Python Cookbook, 2nd Edition

Preface

This book is not a typical O’Reilly book, written as a cohesive manuscript by one or two authors. Instead, it is a new kind of book—a bold attempt at applying some principles of open source development to book authoring. Over 300 members of the Python community contributed materials to this book. In this Preface, we, the editors, want to give you, the reader, some background regarding how this book came about and the processes and people involved, and some thoughts about the implications of this new form.

The Design of the Book

In early 2000, Frank Willison, then Editor-in-Chief of O’Reilly & Associates, contacted me (David Ascher) to find out if I wanted to write a book. Frank had been the editor for Learning Python, which I cowrote with Mark Lutz. Since I had just taken a job at what was then considered a Perl shop (ActiveState), I didn’t have the bandwidth necessary to write another book, and plans for the project were gently shelved. Periodically, however, Frank would send me an email or chat with me at a conference regarding some of the book topics we had discussed. One of Frank’s ideas was to create a Python Cookbook, based on the concept first used by Tom Christiansen and Nathan Torkington with the Perl Cookbook. Frank wanted to replicate the success of the Perl Cookbook, but he wanted a broader set of people to provide input. He thought that, much as in a real cookbook, a larger set of authors would provide for a greater range of tastes. The quality, in his vision, would be ensured by the oversight of a technical editor, combined with O’Reilly’s editorial review process.

Frank and Dick Hardt, ActiveState’s CEO, realized that Frank’s goal could be combined with ActiveState’s goal of creating a community site for open source programmers, called the ActiveState Programmer’s Network (ASPN). ActiveState had a popular web site, with the infrastructure required to host a wide variety of content, but it wasn’t in the business of creating original content. ActiveState always felt that the open source communities were the best sources of accurate and up-to-date content, even if sometimes that content was hard to find.

The O’Reilly and ActiveState teams quickly realized that the two goals were aligned and that a joint venture would be the best way to achieve the following key objectives:

• Creating an online repository of Python recipes by Python programmers for Python programmers

• Publishing a book containing the best of those recipes, accompanied by overviews and background material written by key Python figures

• Learning what it would take to create a book with a different authoring model

At the same time, two other activities were happening. First, those of us at ActiveState, including Paul Prescod, were actively looking for “stars” to join ActiveState’s development team. One of the candidates being recruited was the famous (but unknown to us, at the time) Alex Martelli. Alex was famous because of his numerous and exhaustive postings on the Python mailing list, where he exhibited an unending patience for explaining Python’s subtleties and joys to the increasing audience of Python programmers. He was unknown because he lived in Italy and, since he was a relative newcomer to the Python community, none of the old Python hands had ever met him—their paths had not happened to cross back in the 1980s when Alex lived in the United States, working for IBM Research and enthusiastically using and promoting other high-level languages (at the time, mostly IBM’s Rexx).

ActiveState wooed Alex, trying to convince him to move to Vancouver. We came quite close, but his employer put some golden handcuffs on him, and somehow Vancouver’s weather couldn’t compete with Italy’s. Alex stayed in Italy, much to my disappointment. As it happened, Alex was also at that time negotiating with O’Reilly about writing a book. Alex wanted to write a cookbook, but O’Reilly explained that the cookbook was already signed. Later, Alex and O’Reilly signed a contract for Python in Nutshell.

The second ongoing activity was the creation of the Python Software Foundation. For a variety of reasons, best left to discussion over beers at a conference, everyone in the Python community wanted to create a non-profit organization that would be the holder of Python’s intellectual property, to ensure that Python would be on a legally strong footing. However, such an organization needed both financial support and buy-in from the Python community to be successful.

Given all these parameters, the various parties agreed to the following plan:

• ActiveState would build an online cookbook, a mechanism by which anyone could submit a recipe (i.e., a snippet of Python code addressing a particular problem, accompanied by a discussion of the recipe, much like a description of why one should use cream of tartar when whipping egg whites). To foster a community of authors and encourage peer review, the web site would also let readers of the recipes suggest changes, ask questions, and so on.

• As part of my ActiveState job, I would edit and ensure the quality of the recipes. Alex Martelli joined the project as a co-editor when the material was being prepared for publication, and, with Anna Martelli Ravenscroft, took over as primary editor for the second edition.

• O’Reilly would publish the best recipes as the Python Cookbook.

• In lieu of author royalties for the recipes, a portion of the proceeds from the book sales would be donated to the Python Software Foundation.

The Implementation of the Book

The online cookbook (at http://aspn.activestate.com/ASPN/Cookbook/Python/) was the entry point for the recipes. Users got free accounts, filled in a form, and presto, their recipes became part of the cookbook. Thousands of people read the recipes, and some added comments, and so, in the publishing equivalent of peer review, the recipes matured and grew. While it was predictable that the chance of getting your name in print would get people attracted to the online cookbook, the ongoing success of the cookbook, with dozens of recipes added monthly and more and more references to it on the newsgroups, is a testament to the value it brings to the readers—value which is provided by the recipe authors.

Starting from the materials available on the site, the implementation of the book was mostly a question of selecting, merging, ordering, and editing the materials. A few more details about this part of the work are in the “Organization” section of this Preface.

Using the Code from This Book

This book is here to help you get your job done. In general, you may use the code in this book in your programs and documentation. You do not need to contact us for permission unless you’re reproducing a significant portion of the code. For example, writing a program that uses several chunks of code from this book does not require permission. Selling or distributing a CD-ROM of code taken from O’Reilly books does require permission. Answering a question by citing this book and quoting example code does not require permission. Incorporating a significant amount of code from this book into your product’s documentation does require permission. We appreciate, but do not require, attribution. An attribution usually includes the title, author, publisher, and ISBN. For example: “Python Cookbook, 2d ed., by Alex Martelli, Anna Martelli Ravenscroft, and David Ascher (O’Reilly Media, 2005) 0-596-00797-3.” If you feel your use of code from this book falls outside fair use or the permission given above, feel free to contact us at [e-mail].

Audience

We expect that you know at least some Python. This book does not attempt to teach Python as a whole; rather, it presents some specific techniques and concepts (and occasionally tricks) for dealing with particular tasks. If you are looking for an introduction to Python, consider some of the books described in the Further Reading section of this Preface. However, you don’t need to know a lot of Python to find this book helpful. Chapters include recipes demonstrating the best techniques for accomplishing some elementary and general tasks, as well as more complex or specialized ones. We have also added sidebars, here and there, to clarify certain concepts which are used in the book and which you may have heard of, but which might still be unclear to you. However, this is definitely not a book just for beginners. The main target audience is the whole Python community, mostly made up of pretty good programmers, neither newbies nor wizards. And if you do already know a lot about Python, you may be in for a pleasant surprise! We’ve included recipes that explore some the newest and least well-known areas of Python. You might very well learn a few things—we did! Regardless of where you fall along the spectrum of Python expertise, and more generally of programming skill, we believe you will get something valuable from this book.

If you already own the first edition, you may be wondering whether you need this second edition, too. We think the answer is “yes.” The first edition had 245 recipes; we kept 146 of those (with lots of editing in almost all cases), and added 192 new ones, for a total of 338 recipes in this second edition. So, over half of the recipes in this edition are completely new, and all the recipes are updated to apply to today’s Python—releases 2.3 and 2.4. Indeed, this update is the main factor which lets us have almost 100 more recipes in a book of about the same size. The first edition covered all versions from 1.5.2 (and sometimes earlier) to 2.2; this one focuses firmly on 2.3 and 2.4. Thanks to the greater power of today’s Python, and, even more, thanks to the fact that this edition avoids the “historical” treatises about how you had to do things in Python versions released 5 or more years ago, we were able to provide substantially more currently relevant recipes and information in roughly the same amount of space.

Organization

This book has 20 chapters. Each chapter is devoted to a particular kind of recipe, such as algorithms, text processing, databases, and so on. The 1st edition had 17 chapters. There have been improvements to Python, both language and library, and to the corpus of recipes the Python community has posted to the cookbook site, that convinced us to add three entirely new chapters: on the iterators and generators introduced in Python 2.3; on Python’s support for time and money operations, both old and new; and on new, advanced tools introduced in Python 2.2 and following releases (custom descriptors, decorators, metaclasses). Each chapter contains an introduction, written by an expert in the field, followed by recipes selected from the online cookbook (in some cases—about 5% of this book’s recipes—a few new recipes were specially written for this volume) and edited to fit the book’s formatting and style requirements. Alex (with some help from Anna) did the vast majority of the selection—determining which recipes from the first edition to keep and update, and selecting new recipes to add, or merge with others, from the nearly 1,000 available on the site (so, if a recipe you posted to the cookbook site didn’t get into this printed edition, it’s his fault!). He also decided which subjects just had to be covered and thus might need specially written recipes—although he couldn’t manage to get quite all of the specially written recipes he wanted, so anything that’s missing, and wasn’t on the cookbook site, might not be entirely his fault.

Once the selection was complete, the work turned to editing the recipes, and to merging multiple recipes, as well as incorporating important contents from many significant comments posted about the recipes. This proved to be quite a challenge, just as it had been for the first edition, but even more so. The recipes varied widely in their organization, level of completeness, and sophistication. With over 300 authors involved, over 300 different “voices” were included in the text. We have striven to maintain a variety of styles to reflect the true nature of this book, the book written by the entire Python community. However, we edited each recipe, sometimes quite considerably, to make it as accessible and useful as possible, ensuring enough uniformity in structure and presentation to maximize the usability of the book as a whole. Most recipes, both from the first edition and from the online site, had to be updated, sometimes heavily, to take advantage of new tools and better approaches developed since those recipes were originally posted. We also carefully reconsidered (and slightly altered) the ordering of chapters, and the placement and ordering of recipes within chapters; our goal in this reordering was to maximize the book’s usefulness for both newcomers to Python and seasoned veterans, and, also, for both readers tackling the book sequentially, cover to cover, and ones just dipping in, in “random access” fashion, to look for help on some specific area.

While the book should thus definitely be accessible “by hops and jumps,” we nevertheless believe a first sequential skim will amply repay the modest time you, the reader, invest in it. On such a skim, skip every recipe that you have trouble following or that is of no current interest to you. Despite the skipping, you’ll still get a sense of how the whole book hangs together and of where certain subjects are covered, which will stand you in good stead both for later in-depth sequential reading, if that’s your choice, and for “random access” reading. To further help you get a sense of what’s where in the book, here’s a capsule summary of each chapter’s contents, and equally capsule bios of the Python experts who were so kind as to take on the task of writing the chapters’ “Introduction” sections.

Chapter 1 , introduction by Fred L. Drake, Jr.

This chapter contains recipes for manipulating text in a variety of ways, including combining, filtering, and formatting strings, substituting variables throughout a text document, and dealing with Unicode.

Fred Drake is a member of the PythonLabs group, working on Python development. A father of three, Fred is best known in the Python community for single-handedly maintaining the official documentation. Fred is a co-author of Python & XML (O’Reilly).

Chapter 2 , introduction by Mark Lutz

This chapter presents techniques for working with data in files and for manipulating files and directories within the filesystem, including specific file formats and archive formats such as tar and zip.

Mark Lutz is well known to most Python users as the most prolific author of Python books, including Programming Python, Python Pocket Reference, and Learning Python (all from O’Reilly), which he co-authored with David Ascher. Mark is also a leading Python trainer, spreading the Python gospel throughout the world.

Chapter 3 , introduction by Gustavo Niemeyer and Facundo Batista

This chapter (new in this edition) presents tools and techniques for working with dates, times, decimal numbers, and some other money-related issues.

Gustavo Niemeyer is the author of the third-party dateutil module, as well as a variety of other Python extensions and projects. Gustavo lives in Brazil. Facundo Batista is the author of the Decimal PEP 327, and of the standard library module decimal, which brought floating-point decimal support to Python 2.4. He lives in Argentina. The editors were delighted to bring them together for this introduction.

Chapter 4 , introduction by David Ascher

This chapter includes recipes for many common techniques that can be used anywhere, or that don’t really fit into any of the other, more specific recipe categories.

David Ascher is a co-editor of this volume. David’s background spans physics, vision research, scientific visualization, computer graphics, a variety of programming languages, co-authoring Learning Python (O’Reilly), teaching Python, and these days, a slew of technical and nontechnical tasks such as managing the ActiveState team. David also gets roped into organizing Python conferences on a regular basis.

Chapter 5 , introduction by Tim Peters

This chapter covers techniques for searching and sorting in Python. Many of the recipes explore creative uses of the stable and fast list.sort in conjunction with the decorate-sort-undecorate (DSU) idiom (newly built in with Python 2.4), while others demonstrate the power of heapq, bisect, and other Python searching and sorting tools.

Tim Peters, also known as the tim-bot, is one of the mythological figures of the Python world. He is the oracle, channeling Guido van Rossum when Guido is busy, channeling the IEEE-754 floating-point committee when anyone asks anything remotely relevant, and appearing conservative while pushing for a constant evolution in the language. Tim is a member of the PythonLabs team.

Chapter 6 , introduction by Alex Martelli

This chapter offers a wide range of recipes that demonstrate the power of object-oriented programming with Python, including fundamental techniques such as delegating and controlling attribute access via special methods, intermediate ones such as the implementation of various design patterns, and some simple but useful applications of advanced concepts, such as custom metaclasses, which are covered in greater depth in Chapter 20.

Alex Martelli, also known as the martelli-bot, is a co-editor of this volume. After almost a decade with IBM Research, then a bit more than that with think3, inc., Alex now works as a freelance consultant, most recently for AB Strakt, a Swedish Python-centered firm. He also edits and writes Python articles and books, including Python in a Nutshell (O’Reilly) and, occasionally, research works on the game of contract bridge.

Chapter 7 , introduction by Aaron Watters

This chapter presents Python techniques for persistence, including serialization approaches and interaction with various databases.

Aaron Watters was one of the earliest advocates of Python and is an expert in databases. He’s known for having been the lead author on the first book on Python (Internet Programming with Python, M&T Books, now out of print), and he has authored many widely used Python extensions, such as kjBuckets and kwParsing. Aaron currently works as a freelance consultant.

Chapter 8 , introduction by Mark Hammond

This chapter includes a collection of recipes that assist with the debugging and testing process, from customizing error logging and traceback information, to unit testing with custom modules, unittest and doctest.

Mark Hammond is best known for his work supporting Python on the Windows platform. With Greg Stein, he built an incredible library of modules interfacing Python to a wide variety of APIs, libraries, and component models such as COM. He is also an expert designer and builder of developer tools, most notably Pythonwin and Komodo. Finally, Mark is an expert at debugging even the most messy systems—during Komodo development, for example, Mark was often called upon to debug problems that spanned three languages (Python, C++, JavaScript), multiple threads, and multiple processes. Mark is also co-author, with Andy Robinson, of Python Programming on Win32 (O’Reilly).

Chapter 9 , introduction by Greg Wilson

This chapter covers a variety of techniques for concurrent programming, including threads, queues, and multiple processes.

Greg Wilson writes children’s books, as well as books on parallel programming and data crunching. When he’s not doing that, he’s a contributing editor with Doctor Dobb’s Journal, an adjunct professor in Computer Science at the University of Toronto, and a freelance software developer. Greg was the original driving force behind the Software Carpentry project, and he recently received a grant from the Python Software Foundation to develop Pythonic course material for computational scientists and engineers.

Chapter 10 , introduction by Donn Cave

This chapter includes recipes for a number of common system administration tasks, from generating passwords and interacting with the Windows registry, to handling mailbox and web server issues.

Donn Cave is a software engineer at the University of Washington’s central computer site. Over the years, Donn has proven to be a fount of information on comp.lang.python on all matters related to system calls, Unix, system administration, files, signals, and the like.

Chapter 11 , introduction by Fredrik Lundh

This chapter contains recipes for common GUI tasks, mostly with Tkinter, but also a smattering of wxPython, Qt, image processing, and GUI recipes specific to Jython (for JVM—Java Virtual Machine), Mac OS X, and IronPython (for dotNET).

Fredrik Lundh, also known as the eff-bot, is the CTO of Secret Labs AB, a Swedish Python-focused company providing a variety of products and technologies. Fredrik is the world’s leading expert on Tkinter (the most popular GUI toolkit for Python), as well as the main author of the Python Imaging Library (PIL). He is also the author of Python Standard Library (O’Reilly), which is a good complement to this volume and focuses on the modules in the standard Python library. Finally, he is a prolific contributor to comp.lang.python, helping novices and experts alike.

Chapter 12 , introduction by Paul Prescod

This chapter offers techniques for parsing, processing, and generating XML using a variety of Python tools.

Paul Prescod is an expert in three technologies: Python, which he need not justify; XML, which makes sense in a pragmatic world (Paul is co-author of the XML Handbook, with Charles Goldfarb, published by Prentice Hall); and Unicode, which somehow must address some deep-seated desire for pain and confusion that neither of the other two technologies satisfies. Paul is currently a product manager at Blast Radius.

Chapter 13 , introduction by Guido van Rossum

This chapter covers a variety of network programming techniques, from writing basic TCP clients and servers to manipulating MIME messages.

Guido created Python, nurtured it throughout its infancy, and is shepherding its growth. Need we say more?

Chapter 14 , introduction by Andy McKay

This chapter presents a variety of web-related recipes, including ones for CGI scripting, running a Java servlet with Jython, and accessing the content of web pages.

Andy McKay is the co-founder and vice president of Enfold Systems. In the last few years, Andy went from being a happy Perl user to a fanatical Python, Zope, and Plone expert. He wrote the Definitive Guide to Plone (Apress) and runs the popular Zope discussion site, http://www.zopezen.org.

Chapter 15 , introduction by Jeremy Hylton

This chapter provides recipes for using Python in simple distributed systems, including XML-RPC, CORBA, and Twisted’s Perspective Broker.

Jeremy Hylton works for Google. In addition to young twins, Jeremy’s interests including programming language theory, parsers, and the like. As part of his work for CNRI, Jeremy worked on a variety of distributed systems.

Chapter 16 , introduction by Paul F. Dubois

This chapter contains Python techniques that involve program introspection, currying, dynamic importing, distributing programs, lexing and parsing.

Paul Dubois has been working at the Lawrence Livermore National Laboratory for many years, building software systems for scientists working on everything from nuclear simulations to climate modeling. He has considerable experience with a wide range of scientific computing problems, as well as experience with language design and advanced object-oriented programming techniques.

Chapter 17 , introduction by David Beazley

This chapter offers techniques for extending Python and recipes that assist in the development of extensions.

David Beazley’s chief claim to fame is SWIG, an amazingly powerful hack that lets one quickly wrap C and other libraries and use them from Python, Tcl, Perl, and myriad other languages. Behind this seemingly language-neutral tool lies a Python supporter of the first order, as evidenced by his book, Python Essential Reference (New Riders). David Beazley is a fairly sick man (in a good way), leading us to believe that more scarily useful tools are likely to emerge from his brain. He’s currently inflicting his sense of humor on computer science students at the University of Chicago.

Chapter 18 , introduction by Tim Peters

This chapter provides a collection of fascinating and useful algorithms and data structures implemented in Python.

See the discussion of Chapter 5 for information about Tim Peters.

Chapter 19 , introduction by Raymond Hettinger

This chapter (new in this edition) contains recipes demonstrating the variety and power of iterators and generators—how Python makes your loops’ structures simpler, faster, and reusable.

Raymond Hettinger is the creator of the itertools package, original proposer of generator expressions, and has become a major contributor to the development of Python—if you don’t know who originated and implemented some major novelty or important optimization in the 2.3 and 2.4 releases of Python, our advice is to bet it was Raymond!

Chapter 20 , introduction by Raymond Hettinger

This chapter (new in this edition) provides an in-depth look into the infrastructural elements which make Python’s OOP so powerful and smooth, and how you can exploit and customize them for fun and profit. From handy idioms for building properties, to aliasing and caching attributes, all the way to decorators which optimize your functions by hacking their bytecode and to a factory of custom metaclasses to solve metatype conflicts, this chapter shows how, while surely “there be dragons here,” they’re the wise, powerful and beneficent Chinese variety thereof...!

See the discussion of Chapter 19 for information about Raymond Hettinger.

Further Reading

There are many texts available to help you learn Python or refine your Python knowledge, from introductory texts all the way to quite formal language descriptions.

We recommend the following books for general information about Python (all these books cover at least Python 2.2, unless otherwise noted):

• Python Programming for the Absolute Beginner, by Michael Dawson (Thomson Course Technology), is a hands-on, highly accessible introduction to Python for people who have never programmed.

• Learning Python, by Mark Lutz and David Ascher (O’Reilly), is a thorough introduction to the fundamentals of Python.

• Practical Python, by Magnus Lie Hetland (APress), is an introduction to Python which also develops, in detail, ten fully worked out, substantial programs in many different areas.

• Dive into Python, by Mark Pilgrim (APress), is a fast-paced introduction to Python for experienced programmers, and it is also freely available for online reading and downloading (http://diveintopython.org/).

• Python Standard Library, by Fredrik Lundh (O’Reilly), provides a use case for each module in the rich library that comes with every standard Python distribution (in the current first edition, the book only covers Python up to 2.0).

• Programming Python, by Mark Lutz (O’Reilly), is a thorough rundown of Python programming techniques (in the current second edition, the book only covers Python up to 2.0).

• Python Essential Reference, by David Beazley (New Riders), is a quick reference that focuses on the Python language and the core Python libraries (in the current second edition, the book only covers Python up to 2.1).

• Python in a Nutshell, by Alex Martelli (O’Reilly), is a comprehensive quick reference to the Python language and the key libraries used by most Python programmers.

In addition, several more special-purpose books can help you explore particular aspects of Python programming. Which books you will like best depends a lot on your areas of interest. From personal experience, the editors can recommend at least the following:

• Python and XML, by Christopher A. Jones and Fred L. Drake, Jr. (O’Reilly), offers thorough coverage of using Python to read, process, and transform XML.

• Jython Essentials, by Samuele Pedroni and Noel Rappin (O’Reilly), is the authoritative book on Jython, the port of Python to the JVM. Particularly useful if you already know some (or a lot of) Java.

• Game Programming with Python, by Sean Riley (Charles River Media), covers programming computer games with Python, all the way from advanced graphics to moderate amounts of “artificial intelligence.”

• Python Web Programming, by Steve Holden (New Riders), covers building networked systems using Python, with introductions to many other related technologies (databases, HTTP, HTML, etc.). Very suitable for readers with none to medium experience with these fields, but has something to teach everyone.

In addition to these books, other important sources of information can help explain some of the code in the recipes in this book. We’ve pointed out the information that seemed particularly relevant in the “See Also” sections of each recipe. In these sections, we often refer to the standard Python documentation: most often the Library Reference, sometimes the Reference Manual, and occasionally the Tutorial. This documentation is freely available in a variety of forms:

• On the python.org web site (at http://www.python.org/doc/), which always contains the most up-to-date documentation about Python.

• On the pydoc.org web site (at http://pydoc.org/), accompanied by module-by-module documentation of the standard library automatically generated by the very useful pydoc tool.

• In Python itself. Recent versions of Python boast a nice online help system, which is worth exploring if you’ve never used it. Just type help( ) at the interactive Python interpreter prompt to start exploring.

• As part of the online help in your Python installation. ActivePython’s installer, for example, includes a searchable Windows help file. The standard Python distribution currently includes HTML pages, but there are plans to include a similar Windows Help file in future releases.

We have not included specific section numbers in our references to the standard Python documentation, since the organization of these manuals can change from release to release. You should be able to use the table of contents and indexes to find the relevant material. For the Library Reference, in particular, the Module Index (an alphabetical list of all standard library modules, each module name being a hyperlink to the Library Reference documentation for that module) is invaluable. Similarly, we have not given specific pointers in our references to Python in a Nutshell: that book is still in its first edition (covering Python up to 2.2) at the time of this writing, but by the time you’re reading, a second edition (covering Python 2.3 and 2.4) is likely to be forthcoming, if not already published.

Conventions Used in This Book

Pronouns: the first person singular is meant to convey that the recipe’s or chapter introduction’s author is speaking (when multiple credits are given for a recipe, the author is the first person credited); however, even such remarks have at times had to be edited enough that they may not reflect the original author’s intended meaning (we, the editors, tried hard to avoid that, but we know we must have failed in some cases, since there were so many remarks, and authorial intent was often not entirely clear). The second person is meant to refer to you, the reader. The first person plural collectively indicates you, the reader, plus the recipe’s author and co-authors, the editors, and my friend Joe (hi Joe!)—in other words, it’s a very inclusive “we” or “us.”

Code: each block of code may indicate a complete module or script (or, often, a Python source file that is usable both as a script and as a module), an isolated snippet from some hypothetical module or script, or part of a Python interactive interpreter session (indicated by the prompt >>>).

The following typographical conventions are used throughout this book:

Italic for commands, filenames, for emphasis, and for first use of a term.

Constant width for general code fragments and keywords (mostly Python ones, but also other languages, such as C or HTML, where they occur). Constant width is also used for all names defined in Python’s library and third-party modules.

Constant width bold is used to emphasize particular lines within code listings and show output that is produced.

How to Contact Us

We have tested and verified all the information in this book to the best of our abilities, but you may find that some features have changed, or that we have let errors slip through the production of the book. Please let us know of any errors that you find, as well as any suggestions for future editions, by writing to:

O’Reilly Media

1005 Gravenstein Highway North

Sebastopol, CA 95472

[phone] (in the United States or Canada)

[phone] (international/local)

[phone] (fax)

We have a web site for the book, where we’ll list examples, errata, and any plans for future editions. You can access this page at:

http://www.oreilly.com/catalog/pythoncook2

To ask technical questions or comment on the book, send email to:

[e-mail]

For more information about our books, conferences, Resource Centers, and the O’Reilly Network, see our web site at:

http://www.oreilly.com/

The online cookbook from which most of the recipes for this book were taken is available at:

http://aspn.activestate.com/ASPN/Cookbook/Python

Safari® Enabled

When you see a Safari Enabled icon on the cover of your favorite technology book, that means the book is available online through the O’Reilly Network Safari Bookshelf.

Safari offers a solution that’s better than e-books. It’s a virtual library that lets you easily search thousands of top tech books, cut and paste code samples, download chapters, and find quick answers when you need the most accurate, current information. Try it for free at http://safari.oreilly.com.

Acknowledgments

Most publications, from mysteries to scientific papers to computer books, claim that the work being published would not have been possible without the collaboration of many others, typically including local forensic scientists, colleagues, and children, respectively. This book makes this claim to an extreme degree. Most of the words, code, and ideas in this volume were contributed by people not listed on the front cover. The original recipe authors, readers who submitted useful and insightful comments to the cookbook web site, and the authors of the chapter introductions, are the true authors of the book, and they deserve the credit.

David Ascher

The software that runs the online cookbook was the product of Andy McKay’s constant and diligent effort. Andy was ActiveState’s key Zope developer during the online data-collection phase of this project, and one of the key developers behind ASPN (http://aspn.activestate.com), ActiveState’s content site, which serves a wide variety of information for and by programmers of open source languages such as Python, Perl, PHP, Tcl, and XSLT. Andy McKay used to be a Perl developer, by the way. At about the same time that I started at ActiveState, the company decided to use Zope to build what would become ASPN. In the years that followed, Andy has become a Zope master and somewhat of a Python fanatic (without any advocacy from me!), and is currently a Zope and Plone author, consultant and entrepreneur. Based on an original design that I put together with Diane Mueller, also of ActiveState, Andy single-handedly implemented ASPN in record time, then proceeded to adjust it to ever-changing requirements for new features that we hadn’t anticipated in the early design phase, staying cheerful and professional throughout. It’s a pleasure to have him as the author of the introduction to the chapter on web recipes. Since Andy’s departure, James McGill has taken over as caretaker of the online cookbook—he makes sure that the cookbook is live at all hours of the day or night, ready to serve Pythonistas worldwide.

Paul Prescod, then also of ActiveState, was a kindred spirit throughout the project, helping with the online editorial process, suggesting changes, and encouraging readers of comp.lang.python to visit the web site and submit recipes. Paul also helped with some of his considerable XML knowledge when it came to figuring out how to take the data out of Zope and get it ready for the publication process.

The last activator I’d like to thank, for two different reasons, is Dick Hardt, founder and CEO of ActiveState. The first is that Dick agreed to let me work on the cookbook as part of my job. Had he not, I wouldn’t have been able to participate in it. The second reason I’d like to thank Dick is for suggesting at the outset that a share of the book royalties go to the Python Software Foundation. This decision not only made it easier to enlist Python users into becoming contributors but has also resulted in some long-term revenue to an organization that I believe needs and deserves financial support. All Python users will benefit.

Writing a software system a second time is dangerous; the “second-system” syndrome is a well-known engineering scenario in which teams that are allowed to rebuild systems “right” often end up with interminable, over-engineered projects. I’m pleased to say that this didn’t happen in the case of this second edition, for two primary reasons. The first was the decision to trim the scope of the cookbook to cover only truly modern Python—that made the content more manageable and the book much more interesting to contemporary audiences. The second factor was that everyone realized with hindsight that I would have no time to contribute to the day-to-day editing of this second edition. I’m as glad as ever to have been associated with this book, and pleased that I have no guilt regarding the amount of work I didn’t contribute. When people like Alex and Anna are willing to take on the work, it’s much better for everyone else to get out of the way.

Finally, I’d like to thank the O’Reilly editors who have had a big hand in shaping the cookbook. Laura Lewin was the original editor for the first edition, and she helped make sure that the project moved along, securing and coordinating the contributions of the introduction authors. Paula Ferguson then took the baton, provided a huge amount of precious feedback, and copyedited the final manuscript, ensuring that the prose was as readable as possible given the multiplicity of voices in the book. Jonathan Gennick was the editor for the second edition, and as far as I can tell, he basically let Alex and Anna drive, which was the right thing to do. Another editor I forgot to mention last time was Tim O’Reilly, who got more involved in this book than in most, in its early (rough) phases, and provided very useful input.

Each time I review this acknowledgments section, I can’t help but remember O’Reilly’s Editor-in-Chief at the inception of the project, Frank Willison. Frank died suddenly on a black day, July 30, 2001. He was the person who most wanted to see this book happen, for the simple reason that he believed the Python community deserved it. Frank was always willing to explore new ideas, and he was generous to a fault. The idea of a book with over a hundred authors would have terrified most editors. Frank saw it as a challenge and an experiment. I still miss Frank.

Alex Martelli

I first met Python thanks to the gentle insistence of a former colleague, Alessandro Bottoni. He kept courteously repeating that I really should give Python a try, in spite of my claims that I already knew more programming languages than I knew what to do with. If I hadn’t trusted his technical and aesthetic judgment enough to invest the needed time and energy on the basis of his suggestion, I most definitely wouldn’t be writing and editing Python books today. Thanks for your well-placed stubbornness, Alessandro!

Of course, once I tasted Python, I was irretrievably hooked—my lifelong taste for very high-level (often mis-named “scripting”) languages at last congealed into one superb synthesis. Here, at long last, was a language with the syntactic ease of Rexx (and then some), the semantic simplicity of Tcl (and then some), the intellectual rigor of Scheme (and other Lisp variants), and the awesome power of Perl (and then some). How could I resist? Still, I do owe a debt to Mike Cowlishaw (inventor of Rexx), who I had the pleasure of having as a colleague when I worked for IBM Research, for first getting me hooked on scripting. I must also thank John Ousterhout and Larry Wall, the inventors of Tcl and Perl, respectively, for later reinforcing my addiction through their brainchildren.

Greg Wilson first introduced me to O’Reilly, so he must get his share of thanks, too—and I’m overjoyed at having him as one of the introduction authors. I am also grateful to David Ascher, and several people at O’Reilly, for signing me up as co-editor of the first edition of this book and supporting so immediately and enthusiastically my idea that, hmmm, the time had sure come for a second edition (in dazed retrospect, I suspect what I meant was mostly that I had forgotten how deuced much work it had been to do the first one . . . and failed to realize that, with all the new materials heaped on ActiveState’s site, as well as Python’s wonderful progress over three years, the second edition would take more work than the first one. . . !).

I couldn’t possibly have done the job without an impressive array of technology to help me. I don’t know the names of all the people I should thank for the Internet, ADSL, and Google’s search engines, which, together, let me look things up so easily—or for many of the other hardware and software technologies cooperating to amplify my productivity. But, I do know I couldn’t have made it without Theo de Raadt’s OpenBSD operating system, Steve Jobs’ inspiration behind Mac OS X and the iBook G4 on which I did most of the work, Bram Moolenaar’s VIM editor, and, of course, Guido van Rossum’s Python language. So, I’ll single out Theo, Steve, Bram, and Guido for special thanks!

Nor, as any book author will surely confirm, could I have done it without patience and moral support from friends and family—chiefly my children Lucio and Flavia, my sister Elisabetta, my father Lanfranco. But the one person who was truly indispensable to this second edition was my wife and co-editor Anna. Having reconnected (after many years apart) thanks to Python, taken our honeymoon at the Open Source Convention, given a joint Lightning Talk about our “Pythonic Marriage,” maybe I should have surmised how wonderful it would be to work so closely with her, day in and day out, on such a large and complex joint project. It was truly incredible, all the way through, fully including the heated debates about this or that technical or organizational point or exact choice of wording in delicate cases. Throughout the effort and the stress, her skill, her love, her joy, always shined through, sustained me, and constantly renewed my energies and my determination. Thanks, Anna!

Anna Martelli Ravenscroft

I discovered Python about two years ago. I fell in love, both with Python and (concurrently) with the martelli-bot. Python is a language that is near to my heart, primarily because it is so quickly usable. It doesn’t require you to become a hermit for the next four years in order to do anything with the language. Thank you to Guido. And thanks to the amazing Python community for providing such a welcoming atmosphere to newcomers.

Working on this book was quite the learning experience for me. Besides all the Python code, I also learned both XML and VI, as well as reacquainting myself with Subversion. Thanks go to Holger Krekel and codespeak, for hosting our subversion repository while we travelled. Which brings us to a group of people who deserve special thanks: our reviewers. Holger Krekel, again, was exceptionally thorough, and ensured, among other things, that we had solid Unicode support. Raymond Hettinger gave us a huge amount of valuable, detailed insight throughout, particularly where iterators and generators were concerned. Both Raymond and Holger often offered alternatives to the presented “solutions” when warranted. Valentino Volonghi pointed out programming style issues as well as formatting issues and brought an incredible amount of enthusiasm to his reviews. Ryan Alexander, a newcomer to Python with a background in Java, provided extremely detailed recommendations on ordering and presenting materials (recipes and chapters), as well as pointing out explanations that were weak or missing altogether. His perspective was invaluable in making this book more accessible and useful to new Pythonistas. Several other individuals provided feedback on specific chapters or recipes, too numerous to list here. Your work, however, is greatly appreciated.

Of course, thanks go to my husband. I am amazed at Alex’s patience with questions (and I questioned a lot). His dedication to excellence is a co-author’s dream. When presented with feedback, he consistently responded with appreciation and focus on making the book better. He’s one of the least ego-istical writers I’ve ever met.

Thank you to Dan, for encouraging my geekiness by starting me on Linux, teaching me proper terminology for the stuff I was doing, and for getting me hooked on the Internet. And finally, an extra special thanks to my children, Inanna and Graeme, for their hugs, understanding, and support when I was in geekmode, particularly during the final push to complete the book. You guys are the best kids a mother could wish for.