How can I find out how many capturing brackets are in a Perl regexp?

Question

If I am given a rexexp in Perl, can I find out how many capturing brackets there are? So, for example:

\w       -> 0
(\w)     -> 1
(\w(\w)) -> 2

Answer 1

Do you want to know how many matches there are or how many sets of brackets? If you want to be able to count the brackets then you might want to take a look at a module like Text::Balanced which parses delimited text.

On the other hand, if you want to know how matches there are you would be better off executing your regular expression in list context:

my @matches = $string_to_match_on =~ /(\w(\w))/;

The size of the list will give you the number of matches:

my $count = @matches;

(as a list or array in scalar context gives the size of the list or array).

Answer 2

There are two special array @- and @+ containing the start positions and the end positions of successful matches. Use the array length when matching is done.

Static analysis: To know all pairs you need to parse the regex string. Count all unescaped opening bracktes which have a closing one.

Answer 3

It's not really trivial as not all parentheses are not capturing - for example (?:...), (?=...) and so on.

Generally, remember you can always:

my @catch_all = $string =~ m/......................./;

and then just check @catch_all;

Answer 4

It is important to know why you need this.

Does YAPE::Regex help?

Edit: Here is demonstration:

#!/usr/bin/perl

use strict;
use warnings;

use Data::Dumper;
use YAPE::Regex;

my $regex = qr/^(A)(B)(C)[0-9]+(\w+)$/;

my $parser = YAPE::Regex->new($regex);

my $n_captures;

while (my $node = $parser->next) {
    if ( $parser->state =~ /^capture\(([0-9]+)\)$/ ) {
        $n_captures = $1;
    }
}

print "$n_captures\n";


C:\Temp> t
4