transform”>, declares the document to be an XSL stylesheet that conforms to the specifica-
tion at http://www.w3.org/1999/XSL/Transform and associates the xsl: prefix with that URI.
Each xsl:template element is triggered whenever a node that matches a certain XPath is encountered.
For instance, <xsl:template match=”author[1]”> is triggered every time an <author> node is
found that is the first in a list of authors.
XML tags that don’t start with the xsl: prefix are not parsed and are written to the output, as is plain-
text in the body of a template. Therefore, the following template returns the skeleton of an HTML page,
with a <head>, <body>, and an <h1> with the title of the library:
<xsl:template match=”/library”>
<html>
<head>
<xsl:value-of select=”@owner”/>’s Library
</head>
295

34.7 Page 337

▲back to top
Chapter 15
<body>
<h1><xsl:value-of select=”@owner”/>’s Library</h1>
<xsl:apply-templates/>
</body>
</html>
</xsl:template>
The xsl:value-of element returns the text value of an XPath expression. If the XPath selects more than
one node, each node is converted to text according to XSL’s conversion rules and then concatenated and
returned. <xsl:value-of select=”@owner”/>, for instance, will return the text value of the owner
attribute on the current context node, which in this case is the <library> node. Because the attribute is
a string, it will return John Q. Reader unchanged.
The xsl:apply-templates element is where the power of XSL occurs. When called with no argu-
ments, it selects all child nodes of the current node, triggers the templates that match each of them, and
then inserts the resulting nodes into the results of the current template. It can also be called with a
select argument in the form of an XPath that will apply templates only to the nodes selected.
Putting It All Together: Working with RSS
Now that you’ve learned how to work with XML in Python, it’s time for a real-world example that
shows you how you might want to use these modules to create your own RSS feed and how to take an
RSS feed and turn it into a web page for reading.
RSS Overview and Vocabulary
Depending on who you ask, RSS stands for Really Simple Syndication, or Rich Site Summary, or RDF
Site Summary. Regardless of what you want to call it, RSS is an XML-based format for syndicating con-
tent from news sites, blogs, and anyone else who wants to share discrete chunks of information over
time. RSS’s usefulness lies in the ease with which content can be aggregated and republished. RSS makes
it possible to read all your favorite authors’ blogs on a single web page, or, for example, to see every arti-
cle from a news agency containing the word “Tanzania” first thing every day.
RSS originally started as part of Netscape’s news portal and has released several versions since then.
After Netscape dropped development on RSS and released it to the public, two different groups began
developing along what they each felt was the correct path for RSS to take. At present, one group has
released a format they are calling RSS 1.0, and the other has released a format they are calling 2.0,
despite the fact that 2.0 is not a successor to 1.0. At this point, RSS refers to seven different and some-
times incompatible formats, which can lead to a great deal of confusion for the newcomer to RSS.
Making Sense of It All
The following table summarizes the existing versions of RSS. As a content producer, the choice of ver-
sion is fairly simple, but an RSS aggregator, which takes content from multiple sites and displays it in a
single feed, has to handle all seven formats.
296

34.8 Page 338

▲back to top
Using Python for XML
Version
0.90.
0.91.
0.92.
0.93.
0.94.
1.0.
2.0.
Owner
Netscape
Userland
Userland
Userland
Userland
RSS-DEV Working
Group
Userland
Notes
The original format. Netscape decided this format
was overly complex and began work on 0.91 before
dropping RSS development. Obsolete by 1.0.
Partially developed by Netscape before being picked
up by Userland. Incredibly simple and still very
popular, although officially obsolete by 2.0.
More complex than .91. Obsolete by 2.0.
More complex than .91. Obsolete by 2.0.
More complex than .91. Obsolete by 2.0.
RDF-based. Stable, but with modules still under
development. Successor to 0.90.
Does not use RDF. Successor to 0.94. Stable, with
modules still under development.
RSS Vocabulary
RSS feeds are composed of documents called channels, which are feeds from a single web site. Each
channel has a title, a link to the originating web site, a description, and a language. It also contains one
or more items, which contain the actual content of the feed. An item must also have a title, a description,
and a unique link back to the originating web site.
RSS 1.0 adds optional elements for richer content syndication, such as images, and a text input element
for submitting information back to the parent site.
An RSS DTD
The DTD Netscape released for RSS 0.91 is freely available at http://my.netscape.com/publish/
formats/rss-0.91.dtd. It’s the simplest of the RSS document models, and it’s the one that will be
used in the RSS examples in this chapter. To use it, include a DTD reference to that URI at the top of
your XML file.
A Real-World Problem
With the increasing popularity of blogging, fueled by easy-to-use tools like Blogger and Moveable Type,
it would be nice to be able to syndicate your blog out, so that other people could aggregate your posts
into their portal pages. To do this, you’d like a script that reads your blogs and turns them into an RSS
feed to which other people can then subscribe.
297

34.9 Page 339

▲back to top
Chapter 15
Try It Out Creating an RSS Feed
1. Either download the following from the web site for this book, or type it into a file called
myblog.html:
<html>
<head>
<title>My Daily Blog</title>
</head>
<body>
<h1>My Daily Blog</h1>
<p>This blog contains musings and news</p>
<div class=”story”>
<a name=”autogen4”/>
<h2>Really Big Corp to buy Slightly Smaller Corp</h2>
<div class=”date”>10:00 PM, 1/1/2005</div>
<span class=”content”>
Really Big Corp announced it’s intent today to buy Slightly Smaller Corp. Slightly
Smaller Corp is the world’s foremost producer of lime green widgets. This will
clearly impact the world’s widget supply.
</span>
</div>
<div class=”story”>
<a name=”autogen3”/>
<h2>Python Code now easier than ever</h2>
<div class=”date”>6:00 PM, 1/1/2005</div>
<span class=”content”>
Writing Python has become easier than ever with the release of the new book,
Beginning Python, from Wrox Press.
</span>
</div>
<div class=”story”>
<a name=”autogen2”/>
<h2>Really Famous Author to speak at quirky little bookstore</h2>
<div class=”date”>10:00 AM, 1/1/2005</div>
<span class=”content”>
A really good author will be speaking tomorrow night at a charming little bookstore
in my home town. It’s a can’t miss event.
</span>
</div>
<div class=”story”>
<a name=”autogen1”/>
<h2>Blogging more popular than ever</h2>
<div class=”date”>2:00 AM, 1/1/2005</div>
<span class=”content”>
More people are blogging now than ever before, leading to an explosion of opinions
and timely content on the internet. It’s hard to say if this is good or bad, but
it’s certainly a new method of communication.
</span>
</div>
</body>
</html>
298

34.10 Page 340

▲back to top
Using Python for XML
2. Type or download the following XSLT from the web site for this book into a file called
HTML2RSS.xsl:
<?xml version=”1.0”?>
<xsl:stylesheet version=”1.0”
xmlns:xsl=”http://www.w3.org/1999/XSL/Transform”>
<xsl:output method=”xml” doctype-
system=”http://my.netscape.com/publish/formats/rss-0.91.dtd”
doctype-public=”-//Netscape Communications//DTD RSS 0.91//EN”/>
<xsl:template match=”/”>
<rss version=”0.91”>
<channel>
<xsl:apply-templates select=”html/head/title”/>
<link>http://server.mydomain.tld</link>
<description>This is my blog. There are others like it, but this one is
mine.</description>
<xsl:apply-templates select=”html/body/div[@class=’story’]”/>
</channel>
</rss>
</xsl:template>
<xsl:template match=”head/title”>
<title>
<xsl:apply-templates/>
</title>
</xsl:template>
<xsl:template match=”div[@class=’story’]”>
<item>
<xsl:apply-templates/>
<link>
http://server.mydomain.tld/myblog.html#<xsl:value-of select=”a/@name”/>
</link>
</item>
</xsl:template>
<xsl:template match=”h2”>
<title><xsl:apply-templates/></title>
</xsl:template>
<xsl:template match=”div[@class=’story’]/span[@class=’content’]”>
<description>
<xsl:apply-templates/>
</description>
</xsl:template>
<xsl:template match=”div[@class=’date’]”/>
</xsl:stylesheet>
299

35 Pages 341-350

▲back to top

35.1 Page 341

▲back to top
Chapter 15
3. The same instructions go for this file – either type it in, or download it from the web site for the
book into a file called HTML2RSS.py:
#!/usr/bin/python
from Ft.Xml import InputSource
from Ft.Xml.Xslt.Processor import Processor
from xml.parsers.xmlproc import xmlval
class docErrorHandler(xmlval.ErrorHandler):
def warning(self, message):
print message
def error(self, message):
print message
def fatal(self, message):
print message
#Open the stylesheet as a stream
html = open(‘myblog.html’)
xsl = open(‘HTML2RSS.xsl’)
#Parse the streams and build input sources from them
parsedxml = InputSource.DefaultFactory.fromStream(html, “myblog.html”)
parsedxsl = InputSource.DefaultFactory.fromStream(xsl, “HTML2RSS.xsl”)
#Create a new processor and attach stylesheet, then transform XML
processor = Processor()
processor.appendStylesheet(parsedxsl)
HTML = processor.run(parsedxml)
#Write RSS out to a file
output = open(“rssfeed.xml”, ‘w’)
output.write(HTML)
output.close
#validate the RSS produced
parser=xmlval.XMLValidator()
parser.set_error_handler(docErrorHandler(parser))
parser.parse_resource(“rssfeed.xml”)
How It Works
Similarly to the XSLT example, this example opens a document and an XSLT, creates a processor, and
uses the processor to run the XSLT on the source document. This is slightly different, however. The doc-
ument being transformed is HTML. However, any XHTML-compliant document can be transformed,
just like any other kind of XML.
Creating the Document
There’s an additional line in the XSL this time, one that reads <xsl:output method=”xml” doctype-
system=”http://my.netscape.com/publish/formats/rss-0.91.dtd” doctype-public=”-
//Netscape Communications//DTD RSS 0.91//EN”/> . The xsl:output element is used to control
the format of the output document. It can be used to output HTML instead of XML, and it can also be
used to set the doctype of the resulting document. In this case, the doctype is being set to
300

35.2 Page 342

▲back to top
Using Python for XML
http://my.netscape.com/publish/formats/rss-0.91.dtd, which means the document can be
validated after it’s produced to make sure that the resulting RSS is correct.
The stylesheet selects the title of the web page as the title of the RSS feed and creates a description for it,
and then pulls story content from the body of the document. To make the example less complex, the
HTML has been marked up with div tags to separate stories, but that isn’t strictly necessary.
Checking It Against the DTD
As in the validation example, a validating parser is being created, and an ErrorHandler class is being
created. The result document already has the document type set, so all that’s required to validate it is to
parse it with a validating parser and then print any errors encountered with the validation.
Another Real-World Problem
Now that you’ve started publishing your own content, it would be nice to look at everyone else’s while
you’re at it. If you built your own aggregator, then you could create a personalized web page of the
news feeds you like to read.
Try It Out Creating An Aggregator
1. Type or download the following into a file called RSS2HTML.xsl:
<?xml version=”1.0”?>
<xsl:stylesheet version=”1.0”
xmlns:xsl=”http://www.w3.org/1999/XSL/Transform”>
<xsl:template match=”/”>
<html>
<head>
<title>
My Personal News Feed
</title>
</head>
<body>
<h1>My Personal News Feed</h1>
<xsl:apply-templates select=”//channel/item[1]”/>
</body>
</html>
</xsl:template>
<xsl:template match=”item”>
<xsl:apply-templates/>
</xsl:template>
<xsl:template match=”title”>
<h2><xsl:value-of select=”.”/></h2>
</xsl:template>
<xsl:template match=”description”>
<xsl:apply-templates/>
</xsl:template>
301

35.3 Page 343

▲back to top
Chapter 15
<xsl:template match=”link”>
<a>
<xsl:attribute name=”href”>
<xsl:value-of select=”.”/>
</xsl:attribute>
<xsl:value-of select=”.”/>
</a>
</xsl:template>
</xsl:stylesheet>
2. Download or type the following to a file called RSS2HTML.py:
#!/usr/bin/python
from Ft.Xml import InputSource
from Ft.Xml.Xslt.Processor import Processor
#Open the stylesheet as a stream
xsl = open(‘RSS2HTML.xsl’)
#Parse the streams and build input sources from them
parsedxml =
InputSource.DefaultFactory.fromUri(“http://www.newscientist.com/feed.ns?index=mars-
rovers&type=xml “)
parsedxsl = InputSource.DefaultFactory.fromStream(xsl, “RSS2HTML.xsl”)
#Create a new processor and attach stylesheet, then transform XML
processor = Processor()
processor.appendStylesheet(parsedxsl)
HTML = processor.run(parsedxml)
#Write HTML out to a file
output = open(“aggregator.html”, ‘w’)
output.write(HTML)
output.close
3. Run python RSS2HTML.py. Then open aggregator.html in a browser or text editor and view
the resulting HTML.
How It Works
The example RSS feed is a 0.91 RSS feed, for simplicity’s sake. Much like the example for using XSLTs,
the Python script opens and parses a feed, parses the XSL to be applied, and then creates a processor and
associates the stylesheet with it and processes the contents of the feed. In this case, however, the script is
processing a feed from a URL using InputSource.DefaultFactory.fromUri. Fortunately, the mod-
ule takes care of all of the details of getting the data from the remote server. You simply need to specify
the URL for the feed and have a working Internet connection.
302

35.4 Page 344

▲back to top
Using Python for XML
Summary
In this chapter, you’ve learned the following:
❑ How to parse XML using both SAX and DOM
❑ How to validate XML using xmlproc
❑ How to transform XML with XSLT
❑ How to parse HTML using either HTMLParser or htmllib
❑ How to manipulate RSS using Python
In Chapter 16, you learn more about network programming and e-mail. Before proceeding, however, try
the exercises that follow to test your understanding of the material covered in this chapter. You can find
the solutions to these exercises in Appendix A.
Exercises
1. Given the following configuration file for a Python application, write some code to extract the
configuration information using a DOM parser:
<?xml version=”1.0”?>
<!DOCTYPE config SYSTEM “configfile.dtd”>
<config>
<utilitydirectory>/usr/bin</utilitydirectory>
<utility>grep</utility>
<mode>recursive</mode>
</config>
2. Given the following DTD, named configfile.dtd, write a Python script to validate the previ-
ous configuration file:
<!ELEMENT config (utilitydirectory, utility, mode)>
<!ELEMENT utilitydirectory
(#PCDATA)*>
<!ELEMENT utility
(#PCDATA)*>
<!ELEMENT mode (#PCDATA)*>
3. Use SAX to extract configuration information from the preceding config file instead of DOM.
303

35.5 Page 345

▲back to top

35.6 Page 346

▲back to top
16
Network Programming
For more than a decade at the time this book is being written, one of the main reasons driving
the purchase of personal computers is the desire to get online: to connect in various ways to other
computers throughout the world. Network connectivity — specifically, Internet connectivity — is
the “killer app” for personal computing, the feature that got a computer-illiterate general popula-
tion to start learning about and buying personal computers en masse.
Without networking, you can do amazing things with a computer, but your audience is limited
to the people who can come over to look at your screen or who can read the printouts or load the
CD’s and floppy disks you distribute. Connect the same computer to the Internet and you can
communicate across town or across the world.
The Internet’s architecture supports an unlimited number of applications, but it boasts two
killer apps of its own — two applications that people get online just to use. One is, of course, the
incredibly popular World Wide Web; which is covered in Chapter 21, “Web Applications and
Web Services.”
The Internet’s other killer app is e-mail, which is covered in depth in this chapter. Here, you’ll use
standard Python libraries to write applications that compose, send, and receive e-mail. Then, for
those who dream of writing their own killer app, you’ll write some programs that use the Internet
to send and receive data in custom formats.
Try It Out Sending Some E-mail
Jamie Zawinski, one of the original Netscape programmers, has famously remarked, “Every pro-
gram attempts to expand until it can read mail.” This may be true (it certainly was of the Netscape
browser even early on when he worked on it), but long before your program becomes a mail reader,
you’ll probably find that you need to make it send some mail. Mail readers are typically end-user
applications, but nearly any kind of application can have a reason to send mail: monitoring soft-
ware, automation scripts, web applications, even games. E-mail is the time-honored way of sending
automatic notifications, and automatic notifications can happen in a wide variety of contexts.
Python provides a sophisticated set of classes for constructing e-mail messages, which are covered
a bit later. Actually, an e-mail message is just a string in a predefined format. All you need to send

35.7 Page 347

▲back to top
Chapter 16
an e-mail message is a string in that format, an address to send the mail to, and Python’s smtplib mod-
ule. Here’s a very simple Python session that sends out a bare-bones e-mail message:
>>> fromAddress = ‘sender@example.com’
>>> toAddress = ‘me@my.domain’
>>> msg = “Subject: Hello\\n\\nThis is the body of the message.”
>>> import smtplib
>>> server = smtplib.SMTP(“localhost”, 25)
>>> server.sendmail(fromAddress, toAddress, msg)
{}
smtplib takes its name from SMTP, the Simple Mail Transport Protocol. That’s the protocol, or stan-
dard, defined for sending Internet mail. As you’ll see, Python comes packaged with modules that help
you speak many Internet protocols, and the module is always named after the protocol: imaplib,
poplib, httplib, ftplib, and so on.
Put your own e-mail address in me@mydomain, and if you’ve got a mail server running on your machine,
you should be able to send mail to yourself, as shown in Figure 16-1.
Figure 16-1
However, you probably don’t have a mail server running on your machine. (You might have one if you’re
running these scripts on a shared computer, or if you set the mail server up yourself, in which case you
probably already know a bit about networking and are impatiently waiting for the more advanced parts
of this chapter.) If there’s no mail server on the machine where you run this script, you’ll get an exception
when you try to instantiate the remote SMTP mail server object, something similar to this:
Traceback (most recent call last):
File “<stdin>”, line 1, in ?
File “/usr/lib/python2.4/smtplib.py”, line 241, in __init__
(code, msg) = self.connect(host, port)
File “/usr/lib/python2.4/smtplib.py”, line 303, in connect
raise socket.error, msg
socket.error: (111, ‘Connection refused’)
What’s going on here? Look at the line that caused the exception:
>>> server = smtplib.SMTP(“localhost”, 25)
The constructor for the smtplib class is trying to start up a network connection using IP, the Internet
Protocol. The string “localhost” and the number 25 identify the Internet location of the putative mail
server. Because you’re not running a mail server, there’s nothing at the other end of the connection, and
when Python discovers this fact, it can’t continue.
306

35.8 Page 348

▲back to top
Network Programming
To understand the mystical meanings of “localhost” and 25, it helps to know a little about protocols,
and the Internet Protocol in particular.
Understanding Protocols
A protocol is a convention for structuring the data sent between two or more parties on a network. It’s
analogous to the role of protocol or etiquette in relationships between humans. For instance, suppose
that you wanted to go out with friends to dinner or get married to someone. Each culture has defined
conventions describing the legal and socially condoned behavior in such situations. When you go out for
dinner, there are conventions about how to behave in a restaurant, how to use the eating utensils, and
how to pay. Marriages are carried out according to conventions regarding rituals and contracts, conven-
tions that can be very elaborate.
These two activities are very different, but the same lower-level social protocols underlie both of
them. These protocols set standards for things such as politeness and the use of a mutually understood
language. On the lowest level, you may be vibrating your vocal cords in a certain pattern, but on a
higher level you’re finalizing your marriage by saying “I do.” Violate a lower-level protocol (say, by act-
ing rudely in the restaurant) and your chances of carrying out your high-level goal can be compromised.
All of these aspects of protocols for human behavior have their correspondence in protocols for com-
puter networking.
Comparing Protocols and Programming Languages
Thousands of network protocols for every imaginable purpose have been invented over the past few
decades; it might be said that the history of networking is the history of protocol design. Why so many
protocols? To answer this question, consider another analogy to the world of network protocols: Why so
many programming languages? Network protocols have the same types of interrelation as programming
languages, and people create new protocols for the same reasons they create programming languages.
Different programming languages have been designed for different purposes. It would be madness to
write a word processor in the FORTRAN language, not because FORTRAN is objectively “bad,” but
because it was designed for mathematical and scientific research, not end-user GUI applications.
Similarly, different protocols are intended for different purposes. SMTP, the protocol you just got a brief
look at, could be used for all sorts of things besides sending mail. No one does this because it makes
more sense to use SMTP for the purpose for which it was designed, and use other protocols for other
purposes.
A programming language may be created to compete with others in the same niche. The creator of a new
language may see technical or aesthetic flaws in existing languages and want to make their own tasks
easier. A language author may covet the riches and fame that come with being the creator of a popular
language. A person may invent a new protocol because they’ve come up with a new type of application
that requires one.
Some programming languages are designed specifically for teaching students how to program, or, at the
other end of programming literacy, how to write compilers. Some languages are designed to explore
new ideas, not for real use, and other languages are created as a competitive tool by one company for
use against another company.
307

35.9 Page 349

▲back to top
Chapter 16
These factors also come into play in protocol design. Companies sometimes invent new, incompatible
protocols to try to take business from a competitor. Some protocols are intended only for pedagogical
purposes. For instance, this chapter will, under the guise of teaching network programming, design
protocols for things like online chat rooms. There are already perfectly good protocols for this, but
they’re too complex to be given a proper treatment in the available space.
The ADA programming language was defined by the U.S. Department of Defense to act as a common
language across all military programming projects. The Internet Protocol was created to enable multiple
previously incompatible networks to communicate with one another (hence the name “Internet”).
Nowadays, even internal networks (intranets) usually run atop the Internet Protocol, but the old motives
(the solving of new problems, competition, and so on) remain in play at higher and lower levels, which
brings us to the most interesting reason for the proliferation of programming languages and protocols.
The Internet Protocol Stack
Different programming languages operate at different levels of abstraction. Python is a very high-level
language capable of all kinds of tasks, but the Python interpreter itself isn’t written in Python: It’s
written in C, a lower-level language. C, in turn, is compiled into a machine language specific to your
computer architecture. Whenever you type a statement into a Python interpreter, there is a chain of
abstraction reaching down to the machine code, and even lower to the operation of the digital circuits
that actually drive the computer.
There’s a Python interpreter written in Java (Jython), but Java is written in C. PyPy is a project that
aims to implement a Python interpreter in Python, but PyPy runs on top of the C or Java implementa-
tion. You can’t escape C!
In one sense, when you type a statement into the Python interpreter, the computer simply “does what
you told it to.” In another, it runs the Python statement you typed. In a third sense, it runs a longer series
of C statements, written by the authors of Python and merely activated by your Python statement. In a
fourth sense, the computer runs a very long, nearly incomprehensible series of machine code statements.
In a fifth, it doesn’t “run” any program at all: You just cause a series of timed electrical impulses to be
sent through the hardware. The reason we have high-level programming languages is because they’re
easier to use than the lower-level ones. That doesn’t make lower-level languages superfluous, though.
English is a very high-level human language capable of all kinds of tasks, but one can’t speak English
just by “speaking English.” To speak English, one must actually make some noises, but a speaker can’t
just “make some noises” either: We have to send electrical impulses from our brains that force air out of
the lungs and constantly reposition the tongues and lips. It’s a very complicated process, but we don’t
even think about the lower levels, only the words we’re saying and the concepts we’re trying to convey.
The soup of network protocols can be grouped into a similar hierarchical structure based on levels of
abstraction, or layers. On the physical layer, the lowest level, it’s all just electrical impulses and EM
radiation. Just above the physical layer, every type of network hardware needs its own protocol, imple-
mented in software (for instance, the Ethernet protocol for networks that run over LAN wires). The elec-
tromagnetic phenomena of the physical layer can now be seen as the sending and receiving of bits from
one device to another. This is called the data link layer. As you go up the protocol stack, these raw bits
take on meaning: They become routing instructions, commands, responses, images, web pages.
308

35.10 Page 350

▲back to top
Network Programming
Because different pieces of hardware communicate in different ways, connecting (for example) an
Ethernet network to a wireless network requires a protocol that works on a higher level then the data
link layer. As mentioned earlier, the common denominator for most networks nowadays is the Internet
Protocol (IP), which implements the network layer and connects all those networks together. IP works
on the network layer.
Directly atop the network layer is the transport layer, which makes sure the information sent over IP
gets to its destination reliably, in the right order, and without errors. IP doesn’t care about reliability or
error-checking: It just takes some data and a destination address, sends it across the network, and
assumes it gets to that address intact.
TCP, the Transmission Control Protocol, does care about these things. TCP implements the transport
layer of the protocol stack, making reliable, orderly communication possible between two points on the
network. It’s so common to stack TCP on top of IP that the two protocols are often treated as one and
given a unified name, TCP/IP.
All of the network protocols you’ll study and design in this chapter are based on top of TCP/IP.
These protocols are at the application layer and are designed to solve specific user problems. Some
of these protocols are known by name even to nonprogrammers: You may have heard of HTTP, FTP,
BitTorrent, and so on.
When people think of designing protocols, they usually think of the application layer, the one best suited
to Python implementations. The other current field of interest is at the other end in the data link layer:
embedded systems programming for connecting new types of devices to the Internet. Thanks to the over-
whelming popularity of the Internet, TCP/IP has more or less taken over the middle of the protocol stack.
A Little Bit About the Internet Protocol
Now that you understand where the Internet Protocol fits into the protocol stack your computer uses,
there are only two things you really need to know about it: addresses and ports.
Internet Addresses
Each computer on the Internet (or on a private TCP/IP network) has one or more IP addresses, usually
represented as a dotted series of four numbers, like “208.215.179.178.” That same computer may also
have one or more hostnames, which look like “wrox.com.”
To connect to a service running on a computer, you need to know its IP address or one of its hostnames.
(Hostnames are managed by DNS, a protocol that runs on top of TCP/IP and silently turns hostnames
into IP addresses). Recall the script at the beginning of this chapter that sent out mail. When it tried to
connect to a mail server, it mentioned the seemingly magic string “localhost”:
>>> server = smtplib.SMTP(“localhost”, 25)
“localhost” is a special hostname that always refers to the computer you’re using when you mention
it (each computer also has a special IP address that does the same thing: 127.0.0.1). The hostname is
how you tell Python where on the Internet to find your mail server.
309

36 Pages 351-360

▲back to top

36.1 Page 351

▲back to top
Chapter 16
It’s generally better to use hostnames instead of IP addresses, even though the former immediately gets
turned into the latter. Hostnames tend to be more stable over time than IP addresses. Another example
of the protocol stack in action: The DNS protocol serves to hide the low-level details of IP’s addressing
scheme.
Of course, if you don’t run a mail server on your computer, “localhost” won’t work. The organization
that gives you Internet access should be letting you use their mail server, possibly located at mail.[your
ISP].com or smtp.[your ISP].com. Whatever mail client you use, it probably has the hostname of a mail
server somewhere in its configuration, so that you can use it to send out mail. Substitute that for
“localhost” in the example code listed previously, and you should be able to send mail from Python:
>>> fromAddress = ‘sender@example.com’
>>> toAddress = ‘[your email address]’
>>> msg = “Subject: Hello\\n\\nThis is the body of the message.”
>>> import smtplib
>>> server = smtplib.SMTP(“mail.[your ISP].com”, 25)
>>> server.sendmail(fromAddress, toAddress, msg)
{}
Unfortunately, you still might not be able to send mail, for any number of reasons. Your SMTP server
might demand authentication, which this sample session doesn’t provide. It might not accept mail from
the machine on which you’re running your script (try the same machine you normally use to send
mail). It might be running on a nonstandard port (see below). The server might not like the format of
this bare-bones message, and expect something more like a “real” e-mail message; if so, the email mod-
ule described in the following section might help. If all else fails, ask your system administrator for help.
Internet Ports
The string “localhost” has been explained as a DNS hostname that masks an IP address. That leaves
the mysterious number 25. What does it mean? Well, consider the fact that a single computer may host
more than one service. A single machine with one IP address may have a web server, a mail server, a
database server, and a dozen other servers. How should clients distinguish between an attempt to con-
nect to the web server and an attempt to connect to the database server?
A computer that implements the Internet Protocol can expose up to 65536 numbered ports. When you
start an Internet server (say, a web server), the server process “binds” itself to one or more of the ports
on your computer (say, port 80, the conventional port for a web server) and begins listening for
outside connections to that port. If you’ve ever seen a web site address that looked like “http://www.
example.com:8000/”, that number is the port number for the web server — in this case, a port number
that violates convention. The enforcer of convention in this case is the Internet Assigned Numbers
Authority.
The IANA list of protocols and conventional port numbers is published at www.iana.org/
assignments/port-numbers.
According to the IANA, the conventional port number for SMTP is 25. That’s why the constructor to the
SMTP object in that example received 25 as its second argument (if you don’t specify a port number at
all, the SMTP constructor will assume 25):
>>> server = smtplib.SMTP(“localhost”, 25)
310

36.2 Page 352

▲back to top
Network Programming
The IANA divides the port numbers into “well-known ports” (ports from 0 to 1023), “registered ports”
(from 1024 to 49151), and “dynamic ports” (from 49152 to 65535). On most operating systems, you must
have administrator privileges to bind a server to a well-known port because processes that bind to those
ports are often themselves given administrator privileges. Anyone can bind servers to ports in the regis-
tered range, and that’s what we’ll do for the custom servers written in this chapter. The dynamic range is
used by clients, not servers; we’ll cover that later when talking about sockets.
Sending Internet E-mail
With a basic understanding of how TCP/IP works, the Python session from the beginning of this chapter
should now make more sense:
>>> fromAddress = ‘sender@example.com’
>>> toAddress = ‘recipient@example.com’
>>> msg = “Subject: Hello\\n\\nThis is the body of the message.”
>>> import smtplib
>>> server = smtplib.SMTP(“localhost”, 25)
>>> server.sendmail(fromAddress, toAddress, msg)
{}
If you don’t have an SMTP server running on your machine, you should now be able to find out a host-
name and port number that will work for you. The only piece of the code I haven’t explained is why the
e-mail message looks the way it does.
The E-mail File Format
In addition to the large number of e-mail-related protocols, Internet engineers have designed a couple of
file formats for packaging the parts of an e-mail message. Both of these protocols and file formats have
been published in numbered documents called RFCs.
Throughout this chapter, until you start writing your own protocols, you’ll be working with protocols
and formats designed by others and specified in RFCs. These documents often contain formal language
specifications and other not-quite-light reading, but for the most part they’re pretty readable.
The current standard defining the format of e-mail messages is RFC 2822. Published in 2001, it updated
the venerable RFC 822, which dates from 1982 (Maybe RFC 2822 would have been published earlier
if they hadn’t had to wait for the numbers to match up). You may still see references to “RFC 822” as
shorthand for “the format of e-mail messages,” such as in Python’s now deprecated rfc822 module.
To find a particular RFC, you can just search the web for “RFC x”, or look on the official
site at www.ietf.org/rfc.html. RFC 2822 is hosted at (among other places)
www.ietf.org/rfc/rfc2822.txt.
An e-mail message consists of a set of headers (metadata describing the message) and a body (the mes-
sage itself). The headers are actually sent in a form like key-value pairs in which a colon and a space sep-
arate the key and the value (for instance, “Subject: Hello”). The body is just that: the text of the message.
311

36.3 Page 353

▲back to top
Chapter 16
You can create RFC2822-compliant messages with Python using the Message class in Python’s email
module. The Message object acts like a dictionary that maps message header names to their values. It
also has a “payload,” which is the body text:
>>> from email import Message
>>> from email.Message import Message
>>> message = Message()
>>> message[‘Subject’] = ‘Hello’
>>> message.set_payload(‘This is the body of the message’)
>>> print str(message)
From nobody Fri Mar 25 20:08:22 2005
Subject: Hello
This is the body of the message
That’s more code than just specifying the e-mail string, but it’s less error-prone, especially for a complex
message. Also, you’ll notice that you got back information that you didn’t put into the message. This is
because the smtplib adds some required headers onto your message when you send it.
RFC2822 defines some standard message headers, described in the following table. It also defines data
representation standards for some of the header values (for instance, it defines a way of representing
e-mail addresses and dates). The standard also gives you space to define custom headers for use in your
own programs that send and receive e-mail.
Header
To
From
Date
Subject
Cc
Example
To: Leonard Richardson <leonardr@example.com>
From: Peter C. Norton <peter@example.com>
Date: Wed, 16 Mar 2005 14:36:07 -0500 (EST)
Subject: Python book
Cc: michael@example.com,
Jason Diamond <jason@example.com>
Purpose
Addresses of people who
should receive the message
The e-mail address of the
person who (allegedly) sent
the message
The date the message was
sent
A summary or title of the
message, intended for
human consumption
Addresses of people who
should receive the message,
even though it’s not
addressed to them
Note a few restrictions on the content of the body. RFC2822 requests that there be fewer than 1000 char-
acters in each line of the body. A more onerous restriction is that your headers and body can only contain
U.S. ASCII characters (that is, the first 127 characters of ASCII): no “international” or binary characters
are allowed. By itself this doesn’t make sense because you’ve probably already seen e-mail messages in
other languages. How that happens is explained next.
312

36.4 Page 354

▲back to top
Network Programming
MIME Messages
If RFC 2822 requires that your e-mail message contain only U.S. ASCII characters, how is it possible that
people routinely send e-mail with graphics and other binary files attached? This is achieved with an
extension to the RFC2822 standard called MIME, the Multi-purpose Internet Mail Extension.
MIME is a series of standards designed around fitting non-U.S.-ASCII data into the 127 seven-bit charac-
ters that make up U.S. ASCII. Thanks to MIME, you can attach binary files to e-mail messages, write mes-
sages and even headers (such as your name) using non-English characters, and have it all come out right
on the other end (assuming the other end understands MIME, which almost everyone does nowadays).
The main MIME standard is RFC 1521, which describes how to fit binary data into the body of e-mail
messages. RFC 1522 describes how to do the same thing for the headers of e-mail messages.
MIME Encodings: Quoted-printable and Base64
The most important parts of MIME are its encodings, which provide ways of encoding 8-bit characters
into seven bits. MIME defines two encodings: quoted-printable encoding and Base64 encoding. Python
provides a module for moving strings into and out of each encoding,
The quoted-printable encoding is intended for text that contains only a few 8-bit characters, with the
majority of characters being U.S. ASCII. The advantage of the quoted-printable encoding is that the text
remains mostly legible once encoded, making it ideal for text written in or borrowing words from
Western European languages (languages that can be represented in U.S. ASCII except for a few charac-
ters that use diacritical marks). Even if the recipient of your message can’t decode the quoted-printable
message, they should still be able to read it. They’ll just see some odd-looking equal signs and hexadeci-
mal numbers in the middle of words.
The Python module for encoding and decoding is quopri:
>>> import quopri
>>> encoded = quopri.encodestring(“I will have just a soupçon of soup.”)
>>> print encoded
I will have just a soup=E7on of soup.
>>> print quopri.decodestring(encoded)
I will have just a soup\\xe7on of soup.
Depending on your terminal settings, you might see the actual “ç” character in the last line, or you
might see “\\xe7”. “\\xe7” is the Python string representation of the “ç” character, just as “\\E7” is the
quoted-printable representation. In the session reproduced above, that string was decoded into a Python
string, and then re-encoded in a Python-specific form for display!
The Base64 encoding, on the other hand, is intended for binary data. It should not be used for human-
readable text, because it totally obscures the text:
>>> import base64
>>> encoded = base64.encodestring(“I will have just a soupçon of soup.”)
>>> print encoded
SSB3aWxsIGhhdmUganVzdCBhIHNvdXBvbiBvZiBzb3VwLg==
>>> print base64.decodestring(encoded)
I will have just a souçpon of soup.
313

36.5 Page 355

▲back to top
Chapter 16
Why bother with base64 when quoted-printable works on anything and doesn’t mangle human-readable
text? Apart from the fact that it would be kind of misleading to encode something as “quoted-printable”
when it’s not “printable” in the first place, Base64 encoding is much more efficient at representing binary
data than quoted-printable encoding. Here’s a comparison of the two encodings against a long string of
random binary characters:
>>> import random
>>> import quopri
>>> import base64
>>> length = 10000
>>> randomBinary = ‘’.join([chr(random.randint(0,255)) for x in range(0, length)])
>>> len(quopri.encodestring(randomBinary)) / float(length)
2.0663999999999998
>>> len(base64.encodestring(randomBinary)) / float(length)
1.3512
Those numbers will vary slightly across runs because the strings are randomly generated, but if you try
this experiment you should get similar results to these every time. A binary string encoded as quoted-
printable encoding is safe to send in an e-mail, but it’s (on average) about twice as long as the original,
unsendable string. The same binary string, encoded with Base64 encoding, is just as safe, but only about
1.35 times as long as the original. Using Base64 to encode mostly binary data saves space and bandwidth.
At the same time, it would be overkill to encode an ASCII string with Base64 just because it contains a
few characters outside of the U.S. ASCII range. Here’s the same comparison done with a long random
string that’s almost entirely composed of U.S. ASCII characters:
>>> import random
>>> import quopri
>>> import base64
>>> length = 10000
>>> randomBinary = ‘’.join([chr(random.randint(0,128)) for x in range(0, length)])
>>> len(quopri.encodestring(randomBinary)) / float(length)
1.0661
>>> len(base64.encodestring(randomBinary)) / float(length)
1.3512
Here, the quoted-printable representation is barely larger than the original text (it’s almost the same as
the original text), but the Base64 representation is 1.35 times as long as the original, just as before. This
demonstrates why MIME supports two different encodings: to quote RFC1521, “a ‘readable’ encoding
[quoted-printable] and a ‘dense’ encoding [Base64].”
MIME is more “multi-purpose” than its name implies. Many features of MIME have been picked up for
use outside of e-mail applications. The idea of using Base64 or quoted-printable to turn non-ASCII
characters into ASCII shows up in other domains. Base64 encoding is also sometimes used to obscure
text from human readability without actually encrypting it.
MIME Content Types
The other important part of MIME is its idea of a content type. Suppose that you send your friend an
e-mail message: “Here’s that picture I took of you.”, and attach an image. Thanks to Base64 encoding,
the recipient will get the encoded data as you sent it, but how is their mail reader supposed to know that
it’s an image and not some other form of binary data?
314

36.6 Page 356

▲back to top
Network Programming
MIME solves this problem by defining a custom RFC2822-format header called Content-Type. This
header describes what kind of file the body is, so that the recipient’s mail client can figure out how to dis-
play it. Content types include text/plain (what you’d get if you put a normal e-mail message into a MIME
envelope), text/html, image/jpeg, video/mpeg, audio/mp3, and so on. Each content type has a “major
type” and a “minor type”, separated by a slash. The major types are very general and there are only seven
of them, defined in the MIME standard itself. The minor types usually designate particular file formats.
The idea of a string having a “Content-Type”, which tells the recipient what to do with it, is another
invention of MIME used outside of the e-mail world. The most common use is in HTTP, the protocol
used by the World Wide Web and covered in Chapter 22. Every HTTP response is supposed to have a
“Content-Type” header (just like a MIME e-mail message), which tells the web browser how to display
the response.
Try It Out Creating a MIME Message with an Attachment
So far, so good. Python provides many submodules of the e-mail module for constructing MIME mes-
sages, including a module for each of the major content types. It’s simple to use these to craft a MIME
message containing an encoded image file.
>>> from email.MIMEImage import MIMEImage
>>> filename = ‘photo.jpg’
>>> msg = MIMEImage(open(filename).read(), name=filename)
>>> msg[‘To’] = ‘You <you@example.com>’
>>> msg[‘From’] = ‘Me <me@example.com>’
>>> msg[‘Subject’] = ‘Your picture’
>>> print str(msg)
From nobody Sun Mar 20 15:15:27 2005
Content-Type: image/jpeg; name=”photo.jpg”
MIME-Version: 1.0
Content-Transfer-Encoding: base64
From: Me <me@example.com>
To: You <you@example.com>
Subject: Your picture
/4AAQSkZJRgABAQEASABIAAD//gAXQ3JlYXRlZCB3aXRoIFRoZSBHSU1Q/9sAQwAIBgYHBgUI
...
[Much base64 encoded text omitted.]
...
3f7kklh4dg+UTZ1TsAAv1F69UklmZ9hrzogZibOqSSA8gZySSSJI/9k=
Of course, for ‘photo.jpg’, you should substitute the filename of any other image file you have
handy. Just put the file into the directory from which you invoke the Python session.
Send this message using smtplib (as per the first example in this chapter), and it’ll show up at the other
end looking something like what is shown in Figure 16-2.
Because we told the MIMEImage constructor that the picture was called photo.jpg, the mail client on
the other end will be able to save it under that filename. Note that MIMEImage automatically figured out
the minor type of the JPEG data, and transformed it into base64.
315

36.7 Page 357

▲back to top
Chapter 16
Figure 16-2
MIME Multipart Messages
There’s just one problem. This isn’t quite the e-mail message described earlier. That message was a short
piece of text (“Here’s that picture I took of you.”) and an attached image. This message is just the
image. There’s no space for the text portion in the body of the message; putting it there would compro-
mise the image file. The Content-Type header of a mail message can be text/plain or image/jpeg; it
can’t be both. So how do mail clients create messages with attachments?
In addition to classifying the file formats defined by other standards (for instance, image for image file
formats), MIME defines a special major type called multipart. A message with a major content type of
multipart can contain other MIME messages in its body, each with its own set of headers and its own
content type.
The best way to see how this works is to create a multipart message using the email.MIMEMultipart
module, in conjunction with the email.MIME* modules for the files you want to attach. Here is a script
called FormatMimeMultipartMessage.py, a slightly more complicated version of the previous example:
#!/usr/bin/python
from email.MIMEMultipart import MIMEMultipart
import os
import sys
filename = sys.argv[1]
msg = MIMEMultipart()
msg[‘From’] = ‘Me <me@example.com>’
msg[‘To’] = ‘You <you@example.com>’
msg[‘Subject’] = ‘Your picture’
from email.MIMEText import MIMEText
316

36.8 Page 358

▲back to top
Network Programming
text = MIMEText(“Here’s that picture I took of you.”)
msg.attach(text)
from email.MIMEImage import MIMEImage
image = MIMEImage(open(filename).read(), name=os.path.split(filename)[1])
msg.attach(image)
Run this script, passing in the path to an image file, and you’ll see a MIME multipart e-mail message
that includes a brief text message and the image file, encoded in base64:
# python FormatMimeMultipartMessage.py ./photo.jpg
From nobody Sun Mar 20 15:41:23 2005
Content-Type: multipart/mixed; boundary=”===============1011273258==”
MIME-Version: 1.0
From: Me <me@example.com>
To: You <you@example.com>
Subject: Your picture
--===============1011273258==
Content-Type: text/plain; charset=”us-ascii”
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Here’s that picture I took of you.
--===============1011273258==
Content-Type: image/jpeg; name=”photo.jpg”
MIME-Version: 1.0
Content-Transfer-Encoding: base64
/4AAQSkZJRgABAQEASABIAAD//gAXQ3JlYXRlZCB3aXRoIFRoZSBHSU1Q/9sAQwAIBgYHBgUI
...
[As before, much base64 encoded text omitted.]
...
3f7kklh4dg+UTZ1TsAAv1F69UklmZ9hrzogZibOqSSA8gZySSSJI/9k=
--===============1011273258==
When you send this message, it will show up at the other end looking more like you expect a message
with an attachment to look (see Figure 16-3). This is the kind of e-mail your e-mail client creates when
you send a message with attachments.
Several features of this e-mail bear mentioning:
❑ The content type (multipart/mixed) isn’t enough, by itself, to make sense of the message
body. MIME also requires the definition of a “boundary”, a string generated semi-randomly by
Python and used in the body of the message to note where one part stops and another begins.
❑ The message as a whole has all the headers we associate with e-mail messages: Subject, From,
To, and the MIME-specific Content-Type header. In addition to this, each part of the message
has a separate set of headers. These are not message headers, although they’re in the RFC2822
header format; and some headers (MIME-Version and Content-Type) show up in both the
message headers and the body. These are MIME message body headers, interpreted by the
MIME parser. As far as RFC 2822 is concerned, they’re part of the message body, just like the
317

36.9 Page 359

▲back to top
Chapter 16
files they describe, the boundaries that separate MIME parts, and the text “Here’s that
picture I took of you.”
❑ The MIME part containing the body of the message has an encoding of 7bit. This just means
that the part is not encoded at all. Every character in the part body was U.S. ASCII, so there was
no need to encode it.
Figure 16-3
Python’s mail classes are very useful once you know what kind of mail you want to construct: for text-only
messages, use the simple email.Message class. To attach a file to a message, use one of the email.Mime*
classes. To send multiple files, or a combination of text and files, use email.MimeMultipart in conjunc-
tion with the other email.Mime* classes.
A problem arises when you’re not sure ahead of time which class to use to represent your e-mail mes-
sage. Here’s a class called SmartMessage for building e-mail messages that starts out keeping body text
in a simple Message representation, but which will switch to MimeMultipart if you add an attachment.
This strategy will generate the same range of e-mail message bodies as a typical end-user mail applica-
tion: simple RFC 2822 bodies for simple messages, and complex MIME bodies for messages with attach-
ments. Put this class in a file called SendMail.py:
from email import Encoders
from email.Message import Message
from email.MIMEText import MIMEText
from email.MIMEMultipart import MIMEMultipart
from email.MIMENonMultipart import MIMENonMultipart
import mimetypes
class SmartMessage:
“””A simplified interface to Python’s library for creating email
318

36.10 Page 360

▲back to top
Network Programming
messages, with and without MIME attachments.”””
def __init__(self, fromAddr, toAddrs, subject, body):
“””Start off on the assumption that the message will be a simple RFC
2822 message with no MIME.”””
self.msg = Message()
self.msg.set_payload(body)
self[‘Subject’] = subject
self.setFrom(fromAddr)
self.setTo(toAddrs)
self.hasAttachments = False
def setFrom(self, fromAddr):
“Sets the address of the sender of the message.”
if not fromAddr or not type(fromAddr)==type(‘’):
raise Exception, ‘A message must have one and only one sender.’
self[‘From’] = fromAddr
def setTo(self, to):
“Sets the address or addresses that will receive this message.”
if not to:
raise Exception, ‘A message must have at least one recipient.’
self._addresses(to, ‘To’)
#Also store the addresses as a list, for the benefit of future
#code that will actually send this message.
self.to = to
def setCc(self, cc):
“””Sets the address or addresses that should receive this message,
even though it’s not addressed directly to them (“carbon-copy”).”””
self._addresses(cc, ‘Cc’)
def addAttachment(self, attachment, filename, mimetype=None):
“Attaches the given file to this message.”
#Figure out the major and minor MIME type of this attachment,
#given its filename.
if not mimetype:
mimetype = mimetypes.guess_type(filename)[0]
if not mimetype:
raise Exception, “Couldn’t determine MIME type for “, filename
if ‘/’ in mimetype:
major, minor = mimetype.split(‘/’)
else:
major = mimetype
minor = None
#The message was constructed under the assumption that it was
#a single-part message. Now that we know there’s to be at
#least one attachment, we need to change it into a multi-part
#message, with the first part being the body of the message.
if not self.hasAttachments:
body = self.msg.get_payload()
newMsg = MIMEMultipart()
319

37 Pages 361-370

▲back to top

37.1 Page 361

▲back to top
Chapter 16
newMsg.attach(MIMEText(body))
#Copy over the old headers to the new object.
for header, value in self.msg.items():
newMsg[header] = value
self.msg = newMsg
self.hasAttachments = True
subMessage = MIMENonMultipart(major, minor, name=filename)
subMessage.set_payload(attachment)
#Encode text attachments as quoted-printable, and all other
#types as base64.
if major == ‘text’:
encoder = Encoders.encode_quopri
else:
encoder = Encoders.encode_base64
encoder(subMessage)
#Link the MIME message part with its parent message.
self.msg.attach(subMessage)
def _addresses(self, addresses, key):
“””Sets the given header to a string representation of the given
list of addresses.”””
if hasattr(addresses, ‘__iter__’):
addresses = ‘, ‘.join(addresses)
self[key] = addresses
#A few methods to let scripts treat this object more or less like
#a Message or MultipartMessage, by delegating to the real Message
#or MultipartMessage this object holds.
def __getitem__(self, key):
“Return a header of the underlying message.”
return self.msg[key]
def __setitem__(self, key, value):
“Set a header of the underlying message.”
self.msg[key] = value
def __getattr__(self, key):
return getattr(self.msg, key)
def __str__(self):
“Returns a string representation of this message.”
return self.msg.as_string()
Try It Out Building E-mail Messages with SmartMessage
To test out SmartMessage, put it into a file called SendMail.py and run a Python session like this one:
>>> from SendMail import SmartMessage
>>> msg = SmartMessage(“Me <me@example.com>”, “You <you@example.com>”, “Your
picture”, “Here’s that picture I took of you.”)
>>> print str(msg)
320

37.2 Page 362

▲back to top
Network Programming
Subject: Your picture
From: Me <me@example.com>
To: You <you@example.com>
Here’s that picture I took of you.
>>> msg.addAttachment(open(“photo.jpg”).read(), “photo.jpg”)
>>> print str(msg)
Content-Type: multipart/mixed; boundary=”===============1077328303==”
MIME-Version: 1.0
Subject: Your picture
From: Me <me@example.com>
To: You <you@example.com>
--===============1077328303==
Content-Type: text/plain; charset=”us-ascii”
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Here’s that picture I took of you.
--===============1077328303==
Content-Type: image/jpeg
MIME-Version: 1.0
Content-Transfer-Encoding: base64
/9j/4AAQSkZJRgABAQEASABIAAD//gAXQ3JlYXRlZCB3aXRoIFRoZSBHSU1Q/9sAQwAIBgYHBgUI
...
[Once again, much base64 text omitted.]
...
3f7kklh4dg+UTZ1TsAAv1F69UklmZ9hrzogZibOqSSA8gZySSSJI/9k=
--===============0855656444==--
How It Works
SmartMessage wraps the classes in Python’s email module. When the SmartMessage object is first
created, it keeps its internal representation in a Message object. This message has a simple string
representation.
When a file is attached to the SmartMessage, though, a Message object won’t do the job anymore.
Message objects know only about RFC2822, nothing about the MIME extensions. At this point,
SmartMessage transparently swaps out the Message object for a MimeMultipart object with the
same headers and payload.
This transparent swap avoids forcing the user to decide ahead of time whether or not a message should
be MIME encoded. It also avoids a lowest-common-denominator strategy of MIME-encoding each and
every message, which is a wasteful operation for messages that are just one text part.
Sending Mail with SMTP and smtplib
Now that you know how to construct e-mail messages, it’s appropriate to revisit in a little more detail
the protocol used to send them. This is SMTP, another TCP/IP-based protocol, defined in RFC 2821.
321

37.3 Page 363

▲back to top
Chapter 16
Let’s look at the original example one more time:
>>> fromAddress = ‘sender@example.com’
>>> toAddress = [your email address]
>>> msg = “Subject: Hello\\n\\nThis is the body of the message.”
>>> import smtplib
>>> server = smtplib.SMTP(“localhost”, 25)
>>> server.sendmail(fromAddress, toAddress, msg)
{}
You connect to an SMTP server (at port 25 on localhost) and send a string message from one address to
another. Of course, the location of the SMTP server shouldn’t be hard-coded, and because some servers
require authentication, it would be nice to be able to accept authentication information when creating
the SMTP object. Here’s a class that works with the SmartMessage class defined in the previous section
to make it easier to send mail. Because the two classes go together, add this class to SendMail.py, the
file that also contains the SmartMessage class:
from smtplib import SMTP
class MailServer(SMTP):
“A more user-friendly interface to the default SMTP class.”
def __init__(self, server, serverUser=None, serverPassword=None, port=25):
“Connect to the given SMTP server.”
SMTP.__init__(self, server, port)
self.user = serverUser
self.password = serverPassword
#Uncomment this line to see the SMTP exchange in detail.
#self.set_debuglevel(True)
def sendMessage(self, message):
“Sends the given message through the SMTP server.”
#Some SMTP servers require authentication.
if self.user:
self.login(self.user, self.password)
#The message contains a list of destination addresses that
#might have names associated with them. For instance,
#”J. Random Hacker <jhacker@example.com>”. Some mail servers
#will only accept bare email addresses, so we need to create a
#version of this list that doesn’t have any names associated
#with it.
destinations = message.to
if hasattr(destinations, ‘__iter__’):
destinations = map(self._cleanAddress, destinations)
else:
destinations = self._cleanAddress(destinations)
self.sendmail(message[‘From’], destinations, str(message))
def _cleanAddress(self, address):
“Transforms ‘Name <email@domain>’ into ‘email@domain’.”
parts = address.split(‘<’, 1)
322

37.4 Page 364

▲back to top
Network Programming
if len(parts) > 1:
#This address is actually a real name plus an address:
newAddress = parts[1]
endAddress = newAddress.find(‘>’)
if endAddress != -1:
address = newAddress[:endAddress]
return address
Try It Out Sending Mail with MailServer
This chapter’s initial example constructed a message as a string and sent it through SMTPlib. With the
SmartMessage and MailServer classes, you can send a much more complex message, using simpler
Python code:
>>> from SendMail import SmartMessage, MailServer
>>> msg = SmartMessage(“Me <me@example.com>”,
“You <you@example.com>”,
“Your picture”,
“Here’s that picture I took of you.”)
>>> msg.addAttachment(open(“photo.jpg”).read(), “photo.jpg”)
>>> MailServer(“localhost”).sendMessage(msg)
>>>
Run this code (substituting the appropriate e-mail addresses and server hostname), and you’ll be able to
send mail with MIME attachments to anyone.
How It Works
SmartMessage wraps the classes in Python’s email module. As before, the underlying representation
starts out as a simple Message object but becomes a MimeMultipart object once photo.jpg is attached.
This time, the message is actually sent through an SMTP server. The MailServer class hides the fact that
smtplilb expects you to specify the “To” and “From” headers twice: one in the call to the sendmail
method and again in the body of the mail message. It also takes care of sanitizing the destination
addresses, putting them into a form that all SMTP servers can deal with. Between the two wrapper
classes, you can send complex e-mail messages from a Python script almost as easily as from a mail client.
Retrieving Internet E-mail
Now that you’ve seen how to send mail, it’s time to go all the way toward fulfilling Jamie Zawinski’s
prophecy and expand your programs so that they can read mail. There are three main ways to do this,
and the choice is probably not up to you. How you retrieve mail depends on your relationship with the
organization that provides your Internet access.
Parsing a Local Mail Spool with mailbox
If you have a Unix shell account on your mail server (because, for instance, you run a mail server on your
own computer), mail for you is appended to a file (probably /var/spool/mail/[your username]) as it
comes in. If this is how your mail setup works, your existing mail client is probably set up to parse that
323

37.5 Page 365

▲back to top
Chapter 16
file. It may also be set up to move messages out of the spool file and into your home directory as they
come in.
The incoming mailbox in /var/spool/mail/ is kept in a particular format called “mbox format”. You
can parse these files (as well as mailboxes in other formats such as MH or Maildir) by using the classes
in the mailbox module.
Here’s a simple script, MailboxSubjectLister.py, that iterates over the messages in a mailbox file,
printing out the subject of each one:
#!/usr/bin/python
import email
import mailbox
import sys
if len(sys.argv) < 2:
print ‘Usage: %s [path to mailbox file]’ % sys.argv[0]
sys.exit(1)
path = sys.argv[1]
fp = open(path, ‘rb’)
subjects = []
for message in mailbox.PortableUnixMailbox(fp, email.message_from_file):
subjects.append(message[‘Subject’])
print ‘%s message(s) in mailbox “%s”:’ % (len(subjects), path)
for subject in subjects:
print ‘’, subject
UnixMailbox (and the other Mailbox classes in the mailbox module) take as their constructor a
file object (the mailbox file), and a function that reads the next message from the file-type object.
In this case, the function is the email module’s message_from_file. The output of this useful
function is a Message object, or one of its MIME* subclasses, such as MIMEMultipart. This and the
email.message_from_string function are the most common ways of creating Python representations
of messages you receive.
You can work on these Message objects just as you could with the Message objects created from scratch
in earlier examples, where the point was to send e-mail messages. Python uses the same classes to repre-
sent incoming and outgoing messages.
Try It Out Printing a Summary of Your Mailbox
If you have a Unix account on your e-mail server, you can run the mailbox subject lister against your
mail spool file, and get a list of subjects. If you don’t have a Unix account on your e-mail server, or if you
use a web-based mail service, you won’t be able to get your mail this way:
$ python MailboxSubjectLister.py /var/spool/mail/leonardr
4 message(s) in mailbox “/var/spool/mail/leonardr”:
DON’T DELETE THIS MESSAGE -- FOLDER INTERNAL DATA
This is a test message #1
This is a test message #2
This is a test message #3
324

37.6 Page 366

▲back to top
Network Programming
The first message isn’t a real message; it’s a dummy message sometimes created when you use a mail
client to read your spool file. If your application works on spool files that are sometimes accessed
through other means, you’ll need to recognize and deal with that kind of message.
Fetching Mail from a POP3 Server with poplib
Parsing a local mail spool didn’t require going over the network, because you ran the script on the same
machine that had the mail spool. There was no need to involve a network protocol, only a file format
(the format of Unix mailboxes, derived mainly from RFC 2822).
However, most people don’t have a Unix shell account on their mail server (or if they do, they want to
read mail on their own machine instead of on the server). To fetch mail from your mail server, you need
to go over a network, which means you must use a protocol. There are two popular protocols for doing
this. The first, which was once near-universal though now waning in popularity, is POP3, the third revi-
sion of the Post Office Protocol.
POP3 is defined in RFC 1939, but as with most popular Internet protocols, you don’t need to delve
very deeply into the details, because Python includes a module that wraps the protocol around a
Python interface.
Here’s POP3SubjectLister, a POP3-based implementation of the same idea as the mailbox parser
script. This script prints the subject line of each message on the server:
#!/usr/bin/python
from poplib import POP3
import email
class SubjectLister(PpOP3):
“””Connect to a POP3 mailbox and list the subject of every message
in the mailbox.”””
def __init__(self, server, username, password):
“Connect to the POP3 server.”
POP3.__init__(self, server, 110)
#Uncomment this line to see the details of the POP3 protocol.
#self.set_debuglevel(2)
self.user(username)
response = self.pass_(password)
if response[:3] != ‘+OK’:
#There was a problem connecting to the server.
raise Exception, response
def summarize(self):
“Retrieve each message, parse it, and print the subject.”
numMessages = self.stat()[0]
print ‘%d message(s) in this mailbox.’ % numMessages
parser = email.Parser.Parser()
for messageNum in range(1, numMessages+1):
messageString = ‘\\n’.join(self.top(messageNum, 0)[1])
message = parser.parsestr(messageString)
325

37.7 Page 367

▲back to top
Chapter 16
#Passing in True to parser.parsestr() will only parse the headers
#of the message, not the body. Since all we care about is the
#body, this will save some time. However, this is only
#supported in Python 2.2.2 and up.
#message = parser.parsestr(messageString, True)
print ‘’, message[‘Subject’]
After the data is on this side of the network, there’s no fundamental difference between the way it’s
handled with this script and the one based on the UnixMailbox class. As with the UnixMailbox
script, we use the email module to parse each message into a Python data structure (although here,
we use the Parser class, defined in the email.Parser module, instead of the message_from_file
convenience function).
The downside of using POP3 for this purpose is that the POP3.retr method has side effects. When you
call retr on a message on the server, the server marks that message as having been read. If you use a
mail client or a program like fetchmail to retrieve new mail from the POP3 server, then running this
script might confuse the other program. The message will still be on the server, but your client might not
download it if it thinks the message has already been read.
POP3 also defines a command called top, which doesn’t mark a message as having been read and which
only retrieves the headers of a message. Both of these – top and retr – are ideal for the purposes of
this script; we’ll save bandwidth (not having to retrieve the whole message just to get the subject) and
your script won’t interfere with the operation of other programs that use the same POP3 mailbox.
Unfortunately, not all POP3 servers implement the top command correctly. Because it’s so useful when
implemented correctly, though, here’s a subclass of the SubjectLister class which uses the top com-
mand to get message headers instead of retrieving the whole message. If you know your server supports
top correctly, this is a better implementation:
class TopBasedSubjectLister(SubjectLister):
def summarize(self):
“””Retrieve the first part of the message and find the ‘Subject:’
header.”””
numMessages = self.stat()[0]
print ‘%d message(s) in this mailbox.’ % numMessages
for messageNum in range(1, numMessages+1):
#Just get the headers of each message. Scan the headers
#looking for the subject.
for header in self.top(messageNum, 0)[1]:
if header.find(‘Subject:’) == 0:
print header[len(‘Subject:’):]
break
Both SubjectLister and TopBasedSubjectLister will yield the same output, but you’ll find that
TopBasedSubjectLister runs a lot faster (assuming your POP3 server implements top correctly).
Finally, we’ll create a simple command-line interface to the POP3-based SubjectLister class, just as
we did for the MailboxSubjectLister.py. This time, however, you need to provide a POP3 server
and credentials on the command line, instead of the path to a file on disk:
326

37.8 Page 368

▲back to top
Network Programming
if __name__ == ‘__main__’:
import sys
if len(sys.argv) < 4:
print ‘Usage: %s [POP3 hostname] [POP3 user] [POP3 password]’ % sys.argv[0]
sys.exit(0)
lister = TopBasedSubjectLister(sys.argv[1], sys.argv[2], sys.argv[3])
lister.summarize()
Try It Out Printing a Summary of Your POP3 Mailbox
Run POP3SubjectLister.py with the credentials for a POP server, and you’ll get a list of subjects:
$ python POP3SubjectLister.py pop.example.com [username] [password]
3 message(s) in this mailbox.
This is a test message #1
This is a test message #2
This is a test message #3
When you go through the POP3 server, you won’t get the dummy message you might get when parsing
a raw Unix mailbox file, as shown previously. Mail servers know that that message isn’t really a mes-
sage; the Unix mailbox parser treats it as one.
How It Works
The SubjectLister object (or its TopBasedSubjectLister subclass) connects to the POP3 server and
sends a “stat” command to get the number of messages in the mailbox. A call to stat returns a tuple con-
taining the number of messages in the mailbox, and the total size of the mailbox in bytes. The lister then
iterates up to this number, retrieving every message (or just the headers of every message) as it goes.
If SubjectLister is in use, the message is parsed with the email module’s Parser utility class, and
the Subject header is extracted from the resulting Message or MIMEMultipart object. If
TopBasedSubjectLister is in use, no parsing is done: The headers are retrieved from the server as a
list and scanned for a “Subject” header.
Fetching Mail from an IMAP Server with imaplib
The other protocol for accessing a mailbox on a remote server is IMAP, the Internet Message Access
Protocol. The most recent revision of IMAP is defined in RFC 3501, and it has significantly more features
than POP3. It’s also gaining in popularity over POP3.
The main difference between POP3 and IMAP is that POP3 is designed to act like a mailbox: It just holds
your mail for a while until you collect it. IMAP is designed to keep your mail permanently stored on the
server. Among other things, you can create folders on the server, sort mail into them, and search them.
These are more complex features that are typically associated with end-user mail clients. With IMAP, a
mail client only needs to expose these features of IMAP; it doesn’t need to implement them on its own.
Keeping your mail on the server makes it easier to keep the same mail setup while moving from com-
puter to computer. Of course, you can still download mail to your computer and then delete it from the
server, as with POP3.
327

37.9 Page 369

▲back to top
Chapter 16
Here’s IMAPSubjectLister.py, an IMAP version of the script we’ve already written twice, which
prints out the subject lines of all mail on the server. IMAP has more features than POP3, so this script
exercises proportionately fewer of them. However, even for the same functionality, it’s a great improve-
ment over the POP3 version of the script. IMAP saves bandwidth by retrieving the message subjects and
nothing else: a single subject header per message. Even when POP3’s top command is implemented cor-
rectly, it can’t do better than fetching all of the headers as a group.
What’s the catch? As the imaplib module says of itself, “to use this module, you must read the RFCs
pertaining to the IMAP4 protocol.” The imaplib module provides a function corresponding to each of
the IMAP commands, but it doesn’t do many transformations between the Python data structures you’re
used to creating and the formatted strings used by the IMAP protocol. You’ll need to keep a copy of RFC
3501 on hand or you won’t know what to pass into the imaplib methods.
For instance, to pass a list of message IDs into imaplib, you need to pass in a string like “1,2,3”, not the
Python list (1,2,3). To make sure only the subject is pulled from the server, IMAPSubjectLister.py passes
the string “(BODY[HEADER.FIELDS (SUBJECT)])” as an argument to an imaplib method. The result of
that command is a nested list of formatted strings, only some of which are actually useful to the script.
This is not exactly the kind of intuitiveness one comes to expect from Python. imaplib is certainly use-
ful, but it doesn’t do a very good job of hiding the details of IMAP from the programmer:
#!/usr/bin/python
from imaplib import IMAP4
class SubjectLister(IMAP4):
“””Connect to an IMAP4 mailbox and list the subject of every message
in the mailbox.”””
def __init__(self, server, username, password):
“Connect to the IMAP server.”
IMAP4.__init__(self, server)
#Uncomment this line to see the details of the IMAP4 protocol.
#self.debug = 4
self.login(username, password)
def summarize(self, mailbox=’Inbox’):
“Retrieve the subject of each message in the given mailbox.”
#The SELECT command makes the given mailbox the ‘current’ one,
#and returns the number of messages in that mailbox. Each message
#is accessible via its message number. If there are 10 messages
#in the mailbox, the messages are numbered from 1 to 10.
numberOfMessages = int(self._result(self.select(mailbox)))
print ‘%s message(s) in mailbox “%s”:’ % (numberOfMessages, mailbox)
#The FETCH command takes a comma-separated list of message
#numbers, and a string designating what parts of the
#message you want. In this case, we want only the
#’Subject’ header of the message, so we’ll use an argument
#string of ‘(BODY[HEADER.FIELDS (SUBJECT)])’.
#
#See section 6.4.5 of RFC3501 for more information on the
#format of the string used to designate which part of the
#message you want. To get the entire message, in a form
328

37.10 Page 370

▲back to top
Network Programming
#acceptable to the email parser, ask for ‘(RFC822)’.
subjects = self._result(self.fetch(‘1:%d’ % numberOfMessages,
‘(BODY[HEADER.FIELDS (SUBJECT)])’))
for subject in subjects:
if hasattr(subject, ‘__iter__’):
subject = subject[1]
print ‘’, subject[:subject.find(‘\\n’)]
def _result(self, result):
“””Every method of imaplib returns a list containing a status
code and a set of the actual result data. This convenience
method throws an exception if the status code is other than
“OK”, and returns the result data if everything went all
right.”””
status, result = result
if status != ‘OK’:
raise status, result
if len(result) == 1:
result = result[0]
return result
if __name__ == ‘__main__’:
import sys
if len(sys.argv) < 4:
print ‘Usage: %s [IMAP hostname] [IMAP user] [IMAP password]’ % sys.argv[0]
sys.exit(0)
lister = SubjectLister(sys.argv[1], sys.argv[2], sys.argv[3])
lister.summarize()
Try It Out Printing a Summary of Your IMAP Mailbox
Just execute IMAPSubjectLister.py with your IMAP credentials (just as with POP3SubjectLister),
and you’ll get a summary similar to the two shown earlier in this chapter:
$ python IMAPSubjectLister.py imap.example.com [username] [password]
3 message(s) in mailbox “Inbox”:
This is a test message #1
This is a test message #2
This is a test message #3
How It Works
As with the POP3 example, the first thing to do is connect to the server. POP3 servers provide only
one mailbox per user, but IMAP allows one user any number of mailboxes, so the next step is to select
a mailbox.
The default mailbox is called “Inbox”, and selecting a mailbox yields the number of messages in that
mailbox (some POP3 servers, but not all, return the number of messages in the mailbox when you
connect to the server).
Unlike with POP3, IMAP lets you retrieve more than one message at once. It also gives you a lot of flexi-
bility in defining which parts of a message you want. The IMAP-based SubjectLister makes just one
IMAP call to retrieve the subjects (and only the subjects) of every message in the mailbox. Then it’s just a
329

38 Pages 371-380

▲back to top

38.1 Page 371

▲back to top
Chapter 16
matter of iterating over the list and printing out each subject. The real trick is knowing what arguments
to pass into imaplib and how to interpret the results.
IMAP’s Unique Message IDs
Complaints about imaplib’s user-friendliness aside, you might have problems writing IMAP scripts if you
assume that the message numbers don’t change over time. If another IMAP client deletes messages from a
mailbox while this script is running against it (suppose you have your mail client running, and you use it to
delete some spam while this script is running), the message numbers will be out of sync from that point on.
The IMAP-based SubjectLister class minimizes this risk by getting the subject of every message in
one operation, immediately after selecting the mailbox:
self.fetch(‘1:%d’ % numberOfMessages, ‘(BODY[HEADER.FIELDS (SUBJECT)])’)
If there are 10 messages in the inbox, the first argument to fetch will be “1:10”. This is a slice of the
mailbox, similar to a slice of a Python list, which returns all of the messages: message 1 through message
10 (IMAP and POP3 messages are numbered starting from 1).
Getting the data you need as soon as you connect to the server minimizes the risk that you’ll pass a no-
longer-valid message number onto the server, but you can’t always do that. You may write a script that
deletes a mailbox’s messages, or that files them in a second mailbox. After you change a mailbox, you
may not be able to trust the message numbers you originally got.
Try It Out Fetching a Message by Unique ID
To help you avoid this problem, IMAP keeps a unique ID (UID) for every message under its control.
You can fetch the unique IDs from the server and use them in subsequent calls using imaplib’s uid
method. Unfortunately, this brings you even closer to the details of the IMAP protocol. The IMAP4 class
defines a separate method for each IMAP command (e.g. IMAP4.fetch, IMAP4.search, etc.), but when
you’re dealing with IDs, you can’t use those methods. You can use only the IMAP4.uid method, and
you must pass the IMAP command you want as the first argument. For instance, instead of calling
IMAP4.fetch([arguments]), you must call IMAP4.uid(‘FETCH’, [arguments]).
>>> import imaplib
>>> import email
>>> imap = imaplib.IMAP4(‘imap.example.com’)
>>> imap.login(‘[username]’, ‘[password]’)
(‘OK’, [‘Logged in.’])
>>> imap.select(‘Inbox’)[1][0]
‘3’
>>>
>>> #Get the unique IDs for the messages in this folder.
... uids = imap.uid(‘SEARCH’, ‘ALL’)
>>> print uids
(‘OK’, [‘49532 49541 49563’])
>>>
>>> #Get the first message.
... uids = uids[1][0].split(‘ ‘)
>>> messageText = imap.uid(‘FETCH’, uids[0], “(RFC822)”)[1][0][1]
>>> message = email.message_from_string(messageText)
>>> print message[‘Subject’]
This is a test message #1
330

38.2 Page 372

▲back to top
Network Programming
How It Works
Getting a message by unique ID requires four IMAP commands. First and second, the client must con-
nect to the server and select a mailbox, just as in the previous IMAP example. Third, the client needs to
run a SEARCH command that returns a list of message UIDs. Finally, the client can pass in one of the
UIDs to a FETCH command and get the actual message.
The last two steps both go through the IMAP4.uid method; if UIDs weren’t involved, they would use
the search and fetch methods, respectively.
Using imaplib to interact with an IMAP server can be a pain, but it’s not as bad as communicating
directly with the server.
POP3 servers also support UIDs, though it’s less common for multiple clients to access a single POP3
mailbox simultaneously. A POP3 object’s uidl method will retrieve the UIDs of the messages in its mail-
box. You can then pass a UID into any of a POP3 object’s other methods that take message IDs: for
instance, retr and top. IMAP’s UIDs are numeric; POP3’s are the “message digests”: hexadecimal sig-
natures derived from the contents of each message.
Secure POP3 and IMAP
Both the POP3 or IMAP examples covered earlier in this section have a security problem: They send
send your username and password over the network without encrypting it. That’s why both POP and
IMAP are often run atop the Secure Socket Layer (SSL). This is a generic encryption layer also used to
secure HTTP connections on the World Wide Web. POP and IMAP servers that support SSL run on dif-
ferent ports from the ones that don’t: The standard port number for POP over SSL is 995 instead of 23,
and IMAP over SSL uses port 993 instead of port 143.
If your POP3 or IMAP server supports SSL, you can get an encrypted connection to it by just swapping
out the POP3 or IMAP4 class for the POP3_SSL or IMAP4_SSL class. Each SSL class is in the same module
and has the same interface as its insecure counterpart but encrypts all data before sending it over the
network.
Webmail Applications Are Not E-mail Applications
If you use a webmail system such as Yahoo! Mail or Gmail, you’re not technically using a mail applica-
tion at all: You’re using a web application that happens to have a mail application on the other side. The
scripts in this section won’t help you fetch mail from or send mail through these services, because they
implement HTTP, not any of the e-mail protocols (however, Yahoo! Mail offers POP3 access for a fee).
Instead, you should look at Chapter 21 for information on how web applications work.
The libgmail project aims to create a Python interface to Gmail, one that can treat Gmail as an SMTP,
POP3, or IMAP server. The libgmail homepage is at http://libgmail.sourceforge.net/.
Socket Programming
So far, we’ve concerned ourselves with the protocols and file formats surrounding a single Internet
application: e-mail. E-mail is certainly a versatile and useful application, but e-mail-related protocols
account for only a few of the hundreds implemented atop the Internet Protocol. Python makes it easier
331

38.3 Page 373

▲back to top
Chapter 16
to use the e-mail-related protocols (and a few other protocols not covered in this chapter) by providing
wrapper libraries, but Python doesn’t come with a library for every single Internet protocol. It certainly
won’t have one for any new protocols you decide to create for your own Internet applications.
To write your own protocols, or to implement your own Python libraries along the lines of imaplib or
poplib, you’ll need to go down a level and learn how programming interfaces to IP-based protocols
actually work. Fortunately, it’s not hard to write such code: smtplib, poplib, and the others do it with-
out becoming too complicated. The secret is the socket library, which makes reading and writing to a
network interface look a lot like reading and writing to files on disk.
Introduction to Sockets
In many of the previous examples, you connected to a server on a particular port of a particular machine
(for instance, port 25 of localhost for a local SMTP server). When you tell imaplib or smtplib to con-
nect to a port on a certain host, behind the scenes Python is opening a connection to that host and port.
Once the connection is made, the server opens a reciprocal connection to your computer. A single Python
“socket” object hides the outgoing and incoming connections under a single interface. A socket is like a
file you can read to and write from at the same time.
To implement a client for a TCP/IP-based protocol, you open a socket to an appropriate server. You
write data to the socket to send it to the server, and read from the socket the data the server sends you.
To implement a server, it’s just the opposite: You bind a socket to a hostname and a port and wait for a
client to connect to it. Once you have a client on the line, you read from your socket to get data from
the client, and write to the socket to send data back.
It takes an enormous amount of work to send a single byte over the network, but between TCP/IP and
the socket library, you get to skip almost all of it. You don’t have to figure out how to get your data
halfway across the world to its destination, because TCP/IP handles that for you. Nor need you worry
about turning your data into TCP/IP packets, because the socket library handles that for you.
Just as e-mail and the web are the killer apps for the use of the Internet, sockets might be considered the
killer app for the adoption of TCP/IP. Sockets were introduced in an early version of BSD UNIX, but
since then just about every TCP/IP implementation has used sockets as its metaphor for how to write
network programs. Sockets make it easy to use TCP/IP (at least, easier than any alternative), and this
has been a major driver of TCP/IP’s popularity.
As a first socket example, here’s a super-simple socket server, SuperSimpleSocketServer.py:
#!/usr/bin/python
import socket
import sys
if len(sys.argv) < 3:
print ‘Usage: %s [hostname] [port number]’ % sys.argv[0]
sys.exit(1)
hostname = sys.argv[1]
port = int(sys.argv[2])
#Set up a standard Internet socket. The setsockopt call lets this
332

38.4 Page 374

▲back to top
Network Programming
#server use the given port even if it was recently used by another
#server (for instance, an earlier incarnation of
#SuperSimpleSocketServer).
sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
sock.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
#Bind the socket to a port, and bid it listen for connections.
sock.bind((hostname, port))
sock.listen(1)
print “Waiting for a request.”
#Handle a single request.
request, clientAddress = sock.accept()
print “Received request from”, clientAddress
request.send(‘-=SuperSimpleSocketServer 3000=-\\n’)
request.send(‘Go away!\\n’)
request.shutdown(2) #Stop the client from reading or writing anything.
print “Have handled request, stopping server.”
sock.close()
This server will serve only a single request. As soon as any client connects to the port to which it’s
bound, it will tell the client to go away, close the connection, stop serving requests, and exit.
Try It Out Connecting to the SuperSimpleSocketServer with Telnet
The telnet program is a very simple client for TCP/IP applications. You invoke it with a hostname and a
port; it connects you to that port; and then you’re on your own. Anything you type is sent over a socket
to the server, and anything the server sends over the socket is printed to your terminal. Telnet is included
as a command-line program in Windows, Mac OS X, and Unix installations, so you shouldn’t have trou-
ble getting it.
Because our example socket server doesn’t really do anything, there’s little point in writing a custom
client for it. To test it out, just start up the server:
$ python SuperSimpleSocketServer.py localhost 2000
Waiting for a request.
Then, in a separate terminal, telnet into the server:
$ telnet localhost 2000
Trying 127.0.0.1...
Connected to rubberfish.
Escape character is ‘^]’.
-=SuperSimpleSocketServer 3000=-
Go away!
Connection closed by foreign host.
Go back to the terminal on which you ran the server and you should see output similar to this:
Received request from (‘127.0.0.1’, 32958)
Have handled request, stopping server.
333

38.5 Page 375

▲back to top
Chapter 16
How It Works
When you started the SuperSimpleSocketServer, you bound the process to port 2000 of the “local-
host” hostname. When that script called socket.accept, it stopped running and began to “block” on
socket input, waiting for someone to connect to the server.
When your telnet command opens up a TCP/IP connection to the SuperSimpleSocketServer, the
socket.accept method call returns from its wait. At last, someone has connected to the server! The
return values of socket.accept give the server the tools it needs to communicate with this client: a
socket object and a tuple describing the network address of the client. The server sends some data to the
client through the socket and then shuts down. No further socket connections will be accepted.
The only obscure thing here is that client address tuple: (‘127.0.0.1’, 32958). You’ve seen 127.0.0.1
already; it is a special IP address that refers to “this computer”: it’s the IP address equivalent of “local-
host”. A connection to the server from 127.0.0.1 means that the client is coming from the same computer
that’s running the server. If you’d telnetted in from another machine, that machine’s IP address would
have shown up instead.
32958 is a temporary or “ephemeral” port number for the client. Recall that what looks like a single,
bidirectional “socket” object actually contains two unidirectional connections: one from the client to the
server and one from the server to the client. Port 2000 on localhost, the port to which the server was
bound when we started it up, is the destination for all client data (not that this client got a chance to
send any data). The data sent by the server must also have a destination hostname and port, but not a
predefined one. While a server port is usually selected by the human in charge of the server, ephemeral
ports are selected by the client’s operating system. Run this exercise again and you’ll see that each indi-
vidual TCP/IP connection is given a different ephemeral port number.
Binding to an External Hostname
If you tried to telnet into the SuperSimpleSocketServer from another machine, as suggested above,
you might have noticed that you weren’t able to connect to the server. If so, it may be because you
started the server by binding it to localhost. The special “localhost” hostname is an internal host-
name, one that can’t be accessed from another machine. After all, from someone else’s perspective,
“localhost” means their computer, not yours.
This is actually very useful because it enables you to test out the servers from this chapter (and Chapter 21)
without running the risk of exposing your computer to connections from the Internet at large (of course, if
you are running these servers on a multiuser machine, you might have to worry about the other users on
the same machine, so try to run these on a system that you have to yourself). However, when it comes time
to host a server for real, and external connections are what you want, you need to bind your server to an
external hostname.
If you can log into your computer remotely via SSH, or you already run a web server, or you ever make
a reference to your computer from another one, you already know an external hostname for your com-
puter. On the other hand, if you have a dial-up or broadband connection, you’re probably assigned a
hostname along with an IP address whenever you connect to your ISP. Find your computer’s IP address
and do a DNS lookup on it to find an external hostname for your computer. If all else fails, you can bind
servers directly to your external IP address (not 127.0.0.1, as that will have the same problem as binding
to “localhost”).
334

38.6 Page 376

▲back to top
Network Programming
If you bind a server to an external hostname and still can’t connect to it from the outside, there may be a
firewall in the way. Fixing that is beyond what this book can cover. You should ask your local computer
guru to help you with this.
The Mirror Server
Here’s a server that’s a little more complex (though not more useful) and that shows how Python enables
you to treat socket connections like files. This server accepts lines of text from a socket, just as a script
might on standard input. It reverses the text and writes the reversed version back through the socket, just
as a script might on standard output. When it receives a blank line, it terminates the connection:
#!/usr/bin/python
import socket
class MirrorServer:
“””Receives text on a line-by-line basis and sends back a reversed
version of the same text.”””
def __init__(self, port):
“Binds the server to the given port.”
self.socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
self.socket.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
self.socket.bind(port)
#Queue up to five requests before turning clients away.
self.socket.listen(5)
def run(self):
“Handles incoming requests forever.”
while True:
request, client_address = self.socket.accept()
#Turn the incoming and outgoing connections into files.
input = request.makefile(‘rb’, 0)
output = request.makefile(‘wb’, 0)
l = True
try:
while l:
l = input.readline().strip()
if l:
output.write(l[::-1] + ‘\\r\\n’)
else:
#A blank line indicates a desire to terminate the
#connection.
request.shutdown(2) #Shut down both reads and writes.
except socket.error:
#Most likely the client disconnected.
pass
if __name__ == ‘__main__’:
import sys
if len(sys.argv) < 3:
print ‘Usage: %s [hostname] [port number]’ % sys.argv[0]
sys.exit(1)
hostname = sys.argv[1]
port = int(sys.argv[2])
MirrorServer((hostname, port)).run()
335

38.7 Page 377

▲back to top
Chapter 16
Try It Out Mirroring Text with the MirrorServer
As with the SuperSimpleSocketServer, you can use this without writing a specialized client. You can
just telnet into the MirrorServer and enter some text. Enter a blank line and the server will disconnect
you. In one terminal, start the server:
$ python MirrorServer.py localhost 2000
In another, telnet into the server as a client:
$ telnet localhost 2000
Trying 127.0.0.1...
Connected to rubberfish.
Escape character is ‘^]’.
Hello.
.olleH
Mirror this text!
!txet siht rorriM
Connection closed by foreign host.
$
The Mirror Client
Though you’ve just seen that the mirror server is perfectly usable through telnet, not everyone is com-
fortable using telnet. What we need is a flashy mirror server client with bells and whistles, so that even
networking novices can feel the thrill of typing in text and seeing it printed out backward. Here’s a sim-
ple client that takes command-line arguments for the server destination and the text to reverse. It con-
nects to the server, sends the data, and prints the reversed text:
#!/usr/bin/python
import socket
class MirrorClient:
“A client for the mirror server.”
def __init__(self, server, port):
“Connect to the given mirror server.”
self.socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
self.socket.connect((server, port))
def mirror(self, s):
“Sends the given string to the server, and prints the response.”
if s[-1] != ‘\\n’:
s += ‘\\r\\n’
self.socket.send(s)
#Read server response in chunks until we get a newline; that
#indicates the end of the response.
buf = []
input = ‘’
336

38.8 Page 378

▲back to top
Network Programming
while not ‘\\n’ in input:
try:
input = self.socket.recv(1024)
buf.append(input)
except socket.error:
break
return ‘’.join(buf)[:-1]
def close(self):
self.socket.send(‘\\r\\n’) #We don’t want to mirror anything else.
self.socket.close()
if __name__ == ‘__main__’:
import sys
if len(sys.argv) < 4:
print ‘Usage: %s [host] [port] [text to be mirrored]’ % sys.argv[0]
sys.exit(1)
hostname = sys.argv[1]
port = int(sys.argv[2])
toMirror = sys.argv[3]
m = MirrorClient(hostname, port)
print m.mirror(toMirror)
m.close()
The mirror server turns its socket connection into a pair of files, but this client reads from and writes to
the socket directly. There’s no compelling reason for this; I just felt this chapter should include at least
one example that used the lower-level socket API. Note how the server response is read in chunks, and
each chunk is scanned for the newline character that indicates the end of the response. If this example
had created a file for the incoming socket connection, that code would have been as simple as calling
input.readline.
It’s important to know when the response has ended, because calling socket.recv (or input.readline)
will block your process until the server sends some more data. If the server is waiting for more data from the
client, your process will block forever. (See the sections below on select “Single-Threaded Multitasking
with select” and “The Twisted Framework” for ways of avoiding this problem.)
SocketServer
Sockets are very useful, but Python isn’t satisfied with providing the same C-based socket interface you
can get with most languages on most operating systems. Python goes one step further and provides
SocketServer, a module full of classes that let you write sophisticated socket-based servers with
very little code.
Most of the work in building a SocketServer is defining a request handler class. This is a subclass of
the SocketServer module’s BaseRequestHandler class, and the purpose of each request handler
object is to handle a single client request for as long as the client is connected to the server. This is
implemented in the handler’s handle method. The handler may also define per-request setup and
tear-down code by overriding setup and finish.
337

38.9 Page 379

▲back to top
Chapter 16
The methods of a BaseRequestHandler subclass have access to the following three members:
❑ request: A socket object representing the client request: the same object obtained from
socket.accept in the MirrorServer example.
❑ client_address: A 2-tuple containing the hostname and port to which any data the server outputs
will be sent. The other object obtained from socket.accept in the MirrorServer example.
❑ server: A reference to the SocketServer that created the request handler object.
By subclassing StreamRequestHandler instead of BaseRequestHandler, you also get access to the
file-like objects that let you read from and write to the socket connection. BaseRequestHandler gives
you access to two other members:
❑ rfile: The file corresponding to the data that comes in over the socket (from the client if you’re
writing a server, from the server if you’re writing a client). Equivalent to what you get when
you call request.makefile(‘rb’).
❑ wfile: The file corresponding to the data that you send over the socket (to the client if you’re
writing a server, to the server if you’re writing a client). Equivalent to what you get when you
call request.makefile(‘wb’).
By rewriting the MirrorServer as a SocketServer server (specifically, a TCPServer), you can elimi-
nate a lot of code to do with socket setup and teardown, and focus on the arduous task of reversing text.
Here’s MirrorSocketServer.py:
#!/usr/bin/python
import SocketServer
class RequestHandler(SocketServer.StreamRequestHandler):
“Handles one request to mirror some text.”
def handle(self):
“””Read from StreamRequestHandler’s provided rfile member,
which contains the input from the client. Mirror the text
and write it to the wfile member, which contains the output
to be sent to the client.”””
l = True
while l:
l = self.rfile.readline().strip()
if l:
self.wfile.write(l[::-1] + ‘\\n’)
if __name__ == ‘__main__’:
import sys
\\ if len(sys.argv) < 3:
print ‘Usage: %s [hostname] [port number]’ % sys.argv[0]
sys.exit(1)
hostname = sys.argv[1]
port = int(sys.argv[2])
SocketServer.TCPServer((hostname, port), RequestHandler).serve_forever()
338

38.10 Page 380

▲back to top
Network Programming
Almost all of the socket-specific code is gone. Whenever anyone connects to this server, the TCPServer
class will create a new RequestHandler with the appropriate members and call its handle method to
handle the request.
The MirrorClient we wrote earlier will work equally well with this server, because across the network
both servers take the same input and yield the same output. The same principle applies as when you
change the implementation of a function in a module to get rid of redundant code but leave the interface
the same.
Multithreaded Servers
One problem with both of these implementations of the mirror server is that only one client at a time can
connect to a running server. If you open two telnet sessions to a running server, the second session won’t
finish connecting until you close the first one. If real servers worked this way, nothing would ever get
done. That’s why most real servers spawn threads or subprocesses to handle multiple connections.
The SocketServer module defines two useful classes for handling multiple connections at once:
ThreadingMixIn and ForkingMixIn. A SocketServer class that subclasses ThreadingMixIn will
automatically spawn a new thread to handle each incoming request. A subclass of ForkingMixIn
will automatically fork a new subprocess to handle each incoming request. I prefer ThreadingMixIn
because threads are more efficient and more portable than subprocesses. It’s also much easier to write
code for a thread to communicate with its parent than for a subprocess to communicate with its parent.
See Chapter 9 for an introduction to threads and subprocesses.
Here’s MultithreadedMirrorServer.py, a multithreaded version of the MirrorSocketServer. Note
that it uses the exact same RequestHandler definition as MirrorSocketServer.py. The difference
here is that instead of running a TCPServer, we run a ThreadingTCPServer, a standard class that
inherits both from ThreadingMixIn and TCPServer:
#!/usr/bin/python
import SocketServer
class RequestHandler(SocketServer.StreamRequestHandler):
“Handles one request to mirror some text.”
def handle(self):
“””Read from StreamRequestHandler’s provided rfile member,
which contains the input from the client. Mirror the text
and write it to the wfile member, which contains the output
to be sent to the client.”””
l = True
while l:
l = self.rfile.readline().strip()
if l:
self.wfile.write(l[::-1] + ‘\\n’)
if __name__ == ‘__main__’:
import sys
if len(sys.argv) < 3:
print ‘Usage: %s [hostname] [port number]’ % sys.argv[0]
339

39 Pages 381-390

▲back to top

39.1 Page 381

▲back to top
Chapter 16
sys.exit(1)
hostname = sys.argv[1]
port = int(sys.argv[2])
server = SocketServer.ThreadingTCPServer((hostname, port), RequestHandler)
server.serve_forever()
With this server running, you can run a large number of telnet sessions and MirrorClient sessions in
parallel. ThreadingMixIn hides the details of spawning threads, just as TCPServer hides the details of
sockets. The goal of all these helper classes is to keep your focus on what you send and receive over the
network.
The Python Chat Server
For the mirror server, the capability to support multiple simultaneous connections is useful but it doesn’t
change what the server actually does. Each client interacts only with the server, and not even indirectly
with the other clients. This model is a popular one; web servers and mail servers use it, among others.
There is another type of server, though, that exists to connect clients to each other. For many applica-
tions, it’s not the server that’s interesting: it’s who else is connected to it. The most popular applications
of this sort are online chat rooms and games. In this section, you’ll design and build a simple chat server
and client.
Perhaps the original chat room was the (non-networked) Unix wall command, which enables you to
broadcast a message to everyone logged in on a Unix system. Internet Relay Chat, invented in 1988 and
described in RFC 1459, is the most popular TCP/IP-based chat room software. The chat software you
write here will have some of the same features as IRC, although it won’t be compatible with IRC.
Design of the Python Chat Server
In IRC, a client that connects to a server must provide a nickname: a short string identifying the person
who wants to chat. A nickname must be unique across a server so that users can’t impersonate one
another. Our server will carry on this tradition.
An IRC server provides an unlimited number of named channels, or rooms, and each user can join any
number of rooms. Our server will provide only a single, unnamed room, which all connected users will
inhabit.
Entering a line of text in an IRC client broadcasts it to the rest of your current room, unless it starts with
the slash character. A line starting with the slash character is treated as a command to the server. Our
server will act the same way.
IRC implements a wide variety of server commands: For instance, you can use a server command to
change your nickname, join another room, send a private message to another user, or try to send a file to
another user.
340

39.2 Page 382

▲back to top
Network Programming
For example, if you issue the command /nick leonardr to an IRC server, you’re attempting to change
your nickname from its current value to leonardr. Your attempt might or might not succeed, depend-
ing on whether or not there’s already a leonardr on the IRC server.
Our server will support the following three commands, taken from IRC and simplified:
❑ /nick [nickname]: As described above, this attempts to change your nickname. If the nickname
is valid and not already taken, your nickname will be changed and the change will be
announced to the room. Otherwise, you’ll get a private error message.
❑ /quit [farewell message]: This command disconnects the user from the chat server. Your
farewell message, if any, will be broadcast to the room.
❑ /names: This retrieves the nicknames of the users in the chat room as a space-separated string.
The Python Chat Server Protocol
Having decided on a feature set and a design, we must now define an application-specific protocol for
our Python Chat Server. This protocol will be similar to SMTP, HTTP, and the IRC protocol in that it will
run atop TCP/IP to provide the structure for a specific type of application. However, it will be much
simpler than any of those protocols.
The mirror server also defined a protocol, though it was so simple it may have escaped notice. The mir-
ror server protocol consists of three simple rules:
1. Send lines of text to the server.
2. Every time you send a newline, the server will send you back that line of text, reversed, with a
newline at the end.
3. Send a blank line to terminate the connection.
The protocol for the Python Chat Server will be a little more complex than that, but by the standards of
protocol design it’s still a fairly simple protocol. The following description is more or less the informa-
tion that would go into an RFC for this protocol. If we were actually writing an RFC, we would go into a
lot more detail and provide a formal definition of the protocol; that’s not as necessary here, because the
protocol definition will be immediately followed by an implementation in Python.
Of course, if we did write an RFC for this, it wouldn’t be accepted. The IRC protocol already has an
RFC, and it’s a much more useful protocol than this example one.
Our Hypothetical Protocol in Action
One good way to figure out the problems involved in defining a protocol is to write a sample session
to see what the client and server need to say to each other. Here’s a sample session of the Python Chat
Server. In the following transcript, a user nicknamed leonardr connects to a chat room in which a
shady character nicknamed pnorton is already lurking. The diagram shows what leonardr might send
to the server, what the server would send to him in response, and what it would send to the other client
(pnorton) as a result of leonardr’s input.
341

39.3 Page 383

▲back to top
Chapter 16
Me to the Server
leonardr
/names
Hello!
/nick pnorton
/nick leonard
Hello again!
/quit Goodbye
The Server to Me
Who are you?
Hello, leonardr, welcome to the
Python Chat Server.
pnorton leonardr
<leonardr> Hello!
There’s already a user named
pnorton here.
leonardr is now known as leonard
<leonard> Hello again!
The Server to pnorton
leonardr has joined the chat.
<leonardr> Hello!
leonardr is now known as leonard
<leonard> Hello again!
leonard has quit: Goodbye
Initial Connection
After establishing a connection between the client and server, the first stage of the protocol is to get a
nickname for the client. A client can’t be allowed into a chat room without a nickname because that
would be confusing to the other users. Therefore, the server will ask each new client: “Who are you?”
and expect a nickname in response, terminated by a newline. If what’s sent is an invalid nickname or the
nickname of a user already in the chat room, the server will send an error message and terminate the
connection. Otherwise, the server will welcome the client to the chat room and broadcast an announce-
ment to all other users that someone has joined the chat.
Chat Text
After a client is admitted into the chat room, any line of text they send will be broadcast to every user in
the room, unless it’s a server command. When a line of chat is broadcast, it will be prefaced with the nick-
name of the user who sent it, enclosed in angle brackets (e.g., “<leonardr> Hello, all.”). This will pre-
vent confusion about who said what, and visually distinguish chat messages from system messages.
Server Commands
If the client sends a recognized server command, the command is executed and a private system mes-
sage may be sent to that client. If the execution of the command changes the state of the chat room (for
instance, a user changes his nickname or quits), all users will receive a system message notifying them of
342

39.4 Page 384

▲back to top
Network Programming
the change (e.g., “leonardr is now known as leonard”). An unrecognized server command will result
in an error message for the user who sent it.
General Guidelines
For the sake of convenience and readability, the chat protocol is designed to have a line-based and
human-readable format. This makes the chat application usable even without a special client (although
we will write a special client to make chatting a little easier). Many TCP/IP protocols work in similar
ways, but it’s not a requirement. Some protocols send only binary data, to save bandwidth or because
they encrypt data before transmitting it.
Here’s the server code, in PythonChatServer.py. Like MultithreadedMirrorServer, its actual
server class is a ThreadingTCPServer. It keeps a persistent map of users’ nicknames that point to the
wfile members. That lets the server send those users data. This is how one user’s input can be broad-
cast to everyone in the chat room:
#!/usr/bin/python
import SocketServer
import re
import socket
class ClientError(Exception):
“An exception thrown because the client gave bad input to the server.”
pass
class PythonChatServer(SocketServer.ThreadingTCPServer):
“The server class.”
def __init__(self, server_address, RequestHandlerClass):
“””Set up an initially empty mapping between a user’s nickname
and the file-like object used to send data to that user.”””
SocketServer.ThreadingTCPServer.__init__(self, server_address,
RequestHandlerClass)
self.users = {}
class RequestHandler(SocketServer.StreamRequestHandler):
“””Handles the life cycle of a user’s connection to the chat
server: connecting, chatting, running server commands, and
disconnecting.”””
NICKNAME = re.compile(‘^[A-Za-z0-9_-]+$’) #Regex for a valid nickname
def handle(self):
“””Handles a connection: gets the user’s nickname, then
processes input from the user until they quit or drop the
connection.”””
self.nickname = None
self.privateMessage(‘Who are you?’)
nickname = self._readline()
done = False
try:
self.nickCommand(nickname)
343

39.5 Page 385

▲back to top
Chapter 16
self.privateMessage(‘Hello %s, welcome to the Python Chat Server.’\\
% nickname)
self.broadcast(‘%s has joined the chat.’ % nickname, False)
except ClientError, error:
self.privateMessage(error.args[0])
done = True
except socket.error:
done = True
#Now they’re logged in; let them chat.
while not done:
try:
done = self.processInput()
except ClientError, error:
self.privateMessage(str(error))
except socket.error, e:
done = True
def finish(self):
“Automatically called when handle() is done.”
if self.nickname:
#The user successfully connected before disconnecting.
#Broadcast that they’re quitting to everyone else.
message = ‘%s has quit.’ % self.nickname
if hasattr(self, ‘partingWords’):
message = ‘%s has quit: %s’ % (self.nickname,
self.partingWords)
self.broadcast(message, False)
#Remove the user from the list so we don’t keep trying to
#send them messages.
if self.server.users.get(self.nickname):
del(self.server.users[self.nickname])
self.request.shutdown(2)
self.request.close()
def processInput(self):
“””Reads a line from the socket input and either runs it as a
command, or broadcasts it as chat text.”””
done = False
l = self._readline()
command, arg = self._parseCommand(l)
if command:
done = command(arg)
else:
l = ‘<%s> %s\\n’ % (self.nickname, l)
self.broadcast(l)
return done
Each server command is implemented as a method. The _parseCommand method, defined later, takes a
line that looks like “/nick” and calls the corresponding method (in this case, nickCommand):
344

39.6 Page 386

▲back to top
Network Programming
#Below are implementations of the server commands.
def nickCommand(self, nickname):
“Attempts to change a user’s nickname.”
if not nickname:
raise ClientError, ‘No nickname provided.’
if not self.NICKNAME.match(nickname):
raise ClientError, ‘Invalid nickname: %s’ % nickname
if nickname == self.nickname:
raise ClientError, ‘You are already known as %s.’ % nickname
if self.server.users.get(nickname, None):
raise ClientError, ‘There\\’s already a user named “%s” here.’ %
nickname
oldNickname = None
if self.nickname:
oldNickname = self.nickname
del(self.server.users[self.nickname])
self.server.users[nickname] = self.wfile
self.nickname = nickname
if oldNickname:
self.broadcast(‘%s is now known as %s’ % (oldNickname, self.nickname))
def quitCommand(self, partingWords):
“””Tells the other users that this user has quit, then makes
sure the handler will close this connection.”””
if partingWords:
self.partingWords = partingWords
#Returning True makes sure the user will be disconnected.
return True
def namesCommand(self, ignored):
“Returns a list of the users in this chat room.”
self.privateMessage(‘, ‘.join(self.server.users.keys()))
# Below are helper methods.
def broadcast(self, message, includeThisUser=True):
“””Send a message to every connected user, possibly exempting the
user who’s the cause of the message.”””
message = self._ensureNewline(message)
for user, output in self.server.users.items():
if includeThisUser or user != self.nickname:
output.write(message)
def privateMessage(self, message):
“Send a private message to this user.”
self.wfile.write(self._ensureNewline(message))
def _readline(self):
“Reads a line, removing any whitespace.”
return self.rfile.readline().strip()
345

39.7 Page 387

▲back to top
Chapter 16
def _ensureNewline(self, s):
“Makes sure a string ends in a newline.”
if s and s[-1] != ‘\\n’:
s += ‘\\r\\n’
return s
def _parseCommand(self, input):
“””Try to parse a string as a command to the server. If it’s an
implemented command, run the corresponding method.”””
commandMethod, arg = None, None
if input and input[0] == ‘/’:
if len(input) < 2:
raise ClientError, ‘Invalid command: “%s”’ % input
commandAndArg = input[1:].split(‘ ‘, 1)
if len(commandAndArg) == 2:
command, arg = commandAndArg
else:
command, = commandAndArg
commandMethod = getattr(self, command + ‘Command’, None)
if not commandMethod:
raise ClientError, ‘No such command: “%s”’ % command
return commandMethod, arg
if __name__ == ‘__main__’:
import sys
if len(sys.argv) < 3:
print ‘Usage: %s [hostname] [port number]’ % sys.argv[0]
sys.exit(1)
hostname = sys.argv[1]
port = int(sys.argv[2])
PythonChatServer((hostname, port), RequestHandler).serve_forever()
The Python Chat Client
As with the mirror server, this chat server defines a simple, human-readable protocol. It’s possible to use
the chat server through telnet, but most people would prefer to use a custom client.
Here’s PythonChatClient.py, a simple text-based client for the Python Chat Server. It has a few
niceties that are missing when you connect with telnet. First, it handles the authentication stage on its
own: If you run it on a Unixlike system, you won’t even have to specify a nickname, because it will use
your account name as a default. Immediately after connecting, the Python Chat Client runs the /names
command and presents the user with a list of everyone in the chat room.
After connecting, this client acts more or less like a telnet client would. It spawns a separate thread to
handle user input from the keyboard even as it reads the server’s output from the network:
#!/usr/bin/python
import socket
import select
import sys
import os
from threading import Thread
346

39.8 Page 388

▲back to top
Network Programming
class ChatClient:
def __init__(self, host, port, nickname):
self.socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
self.socket.connect((host, port))
self.input = self.socket.makefile(‘rb’, 0)
self.output = self.socket.makefile(‘wb’, 0)
#Send the given nickname to the server.
authenticationDemand = self.input.readline()
if not authenticationDemand.startswith(“Who are you?”):
raise Exception, “This doesn’t seem to be a Python Chat Server.”
self.output.write(nickname + ‘\\r\\n’)
response = self.input.readline().strip()
if not response.startswith(“Hello”):
raise Exception, response
print response
#Start out by printing out the list of members.
self.output.write(‘/names\\r\\n’)
print “Currently in the chat room:”, self.input.readline().strip()
self.run()
def run(self):
“””Start a separate thread to gather the input from the
keyboard even as we wait for messages to come over the
network. This makes it possible for the user to simultaneously
send and receive chat text.”””
propagateStandardInput = self.PropagateStandardInput(self.output)
propagateStandardInput.start()
#Read from the network and print everything received to standard
#output. Once data stops coming in from the network, it means
#we’ve disconnected.
inputText = True
while inputText:
inputText = self.input.readline()
if inputText:
print inputText.strip()
propagateStandardInput.done = True
class PropagateStandardInput(Thread):
“””A class that mirrors standard input to the chat server
until it’s told to stop.”””
def __init__(self, output):
“””Make this thread a daemon thread, so that if the Python
interpreter needs to quit it won’t be held up waiting for this
thread to die.”””
Thread.__init__(self)
self.setDaemon(True)
self.output = output
self.done = False
347

39.9 Page 389

▲back to top
Chapter 16
def run(self):
“Echo standard input to the chat server until told to stop.”
while not self.done:
inputText = sys.stdin.readline().strip()
if inputText:
self.output.write(inputText + ‘\\r\\n’)
if __name__ == ‘__main__’:
import sys
#See if the user has an OS-provided ‘username’ we can use as a default
#chat nickname. If not, they have to specify a nickname.
try:
import pwd
defaultNickname = pwd.getpwuid(os.getuid())[0]
except ImportError:
defaultNickname = None
if len(sys.argv) < 3 or not defaultNickname and len(sys.argv) < 4:
print ‘Usage: %s [hostname] [port number] [username]’ % sys.argv[0]
sys.exit(1)
hostname = sys.argv[1]
port = int(sys.argv[2])
if len(sys.argv) > 3:
nickname = sys.argv[3]
else:
#We must be on a system with usernames, or we would have
#exited earlier.
nickname = defaultNickname
ChatClient(hostname, port, nickname)
A more advanced chat client might have a GUI that put incoming text in a separate window from the
text the user types, to keep input from being visually confused with output. As it is, in a busy chat room,
you might be interrupted by an incoming message while you’re typing, and lose your place.
Single-Threaded Multitasking with select
The reason PythonChatClient spawns a separate thread to gather user input is that a call to sys.
stdin.readline won’t return until the user enters a chat message or server command. A naïve chat
client might call sys.stdin.readline and wait for the user to type something in, but while it was
waiting the other users would keep chatting and the socket connection from the server would fill up
with a large backlog of chat. No chat messages would be displayed until the user pressed the Enter key
(causing sys.stdin.readline to return), at which time the whole backlog would come pouring onto
the screen. Trying to read from the socket connection would cause the opposite problem: The user would
be unable to enter any chat text until someone else in the chat room said something. Using two threads
avoids these problems: One thread can keep an eye on standard input while the other keeps an eye on
the socket connection.
348

39.10 Page 390

▲back to top
Network Programming
However, it’s possible to implement the chat client without using threads. (After all, telnet works more
or less the same way as PythonChatClient, and the telnet program is older than the idea of threads.)
The secret is to just peek at standard input and the socket connection — not trying to read from them,
just seeing if there’s anything to read. You do this by using the select function, provided by Python’s
select module.
select takes three lists of lists, and each second-level list contains file-type objects: one for objects you
read (like sys.stdin), one for objects to which you write (like sys.stdout), and one for objects to which
you write errors (like sys.stdout). By default, a call to select will block (wait for input) but only until
at least one of the file-type objects you passed in is ready to be used. It will then return three lists of lists,
which contain a subset of the objects you passed in: only the ones that are ready and have some data for
the program to pay attention to. You might think of select as acting sort of like Python’s built-in filter
function, filtering out the objects that aren’t ready for use. By using select, you can avoid the trap of
calling read on a file-type object that doesn’t have any data to read.
Here’s a subclass of ChatClient that uses a loop over select to check whether standard input or the
server input have unread data:
class SelectBasedChatClient(ChatClient):
def run(self):
“””In a tight loop, see whether the user has entered any input
or whether there’s any from the network. Keep doing this until
the network connection returns EOF.”””
socketClosed = False
while not socketClosed:
toRead, ignore, ignore = select.select([self.input, sys.stdin],
[], [])
#We’re not disconnected yet.
for input in toRead:
if input == self.input:
inputText = self.input.readline()
if inputText:
print inputText.strip()
else:
#The attempt to read failed. The socket is closed.
socketClosed = True
elif input == sys.stdin:
input = sys.stdin.readline().strip()
if input:
self.output.write(input + ‘\\r\\n’)
We must pass in three lists to select, but we pass in empty lists of output files and error files. All we
care about are the two sources of input (from the keyboard and the network), as those are the ones that
might block and cause problems when we try to read them.
In one sense, this code is more difficult to understand than the original ChatClient, because it uses a
trick to rapidly switch between doing two things, instead of just doing both things at once. In another
sense, it’s less complex than the original ChatClient because it’s less code and it doesn’t involve multi-
threading, which can be difficult to debug.
349

40 Pages 391-400

▲back to top

40.1 Page 391

▲back to top
Chapter 16
It’s possible to use select to write servers without forking or threading, but I don’t recommend writing
such code yourself. The Twisted framework (described in the section “The Twisted Framework,” later
in this chapter) provides a select-based server framework that will take care of the details for you, just
as the classes in SocketServer take care of the details of forking and threading.
Other Topics
Many aspects of network programming are not covered in this chapter. The most obvious omission (the
technologies and philosophies that drive the World Wide Web) will be taken up Chapter 21. The follow-
ing sections outline some other topics in networking that are especially interesting or important from the
perspective of a Python programmer.
Miscellaneous Considerations for Protocol Design
The best way to learn about protocol design is to study existing, successful protocols. Protocols are usu-
ally well documented, and you can learn a lot by using them and reading RFCs. Here are some common
design considerations for protocol design not covered earlier in this chapter.
Trusted Servers
The Python Chat Server is used by one client to broadcast information to all other clients. Sometimes,
however, the role of a server is to mediate between its clients. To this end, the clients are willing to trust
the server with information they wouldn’t trust to another client.
This happens often on web sites that bring people together, such as auction sites and online payment sys-
tems. It’s also implemented at the protocol level in many online games, in which the server acts as referee.
Consider a game in which players chase each other around a map. If one player knew another’s location
on the map, that player would gain an unfair advantage. At the same time, if players were allowed to
keep their locations secret, they could cheat by teleporting to another part of the map whenever a pur-
suer got too close. Players give up the ability to cheat in exchange for a promise that other players won’t
be allowed to cheat either. A trusted server creates a level playing field.
Terse Protocols
Information that can be pieced together by a client is typically not put into the protocol. It would be
wasteful for a server that ran chess games to transfer a representation of the entire board to both players
after every successful move. It would suffice to send “Your move was accepted.” to the player who
made the move, and describe the move to the other player. State-based protocols usually transmit the
changes in state, rather than send the whole state every time it changes.
The protocol for the Python Chat Server sends status messages in complete English sentences. This
makes the code easier to understand and the application easier to use through telnet. The client behavior
depends on those status messages: For instance, PythonChatClient expects the string “Who are you?”
as soon as it connects to the server. Doing a protocol this way makes it difficult for the server to cus-
tomize the status messages, or for the client to translate them into other languages. Many protocols
define numeric codes or short abbreviations for status messages and commands, and explain their
meanings in the protocols’ RFC or other definition document.
350

40.2 Page 392

▲back to top
Network Programming
The Twisted Framework
The Twisted framework is an alternative way of writing networked applications in Python. While the
classes in SocketServer are designed around spawning new threads and processes to handle requests,
Twisted uses a loop over the select function (as in the client example above) to timeshare between all
pending processes.
Download the Twisted framework libraries from the Twisted web site at http://twistedmatrix.
com/projects/core/, and install them.
For simple applications, programming in Twisted can be almost exactly like programming using
the SocketServer classes. Below is TwistedMirrorServer.py, a Twisted implementation of the
mirror server defined earlier in this chapter. Note that it looks a lot like the implementation that used
SocketServer classes, once we account for the fact that Twisted uses different names for the objects
provided by both it and the SocketServer framework (for instance, Twisted uses “factory” instead of
“server” and “transport” instead of “wfile”):
from twisted.internet import protocol, reactor
from twisted.protocols import basic
class MirrorProtocol(basic.LineReceiver):
“Handles one request to mirror some text.”
def lineReceived(self, line):
“””The client has sent in a line of text. Write out the
mirrored version.”””
self.transport.write(line[::-1]+ ‘\\r\\n’)
class MirrorFactory(protocol.ServerFactory):
protocol = MirrorProtocol
if __name__ == ‘__main__’:
import sys
if len(sys.argv) < 3:
print ‘Usage: %s [hostname] [port number]’ % sys.argv[0]
sys.exit(1)
hostname = sys.argv[1]
port = int(sys.argv[2])
reactor.listenTCP(port, MirrorFactory(), interface=hostname)
reactor.run()
This works just the same as the other MirrorServer implementations, but it runs faster because there’s
no need to spawn new threads.
Deferred Objects
Because Twisted servers run all of their code in a single thread, it’s very important that you write your
Twisted code so that it never blocks waiting for something to happen. It’s bad enough when a single
request drags on because the server has to consult a slow database to fulfill it — imagine what it would
be like if every request were stopped in its tracks just because one of them caused a database call.
351

40.3 Page 393

▲back to top
Chapter 16
The Twisted team has implemented new, blocking-free ways to do just about anything that might cause
a process to block: accessing a database, getting output from a subprocess, and using most of the popu-
lar Internet protocols. Behind the scenes, these implementations either feed into the same select loop
that drives your main application, or they use threads.
In Twisted, the standard way to do something that might take a while is to obtain a Deferred object
that knows how to do it and then register what you want to do next as a callback of the Deferred object.
Suppose you have some users who use your TwistedMirrorServer so much that it’s putting a load
on your CPU. You decide to change the mirror server so that any given user can only mirror one line of
text every five seconds. You might be tempted to implement this feature by calling time.sleep for the
appropriate interval if a user tries to use the server too often, like this:
#!/usr/bin/python
#This example is BAD! Do not use it!
from twisted.internet import protocol, reactor
from twisted.protocols import basic
import time
class MirrorProtocol(basic.LineReceiver):
“Handles one request to mirror some text.”
def __init__(self):
“””Set the timeout counter to a value that will always let a
new user’s first request succeed immediately.”””
self.lastUsed = 0
def lineReceived(self, line):
“””The client has sent in a line of text. Write out the
mirrored version, possibly waiting for a timeout to expire
before we do. Note: this is a very bad implementation because
we’re running this in a Twisted server, but time.sleep() is a
blocking call.”””
elapsed = time.time() - self.lastUsed
print elapsed
if elapsed < (self.factory.PER_USER_TIMEOUT * 1000):
time.sleep(self.factory.PER_USER_TIMEOUT-elapsed)
self.transport.write(line[::-1]+ ‘\\r\\n’)
self.lastUsed = time.time()
class MirrorFactory(protocol.ServerFactory):
“A server for the Mirror protocol defined above.”
protocol = MirrorProtocol
PER_USER_TIMEOUT = 5
The problem is that time.sleep blocks the thread until it’s complete. Since Twisted servers run in
a single thread, calling time.sleep will prevent any client from having their text mirrored until that
time.sleep call returns.
Fortunately, the Twisted team has implemented a non-blocking equivalent to time.sleep, called
callLater. This method returns a Deferred object that will call the given function after a certain
amount of time has elapsed. This gives you the equivalent functionality of time.sleep, but it doesn’t
block, so the ability of the Twisted server to deal with other clients is not impaired:
352

40.4 Page 394

▲back to top
Network Programming
from twisted.internet import protocol, reactor
from twisted.protocols import basic
import time
class MirrorProtocol(basic.LineReceiver):
“Handles one request to mirror some text.”
def __init__(self):
“””Set the timeout counter to a value that will always let a
new user’s first request succeed immediately.”””
self.lastUsed = 0
def lineReceived(self, line):
“””The client has sent in a line of text. Write out the
mirrored version, possibly waiting for a timeout to expire
before we do. This is a good implementation because it uses
a method that returns a Deferred object (reactor.callLater())
and registers a callback (writeLine) with that object.”””
elapsed = time.time() - self.lastUsed
if elapsed < self.factory.PER_USER_TIMEOUT:
reactor.callLater(self.factory.PER_USER_TIMEOUT-elapsed,
self.writeLine, line)
else:
self.writeLine(line)
def writeLine(self, line):
“Writes the given line and sets the user’s timeout.”
self.transport.write(line[::-1] + ‘\\r\\n’)
self.lastUsed = time.time()
class MirrorFactory(protocol.ServerFactory):
“A server for the Mirror protocol defined above.”
protocol = MirrorProtocol
PER_USER_TIMEOUT = 5
if __name__ == ‘__main__’:
import sys
if len(sys.argv) < 3:
print ‘Usage: %s [hostname] [port number]’ % sys.argv[0]
sys.exit(1)
hostname = sys.argv[1]
port = int(sys.argv[2])
reactor.listenTCP(port, MirrorFactory(), interface=hostname)
reactor.run()
This is not a general example: It only works because callLater has already been implemented as a
non-blocking equivalent to the blocking sleep function. If you’re going to write Twisted code, you’ll
need to find or write a non-blocking equivalent to every blocking function you want to call. Using
Twisted requires a different way of thinking about programming, but its unique approach offers a
higher-performance way to write network clients and servers.
353

40.5 Page 395

▲back to top
Chapter 16
The Peer-to-Peer Architecture
All of the protocols developed in this chapter were designed according to the client-server architecture.
This architecture divides the work of networking between two different pieces of software: the clients,
who request data or services, and the servers, which provide the data or carry out the services. This
architecture assumes a few powerful computers will act as servers, and a large number of computers
will act as clients. Information tends to be centralized on the server: to allow for central control, to
ensure fairness (for instance, in a game with hidden information), to make it unnecessary for clients to
trust each other, or just to make information easier to find.
The other popular architecture is the peer-to-peer architecture. In this architecture, every client is also a
server. A peer-to-peer protocol may define “client” actions and “server” actions, but every process that
makes requests is also capable of serving them.
Though most of the protocols implemented on top of it use the client-server architecture, TCP/IP is a
peer-to-peer protocol. Recall that a socket connection actually covers two unidirectional TCP/IP connec-
tions: one from you to your destination and one going the other way. You can’t be a TCP/IP client with-
out also being a TCP/IP server: you’d be sending data without any way of receiving a response.
At the application level, the most popular peer-to-peer protocol is BitTorrent. BitTorrent makes it easy
to distribute a large file by sharing the cost of the bandwidth across all of the people who download it.
Under the client-server architecture, someone who wanted to host a file would put it on her server and
bear the full cost of the bandwidth for every download. The original BitTorrent implementation is writ-
ten in Python, and the first release was in 2002. BitTorrent is proof positive that there’s still room for
clever new TCP/IP protocols, and that it’s possible to implement high-performance protocols in Python.
Summary
Python provides high-level tools for using existing TCP/IP-based protocols, making it easy to write cus-
tom clients. It also comes packaged with tools that help you design your own networked applications.
Whether you just want to send mail from a script, or you have an idea for the Internet’s next killer app,
Python can do what you need.
Exercises
1. Distinguish between the following e-mail-related standards: RFC 2822, SMTP, IMAP, MIME,
and POP.
2. Write a script that connects to a POP server, downloads all of the messages, and sorts the mes-
sages into files named after the sender of the message. (For instance, if you get two e-mails from
user@example.com, they should both go into a file called “user@example.com”).
3.
4.
5.
354
What would be the corresponding behavior if you had an IMAP server instead? Write that
script, too (use RFC 3501 as a reference).
Suppose that you were designing an IRC-style protocol for low-bandwidth embedded devices
such as cell phones. What changes to the Python Chat Server protocol would it be useful to make?
A feature of IRC not cloned in the Python Chat Server is the /msg command, which enables one
user to send a private message to another instead of broadcasting it to the whole room. How
could the /msg command be implemented in the Python Chat Server?
When does it make sense to design a protocol using a peer-to-peer architecture?

40.6 Page 396

▲back to top
17
Extension Programming
with C
Don’t let anybody mislead you: well-written code in C will always execute faster than code writ-
ten in Python. Having said that, don’t be misled: Developing code in Python will always be faster
than developing code in C.
This may seem like a dilemma at first. You want to have fast code, and you want to produce it
quickly. Balancing these, and the problem it creates, is actually easily solved. Develop your code
in Python. After all, developer’s time is much more expensive than the computer’s time. Plus,
humans have a miserable track record of predicting where a bottleneck is going to occur in a sys-
tem. Spending time optimizing code up front by doing things like taking a lot of time to write a
new program in C is usually wasted time. This is what led the esteemed computer scientist, C. A.
R. Hoare, to say, “Premature optimization is the root of all evil.” Of course, he was only talking
about computer programs, but the point is there.
If you’ve written your code, optimized your algorithms, and still find performance is unaccept-
able, you should profile your application by finding out where it’s spending its time, determine
where the bottlenecks are, and reimplement those small parts in C as a Python extension module.
That’s part of what this chapter is about.
Or if you already have an existing body of code written in C and you want to leverage that from
within Python, you can create a small Python extension module exposing that C code to your
Python code so it can be called as though it were written in Python. This is probably the more
common reason for implementing an extension module (a module written in a language other
than Python).
In this chapter, you’ll learn how to create an extension module in C for the standard Python inter-
preter, but you have to promise that you’ll do so only if you have absolutely no other option. This
chapter assumes you are already familiar with C. If you’re not, you need to rope someone who is
familiar with C into helping you out.
This chapter starts out with some basics and then shows a real-world, practical example in which
you define, in C, a class that can encode raw audio data into MP3-encoded data. Your class will be

40.7 Page 397

▲back to top
Chapter 17
usable from Python and will make method calls on pure Python objects, demonstrating how you can
communicate both ways.
This chapter is just an introduction to using the Python API from C and is no way a substitute for the
API documentation found at http://docs.python.org/. You should look up the function definitions
you’ll be using as they’re mentioned throughout the examples.
Extension Module Outline
First of all, a Python extension module is nothing more than a normal C library. On Unix machines, these
libraries usually end in “.so” (for shared object). On Windows machines, you typically see “.dll” (for
dynamically linked library).
Before you get started, you’re going to need the Python header files. On Unix machines, this usually
requires installing a developer-specific package such as python2.4-dev. Windows users get these headers
as part of the package when they use the binary Python installer.
For your first look at a Python extension module, you’ll be grouping your code into three parts: the C
functions you want to expose as the interface from your module, a table mapping the names of your
functions as Python developers will see them to C functions inside the extension module, and an initial-
ization function.
Most extension modules can be contained in a single C source file, sometimes called the glue. Start the
file out including Python.h, which will give you access to the internal Python API used to hook your
module into the interpreter. Be sure to include Python.h before any other headers you might need. You’ll
follow the includes with the functions you want to call from Python.
Interestingly, the signatures of the C implementations of your functions will always take one of the fol-
lowing three forms:
PyObject *MyFunction(PyObject *self, PyObject *args);
PyObject *MyFunctionWithKeywords(PyObject *self,
PyObject *args,
PyObject *kw);
PyObject *MyFunctionWithNoArgs(PyObject *self);
Typically, your C functions will look like the first of the preceding three declarations. The arguments
passed into your functions are packed into a tuple that you’ll have to break apart in order to use, which
explains how you can implement a function in C that takes only two arguments but can accept any num-
ber of arguments as called from Python.
Notice how each one of the preceding declarations returns a Python object. There’s no such thing as a
“void” function in Python as there is in C. If you don’t want your functions to return a value, return the
C equivalent of Python’s None value. The Python headers define a macro, Py_RETURN_NONE, that does
this for us.
356

40.8 Page 398

▲back to top
Extension Programming with C
Seeing these declarations should make it obvious how object-oriented Python is. Everything is an object.
In C, you’ll be using the Python API to work with these objects, but the concepts you know from Python
still hold.
The names of your C functions can be whatever you like as they’ll never be seen outside of the extension
module. In fact, the functions are usually declared with the static keyword (which in C means they’re
not visible outside of the current source file). In the example code, functions usually are named by com-
bining the Python module and function names together, as shown here:
static PyObject *foo_bar(PyObject *self, PyObject *args) {
/* Do something interesting here. */
Py_RETURN_NONE;
}
This would be a Python function called bar inside of the module foo. You’ll be putting pointers to your
C functions into the method table for the module that usually comes next in your source code.
This method table is a simple array of PyMethodDef structures. That structure looks something like this:
struct PyMethodDef {
char
*ml_name;
PyCFunction ml_meth;
int
ml_flags;
char
*ml_doc;
};
That first member, ml_name, is the name of the function as the Python interpreter will present it when
it’s used in Python programs. The PyCFunction member must be the address to a function that has any
one of the signatures described previously. ml_flags tells the interpreter which of the three signatures
ml_meth is using. ml_flags will usually have a value of METH_VARARGS. This value can be bitwise
or’ed with METH_KEYWORDS if you want to allow keyword arguments into your function. It can also have
a value of METH_NOARGS that indicates you don’t want to accept any arguments. Finally, the last member
in the PyMethodDef structure, ml_doc, is the docstring for the function, which can be NULL if you don’t
feel like writing one — shame on you.
This table needs to be terminated with a sentinel that consists of NULL and 0 values for the appropriate
members.
This is what a table containing an entry for our foo_bar function would look like:
static PyMethodDef foo_methods[] = {
{ “bar”, (PyCFunction)foo_bar, METH_NOARGS, “My first function.” },
{ NULL, NULL, 0, NULL }
};
Casting the address of foo_bar to a PyCFunction is necessary to get the compiler to not warn us about
incompatible pointer types. This is safe because of the METH_NOARGS flag for the ml_flags member,
which indicates to the Python interpreter that it should only call your C function with one PyObject *
as an argument (and not two as would be the case if you used METH_VARARGS, or three if you used
METH_VARARGS|METH_KEYWORDS).
357

40.9 Page 399

▲back to top
Chapter 17
The last part of your extension module is the initialization function. This function is called by the Python
interpreter when the module is loaded. It’s required that the function be named initfoo, where foo is
the name of the module.
The initialization function needs to be exported from the library you’ll be building. The Python headers
define PyMODINIT_FUNC to include the appropriate incantations for that to happen for the particular
environment in which we’re compiling. All you have to do is use it when defining the function.
Putting this all together looks like the following:
#include <Python.h>
static PyObject *foo_bar(PyObject *self, PyObject *args) {
/* Do something interesting here. */
Py_RETURN_NONE;
}
static PyMethodDef foo_methods[] = {
{ “bar”, (PyCFunction)foo_bar, METH_NOARGS, NULL },
{ NULL, NULL, 0, NULL }
};
PyMODINIT_FUNC initfoo() {
Py_InitModule3(“foo”, foo_methods, “My first extension module.”);
}
The Py_InitModule3 function is typically what you use to define a module because it lets you define a
docstring for a module, which is always a nice thing to do.
Building and Installing Extension Modules
You can build the extension module in a couple of different ways. The obvious way is to build it the way
you build all of the libraries on your platform. Save the previous example as foo.c. Then, to compile
your module on Linux, you could do something like this:
gcc -shared -I/usr/include/python2.4 foo.c -o foo.so
Building the extension module on Windows would look something like this:
cl /LD /IC:\\Python24\\include foo.c C:\\Python24\\libs\\python24.lib
For either of these commands to work, you’ll need to have a C compiler installed and have it available in
your path (if you’re reading this chapter, you probably do). The Python headers need to be installed and
accessible to the compiler. In both of these examples, the directory containing the Python headers is speci-
fied on the command line (as is the path to the Python library for the Windows compiler). If your headers
and libraries are located in a different location, the commands will have to be modified accordingly.
The name of the actual shared object (or DLL on Windows) needs to be the same as the string passed in
to Py_InitModule3 (minus the .so or .dll extension). Optionally, you can suffix the base name of the
library with module. So our foo extension module could be called foo.so or foomodule.so.
358

40.10 Page 400

▲back to top
Extension Programming with C
This works, but it’s not the only way to do it. The new and improved way of building extension modules
is to use distutils, which is included in all recent versions of Python.
The distutils package makes it possible to distribute Python modules, both pure Python and exten-
sion modules, in a standard way. Modules are distributed in source form and built and installed via a
setup script (usually called setup.py). As long as your users have the required compiler packages and
Python headers installed, this usually works.
The setup script is surprisingly succinct:
from distutils.core import setup, Extension
setup(name=’foo’, version=’1.0’, ext_modules=[Extension(‘foo’, [‘foo.c’])])
Running this script through the Python interpreter demonstrates that you’re getting quite a bit more
than initially expected with just two lines of code:
$ python setup.py
usage: setup.py [global_opts] cmd1 [cmd1_opts] [cmd2 [cmd2_opts] ...]
or: setup.py --help [cmd1 cmd2 ...]
or: setup.py --help-commands
or: setup.py cmd --help
error: no commands supplied
Trying again with the --help-commands argument displays all of the commands your setup script can
respond to:
$ python setup.py --help-commands
Standard commands:
build
build everything needed to install
build_py
“build” pure Python modules (copy to build directory)
build_ext
build C/C++ extensions (compile/link to build directory)
build_clib
build C/C++ libraries used by Python extensions
build_scripts “build” scripts (copy and fixup #! line)
clean
clean up output of ‘build’ command
install
install everything from build directory
install_lib
install all Python modules (extensions and pure Python)
install_headers install C/C++ header files
install_scripts install scripts (Python or otherwise)
install_data
install data files
sdist
create a source distribution (tarball, zip file, etc.)
register
register the distribution with the Python package index
bdist
create a built (binary) distribution
bdist_dumb
create a “dumb” built distribution
bdist_rpm
create an RPM distribution
bdist_wininst create an executable installer for MS Windows
usage: setup.py [global_opts] cmd1 [cmd1_opts] [cmd2 [cmd2_opts] ...]
or: setup.py --help [cmd1 cmd2 ...]
or: setup.py --help-commands
or: setup.py cmd --help
359

41 Pages 401-410

▲back to top

41.1 Page 401

▲back to top
Chapter 17
There’s a lot going on here, but all you need for now is the build command. Executing that will compile
foo.c into foo.so (on Linux) or foo.dll (on Windows). This file will end up in a subdirectory of the
build directory in your current directory unless you change that with more command-line options.
For the module to be importable by the Python interpreter, it needs to be in the current directory or in a
directory listed in the PYTHONPATH environmental variable or in a directory listed in the sys.path list,
which you can modify at runtime, although I wouldn’t recommend it.
The easiest way to get this to happen is to use another one of the setup script commands:
$ python setup.py install
If you hadn’t already built the module, this would have done that for you because building is a pre-requisite
for installing (much like a make file). The install command also copies the module to the site-packages
directory for our Python installation. This site-packages directory is listed in sys.path, so after this is done,
you can start using the module.
On Unix-based systems, you’ll most likely need to run this command as root in order to have permis-
sions to write to the site-packages directory. This usually isn’t a problem on Windows. It’s also possible
to install modules in alternative locations using the --home or --prefix command-line options, but
doing this leaves you responsible for ensuring they’re put in a directory the Python interpreter knows
about when it’s run.
Passing Parameters from Python to C
After you have everything built and installed, importing your new extension module and invoking its
one function is easy:
>>> import foo
>>> dir(foo)
[‘__doc__’, ‘__file__’, ‘__name__’, ‘bar’]
>>> foo.bar()
If you tried to pass in any arguments to your function, the interpreter will rightfully complain:
>>> foo.bar(1)
Traceback (most recent call last):
File “<stdin>”, line 1, in ?
TypeError: bar() takes no arguments (1 given)
Because you’ll most likely want to define functions that do accept arguments, you can use one of the
other signatures for your C functions. For example, a “normal” function (one that accepts some number
of parameters) would be defined like this:
static PyObject *foo_baz(PyObject *self, PyObject *args) {
/* Parse args and do something interesting here. */
Py_RETURN_NONE;
}
360

41.2 Page 402

▲back to top
Extension Programming with C
The method table containing an entry for the new function would look like this:
static PyMethodDef foo_methods[] = {
{ “bar”, (PyCFunction)foo_bar, METH_NOARGS, NULL },
{ “baz”, foo_baz, METH_VARARGS, NULL },
{ NULL, NULL, 0, NULL }
};
After making those changes to foo.c and saving them, you’re going to want to close any open Python
interpreters that imported the old version of the extension module so that you can recompile the source,
start a new interpreter, and import the new version of the extension module. It’s easy forget to do this if
you’re compiling in one window and invoking Python in another.
Compiling the new version of your module and importing it will enable you to invoke the new function
with any number of arguments of any type:
>>> foo.baz()
>>> foo.baz(1)
>>> foo.baz(1, 2.0)
>>> foo.baz(1, 2.0, “three”)
The reason why anything goes is that you haven’t written the C code to enforce a certain number and
type of arguments.
The Python API gives you the PyArg_ParseTuple function to extract the arguments from the one
PyObject pointer passed into your C function. This is a variadic function much like the standard
sscanf function with which you might be familiar.
The first argument to PyArg_ParseTuple is the args argument. This is the object you’ll be “parsing”.
The second argument is a format string describing the arguments as you expect them to appear. Each
argument is represented by one or more characters in the format string. An i indicates that you expect
the argument to be an integer-like object, which PyArg_ParseTuple will convert into a int as known in
C. Specifying a d in the format string will give you a double, while s will give you a string (char *). For
example, if you expected the baz function to be passed one integer, one double, and one string, our for-
mat string would be “ids”. The full list of indicators that you can include in a format string can be
found at http://docs.python.org/api/arg-parsing.html.
The remaining arguments to PyArg_ParseTuple are pointers to storage space of the appropriate type
for your arguments, just like sscanf. Knowing this, you might rewrite baz to look like the following:
static PyObject *foo_baz(PyObject *self, PyObject *args) {
int
i;
double d;
char *s;
if (!PyArg_ParseTuple(args, “ids”, &i, &d, &s)) {
return NULL;
}
/* Do something interesting here. */
Py_RETURN_NONE;
}
361

41.3 Page 403

▲back to top
Chapter 17
PyArg_ParseTuple will return 0 if it fails to extract exactly what was specified in the format string. It’s
important that you return NULL from your function when this happens so that the interpreter can gener-
ate an exception for your caller.
What about optional arguments? If you include a | (the vertical bar) character in your format string, the
indicators to the left of the | will be required, but the indicators to the right will be optional. You’re going
to want to give your local storage for the optional arguments a default value because PyArg_ParseTuple
won’t write anything to those variables if the caller didn’t specify the necessary arguments.
For example, if baz required one int, one double, and one string but also allowed an optional int, dou-
ble, and then a string, you might rewrite it to look like this:
static PyObject *foo_baz(PyObject *self, PyObject *args) {
int
i;
double d;
char *s;
int
i2 = 4;
double d2 = 5.0;
char *s2 = “six”;
if (!PyArg_ParseTuple(args, “ids|ids”, &i, &d, &s, &i2, &d2, &s2)) {
return NULL;
}
/* Do something interesting here. */
Py_RETURN_NONE;
}
Lastly, this next and final form your C functions might take will only be necessary when your functions
accept keyword arguments. In this case, you’ll use the signature that accepts three PyObject * argu-
ments and set the ml_flags member in your method table entry to METH_VARARGS|METH_KEYWORDS.
Instead of using the PyArg_ParseTuple function to extract your arguments, you’ll use
PyArg_ParseTupleAndKeywords.
This is what the function might look like:
static PyObject *foo_quux(PyObject *self, PyObject *args, PyObject *kw) {
char *kwlist[] = { “i”, “d”, “s”, NULL };
int
i;
double d = 2.0;
char *s = “three”;
if (!PyArg_ParseTupleAndKeywords(args, kw, “i|ds”, kwlist, &i, &d, &s)) {
return NULL;
}
/* Do something interesting here. */
Py_RETURN_NONE;
}
This would be its entry in the method table right after the entry for the baz function but before the sen-
tinel entry:
{ “quux”, (PyCFunction)foo_quux, METH_VARARGS|METH_KEYWORDS, NULL },
362

41.4 Page 404

▲back to top
Extension Programming with C
PyArg_ParseTupleAndKeywords works just like PyArg_ParseTuple with the exception of two extra
arguments. First, you need to pass in the pointer to the Python object containing the keyword arguments.
Second, you need to indicate what keywords you’re interested in. You do that with a NULL-terminated list
of strings. In the preceding example, you’re saying that your keywords are “i”, “d”, and “s”.
Each keyword needs to correspond with one indicator in the format string even if you don’t ever intend
to have your callers use a keyword for certain arguments. Notice how the preceding example includes
three indicators in the format string. The first, “i”, is required while the other two, “d” and “s”, are
optional. You could call this function (from Python) in any of the following ways:
>>> foo.quux(1)
>>> foo.quux(i=1)
>>> foo.quux(1, 2.0)
>>> foo.quux(1, 2.0, “three”)
>>> foo.quux(1, 2.0, s=”three”)
>>> foo.quux(1, d=2.0)
>>> foo.quux(1, s=”three”)
>>> foo.quux(1, d=2.0, s=”three”)
>>> foo.quux(1, s=”three”, d=2.0)
>>> foo.quux(i=1, d=2.0, s=”three”)
>>> foo.quux(s=”three”, d=2.0, i=1)
You can probably come up with even more variations.
Returning Values from C to Python
PyArg_ParseTuple and PyArg_ParseTupleAndKeywords convert from Python objects into C values
but what about going the other way? How would you return a value from a function implemented in C
back into Python?
All of the function signatures you saw previously return a PyObject *, so you need to use whatever the
opposite of PyArg_ParseTuple is in order to turn a C value into a Python object. That function is called
Py_BuildValue.
Py_BuildValue takes in a format string much like PyArg_ParseTuple does. Instead of passing in the
addresses of the values you’re building, you pass in the actual values. Here’s an example showing how
to implement an add function:
static PyObject *foo_add(PyObject *self, PyObject *args) {
int a;
int b;
if (!PyArg_ParseTuple(args, “ii”, &a, &b)) {
return NULL;
}
return Py_BuildValue(“i”, a + b);
}
The Python equivalent of this function would look like this:
def add(a, b):
return a + b
363

41.5 Page 405

▲back to top
Chapter 17
What if you want to return more than one value from your function? In Python, you do that by return-
ing a tuple. To do that in C, you do that by building a tuple with Py_BuildValue. If your format string
has more than one indicator, you’ll get a tuple. You can also be explicit and surround your indicators
with parentheses:
static PyObject *foo_add_and_subtract(PyObject *self, PyObject *args) {
int a;
int b;
if (!PyArg_ParseTuple(args, “ii”, &a, &b)) {
return NULL;
}
return Py_BuildValue(“(ii)”, a + b, a - b);
}
To help visualize what this function is doing, this is what it would look like if implemented in Python:
def add_and_subtract(a, b):
return (a + b, a - b)
Now, armed with just this knowledge, it’s possible for you to create a wide variety of extension modules.
Let’s put this to good use and work on a real example.
The LAME Project
LAME is (or was) an acronym that originally stood for “LAME Ain’t an MP3 Encoder.” Whether it’s offi-
cially considered an MP3 encoder or not isn’t important to you, because it functions as a (most excellent)
free and open-source library that is capable of encoding MP3s.
Dozens of software projects use LAME but not many are implemented in Python, which is why you’ll be
using it as an example to demonstrate just how easy it is to create extension modules for Python that
leverage an existing C code base, even when the C code wasn’t written to be interfaced with Python.
This example is also a very practical one. Consider how many years went into developing the LAME
code base, which in case you don’t know is many, many, many person-years. Would you really want to
duplicate that work by reimplementing it in Python? Now consider what your answer would be if you
were told that it would be unbelievably slow it would run if you had a Python-only encoder! This isn’t
anything against Python, by the way. This would old true of any language that is higher-level than C.
Languages such as Java, Perl, etc. would have the same limitation. This is a perfect example of code that
you would not want to use Python to develop (there are very few examples where this is true).
Before creating an extension module that wraps the LAME library, you need to learn how to use the API
exposed by that library. The core of the LAME API is small enough to create a quick demonstration with
only a page or so of C code.
You need the LAME headers and libraries installed on your machine before you can write any code
that uses its API, of course. The LAME Project’s web site is located on SourceForge at http://lame.
sourceforge.net/. You can download the source code from there. While you can download and com-
pile and install the libraries for any part of the LAME package from there, you won’t find any pre-built
364

41.6 Page 406

▲back to top
Extension Programming with C
binaries on this site (presumably to avoid the potential legal issues of distributing an mp3 encoder).
However, you can find links to sites that do provide downloadable binaries by looking for them on the
LAME Project’s web site (if you’d rather not build from source).
Packages can easily be found on the web for most Linux distributions. Some names these packages may
be listed under are lame, liblame, or the liblame-dev package. If you can’t find a package or would rather
build from source, ./configure, make, and make install will work to build a complete working instal-
lation of LAME, just as they do with almost every other project you build from source on Linux.
Windows users can use any of the pre-built binaries but those don’t usually come with the headers,
so you’ll have to download those from the main site. If you’re doing that, you might as well build the
libraries yourself. The LAME source code includes a Visual Studio workspace that can build everything
you need to get through the rest of this chapter. There will be errors (there were for the author), but the
build process makes it far enough to finish building just what we need, so you can ignore those errors
and be OK.
The general overview of creating an MP3 file with LAME is described here:
1. Initialize the library.
2. Set up the encoding parameters.
3. Feed the library one buffer of audio data at a time (returning another buffer of MP3-encoded
bytes of that data).
4. Flush the encoder (possibly returning more MP3 data).
5. Close the library.
That’s it!
Here’s an example written in C that uses the LAME API. It can encode any raw audio file into an MP3-
encoded audio file. If you want to compile it to make sure it works, save it in a file called clame.c:
#include <stdio.h>
#include <stdlib.h>
#include <lame.h>
#define INBUFSIZE 4096
#define MP3BUFSIZE (int)(1.25 * INBUFSIZE) + 7200
int encode(char *inpath, char *outpath) {
int status = 0;
lame_global_flags *gfp;
int ret_code;
FILE *infp;
FILE *outfp;
short *input_buffer;
365

41.7 Page 407

▲back to top
Chapter 17
366
int input_samples;
char *mp3_buffer;
int mp3_bytes;
/* Initialize the library. */
gfp = lame_init();
if (gfp == NULL) {
printf(“lame_init returned NULL\\n”);
status = -1;
goto exit;
}
/* Set the encoding parameters. */
ret_code = lame_init_params(gfp);
if (ret_code < 0) {
printf(“lame_init_params returned %d\\n”, ret_code);
status = -1;
goto close_lame;
}
/* Open our input and output files. */
infp = fopen(inpath, “rb”);
outfp = fopen(outpath, “wb”);
/* Allocate some buffers. */
input_buffer = (short*)malloc(INBUFSIZE*2);
mp3_buffer = (char*)malloc(MP3BUFSIZE);
/* Read from the input file, encode, and write to the output file. */
do {
input_samples = fread(input_buffer, 2, INBUFSIZE, infp);
if (input_samples > 0) {
mp3_bytes = lame_encode_buffer_interleaved(
gfp,
input_buffer,
input_samples / 2,
mp3_buffer,
MP3BUFSIZE
);
if (mp3_bytes < 0) {
printf(“lame_encode_buffer_interleaved returned %d\\n”, mp3_bytes);
status = -1;
goto free_buffers;
} else if (mp3_bytes > 0) {
fwrite(mp3_buffer, 1, mp3_bytes, outfp);
}
}
} while (input_samples == INBUFSIZE);
/* Flush the encoder of any remaining bytes. */
mp3_bytes = lame_encode_flush(gfp, mp3_buffer, sizeof(mp3_buffer));
if (mp3_bytes > 0) {
printf(“writing %d mp3 bytes\\n”, mp3_bytes);
fwrite(mp3_buffer, 1, mp3_bytes, outfp);
}

41.8 Page 408

▲back to top
Extension Programming with C
/* Clean up. */
free_buffers:
free(mp3_buffer);
free(input_buffer);
fclose(outfp);
fclose(infp);
close_lame:
lame_close(gfp);
exit:
return status;
}
int main(int argc, char *argv[]) {
if (argc < 3) {
printf(“usage: clame rawinfile mp3outfile\\n”);
exit(1);
}
encode(argv[1], argv[2]);
return 0;
}
To compile the file on Linux, this command should work (assuming you installed a package like
liblame-dev or that the lame development components have installed the appropriate header files in
/usr/include/lame):
gcc -I/usr/include/lame clame.c -lmp3lame -o clame
On Windows, you’ll probably have to use a command like this (assuming you built from source):
cl /IC:\\lame-3.96.1\\include clame.c \\
C:\\lame-3.96.1\\libmp3lame\\Release\\libmp3lame.lib \\
C:\\lame-3.96.1\\mpglib\\Release\\mpglib.lib
Those command-line parameters are telling the compiler where to find the LAME headers and necessary
libraries. You’ll probably have to adjust them to point to the correct directories.
That wasn’t too bad, was it? Of course, this code doesn’t know how to extract data out of a WAV or any
other sort of audio file. It is assumed here that the input file contains nothing but raw, 16-bit, signed
samples at 44.1 kHz. Turning a WAV file into one of these raw files is a simple command on most Unix-
based machines (assuming you have the sox program, which should also be available as a package):
sox test.wav -t raw test.raw
367

41.9 Page 409

▲back to top
Chapter 17
The LAME Extension Module
To create an extension module that enables you to encode a raw audio file into an MP3 could be as simple
as creating a simple function that invokes the encode function you defined in the preceding example:
#include <Python.h>
#include <lame.h>
/* defined in clame.c */
int encode(char *, char *);
static PyObject *pylame1_encode(PyObject *self, PyObject *args) {
int status;
char *inpath;
char *outpath;
if (!PyArg_ParseTuple(args, “ss”, &inpath, &outpath)) {
return NULL;
}
status = encode(inpath, outpath);
return Py_BuildValue(“i”, status);
}
static PyMethodDef pylame1_methods[] = {
{ “encode”, pylame1_encode, METH_VARARGS, NULL },
{ NULL, NULL, 0, NULL }
};
PyMODINIT_FUNC initpylame1() {
Py_InitModule3(“pylame1”, pylame1_methods, “My first LAME module.”);
}
Here the encode function accepts two string arguments — the input path and the output path.
Try saving the preceding code in a file called pylame1.c and compiling it with this command:
gcc -shared -I/usr/include/python2.4 -I/usr/include/lame \\
pylame1.c clame.c \\
-lmp3lame -o pylame1.so
On Windows, you’ll need something like this:
cl /LD /IC:\\Python24\\include /IC:\\lame-3.96.1\\include \\
pylame1.c clame.c \\
C:\\Python24\\libs\\python24.lib \\
C:\\lame-3.96.1\\libmp3lame\\Release\\libmp3lame.lib \\
C:\\lame-3.96.1\\mpglib\\Release\\mpglib.lib
Note that we’re compiling the same clame.c example we used in the previous section into this DLL by
including it on the command line.
This works, but it’s not ideal; you have no way of influencing how the encode function works other
than by passing in two strings. What if you wanted to encode something other than a raw audio file?
368

41.10 Page 410

▲back to top
Extension Programming with C
How about a WAV file or perhaps some audio data you’re streaming off the network? There’s no reason
why you couldn’t implement that functionality in Python, where it would be much easier to do.
You have two options: You can have the Python code pass the audio data into the encode function, one
chunk at a time, just like you do in the C function. Or, you can pass some object with a read method in
to encode, which would then read its data from that object.
Although the second option might sound more object oriented, the first is the better choice, as it pro-
vides more flexibility. You could always define some sort of object that reads from some source and
passes it on to the encoder, but it would be a lot harder to go the other way around.
Using this design is going to require that you make some changes in the extension module. Right now,
there’s just one function, and that’s fine because that function is doing all of the work for you. With the
new approach, however, you’ll be making multiple calls to the function that you’ll be using to encode
the audio data as MP3 data. You can’t have the function re-open the file every time it’s called, so you’re
going to need to maintain some state information about where you are in the file somewhere. You can
have the caller maintain that state, or you can encapsulate it inside some object defined by your module,
which is the approach you’ll be taking here.
The new version of your extension module needs to expose a class so that your clients can create
instances of this class and invoke methods on them. You’ll be hiding a small amount of state in those
instances so they can remember which file they’re writing to between method calls.
As you learn what you need to do for this new module, you’ll see the snippets of code relevant to what
is being explained. The entire source for pylame2.c will be shown later so you can see the snippets
together in all of their glory.
The C language syntax doesn’t directly support defining a new class, but it does have structures; and in C
structures can contain function pointers, which is good enough for what you’re trying to do right now.
When the Python interpreter creates a new instance of your class, it will actually be allocating enough space
to store a new instance of your structure. It’s that structure that will contain all of your state for each object.
The Python interpreter needs to store some information in your objects as well. Every object has a refer-
ence count and a type, so the first part of your structure has to contain these in order for the Python
interpreter to find them:
typedef struct {
PyObject_HEAD
/* State goes here. */
} pylame2_EncoderObject;
The PyObject_HEAD macro will add the appropriate members to the structure — you just have to make
sure that it’s the first thing you add.
You need to provide a function to create the new instances of this structure:
static PyObject *Encoder_new(PyTypeObject *type, PyObject *args, PyObject *kw) {
pylame2_EncoderObject *self = (pylame2_EncoderObject *)type->tp_alloc(type, 0);
/* Initialize object here. */
return (PyObject *)self;
}
369

42 Pages 411-420

▲back to top

42.1 Page 411

▲back to top
Chapter 17
Think of this as equivalent to Python’s __new__ method. This function will be called by the interpreter
when it needs to create a new instance of your type. Notice how you’re not calling malloc directly but
are instead invoking some other function as indicated by the tp_alloc member of the PyTypeObject
that was passed in to your function. You’ll see what function that is in a bit.
You also need a function to free your instances:
static void Encoder_dealloc(PyObject *self) {
self->ob_type->tp_free(self);
}
Think of this function as equivalent to Python’s __del__ method and being a counterpart to Encoder_new.
Because you’re calling tp_free on your object’s type object here, you’re probably assuming that the
tp_free function is the counterpart to the tp_alloc function and you’re right.
What about the other methods your object is supposed to support? Do you add function pointers to
your structure to represent those? If you did, each instance would be eating up memory with the exact
same set of pointers, which would be a waste. Instead, you’re going to store the function pointers for
your methods in a separate structure and your objects will refer to that structure.
Remember that each object knows its type — there’s a pointer to a type object hiding inside the
PyObject_HEAD macro. Therefore, you need another structure to represent that:
static PyTypeObject pylame2_EncoderType = {
PyObject_HEAD_INIT(NULL)
0,
/* ob_size */
“pylame2.Encoder”,
/* tp_name */
sizeof(pylame2_EncoderObject), /* tp_basicsize */
0,
/* tp_itemsize */
Encoder_dealloc,
/* tp_dealloc */
0,
/* tp_print */
0,
/* tp_getattr */
0,
/* tp_setattr */
0,
/* tp_compare */
0,
/* tp_repr */
0,
/* tp_as_number */
0,
/* tp_as_sequence */
0,
/* tp_as_mapping */
0,
/* tp_hash */
0,
/* tp_call */
0,
/* tp_str */
0,
/* tp_getattro */
0,
/* tp_setattro */
0,
/* tp_as_buffer */
Py_TPFLAGS_DEFAULT,
/* tp_flags */
“My first encoder object.”, /* tp_doc */
0,
/* tp_traverse */
0,
/* tp_clear */
0,
/* tp_richcompare */
0,
/* tp_weaklistoffset */
370

42.2 Page 412

▲back to top
Extension Programming with C
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
0,
Encoder_new,
0,
};
/* tp_iter */
/* tp_iternext */
/* tp_methods */
/* tp_members */
/* tp_getset */
/* tp_base */
/* tp_dict */
/* tp_descr_get */
/* tp_descr_set */
/* tp_dictoffset */
/* tp_init */
/* tp_alloc */
/* tp_new */
/* tp_free */
This is going to be the structure for what you’re going to get a pointer to when your Encoder_new func-
tion is called. There’s a lot to that structure (and even more that you can’t see yet), but you’re letting
most of the members default to NULL for now. You’ll go over the important bits before moving on.
The PyObject_HEAD_INIT macro adds the members that are common to all types. It must be the first
member in the structure. It’s like PyObject_HEAD except that it initializes the type pointer to whatever
you pass in as an argument.
Remember: In Python, types are objects, too, so they also have types. You could call a type’s type a “type
type”. The Python API calls it PyType_Type. It’s the type of type objects. You really want to be able to
pass &PyType_Type into this macro but some compilers won’t let you statically initialize a structure
member with a symbol defined in some other module, so you’ll have to fill that in later.
The next member, ob_size, might look important but it’s a remnant from an older version of the Python
API and should be ignored. The member after the name of your type, tp_basicsize, represents the size
of all your object instances. When the interpreter needs to allocate storage space for a new instance, it
will request tp_basicsize bytes.
Most of the rest of the members are currently NULL, but you’ll be filling them in later. They’ll hold func-
tion pointers for some of the more common operations that many objects support.
The tp_flags member specifies some default flags for the type object, which all type objects need; and
the tp_doc member holds a pointer to the docstring for the type, which you always want to provide
because we’re good Python citizens.
Notice the tp_alloc and tp_free members, which are set to NULL. Aren’t those the members we’re
calling from Encoder_new and Encoder_dealloc? Yes, they are, but you’re going to use a Python API
function to fill them in with the appropriate addresses later on because some platforms don’t like it
when you statically initialize structure members with addresses of functions in other libraries.
At this point, you’ve defined two structures. To actually make them available via your extension mod-
ule, you need to add some code to your module’s initialization function:
371

42.3 Page 413

▲back to top
Chapter 17
PyMODINIT_FUNC initpylame2() {
PyObject *m;
if (PyType_Ready(&pylame2_EncoderType) < 0) {
return;
}
m = Py_InitModule3(“pylame2”, pylame2_methods, “My second LAME module.”);
Py_INCREF(&pylame2_EncoderType);
PyModule_AddObject(m, “Encoder”, (PyObject *)&pylame2_EncoderType);
}
PyType_Ready gets a type object “ready” for use by the interpreter. It sets the type of the object to
PyType_Type and sets a number of the function pointer members that you had previously left NULL,
along with a number of other bookkeeping tasks necessary in order to hook everything up properly,
including setting your tp_alloc and tp_free members to suitable functions.
After you get your type object ready, you create your module as usual, but this time you’re saving the
return value (a pointer to a module object) so you can add your new type object to the module. Previously,
you had been ignoring the return value and letting the method table define all of the members of the
module. Because there’s no way to fit a PyObject pointer into a method table, you need to use the
PyModule_AddObject function to add your type object to the module. This function takes in the pointer
to the module returned from Py_InitModule3, the name of your new object as it should be known in
the module, and the pointer to the new object itself.
If you were to compile what you had so far, you’d be able to create new Encoder instances:
>>> import pylame2
>>> e = pylame2.Encoder()
That object doesn’t do you much good, however, as it doesn’t have any useful behavior yet.
To make these objects useful, you have to allow for some information to be passed into their initialization
functions. That information could simply be the path to the file to which you want to write. Your initial-
ization function could use that path to open a file handle that would enable you to write to it, but there’ll
be no writing until somebody invokes the encode method on your object. Therefore, your object needs to
retain the handle for the file it opened.
You’re also going to be invoking functions defined in the LAME library, so your objects will also need to
remember the pointer to the lame_global_flags structure returned by lame_init.
Here’s your structure with state and a modified Encoder_new function to initialize it:
typedef struct {
PyObject_HEAD
FILE *outfp;
lame_global_flags *gfp;
} pylame2_EncoderObject;
static PyObject *Encoder_new(PyTypeObject *type, PyObject *args, PyObject *kw) {
pylame2_EncoderObject *self = (pylame2_EncoderObject *)type->tp_alloc(type, 0);
self->outfp = NULL;
self->gfp = NULL;
return (PyObject *)self;
}
372

42.4 Page 414

▲back to top
Extension Programming with C
You’re not checking args and kw here, because this is the equivalent of Python’s __new__ method, not
__init__. It’s in your C implementation of __init__ that you’ll be opening the file and initializing the
LAME library:
static int Encoder_init(pylame2_EncoderObject *self,
PyObject *args, PyObject *kw) {
char *outpath;
if (!PyArg_ParseTuple(args, “s”, &outpath)) {
return -1;
}
if (self->outfp || self->gfp) {
PyErr_SetString(PyExc_Exception, “__init__ already called”);
return -1;
}
self->outfp = fopen(outpath, “wb”);
self->gfp = lame_init();
lame_init_params(self->gfp);
return 0;
}
Your __init__ implementation is checking two things. The first you’ve already seen. You’re using
PyArg_ParseTuple to ensure that you were passed in one string parameter. The second check is ensur-
ing that the outfp and gfp members of your instance are NULL. If they’re not, this function must already
have been called for this object, so return the appropriate error code for this function after using the
PyErr_SetString function to “set” an exception. After you return into the Python interpreter, an
exception will be raised and your caller is going to have to catch it or suffer the consequences. You need
to do this because it’s always possible to call __init__ twice on an object. With this code in place,
calling __init__ twice on your objects might look like this:
>>> import pylame2
>>> e = pylame2.Encoder(“foo.mp3”)
>>> e.__init__(“bar.mp3”)
Traceback (most recent call last):
File “<stdin>”, line 1, in ?
Exception: __init__ already called
Of course, you could be nice and reinitialize the object, but that’s not necessary for what you want to get
done today. You should also be checking for errors, of course.
To indicate that you want this initialization function to be called for each new instance of your class, you
need to add the address this function needs to your type object:
(initproc)Encoder_init,
/* tp_init */
You’re casting it here because we cheated and declared that Encoder_init accepted a pylame2_
EncoderObject * as its first argument instead of the more generic PyObject *. You can get away with
this type of stuff in C, but you have to be absolutely certain that you know what you’re doing.
Because your instances now contain state that reference resources, you need to ensure that those resources
are properly disposed of when the object is released. To do this, update your Encoder_dealloc function:
373

42.5 Page 415

▲back to top
Chapter 17
static void Encoder_dealloc(pylame2_EncoderObject *self) {
if (self->gfp) {
lame_close(self->gfp);
}
if (self->outfp) {
fclose(self->outfp);
}
self->ob_type->tp_free(self);
}
If you were to build your module with the code you have so far, import it, create an encoder object, and
then delete it (using the del keyword or rebinding the variable referencing your object to None or some
other object), you would end up with an empty file in the current directory because all you did was open
and then close it without writing anything to it. You’re getting closer!
You now need to add support for the encode and close methods to your type. Previously, you had
created what was called a method table, but that was really defining module-level functions. Defining
methods for classes is just as easy but different. You define the methods just like the module-level func-
tions and then create a table listing them:
static PyObject *Encoder_encode(PyObject *self, PyObject *args) {
Py_RETURN_NONE;
}
static PyObject *Encoder_close(PyObject *self) {
Py_RETURN_NONE;
}
static PyMethodDef Encoder_methods[] = {
{ “encode”, Encoder_encode, METH_VARARGS,
“Encodes and writes data to the output file.” },
{ “close”, (PyCFunction)Encoder_close, METH_NOARGS,
“Closes the output file.” },
{ NULL, NULL, 0, NULL }
};
Then the address of the table is used to initialize the tp_methods member of your type object:
Encoder_methods,
/* tp_methods */
With those stubs in place, you could build the module and see the methods and even call them on your
objects:
>>> import pylame2
>>> e = pylame2.Encoder(‘foo.mp3’)
>>> dir(e)
[‘__class__’, ‘__delattr__’, ‘__doc__’, ‘__getattribute__’, ‘__hash__’,
‘__init__’, ‘__new__’, ‘__reduce__’, ‘__reduce_ex__’, ‘__repr__’,
‘__setattr__’, ‘__str__’, ‘close’, ‘encode’]
>>> e.encode()
>>> e.close()
374

42.6 Page 416

▲back to top
Extension Programming with C
All you have to do now is implement the functions. Here’s Encoder_encode (sans complete error-
checking):
static PyObject *Encoder_encode(pylame2_EncoderObject *self, PyObject *args) {
char *in_buffer;
int in_length;
int mp3_length;
char *mp3_buffer;
int mp3_bytes;
if (!(self->outfp && self->gfp)) {
PyErr_SetString(PyExc_Exception, “encoder not open”);
return NULL;
}
if (!PyArg_ParseTuple(args, “s#”, &in_buffer, &in_length)) {
return NULL;
}
in_length /= 2;
mp3_length = (int)(1.25 * in_length) + 7200;
mp3_buffer = (char *)malloc(mp3_length);
if (in_length > 0) {
mp3_bytes = lame_encode_buffer_interleaved(
self->gfp,
(short *)in_buffer,
in_length / 2,
mp3_buffer,
mp3_length
);
if (mp3_bytes > 0) {
fwrite(mp3_buffer, 1, mp3_bytes, self->outfp);
}
}
free(mp3_buffer);
Py_RETURN_NONE;
}
You expect this argument to be passed a string. Unlike strings in C, which are simple NUL-terminated
arrays of characters, you expect that this string will contain embedded NUL characters (the NUL character,
which is simple the end-of-string indication in C has the value of ‘\\0’ in C. Note the single quotes — in C
remember that the different quotes have different meanings. NUL can also be shown as “” in C.) Therefore,
instead of using the “s” indicator when parsing the arguments, you use “s#”, which allows for embedded
NUL characters. PyArg_ParseTuple will return both the bytes in a buffer and the length of the buffer
instead of tacking a NUL character on the end. Other than that, this function is pretty straightforward.
Here’s Encoder_close:
static PyObject *Encoder_close(pylame2_EncoderObject *self) {
int mp3_length;
char *mp3_buffer;
int mp3_bytes;
if (!(self->outfp && self->gfp)) {
PyErr_SetString(PyExc_Exception, “encoder not open”);
return NULL;
}
375

42.7 Page 417

▲back to top
Chapter 17
mp3_length = 7200;
mp3_buffer = (char *)malloc(mp3_length);
mp3_bytes = lame_encode_flush(self->gfp, mp3_buffer, sizeof(mp3_buffer));
if (mp3_bytes > 0) {
fwrite(mp3_buffer, 1, mp3_bytes, self->outfp);
}
free(mp3_buffer);
lame_close(self->gfp);
self->gfp = NULL;
fclose(self->outfp);
self->outfp = NULL;
Py_RETURN_NONE;
}
You need to make sure you set outfp and gfp to NULL here to prevent Encoder_dealloc from trying to
close them again.
For both Encoder_encode and Encoder_close, you’re checking to make sure your object is in a valid
state for encoding and closing. Somebody could always call close and then follow that up with another
call to close or even a call to encode. It’s better to raise an exception than to bring down the process
hosting your extension module.
We went over a lot to get to this point, so it would probably help if you could see the entire extension
module in one large example:
#include <Python.h>
#include <lame.h>
typedef struct {
PyObject_HEAD
FILE *outfp;
lame_global_flags *gfp;
} pylame2_EncoderObject;
static PyObject *Encoder_new(PyTypeObject *type, PyObject *args, PyObject *kw) {
pylame2_EncoderObject *self = (pylame2_EncoderObject *)type->tp_alloc(type, 0);
self->outfp = NULL;
self->gfp = NULL;
return (PyObject *)self;
}
static void Encoder_dealloc(pylame2_EncoderObject *self) {
if (self->gfp) {
lame_close(self->gfp);
}
if (self->outfp) {
fclose(self->outfp);
}
self->ob_type->tp_free(self);
}
static int Encoder_init(pylame2_EncoderObject *self, PyObject *args, PyObject *kw) {
376

42.8 Page 418

▲back to top
Extension Programming with C
char *outpath;
if (!PyArg_ParseTuple(args, “s”, &outpath)) {
return -1;
}
if (self->outfp || self->gfp) {
PyErr_SetString(PyExc_Exception, “__init__ already called”);
return -1;
}
self->outfp = fopen(outpath, “wb”);
self->gfp = lame_init();
lame_init_params(self->gfp);
return 0;
}
static PyObject *Encoder_encode(pylame2_EncoderObject *self, PyObject *args) {
char *in_buffer;
int in_length;
int mp3_length;
char *mp3_buffer;
int mp3_bytes;
if (!(self->outfp && self->gfp)) {
PyErr_SetString(PyExc_Exception, “encoder not open”);
return NULL;
}
if (!PyArg_ParseTuple(args, “s#”, &in_buffer, &in_length)) {
return NULL;
}
in_length /= 2;
mp3_length = (int)(1.25 * in_length) + 7200;
mp3_buffer = (char *)malloc(mp3_length);
if (in_length > 0) {
mp3_bytes = lame_encode_buffer_interleaved(
self->gfp,
(short *)in_buffer,
in_length / 2,
mp3_buffer,
mp3_length
);
if (mp3_bytes > 0) {
fwrite(mp3_buffer, 1, mp3_bytes, self->outfp);
}
}
free(mp3_buffer);
Py_RETURN_NONE;
}
static PyObject *Encoder_close(pylame2_EncoderObject *self) {
int mp3_length;
char *mp3_buffer;
int mp3_bytes;
if (!(self->outfp && self->gfp)) {
PyErr_SetString(PyExc_Exception, “encoder not open”);
return NULL;
}
377

42.9 Page 419

▲back to top
Chapter 17
mp3_length = 7200;
mp3_buffer = (char *)malloc(mp3_length);
mp3_bytes = lame_encode_flush(self->gfp, mp3_buffer, sizeof(mp3_buffer));
if (mp3_bytes > 0) {
fwrite(mp3_buffer, 1, mp3_bytes, self->outfp);
}
free(mp3_buffer);
lame_close(self->gfp);
self->gfp = NULL;
fclose(self->outfp);
self->outfp = NULL;
Py_RETURN_NONE;
}
static PyMethodDef Encoder_methods[] = {
{ “encode”, (PyCFunction)Encoder_encode, METH_VARARGS,
“Encodes and writes data to the output file.” },
{ “close”, (PyCFunction)Encoder_close, METH_NOARGS,
“Closes the output file.” },
{ NULL, NULL, 0, NULL }
};
static PyTypeObject pylame2_EncoderType = {
PyObject_HEAD_INIT(NULL)
0,
/* ob_size */
“pylame2.Encoder”,
/* tp_name */
sizeof(pylame2_EncoderObject), /* tp_basicsize */
0,
/* tp_itemsize */
(destructor)Encoder_dealloc, /* tp_dealloc */
0,
/* tp_print */
0,
/* tp_getattr */
0,
/* tp_setattr */
0,
/* tp_compare */
0,
/* tp_repr */
0,
/* tp_as_number */
0,
/* tp_as_sequence */
0,
/* tp_as_mapping */
0,
/* tp_hash */
0,
/* tp_call */
0,
/* tp_str */
0,
/* tp_getattro */
0,
/* tp_setattro */
0,
/* tp_as_buffer */
Py_TPFLAGS_DEFAULT,
/* tp_flags */
“My first encoder object.”, /* tp_doc */
0,
/* tp_traverse */
0,
/* tp_clear */
0,
/* tp_richcompare */
0,
/* tp_weaklistoffset */
0,
/* tp_iter */
0,
/* tp_iternext */
Encoder_methods,
/* tp_methods */
0,
/* tp_members */
0,
/* tp_getset */
0,
/* tp_base */
378

42.10 Page 420

▲back to top
Extension Programming with C
0,
0,
0,
0,
(initproc)Encoder_init,
0,
Encoder_new,
0,
};
/* tp_dict */
/* tp_descr_get */
/* tp_descr_set */
/* tp_dictoffset */
/* tp_init */
/* tp_alloc */
/* tp_new */
/* tp_free */
static PyMethodDef pylame2_methods[] = {
{ NULL, NULL, 0, NULL }
};
PyMODINIT_FUNC initpylame2() {
PyObject *m;
if (PyType_Ready(&pylame2_EncoderType) < 0) {
return;
}
m = Py_InitModule3(“pylame2”, pylame2_methods, “My second LAME module.”);
Py_INCREF(&pylame2_EncoderType);
PyModule_AddObject(m, “Encoder”, (PyObject *)&pylame2_EncoderType);
}
You can now save this file as pylame2.c and compile it.
On Linux:
gcc -shared -I/usr/include/python2.4 -I/usr/include/lame pylame2.c \\
-lmp3lame -o pylame2.so
On Windows:
cl /LD /IC:\\Python24\\include /IC:\\lame-3.96.1\\include pylame2.c \\
C:\\Python24\\libs\\python24.lib \\
C:\\lame-3.96.1\\libmp3lame\\Release\\libmp3lame.lib \\
C:\\lame-3.96.1\\mpglib\\Release\\mpglib.lib
Once that’s done, you can exercise your new extension module with a simple driver script written
entirely in Python:
import pylame2
INBUFSIZE = 4096
encoder = pylame2.Encoder(‘test.mp3’)
input = file(‘test.raw’, ‘rb’)
data = input.read(INBUFSIZE)
while data != ‘’:
encoder.encode(data)
data = input.read(INBUFSIZE)
input.close()
encoder.close()
379

43 Pages 421-430

▲back to top

43.1 Page 421

▲back to top
Chapter 17
That completes version 2 of your extension module. You’re able to read data from anywhere. Your sam-
ple driver is still reading from the raw input file you created earlier, but there’s nothing stopping it from
extracting that information out of a WAV file or reading it from a socket.
The only deficiency with this version of the module is that you can’t customize how the encoded data
is written. You’re going to fix that in the next revision of the module by “writing” to an object and not
directly to the file system. Intrigued? Read on.
Using Python Objects from C Code
Python’s a dynamically typed language, so it doesn’t have a formal concept of interfaces even though
we use them all the time. The most common interface is the “file” interface. Terms like “file-like object”
describe this interface. It’s really nothing more than an object that “looks like” a file object. Usually, it
can get by with only either a read or write method and nothing more.
For the next version of your extension module, you’re going to allow your users to pass in any file-like
object (supporting a write method) when constructing new encoder objects. Your encoder object will sim-
ply call the write method with the MP3-encoded bytes. You don’t have to be concerned about whether it’s
a real file object or a socket or anything else your users can dream up. This is polymorphism at its finest.
In the last version of the module, your object held a FILE *. You need to change this by adding a refer-
ence to a PyObject and removing the FILE *:
typedef struct {
PyObject_HEAD
PyObject *outfp;
lame_global_flags *gfp;
} pylame3_EncoderObject;
Encoder_new can stay the same because all it does is set outfp to NULL. Encoder_dealloc, however,
needs to be modified:
static void Encoder_dealloc(pylame3_EncoderObject *self) {
if (self->gfp) {
lame_close(self->gfp);
}
Py_XDECREF(self->outfp);
self->ob_type->tp_free(self);
}
Instead of calling fclose, you use the Py_XDECREF macro to decrement the reference count by one. You
can’t delete the object, because there might be other references to it. In fact, other references to this object
are likely because the object came from outside of this module. You didn’t create it, but somebody else
did and passed it in to you. They probably still have a variable bound to that object.
If you’re decrementing the reference count here in Encoder_dealloc, you must be incrementing it
someplace else. You’re doing that in Encoder_init:
380

43.2 Page 422

▲back to top
Extension Programming with C
static int Encoder_init(pylame3_EncoderObject *self,
PyObject *args, PyObjecti*kw) {
PyObject *outfp;
if (!PyArg_ParseTuple(args, “O”, &outfp)) {
return -1;
}
if (self->outfp || self->gfp) {
PyErr_SetString(PyExc_Exception, “__init__ already called”);
return -1;
}
self->outfp = outfp;
Py_INCREF(self->outfp);
self->gfp = lame_init();
lame_init_params(self->gfp);
return 0;
}
You’ve modified the format string for PyArg_ParseTuple to contain “O” instead of “s”. “O” indi-
cates that you want an object pointer. You don’t care what type of object it is; you just don’t want
PyArg_ParseTuple to do any kind of conversion from the object to some primitive C data type.
After you’re sure you were passed the correct number of arguments and __init__ hasn’t been called
before, you can store the object argument for later use. Here you’re using the Py_INCREF macro to
increment the reference count. This will keep the object alive until you decrement the count.
Why did the previous macro, Py_XDECREF, have an “X” in it, while this one did not? There are actually two
forms of these macros. The “X” versions check to ensure that the pointer isn’t NULL before adjusting the refer-
ence count. The other two don’t do that check. They’re faster, but you have to know what you’re doing in
order to use them correctly. The documentation for PyArg_ParseTuple tells us that if it succeeds, the output
pointer will be valid, so I felt safe using Py_INCREF here, but I didn’t feel that safe with Encoder_dealloc.
Making sure that you perfectly balance your increments with your decrements is the trickiest part of
implementing extension modules, so be careful. If you don’t, you could leak memory, or you might
access an object that’s already been deleted, which is never a good thing.
It’s also very important to pay attention to the documentation for the different API functions you use in
terms of references. Some functions will increase the reference count before returning it. Others won’t.
The documentation for PyArg_ParseTuple states that the reference count is not increased, which is
why we have to increment it if we expect it to stick around for as long as we need it.
Now that you have an object (that hopefully has a write method), you need to use it. Instead of calling
fwrite in Encoder_encode and Encoder_close, you want to call the write method on your object.
The Python API has a function called PyObject_CallMethod that will do exactly what you need it to
do. Here’s the snippet of code you would use in both Encoder_encode and Encoder_close to call the
write method on your object:
PyObject* write_result = PyObject_CallMethod(
self->outfp, “write”, “(s#)”, mp3_buffer, mp3_bytes);
if (!write_result) {
free(mp3_buffer);
return NULL;
}
Py_DECREF(write_result);
381

43.3 Page 423

▲back to top
Chapter 17
PyObject_CallMethod requires three parameters. The first is the object on which you’re invoking the
method. This object will be the first argument into the method, usually called self. The second argu-
ment to PyObject_CallMethod is the name of the method. The third argument is a format string
describing the arguments. This can be NULL if there are no arguments. When it’s not NULL, it looks very
similar to a PyArg_ParseTuple format string except it’s always surrounded with parentheses.
PyObject_CallMethod is basically calling Py_BuildValue for you with these parameters, and the
tuple that results is being passed in to your method.
PyObject_CallMethod returns a PyObject *. All write method implementations probably return
None, but you’re still responsible for decrementing the reference count.
Because most of pylame3.c hasn’t changed from pylame2.c, I won’t include the entire file here. It
shouldn’t be too difficult to insert the changes described in this section.
Once the new version of the module is compiled, you can use any file-like object you want as a parame-
ter to the Encoder object. Here’s an example that demonstrates this:
import pylame3
INBUFSIZE = 4096
class MyFile(file):
def __init__(self, path, mode):
file.__init__(self, path, mode)
self.n = 0
def write(self, s):
file.write(self, s)
self.n += 1
output = MyFile(‘test3.mp3’, ‘wb’)
encoder = pylame3.Encoder(output)
input = file(‘test.raw’, ‘rb’)
data = input.read(INBUFSIZE)
while data != ‘’:
encoder.encode(data)
data = input.read(INBUFSIZE)
input.close()
encoder.close()
output.close()
print ‘output.write was called %d times’ % output.n
This example includes a class derived from the built-in file object to show off some of the stuff you can
do. OK, it’s not that impressive, but it at least shows how flexible your new extension module can be. As
long as you pass in an object that has a write method, your extension module is happy.
382

43.4 Page 424

▲back to top
Extension Programming with C
Summary
In this chapter, you learned how to expose simple functions implemented in C to Python developers by
creating an extension module and defining a method table. Converting Python objects to C values is
done using PyArg_ParseTuple. Going the opposite way, turning a C value into a Python object is done
using Py_BuildValue.
You also looked at how to define new types in an extension module by defining the object and type
structures. You set up the type object so that it could create new instances of your type and later destroy
them. Making sure that you correctly increment and decrement the reference counts of objects that you
use requires careful consideration.
There’s a lot more to writing extension modules, of course, but not enough room in one chapter to
cover it all. Be sure to consult the documentation at http://docs.python.org/ext/ext.html and
http://docs.python.org/api/api.html.
Exercises
1. Add a new module-level function to the foo module you created earlier in the chapter. Call the
function reverse_tuple and implement it so that it accepts one tuple as an argument and
returns a similarly sized tuple with the elements in reverse order. Completing this exercise is going
to require research on your part because you need to know how to “unpack” a tuple. You already
know one way to create a tuple (using Py_BuildValue), but that’s not going to work for this exer-
cise, because you want your function to work with tuples of arbitrary size. The Python/C API
documentation for tuples (at http://docs.python.org/api/tupleObjects.html) lists all of
the functions you need to accomplish this. Be careful with your reference counting!
2. List and dictionary objects are an extremely important part of nearly all Python applications
so it would be useful to learn how to manipulate those objects from C. Add another function to
the foo module called dict2list that accepts a dictionary as a parameter and returns a list.
The members of the list should alternate between the keys and the values in the dictionary. The
order isn’t important as long as each key is followed by its value. You’ll have to look up how to
iterate over the items in the dictionary (hint: look up PyDict_Next) and how to create a list and
append items to it (hint: look up PyList_New and PyList_Append).
383

43.5 Page 425

▲back to top

43.6 Page 426

▲back to top
18
Writing Shareware and
Commercial Programs
Python is not just for open-source applications, even though it is a successful open-source project.
Two of Python’s greatest advantages are rapid application development because of how easy it is to
develop in it, and how it can integrate with existing code even when it’s written in another language.
Commercial entities also want to take advantage of being able to develop application features
faster than their competition does. This is not an advantage that can be overstated in the commer-
cial software environment. Anything that one programmer can do another programmer can dupli-
cate, but it takes time. The faster and more efficiently a company can develop its products, the
more likely that company is to be able to meet their customers’ needs before one of their competi-
tors manages to do the same.
Conservative estimates place Python’s productivity at one level of magnitude greater than that of
other cross-platform development environments, such as .NET and Java, for large programs. That’s
a lot of saved time and money for a development organization and a significant competitive advan-
tage for the company that chooses to use Python rather than Java or .NET, or, of course, C or C++
(which can’t be considered either fast to develop in or cross-platform by any modern definition).
With these advantages in mind, this chapter should help you decide how to use Python in a com-
mercial context, whether as a smaller company doing shareware or small applications or a large
company doing large-scale development.
A Case Study: Background
Think of this section as a case study to which you can refer when designing your own solutions, so
you can frame some of the decisions you will likely have to make when starting a new commercial
project in Python.
A software security company, Immunity, was formed in 2002 in New York City, shortly after
the attacks on the World Trade Center. This company had the traditional specialized software

43.7 Page 427

▲back to top
Chapter 18
company’s tripod of services: training, consulting, and a proprietary software platform. The proprietary
software, Immunity CANVAS, needed to do several complex things:
❑ Replicate many complex network protocols, including SSL, OncRPC, DCE-RPC, FTP, HTTP,
and so on
❑ Conduct mathematical operations on data as if that data were composed of C-like primitives
(unsigned integers, signed integers, C-style strings, and so on)
❑ Port to at least Windows and Linux
❑ Have a nice, commercial-quality GUI
Currently, some advantages to selling software to the software security market do exist; chief among them
is the fact that most purchasers of CANVAS are likely to be highly technical. In fact, a large majority of
them will be used to using Linux and installing software for Linux, making it a viable platform to target.
How Much Python Should You Use?
Commercial programs are not written in a vacuum. Ideally, when a commercial project is launched, it has
been created to meet the requirements of clients who have a valid business need, and who have acknowl-
edged that need and are looking for a solution. While an open-source program doesn’t face the threat of
disappearing due to a lack of unit sales (as it can only be killed off by a lack of interest from its developers,
not a lack of customers), a commercial enterprise represents an ongoing investment, often in the face of
other software companies making counter investments with the hopes of occupying the same niche. It’s
worth noting that the main competition of Immunity CANVAS, CORE Impact, also uses Python, although
the bulk of the product is written in C++. Both CANVAS and Impact are exploitation frameworks, which
means that they allow companies to test the security of their systems and networks by simulating attacks
on known and suspected vulnerabilities, and both have their exploits written in Python.
Impact uses a closed-source engine written in C/C++ to help manage and run the exploits and to
manage the GUI. This set the stage for the first decision the Immunity developers had to make: Given
that their competition was also using Python and it provided obvious benefits, how could they maxi-
mize their advantage by using Python? Immunity chose to do two things at the design level to maintain
a rapid application development advantage over CORE:
1. Move all of the CANVAS codebase to Python, including the GUI and underlying exploitation
framework
2. Support a command-line interface, which enables a faster development cycle by isolating the
engine from the GUI, and which enables the use of standard debuggers and profiling tools
The trade-off here is that by using a pure Python underlying framework, it becomes extremely difficult
to inhibit software “piracy.” Python, as a (mainly) interpreted language, creates program files that are far
easier to reverse-engineer than files created in a language like C or C++, where the source code is com-
piled to a binary format after being heavily mangled in the process by intermediary optimization and
translation phases. Therefore, the decision to use Python has consequences that reach into every level of
the development plan, and often reflects strongly in the production of the business plan. The first issue
is keeping the customers honest. It’s common knowledge that very few companies, no matter how well-
meaning, are 100 percent honest about how many places they’ve installed software. It’s understood that,
386

43.8 Page 428

▲back to top
Writing Shareware and Commercial Programs
typically, one paid copy of software will be used by 10 people at the same time. If this is something that
you’re going to need to address, see the section “Pure Python Licensing” that follows for some ideas on
how you can offset this.
If your competition is not using Python, or if you have a larger developer team than your competition,
you can package and embed a Python interpreter in your program and still maintain some of the advan-
tages that Python offers for rapid application development, while using traditional mechanisms to allow
flexible software licensing, such as integrating a commercial license manager into your package. However,
this isn’t possible if you’re expecting to use the Python interpreter that is already installed on your
clients’ systems.
Even if you’ve decided to write all your code in Python, you still have another decision to make: Should
you allow your developers the capability to import and use modules that are not from the base modules in
the Python distribution? The base modules are guaranteed to be installed with every Python installation,
and they have liberal BSD-like licenses. However, you might want to include features that already exist in
another library in your software, but do it without the costly investment of having to re-implement these
features. Including them directly in your software places the onus on you to determine the licensing
requirements and handle any related issues yourself, whereas going the other route and relying on external
libraries and modules having to exist on the clients’ systems means having to find a way of ensuring that
your users can obtain the required libraries, if you’re not going to distribute them yourself.
When you write software that relies on any other software that you haven’t written,
it’s important to get the licensing right. A BSD-style license allows you to use the
source code in free or commercial products without requiring you to pay a license
fee, and grants this right irrevocably. However, it does not grant you the right to
claim any copyright on the product you are using. To understand some of the
licenses that come with open-source software, you can read the following document:
http://cyber.law.harvard.edu/openlaw/gpl.pdf. For commercial products,
you should refer to the license that you received when you purchased it, as there are
as many licenses as there are products.
Immunity decided to go with a pure-Python approach for CANVAS, except with regard to their GUI
library, for which they chose pyGTK (although this means that the development is still done in Python;
see Chapter 13 for information on pyGTK). Key to making your choice should be a survey of projects that
are currently using Python to do commercial work, and determining whether the problems that you will
be facing have already been solved by others. If they have, then there is a high probability that you will be
able to solve your problems with Python as well. For some ideas about what has been done, look at Zope,
Immunity CANVAS, WingIDE, and Ximian RedCarpet. These are all good products to look at as you
frame your software and business architectures, and each one solves a different set of problems.
Pure Python Licensing
If you have chosen to go with a pure-Python model, you can still take steps to allow for lucrative licens-
ing models. Your goal is as follows: You don’t want to turn off your customers or give up too many of
the advantages of using Python in the first place, but you still want to be able to control the distribution
of your software and restrict users to using what they’ve paid for.
387

43.9 Page 429

▲back to top
Chapter 18
Whatever you do, don’t forget to market the advantages to your customers of your open development
platform as strongly as you can. Immunity has found this to be a key differentiation between themselves
and their competition; Immunity’s CANVAS is transparent from front to back, so customers can add to
and customize the product, which enables them to do more with the product than they could have ever
imagined. No business can truly predict what their customers are going to do with their product, and
your business is no exception. That’s why one of the hardest decisions you have to make is how to
create your software license, because it can affect how your customers can use your product.
Licenses are all about restrictions. The problem with restrictions regarding usage of your product is that
they typically require a binary copy protection module that cannot be removed from your product, as
shown in the following example:
if licensedDate < todaysDate:
print “You have run out of license time!”
sys.exit(1)
Although this bit of code may work well in principle within a compiled C program because by its nature
the code created from it is hard to discern, it’s trivial for even the most basic Python programmer to
remove it from your program because it appears in readable text.
To avoid this obviousness, you could distribute your Python as a .pyc or .pyo if you like, but these are
easily reverse-engineered as well. In other words, enforcing something like an annually renewed license
scheme by restricting based on the date, as above, is nearly impossible. Likewise, programs that require
“activation” are nearly impossible to create without a binary module in which you can hide the work-
ings of this protection.
WingIDE 2.0 manages to do this form of activation-based license control well, so it might be worth
your time to take a look at how they do it if you decide to go down the product activation road. The key
to their success is that they distribute their own Python engine, rather than use a provided Python on the
host system. However, this style of distribution may not work for you. Keep in mind that even with this
system in place, WingIDE is also distributed as a pure-source license on certain platforms.
Not being able to have binary modules important for the functioning of your program leaves you at a
loss for restricting the usage of your product based on time or the number of installations at a site.
Web Services Are Your Friend
As you’ll see in Chapter 21, Python support for web services is thorough, complete, and once you’ve
learned something about using it, drop-dead simple to use. This is a major advantage over almost every
language except C#, and Python’s dynamic typing makes it even better for this sort of use. If you think
of your product as a rich client to a number of server services implemented on a web server you control
(which you should also in write in Python, keeping your Rapid Application Development cycle advan-
tage), you have almost all the advantages of a binary-only model when it comes to licensing because you
can control use of the product from the central web site, without the disadvantage of having to use a
compiled language.
This, of course, requires that you have some kind of functionality that you can centralize efficiently and
that your users are always able to contact your main server. You could also consider selling (at an appro-
priately high price) a version of the central server that can be installed on-site, in an enterprise client’s
network.
388

43.10 Page 430

▲back to top
Writing Shareware and Commercial Programs
Exploits
An exploit is a program that takes advantage of a buffer overflow or similar vulnera-
bility to obtain unauthorized access to a remote computer. The most famous exploits,
and those that are covered the most in the media, are those used by worms such as
Code Red to spread themselves.
One example of this technique is an upcoming product from Immunity based on their flagship product
CANVAS. CANVAS is an information security product that automates phases of security penetration
tests. A penetration test often involves using a number of known security exploits against machines in
your organization to determine whether those exploits will work.
One particular problem when running penetration tests is that often the targeted machine will need to
connect back to the tester’s machine, and firewalls in between the tester’s machine and the target will
have filters preventing outbound connections to the tester, leaving the target still vulnerable but untested.
A simple solution is to have the target make fake DNS queries (a protocol that will usually be passed
through firewalls internal to an organization). However, this requires a specialized and customized DNS
server that enables you to use DNS as a covert channel and can integrate into the full CANVAS frame-
work. If you put that DNS server on a central server that also covers licensing, it’s a perfect fit for the
web services licensing model because it’s a publicly available central server, enabling Immunity to track
compliance with per-seat licenses. Of course, for an additional charge, the company can sell the modi-
fied DNS server and have customers integrate with them locally. This is the sort of opportunity you
should look for when trying to maximize your programming advantage using Python. Although you
could still do product-activation-style licensing using a binary component from a compiled language,
it’s better in the long run for your business to evolve past the need for restrictive licensing on the client
side, which is always difficult and can be hard to troubleshoot if it misbehaves.
Pricing Strategies
Realistic pricing for software has to take into account the actual size of your market. If you assume your
product will be widely pirated, as is true with most programs that don’t have per-computer activation,
then your pricing needs to rise accordingly or depend on external factors such as support and training,
while being weighed against the value it offers to those clients who will be paying.
Per-company licensing is a common tactic used by software vendors looking to capitalize on an “all you
can eat buffet” desire among their customers. Selling an Enterprise license, which allows for unlimited
use within an organization, can be an easy way for you to compensate for the lack of control your pro-
gram inherently has by not placing any technical restrictions on the software, instead relying on the
clients to not compromise the valuable product for which they’ve paid.
Likewise, you can compensate for your technology’s inability to protect itself from being copied by
using legal means. An NDA with appropriately strict penalties can scare companies straight faster than
a shrink-wrap license can. However, to truly enforce an NDA, you need a way to prove that a company
is leaking your information. This is where the technology comes to your aid again.
389

44 Pages 431-440

▲back to top

44.1 Page 431

▲back to top
Chapter 18
Watermarking
Proving someone is leaking your software is harder than it sounds. Unless you have an inside source,
you need to be able to take a leaked piece of software and say “This came from Company X.” This can
be more difficult to prove than it is to say. It means you have to manage a database of all your users, dis-
tribute uniquely different software packages to each one, and have a mechanism for dealing with leaks
(some way to enforce the revocation of the right to use the software).
One way to make each distributed package different is through watermarking. Watermarking is a
generic term for any process that is not observable to the user but that enables you to track down the
origin of any particular piece of software traded on the Internet at large. The simplest form used is to
include the name of the user in the executable somewhere. This sort of simplistic measure is easily
detected and bypassed by the user, however.
What you want is something provable and hard to detect. In mathematical terms, you would want a
reversible function that you can apply to the input text (your program), one that is a one-to-one mapping
on your user database, so you can know with 100 percent confidence to whom the software was given
and therefore who is responsible for its being leaked in violation of its license.
Python is a whitespace-sensitive language, which somewhat restricts your ability to change the contents
of a file while still having a functional program, something most watermarking programs would nor-
mally do on an executable. You can still modify many parts of the Python program you’re distributing,
however, and you have the advantage that Python is very good at parsing programs written in Python,
so many changes would be harmless. Still, the basic techniques for doing this don’t require advanced
and customized parsing of any kind. The following example comes from the Immunity CANVAS distri-
bution code and is normally installed as a CGI. This will do the right thing when it comes to delivering
the file to the client via their browser, which is not as easy as it might originally seem. If you’re running
under Apache, and you do a print statement, Apache will sometimes add a \\n to the end of the file,
which corrupts it. Use sys.stdout.write instead of print to avoid this.
Security is an important part of your distribution scheme. Note that the use of md5 hashes (nicely
importable in Python) and the general security of the Python system allows for a higher level of confi-
dence than most web frameworks have. You’ll rarely see the source code to a production C CGI being
distributed publicly in a book!
You’ll see chunks of code that perform these functions, interspersed with comments about what the CGI
is doing, as follows:
def normalizeDate(date):
if date.count(“/”) == 0:
return date
dates = date.split(“/”)
#add the year if just a 05 -> 2005
if len(dates[2]) == 2:
dates[2] = “20” + dates[2]
newdate = “%4.4d%2.2d%2.2d” % (int(dates[2]), int(dates[0]), int(dates[1]))
return newdate
390

44.2 Page 432

▲back to top
Writing Shareware and Commercial Programs
The preceding code fragment demonstrates some easy ways to handle dates with Python. The input to
this function is a string in the format of “5/12/2004” (May 12, 2004). Typically, this would be the expira-
tion date of the user’s subscription. All such dates are normalized by this function to be in the format
“20050512”. The advantage of this is that later, dates can be easily compared with simple subtractions:
#md5 sums a file
def getmd5sum(filename):
m = md5.new()
f = open(filename)
#insert error checking here
s = f.read()
f.close()
m.update(s)
return m.hexdigest()
#md5 sums data
def getmd5(data):
m = md5.new()
m.update(data)
#sys.stderr.write(“%s”%m)
return m.hexdigest()
These two functions show how to use the md5 module in your programs to verify a file’s integrity by
producing a small enough number that can be compared against a known value.
Of course, there is a simple gotcha in the getmd5sum function for the sake of simplification. If you use
the read method of the f file, and read it all into s, and if the file you’re reading is very, very large, it can
cause s to take up all of your free memory, causing the Python program to crash or, even worse, make
your system unusable by hogging that important resource:
def getresult(who,ip,passwd):
userdb = {}
filedb = {}
data = file(“newuserdb.txt”).readlines()
for line in data:
line = line.replace(“\\r”,””).replace(“\\n”,””)
if line == “”:
continue
#print “Doing %s”%line
stuff=line.split(“ “)
try:
name = stuff[0]
date = stuff[1]
passwordhash = stuff[2]
number = stuff[3]
userdb[name] = [date,passwordhash,number]
except IndexError:
#print “IndexError on *%s*”%line
pass
From reading the preceding fragment, you should have a clear sense of the format of the CANVAS cus-
tomer database file. Your own customer database file may be similar, assuming you keep it as a plaintext
file. It’s also perfectly acceptable to use something like Pickle() here for code as small as this.
391

44.3 Page 433

▲back to top
Chapter 18
Alternatively, you could use Python’s excellent database support to connect to a mySQL or other rela-
tional database that is more suitable for large data sets. One good reason to keep it as a text file as we do
here is that you’ll invariably find yourself looking at it and editing it manually to fix the kind of data
entries problems that lead to common customer billing problems.
In this case, a sample line we’re reading in might look like this:
JUSTINE 5/12/2005 6372a7c27d981de94e464b934e9a6ebc 279321
The preceding code says that there is a user JUSTINE whose subscription expires on 5/12/2005, and
who has a user ID of 279321. The long hexadecimal string between the date and JUSTINE’s user ID is the
password hash (in a different format than what was used in the example in Chapter 9). Remember that
this is a continuation of the function, so make sure that your indentation is correct:
#check existence of user
if not userdb.has_key(who):
#print “No %s in userdb!”%who
error(ip=ip,who=who)
At any point in the program, if we find that there is an error of any kind, we print an error message and
bail. CGI programs need to be careful about what they print out — anything they print may show up on
the user’s screen as the result of their query! Therefore, print statements are commented out here, but
they can be enabled if you are testing this from the command line.
#check password
hash = userdb[who][1]
if hash != getmd5(passwd):
#print “Passwords did not match (%s)!”%hash
error(ip = ip, who = who)
Note here that we give no indication as to whether it’s the user name or the password that was incorrect.
This is proper security best practices for web applications. While the webmaster and application devel-
oper should be able to get the information as it’s printed in the error function, the user reading the error
message in his or her browser shouldn’t be given information that could indicate to a hostile attacker
that they’ve found the name of a valid user.
date = userdb[who][0]
The preceding line will return the date of this user (now authenticated with user name and password)
from our customer database.
number = int(userdb[who][2])
In assigning the values of date and number, we pull the customer number from the user database. This
line assumes that the integers are in base 10. If you think you’ll be dealing with numbers such as
0x01020304, you’ll want to use int(number,0) to automatically convert them.
date_normalized = normalizeDate(date)
import time
today = “%4.4d%2.2d%2.2d” % time.localtime()[:3]
392

44.4 Page 434

▲back to top
Writing Shareware and Commercial Programs
Note that you can import modules at any point, and you may find yourself doing this in production
code, although it’s not the best form.
if int(normalizeDate(date)) < int(today):
#customer has run out of canvas updates
#Note - in production this is formatted as html
print “Your subscription has run out on %s” % date
sys.exit(1)
The ability to compare dates with a simple integer operation is why we have normalizeDate. It enables
a human-readable date to be converted to a format that Python can work with easily.
logfd=open(“/var/CANVAS/log.txt”,”a+”)
if logfd != None:
import time
logfd.write(“CANVAS downloaded:
%s %s %s\\n”%(time.ctime(), ip, who))
logfd.close()
try:
import os
#remove old directory, if one exists
os.system(“rm -rf /var/CANVAS/CANVAS_%s”%who )
#create a new CANVAS directory for this person
os.system(“cp -R /var/CANVAS/CANVAS_DEMO /var/CANVAS/CANVAS_%s”%who)
#then add watermark
The watermarking code has been removed for public release — you can easily come up with your own.
We’re careful here to protect the input to os.system(). If you get lazy, using os.system() can have
severe consequences for your security. In Python 2.4 and later, the subprocess.call() function has
been added specifically to help address this.
#then compress and archive.
#note: only works on linux or places with gnu tar
os.system(“cd /var/CANVAS; tar -czf CANVAS_%s.tgz CANVAS_%s > /dev/null” %
(who, who))
os.system(“rm -rf /var/CANVAS/CANVAS_%s” % who)
#then serve up
fd = open(“/var/CANVAS/CANVAS_” + who + “.tgz”)
except:
import traceback
traceback.print_exc(file = sys.stderr)
The traceback module is useful for your own debugging — we’re careful to send its output to stderr here
so that the user doesn’t learn too much about our CGI script. (Of course, it’s now been printed — the
whole script — in a book!)
print “CANVAS not found!”
error()
sys.exit(1)
data = fd.read()
fd.close()
os.system(“rm -rf /var/CANVAS/CANVAS_%s.tgz” % who)
393

44.5 Page 435

▲back to top
Chapter 18
print “Content-Type: application/octet-stream”
print “Content-Disposition: attachment; filename=\\”%s\\”” % (“CANVAS_” + who +
“.tgz”)
print “Content-Size: %d” % len(data)
print “”
sys.stdout.write(data)
sys.exit(1)
This next, and final, code fragment starts the whole script off by loading in variables passed from the
web server using the cgi.FieldStorage method.
def run_as_cgi():
form = cgi.FieldStorage()
#print “””Content-Type: text/html
#
#”””
#print “form=*%s*”%os.getenv(“REMOTE_ADDR”)
if form.has_key(‘username’):
if not form.has_key(‘password’):
error()
who = form[‘username’].value
passwd = form[‘password’].value
ip = os.getenv(“REMOTE_ADDR”)
getresult(who, ip, passwd)
else:
# If there are no form values, return a page with the form itself.
# See chapter 22 for how to write a form for yourself. Remember
# that you need to include all of the form elements that will allow this to
succeed!
print formhtml
Watermarking is a funny thing — often, simply saying you do watermarking is as effective as doing it. If
you do decide to go this route, you’ll want to change your watermark every six months or so. But you
should note that anyone with two copies of your program will be able to erase it, because they’ll be able
to compare the two versions.
And that’s it! That’s a working starting point for watermarking software.
Other Models
Looking at your product as more than just licensing revenue enables you to structure it to take advan-
tage of Python in many other ways. This simple set of code exemplifies the main revenue stream of
CANVAS: recurring subscriptions:
def registerModule(name):
“imports and adds a exploit module to our list, returns 1 on success”
#print “RegisterModule %s”%name
if name in exploitnames:
return
sys.path.append(“exploits/%s”%name)
try:
code=”import %s as exploitmod”%name
exec code
394

44.6 Page 436

▲back to top
Writing Shareware and Commercial Programs
except:
print “Was unable to import %s”%name
return 0 #failure
#go on to add it to our list, since we were able to import it
exploitnames.append(name)
exploitmods.append(exploitmod)
return 1 #success
This code uses a directory name from the CANVAS_name/exploits/ directory and then assumes that
the directory contains a file with the same name as the directory, with .py. It then creates a mini-script to
import that module into the current namespace, and adds it to our internal list of modules.
As you can see, the CANVAS product is written so that it can continually be expanded with additional
modules, which run within the framework. These enable the company selling it as a service to provide a
continually expanding value to customers willing to pay for a subscription.
Therefore, in your own endeavors, take advantage of Python’s dynamic nature wherever possible to pro-
vide this kind of functionality in your product. While a C++ programmer can look to a shared library
(DLL or .so files) loading as a way of expanding their features, only Python can do this so quickly and
easily, while also offering a variety of important buzzwords that actually mean something: introspection,
comprehensive exception handling, and portability built in!
Selling as a Platform, Rather Than a Product
Likewise, now that you have extensibility built into your product in fifteen lines or less, you can offer
your product as a framework on which other people can build. Microsoft calls this “developing an
ecosystem,” whereby they sell one copy of Microsoft Windows to everyone on Earth, and everyone else
has to build to that standard. Not that this will automatically make you into Microsoft, but you don’t
have to look too far to see that this model can work.
This is where you may find that having a completely open codebase at each of your customer sites is a
huge advantage. They can build it into their own processes, doing things you would never have imag-
ined. For instance, early on in CANVAS development, Immunity sold a copy to a large software com-
pany that then developed a process of their own using CANVAS. They were using CANVAS to scan
their entire network once a day. They would use the results to automatically upload the patches needed
to address the issue that had permitted them to break in. This is the sort of automation you allow your
customers to have when your product is Pure-Python. They will find a lot of value in that.
Additionally, Python is so easy to debug that many non-programmers have sent Immunity patches
they’ve figured out on their own in their environment. Although these patches may not always have
been of the quality you wanted, it’s important to note that you won’t see customers going to that sort of
effort for any compiled product or even a large open- source C codebase.
Your Development Environment
The right IDE makes all the difference with Python. Although many Python programmers stick to basic
editors like vim, it and other vi variants tend to deal poorly with tabs and spaces, and intermixed tabs
and spaces can make it nearly impossible to find errors in your programs. Besides, it makes sense to take
full advantage of anything that can make your development team more agile than the competition —
that’s why you went with Python in the first place!
395

44.7 Page 437

▲back to top
Chapter 18
To make everything work as well as possible, you should have your entire development team working
with the same IDE and have them use the same number of spaces for their indentation. This will save
you reams of problems down the road.
Immunity standardized on WingIDE, although BlackAddr and Emacs also work. Figure 18-1 shows
WingIDE 2.0 running on Windows XP. WingIDE has several other features specific to Python that you
may not have needed in your C/C++/Java IDE. Primary among these features is a “Source Assistant,”
which will attempt to guess the type of any variable you click. This can be a time-saving feature for
Python, which is dynamically typed, which essentially means that the value of a variable may be differ-
ent at different times. The key to selecting an IDE is to choose one that runs on all of the platforms your
developers use — in this case study, Linux and Windows.
Figure 18-1
Part of choosing your IDE is acknowledging that your IDE should come from the same community as
your product. In this case, WingIDE developers can often be found on the pyGTK mailing lists. This lets
you know that you’ll be well supported when you have problems debugging our own pyGTK-based
program. If your product is QT-based, BlackAddr might be a good choice. If you’re already familiar with
it, Emacs is also quite good with Python, and IDLE is always available as a backup because it comes
with Python. Of course, each Python programmer has his or her favorite, and as Python becomes more
and more popular, the field of editors and IDEs gets more crowded. CodeEditor (http://pythoncard.
sourceforge.net/tools/codeEditor.html) is another favorite.
Finding Python Programmers
Whenever a company decides to build using a language or platform, they need to look at how difficult
it is to find developers to work on that product as they expand. Even if your initial team are all master
Python programmers, unless you can replace them as they come and go, and unless you can grow at a
reasonable cost, Python might not be your best choice of languages. Thankfully, Python programmers
are quite easy to find — often, they’re willing to leave other jobs to program in Python. There are really
two major ways to grow your team, as described in the following sections.
396

44.8 Page 438

▲back to top
Writing Shareware and Commercial Programs
Training non-Python Programmers
Python is reasonably C-like. Once you can get your new hire past the idea that he has to work within
WingIDE (or your IDE of choice) and he has to work within a whitespace-sensitive language, it’s easy to
get him up to speed on Python itself. This book might even help get him past the initial hump of learn-
ing new APIs.
The IDE shouldn’t be a big deal, but programmers can be creatures of habit. This is understandable, as
once you’ve dedicated time to learning a complex set of tools like an editor, it’s hard to take baby steps
again. You need to emphasize the benefits of the environment you’re offering, and provide new Python
programmers the time they need to learn their environment.
Python Employment Resources
Finding Python programmers specifically can be done on any technical mailing list. You’ll be surprised
how many skilled Python programmers there are in your direct community. Immunity draws from
security mailing lists and conferences (Python has made significant inroads into the security community)
and from the local Linux User Group mailing lists. Finding a list of all Python projects on SourceForge
and e-mailing the developers of those will get you more responses than you might think.
Largely, however, because Python is so easy to learn, look for domain expertise first — it’s often harder
to get someone who can think about the problems you’re working on. Once you’ve got that, you’ll
develop their Python experience later.
Python Problems
Many people initially go through an infatuation phase with Python. Then, like in any relationship, they
realize that Python is not perfect. In fact, the more you use it, the more you realize there are gremlins at
every corner. Although some of those gremlins are covered here, it helps to know that not all is perfect in
the promised land.
Porting to Other Versions of Python
Python can feel like it is not a stable platform for development if you use portions of it that are changing
and you aren’t forewarned. Compared to a stable C software stack, Python applications that rely on
these changing parts of the language can appear to be fragile. If you want to build something large, you
will most likely have to stay with one version of Python, rather that let your users and developers use
the most recent version of Python. This is often a problem, as you now face a dilemma:
1. The binary modules (such as pyGTK) that you rely on will usually only support the latest ver-
sion of Python.
2. Some things in the new Python are guaranteed to break your application in subtle ways, until
you and other users have a chance to shake this out. This process does not directly benefit your
business.
This is not a dilemma any company wants to face with their development platform, though some form
of this dilemma will often turn up no matter what development language or environment you pick. The
only real solution is more engagement with the open-source community that develops the software
397

44.9 Page 439

▲back to top
Chapter 18
stack. If you have the skills on hand, you can become a pyGTK (or whatever module you use) developer
and do testing and maintenance on old versions for them, and receive a quid pro quo of getting better
support than you could possibly buy.
Making the decision to give up the benefit of moving to new versions of Python or paying the size and
complexity price of distributing your own Python with your product may initially seem counter-intuitive,
but take a quick look at some of the things the Python development process brings to you that you may
not be expecting if you are used to developing to a more commercialized platform:
On Python 2.3.4 or previous Python versions, the following code will produce the result you would
expect from a C program:
>>> print “%x” % -2
fffffffe
However, a later version of Python will behave differently, as seen below with Python 2.4:
>>> print “%x” % -2
-2
In this trivial example, the problem is obvious, and somewhat documented in “future” warnings in ver-
sions of Python prior to 2.4. (When you see warnings pertaining to future changes to the language, if
you expect to use that features you should get onto a Python mailing list and ask what will change! It
can save you a lot of work down the line.) Once your code is written and tested, and has stabilized into
everyday use, however, this kind of problem may be hidden under many layers of complexity.
This kind of language wart may seem like a small thing. But what happened is that everything under
the covers regarding how unsigned integer types are implemented has changed. Integers are a base type
you may be using in more than a few places in your program, so this kind of change can be very signifi-
cant. This is the sort of thing that introduces subtle bugs, costing development time that could otherwise
be going into new and cool features. This is not the only thing that changed in Python 2.4 that will affect
you, and discovering every new incompatibility is going to potentially destroy your schedule if you
don’t keep an eye out for them.
The Python community doesn’t seem to see this as a problem — they’re simply not as conservative about
such changes as other platform stacks are, but it is a huge reason for other platforms’ commercial accep-
tance, such as Java’s. Stability is more important than almost anything else when you’re looking at a
software development life cycle of ten years or more, and Python still doesn’t have that mindset.
As another example, the struct.pack(“L”,1) function will return 32 bits on 32-bit systems, and 64
bits on 64-bit systems (such as the new AMD processor). This can completely break your network proto-
col support silently and irrevocably. For this reason, large parts of CANVAS eschew struct.pack and
struct.unpack altogether, functions that are prolific in common networking code.
Porting to Other Operating Systems
Python is not just a virtual machine, an API, and a set of syntax. It’s also a set of libraries that wraps each
operating system’s internal API and converts the OS concepts to Python concepts. This is important
when dealing with threading, certainly, but a lot of other APIs can cause you problems because the
Python community wrapped them thinly, or without a lot of work to isolate you, the Python program-
mer, from these differences between platforms.
398

44.10 Page 440

▲back to top
Writing Shareware and Commercial Programs
Marshalling
In computer science terms, to marshall or unmarshall is to take a program’s variables
and transform them into a binary representation, which can then be sent over the net-
work or used by another API. The Pickle module’s capability to turn Python objects
into a string is an example of marshalling. This is commonly used in Remote Procedure
Call (RPC) libraries, and is also used here to format arguments for the Linux system
call ioctl. Ioctl takes different arguments on each Unix-like operating system, and
can also be endian dependent, such that Linux on the SPARC would need different
arguments than Linux on x86.
For example, the socket API, a commonly used API, will throw different exceptions on Windows than it
will on Linux. This means you’ll often need a blanket try: and except:, or you’ll have to catch both
possible exceptions, making your code more complex than it seems like it should be — you’d expect that
the same error could be handled the same way on two different platforms. Not being able to do this can
make you feel that the code to do this looks and feels ugly, because it can be puzzling to see apparent
duplication of effort.
Here’s some CANVAS code that gets the IP address of a particular interface:
def getLinuxIPFromInterface(interface):
import fcntl
SIOCGIFADDR = 0x8915
s = socket.socket(socket.AF_INET, socket.SOCK_DGRAM,0)
r = fcntl.ioctl(s.fileno(), SIOCGIFADDR,interface+256*’\\x00’)
IP = socket.inet_ntoa(r[20:24])
return IP
You’ll notice that this code has to manually marshall the arguments into a poorly documented system
call using a magic number (ioctl and 0x8915, respectively) available only on Unix and Linux; and the
magic number is different on different Unix systems (other Unix platforms will have a value that is dif-
ferent from Linux, and from each other). As you can imagine, this code does not work on Mac OS X,
Solaris, Linux for SPARC or any other Unix.
The basic rule is that where Python has thin wrappers over the OS layer, your development is going to
suffer. Unfortunately, sometimes where Python has a thick, well-tested, and mature wrapper, such as for
threads, you’re going to suffer from third-party binary modules using it improperly (again, such as pyGTK).
Debugging Threads
Put simply, you can’t debug threads in Python. Print statements are as good as it gets for now. This is a
significant omission in the environment, as all real programs are heavily threaded. Programmers who
come from other backgrounds may find this difficult, so program carefully around threads.
Common Gotchas
Large books have been written on “gotchas” in various languages, but Immunity found that one gotcha
in particular riddled their early code. While many experienced Python programmers know about this
particular feature of the language, it is not intuitively obvious to beginner programmers and can cause
399

45 Pages 441-450

▲back to top

45.1 Page 441

▲back to top
Chapter 18
otherwise fast programs to grind to a halt on large data sets. The problem stems from the fact that in
Python, many data types, such as strings, are considered immutable (that is, they cannot be changed or
added to). Programmers used to C’s pointer types are often prone to this mistake more than other pro-
grammers but are also equipped with the implementation knowledge to avoid it.
Take, for example, the following Python program fragment:
A=”A”
B=”B”
A+=B is really equivalent to the C code fragment:
A = malloc(strlen(A) + strlen(B)); sprintf(A, “%s%s”, A,B);
Hence, this is an O(N) operation, which means that it takes a linear time to run. This means that, for
example, if you have 10 items, it takes 10 operations to work. If you put in 100, it takes 100. There are
faster ways to do this; and in computer science, considerable research is done to identify and avoid situ-
ations where you’re stuck with performance this bad. If you run this in a loop — say, as part of a file
download program — you will suffer horrible performance.
Therefore, this code
A = “A”
B = “B” * 50
for i in range(0,2000):
A += B
should really be this much, much faster version:
A = “A”
B = “B” * 50
alist = [A]
for i in range(0,2000):
alist.append(B)
A = “”.join(alist)
Fixing this simple gotcha may improve your performance a thousand times. Other than simple fixes like
that, Immunity tries to maintain code that is as readable as possible:
❑ Eschew the complex parts of Python — for example, lambda constructions or map statements —
wherever you can unless you know why you want them.
❑ Keeping your Python self-documenting is a matter of discipline.
❑ Always assume that the next person to debug your code is looking at Python code for the first
time. Doing this can help you write your code so that you will understand it better.
Portable Distribution
Distributing your program is more than just sending people a tarball (Unix-speak for a tar.gz file, simi-
lar in Windows to a .zip file) — it’s also about having that tarball work the way you expect it to when it
400

45.2 Page 442

▲back to top
Writing Shareware and Commercial Programs
gets there. Most users prefer having an executable that handles the heavy lifting like installation, verifi-
cation that the install works, and that all system dependencies — things that are needed to work — are
present. Although Immunity hasn’t chosen to go this route, it might be something to consider if your
customer base is not as technical as the security community tends to be.
Having a redistributable Python program means more than just having portable Python code: It means
being able to deliver that code and the software stack it relies on portably. Ideally, this functionality
would be built into the platform itself with nice GUI installers, but most organizations use either py2exe
or cx_freeze, which are much better. However, rest assured that distributing your Python program as
one large binary is possible, as long as you’re willing to pay the size limit.
❑ www.python.org/moin/Freeze — The original Python Freeze.
❑ http://starship.python.net/crew/atuining/cx_Freeze/ cx_freeze — This does not
require a compiler to modify your source Python distribution.
❑ http://starship.python.net/crew/theller/py2exe/ — The more Windows-centric
py2exe has been used by people looking for a more complex setup.
The option of the future for some applications may well be Live-CDs with built-in Linux and Python
distributions. With these, you could even control the operating system and save clients costs such as OS
licensing costs and patching and maintenance costs.
As virtual systems become more powerful (QEMU comes to mind), and Cooperative Linux (www.
colinux.org/) makes running a full Linux distribution under Windows more mainstream, you’ll find
it potentially ideal to distribute your product as an ISO or virtual machine package.
Of course, if you have a very technical customer base and you’re technical too, tarballs with INTSTALL.TXT
files work well enough to get you started, though most customers will demand more later.
Essential Libraries
Occasionally, you’ll want to import modules into your commercial program as if they were part of the
base Python distribution. The following sections describe modules that Immunity has found invaluable.
As your project matures, you’ll no doubt have a list of external, free modules you just can’t do without.
Timeoutsocket
Timeoutsocket is the first indispensable module. In addition to including some neat functionality for
wrapping the socket module, it adds timeouts to nearly all socket operations. Using timeoutsocket is
as simple as adding an import timeoutsocket and then calling mysocket.set_timeout(4) on any
newly created TCP sockets. This even affects sockets used from within libraries that know nothing about
timeoutsocket. When a socket operation times out (which is something they may not do by default, but
that you always want them to do, and preferably in a way that you control), it will throw an exception,
which you can catch. Of course, mysocket.set_timeout(None) will emulate the standard behavior
and never time out.
401

45.3 Page 443

▲back to top
Chapter 18
This fragment from timeoutsocket.py demonstrates how you can do something similar with your code:
# From timeoutsocket.py
# Silently replace the standard socket module
#
import sys
if sys.modules[“socket”].__name__ != __name__:
me = sys.modules[__name__]
sys.modules[“_timeoutsocket”] = sys.modules[“socket”]
sys.modules[“socket”] = me
for mod in sys.modules.values():
if hasattr(mod, “socket”) and type(mod.socket) == type(me):
mod.socket = me
Being able to call s.set_timeout(5) has prevented quite a few painful sections of code inside of
CANVAS. Again, if socket operations and network protocols are something on which your product
relies, consider a good strong look at the Twisted Python architecture, an entire framework that
gives you a way of approaching complex network application designs. It can be found at http://
twistedmatrix.com/projects/twisted/.
PyGTK
This module is cross-platform, free, and of an extremely high quality. As mentioned in Chapter 13,
separating your GUI from your code is a key factor in rapid application development (RAD). Immunity
wishes to spend as little time as possible writing GUI code, and as much time as possible writing appli-
cation code. PyGTK is a natural fit if you can use it.
GEOip
GEOip is a free library you can install on your server that enables you to programmatically map IP
addresses to countries. The following code block shows the basic usage of this in a CGI script, but
Immunity uses it inside CANVAS as well. Having an IP-to-country mapping is useful in many cases,
and for some extra cash, GEOip can offer you the city and state level.
#!/usr/bin/python
import os,cgi,sys,md5
os.putenv(“LD_LIBRARY_PATH”, “/usr/local/lib”) #for GeoIP
sys.path.append(“/usr/local/lib”)
os.environ[“LD_LIBRARY_PATH”] = “/usr/local/lib”
#print os.environ
def getresult(who, ip):
import GeoIP
gi = GeoIP.new(GeoIP.GEOIP_MEMORY_CACHE)
country = gi.country_code_by_addr(ip)
if country != “US”:
error(ip = ip)
# [ ... ]
402

45.4 Page 444

▲back to top
Writing Shareware and Commercial Programs
Summary
Using Python in a commercial setting as part of consumer software can be trying. While Python is great
on one computer, supporting it on thousands of computers requires a level of infrastructure you may not
have expected. In addition, unlike thicker software stacks such as Java, operating system and platform
differences leak through to the developer.
In the end, though, being able to develop your application ten times faster than the Java farm next door
may mean the difference between success and failure. Immunity has found that even a small difference
in the amount of Python you use can make a huge difference in time-to-market.
As with any technology, it helps to have your business model oriented correctly around the limitations
of the platform. Copy protection is made harder, but customer support is made easier. The trade-offs are
there, but as long as you understand them, you can use Python to deadly effect in the business world.
403

45.5 Page 445

▲back to top

45.6 Page 446

▲back to top
19
Numerical Programming
In this chapter, you will learn how to use Python to work with numbers. You’ve already seen some
arithmetic examples, but after reading this chapter, you’ll have a better understanding of the dif-
ferent ways you can represent numbers in Python, of how to perform mathematical computations,
and of efficient ways of working with large numerical data sets.
Numerical code lies at the heart of technical software, and is used widely in science, engineering,
finance, and related fields. Almost any substantial program does some nontrivial numerical com-
putation, so it pays to be familiar with some of the contents of this chapter even if you are not
working in one of these fields. For instance, if you are writing a script to analyze web logs, you
might want to compute statistics on the rate of hits on your web server; if you are writing a pro-
gram with a graphical user interface, you might need math functions to compute the coordinates
of the graphics in your GUI.
Parts of this chapter require some understanding of math beyond simple arithmetic. Feel free to
skip over these if you have forgotten the math being used. The last section of this chapter, which
discusses numerical arrays, is technically more advanced than most of the material in this book,
but it’s important reading if you plan to use Python for handling large sets of numbers.
Designing software that performs complex numerical computation, known as numerical analysis,
is both a science and an art. There are often many ways of doing a computation, and numerical
analysis tells you which of these will produce an answer closest to the correct result. Things can
get tricky, especially when working with floating-point numbers, because, as you will see, a
floating-point number is merely an approximation of a real number. This chapter mentions numer-
ical precision but doesn’t go into the finer points, so if you are embarking on writing software that
performs extensive floating-point computations, consider flipping through a book on numerical
analysis to get a sense of the kind of problems you might run into.
Numbers in Python
A number, like any object in Python, has a type. Python has four basic numerical types. Two of
these, int and long, represent integers, and float represents floating-point numbers. The fourth
numeric type, which is covered later in this chapter, represents complex floating-point numbers.

45.7 Page 447

▲back to top
Chapter 19
Integers
You’ve already seen the simplest integer type, int. If you write an ordinary number in your program
like 42, called a literal number, Python creates an int object for it:
>>> x = 42
>>> type(x)
<type ‘int’>
You didn’t have to construct the int explicitly, but you could if you want, like this:
>>> x = int(42)
You can also use the int constructor to convert other types, such as strings or other numerical types, to
integers:
>>> x = int(“17”)
>>> y = int(4.8)
>>> print x, y, x - y
17 4 13
In the first line, Python converts a string representing a number to the number itself; you can’t do math
with “17” (a string), but you can with 17 (an integer). In the second line, Python converted the floating-
point value 4.8 to the integer 4 by truncating it — chopping off the part after the decimal point to make
it an integer.
When you convert a string to an int, Python assumes the number is represented in base 10. You can
specify another base as the second argument. For instance, if you pass 16, the number is assumed to be
hexadecimal:
>>> hex_number = “a1”
>>> print int(hex_number, 16)
161
You can specify hexadecimal literals by prefixing the number with 0x. For example, hexadecimal 0xa1
is equivalent to decimal 161. Similarly, literals starting with just a 0 are assumed to be octal (base 8), so octal
0105 is equivalent to decimal 69. These conventions are used in many other programming languages, too.
Long Integers
What’s the largest number Python can store in an int? Python uses at least 32 bits to represent integers,
which means that you can store numbers at least as large as 231–1 and negative numbers as small as –231.
If you need to store a larger number, Python provides the long type, which represents arbitrarily large
integers.
For example, long before the search engine Google existed, mathematicians defined a googol, a one
followed by 100 zeros. To represent this number in Python, you could type out the hundred zeros, or
you can save yourself the trouble by using the exponentiation operator, **:
>>> googol = 10 ** 100
>>> print googol
406

45.8 Page 448

▲back to top
Numerical Programming
10000000000000000000000000000000000000000000000000000000000000000000000000000000000
000000000000000000
This is an example of a long object:
>>> type(googol)
<type ‘long’>
Note that when you computed the value of googol, you used only int literals — namely, 10 and 100.
Python converted the result to a long automatically because it didn’t fit in an int.
If you enter a literal that is too large for an int, Python uses a long automatically:
>>> type(12345678900)
<type ‘long’>
You can also construct a long object for a number that would fit in an int. Either call the long construc-
tor explicitly, as in long(42), or append an L to the literal, as in 42L.
Floating-point Numbers
In Python, a floating-point number is represented by a float object. A floating-point number is only an
approximation to a real number, so you may sometimes see results that look strange. For example:
>>> x = 1.1
>>> x
1.1000000000000001
>>> print x
1.1
What’s going on here? You assigned to x the floating-point approximation to the number 1.1. The floating-
point number that Python can represent that is closest to 1.1 is actually a tiny bit different, and Python is
honest with you and shows this number when you ask for the full representation of x. When you print
x, however, Python provides you with a “nice” depiction of the number, which doesn’t show enough
decimal places to illustrate the floating-point approximation.
Simply entering x at the command prompt prints what you would get by calling repr(x). Entering
print x prints what you would get by calling str(x).
Representation of int and long
Internally, Python uses the C type long to represent int objects. If you are using a
64-bit architecture, Python can represent numbers between –263 and 263–1 as int
objects. However, it’s best to assume that an int is only 32 bits, in case you later decide
to run your program on another architecture.
Use the Python long type for larger integers. For these, Python uses a internal represen-
tation that isn’t fixed in size, so there’s no limit. Be aware, however, that long objects
take up more memory than int objects, and computations involving them are much
slower than those using only int objects.
407

45.9 Page 449

▲back to top
Chapter 19
Floating-point Precision
A floating-point number is an approximation. As you have seen, it can carry only a
limited number of digits of precision.
Formally, Python does not make any promises about the number of digits of precision
retained in float variables. However, internally Python uses the C type double to store
the contents of float objects, so if you know the precision of a C double variable on a
platform, you’ll know the precision of a Python float when running on that platform.
Most systems store a double in 64 bits and provide about 16 digits of precision.
As with integers, you can use the float constructor to covert strings to numbers (but only in base 10).
For example:
>>> x = float(“16.4”)
Very large and very small floating-point numbers are represented with exponential notation, which
separates out the power of ten. A googol as a floating-point number would be 1e+100, which means the
number 1 times ten raised to the power 100. The U.S. national debt at the time this was written, accord-
ing to the Treasury Department web site, was:
>>> debt = 7784834892156.63
Python prefers exponential notation to print a number this large:
>>> print debt
7.78483489216e+012
You can also enter literals with exponential notation.
Formatting Numbers
You can convert any Python number to a string using the str constructor. This produces the text that
would be printed by the print statement, as a string object. For simple applications, this is adequate.
For better control of the output format, use Python’s built-in string formatting operator, %.
Note that this has nothing to do with the remainder operator. If you use % after a string, that’s the
string formatting operator. If you use % between two numbers, then you get the remainder operator.
Following are some details on formatting numbers. If you are familiar with the printf function in C,
you already know much of the syntax for formatting numbers in Python.
To format an integer (int or long), use the %d conversion in the format string. For a floating-point num-
ber, use %f. If you use %d with a floating-point number or %f with an integer, Python will convert the
number to the type indicated by the conversion. For example:
>>> print “%d” % 100
100
408

45.10 Page 450

▲back to top
Numerical Programming
>>> print “%d” % 101.6
101
You probably didn’t really notice, since it’s so obvious, that Python formatted these integers in base 10.
For some applications, you might prefer your output in hexadecimal. Use the %x conversion to produce
this. If you use %#x, Python puts 0x before the output to make it look just like a hexadecimal literal
value, like so:
>>> print “%#x” % 100
0x64
Similarly, %o (that’s the letter “o,” not a zero) produces output in octal, and %#o produces octal output
preceded by a 0.
For integers, you can specify the width (number of digits) of the output by placing a number after the %
in the format string. If the number starts with 0, the output will be left-padded with zeros; otherwise, it
will be padded with spaces. In the examples that follow, you surrounded the output with parentheses so
you can see exactly what Python generates for the %d conversions:
>>> print “z is (%6d)” % 175
z is ( 175)
>>> print “z is (%06d)” % 175
z is (000175)
When you format floating-point numbers, you can specify the total width of the output, and/or the
number of digits displayed after the decimal place. If you want the output to have total width w and to
display p decimal places, use the conversion %w.pf in the format string. The total width includes the
decimal point and digits after the decimal point. Unlike converting a float to an integer value, Python
rounds to the nearest digit in last decimal place:
>>> x = 20.0 / 3
>>> print “(%6.2f)” % x
( 6.67)
If you omit the number before the decimal point, Python uses as much room as necessary to print the
integer part and the decimal places you asked for:
>>> print “(%.4f)” % x
(6.6667)
You can demand as many digits as you want, but remember that a float carries a limited precision and,
therefore, contains information for only 16 digits or so. Python will add zero digits to fill out the rest:
>>> two_thirds = 2.0 / 3
>>> print “%.40f” % two_thirds
0.6666666666666666300000000000000000000000
The number you see may be slightly different, as architectures handle the details of floating-point com-
putations differently.
409

46 Pages 451-460

▲back to top

46.1 Page 451

▲back to top
Chapter 19
If you omit the number after the decimal point (or specify zero decimal places), Python doesn’t show
any decimal places and omits the decimal point, too:
>>> print “(%4.f)” % x
( 7)
For example, the following function formats the ratio of its arguments, num and den, as a percentage,
showing one digit after the decimal point:
>>> def as_percent(num, den):
...
if den == 0:
...
ratio = 0
...
else:
...
ratio = float(num) / den
...
return “%5.1f%%” % (100 * ratio)
...
>>> print “ratio = “ + as_percent(6839, 13895)
ratio = 49.2%
One nice thing about this function is that it confirms that the denominator is not zero, to avoid division-
by-zero errors. Moreover, look closely at the format string. The first % goes with the f as part of the
floating-point conversion. The %% at the end is converted to a single % in the output: Because the percent
symbol is used to indicate a conversion, Python requires you to use two of them in a format string if you
want one in your output.
You don’t have to hard-code the width or number of decimal places in the format string. If you use an
asterisk instead of a number in the conversion, Python takes the value from an extra integer argument in
the argument tuple (positioned before the number that’s being formatted). Using this feature, you can
write a function that formats U.S. dollars. Its arguments are an amount of money and the number of
digits to use for the dollars part, not including the two digits for cents:
>>> def format_dollars(dollars, places):
...
return “$%*.2f” % (places + 3, dollars)
...
>>> print format_dollars(499.98, 5)
$ 499.98
In the format string, you use * instead of the total width in the floating-point conversion. Python looks
at the argument tuple and uses the first value as the total width of the conversion. In this case, you spec-
ify three more than the desired number of digits for dollars, to leave room for the decimal point and the
two digits for cents.
Even more options are available for controlling the output of numbers with the string formatting opera-
tor. Consult the Python documentation for details, under the section on sequence types (because strings
are sequences) in the Python Library Reference.
Characters as Numbers
What about characters? C and C++ programmers are used to manipulating characters as numbers, as C’s
char type is just another integer numeric type. Python doesn’t work like this, though. In Python, a char-
acter is just a string of length one, and cannot be used as a number.
410

46.2 Page 452

▲back to top
Numerical Programming
Occasionally, you might need to convert between characters and their numeric values. Python provides
the built-in function ord to convert a single character to its numeric code and the function asc to con-
vert back from a numeric code to a character. The numeric code must be between 0 and 255.
Strictly speaking, this code is not ASCII, as ASCII only goes up to 127. However, the first 127 values
converted by ord and asc are ASCII code values.
If you are a Usenet regular, you are probably familiar with the rot13 cipher. It’s not particularly secure;
all it does is rotate letters of the alphabet 13 positions forward, wrapping around from “z” to “a”. Using
chr and ord functions, it’s not hard to implement in Python:
def rot13_character(character):
# Look up codes for ends of the alphabet.
a = ord(‘a’)
z = ord(‘z’)
A = ord(‘A’)
Z = ord(‘Z’)
code = ord(character)
# Rotate lower-case characters.
if a <= code <= z:
code = a + (code - a + 13) % 26
# Rotate upper-case characters.
elif A <= code <= Z:
code = A + (code - A + 13) % 26
# Leave other characters alone.
else:
pass
return chr(code)
def rot13(plaintext):
# Loop over letters in the text.
ciphertext = “”
for character in plaintext:
ciphertext += rot13_character(character)
return ciphertext
The program is composed of two functions. The first, rot13_character, applies rot13 to a single char-
acter. If it’s an uppercase or lowercase letter, it is rotated 13 places; otherwise, it is left alone. (In case you
are not familiar with the remainder operator, %, it is described in the next section.) The main function,
rot13, takes the message to be coded (the “plaintext”) and creates the encoded message (the “cipher-
text”) by rotating each letter.
Save the preceding code into a module file named rot13.py. In Python, import the module, and try
it out:
>>> import rot13
>>> message = rot13.rot13(“This is a TOP-SECRET encoded message.”)
>>> print message
Guvf vf n GBC-FRPERG rapbqrq zrffntr.
411

46.3 Page 453

▲back to top
Chapter 19
Rot13 has the nice property that it is its own inverse: To decode a rot13-encoded message, you just apply
rot13 to it again:
>>> print rot13.rot13(message)
This is a TOP-SECRET encoded message.
Mathematics
In addition to the usual complement of arithmetic operations, Python includes some handy built-in
math functions, and a math module that provides other commonly used functions. Coverage of arith-
metic operators may seem obvious, but you should also understand some subtle points about how
Python handles certain numeric types.
Arithmetic
Python provides the normal arithmetic operators + (addition), _ (subtraction), * (multiplication), and /
(division) for numerical types. You can mix numerical types when using these operators; Python auto-
matically chooses the more flexible type for the result:
>>> i = 10
>>> f = 6.54
>>> print i + f
16.54
When adding an integer, i, and a floating-point number f, Python chose a float for the result.
These operators all have special forms for updating the values of variables, written by adding an equals
sign right after the operator. Instead of writing
>>> total = total + 6
>>> coefficient = coefficient / 2
you can simply write
>>> total += 6
>>> coefficient /= 2
and so forth.
Be careful when dividing two integers (whether they are int or long objects). Python always uses an
integer type for the result, even if the “correct” result is fractional. Don’t get caught by computations
like this one:
>>> print 10 / 3
3
412

46.4 Page 454

▲back to top
Numerical Programming
Oops! The “correct” answer is three and one third, or 3.3333, but Python uses an integer for the result,
rounding down to three. This doesn’t work either:
>>> quarter = float(1 / 4)
>>> print quarter
0.0
What you’ve asked Python to do here is to divide two integers, and then to convert the result to a float.
What you really want is division of the numbers as floating-point values. Adding a decimal point to either
number does the trick: That makes the number a float literal, so Python uses float for the quotient:
>>> quarter = 1.0 / 4
>>> print quarter
0.25
Similarly, if you are writing a function that takes the average of two numbers, you probably want the
result as a float even if both arguments are integers. Use the float constructor to do the conversion:
>>> def average(x, y):
...
return float(x + y) / 2
...
The exponentiation operator ** is used previously in examples. It, too, works for integer and floating-
point values. The function that follows uses it to compute compounded interest. The function returns
the amount of money you would have if you put starting_balance in a bank account with APR
annual_rate and waited for years:
>>> def compound(starting_balance, annual_rate, years):
...
return starting_balance * ((1 + annual_rate) ** years)
...
Ten grand in the bank at 4 percent APR for a century yields:
>>> print compound(10000, 0.04, 100)
505049.481843
That’s half a million bucks. Start saving now.
Floor Division
Python provides another division operator, called floor division, which explicitly
rounds down the quotient to an integer, like the default division behavior for int and
long. Floor division is represented by //. You can use it with float objects as well: for
instance, 6.6//3.0 evaluates to 2.0.
When you divide two integers, it’s a good idea to use floor division when that’s what
you mean, or explicitly to convert one of the arguments to a float otherwise. This will
prevent you from getting burned by surprising division results.
413

46.5 Page 455

▲back to top
Chapter 19
Also useful is the remainder operator %. It’s like floor division, but instead of returning the quotient, it
returns the remainder. Using it, you can format a number of months into whole years and remaining
months:
>>> def format_months(months):
...
print “%d years, %d months” % (months // 12, months % 12)
...
>>> format_months(100)
8 years, 4 months
Built-in Math Functions
A few very common mathematical functions are available as built-in functions. The simplest is abs,
which returns the absolute value of a number. The number that abs returns is the same type as the
number you pass it:
>>> print abs(-6.5)
6.5
Also useful are min and max, which return the smallest or largest of several values. You can call them
either with several numeric arguments or with a single argument that is a sequence (such as a list or
tuple) of numbers. The values needn’t all be the same type:
>>> print min(6, 7, 2, 8, 5)
2
>>> print max([0, 43.5, 19, 5, -6])
43.5
The round function rounds a floating-point value to a specified number of digits. This is similar to the
behavior you saw before in the %f conversions, except the result is not a string but rather another floating-
point number with which you can perform further computations. Specify the number to round, and the
number of decimal places you want to keep:
>>> print round(1234.56789, 2)
1234.57
You can even specify a negative number of decimal places, which rounds to that multiple of 10:
>>> print round(1234.56789, -2)
1200.0
Lastly, the sum function adds numbers in a sequence. Together with range, you can compute the sum of
the first 100 positive integers:
>>> print sum(range(1, 101))
5050
Suppose in your Python programming class you got a 96 percent and 90 percent on the two homework
assignments, a perfect score on the final project, and an 88 percent on the final exam. What’s your
414

46.6 Page 456

▲back to top
Numerical Programming
average for the class? Of course, you would write a Python function to compute it. The function uses
sum and computes the mean, or average, value of a sequence of numbers:
>>> def mean(numbers):
...
if numbers:
...
return float(sum(numbers)) / len(numbers)
...
else:
...
raise ValueError, “no numbers specified”
...
>>> print mean([96, 90, 100, 88])
93.5
It’s a good idea to make sure that the sequence of numbers isn’t empty, to avoid dividing by zero. In this
case, the function raises an exception if the sequence is empty.
The math Module
The math module contains the standard transcendental functions listed here. All these functions take
float arguments and return float values:
❑ square root: sqrt
❑ exponentiation: exp
❑ logarithms: log (natural logarithm), log10 (base 10 logarithm)
❑ trigonometric functions: sin, cos, and tan; arguments are in radians
❑ inverse trigonometric functions: asin, acos, and atan; results are in radians
❑ hyperbolic functions: sinh, cosh, and tanh
A few other useful math functions are included:
❑ hypot(x, y) is equivalent to sqrt(x ** 2 + y ** 2)
❑ atan2(x, y) is like atan(x / y) but gets the quadrant right and handles a zero denominator
❑ floor and ceil are the standard floor and ceiling functions; their results are integers but repre-
sented as float values
The math package also contains the constants pi and e.
Here’s some sample code that uses the math module. It will give you flashbacks to your freshman
physics class. It’s a function that computes the time of flight and range of a projectile launched into the
air (such as a cannonball), neglecting friction. Examine it at least long enough to understand how the
Python code works. Pay attention to how sin, cos, and pi are imported from math, which saves you
from having to refer to them as math.sin and so on. It’s a handy technique for commonly used func-
tions. Note also how carefully the units used in the arguments and results are documented. Many failed
rocket launches attest to the importance of this practice.
415

46.7 Page 457

▲back to top
Chapter 19
from math import sin, cos, pi
def trajectory(velocity, angle):
“””Compute time of flight and range of a projectile.
For a projectile with initial ‘velocity’ in meters/sec launched at
‘angle’ from horizontal in degrees, returns time of flight in sec
and range in meters, neglecting friction.”””
# Gravitational acceleration in meters/sec^2.
g = 9.8
# Convert ‘angle’ to radians.
angle = angle * pi / 180
# Compute horizontal and vertical components of velocity.
v_h = velocity * cos(angle)
v_v = velocity * sin(angle)
# Compute the time of flight and range.
tof = 2 * v_v / g
range = tof * v_h
return tof, range
Suppose you throw a ball into the air at 40 m/sec (about 90 mph) at a 45° angle. How long will it stay in
the air, and how far away will it land? Save the preceding code into a file named ballistic.py, and
then call the function like this:
>>> from ballistic import trajectory
>>> tof, range = trajectory(40, 45)
>>> print “time of flight: %.1f sec, range: %.0f meters” % (tof, range)
time of flight: 5.8 sec, range: 163 meters
Complex Numbers
A complex number is the sum of a real number and an imaginary number. In case you need a refresher,
an imaginary number is a multiple of the imaginary unit, which is the square root of –1. Mathematicians
(and math teachers) usually use the symbol i for the imaginary unit, while engineers often use j.
In Python, an imaginary number is written as a number followed by j (with no intervening spaces):
>>> imaginary_number = 16j
To create a complex number, add (or take the difference of) a real number and an imaginary number:
>>> complex_number = 6 + 4j
Python stores the complex number as a single object, whose type is complex:
>>> print complex_number
(6+4j)
>>> print type(complex_number)
<type ‘complex’>
416

46.8 Page 458

▲back to top
Numerical Programming
If you prefer, you can use the complex constructor to construct complex number objects. This assign-
ment is equivalent to the preceding one:
>>> complex_number = complex(6, 4)
Let’s make sure that 1j is really the imaginary unit:
>>> print 1j ** 2
(-1+0j)
This verifies that j2 is in fact –1, and also demonstrates that the result of an arithmetic operation involv-
ing complex values is itself a complex, even if the result happens to be a real number (that is, has a zero
imaginary part).
You can’t write j by itself to represent the imaginary unit. You must write 1j. By itself, j represents
the variable named “j.”
Both the real and imaginary parts of a complex object are stored as floating-point values, even if you
specified them as integers. Remember that 1/3 in Python returns zero? Not so for complex numbers:
>>> print (1+0j)/3
(0.333333333333+0j)
Arithmetic works for complex numbers as you would expect, and you can mix int, long, float, and
complex in the same expression:
>>> print 2 * (10 + 3j) * (6.5 - 4j) / (1 - 1j) + 30L
(127.5+56.5j)
A few other operations round out Python’s handling of complex numbers. First, the mathematical
operations Re and Im return the real and imaginary parts of a complex number, respectively. These are
provided in Python by attributes named real and imag that every complex object has. The value of
each is a float:
>>> x = 5 - 6j
>>> print x.real
5.0
>>> print x.imag
-6.0
You saw before that the built-in abs function returns the absolute value of an int, long, or double object.
For complex numbers, it returns the magnitude, which is the square root of the sum of the squares of the
real and imaginary parts. You can verify this by using the hypot function discussed previously:
>>> print abs(x)
7.81024967591
>>> import math
>>> print math.hypot(x.real, x.imag)
7.81024967591
417

46.9 Page 459

▲back to top
Chapter 19
Precision of Complex Numbers
Let’s verify the famous and very fundamental mathematical identity eiπ + 1 = 0:
>>> print cmath.exp(1j * cmath.pi) + 1
1.22460635382e-016j
What’s this? It’s a complex number with real part of zero and imaginary part approxi-
mately 1.225 × 10_16. That’s close, but not quite equal to zero.
Python stores both the real part and the complex part with the same precision as a
float value, about 16 digits on most systems. That means Python’s representation of
eiπ is equal to –1 only to about 16 digits. Therefore, you shouldn’t be surprised if the
result after adding +1 is off by about 10–16.
Finally, every complex object has a method conjugate, which returns the complex conjugate. This is
the complex number with the same real part and negated imaginary part. Keep in mind that while real
and imag are attributes (you don’t call them), conjugate is a method (you must call it):
>>> print x.conjugate()
(5+6j)
The transcendental functions in the math package work only on and return float values. For instance,
you can’t actually take the square root of –1 to obtain 1j:
>>> print math.sqrt(-1)
Traceback (most recent call last):
File “<interactive input>”, line 1, in ?
ValueError: math domain error
That’s a shame, because square roots and most of the other functions in math can be defined on complex
numbers, too. Fortunately, Python provides a parallel module named cmath, which contains versions of
the same functions that operate on and return complex objects. Its version of the square root function
can handle –1:
>>> import cmath
>>> print cmath.sqrt(-1)
1j
Arrays
You’ve learned how to perform computations with individual numbers, be they integers, floating-point
numbers, or even complex numbers. What if you want to perform computations on many numbers? A
group of numbers is typically arranged into an array. In this section, you will learn different ways of
implementing arrays in Python.
Keep in mind that arrays may be multidimensional. If you arrange numbers in a row, you have a one-
dimensional array. A vector in linear algebra is an example; another is a list of daily closing prices of your
418

46.10 Page 460

▲back to top
Numerical Programming
favorite stock. You can also arrange your numbers on a rectangular grid, to produce a two-dimensional
array. A grayscale image is often represented as a two-dimensional array, where each value is the lightness
of one pixel in the image. In some applications, you may want to arrange your numbers into higher-
dimensional arrays as well.
You’ve already seen one technique for constructing arrays in Python, when you wrote the mean function
earlier. That function takes a sequence of numbers (of arbitrary length) and computes the numbers’
mean. You can think of this sequence of numbers as an array and can think of mean as a function that
acts on an array. You can invoke the function with a list of numbers, but it works with any sequence
type, including tuples. These built-in types are the simplest way of building arrays.
Let’s take another example of a function that operates on an array. You already wrote a function that
computes the mean of an array of numbers. Now write a function that computes the standard deviation.
To remind you, the standard deviation is an indication of how much the numbers vary among them-
selves. If they’re all almost the same, the standard deviation will be small, whereas if they are all over
the place, the standard deviation will be large. The formula for the standard deviation that you will use
is shown as follows:
! v =
1
N
N
i=1
x
2
i
-
n2
Here x1,..., xN are the numbers in the array, µ is their mean, and N is the length of the array.
You could implement a standard deviation function several different ways. Here’s one of them:
from math import sqrt
def stddev(numbers):
n = len(numbers)
sum = 0
sum_of_squares = 0
for number in numbers:
sum += number
sum_of_squares += number * number
return sqrt(sum_of_squares / n - (sum / n) ** 2)
This function loops over the numbers to compute their sum of squares. Simultaneously, it computes
their sum, as it needs that to compute the mean. The last line computes the standard deviation according
to the preceding formula. You might have noticed that the function uses number * number when com-
puting the sum of squares instead of number ** 2; that’s because squaring a number by multiplying it
by itself is faster than using the general exponentiation operator.
Lists or Tuples?
Which should you use for arrays: lists or tuples? Remember that lists can be modified,
whereas tuples cannot. Therefore, if you need to add to, remove from, or change the
array, use a list. While you can perform these operations on a tuple by creating a new
tuple with numbers added, removed, or changed, this is more difficult to code and
often runs more slowly. For fixed sequences of numbers, you can use tuples.
419

47 Pages 461-470

▲back to top

47.1 Page 461

▲back to top
Chapter 19
Watch stddev in action. Remember that it takes one argument, a sequence of numbers (not several
numerical arguments):
>>> print stddev((5.6, 3.2, -1.0, 0.7))
2.50137462208
Think for a moment about some advantages and drawbacks of using lists of numbers for arrays:
❑ The elements of a Python list need not be of the same type. You can create a list for which some
elements are int, float, long, and double, or other objects like strings or even other sequences.
For some applications, this is very handy. For instance, you may want to store None in a sequence
to indicate that a value is not known. For other applications, it’s important to make sure that all
of the values in an array are of the same type. In that case, you’ll have to write extra code to
ensure this.
❑ Lists are single-dimensional, which makes them natural for expressing one-dimensional arrays.
You can create two-dimensional arrays as lists of lists and higher-dimensional arrays analogously,
but this can get complicated.
❑ Lists are a standard part of Python. They’re always available (you don’t even have to import a
module), and you already know how to use them.
❑ Lists can be pickled. That makes it easy to store your list in a file for later use.
❑ Internally, Python represents each element in a list as a separate object. Therefore, if you have a
list of a million numbers (not at all unusual in many fields), you force Python to keep track of
1,000,001 objects: the list itself and all of its elements. This both wastes a lot of memory and
makes Python work pretty hard whenever you access or modify the array.
This last point is a major limitation in many types of numerical work. To address it, you can use one of
two other array implementations that store numbers more efficiently.
The array Module
The Python standard library has just the ticket: a module array for one-dimensional arrays of numbers.
The array type in this module stores numbers all together in memory as one object, subject to the con-
straint that all of them must be of the same type. The numerical types supported by array are not the
same as Python’s numeric types. (In fact, they correspond to the numerical types in the C language.) An
array can store numbers equivalent to Python’s int and float, as well as integers of other sizes, and
floating-point numbers of other precisions. (An array can store long values, but not arbitrarily large
ones, and cannot store complex values at all.)
When you create an array, you have to specify the numerical type to store in the array. The type is speci-
fied by a single character. To store numbers as Python int objects, use “l”; for float use “d”. (There
are other options available; see the documentation for the array module for a list of them.) If you don’t
specify anything else, you’ll get an empty array:
>>> import array
>>> a = array.array(“l”)
>>> print a
array(‘l’)
420

47.2 Page 462

▲back to top
Numerical Programming
Generally, you can use an array object just as you would an ordinary list. You can insert, append, or
delete elements, and the indexing syntax is the same. (Note that in versions of Python earlier than 2.4,
an array object is somewhat more limited than a list object.) For example:
>>> a.append(15)
>>> a.extend([20, 17, 0])
>>> print a
array(‘l’, [15, 20, 17, 0])
>>> a[1] = 42
>>> print a
array(‘l’, [15, 42, 17, 0])
>>> del a[2]
>>> print a
array(‘l’, [15, 42, 0])
You can also convert a list or tuple to an array object by passing it to the constructor:
>>> t = (5.6, 3.2, -1.0, 0.7)
>>> a = array.array(“d”, t)
>>> print a
array(‘d’, [5.5999999999999996, 3.2000000000000002, -1.0, 0.69999999999999996])
Here again you see the approximate nature of floating-point values.
In fact, because an array object behaves very much like a list, you can pass it to the same stddev func-
tion you wrote previously, and it works just fine:
>>> print stddev(a)
2.50137462208
If you ever need to convert back to an ordinary tuple or list, just past the array to the tuple or list
constructor:
>>> back_again = tuple(a)
>>> print back_again
(5.5999999999999996, 3.2000000000000002, -1.0, 0.69999999999999996)
Compared to lists, array objects have the following advantages and disadvantages:
❑ All elements of an array are the same type.
❑ Like a list, an array is one-dimensional.
❑ The array module is part of Python’s standard library (but don’t forget to import it).
❑ An array object cannot automatically be pickled.
❑ An array object stores its values much more efficiently than a list of numbers does. However,
computations on the numbers are generally not much faster, as computations are performed
using Python’s normal number objects.
421

47.3 Page 463

▲back to top
Chapter 19
The numarray Package
This chapter ends with a brief look at one more array package, numarray. The numarray package is
much more sophisticated than the array module, and supports multidimensional arrays and operations
on entire arrays. If you are familiar with an array manipulation package such as Matlab, you will recog-
nize many of the features of numarray.
Unfortunately, numarray is not part of Python’s standard library, so you must install it yourself.
Fortunately, it is also free software, and easy to download and install. If you work with arrays of num-
bers and would like to use Python for this, it’s definitely worth the trouble of installing numarray
because of the rich set of features it provides.
The web page for numarray is at www.stsci.edu/resources/software_hardware/numarray.
You can browse or download the documentation, which is quite extensive. The Download link takes you to
the package’s SourceForge page, from which you can download the full source code, including installation
instructions. If you are using Windows, your life is easier: You can download an installer from the same
place. If you are using GNU/Linux, you can download RPM packages from www.python.org/pyvault/.
After you have installed numarray correctly, you should be able to import it:
>>> import numarray
To explain all the features of numarray would fill an entire book by itself. Instead, here is a brief tour,
which demonstrates how to write the stddev function using numarray. It is hoped that this will whet
your appetite to learn more about the package on your own.
Using Arrays
The array type in numarray is also called array. You can convert a tuple or list to an array object by
passing it to the constructor:
>>> a = numarray.array([5.6, 3.2, -1.0, 0.7])
>>> print a
[ 5.6 3.2 -1. 0.7]
Notice that when it printed the array, numarray omitted the commas between elements, which produces
output similar to the notation used in mathematics. The elements of an array object must all be of the
same type, just as in the array module, but numarray guesses the type from the elements you give it.
You can ask to see what type it chose with the type method:
>>> print a.type()
Float64
The “Float64” type stores a 64-bit floating-point value, suitable for Python’s float objects.
A key feature of numarray is that you can perform operations on entire arrays. For example, to double
all of the values in an array, just multiply by two, like so:
>>> print 2 * a
[ 11.2 6.4 -2.
1.4]
422

47.4 Page 464

▲back to top
Numerical Programming
You can perform operations on two arrays, too. The operation is performed elementwise — that is, on
pairs of corresponding elements from the two arrays:
>>> b = numarray.array([0.5, 0.0, -1.0, 2.0])
>>> print 2 * a + b
[ 11.7 6.4 -3. 3.4]
You can also create multidimensional arrays by passing to the constructor lists of lists, lists of lists of
lists, and so on. A two-dimensional, three-by-three array looks like this:
>>> m = numarray.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
>>> print m
[[1 2 3]
[4 5 6]
[7 8 9]]
Observe how each of the inner lists became a row in the array. Here, because all of the numbers in the
array are integers, numarray uses an integer type:
>>> print m.type()
Int32
You can index a one-dimensional array just as you would a list. To index a higher-dimensional array,
separate the indices by commas:
>>> print m[0,1]
2
>>> m[1,2] = 12
>>> print m
[[ 1 2 3]
[ 4 5 12]
[ 7 8 9]]
The shape of an array is a tuple containing the number of elements along each dimension. A one-dimensional
array has a shape of length one, a two-dimensional array has a shape of length two, and so on. Use the
shape attribute to obtain this:
>>> print a.shape
(4,)
>>> print m.shape
(3, 3)
Computing the Standard Deviation
Now write stddev using numarray:
from math import sqrt
import numarray
def stddev2(numbers):
n, = numbers.shape
sum = numarray.sum(numbers)
sum_of_squares = numarray.sum(numbers * numbers)
return sqrt(sum_of_squares / n - (sum / n) ** 2)
423

47.5 Page 465

▲back to top
Chapter 19
Here is a line-by-line description of how this function works:
1. The first line extracts the length of the array. We could have instead written n = len(numbers),
but we chose to exploit the fact that the shape of a one-dimensional array is a tuple whose only
element is the length of the array.
2. The second line uses the sum function in numarray, which adds numbers in the array.
3. The third line uses the sum function again, but this time on numbers * numbers. What is this
object? It’s the elementwise product of the array with itself, which, if you think about it, is an
array whose elements are the squares of the elements of numbers. Calling sum on this array
computes the sum of squares.
4. The last line computes the standard deviation as before.
Make sure it produces the same result:
>>> print stddev2(a)
2.50137462208
Why should you bother installing, learning, and using numarray?
❑ As with the array module, its arrays are stored efficiently. This matters a lot if you are working
with very large data sets.
❑ It supports multidimensional arrays.
❑ An array object from numarray can be pickled. (Make sure the numarray package is available
when unpickling.)
❑ Notice that our numarray version of stddev2 contains no explicit operations on the individual
elements of the array. Computations are performed by the sum function and the multiplication
operator * acting on arrays. As a result, numarray performs the computations on individual ele-
ments internally, without resorting to arithmetic using Python objects. For large arrays, this exe-
cutes very much faster. (For example, on one of the authors’ computers, stddev2 runs about 20
times faster than stddev for an array of a million numbers.)
❑ In addition, the code for stddev2 is somewhat shorter than the code for stddev, because the
former doesn’t contain an explicit loop. Computations on array elements can usually be
expressed more simply using the functions and methods in numarray — in some cases, very
much more simply.
Summary
In this chapter, you learned how to perform many kinds of numerical computations in Python. You
experimented first with Python’s built-in integer and floating point number types and saw how to use
Python’s built-in arithmetic operations. Then you moved on to higher mathematics, using the special
functions in the math module and Python’s complex number type.
Finally, you learned three different ways of representing arrays of numbers: The simplest method is to
use a list or tuple of numbers. For more efficient storage, use the array module included with Python.
For greatest flexibility and a host of sophisticated features for programming with arrays, download and
install the numarray package.
424

47.6 Page 466

▲back to top
Numerical Programming
Exercises
1. Write a function that expresses a number of bytes as the sum of gigabytes, megabytes, kilobytes,
and bytes. Remember that a kilobyte is 1024 bytes, a megabyte is 1024 kilobytes, and so on. The
number of each should not exceed 1023. The output should look something like this:
>>> print format_bytes(9876543210)
9 GB + 203 MB + 5 KB + 746 bytes
2. Write a function that formats an RGB color in the color syntax of HTML. The function should
take three numerical arguments, the red, green, and blue color components, each between zero
and one. The output is a string of the form #RRGGBB, where RR is the red component as a value
between 0 and 255, expressed as a two-digit hexadecimal number, and GG and BB likewise for
the green and blue components.
For example:
>>> print rgb_to_html(0.0, 0.0, 0.0)
#000000
>>> print rgb_to_html(1.0, 1.0, 1.0)
#ffffff
>>> print rgb_to_html(0.8, 0.5, 0.9)
#cc80e6
# black
# white
# purple
3. Write a function named normalize that takes an array of float numbers and returns a copy of
the array in which the elements have been scaled such that the square root of the sum of their
squares is one. This is an important operation in linear algebra and other fields.
Here’s a test case:
>>> for n in normalize((2.2, 5.6, 4.3, 3.0, 0.5)):
... print “%.5f” % n,
...
0.27513 0.70033 0.53775 0.37518 0.06253
For bonus geek credit, implement it using numarray.
425

47.7 Page 467

▲back to top

47.8 Page 468

▲back to top
20
Python in the Enterprise
Enterprise applications are software that address the needs of a company or other organization
that has a need for a portion of the business to be run through that software. Of course, this defini-
tion could encompass nearly any kind of software, but the software usually thought of as enter-
prise business software is that which supports the modeling of business processes. Enterprise
software relies on infrastructure platforms — the flexible components that applications are built to
rely on, such as relational databases, third-party libraries, high-availability suites, and more.
A business process is any repeatable way for an organization to accomplish a task that is part of its
business or that supports its business. Business processes usually start as undocumented, ad hoc
actions that have grown out of trial-and-error and at some point have become essential to getting
work done; sometimes they exist only in the heads of particular individuals who simply know
that, for instance, “the ABC forms are always sent to Mr. Kofax in HR on Thursday afternoons.”
This kind of undocumented process is dangerous when an organization has grown, because if it is
lost for any reason (due to a long vacation, for instance, or a physical disaster), the business will
suffer. This loss will involve associated costs due to missed deadlines, confusion, and disorganiza-
tion when employees leave the company or simply move on to new duties, and their replacements
are forced to resolve problems reactively instead of being able to rely on documented processes to
get work done. In addition, brand-new regulatory requirements for business auditing may even
make undocumented processes illegal in some cases. Therefore, it’s not at all surprising that an
entire industry is devoted to the understanding, documentation, and formalization of business
processes.
There are benefits beyond legal compliance. After a process has been documented, it becomes
amenable to being supported by software and to being modeled using relatively standard plat-
forms. Of course, several legacy applications support business processes — the larger the company,
the more elaborate and often older they are. These legacy applications include standard infrastruc-
ture, such as relational database systems and e-mail systems, and custom software ranging from
simple periodic processing scripts and standardized batch jobs, all the way through vast online
architectures and bullet-proofed enterprise platforms.

47.9 Page 469

▲back to top
Chapter 20
This chapter will show you the following:
❑ Some of the main types of enterprise applications and how they can be useful, not only to the
Fortune-500 CEO, but to the sole proprietor, and to any organization in between
❑ Some of the regulatory frameworks out there that are making people think harder about their
business processes lately
❑ How you can use Python and the open-source workflow toolkit wftk, either to talk to any exist-
ing enterprise platform already available or simply to roll your own if you’re on a tight budget
but still need to get organized
You’ll also be introduced to a few fairly simple applications of wftk and Python, which you can easily
use as a starting point for more elaborate enterprise architecture.
Enterprise Applications
You now know (if you didn’t already) that enterprise applications take advantage of an infrastructure
for modeling the organization and its activities. In general, things like organization charts, reporting
trees, and activities that are shown as flowcharts are the familiar forms of this kind of modeling. When
they’re being created, these models involve three main types of objects, which are then reflected in the
infrastructure involved:
❑ Documents and data
❑ People, their authority to take action, and their groups (organizational structure)
❑ Actions to be taken
Keep these in mind while you take a closer look at the subdivisions of enterprise infrastructure systems.
Document Management
A document is basically anything you can think of as being printed on a piece of paper. It might be a let-
ter or a fax; it could be a 100-page Word document or an Excel spreadsheet; it might be a file of program-
ming code; or it could be a simple e-mail message.
The default way to handle documents when you’re actually producing them is by storing them in files in
your local computer or on a shared network folder, but there are problems with this approach:
❑ The computers’ file systems are often on local machines or have different access paths from dif-
ferent areas of an organization (so a shared file in New York could be in \\\\local_server\\file,
but in Seattle it is in \\\\NYC\\public\\local_server\\file). This creates a barrier to sharing
files stored in them because each office, or even each department within a company, needs to
know various, different, and specific information that they can’t usefully share with others in the
organization.
❑ The name of a file in a file system is arbitrary. If some system exists only in a person’s head, then
files could be named something like FaxFromBob20050105.tiff — and this might be easy to
figure out. But what if you just clicked Save and the file automatically received the name
428

47.10 Page 470

▲back to top
Python in the Enterprise
7EF78S.tiff? You’ll almost certainly lose track of it; and, of course, anybody else who needs
your files doesn’t stand a chance (think about what it would mean if this fax had great legal sig-
nificance or was confirmation of a multi-million-dollar order!)
❑ A file system has no way to store extra information about a document, such as comments, date
of origination, process status (“needs handling” or “old business”, for instance), which means
that it also can’t be searched for documents based on this desirable information. Nor is there a
way to place it in its context; given the document, why do you have it? What is it for? Is it part
of an ongoing customer complaint? A tax audit? Speaking of audits, do you know which of your
files an auditor needs to see? Are you sure?
❑ If the document is something you’re working on, or is something that changes (think of your
organization’s employee manual, or the current list of catering service vendors — anything that
can change over time), there is no way to tell when it changed, why it changed, whether it
changed, or how it changed. You can’t tell what it might have looked like last November when
you had that meeting about it, and you can’t tell why or when somebody added or removed
content from it.
All of these business-level problems with simply storing files on file systems are reasons for you to get
interested in thinking about document management, which is really a tautology. If you work on documents
at all, you’re doing document management just by naming, writing, changing, and saving them. Even if
you’re writing programs, source code is simply another type of document.
Therefore, the question is really this: How well are you doing document management? As soon as you’re
talking about any organization of more than one person, the complexity and difficulty of managing doc-
uments well increases, multiplying many times for each person you add to the mix.
The Evolution of Document Management Systems
Documents aren’t the only data managed on an enterprise level. You also need to provide care and feed-
ing of the infrastructure components. One of the first types of enterprise infrastructure software was the
database; and the modern databases, relational databases, are a type of software that nearly all program-
mers have dealt with (and so have you if you’ve followed through these chapters in order, as Chapter 14
introduces you to databases). By standardizing the storage of record-based data, the advent of database
management systems changed the way software was designed, by making it possible to concentrate on
the specific ways in which data could be manipulated, related, and reported, while leaving the nitty-
gritty of actually storing it to the database software.
Relational databases originally hit the market in the late 70s; and during the 80s, document management
software that worked with this class of software also made its debut. Many early document management
systems were basically standalone software. Sometimes you’ll see the acronym EDMS; the “E” stands
for “electronic,” because the marketing strategy at the time was to replace paper filing systems with
technology solutions that were purely electronic filing systems. As such, the design of these systems
often reflected the assumption that document management would be the motivation for a given organi-
zation to buy computers at all! It wasn’t really a consideration that the organization had any separate
interest in databases or their management.
By the early 90s, computers had gone through the global workplace like wildfire; and at the same time,
all document management systems had moved to the model of utilizing the relational database for their
storage of data about documents and leveraging their ability to store data, index it, and retrieve it easily.
429

48 Pages 471-480

▲back to top

48.1 Page 471

▲back to top
Chapter 20
Document management systems usually use the term metadata to refer to data about
documents — that is, information specifying things such as “who wrote it,” “when
was it created,” “who modified it last,” “who authorized it,” and so on, as they con-
sider the “data” they manage to be the documents themselves.
What You Want in a Document Management System
In short, a document management system can be thought of as a special case of a relational database sys-
tem. This is because a document management system stores information about the document in one or
more database records, while the content of the document will be attached to the record and can usually
be retrieved. This much functionality can easily be implemented in, say, Oracle, by using Oracle’s stan-
dard BLOB or CLOB mechanism for storing large amounts of binary data (BLOB means Binary Large
OBject) or character data (CLOB means Character Large OBject). If you store your documents in an
Oracle database (or any RDBMS with LOB support — that is, Large Object support), you already have
some advantages over the file system. What you gain by doing this is that you don’t need to worry as
much about naming your files and your metadata (there’ll be a structure in the database) or the capabil-
ity to store extra data about your documents (the database can store metadata in a column alongside the
file), and your storage location is global (all enterprise databases can be connected to remotely), search-
able, and, from the point of view of an organization that can afford to ensure that someone with the right
expertise is on staff, easy to manage.
Of course, you still don’t have that list of versions of each document or a way to keep track of who is
currently working on a new version. These features are the final basic requirements for realizing true
document management as it’s practiced today.
Maintaining the version list of a document is pretty straightforward (and if you use CVS, Subversion, or
some other version control system for your programming work, you already know how this can work).
The principle is simple: When a user wants to work on a document, she registers herself as the user who
is working on a new version. When this has been done, the user is referred to as having checked out the
document, which indicates that she can work on it; and when she’s done, she performs a corresponding
check in action when the new version is no longer being edited. The combination of the checkout and
checkin enables the management system to know what version the document started at, and if other
people have also, in parallel, checked out, edited, and checked in the same document, both users can be
informed of the conflict and they can resolve it. There may be various draft versions of a document, with
the currently valid production version being flagged in some way. In addition, there are often more or
less elaborate schemes for avoiding or settling “conflicts” that result when more than one person is try-
ing to work on the document, such as branching of versions, merging variants, and so on.
The key functionality inherent in document management is simply this: A document management sys-
tem provides an organization within one unified system to track all of an organization’s documents,
whether they’re computer-generated text files such as program source code, more complex textlike doc-
uments such as Word documents, scanned images from paper, or even just information about actual
paper stored in a box somewhere. Once it’s actively used, that unified system can tell you the current
status of any document, including when it was created, who created it, and when and why and can
store whatever other information may be pertinent.
Because a document management system knows everything that an organization considers to be impor-
tant about a document, including its history, it can also be used to control the retention of documents.
430

48.2 Page 472

▲back to top
Python in the Enterprise
Document retention is a subject in the news lately, again because of the new requirements imposed by
Sarbanes-Oxley (you’ll look at them in more detail later). A document retention framework is an orga-
nized way of ensuring that particular documents are kept around in storage for a given period of time,
after which their disposition is further specified (the terms disposition and archival are more good ter-
minology you can use to impress your friends). Documents may be off-loaded to less accessible, but
more permanent, storage (such as optical media), or they may be shredded — that is, their content dis-
carded. Sometimes, they may leave a permanent archival footprint — that is, their metadata may be
retained permanently in some way.
One of the Python case studies is a relatively simple document retention framework based on the wftk
workflow toolkit. You can use this example right out of the box as a full-fledged document management
system, and you can use it as a base from which to build more elaborate support systems for your orga-
nization’s business processes. You can see that enterprise software is pretty powerful stuff.
People in Directories
The management of users and groups is a separate area of enterprise software infrastructure that is
heavily relied upon, but is often a bit more complex than it seems it should be. Most earlier platforms
included their own idiosyncratic systems for keeping track of who’s who. Fortunately, in the 90s, with
the advent of open-directory standards and well-understood platforms such as X.500 and LDAP, directo-
ries started to be understood as a separate area of software infrastructure that needed to be standard.
One of the reasons why directories were a slow starter is that keeping track of users and groups is sim-
ply not very complicated. You can easily do good user and group management in a plain old database,
so it’s not surprising that the writers of, say, document management systems would simply include
users and groups into their database structures and not think twice about using an external system.
(The fact that external global systems for user management didn’t exist until recently was another
strong motivation.)
Organizationwide directories make sense for a few reasons. The first is simply the fact that they exist
and have become a standard, and they can become the de facto location of information about people
across one entire organization, or across many sub-organizations. The second reason is that they are so
ubiquitous, because when they work well and are structured in such a way that they make a lot of sense
for storing directory information, they fill a niche that every organization has. The third reason that
modern directory servers are becoming so popular is that directory servers are now designed and writ-
ten to be blindingly fast and low-overhead for reading data from them. Because organizational data
tends to change very slowly in comparison to the data stored in most relational databases, you can
think of a directory server as a sort of write-once read-often database — retrieving user information
from a directory, even performing complex searches on directory information, can be cheaper than from
a relational database. This makes it suitable for use by even large organizations with many dependent
applications. Directory concepts can be better understood if they’re compared to relational databases.
Directories contain entries, which are like database records. Entries have attributes that are named,
searchable values, like a row of data in a database as well. Unlike relational databases, a directory sys-
tem arranges these entries in a hierarchical, or tree, structure, a lot like the files and folders on a regular
disk drive. The tree structure works well because it often corresponds to the organizational structure of
a company, or to geographical relationships between entities, so that you might have an entry for the
whole company, and sub-entries (branches from the tree) for divisions, departments, working groups,
and finally individual employees.
431

48.3 Page 473

▲back to top
Chapter 20
Beyond simply storing contact information, the directory can also store information about the functional
groups to which a particular person belongs. These are usually just called groups, although this can
result in some confusion with organizational groups. The functional groups to which a particular user
belongs can be used to determine the roles that person can play in particular business processes. You’ll
find out more about this, and in more detail, when the discussion turns to workflow systems.
The simplest and possibly the most widespread access API for directory systems, and the one you’ll be
looking at in this chapter, is LDAP (Lightweight Directory Access Protocol.) The OpenLDAP project has
standalone servers, which can talk to LDAP clients, and LDAP can also be used to talk to other directo-
ries, such as X.500. The API is simple and easy to understand when you get past the newness of direc-
tory names, and it’s well-supported in Python, so it’s an easy choice.
In the following sections, you’ll look at some simple ways to use the Python-ldap module from Python
to retrieve information about users stored in an OpenLDAP directory, but first you’ll be introduced to
the last of the three main categories of enterprise software covered in this chapter.
Taking Action with Workflow
The third category of enterprise software is workflow. Workflow systems model the processes that are
used in the organization. From a programming perspective, a process is usually thought of as being a
running program, but the processes modeled by workflow may be active for days, months, or even
years; these are business processes, not system processes.
Whenever people — that is, actual human beings — must make decisions or take action during the
course of a process, that process is an example of a workflow. Workflow processes can also be fully auto-
matic, with any decisions made based on pre-set rules, but in general it’s the involvement of human
beings that’s key.
An example of a simple workflow case would be as follows:
1. A user on the Internet submits a form.
2. The results of the form are used to add a record to a database — a situation with which you’re
familiar.
3. The results need to be approved, depending on the contents. A workflow system makes it easy
to model an approval step:
a. After the form is submitted, it is added to a queue
b. A clerk checks it to make sure it’s appropriate.
4. The record added to the database.
Only after every step has been completed is the workflow complete. If any step doesn’t succeed (the
form isn’t fully filled out, the clerk can’t validate the data, or some other event interferes), the workflow
doesn’t complete, and has to be expanded to accommodate that. For instance, do you get in touch with
the user (you probably want this for a business lead) or do you just discard the data (it seems many
companies choose to do this with customer feedback anyway).
Between the form submission and the addition of the database record, the situation is active in the work-
flow system as a (workflow) process. Any time an actor is involved and must perform an action, there’s
432

48.4 Page 474

▲back to top
Python in the Enterprise
a task involved. For instance, in step 3a, there is a task active for the clerk. Now consider this: “the clerk”
may be a single person or any one of several people — for example, the staff of a department, or even a
program that automates the validation.
In the workflow world, you talk about a role — and then you can specify, perhaps by means of an
attribute value in a directory, who can fill that role in any specific case by placing that user, another
process or program, into that role.
The workflow system records each event relevant to the process in the process’s enactment. This
includes things such as who (the actor) completed a particular task, who is assigned to a given role for
the current process, and so forth.
This functionality makes workflow systems particularly useful for demonstrating
compliance with regulatory requirements — a workflow system not only ensures
that a documented process is followed; it enables you to demonstrate that this is
actually the case.
Auditing, Sarbanes-Oxley, and
What You Need to Know
Auditing is a word to strike fear into the hearts of nearly everyone, but that’s only because the natural
state of most people is one of disorganization. The point of auditing is to ensure the following:
1. Things are documented.
2. The documentation is correct and complete.
3. The documentation can actually be found.
As you can well imagine, the features of enterprise software and the requirements of auditors coincide
to a great deal, as they’re both intended to ensure that the people who own and run companies actually
know what those companies are doing. By knowing more about enterprise software, you can be better
prepared to help your organization meet auditing requirements.
This is a disclaimer:
While you will gain some understanding of auditing here (and the thinking that goes
into it), the practice of auditing is a lot like the practice of law. You can only learn
enough in one chapter to be dangerous. It is highly specific to the situation at hand;
and in spite of the rules involved, it is as much an art as a science. The company, orga-
nization, and country in which an audit is being conducted, and even the people and
regulators involved, combine to make the rules that set the conditions for the success
or the failure of the audit. No two will be exactly alike!
433

48.5 Page 475

▲back to top
Chapter 20
If you’re interested in working with products that will be used in an audit, or that need to be bulletproof
for an audit, you need to speak with the people who will be involved with that audit — lawyers, accoun-
tants, business stakeholders, and anyone else relevant, and get their input! Without that, you shouldn’t
have a feeling that you’re on firm ground in developing a program or process to meet those needs.
Auditing requirements are prominent in the news lately. The Sarbanes-Oxley Act (often referred to as SOX
by the people who love it) was signed into law in 2002 in order to strengthen the legal auditing require-
ments for publicly traded companies, although its different sections didn’t go into effect until later —
some parts as late as the end of 2004. As is the case for most legislation, it is a sprawling 150 pages in
length, and only two sections (Section 302 and Section 404) are really of interest to us as programmers, as
they involve the “control frameworks” in place in a company, which includes IT installations.
It’s important to realize that while auditing started out as a discipline for enforcing financial controls
(an “audit” to most people is something the IRS does when they scrutinize your financial records), in
fact, auditing now concerns documentation and verification of all sorts. It encompasses disciplines from
traditional financial documents to the documentation of compliance with any legal requirements, which
might include OSHA or EPA regulations, for instance.
Some industries, such as medical care, have industry-specific regulatory frameworks (HIPAA is the
example for the medical industry), and an audit must be able to show that a company is in compliance
with those requirements. Another example of auditing guidelines is the ISO quality management series,
such as ISO 9000 and ISO 9001. Compliance with these guidelines is typically not required by law, but
can be a very potent factor in competitiveness, because many companies will not deal with suppliers
who are not ISO compliant.
Auditing and Document Management
The central point of all these auditing requirements is document management. The ISO defines quality
auditing, for instance, as a “systematic, independent and documented process for obtaining audit evi-
dence and evaluating it objectively to determine the extent to which audit criteria are fulfilled.” Audit
criteria are the requirements imposed by legal requirements, industry guidelines, or company policy;
ISO defines them as a “set of policies, procedures, or other requirements against which collected audit
evidence is compared.” The real crux of the matter, however, is the audit evidence, which ISO defines as
“records, statements of fact or other information, relevant to the audit and which are verified.” Thus,
the task of document management is to store audit evidence, and to make it available to auditors in an
organized way whenever any of it is required.
Simple storage, however, is not the only requirement you have for document management. Document
retention, which you’ve read about, is the flip side of the coin. Documentation retention policies specify,
for different classes of documents, how long each one must be held and be retrievable, and what is to be
done with it after a particular period of time (burned to optical storage, discarded, etc.) For instance, there
are legal requirements in some countries, such as Italy, for ISPs to retain all router logs for a certain time,
such as 90 days or three years. Once that period has elapsed, not only is there no reason to keep those
logs, they may actually represent a liability because the ISP has to pay for storage space for all of that data.
A document retention policy might specify that documents should be discarded after the specified period
elapses.
Another example is the retention of e-mail logs to and from corporate users. There are some require-
ments for retention of internal communications, but in general, companies don’t want to keep them any
longer than absolutely necessary from a legal standpoint.
434

48.6 Page 476

▲back to top
Python in the Enterprise
Document retention is one of those subjects that people talk about in the abstract, around the water
cooler with their friends, but it’s really not all that difficult in principle to build a simple document
retention system (where it becomes complicated, of course, is when complex logic is required to decide
when a document should be affected, which is part of the large job of document management). One of
the programming examples that follow is actually the construction of a simple document retention
framework in Python.
Working with Actual Enterprise Systems
The actual coding of most enterprise systems is relatively simple when you get right down to it. When
you’ve documented the way your organization actually works, and you’ve modeled the system context
and the business processes, all you really need to do is write various bits of glue code to move informa-
tion around, which Python makes really easy.
You’ll be able to work through all three of the categories of software covered earlier (document manage-
ment, directories, and workflow systems), but you’ll only be using two different packages to do all of
that. The wftk open-source workflow toolkit is the package used to demonstrate data and document
management and workflow systems; then the Python-ldap module is introduced to cover a few simple
operations against LDAP directories.
Introducing the wftk Workflow Toolkit
The wftk is an open-source workflow toolkit written by Michael Roberts of vivtek.com, who has lav-
ished much effort on it since its inception in 2000. It is written in ANSI C and has a well-tested interface
into Python, so it’s a strong choice for writing enterprise-oriented code in Python.
The one thing you have to keep in mind when writing to the wftk, however, is that you’re not just sit-
ting down to write code. As explained at the outset of this chapter, enterprise programming is all about
modeling business processes, and the wftk reflects that basic approach. Before doing anything else, you
have to describe the context of the actions and objects with which you’ll be working. After a while, this
gets to be second nature, but until it does, it may feel unnatural.
It makes sense to talk about a repository first when starting to build a system using the wftk. The repos-
itory defines the context for everything else in a wftk system. The key part of the repository definition is
the lists it defines. A list is simply a data source — in terms of relational databases, a list can be seen as a
table, because it contains entries that have corresponding records. However, there are some differences:
❑ An entry always has a unique key that can be used to reference it.
❑ An entry need not consist of a fixed number or assortment of fields (data values).
❑ An entry can also contain arbitrary XML.
❑ An entry can include any number of named attachments, each of which can have a version list.
It’s important to realize that these are simply the capabilities of a general list; real lists stored in real
actual places may only offer a portion of these capabilities based on your decisions.
435

48.7 Page 477

▲back to top
Chapter 20
The other salient feature of the repository and the repository manager is that everything is managed by
adapters. An adapter is a fairly compact set of code that defines an external interface. For instance, there
are adapters to store list entries in a local directory as XML files, or in a relational database table (MySQL,
Oracle, or a generic ODBC database) as lines in a text file, and so on. In addition, it’s simple to write new
adapters to treat other resources as suitable locations for the storage of repository lists.
In other words, when the task is to write some simple code to do something in the context of an already
existing set of information resources, the first task is to describe those resources as lists so that the repos-
itory manager knows what’s where. The bigger examples will step through a couple of systems in detail,
but initially you can look at a few much simpler examples so you can see how it’s done. All of these
examples assume that you have already installed the wftk and it is running in your Python installation
(see Appendix B for the wftk home page, and this book’s web site at www.wrox.com for more thorough
instructions on installing the wftk on your computer).
For your first foray into the wftk, set up an extremely simple repository and write a few scripts to save
and retrieve objects. Generally, a wftk repository is a local directory in which the repository definition is
saved as an XML file named system.defn. Subdirectories can then be used as default list storage (with
each entry in an XML file of its own), and a convenient event log will be written to repository.log in
the repository directory. If you haven’t read Chapter 15 on XML and Python yet, you can look there if
you are confused. In addition, remember that all of this text will be on the book’s web site.
Try It Out Very Simple Record Retrieval
1. Create a directory anywhere you want; this directory will be your repository (all the following
examples will reside in this directory). Using the text editor of your choice, open the file
system.defn in that directory, and define the repository’s structure like this:
<repository loglevel=”6”>
<list id=”simple”>
<field id=”field1” special=”key”/>
<field id=”field2”/>
</list>
</repository>
2. Save the file and then create a subdirectory named simple. Just to make things easy, use your text
editor to add a file in the new subdirectory named first and add the following contents to it:
<rec>
<field id=”field1”>value1</field>
<field id=”field2”>value2</field>
</rec>
3. Save the record and then return to the main directory and add a new Python script named sim-
ple.py. This script is going to retrieve lists of keys and then the object you just created. Again
using your text editor, enter the following code:
import wftk
repos = wftk.repository(‘site.opm’)
l = repos.list()
print l
l = repos.list(‘simple’)
436

48.8 Page 478

▲back to top
Python in the Enterprise
print l
e = repos.get(‘simple’, ‘first’)
print e
print e.string()
print
print e[‘field1’]
4. Now run the simple retrieval script:
C:\\projects\\simple>python simple.py
[‘simple’]
[‘first’]
<rec key=”first” list=”simple”>
<field id=”field1”>value1</field>
<field id=”field2”>value2</field>
</rec>
<rec key=”first” list=”simple”>
<field id=”field1”>value1</field>
<field id=”field2”>value2</field>
</rec>
value1
How It Works
What did you just do? In step 1, you described a simple repository for the wftk. In step 2, you created a
schema for a two-element list (within the <rec></rec> tags). In step 3, you then wrote a small program
that accesses data from the fields you defined in your new repository. In step 4, you used that script to
get some data.
Specifically, the repos.list call returns a list of keys that reside in that particular list. If you don’t
provide a list name, though, all of the names of all of the lists defined in the system will be returned.
Therefore, you can see that your file first.xml shows up as a record with the key “first”, in the second
call to repos.list.
When that’s confirmed, the repos.get invocation is used to retrieve an entry. Its first parameter is the
repository, “simple”, and the second parameter is the key, “first”. This entry is an xmlobj object,
which is a specific wftk object:
The upshot is that it is an XML record that the wftk can use like a Python dictionary from which it can
retrieve values, which is done in the final line of the example. (The xmlobj library has some additional
extended capabilities, which you can explore on your own if you like.) If you choose to print just the
entry (either by calling it or calling it with the string method), it renders itself back into XML.
The most important point to take away from this first opportunity to test the waters is this: Because the
repository definition describes a context, this description can also be used to indicate storage of data in
other locations, such as MySQL databases.
That flexibility makes it a lot more useful and exciting: By offering you the capability to describe data
sources, you make any data source amenable to being integrated with your workflow. Even better, the
adapter concept means that even if you have data in some idiosyncratic data format, all you have to do
is write a few lines of relatively simple C code, and the wftk can use it — it’s as simple as that.
437

48.9 Page 479

▲back to top
Chapter 20
Note a few things in the server definition in the first file: First is the log level set on the very first line.
The log level determines how verbose the wftk will be (how much information will be printed) when
logging events to the repository.log file (you can look at this now; it’ll be there after you’ve run the
example script). A log level of 6 is very high, which means that it will print a lot of information; under
normal circumstances, you probably won’t want to use it for anything except development and debug-
ging. Instead, try a log level of 2 or 3 after you’ve seen the messages once; this will help you to get a
general sense of what’s going on. If you don’t need the log at all, turn it off with a log level of 0.
Try It Out Very Simple Record Storage
Record retrieval alone will only take you so far. Let’s create another simple bit of code, this time to
store a new record into the same list that you just created, and from which you’ve already retrieved the
record “first”:
1. Go to the repository directory you set up in the last example and add another new Python script
named simple2.py. This script is going to store a record and then display the newly changed
record to show the effect on the list:
import wftk
repos = wftk.repository()
e = wftk.entry (repos, ‘simple’)
e.parse (“””
<rec>
<field id=”field2”>this is a test value</field>
<field id=”extra”>here is an extra value!</field>
</rec>
“””)
print “BEFORE SAVING:”
print e
e.save()
print “AFTER SAVING:”
print e
print
l = repos.list(‘simple’)
print l
2. Now run the storage script:
C:\\projects>python simple2.py
BEFORE SAVING:
<rec>
<field id=”field2”>this is a test value</field>
<field id=”extra”>here is an extra value!</field>
</rec>
AFTER SAVING:
<rec list=”simple” key=”20050312_17272900”>
<field id=”field1”>20050312_17272900</field>
<field id=”field2”>this is a test value</field>
<field id=”extra”>here is an extra value!</field>
</rec>
[‘20050312_17151900’, ‘first’]
438

48.10 Page 480

▲back to top
Python in the Enterprise
How It Works
In step 1, the program code creates a new entry for a given list, parses XML for a new entry into the list,
and then saves it with the e.save method. Here, this is accomplished in two steps so you can see how
it can be done in case you ever need to do something between these steps. For a real-world application,
you have a simpler option: You can use the alternative repos.add(‘simple’, contents), which
would have the same ultimate effect except that you wouldn’t have the record to print before and after
storage because it’s all wrapped into one step. Here, the default storage adapter has found the column
flagged for special key handling; and because a key requires a unique value in the repository (similar to
how a primary key in a relational database works), it has created a unique value for that field.
In the following workflow examples, you can see another interesting application of this same behavior.
Just as a record is extended by adding a new key, workflow actions can also be defined, which are trig-
gered on particular events, specifically when a record is added. Because the workflow actions can be
pretty advanced, you can take advantage of these triggers as a convenient way of running a script to
perform maintenance, perform backups, or do other things you need, at periodic intervals.
Another feature demonstrated in the code example is the extra field in the stored object that isn’t defined
in the list definition (see the file ‘server.defn’ in the previous code example to confirm this).
Unlike a database, the wftk really doesn’t care and will handle this without complaint. This underscores
the fact that the server definition is descriptive — it simply tells the wftk what is out there in its environ-
ment. The philosophy of the wftk is to make as few further assumptions as possible and still do some-
thing sensible with what it finds.
Try It Out Data Storage in MySQL
Next in this series of wftk building blocks is to store some data in your MySQL database and then use the
same set of code to work with it in the database as you just used to work with files in the local directory.
The point here is to emphasize that the wftk works well with data no matter where it might reside — as
long as it can be organized into lists and records. Once you have your data in this form, you’re good.
This exercise assumes that you have MySQL installed and available to you, and that you have compiled
the wftk with MySQL capabilities (on Unix) or you have the LIST_mysql.dll adaptor installed (on
Windows). If you don’t have these, see the web site for this book for more detailed instructions on find-
ing and installing them.
1. From the MySQL prompt, define a table and add a record to it like this:
mysql> create table test (
-> id int(11) not null primary key auto_increment,
-> entry datetime,
-> body text);
Query OK, 0 rows affected (0.53 sec)
mysql> insert into test (entry, body) values (now(), ‘this is a test’);
Query OK, 1 row affected (0.08 sec)
mysql> select * from test;
439

49 Pages 481-490

▲back to top

49.1 Page 481

▲back to top
Chapter 20
+----+---------------------+----------------+
| id | entry
| body
|
+----+---------------------+----------------+
| 1 | 2005-03-11 23:56:59 | this is a test |
+----+---------------------+----------------+
1 row in set (0.02 sec)
2. Now go to the repository directory from the first code example and modify server.defn to
look like the following. All you’re doing is adding a list definition for your new table, and
adding a connection definition for MySQL. Note that if this is anything other than a private test
installation of MySQL, you will want to change the userid and password that wftk will use to
log into your test database! Make it something that only you know, and use it wherever the
userid “root” and password “root” is shown here in the examples.
<site loglevel=”6”>
<connection storage=”mysql:wftk”
host=”localhost” database=”test” user=”root” password=”root”/>
<list id=”simple”>
<field id=”field1” special=”key”/>
<field id=”field2”/>
</list>
<list id=”mtest” storage=”mysql:wftk” table=”test” key=”id”>
<field id=”id” special=”key”/>
<field id=”entry” special=”now”/>
<field id=”body”/>
</list>
</site>
3. Now add yet another script file; call it simplem.py, and add the following content:
import wftk
repos = wftk.repository()
l = repos.list()
print l
l = repos.list(‘mtest’)
print l
e = repos.get(‘mtest’, ‘1’)
print e
e = repos.add (‘mtest’, “””
<rec>
<field id=”body”>this is a test value</field>
</rec>
“””)
l = repos.list(‘mtest’)
print l
print e
print
print e[‘body’]
440

49.2 Page 482

▲back to top
Python in the Enterprise
4. Now run the storage script:
C:\\projects>python simplemysql.py
[‘simple’, ‘mtest’]
[‘1’]
<record id=”1” list=”mtest” key=”1”>
<field id=”id”>1</field>
<field id=”entry”>2005-03-12 20:25:51</field>
<field id=”body”>this is a test</field>
</record>
[‘1’, ‘2’]
<rec list=”mtest”>
<field id=”body”>this is a test value</field>
<field id=”entry”>2005-03-12 20:42:09</field>
</rec>
this is a test value
How It Works
Except for the specification of the list storage, this example is simply a combination of the code from the
previous two examples, cleaned up a little because you already know what it’s doing. This can be pretty
much the same because after you have defined a context for the wftk, you no longer have to worry
about where things are stored because the wftk interfaces make them all work for you.
Although this example doesn’t demonstrate it, there is one significant difference between storage in
MySQL and storage in a local directory: If you added an extra field not defined in the table, similar to
the “extra” field in the storage example shown earlier, the MySQL adapter will simply ignore it when
writing it to the database. This may have surprising consequences later when the record is retrieved
again — simply put, your data would be lost.
There is a feature of the MySQL adaptor that can improve the situation: Any fields not included in the
table definition can be stored as XML in a designated BLOB field. If you are defining your MySQL tables
to suit wftk, this can be a good solution, but if you need to work with an existing schema, adding a
BLOB field for this purpose may not be an option. The next example shows an alternative way of deal-
ing with this kind of requirement.
Try It Out Storing and Retrieving Documents
Storage in relational databases gives you some nice advantages, at the cost of some disadvantages. The
most obvious advantage is that you have the selection and ordering facilities of the database available
when searching for data, as you saw in the last code example. The default local directory storage adapter
is very weak in this regard; although it does have some rudimentary logic for searching and sorting,
there’s not much point in rewriting a database when it’s so simple to install a professionally designed
database system initially.
However, a relational database has its own weaknesses. Chief among them is the rigidity of its
schema — its data format. A table in an RDBMS must always have the same records, and there are only
limited facilities available for storage of attachments. Of course, BLOBs or similar structures can often be
used, depending on requirements, but sometimes it’s convenient to combine the strengths of the default
storage format (completely flexible storage in local files) with the relational database (good searching
capabilities.) The mechanism for doing this is called an index.
441

49.3 Page 483

▲back to top
Chapter 20
An index is an auxiliary storage mechanism to store selected fields from the main list. If an index is set
on a list, the index is used to perform key generation and can also be used to perform searching and
sorting. Thus, using a relational database to store the index, which refers to data that is in directory
storage, can give you some real advantages, depending on how scalable the result needs to be and the
resources you have available. In this code example, you’ll use just such a setup to create a simple
document management system:
1. Using the text editor of your choice, open the file system.defn in the example repository from
the previous examples and add a new document storage list to the repository as follows (you
can leave in place the lists that you defined previously, but you’re not going to be using them
anymore):
<repository loglevel=”6”>
<list id=”docs” list-from=”docindex”>
<field id=”id”/>
<field id=”title”/>
<field id=”descr” type=”text”/>
<field id=”content” type=”document”/>
<index id=”docindex” storage=”mysql:wftk” table=”docindex” order=”title”>
<field id=”id” special=”key”/>
<field id=”title”/>
<field id=”descr”/>
<field id=”created_by” from=”content.created_by”/>
<field id=”created_on” from=”content.created_on”/>
<field id=”edited_by” from=”content.edited_by”/>
<field id=”edited_on” from=”content.edited_on”/>
<field id=”size” from=”content.size”/>
</index>
</list>
<list id=”_users” storage=”here”>
<user id=”me” password=”x” name=”John Q. User”/>
<user id=”you” password=”x” name=”Jane D. User”/>
</list>
</repository>
2. Start your MySQL client and add a document index table like this:
mysql> create table docindex (
-> id int(11) not null primary key auto_increment,
-> created_by text,
-> created_on datetime,
-> edited_by text,
-> edited_on datetime,
-> title text,
-> descr text);
Query OK, 0 rows affected (0.43 sec)
442

49.4 Page 484

▲back to top
Python in the Enterprise
3. Now create a new script called docs.py and enter the following code:
import wftk
repos = wftk.repository()
repos.user_auth (‘me’, ‘x’)
e = wftk.entry (repos, ‘docs’)
e.parse (“””
<rec>
<field id=”title”>Code file simple.py</field>
<field id=”descr”>This script demonstrates wftk reads.</field>
</rec>
“””)
print “BEFORE SAVING:”
print e
e.save()
print “AFTER SAVING:”
print e
print
e.attach_file(‘content’, ‘simple.py’)
print “AFTER ATTACHMENT:”
print e
print
l = wftk.list (repos, ‘docs’)
l.query (“edited_by=’me’”)
print l.keys()
print l[l.keys()[0]]
print e.retrieve(‘content’)
How It Works
Here is where things start to get interesting with document management; and as always, when things
get interesting, they start to get more complicated. The main thing to note is that this example saves a
single document (using the code from the first example) into the document management system, does
some simple searching, and then retrieves the document. On the way, of course, is a lot more detail than
in the previous examples, so let’s look at everything in turn, starting at the top.
First, notice that the docs list is more complex than anything you’ve seen so far. In addition to having
a field of type ‘content’ to store the document outside the main object, it also defines an index. The
index, as explained previously, is a way of having your cake and eating it, too. It enables you to have
a simple, easy-to-search table in MySQL that is always synchronized with the main list storage imple-
mented as a directory containing XML files and attachment documents. Whenever a change is made to
this list, the change is executed for both the main list storage and the MySQL table that mirrors it.
Here, the index uses a from attribute on several of the fields to extract information about the main
attachment. This attachment field is named “content”, and so, for example, “content.size” refers to
the “size” attribute of the final field XML (you can scan down to the output to see the structure of that
field; we’ll come back to this later.) This means that you can build simple SQL query, one that uses a
443

49.5 Page 485

▲back to top
Chapter 20
WHERE clause to find particular objects that have been recorded in the repository, such as only those with
particular sizes, or those created or edited before or after particular dates, or by particular users. All of
this information is always saved by the attachment process so you know that it will be available for you
to query.
Note that the storage setup for this simple document management system, although it may encounter
issues in scaling, can easily be a test bed for a more industrial-strength version at some later date.
For instance, you could easily replace the back-end with an Oracle database, storing documents in a BLOB
field, and continue to use any Python scripts you’d written against the test system with no changes at all.
The same applies to whatever workflow you define to be used with this system. Moreover, if you move to
a commercial document management system, at most you would have to write an adaptor module to
interface the wftk to the new storage system, and continue to use the same scripts and workflow.
The second feature of this new repository definition, and something you haven’t learned in this book
yet, is that it contains a user list. This user list is very simple, and obviously it isn’t built with security in
mind; in a real system, you would want to have some more convincing security. However, for demon-
stration purposes, and in other limited circumstances, this can be valid solution.
This list uses “here” storage, meaning it’s a read-only list that is defined and stored right in the repository
definition. It defines two users, me and you. Of course, you need a user list because next you’re going to
register a user before attaching documents, so that the attaching user can be used as a parameter to be
searched for. This is needed in most real environments.
Moving along to the SQL definition of the docindex table, note that the primary key of the table has an
auto_increment attribute. This is a nifty MySQL feature that assigns a unique key to a record when it’s
saved; the wftk, after the key field is defined as a key field with attribute keygen set to “auto”, will
first save the record to the index, ask MySQL what key it was assigned, and then will modify the record
in the main storage to match.
Now take a look at the code. There are several new features to study here, the first being the call to
user_auth to assign a user to the current session. The current assigned user has very little obvious
effect, but it allows workflow, permissions, and the attachment process to note who is taking action.
You’ll come back to user authorization when you look at more advanced workflow later.
The document object is created and saved in the same way you’ve created and saved your objects so far,
but now you also attach a file to it. Note that attachments are named by field, and that you can attach
arbitrary numbers and types of documents to an object. Objects don’t all have to be predefined in the
list definition.
Because you’ve already defined an attachment field and have indexed some of its attributes, only
attachments with that name will affect the index.
You aren’t restricted to attaching the contents of files, though. The attach method can specify an arbi-
trary string to be written to the attachment. In Python, this gives you a lot of flexibility because you
could even attach a pickled object that you wanted to be run on the object when it’s retrieved!
When the file is attached, things get interesting. Because this list has MySQL behind it, you can use the
power of SQL to search on it. The next few lines build a special list object to perform the query on, and
then call the query with the text of an SQL WHERE clause. After the query is done, there is a little manipu-
lation of the data, and then you can retrieve the attachment again and print it as a demonstration.
444

49.6 Page 486

▲back to top
Python in the Enterprise
Looking at the output from this example, you can first see three versions of the object as it’s built, saved,
and then attached. Remember that after it has been saved, it is given an id field. This field is written by
MySQL with a unique key, and this is done automatically by the “autoincrement” option that was set
when the table was defined.
After attachment happens, you can see that a new field, the content field, has been written to the object.
This field doesn’t store the attachment itself, but rather specifies where the adapter can retrieve it when
you do want it. Obviously, because attached files can be of any size, it’s better not to retrieve them when-
ever the object is used, because if it’s a large file and you don’t want it every time, that would slow
down your programs a lot.
The descriptor field for an attachment is decorated with various useful attributes, which the wftk can
use to retrieve information about the attachment during the attachment process — things that you’ve
seen discussed and that you already know are important, such as information about the time at which
events occurred to the document, about the user, and about the size of the file itself.
This is the data that you abstract out for the MySQL index summary, and you’ll see it and use it again
later in the output. You can also see it by running queries using your MySQL client with what you
already know about mysql — for instance, querying with SELECT * FROM docindex.
After the file is attached to the object, the results of the query are returned. The query searches on objects
created by user “me,” so if you run this example code several times, you’ll see that all of those objects
are retrieved by this query, which could be more useful in the future when you are looking for multiple
results. Of course, you can easily modify the attachment code to do something else, and then the results
of this query will change based on what you’ve done.
Here is the result of running the script:
C:\\projects\\simple>python docs.py
BEFORE SAVING:
<rec>
<field id=”title”>Code file simple.py</field>
<field id=”descr”>This script demonstrates wftk reads.</field>
</rec>
AFTER SAVING:
<rec list=”docs” key=”1”>
<field id=”title”>Code file simple.py</field>
<field id=”descr”>This script demonstrates wftk reads.</field>
<field id=”id”>1</field>
</rec>
AFTER ATTACHMENT:
<rec list=”docs” key=”1”>
<field id=”title”>Code file simple.py</field>
<field id=”descr”>This script demonstrates wftk reads.</field>
<field id=”id”>1</field>
<field id=”content” type=”document” created_on=”2005-03-19 20:16:27” edited_on=”
2005-03-19 20:16:27” created_by=”me” edited_by=”me” size=”272” mimetype=”” locat
ion=”_att_1_content_.dat”/>
</rec>
[‘1’]
<record id=”1”>
445

49.7 Page 487

▲back to top
Chapter 20
<field id=”id”>1</field>
<field id=”created_by”>me</field>
<field id=”created_on”>2005-03-19 20:16:27</field>
<field id=”edited_by”>me</field>
<field id=”edited_on”>2005-03-19 20:16:27</field>
<field id=”size”>272</field>
<field id=”title”>Code file simple.py</field>
<field id=”descr”>This script demonstrates wftk reads.</field>
</record>
import wftk
repos = wftk.repository(‘site.opm’)
e = wftk.entry (repos, ‘simple’)
e.parse (“””
<rec>
<field id=”field2”>this is a test value</field>
</rec>
“””)
print “BEFORE SAVING:”
print e
e.save()
print “AFTER SAVING:”
print e
print
l = repos.list(‘simple’)
print l
The results of the query show first the list of keys returned by the query, and then an example record
after it has been returned. Note that these return records are the returns from MySQL; they have a differ-
ent structure from the records actually saved to the main list storage. Specifically, you can see that the
attachment-specific fields such as size and created_on have been stored as separate fields in the
database and that they remain separate fields here in the XML output.
Finally, the output dumps the content of the attachment, which is just the code from the first sample,
which was saved.
There are now a hundred different things you could do to make this script serve a specific, useful
purpose in your work. One of those is to manage your stored documents in a document retention
framework, so let’s look at that.
Try It Out A Document Retention Framework
Ready to get your feet wet with something really useful? Try putting together a simple document reten-
tion manager. You already know nearly everything you need from the preceding examples; all that’s
needed is to roll it all up into one program. As noted, you shouldn’t be terribly worried at the lack of
scalability of this test system; you can easily swap out the back-end for applications with more heavy-
duty performance requirements.
This example assumes that you’ve worked through the last one and that you have a few documents
already stored in the docs list. If you didn’t define a docs list, this example isn’t going to work. Of
446

49.8 Page 488

▲back to top
Python in the Enterprise
course, even if you did define a docs list, its contents are going to be pretty bland if you haven’t modi-
fied the example code, but you’ll still get a feel for how this works.
1. Using the text editor of your choice, open the file system.defn in the example repository from
the previous examples and add a new list for storing the retention rules you’ll be defining:
<repository loglevel=”6”>
<list id=”rules” storage=”mysql:wftk” table=”rules” order=”sort”>
<field id=”id”/>
<field id=”sort”/>
<field id=”name”/>
<field id=”rule”/>
</list>
Make sure you don’t change the rest of the repository definition!
2. Start your MySQL client and add the rule table and a couple of rules, like this:
mysql> create table rules (
-> id int(11) not null primary key auto_increment,
-> sort int(11),
-> name text,
-> rule text);
Query OK, 0 rows affected (0.01 sec)
mysql> insert into rules (sort, name, rule) values (1, ‘files by me’,
“created_by=’me’ and to_days(now()) – to_days(edited_on) > 4”);
Query OK, 1 row affected (0.01 sec)
mysql> insert into rules (sort, name, rule) values (0, ‘files by you’, “created_
by=’you’ and to_days(now()) – to_days(edited_on) > 3”);
Query OK, 1 rows affected (0.01 sec)
3. Create a new script called trash.py and enter the following code:
import wftk
repos = wftk.repository()
rlist = wftk.list (repos, ‘rules’)
rlist.query (“”)
for r in rlist.keys():
rule = wftk.xmlobj(xml=rlist[r])
print “Running rule ‘%s’” % rule[‘name’]
docs = wftk.list (repos, ‘docs’)
docs.query (rule[‘rule’])
for d in docs.keys():
print “ -- Deleting document %s” % d
doc = repos.get (‘docs’, d)
doc.delete()
447

49.9 Page 489

▲back to top
Chapter 20
4. Run it (naturally, your output will look different from this):
C:\\projects\\simple>python trash.py
C:\\projects\\articles\\python_book\\ch20_enterprise>python trash.py
Running rule ‘files by you’
Running rule ‘files by me’
-- Deleting document 2
How It Works
Given that this program is actually the first one in this chapter that may have some real-world application,
you might find it surprising that it’s so simple. Here, the approach has been to define the working environ-
ment (the repository) first and then proceed to define the actions to take. Let’s look at things in detail.
The definition of the rules list is deliberately simplified so that the structure of the program will be easy
to see. For instance, there is no way to define the action to be taken or the archival level to be used for
any rule that fires; the only action, by default, is deletion. All you’ve really done is to define a list of
“WHERE” clauses, give them convenient names, and make it possible to define a sort order. The order
attribute on the rules list ensures that the wftk will retrieve them in the right order when you want to
use them.
In step 2, you define the rules table in MySQL and then insert two rules. Files edited by the user “you”
are deleted in three days, but files defined by the user “me” aren’t deleted until they’re four days old —
this is an example of how roles can be used to define a retention policy.
Just to demonstrate a point, the sort order of the two rules is in the opposite of the order they’re defined
(and, thus, the opposite of the numeric order of their keys.) The point is that the wftk will retrieve them
anyway, and in the sort order that you specified, without needing any additional prompting for input
from you.
To start the program, then, connect to the repository and define a list of rules to be retrieved; then use the
query method to retrieve them. Using the keys method to iterate along the rules, define an fwtk XML
object for each retrieved record to make them available to work with. An XML object is the standard
record storage structure of the wftk, so after you’ve defined the object, you can address named fields
directly, making it easy to work with list query results. Now, proceed to iterate through the keys and pro-
cess each rule against each list in turn. After telling the user which rule is active, a second query is built to
retrieve all of the documents matching the rule and delete those documents. It’s as simple as that.
The python-ldap Module
Now that you’ve gotten your feet wet with enterprise-style document management programming, look
at the second open-source project of the chapter, OpenLDAP and the python-ldap module.
As mentioned earlier, OpenLDAP is a directory service that comes in a packaged form. It’s a convenient
package because it runs well under Linux or Windows, and it implements the standard LDAP protocol.
LDAP in general is a very solid, well-understood protocol for directory access, and it’s the long-term
strategy for many key computer platforms, including Microsoft’s Active Directory, Sun’s SunOne direc-
tory server, and offerings from other vendors as well. LDAP, as an infrastructure component, is not
going to go away soon, and using it from Python is incredibly simple.
448

49.10 Page 490

▲back to top
Python in the Enterprise
When working through the LDAP examples, you have two options. You can use an existing LDAP
server in your organization to try scripts out on (in this case, you’ll need to modify the scripts to match
your server’s schema, authentication, and other values that will already be defined for you), or you can
set up a test LDAP server, load some data into it, and use that to work with. The first Try It Out that fol-
lows explains how to set up your own LDAP server for these examples, if that’s the way you want to go,
and it’s a good exercise to help you understand a little more about how LDAP works.
Unfortunately, although there are high-quality precompiled Windows distributions for the OpenLDAP
server software itself, there are currently no recent Windows builds of the client software used by the
python-ldap modules, and no interfaces from python-ldap to the standard Microsoft LDAP client
in wldap32.dll. Therefore, for all of these LDAP examples, as of the time this book is published,
you’ll need to have a Linux or Unix system to build an OpenLDAP instance. Fortunately, if you are a
Windows user, you can use the cygwin toolkit to create an environment that you can use to build and
run OpenLDAP. For instructions on downloading and installing OpenLDAP and the python-ldap
module, see the web site for this book.
Try It Out Using Basic OpenLDAP Tools
1. After you’ve downloaded and installed an OpenLDAP package and have followed this book’s
web site instructions for how to set up a basic server, make sure that the domain name it serves
in the slapd.conf file is “wftk.org” if you want to use the following examples without
modifying them. When OpenLDAP is running on your system, use a text editor to create the
following LDIF file anywhere you want:
# Add a simple, generic user
dn: cn=Different Person,dc=wftk,dc=org
objectClass: person
sn: Different Person
cn: Different Person
# Add another user
dn: cn=Michael Roberts,dc=wftk,dc=org
objectClass: person
sn: Roberts
cn: Michael Roberts
# Add a workflow group: wfstarter
dn: cn=wfstarter,dc=wftk,dc=org
objectclass: organizationalRole
cn: wfstarter
roleOccupant: cn=Michael Roberts
roleOccupant: cn=Different Person</repository>
2. Save the file as testldif.txt, and then use ldapadd to add the data you just entered:
[michael@me michael]$ ldapadd -x -D “cn=Manager,dc=wftk,dc=org” -W -f testldif.txt
Enter LDAP Password:
adding new entry “cn=Different Person,dc=vivtek,dc=com”
adding new entry “cn=Michael Roberts,dc=vivtek,dc=com”
adding new entry “cn=wfstarter,dc=vivtek,dc=com”
449

50 Pages 491-500

▲back to top

50.1 Page 491

▲back to top
Chapter 20
3. Now, use ldapsearch to see what happened (note the command on the first line):
[michael@me michael]$ ldapsearch -x -b ‘dc=vivtek,dc=com’ ‘(objectclass=*)’
# extended LDIF
#
# LDAPv3
# base <dc=wftk,dc=org> with scope sub
# filter: (objectclass=*)
# requesting: ALL
#
# Different Person, wftk.org
dn: cn=Different Person,dc=wftk,dc=org
objectClass: person
sn: Different Person
cn: Different Person
# Michael Roberts, wftk.org
dn: cn=Michael Roberts,dc=wftk,dc=org
objectClass: person
sn: Roberts
cn: Michael Roberts
# wfstarter, wftk.org
dn: cn=wfstarter,dc=wftk,dc=org
objectClass: organizationalRole
cn: wfstarter
roleOccupant: cn=Michael Roberts
roleOccupant: cn=Different Person
# search result
search: 2
result: 0 Success
# numResponses: 4
# numEntries: 3
How It Works
The LDIF format (LDAP Data Interchange Format) is defined in RFC 2849 and is the standard text-based
format for defining data and dumping data to and from LDAP databases — and even specifying other
arbitrary data changes, such as changes to and deletions of records. An LDIF file is divided into records
by blank lines, and the first line of each record gives the distinguishing name (the DN, or the key) of the
record affected. The default operation is addition, and the file defined previously simply adds a few
test records to an otherwise empty database. Use this test database for the rest of the examples.
The ldapadd utility is used to interpret LDIF files and make the LDAP API calls appropriate to carry
out the instructions they contain. In addition, the ldapsearch utility can be used to search the database
from the command line and format the results in a more or less LDIF format. These are handy tools to
have at your disposal when you’re working with LDAP, but to do any more involved work, you’ll want
to write your own code in Python, and that’s what the next example is all about.
450

50.2 Page 492

▲back to top
Python in the Enterprise
Try It Out Simple LDAP Search
OK, now that you either have an LDAP server installed or one already available in your organization,
try some basic LDAP accesses from Python so you can see how it all works. When this book went to
press, it was difficult to get python-ldap working under Windows, so this presumes you’re scripting
on a Linux machine:
1. Create a file named simpleldap.py and enter the following:
import ldap
l = ldap.open(‘127.0.0.1’)
l.simple_bind_s (‘’, ‘’)
print “Search for everything:”
ldap_result = l.search_s(“dc=vivtek,dc=com”, ldap.SCOPE_SUBTREE, “cn=*”, None)
print ldap_result
print
print “Search for objects with names containing ‘Michael’:”
ldap_result = l.search_s(“dc=vivtek,dc=com”, ldap.SCOPE_SUBTREE, “cn=*Michael*”, None)
print ldap_result
print
print “Retrieve organizational role ‘wfstarter’:”
ldap_result = l.search_s(“dc=vivtek,dc=com”, ldap.SCOPE_SUBTREE, “cn=wfstarter”,
[“organizationalRole”])
print ldap_result
print
print “Search for everything again, but this time with an asynchronous search:”
ldap_result_id = l.search(“dc=wftk,dc=org”, ldap.SCOPE_SUBTREE, “cn=*”, None)
while 1:
result_type, result_data = l.result(ldap_result_id, 0)
if (result_data == []):
break
else:
if result_type == ldap.RES_SEARCH_ENTRY:
print result_data
2. Now run it:
[michael@me michael]$ python simpleldap.py
Search for everything:
[(‘cn=Different Person,dc=vivtek,dc=com’, {‘objectClass’: [‘person’], ‘sn’:
[‘Different Person’], ‘cn’: [‘Different Person’]}), (‘cn=Michael
Roberts,dc=vivtek,dc=com’, {‘objectClass’: [‘person’], ‘sn’: [‘Roberts’], ‘cn’:
[‘Michael Roberts’]}), (‘cn=wfstarter,dc=vivtek,dc=com’, {‘objectClass’:
[‘organizationalRole’], ‘roleOccupant’: [‘cn=Michael Roberts’, ‘cn=Different
Person’], ‘cn’: [‘wfstarter’]})]
Search for objects with names containing ‘Michael’:
[(‘cn=Michael Roberts,dc=vivtek,dc=com’, {‘objectClass’: [‘person’], ‘sn’:
[‘Roberts’], ‘cn’: [‘Michael Roberts’]})]
451

50.3 Page 493

▲back to top
Chapter 20
Retrieve organizational role ‘wfstarter’:
[(‘cn=wfstarter,dc=vivtek,dc=com’, {‘objectClass’: [‘organizationalRole’],
‘roleOccupant’: [‘cn=Michael Roberts’, ‘cn=Different Person’], ‘cn’: [‘wfstarter’]})]
Search for everything again, but this time with an asynchronous search:
[(‘cn=Different Person,dc=vivtek,dc=com’, {‘objectClass’: [‘person’], ‘sn’:
[‘Different Person’], ‘cn’: [‘Different Person’]})]
[(‘cn=Michael Roberts,dc=vivtek,dc=com’, {‘objectClass’: [‘person’], ‘sn’:
[‘Roberts’], ‘cn’: [‘Michael Roberts’]})]
[(‘cn=wfstarter,dc=vivtek,dc=com’, {‘objectClass’: [‘organizationalRole’],
‘roleOccupant’: [‘cn=Michael Roberts’, ‘cn=Different Person’], ‘cn’: [‘wfstarter’]})]
How It Works
This simple example really only scratches the surface of the python-ldap module. The intent is just to
get a feel for how the module is used and to look at the data structures used by the module to represent
LDAP results.
The first line to note is the call to the simple_bind_s method. This method logs into the server, using —
in this case — an anonymous connection. Here you would normally give the distinguished name of your
user record and the password associated with that record. Note also the “_s” ending on the function. In
the LDAP API, most functions come in both synchronous and asynchronous forms, as there are no guar-
antees that the answer from the server will come back quickly. For simplicity, it’s often easiest to use the
synchronous API for utilities, but in any environment where the user needs responsiveness, you’ll want
to look into the asynchronous varieties. The asynchronous API sends off requests to the server and
returns immediately with a handle you can use to call monitoring functions. The monitoring functions
can then tell you whether the server has answered. More on this later.
In the meantime, look at the next three blocks, which illustrate various things you can search on using the
search_s method. The four inputs to the search are the distinguished name of an object to start searching
at, the scope of the search (whether you’re just interested in that object, its immediate children, or all
objects in its descent), a filter for the search, and the types of objects to retrieve. You can see how the filters
work fairly easily from the examples; use the asterisk (*) as a wildcard for the filter parameter. If you
don’t supply an object type for the last parameter, any objects matching the filter will be retrieved.
The final LDAP call is an example of how to call an asynchronous search and then poll the API to get all of
the search results. This search performs the same search as the first (synchronous) search in the example.
The output is basically just the return values from all four searches. The normal formatting (or lack thereof)
of the output is hard to see, so rearrange the first search result a little:
Search for everything:
[(‘cn=Different Person,dc=vivtek,dc=com’,
{‘objectClass’: [‘person’],
‘sn’
: [‘Different Person’],
‘cn’
: [‘Different Person’]
}
),
(‘cn=Michael Roberts,dc=vivtek,dc=com’,
{‘objectClass’: [‘person’],
‘sn’
: [‘Roberts’],
‘cn’
: [‘Michael Roberts’]
452

50.4 Page 494

▲back to top
Python in the Enterprise
}
),
(‘cn=wfstarter,dc=vivtek,dc=com’,
{‘objectClass’ : [‘organizationalRole’],
‘roleOccupant’: [‘cn=Michael Roberts’,
‘cn=Different Person’],
‘cn’
: [‘wfstarter’]
}
)
]
As you can see, the structure of the search return is a list of tuples. Each tuple has two elements: The first
element is the distinguished name of the object returned, and the second element is a dictionary contain-
ing data fields from the object. Note that in LDAP, any field may have multiple values, so the value half
of each data field is itself a list. You can see this most clearly in the value for roleOccupant in the
wfstarter object; this field has two values.
More LDAP
This little introduction to the wonders of directory programming has naturally barely scratched the sur-
face of LDAP programming. You can find several good books on working with LDAP, but as long as it’s
fresh in your mind, it might be a good idea to take in a couple of pointers.
First, while the core LDAP schema used in the preceding example can do a lot of basic things, nearly any
LDAP application of any scope will define its own schema. An LDAP schema does the same thing as a
schema in any database system: It defines objects and the attributes they must have (and in LDAP, you
can also define optional attributes) and the types of values those attributes can contain. Thus, if you’re
working with an existing LDAP server, you will almost certainly encounter various schema definitions
with which you need to work.
You can still store useful information in the core schema. For instance, the example database shown in
the preceding example defines a couple of organizational roles (object type organizationalRole),
each of which has multiple occupants. This is a simple (although not very scalable) way to define user
groups, which requires absolutely no schema design or definition work at all. Therefore, it’s possible to
use an LDAP installation to note roles and responsibilities for a small-to-medium organization or group.
For the reasons of flexibility talked about earlier — the fact that writing a truly flexible mapping from
LDAP functionality onto the wftk’s language is more complicated than it looks — the wftk does not yet
have a finished LDAP adaptor (at least, that’s the situation as this book goes to press.) But if it did, or
if you wrote a simplified one customized for your own organizational needs, LDAP would be a good
choice to store the user and permissions information necessary for any real-world enterprise application.
Back to the wftk
For the third section of example programming, look at the wftk again, this time in its originally
intended capacity of being a workflow toolkit. Essentially, there are two things to keep in mind about
applications of the wftk’s workflow functionality: First, workflow is activated when events occur in the
system — basically, the addition or modification of objects. After a list is defined, you can define a work-
flow to be activated when an object is added (for instance, the addition could contain data from a form
submission on a web site, or it could be as simple as a cron job on a Unix server firing at a particular
time of day in order to cause some workflow action to take place.)
453

50.5 Page 495

▲back to top
Chapter 20
Second, the entire point of workflow is to coordinate the actions of a diverse group of people and/or
programs. If an action to be taken for data involves any kind of scheduling of resources, for instance, this
is an ideal application for workflow. The wftk excels in modeling this kind of system.
This chapter demonstrates two examples of wftk workflow programming in Python: a basic approval
workflow used to intercept and process records submitted (perhaps originally from a web page — where
they’re coming from is really not terribly important), and an action processing queue in which a periodic
process checks for work to be done and then runs a program to do it, storing the information back into
the original task object. Naturally, either of these examples could be ramified endlessly, but they’re good
for getting you started.
Try It Out Simple Workflow Trigger
For the first real workflow in this chapter, you’ll add a simple state machine and approval workflow to
the repository from the very first example. This will accept submissions into a staging area list and then
start a workflow process to get approval from an administrator. When the submission is approved, the
record is written to the main list; if approval isn’t granted, the code deletes it and forgets the whole thing:
1. In the repository directory from the first example, open server.defn and add the following
list definition; this defines a list to be used as a staging area for proposed additions to the list
“simple”:
<list id=”staging”>
<field id=”field1” special=”key”/>
<field id=”field2”/>
<state id=”proposed”/>
<state id=”approved” archive-to=”simple”/>
<state id=”rejected” archive-to=”_trash”/>
<on action=”add”>
<task role=”me” label=”Check anonymous submission”>
<data id=”state”/>
</task>
</on>
</list>
2. Add subdirectories to the repository for the task index and the staging area; these must be
named “_tasks” and “staging”, respectively. Both are treated as perfectly normal lists, so
they can just as easily be stored in MySQL or wherever is appropriate for your needs, but it’s a
little simpler for now to leave them in the default directory-based storage.
3. Now add some scripts. First, use your favorite word processor to add a file called submit.py:
import wftk
repos = wftk.repository()
e = wftk.entry (repos, ‘staging’)
e.parse (“””
<rec>
<field id=”field2”>this is an anonymous submission</field>
</rec>
“””)
e.save()
454

50.6 Page 496

▲back to top
Python in the Enterprise
4. Now add another file called approve.py:
import wftk
repos = wftk.repository()
repos.user_auth (‘me’, ‘x’)
l = wftk.list (repos, ‘_todo’)
l.query ()
e = repos.get(‘_todo’, l.keys()[0])
e.set (“state”, “approved”)
e.save()
How It Works
This example is deceptively simple, but it actually does some useful work. In the first step, you did the
greatest part of the work, by defining the staging area list. That definition has exactly the same fields as
the list “simple”, because it stores suggested additions to the “simple” list. However, the staging area
is a little smarter, because it has a state machine.
A state machine is a data structure that defines different states for an object; it also defines what valid
transitions between states that object can take. In the wftk context, you can also specify actions to be
taken by the workflow engine when objects actually transition between states, and that’s where things
get interesting. The three “state” elements define the start state (proposed), and two other states,
“approved” and “rejected”. Each of these two states is treated as an archival (that is, as an action to
be taken after completion of workflow, which is signaled by entry of the respective state). The archival
target for the “approved” state is the “simple” list — which means that if an object in the staging area
enters the approved state, all outstanding workflow is removed, it is deleted from the staging area, and
it is added automatically to the “simple” list.
If, on the other hand, it enters the “rejected” state, it’s still cleaned up, but it is “added” to the _trash
list — a virtual list, which simply means it is deleted. This is a convenient way to specify that a record
should be discarded.
After the state definitions, there is a simple workflow process definition in the “on” element to ensure
that it’s activated as soon as a record is entered into the staging area. All this workflow does is to create
one task on the active task list; that task is then given the state value as input/output value. This is not
strictly necessary from the point of view of the code, but it’s useful to document the fact that the task is
going to be used to change the object’s state.
Looking at the submit.py script, you can see that it is essentially identical to the example script to
add an entry to the simple list shown earlier in the chapter. The only difference is that it specifies the
“staging” list, and it has been cleaned up to remove demonstration printouts. When you run it, it will
create an entry in the staging area, but there’s an important difference: Because workflow is involved, an
entry is also added to the _tasks list.
Now look at the approve.py script, which is where everything interesting happens. Here, after import-
ing the wftk and opening the repository, the script immediately authenticates itself as the user “me”.
This allows the next line to retrieve the “_todo” list (another magical list, which contains all active tasks
for the current user) and get the first key from that list. Naturally, in a real-world application, you would
455

50.7 Page 497

▲back to top
Chapter 20
do something other than blindly approve the first task in the list (like actually ask a human being to
make a decision), but this illustrates the basic operation of workflow.
Once the first task on the list is retrieved, its state is changed to “approved” and it is saved again. This
simple action hides a great deal of work on the part of the system — once the state is changed, this trig-
gers the state transition, and the record is now moved into the “simple” list. Thus, the function of the
staging area is fulfilled.
You could easily write an equivalent script reject.py that would reject the first submission on the task
list — simply copy approve.py exactly, and then replace the new state value with “rejected”, and you
have written a simple rejection script that will delete the submission. Another interesting variant would
be a separate archival list for rejected submissions (perhaps you want to give people a second chance). To
do this, you would simply define the list and name it in the archive-to attribute of the rejected state.
The point is that with the wftk, you don’t have to write much code to specify many rather complicated
workflow tasks. You simply define the system, say how it should behave, and you’re finished.
Try It Out Action Queue Handler
The previous example illustrated the involvement of a human being in a programmed process, by creat-
ing a task that could be inspected by a human and allowing that person to make a decision regarding
further processing. Another useful application of workflow, though, is as glue for disparate software sys-
tems, or as an organizational framework for programmatic tasks. In this type of system, you can imagine
both humans and programs as agents; and automatic task processing programs can be called action
queue handlers, because their to-do list is effectively an action queue.
Actually, the task completion script you wrote in the previous example can already be seen as an action
queue handler! It checks a task index for actions it needs to take and then completes tasks as directed,
so it already fits the bill. In addition, however, it can do whatever else you need. The following exercise
is a cobbled together simple task it can complete, and it works with a modified version of the previous
example’s system.
1. In the repository directory from the first example, open server.defn and add or modify the
staging area definition from the last example (the changed portion is in italics):
<list id=”staging”>
<field id=”field1” special=”key”/>
<field id=”field2”/>
<state id=”incoming”/>
<state id=”proposed”/>
<state id=”approved” archive-to=”simple”/>
<state id=”rejected” archive-to=”_trash”/>
<on action=”add”>
<task role=”you” label=”Automatic incoming task”>
<data id=”state”/>
<data id=”field2”/>
</task>
<task role=”me” label=”Check anonymous submission”>
<data id=”state”/>
</task>
</on>
</list>
456

50.8 Page 498

▲back to top
Python in the Enterprise
2. If you skipped the last example, add subdirectories to the repository for the task index and the
staging area; these must be named “_tasks” and “staging”, respectively. You’ll also need the
scripts defined there, so enter them as well.
3. Use your favorite word processor to add a file called autocheck.py:
import wftk
repos = wftk.repository()
repos.user_auth (‘you’, ‘x’)
l = wftk.list (repos, ‘_todo’)
l.query ()
e = repos.get(‘_todo’, l.keys()[0])
e.set (“field2”, e.get(“field2”) + “ (checked by automatic processor)”)
e.set (“state”, “proposed”)
e.save()
How It Works
Again, this example only illustrates general possibilities for using the wftk; in any real-world situation,
you would be doing something more elaborate, but take a look at what this setup does. The two
key differences in the definition of the staging area are that a new state has been added before the
“proposed” state, and that a new task has been added to the workflow sequence before human
approval is necessary. That new first task is your action queue, and the autocheck.py script is the
action queue processor.
As promised, the autocheck.py script is very similar to the approval script of the previous example,
with one significant difference: It makes a data change to “field2” before updating the entry’s state to
“proposed”. The actual data change is trivial in this case, but it could be anything you need it to be.
The script becomes an action queue processor when you set up some scheduler in the operating system
(under Unix, this is called a cron job, and under Windows, a scheduled task) to execute it periodically.
This kind of queue processor can serve several related purposes: Because it effectively forces records to
be handled one by one, it can be used to set a limit on the number of requests processed in a given time
frame. In addition, because the submission and autocheck processes are completely separate, it enables
the system to respond very quickly to the submitter, and then execute a (possibly lengthy) autocheck
procedure later. Only after the autocheck procedure is finished is a human asked to make a decision
about the submission.
Note that there’s nothing in this definition that forbids the autocheck procedure from changing the
entry’s state to “approved” or “rejected” — this means that the autocheck procedure might also be
used to handle certain automatic cases on its own, and the remainder of the workflow (that is, the
human involvement) thus becomes unnecessary.
457

50.9 Page 499

▲back to top
Chapter 20
Summary
Enterprise applications make use of software infrastructure to model and support business processes.
This chapter covered three general categories of enterprise software:
❑ Document management, to keep track of the documents that make up the knowledge of a business
❑ Directories, which store information about the people who run a business
❑ Workflow systems, which model and store information about the ongoing processes of a business
You were introduced to some of the current state of the auditing art so that you can put your program-
ming efforts into perspective and see how these categories of software are related, not only to the needs
of business organizations, but to the new requirements for documentation and validation imposed by
regulatory frameworks.
Then you saw two open-source packages: the python-ldap module for talking to LDAP directories and
the wftk open-source workflow toolkit for document management and workflow applications. Some
useful snippets of Python showed how easy it is to write code to keep your organization organized, and
your auditors as happy as auditors can be expected to be. However, the key to keeping auditors happy,
just as any other category of software users, is to ensure that they specify their own needs and that they
have as much input as possible in the evolution of the software that makes their lives easy. Now, though,
you should have a few ways to make your own life easy in that whole process.
Exercises
1. What documentation of existing business processes does your organization already have in
place? Is this documentation machine-usable, or does it simply consist of English descriptions
of the way you do business (the latter is enough to satisfy ISO 9000 requirements, and is already
a great deal better than a system in which people just “know” what should be done). Find out
how your company does its auditing, and think about how the techniques outlined in this
chapter could make it easier for the employees do their jobs.
2. Think of some processes you can model in your life outside the office or in other businesses.
If you have a small business and skipped question 1 because it didn’t apply to you, you’re in
luck: You can start modeling your business processes right now!
3. The document retention framework stores it rules in a list in wftk. Devise a set of workflow and
document management tools based on the wftk that would enable you to modify those rules in
a controlled way.
458

50.10 Page 500

▲back to top
21
Web Applications
and Web Services
If you’ve ever surfed the web, you’ve probably used web applications: to do research, to pay your
bills, to send e-mail, or to buy from an online store. As a programmer, you may even have written
web applications in other languages. If you have, you’ll find the experience of doing so in Python
comfortingly familiar, and probably easier. If you’re just starting out, then rest assured there’s no
better way to enter this field than with Python.
When the World Wide Web was invented in the early 1990s, the Internet was used mainly by uni-
versity students, researchers, and employees of technology companies. Within a few years, the
web had brought the Internet into popular use and culture, co-opting proprietary online services
or driving them into bankruptcy. Its triumph is so complete that for many people, the web is syn-
onymous with the Internet, a technology that predates it by more than 20 years.
Our culture became dependent on the web so quickly that it hardly seems necessary to evangelize
the benefits for the user of web applications over traditional client-server or standalone applica-
tions. Web applications are accessible from almost anywhere in the world. Installing one piece of
software on your computer — a web browser — gives you access to all of them. Web applications
present a simple user interface using a limited set of widgets. They are (usually) platform indepen-
dent, usable from any web browser on any operating system — including ones not yet created
when the application was written.
If you haven’t yet written your own web applications, however, you might not know about the ben-
efits of developing for the web platform. In many respects, the benefits for the developer are the flip
side of the benefits for the user. A web application doesn’t need to be distributed; its users come to
it. Updates don’t have to be distributed either: When you upgrade the copy of the program on your
server, all of your users start using the new version. Web applications are by convention easy to
pick up and use, and because others can link to a web application from their own web sites, driving
traffic there, buzz and word-of-mouth spread much more quickly. As the developer, you also have
more freedom to experiment and more control over the environment in which your software runs.
The virtues of the web are the virtues of Python: its flexibility, its simplicity, and its inclusive spirit.
Python applications are written on Internet time; a hobbyist’s idea can be explored in an evening
and become a web fad the next day.

51 Pages 501-510

▲back to top

51.1 Page 501

▲back to top
Chapter 21
Python also comes packaged with simple, useful modules for interacting with web clients and servers:
urlparse, urllib, and its big brother, urllib2, htmllib, cgi, even SimpleHttpServer. There are
also many (some would say too many) open-source frameworks that make it easy to build a complex
Python web application. Frameworks such as Zope, Quixote, CherryPy, and Subway provide templat-
ing, authentication, access control, and more, freeing you up to work on the code that makes your
application special.
It’s a huge field, perhaps the most active in the Python community, but this chapter gets you started.
You’ll learn how to use basic, standard Python modules to make web applications people will find use-
ful. You’ll also learn how to make them even more useful by creating “web service” interfaces that make
it possible for your users to use your applications as elements in their own programs. In addition, you
will learn how to write scripts of your own to consume popular web services and turn the knowledge
gained to your advantage.
If you’re reading this chapter, you’ve probably used web applications before and perhaps have written a
web page or two, but you probably don’t know how the web is designed or how web applications work
behind the scenes. If your experience is greater, feel free to skip ahead, although you may find the next
section interesting. If you’ve been writing web applications, you might not have realized that the web
actually implements a specific architecture, and that keeping the architecture in mind leads to better,
simpler applications.
REST: The Architecture of the Web
It might seem strange to think of the web as having an architecture at all, especially for anyone who
started programming as or after the web became popular. Because it’s so tightly integrated into your
daily life, the assumptions that drive the web might seem invisible or have the flavor of defaults. They
are out there, though, differing from what came before and arranged into a coherent architecture. The
architecture of the web was formally defined in 2000 by Roy Fielding, one of its founders. He calls the
web architecture Representational State Transfer, or REST. This section briefly summarizes the most
important concepts behind REST, while connecting them to the workings of HTTP (the protocol that
implements REST) and providing examples of architectures that made the same decisions differently.
Characteristics of REST
Much of this chapter is dedicated to writing applications that use the features of the REST architecture to
best advantage. As a first step toward learning about those features, here’s a brief look at some of the
main aspects of REST.
REST Resources
Fielding’s dissertation on architectural styles and REST is available at www.ics.uci.
edu/~fielding/pubs/dissertation/top.htm. Chapter 5 describes REST.
Introductions that are more informal are available at the REST Wiki, at http://rest.
blueoxen.net/, and at the Wikipedia entry for REST, at http://en.wikipedia.
org/wiki/REST.
460

51.2 Page 502

▲back to top
Web Applications and Web Services
A Distributed Network of Interlinked Documents
The most fundamental characteristic of an architecture is the purpose it serves. Without this to use as a
guideline, there would be no way to distinguish good architectures from bad ones. Therefore, the first
characteristic of REST is the problem it solves: the creation of a “distributed hypermedia system,” to
quote the Fielding dissertation. REST drives the web: a network of documents that link to one another,
dispersed over a large number of geographically scattered computers under varied ownership. All of
REST’s other characteristics must be evaluated against this one.
A Client-Server Architecture
The second characteristic of REST is the nature of the actors in a REST architecture. REST defines a
client-server relationship between actors, as opposed to, say, the peer-to-peer relationship defined by
BitTorrent or other file-sharing programs. A document on the web is stored on (or generated by) a partic-
ular server and delivered upon request to a client who asks for it. A client talks only to servers, and a
server only to its clients. In HTTP, the server is a web server, and the client is typically a web browser.
Servers Are Stateless
The third characteristic of REST is that no session state is kept on the server. Every request made by a
client must contain all of the information necessary to carry out that request. The web server need not
know anything about previous requests the client may have made. This requirement is why web
browsers pass cookies and authentication credentials to a site with every single request, rather than only
once at the beginning of a long session.
An HTTP session only lasts as long as one back-and-forth transaction between client and server: The
client requests a document from the server, and the server delivers the response, which either contains
the requested document or explains why the server couldn’t deliver it. Protocols like FTP and SSH, in
which the client and server communicate more than once per session, must keep state on the server side
so that each communication can be understood in the context of the previous one. REST puts this burden
on the client instead.
Many frameworks and applications build sessions on top of HTTP by using cookies, special URLs, or
some other trick. There’s nothing wrong with this — it’s not illegal or immoral, and it has its benefits —
but by doing this, the application forfeits the benefits of the stateless server. A user might find it impossi-
ble to come back to a particular document or might get stuck in a bad state and be unable to do anything
about it because the problem is on the server.
Resources
Because the problem REST solves is that of managing a distributed network of documents, its unit of
storage is the document, or in REST terms the resource. A static web page is a resource according to
REST, but so is one that’s dynamically generated by a web application. On the web, anything interesting
you can get with your web browser is a resource.
Each resource has at least one unique identifier, a string that names it and no other resource. In the world
of HTTP, this is the resource’s URL. The resource identifier http://www.python.org/ identifies a well-
known resource that talks about Python. http://python.org/ is another identifier for the same
resource. http://www.google.com/search?q=Python is an identifier denoting a dynamic resource:
one created upon request by a web application. This custom-made resource is an index full of references
to other resources; all of which should pertain in some way to Python (the language or the snake). It
didn’t have to be this way: WAIS, one of the technologies subsumed by the web, treated searches and
461

51.3 Page 503

▲back to top
Chapter 21
search results as first-class objects. In the REST architecture, these things only exist within resources and
their identifiers.
A web object that can’t be reached by typing an address is not technically a REST resource, because it has
no identifier. If you can only get to a web page by submitting a form in your web browser, that page is
not a resource; it’s a side effect of your form submission. It’s generally a good idea to make your web
pages real resources. A resource is more useful than a nonresource that contains the same information: It
can be bookmarked, passed around to others, accessed automatically, and used as input to scripts that
manipulate resources.
Representations
When you request a resource with your web browser, what you actually get back is a representation of
that resource. In the most common case, a resource has only one representation: The resource is a file on
the disk of the web server, and its representation is byte-for-byte the same as that file. However, a single
resource may have multiple representations. A news site may make each of its stories available in an
HTML file, a stripped-down printer-friendly HTML file, a plaintext file, a PDF file, and so on.
A web client may choose a representation for a resource by choosing between that resource’s identifiers
(for instance, story.html or story.html?printable), or it may simply tell the server which format it
prefers and let the server decide which representation is most appropriate.
REST Operations
We normally think of web pages as things we read, but we act on the web as well, creating and changing
pages through the same tool we use to retrieve them. If you have a weblog, you’re familiar with creating
new web resources by using your web browser, but it also happens in other contexts. When you send
e-mail through a webmail application, an archive page is created that contains the message you sent.
When you buy something from an online store, a receipt page is made available, and other pages on the
site change to show the outstanding order.
The action of retrieving a resource should be idempotent: The fact that you made the request should not
change the contents of the resource. Resource modification is a different operation altogether. In addition
to retrieving a resource, REST also enables a client to create, modify, and delete a server’s resources (given
the proper authorization, of course). A client creates a new resource by specifying, in some format, a rep-
resentation for the resource, and modifies an existing resource by specifying the desired new representa-
tion. It’s up to the web application to render to the exact format of the representation it wants.
In HTTP, the four basic operations are implemented by four commands, or verbs, as described in the
following table.
HTTP Verb
GET
POST
PUT
DELETE
Purpose
Retrieve a resource’s representation
Modify a resource to bring it in line with the provided new representation
Create a new resource from the provided representation
Delete an existing resource
462

51.4 Page 504

▲back to top
Web Applications and Web Services
These four commands are often compared to the basic file system operations (read, write, create, and
delete) and to the four basic SQL commands (SELECT, UPDATE, INSERT, and DELETE). Unfortunately,
as you’ll see in a bit, web browsers support only the first two commands.
HTTP: Real-World REST
Although REST’s principles are generally applicable, it’s realized primarily in HTTP, the protocol that
drives the web. The best way to understand HTTP is to see it in action. To that end, you’re going to
write a web server.
No, really. It’s easy to write a web server in Python. In fact, the simplest one takes just three lines of
code, because Python is packaged with a web server, and all you have to do is activate it.
Try It Out Python’s Three-Line Web Server
Enter this script into a file called EasyWebServer.py:
#!/usr/bin/python
from BaseHTTPServer import HTTPServer
from SimpleHTTPServer import SimpleHTTPRequestHandler
HTTPServer((‘localhost’, 8000), SimpleHTTPRequestHandler).serve_forever()
Run the script and you’ll be able to access your new web server by visiting the URL http://
localhost:8000/.
If another server is already running on port 8000 on your machine, just change the port number in the
script and in the URL when you check it and viola!
How It Works
The script drives an HTTPServer object, which listens on port 8000 for HTTP requests. Every time you
hit the web server with a web browser, a SimpleHTTPRequestHandler object will be spawned to han-
dle your request. The server will serve pages forever until you interrupt it by killing the script.
When you run this script, the directory in which you ran it becomes a REST-accessible resource, as do all
of its files and subdirectories. When you use your web browser to make an HTTP request for one of
those resources, the server looks on disk for a file corresponding to the resource you requested and
serves it to you as part of the HTTP response.
Binding the web server to the special hostname localhost prevents people on the Internet at large, or
elsewhere on your local network, from using your web server (see Chapter 16 for more details). However,
anyone else on the computer you’re using can visit http://localhost:8000/ and see everything
you’re serving. If you’re running this script on a shared machine, make sure you run it from a directory
that doesn’t contain documents you don’t want to share.
When you’re ready to start serving web pages and applications to everyone on the Internet, you’ll need
to bind the web server to an external-facing hostname or IP address. Again, Chapter 16 has more infor-
mation on this.
463

51.5 Page 505

▲back to top
Chapter 21
The Visible Web Server
Because you’re already programming your own web servers, it’s not difficult to write one that enables you
to see your own sample HTTP request and response. Here’s a script called VisibleWebServer.py. It
includes a subclass of SimpleHTTPRequestHandler that does everything SimpleHTTPRequestHandler
does, but that also captures the text of the HTTP request and response and prints them to standard output.
When you make a request to EasyHTTPServer, it just prints out a little log message to the server’s stan-
dard output. When you hit the Visible Web Server, you get everything:
#!/usr/bin/python
from BaseHTTPServer import HTTPServer
from SimpleHTTPServer import SimpleHTTPRequestHandler
#The port of your local machine on which you want to run this web
#server. You’ll access the web server by visiting,
#e.g. “http://localhost:8000/”
PORT = 8000
class VisibleHTTPRequestHandler(SimpleHTTPRequestHandler):
“””This class acts just like SimpleHTTPRequestHandler, but instead
of logging only a summary of each hit to standard output, it logs
the full HTTP request and response.”””
def log_request(self, code=’-’, size=’-’):
“””Logs a request in great detail. This method is called by
SimpleHTTPRequestHandler.do_GET().”””
print self._heading(“HTTP Request”)
#First, print the resource identifier and desired operation.
print self.raw_requestline,
#Second, print the request metadata
for header, value in self.headers.items():
print header + “:”, value
def do_GET(self, method=’GET’):
“””Handles a GET request the same way as
SimpleHTTPRequestHandler, but also prints the full text of the
response to standard output.”””
#Replace the file object being used to output response with a
#shim that copies all outgoing data into a place we can see
#later. Then, give the actual work of handling the request to
#SimpleHTTPRequestHandler.
self.wfile = FileWrapper(self.wfile)
SimpleHTTPRequestHandler.do_GET(self)
#By this time, the shim file object we created previously is
#full of the response data, and is ready to be displayed. The
#request has also been displayed, since it was logged by
#log_request() (called by SimpleHTTPRequestHandler’s do_GET)
print “”
print self._heading(“HTTP Response”)
print self.wfile
464

51.6 Page 506

▲back to top
Web Applications and Web Services
def _heading(self, s):
“””This helper method formats a header string so it stands out
from the data beneath it.”””
line = ‘=’ * len(s)
return line + ‘\\n’ + s + ‘\\n’ + line
class FileWrapper:
“””This class wraps a file object, such that everything written to
the file is also silently appended to a buffer that can be printed
out later.”””
def __init__(self, wfile):
“””wfile is the file object to which the response is being
written, and which this class silently replaces.”””
self.wfile = wfile
self.contents = []
def __getattr__(self, key):
“””If someone tries and fails to get an attribute of this
object, they’re probably trying to use it as the file object
it replaces. Delegate to that object.”””
return getattr(self.wfile, key)
def write(self, s):
“””Write a string to the ‘real’ file and also append it to the
list of strings intended for later viewing.”””
self.contents.append(s)
self.wfile.write(s)
def __str__(self):
“””Returns the output so far as a string.”””
return ‘’.join(self.contents)
if __name__ == ‘__main__’:
httpd = HTTPServer((‘localhost’, PORT), VisibleHTTPRequestHandler)
httpd.serve_forever()
Note how even though SimpleHTTPRequestHandler wasn’t designed for its output to be intercepted,
it wasn’t terribly difficult to replace its output file with an impostor that does what you need. Python’s
operator overloading makes it easy for one object to impersonate another. Now let’s actually use this
script and consider a sample request and response.
Try It Out Seeing an HTTP Request and Response
Create a file called hello.html in the directory in which you put VisibleWebServer.py. Put the
following HTML code into the file:
<html>
<body>Hello, world!</body>
</html>
465

51.7 Page 507

▲back to top
Chapter 21
Start up VisibleWebServer.py and, using a web browser, visit the URL http://localhost:8000/
hello.html. In the standard output of the VisibleWebServer.py process, you should see output
much like the following:
============
HTTP Request
============
GET /hello.html HTTP/1.0
host: localhost:8000
accept-language: en
accept-encoding: gzip, compress
accept: text/*, */*;q=0.01
user-agent: Lynx/2.8.5rel.1 libwww-FM/2.14
=============
HTTP Response
=============
HTTP/1.0 200 OK
Server: SimpleHTTP/0.6 Python/2.3.4
Date: Thu, 24 Feb 2005 00:47:25 GMT
Content-type: text/html
Content-Length: 42
<html>
<body>Hello, world!</body>
</html>
How It Works
When you request hello.html, the HTTPServer object created by VisibleWebServer.py
spawns a VisibleHTTPRequestHandler object to handle your request. This does everything
that a SimpleHTTPRequestHandler spawned by EasyWebServer.py would do, but it also
makes sure the full text of the HTTP request and response are printed to standard output.
SimpleHTTPRequestHandler would have just printed a summary of the request.
If you use the FireFox or Mozilla web browser, you can install an extension that
will let you see portions of every HTTP request you make and every response you
get. The extension is called LiveHTTPHeaders, and it’s available from http://
livehttpheaders.mozdev.org/. This can be very useful in debugging web appli-
cations, but you can see only the headers, not the actual request or response data.
There are also several web applications that will make an HTTP request on your
behalf and show you the request and response. The most full-featured application
of this sort is Web-Sniffer, at http://web-sniffer.net/.
The HTTP Request
A HTTP request has two parts. The first line of the request is the command; it contains an HTTP verb, a
resource identifier, and (optionally) the version of HTTP being used:
GET /hello.html HTTP/1.0
466

51.8 Page 508

▲back to top
Web Applications and Web Services
Here the verb is GET and the resource identifier is /hello.html. (For these purposes, HTTP 1.0 and
HTTP 1.1 are basically the same, so don’t worry about that part.)
The second part of the HTTP request is a series of headers: key-value pairs describing the client and
providing additional information about the request:
host: localhost:8000
accept-language: en
accept-encoding: gzip, compress
accept: text/*, */*;q=0.01
user-agent: Lynx/2.8.5rel.1 libwww-FM/2.14
In the REST architecture, all information necessary to identify the resource should be kept in the identi-
fier. Because SimpleHTTPServer only serves static files, you’ll use /foo.html to uniquely identify one
file on disk. Another web server might be able to dynamically generate a representation of /foo.html
instead of just looking for a file on disk, but /foo.html would still identify one particular resource.
Though the identifier should completely identify the resource, the key-value pairs can be used to make
smaller-scale decisions about which representation of the resource to show — for instance, to send a local-
ized version of a document in response to the Accept-Language header. HTTP headers are also used to
regulate caching and to transmit persistent client state (that is, cookies) and authentication information.
Web browsers generally send HTTP headers with capitalized names like “User-Agent”, and that’s how
this chapter refers to particular headers. A quirk of the SimpleHTTPRequestHandler class means
that the Visible Web Server prints out header names in lowercase even if that’s not how they were
received, but it doesn’t matter much: HTTP headers are not case-sensitive. “User-Agent” and “user-
agent”are the same header.
The HTTP Response
The HTTP response tells the story of how the web server tried to fulfill the corresponding request. It
begins with a status code, which summarizes the response:
HTTP/1.0 200 OK
In this case, the response code was 200 (OK), which means everything went fine and your resource is
enclosed. Less desirable status codes you may have seen in your web browsing include the following:
❑ 403 (Forbidden), which means the resource might or might not exist but you’re not allowed to
receive it anyway
❑ 404 (File Not Found), the most famous HTTP status code that you’ll actually see in your
browser, this means the resource is just gone and has left no forwarding address, or was never
there
❑ 500 (Internal Server Error), which is often caused by a bug in a web application
All forty standard error codes are defined and categorized in RFC 2616, available at www.w3.org/
Protocols/rfc2616/rfc2616-sec10.html. Some of them are obscure, but it pays to know them.
For instance, the 204 response code, “No Content”, can be used in a web application to take action
when the user clicks a link, without making the user’s web browser load another page.
467

51.9 Page 509

▲back to top
Chapter 21
Following the status code are a number of headers, in the same key-value format as HTTP request headers:
Server: SimpleHTTP/0.6 Python/2.3.4
Date: Thu, 24 Feb 2005 00:47:25 GMT
Content-type: text/html
Content-Length: 42
Just as request headers contain information potentially useful to the web server, response headers contain
information potentially useful to the web browser. By far the most important HTTP response header is
“Content-Type”. Without this header, the web browser wouldn’t know how to display the document
being sent. The content type of /foo.html is text/html, which tells the web browser to render the rep-
resentation it receives as HTML. If the client had requested /foo.jpg instead, the content type would
have been image/jpeg, and the browser would have known to render the document as a graphic instead.
A blank line follows the response headers, and the rest of the response consists of the document being
delivered (if any). For a successful GET request, the document is the resource that was requested. For a
successful POST, PUT, or DELETE request, the result document is often the new version of the resource that
was changed, or a status message talking about the success of the operation. An unsuccessful operation
often results in an HTTP response containing a document describing the error and possibly offering help.
Web applications are considered more or less “RESTful” depending on how well they employ the fea-
tures of HTTP. There are no hard-and-fast rules for this, and sometimes convenience wins out over
RESTfulness, but HTTP has conventions, and you might as well use them to your advantage instead of
reinventing them unnecessarily. Some rules of thumb for designing RESTful interfaces follow:
❑ Keep resource identifiers transparent. A user should be able to figure out what kind of resource
is on the other end of a resource identifier just by looking at it. The biggest challenge to achiev-
ing this is designing the resource identifier so that it holds all of the information necessary to
uniquely identify the resource.
❑ On the other hand, don’t put something into the resource identifier if it doesn’t help identify a
resource. Ask the user to provide that information in an HTTP header instead, or in the data of
a POST, DELETE, or PUT request.
❑ Don’t put something into the data of a POST, DELETE, or PUT request if it makes sense to put it
into one of the standard HTTP headers. For instance, authentication information can be submit-
ted through HTTP authentication. If you make a resource available in multiple formats, you can
have clients use the HTTP header “Accept” to specify which one they want.
❑ Don’t return a status code of 200 (“OK”) on an error, unless there’s really no HTTP error that
conveys the problem. 500 (problem on the server end) and 400 (problem on the user end) are
good general-purpose errors. One problem with this rule is that browsers such as Internet
Explorer may show their own generic error screen if they receive an error code other than 200,
blocking a document you might have generated to help the user with her specific problem.
CGI: Turning Scripts into Web Applications
Using different web browsers and resources, experiment with the Visual Web Server until it becomes
boring. Unless you find this whole topic boring, this encroaching ennui probably means you’re pushing
the limits of what’s to be learned from examining HTTP requests and responses. Fortunately, it gets
much more interesting very quickly: The next phase is the dynamic world of web applications.
468

51.10 Page 510

▲back to top
Web Applications and Web Services
REST is easy to implement when you’re just serving files off of a hard disk, but that only covers the part
of REST whereby you request resources. Representations, the means by which you create, modify, and
delete resources, don’t come into the picture at all. Although a set of static HTML files is technically a
web application, it’s not a very interesting one.
You can handle the transfer of representations and the creation of dynamic applications in a number of
ways, but the venerable standard is the Common Gateway Interface (CGI). CGI was developed in the
early 1990s and has remained more or less the same since its creation. The goal of CGI is to enable some-
one to write a script that can be invoked from an HTTP request, without having to know anything about
web server programming. A web server that supports CGI is capable of transforming certain HTTP
requests into script invocations.
The CGI standard is hosted at http://hoohoo.ncsa.uiuc.edu/cgi/. The page hasn’t changed
since 1996, but neither has CGI.
Because CGI is implemented inside the web server, it must be enabled through web server configura-
tion. The setup of CGI is highly dependent on the brand of web server and on your system administra-
tor’s idea of how a system should be administrated. Even different Linux distributions have different
out-of-the-box setups for CGI. Rather than give comprehensive instructions for all contingencies, or
evade the issue altogether and assume you can get it working, following are a few lines of Python that
implement a simple CGI server; save this under the name of EasyCGIServer.py. This server can be
used for all of the CGI examples in this chapter. Once again, a built-in Python module makes it easy.
from BaseHTTPServer import HTTPServer
from CGIHTTPServer import CGIHTTPRequestHandler
HTTPServer((‘localhost’, 8000), CGIHTTPRequestHandler).serve_forever()
The code is as simple as that for EasyWebServer; in fact, it’s nearly identical. The only new feature
EasyCGIServer supports is special treatment of the cgi-bin directory, which is where CGI scripts are
kept.
Try It Out Running a CGI Script
Create a directory called cgi-bin beneath the directory in which you keep EasyWebServer.py and
EasyCGIServer.py. Put the following code in the file cgi-bin/hello.cgi:
#!/usr/bin/python
print “Content-type: text/plain\\n”
print “Hello, world!”
The filenames of all the CGI scripts in this chapter will have the .cgi extension. This visually distin-
guishes the CGI scripts from the regular Python scripts, and makes it possible to run them on web
servers that will only execute a CGI script if it has a .cgi extension.
If you’re on a Unix-based system, you’ll also need to make hello.cgi editable with the chmod command:
# chmod u+x ./cgi-bin/hello.cgi
Run hello.cgi from the command line to make sure the script works:
469

52 Pages 511-520

▲back to top

52.1 Page 511

▲back to top
Chapter 21
# ./cgi-bin/hello.cgi
Content-type: text/plain
Hello, world!
Start EasyWebServer.py and use a web browser to visit http://localhost:8000/cgi-bin/hello.cgi.
Either your web browser will invite you to download hello.cgi as a Python script, or you will see the
source code to hello.cgi as plaintext in your web browser:
#!/usr/bin/python
print “Content-type: text/plain\\n”
print “Hello, world!”
Kill EasyWebServer.py and start up EasyCGIServer.py instead. In your web browser, reload
http://localhost:8000/cgi-bin/hello.cgi. You should see the string “Hello, world!” as plaintext
in your web browser:
Hello, world!
How It works
When you requested /cgi-bin/hello.cgi through EasyWebServer, the server interpreted it the way
EasyWebServer interprets every request: as a request for a static file to be found on disk. What you
received was the contents of the static file /cgi-bin/hello.cgi.
When you requested the same resource through EasyCGIServer, the server interpreted it differently.
Instead of treating hello.cgi as a file to be read, EasyCGIServer treated it as a script to be run. The
script was executed as from the command line, and its output was used to create the HTTP response.
What you saw in your web browser was the content part of the HTTP response, rendered according to the
Content-Type header provided by the script. Any executable .py or .cgi script you put into cgi-bin/
will be run by EasyWebServer when requested, and its output will be used to create an HTTP response.
The Web Server Makes a Deal with the CGI Script
The CGI standard specifies a deal that a CGI-enabled web server makes with any file it chooses to inter-
pret as a CGI script. The web server is responsible for receiving and parsing the HTTP request, for rout-
ing the request to the correct script, and for executing that script just as you might execute a Python
script from the command line. It’s also responsible for modifying the script’s runtime environment to
include CGI-specific variables, whose values correspond to information about the runtime environment,
and information found in the HTTP request. For instance, the User-Agent header becomes the environ-
ment variable HTTP_USER_AGENT, and the HTTP verb invoked by the request becomes the environment
variable HTTP_METHOD. As with any other environment variables, these special variables can be accessed
through the os.environ dictionary, and the script can use them to evaluate the HTTP request.
In return for this service, the CGI script is expected to take over the duties of the web server for the dura-
tion of that HTTP session. Anything the script writes to standard output is output as part of the HTTP
response. This means that in addition to producing a document of some kind, the script needs to output
any necessary HTTP headers as a preface to the document. At the very least, every CGI script must out-
put the Content-type HTTP header.
470

52.2 Page 512

▲back to top
Web Applications and Web Services
If you’re having trouble getting a script to work through the web browser, you can try setting the appro-
priate CGI environment variables manually and executing the script from the command line.
CGI’s Special Environment Variables
Your script might find more than 20 special CGI variables in its environment. The important ones are
covered a bit later, but first look at a very simple CGI script that gives you the tools you need to explore
the variables yourself. It’s called PrintEnvironment.cgi:
#!/usr/bin/python
import os
import cgitb
cgitb.enable()
The cgitb module will give you exception reporting and stack tracebacks in your web browser, similar
to what you see when a command-line Python script throws an exception. It’ll save you from getting
mysterious 500 error codes, and from having to look through web server logs to find the actual error
message. The cgitb module is available only in Python versions 2.2 and later:
#Following is a list of the environment variables defined by the CGI
#standard. In addition to these 17 predefined variables, each HTTP
#header in the request has a corresponding variable whose name begins
#with “HTTP_”. For instance, the value of the “User-Agent” header is
#kept in “HTTP_USER_AGENT”.
CGI_ENVIRONMENT_KEYS = [ ‘SERVER_SOFTWARE’,
‘SERVER_NAME’,
‘GATEWAY_INTERFACE’,
‘SERVER_PROTOCOL’,
‘SERVER_PORT’,
‘REQUEST_METHOD’,
‘PATH_INFO’,
‘PATH_TRANSLATED’,
‘SCRIPT_NAME’,
‘QUERY_STRING’,
‘REMOTE_HOST’,
‘REMOTE_ADDR’,
‘AUTH_TYPE’,
‘REMOTE_USER’,
‘REMOTE_IDENT’,
‘CONTENT_TYPE’,
‘CONTENT_LENGTH’ ]
#First print the response headers. The only one we need is Content-type.
print “Content-type: text/plain\\n”
#Next, print the environment variables and their values.
print “Here are the headers for the request you just made:”
for key, value in os.environ.items():
if key.find(‘HTTP_’) == 0 or key in CGI_ENVIRONMENT_KEYS:
print key, “=>”, value
471

52.3 Page 513

▲back to top
Chapter 21
Put this file in your cgi-bin/ directory, make it executable, and visit http://localhost:8000/
cgi-bin/PrintEnvironment.cgi. You should see something like the following:
Here are the headers for the request you just made:
SERVER_SOFTWARE => SimpleHTTP/0.6 Python/2.3.4
REQUEST_METHOD => GET
PATH_INFO =>
SERVER_PROTOCOL => HTTP/1.0
QUERY_STRING =>
CONTENT_LENGTH =>
SERVER_NAME => rubberfish
PATH_TRANSLATED => /home/leonardr/LearningPython/listings
SERVER_PORT => 8000
CONTENT_TYPE => text/plain
HTTP_USER_AGENT => Lynx/2.8.5rel.1 libwww-FM/2.14
HTTP_ACCEPT => text/html, text/plain, text/rtf, text/*, */*;q=0.01
GATEWAY_INTERFACE => CGI/1.1
SCRIPT_NAME => /cgi-bin/PrintEnvironment.py
REMOTE_ADDR => 127.0.0.1
REMOTE_HOST => rubberfish
With the PrintEnvironment.py file in place, you’re defining a resource with the identifier http://
localhost:8000/cgi-bin/PrintEnvironment.cgi. When you run EasyCGIServer, this resource is
defined by the output you get when you run the Python code in PrintEnvironment.cgi; and, depend-
ing on the content of your request, it can be different every time you hit that URL.
PrintEnvironment.cgi contains an enumeration of the defined CGI environment
variables and only prints the values of those variables. The purpose of this is
twofold: to put that information where you’ll see it and to avoid leaking information
that might be contained in other irrelevant environment variables.
EasyCGIServer inherits the environment of the shell you used to run it; this means
that if you run EasyCGIServer instead of Apache or another web server, a version
of PrintEnvironment.cgi that printed the whole environment would print PATH
and all the other environment variables in your shell. This information would
swamp the legitimate CGI variables and possibly disclose sensitive information
about your user account. Remember that any web servers you set up on your com-
puter can be accessed by anyone else on the same machine, and possibly by the
Internet at large. Don’t expose information about yourself unnecessarily.
A few of the CGI-specific environment variables deserve further scrutiny here:
❑ REQUEST_METHOD is the HTTP verb corresponding to the REST method you used against
this resource. Because you were just trying to retrieve a representation of the resource, you used
the GET HTTP verb.
❑ QUERY_STRING and PATH_INFO are the two main ways in which a resource identifier
makes it into a CGI script. You can experiment with these two variables by accessing
PrintEnvironment.cgi in different ways. For instance, GETting the resource identifier
/cgi-bin/PrintEnvironment.cgi/pathInfo/?queryString will set PATH_INFO to
472

52.4 Page 514

▲back to top
Web Applications and Web Services
pathInfo/ and QUERY_STRING to queryString. The strange-looking, hard-to-understand
URLs you often see when using web applications are usually long QUERY_STRINGs.
❑ HTTP_USER_AGENT is a string provided by the web browser you used to access the page,
which corresponds to the “User-Agent” HTTP header and which is supposed to identify the
web browser you’re using. It’s interesting as an example of an HTTP header being transformed
into a CGI environment variable. Another such variable is HTTP_REFERER, derived from the
“Referer” HTTP header. The “Referer” header is provided whenever you click a link from one
page to another, so that the second page knows how you accessed it.
Accepting User Input through HTML Forms
It’s possible to manipulate the output of PrintEnvironment.cgi enough to prove that it serves
dynamic resources, but the interface to it isn’t that good. To get different text back, you have to use
different web browsers, hack the URL (that is, request different resources) or do even weirder things.
Most web applications eschew this type of interface in favor of one based on HTML forms. You can
make a lot of useful web applications just by writing simple CGIs that print HTML forms and read the
QUERY_STRING and PATH_INFO variables.
A brief recap of HTML forms seems appropriate here, as the forms are only relevant to web applications.
Even if you already know HTML, it’s useful to place HTML forms in the context of the REST architecture.
An HTML form is enclosed within <FORM> tags. The opening <FORM> tag has two main attributes:
action, which contains the identifier of the CGI script to call or the resource to be operated upon, and
method, which contains the HTTP verb to be used when submitting the form.
Between the opening <FORM> tag and the closing </FORM> tab, special HTML tags can be used, which
a web browser renders as GUI controls. The GUI controls available include text boxes, checkboxes,
radio button groups, buttons that activate form submission (all achieved with the INPUT tag), large text
entry fields (the TEXTAREA tag), and drop-down selection boxes (the SELECT tag). Figure 21-1 shows an
example of a very simple HTML form, along with the set of GUI controls it causes to be rendered in a
web browser.
If you put that HTML in a file called SimpleHTMLForm.html in the root directory of your EasyCGIServer
installation, you can retrieve it via the URL http://localhost:8000/SimpleHTMLForm.html. Because
it’s not a CGI script, EasyCGIServer will serve it as a static file, just as EasyWebServer would. If you
then click the Submit button, the form data will be encoded by the web browser into a GET request, and
submitted to a resource with a long identifier beginning with /cgi-bin/PrintFormSubmission.cgi.
Unfortunately, there’s nothing on disk — no file and no script — corresponding to that resource identifier,
so instead of doing anything useful, the web server is going to return a “page not found” error document
(status code: the famous 404). With Python’s cgi module, though, it’s easy to put a script in place that will
take the form submission and do something with it.
Html forms’ limited vocabulary
The only HTTP verbs supported by HTML forms are GET, for reading a resource, and
POST, for writing to a resource. A form action of PUT or DELETE is invalid HTML, and
most web browsers will submit a POST request instead. As you’ll see, this puts a bit of a
kink in the implementation of REST-based web applications, but it’s not too bad.
473

52.5 Page 515

▲back to top
Chapter 21
Figure 21-1
The cgi Module: Parsing HTML Forms
When you click one of the Submit buttons on SimpleHTMLForm.html, notice that you’re not exactly
GETting the resource /cgi-bin/PrintFormSubmission.cgi, the resource specified in the action
attribute of the <FORM> tag. You’re GETting a slightly different resource, something with the long, unwieldy
identifier of /cgi-bin/PrintFormSubmission.cgi?textField=Some+text&radioButton=
2&button=Submit.
This is how a GET form submission works: The web browser gathers the values of the fields in the form
you submitted and encodes them so they don’t contain any characters not valid in a URL (for instance,
spaces are replaced by plus signs). It then appends the field values to the form destination, to get the
actual resource to be retrieved. Assuming there’s a CGI at the other end to intercept the request, the CGI
will see that encoded form information in its QUERY_STRING environment variable. A similar encoding
happens when you submit a form using the POST verb, but in that case the form data is sent as part of
the data, not as part of the resource identifier. Instead of being made available to the script in environ-
ment variables, POSTed data is made available on standard input.
The cgi module knows how to decode the form data present in HTTP requests, whether the request
uses GET or POST. The cgi module can obtain the data from environment variables (GET) or standard
input (POST), and use it to create a reconstruction of the original HTML form in a class called
FieldStorage.
FieldStorage can be accessed just like a dictionary, but in Python 2.2 and later, the safest way to use it
is to call its getfirst() method, passing in the name of the field whose value you want.
In versions of Python prior to 2.2, the getfirst method is not available. Instead, to be safe you need to
simulate getfirst with code like the following:
fieldVal = form.getValue(“field”)
if isinstance(fieldVal, list): #More than one “field” was submitted.
fieldVal = fieldVal[0]
When you’re actually expecting multiple values for a single CGI variable, use the _getlist_ method
instead of getfirst to get all the set values.
474

52.6 Page 516

▲back to top
Web Applications and Web Services
Safety when accessing form values
Why is form.getfirst(‘fieldName’) safer than form[‘fieldName’]? The root of
the problem is that sometimes a single form submission can legitimately provide two
or more values for the same field (for instance, this happens when a user selects more
than one value of a selection box that allows multiple selections). If this happens,
form[‘fieldName’] will return a list of values (e.g., all the selected values in the
multiple-selection box) instead of a single value. This is fine as long as your script is
expecting it to happen, but because users have complete control of the data they sub-
mit to your CGI script, a malicious user could easily submit multiple values for a field
in which you were only expecting one.
If someone pulls that trick on you and your script is using form[‘fieldName’], you’ll
get a list where you were expecting a single object. If you treat a list as though it were a
single object your script will surely crash. That’s why it’s safer to use getfirst: It is
always guaranteed to return only the first submitted value, even if a user is trying to
crash your script with bad data.
Now that you know about the FieldStorage object, it’s easy to write the other half of SimpleHTMLForm.
html: PrintFormSubmission.cgi, a CGI script that prints the values it finds in the form’s fields:
#!/usr/bin/python
import cgi
import cgitb
cgitb.enable()
form = cgi.FieldStorage()
textField = form.getfirst(“textField”)
radioButton = form.getfirst(“radioButton”)
submitButton = form.getfirst(“button”)
print ‘Content-type: text/html\\n’
print ‘<html>’
print ‘<body>’
print ‘<p>Here are the values of your form submission:</p>’
print ‘<ul>’
print ‘<li>In the text field, you entered “%s”.</li>’ % textField
print ‘<li>Of the radio buttons, you selected “%s”.’ % radioButton
print ‘<li>The name of the submit button you clicked is “%s”.’ % submitButton
print ‘</ul>’
print ‘</body>’
print ‘</html>’
Now, when you click the submit button on SimpleHTMLForm.html, instead of getting a 404 Not Found
error, you’ll see something similar to what is shown in Figure 21-2.
Figure 21-2
475

52.7 Page 517

▲back to top
Chapter 21
So far so good. Let’s go a little further, though, and create a script capable of printing out any form sub-
mission at all. That way, you can experiment with HTML forms of different types. To get you started,
let’s have the new script print out a fairly complex HTML form when you hit it without submitting a
form to it. The script that follows deserves to be called PrintAnyFormSubmission.cgi:
#!/usr/bin/python
import cgi
import cgitb
import os
cgitb.enable()
form = cgi.FieldStorage()
print ‘Content-type: text/html\\n’
print ‘<html>’
print ‘<body>’
if form.keys():
verb = os.environ[‘REQUEST_METHOD’]
print ‘<p>Here are the values of your %s form submission:</p>’ % verb
print ‘<ul>’
for field in form.keys():
valueObject = form[field]
if isinstance(valueObject, list):
#More than one value was submitted. We therefore have a
#whole list of ValueObjects. getlist() would have given us
#the string values directly.
values = [v.value for v in valueObject]
if len(values) == 2:
connector = ‘“ and “‘ #’”Foo” and “bar”’
else:
connector = ‘“, and “‘ #’”Foo”, “bar”, and “baz”’
value = ‘“, “‘.join(values[:-1]) + connector + values[-1]
else:
#Only one value was submitted. We therefore have only one
#ValueObject. getfirst() would have given us the string
#value directly.
value = valueObject.value
print ‘<li>For <var>%s</var>, I got “%s”</li>’ % (field, value)
else:
print ‘’’<form method=”GET” action=”%s”>
<p>Here’s a sample HTML form.</p>
<p><input type=”text” name=”textField” value=”Some text” /><br />
<input type=”password” name=”passwordField” value=”A password” />
<input type=”hidden” name=”hiddenField” value=”A hidden field” /></p>
<p>Checkboxes:
<input type=”checkbox” name=”checkboxField1” checked=”checked” /> 1
<input type=”checkbox” name=”checkboxField2” selected=”selected” /> 2
</p>
<p>Choose one:<br />
<input type=”radio” name=”radioButton” value=”1” /> 1<br />
476

52.8 Page 518

▲back to top
Web Applications and Web Services
<input type=”radio” name=”radioButtons” value=”2” checked=”checked” /> 2<br />
<input type=”radio” name=”radioButtons” value=”3” /> 3<br /></p>
<textarea name=”largeTextEntry”>A lot of text</textarea>
<p>Choose one or more: <select name=”selection” size=”4” multiple=”multiple”>
<option value=”Option 1”>Option 1</option>
<option value=”Option 2” selected=”selected”>Option 2</option>
<option value=”Option 3” selected=”selected”>Option 3</option>
<option value=”Option 4” selected=”selected”>Option 4</option>
</select></p>
<p><input type=”Submit” name=”button” value=”Submit this form” />
<p><input type=”Submit” name=”button” value=”Submit this form (Button #2)” />
</form>’’’ % os.environ[‘SCRIPT_NAME’]
print ‘</body>’
print ‘</html>’
Try It Out Printing Any HTML Form Submission
Put PrintAnyFormSubmission.cgi in your cgi-bin/ directory and start up EasyCGIServer. Visit
http://localhost:8000/cgi-bin/PrintAnyFormSubmission.cgi. You’ll be given an HTML form
that looks something like what is shown in Figure 21-3.
Figure 21-3
477

52.9 Page 519

▲back to top
Chapter 21
Change any of the form data you want and click one of the Submit buttons. You’ll be taken to a screen
that looks like the one shown in Figure 21-4.
Figure 21-4
How It Works
When you first request the resource identified by /cgi-bin/PrintAnyFormSubmission.cgi, the
script uses the cgi module to look for a form submission. Because there are no form variables, it
assumes you didn’t submit a form at all and presents the default resource: a fairly complex HTML form
for you to play with.
When you click one of the Submit buttons, you request a very different resource: something like /cgi-
bin/PrintAnyFormSubmission.cgi?textField=Some+text&passwordField=A+password&hidden
Field=A+hidden+field&checkboxField1=on&radioButtons=2&largeTextEntry=A+lot+of+text
&selection=Option+2&selection=Option+3&selection=Option+4&button=Submit+this+form+
%28Button+%232%29. This time, the cgi module picks up a lot of form variables and outputs a dynami-
cally generated resource that iterates over the submitted form variables to describe the form you submit-
ted. If you submit the form again with different values, you’re requesting a slightly different resource and
the HTML output by the script will be different in corresponding ways.
If you’re new to web programming, note especially that even though there was a checkboxField2
field in the form submitted, there’s no mention of it in the description of the form submission. Web
browsers don’t encode unchecked checkboxes into the form submission, so they don’t show up at all in
the FieldStorage object. This can be a little annoying.
You can use SimpleHTMLForm.html against this script as well as against PrintFormSubmission.cgi.
In fact, you can use any form at all against this script, including forms designed for other web applications,
as long as you change the form’s action attribute to point to /cgi-bin/PrintFormSubmission.cgi.
However, if you don’t provide any inputs at all (i.e., you GET the base resource /cgi-bin/Print
FormSubmission.cgi), you’ll be given the default HTML form. This pattern — a CGI script that, when
invoked with no arguments, prints its own form — is a powerful tool for building self-contained applica-
tions. Note also how the script uses the special CGI-provided environment variable SCRIPT_NAME to refer
to itself. Even if you name this script something else or put it in another directory, the form it generates will
still refer to itself.
478

52.10 Page 520

▲back to top
Web Applications and Web Services
Like the EasyHTTPServer, PrintAnyFormSubmission.cgi is a good way to experiment with a new
concept, but it gets boring quickly. It’s time to move on to something more interesting: a real web
application.
Building a Wiki
With a basic knowledge of REST, the architecture of the web; and CGI, the main way of hooking up pro-
grams to that architecture, you’re ready to design and build a basic application. The next few pages will
detail the construction of a simple content management system called a wiki.
The wiki was invented in 1995 by Ward Cunningham and is best known today as the base for Wikipedia
(www.wikipedia.org), a free online encyclopedia (see Figures 21-5 and 21-6). Cunningham’s original
wiki (http://c2.com/cgi/wiki/) is still popular among programmers, containing information on
and discussion of technical and professional best practices. Of course, there’s also the REST wiki men-
tioned earlier.
Figure 21-5
479

53 Pages 521-530

▲back to top

53.1 Page 521

▲back to top
Chapter 21
Figure 21-6
The most distinctive features of wikis are as follows:
❑ Open, web-based editing — Some content management systems require special software or a
user account to use, but wiki pages are editable through any web browser. On most wikis, every
page is open to editing by anyone at all. Because of problems with spam and vandalism, some
wikis have begun to require user accounts. Even in wikis that distinguish between members
and nonmembers, though, the norm is that any member can edit any page. This gives wikis an
informal feel, and the near lack of barriers to entry encourages people to contribute.
❑ A flat namespace of pages — Each page in a wiki has a unique name. Page names are often
WikiWords, strings formed by capitalizing several words (the title of the page) and pushing
them together. That is, WikiPageNames OftenLookLikeThis. There is no directory structure in a
wiki; all pages are served from the top level. Pages are organized through the creation of addi-
tional pages to serve as indexes and portals.
❑ Linking through citing — One wiki page can link to another simply by mentioning its
WikiWord name in its own body. When a page is rendered, all WikiWords cited therein are
linked to the corresponding pages. A page may reference a WikiWord for which no page yet
exists: At rendering time, such a reference is linked to a form for creating the nonexistent page.
Wikis that don’t name their pages with WikiWords must define some other convention for link-
ing to another page in the same wiki.
❑ Simple, text-based markup — Rather than require the user to input HTML, wikis employ a few
simple rules for transforming ASCII text into the HTML displayed when a page is rendered.
480

53.2 Page 522

▲back to top
Web Applications and Web Services
Sample rules include the use of a blank line to signify a new paragraph, and the use of *aster-
isks* to bold a selection. Unfortunately, these conventions are only informal, and there are no
hard-and-fast rules So, the specific rules differ widely across the various wiki implementations.
See http://c2.com/cgi/wiki?WikiDesignPrinciples for Cunningham’s original Wiki
design principles.
Sample applications often lack important features necessary to make the application fit for actual use.
An online store application presented within the context of this chapter would be too complex to be
easily understood, yet not complete enough to actually use to run an online store. Because the defining
features of a wiki are so few and simple, it’s possible to design, build, and explain a fully fledged wiki in
just a few pages. BittyWiki, the application designed and built in this chapter according to the principles
just described, weighs in at under 10 kilobytes, but it’s not the shortest wiki written in Python.
See http://infomesh.net/2003/wypy/wypy.txt for a wiki written in only 814 characters and
11 lines of Python. It’s acutely painful to behold.
The BittyWiki Core Library
Before writing any code, you need to make a couple of design decisions about the nature of the wiki you
want to create. In the following examples, the design decisions made are the ones that lead to the sim-
plest wiki back-end: after all, for the purposes of this discussion, the important part of BittyWiki is the
interface it presents to the web, not the back-end.
Back-end Storage
Wiki implementations store their pages in a variety of ways. Some keep their files on disk, some in a
database, and some in a version control repository so that users can easily revert vandalism. For simplic-
ity’s sake, a BittyWiki installation will keep a page on a disk file named after that page. All of a given
wiki’s pages will be kept in the same directory. Because the wiki namespace is flat, no subdirectories
are needed.
WikiWords
Each wiki implementation that uses WikiWords must decide which strings are valid names of wiki
pages, so that it can automatically link citations of those pages. BittyWiki will use one of the simplest
WikiWord definitions: It will treat as a WikiWord any string of letters and numbers that begins with a
capital letter and contains at least two capitals. “WikiWord” is itself a WikiWord, as are “WikiWord2,”
“WikiworD,” “WWW,” and “AI.”
Any wiki page can be retrieved by name, but you also need a default page for when no name is speci-
fied. The default page will be the one called “HomePage.”
Writing the BittyWiki Core
On the basis of those design decisions, it’s now possible to write the core of BittyWiki: the code that
reads from and writes to the back-end, and that processes the WikiWord links. Put this code into
BittyWiki.py, in your cgi-bin/ directory or somewhere in your PYTHON_PATH:
481

53.3 Page 523

▲back to top
Chapter 21
“””This module implements the BittyWiki core code: that which is not
bound to any particular interface.”””
import re
import os
class Wiki:
“A class representing a wiki as a whole.”
HOME_PAGE_NAME = “HomePage”
def __init__(self, base):
“Initializes a wiki that uses the provided base directory.”
self.base = base
if not os.path.exists(self.base):
os.makedirs(self.base)
elif not os.path.isdir(self.base):
raise IOError(‘Wiki base “%s” is not a directory!’ % self.base)
def getPage(self, name=None):
“””Retrieves the given page for this wiki, which may or may not
currently exist.”””
if not name:
name = self.HOME_PAGE_NAME
return Page(self, name)
class Page:
“””A class representing one page of a wiki, containing all the
logic necessary to manipulate that page and to determine which other
pages it references.”””
#We consider a WikiWord any word beginning with a capital letter,
#containing at least one other capital letter, and containing only
#alphanumerics.
WIKI_WORD_MATCH = “(([A-Z][a-z0-9]*){2,})”
WIKI_WORD = re.compile(WIKI_WORD_MATCH)
WIKI_WORD_ALONE = re.compile(‘^%s$’ % WIKI_WORD_MATCH)
def __init__(self, wiki, name):
“””Initializes the page for the given wiki with the given
name, making sure the name is valid. The page may or may not
actually exist right now in the wiki.”””
#WIKI_WORD matches a WikiWord anywhere in the string. We want to make
#sure the page is a WikiWord and nothing else.
if not self.WIKI_WORD_ALONE.match(name):
raise NotWikiWord, name
self.wiki = wiki
self.name = name
self.path = os.path.join(self.wiki.base, name)
def exists(self):
“Returns true if there’s a page for the wiki with this name.”
return os.path.isfile(self.path)
482

53.4 Page 524

▲back to top
Web Applications and Web Services
def load(self):
“Loads this page from disk, if it exists.”
if not hasattr(self, ‘text’):
self.text = ‘’
if self.exists():
self.text = open(self.path, ‘r’).read()
def save(self):
“Saves this page. If it didn’t exist before, it does now.”
if not hasattr(self, ‘text’):
self.text = ‘’
out = open(self.path, ‘w’)
out.write(self.text)
out.close()
def delete(self):
“Deletes this page, assuming it currently exists.”
if self.exists():
os.remove(self.path)
def getText(self):
“Returns the raw text of this page.”
self.load()
return self.text
class NotWikiWord(Exception):
“””Exception thrown when someone tries to pass off a non-WikiWord
as a WikiWord.”””
pass
Try It Out Creating Wiki Pages from an Interactive Python Session
In just a bit, you’re going to give BittyWiki a web interface, and spend much of the rest of the chapter
accessing it via HTTP. The easiest way to get used to the basic API, however, is to play with BittyWiki
from an interactive Python session — no web interface needed:
>>> from BittyWiki import Wiki
>>> wiki = Wiki(“localwiki”)
>>> homePage = wiki.getPage()
>>> homePage.text = “Here’s the home page.\\n\\nIt links to PageTwo and PageThree.”
>>> homePage.save()
The localwiki directory now contains your wiki’s files:
>>> #The “localwiki” directory now contains your wiki’s files.
>>> import os
>>> open(os.path.join(“localwiki”,”HomePage”)).read()
“Here’s the home page.\\n\\nIt links to PageTwo and PageThree.”
HomePage references other pages in the wiki, but none of them exist yet:
>>> page2 = wiki.getPage(“PageTwo”)
>>> page2.exists()
False
483

53.5 Page 525

▲back to top
Chapter 21
Of course, we can create one of those pages:
>>> page2.text = “Here’s page 2.\\n\\nIt links back to HomePage.”
>>> page2.save()
>>> page2.exists()
True
Finally, a look at the NotWikiWord exception:
>>> wiki.getPage(“Wiki”)
Traceback (most recent call last):
File “<stdin>”, line 1, in ?
File “BittyWiki.py”, line 25, in getPage
return Page(self, name)
File “BittyWiki.py”, line 47, in __init__
raise NotWikiWord, name
BittyWiki.NotWikiWord: Wiki
The BittyWiki Web Interface
The BittyWiki library provides a way to manipulate the wiki, but it has no user interface. You can write
standalone scripts to manipulate the repository, or create pages from an interactive prompt, but wikis
were intended to be used over the web. Another set of design decisions awaits, related to how BittyWiki
should expose the wiki pages and operations as REST resources.
Resources
Because REST is based on resources, the first thing to consider when designing a web application is the
nature of the resources to provide. A wiki provides only one type of resource: pages out of a flat namespace.
Information in the URL path is easier to read than keeping it in the string, so a wiki page should be
retrieved by sending a GET request to the CGI, appending the page name to the CGI path. The resulting
resource identifier looks like /bittywiki.cgi/PageName. To modify a page, a POST request should be
sent to its resource identifier.
The allowable operations on a wiki page are as follows: creating one, reading one, updating one, and
deleting one. These four operations are so common to different types of resource that they have their
own acronym (CRUD), used to describe the many applications designed for performing those opera-
tions. A wiki is a web-based CRUD application for named pages of text kept in a flat namespace.
Most wikis either implement page delete as a special administrator command, or don’t implement it at
all; this is because a page delete command makes vandalism very easy. BittyWiki’s naivete with respect
to the delete command is perhaps its least realistic feature.
Request Structure
Not by coincidence, the CRUD operations correspond to the four main HTTP verbs: Recall that the same
four operations show up repeatedly, whether the subject is databases, file system access, or web resources.
Ideally, one CRUD operation would map to one HTTP verb.
484

53.6 Page 526

▲back to top
Web Applications and Web Services
When users request a page for reading, the only information they must provide is the page name.
Therefore, for the read operation, no additional information must be tacked on to the resource identifier
defined in the previous section. A simple GET to the resource identifier will suffice.
When modifying a page, it’s necessary to send not only the name of the page but its desired new con-
tents. POSTing the data to the resource identifier should suffice to do that.
Now you run into a problem: You have two more operations (create and delete), but only one HTTP
method (POST) is both suitable for those operations and also supported by the HTML forms that will
make up your interface. These operations must be consolidated somehow.
It makes no sense to “create” a page that already exists or to “edit” a nonexistent page, so those two
operations could be combined into a single write operation. There are still two actions (write and
delete) to go through POST, so the problem remains.
The solution is to have users put a marker in their POST data to indicate which operation they want to
perform, rather than just post the data they want to use in the operation. The key for this marker will be
operation, and the allowable values will be write and delete.
But Wait — There’s More (Resources)
So far, the design assumes that the write and delete actions are triggered in response to HTML form
submissions. Where are those HTML forms going to come from? Because the forms need to be dynami-
cally generated based on the name of the page they’re modifying, they must be generated by the wiki
program. This makes them a new type of resource. Contrary to what was stated earlier, BittyWiki actu-
ally serves two types of resources. Its primary job is to serve pages, but it must also serve HTML forms
for manipulating those pages.
Unlike pages, forms can’t be created, updated, or deleted by the user: they can only be read. (After
they’re read, however, they can be used to create, update or delete pages.) The forms should therefore
be accessible through GET URLs.
Because the user will be requesting a form to write or delete a particular page, it makes sense to base the
resource identifier for the form on that of the page. There are two ways of doing this. The first is to continue
to append to the PATH_INFO of the identifier, so that the form to delete the page at /bittywiki.cgi/
MyPage is located at /wiki.cgi/MyPage/delete. The other way is to use the QUERY_STRING, so that that
form is located at /wiki.cgi/MyPage?operation=delete.
There’s no general right or wrong solution. However, because the “operation” keyword is already in
use for the POST form submissions, and because the pages (not the forms) are the real point of a wiki,
BittyWiki will implement the second strategy. The possible values will be the same as for the POST com-
mands: write and delete.
To summarize: Each wiki page in BittyWiki boasts three associated resources. Each resource might
behave differently in response to a GET and a POST, as shown in the following table.
485

53.7 Page 527

▲back to top
Chapter 21
Resource
/bittywiki.cgi/PageName
/bittywiki.cgi/PageName?operation=write
/bittywiki.cgi/PageName?operation=delete
What GET does
What POST does
Displays the page if
it exists; displays create
form if not
Displays edit form
Displays delete form
Nothing
Writes page, provides
status
Deletes page,
provides status
If no page name is specified (that is, someone GETs the bare resource /bittywiki.cgi/), the CGI will
ask the core wiki code to retrieve the default page.
There are tradeoffs to consider when you’re designing your resource identifiers and weighing PATH_INFO
against QUERY_STRING. Both “/foo.cgi/clients/MegaCorp” and “/foo.cgi?client=MegaCorp”
are legitimate REST identifiers for the same resource. The advantage of the first one is that it looks a lot
nicer, more like a “real” resource. If you want to give the appearance of hierarchy in your data structure,
nothing does it as well as a PATH_INFO-based identifier scheme.
The problem is that you can’t use that scheme in conjunction with an HTML form that lets you, for
example, select MegaCorp from a list of clients. The destination of an HTML form needs to be defined at
the time the form is printed, so the best you can do ahead of time would be /foo.cgi/, letting the web
browser tack on “?client=MegaCorp” when the user submits the form. If your application has this
problem, you might consider defining two resource identifiers for each of your resources: an identifier
that uses PATH_INFO, and one that uses QUERY_STRING.
Wiki Markup
The final question is to consider how to transform the plaintext typed in by writers into the HTML dis-
played to readers. Some wikis are extravagant and let writers do things like draw tables and upload
images. BittyWiki will support only a few very basic types of text-to-HTML markup:
❑ To ensure valid HTML, all pages will be placed within paragraph (<p>) tags.
❑ Two consecutive newlines will be treated as a paragraph break.
❑ Any HTML manually typed into a wiki page will be escaped, so that it’s displayed to the viewer
instead of being interpreted by the web browser.
Because there are so few markup rules, BittyWiki pages will look a little bland, but prohibiting raw
HTML will limit the capabilities of any vandals that happen along.
With these design decisions made, it’s now possible to create the CGI web interface to BittyWiki. This
code should go into bittywiki.cgi, in the same cgi-bin/ directory where you put BittyWiki.py:
486

53.8 Page 528

▲back to top
Web Applications and Web Services
#!/usr/bin/python
import cgi
import cgitb
import os
import re
from BittyWiki import Wiki, Page, NotWikiWord
cgitb.enable()
#First, some HTML templates.
MAIN_TEMPLATE = ‘’’<html>
<head><title>%(title)s</title>
<body>%(body)s<hr />%(navLinks)s</body>
</html>’’’
VIEW_TEMPLATE = ‘’’%(banner)s
<h1>%(name)s</h1>
%(processedText)s’’’
WRITE_TEMPLATE = ‘’’%(banner)s
<h1>%(title)s</h1>
<form method=”POST” action=”%(pageURL)s”>
<input type=”hidden” name=”operation” value=”write”>
<textarea rows=”15” cols=”80” name=”data”>%(text)s</textarea><br />
<input type=”submit” value=”Save”>
</form>’’’
DELETE_TEMPLATE = ‘’’<h1>%(title)s</h1>
<p>Are you sure %(name)s is the page you want to delete?</p>
<form method=”POST” action=”%(pageURL)s”>
<input type=”hidden” name=”operation” value=”delete”>
<input type=”submit” value=”Delete %(name)s!”>
</form>’’’
ERROR_TEMPLATE = ‘<h1>Error: %(error)s</h1>’
BANNER_TEMPLATE = ‘<p style=”color:red;”>%s</p><hr />’
#A snippet for linking a WikiWord to the corresponding wiki page.
VIEW_LINK = ‘<a href=”%s”>%%(wikiword)s</a>’
#A snippet for linking a WikiWord with not corresponding page to a
#form for creating that page.
ADD_LINK = ‘%%(wikiword)s<a href=”%s”>?</a>’
Rather than print out HTML pages from inside the CGI script, it’s often useful to define HTML tem-
plates as strings ahead of time and use Python’s string interpolation to fill them with dynamic values.
This helps to separate presentation and content, making it much easier to customize the HTML.
Separating the HTML out from the Python code makes it possible to hand the templates over to a web
designer who doesn’t know Python.
487

53.9 Page 529

▲back to top
Chapter 21
One feature of Python that deserves wider recognition is its capability to do string interpolation with a
map instead of a tuple. If you have a string “A %(foo)s string”, and a map containing an item keyed
to foo, then interpolating the string with the map will replace “%(foo)s” with the string value of the
item keyed to foo:
class WikiCGI:
#The possible operations on a wiki page.
VIEW = ‘’
WRITE = ‘write’
DELETE = ‘delete’
def __init__(self, wikiRoot):
self.wiki = Wiki(wikiRoot)
def run(self):
toDisplay = None
try:
#Retrieve the wiki page the user wants.
page = os.environ.get(‘PATH_INFO’, ‘’)
if page:
page = page[1:]
page = self.wiki.getPage(page)
except NotWikiWord, badName:
page = None
error = ‘“%s” is not a valid wiki page name.’ % badName
toDisplay = self.makeError(error)
if page:
#Determine what the user wants to do with the page they
#requested.
makeChange = os.environ[‘REQUEST_METHOD’] == ‘POST’
if makeChange:
defaultOperation = self.WRITE
else:
defaultOperation = ‘’
form = cgi.FieldStorage()
operation = form.getfirst(‘operation’, defaultOperation)
#We now know which resource the user was trying to access
#(“page” in conjunction with “operation”), and “form”
#contains any representation they were submitting. Now we
#delegate to the appropriate method to handle the operation
#they requested.
operationMethod = self.OPERATION_METHODS.get(operation)
if not operationMethod:
error = ‘“%s” is not a valid operation.’ % operation
toDisplay = self.makeError(error)
if not page.exists() and operation and not \\
(makeChange and operation == self.WRITE):
#It’s okay to request a resource based on a page that
#doesn’t exist, but only if you’re asking for the form to
#create it, or actually trying to create it.
488

53.10 Page 530

▲back to top
Web Applications and Web Services
toDisplay = self.makeError(‘No such page: “%s”’ % page.name)
if operationMethod:
toDisplay = operationMethod(self, page, makeChange, form)
#All the operation methods, as well as makeError, are expected
#to return a set of values that can be used to render the HTML
#response: the title of the page, the body template to use, a
#map of variables to interpolate into the body template, and a
#set of navigation links to put at the bottom of the page.
title, bodyTemplate, bodyArgs, navLinks = toDisplay
if page and page.name != Wiki.HOME_PAGE_NAME:
backLink = ‘<a href=”%s”>Back to wiki homepage</a>’
navLinks.append(backLink % self.makeURL())
print “Content-type: text/html\\n”
print MAIN_TEMPLATE % {‘title’ : title,
‘body’ : bodyTemplate % bodyArgs,
‘navLinks’ : ‘ | ‘.join(navLinks)}
When the WikiCGI class is instantiated, it finds out which resource is being requested, and what the
user wants to do with that resource. It delegates to one of a number of methods (yet to be defined) that
handle the various possible operations.
Each of these methods is expected to return the skeleton of a web page: the title, a template string (one of
the templates defined earlier: VIEW_TEMPLATE, WRITE_TEMPLATE, etc.), a map of variables to use when
interpolating that template, and a set of links to help the user navigate the wiki.
The last act of WikiCGI instantiation is to fill out this skeleton: to interpolate the provided variable map
into the page-specific template string and then to interpolate that into the overarching main template.
The result, a complete HTML page, is simply printed to standard output.
The next part of the CGI defines the three operation-specific methods, which take a page and (possibly)
a resource representation stored in form data; makes any appropriate changes; and returns the raw
materials for a document:
def viewOperation(self, page, makeChange, form=None, banner=None):
“””Renders a page as HTML, either as the result of a request
for it as a resource, or as a side effect of some other
operation.”””
if banner:
banner = BANNER_TEMPLATE % banner
else:
banner = ‘’
if not page.exists():
title = ‘Creating %s’ % page.name
toDisplay = (title, WRITE_TEMPLATE,
{‘title’ : title,
‘banner’ : banner,
‘pageURL’ : self.makeURL(page),
‘text’ : ‘’},
[])
else:
writeLink = ‘<a href=”%s”>Edit this page</a>’ \\
489

54 Pages 531-540

▲back to top

54.1 Page 531

▲back to top
Chapter 21
490
% self.makeURL(page, self.WRITE)
deleteLink = ‘<a href=”%s”>Delete this page</a>’ \\
% self.makeURL(page, self.DELETE)
toDisplay = (page.name, VIEW_TEMPLATE,
{‘name’ : page.name,
‘banner’ : banner,
‘processedText’ : self.renderPage(page)},
[writeLink, deleteLink])
return toDisplay
def writeOperation(self, page, makeChange, form):
“Saves a page, or displays its create or edit form.”
if makeChange:
data = form.getfirst(‘data’)
page.text = data
page.save()
#The operation is done, but we still need a document to
#return to the user. Display the new version of this page,
#with a banner.
toDisplay = self.viewOperation(page, 0, banner=’Page saved.’)
else:
navLinks = []
pageURL = self.makeURL(page)
if page.exists():
title = ‘Editing ‘ + page.name
navLinks.append(‘<a href=”%s”>Back to %s</a>’ % (pageURL,
page.name))
else:
title = ‘Creating ‘ + page.name
toDisplay = (title, WRITE_TEMPLATE, {‘title’ : title,
‘banner’ : ‘’,
‘pageURL’ : pageURL,
‘text’ : page.getText()},
navLinks)
return toDisplay
def deleteOperation(self, page, makeChange, form=None):
“Deletes a page, or displays its delete form.”
if makeChange:
page.delete()
banner = ‘Page “%s” deleted.’ % page.name
#The page is deleted, but we still need a document to
#return to the user. Display the wiki homepage, with a banner.
toDisplay = self.viewOperation(self.wiki.getPage(), 0,
banner=banner)
else:
if page.exists():
title = ‘Deleting ‘ + page.name
pageURL = self.makeURL(page)
backLink = ‘<a href=”%s”>Back to %s</a>’
toDisplay = (title, DELETE_TEMPLATE, {‘title’ : title,
‘name’ : page.name,
‘pageURL’ : pageURL},
[backLink % (pageURL, page.name)])
else:
error = “You can’t delete a page that doesn’t exist.”

54.2 Page 532

▲back to top
Web Applications and Web Services
toDisplay = self.makeError(error)
return toDisplay
#A registry mapping ‘operation’ keys to methods that perform the operations.
OPERATION_METHODS = { VIEW : viewOperation,
WRITE: writeOperation,
DELETE: deleteOperation }
def makeError(self, errorMessage):
“Creates a set of return values indicating an error.”
return (ERROR_TEMPLATE, “Error”, {‘error’ : errorMessage,
‘mainURL’ : self.makeURL(“”)}, [])
def makeURL(self, page=””, operation=None):
“Creates a URL to the resource defined by the given page and resource.”
if hasattr(page, ‘name’):
#A Page object was passed in, instead of a page name.
page = page.name
url = os.environ[‘SCRIPT_NAME’] + ‘/’ + page
if operation:
url += ‘?operation=’ + operation
return url
The last main section of this CGI is the code that transforms the raw wiki text into HTML, linking
WikiWords to BittyWiki resources and creating paragraph breaks:
#A regular expression for use in turning multiple newlines
#into paragraph breaks.
MULTIPLE_NEWLINES = re.compile(“(\\r?\\n){2,}”)
def renderPage(self, page):
“””Returns the text of the given page, with transforms applied
to turn BittyWiki markup into HTML: WikiWords linked to the
appropriate page or add form, and double newlines turned into
paragraph breaks.”””
#First, escape any HTML present in the bare text so that it is
#shown instead of interpreted.
text = page.getText()
for find, replace in ((‘<’, ‘&lt;’), (‘>’, ‘&gt;’), (‘&’, ‘&amp;’)):
text = text.replace(find, replace)
#Link all WikiWords in the text to their view or add resources.
html = ‘<p>’ + page.WIKI_WORD.sub(self._linkWikiWord, text) \\
+ ‘</p>’
#Turn multiple newlines into paragraph breaks.
html = self.MULTIPLE_NEWLINES.sub(‘</p>\\n<p>’, html)
return html
def _linkWikiWord(self, match):
“””A helper method used to replace a WikiWord with a link to view
the corresponding page (if it exists), or a link to create the
corresponding page (if it doesn’t).”””
491

54.3 Page 533

▲back to top
Chapter 21
linkedPage = self.wiki.getPage(match.group(0))
link = ADD_LINK
if linkedPage.exists():
link = VIEW_LINK
link = link % self.makeURL(“%(wikiword)s”)
#The link now looks something like:
# <a href=”/cgi-bin/bittywiki.cgi/%(wikiword)s”>%(wikiword)s</a>
#We’ll interpolate ‘wikiword’ to fill in the actual page name.
return link % {‘wikiword’ : linkedPage.name}
Finally, here is the code that invokes WikiCGI against a particular wiki when this file is run as a script:
if __name__ == ‘__main__’:
WikiCGI(“wiki/”).run()
Once you’re underway, you’ll be able to start editing pages of your own. You can see an example page in
Figure 21-7 and its corresponding editing page in Figure 21-8.
Figure 21-7
Figure 21-8
492

54.4 Page 534

▲back to top
Web Applications and Web Services
Make this code executable and try it out in conjunction with EasyCGIServer or with your web host’s
CGI setup. Hitting http://localhost:8000/cgi-bin/bittywiki.cgi (or the equivalent on your web
host) will send you to the form for creating the wiki’s home page. You can write a home page, making ref-
erences to other pages that don’t exist yet, and then click on the question marks near their names to create
them. You can build your wiki starting from there; this is how real wikis grow. A wiki is an excellent tool
for managing collaboration with other members of a development team, or just for keeping track of your
own notes. They’re also easy and fun to build, which is why there are so many implementations.
BittyWiki is a simple but fully functional wiki with a simple but flexible design. The presentation HTML is
separated from the logic, and the job of identifying the resource is done by a method that then dispatches
to one of several handler methods. The handler methods identify the provided representation (if any),
take appropriate action, and return the resource representation or other document to be rendered. The
resources and operations were designed by considering the problem according to the principles of REST.
This type of design and architecture are a very useful way of building standalone web applications.
Web Services
So far, the web applications developed in this chapter share one unstated underlying assumption: their
intended audience is human. The same is true of most applications available on the web. The resource
representations served by the typical web application (the wiki we just wrote being no exception) are a
conglomeration of data, response messages, layout code, and navigation, all bundled together in an
HTML file intended to be rendered by a web browser in a form pleasing to humans. When interaction is
needed, applications present GUI forms for you to fill out through a human-computer interface; and
when you submit the forms, you get more pretty HTML pages. In short, web applications are generally
written by humans for humans.
Yet web applications, even the most humancentric, have always had nonhuman users: software clients
not directly under the direction of a human. To give them a catchy name, robots. From search engine spi-
ders to automatic auction bidding scripts to real-time weather display clients, all sorts of scripted clients
consume web applications, often without the knowledge of the people who originally wrote those appli-
cations. If a web application proves useful, someone will eventually write a robot that uses it.
In the old days, robots had no choice but to impersonate web browsers with humans driving them. They
would make HTTP requests just like a web browser would, and parse the resulting HTML to find the
interesting parts. While this is still a common technique, more and more web applications are exposing
special interfaces solely for the benefit of robots. Doing so makes it easier to write robots, and frees the
server from using its bandwidth to send data that won’t be used. These interfaces are called web ser-
vices. Big-name companies like Google, Yahoo!, Amazon, and eBay have exposed web service APIs to
their web applications, as have many lesser-known players.
Many fancy standards have been created around web services, some of which are covered later in this
chapter, but the basic fact is that web services are just web applications for robots. A web service usually
corresponds to a web application, and makes some of the functionality of that application available in
robot-friendly form. The only reason these fancy standards exist is to make it easier to write robots or to
expose your application to robots.
Robots have different needs than humans. Humans can glance at an HTML rendering of a page and
separate the important page-specific data from the navigation, logos, and clutter. A robot has no such
493

54.5 Page 535

▲back to top
Chapter 21
ability: It must be programmed to parse out the data it needs. If a redesign changes the HTML a site pro-
duces, any robot that reads and parses that HTML must be reprogrammed. A human can recall or make
up the input when a web application requires it; a robot must be programmed ahead of time to provide
the right input. Because of this, it’s no surprise that web services tend to have better usage documentation
than their corresponding web applications, nor that they serve more structured resource representations.
Web services and the scripts that use them can exist in symbiotic relationships. If you provide web ser-
vices that people want to use, you form a community around your product and get favorable publicity
from what they create. You can give your users the freedom to base new applications on yours, instead
of having to implement their feature requests yourself. Remember that if your application is truly useful,
people are going to write robots that use it no matter what you do. You might as well bless this use,
monitor it, and track it.
The benefits of consuming others’ web services are more obvious: You gain access to data sets and algo-
rithms you’d otherwise have to implement yourself. You don’t need to get permission to use these data
sets, because web services are prepackaged permission.
Even if you control both the producers and the consumers of data, there are advantages to bridging the
gap with web services. Web services enable you to share code across machines and programming lan-
guages, just as web applications can be accessed from any browser or operating system.
Python is well suited to using and providing web services. Its loose typing is a good match for the various
web service standards, which provide limited or nonexistent typing. Because Python lets you overload a
class’ method call operator, it’s possible to make a web service call look exactly like an ordinary method
call. Finally, Python’s standard library provides good basic web support. If a high-level protocol won’t
meet your needs or its library has a bug, you can drop to the next lowest level and still get the job done.
How Web Services Work
Web services are just web applications for robots, so it’s natural that they should operate just like normal
web applications: You send an HTTP request and you get some structured data back in the response. A web
service is supposed to be used by a script, though, so the request that goes in and the response that comes
out need to be more formally defined. Whereas a web application usually returns a full-page image that is
rendered by a browser and parsed by the human brain, a web service returns just the “important” data in
some easily parseable format, usually XML. There’s also usually a human-readable or machine-parseable
description of the methods being exposed by the web service, to make it easier for users to write a script
that does what they want.
There are three main standards for web services: REST, XML-RPC, and SOAP. For each standard, this chap-
ter will show you how to use an existing public web service to do something useful, how to expose the
BittyWiki API as a web service, and how to make a robot manipulate the wiki through that web service.
REST Web Services
If REST is so great for the web that humans use, why shouldn’t it also work for robots? The answer is
that it works just fine. The hypertext links and HTML forms you designed for your human users are
access points into a REST API that can just as easily be used by a properly programmed robot. All you
494

54.6 Page 536

▲back to top
Web Applications and Web Services
need to add is a way to provide robot-friendly representations of your resources, and a way for robots to
get at those representations.
If you’re designing a web application from scratch, keep in mind the needs of both humans and robots.
You should end up able to expose similar APIs to your HTML forms and to external scripts. It’s unlikely
you’ll expose the exact same features to humans and to robots, but you’ll be able to reuse a lot of archi-
tecture and code.
In some situations you might want to create a new, simpler API and expose that as your web service
instead. This might happen if you’re working on an application with an ugly API that was never meant
to be seen by outsiders, if your web application is very complex, or if the people writing robots only
want to use part of the full API.
REST Quick Start: Finding Bargains on Amazon.com
Amazon.com, the popular online store, makes much of their data available through a REST web service
called Amazon Web Services. Perhaps the most interesting feature of this web service is the capability it
offers to search for books or other items and then retrieve metadata, pictures, and reviews for an item.
Amazon effectively gives you programmatic access to their product database, something that would be
difficult to duplicate or obtain by other means.
The Amazon Web Services homepage is at www.amazon.com/gp/aws/landing.html.
To use Amazon Web Services you need a subscription ID. This is a 13-character string that identifies
your account. You can get one for free by signing up at www.amazon.com/gp/aws/registration/
registration-form.html/. After you’ve got an API key, you can use it to query Amazon Web Services.
Because the AWS interface is RESTful, you invoke it by sending a GET request to a particular resource:
The results are returned within an XML document. It’s the web service equivalent of Amazon’s search
engine web application. Instead of a user interface based on HTML forms, AWS has rules for constructing
resources. Instead of a pretty HTML document containing your search results, it gives you a structured
XML representation of them.
The Amazon Web Services are actually something of a REST heretic. Though most
of AWS’s design is RESTful, it defines a few operations that make changes on the
server side when you GET them. For instance, the AWS “CartModify” operation
enables you to add or remove items from your Amazon shopping cart just by
making a GET request. Recall that GET requests shouldn’t change any resources on
the server side; you should use POST, PUT, or DELETE for such operations.
Presumably, the AWS designers chose consistency (using GET for everything) over
RESTfulness.
Because the AWS API isn’t purely RESTful, it’s not necessarily safe to pass around
the resource identifiers AWS gives you. Someone else might end up adding books to
your shopping cart by mistake! This is exactly the sort of thing to avoid when
designing your own REST API.
495

54.7 Page 537

▲back to top
Chapter 21
Try It Out Peeking at an Amazon Web Services Response
You can invoke Amazon Web Services using the same urllib module you’d use to download a web
page. Here’s an interactive Python session that searches for books by James Joyce (slightly reformatted
and edited for brevity):
>>> import urllib
>>> author = “Joyce, James”
>>> subscriptionID = [your subscription id]
>>> url = “http://xml.amazon.com/onca/xml3?f=xml&t=webservices-20&dev-
t=%s&type=lite&mode=books&AuthorSearch=%s” % (subscriptionID, urllib.quote(author))
>>> print urllib.urlopen(url).read()
<?xml version=”1.0” encoding=”UTF-8”?>
<ProductInfo xmlns:xsi=”http://www.w3.org/2001/XMLSchema-instance”
xsi:noNamespaceSchemaLocation=”http://xml.amazon.com/schemas3/dev-lite.xsd”>
...
<Details url=”http://www.amazon.com/exec/obidos/ASIN/0142437344/webservices-20?dev-
t=D8O1OTR10IMN7%26camp=2025%26link_code=xm2”>
<Asin>0142437344</Asin>
<ProductName>A Portrait of the Artist As a Young Man (Penguin
Classics)</ProductName>
<Catalog>Book</Catalog>
<Authors>
<Author>James Joyce</Author>
</Authors>
<ReleaseDate>25 March, 2003</ReleaseDate>
<Manufacturer>Penguin Books</Manufacturer>
<ImageUrlSmall>http://images.amazon.com/images/P/0142437344.01.THUMBZZZ.jpg</ImageU
rlSmall>
<ImageUrlMedium>http://images.amazon.com/images/P/0142437344.01.MZZZZZZZ.jpg</Image
UrlMedium>
<ImageUrlLarge>http://images.amazon.com/images/P/0142437344.01.LZZZZZZZ.jpg</ImageU
rlLarge>
<Availability>Usually ships in 24 hours</Availability>
<ListPrice>$9.00</ListPrice>
<OurPrice>$8.10</OurPrice>
<UsedPrice>$1.95</UsedPrice>
</Details>
...
</ProductInfo>
How It Works
All we did there was open a URL and read it. You can visit the same URL in a web browser (treating the web
service as a web application) and get the exact same data as we did from the interactive Python session. The
differences between web applications and web services have nothing to do with architecture; both use the
architecture of the web. The only differences are related to the format of the requests and responses.
There are two problems with just opening that resource and reading it, however (whether from a script
or from a web browser), and they should be obvious from that session log. The AWS URL to do a search
496

54.8 Page 538

▲back to top
Web Applications and Web Services
is really long and difficult to remember. Even with a reference guide, it’s hard to keep all the URL
parameters straight. Second, the response is a lot of XML data. It’ll take some work to parse it or trans-
form it into a more human-friendly form. Fortunately, that work has already been done for us.
A popular web service will eventually sprout clients written in every major programming language. For
Amazon Web Services, the standard Python client is PyAmazon, originally written by Mark Pilgrim and
now maintained by Michael Josephson. This module abstracts the details of the Amazon Web Services
REST API. It enables you to request one of those complex resources just by making a method call, and
retrieve a list of Python objects instead of a mass of XML. Behind the scenes, it uses urllib to retrieve
a resource (just like we did), and then parses the XML response into a Python data structure. Thanks to
PyAmazon, it’s easy to have Pythonic fun with Amazon Web Services.
Download PyAmazon from www.josephson.org/projects/pyamazon/ and install it into your
PYTHON_PATH or into the directory in which you plan to write your scripts that use AWS. While
you’re at it, also download OnDemandAmazonList, a class that lets you iterate over paginated lists of
AWS search results as though they were normal Python lists. The sample application that follows uses
OnDemandAmazonList to make the code more natural.
Introducing WishListBargainFinder
Amazon lets individuals and booksellers advertise their used copies of books on their site, and Amazon
presents the lowest used price for a book alongside its own price for a new book. If you look back at that
XML search result for James Joyce, you’ll see that A Portrait of the Artist as a Young Man is available new
from Amazon for $8.10 (“OurPrice”), but there are also people selling used copies for as low as $1.95
(“UsedPrice”). That’s a pretty good price, even when you factor in shipping. Many of the books listed on
Amazon are available used for as little as one cent. Amazon will show you the lowest used price for any
individual book, but it’s not so easy to scan a whole list looking for bargains.
Amazon users can keep “wish lists” of things they’d like to own. If you keep one yourself, you’ve selected
out of the millions of items on Amazon a few that you’d be especially interested in buying for a bargain.
Amazon Web Services provides a wish list search, so it’s possible to write a script that uses AWS to go
through a wish list and identify the bargains. If you don’t mind buying used, this could save you a lot
of money.
Here’s a class, BargainFinder, that accepts a list obtained from an AWS query and scans it for second-
hand bargains. Bargains can be defined as costing less than a certain amount (say, $3), or as costing a
certain amount less than the corresponding items new from Amazon (say, 75% less). It, and the code
fragments that follow it, are part of a file I call WishListBargainFinder.py:
import copy
import re
import amazon
class BargainFinder:
“””A class that, given a list of Amazon items, finds out which
items in the list are available used at a bargain price.”””
def __init__(self, bargainCoefficient=.25, bargainCutoff=3.00):
“””The bargainCoefficient is how little an item must cost
used, versus its new price, to be considered a bargain. The
497

54.9 Page 539

▲back to top
Chapter 21
default bargain coefficient is .25, meaning that an item
available used for less than 25% of its Amazon price is
considered a bargain.
The bargainCutoff is for finding bargains among items that are
cheap to begin with. The default bargainCutoff is 5, meaning
that any item available used for less than $3.00 is considered
a bargain, even if it’s available new for only a little more
than $3.00.”””
if bargainCoefficient >= 1:
raise Exception, ‘It makes no sense to look for “bargains” that ‘ \\
+ ‘cost more used than new!’
self.coefficient = bargainCoefficient
self.cutoff = bargainCutoff
def printBargains(self, items):
“””Find the bargains in the given list and present them in a
textual list.”””
bargains = self.getBargains(items)
printedHeader = 0
if bargains:
print (‘Here are items available used for less than $%.2f, ‘ + \\
‘or for less than %.2d%% of their Amazon price:’) \\
% (self.cutoff, self.coefficient*100)
prices = bargains.keys()
prices.sort()
for usedPrice in prices:
for bargain, amazonPrice in bargains[usedPrice]:
savings = ‘’
if amazonPrice:
percentageSavings = (1-(usedPrice/amazonPrice)) * 100
savings = ‘(Save %.2d%% off $%.2f) ‘ \\
% (percentageSavings, amazonPrice)
print ‘ $%.2f %s%s’ % (usedPrice, savings,
bargain.ProductName)
else:
print “Sorry, I couldn’t find any bargains in that list.”
def getBargains(self, items):
“Scan the given list, looking for bargains.”
bargains = {}
for item in items:
bargain = False
amazonPrice = self.getPrice(item, “OurPrice”)
usedPrice = self.getPrice(item, “UsedPrice”)
if usedPrice:
if usedPrice < self.cutoff:
bargain = True
if amazonPrice:
if (amazonPrice * self.coefficient) > usedPrice:
bargain = True
if bargain:
#We sort the bargains by the used price, so the
#cheapest items are displayed first.
498

54.10 Page 540

▲back to top
Web Applications and Web Services
bargainsForPrice = bargains.get(usedPrice, None)
if not bargainsForPrice:
bargainsForPrice = []
bargains[usedPrice] = bargainsForPrice
bargainsForPrice.append((item, amazonPrice))
return bargains
def getPrice(self, item, priceField):
“””Retrieves the named price field (eg. “OurPrice”,
“UsedPrice”, and attempts to parse its currency string into a
number.”””
price = getattr(item, priceField, None)
if price:
price = self._parseCurrency(price)
return price
def _parseCurrency(self, currency):
“””A cheap attempt to parse an amount of currency into a
floating-point number: Strip out everything but numbers,
decimal point, and negative sign.”””
return float(self.IRRELEVANT_CURRENCY_CHARACTERS.sub(‘’, currency))
IRRELEVANT_CURRENCY_CHARACTERS = re.compile(“[^0-9.-]”)
This class won’t quite work as is, because it assumes that a list of query results obtained from PyAmazon
(the items argument to getBargains) works just like a Python list. Actually, AWS query results are
delivered in pages of ten. Making a single AWS query will return only the single page you request, and
you’ll need extra logic to iterate from the last item on the first page to the first item of the second.
That’s why OnDemandAmazonList was invented. This class, available from the same web site as
PyAmazon itself, hides the complexity of retrieving successive AWS result pages behind an interface that
looks just like a Python list. You iterate over an OnDemandAmazonList as you would any other list, and
behind the scenes it makes the necessary web service calls to get the data you want. This is another
example of why Python excels at web services: It makes it easy to hide this kind of inconvenient detail.
With OnDemandAmazonList, it’s a simple matter to put an interface on the BargainFinder class with
code that retrieves a wish list as an OnDemandAmazonList, and runs it through the BargainFinder to
find the items on the wish list that are available used for a bargain price. You could just as easily use the
BargainFinder to find bargains in the result set of any other AWS query, so long as you made sure to
wrap the query in an OnDemandAmazonList:
from OnDemandAmazonList import OnDemandAmazonList
def getWishList(subscriptionID, wishListID):
“Returns an iterable version of the given wish list.”
kwds = {‘license_key’ : subscriptionID,
‘wishlistID’ : wishListID,
‘type’ : ‘lite’}
return OnDemandAmazonList(amazon.searchByWishlist, kwds)
if __name__ == ‘__main__’:
import sys
if len(sys.argv) != 3:
print ‘Usage: %s [AWS subscription ID] [wish list id]’ % sys.argv[0]
499

55 Pages 541-550

▲back to top

55.1 Page 541

▲back to top
Chapter 21
sys.exit(1)
subscriptionID, wishListID = sys.argv[1:]
wishList = getWishList(subscriptionID, wishListID)
BargainFinder().printBargains(wishList)
Here’s the WishListBargainFinder running against my mother’s wish list:
# python WishListBargainFinder.py [My subscription ID] 1KT0ATF9MM4FT
Here are items available used for less than $3.00, or for less than 25% of their
Amazon price:
$0.29 (Save 94% off $4.99) Clockwork : Or All Wound Up
$1.99 (Save 68% off $6.29) The Fifth Elephant: A Novel of Discworld
$2.95 (Save 57% off $6.99) Interesting Times (Discworld Novels (Paperback))
$2.96 (Save 52% off $6.29) Jingo: A Novel of Discworld
A quick word about Amazon wish list IDs: The WishListBargainFinder takes a wish list ID as
command-line input, but Wish list IDs are a little bit hidden in the Amazon web application. To find a
person’s wish list ID, you need to go to their wish list and then look at the id field of the URL. The wish
list ID is a twelve-character series of letters and numbers that looks like BUWBWH9K2H77.
You can programmatically search for a user’s wish list by making an AWS call (using the ListSearch
operation), but because that method is not yet supported by PyAmazon, you’ll have to construct the URL
and parse the XML yourself. For guidance, look at the examples on Amazon’s site: www.amazon.com/
gp/aws/sdk/104-5260879-5653550
Giving BittyWiki a REST API
Let’s revisit BittyWiki, the simple wiki application we created in the last section as a sample web applica-
tion. By design, BittyWiki already exposes a very simple REST API. Recall that in addition to the name
of the page, which is always part of the resource identifier, there are only two variables to consider:
operation and data. operation tells BittyWiki what you want to do to the page you named, and
data contains the data you want to shove into the page. Now let’s consider this API from a robot’s
point of view.
The first thing to consider is how to even determine whether a given request comes from a human
(more accurately, a web browser) or a robot. You might think this is easy; after all, the User-Agent
HTTP header you saw earlier is supposed to identify the software that’s making the request. The
problem is that there’s no definitive list of web browsers. New browsers and robots are being created all
the time, and some use the same underlying libraries (a web browser and a robot written in Python
might both claim to be urllib). The User-Agent string isn’t reliable enough to be used as a basis for
this decision.
Most web services solve this problem by creating a second set of resource identifiers that mirror the
resource identifiers used by the web application but serve up robot-friendly resource representations.
The “robot’s entrance” for your application might be an entirely separate script (app-api.cgi instead of
500

55.2 Page 542

▲back to top
Web Applications and Web Services
app.cgi) or a standard string prepended to the PATH_INFO of a resource identifier (app.cgi/api/foo
instead of app.cgi/foo). The PATH_INFO solution yields nicer-looking resource identifiers, but TiniWiki’s
REST web service will be implemented as a separate CGI, just because it’s easier to present.
One final note with respect to PUT and DELETE. Web services are free from dependence on HTML forms.
While the PUT and DELETE HTTP verbs aren’t supported by web browsers, they are supported by many
(but not all) programmable clients. We could simplify the preexisting BittyWiki interface a little by bring-
ing in PUT and DELETE. Doing this would let us get rid of the operation argument, which is only used
to distinguish a PUT or POST-style POST request from a DELETE-style POST request. However, for the
sake of correspondence with the web application, and because not all programmable clients support PUT
and DELETE, the BittyWiki REST web service won’t take this route.
The second thing to consider is which features of the web application it makes sense to expose through
an external API. Why would someone want programmatic access to the contents of a wiki? A wiki’s
users might create two types of robot:
❑ A robot that modifies or creates wiki pages — for instance, an automated test system that posts a
daily status report to a particular wiki page
❑ A robot that retrieves wiki pages — to archive or mirror a wiki or to render wiki pages to an
end-user in some format besides HTML
The first type of robot might need to create, edit, and delete a wiki page. That functionality can remain
more or less intact, but unlike in a web application, there’s no need to present a nice-looking document
after taking a requested action. All the robot needs to know is whether or not its request was carried out.
The document returned for a POST operation need only contain a status message.
Both types of robot need to retrieve pages from the wiki. What they actually need, though, is not the HTML
rendering of the page (the thing you get when you GET /bittywiki.cgi/PageName), but the raw page
data (the thing that shows up in the edit box when you GET /bittywiki.cgi/PageName?operation=
write). The first type of robot needs the data in this format because it’s going to do its own rendering, and
it’s easier to render from the raw data than from HTML. The second type of robot needs it in this format for
a similar reason; it’s because that’s what shows up in the edit box because that’s how it’s stored on the
back-end.
BittyWiki’s REST API for robots is therefore basically similar to the REST API for web browsers. The
only difference is the format of the responses: Instead of human-readable HTML documents, the REST
web service outputs plaintext documents. A more complicated REST web service, like Amazon’s, would
probably output documents formatted in XML or sparse HTML, expecting the client to parse them.
Here’s the plaintext result of GETting http://localhost:8000/cgi-bin/bittywiki-rest.cgi;
compare it to the HTML output when you GET http://localhost:8000/cgi-bin/bittiwiki.cgi:
This is the home page for my BittyWiki installation.
Here you can learn about the philosophy and technologies that drive web
applications: REST, CGI, and the PythonLanguage.
The structure of bittywiki-rest.cgi is also similar to bittywiki.cgi:
501

55.3 Page 543

▲back to top
Chapter 21
#!/usr/bin/python
import cgi
import cgitb
cgitb.enable()
import os
import re
from BittyWiki import Wiki, Page, NotWikiWord
class WikiRestApiCGI:
#The possible operations on a wiki page.
VIEW = ‘’
WRITE = ‘write’
DELETE = ‘delete’
#The possible response codes this application might return.
RESPONSE_CODES = { 200 : ‘OK’,
400 : ‘Bad Request’,
404 : ‘Not Found’}
def __init__(self, wikiBase):
“Initialize with the given wiki.”
self.wiki = Wiki(wikiBase)
def run(self):
“””Determine the command, dispatch to the appropriate handler,
and print the results as an XML document.”””
toDisplay = None
try:
page = os.environ.get(‘PATH_INFO’, ‘’)
if page:
page = page[1:]
page = self.wiki.getPage(page)
except NotWikiWord, badName:
toDisplay = 400, ‘“%s” is not a valid wiki page name.’ % badName
if not toDisplay:
form = cgi.FieldStorage()
operation = form.getfirst(‘operation’, self.VIEW)
operationMethod = self.OPERATION_METHODS.get(operation)
if operationMethod:
if not page.exists() and operation != self.WRITE:
toDisplay = 404, ‘No such page: “%s”’ % page.name
else:
toDisplay = operationMethod(self, page, form)
else:
toDisplay = 400, ‘“%s” is not a valid operation.’ % operation
#Print the response.
responseCode, payload = toDisplay
print ‘Status: %s %s’ % (responseCode,
self.RESPONSE_CODES.get(responseCode))
print ‘Content-type: text/plain\\n’
print payload
502

55.4 Page 544

▲back to top
Web Applications and Web Services
The main code figures out the resource and the desired operation and hands this off (along with any
provided representation) to a handler method. The result is then rendered — but this time as plaintext:
def viewOperation(self, page, form=None):
“Returns the raw text of the given wiki page.”
return 200, page.getText()
def writeOperation(self, page, form):
“Writes the specified page.”
page.text = form.getfirst(‘data’)
page.save()
return 200, “Page saved.”
def deleteOperation(self, page, format, form=None):
“Deletes the specified page.”
if not page.exists():
toDisplay = 404, “You can’t delete a page that doesn’t exist.”
else:
page.delete()
toDisplay = 200, “Page deleted.”
return toDisplay
#A registry mapping ‘operation’ keys to methods that perform the operations.
OPERATION_METHODS = { VIEW : viewOperation,
WRITE: writeOperation,
DELETE: deleteOperation }
The three operation handler methods are also similar to their counterparts in bittywiki.cgi, though
simpler because they produce less data.
Wiki Search-and-Replace Using the REST Web Service
What good is this web service for BittyWiki? Well, here’s an only slightly contrived example: Suppose
that you get someone to host a BittyWiki installation for an open-source project you’re working on,
called Foo. You create a lot of wiki pages that mention the name of the project in their text (“Foo is a
triphasic degausser for semantic defribulation”) and in the titles of the pages (BenefitsOfFoo, FooDesign,
etc.). All is going well until one day when you decide to change the name of your project to Bar. It would
take a long time to manually change those wiki pages (including renaming many of them), and you
don’t have access to the server on which the wiki is actually hosted, so you can’t write a script to crawl
the file system. What do you do?
Here’s a Python script, WikiSpiderREST.py, which acts as a wiki search-and-replace spider. Starting at
the HomePage of the wiki (which is a WikiWord), it crawls the wiki by following WikiWord links, and
replaces all of the instances of one string (e.g., “Foo”) with another string (e.g., “Bar”). A page whose
name contains the old string (e.g., “FooDesign”) is deleted and recreated under a different name (e.g.,
“BarDesign”). WikiSpiderREST.py keeps track of the pages it has processed so as not to waste time or
get stuck in a loop:
503

55.5 Page 545

▲back to top
Chapter 21
#!/usr/bin/python
import re
import urllib
class WikiReplaceSpider:
“A class for running search-and-replace against a web of wiki pages.”
WIKI_WORD = re.compile(‘(([A-Z][a-z0-9]*){2,})’)
def __init__(self, restURL):
“Accepts a URL to a BittyWiki REST API.”
self.api = BittyWikiRestAPI(restURL)
def replace(self, find, replace):
“””Spider wiki pages starting at the front page, accessing them
and changing them via the provided API.”””
processed = {} #Keep track of the pages already processed.
todo = [‘HomePage’] #Start at the front page of the wiki.
while todo:
for pageName in todo:
print ‘Checking “%s”’ % pageName
try:
pageText = self.api.getPage(pageName)
except RemoteApplicationException, message:
if str(message).find(“No such page”) == 0:
#Some page mentioned a WikiWord that doesn’t exist
#yet; not a big deal.
pass
else:
#Some other problem; pass it on up.
raise RemoteApplicationException, message
else:
#This page actually exists; process it.
#First, find any WikiWords in this page: they may
#reference other existing pages.
for wikiWord in self.WIKI_WORD.findall(pageText):
linkPage = wikiWord[0]
if not processed.get(linkPage) and linkPage not in todo:
#We haven’t processed this page yet: put it on
#the to-do list.
todo.append(linkPage)
#Run the search-and-replace on the page text to get the
#new text of the page.
newText = pageText.replace(find, replace)
#Check to see if this page name matches
#search and replace. If it does, delete it and
#recreate it with the new text; otherwise, just
#save the new text.
newPageName = pageName.replace(find, replace)
if newPageName != pageName:
print ‘ Deleting “%s”, will recreate as “%s”’ \\
% (pageName, newPageName)
504

55.6 Page 546

▲back to top
Web Applications and Web Services
self.api.delete(pageName)
if newPageName != pageName or newText != pageText:
print ‘ Saving “%s”’ % newPageName
self.api.save(newPageName, newText)
#Mark the new page as processed so we don’t go through
#it a second time.
if newPageName != pageName:
processed[newPageName] = True
processed[pageName] = True
todo.remove(pageName)
So far, there’s been nothing REST-specific except the reference to a BittyWikiRestAPI class. That’s
about to change as we go ahead and define that class, as well as others that implement a general Python
interface to the BittyWiki REST API:
class BittyWikiRestAPI:
“A Python interface to the BittyWiki REST API.”
def __init__(self, restURL):
“Do all the work starting from the base URL of the REST interface.”
self.base = restURL
def getPage(self, pageName):
“Returns the raw markup of the named wiki page.”
return self._doGet(pageName)
def save(self, pageName, data):
“Saves the given data to the named wiki page.”
return self._doPost(pageName, { ‘operation’ : ‘write’,
‘data’ : data })
def delete(self, pageName):
“Deletes the named wiki page.”
return self._doPost(pageName, { ‘operation’ : ‘delete’ })
def _doGet(self, pageName):
“”””Does a generic HTTP GET. Returns the response body, or
throws an exception if the response code indicates an error.”””
url = self._makeURL(pageName)
return self.Response(urllib.urlopen(url)).body
def _doPost(self, pageName, data):
“””Does a generic HTTP POST. Returns the response body, or
throws an exception if the response code indicates an error.”””
url = self._makeURL(pageName)
return self.Response(urllib.urlopen(url, urllib.urlencode(data))).body
def _makeURL(self, pageName):
“Returns the URL to the named wiki page.”
url = self.base
if url[-1] != ‘/’:
url += ‘/’
return url + pageName
505

55.7 Page 547

▲back to top
Chapter 21
class Response:
“””This class handles the HTTP response returned by the REST
web service.”””
def __init__(self, inHandle):
self.body = None
statusCode = None
info = inHandle.info()
#The status has automatically been read into an object
#that also contains all the HTTP headers. The status
#string looks like ‘200 OK’
statusHeader = info[‘status’]
statusCode = int(statusHeader.split(‘ ‘)[0])
#The remaining data is the plain-text response. In a more
#complex application, this might be structured text or
#XML, and at this point it would need to be parsed.
self.body = inHandle.read()
#The response codes in the 2xx range are the only good
#ones. Getting any other response code should result in
#an exception.
if statusCode / 100 != 2:
raise RemoteApplicationException, self.body
class RemoteApplicationException(Exception):
“””A simple exception class for use when the REST API returns an
error condition.”””
pass
The BittyWikiRestAPI class uses the urllib library to GET and POST things to BittyWiki’s REST inter-
face CGI. It interprets the response as a status message, an exception message, or the text of a requested
page. This class could be distributed in a standalone module to encourage developers to write BittyWiki
add-ons in Python.
Note that the Response class is defined within the BittyWikiRestAPI class: No one else is going to use
it, and putting it here makes it invisible outside the class. This is completely optional, but it makes the
top-level view neater.
Finally, some code that implements a command-line interface to the spider:
if __name__ == ‘__main__’:
import sys
if len(sys.argv) == 4:
restURL, find, replace = sys.argv[1:]
else:
print ‘Usage: %s [URL to BittyWiki REST API] [find] [replace]’ \\
% sys.argv[0]
sys.exit(1)
WikiReplaceSpider(restURL).replace(find, replace)
506

55.8 Page 548

▲back to top
Web Applications and Web Services
Try It Out Wiki Searching and Replacing
Use your BittyWiki installation to create a few wiki pages around a particular topic. In the example
shown in Figure 21-9, a few pages have been created for the mythical Foo project.
Figure 21-9
Run the WikiSpiderREST.py command to change your topic to another one. You should see output
similar to this:
$ python WikiSpiderREST.py http://localhost:8000/cgi-bin/bittywiki-rest.cgi Foo Bar
Checking “HomePage”
Saving “HomePage”
Checking “FooCaseStudies”
Deleting “FooCaseStudies”, will recreate as “BarCaseStudies”
Saving “BarCaseStudies”
Checking “CVSRepository”
Saving “CVSRepository”
Checking “CaseStudy2”
Checking “BenefitsOfFoo”
Deleting “BenefitsOfFoo”, will recreate as “BenefitsOfBar”
Saving “BenefitsOfBar”
Checking “CaseStudy1”
Saving “CaseStudy1”
Checking “FooDesign”
Deleting “FooDesign”, will recreate as “BarDesign”
Saving “BarDesign”
Lo and behold: The wiki pages have been changed and, where necessary, renamed (see Figure 21-10).
How It Works
WikiSpiderREST.py keeps a list of WikiWords to check and possibly subject to search-and-replace. To pro-
cess one of the WikiWords, it retrieves the corresponding page through the BittyWiki web service API. If the
page actually exists, its text is scanned, and all of its WikiWords are put on the list of items to check later.
The page then has its text modified using string search-and-replace, and is saved through the web service
API. If the page name contains the string to be replaced, it’s deleted and a new page with the same content
is created — again, through the web service API. The next WikiWord in the list is then checked, and so on.
507

55.9 Page 549

▲back to top
Chapter 21
Figure 21-10
Because WikiSpiderREST.py has no knowledge of wiki pages that are inaccessible from the HomePage,
it’s not guaranteed to get all of the pages on the wiki. It only gets the ones a human user would see if they
started at the HomePage and clicked all of the links.
XML-RPC
XML-RPC is a protocol that does the same job as REST: It makes it easy to write a robot that accesses
and/or modifies some remote application just by making HTTP requests. There are some important dif-
ferences, though. Whereas a REST call looks like manipulation of a document repository, an XML-RPC
looks like a function call (in fact, in Python implementations the call to the web service is disguised as a
function call). Instead of sending a GET or POST to the resource you want to retrieve or modify, as with
REST, XML-RPC traditionally has you do all your calls by POSTing to one special “server” resource. The
data you POST contains an XML representation of a function you’d like to call, and any arguments to
that function. As with REST, the response to your call is a document containing any information you
requested, any status messages, and so on.
BittyWiki is simple enough that everything you pass in or get out is a mere string. We’re fortunate in this
regard because strings are the only data type supported by REST. If you need to pass an integer into a
REST application, you need to encode it as a string and trust that the resource handler will know to turn it
back into an integer. If you need to pass in an ordered list, you need to learn the server’s preferred way of
representing an ordered list as a string. One REST application might represent lists as “item1,item2,item3”;
another might represent them as “item1|item2|item3|”; a third might represent them as a custom-defined
XML data structure. The major shortcoming of REST is that there’s no standard way of marshalling differ-
ent data types into strings, or of unmarshalling a string into typed data. You need to relearn the request and
response format for every REST web service you use.
Here’s the canonical sample XML-RPC client application. The public XML-RPC server betty.userland.com
provides some example methods, including one that returns the name of a U.S. state, given an index into
an alphabetical list:
>>> import xmlrpclib
>>> server = xmlrpclib.ServerProxy(“http://betty.userland.com”)
>>> server.examples.getStateName(41)
‘South Dakota’
508

55.10 Page 550

▲back to top
Web Applications and Web Services
If this were a REST web service, the forty-first state in the list would be accessible as a distinct resource,
perhaps “http://betty.userland.com/StateNames/41”. You’d get the name of a state by GETting
the appropriate resource. You might have access to a Python library that handles the request and response
details (the way the PyAmazon library handles the details of Amazon Web Services), but such libraries
need to be written anew for each REST web service, as there’s no REST standard for data structure
representation.
XML-RPC’s main advantage over REST is that it provides a standard way of encoding simple data struc-
tures into request and response data. XML-RPC specifies different XML strings for encoding the integer
4, the floating-point value 4.0, the string “4”, and a list containing only the string “4”. What you get
back from an XML-RPC call is not a document that you have to parse, but a description of a data struc-
ture that can be automatically created for you by xmlrpclib, the XML-RPC library that comes with
Python. It’s possible to make any kind of XML-RPC call using just one library (xmlrpclib).
By now, you’ll have noticed that Python is not very fastidious about types, and it will work with you on
transforming one type to another. That said, its built-in types cover just about everything for which
XML-RPC defines a representation: Booleans, integers, floating-point numbers, strings, arrays, and
dictionaries. For binary data and dates, xmlrpclib provides wrapper classes (Python got date/time
object support in version 2.3, but xmlrpclib hasn’t yet been updated to use it).
The XML-RPC spec, at www.xml-rpc.com/spec/, is short and sweet.
XML-RPC Quick Start: Get Tech News from Meerkat
Meerkat is a public web application that aggregates technology news from hundreds of weblogs and
news sites. It was one of the first web applications to expose a web service interface: first a RESTlike
interface and then an XML-RPC interface.
Meerkat’s XML-RPC interface is described at www.oreillynet.com/pub/a/rss/2000/11/14/
meerkat_xmlrpc.html.
Meerkat’s API exposes access to three types of objects: channels (weblogs and news sites), categories
(groupings of channels), and items (stories published by channels). Unfortunately, to use any of the func-
tions that deal with channels or categories, you must do some legwork ahead of time to ascertain the
numeric channel or category IDs. The most generally useful method is therefore getItems, a search func-
tion that tries to match your search criteria against Meerkat’s database of recently posted news items.
Here’s a simple script, MeerkatSummary.py, that takes a search criterion as input and determines which
Meerkat channels have the most stories that match the search:
import xmlrpclib
class MeerkatSummary:
“””Lists channels that match a search term, in order of how many
stories match.”””
SERVER_URL = ‘http://www.oreillynet.com/meerkat/xml-rpc/server.php’
def __init__(self):
“Set up a reference to the Meerkat server.”
509

56 Pages 551-560

▲back to top

56.1 Page 551

▲back to top
Chapter 21
#Passing ‘verbose=True’ to the server constructor will make it
#print the text of the request and response for each XML-RPC
#call, letting you see the internal workings of the protocol.
#verbose = True
verbose = False
server = xmlrpclib.ServerProxy(self.SERVER_URL, verbose=verbose)
self.meerkat = server.meerkat
def findChannels(self, searchTerm):
“Given a search term, find out which channels have the most hits.”
channelTotals = {}
items = self.meerkat.getItems({‘search’ : searchTerm,
‘channels’ : True})
for item in items:
channel = item[‘channel’]
totalForChannel = channelTotals.get(channel, 0)
totalForChannel += 1
channelTotals[channel] = totalForChannel
#Turn the map into a list of (matches, channel name) tuples, and sort it.
totalAndChannel = [(a,b) for b,a in channelTotals.items()]
totalAndChannel.sort()
totalAndChannel.reverse()
print ‘Meerkat report for “%s”:’ % searchTerm
for total, channel in totalAndChannel:
print “%2d %s” % (total, channel)
The actual web service call is self.meerkat.getItems, on the third line of MeerkatSummary.
findChannels. If you blink you’ll miss it, because as far as Python is concerned, it’s just another method
call — albeit one that’s implemented differently than a local method call. xmlrpclib defines a __call__
method for ServerProxy that handles the XML-RPC for getItems.
The previous section’s WishListBargainFinder also hid the complexity of a web service behind a
standard Python method: In that case, it was amazon.searchWishList that activated the REST web
service. The difference is that someone had to write a Python method called “searchWishList” that made
an AWS-specific REST request and processed the AWS-specific response. The getItems method is han-
dled by xmlrpclib — there’s no special code for dealing with the Meerkat XML-RPC server, no need for
an actual Python method called getItems:
if __name__ == ‘__main__’:
import sys
if len(sys.argv) != 2:
print “Usage: %s [search term]” % sys.argv[0]
sys.exit(1)
else:
MeerkatSummary().findChannels(sys.argv[1])
Run the script, and you’ll see a variety of news channels that have mentioned Python:
$ python MeerkatSummary.py Python
Meerkat report for “Python”:
22 Freshmeat Daily News
10 Python URL (daily updates)
8 Vaults of Parnassus
510

56.2 Page 552

▲back to top
Web Applications and Web Services
2 Python Cookbook
1 Zope org
1 SourceForge Project News
1 Python org latest headlines
1 NetBSD Packages
1 Linux Weekly News
1 Linux Today
1 IceWalkers
1 Developer Shed
James Joyce, who got lots of results from an Amazon Web Services query, doesn’t do as well here:
$ python MeerkatSummary.py “James Joyce”
Meerkat report for “James Joyce”:
1 kottke.org
1 Beyond the Beyond
Note that Meerkat’s getItems method will never return more than 50 results; and unlike with Amazon
Web Services, there’s no pagination interface that lets you get more. Any script you write that runs
against Meerkat will be limited to delving into the recent past.
The XML-RPC Request
The XML-RPC request body is the body of an HTTP POST request. It’s an XML document containing a
methodCall element. The methodCall element contains two elements of its own: methodName, which
designates the method to be called; and params, which contains a list of the parameters to be passed as
arguments into the method.
Here’s a sample XML-RPC request for a hypothetical method that sorts a list of numbers in either
ascending or descending order:
<?xml version=”1.0”?>
<methodCall>
<methodName>searchsort.sortList</methodName>
<params>
<param>
<value>
<array>
<data>
<value><i4>10</i4></value>
<value><i4>2</i4></value>
</data>
</array>
</param>
<param><value><boolean>1</boolean></param>
</params>
</methodCall>
This is the XML-RPC equivalent of invoking a hypothetical local method with the following code:
import searchsort
searchsort.sortList([10, 2], True)
511

56.3 Page 553

▲back to top
Chapter 21
Given what you know about xmlrpclib, it’s no surprise that this method request would be generated
and POSTed when you ran code like this:
import xmlrpclib
xmlrpclib.ServerProxy(“http://sortserver/RPC”).searchsort.sortList([10, 2], True)
Representation of Data in XML-RPC
The XML-RPC methodName can be any string, but XML-RPC methods are traditionally grouped into
named packages, such as searchsort in the preceding example. In a Python implementation, this
makes it look like a module called searchsort that contains the functions to expose, like sortList.
XML-RPC parameters can be any of the following types:
Data Type
Boolean True or False
A string
An integer
A floating-point number
An array (items can be of
any type, or mixed type)
A dictionary (keys must be
strings; values can be any type)
Sample XML-RPC Representation
<boolean>1</boolean>
<string>James Joyce</string>
<i4>10</i4>
<double>5.1</double>
<array>
<data>
<value><i4>10</i4></value>
<value><i4>2</i4></value>
</data>
</array>
<struct>
<member>
<name>search</name>
<value><string>James Joyce</string></value>
</member>
<member>
<name>channels</name>
<value><boolean>1</boolean></value>
</member>
</struct>
512

56.4 Page 554

▲back to top
Web Applications and Web Services
Data Type
A date and time
Binary data
Sample XML-RPC Representation
<dateTime.iso8601>20050614T19:11:20 </dateTime.iso8601>
(Use xmlrpclib’s DateTime wrapper class, which can be
instantiated from a time tuple, seconds since epoch, etc.)
<base64>AVRoaXMgaXMgYmluYXJ5IGRhdGEu</base64>
(Use xmlrpclib’s Binary wrapper class, which can be instan-
tiated from a string.)
Strongly typed languages can have problems with some of these: mixed-type arrays, for example.
Dynamic languages like Python handle these in stride.
The XML-RPC Response
The body of the XML-RPC response is an XML document describing the return value of the function
invoked by the request.
Assuming our hypothetical searchsort.sortList method does what it says, when invoked with the
sample body given earlier it’ll return a response that looks like this:
<?xml version=”1.0”?>
<methodResponse>
<params>
<param>
<value>
<array>
<data>
<value><i4>2</i4></value>
<value><i4>10</i4></value>
</data>
</array>
</value>
</param>
</params>
</methodResponse>
The response has the same basic structure as the request, but it’s sparser. It’s missing a methodName ele-
ment because it’s assumed you know which method you just called. It has a params element, just like the
request; but whereas the request’s params element could contain any number of param children (the argu-
ments to the method), the response list is only allowed to contain a single param child: the return value.
If Something Goes Wrong
A REST web service is expected to flag error conditions using HTTP status codes, in conjunction with
error documents that describe the problem. As you might expect, XML-RPC does a similar thing in a
more structured way.
513

56.5 Page 555

▲back to top
Chapter 21
If an XML-RPC server can’t complete a request for any reason, it will return a response containing a
fault, instead of one containing a return value in params. A sample fault response is as follows:
<?xml version=”1.0”?>
<methodResponse>
<fault>
<value>
<struct>
<member>
<name>faultCode</name>
<value><int>4</int></value>
</member>
<member>
<name>faultString</name>
<value><string>No such method: “searchSort.sortList”.</string></value>
</member>
</struct>
</value>
</fault>
</methodResponse>
The fault element describes an XML-RPC struct (i.e., a Python dictionary) with two members:
faultString, which contains a human-readable description of what went wrong, and faultCode, the
equivalent to the HTTP status code used to signify failure in REST contexts (even an XML-RPC call that
results in a fault response will have an HTTP status code of 200). The advantage of faultCodes is that
you can define them as you please for whatever problems are specific to your application. The disadvan-
tage is that, unlike with HTTP status codes, there’s no consensus as to what faultCodes mean. You’ll
need to reach an understanding with your users about the meanings of your service’s faultCodes.
Within Python, a response with a fault corresponds to an xmlrpclib.Fault object, a subclass of
Error. If you’re using Python’s XML-RPC libraries, you can just raise and catch exceptions normally,
instead of having to worry about creating or parsing XML-RPC faults.
Exposing the BittyWiki API through XML-RPC
If you doubt that Python programmers are spoiled, consider this: Not only does the language come bun-
dled with a library that makes it easy to write XML-RPC clients; it also comes bundled with an XML-RPC
server along the lines of the bundled SimpleHTTPServer and SimpleCGIServer. As with the other
Simple*Server classes, SimpleXMLRPCServer runs as a standalone web server on its own port.
However, the XML-RPC functionality is also available as a CGI program that accepts HTTP POSTs in
XML-RPC format. This is implemented in another class, CGIXMLRPCRequestHandler, the name of which
probably has more consecutive capital letters than any other class name in the Python standard library.
CGIXMLRPCRequestHandler is only included with Python versions 2.3 and later. However, you
should be able to use Python 2.3’s copy of SimpleXMLRPCServer.py with Python version 2.2. Just
copy it into a local directory, give it a different name, and import it from 2.2. Remember, of course, when
you upgrade to Python 2.3, to get rid of your copy and start using the “real” copy.
514

56.6 Page 556

▲back to top
Web Applications and Web Services
Here’s a script, bittywiki-xmlrpc.cgi, that will expose the BittyWiki API either through an XML-
RPC CGI (if you invoke it without command-line arguments, the way a CGI-enabled web server would)
or through a standalone XML-RPC server (if you pass it the port to use on the command line):
If you’re using the EasyCGIServer presented earlier, or another server based on Python’s
CGIHTTPServer, using this script as a CGI may not work for you. If you run into problems with the
CGI, try using another web server, such as Apache, or running a standalone XML-RPC server instead
of going through a CGI.
import sys
import SimpleXMLRPCServer
from BittyWiki import Wiki
class BittyWikiAPI:
“””A simple wrapper around the basic BittyWiki functionality we
want to expose to the API.”””
def __init__(self, wikiBase):
“Initialize a wiki located in the given directory.”
self.wiki = Wiki(wikiBase)
def getPage(self, pageName):
“Returns the text of the given page.”
page = self.wiki.getPage(pageName)
if not page.exists():
raise NoSuchPage, page.name
return page.getText()
def save(self, pageName, newText):
“Saves a page of the wiki.”
page = self.wiki.getPage(pageName)
page.text = newText
page.save()
return “Page saved.”
def delete(self, pageName):
“Deletes a page of the wiki.”
page = self.wiki.getPage(pageName)
if not page.exists():
raise NoSuchPage, pageName
page.delete()
return “Page deleted.”
class NoSuchPage(Exception):
pass
So far, nothing XML-RPC specific — just a nicely packaged interface to the three basic functions of the
BittyWiki API. Next, we write a function that exposes those three functions to XML-RPC. There are two
ways of doing this: You can register functions one at a time or register an object instance, which registers
all that object’s methods at once. This example provides code for both ways of registering the methods,
but the instance registration is commented out, because in earlier versions of Python it exposed a secu-
rity vulnerability:
515

56.7 Page 557

▲back to top
Chapter 21
def handlerSetup(handler, api):
“””This function registers the methods of the BittyWiki API
as functions of an XML-RPC handler.”””
#Register the standard functions used by XML-RPC to advertise which methods
#are available on a given server.
handler.register_introspection_functions()
#Register the BittyWiki API methods as XML-RPC functions in the
#’bittywiki’ namespace.
handler.register_function(api.getPage, ‘bittywiki.getPage’)
handler.register_function(api.save, ‘bittywiki.save’)
handler.register_function(api.delete, ‘bittywiki.delete’)
#Here’s a way of registering all three methods in one line of
#code, by registering the API object itself. It’s commented out
#because it exposes a security hole in older versions of Python
#(specifically: 2.2, 2.3.4 and 2.4.0). See
#http://www.python.org/security/PSF-2005-001/ for details.
#
#handler.register_instance(api)
#
#Note also that when you do this, the functions exposed to XML-RPC
#will not have the ‘bittywiki.’ prefix: they’ll be named,
#e.g. “getPage” instead of “bittywiki.getPage”. If this disturbs
#you, try something like this instead:
#class Container:
# pass
#container = Container()
#container.bittywiki = api
#handler.register_instance(container)
Finally, the script portion, which starts up either a standalone XML-RPC server that can serve any num-
ber of requests, or a CGI-based XML-RPC script, which serves only the current request:
if __name__ == ‘__main__’:
WIKI_BASE = ‘wiki/’
api = BittyWikiAPI(WIKI_BASE)
standalonePort = None
if len(sys.argv) > 1:
#The user provided a port number; that means they want to
#run a standalone server.
standalonePort = sys.argv[1]
try:
standalonePort = int(standalonePort)
except ValueError:
#Oops, that wasn’t a port number. Chide the user and exit.
scriptName = sys.argv[0]
print ‘Usage:’
print ‘ “%s [port number]” to start a standalone server.’ \\
% scriptName
print ‘ “%s” to invoke as a CGI.’ % scriptName
sys.exit(1)
isStandalone = 1
print “Starting up standalone XML-RPC server on port %s.” \\
516

56.8 Page 558

▲back to top
Web Applications and Web Services
% standalonePort
handler = SimpleXMLRPCServer.SimpleXMLRPCServer\\
((‘localhost’, standalonePort))
else:
#No port number specified; this is a CGI invocation.
handler = SimpleXMLRPCServer.CGIXMLRPCRequestHandler()
#The function registration step is identical for
#SimpleXMLRPCServer and CGIXMLRPCRequestHandler.
handlerSetup(handler, api)
if standalonePort:
handler.serve_forever()
else:
handler.handle_request()
Try It Out Manipulating BittyWiki through XML-RPC
It’s now possible to make XML-RPC calls against BittyWiki from other machines and even other
programming languages, just as we were earlier making XML-RPC calls against Meerkat (which is
written in PHP).
In one window, start the standalone XML-RPC server (alternatively, make sure the web server that
serves the XML-RPC CGI is running):
# python BittyWiki-XMLRPC.py 8001
Starting up standalone XML-RPC server on port 8001.
In another, start an interactive Python session:
>>> import xmlrpclib
>>> server = xmlrpclib.ServerProxy(“http://localhost:8001/”)
>>> #Alternatively, something like:
... #server = xmlrpclib.ServerProxy(“http://localhost/cgi-bin/bittywiki-
xmlrpc.cgi”)
>>> bittywiki = server.bittywiki
>>> bittywiki.getPage(“CreatedByXMLRPC”)
Traceback (most recent call last):
File “<stdin>”, line 1, in ?
...
raise Fault(**self._stack[0])
xmlrpclib.Fault: <Fault 1: ‘No such page:CreatedByXMLRPC’>
>>> bittywiki.save(“CreatedByXMLRPC”, “This page was created through the XML-RPC
interface.”)
‘Page saved.’
>>> bittywiki.getPage(“CreatedByXMLRPC”)
‘This page was created through the XML-RPC interface.’
You’re using web services, but you didn’t have to write special client code or (except at the beginning,
when you connected to the server) even be aware that you’re using web services. Of course, the changes
you make to the wiki through this interface will also show up for people using the web application or
BittyWiki’s REST-based web service.
517

56.9 Page 559

▲back to top
Chapter 21
Wiki Search-and-Replace Using the XML-RPC Web Service
Remember WikiSpiderREST.py, the script that crawled BittyWiki pages using its REST API to perform
search-and-replace operations? You had to write a custom class (BittyWikiRESTAPI) to construct the
right URLs to use against the REST interface, and a custom XML parser to process the response docu-
ments you got in return. Of course, once you have written that stuff, it can be reused in any application
that uses BittyWiki’s REST API, but the main selling point of XML-RPC is that such classes aren’t neces-
sary: xmlrpclib handles everything. Let’s put that to the test by rewriting WikiSpiderREST.py as
WikiSpiderXMLRPC.py:
#!/usr/bin/python
import re
import xmlrpclib
class WikiReplaceSpider:
“A class for running search-and-replace against a web of wiki pages.”
WIKI_WORD = re.compile(‘(([A-Z][a-z0-9]*){2,})’)
def __init__(self, rpcURL):
“Accepts a URL to a BittyWiki XML-RPC API.”
server = xmlrpclib.ServerProxy(rpcURL)
self.api = server.bittywiki
def replace(self, find, replace):
“””Spider wiki pages starting at the front page, accessing them
and changing them via the XML-RPC API.”””
processed = {} #Keep track of the pages already processed.
todo = [‘HomePage’] #Start at the front page of the wiki.
while todo:
for pageName in todo:
print ‘Checking “%s”’ % pageName
try:
pageText = self.api.getPage(pageName)
except xmlrpclib.Fault, fault:
if fault.faultString.find(“No such page”) == 0:
#We tried to access a page that doesn’t exist;
#not a big deal.
pass
else:
#Some other problem; pass it on up.
raise xmlrpclib.Fault, fault
else:
#This page actually exists; process it.
#First, find any WikiWords in this page: they may
#reference other pages.
for wikiWord in self.WIKI_WORD.findall(pageText):
linkPage = wikiWord[0]
if not processed.get(linkPage) and linkPage not in todo:
#We haven’t processed this page yet: put it on
#the to-do list.
todo.append(linkPage)
518

56.10 Page 560

▲back to top
Web Applications and Web Services
#Run the search-and-replace on the page text to get the
#new text of the page.
newText = pageText.replace(find, replace)
#Check to see if this page name matches the search
#string. If it does, delete it and recreate it
#with the new text; otherwise, just save the new
#text in the existing page.
newPageName = pageName.replace(find, replace)
if newPageName != pageName:
print ‘ Deleting “%s”, will recreate as “%s”’ \\
% (pageName, newPageName)
self.api.delete(pageName)
if newPageName != pageName or newText != pageText:
print ‘ Saving “%s”’ % newPageName
saveResponse = self.api.save(newPageName, newText)
#Mark the new page as processed so we don’t go through
#it a second time.
if newPageName != pageName:
processed[newPageName] = True
processed[pageName] = True
todo.remove(pageName)
The WikiReplaceSpider class looks almost exactly the same as before. The only big difference is that,
whereas before a method call like api.getPage moved into custom REST code you had to write, it
now moves into pre-existing xmlrpclib code. Without those API-specific classes to implement, the
WikiReplaceSpider class is pretty much all the code:
if __name__ == ‘__main__’:
import sys
if len(sys.argv) == 4:
rpcURL, find, replace = sys.argv[1:]
else:
print ‘Usage: %s [URL to BittyWiki XML-RPC API] [find] [replace]’ \\
% sys.argv[0]
sys.exit(1)
WikiReplaceSpider(rpcURL).replace(find, replace)
That’s it. This spider works just like the REST version, but it takes less code because there’s no one-off
code to deal with the specifics of the REST API. This script is run just like the REST version, but the URL
passed in is the URL to the XML-RPC interface, instead of the URL to the REST interface:
$ python WikiSpiderXMLRPC.py http://localhost:8000/cgi-bin/bittywiki-xmlrpc.cgi
Foo Bar
Checking “HomePage”
Saving “HomePage”
Checking “FooCaseStudies”
...
519

57 Pages 561-570

▲back to top

57.1 Page 561

▲back to top
Chapter 21
SOAP
XML-RPC solves REST’s main problem by defining a standard way to represent data types such as inte-
gers, dates, and lists. However, while XML-RPC was being defined, the W3C’s XML working group was
working on its own representation of those data types and many others. After XML-RPC became popu-
lar, the W3C turned their attention to it, and started redesigning it to use their preexisting standards.
Along the way, ambition broadened the scope of their project to include any kind of message exchange,
not just procedure calls and their return values. The result was SOAP. The acronym originally stood for
Simple Object Access Protocol, but because the standard’s scope has been expanded so far beyond sim-
ple remote procedure calls, the acronym itself is no longer applicable.
SOAP may still be simple compared to COM or CORBA, but it’s a lot more complicated than XML-RPC.
Fortunately, you don’t need all of SOAP just to expose a web application as a web service. The part you
do need looks basically like XML-RPC with a more general XML encoding scheme. SOAP gives you
access to a broader range of data types than XML-RPC, and it lets you define your own, but this intro-
ductory treatment will just treat it as an analogue of XML-RPC.
SOAP Quick Start: Surfing the Google API
Just as with REST and XML-RPC, a SOAP message is typically sent as the data portion of an HTTP POST
request. Just as with those other protocols, then, it’s technically possible to use a SOAP web service with-
out any SOAP-specific tools: Just construct the message by hand, send it off with urllib, and parse the
response with the xml.sax module. Realistically, though, you need a SOAP library to use SOAP with
Python. A SOAP library will deal with transforming Python data structures to SOAP’s XML representa-
tions and back, just as xmlrpclib does for XML-RPC.
Unfortunately, there’s no “soaplib” bundled with Python, but you can download one. There are two
SOAP libraries for Python. The one library we’ll use in this chapter is SOAPpy, which provides an
xmlrpclib-like version of a SOAP client and a SOAP server.
If you’re running Debian GNU/Linux, you can just install the “soappy” package; if not, you can down-
load the distribution from http://pywebsvcs.sourceforge.net/. ZSI, the other Python SOAP
package, is also available from that site. Be warned that SOAPpy requires two other packages: fpconst,
a floating-point library, and PyXML, a set of XML utilities. More information and links to the packages
are available in the SOAPpy README file.
To start using SOAPpy, you’ll need a SOAP service to call. As it happens, Google Web Services, one of the
most celebrated web service APIs, uses SOAP. Google’s SOAP API gives you access to the data set used
by their search engine.
To use the APIs, you’ll need to sign up with Google for an API key, just as with Amazon in the REST
example. Also as with Amazon, Google puts limitations on your use of their API. Whereas Amazon
allows only one API call per second per IP address, Google limits you to one thousand API calls per day
per API key. Be careful while experimenting with this service; a script that goes out of control could
leave you unable to make any more API calls until the next day.
Sign up for a Google API key at http://api.google.com/createkey.
520

57.2 Page 562

▲back to top
Web Applications and Web Services
While you’re at Google’s site, you can also download the API developers’ kit. The kit is oriented towards
Java and .NET programmers, but it does contain a useful human-readable API reference, as well as a
WSDL file, which describes the API in machine-parseable form. (See the “WSDL” section later for more
information on WSDL.) After you’ve got an API key, you’re ready to write some scripts. Here’s a simple
one, GoogleSearch.py, that does a Google search from the command line and prints the results:
import SOAPpy
class GoogleAPI:
“Implements part of the Google Web API as a simple Python class.”
URL = ‘http://api.google.com/search/beta2’
NAMESPACE = ‘urn:GoogleSearch’
def __init__(self):
self.server = SOAPpy.SOAPProxy(self.URL, self.NAMESPACE)
SOAPProxy acts a lot like xmlrpclib’s ServerProxy class: It overrides __call__ to transform method
calls into web service calls. The only difference so far is that whereas a ServerProxy operates against a
server URL, a SOAPProxy also operates against a namespace. SOAP is big on namespaces, as you’ll see.
Here, the namespace is used to identify a provider of web service methods. One server might provide
two completely different web services, both of which implement a method called doFoo. Each web ser-
vice has its own namespace, though, which enables the server to distinguish between your doFoo and
your neighbor’s doFoo.
Recall that in XML-RPC interfaces, this problem is traditionally solved by qualifying the method names
with package names, as in “bittywiki.getPage”. In the SOAP interface, that method would be called
just plain “getPage” but it would be executed in a namespace like ‘urn:BittyWiki’ to distinguish it
from the “getPage” methods provided by other web services:
#These two commands will make SOAPpy print the raw request and
#response for each SOAP call, letting you see the internal
#workings of the protocol.
#self.server.config.dumpSOAPOut=1
#self.server.config.dumpSOAPIn=1
def doGoogleSearch(self, key, searchString, resultOffset=0, maxResults=10,
filter=True, restrict=””, safeSearch=True,
languageRestrict=”en”):
“””A convenience method to hide the fact that a call to
doGoogleSearch requires ten arguments, two of which are
deprecated and shouldn’t be used. By calling this method you
can do a search by providing only your Google API key and the
search string. For the meanings of the other arguments,
see the reference for the Google Web APIs.”””
return self.server.doGoogleSearch(key, searchString, resultOffset,
maxResults, filter, restrict,
safeSearch, languageRestrict, “”, “”)
521

57.3 Page 563

▲back to top
Chapter 21
Just as with XML-RPC, a method call on the server object silently spawns an XML document depicting
the method you want to call, and the values you’re providing for that method’s arguments. This XML
document is submitted to the server via HTTP POST. The HTTP response is a second XML document
depicting a data structure, which is parsed and used to construct an actual Python data structure. The
data structure is what is returned to you:
if __name__ == ‘__main__’:
import sys
if len(sys.argv) != 3:
print “Usage: %s [Google API key] [Search term]” % sys.argv[0]
sys.exit(1)
key, term = sys.argv[1:3]
resultObj = GoogleAPI().doGoogleSearch(key, term)
results = resultObj.resultElements
print ‘First %s result(s) for “%s”:’ % (len(results), term)
for result in results:
print “ %s: %s” % (result.title, result.URL)
Running this script gives you search results on standard output:
$ python GoogleSearch.py [Your Google API key] “Python and SOAP”
First 10 result(s) for “Python and SOAP”:
The <b>Python</b> Web services developer: <b>Python</b> <b>SOAP</b> libraries
http://www-106.ibm.com/developerworks/library/ws-pyth5/
The <b>Python</b> Web services developer: <b>Python</b> <b>SOAP</b> libraries, Part 2
http://www-106.ibm.com/developerworks/library/ws-pyth6/
...
Scripting the Web with <b>Python</b> : Specs &gt; <b>SOAP</b> &gt; Implementations
http://python.scripting.com/directory/13/specs/soap/implementations
[PythonCE] using <b>Python</b> 2.2 + <b>SOAP</b> on Windows CE
http://mail.python.org/pipermail/pythonce/2002-May/000069.html
It’s useful for didactic purposes to show you how to make SOAP calls directly, and often you may have
no other choice. If you’re serious about using the Google API with Python, though, it’s better to use
googlepy, a Google-specific SOAP client library for Python. It’s available at http://pygoogle.
sourceforge.net/, and it’s a lot simpler than calling the SOAP methods directly.
The SOAP Request
Here’s a transcript of a hypothetical SOAP RPC call that tries to sort a list in ascending order; compare it
to the XML-RPC transcript earlier that called an XML-RPC version of the same method:
<?xml version=”1.0” encoding=”UTF-8”?>
<SOAP-ENV:Envelope SOAP-
ENV:encodingStyle=”http://schemas.xmlsoap.org/soap/encoding/”
xmlns:SOAP-ENV=”http://schemas.xmlsoap.org/soap/envelope/”
xmlns:SOAP-ENC=”http://schemas.xmlsoap.org/soap/encoding/”
522

57.4 Page 564

▲back to top
Web Applications and Web Services
xmlns:xsi=”http://www.w3.org/1999/XMLSchema-instance”
xmlns:xsd=”http://www.w3.org/1999/XMLSchema”>
<SOAP-ENV:Body>
<ns1:sortList xmlns:ns1=”urn:SearchSort” SOAP-ENC:root=”1”>
<v1 SOAP-ENC:arrayType=”xsd:int[2]” xsi:type=”SOAP-ENC:Array”>
<item>10</item>
<item>2</item>
</v1>
<v2 xsi:type=”xsd:boolean”>True</v2>
</ns1:sortList>
</SOAP-ENV:Envelope>
The first thing to notice is all those xmlns declarations. SOAP is very particular about XML namespaces,
whereas XML-RPC is much more informal and serves standalone XML documents. SOAP uses XML
namespaces to define the format of the SOAP message itself (SOAP-ENV), the data types (such as
xsd:boolean and the SOAP-specific SOAP-ENC:Array), and the very concept of a data type (xsi:type).
This gives SOAP a lot more flexibility in how its data is encoded, but between XML Schema (xsd) and
the SOAP data encoding schema (SOAP-ENC), most of the basic data types are already defined for you.
Only in more complicated cases will you need to define custom data types.
The other namespace mentioned in this message is urn:SearchSort. That’s the namespace of the
method you’re trying to call. As mentioned before, this is like the way the XML-RPC version of this
request named its method searchsort.sortList, instead of just sortList. SOAP has formalized the
XML-RPC convention, and uses XML namespaces to distinguish between different methods with the
same name. Your SOAP call must be executed in a particular XML namespace. If you use a Python SOAP
library to make SOAP calls, this is probably the only namespace you’ll actually have to worry about.
If you ignore the namespaces, this message looks a lot like the XML-RPC message you saw earlier.
There’s a method call tag that contains a list of tags for the arguments to be passed into the method.
Instead of the method call tag containing a child tag with the method name, here the tag is simply
named after the method to be called. In XML-RPC, the arguments were listed inside a separate params
tag. Here, they’re direct children of the method call tag. The SOAP message is a little more concise, but
(again, disregarding the namespace declarations) just as easy to read.
Compare the XML-RPC representation of the array to be sorted, which you saw earlier, to the SOAP rep-
resentation of the same array:
<array>
<data>
<value><i4>2</i4></value>
<value><i4>10</i4></value>
</data>
</array>
<v1 SOAP-ENC:arrayType=”xsd:int[2]” xsi:type=”SOAP-ENC:Array”>
<item>10</item>
<item>2</item>
</v1>
523

57.5 Page 565

▲back to top
Chapter 21
This difference between the two protocols is typical. There’s more up-front definition in SOAP and more
references to external documents that formally define the data types. The upside of that is that once the
definition is done, it takes fewer bytes to actually define a data structure. It doesn’t make much differ-
ence with a small array like this, but consider an array with thousands or millions of elements. SOAP is
more efficient than XML-RPC at representing large data structures.
The SOAP Response
Here’s a possible response you might get from a SOAP server after sending it the sortList request:
<?xml version=”1.0” encoding=”UTF-8”?>
<SOAP-ENV:Envelope SOAP-
ENV:encodingStyle=”http://schemas.xmlsoap.org/soap/encoding/”
xmlns:SOAP-ENV=”http://schemas.xmlsoap.org/soap/envelope/”
xmlns:SOAP-ENC=”http://schemas.xmlsoap.org/soap/encoding/”
xmlns:xsi=”http://www.w3.org/1999/XMLSchema-instance”
xmlns:xsd=”http://www.w3.org/1999/XMLSchema”>
<SOAP-ENV:Body>
<ns1:sortList xmlns:ns1=”urn:SearchSort” SOAP-ENC:root=”1”>
<return SOAP-ENC:arrayType=”xsd:int[2]” xsi:type=”SOAP-ENC:Array”>
<item>2</item>
<item>10</item>
</return>
</ns1:sortList>
</SOAP-ENV:Envelope>
Just as with XML-RPC, a SOAP response has the same basic structure as a SOAP request. Where the
SOAP request had a list of arguments, the SOAP response has a single return value. This, too, is similar
to XML-RPC: Recall that an XML-RPC response contained a params list, which was only allowed to con-
tain one param — the return value. SOAP makes this convention more natural by eliminating the
params tag and just returning the return value.
If Something Goes Wrong
If you make a SOAP request that makes the server code throw an exception, the Body of the response
you get back will contain a Fault element. It might look something like this:
</SOAP-ENV:Body>
<SOAP-ENV:Fault SOAP-ENC:root=”1”>
<faultcode>SOAP-ENV:Client</faultcode>
<faultstring>No method urn:SearchSort:sortList found</faultstring>
<detail xsi:type=”xsd:string”>
There’s no method “sortList” in the urn:SearchSort namespace.
</detail>
</SOAP-ENV:Fault>
</SOAP-ENV:Body>
The faultstring and detail sub-elements of Fault are for human-readable descriptions, and the
faultcode element describes the type of error. Whereas XML-RPC says nothing about the fault code
except that it must be an integer, SOAP defines four standard strings to serve as fault codes. Two of them
524

57.6 Page 566

▲back to top
Web Applications and Web Services
(mustUnderstand and VersionMismatch) you probably won’t encounter in basic SOAP use. The other
two fault codes serve, appropriately enough, to identify who caused the fault. If you’re writing a SOAP
client and you get a faultcode of Client (Receiver in SOAP version 1.2), that means you caused the
error (for instance, in the preceding, by calling a method that doesn’t exist in the namespace you speci-
fied). If the faultcode is Server (Sender in SOAP version 1.2), that means there’s nothing wrong with
your request but the server can’t fulfill it at the moment — perhaps the server code can’t access a
database or some other necessary resource.
Within a Python interface, the details of a response with a Fault are hidden from you, pretty much as in
XML-RPC. If a Python method you’ve exposed through SOAP throws an exception, the SOAP server
will automatically transform the exception into a SOAP response with a Fault element. If you’re using
SOAPpy and you call a remote method that responds with a Fault, it’ll be transformed into a subclass of
Error: SOAPpy.Types.faultType.
Exposing a SOAP Interface to BittyWiki
In principle, there’s no reason why you shouldn’t be able to run a SOAP server from a CGI script:
Remember that despite all the additional complexity and mystique of SOAP, it’s just like REST and
XML-RPC in that it’s just a document being POSTed to a URL and another document being sent in
return. Unfortunately, SOAPpy doesn’t provide a CGI script that serves SOAP requests, only a standalone
server, SOAPServer.
ZSI, the other SOAP implementation for Python, does offer a CGI-based server.
The following sample script, BittyWiki-SOAPServer.py, exposes the BittyWiki interface to SOAP
using a standalone server. This file should go into the same directory as the file BittyWiki.py, so that
you can use the core BittyWiki engine. Alternatively, you can put BittyWiki.py into one of the directo-
ries in your PYTHON_PATH so you can use it from anywhere:
#!/usr/bin/python
import sys
import SOAPpy
from BittyWiki import Wiki
class BittyWikiAPI:
“””A simple wrapper around the basic BittyWiki functionality we
want to expose to the API.”””
def __init__(self, wikiBase):
“Initialize a wiki located in the given directory.”
self.wiki = Wiki(wikiBase)
def getPage(self, pageName):
“Returns the text of the given page.”
page = self.wiki.getPage(pageName)
if not page.exists():
raise NoSuchPage, page.name
return page.getText()
def save(self, pageName, newText):
“Saves a page of the wiki.”
525

57.7 Page 567

▲back to top
Chapter 21
page = self.wiki.getPage(pageName)
page.text = newText
page.save()
return “Page saved.”
def delete(self, pageName):
“Deletes a page of the wiki.”
page = self.wiki.getPage(pageName)
if not page.exists():
raise NoSuchPage, page.name
page.delete()
return “Page deleted.”
class NoSuchPage(Exception):
“””An exception thrown when a caller tries to access a page that
doesn’t exist.”””
pass
The actual API code is exactly the same as for the XML-RPC server; it could even be moved into
a common library. The only difference is that now we register it with a SOAPServer instead of a
SimpleXMLRPCServer:
DEFAULT_PORT = 8002
NAMESPACE = ‘urn:BittyWiki’
WIKI_BASE = ‘wiki/’
if __name__ == ‘__main__’:
api = BittyWikiAPI(WIKI_BASE)
port = DEFAULT_PORT
if len(sys.argv) > 1:
port = sys.argv[1]
try:
port = int(port)
except ValueError:
#Oops, that wasn’t a port number. Chide the user and exit.
print ‘Usage: “%s [optional port number]”’ % sys.argv[0]
sys.exit(1)
print “Starting up standalone SOAP server on port %s.” % port
handler = SOAPpy.SOAPServer((‘localhost’, port))
handler.registerObject(api, NAMESPACE)
handler.serve_forever()
Try It Out Manipulating BittyWiki through SOAP
In one window, start the standalone SOAP server:
$ python BittyWiki-SOAPServer.py 8002
Starting up standalone XML-RPC server on port 8002.
In another, start an interactive Python session:
526

57.8 Page 568

▲back to top
Web Applications and Web Services
>>> import SOAPpy
>>> bittywiki = SOAPpy.SOAPProxy(“http://localhost:8002/”, “urn:BittyWiki”)
>>> bittywiki.getPage(“CreatedBySOAP”)
<Fault SOAP-ENV:Server: Method urn:BittyWiki:getPage failed.: __main__.NoSuchPage
CreatedBySOAP>
Traceback (most recent call last):
File “<stdin>”, line 1, in ?
...
SOAPpy.Types.faultType: <Fault SOAP-ENV:Server: Method urn:BittyWiki:getPage
failed.: __main__.NoSuchPage CreatedBySOAP>
>>> bittywiki.save(“CreatedBySOAP”, “This page was created through the SOAP
interface.”)
‘Page saved.’
>>> bittywiki.getPage(“CreatedBySOAP”)
‘This page was created through the SOAP interface.’
The experience of using SOAP, hidden behind SOAPpy, is similar to the experience of using XML-RPC,
hidden behind xmlrpclib. You can make method calls, passing in standard Python objects, and let the
library take care of all the details.
Wiki Search-and-Replace Using the SOAP Web Service
Here’s WikiSpiderSOAP.py, another wiki search-and-replace client similar to the ones described earlier
for BittyWiki’s REST and XML-RPC interfaces. By now, this code should be familiar. The pattern is
always the same: Set up some reference to the basic BittyWiki API and run the basic search-and-replace
spider algorithm using it. The only major difference between this version and the XML-RPC version is
the exception handling: xmlrpclib and SOAPpy act differently when something goes wrong on the
server side, so the exception handling code must be different. Other than that, the SOAP-based search-
and-replace spider looks more or less the same as the XML-RPC one:
#!/usr/bin/python
import re
import SOAPpy
class WikiReplaceSpider:
“A class for running search-and-replace against a web of wiki pages.”
WIKI_WORD = re.compile(‘(([A-Z][a-z0-9]*){2,})’)
def __init__(self, rpcURL):
“Accepts a URL to a BittyWiki SOAP API.”
self.api = SOAPpy.SOAPProxy(rpcURL, “urn:BittyWiki”)
self.api.config.dumpSOAPIn=1
def replace(self, find, replace):
“””Spider wiki pages starting at the front page, accessing them
and changing them via the XML-RPC API.”””
processed = {} #Keep track of the pages already processed.
todo = [‘HomePage’] #Start at the front page of the wiki.
while todo:
for pageName in todo:
print ‘Checking “%s”’ % pageName
527

57.9 Page 569

▲back to top
Chapter 21
try:
pageText = self.api.getPage(pageName)
except SOAPpy.Types.faultType, fault:
if fault.detail.find(“NoSuchPage”) != -1:
#Some page mentioned a WikiWord that doesn’t exist
#yet; not a big deal.
pass
else:
#Some other problem; pass it on up.
raise SOAPpy.Types.faultType, fault
else:
#This page actually exists; process it.
#First, find any WikiWords in this page: they may
#reference other existing pages.
for wikiWord in self.WIKI_WORD.findall(pageText):
linkPage = wikiWord[0]
if not processed.get(linkPage) and linkPage not in todo:
#We haven’t processed this page yet: put it on
#the to-do list.
todo.append(linkPage)
#Run the search-and-replace on the page text to get the
#new text of the page.
newText = pageText.replace(find, replace)
#Check to see if this page name matches the search
#string. If it does, delete it and recreate it
#with the new text; otherwise, just save the new
#text in the existing page.
newPageName = pageName.replace(find, replace)
if newPageName != pageName:
print ‘ Deleting “%s”, will recreate as “%s”’ \\
% (pageName, newPageName)
self.api.delete(pageName)
if newPageName != pageName or newText != pageText:
print ‘ Saving “%s”’ % newPageName
self.api.save(newPageName, newText)
#Mark the new page as processed so we don’t go through
#it a second time.
if newPageName != pageName:
processed[newPageName] = True
processed[pageName] = True
todo.remove(pageName)
if __name__ == ‘__main__’:
import sys
if len(sys.argv) == 4:
rpcURL, find, replace = sys.argv[1:]
else:
print ‘Usage: %s [URL to BittyWiki SOAP API] [find] [replace]’ \\
% sys.argv[0]
sys.exit(1)
WikiReplaceSpider(rpcURL).replace(find, replace)
528

57.10 Page 570

▲back to top
Web Applications and Web Services
This spider works just like the REST and the XML-RPC versions described earlier in this chapter:
$ python WikiSpiderSOAP.py http://localhost:8002/ Foo Bar
Checking “HomePage”
Saving “HomePage”
Checking “FooCaseStudies”
...
Note that because BittyWiki-SOAPServer.py runs its own web server, there’s no need to point to a
script somewhere on the web server that handles the SOAP interface. The entire web server is the SOAP
interface.
Documenting Your Web Ser vice API
Exposing a web service API won’t do any good unless the people who want to write robots can figure out
how to use it. If you were to distribute a Python module with inadequate documentation (shame on you),
a determined user could try to figure out the API by looking at the source code and, if necessary, making
experimental changes, learning through trial and error. That isn’t possible when you expose a web ser-
vice, so it’s especially important that you have a real way of getting the API information to your users.
Human-Readable API Documentation
In my opinion, no matter which web service protocol you’re using, nothing beats an up-to-date human-
readable description of an API. This can be written manually or generated through introspection and the
use of Python docstrings. Up next are three sample documents that describe the three web service APIs
for the BittyWiki application created in this chapter. They’re all extremely short, but they contain all the
information a user needs to write an application using any of them.
The BittyWiki REST API Document
To get the raw wiki markup for the page “WikiPage”, GET the URL http://localhost:8000/
cgi-bin/bittywiki-rest.cgi/WikiPage. You’ll get an XML data structure in which the <data> tag
contains the wiki markup of the WikiPage page. If the WikiPage page doesn’t exist, you’ll get an error.
To modify the contents of the page “WikiPage”, POST to the URL http://localhost:8000/cgi-bin/
bittywiki-rest.cgi/WikiPage. Set data equal to the wiki markup you want to write to the page,
and operation to the string write. You’ll receive an XML data structure in which the <message> tag
contains a status message. If the WikiPage page doesn’t exist, it will be automatically created.
To delete the page “WikiPage”, POST to the URL http://localhost:8000/cgi-bin/bittywiki-rest.
cgi/WikiPage. Set “operation” to the string delete. You’ll receive an XML data structure in which the
<message> tag contains a status message. If the WikiPage page doesn’t exist, you’ll get an error.
The BittyWiki XML-RPC API Document
The BittyWiki API server is located at http://localhost:8001/. It exposes three methods:
❑ bittywiki.getPage(string pageName): Returns the text of the named page. Passing an empty string
designates the wiki home page. This will throw a fault if you request a page that doesn’t exist.
529

58 Pages 571-580

▲back to top

58.1 Page 571

▲back to top
Chapter 21
❑ bittywiki.save(string pageName, string text): Sets the text of the named page. If the page
doesn’t already exist, it’ll be automatically created.
❑ bittywiki.delete(string pageName): Deletes the named page. This will throw a fault if you try
to delete a page that doesn’t exist.
The BittyWiki SOAP API Document
The BittyWiki SOAP server is located at http://localhost:8002/. It exposes three methods in the
namespace “urn:BittyWiki”:
❑ getPage(string pageName): Returns the text of the named page. Passing an empty string desig-
nates the wiki homepage. This will throw a fault if you request a page that doesn’t exist.
❑ save(string pageName, string text): Sets the text of the named page. If the page doesn’t already
exist, it will be automatically created.
❑ delete(string pageName): Deletes the named page. This will throw a fault if you try to delete a
page that doesn’t exist.
The XML-RPC Introspection API
An unofficial addendum to the XML-RPC specification defines three special functions in the “system”
namespace, as a convenience to users who might not know which functions an XML-RPC server supports,
or what those functions might do. These special functions are the web service equivalent of Python’s ever-
useful dir and help commands. Both SimpleXMLRPCServer and CGIXMLRPCRequestHandler support
two of the three introspection functions, assuming you call the register_introspection_functions
method on the server or handler object after instantiating it:
handler = SimpleXMLRPCServer.SimpleXMLRPCServer((host, port))
handler.register_introspection_functions()
Method Name
system.listMethods()
system.methodHelp(string funcName)
system.methodSignature(string funcName)
What It Does
Returns the names of all the functions the server
makes available
Returns a string with documentation for the named
function. Implemented in Python by returning the
function’s Python docstring.
Returns the signature and return type of the
named function. Not automatically supported by
the Python implementation because Python func-
tion definitions don’t include type information.
Try It Out Using the XML-RPC Introspection API
Start up and connect to the BittyWiki XML-RPC server (or CGI) as before. In addition to the BittyWiki
methods shown earlier, you can use the XML-RPC introspection methods:
530

58.2 Page 572

▲back to top
Web Applications and Web Services
>>> import xmlrpclib
>>> server = xmlrpclib.ServerProxy(“http://localhost:8001/”)
>>> server.system.listMethods()
[‘bittywiki.delete’, ‘bittywiki.getPage’, ‘bittywiki.save’, ‘system.listMethods’,
‘system.methodHelp’, ‘system.methodSignature’]
>>> server.system.methodHelp(“bittywiki.save”)
‘Saves a page of the wiki.’
>>> server.system.methodSignature(“bittywiki.save”)
‘signatures not supported’
XML-RPC introspection isn’t meant as a substitute for a human-readable API document. For one thing,
it’s hard to get people excited about using your API if they must use XML-RPC method calls to even see
what it is. However, the introspection API does make it a lot easier to experiment with an XML-RPC web
service from an interactive Python shell.
A Python-centric reference to the XML-RPC introspection API is at www.python.org/doc/lib/
serverproxy-objects.html.
WSDL
Many SOAP-based web services define their interface in a WSDL file. WSDL is basically a machine-
parseable version of the human-readable API document shown earlier in this section.
Recall that XML-RPC defines a set of rules for transforming a few basic data structures into XML docu-
ments and back into data structures. WSDL allows such rules to be constructed on the fly. It’s more or less a
programming-language-agnostic schema for describing functions: their names, the data types of their argu-
ments, and the data types of their return values. Although WSDL is associated with SOAP, it’s possible to
use SOAP without using WSDL (in fact, we did just that throughout this chapter’s section on SOAP).
A WSDL file is an XML document (of course!), which defines the following aspects of your web service
inside its definitions element:
❑ Any custom data types defined by your web service. These go into complexType elements of a
types list.
❑ The formats of the messages sent and received by your web service; that is, the signatures and
return values of the functions your web service defines. These are defined in a series of message
elements, and may make reference to any custom data types you defined earlier.
❑ The names of the functions your web service provides, along with the input and output mes-
sages expected by each. This is in the portType element, which contains an operation element
for each of the web service’s functions.
❑ A binding of your web service’s functions to a specific protocol — that is, HTTP. For simple
SOAP applications, this section is an exercise in redundancy: You end up just listing all of your
functions again. It exists because SOAP is protocol-independent; you need to explicitly state that
you’re exposing your methods over HTTP. This goes in the binding element.
❑ Finally, the URL to your web service. This is defined in the service element.
531

58.3 Page 573

▲back to top
Chapter 21
Here’s BittyWiki.wsdl, a WSDL file for the SOAP API exposed by BittyWiki:
<?xml version=”1.0”?>
<definitions name=”BittyWiki”
targetNamespace=”urn:BittyWiki”
xmlns:xsd=”http://www.w3.org/2001/XMLSchema”
xmlns:soap=”http://schemas.xmlsoap.org/wsdl/soap/”
xmlns=”http://schemas.xmlsoap.org/wsdl/”>
<portType name=”BittyWikiPortType”>
<operation name=”getPage”>
<input message=”sendPageName”/>
<output message=”getPageText”/>
</operation>
<operation name=”save”>
<input message=”sendPageNameAndText”/>
<output message=”getStatusMessage”/>
</operation>
<operation name=”delete”>
<input message=”sendPageName”/>
<output message=”getStatusMessage”/>
</operation>
</portType>
The WSDL parser now knows which functions are exposed by BittyWiki, but nothing about the signa-
tures or return types of those functions. Those come next:
<message name=”sendPageName”>
<part name=”pageName” type=”xsd:string”/>
</message>
<message name=”sendPageNameAndText”>
<part name=”pageName” type=”xsd:string”/>
<part name=”pageText” type=”xsd:string”/>
</message>
<message name=”getPageText”>
<part name=”pageText” type=”xsd:string”/>
</message>
<message name=”getStatusMessage”>
<part name=”message” type=”xsd:string”/>
</message>
532

58.4 Page 574

▲back to top
Web Applications and Web Services
A rather redundant section follows, as the four SOAP functions are bound to SOAP-over-HTTP:
<binding type=”BittyWikiPortType” name=”BittyWikiSOAPBinding”>
<soap:binding style=”rpc” transport=”http://schemas.xmlsoap.org/soap/http” />
<operation name=”getPage”>
<input><soap:body use=”literal” namespace=”urn:BittyWiki” /></input>
<output><soap:body use=”literal” namespace=”urn:BittyWiki” /></output>
</operation>
<operation name=”save”>
<input><soap:body use=”literal” namespace=”urn:BittyWiki” /></input>
<output><soap:body use=”literal” namespace=”urn:BittyWiki” /></output>
</operation>
<operation name=”delete”>
<input><soap:body use=”literal” namespace=”urn:BittyWiki” /></input>
<output><soap:body use=”literal” namespace=”urn:BittyWiki” /></output>
</operation>
</binding>
Finally, the code to let WSDL know where to find the BittyWiki web service:
<service name=”BittyWiki”>
<port name=”BittyWikiPort” binding=”BittyWikiSOAPBinding”>
<soap:address location=”http://localhost:8002/”/>
</port>
</service>
</definitions>
The BittyWiki API doesn’t define any custom data types, so there’s no types element in its WSDL file.
If you want to see a types element that has some complexTypes, look at the WSDL file for the Google
Web APIs.
WSDL is pretty complicated: That WSDL file is bigger than the Python script implementing the web ser-
vice it describes. WSDL files are usually generated from the corresponding web service source code, so
that humans don’t have to specify them. It’s not possible to do this from Python code because a big part
of WSDL is defining the data types, and Python functions don’t have predefined data types. Both the
SOAPpy and ZSI libraries can parse WSDL (in fact, they share a WSDL library: wstools), but there’s not
much in the way of Python-specific resources for generating WSDL.
Try It Out Manipulating BittyWiki through a WSDL Proxy
The following looks more or less like the previous example of BittyWiki manipulation through direct
SOAP calls:
533

58.5 Page 575

▲back to top
Chapter 21
>>> import SOAPpy
>>> proxy = SOAPpy.WSDL.Proxy(open(“BittyWiki.wsdl”))
>>> proxy.getPage(“SOAPViaWSDL”)
<Fault SOAP-ENV:Server: Method urn:BittyWiki:getPage failed.: __main__.NoSuchPage
SOAPViaWSDL>
Traceback (most recent call last):
...
SOAPpy.Types.faultType: <Fault SOAP-ENV:Server: Method urn:BittyWiki:getPage
failed.: __main__.NoSuchPage SOAPViaWSDL>
>>> proxy.save(“SOAPViaWSDL”, “This page created through SOAP via WSDL.”)
‘Page saved.’
>>> proxy.getPage(“SOAPViaWSDL”)
‘This page created through SOAP via WSDL.’
The main difference here is that going through WSDL will stop you from calling web service methods
that don’t exist:
>>> proxy.noSuchMethod()
Traceback (most recent call last):
...
AttributeError: noSuchMethod
>>>
>>> server = SOAPpy.SOAPProxy(“http://localhost:8002/”, “urn:BittyWiki”)
>>> server.noSuchMethod()
<Fault SOAP-ENV:Client: No method urn:BittyWiki:noSuchMethod found:
exceptions.AttributeError BittyWikiAPI instance has no attribute ‘noSuchMethod’>
Traceback (most recent call last):
...
SOAPpy.Types.faultType: <Fault SOAP-ENV:Client: No method urn:BittyWiki:noSuchMethod
found: exceptions.AttributeError BittyWikiAPI instance has no attribute
‘noSuchMethod’>
Both attempts to call noSuchMethod raised an exception, but going through WSDL meant the problem
was caught on the local machine instead of the server. This ability is a lot more interesting to a compiled
language: WSDL makes it possible to apply the same compile-time checks to web service calls as to local
function calls.
Choosing a Web Service Standard
This chapter described three standards for web services, each with a different philosophy, each with
advantages and drawbacks. REST aims to get the most out of the facilities provided by HTTP, but it
lacks a standard encoding for even simple data types. XML-RPC provides that encoding, but it’s verbose
and only deals with simple data types and compositions of simple data types. SOAP offers the structured
data types of XML-RPC with the flexibility of REST, but its added complexity makes hard cases more
difficult to understand than if they’d just been implemented with REST.
Industry trends favor REST and SOAP over XML-RPC. SOAP has the backing of large software compa-
nies such as IBM and Microsoft; REST has the backing of independent web service users and developers.
534

58.6 Page 576

▲back to top
Web Applications and Web Services
That’s because APIs based around REST and XML-RPC are generally easier to learn and use. Whenever
web services expose the same API through different protocols, the simplest one generally wins. For
instance, Amazon exposes a SOAP API in addition to the REST API covered in this chapter, but about 80
percent of its users choose REST over SOAP.
Which should you choose? Well, if you were a big fan of large software companies like IBM and
Microsoft, you probably wouldn’t be using Python in the first place. You would be using Java or .NET:
two strongly typed languages with good SOAP tool support. In most cases, the extra functionality of
SOAP isn’t needed, and Python’s support for SOAP isn’t consummate with the added complexity, so
why choose it unnecessarily?
You should start off by planning to expose a well-designed REST or XML-RPC API. If, during the design
or implementation stage, you start running into problems with your choice, look into using SOAP.
Unless you’re doing heavy-duty automatic business process software, or interfacing with a statically
typed language like Java or .NET, you’ll probably be able to see the REST or XML-RPC API through to
the end. Your users will thank you for the simpler interface.
My ideal web service would have a RESTful interface in which each resource could accept POST data in
the format defined by XML-RPC (or some simple subset of SOAP). The web service could then be
designed along REST principles, but some variant of xmlrpclib or SOAPpy could be used to marshal
and unmarshal the data without requiring the creation of custom parsers.
Whatever you choose, please try to keep web services in mind from the moment you begin the design: A
web service is just a web application for robots. If you want your application to inspire creativity and not just
meet a predefined need, you must give up some of the control to your users.
Web Service Etiquette
A web service may have users who skirt the rules, or administrators who feel the users are ungrateful
for the service they’re being provided. In the interests of harmony, here are a few basic pieces of advice
for managing the social aspect of web services.
For Consumers of Web Services
If you write a robot to consume someone else’s web services, it’s important to play by the rules. In par-
ticular, don’t try to evade any limitations such as license keys or daily limits on your access to the API.
Access to a web service is a privilege, not a right. It’s better to run out of API calls and have to complete
a task later than you planned than to have your access taken away altogether.
For Producers of Web Services
If you’re planning to expose your web application through a web service, you need to consider the flip
side of these issues. If your audience is already scripting your application, you’ve got a leg up because
you don’t have to guess what people might do with it. Before you design your web services, poll your
robot-writing users to see what parts of your application they’re using. Make your web services avail-
able on terms that allow users to move over to the new system, or they’ll have no incentive to switch.
535

58.7 Page 577

▲back to top
Chapter 21
As producer of a public web service, you might feel like the burden of etiquette falls completely on your
users. After all, you’re providing a service to them and not expecting anything in return. Nonetheless,
it’s important to make your terms of use palatable because the people writing the robots have the final
advantage: So long as you provide a web application with the same functionality as the web service,
determined users can always write a robot to use the web application however they want. There’s no
foolproof way you can distinguish between a robot that uses your site and the web browser a human
might use to use your site. They’re both pieces of software running on someone’s computer, making an
HTTP request. All the HTTP headers, including the User-Agent and the authentication headers, can be
forged by a robot.
That said, if a particular robot is causing you trouble, you can solve the problem with the same tools
you’d use against a troublesome human user. In particular, the Apache module mod_throttle is useful
for dealing with users who are taking more than their fair share of resources.
Download mod_throttle at www.snert.com/Software/mod_throttle/. Note that as of this
writing, it’s only available for Apache 1.3.
Using Web Applications as Web Services
It’s possible to write scripts that consume web applications as though they were web services. After all,
that’s how the idea of web services got started in the first place. Some sites still haven’t gotten the web
services religion, or might have web services that don’t expose the functionality you need. To write the
robot you have in mind, you’d have to go through the application.
This chapter doesn’t cover how to write such scripts, but the general principles are similar to web ser-
vices; and if this topic interests you, you’ll eventually find yourself doing it. When you do, don’t do any-
thing that violates the site’s terms of service. In addition, don’t access the site more than a human user
would. If you can, run your script in off hours so you don’t add to the load on the system. Finally, ask
the site administrators for a web service interface so you can work against a more stable interface that
uses less bandwidth.
A Sampling of Publicly Available
Web Services
Many popular web applications are also available as web services. The two most popular are Amazon
Web Services and the Google Web API (both covered in this chapter). However, most web services are
made available by smaller companies as differentiators, or as an extension of noncorporate or volunteer
services. Some of the most popular Web APIs fall into this category: the Technorati web API, which acts
as a near-real-time search for weblogs, and the API to the social bookmarking site del.icio.us.
Most web services are made available under official or informal restrictions. Often, users are asked to
sign up for a license key, and each user is allowed only a certain number of API calls per unit time (as
mentioned earlier, the Google and Amazon APIs work this way). Some web services, such as eBay’s, are
available only by paid subscription.
536

58.8 Page 578

▲back to top
Web Applications and Web Services
Knowledge of public web services is a useful thing to have in your programmer’s toolkit. Some are mere
toys, but others can be very useful if you’ve got an idea but not the data set you would need to realize it.
You can also follow their examples when exposing your web applications as web services. The following
table describes some of the more interesting specimens, as well as some sites that aggregate all manner
of public web services.
Site
Amazon.com
del.icio.us
Flickr
Google
Meerkat
Mail To
The Future
Speller
Technorati
Xmethods
WebserviceX
Yahoo!
What You Can Do
with the API
Search books and
other items for sale.
Store and search
your bookmarks.
Store and search
your photos (and
other peoples’!).
Search the web.
Search recent
tech news.
Send email
reminders to
yourself.
Spell-check text.
Search weblogs
in near-real-time.
A repository
aggregating a
wide variety of
web services.
Another repository
of web services
Search web
pages, images, etc.
Protocol(s)
Supported
To Learn More
REST, SOAP www.amazon.com/gp/aws/landing.html
REST
REST,
XML-RPC,
SOAP
SOAP
REST,
XML-RPC
XML-RPC
http://del.icio.us/api/
www.flickr.com/
www.google.com/apis/
www.oreillynet.com/pub/a/rss/
2000/05/09/meerkat_api.html
www.oreillynet.com/pub/a/rss/
2000/11/14/meerkat_xmlrpc.html
www.xmlrpc.com/stories/
storyReader$2598
XML-RPC
REST
SOAP
www.stuffeddog.com/speller/
doc/rpc.html
http://developers.technorati.com/wiki/
TechnoratiApi
www.xmethods.net/
SOAP
REST
www.webservicex.net/
http://developer.yahoo.net/
537

58.9 Page 579

▲back to top
Chapter 21
Summary
Web applications are powerful and popular; with Python, they’re also easy to write. The REST architec-
ture made the web usable and successful: Employing it when designing your application gives you a
head start. Web applications are designed for humans; a web service is just a web application designed
for use by software scripts instead. Expose REST and XML-RPC web services for simplicity and easy
adoption, SOAP for heavy-duty tasks or when interfacing with Java or .NET applications. Make use of
the web services provided by others: They’re opening up their data sets and algorithms for your benefit.
Exercises
1. What’s a RESTful way to change BittyWiki so that it supports hosting more than one Wiki?
2. Write a web application interface to WishListBargainFinder.py. (That is, a web application
that delegates to the Amazon Web Services.)
3. The BittyWiki API makes the raw wiki markup of a page available when you GET a page (REST
interface) or call the getPage method (XML-RPC and SOAP interfaces). Suppose your web ser-
vice users start wanting to get BittyWiki pages rendered as HTML fragments so that they don’t
have to write their own transformations from BittyWiki markup to HTML. What’s a RESTful
way to solve the problem? What’s a solution that’s more in keeping with the XML-RPC or
SOAP philosophies?
4. The wiki search-and-replace spider looks up every new WikiWord it encounters to see whether
it corresponds to a page of the wiki. If it finds a page by that name, that page is processed.
Otherwise, nothing happens and the spider has wasted a web service request. How could the
web service API be changed so that the spider could avoid those extra web service requests for
nonexistent pages?
5. Suppose that, to prevent vandalism, you change BittyWiki so that pages can’t be deleted.
Unfortunately, this breaks the wiki search-and-replace spider, which sometimes deletes a page
before recreating it with a new name. What’s a solution that meets both your needs and the
needs of the spider’s users?
538

58.10 Page 580

▲back to top
22
Integrating Java with Python
Java is an object-oriented programming language. Java programs are compiled from source code
into byte codes. The Java runtime engine, called a Java virtual machine, or JVM, runs the com-
piled byte codes. Sound familiar? At an abstract level at least, Java and Python are very similar.
Like Java, Python programs are compiled into byte codes, although this can be done at runtime.
Despite these similarities, there are some differences between the languages:
❑ With Python, you can run scripts directly from the source code. Compiling is optional. If
you don’t compile your Python code in advance, the python command will take care of
this for you.
❑ Java syntax is based on C and C++, two very popular programming languages. This
makes it easy for developers using C++ to migrate to Java. Consequently, Java is consid-
ered a more serious and businesslike language than Python.
❑ Python syntax is very simple and easy to learn, but the syntax has diverged far from C.
❑ With its simple syntax and built-in support for lists, dictionaries, and tuples, you’ll find
Python code much easier to write than Java code. Generally, Python programs require a
lot less code than the corresponding Java code.
❑ Java has an advantage over Python in terms of standard APIs, though. The base Java lan-
guage includes a mature database API, an API for parsing XML documents, an API for
remote communication, and even an API to access LDAP directory servers. You can do all
of this in Python, but Python lacks the richness, and standardization, of the many Java
APIs. This becomes more apparent when you write enterprise applications in Python.
Java’s enterprise APIs, called J2EE, enable Java to be a player in the enterprise market.
Python, unfortunately, has been relegated to a minimal role in the enterprise market.
When writing enterprise applications, you’ll likely need to write them in Java. Even though
Python can work well in this space, as shown in Chapters 20 and 21, Java controls the mind share
for the enterprise. Luckily, you can get the best of both worlds with Jython, an implementation of
Python in Java.

59 Pages 581-590

▲back to top

59.1 Page 581

▲back to top
Chapter 22
Jython enables you to execute Python code from within a Java virtual machine — that is, from within any
Java application. This chapter covers that topic, including the following:
❑ Reasons for scripting within Java applications
❑ Comparing Jython with the regular C-based Python implementations
❑ Installing Jython
❑ Running Python scripts from Jython
❑ Calling Java code from Python scripts
❑ Extending Java classes with Python classes
❑ Writing J2EE Servlets in Python
❑ Embedding the Jython interpreter in your Java applications
Note that you’ll want to have some familiarity with both Java and Python to be able to integrate Python
and Java.
Scripting within Java Applications
Most software developers consider Java to be a large systems programming language, a serious language
for serious tasks. Python, in this way of thinking, comes from the realm of scripting languages such as
Perl and Tcl. As such, many developers typically don’t respect Python because scripting languages are, of
course, created for people who cannot program. You know this isn’t true, but the split between program-
ming and scripting languages remains, even though Python gracefully bridges this gap.
Despite this lack of respect, scripting languages have proven to be very productive and are widely
deployed as critical parts of companies small and large (and huge and gigantic, too). You can generally
accomplish a lot more in less time with less code using a scripting language than you can with a system
programming language like Java.
With Java applications, scripting comes in handy for a number of reasons, including the following:
❑ The scripting language can act as a macro extension language. Much like Visual Basic for
Applications (VBA) enables you to script extensions to Microsoft Office, you can enable
users to extend your own Java applications using Jython. Complex text editors such as jEdit
(www.jedit.org) enable you to write scripts in this fashion.
❑ Use Jython to speed the development of Java applications. As a high-level scripting language,
you can take advantage of the productivity features of Python when compared to the complex-
ity of Java.
❑ Explore and debug running systems. Using the interactive capabilities of Jython, you can
explore a running Java application. You can execute code as needed, all interactively. You
already take this for granted in Python, but it’s something that Java just doesn’t have.
❑ You can script unit tests much faster than writing them in Java. Many organizations feel uncom-
fortable about introducing scripting languages, especially open-source ones. Using scripts for
540

59.2 Page 582

▲back to top
Integrating Java with Python
testing provides the advantages of scripting without shipping the scripting packages in your
application or using the scripting packages in production.
❑ In addition to unit testing, scripting works well for full system testing. A system-testing package
called the Grinder uses Jython to create test scripts. See http://grinder.sourceforge.net/
for more on the Grinder.
❑ You can create one-off scripts for tasks such as data migration. If you just need to update a par-
ticular row in a database table, or fix a particular setting, you can do this a lot quicker using a
script.
❑ You can extend enterprise applications without having to redeploy the application. This is very
handy if you need to keep your system running all the time. In addition, developers can extend
the functionality of the system without requiring the security permissions to redeploy the
application.
Jython, being based on the very popular Python language, enables you to do all of this, and more.
Comparing Python Implementations
The traditional Python implementation, often called C-Python, compiles and runs on a huge number
of platforms, including Windows, Linux, and Mac OS X. C-Python is written in the C programming
language. The Java-based Jython compiles and runs on any platform that supports the Java virtual
machine (technically, any platform that supports J2SE, or Java 2 Standard Edition). This includes
Windows, Linux, and Mac OS X. In this respect, the two Python implementations are very similar in
how cross-platform they are.
However, Jython isn’t up to date compared to the traditional C-Python implementation. The C-Python
implementation sports new features that have not yet been written in the Java implementation. That’s
understandable, as the C-Python is where the first development happens, and the Jython developers
have to re-implement every Python feature in Java.
Which foundation you use for Python scripting, C-Python or Jython, doesn’t really matter, because both
support the Python language. In general, you’ll want to use C-Python unless you have a specific need to
work within the Java platform. In that case, obviously, use Jython!
The rest of this chapter shows you how to do just that.
Installing Jython
As an open-source project, Jython doesn’t follow a set release cycle. Your best bet is to download the lat-
est release from www.jython.org. Then, follow the instructions on the web site for installing Jython.
Older versions of Jython, such as 2.1, are packaged as a Java .class file of the installation program.
When you run the file, the program will install Jython on your hard disk. Newer pre-release versions
come packaged as a Zip file. Unzip the file to install Jython.
541

59.3 Page 583

▲back to top
Chapter 22
After installing Jython, you should have two executable scripts in the Jython installation directory:
jython and jythonc, similar in purpose to python and pythonc. The jythonc script, though, is
intended to compile Python code into Java .class files. You need to have the jython script in your
path, or available so you can call it.
On Windows, you will get DOS batch files jython.bat and jythonc.bat.
Running Jython
The jython script runs the Jython interpreter. The jythonc script runs the Jython compiler, which com-
piles Jython code to Java .class files. In most cases, you’ll want to use the jython script to run Jython.
Running Jython Interactively
Like Python with the python command, Jython supports an interactive mode. In this mode, you can
enter Jython expressions, as you’d expect. Jython expressions are for the most part the same as Python
expressions, except you can call upon the Java integration as well.
To run the Jython interpreter, run the jython script (jython.bat on Windows).
Try It Out Running the Jython Interpreter
Run the interpreter and then enter in the following expressions:
$ ./jython
Jython 2.1 on java1.4.2_05 (JIT: null)
Type “copyright”, “credits” or “license” for more information.
>>> 44 / 11
4
>>> 324 / 101
3
>>> 324.0 / 101.0
3.207920792079208
>>> 324.0 / 101
3.207920792079208
>>> import sys
>>> sys.executable
>>> sys.platform
‘java1.4.2_05’
>>> sys.version_info
(2, 1, 0, ‘final’, 0)
>>>
How It Works
As shown in this example, the Jython interpreter appears and acts like the Python interpreter. This is just
what you’d expect, as Jython is supposed to be an implementation of the Python language on top of the
Java platform.
542

59.4 Page 584

▲back to top
Integrating Java with Python
Math operations should work mostly the same as with Python. (“Mostly the same” means that some
floating-point operations will create slightly different results.)
With the sys module, note how the property sys.executable is not set when you run Jython interac-
tively. Also note that this example is using Jython 2.1.
On the same platform, you can see the differences when you run the same expressions with the python
command, the C-Python interpreter. For example:
$ python
Python 2.3 (#1, Sep 13 2003, 00:49:11)
[GCC 3.3 20030304 (Apple Computer, Inc. build 1495)] on darwin
Type “help”, “copyright”, “credits” or “license” for more information.
>>> 44 / 11
4
>>> 324 / 101
3
>>> 324.0 /101.0
3.2079207920792081
>>> 324.0 / 101
3.2079207920792081
>>> import sys
>>> sys.executable
‘/usr/bin/python’
>>> sys.platform
‘darwin’
>>> sys.version_info
(2, 3, 0, ‘final’, 0)
>>>
Note how the floating-point numbers show an extra digit by default. This example was run under
Python 2.3 on Mac OS X 10.3, but the results should appear the same on Python 2.4 as well.
Running Jython Scripts
As with the python command, jython can also run your scripts, as shown in the following example.
Try It Out Running a Python Script
Enter the following simple script and name the file jysys.py:
import sys
print ‘Python sys.path:’
print sys.path
print ‘Script arguments are:’
print sys.argv
543

59.5 Page 585

▲back to top
Chapter 22
When you run this script with jython, you should see output like the following:
$ jython jysys.py 1 2 3 4
Python sys.path:
[‘’, ‘/Users/ericfj/writing/python/.’,
‘/Users/ericfj/Documents/java/jython-2.1/Lib’]
Script arguments are:
[‘jysys.py’, ‘1’, ‘2’, ‘3’, ‘4’]
The filepaths will differ depending on where you installed Jython.
How It Works
In this example, you can pass any command-line arguments to the script. The 1 2 3 4 shown here just
helps to see the arguments held in the sys.argv property.
The sys.path property holds a very small number of directories, especially when compared to the stan-
dard C-Python distribution. For example, you can run the same script with the python interpreter as
shown here:
$ python jysys.py 1 2 3 4
Python sys.path:
[‘/Users/ericfj/writing/python’,
‘/System/Library/Frameworks/Python.framework/Versions/2.3/lib/python23.zip’,
‘/System/Library/Frameworks/Python.framework/Versions/2.3/lib/python2.3’,
‘/System/Library/Frameworks/Python.framework/Versions/2.3/lib/python2.3/
plat-darwin’,
‘/System/Library/Frameworks/Python.framework/Versions/2.3/lib/python2.3/plat-mac’,
‘/System/Library/Frameworks/Python.framework/Versions/2.3/lib/python2.3/
plat-mac/lib-scriptpackages’,
‘/System/Library/Frameworks/Python.framework/Versions/2.3/lib/python2.3/lib-tk’,
‘/System/Library/Frameworks/Python.framework/Versions/2.3/lib/python2.3/
lib-dynload’,
‘/System/Library/Frameworks/Python.framework/Versions/2.3/lib/python2.3/
site-packages’]
Script arguments are:
[‘jysys.py’, ‘1’, ‘2’, ‘3’, ‘4’]
In this case, note the larger number of directories in the sys.path property.
These examples were run on Mac OS X version 10.3.8. The paths will differ on other operating systems.
You’ll notice that the startup time for jython-run scripts is a lot longer than that for python-run scripts.
That’s because of the longer time required to start the java command and load the entire Java environment.
Controlling the jython Script
The jython script itself merely acts as a simple wrapper over the java command. The jython script
sets up the Java classpath and the python.home property. You can also pass arguments to the jython
544

59.6 Page 586

▲back to top
Integrating Java with Python
script to control how Jython runs, as well as arguments to your own scripts. The basic format of the
jython command line follows:
jython jython_arguments what_to_run arguments_for_your_script
The jython_arguments can be -S to not imply an import site when Jython starts and -i to run Jython
in interactive mode. You can also pass Java system properties, which will be passed in turn to your
Jython scripts. The format for this is -Dproperty=value, which is a standard Java format for passing
property settings on the command line.
You’ll normally pass the name of a Jython script file as the what_to_run section of the command — for
example, the jysys.py script used in the previous example. The jython script offers more options,
though, as shown in the following table.
Option
filename.py
-c command
-jar jarfile
-
Specifies
Runs the given Jython script file
Runs the command string passed on the command line
Runs the Jython script __run__.py in the given jar file
Reads the commands to run from stdin. This allows you to pipe Jython com-
mands to the Jython interpreter
You can choose any one of the methods listed in the table.
In addition, the arguments_for_your_script are whatever arguments you want to pass to your script.
These will be set into sys.argv[1:] as you’d expect.
Making Executable Commands
Note that because jython is a script, you cannot use the traditional shebang comment line to run Jython
scripts. (On Unix and Linux systems, that’s the line that starts with the hash, or sharp, symbol and then
has the exclamation point, or “bang,” so you get “sh(arp)-bang.” This tells the system that this com-
mand is how the program you’re running should be invoked.) For example, with a Python script, you
can add the following line as the first line of your script:
#! /usr/bin/python
If your script has this line as the first line, and if the script is marked with execute permissions, the oper-
ating system can run your Python scripts as commands.
Note that Windows is the lone exception. Windows uses a different means to associate files ending in
.py with the Python interpreter.
With Jython scripts, though, you cannot use this mechanism. That’s because many operating systems
require that the program that runs a script be a binary executable program, not a script itself. That is,
you have a script you wrote that you want run by the jython script.
545

59.7 Page 587

▲back to top
Chapter 22
To get around this problem, use the env command. For example, change the shebang line to the following:
#! /usr/bin/env jython
For this line to work, the jython script must be in your path.
Try It Out Making an Executable Script
Insert the following lines into the previous jysys.py script. The new line is marked in bold.
#! /usr/bin/env jython
import sys
print ‘Python sys.path:’
print sys.path
print ‘Script arguments are:’
print sys.argv
Save this new file under the name jysys, with no extension. Use the chmod command to add execute
permissions, as shown in the following example:
$ chmod a+x jysys
You can then run this new command:
$ ./jysys 1 2 3 4
Python sys.path:
[‘.’, ‘/Users/ericfj/writing/python/.’, ‘/Users/ericfj/Documents/java/jython-
2.1/Lib’]
Script arguments are:
[‘./jysys’, ‘1’, ‘2’, ‘3’, ‘4’]
How It Works
The shebang comment works the same for Jython as it does for all other scripting languages. The only
quirk with Jython is that the jython command itself is a script that calls the java command.
In the next section, you’ll learn more about how the java command runs Jython scripts.
Running Jython on Your Own
You don’t have to use the jython script, though, to execute Jython scripts. You can call the Jython inter-
preter just like any other Java application.
546

59.8 Page 588

▲back to top
Integrating Java with Python
The jython script itself is fairly short. Most of the action occurs by calling the java command with a
large set of arguments, split up here for clarity:
java -Dpython.home=”/Users/ericfj/Documents/java/jython-2.1” \\
-classpath “/Users/ericfj/Documents/java/jython-2.1/jython.jar:$CLASSPATH” \\
“org.python.util.jython” “$@”
The paths will differ depending on where you installed Jython. The jython script, though, does nothing
more than run the class org.python.util.jython from the jar file jython.jar (which the script adds
to the Java classpath). The script also sets the python.home system property, necessary for Jython to find
support files.
To run Jython on your own, you just need to ensure that jython.jar is in the classpath. Execute an
interpreter class, such as org.python.util.jython. In addition, you need to set the python.home sys-
tem property.
You also need to ensure that Jython is properly installed on every system that will run your Jython scripts.
Packaging Jython-Based Applications
Jython isn’t a standalone system. It requires a large number of Python scripts that form the Jython
library. Thus, you need to include the jython.jar file as well as the Jython library files. At a bare mini-
mum, you need the Lib and cachedir directories that come with the Jython distribution.
Jython needs to be able to write to the cachedir directory.
Java applications, especially J2EE enterprise applications, usually don’t require a set of files stored in a
known location on the file system. If you include Jython, though, you’ll need to package the files, too.
Up to now, you can see that Jython really is Python, albeit an older version of Python. The real advan-
tage of Jython, though, lies in the capability to integrate Python with Java, offering you the best of both
worlds.
Integrating Java and Jython
The advantage of Jython comes into play when you integrate the Jython interpreter into your Java appli-
cations. With this combination, you can get the best of both the scripting world and the rich set of Java
APIs. Jython enables you to instantiate objects from Java classes and treat these objects as Python objects.
You can even extend Java classes within Jython scripts.
Jython actively tries to map Java data types to Python types and vice versa. This mapping isn’t always
complete because the feature is under active development. For the most part, however, you’ll find that
Jython does the right thing when converting to and from Python types.
547

59.9 Page 589

▲back to top
Chapter 22
Using Java Classes in Jython
In general, treat Java classes as Python classes in your scripts. Jython uses the Python syntax for import-
ing Java classes. Just think of Java packages as a combination of Python modules and classes. For exam-
ple, to import java.util.Map into a Jython script, use the following code:
from java.util import Map
Note how this looks just like a Python import. You can try this out in your own scripts, as shown in the
following example.
Try It Out Calling on Java Classes
Enter the following script and name the file jystring.py:
import sys
from java.lang import StringBuffer, System
sb = StringBuffer(100) # Preallocate StringBuffer size for performance.
sb.append(‘The platform is: ‘)
sb.append(sys.platform) # Python property
sb.append(‘ time for an omelette.’)
sb.append(‘\\n’)
# Newline
sb.append(‘Home directory: ‘)
sb.append( System.getProperty(‘user.home’) )
sb.append(‘\\n’)
# Newline
sb.append(‘Some numbers: ‘)
sb.append(44.1)
sb.append(‘, ‘)
sb.append(42)
sb.append(‘ ‘)
# Try appending a tuple.
tup=( ‘Red’, ‘Green’, ‘Blue’, 255, 204, 127 )
sb.append(tup)
print sb.toString()
# Treat java.util.Properties as Python dictionary.
props = System.getProperties()
print ‘User home directory:’, props[‘user.home’]
548

59.10 Page 590

▲back to top
Integrating Java with Python
When you run this script, you should see the following output:
$ jython jystring.py
The platform is: java1.4.2_05 time for an omelette.
Home directory: /Users/ericfj
Some numbers: 44.1, 42 (‘Red’, ‘Green’, ‘Blue’, 255, 204, 127)
User home directory: /Users/ericfj
Note that your output will depend on where your home directory is located and which version of Java
you have installed.
How It Works
This script imports the Java StringBuffer class and then calls a specific constructor for the class:
from java.lang import StringBuffer
sb = StringBuffer(100)
The Jython interpreter converts the value 100 from a Python number to a Java number.
In Java programs, you do not need to import classes from the java.lang package. In Jython, import
every Java class you use.
You can pass literal text strings as well as Python properties to the StringBuffer append method:
sb.append(‘The platform is: ‘)
sb.append(sys.platform) # Python property
This example shows that Jython will correctly convert Python properties into Java strings for use in a
Java object. You can also pass the data returned by a Java method:
sb.append( System.getProperty(‘user.home’) )
In this case, the System.getProperty method returns an object of the Java type Object. Again, Jython
properly handles this case, as Jython does with numbers:
sb.append(44.1)
sb.append(42)
You can even append a Python tuple:
tup=( ‘Red’, ‘Green’, ‘Blue’, 255, 204, 127 )
sb.append(tup)
The preceding example shows that Jython does the right thing when converting the tuple to a Java text
string.
549

60 Pages 591-600

▲back to top

60.1 Page 591

▲back to top
Chapter 22
In addition to converting Python types to Java types, Jython works the other way as well. You can pass a
Java String object, returned by the toString method, to the Python print function:
print sb.toString()
This shows how you can treat Java strings as Python strings. You can also treat Java hash maps and hash
tables as Python dictionaries, as shown in the following example:
props = System.getProperties()
print ‘User home directory:’, props[‘user.home’]
The Java System.getProperties method returns an object of type java.util.Properties, which
Jython automatically converts into a Python dictionary.
Data type conversions as shown by this example are just what you’d expect when you integrate Java and
Python. Jython does a lot of work under the covers, though. Java has a class hierarchy, as does Python. A
large part of Jython is an attempt to merge these two large hierarchies together. Ultimately, you tend to
get the best of both worlds.
For example, Python has the ability to pass named properties to a constructor. This proves especially
useful when you work with APIs such as Swing for user interfaces. The Swing API has many, many
classes. Each class supports a large number of properties on objects. Working with Java alone, you can
only call the constructors that have been defined, and the parameters must be placed in a particular
order. With Python, though, you can pass named properties to the object’s constructor and set as many
properties as needed within one call.
The following example shows this technique.
Try It Out Creating a User Interface from Jython
Enter the following script and name the file jyswing.py:
from java.lang import System
from javax.swing import JFrame, JButton, JLabel
from java.awt import BorderLayout
# Exit application
def exitApp(event):
System.exit(0)
# Use a tuple for size
frame = JFrame(size=(500,100))
# Use a tuple for RGB color values.
frame.background = 127,255,127
button = JButton(label=’Push to Exit’, actionPerformed=exitApp)
label = JLabel(text=’A Pythonic Swing Application’,
550

60.2 Page 592

▲back to top
Integrating Java with Python
horizontalAlignment=JLabel.CENTER)
frame.contentPane.add(label, BorderLayout.CENTER)
frame.contentPane.add(button, BorderLayout.WEST)
frame.setVisible(1)
When you run this script, you should see a window like the one shown in Figure 22-1.
Figure 22-1
Click the button to exit the application.
How It Works
This script shows how you can use Jython with the complex Swing APIs. While this example is almost
all calls to Java APIs, it is much shorter than the corresponding Java program would be. That’s because
of the handy built-in features that come with Python, such as support for tuples and setting properties.
The script starts by importing several classes in the AWT and Sing APIs. The JFrame class acts as a top-
level window in an application. You can create a JFrame widget with the following statements:
frame = JFrame(size=(500,100))
The size property on a JFrame widget is an instance of another Java class, java.awt.Dimension. In
this example, you can make a Dimension object from a tuple and then pass this object to set the size
property of the JFrame.
This shows how Jython can make working with the Swing APIs palatable. Creating a user interface
with Swing usually involves a lot of tedious coding. Jython greatly reduces this effort.
You can use the Python support for tuples and the Jython-provided integration with Java APIs to set col-
ors as well:
frame.background = 127,255,127
This sets the background color to a light green.
This example uses an eight-bit color definition, with values of zero to 255 for each of the red, green, and
blue components of the color. Therefore, 255 means that the green value is set to all on, and the red and
blue values are set to half on.
551

60.3 Page 593

▲back to top
Chapter 22
Jython makes it easy to create interactive widgets on the screen. For example, the following code creates
a JButton widget and sets the widget to call the function exitApp when the user clicks the button:
def exitApp(event):
System.exit(0)
button = JButton(label=’Push to Exit’, actionPerformed=exitApp)
In this case, the exitApp function calls the Java method System.exit to exit the Java engine and
therefore quit the application. Jython enables you to set Java properties to Python functions, such as
exitApp in this example. In Java, you would need to make a class that implements the methods in the
java.awt.event.ActionListener interface and then pass in an instance of this class as the action
listener for the JButton. The Jython approach makes this much easier.
The example also creates a JLabel widget, which displays a text message, an image, or both. The
jyswing.py script sets the horizontal alignment so that the text displays in the center of the widget’s
bounds:
label = JLabel(text=’A Pythonic Swing Application’,
horizontalAlignment=JLabel.CENTER)
In this example, the value JLabel.CENTER is a constant on the JLabel class.
In Java terms, JLabel.CENTER is a public static final value on the class.
Once created, you need to place the widgets within a container. In the example script, you need to place
the JButton and JLabel widgets in the enclosing JFrame widget, as shown by the following code:
frame.contentPane.add(label, BorderLayout.CENTER)
frame.contentPane.add(button, BorderLayout.WEST)
In Swing applications, you add widgets to the content pane of the JFrame, not directly to the JFrame
itself.
Finally, the script makes the JFrame widget visible:
frame.setVisible(1)
Note that the Java setVisible method expects a Java Boolean value, but using the Python True would
be flagged as a syntax error because the Java boolean objects aren’t 0 and 1, as they are in Python;
they’re a class that gets used sometimes, while 0 and 1 get used at other times in Java. This is one area
where Python data types and constants are not yet mapped to their Java equivalents.
Accessing Databases from Jython
JDBC, or Java Database Connectivity, provides a set of APIs to access databases in a consistent manner.
Most, but not all, differences between databases can be ignored when working with JDBC.
Python has a set of database APIs as well, as described in Chapter 14. A large difference between the
Python APIs and the Java APIs is that the Java JDBC drivers are almost all written entirely in Java.
552

60.4 Page 594

▲back to top
Integrating Java with Python
Furthermore, almost all JDBC drivers are written by the database vendors. Most Python DB drivers,
such as the ones for Oracle, are written in C with a Python layer on top. Most are written by third par-
ties, and not by the database vendors. The Java JDBC drivers, then, can be used on any platform that
supports Java. The Python DB drivers, though, must be recompiled on each platform and may not work
on all systems that support Python.
With Jython, the zxJDBC package provides a Python DB-compliant driver that works with any JDBC
driver. That is, zxJDBC bridges between the Python and Java database APIs, enabling your Jython scripts
to take advantage of the many available JDBC drivers and to use the simpler Python DB APIs.
When working with JDBC drivers, you need the value of four properties to describe the connection to
the database, shown in the following table.
Property
JDBC URL
User name
Password
Driver
Holds
Description of the connection to the database in a format defined by the
driver
Name of a user who has access rights to the database
Password of the user. This is the password to the database, not to an
operating system.
Name of the Java class that provides the JDBC driver
You need to gather these four values for any database connection you need to set up using JDBC. The
zxJDBC module requires these same values. To connect to a database using the zxJDBC driver, you can
use code like the following:
from com.ziclix.python.sql import zxJDBC
url=’jdbc:hsqldb:hsql://localhost/xdb’
user=’sa’
pw=’’
driver=’org.hsqldb.jdbcDriver’
db = zxJDBC.connect(url, user, pw, driver)
The values shown here for the JDBC connection come from the default values for the HSqlDB database,
covered in the section “Setting Up a Database.”
Working with the Python DB API
Once you have a connection, you can use the same techniques shown in Chapter 14. The zxJDBC mod-
ule provides a DB 2.0 API–compliant driver. (Well, mostly compliant.) For example, you can create a
database table using the following code:
cursor = db.cursor()
cursor.execute(“””
create table user
553

60.5 Page 595

▲back to top
Chapter 22
(userid integer,
username varchar,
firstname varchar,
lastname varchar,
phone varchar)
“””)
cursor.execute(“””create index userid on user (userid)”””)
After creating a table, you can insert rows using code like the following:
cursor.execute(“””
insert into user (userid,username,firstname,lastname,phone)
values (4,’scientist’,’Hopeton’,’Brown’,’555-5552’)
“””)
Be sure to commit any modifications to the database:
db.commit()
You can query data using code like the following:
cursor.execute(“select * from user”)
for row in cursor.fetchall():
print row
cursor.close()
See Chapter 14 for more on the Python DB APIs.
Setting Up a Database
If you already have a database that includes a JDBC driver, you can use that database. For example,
Oracle, SQL Server, Informix, and DB2 all provide JDBC drivers for the respective databases.
Unfortunately, the Gadfly database used in Chapter 14 does not include a JDBC driver. You can use the
Python DB API driver, though, if you wish.
If you have a database set up, try to use it. If you have no database, a handy choice is HSqlDB. HSqlDB
provides a small, fast database. A primary advantage of HSqlDB is that because it is written in Java, you
can run it on any platform that runs Java.
See http://hsqldb.sourceforge.net/ for more on the HSqlDB database. You can download this
open-source free package from this site.
You’ll find installing HSqlDB quite simple. Just unzip the file you download and then change to the new
hsqldb directory. To run the database in server mode, with the default parameters, use a command like
the following:
$ java -cp ./lib/hsqldb.jar org.hsqldb.Server -database.0 mydb -dbname.0 xdb
[Server@922804]: [Thread[main,5,main]]: checkRunning(false) entered
[Server@922804]: [Thread[main,5,main]]: checkRunning(false) exited
554

60.6 Page 596

▲back to top
Integrating Java with Python
[Server@922804]: Startup sequence initiated from main() method
[Server@922804]: Loaded properties from
[/Users/ericfj/writing/python/chap22/server.properties]
[Server@922804]: Initiating startup sequence...
[Server@922804]: Server socket opened successfully in 160 ms.
[Server@922804]: Database [index=0, id=0, db=file:mydb, alias=xdb] opened
sucessfully in 1168 ms.
[Server@922804]: Startup sequence completed in 1444 ms.
[Server@922804]: 2005-03-12 20:09:33.417 HSQLDB server 1.7.3 is online
[Server@922804]: To close normally, connect and execute SHUTDOWN SQL
[Server@922804]: From command line, use [Ctrl]+[C] to abort abruptly
You can stop this database by typing Ctrl+C in the shell window where you started HSqlDB. You now
have a database that you can connect to using the default properties shown in the following table.
Property
JDBC URL
User name
Password
Driver
Value
driver.jdbc:hsqldb:hsql://localhost/xdb
sa
‘’ (two single quotes, an empty string)
org.hsqldb.jdbcDriver
Working with JDBC drivers requires that you add the JDBC jar or jars to the Java classpath. The
jython script doesn’t handle this case, so you need to modify the script. For example, to use the HSqlDB
database, modify the script to add the hsqldb.jar jar file. For example:
#!/bin/sh
###############################################################################
#
# This file generated by Jython installer
java -Dpython.home=”/Users/ericfj/Documents/java/jython-2.1” \\
-classpath \\
“/Users/ericfj/Documents/java/jython-2.1/jython.jar:$CLASSPATH:./hsqldb.jar” \\
“org.python.util.jython” “$@”
The bold text shows the additional jar file. This example assumes that the file hsqldb.jar will be
located in the current directory. That may not be true. You may need to enter the full path to this jar file.
To pull all this together, try the following example, built using the HSqlDB database.
Try It Out Create Tables
Enter the following script and name the file jyjdbc.py:
from com.ziclix.python.sql import zxJDBC
# Modify as needed for your database.
555

60.7 Page 597

▲back to top
Chapter 22
url=’jdbc:hsqldb:hsql://localhost/xdb’
user=’sa’
pw=’’
driver=’org.hsqldb.jdbcDriver’
db = zxJDBC.connect(url, user, pw, driver)
cursor = db.cursor()
cursor.execute(“””
create table user
(userid integer,
username varchar,
firstname varchar,
lastname varchar,
phone varchar)
“””)
cursor.execute(“””create index userid on user (userid)”””)
cursor.execute(“””
insert into user (userid,username,firstname,lastname,phone)
values (1,’ericfj’,’Eric’,’Foster-Johnson’,’555-5555’)
“””)
cursor.execute(“””
insert into user (userid,username,firstname,lastname,phone)
values (2,’tosh’,’Peter’,’Tosh’,’555-5554’)
“””)
cursor.execute(“””
insert into user (userid,username,firstname,lastname,phone)
values (3,’bob’,’Bob’,’Marley’,’555-5553’)
“””)
cursor.execute(“””
insert into user (userid,username,firstname,lastname,phone)
values (4,’scientist’,’Hopeton’,’Brown’,’555-5552’)
“””)
db.commit()
cursor.execute(“select * from user”)
for row in cursor.fetchall():
print row
cursor.close()
db.close()
556

60.8 Page 598

▲back to top
Integrating Java with Python
When you run this script, you will see output like the following:
$ jython jyjdbc.py
(1, ‘ericfj’, ‘Eric’, ‘Foster-Johnson’, ‘555-5555’)
(2, ‘tosh’, ‘Peter’, ‘Tosh’, ‘555-5554’)
(3, ‘bob’, ‘Bob’, ‘Marley’, ‘555-5553’)
(4, ‘scientist’, ‘Hopeton’, ‘Brown’, ‘555-5552’)
How It Works
This script is almost the same as the createtable.py script from Chapter 14. This shows the freedom
the Python DB API gives you, as you are not tied to any one database vendor. Other than the code to
establish the connection to the database, you’ll find your database code can work with multiple
databases.
To establish a connection to HSqlDB, you can use code like the following:
from com.ziclix.python.sql import zxJDBC
# Modify as needed for your database.
url=’jdbc:hsqldb:hsql://localhost/xdb’
user=’sa’
pw=’’
driver=’org.hsqldb.jdbcDriver’
db = zxJDBC.connect(url, user, pw, driver)
This code uses the default connection properties for HSqlDB for simplicity. In a real-world scenario, you
never want to use the default user name and password. Always change the database administrator user
and password. Furthermore, HSqlDB defaults to having no password for the administration user, sa
(short for system administrator). This, of course, provides a large hole in security.
The following code, taken from Chapter 14, creates a new database table:
cursor = db.cursor()
cursor.execute(“””
create table user
(userid integer,
username varchar,
firstname varchar,
lastname varchar,
phone varchar)
“””)
cursor.execute(“””create index userid on user (userid)”””)
While SQL does not standardize the commands necessary to create databases and database tables, this
table sports a rather simple layout, so you should be able to use these commands with most SQL
databases.
557

60.9 Page 599

▲back to top
Chapter 22
The code to insert rows also comes from Chapter 14, as does the query code. In this, it is Python, with
the DB 2.0 API, that provides this commonality. The Jython zxJDBC module follows this API. For exam-
ple, the code to query all the rows from the user table follows:
cursor = db.cursor()
cursor.execute(“select * from user”)
for row in cursor.fetchall():
print row
cursor.close()
The zxJDBC module, though, extends the Python DB API with the concept of static and dynamic cur-
sors. (This ties to the concepts in the java.sql.ResultSet API.) In the Python standard API, you
should be able to access the rowcount attribute of the Cursor object. In Java, a ResultSet may not know
the full row count for a given query, which may have returned potentially millions of rows. Instead, the
JDBC standard allows the ResultSet to fetch data as needed, buffering in the manner determined by the
database vendor or JDBC driver vendor. Most Java code that reads database data will then iterate over
each row provided by the ResultSet.
To support the Python standard, the zxJDBC module needs to read in all the rows to properly determine
the rowcount value. This could use a huge amount of memory for the results of a query on a large table.
This is what the zxJDBC documentation calls a static database cursor.
To avoid the problem of using too much memory, you have the option of getting a dynamic cursor. A
dynamic cursor does not set the rowcount value. Instead, a dynamic cursor fetches data as needed. If
you request a dynamic cursor, you cannot access the rowcount value, but, you can iterate through the
cursor to process all the rows returned by the query. To request a dynamic cursor, pass a 1 to the cursor
method:
cursor = db.cursor(1)
Dynamic cursors are not part of the Python DB API, so code using this technique will not work with any
DB driver except for the Jython zxJDBC driver.
Database access is essential if you are writing enterprise applications. You also need it to be able to create
robust Web applications.
Writing J2EE Servlets in Jython
Most Java development revolves around enterprise applications. To help (or hinder, depending on your
view), Java defines a set of standards called J2EE, or Java 2 Enterprise Edition. The J2EE standards define
an application server and the APIs such a server must support. Organizations can then choose applica-
tion servers from different vendors, such as WebSphere from IBM, WebLogic from Bea, JBoss from the
JBoss Group, and Tomcat from the Apache Jakarta project. Java developers write enterprise applications
that are hosted on one of these application servers.
558

60.10 Page 600

▲back to top
Integrating Java with Python
A servlet is defined as a small server-based application. The term servlet is a play on applet, which
describes a small application. Because in the Java arena applets always run on the client, the server
equivalent needed a new name, hence servlet. Each servlet provides a small piece of the overall applica-
tion, although the term small may be defined differently than you are used to, as most enterprise appli-
cations are huge.
Within a J2EE application server, servlets are passive request-response applications. The client, typically
a web browser such as Internet Explorer or Firefox, sends a request to the application server. The appli-
cation server passes the request to a servlet. The servlet then generates the response, usually an HTML
document, that the server sends back to the client. In virtually all cases, the HTML document sent back
to the client is created dynamically. For example, in a web ordering system, the HTML document sent
back may be the results of a search or the current prices for a set of products.
The benefit of writing servlets is that J2EE provides a well-defined API for writing your servlets, and
multiple vendors support this API. Contrast this situation with the Python situation where you can
choose from many Python Web APIs, but you won’t find anywhere near the vendor support you find in
the J2EE arena.
With Jython, you can write Java servlets in Python, simplifying your work immensely. To do so, though,
you need an application server that supports servlets.
Setting Up an Application Server
If you already have a J2EE application server, use that. If not, try Tomcat. Tomcat, from the Apache
Jakarta project, provides a free open-source servlet engine (called a servlet container in J2EE-speak).
Download Tomcat from http://jakarta.apache.org/tomcat/. Be careful which version you down-
load. Starting with version 5.5, Tomcat requires a 1.5 Java Development Kit, or JDK. Only download a
version of Tomcat that will work with the JDK you have installed. Download a Zip file of the binary
(pre-built) release of the version you select.
To install, unzip the file you downloaded in a directory. You should see a Tomcat directory based on the
version you downloaded, such as jakarta-tomcat-5.0.28. Change to this directory. In this directory,
you will see a number of files and subdirectories. The two most important subdirectories are bin, which
contains scripts for starting and stopping Tomcat, and webapps, which is where you need to place any
Jython scripts you create (in a special subdirectory covered in the next section).
To run Tomcat, change to the bin subdirectory and run the startup.sh script (startup.bat on
Windows). For example:
$ ./startup.sh
Using CATALINA_BASE: /Users/ericfj/servers/jakarta-tomcat-5.0.28
Using CATALINA_HOME: /Users/ericfj/servers/jakarta-tomcat-5.0.28
Using CATALINA_TMPDIR: /Users/ericfj/servers/jakarta-tomcat-5.0.28/temp
Using JAVA_HOME:
/Library/Java/Home
You must ensure that the JAVA_HOME environment variable is set, or Tomcat will not start. To verify
Tomcat is running, enter the following URL into a Web browser: http://localhost:8080/. You
should see a document like the one shown in Figure 22-2.
559

61 Pages 601-610

▲back to top

61.1 Page 601

▲back to top
Chapter 22
Figure 22-2
Once you have an application server such as Tomcat running, the next step is to deploy an application —
in this case, a special Python servlet called PyServlet.
Adding the PyServlet to an Application Server
Jython includes a class called org.python.util.PyServlet that acts as a front end for Python scripts.
The PyServlet class will load Python scripts, compile these scripts, and then execute the scripts as if
they were Java servlets (which, in fact, they are, as shown in the section “Extending HttpServlet,”
following).
To make all this magic work, though, you need to create a bona fide J2EE Web application. Luckily, this
isn’t that hard. Change to the directory in which you installed Tomcat and run the following commands,
which create directories:
$ cd webapps
$ mkdir jython
This command creates a directory under webapps with the name of jython. This means the name of our
Web application will be jython:
$ mkdir webapps/jython/WEB-INF
560

61.2 Page 602

▲back to top
Integrating Java with Python
This command creates a WEB-INF directory. The name and case of this directory are very important. In J2EE,
the WEB-INF directory contains the libraries and deployment information about your Web application:
$ mkdir webapps/jython/WEB-INF/lib
The lib subdirectory holds any jar files needed by your Web application. You need one jar file, jython.
jar, from the Jython installation. Copy this file into the webapps/jython/WEB-INF/lib directory that
you just created.
Next, you need to create a file named web.xml in the webapps/jython/WEB-INF directory. Enter the
following text into web.xml:
<?xml version=”1.0” encoding=”ISO-8859-1”?>
<web-app>
<servlet>
<servlet-name>PyServlet</servlet-name>
<servlet-class>org.python.util.PyServlet</servlet-class>
<init-param>
<param-name>python.home</param-name>
<param-value>/Users/ericfj/Documents/java/jython-2.1</param-value>
</init-param>
</servlet>
<servlet-mapping>
<servlet-name>PyServlet</servlet-name>
<url-pattern>*.py</url-pattern>
</servlet-mapping>
</web-app>
Change the path in bold to the full path to the directory in which you installed Jython.
You should now have two files in your new Web application:
webapps/jython/WEB-INF/lib/jython.jar
webapps/jython/WEB-INF/web.xml
These two files are essential. Next, you need to create some Python scripts within the new Web application.
This chapter presents a whirlwind introduction to J2EE, a frightfully complicated subject. If you’re not
familiar with J2EE, you can look up more information in a J2EE tutorial, such as the one located at
http://java.sun.com/j2ee/1.4/docs/tutorial/doc/index.html.
Extending HttpServlet
The javax.servlet.http.HttpServlet class provides the main hook for J2EE developers to create
servlets. J2EE developers extend HttpServlet with their own classes to create servlets. With the
PyServlet class, you can do the same with Jython. With Jython, however, this task becomes a lot easier
than writing everything in Java.
561

61.3 Page 603

▲back to top
Chapter 22
Use the following code as a template for creating your servlet classes in Jython:
from javax.servlet.http import HttpServlet
class verify(HttpServlet):
def doGet(self, request, response):
self.handleRequest(request, response)
def doPost(self, request, response):
self.handleRequest(request, response)
def handleRequest(self, request, response):
response.setContentType(“text/html”);
out = response.getOutputStream()
print >>out, “YOUR OUTPUT HERE”
out.close()
return
Your classes must inherit from HttpServlet. In addition, you need to create two methods, doGet and
doPost, as described in the following table.
Method
DoGet
DoPost
Usage
Handles GET requests, which place all the parameters on the URL
Handles POST requests, usually with data from a form
In virtually all cases, you want both methods to perform the same task. Any differences in these meth-
ods only serve to make your Web applications harder to debug. Therefore, write another method that
both can call, such as the handleRequest method shown in the previous template.
In your handleRequest method, you must perform a number of tasks. All must be correct, or you will
see an error or no output. These tasks include the following:
❑ Set the proper content type on the response object. In most cases, this will be text/html.
❑ Get an output stream from the response object.
❑ Write all output to this stream.
❑ Close the stream.
The following example shows how to create a real servlet from this code template.
Try It Out Writing a Python Servlet
Enter the following and save the file as webapps/jython/verify.py:
import sys
from javax.servlet.http import HttpServlet
class verify(HttpServlet):
562

61.4 Page 604

▲back to top
Integrating Java with Python
def doGet(self, request, response):
self.handleRequest(request, response)
def doPost(self, request, response):
self.handleRequest(request, response)
def handleRequest(self, request, response):
response.setContentType(“text/html”);
out = response.getOutputStream()
print >>out, “<html><head><title>”
print >>out, “Jython Is Running</title></head>”
print >>out, “<body>”
print >>out, “<h2>Jython is running</h2>”
print >>out, “<p>”
print >>out, “Version:”, sys.version, “ verified.”
print >>out, “</p>”
print >>out, “</body></html>”
out.close()
return
You must place this file within your Web application in the webapps/jython directory. After saving the
file, stop and then restart Tomcat to ensure that your changes are recognized.
Test your new servlet by entering the following URL in your Web browser: http://localhost:8080/
jython/verify.py. Figure 22-3 shows the results you should see.
Figure 22-3
How It Works
Three crucial parts make this servlet work:
❑ Tomcat must be running.
❑ You must have the correct directory structure and contents for your Web application.
❑ The URL must name a Python script in your Web application. The script must have a .py file-
name extension.
563

61.5 Page 605

▲back to top
Chapter 22
In the web.xml file created previously, you registered the servlet PyServlet for all files ending with a
.py extension. Thus, with a URL of http://localhost:8080/jython/verify.py, Tomcat will direct
the servlet PyServlet to handle the request. The following table splits this URL into its important
components.
Component
http://
jython
verify.py
Usage
Defines the protocol used, HTTP in this case
This is the name of your Web application (it could be any name you wanted).
With Tomcat, there must be a webapps/jython directory.
Name of a file within the Web application. The .py extension signals that the
PyServlet should handle the request.
The actual servlet class itself is rather small and follows the code template shown previously. The main
action of this servlet occurs in the handleRequest method:
def handleRequest(self, request, response):
response.setContentType(“text/html”);
out = response.getOutputStream()
print >>out, “<html><head><title>”
print >>out, “Jython Is Running</title></head>”
print >>out, “<body>”
print >>out, “<h2>Jython is running</h2>”
print >>out, “<p>”
print >>out, “Version:”, sys.version, “ verified.”
print >>out, “</p>”
print >>out, “</body></html>”
out.close()
return
Most of this method is a number of print statements, sending HTML-formatted text to the output stream.
Compare this method for creating Web applications with the technologies introduced in Chapter 21.
As you can see, you really need to know both Python and Java, at least a bit, to be able to work with
Jython. That’s why choosing the right tools is important.
Choosing Tools for Jython
Because Jython focuses on working with Java as well as Python, the best choice for Jython tools comes
from the Java arena. The following tools can help with your Jython work:
❑ The jEdit text editor (www.jedit.org/) includes a number of plugins for working with Python.
The editor highlights Python syntax, whether you are working with Python or Jython. In addi-
tion, the JythonInterpreter plugin includes an embedded Jython interpreter. See http://
plugins.jedit.org/ for more on jEdit plugins.
564

61.6 Page 606

▲back to top
Integrating Java with Python
❑ The Eclipse Integrated Development Environment, or IDE, provides excellent support for Java
development. In addition, two Eclipse plugins stand out for Jython usage: PyDev, for working
with Python, at http://sourceforge.net/projects/pydev/; and the RedRobin Jython
development tools plugin, at http://home.tiscali.be/redrobin/jython/. The latter
plugin focuses specifically on Jython. Download Eclipse itself from www.eclipse.org.
Whichever tools you choose, all you really need is a text editor and a command-line shell. Furthermore,
the tools you choose can help with testing, especially testing Java applications.
Testing from Jython
Because Jython provides an interactive environment on top of the Java platform, Jython makes an excel-
lent tool for interactive testing. The following examples show how you can use Jython’s interactive
mode to explore your Java environment.
Try It Out Exploring Your Environment with Jython
Enter the following commands to see information on the Java Map interface:
$ jython
Jython 2.1 on java1.4.2_05 (JIT: null)
Type “copyright”, “credits” or “license” for more information.
>>> from java.util import Map
>>> print dir(Map)
[‘clear’, ‘containsKey’, ‘containsValue’, ‘empty’, ‘entrySet’, ‘equals’, ‘get’,
‘hashCode’, ‘isEmpty’, ‘keySet’, ‘put’, ‘putAll’, ‘remove’, ‘size’, ‘values’]
>>>
How It Works
This example uses the Python dir function to display information about the java.util.Map interface
in Java. You can list information on any Java class or interface.
As another example, you can examine the JNDI, or Java Naming and Directory Interface, classes such as
InitialContext, as shown here:
$ jython
Jython 2.1 on java1.4.2_05 (JIT: null)
Type “copyright”, “credits” or “license” for more information.
>>> from javax.naming import InitialContext
>>> print dir(InitialContext)
[‘bind’, ‘composeName’, ‘createSubcontext’, ‘destroySubcontext’, ‘getNameParser’,
‘list’, ‘listBindings’, ‘lookup’, ‘lookupLink’, ‘rebind’, ‘rename’, ‘unbind’]
>>>
Combine this technique with an embedded Jython interpreter to examine a running application. See the
following section, “Embedding the Jython Interpreter,” for more information on embedding the Jython
interpreter.
565

61.7 Page 607

▲back to top
Chapter 22
In addition to using Jython’s interactive mode, you can also write tests in Jython.
Many organizations shy away from open-source software such as Jython. You may find it much easier
to introduce Jython just for writing tests, something that will not go into production. Once your orga-
nization gains some experience with Jython, people may be more receptive to using Jython in more
areas.
The examples so far have all used the jython script to run Jython scripts, except for the PyServlet
servlet example. With the PyServlet class, you have a Java class with the Jython interpreter. You can
add the Jython interpreter to your own classes as well.
Embedding the Jython Interpreter
By embedding the Jython interpreter in your own Java classes, you can run scripts from within your
application, gaining control over the complete environment. That’s important because few Java applica-
tions run from the command line.
You can find the Jython interpreter in the class org.python.util.PythonInterpreter.
You can use code like the following to initialize the Jython interpreter:
Properties props = new Properties();
props.put(“python.home”, pythonHome);
PythonInterpreter.initialize(
System.getProperties(),
props,
new String[0]);
interp = new PythonInterpreter(null, new PySystemState());
Note that this is Java code, not Python code.
You must set the python.home system property.
Calling Jython Scripts from Java
After initializing the interpreter, you can execute a Jython script with a call to the execfile method. For
example:
interp.execfile(fileName);
You need to pass the full name of the file to execute. You can see this in action with the following
example.
566

61.8 Page 608

▲back to top
Integrating Java with Python
Try It Out Embedding Jython
Enter the following Java program and name the file JyScriptRunner.java:
package jython;
import java.util.Properties;
import org.python.util.PythonInterpreter;
import org.python.core.PySystemState;
/**
* Runs Jython scripts.
*/
public class JyScriptRunner {
private PythonInterpreter interp;
/**
* Initializes the Jython interpreter.
*/
public void initialize(String pythonHome) {
Properties props = new Properties();
props.put(“python.home”, pythonHome);
PythonInterpreter.initialize(
System.getProperties(),
props,
new String[0]);
interp = new PythonInterpreter(null, new PySystemState());
}
/**
* Runs the given script.
*/
public void run(String fileName) {
interp.execfile(fileName);
}
public static void main(String[] args) {
String fileName = args[0];
JyScriptRunner runner = new JyScriptRunner();
String pythonHome = System.getProperty(“python.home”);
runner.initialize(pythonHome);
runner.run(fileName);
}
}
567

61.9 Page 609

▲back to top
Chapter 22
Because this is a Java program, you will need to compile the program with a command like the
following:
$ javac -classpath ./jython.jar JyScriptRunner.java
When you run this Java program, you will see output like the following:
$ java -cp ./jython.jar:. \\
-Dpython.home=”/Users/ericfj/Documents/java/jython-2.1” \\
jython.JyScriptRunner jystring.py
The platform is: java1.4.2_05 time for an omelette.
Home directory: /Users/ericfj
Some numbers: 44.1, 42 (‘Red’, ‘Green’, ‘Blue’, 255, 204, 127)
User home directory: /Users/ericfj
This example runs the jystring.py example script. You will need to change the -Dpython.home set-
ting to the location where you have installed Jython. Also change the ./jython.jar to the location
where you have the file jython.jar.
How It Works
The program expects the caller to pass two values: the setting for the python.home system property
and the name of a script to execute. You must have the jython.jar located in the current directory (or
change the command line to refer to the location of your jython.jar file).
The JyScriptRunner class includes a main method, called when you run the program. The main
method extracts the system property python.home as well as the filename from the command line (held
in the args array). The main method then instantiates a JyScriptRunner object.
The main method initializes the JyScriptRunner object and then calls the run method to execute the
script. Any errors encountered in the Jython script will result in exceptions that stop the program.
This is probably about the simplest Jython interpreter you can create. In your applications, you’ll likely
want to control the location of the Python home directory, perhaps placing this under an application
directory.
Compiling Python Code to Java
You can use the jythonc script to compile your Jython code into Java code, and then, into Java .class
files. At its simplest, you can call jythonc with the following arguments:
jythonc --package package_name file.py
568

61.10 Page 610

▲back to top
Integrating Java with Python
This will create a directory called jpywork, and inside that directory, .java and .class files. For example,
you can compile the jystring.py script introduced previously in this chapter with a command like the
following:
$ jythonc --package jy jystring.py
processing jystring
Required packages:
java.lang
Creating adapters:
Creating .java files:
jystring module
Compiling .java to .class...
Compiling with args:
[‘/System/Library/Frameworks/JavaVM.framework/Versions/1.4.2/Home/bin/javac’,
‘-classpath’, ‘/Users/ericfj/Documents/java/jython-
2.1/jython.jar::./hsqldb.jar:./jpywork::
/Users/ericfj/Documents/java/jython2.1/Tools/jythonc:
/Users/ericfj/writing/python/chap22/.:
/Users/ericfj/Documents/java/jython-2.1/Lib’,
‘./jpywork/jy/jystring.java’]
0
When complete, you should see the following files in the jpywork directory:
$ ls jpywork/jy
jystring$_PyInner.class
jystring.class
jystring.java
Handling Differences between C Python and
Jython
The C Python platform creates a complete environment based on Python standards and conventions.
Jython, on the other hand, tries to create a complete Python environment based on the Java platform.
Because of this, there are bound to be differences between the two implementations. These differences
are compounded when you mix Java code into your Jython scripts.
The Jython interpreter will attempt to convert Python types into the necessary Java types to call methods
on Java classes. Wherever possible, the Jython interpreter tries to do the right thing, so in most cases you
don’t have to pay much attention to these type conversions. If you are unsure which Python types are
needed to call a particular Java method, look at the types listed in the following table.
569

62 Pages 611-620

▲back to top

62.1 Page 611

▲back to top
Chapter 22
Python Type
None
Integer (any non-zero
value is true)
Integer
String
String of length 1
Float
String
Any
Class or JavaClass
Array (must contain
objects of a given type
or subclasses of the
given type)
Java Type
Null
Boolean
short, int, long, byte
byte[], char[], java.lang.String
Char
float, double
java.lang.Object, converted to java.lang.String
java.lang.Object
java.lang.Class
Array of a particular type
For example, if a Java method expects a type of java.lang.Object and you pass a Python String,
Jython will convert the Python String to a java.langString object. Jython will pass any other Python
object type unchanged.
You can do many more things with Jython beyond the introduction provided in this chapter. For exam-
ple, you can create classes in Jython and then call those classes from Java. (Look in the source code for
the PyServlet class to see an example of this.)
Summary
Jython provides the capability to combine the scripting power of Python with the enterprise infrastruc-
ture of Java. Using Jython can make you a much more productive Java developer, especially in organiza-
tions where Python is not accepted but Java is.
Jython allows you to do the following:
❑ Run Python scripts from the Java platform. Because these scripts differ from Python, they are
usually called Jython scripts.
❑ Call on Java code and classes from within your scripts. This enables you to take advantage of
the rich set of Java libraries from Jython scripts.
❑ Create user interfaces with the Java Swing API. Jython scripts can use Python’s tuple and prop-
erty support to dramatically reduce the code required to create Swing-based user interfaces.
❑ Compile Jython code to Java .class files with the jythonc script.
570

62.2 Page 612

▲back to top
Integrating Java with Python
❑ Access any database that provides a JDBC driver. The zxJDBC driver bridges from the Python
DB API to the Java JDBC API.
❑ Run Jython scripts as Java servlets by using the handy PyServlet class from your J2EE web
applications.
❑ Interactively gather information on Java classes and execute methods on those classes. This is
very useful for testing.
❑ Embed the Jython interpreter in your own Java classes, enabling you to execute Jython scripts
and expressions from your Java code.
This chapter wraps up this tutorial on Python. The appendixes provide answers to the chapter exercises,
links to Python resources, as well as a discussion of new features in the Python 2.4 release.
Exercises
1. If Python is so cool, why in the world would anyone ever use another programming language
such as Java, C++, C#, Basic, or Perl?
2. The Jython interpreter is written in what programming language? The python command is
written in what programming language?
3. When you package a Jython-based application for running on another system, what do you
need to include?
4. Can you use the Python DB driver modules, such as those described in Chapter 14, in your
Jython scripts?
5. Write a Jython script that creates a window with a red background using the Swing API.
571

62.3 Page 613

▲back to top

62.4 Page 614

▲back to top
A
Answers to Exercises
Chapter 1
Exercise 1 solution
‘Rock a by baby,\\n\\ton the tree top,\\t\\twhen the wind blows\\n\\t\\t\\t the cradle
will drop.’
Because this is not being printed, the special characters (those preceded with a backslash) are not
translated into a form that will be displayed differently from how you typed them.
Exercise 2 solution
Rock a by baby,
on the tree top,
when the wind blows
the cradle will drop.
When they are printed, “\\n” and “\\t” produce a newline and a tab character, respectively. When
the print function is used, it will render them into special characters that don’t appear on your
keyboard, and your screen will display them.
Chapter 2
Exercise 1 solution
>>> 5 * 10
50
Exercise 2 solution
>>> 122/3
40
>>> 122.0/3
40.666666666666664

62.5 Page 615

▲back to top
Appendix A
Exercise 3 solution
>>> print “%o” % 6
6
>>> print “%o” % 7
7
>>> print “%o” % 8
10
>>> print “%o” % 9
11
>>> print “%o” % 10
12
>>> print “%o” % 11
13
>>> print “%o” % 12
14
>>> print “%o” % 13
15
>>> print “%o” % 14
16
Exercise 4 solution
>>> print “%x” % 9
9
>>> print “%x” % 10
a
>>> print “%x” % 11
b
>>> print “%x” % 12
c
>>> print “%x” % 13
d
>>> print “%x” % 14
e
>>> print “%x” % 15
f
>>> print “%x” % 16
10
>>> print “%x” % 17
11
>>> print “%x” % 18
12
>>> print “%x” % 19
13
Exercise 5 solution
When an unknown function is called, Python doesn’t know that the name that’s been typed in is neces-
sarily a function at all, so it just flags a general syntax error:
>>> pintr “%x” & x
File “<input>”, line 1
pintr “%x” & x
^
SyntaxError: invalid syntax
574

62.6 Page 616

▲back to top
Answers to Exercises
You’ll notice, however, that codeEditor will display print in bold when you type it. This is because print
is a special word to Python, and codeEditor knows this. You can help yourself catch errors by paying
attention to how the editor reacts to what you’ve typed.
Exercise 6 solution
>>> print “%u” % -10
-10
Python 2.4 will turn the unsigned integer into a signed integer instead of confusing you. In prior ver-
sions, it would have given you a very large number — somewhere a bit more than four billion.
Chapter 3
Exercise 1 solution
>>> dairy_section = [“milk”, “cottage cheese”, “butter”, “yogurt”]
Exercise 2 solution
>>> print “First: %s and Last %s” % (dairy_section[0], dairy_section[1])
First: milk and Last cottage cheese
Exercise 3 solution
>>> milk_expiration = (10, 10, 2005)
Exercise 4 solution
>>> print “This milk will expire on %d/%d/%d” % (milk_expiration[0],
milk_expiration[1], milk_expiration[2])
This milk will expire in 10/10/2005
Exercise 5 solution
>>> milk_carton = {}
>>> milk_carton[“expiration_date”] = milk_expiration
>>> milk_carton[“fl_oz”] = 32
>>> milk_carton[“cost”] = 1.50
>>> milk_carton[“brand_name”] = “Milk”
Exercise 6 solution
>>> print “The expiration date is %d/%d/%d” % (milk_carton[“expiration_date”][0],
milk_carton[“expiration_date”][1], milk_carton[“expiration_date”][2])
The expiration date is 10/10/2005
Exercise 7 solution
>>> print “The cost for 6 cartons of milk is %.02f” % (6* milk_carton[“cost”])
The cost for 6 cartons of milk is 9.00
575

62.7 Page 617

▲back to top
Appendix A
Exercise 8 solution
>>> cheeses = [“cheddar”, “american”, “mozzarella”]
>>> dairy_section.append(cheeses)
>>> dairy_section
[‘milk’, ‘cottage cheese’, ‘butter’, ‘yogurt’, [‘cheddar’, ‘american’,
‘mozzarella’]]
>>> dairy_section.pop()
[‘cheddar’, ‘american’, ‘mozzarella’]
Exercise 9 solution
>>> len(dairy_section)
4
Exercise 10 solution
>>> print “Part of some cheese is %s” % cheeses[0][0:5]
Part of some cheese is chedd
Chapter 4
At this point, the examples can become long. You should consider typing the code into codeEditor’s
main window and cutting and pasting lines so that you don’t have too much to type if there is a prob-
lem. Even if you don’t choose to do this, you can use the Shift and up and down arrows to bring back
lines you’ve already typed, and you can edit these lines. Try it out.
Exercise 1 solution
The key theme here is that 0 is False, and everything else is considered not False, which is the same as
True:
>>> if 0:
...
print “0 is True”
...
>>> if 1:
...
print “1 is True”
...
1 is True
>>> if 2:
...
print “2 is True”
...
2 is True
>>> if 3:
...
print “3 is True”
...
3 is True
>>> if 4:
...
print “4 is True”
...
4 is True
>>> if 5:
576

62.8 Page 618

▲back to top
...
print “5 is True”
...
5 is True
Answers to Exercises
Exercise 2 solution
>>> number = 3
>>> if number >= 0 and number <= 9:
... print “The number is between 0 and 9: %d” % number
...
The number is between 0 and 9: 3
Exercise 3 solution
>>> test_tuple = (“this”, “little”, “piggie”, “went”, “to”, “market”)
>>> search_string = “toes”
>>> if test_tuple[0] == search_string:
...
print “The first element matches”
... elif test_tuple[1] == search_string:
...
print “the second element matches”
... else:
...
print “%s wasn’t found in the first two elements” % search_string
...
toes wasn’t found in the first two elements
Exercise 4 solution
>>> fridge = {“butter”:”Dairy spread”, “peanut butter”:”non-dairy spread”,
“cola”:”fizzy water”}
>>> food_sought = “chicken”
>>> for food_key in fridge.keys():
...
if food_key == food_sought:
...
print “Found what I was looking for: %s is %s” % (food_sought,
fridge[food_key])
...
break
... else:
...
print “%s wasn’t found in the fridge” % food_sought
...
chicken wasn’t found in the fridge
Exercise 5 solution
>>> fridge = {“butter”:”Dairy spread”, “peanut butter”:”non-dairy spread”,
“cola”:”fizzy water”}
>>> fridge_list = fridge.keys()
>>> current_key = fridge_list.pop()
>>> food_sought = “cola”
>>> while len(fridge_list) > 0:
...
if current_key == food_sought:
...
print “Found what I was looking for: %s is %s” % (food_sought,
fridge[current_key])
...
break
...
current_key = fridge_list.pop()
577

62.9 Page 619

▲back to top
Appendix A
... else:
...
print “%s wasn’t found in the fridge” % food_sought
...
Found what I was looking for: cola is fizzy water
Exercise 6 solution
>>> fridge = {“butter”:”Dairy spread”, “peanut butter”:”non-dairy spread”,
“cola”:”fizzy water”}
>>> food_sought = “chocolate milk”
>>> try:
...
fridge[food_sought]
... except KeyError:
...
print “%s wasn’t found in the fridge” % food_sought
... else:
...
print “Found what I was looking for: %s is %s” % (food_sought,
fridge[food_key])
...
chocolate milk wasn’t found in the fridge
Chapter 5
Exercise 1 solution
def do_plus(first, second):
return first + second
Exercise 2 solution
def do_plus(first, second):
for param in (first, second):
if (type(param) != type(“”)) and (type(param) != type(1)):
raise TypeError, “This function needs a string or an integer”
return first + second
Exercise 3 solution
# Part 1 - fridge has to go before the omelet_type. omelet_type is an
# optional parameter with a default parameter, so it has to go at the end.
# This can be used with a fridge such as:
# f = {‘eggs’:12, ‘mozzarella cheese’:6,
#
‘milk’:20, ‘roast red pepper’:4, ‘mushrooms’:3}
# or other ingredients, as you like.
def make_omelet_q3(fridge, omelet_type = “mozzarella”):
“””This will make an omelet. You can either pass in a dictionary
that contains all of the ingredients for your omelet, or provide
a string to select a type of omelet this function already knows
about
The default omelet is a mozerella omelet”””
def get_omelet_ingredients(omelet_name):
578

62.10 Page 620

▲back to top
Answers to Exercises
“””This contains a dictionary of omelet names that can be produced,
and their ingredients”””
# All of our omelets need eggs and milk
ingredients = {“eggs”:2, “milk”:1}
if omelet_name == “cheese”:
ingredients[“cheddar”] = 2
elif omelet_name == “western”:
ingredients[“jack_cheese”] = 2
ingredients[“ham”]
=1
ingredients[“pepper”]
=1
ingredients[“onion”]
=1
elif omelet_name == “greek”:
ingredients[“feta_cheese”] = 2
ingredients[“spinach”]
=2
# Part 5
elif omelet_name == “mozzarella”:
ingredients[“mozzarella cheese”] = 2
ingredients[“roast red pepper”] = 2
ingredients[“mushrooms”] = 1
else:
print “That’s not on the menu, sorry!”
return None
return ingredients
# part 2 - this version will use the fridge that is available
# to the make_omelet function.
def remove_from_fridge(needed):
recipe_ingredients = {}
# First check to ensure we have enough
for ingredient in needed.keys():
if needed[ingredient] > fridge[ingredient]:
raise LookupError, “not enough %s to continue” % ingredient
# Then transfer the ingredients.
for ingredient in needed.keys():
# Remove it from the fridge
fridge[ingredient] = fridge[ingredient] - needed[ingredient]
# and add it to the dictionary that will be returned
recipe_ingredients[ingredient] = needed[ingredient]
# Part 3 - recipe_ingredients now has all the needed ingredients
return recipe_ingredients
# Part 1, continued - check the type of the fridge
if type(fridge) != type({}):
raise TypeError, “The fridge isn’t a dictionary!”
if type(omelet_type) == type({}):
print “omelet_type is a dictionary with ingredients”
return make_food(omelet_type, “omelet”)
elif type(omelet_type) == type(“”):
needed_ingredients = get_omelet_ingredients(omelet_type)
omelet_ingredients = remove_from_fridge(needed_ingredients)
return make_food(omelet_ingredients, omelet_type)
else:
print “I don’t think I can make this kind of omelet: %s” % omelet_type
579

63 Pages 621-630

▲back to top

63.1 Page 621

▲back to top
Appendix A
Exercise 4 solution
The get_omelet_ingredient from make_omelet_q3 could be changed to look like the following:
def get_omelet_ingredients(omelet_name):
“””This contains a dictionary of omelet names that can be produced,
and their ingredients”””
# All of our omelets need eggs and milk
ingredients = {“eggs”:2, “milk”:1}
if omelet_name == “cheese”:
ingredients[“cheddar”] = 2
elif omelet_name == “western”:
ingredients[“jack_cheese”] = 2
ingredients[“ham”]
=1
ingredients[“pepper”]
=1
ingredients[“onion”]
=1
elif omelet_name == “greek”:
ingredients[“feta_cheese”] = 2
ingredients[“spinach”]
=2
# Part 5
elif omelet_name == “mozerella”:
ingredients[“mozerella cheese”] = 2
ingredients[“roast red pepper”] = 2
ingredients[“mushrooms”] = 1
# Question 4 - we don’ want anyone hurt in our kitchen!
elif omelet_name == “salmonella”:
raise TypeError, “We run a clean kitchen, you won’t get this here”
else:
print “That’s not on the menu, sorry!”
return None
return ingredients
When run, the error raised by trying to get the salmonella omelet will result in the following error:
>>> make_omelet_q3({‘mozzarella cheese’:5, ‘eggs’:5, ‘milk’:4, ‘roast red
pepper’:6, ‘mushrooms’:4}, “salmonella”)
Traceback (most recent call last):
File “<stdin>”, line 1, in ?
File “ch5.py”, line 209, in make_omelet_q3
omelet_ingredients = get_omelet_ingredients(omelet_type)
File “ch5.py”, line 179, in get_omelet_ingredients
raise TypeError, “We run a clean kitchen, you won’t get this here”
TypeError: We run a clean kitchen, you won’t get this here
>>>
You can see from this that the program was run from <stdin>, which means it was run interactively,
with python -i ch5.py or by using codeEditor’s Run with Interpreter option. The function make_
omelet_q3 was called, and the function get_omelet_ingredients was called on line 209 in ch5.py.
Python shows you the actual line that was run, in case you needed more information at first glance.
Note that depending on the contents of your ch5.py file, the exact line numbers shown in your stack
trace will be different from those shown here.
You can next see that line 179 is where get_omelet_ingredients raised the error (though it may be at
a different line in your own file).
580

63.2 Page 622

▲back to top
Answers to Exercises
If you called this from within another function, the stack would be one layer deeper, and you would see
the information relating to that extra layer as well.
Chapter 6
Exercise 1 solution
def mix(self, display_progress = True):
“””
mix(display_progress = True) - Once the ingredients have been obtained from
a fridge call this
to prepare the ingredients. If display_progress is False do not print
messages.
“””
for ingredient in self.from_fridge.keys():
if display_progress == True:
print “Mixing %d %s for the %s omelet” %
(self.from_fridge[ingredient], ingredient, self.kind)
self.mixed = True
Exercise 2 solution
Note that you could go one step further and make the quiet setting of the mix function an option, too. As
it is, this doesn’t give you much feedback about what’s going on, so when you test it, it may look a bit
strange.
def quick_cook(self, fridge, kind = “cheese”, quantity = 1):
“””
quick_cook(fridge, kind = “cheese”, quantity = 1) -
performs all the cooking steps needed. Turns out an omelet fast.
“””
self.set_kind(kind)
self.get_ingredients(fridge)
self.mix(False)
self.make()
Exercise 3 solution
Just the documentation, not the functions, would look something like this. However, you should find a
format that suits you.
Note that only undocumented functions will have their docstrings described here.
class Omelet:
“””This class creates an omelet object. An omelet can be in one of
two states: ingredients, or cooked.
An omelet object has the following interfaces:
get_kind() - returns a string with the type of omelet
set_kind(kind) - sets the omelet to be the type named
set_new_kind(kind, ingredients) - lets you create an omelet
mix() - gets called after all the ingredients are gathered from the fridge
581

63.3 Page 623

▲back to top
Appendix A
cook() - cooks the omelet
“””
def __init__(self, kind=”cheese”):
“””__init__(self, kind=”cheese”)
This initializes the Omelet class to default to a cheese omelet.
Other methods
“””
self.set_kind(kind)
return
def set_kind(self, kind):
“””
set_kind(self, kind) - changes the kind of omelet that will be created
if the type of omelet requested is not known then return False
“””
def get_kind(self):
“””
get_kind() - returns the kind of omelet that this object is making
“””
def set_kind(self, kind):
“””
set_kind(self, kind) - changes the kind of omelet that will be created
if the type of omelet requested is not known then return False
“””
def set_new_kind(self, name, ingredients):
“””
set_new_kind(name, ingredients) - create a new type of omelet that is
called “name” and that has the ingredients listed in “ingredients”
“””
def __known_kinds(self, kind):
“””
__known_kinds(kind) - checks for the ingredients of “kind” and returns them
returns False if the omelet is unknown.
“””
def get_ingredients(self, fridge):
“””
get_ingredients(fridge) - takes food out of the fridge provided
“””
def mix(self):
“””
mix() - Once the ingredients have been obtained from a fridge call this
to prepare the ingredients.
“””
def make(self):
“””
make() - once the ingredients are mixed, this cooks them
“””
582

63.4 Page 624

▲back to top
Answers to Exercises
Exercise 4 solution
>>> print “%s” % o.__doc__
This class creates an omelet object. An omelet can be in one of
two states: ingredients, or cooked.
An omelet object has the following interfaces:
get_kind() - returns a string with the type of omelet
set_kind(kind) - sets the omelet to be the type named
set_new_kind(kind, ingredients) - lets you create an omelet
mix() - gets called after all the ingredients are gathered from the fridge
cook() - cooks the omelet
>>> print “%s” % o.set_new_kind.__doc__
set_new_kind(name, ingredients) - create a new type of omelet that is
called “name” and that has the ingredients listed in “ingredients”
You can display the remaining docstrings in the same way.
Exercise 5 solution
class Recipe:
“””
This class houses recipes for use by the Omelet class
“””
def __init__(self):
self.set_default_recipes()
return
def set_default_recipes(self):
self.recipes = {“cheese” : {“eggs”:2, “milk”:1, “cheese”:1},
“mushroom” : {“eggs”:2, “milk”:1, “cheese”:1,
“mushroom”:2},
“onion” : {“eggs”:2, “milk”:1, “cheese”:1, “onion”:1}}
def get(self, name):
“””
get(name) - returns a dictionary that contains the ingredients needed to
make the omelet in name.
When name isn’t known, returns False
“””
try:
recipe = self.recipes[name]
return recipe
except KeyError:
return False
def create(self, name, ingredients):
“””
create(name, ingredients) - adds the omelet named “name” with the
ingredients
“ingredients” which is a dictionary.
“””
self.recipes[name] = ingredients
583

63.5 Page 625

▲back to top
Appendix A
Exercise 6 solution
Note that the order of parameters in the interface for the class has now been changed, because you can’t
place a required argument after a parameter that has an optional default value.
When you test this, remember that you now create an omelet with a recipe as its mandatory parameter.
def __init__(self, recipes, kind=”cheese”):
“””__init__(self, recipes, kind=”cheese”)
This initializes the omelet class to default to a cheese omelet.
“””
self.recipes = recipes
self.set_kind(kind)
return
def set_new_kind(self, name, ingredients):
“””
set_new_kind(name, ingredients) - create a new type of omelet that is
called “name” and that has the ingredients listed in “ingredients”
“””
self.recipes.create(name, ingredients)
self.set_kind(name)
return
def __known_kinds(self, kind):
“””
__known_kinds(kind) - checks for the ingredients of “kind” and returns them
returns False if the omelet is unknown.
“””
return self.recipes.get(kind)
Chapter 7
Exercise 1 solution
Remember that you’re not a regular user of your class when you write tests. You should feel free to
access internal names if you need to!
if __name__ == ‘__main__’:
r = Recipe()
if r.recipes != {“cheese” : {“eggs”:2, “milk”:1, “cheese”:1},
“mushroom” : {“eggs”:2, “milk”:1, “cheese”:1,
“mushroom”:2},
“onion” : {“eggs”:2, “milk”:1, “cheese”:1, “onion”:1}}:
print “Failed: the default recipes is not the correct list”
cheese_omelet = r.get(“cheese”)
if cheese_omelet != {“eggs”:2, “milk”:1, “cheese”:1}:
print “Failed: the ingredients for a cheese omelet are wrong”
western_ingredients = {“eggs”:2, “milk”:1, “cheese”:1, “ham”:1, “peppers”:1,
“onion”:1}
r.create(“western”, western_ingredients)
if r.get(“western”) != western_ingredients:
584

63.6 Page 626

▲back to top
Answers to Exercises
print “Failed to set the ingredients for the western”
else:
print “Succeeded in getting the ingredients for the western.”
Exercise 2 solution
At the end of the Fridge module, insert the following code. Note the comment about changing the
add_many function to return True. If you don’t do that, add_many will return None, and this test will
always fail!
if __name__ == ‘__main__’:
f = Fridge({“eggs”:10, “soda”:9, “nutella”:2})
if f.has(“eggs”) != True:
print “Failed test f.has(‘eggs’)”
else:
print “Passed test f.has(‘eggs’)”
if f.has(“eggs”, 5) != True:
print “Failed test f.has(‘eggs’, 5)”
else:
print “Passed test f.has(‘eggs’, 5)”
if f.has_various({“eggs”:4, “soda”:2, “nutella”:1}) != True:
print ‘Failed test f.has_various({“eggs”:4, “soda”:2, “nutella”1})’
else:
print ‘Passed test f.has_various({“eggs”:4, “soda”:2, “nutella”1})’
# Check to see that when we add items, that the number of items in the fridge
# is increased!
item_count = f.items[“eggs”]
if f.add_one(“eggs”) != True:
print ‘Failed test f.add_one(“eggs”)’
else:
print ‘Passed test f.add_one(“eggs”)’
if f.items[“eggs”] != (item_count + 1):
print ‘Failed f.add_one() did not add one’
else:
print ‘Passed f.add_one() added one’
item_count = {}
item_count[“eggs”] = f.items[“eggs”]
item_count[“soda”] = f.items[“soda”]
# Note that the following means you have to change add_many to return True!
if f.add_many({“eggs”:3,”soda”:3}) != True:
print ‘Failed test f.add_many({“eggs”:3,”soda”:3})’
else:
print ‘Passed test f.add_many({“eggs”:3,”soda”:3})’
if f.items[“eggs”] != (item_count[“eggs”] + 3):
print “Failed f.add_many did not add eggs”
else:
print “Passed f.add_many added eggs”
if f.items[“soda”] != (item_count[“soda”] + 3):
print “Failed f.add_many did not add soda”
else:
print “Passed f.add_many added soda”
item_count = f.items[“eggs”]
if f.get_one(“eggs”) != True:
585

63.7 Page 627

▲back to top
Appendix A
print ‘Failed test f.get_one(“eggs”)’
else:
print ‘Passed test f.get_one(“eggs”)’
if f.items[“eggs”] != (item_count - 1):
print “Failed get_one did not remove an eggs”
else:
print “Passed get_one removed an eggs”
item_count = {}
item_count[“eggs”] = f.items[“eggs”]
item_count[“soda”] = f.items[“soda”]
eats = f.get_many({“eggs”:3, “soda”:3})
if eats[“eggs”] != 3 or eats[“soda”] != 3:
print ‘Failed test f.get_many({“eggs”:3, “soda”:3})’
else:
print ‘Passed test f.get_many({“eggs”:3, “soda”:3})’
if f.items[“eggs”] != (item_count[“eggs”] - 3):
print “Failed get many didn’t remove eggs”
else:
print “Passed get many removed eggs”
if f.items[“soda”] != (item_count[“soda”] - 3):
print “Failed get many didn’t remove soda”
else:
print “Passed get many removed soda”
Exercise 3 solution
You can try to generate errors by mistyping the name of a key in one place in the module, and confirm-
ing that this results in your tests warning you. If you find situations that these tests don’t catch, you
should try to code a test for that situation so it can’t ever catch you.
Chapter 8
Exercise 1 solution
Here’s a simple but inefficient way to solve the problem:
import os
def print_dir(dir_path):
# Loop through directory entries, and print directory names.
for name in sorted(os.listdir(dir_path)):
full_path = os.path.join(dir_path, name)
if os.path.isdir(full_path):
print full_path
# Loop again, this time printing files.
for name in sorted(os.listdir(dir_path)):
full_path = os.path.join(dir_path, name)
if os.path.isfile(full_path):
print full_path
586

63.8 Page 628

▲back to top
Answers to Exercises
Here’s the extra-credit solution, which only scans and sorts the directory once:
import os
def print_dir(dir_path):
# Loop through directory entries. Since we sort the combined
# directory entries first, the subdirectory names and file names
# will each be sorted, too.
file_names = []
for name in sorted(os.listdir(dir_path)):
full_path = os.path.join(dir_path, name)
if os.path.isdir(full_path):
# Print subdirectory names now.
print full_path
elif os.path.isfile(full_path):
# Store file names for later.
file_names.append(full_path)
# Now print the file names.
for name in file_names:
print name
Exercise 2 solution
import os
import shutil
def make_version_path(path, version):
if version == 0:
return path
else:
return path + “.” + str(version)
def rotate(path, max_keep, version=0):
“””Rotate old versions of file ‘path’.
Keep up to ‘max_keep’ old versions with suffixes .1, .2, etc.
Larger numbers indicate older versions.”””
src_path = make_version_path(path, version)
if not os.path.exists(src_path):
# The file doesn’t exist, so there’s nothing to do.
return
dst_path = make_version_path(path, version + 1)
if os.path.exists(dst_path):
# There already is an old version with this number.
if version < max_keep - 1:
# Renumber the old version.
rotate(path, max_keep, version + 1)
else:
# Too many old versions, so remove it.
os.remove(dst_path)
What to do?
shutil.move(src_path, dst_path)
587

63.9 Page 629

▲back to top
Appendix A
Exercise 3 solution
Here is the program to add an entry to the diary. The diary data is stored in a filename’s diary-
data.pickle.
import time
import os
import pickle
diary_file = “diary-data.pickle”
# Does the diary file exist?
if os.path.exists(diary_file):
# Yes. Load it.
diary_data = pickle.load(file(diary_file))
else:
# No. Start a new diary.
diary_data = []
# Get the current date and time.
date_and_time = time.ctime(time.time())
# Ask the user to input the diary text.
text = raw_input(“Your thoughts today? “)
# Add the entry to the diary.
diary_data.append((date_and_time, text))
# Write the diary data.
pickle.dump(diary_data, file(diary_file, “w”))
Following is the program to print the diary:
import pickle
# Load the diary data.
diary_data = pickle.load(file(“diary-data.pickle”))
# Print the diary entries, most recent first.
for date_and_time, text in reversed(diary_data):
print “Thoughts on %s:” % date_and_time
print “%s” % text
print
Chapter 10
Exercise 1 solution
You get access to the functionality with a module by importing the module or items from the module.
Exercise 2 solution
If you define the variable __all__, you can list the items that make up the public API for the module.
For example:
__all__ = [‘Meal’,’AngryChefException’, ‘makeBreakfast’,
‘makeLunch’, ‘makeDinner’, ‘Breakfast’, ‘Lunch’, ‘Dinner’]
588

63.10 Page 630

▲back to top
Answers to Exercises
If you do not define the __all__ variable (although you should), the Python interpreter looks for all
items with names that do not begin with an underscore.
Exercise 3 solution
The help function displays help on any module you have imported. The basic syntax follows:
help(module)
You can also use the pydoc command, not covered in this chapter.
Exercise 4 solution
Look in the directories listed in the variable sys.path for the locations of modules on your system. You
need to import the sys module first.
Exercise 5 solution
Any Python commands can be placed in a module. Your modules can have Python commands, Python
functions, Python variables, Python classes, and so on. In most cases, though, you want to avoid running
commands in your modules. Instead, the module should define functions and classes and let the caller
decide what to invoke.
Chapter 11
Exercise 1 solution
import os, os.path
import re
def print_pdf (arg, dir, files):
for file in files:
path = os.path.join (dir, file)
path = os.path.normcase (path)
if not re.search (r”.*\\.pdf”, path): continue
if re.search (r” “, path): continue
print path
os.path.walk (‘/’, print_pdf, 0)
Note how this example just changes the name of the directory to start processing with the
os.path.walk function.
Exercise 2 solution
import os, os.path
import re
def print_pdf (arg, dir, files):
for file in files:
589

64 Pages 631-640

▲back to top

64.1 Page 631

▲back to top
Appendix A
path = os.path.join (dir, file)
path = os.path.normcase (path)
if not re.search (r”.*\\.pdf”, path): continue
if not re.search (r”boobah”, path): continue
print path
os.path.walk (‘.’, print_pdf, 0)
This example just includes an additional test in the print_pdf function.
Exercise 3 solution
import os, os.path
import re
def print_pdf (arg, dir, files):
for file in files:
path = os.path.join (dir, file)
path = os.path.normcase (path)
if not re.search (r”.*\\.pdf”, path): continue
if re.search (r”boobah”, path): continue
print path
os.path.walk (‘.’, print_pdf, 0)
Note how this example simply removes the not from the second test.
Chapter 13
Exercise 1 solution
The first step is to write a Glade interface. This should take just a few clicks now that you’re familiar
with the tools. This GUI is very similar to our previous pyRAP GUI, so we’ll just show the finished pro-
ject (see Figure A-1). Save that Glade project to the current directory, and we’ll write some code to use it.
590
Figure A-1

64.2 Page 632

▲back to top
Answers to Exercises
The following code shows you how to do this simple exercise. You’ve seen almost all of this code before.
Of course, you import os.popen to do the work of running a command. This is done in the send
method:
#!/usr/bin/env python
import findgtk
import gtk
import gtk.glade
class ex1gui:
def __init__(self):
self.wTree = gtk.glade.XML (“ex1.glade”, “window1”)
dic={ “on_window1_destroy” : self.quit,
“on_button1_clicked” : self.send,
}
self.wTree.signal_autoconnect (dic)
#setup the log window
self.logwindowview=self.wTree.get_widget(“textview1”)
self.logwindow=gtk.TextBuffer(None)
self.logwindowview.set_buffer(self.logwindow)
return
def send(self,obj):
print “send called”
command=self.wTree.get_widget(“entry1”).get_text()
print “Command: %s”%command
self.log(“Running: “+command,”black”)
import os
fd=os.popen(command)
data=fd.read()
self.log(“Result: %s”%data,”red”)
return
def log(self,message,color,enter=”\\n”):
“””
logs a message to the log window and scrolls the window to the bottom
“””
message=message+enter
buffer = self.logwindow
iter = buffer.get_end_iter()
#gtk versioning avoidance
if color != “black”:
tag = buffer.create_tag()
tag.set_property(“foreground”, color)
self.logwindow.insert_with_tags(buffer.get_end_iter(), message, tag)
else:
self.logwindow.insert(iter, message)
#gtk.FALSE and gtk.TRUE on older pyGTK
mark = buffer.create_mark(“end”, buffer.get_end_iter(), False)
self.logwindowview.scroll_to_mark(mark,0.05,True,0.0,1.0)
#print “Exited log function”
591

64.3 Page 633

▲back to top
Appendix A
return
def quit(self,obj):
import sys
gtk.main_quit()
sys.exit(1)
if __name__ == ‘__main__’:
#do splashscreen here maybe
thegui=ex1gui()
try:
gtk.threads_init()
except:
print “No threading was enabled when you compiled pyGTK!”
sys.exit(1)
gtk.threads_enter()
gtk.main ()
gtk.threads_leave()
If you click your buttons and nothing happens, you probably forgot to add the signals in the widget
properties window in Glade. Or perhaps you forgot to save the glade file after you did so. If you did
everything correctly, you should see the working GUI. Notice how easy it was to change colors in the
text buffer.
Exercise 2 solution
Note that gui_queue.py requires timeoutsocket.py (see Appendix B). Therefore, if you haven’t
downloaded that already, now’s your chance. It is hoped that you’ll end up with a result similar to what
is shown in Figure A-2.
Figure A-2
The working code follows. There are only minor differences between this code and the code for exercise 1.
We’ve added a thread that handles the command running. We’ve also added the code necessary to inter-
act with gui_queue. Other than that, the code should be very familiar by now. Both of these exercises
were completely written in less than an hour — the conversion from nonthreaded to threaded should take
you only around 15 minutes. This is the power of pyGTK. Of course, these are trivial examples. You’ll
come to truly appreciate pyGTK when you scale your applications up to the size of any major application.
Time not spent laying out your GUI manually is time spent making that one killer feature!
592

64.4 Page 634

▲back to top
Answers to Exercises
#!/usr/bin/env python
import findgtk
import gtk
import gtk.glade
import gui_queue
import threading
import os
import gobject
class commandRunner(threading.Thread):
def __init__(self,command,gui):
threading.Thread.__init__(self)
self.command=command
self.gui=gui
return
def run(self):
fd=os.popen(self.command)
data=”A”
while data!=””:
data=fd.readline()
self.gui.gui_queue_append(“log”,[data,”red”])
return
class ex2gui:
def __init__(self):
#we use the same GUI as the other program!
self.wTree = gtk.glade.XML (“ex1.glade”, “window1”)
dic={ “on_window1_destroy” : self.quit,
“on_button1_clicked” : self.send,
}
self.wTree.signal_autoconnect (dic)
#setup the log window
self.logwindowview=self.wTree.get_widget(“textview1”)
self.logwindow=gtk.TextBuffer(None)
self.logwindowview.set_buffer(self.logwindow)
self.gui_queue=gui_queue.gui_queue(self)
gobject.io_add_watch(self.gui_queue.listensocket,gobject.IO_IN,self.clearqueue)
return
def send(self,obj):
print “send called”
command=self.wTree.get_widget(“entry1”).get_text()
print “Command: %s”%command
self.log(“Running: “+command,”black”)
cR=commandRunner(command,self)
cR.start() #start a new thread!
return
def log(self,message,color,enter=”\\n”):
“””
logs a message to the log window and scrolls the window to the bottom
“””
message=message+enter
593

64.5 Page 635

▲back to top
Appendix A
buffer = self.logwindow
iter = buffer.get_end_iter()
#gtk versioning avoidance
if color != “black”:
tag = buffer.create_tag()
tag.set_property(“foreground”, color)
self.logwindow.insert_with_tags(buffer.get_end_iter(), message, tag)
else:
self.logwindow.insert(iter, message)
#gtk.FALSE and gtk.TRUE on older pyGTK
mark = buffer.create_mark(“end”, buffer.get_end_iter(), False)
self.logwindowview.scroll_to_mark(mark,0.05,True,0.0,1.0)
#print “Exited log function”
return
def quit(self,obj):
import sys
gtk.main_quit()
sys.exit(1)
def clearqueue(self,source,condition):
“””Our callback for gui events”””
self.gui_queue.clearqueue(source, condition)
return 1
def handle_gui_queue(self,command, args):
“””
Callback the gui_queue uses whenever it receives a command for us.
command is a string
args is a list of arguments for the command
“””
gtk.threads_enter()
#print “handle_gui_queue”
if command==”log”:
text=args[0]
color=args[1]
self.log(text,color=color)
else:
print “Did not recognize action to take %s: %s”%(command,args)
#print “Done handling gui queue”
gtk.threads_leave()
return 1
def gui_queue_append(self,command,args):
self.gui_queue.append(command,args)
return 1
if __name__ == ‘__main__’:
#do splashscreen here maybe
thegui=ex2gui()
try:
594

64.6 Page 636

▲back to top
Answers to Exercises
gtk.threads_init()
except:
print “No threading was enabled when you compiled pyGTK!”
sys.exit(1)
gtk.threads_enter()
gtk.main ()
gtk.threads_leave()
Chapter 14
Exercise 1 solution
The choice is c, of course. Just joking. The most appropriate choice is b, with the keys being the person’s
name and the values holding the pizza ingredients, perhaps using commas to separate the different
ingredients.
Exercise 2 solution
You can use any alias you like. Here is one example:
select e.firstname, e.lastname, d.name
from employee e, department d
where e.dept = d.departmentid
order by e.lastname desc
Or, in a full example script:
import gadfly.dbapi20
connection = gadfly.dbapi20.connect(‘pydb’, ‘db’)
cursor = connection.cursor()
cursor.execute(“””
select e.firstname, e.lastname, d.name
from employee e, department d
where e.dept = d.departmentid
order by e.lastname desc
“””)
for row in cursor.fetchall():
print row
cursor.close()
connection.close()
Exercise 3 solution
You don’t have to change much. The changes are in bold:
import sys
import gadfly.dbapi20
595

64.7 Page 637

▲back to top
Appendix A
connection = gadfly.dbapi20.connect(‘pydb’, ‘db’)
cursor = connection.cursor()
employee = sys.argv[1]
# Query to find the employee ID.
query = “””
select e.empid
from user u, employee e
where username=? and u.employeeid = e.empid
“””
cursor.execute(query,(employee,));
for row in cursor.fetchone():
if (row != None):
empid = row
# Now, modify the employee.
cursor.execute(“delete from employee where empid=?”, (empid,))
cursor.execute(“delete from user where employeeid=?”, (empid,))
connection.commit()
cursor.close()
connection.close()
Chapter 15
Exercise 1 solution
from xml.dom.minidom import parse
import xml.dom.minidom
# open an XML file and parse it into a DOM
myDoc = parse(‘config.xml’)
myConfig = myDoc.getElementsByTagName(“config”)[0]
#Get utility directory
myConfig.getElementsByTagName(“utilitydirectory”)[0].childNodes[0].data
#Get utility
myConfig.getElementsByTagName(“utility”)[0].childNodes[0].data
#get mode
myConfig.getElementsByTagName(“mode”)[0].childNodes[0].data
#.....Do something with data.....
Exercise 2 solution
#!/usr/bin/python
from xml.parsers.xmlproc import xmlval
class docErrorHandler(xmlval.ErrorHandler):
596

64.8 Page 638

▲back to top
def warning(self, message):
print message
def error(self, message):
print message
def fatal(self, message):
print message
parser=xmlval.XMLValidator()
parser.set_error_handler(docErrorHandler(parser))
parser.parse_resource(“configfile.xml”)
Exercise 3 solution
#!/usr/bin/python
from xml.sax
import make_parser
from xml.sax.handler import ContentHandler
#begin configHandler
class configHandler(ContentHandler):
inUtildir = False
utildir = ‘’
inUtil = False
util = ‘’
inMode = False
mode = ‘’
def startElement(self, name, attributes):
if name == “utilitydirectory”:
self.inUtildir = True
elif name == “utility”:
self.inUtil = True
elif name == “mode”:
self.inMode = True
def endElement(self, name):
if name == “utilitydirectory”:
self.inTitle = False
elif name == “utility”:
self.inUtil = False
elif name == “mode”:
self.inMode = False
def characters(self, content):
if self.inUtildir:
utildir = utildir + content
elif self.inUtil:
util = util + content
elif self.inMode:
mode = mode + content
Answers to Exercises
597

64.9 Page 639

▲back to top
Appendix A
#end configHandler
parser = make_parser()
parser.setContentHandler(configHandler())
parser.parse(“configfile.xml”)
#....Do stuff with config information here
Chapter 16
Exercise 1 solution
RFC 2822 is a file format standard that describes what e-mail messages should look like.
MIME is a file format standard that describes how to create e-mail messages that contain binary data
and multiple parts, while still conforming to RFC 2822.
SMTP is a protocol used to deliver an e-mail message to someone else.
POP is a protocol used to pick up your e-mail from your mail server.
IMAP is a newer protocol that does the same job as POP. It’s intended to keep the e-mail on the server
permanently, instead of just keeping it until you pick it up.
Exercise 2 solution
Here’s a script that uses POP:
#!/usr/bin/python
from poplib import POP3
from email.Parser import Parser
#Connect to the server and parse the response to see how many messages there
#are, as in this chapter’s previous POP example.
server = POP3(“pop.example.com”)
server.user(“[user]”)
response = server.pass_(“[password]”)
numMessages = response[response.rfind(‘, ‘)+2:]
numMessages = int(numMessages[:numMessages.find(‘ ‘)])
#Parse each email and put it in a file named after the From: header of
#the mail.
parser = Parser()
openFiles = {}
for messageNum in range(1, numMessages+1):
messageString = ‘\\n’.join(server.retr(messageNum)[1])
message = email.parsestr(messageString, True)
fromHeader = message[‘From’]
mailFile = openFiles.get(fromHeader)
if not mailFile:
mailFile = open(fromHeader, ‘w’)
598

64.10 Page 640

▲back to top
Answers to Exercises
openFiles[fromHeader] = mailFile
mailFile.write(messageString)
mailFile.write(‘\\n’)
#Close all the files to which we wrote mail.
for openFile in openFiles.values():
openFile.close()
Because IMAP enables you to sort messages into folders on the server, an IMAP version of this script can
simply create new mailboxes and move messages into them. Here’s a script that does just that:
#!/usr/bin/python
from imaplib import IMAP4
import email
import re
#Used to parse the IMAP responses.
FROM_HEADER = ‘From: ‘
IMAP_UID = re.compile(‘UID ([0-9]+)’)
#Connect to the server.
server = IMAP4(‘imap.example.com’)
server.login(‘[username]’, ‘[password]’)
server.select(‘Inbox’)
#Get the unique IDs for every message.
uids = server.uid(‘SEARCH’, ‘ALL’)[1][0].split(‘ ‘)
uidString = ‘,’.join(uids)
#Get the From: header for each message
headers = server.uid(‘FETCH’, ‘%s’ % uidString,
‘(BODY[HEADER.FIELDS (FROM)])’)
for header in headers[1]:
if len(header) > 1:
uid, header = header
#Parse the IMAP response into a real UID and the value of the
#’From’ header.
match = IMAP_UID.search(uid)
uid = match.groups(1)[0]
fromHeader = header[len(FROM_HEADER):].strip()
#Create the mailbox corresponding to the person who sent this
#message. If it already exists the server will throw an error,
#but we’ll just ignore it.
server.create(fromHeader)
#Copy this message into the mailbox.
server.uid(‘COPY’, uid, fromHeader)
#Delete the messages from the inbox now that they’ve been filed.
server.uid(‘STORE’, uidString, ‘+FLAGS.SILENT’, ‘(\\\\Deleted)’)
server.expunge()
599

65 Pages 641-650

▲back to top

65.1 Page 641

▲back to top
Appendix A
Exercise 3 solution
In general, move as much text as possible out of the protocol and into the client software, which needs to
be downloaded only once. Some specific suggestions:
❑ Send short status codes instead of English sentences: for instance, send “HELLO” instead of
“Hello [nickname], welcome to the Python Chat Server!”.
❑ Assign a number to every user in the chat room, and send the number instead of their nickname
whenever they do something — for instance, broadcast ‘4 Hello’ instead of ‘<user> Hello’
whenever a user sends a message.
❑ Use a compression technique to make the chat text itself take up less bandwidth.
Exercise 4 solution
The easiest way is to simply define a method ‘msgCommand’ and let the _parseCommand dispatch it.
Here’s a simple implementation of msgCommand:
def msgCommand(self, nicknameAndMsg):
“Send a private message to another user.”
if not ‘ ‘ in nicknameAndMsg:
raise ClientError(‘No message specified.’)
nickname, msg = nicknameAndMsg.split(‘ ‘, 1)
if nickname == self.nickname:
raise ClientError(‘What, send a private message to yourself?’)
user = self.server.users.get(nickname)
if not user:
raise ClientError(‘No such user: %s’ % nickname)
msg = ‘[Private from %s] %s’ % (self.nickname, msg)
user.write(self._ensureNewline(msg))
Exercise 5 solution
❑ The peer-to-peer architecture is more general than the client-server architecture. The peer-to-
peer design of TCP/IP makes it a flexible general-purpose protocol. It’s easier to implement a
client-server protocol atop TCP/IP than it is to implement a peer-to-peer design on top of a
client-server protocol. If you want a general-purpose protocol, try to preserve the peer-to-peer
nature of TCP/IP.
❑ Consider using peer-to-peer when it makes sense for a client to download some data from a
server and then immediately start serving it to other clients. A peer-to-peer architecture for the
distribution of e-mail doesn’t make sense, because most e-mail is addressed to one person only.
Once that person has downloaded the e-mail, it shouldn’t be automatically distributed further.
A peer-to-peer architecture for the distribution of newsletters makes more sense.
❑ Peer-to-peer is most useful when you have some way of searching the network. When a net-
work resource doesn’t have a single, unambiguous location (the way a file hosted on a web
server does), it’s more difficult to find what you want, and search facilities are more important.
600

65.2 Page 642

▲back to top
Answers to Exercises
Chapter 19
Exercise 1 solution
def format_bytes(bytes):
units = (
(“GB”, 1024 ** 3),
(“MB”, 1024 ** 2),
(“KB”, 1024 ** 1),
(“bytes”, 1),
)
terms = []
for name, scale in units:
if scale > bytes:
continue
# Show how many of this unit.
count = bytes // scale
terms.append(“%d %s” % (count, name))
# Compute the leftover bytes.
bytes = bytes % scale
# Construct the full output from the terms.
return “ + “.join(terms)
Exercise 2 solution
def rgb_to_html(red, green, blue):
# Convert floats between zero and one to ints between 0 and 255.
red = int(round(red * 255))
green = int(round(green * 255))
blue = int(round(blue * 255))
# Write out HTML color syntax.
return “#%02x%02x%02x” % (red, green, blue)
Exercise 3 solution
Solution using a list of numbers:
from math import sqrt
def normalize(numbers):
# Compute the sum of squares of the numbers.
sum_of_squares = 0
for number in numbers:
sum_of_squares += number * number
# Copy the list of numbers.
result = list(numbers)
# Scale each element in the list.
scale = 1 / sqrt(sum_of_squares)
for i in xrange(len(result)):
result[i] *= scale
return result
601

65.3 Page 643

▲back to top
Appendix A
This very concise numarray version works only when called with a numarray.array object. You can
convert a different array type with numbers = numarray.array(numbers):
from math import sqrt
import numarray
def normalize(numbers):
return numbers / sqrt(numarray.sum(numbers * numbers))
Chapter 21
Exercise 1 solution
Put the name of the wiki in the resource identifier, before the page name: Instead of “/PageName”, it
would be “/Wikiname/PageName”. This is RESTful because it puts data in the resource identifier, keep-
ing it transparent. Not surprising, this identifier scheme also corresponds to the way the wiki files would
be stored on disk.
Exercise 2 solution
#!/usr/bin/python
import cgi
import cgitb
import os
from WishListBargainFinder import BargainFinder, getWishList
cgitb.enable()
SUBSCRIPTION_ID = ‘[Insert your subscription ID here.]’
SUBSCRIPTION_ID = ‘D8O1OTR10IMN7’
form = cgi.FieldStorage()
wishListID = form.getfirst(‘wishlist’, ‘’)
args = {‘title’ : ‘Amazon Wish List Bargain Finder’,
‘action’ : os.environ[‘SCRIPT_NAME’],
‘wishListID’ : wishListID}
print ‘Content-type: text/html\\n’
print ‘’’<html><head><title>%(title)s</title></head>
<form method=”get” action=”%(action)s”>
<h1>%(title)s</h1>
Enter an Amazon wish list ID:
<input name=”wishlist” length=”13” maxlength=”13” value=”%(wishListID)s” />
<input type=”submit” value=”Find bargains”/>
</form>’’’ % args
if wishListID:
print ‘<pre>’
BargainFinder().printBargains(getWishList(SUBSCRIPTION_ID, wishListID))
print ‘</pre>’
print ‘</body></html>’
602

65.4 Page 644

▲back to top
Answers to Exercises
Note that this points to an improvement in BargainFinder: creating a method that returns the bargain
information in a data structure, which can be formatted in plaintext, HTML, or any other way, instead of
just printing the plaintext of the bargains.
Exercise 3 solution
For REST: The BittyWiki web application already outputs rendered HTML because that’s what web
browsers know how to parse. However, a BittyWiki page served by the web application includes naviga-
tion links and other elements besides just a rendering of the page text. If web service users aren’t happy
scraping away that extraneous HTML to get to the actual page text, or if you want to save bandwidth by
not sending that HTML in the first place, there are two other solutions. The first is to have web service
clients provide the HTTP Accept header in GET requests to convey whether they want the “text/plain”
or “text/html” flavor of the resource. The second is to provide different flavors of the same document
through different resources. For instance, /bittywiki-rest.py/PageName.txt could provide the
plaintext version of a page, and /bittywiki-rest.py/PageName.html could provide the rendered
HTML version of the same page.
For XML-RPC and SOAP, the decision is simpler. Just have clients pass in an argument to getPage spec-
ifying which flavor of a page they want.
Exercise 4 solution
This could be fixed by changing the GET resource or getPage API call to return not only the raw text of
the page, but a representation of which WikiWords on the page correspond to existing pages. This could
be a list of WikiWords that have associated pages, or a dictionary that maps all of the page’s referenced
WikiWords to True (if the word has an associated page) or False (if not). The advantage of the second
solution is that it could save the robot side from having to keep its own definition of what constitutes a
WikiWord.
Exercise 5 solution
Create a new API call specifically for renaming a page. In XML-RPC or SOAP, this would be as simple as
creating a rename function and removing the delete function. For a REST API, you might add a capa-
bility to the POST request that creates a new wiki page: Instead of providing the data, let it name another
page of the wiki to use as the data source, with the understanding that the other page will be deleted
afterward.
Chapter 22
Exercise 1 solution
Many organizations have an investment in another programming language. Jython, though, enables you
to use Python in a Java environment.
Exercise 2 solution
Jython is written in Java. The python interpreter is written in C.
603

65.5 Page 645

▲back to top
Appendix A
Exercise 3 solution
You need to include your Jython scripts, of course, but also the following:
❑ The jython.jar Java library
❑ The Jython Lib directory
❑ The Jython cachedir directory. This directory must be writeable.
Exercise 4 solution
No, unless the DB drivers are written in Python or Java. Most Python DB drivers are written in C and
Python, and so cannot run from Jython (without a lot of work with the Java Native Interface, or JNI).
Luckily, the Jython zxJDBC module enables you to call on any JDBC driver from your Jython scripts.
This opens up your options to allow you to access more databases than those for which you can get
Python DB drivers.
Exercise 5 solution
This is probably the simplest way to create such a window:
from javax.swing import JFrame
frame = JFrame(size=(500,100))
# Use a tuple for RGB color values.
frame.background = 255,0,0
frame.setVisible(1)
You can get fancy and add widgets such as buttons and labels, if desired.
604

65.6 Page 646

▲back to top
B
Online Resources
Python is software available from the Internet, and Python’s best day-to-day resources can all be
found there. This appendix describes the software that is used in this book and how to install it.
Most Python-related software can be downloaded for free, and much of it can be downloaded as
source code and compiled — for those of you interested in doing that for yourself. For those read-
ers who begin with the second part of the book, this may be the challenge you’re looking for.
However, the broader audience for this book will be glad to know that everything you need to
follow along with the book’s examples can be installed as packages for the operating systems on
which they are supported.
Software
The examples in this book require that your computer have additional software installed, as well
as an appropriate and functioning operating system such as Windows 2000, XP, XP Pro, or 2003;
Linux (Red Hat’s Fedora RC3 or newer; Debian testing or unstable; or a similarly current distribu-
tion); or Mac OS X 10.2 or newer.
Following is a brief list of the required software, with a description and the URL from which the
software can be downloaded:
❑ Python (www.python.org/) is the home page for the Python language. You can find out
about all things Python there, including additional online tutorials, introductions to the
language, and mailing lists to help you out. The people who write, maintain, change, and
use Python are there. You can find a complete, if terse, set of documentation available
there as well. The version of software used in this book is Python 2.4, and to download it
you can click the Download link at the top of the Python home page, or go directly to
www.python.org/download/. If you’re lucky, maybe you’ll find a more recent version of
Python there that you can use! At the time of publication, Python 2.4.1 has been released.
For Windows, use the Windows .msi installer of the most recent Python 2.4 installations.

65.7 Page 647

▲back to top
Appendix B
For Linux systems, install the package provided for your distribution by the maintainer of the dis-
tribution (for example, the .deb packages from debian.org or the .rpm packages from redhat.com,
such as the information at www.python.org/2.4/). For other Linux distributions, see the home
page for this book for comments from other readers that the authors will be compiling.
It’s important to understand that Apple has incorporated Python 2.3 into Mac OS X 10.3, so
only Python 2.3 can be used there. The authors expect that this situation will change in the
future. More information can be found at http://homepages.cwi.nl/~jack/macpython/
macpython-osx.html.
❑ PythonCard: PythonCard is a lot more than just an editor, and when you are just beginning to
start programming, one of the hardest things to do is get used to the language — how it looks,
behaves, and works. Having a programmers’ editor that can take a lot of the fussy work out
of your hands makes everything a lot faster and easier, and PythonCard comes with an editor
called codeEditor that is very useful for a beginner, in addition to being very useful for more
advanced users who can take advantage of how easy it makes the little tasks, such as running
a program through a test cycle.
PythonCard can be found at http://pythoncard.sourceforge.net/, and it requires a pack-
age called wxWidgets on which it builds. It can be found at www.wxwidgets.org/. The version
of PythonCard used when writing this book was version 0.8.1, although the developers are
rapidly heading toward version 1.0.
For Windows 2000 and later, the installation instructions for PythonCard are described at
http://pythoncard.sourceforge.net/windows_installation.html; however, that
installation will work with the 2.4 version of Python (at the time of writing, the version of
Python mentioned on the web site is 2.3.4).
For Linux distributions, you should install the PythonCard distribution that is appropriate
to your distribution and provided with the distribution by following the instructions at
http://pythoncard.sourceforge.net/linux_installation.html. The installation
instructions for Mandrake Linux should also work for Red Hat’s Fedora Core distribution.
For Mac OS X, the installation instructions for Mac OS X 10.2 (Jaguar) are at http://
pythoncard.sourceforge.net/macosx_jaguar_installation.html; and for 10.3
(Panther), the slightly different instructions are at http://pythoncard.sourceforge.net/
macosx_installation.html.
The instructions for installing wxWidgets are described on the respective PythonCard installa-
tion pages.
❑ Pygtk and pyglade: The GUI programming chapter in this book is written using the pygtk
interface, which gives you access to the GTK+ graphical user interface toolkit from within
Python, along with pyglade, which offers a simple mechanism to design the visual interface
with pygtk.
Pygtk is cross-platform but was originally built for the X Window system, and now is available
natively for Windows and Linux as well as being available on Mac OS X if the X Window
libraries are installed.
The main web site for pygtk is www.pygtk.org/.
For a FAQ, which is also a rapidly updated wiki, for pygtk, and which also answers questions
about pyglade, check out www.async.com.br/faq/pygtk/index.py?req=index.
606

65.8 Page 648

▲back to top
Online Resources
❑ PyUnit: The unit testing framework for Python. This module provides a systematic way of
writing tests within your own source code so that you can verify that your code works as you
expect.
PyUnit’s home page is at http://pyunit.sourceforge.net/.
❑ MySQL: A popular and fast open-source relational database system. Python has robust MySQL
support:
www.mysql.com/ — This is the home page for mysql.com, the company that maintains the
mysql database.
http://sourceforge.net/projects/mysql-python — This is the home page of the mysql-
python module, but there is a minimum amount of documentation online.
❑ Wftk and PyWFTK: The Workflow toolkit. This toolkit allows for the construction of workflows
and the enforcement of those workflows.
www.vivtek.com/wftk/ — The wftk home page.
http://sourceforge.net/projects/wftk/ — The wftk download page.
www.vivtek.com/wftk/doc/code/python/ — The home page for the PyWFTK module.
❑ Jython: An implementation of the python language in pure Java, Jython provides access to all of
the tools available in the commercial Java product space, but it enables you to program using
Python as your language. Visit www.jython.org/.
For More Information
You can find a lot of Python-related information on the Internet. In addition, you can find information
related to the specific components that appear in this book. As a result of the constantly changing nature
of Python and its modules, please look at this book’s web page at www.wrox.com, and follow the instruc-
tions in the introduction to find the specific page for this book. That’s the place to go for help with
installing software, to download samples and provide feedback to the authors, and to receive help with
anything in the book. In addition, more packages and information about the ones that have been men-
tioned can be found online at the web site for this book.
607

65.9 Page 649

▲back to top

65.10 Page 650

▲back to top
C
What’s New in Python 2.4
Python is constantly changing in little ways. Python 2.4 has evolved from version 2.3, but it con-
tains important changes. In this appendix, you’ll be introduced to the changes relevant to the top-
ics covered in this book. This means that this is not an exhaustive treatment by any means but only
a selection of topics touched on in the book — topics that you may want to know as someone new
to Python.
The official list of changes to Python 2.4 can be found at www.python.org/doc/2.4/whatsnew/
whatsnew24.html. If a newer version of Python is available by the time you read this, the list of
changes for that version can be found on the Python web site as well.
Negative Numbers
Numbers that are negative in Python prior to 2.4 would be represented differently when used in
the wrong context. In the past, when unsigned numbers (numbers that are only supposed to be
positive) were manipulated in such a way that they would otherwise have become negative, the
language would instead change the number into a very large positive number. This “wrap”
around to a high positive number is what many programmers expect to happen; however, as of
Python 2.4, the number will silently be transformed into a signed number and be a normal nega-
tive number.
Automatic Promotion of Large Numbers
Prior to 2.4, some operations wouldn’t automatically convert an integer to a long when the integer
was manipulated to become much larger. In Python 2.4, all operations on an integer that would
have resulted in a overflow (a number larger than what the type can hold) will now always turn
the integer into a long.

66 Pages 651-660

▲back to top

66.1 Page 651

▲back to top
Appendix C
A Different String Substitution
As you learned in Chapter 9, there is a new format for creating Templates, which is a method for perform-
ing substitutions from elements within a dictionary into a string based on the key and value pairs that
exist in the dictionary.
The syntax that was added was that when the string.Template or string.SafeTemplate classes
are invoked with a format such as “$first $second $third”, a template object is returned that, when
templateObject.substitute({‘first’: ‘Yes’, ‘second’:’No’, ‘third’:’Maybe’}) is
invoked, the parts of the template that began with the dollar sign are replaced by the values in the dic-
tionary whose key matches the template of the same name — just that the dollar sign is ignored when
looking things up in the dictionary.
Looking at Sequences in Reverse
If you have a long sequence that you want to access in reverse, Python 2.4 has a built-in function called
reverse.
In prior versions of Python, a loop or similar mechanism would have to be explicitly written to proceed
from the end of a sequence to the beginning. Because it works on a special class of sequences called
iterators, reverse works with xrange objects as well as lists and tuples.
reverse is the fastest way of going through something like a large sequence from back to front.
A More Pythonish Subprocess Creation
Chapter 9 looked at working with subprocesses at the same time that threads were first looked at. There
are a few other system-specific ways of creating subprocesses. In version 2.4 a new module called
subprocess was added that has just one class — Popen. Popen provides an interface that is more useful
to experienced programmers because of the flexibility that it offers. In addition, for more experienced
programmers, it offers a way to do things that are often difficult to get right, such as sending data to a
child process or checking to see whether the process is done.
In addition, a shortcut that is considered “safer” than os.system has been provided. Called
subprocess.call, it acts almost the same but eliminates the need to worry about the input you are
sending to the subprocess. This is because os.system invokes the command shell, which can read text
that you pass to it and be induced into misbehaving. Using subprocess.call, this issue can be avoided.
610

66.2 Page 652

▲back to top
What’s New in Python 2.4
Making Import a Bit Easier
Normally, Python is very fussy about where you can insert new lines. One such place in which it has
been fussy in the past is when importing specific names from a module. Normally, importing a long list
of functions or other names from a module would involve using a backslash character at the end of each
line, such as the following:
from SomeModule import this, that, theother, yetanother, andonemore, \\
andonemoreforgoodluck
This makes the import somewhat readable, but the backslash at the end of each line can become difficult
to read.
As an alternative, Python 2.4 allows you to enclose all of the names that you want to import inside of
parentheses, enabling you to do away with the need for the backslash. For example, the preceding code
would become the slightly easier to work with form:
from SomeModule import (this, that, theother, yetanother, andonemore,
andonemoreforgoodluck)
One significant advantage is that this makes it easier for programming editors to automatically keep
track of how a list of names is formatted. With backslashes, this can become messy.
None Really Is None
In the past, it was possible to assign a new value to the special name None. As of Python 2.4, this will
raise an exception, so this mistake can’t be made.
611

66.3 Page 653

▲back to top

66.4 Page 654

▲back to top
Glossary
The following terms are used in the book and are presented here for your convenience.
127.0.0.1 A special IP address used to denote “this computer.” Also see “localhost.”
Anonymous Anonymous functions and variables are not bound to names. Examples of this are the
functions created by the lambda function, a list or tuple created but never associated with a name.
Base64 An encoding strategy defined by MIME that escapes an entire string as a whole. More
efficient than Quoted-printable for binary data.
BitTorrent A peer-to-peer protocol that distributes the cost of hosting a file among all the parties
downloading it.
Call Stack When code is executing, the call stack is the list of functions that your code has exe-
cuted to reach that point in the program. As functions or methods are entered, the location in the
file is noted along with the parameters that the function was called with, and the entry point is
marked in the call stack. When a function is exited, its entry in the call stack is removed. When an
exception occurs, a stack trace is printed that indicates where in the program the problem occurred
by printing out the state of the call stack.
CGI The Common Gateway Interface: A standard for web servers that makes it easy to expose
web interfaces to scripts.
Class A class is a definition that can be used to create objects. A particular class definition con-
tains the declarations of the data and the methods that objects that are instances of that particular
class will have available to them. In Python, functions that appear within the context of a class are
considered to be methods.
Class An object holds data as well as the methods that operate on that data. A class defines what
data are stored and what methods are available. Python is a little looser than most programming lan-
guages, such as Java, C++, or C#, in that Python lets you break rules enforced in other languages. For
example, Python, by default, lets you access data inside a class. This does violate some of the concepts
of object-oriented programming but with good reason: Python aims first and foremost to be practical.

66.5 Page 655

▲back to top
Glossary
Client-server Describes an architecture in which one actor (the server) is a repository for information
requested and acted upon by other actors (the clients). fault: A term used in web services to denote an
error condition. Similar to Python’s exceptions, and generally implemented as exceptions in Python.
goals of REST.
Comment Comments are text in a program that python does not pay attention to. At any point outside
of a string where a hash mark (#) appears, from that point until the end of the, the Python interpreter
ignores all text.
Content type A MIME concept used to indicate the type of a file being sent encoded inside an email
message. Also used by web servers to indicate the type of file being served.
DB API A Python API for accessing databases. The neat thing about this API is that you can use the
same Python code to work with any database for which there is a DB-compliant driver. This includes
Oracle, DB2, and so on. The only differences in your code will likely be the code to connect to the
database, which differs by vendor.
DBM Short for database manager, DBM libraries provide a means to persist Python dictionaries.
Dictionary A data type in python that is indexed by an arbitrary value that is set by the programmer.
The value can be any kind of python object. The index is called the “key” and the object that a key refer-
ences is referred to as it’s “value.”
DNS Domain Name System. A service that runs on top of TCP and resolves hostnames (wrox.com) to
IP addresses (208.215.179.178).
Document Model A way of describing the vocabulary and structure of a document. Defines the data
elements that will be present in a document, what relationship they have to one another, and how many
of them are expected.
DOM The Document Object Model, a tree-based API recommendation from the W3C for working with
XML documents.
DTD Document Type Definition. A specification for producing a Document Model.
Dynamic port See ephemeral port.
Encapsulation Encapsulation is the idea that a class can hide the internal details and data necessary
to perform a certain task. A class holds the necessary data, and you are not supposed to see that data
under normal circumstances. Furthermore, a class provides a number of methods to operate on that
data. These methods can hide the internal details, such as network protocols, disk access, and so on.
Encapsulation is a technique to simplify your programs. At each step in creating your program, you can
write code that concentrates on a single task. Encapsulation hides the complexity.
Encryption The act of hiding information so that it is difficult or impossible to recover without a secret
password. Data is encrypted when it is recoverable. Data which is scrambled and unrecoverable should
be thought of as lost, instead.
Ephemeral port High-numbered IP ports are often created to receive data over TCP/IP as part of a
particular socket connection. Ephemeral ports are administered by the operating system, and have a life-
time of a single socket connection.
614

66.6 Page 656

▲back to top
Glossary
Escape sequences Special characters that begin with the backslash such as \\n for a new line.
Float A floating point number is a number with a fractional or decimal component. Fractions can be
represented as decimal values using a float value. When arithmetic is done with a float and an integer,
the integer will be promoted to being a float.
Function A function is a collection of code defined by using a name. It is invoked by using that name
followed by open and close braces, such as function(). A function can be given parameters that are
enclosed within the braces and separated by commas, such as function(parameter1, parameter2),
that it can use to modify its output.
Header An item of metadata found both in email messages and in HTTP requests and responses. A
header line consists of a key and value separated by a colon and a space. For instance: “Subject: Hello.”
Hexadecimal Base 16 notation, where the numbers are from 0 through 15, and are represented by the
numbers 0-9, and once single digits are exhausted the letters A-F are used. So the number 11 in hex is b.
hostname A human-readable identifier for a computer on an IP network, for instance: wrox.com.
Hostnames are administered through DNS.
HTTP body The data portion of an HTTP request or response.
HTTP headers The metadata portion of an HTTP request or response: a series of key-value pairs.
HTTP defines some standard headers, and CGI defines some more: Applications can define their own.
HTTP HyperText Transfer Protocol, the protocol devised to let web browsers and web servers
communicate.
HTTP request The string sent by an HTTP client to the server, requesting some operation on some
resource.
HTTP response The string sent by an HTTP server to a client, in response to an HTTP request. In REST
terminology, it contains either a representation of a resource or a document describing action taken on a
resource.
HTTP status code A numeric code used in an HTTP response to denote the status of the corresponding
request. There are 40 of these, defined in the HTTP standard.
HTTP verb A string used in an HTTP request to describe what the client wants to do to a resource (for
instance, retrieve a representation of it or modify it).
Idempotent An idempotent action has no side effects. A term taken from mathematics: Multiplying a
number by 1 is an idempotent action. So should be calling an object’s accessor method or (in REST) mak-
ing an HTTP GET request.
Imaginary number a special number that acts like a float but cannot be mixed freely with floats or
integers. If they are mixed, a complex number is the result, not an imaginary number.
IMAP The Internet Message Access Protocol. Also known as IMAP4. A protocol for retrieving and
managing mail. IMAP4 intends for you to store your mail on the server. q.v. POP.
615

66.7 Page 657

▲back to top
Glossary
Infinite loop a loop that has no termination clause, like “while True:”. Often an infinite loop is an acci-
dental situation, but they can be useful as long as there are actions that will happen, and there is code
being executed. One example is a server waiting for connections.
Inheritance Inheritance means that a class can inherit, or gain access to, data and methods defined in a
parent class. This just follows common sense in classifying a problem domain. For example, a rectangle
and a circle are both shapes. In this case, the base class would be Shape. The Rectangle class would then
inherit from Shape, as would the Circle class. Inheritance allows you to treat objects of both the Rectangle
and Circle classes as Shapes, meaning you can write more generic code. For the most part, the base class
should be general and the subclasses specialized. Oftentimes inheritance is called specialization.
Input/Output an umbrella term that covers any kind of operation that reads or writes data. Writing to
screen, input from the keyboard, and network connections are all examples of Input/Output.
Integer whole numbers, without a fractional or decimal component. In Python, if an integer becomes
larger than its range (from -2,147,483,648 to 2,147,483,647 on 32-bit systems) it is automatically converted
to a long integer. The long integer and the plain integer can be mixed freely. If two integers are used in
division where one can’t be evenly divided into the other, an integer will be the result, not a float. The
remainder is discarded. To obtain just the reminder, use the modulus operation. To obtain a float result,
one of the numbers must be a float for the conversion to happen.
I/O See input/output.
IP address The location of a computer on an IP network. For instance, 208.215.179.178.
IP The Internet Protocol. Connects networks based on different technologies (for instance, Ethernet
and wireless) into a single network.
IRC Internet Relay Chat. A protocol for online chat rooms.
Iterator Iterators are objects that you can use in certain contexts that generate a sequence of outputs.
Unlike sequence objects, an iterator like xrange doesn’t have to return a finite list. The object can con-
tinue to create return values when its next method is invoked. Iterators can be used with for loops.
J2EE Java 2 Enterprise Edition, a set of standards for writing enterprise-worthy Java applications. There
are no real corresponding Python standards, but the Twisted framework and others provide enterprise-
worthy features for Python.
JVM Java Virtual Machine, the runtime engine of the Java platform. The java command runs Java
applications similar to how the python command runs Python applications.
Jython An implementation of Python written in the Java language that runs on top of the Java platform.
List A list is a type of sequence, as well as being an iterator. It is similar to a tuple, except that it can be
modified after it is created. A list is created by using the square brackets “[]”.
localhost A special hostname used to denote “this computer”. Also see “127.0.0.1.”
616

66.8 Page 658

▲back to top
Glossary
Long Integer See Integer.
Loop A loop is a form of repetition where a set of operations is performed, and the operations are
repeated until a set of conditions are set.
Method A method is a function inside the context of an object (it is also called a method when you
write it inside of a class). It has automatic access to all of the data within the object that it was invoked
from.
MIME Multipurpose Internet Mail Encoding. A set of standards that make it possible to send multiple
files and international and binary data through email, while still complying with RFC 2822.
Module A module is a collection of code within a file. Modules can contain functions, named vari-
ables, and classes. When a module is used in a program, it is made available using the import built-in
word, and it lives within a scope named after the module. So in a module named “mymodule” the
function “myfunction” would be called by calling “mymodule.myfunction()”. This can be modified by
how the module is imported with import the modifiers “from”, and “as” can modify the behavior of
import so that the module is seen as having a different name. The current module can be found by look-
ing at the variable name “__name__”, which is created locally in each module’s scope. If __name__ is
“__main__” then the scope is currently the top level module — i.e., the program being run.
Module A module is just a Python source file. A module can contain variables, classes, functions, and
any other element available in your Python scripts.
Multipart message A MIME message that contains more than one “document” (for instance, a text
message and an image).
Object An object is an instance of a class. Objects contain data and methods that are defined in the
class. Multiple objects of the same class may exist in the same program at the same time, using different
names. Each object has data that will be different from other objects of the same type. Objects are bound
to a name when they are created.
Octal Base 8 notation, where the numbers range from 0-7.
Package A package is a grouping of modules in a directory that contains a file called __init__.py.
Together, all the files in the directory can act together to implement a combined package that appears,
when it’s used, to act like a single module. The module can contain subdirectories that can also contain
modules. The package offers an organizational structure for distributing more complex program struc-
tures, and it also allows for the conditional inclusion of code that may only work on one platform (for
instance, if one file could not work except on a Mac OS X system, it could be put into its own file and
called only after the correct platform had been verified).
Peer-to-peer Describes an architecture in which all actors have equal standing.
Polymorphism Polymorphism means that subclasses can override methods for more specialized
behavior. For example, a Rectangle and a Circle are both Shapes. You may define a set of common opera-
tions such as move and draw that should apply to all shapes. But the draw method for a Circle will
obviously be different from the draw method for a Rectangle. Polymorphism allows you to name both
methods draw and then call these methods as if the Circle and the Rectangle were both Shapes (which
they are, at least in this example).
617

66.9 Page 659

▲back to top
Glossary
POP The Post Office Protocol. Also known as POP3. A protocol for downloading email from a server.
POP intends that you delete the mail from the server after downloading it. q.v. IMAP.
Port Along with an IP address, a port number identifies a particular service on an Internet network.
Protocol A convention for structuring the data sent between parties on a network. HTTP and TCP/IP
are examples of protocols.
Protocol stack A suite of protocols in which the higher-level protocols delegate to the lower-level ones.
Quoted-printable An encoding strategy defined by MIME that escapes each non-US ASCII character
individually. More efficient than Base64 for text that contains mostly US ASCII characters.
Quotes In Python, strings are defined by being text within quotes. Quotes can be either single (‘), double
(“), or triple quotes (“”” or ‘’’). If a string is started with a single quote, then it must be ended with single
quotes. A string begun with a double quote must be terminated with a double quote. A string begun with a
triple quote must be terminated with a triple quote of the same kind of quote (‘’’ must be matched by a ‘’’,
and a “”” must be matched by a “””). Single and double quotes function in exactly the same way. Triple
quotes are special because they can enclose multi-line strings (strings that contain newlines).
Range Range generates a list of numbers, by default from zero to the number it is given as a parameter,
by one. It can also be instructed to start at a number other than zero and to increment in steps rather
than by one.
RDBMS Relational Datbase Management System. See Relational Database.
Relational database In a relational database, data are stored in tables, two-dimensional data struc-
tures. Each table is made up of rows, also called records. Each row in turn is made up of columns.
Typically, each record holds the information pertaining to one item, such as an audio CD, a person, a
purchase order, an automobile, and so on.
Representation In REST terminology, a depiction of a resource. When you request a resource, what
you get back is a representation. One resource may have multiple representations. For instance, a single
document resource may have HTML, PostScript, and plain-text representations.
Resource identifier A string that uniquely identifies a resource. Generally equivalent to a URL. One
resource may have multiple identifiers.
Resource In REST terminology, an object that can be accessed and/or manipulated from the web. Can
take a number of forms: For instance, it may be a document located on the server, a row in a database, or
even a physical object (such as an item you order in an online store).
RESTfulness An informal metric of how well a web application conforms to the design.
REST REpresentational State Transfer, a name for the architecture of the World Wide Web.
RFC 2822 The standard format for Internet email messages. Requires that email messages be formatted
in US ASCII.
Robot A script that makes HTTP requests while not under the direct control of a human.
618

66.10 Page 660

▲back to top
Glossary
RSS Rich Site Summary, or RDF Site Summary. An XML-based format for syndicating content.
SAX The Simple API for XML. A stream-based XML parser.
Scope Names of data and code; variable names, class names, function names, etc., have different levels
of visibility. Names that are visible within a function or method are either in their scope or come from a
scope that is at a level above the scope of the operation accessing it.
Sequence A sequence is a category of data types. A sequence can refer to any type of object that con-
tains an ordered numerical index, starting from zero, which contains references to values. Each value
referenced from an index number can be any python object that could normally be referenced by a vari-
able name. Elements in a sequence are dereferenced by using the square brackets after the name of the
sequence. So for a sequence named “seq” the fourth element is dereferenced when you see “seq[3]”. It is
3 instead of 4 because the first index number of a sequence is 0.
SMTP Simple Mail Transport Protocol. The standard protocol for sending Internet email.
SOAP Originally stood for Simple Object Access Protocol. A standard for making web service calls,
similar to XML-RPC but more formally defined.
Socket A two-way connection over an IP network. Sockets allow programmers to treat network con-
nections like files.
Spider Robot that, given a starting web page, follows links to find other web pages to operate on. Most
search engines have implemented spiders.
SQL Structured query language, pronounced either sequel or S-Q-L.Language used to access relational
databases.
SSL Secure Socket Layer. A protocol that runs between TCP/IP and some other protocol (such as
SMTP or HTTP), providing end-to-end encryption.
Stack Trace See Call Stack.
String Any combination of letters or numbers enclosed in quotation marks (either single, double, or
a series of three single or double quotes together). Most objects have a string representation that can be
used to print them. Strings can be treated as arrays when you want to access specific letters, or a range of
letters, within the string. Also see unicode.
TCP/IP A term used to describe a very common protocol stack: TCP running on top of IP.
TCP Transport Control Protocol: Makes reliable, orderly communication possible between two points
on an IP network.
Tuple A tuple is a type of sequence as well as an iterator. A tuple is similar to a list, except that once a
tuple has been defined, the number of elements, and the references to elements in it cannot be changed
(however, if it references an object whose data you can change, such as a list or a dictionary, the data
within that other type can still be changed). Tuples are created with the parenthesis “()”. When you cre-
ate a tuple that has only one element, you must put a comma after that single element. Failing to do this
will create a string.
619

67 Pages 661-670

▲back to top

67.1 Page 661

▲back to top
Glossary
Twisted A framework for Python networking. Where other Python networking frameworks use
threads or subprocesses to handle multiple client requests simultaneously, Twisted does everything in a
single thread.
UID Unique ID. Used in a variety of contexts to denote an ID that is unique and stable over time.
Unicode Unicode is a system for encoding strings so that the original letters can be determined, even if
someone using a different character encoding by default reads that string later. (Think of someone using
a computer localized for Russia trying to read a document written in Hebrew — internally characters can
be thought of as numbers in a lookup table, and with different languages and character sets, character
#100 in either character set is likely to be different.) Unicode strings are made by using the construct
u’string’.
User agent A web browser or HTTP-enabled script.
Variable A variable is what data bound to a name is called. The name “variable” usually refers to the
basic types and not more complex objects. This is true even though integers, longs, floats, imaginary
numbers, and strings all objects in Python. This way of thinking is a convention that carries over from
other languages where the distinction is made — some types are not objects.
Web application A program that exposes its interface through HTTP instead of through a command-
line or GUI interface.
Web service A web application designed for use by HTTP-enabled scripts instead of human beings
with web browsers.
Well-known port IP port numbers between 0 and 1023 are well-known ports. Popular services like
web servers tend to run on well-known ports, and services running on well-known ports often run with
administrator privileges.
Whitespace Whitespace are the names of the characters that you can’t see when you are typing or
reading. Newlines, spaces, and tab characters are all whitespace. Python pays attention to whitespaces at
the beginnings of lines, and it is aware of newlines at the ends of lines, except inside of list or tuple defi-
nitions, and except inside of triple-quoted strings.
wiki A web application that allows its users to create and edit web pages through a web interface.
WSDL Web Services Description Language, a way of representing method calls in XML.
XML eXtensible Markup Language. A specification for creating structured markup languages with
customized vocabularies.
XML-RPC The RPC stands for Remote Procedure Call. XML-RPC is a standard for making web service
calls. It defines a way of representing simple data structures in XML, sending data structures over HTTP
as arguments to a function call, and getting another data structure back as a return value.
XML Schema A specification for producing a Document Model.
620

67.2 Page 662

▲back to top
Glossary
XML Validation The process of checking that an XML document is well formed and conforms to its
document model.
XML Wellformedness The process of checking that an XML document comforms to the XML
specification.
Xrange Xrange generates an xrange object, which is an iterable object that behaves similar to a list,
but because a list is not created there is no additional memory used when larger ranges of numbers are
required.
XSL-FO Extensible Style Language Formatting Objects. Markup language for graphical display.
Commonly used for producing documents for final presentation.
XSLT Extensible Style Language for Transformations. A programming language for transforming XML.
621

67.3 Page 663

▲back to top

67.4 Page 664

▲back to top
Index
SYMBOLS AND
NUMERICS
* (asterisk)
* (element grouping operator in DTD), 279
** (exponentiation operator), 406, 413
* (in format specifier), 410
* (in import command), 103
* (multiplication operator), 19, 412
* (regular expression wildcard), 186–187
* (wildcard in glob pattern), 122
* (XPath axis shortcut), 282
@ (at sign), XPath axis shortcut, 282
\\ (backslash)
\\ (escape character), 7, 185
\\ (in Windows path names), 109, 178
\\n (newline character), 7–8
^ (caret), regular expression wildcard, 187
: (colon)
: (preceding code block), 48–49
: (separating dictionary keys and values), 35
{} (curly braces), enclosing dictionaries, 34
$ (dollar sign), regular expression wildcard, 187
“...” (double quotes), enclosing strings, 6, 618
= (equal sign)
= (assignment operator), 28
== (equality operator), 43–44
! (exclamation point)
!= (inequality operator), 45
! (wildcard in glob pattern), 122
# (hash mark)
# (preceding comments), 65–66
#! (shebang comment line), 60, 545
- (hyphen)
- (in format specifier), 10
- (subtraction operator), 19, 412
< (left angle bracket)
< (less than operator), 45–46
<= (less than or equal operator), 47
() (parentheses)
enclosing list of DTD elements, 279
enclosing tuples, 30
. (period), regular expression wildcard, 185
% (percent sign)
% (escaping in strings), 18–19
% (preceding format specifier), 9
% (remainder operator), 20, 414
% (string formatting operator), 408
%d (format specifier), 18, 408
%E (format specifier), 18
%f (format specifier), 18, 22, 408
%o (format specifier), 24, 409
%#o (format specifier), 409
%s (format specifier), 9
%x (format specifier), 24, 409
%X (format specifier), 24
%#x (format specifier), 409
+ (plus sign)
+ (addition operator), 19, 412
+ (combining strings), 8
+ (element grouping operator in DTD), 279
? (question mark)
? (element grouping operator in DTD), 279
? (wildcard in glob pattern), 122
> (right angle bracket)
> (append in print statement), 111
> (greater than operator), 45–46
>= (greater than or equal operator), 47
>> (prompt in Python shell), 5
‘ (single quote), element grouping operator in
DTD, 279
‘...’ (single quotes), enclosing strings, 6, 618
/ (slash)
/ (division operator), 19, 412
// (floor division operator), 413
/ (in Unix path names), 178
// (XPath axis shortcut), 282

67.5 Page 665

▲back to top
[] (square brackets)
[] (square brackets)
dereferencing dictionaries, 35
dereferencing lists, 33
dereferencing tuples, 31
enclosing lists, 33
regular expression wildcard, 187
wildcard in glob pattern, 122
‘’’...’’’ or “””...””” (triple quotes)
enclosing documentation, 156–157
enclosing multi-line strings, 7–8, 618
_ (underscore), in variable names, 30
| (vertical bar)
| (OR between DTD elements), 279
| (preceding optional arguments), 362
0 (False value), 36, 38
0x, preceding hexadecimal literals, 406
1 (True value), 36, 38
4DOM DOM implementation, 285
4Suite library, 294
127.0.0.1, IP address for localhost, 613
200 status code, 467
403 status code, 467
404 status code, 467
500 status code, 467
A
abs function, 414, 417
absolute path, converting relative path to, 116
absolute value, abs function for, 414, 417
abspath function, os.path module, 116
acos function, 415
action queue handler, 456–457
adapters, wftk, 436
addition operator (+), 19, 412
agents, workflow, 456
aggregator, RSS, 301–302
__all__ variable, 103–104, 155
Amazon wish lists, 497–500
Amazon.com web service, 495–497, 537
and operator, 48
anonymous functions, 127–128, 613
anonymous variables, 613
anydbm module, 250
apilevel global, DB API, 272
append method, list, 34
application layer, 309
624
arguments. See parameters
argv list, 99–100
arithmetic
operators, 19–20, 412–414
order of evaluation for, 21–22
precision of, 21–22
array module, 420–421
array object
from array module, 420–421
from numarray module, 422–424
arrays
definition of, 418–419
lists or tuples for, choosing, 419
mean of, 414–415
multidimensional, 418
standard deviation of, 419–420, 423–424
asin function, 415
assert statement, 192–193
AssertionError, 192
assertions, 191–193
assignment operator (=), 28
asterisk (*)
* (element grouping operator in DTD), 279
** (exponentiation operator), 406, 413
* (in format specifier), 410
* (in import command), 103
* (multiplication operator), 19, 412
* (regular expression wildcard), 186–187
* (wildcard in glob pattern), 122
* (XPath axis shortcut), 282
at sign (@), XPath axis shortcut, 282
atan function, 415
atan2 function, 415
attachments, e-mail, 315–316
attributes, XML, 275
auditing, 433–435
axis, XPath, 282
B
backslash (\\)
\\ (escape character), 7, 185
\\ (in Windows path names), 109, 178
\\n, newline character, 7–8
base 16 (hexadecimal) notation
definition of, 24, 615
formatting numbers as, 409
specifying literals in, 406

67.6 Page 666

▲back to top
characters
base 10 notation, 24
base 8 (octal) notation
definition of, 24, 617
formatting numbers as, 409
Base64 encoding, 313–314, 613
base64 module, 141
BaseRequestHandler class, 337–338
Beginning XML (Hunter, et al.), 277
binary copy protection module, 388
Binary Large OBject (BLOB), 430
binary pickle protocol, 125
BitTorrent, 354, 613
BittyWiki example
core library for, 481–484
creating wiki pages from Python session, 483–484
making executable, 493
markup used by, 486–492
resources for, 484–486
REST API document for, 529
REST API for, 500–503
SOAP API document for, 530
SOAP interface for, 525–527
web interface for, 484–493
WSDL proxy for, 533–534
XML-RPC API document for, 529–530
XML-RPC interface for, 514–517
BLOB (Binary Large OBject), 430
blocks
definition of, 2
indentation indicating, 49
Boa Constructor, 224
body
e-mail, 311, 312
HTTP, 615
braces ({}), enclosing dictionaries, 34
brackets, square ([])
dereferencing dictionaries, 35
dereferencing lists, 33
dereferencing tuples, 31
enclosing lists, 33
regular expression wildcard, 187
wildcard in glob pattern, 122
break statement, 53–54
build command, 360
built-in classes, listing, 144
built-in functions
definition of, 10
listing, 144
built-in variables, listing, 144
business processes, modeling of, 427
C
C language
extension modules written in
building and installing, 358–360
definition of, 356–358
passing parameters from Python to, 360–363
returning values to Python, 363–364
performance of, compared to Python, 355
using Python objects from, 380–382
C Python, compared to Jython, 541, 569–570
cachedir directory, Jython, 547
call stack, 74–75, 613
CANVAS example
background of, 385–386
GEOip library used by, 402
IP address, code obtaining, 399
penetration tests by, 389
pure-Python approach used by, 387
subscriptions used by, 394–395
caret (^), regular expression wildcard, 187
ceil function, 415
CGI (Common Gateway Interface)
definition of, 468–469, 613
environment variables for, 471–473
HTML forms and, 473–479
scripts, running, 469–470
web server’s role in executing scripts, 470–471
.cgi files, 469
cgi module, 474–479
cgitb module, 471
CGIXMLRPCRequestHandler class, 514, 530
channels
IRC, 340
RSS, 297
Character Large OBject (CLOB), 430
characters. See also strings
converting to numbers, 410–411
escape character, 6–7
625

67.7 Page 667

▲back to top
chat client, Python
chat client, Python, 346–350
chat server, Python, 340–346
cl command, 358, 367, 368, 379
class keyword, 81–82
classes. See also objects; subclasses
creating objects from, 82
defining, 81–82, 151–152
definition of, 613
documenting, 156–157
extending, 153
interface for, 83
methods for, 83–89
repurposing, 89
when to use, 95
clicked signal, 214–215
client-server architecture
definition of, 354, 614
REST as, 461
CLOB (Character Large OBject), 430
close method
connection object, 266
cursor object, 266
cmath module, 15, 418
code. See blocks; functions; programs
codeEditor
compared to idle editor, 3
definition of, 3
line numbers in, 3–4
Python shell in, 4–5, 16–17
starting, 3
codeEditor command, 3
colon (:)
: (preceding code block), 48–49
: (separating dictionary keys and values), 35
command-line parameters, accessing, 99–100,
134–137
comments. See also docstring
creating, 65–66
definition of, 614
commercial programs, using Python for
development environment for, 395–396
distribution of, 400–401
how much Python to use for, 386–387
libraries for, 401–402
licensing issues, 387–389
piracy issues, 386–387
626
as platforms, 395
pricing strategies, 389–395
Python problems concerning, 397–400
Python programmers for, finding, 396–397
reasons for, 385
Web services and, 388–389
commit method
connection object, 262, 271
cursor object, 266
Common Gateway Interface
definition of, 468–469, 613
environment variables for, 471–473
HTML forms and, 473–479
scripts, running, 469–470
web server’s role in executing scripts, 470–471
comparison operators, 43–47
comparisons
combining, 48
decisions and, 48–50
equality, 43–44
greater than, 45–46
greater than or equal, 47
inequality, 45
less than, 45–46
less than or equal, 47
reversing True or False, 47
compiling, 104
complex method, 417
complex numbers, 15, 416–418
complex object, 417
conjugate method, complex object, 418
connection object
DB API, 263–264
Gadfly, 262
__contains__ method, dictionary, 36
content types, MIME, 314–315, 614
context (environment) of program, 4
continue statement, 54
copy function, shutil module, 119
copy protection module, 388
cos function, 415
cosh function, 415
cPickle module, 125
C-Python, 541
Create, Read, Update, Delete (CRUD), 257
create table statement, 259–260

67.8 Page 668

▲back to top
CRUD (Create, Read, Update, Delete), 257
ctime function, time module, 118
Cunningham, Ward (wiki inventor), 479
curly braces ({}), enclosing dictionaries, 34
cursor method, connection object, 264
Cursors, DB API, 264–271
D
%d format specifier, 18, 408
data link layer, 308
data types
dictionaries
accessing keys or values in, 35–36
creating, 34–35
definition of, 34, 614
equality of, 44
keys in, 35
string substitutions using, 133–134
values in, 35
lists
adding elements to, 34
appending, 40
for arrays, when to use, 419
changing elements of, 33
creating, 33
creating with ranges, 131–133
decisions within (list comprehension), 130
definition of, 33, 616
equality of, 44
last elements of, referencing, 38–39
popping elements from, 40–41
ranges of, referencing (slicing), 39–40
reducing (running function on all elements of),
128–129
treating strings as, 36–37
visiting elements in without loops, 129–130
numbers
arithmetic operators for, 412–414
arithmetic, performing, 19–21
complex numbers, 416–418
converting characters to, 410–411
displaying, 17–19, 21, 22
formatting, 408–410
formatting in base 8 or base 16, 24
math functions for, 414–415
math module for, 415–416
databases
negative numbers, 24, 609
type of, determining, 14
types of, 13–14, 405–408
strings
combining (concatenating), 8
converting to integers, 406
definition of, 5, 619
displaying, 10
displaying numbers as, 17–19
escaping characters in, 6–7
formatting, 9–10
immutable, problems with, 399–400
newline characters in, 7–8
quotes used to enter, 6–8
substitutions using templates, 610
substitutions with dictionaries, 133–134
treating as a list (slicing), 36–37
tuples
appending, 40
for arrays, when to use, 419
creating, 30–31
definition of, 30, 619
equality of, 44
last elements of, referencing, 38–39
layers of, 31–33
one-element tuples, 32
ranges of, referencing (slicing), 39–40
reducing (running function on all elements of),
128–129
visiting elements in without loops, 129–130
database manager (DBM), 614
DatabaseError, DB API, 273
databases
accessing from Jython, 552–558
choosing, 255
DB API
connecting to database, 263–264
Cursors for, 264–271
definition of, 262–263, 614
deleting data, 270–271
downloading modules for, 263
errors, list of, 272–273
globals for, 272
inserting new data, 264–266
querying database, 266–269
transactions, 271–272
updating data, 269–270
627

67.9 Page 669

▲back to top
databases (continued)
databases (continued)
DBM persistent dictionaries
accessing, 252–255
creating, 251–252
definition of, 250
modules for, 250–251
when to use, 255
features for, 249
Gadfly database, 260–262
relational databases
connecting to database, 260–261
defining tables, 259–260
definition of, 255–257, 618
deleting data, 257–259
for document management, 429–431,
441–448
inserting new data, 257–259
querying database, 257–259
SQL (Structured Query Language), 257–261,
619
updating data, 257–259
as wftk repository, 438–446
when to use, 255
DataError, DB API, 273
DB API
connecting to database, 263–264
Cursors for, 264–271
definition of, 262–263, 614
deleting data, 270–271
downloading modules for, 263
errors, list of, 272–273
globals for, 272
inserting new data, 264–266
querying database, 266–269
transactions, 271–272
updating data, 269–270
dbhash module, 250
DBM (database manager), 614
dbm module, 250
DBM persistent dictionaries
accessing, 252–255
creating, 251–252
definition of, 250
modules for, 250–251
when to use, 255
__debug__ variable, 192–193
628
debugging
log level 6 for, 438
threads, 216–217, 225, 399
with walk function, 181
web applications, 466
decisions, 48–50
def statement, 61–62
Deferred object, 351–353
definitions element, WSDL, 531–532
del statement, 111
DELETE command, 462, 473
delete statement, SQL, 257–259
del.icio.us web service, 537
dereference, 30–31
destroy signal, 214–215
dictionaries
accessing keys or values in, 35–36
creating, 34–35
definition of, 34, 614
equality of, 44
keys in, 35
string substitutions using, 133–134
values in, 35
dir function, 80, 144
directories. See also paths
creating, 121
current, determining, 179
determining whether path is, 118
joining into path, 179
listing, 116–118
listing recursively with subdirectories, 118–119,
179
navigation of, 176
removing, 121–122
running functions on all directories in tree, 180
separator characters for, 178
statistics about, 179
for user and group management, 431–432
disutils package, 171–173, 359–360
division operator (/), 19, 412
.dll files, 356
DNS (Domain Name System), 309, 614
__doc__ string, 63–64

67.10 Page 670

▲back to top
docstring (documentation string).
See also comments
for classes, 82, 83, 92
for functions, 63–64
in test cases, 203
document events, SAX, 287–288
document management
auditing and, 434–435
file system for, 428–429
relational databases for, 429–430, 441–448
retention of documents, 430–431, 434–435,
446–448
version control system for, 430
document models
definition of, 614
validating XML against, 285–287
when to use, 278
Document Object Model. See DOM
document objects, DOM, 288
document root, XML, 276
Document Type Definition. See DTD
documentation, for web service API, 529–534.
See also docstring
documentation string. See docstring
documents, REST resources as, 461–462.
See also files
dollar sign ($), regular expression wildcard, 187
DOM (Document Object Model)
definition of, 288–289, 614
implementations of, 285
when to use, 289
Domain Name System (DNS), 309, 614
double quotes (“...”), enclosing strings, 6, 618
DTD (Document Type Definition)
compared to XML, 280
definition of, 277, 278, 614
example of, 278–280
limitations of, 280
for RSS, 297
dumbdbm module, 251
dump function, pickle module, 123–124
dynamic cursor, 558
enterprise applications
E
e constant, 415
%E format specifier, 18
Eclipse IDE, 565
elif statement, 50
else statement
in if block, 50
in repetitions, 54
in try block, 57
e-mail
attachments, 315–316
compared to webmail, 331
file format for, 311–312
MIME messages, 313–321
quote about, by Jamie Zawinski, 305
retrieving, 323–331
retrieving from IMAP server, 327–331
retrieving from POP3 server, 325–327
searching, 178
sending, 305–307, 311–312, 321–323
SMTP and, 321–323
email module, 312, 321, 323, 326
email.Message class, 318
email.Mime* classes, 318
email.MimeMultipart class, 318
enactment, in workflow, 433
encapsulation, 151, 156, 614
encodings, MIME, 313–314
encryption
definition of, 614
of passwords, 140–141
enterprise applications
auditing and, 433–435
definition of, 427–428
document management, 428–431, 434–435
python-ldap module for, 448–453
user and group management, 431–432
wftk workflow toolkit
action queue handler, 456–457
definition of, 435–436
repository, in database, defining and accessing,
438–439
629

68 Pages 671-680

▲back to top

68.1 Page 671

▲back to top
enterprise applications (continued)
enterprise applications (continued)
repository, in database, for document
management, 441–448
repository, in database, storing records in,
439–446
repository, in directory files, defining and
accessing, 436–438
repository, in directory files, storing records in,
438–439
resources for, 607
workflow trigger, 453–456
workflow, 432–433
Enterprise license, 389
environment (context) of program, 4
ephemeral port, 614
ephemeral port number, 334
equal sign (=)
= (assignment operator), 28
== (equality operator), 43–44
Error error, DB API, 273
errors. See exceptions
escape character, 6–7
escape sequences, 615
escaping, 6–7
event-driven parsing, 288
events, widgets, 214–215
examples
BittyWiki example
core library for, 481–484
creating wiki pages from Python session,
483–484
making executable, 493
markup used by, 486–492
resources for, 484–486
REST API document for, 529
REST API for, 500–503
SOAP API document for, 530
SOAP interface for, 525–527
web interface for, 484–493
WSDL proxy for, 533–534
XML-RPC API document for, 529–530
XML-RPC interface for, 514–517
CANVAS example
background of, 385–386
GEOip library used by, 402
IP address, code obtaining, 399
penetration tests by, 389
pure-Python approach used by, 387
subscriptions used by, 394–395
of DTD (Document Type Definition), 278–280
of Glade GUI construction kit, 231–238,
241–248
Google API example, 520–522
LAME project
definition of, 364
extension module for, 368–380
using, 364–367
Meerkat API example, 509–511
mirror client example, 336–337
mirror server example, 335–336
PyRAP example, 231–238, 241–248
search utility examples, 200–207
of Widget Tree, Glade, 241–248
except statement, 55–57
exceptions
AssertionError, 192
from within call stack, 74–75
DB API, 272–273
of DB API, list of, 272–273
definition of, 55
documenting, 156–157
of Fault element of SOAP, 524–525
of fault response of XML-RPC, 513–514
flagging, 73–74
handling, 55–57
IndexError, 33, 34
exceptions (continued)
IOError, 113
KeyError, 55, 56
OSError, 114
from Python shell, 23–24
of Python shell, 23–24
SOAP, 524–525
TypeError, 32, 56, 68
types of, 56
XML-RPC, 513–514
exclamation point (!)
!= (inequality operator), 45
! (wildcard in glob pattern), 122
exec function, os module, 137
630

68.2 Page 672

▲back to top
files
execfile method, PythonInterpreter class, 566
execute method, cursor object, 266
exp function, 415
Expat XML parser, 285
exploit program, 389
exponential notation, 408
exponentiation operator (**), 406, 413
extend method, 34, 40
eXtensible Markup Language (XML)
compared to DTD, 280
created by Glade, 230–231
definition of, 275–277, 620
libraries for, 285
parsing
with DOM, 288–289, 290–292
with SAX, 287–288, 289, 290, 292–293
standards for, 277
transforming with XSLT, 293–296
validating, 285–287, 621
XPath expressions in, 282
Extensible Style Language for Transformations
(XSLT)
definition of, 277, 293–294, 621
transforming XML with, 294–296
Extensible Style Language Formatting Objects
(XSL-FO), 277, 621
extension module
building, 358–360
compiling, 358, 367, 368, 379
definition of, 356–358
include files for, 356
initialization function for, 358
installing, 360
LAME project
definition of, 364
extension module for, 368–380
using, 364–367
method table for, 357, 361
names of C functions in, 357
passing parameters from Python to C, 360–363
returning values from C to Python, 363–364
signatures of C functions in, 356, 360
Extreme Programming (XP), test suites for,
199–205
F
%f format specifier, 18, 22, 408
False value (0)
definition of, 36, 38
as result of comparisons, 43
Fault element, SOAP, 524–525
fault response, XML-RPC, 513–514
feeds, RSS, 297–301
fetchall method, cursor object, 267
fetchone method, cursor object, 267
Fielding, Roy (REST definition), 460
FieldStorage class, 474–475
file object
definition of, 109
deleting, 111, 112
line lengths in, printing, 112
reading, 111–112
writing (creating), 110–111
files
closing, 111, 112
copying, 119
deleting, 119–120
determining whether path is, 118
for document management, 428–429
errors accessing, 113
globbing, 122–123
line lengths in, printing, 112
moving, 119
packages
creating, 101–103
definition of, 95, 101–102, 617
importing all elements of, 103–104
re-importing, 104–106
testing, 106
permissions, 120
pickle files, writing, 123–125
reading, 111–112
renaming, 119
rotating old versions of, 120–121
saving program in, 59–61
searching for
examples of, 200–207
by file name, 176–177
by file type, 181–184, 187–188
splitting extension from end of, 115–116
631

68.3 Page 673

▲back to top
files (continued)
files (continued)
statistics about, 179
writing (creating), 110–111
filter function, using with lambda statement,
127–128, 186
findgtk.py module, 211–212
firewall
effect on binding to external hostname, 335
penetration tests and, 389
500 status code, 467
Flickr web service, 537
floats (floating-point numbers)
converting strings to, 408
converting to integers, 406
definition of, 14, 407, 615
exponential notation for, 408
high-precision, 15
precision of, 407, 408
scientific notation for, 17
floor division operator (//), 413
floor function, 415
for statement, 51–54
foreign key, in relational database, 256
fork function, os module, 137
ForkingMixIn class, 339–340
<FORM> tag, 473
format specifiers
asterisk (*) in, 410
%d format specifier, 18, 408
%E format specifier, 18
%f format specifier, 18, 22, 408
hyphen (-) in, 10
%o format specifier, 24, 409
%#o format specifier, 409
%s format specifier, 9
using, 9
%x format specifier, 24, 409
%X format specifier, 24
%#x format specifier, 409
forward slash
/ (division operator), 19, 412
// (floor division operator), 413
/ (in Unix path names), 178
// (XPath axis shortcut), 282
403 status code, 467
404 status code, 467
4DOM DOM implementation, 285
4Suite library, 294
from modifier, import command, 101, 145
functions. See also specific functions
anonymous functions, 127–128, 613
calling from within other functions, 71–72
creating, 61–62
creating module with, 150
creating within other functions (nesting), 72–73
definition of, 10, 59, 615
documenting, 63–64, 156–157
layers of invocations of, 74–75
naming, 62
parameters for, 66–71
recursive functions, 115
scope of variables and, 64–65
as signal handlers, 214–215
G
Gadfly database, 260–262
gcc command, 358, 367, 368, 379
gdbm module, 251
GEOip library, 402
GET command, 462, 473
getcwd function, os module, 179, 180
getmtime function, os.path module, 118–119
getopt function, getopt module, 135–136
getopt module, 134–137
getsize function, os.path module, 118–119
Glade GUI construction kit
definition of, 223–224
example using, 231–238, 241–248
installing, 225
project, creating, 227
starting, 226
Widget Tree, 238–241
widgets, creating in window, 228–230
window, creating, 227–228
XML created by, 230–231
glob function, glob module, 122–123
glob module, 122–123
globbing, 122–123
glue, 356. See also extension module
632

68.4 Page 674

▲back to top
Gmail, 331
GNU DBM library, 251
gnu_getopt function, getopt module, 136–137
Google API example, 520–522
Google web service, 537
googol, representing, 406
greater than operator (>), 45–46
greater than or equal operator (>=), 47
groups, managing, 431–432
GUI builders, 223–225
GUIs. See also pyGTK toolkit
GUI builders for, 223–225
toolkits for writing with Python, 209–210
H
hash mark (#)
# (preceding comments), 65–66
#! (shebang comment line), 60, 545
headers
definition of, 615
e-mail, 311–312
HTTP
CGI script outputting, 470
definition of, 467–468, 615
help function, 157–162
hexadecimal (base 16) notation
definition of, 24, 615
formatting numbers as, 409
specifying literals in, 406
high-precision floating-point numbers, 15
hostname
binding to, 332
definition of, 615
DNS managing, 309
external, binding to, 334–335
localhost, 309, 613, 616
using instead of IP addresses, 310
HSqlDB database, 554
HTML forms
definition of, 473–474
parsing, 474–479
HTML (HyperText Markup Language)
parsing, 283–285
relationship to XML, 282
htmllib module, 284–285
if statements
HTMLParser class, 283–284
HTTP (HyperText Transfer Protocol)
body, 615
commands for, 462–463, 473
definition of, 615
design guidelines for, 468
headers
CGI script outputting, 470
definition of, 467–468, 615
request
definition of, 466–467
viewing with web server, 464–466
request, definition of, 615
response
definition of, 467–468
viewing with web server, 464–466
response, definition of, 615
session length of, 461
status code, 467–468, 615
verb, 462–463, 473, 615
visible web server using, 464–466
web server using, 463
HTTPServer object, 463
HttpServlet class, 561–564
HTTP_USER_AGENT variable, 473
Hunter, David (Beginning XML), 277
hyperbolic functions, 415
HyperCard, GUI builder for, 225
HyperText Markup Language (HTML)
parsing, 283–285
relationship to XML, 282
HyperText Transfer Protocol. See HTTP
hyphen (-)
- (in format specifier), 10
- (subtraction operator), 19, 412
hypot function, 415
I
IANA (Internet Assigned Numbers Authority),
310–311
IDE (Integrated Development Environment),
395–396, 565
idempotent action, 462, 615
idle editor, compared to codeEditor, 3
if statements, 48–50
633

68.5 Page 675

▲back to top
imag attribute, complex object
imag attribute, complex object, 417
imaginary number
creating complex numbers from, 416
definition of, 14–15, 615
IMAP (Internet Message Access Protocol)
definition of, 615
unique message IDs used by, 330–331
IMAP server
retrieving e-mail from, 327–331
secure, 331
imaplib module, 327–331
immutable
definition of, 32
strings as, problems with, 399–400
import command
from modifier, 101, 145
using, 96–97, 145
version 2.4 enhancements to, 611
indentation, 49
index, database, 441–442
IndexError, 33, 34
inequality operator (!=), 45
inf (infinity), 19
infinite loop, 52–53, 616
infrastructure platforms, 427
inheritance, 151, 616
__init__ method, 152
__init__.py file, 101–103
input, 10. See also Input/Output (I/O)
Input/Output (I/O). See also files; pickles
definition of, 10, 109, 616
file exceptions, 113
manipulating files, 119–121
reading files, 111–112
writing files, 110–111
insert statement, SQL, 257–259
install command, 360
int method, 406
integers. See also longs
automatically converted to long if too large, 609
converting to, from other types, 406
definition of, 14, 406, 616
internal representation of, 407
Integrated Development Environment (IDE),
395–396, 565
IntegrityError, DB API, 273
634
interface for a class, 83
interface methods, 83–84, 85–87
InterfaceError, DB API, 273
internal methods, 84–85
InternalError, DB API, 273
Internet. See network programming
Internet Assigned Numbers Authority (IANA),
310–311
Internet Message Access Protocol (IMAP)
definition of, 615
unique message IDs used by, 330–331
Internet Protocol. See IP
Internet protocol stack, 308–309, 618
Internet Relay Chat (IRC), 340, 616
interpreted languages, 2
introspection API, XML-RPC, 530–531
I/O (Input/Output). See also files; pickles
definition of, 10, 109, 616
file exceptions, 113
manipulating files, 119–121
reading files, 111–112
writing files, 110–111
IOError, 113
IP address
definition of, 309–310, 616
for localhost, 309, 613
IP (Internet Protocol)
addresses, 309–310
definition of, 616
ports, 310–311
IRC (Internet Relay Chat), 340, 616
isdir function, os.path module, 118
isfile function, os.path module, 118
ISO 9000, 434
ISO 9001, 434
iterator, 239, 616
J
j (imaginary number indicator), 15, 416
Java language. See also Jython
classes, calling from Jython, 548–550
classes, embedding Jython interpreter in,
566–568
compared to Python, 539
compiling Jython code into, 568–569
integrating Jython into, 547–552

68.6 Page 676

▲back to top
Java Virtual Machine (JVM), 539, 616
javadom Java DOM implementation, 285
JDBC APIs, 552–553
jEdit text editor, 564
J2EE (Java 2 Enterprise Edition), 616
J2EE servlets, writing, 558–564
join function, os.path module, 114, 179
joins, SQL, 267–269
JVM (Java Virtual Machine), 539, 616
Jython
advantages of, 540–541
compared to C Python, 569–570
compiler, 542
compiling into Java, 568–569
database tables, creating, 555–558
databases, accessing, 552–558
definition of, 539–540, 616
executable commands using, 545–546
IDE for, 565
installing, 541–542
interpreter
calling, 546–547
embedding into Java classes, 566–568
integrating into Java applications, 547–552
running, 542
Java classes, calling, 548–550
jEdit text editor for, 564
J2EE servlets, writing, 558–564
packaging applications based on, 547
platforms supported by, 541
resources for, 607
running interactively, 542–543
running Python scripts using, 543–544
Swing API, using with, 550–552
testing from, 565–566
user interface, creating, 550–552
jython script
passing parameters to, 544–545
running, 542
jythonc script
definition of, 542
running, 568–569
jython.jar file, 547
listdir function, os module
K
KDE builder, 224
KeyError, 55, 56
keys, in dictionaries, 35
L
lambda statement, 127–128
LAME project
definition of, 364
extension module for, 368–380
using, 364–367
languages, programming. See also C language;
Java language; Python
compared to protocols, 307–308
interpreted, 2
Large OBject (LOB), 430
layers (levels of abstraction), 308–309
LDAP (Lightweight Directory Access Protocol),
432, 453. See also python-ldap module
ldapadd utility, 450
left angle bracket (<)
< (less than operator), 45–46
<= (less than or equal operator), 47
len function
for arrays, 34
for strings, 112
for tuples, 31
__len__ method, 80–81
less than operator (<), 45–46
less than or equal operator (<=), 47
levels of abstraction (layers), 308–309
Lib directory, Jython, 547
libglade library, 223, 225, 231
libgmail project, 331
licensing, 387–389
Lightweight Directory Access Protocol (LDAP),
432, 453. See also python-ldap module
line numbers, displayed in codeEditor, 3–4
list comprehension, 130
listdir function, os module, 118–119, 179, 181
635

68.7 Page 677

▲back to top
lists
lists
adding elements to, 34
appending, 40
for arrays, when to use, 419
changing elements of, 33
creating, 33
creating with ranges, 131–133
decisions within (list comprehension), 130
definition of, 33, 616
equality of, 44
last elements of, referencing, 38–39
popping elements from, 40–41
ranges of, referencing (slicing), 39–40
reducing (running function on all elements of),
128–129
treating strings as, 36–37
visiting elements in without loops, 129–130
literal numbers, 406, 407
LiveHTTPHeaders extension, 466
load function, pickle module, 124
LOB (Large OBject), 430
local mail spool, 323–325
localhost
definition of, 616
IP address for, 309, 613
log files
clipping (filtering), 177
rotating old versions of, 120–121
log function, 415
log10 function, 415
logarithm functions, 415
longs (long integers)
definition of, 14, 406–407
integers automatically converted to if too large,
609
internal representation of, 407
loops (repetitions)
breaking out of, 52–54
continuing, 54
creating, 51–52
definition of, 617
else statement in, 54
infinite loops, 52–53, 616
M
mail. See e-mail
mail spool, local, 323–325
Mail To The Future web service, 537
mailbox module, 323–325
__main__ name, 106
makedirs function, os module, 121
map function, 129–130
marshall library, 285
marshalling, 399
match function, re module, 186
math. See arithmetic
math functions, 414–415
max function, 414
mbox format, 324
Meerkat API example, 509–511
Meerkat web service, 537
metadata, 430
methods
definition of, 79–80, 617
documenting, 156–157
interface methods, 83–84, 85–87
internal methods, 84–85
private methods, 84
using, 87–89
MIME (Multipurpose Internet Mail Encoding)
content types, 314–315
definition of, 617
encodings, 313–314
multipart messages, 316–321
standards for, 313
min function, 414
mirror client example, 336–337
mirror server example, 335–336
mkdir function, os module, 121
model, view, controller design, 238
modeling of business processes, 427
modules
accessing elements of, 97–98, 99
creating, 97–99, 150, 165–168
definition of, 95–96, 143–144, 617
documentation for, viewing, 157–162
documenting, 156–157
636

68.8 Page 678

▲back to top
numbers
editing, 98–99
example of, 146–149
exceptions for, defining, 154
importing
all elements of, 103, 145, 168
into program with module scope, 96–97
specific elements of, 145, 611
into top-level scope of program, 101, 103–104,
168–169
installing, 170–173
list of elements in, 144
list of modules that have been called, 105
location of, 97–98, 145–146
public items in, defining, 155–156
re-importing, 104–106
reloading, 145
running as a program, 164
scope of, named for module, 99
testing, 106, 162–163, 164
viewing, 146–149
modulus (remainder) operator (%), 20, 414
move function, shutil module, 119
MP3 example. See LAME project
multidimensional arrays, 418
multipart message, 316–321, 617
multiplication operator (*), 19, 412
Multipurpose Internet Mail Encoding (MIME)
content types, 314–315
definition of, 617
encodings, 313–314
multipart messages, 316–321
standards for, 313
multithreaded servers, 339–340
multithreading. See threads
mutex, 218
MySQL
document retention using, 446–448
modules for, 263
resources for, 607
storing data in, 439–446
N
\\n, newline character, 7–8
__name__ name, 106
names. See variables
namespaces, XML, 276–277
NDA, 389
negative numbers, 24, 609
network layer, 309
network programming
e-mail
attachments, 315–316
MIME messages, 313–321
retrieving, 323–331
sending, 305–307, 311–312
SMTP and, 321–323
peer-to-peer architecture, 354
protocol design, 350
protocols
compared to programming languages, 307–308
definition of, 307
Internet Protocol (IP), 309–311
Internet protocol stack, 308–309, 618
socket programming
binding to external hostname, 334–335
binding to hostname, 332
definition of, 331–332
example of, 332–334
firewalls and, 335
mirror client example, 336–337
mirror server example, 335–336
multithreaded servers, 339–340
Python chat client, 346–350
Python chat server, 340–346
SocketServer module, 337–339
Twisted framework, 351–353
newline character (\\n)
definition of, 7
entering in strings, 7–8
nickname, IRC, 340
node test, XPath, 282
None value, 38, 611
normcase function, os.path module, 180
normpath function, os.path module, 116
not operator, 47
NotSupportedError, DB API, 273
numarray package, 422–424
numbers
arithmetic operators for, 412–414
arithmetic, performing, 19–21
637

68.9 Page 679

▲back to top
numbers (continued)
numbers (continued)
complex numbers, 416–418
converting characters to, 410–411
displaying, 17–19, 21, 22
formatting, 408–410
formatting in base 8 or base 16, 24
math functions for, 414–415
math module for, 415–416
negative numbers, 24, 609
type of, determining, 14
types of, 13–14, 405–408
numerical analysis, 405
numerical programming. See also numbers
arrays for, 418–424
definition of, 405
O
%o format specifier, 24, 409
%#o format specifier, 409
object-oriented programming (OOP)
definition of, 2, 151
uses of, 92
objects. See also classes
creating from classes, 82
definition of, 79–81, 617
listing all elements of, 80
pickling, 123–125
Python, using from C code, 380–382
scope of, 89–92
uses of, 81
octal (base 8) notation
definition of, 24, 617
formatting numbers as, 409
1 (True value), 36, 38
127.0.0.1, IP address for localhost, 613
OOP (object-oriented programming)
definition of, 2, 151
uses of, 92
open function, 112
OpenLDAP project, 432, 448–449
operating systems
directory separator characters specific to, 178
porting to other operating systems, 398–399
OperationalError, DB API, 273
operators
arithmetic, 19–20, 412–414
assignment operator, 28
for combining comparisons, 48
comparison operators, 43–47
or operator, 48
ord function, 411
order of evaluation, for arithmetic, 21–22
os module
definition of, 113
errors raised by, 114
text processing functions, 178–184
OSError, 114
os.listdir module, 116–117
os.path module, 114–116
output, 10. See also Input/Output (I/O)
overloading, 20
P
packages
creating, 101–103
definition of, 95, 101–102, 617
importing all elements of, 103–104
re-importing, 104–106
testing, 106
parameters
command-line parameters, 99–100, 134–137
for functions, 66–71
passing from Python to C, 360–363
passing to jython script, 544–545
paramstyle global, DB API, 272
parentheses (())
enclosing list of DTD elements, 279
enclosing tuples, 30
pass statement, 56–57
passwords, encrypting, 140–141, 331, 392
path variable, sys module, 97–98, 145–146,
170–171
PATH_INFO variable, 472–473
paths. See also directories
assembling directory names into, 114
assigning to strings, 109–110
converting relative to absolute, 116
determining whether file or directory, 118
638

68.10 Page 680

▲back to top
normalizing, 116
normalizing case of, 180
splitting extension from end of, 115–116
splitting into directories, 114–115, 179
pattern matching. See globbing
peer-to-peer architecture, 354, 617
penetration tests, 389
percent sign (%)
% (escaping in strings), 18–19
% (preceding format specifier), 9
% (remainder operator), 20, 414
% (string formatting operator), 408
%d (format specifier), 18, 408
%E (format specifier), 18
%f (format specifier), 18, 22, 408
%o (format specifier), 24, 409
%#o (format specifier), 409
%s (format specifier), 9
%x (format specifier), 24, 409
%X (format specifier), 24
%#x (format specifier), 409
per-company licensing, 389
performance
assertions and, 193
of C code compared to Python code, 355
improving, 355
of loops, 400
of Twisted framework, 353
period (.), regular expression wildcard, 185
physical layer, 308
pi constant, 415
pickle module, 123, 399
pickles, 123–125
piracy, 386–387
plus sign (+)
+ (addition operator), 19, 412
+ (combining strings), 8
+ (element grouping operator in DTD), 279
polymorphism, 151, 617
pop method, list, 40–41
POP (Post Office Protocol), 618
POP3 server
retrieving e-mail from, 325–327
secure, 331
protocol stack
Popen class, 610
poplib module, 325–327
ports
definition of, 310–311, 618
ephermeral port number, 334
well-known port, 311, 620
POST command, 462, 473, 508, 511–513
Post Office Protocol (POP), 618
pound sign (#)
# (preceding comments), 65–66
#! (shebang comment line), 60, 545
predicates, XPath, 282
pricing strategies, 389–395
primary key, in relational database, 256
print function, 10, 21
print statement, 111
print_line_lengths method, file object, 112
private methods, 84
processes. See also subprocess
multiple, using, 137–139
threading, 139–140
programmers, Python, finding, 396–397
programming
changing requirements of, 2–3
consistency in, 2
control in, 2
object-oriented programming, 2
principles of, 1–3
programming languages. See also C language;
Java language; Python
compared to protocols, 307–308
interpreted, 2
ProgrammingError, DB API, 273
programs
blocks of, 2, 49
context (environment) of, 4
current scope of, determining, 106
running, 61
saving in a file, 59–61
as source code, 3
top-level scope of, name for, 106
writing in codeEditor, 3–5
protocol stack, 308–309, 618
639

69 Pages 681-690

▲back to top

69.1 Page 681

▲back to top
protocols
protocols
compared to programming languages, 307–308
definition of, 307, 618
design considerations, 350
Internet Protocol (IP), 309–311
Internet protocol stack, 308–309, 618
terse, 350
pulldom DOM implementation, 285
PUT command, 462, 473
.py files, 60, 146
Py_BuildValue function, 363–364
.pyc files, 104
pydoc command, 173
pyglade module, 606
pygtk module
definition of, 402
importing, 211–212
resources for, 606
pyGTK toolkit
creating widgets, 213–214
definition of, 210
FAQ for, 213
Glade GUI construction kit, 223–224
installing, 211
mailing list for, 213
resources for, 213
running programs, 213–214
testing installation of, 211–212
tutorials for, 213
widgets
creating, 213–214
definition of, 213
handling signals, 214–215, 216–221
multiple, in one window, 222–223
show method for, 214
signals generated by, 214
PyMethodDef structure, 357
PyMODINIT_FUNC function, 358
PyPy project, 308
pyQT toolkit, 209–210, 224
PyRAP example, 231–238, 241–248
Py_RETURN_NONE macro, 356
PySAX libraries, 285
PyServlet class, 560–561
Python. See also Jython
compared to Java, 539
porting to other operating systems, 398–399
porting to other versions of, 397–398
problems with, 397–400
reasons to use
for commercial and shareware programs, 385
for enterprise applications, 427
for extension programming, 355
for GUIs, 209–210
for network programming, 305
for text processing, 175–176
for web applications, 459–460
resources for, 605–606
version 2.4, new features in, 609–611
Python chat client, 346–350
Python chat server, 340–346
python command
h option, 135
i option, 61, 100
running, 4
Python shell
definition of, 4
errors, receiving, 23–24
>> prompt in, 5
starting, 4–5
using with codeEditor, 16–17
PythonCard
codeEditor in, 3
definition of, 3, 225
downloading, 3
resources for, 606
Python.h file, 356
PythonInterpreter class, 566
python-ldap module
defining and accessing data, 449–450
definition of, 448–449
searching data, 451–453
PYTHONPATH variable, 360
PyUnit facility, 191, 607. See also unittest
module
pywftk module, 607
PyXML package, 285, 290
640

69.2 Page 682

▲back to top
reserved words
Q
Query, Update, Insert, Delete (QUID), 257
QUERY_STRING variable, 472–473
question mark (?)
? (element grouping operator in DTD), 279
? (wildcard in glob pattern), 122
queue of threads, 216–221
QUID (Query, Update, Insert, Delete), 257
quoted-printable encoding, 313–314, 618
quotes
double quotes (“...”), enclosing strings, 6, 618
single quote (‘), element grouping operator in
DTD, 279
single quotes (‘...’), enclosing strings, 6, 618
triple quotes (‘’’...’’’ or “””...”””)
enclosing documentation, 156–157
enclosing multi-line strings, 7–8, 618
R
“r”, preceding regular expressions, 185
raise statement, 73–74
range, 618
range function, 131–132
raw strings, in regular expressions, 185
RDBMS (Relational Database Management Sys-
tem), 618. See also relational databases
RDF Site Summary. See RSS
re module, 186–188
read method, file object, 111
readline method, file object, 111
readlines method, file object, 112
real attribute, complex object, 417
Really Simple Syndication. See RSS
(Rich or RDF Site Summary)
recursive functions, 115
reduce function, 128–129
regular expressions
definition of, 176, 184–185
examples of, 186–187
searching for files using, 187–188
uses of, 176
Relational Database Management System
(RDBMS), 618. See also relational databases
relational databases
connecting to database, 260–261
defining tables, 259–260
definition of, 255–257, 618
deleting data, 257–259
for document management, 429–431, 441–448
inserting new data, 257–259
querying database, 257–259
SQL (Structured Query Language), 257–261, 619
updating data, 257–259
as wftk repository, 438–446
when to use, 255
relative path, converting to absolute, 116
reload function, 105–106, 145
remainder (modulus) operator (%), 20, 414
Remote Procedure Call, XML. See XML-RPC
remove function, os module, 119
repetitions (loops)
breaking out of, 52–54
continuing, 54
creating, 51–52
definition of, 617
else statement in, 54
infinite loops, 52–53, 616
repository, wftk
in database
defining and accessing, 439–441
for document management, 441–448
storing records in, 439–446
definition of, 435
in directory files
defining and accessing, 436–438
storing records in, 438–439
repr function, 407
representation, REST, 462, 618
REpresentational State Transfer. See REST
request
HTTP
definition of, 466–467, 615
viewing with web server, 464–466
SOAP, 522–524
XML-RPC, 511–513
REQUEST_METHOD variable, 472
reserved words, 29
641

69.3 Page 683

▲back to top
resource identifier
resource identifier
definition of, 618
HTTP, 466–467, 468
REST, 461, 484, 486, 495, 500–501
resources (information). See also web sites
Beginning XML (Hunter, et al.), 277
pyGTK toolkit, 213
software, 605–607
web services, publicly available, 536–537
resources, REST, 460, 461–462, 618
response
HTTP
definition of, 467–468, 615
viewing, 464–466
SOAP, 524
XML-RPC, 513
REST (REpresentational State Transfer)
definition of, 460–462, 618
design guidelines for, 468
documentation for, 529
operations of, 462–463
resources for, 460, 461–462
web services using
Amazon Web Services, 495–497
Amazon wish lists, 497–500
BittyWiki example of, 500–503
definition of, 494–495
when to use, 534–535
wiki search and replace using, 503–508
RESTfulness, 618
reverse function, 610
RFC 1459, 340
RFC 1521, 313
RFC 1939, 325
RFC 2616, 467
RFC 2822, 311–312, 618
RFCs, 311
Rich Site Summary. See RSS
right angle bracket (>)
> (append in print statement), 111
> (greater than operator), 45–46
>= (greater than or equal operator), 47
>> (prompt in Python shell), 5
rmdir function, os module, 121
rmtree function, shutil module, 122
robot
definition of, 618
web services as, 493–495
roles, 432, 433
rollback method, connection object, 271–272
rooms, IRC, 340
rot13 cipher, 411–412
round function, 414
RSS (Rich or RDF Site Summary)
creating RSS aggregator, 301–302
creating RSS feed, 297–301
definition of, 296, 619
DTD for, 297
versions of, 296–297
runTest method, TestCase class, 193–196
S
%s format specifier, 9
SafeTemplate class, 610
Sarbanes-Oxley Act (SOX), 434
SAX (Simple API for XML)
definition of, 285, 287–288, 619
when to use, 289
schemas
advantages of, 281
definition of, 277, 278, 280, 620
disadvantages of, 281–282
example of, 280–281
scientific notation, 17
scope
definition of, 619
of modules
importing modules into, 96–97
name of, 99
of objects, 89–92
of programs, determining, 106
top-level
importing modules into, 101, 103–104
name for, 106
of variables, 64–65
scripts
CGI, 469–471
running Python scripts using Jython, 543–544
642

69.4 Page 684

▲back to top
search function, re module, 186
search utility examples, 200–207
secure IMAP, 331
secure POP3, 331
Secure Socket Layer (SSL), 331, 619
select function, 349–350
select module, 349–350
select statement, SQL, 257–259
sequences
accessing in reverse, 610
appending, 40
definition of, 619
equality of, 44
last elements of, referencing, 38–39
lists
adding elements to, 34
appending, 40
for arrays, when to use, 419
changing elements of, 33
creating, 33
creating with ranges, 131–133
decisions within (list comprehension), 130
definition of, 33, 616
equality of, 44
last elements of, referencing, 38–39
popping elements from, 40–41
ranges of, referencing (slicing), 39–40
reducing (running function on all elements of),
128–129
treating strings as, 36–37
visiting elements in without loops, 129–130
ranges of, referencing (slicing), 39–40
strings as, 38
tuples
appending, 40
for arrays, when to use, 419
creating, 30–31
definition of, 30, 619
equality of, 44
last elements of, referencing, 38–39
layers of, 31–33
one-element tuples, 32
ranges of, referencing (slicing), 39–40
reducing (running function on all elements of),
128–129
visiting elements in without loops, 129–130
slash (/)
servers, trusted, 350
servlet container, 559
servlets, J2EE, 558–564
setter methods, 152
setUp method, TestCase class, 196–199
setup.py file, 171–172
sgmlop C helper module, 285
sha module, 140–141
shareware, using Python for, 385
development environment for, 395–396
distribution of, 400–401
how much Python to use for, 386–387
libraries for, 401–402
licensing issues, 387–389
piracy issues, 386–387
as platforms, 395
pricing strategies, 389–395
Python problems concerning, 397–400
Python programmers for, finding, 396–397
reasons for, 385
Web services and, 388–389
sharp sign (#)
# (preceding comments), 65–66
#! (shebang comment line), 60, 545
shebang comment line (#!), 60, 545
shell. See Python shell
show method, for widgets, 214
shutil module, 119
signals generated by widgets, 214–215
Simple API for XML. See SAX
Simple Mail Transfer Protocol. See SMTP
Simple Object Access Protocol. See SOAP
SimpleHTTPRequestHandler object, 464
SimpleXMLRPCServer class, 514, 530
sin function, 415
single quote (‘), element grouping operator in
DTD, 279
single quotes (‘...’), enclosing strings, 6, 618
single-threaded multitasking, 348–350
sinh function, 415
site-packages directory, 171
slash (/)
/ (division operator), 19, 412
// (floor division operator), 413
/ (in Unix path names), 178
// (XPath axis shortcut), 282
643

69.5 Page 685

▲back to top
SMTP (Simple Mail Transfer Protocol)
SMTP (Simple Mail Transfer Protocol)
definition of, 619
port number for, 310
sending e-mail with, 321–323
smtplib module, 306, 321–323
.so files, 356
SOAP (Simple Object Access Protocol)
BittyWiki example using, 525–527
definition of, 520, 619
documentation for, 530
Fault element (errors flagged by), 524–525
Google API example, 520–522
request for, 522–524
response for, 524
when to use, 534
wiki search and replace using, 527–529
SOAPpy library, 520
socket library, 332
socket programming
binding to external hostname, 334–335
binding to hostname, 332
definition of, 331–332
example of, 332–334
firewalls and, 335
mirror client example, 336–337
mirror server example, 335–336
multithreaded servers, 339–340
Python chat client, 346–350
Python chat server, 340–346
SocketServer module, 337–339
Timeoutsocket module, 401–402
sockets
binding to hostname, 332
definition of, 332, 619
example of, 332–334
SocketServer module, 337–339
software, resources for, 605–607
sorted function, 117
source code, 3
SOX (Sarbanes-Oxley Act), 434
spaces (whitespace)
for code blocks, 49
definition of, 620
spawn function, os module, 138
specialization (inheritance), 151, 616
Speller web service, 537
spider
definition of, 619
REST example of, 503–508
SOAP example of, 527–529
XML-RPC example of, 518–519
split function, os.path module, 114–115, 179
splitext function, os.path module, 115–116
SQL (Structured Query Language)
connecting to database, 260–261
create table statement, 259–260
definition of, 619
delete statement, SQL, 257–259
insert statement, SQL, 257–259
joins, 267–269
select statement, SQL, 257–259
update statement, SQL, 257–259
sqrt function, 415
square brackets ([])
dereferencing dictionaries, 35
dereferencing lists, 33
dereferencing tuples, 31
enclosing lists, 33
regular expression wildcard, 187
wildcard in glob pattern, 122
SSL (Secure Socket Layer), 331, 619
stack trace, 75
startup method, connection object, 262
stat function, os module, 179, 181
state machine, 455
stateless servers, REST, 461
static database cursor, 558
status code, HTTP, 467–468, 615
steps, XPath, 282
str function, 18, 407, 408
stream-based parsing, 288
StreamRequestHandler class, 338
string formatting operator (%), 408
string module, 134
strings. See also docstring
combining (concatenating), 8
converting to integers, 406
definition of, 5, 619
displaying, 10
displaying numbers as, 17–19
escaping characters in, 6–7
formatting, 9–10
644

69.6 Page 686

▲back to top
immutable, problems with, 399–400
newline characters in, 7–8
quotes used to enter, 6–8
substitutions using templates, 610
substitutions with dictionaries, 133–134
treating as a list (slicing), 36–37
Structured Query Language. See SQL
subclass Error, DB API, 273
subclasses
creating, 139, 153
inheritance and, 151
polymorphism and, 151
subprocess
creating, 610
multithreaded servers and, 339
when to use, 140
subprocess.call function, 393, 610
subscription ID, Amazon Web Services, 495
subtraction operator (-), 19, 412
sum function, 414
Swing API, using Jython with, 550–552
sys module, path variable, 97–98, 145–146,
170–171
sys.modules dictionary, 105
sys.modules.keys list, 105
system function, os module, 138–139
T
tables, in relational databases
creating in Jython, 555–558
defining, 259–260
definition of, 256
tags, XML, 275–276
tan function, 415
tanh function, 415
tarball, 400–401
task, in workflow, 432–433
TCP (Transport Control Protocol), 309, 619
TCP/IP, 309, 619
tearDown method, TestCase class, 196–199
Technorati web service, 537
Template class, 610
templates
string substitutions using, 610
XSLT, 293–294
tree view widget
terse protocols, 350
test cases, 193–196
test fixtures, 196–199
test function, 163, 164
test suites
creating, 194–196
definition of, 193
in Extreme Programming, 199–205
TestCase class
runTest method, 193–196
setUp method, 196–199
tearDown method, 196–199
testing
assertions, 191–193
in Extreme Programming, 199–205
from Jython, 565–566
in software life cycle, 207–208
test cases, 193–196
test fixtures, 196–199
tests. See comparisons; decisions
text processing
definition of, 175
directory navigation for, 176
files, searching for, 176–177
log files, clipping (filtering), 177
mail, searching, 178
os module functions for, 178–184
regular expressions for, 176, 184–188
uses of, 175–178
ThreadingMixIn class, 339–340
threads
debugging, 399
definition of, 139–140
handling signals with, 216–221
Timeoutsocket module, 401–402
TK GUI toolkit, 209
Tomcat server, 559–560
transactions, DB API, 271–272
Transport Control Protocol (TCP), 309, 619
transport layer, 309
tree view widget
creating, 239–241
definition of, 238
example using, 241–248
645

69.7 Page 687

▲back to top
trigonometric frunctions
trigonometric functions, 415
triple quotes (‘’’...’’’ or “””...”””)
enclosing documentation, 156–157
enclosing multi-line strings, 7–8, 618
True value (1)
definition of, 36, 38
as result of comparisons, 43
truncating numbers, 406
trusted servers, 350
try statement, 55–57
tuples
appending, 40
for arrays, when to use, 419
creating, 30–31
definition of, 30, 619
equality of, 44
last elements of, referencing, 38–39
layers of, 31–33
one-element tuples, 32
ranges of, referencing (slicing), 39–40
reducing (running function on all elements of),
128–129
visiting elements in without loops, 129–130
Twisted framework, 351–353, 620
200 status code, 467
type function
checking parameters using, 68–69
definition of, 14
TypeError, 32, 56, 68
types. See data types
U
UID (Unique ID)
definition of, 620
for IMAP, 330–331
underscore (_), in variable names, 30
Unicode, 620
unittest module, 193
unlink function, os module, 120
unsubscriptable, 68
update statement, SQL, 257–259
URL, 461
urllib module, 496
user agent, 470, 473, 500, 620
user interface, creating from Jython, 550–552.
See also GUIs
users, managing, 431–432
V
validation of XML, 285–287
variables
anonymous, 613
assigning values to, 28
changing values of, 29
characters that can’t be used in, 30
conflicting with names in other contexts,
169–170
copying, 29
definition of, 27–28, 620
scope of, 64–65
words that can’t be used as, 29
verb, HTTP, 462–463, 473, 615
version control system, 430
versions of Python
porting to other versions, 397–398
version 2.4, new features in, 609–611
vertical bar (|)
| (OR between DTD elements), 279
| (preceding optional arguments), 362
visible web server, 464–466
W
wait function, os module, 137
walk function, os.path module, 180, 181,
182–183
Warning error, DB API, 273
watermarking, 390–394
W3C (World Wide Web Consortium), 278
web applications. See also web services
advantages of, 459
CGI (Common Gateway Interface)
definition of, 468–469, 613
environment variables for, 471–473
HTML forms and, 473–479
scripts, running, 469–470
web server’s role in executing scripts, 470–471
definition of, 620
frameworks for, 460
646

69.8 Page 688

▲back to top
web sites
HTTP and, 462–468
modules for, 460
REST architecture and, 460–462
as web services, 536
wiki
definition of, 479–481, 620
markup used by, 480–481
search and replace for, using REST, 503–508
search and replace for, using SOAP, 527–529
search and replace for, using XML-RPC,
518–519
storage used by, 481
web server
role in executing CGI script, 470–471
visible (viewing request and response), 464–466
writing, 463
web services
definition of, 388–389, 493–494, 620
designing, 535–536
documenting web service API, 529–534
etiquette for, 535–536
publicly available, list of, 536–537
REST (REpresentational State Transfer)
Amazon Web Services, 495–497
Amazon wish lists, 497–500
BittyWiki example of, 500–503
definition of, 494–495
when to use, 534–535
wiki search and replace using, 503–508
SOAP (Simple Object Access Protocol)
BittyWiki example using, 525–527
definition of, 520, 619
documentation for, 530
Fault element (errors flagged by), 524–525
Google API example, 520–522
request for, 522–524
response for, 524
when to use, 534
wiki search and replace using, 527–529
standards for, 494
standards for, choosing, 534–535
web applications as, 536
WSDL file defining, 531–534
XML-RPC (Remote Procedure Call)
BittyWiki example using, 514–517
data representation, 512–513
definition of, 508–509, 620
documentation for, 529–530
fault response (errors flagged by), 513–514
introspection API, 530–531
Meerkat API example, 509–511
POST request for, 508, 511–513
response, 513
when to use, 534–535
wiki search and replace using, 518–519
web sites
Amazon.com web service, 537
CGI (Common Gateway Interface), 469
DB API, 273
del.icio.us web service, 537
Eclipse IDE, 565
Flickr web service, 537
floating point numbers, 17, 21
4Suite library, 294
Gadfly database, 260
Glade GUI construction kit, 225
Google web service, 520, 522, 537
Grinder, 541
HSqlDB database, 554
jEdit text editor, 564
J2EE, 561
Jython, 541, 607
LAME project, 364
libgmail project, 331
LiveHTTPHeaders extension, 466
Mac Python, 606
Mail To The Future web service, 537
Meerkat web service, 537
MySQL, 607
open source licensing, 387
PyDev, 565
pyglade module, 606
pygtk module, 606
Python, 171, 605–606
Python API from C, 356
PythonCard, 225, 606
PyUnit facility, 607
pywftk module, 607
PyXML package, 285
RedRobin Jython, 565
REST, 460
RSS (Rich or RDF Site Summary), 297
647

69.9 Page 689

▲back to top
web sites (continued)
web sites (continued)
SOAP library, 520, 522
Speller web service, 537
Technorati web service, 537
Tomcat server, 559
Twisted framework, 351, 402
WebserviceX web service, 537
Web-Sniffer, 466
wftk workflow toolkit, 607
wiki, 481
wxPython toolkit, 210
Xmethods web service, 537
XML, 277
XPath, 282
XSLT namespace, 293
Yahoo! web service, 537
webmail, compared to e-mail, 331
WebserviceX web service, 537
well-formedness of XML, 286, 621
well-known port, 311, 620
wftk workflow toolkit
action queue handler, 456–457
definition of, 435–436
repository, in database
defining and accessing, 438–439
for document management, 441–448
storing records in, 439–446
repository, in directory files
defining and accessing, 436–438
storing records in, 438–439
resources for, 607
workflow trigger, 453–456
whichdb module, 251
while statement, 51–54
whitespace
for code blocks, 49
definition of, 620
widget packing, 222–223
Widget Tree, Glade
creating, 239–241
definition of, 238
example using, 241–248
widgets
creating, 213–214, 228–230
definition of, 213
handling signals, 214–215, 216–221
648
initialization required by, 238
multiple, in one window, 222–223
show method for, 214
signals generated by, 214
tree view widget, 238–241
wiki
BittyWiki example
core library for, 481–484
creating wiki pages from Python session,
483–484
making executable, 493
markup used by, 486–492
resources for, 484–486
REST API document for, 529
REST API for, 500–503
SOAP API document for, 530
SOAP interface for, 525–527
web interface for, 484–493
WSDL proxy for, 533–534
XML-RPC API document for, 529–530
XML-RPC interface for, 514–517
definition of, 479–481, 620
markup used by, 480–481
search and replace for
using REST, 503–508
using SOAP, 527–529
using XML-RPC, 518–519
storage used by, 481
Wikipedia encyclopedia, 479
WikiWords, 480, 481
wildcards in glob patterns, 122.
See also regular expressions
WingIDE, 396
workflow
action queue handler, 456–457
definition of, 432–433
workflow trigger, 453–456
World Wide Web Consortium (W3C), 278
write method, file object, 110
WSDL (Web Services Description Language)
definition of, 620
web service interface defined by, 531–534
wxDesigner, 224
wxGlade, 224
wxPython toolkit, 210, 224

69.10 Page 690

▲back to top
X
%x format specifier, 24, 409
%X format specifier, 24
%#x format specifier, 409
XHTML
definition of, 282
parsing, 283–284
Xmethods web service, 537
XML (eXtensible Markup Language)
compared to DTD, 280
created by Glade, 230–231
definition of, 275–277, 620
libraries for, 285
parsing
with DOM, 288–289, 290–292
with SAX, 287–288, 289, 290, 292–293
standards for, 277
transforming with XSLT, 293–296
validating, 285–287, 621
XPath expressions in, 282
XML schema. See schemas
<?xml ...?> tag, 276, 279
XML validation, 285–287, 621
XML well-formedness, 286, 621
xml.dom DOM processor, 285
xml.dom.minidom DOM implementation, 285,
290–292
xmlproc XML parser, 285, 286–287
XML-RPC (Remote Procedure Call)
BittyWiki example using, 514–517
data representation, 512–513
definition of, 508–509, 620
zxJDBC package
documentation for, 529–530
fault response (errors flagged by), 513–514
introspection API, 530–531
Meerkat API example, 509–511
POST request for, 508, 511–513
response, 513
when to use, 534–535
wiki search and replace using, 518–519
xml.sax package, 290, 292–293
XP (Extreme Programming), test suites for,
199–205
XPath, 277, 282
Xrange, 621
xrange object
accessing in reverse, 610
definition of, 132–133
XSL-FO (Extensible Style Language Formatting
Objects), 277, 621
XSLT (Extensible Style Language for Transforma-
tions)
definition of, 277, 293–294, 621
transforming XML with, 294–296
Y
Yahoo! web service, 537
Z
Zawinski, Jamie (quote about e-mail), 305
0 (False value), 36, 38
0x, preceding hexadecimal literals, 406
zxJDBC package, 553
649