Appendix A. Using Regular Expressions in Apache

A number of the Apache web server's configuration directives permit (or require!) the use of what are called regular expressions. Regular expressions are used to determine if a string, such as a URL or a user's name, matches a pattern.

There are numerous resources that cover regular expressions in excruciating detail, so this appendix is not designed to be a tutorial for their use. Instead, it documents the specific features of regular expressions used by Apache—what's available and what isn't. Even though there are quite a number of regular expression packages, with differing feature sets, there are some commonalities among them. The Perl language, for instance, has a particularly rich set of regular expressions but only a small subset of them are available in the Apache regex library, which is different from Perl's.

Regular expressions, as mentioned, are a language that allows you to determine if a particular string or variable looks like some pattern. For example, you may wish to determine if a particular string is all uppercase, or if it contains at least 3 numbers, or perhaps if it contains the word "monkey" or "Monkey." Regular expressions provide a vocabulary for talking about these sort of tests. Most modern programming languages contain some variety of regular expression library, and they tend to have a large number of things in common, although they may differ in small details.

Apache 1.3 uses a regular expression library called hsregex, so called because it was developed by Henry Spencer. Note that this is the same regular expression library used in egrep, which is the same thing as grep on many Unixish platforms.

Apache 2.0 uses a somewhat more full-featured regular expression library called Perl Compatible Regular Expressions (PCRE), so called because it implements many of the features available in the regular expression engine that comes with the Perl programming language. While this appendix does not attempt to communicate all the differences between these two implementations, you should know that hsregex is a subset of PCRE, as far as functionality goes, so everything you can do with regular expressions in Apache 1.3, you can do in 2.0, but not necessarily the other way around.

To grossly simplify, regular expressions implement two kinds of characters. Some characters mean exactly what they say (for example, a G appearing in a regular expression will usually mean the literal character G), while some characters have special significance (for example, the period (.) will match any character at all—a wildcard character). Regular expressions can be composed of these characters to represent (almost) any desired pattern appearing in a string.