Mastering Regular Expressions 208
Simon P. Chappell writes "Classics are funny things, especially in the world of books. There are books that people say "should' be classics (I'll refrain from mentioning names to protect the pretentious) and then there are books that people are too busy actually using to get around to listing as classics. Mastering Regular Expressions, now in it's third edition, is in the second group. It's one of those books that you see on desks in computer departments the world over. This is a real "doers" book." Read the rest of Simon's review.
| Mastering Regular Expressions | |
| author | Jeffrey E.F. Friedl |
| pages | 515 (31 page index) |
| publisher | O'Reilly |
| rating | 11 out of 10 |
| reviewer | Simon P. Chappell |
| ISBN | 0596528124 |
| summary | A classic of modern computer literature. |
This is a book for programmers; managers, project managers and architects need not apply. If you talk about code instead of writing it and have teams of programmers report to you, then consider buying this book and giving it to them. If you're a technical lead or lead programmer, then shame on you if an earlier edition of this book isn't already on your shelves! The majority of examples are written using Perl, but if you can read basic Perl (Pidgin Perl, perhaps?) then you'll be fine with the examples. Programmers in PHP, Java, .NET and Ruby also have dedicated sections of the book, so it's very inclusive and almost platform agnostic.
The book has ten chapters divided into two parts. Chapters one through six are what Mr. Friedl calls the "story" of regular expressions. Chapters seven through ten are an examination of the specific regular expression capabilities of Perl, Java, .NET and PHP.
Chapter one is an introduction to regular expressions. At only 33 pages, you might think that it would be shallow, but rather, it is knowledge dense. The examples in the first chapter use egrep extensively. This makes a lot of sense as it's an advanced tool, easy to use and freely available for most modern operating systems.
Chapter two builds on this introduction with extended introductory examples. These are written in Perl (again, simple and easy to follow), but there is no doubt that the regular expressions are the stars of the show around here. The examples are small Perl programs, but their benefit is that Mr. Friedl talks the reader through the process of creating each of them. This is more useful than just presenting example programs, because with just pure examples, you are out of luck if your specific problem is not covered. With this approach, you're coached towards thinking in regular expressions and are more equipped to address your personal regular expression needs.
Chapter three provides an overview of regular expression features and flavors. It starts with a historical view of the development of regular expressions, including a few asides about the influence that the earlier versions of the book have had on that development. After that, the chapter uses a search and replace example to demonstrate some of the differences between flavors of regular expression capabilities provided by different programming languages. Strings, Unicode and metacharacters round out this overview.
Strap yourself in for chapter four; it's time to talk about the computer science that makes all of that matching work. If you didn't know the difference between an NFA and a DFA regular expression engine before you start this chapter, you most certainly will by the end of it. At first sight, it might seem that this is chapter for the pure propeller heads amongst us. While there is much theory here, it's all presented in the light of how your regular expression engine is trying to do what you asked. By understanding the approaches to regular expression processing, we can learn to help ourselves. We help ourselves when we write regular expressions that run faster and use less memory. We write better regular expressions when we understand the consequences of what we write. For example, the oft written ".*" (dot star) seems like a great way to ignore a bunch of stuff in the middle of an expression, but such simplistic use is just waiting to bite you. This chapter explains why and how to deal with the situations where you'd be tempted to use simplistic expressions and how just a little extra thought can bring you the behavior you want.
Chapter five is a practical counterpoint to the previous theory chapter. Here, Mr. Friedl discusses practical regular expression techniques. There are a number of short examples, before he works through medium sized HTML processing examples and finished up with a look at processing Comma Separated Value (CSV) data.
Chapter six is efficiency. Your regular expression can be as correct as you like, but if it takes what seems like eternity to run, then it's of little use. This chapter mostly addresses NFA based engines, because they have the greatest variability based on how the regular expression is written.
Chapters seven through ten cover the specifics of using regular expressions in Perl, Java, .NET and PHP. They're well written and cover everything you need to apply the content of the first six chapters to your programming language of choice.
Everything about this book is great. This is the kind of book that O'Reilly built its reputation with. A master of the subject matter, writing in a clear, easily understood manner, leaving the reader educated and able to operate comfortably with the subject matter. I may not be a regular expression guru, but I feel that I have a much better grasp of the fundamentals that I would need if I did want to be such a guru.
Mr. Friedl is to be commended for his clear explanations of what is, in all reality, much more complex computer science than many of us are used to dealing with. The fact that his explanations are highly readable and enjoyable is a significant bonus.
There is a website for the book, regex.info and a blog at regex.info/blog, where Mr. Friedl has some wonderful photographs of Japanese gardens with their autumn colors. (Nothing to do with regular expressions, but they appealed to my inner photographer.)
Lastly, while the book is not intended to be an encyclopedia of regular expressions, all of the examples are very relevant to programmers needs and this book can easily serve that reference role.
At the risk of sounding like some kind of O'Reilly shill or a relative of Mr. Friedl, I must report that I don't think that I found a single thing I didn't like about this book.
This is a classic of the first order. Nail it to your desk unless you want to be constantly retrieving it from your co-workers. If I might be permitted a Spinal Tap reference, this one goes to eleven. If you ever use regular expressions, are thinking of using regular expressions or are in the same room as a regular expression, then you need this book.
You can purchase Mastering Regular Expressions from bn.com. Slashdot welcomes readers' book reviews -- to see your own review here, read the book review guidelines, then visit the submission page.
I bought/read this a couple months ago. (Score:5, Interesting)
I read through the whole thing as if it were a novel, and picked up more than a few new things about regexes.
Very handy book, both to read through to really learn how regexes work, and as a day-to-day reference. The score of 11/10 given by the reviewer is bang on.
Re:Agreed! (Score:2, Interesting)
Now that NeXT acquired Apple, the web is indispensible, and the BSD that drives the Mac is settled, I now use regular expressions as in this book. They are not bad, but not the only way, either. Simply put, they won. There will never be another set of regular expressions for 1000 years now, but dont forget, there is More than One Way to Do Things.
an anecdote caused by this good book (Score:4, Interesting)
Like I kid at christmas, I immediately went nuts on my next project with \G and the lookaround operator(s).
Sadly, when a big bundle of code I wrote was delivered to a team in a city on another very large eastern continent, no one could understand what I had written, so they deleted my nifty \G loops and replaced it all with a crappy first-year-college-grad-non-indented parsing state machine using gotos. The complaint was not that I went nuts with regex, but that I was using NONSTANDARD perl version which supported them (instead of their ancient version!), and that it was my duty to deliver a tool using standard versions. I was most angry at the fact that they just replaced the code with a buggy state machine, and then asked me to debug another problem caused by their mess because it was my tool originally. Ugh!
Anyway, my point is: (perl) regex are a far richer tool than meets the eye, but beware The Boneheads: the people who refuse to learn something new that could make their life easier and cling to the old way. Gawd forbid someone learn something new on the job.
Sigh. I was hoping at least ONE programmer over there would have shared my enthusiasm for \G.
Re:For Programmers? NOT! (Score:3, Interesting)
that pissed me off..
A poorly stated case (Score:5, Interesting)
So regular expressions are evil because they're too hard to maintain? If that's you're argument, you need to come up with an alternative that isn't time consuming to code and doesn't require advanced skills that are difficult to master. Good programmers don't hand code fancy solutions any more often than they have to. They rely on well-documented, well-tested language features and APIs. Which describes Perl regular expressions to a T, whatever their shortcomings.
Anyway, Perl regular expressions don't have to be "line noise". That's just the way sloppy people are used to coding. Perl actually allows you to create a clearly formated regular expression in which the structure is pretty obvious, with a little commenting. It does this by providing high-level metacharacters, and by allowing you to use blanks for formatting instead of representing blanks.
Topical plug: Regex Powertoy (Score:4, Interesting)
Give a try to my web-based tool, Regex Powertoy [powertoy.org]. Its interface is all DHTML/CSS/Javascript, but requires a hidden Java (1.5) applet for the advanced and steppable regex engine.
Given that Java core, there are options for adding/removing usual Java literal escaping, which in Java code means lotsa backslashes. Not all Perl advanced features are supported.
I hadn't considered a pick for awk/sed/bash syntax limits/conversion but will consider it. Any handy reference to how their syntax differs from Perl/Java? (The thing that usu. bites me with sed is escaping of parentheses.)
I'll say it again (Score:4, Interesting)
At my previous job (web-based custom market research) we did hundreds of web surveys which had on the average some 400 data points per survey. These had distinct variable names, etc. and were built 100% by hand when I was hired in the company some time in 2002. My first survey project was a disaster, it took me about 20 hours from the final approved survey document to the dynamic version. The process was riddled with manual steps that created an infinite amount of room for errors.
Enter regular expressions.
While fiddling with BBEdit Pro I finally decided to take a shot at regular expressions. After an hour or so of experimenting I started writing a few filters that allowed me to cut down the turnaround from 20 hours per survey to a little over 10 hours. When I got to the point in which I wasn't able to figure things out from the BBEdit documentation and he web, I convinced the boss to buy me Mastering Regular Expressions.
Within the first 50 pages, I had picked up on additional regular expressions concepts that allowed me to eventually cut down the turnaround per survey to less than 8 hours. That's not 8 hours programming, that's 8 hours from the moment the approved survey is handed over to programming to the moment it passes QA checks and is considered ready to go live.
This was a $50 or so book, and it saved us thousands of dollars over the four years I worked at that company. Of course, my reward for saving the company all that money was to lay me off, and I "forgot" to leave instructions on how to use the text filters, so I imagine my replacement is right now writing surveys by hand.
Some of the things that proved to be killer uses for regular expressions within that context:
1. The approved survey would have specific variables that the analysts would need to keep for importing into SPSS later down the process. A text filter picks up those variables and generates a unique list of every variable needed for he survey. The variables are named with specific patterns, so you know which ones are strings, integers, etc.
2. Now that we have a list of variables, it means we can quickly generate the CREATE TABLE statement for the survey data. What used to be done by copying and pasting 400 times is (was?) now done by highlighting the text and running a macro. The output is the SQL command you need.
3. Since you already have the list of variables, you can generate the 400 statements needed to read each form variable into its proper variable in the asp code.
4. The same way you can generate the hidden form fields that you need.
5. The same way you can generate the INSERT statement to send your data to he database.
Little things like that. Eliminating all that copying and pasting really cut down on the QA overhead per project.