Regex lookahead/lookbehind comments

Question

I have a snippet from a config file that I need to be able to match the specified string quote contents, but only when they're not commented out, here's my current regex:

(?<!=#)test\.this\.regex\s+\"(.*?)\"

I feel like this should work? I read it like this:

(?<!=#) lookbehind to make sure it's not preceded by a #

test\.this\.regex\s+\"(.*?)\" matches test.this.regex "sup1"

Here is the config snippet

    test.this.regex "sup1" hi |sup1| # test.this.regex "sup3" hi |sup3|
# test.this.regex "sup2" do |sup2|
    test.this.regex "sup2" do |sup2|

But my regex matches all 4 times:

Match 1
1.  sup1
Match 2
1.  sup3
Match 3
1.  sup2
Match 4
1.  sup2

There seems to be a space after `#` that you haven't accounted for in your regex. — Anirudh Ramanathan, May 17 '14 at 05:12
The syntax of a negative lookbehind is `(?<!...)`, not `(?<!=...)` — Casimir et Hippolyte, May 17 '14 at 05:14
@anubhava I'm using this http://rubular.com/ since it supports lookaheads/lookbehinds — bruchowski, May 17 '14 at 05:18
@DarkCthulhu that is definitely the cause, but how do I check for anything between the `#` and the string I want to match? I tried .* after `#`, and before `test`, neither works — bruchowski, May 17 '14 at 05:27
@bruchowski: it's always good to tag with what language. Tagged [tag:ruby], [tag:rubular], [tag:negative-lookbehind]. Retag as wished. — smci, May 17 '14 at 05:37

anubhava · Accepted Answer · 2014-05-17T05:41:56.697

0

You can use this PCRE regex:

/(?># *(*SKIP)(*FAIL)|(?:^|\s))test\.this\.regex\s+\"[^"]*\"/

Working Demo

(*FAIL) behaves like a failing negative assertion and is a synonym for (?!)
(*SKIP) defines a point beyond which the regex engine is not allowed to backtrack when the subpattern fails later
(*SKIP)(*FAIL) together provide a nice alternative of restriction that you cannot have a variable length lookbehinf in above regex.

UPDATE: Not sure whether ruby supports (*SKIP)(*FAIL) so giving this alternative version:

(?:# *test\.this\.regex\s+\"[^"]*\"|\b(test\.this\.regex\s+\"[^"]*\"))

And look for non-empty matched group #1.

Working Demo 2

edited May 17 '14 at 05:41

answered May 17 '14 at 05:24

anubhava

761,203
64
569
643

This breaks if there is more than one space after the # though – bruchowski May 17 '14 at 05:28
Clarify this in your question and that's precisely the reason why I asked which language you use? – anubhava May 17 '14 at 05:32
I thought I did in the first sentence, sorry if you missed that; I also noted I was using Ruby's regex implementation. It should match if there is anything between the `#` and `test.this.regex`, since that means it's a comment and I want to disregard that match – bruchowski May 17 '14 at 05:34
See updated answer. Your question needs to be tagged with language/tool in addition to regex actually. – anubhava May 17 '14 at 05:35
I had this before, `(#)?.*test\.this\.regex\s+\"(.*)\"` but I was hoping I could match just the strings I wanted with only one capture group – bruchowski May 17 '14 at 05:43
1

`(#)?.*test\.this\.regex\s+\"(.*)\"` won't help since it will match both strings and you **cannot have** dynamic length in lookbehind. – anubhava May 17 '14 at 05:45
`cannot have dynamic length in lookbehind` that's the missing piece, thanks, I didn't know that – bruchowski May 17 '14 at 05:46
I believe other than .NET no other language supports dynamic length `lookbehinds`. – anubhava May 17 '14 at 05:48

Cary Swoveland · Answer 2 · 2014-05-17T19:54:52.540

If you question is embodied in the first sentence (and not specifically about lookarounds) why don't you just use String#split with your regex less the lookbehind?

def doit(str)
  r = /test\.this\.regex\s+\"(.*?)\"/
  str.split('#').first[r,1]
end

doit('test.this.regex "sup1" hi |sup1| # test.this.regex "sup3" hi |sup3|')
  #=> "sup1"
doit('# test.this.regex "sup2" do |sup2|')
  #=> nil
doit('test.this.regex "sup2" do |sup2|')
  #=> "sup2"

Regex lookahead/lookbehind comments

2 Answers2

Working Demo