But I would like to find a mailing list or newsgroup where I can ask
questions about regexps (when things don't work), not specifically
dealing with Python. When I have Python-regexp questions, I'll post
them here of course.
I don't know of any regex-specific newsgroups or mailing lists, but most
old unix hands are pretty good at them, so asking in comp.unix.questions
might be worth trying. Perl makes extensive use of regex, so I'll bet
you'll find some regex gurus on comp.lang.perl too.
Try doing a search on groups.google.com for regex and see what
newsgroups pop up the most with recent dates, then ask your questions
there. Keep in mind that there are many different regex libraries out
there, and they don't all accept exactly the same syntax. Python and
Perl use the same library (I believe).
But, on the other hand, we're a pretty friendly group on
comp.lang.python, and I know there's a bunch of regex wizards who hang
out here as well, so why not just ask here.
One Python-specific tip is to get into the habit of ALWAYS using the r''
style raw strings for regex. It makes life so much simplier.
Joe Francia said:
First, be sure you actually need regexes. See if you can achieve the
same results using string methods. String methods are faster to run,
easier to write, and much easier to debug.
Regex's are an extremely powerful tool, and anybody who's serious about
programming (especially if you do text processing) should have a solid
mastery of them. That being said, Joe is certainly right about them not
always being the best tool for the job. Chainsaws and scalpels are both
useful cutting tools, but hardly interchangable in all applications.
Regex's can lead to extremely fast and compact code, but then can also
lead to stuff which is impossible for anybody to understand 6 months
later.
Also, Joe's comment that string methods are faster to run is a
half-truth. Regex's get compiled. If you use the all-in-one step regex
methods, you compile the expression every time you use it. Compiling is
expensive. But, if you pre-compile the expression, using it is very
fast. Doing the compile step once, and re-using the result can be very
fast. A single regex that looks for a complex pattern may well be
faster than a series of string methods tied together with conditional
logic. Your mileage may vary, so if speed is important to you, try
different ways and profile it.
In any case, for the vast majority of stuff, speed is just not an issue,
and you should pick your weapon based more on clarity and simplicity of
the resulting code than on raw speed.
