Something like this:
http://perl.plover.com/yak/regex/samples/slide083.html
In other words I want to match successfully on { { foo } { bar} } but not on { { foo } .
I see it's possible in perl, and in .NET. Is it possible in emacs regex?
views:
93answers:
2
+4
A:
No, so far Perl/PCRE and .NET are the only regex flavors that support arbitrary nesting (recursive patterns).
Tim Pietzcker
2010-04-24 14:11:36
Mmkay, you get an upvote even though this is not the answer I wanted.
Cheeso
2010-04-24 15:36:22
Yeah, sorry. But I'm not at all sure if recursive regexes are a Good Thing. They can get complicated enough without that already. A parser combined with a regex looks more sensible to me.
Tim Pietzcker
2010-04-24 17:19:42
Meh! "regular expressions" that have been extended to the point that they can do this are not actually "regular" anymore. Not that this is a bad thing, necessarily, but the evolution of the tool has made the terminology obsolete.
dmckee
2010-04-24 18:10:52
@dmckee, What makes a regular expression, *regular*? By "not regular" do you just mean that the extensions are not universally supported? I'll buy that. But I did not know that recursion, or the ability to do balanced matching, was an extension, or was not supported. Which is why I asked. @Tim, I think you're right, and that's what I will likely do.
Cheeso
2010-04-24 19:05:06
@Cheeso: The term "regular expression" has been stretched over the years. A true regular expression would not allow for backreferences or lookaround since those are not "regular" anymore. Recursiveness is just another extra feature that extends what regular expressions can do - a theoretical computer scientist would probably shudder at the terminology...
Tim Pietzcker
2010-04-24 19:21:54
@Cheeso: as Tim said, the "Regular" actually refers to a very specific thing. And the inability to match nesting depths is actually part of the definition of regular. ( http://en.wikipedia.org/wiki/Regular_language#Deciding_whether_a_language_is_regular ). To be fair though, regular expressions engines have been misnamed for at least a decade or two.
quodlibetor
2010-04-27 20:12:41
Ah, ok, now I see. Thanks for the link. I don't understand any of that stuff, the Myhill-Nerode theorem or the pumping lemma or free monoids, and I don't care to delve into it. But at least now I am clear that there is a formal definition for "regular".
Cheeso
2010-04-27 22:44:47
+2
A:
No, but if you have a particular use case to discuss you'll often find that you don't need regexes. Simple state-machines to match parenthases are pretty simple to write in lisp. Looking at the source of Paredit is a good place to start.
Singletoned
2010-04-24 15:40:09
yes, well in my case the paren parsing is already done. It's within cc-mode, so I can just use `(forward-sexp)` with regex matching on either side. I was just hoping to be able to do it with a single regex.
Cheeso
2010-04-24 19:02:02