Regex to match only innermost delimited sequence

Question

I have a string that contains sequences delimited by multiple characters: << and >>. I need a regular expression to only give me the innermost sequences. I have tried lookaheads but they don't seem to work in the way I expect them to.

Here is a test string:

'do not match this <<but match this>> not this <<BUT NOT THIS <<this too>> IT HAS CHILDREN>> <<and <also> this>>'

It should return:

but match this
this too
and <also> this

As you can see with the third result, I can't just use /<<[^>]+>>/ because the string may have one character of the delimiters, but not two in a row.

I'm fresh out of trial-and-error. Seems to me this shouldn't be this complicated.

Theoretically, this kind of problem (stack-ish, depending on some intermediate state, etc) needs a more expressive grammar than a regular. — miku, Aug 09 '11 at 02:20
Regular expressions can only parse regular grammars. This is not a regular grammar. — cdhowie, Aug 09 '11 at 02:21
Perl's regular expression aren't at all regular, and can parse this just fine. — brian d foy, Aug 09 '11 at 09:14
@miku, @cdhowie: Since he wants the inner brackets and not the outer ones, there actually is a regular grammar for it. `/<<(?:[^<>]+|<[^<]|>[^>])*>>/` — ikegami, Aug 09 '11 at 12:43
@ikegami: it's more complicated than that; that doesn't match all of `<<<>>` — ysth, Aug 09 '11 at 15:40
@ysth, Indeed, though still possible to do using a regular grammar. Just a lot more wordy. — ikegami, Aug 09 '11 at 16:39
hmm, does it work if you just end `*?>>` ? no, that still leaves the > alternation potentially getting the < from a << incorrectly — ysth, Aug 09 '11 at 17:59

ikegami · Accepted Answer · 2015-04-06T21:30:48.620

9

@matches = $string =~ /(<<(?:(?!<<|>>).)*>>)/g;

(?:(?!PAT).)* is to patterns as [^CHAR]* is to characters.

edited Apr 06 '15 at 21:30

answered Aug 09 '11 at 12:36

ikegami

367,544
15
269
518

score 6 · Answer 2 · answered Aug 09 '11 at 03:10

6

$string = 'do not match this <<but match this>> not this <<BUT NOT THIS <<this too>> IT HAS CHILDREN>> <<and <also> this>>';
@matches = $string =~ /(<<(?:[^<>]+|<(?!<)|>(?!>))*>>)/g;

answered Aug 09 '11 at 03:10

ysth

96,171
6
121
214

1

Yessss... I knew there was a way. You win the Perl PGA Tour, my friend. – amphetamachine Aug 09 '11 at 03:19

score 0 · Answer 3 · answered Aug 09 '11 at 13:06

Here's a way to use split for the job:

my $str = 'do not match this <<but match this>> not this <<BUT NOT THIS <<this too>> IT HAS CHILDREN>> <<and <also> this>>';
my @a = split /(?=<<)/, $str;
@a = map { split /(?<=>>)/, $_ } @a;

my @match = grep { /^<<.*?>>$/ } @a;

Keeps the tags in there, if you want them removed, just do:

@match = map { s/^<<//; s/>>$//; $_ } @match;

Regex to match only innermost delimited sequence

3 Answers3

Linked