I'd solve this in two steps. In the first one, match the whole repetition, than split it into single expressions:

my @source = $text =~ /(regex1)((?:regex2)+)(regex3)/g;
my @repeated = $source[1] =~ /.../g;  # or split or whatever
[download]

($q=q:Sq=~/;[c](.)(.)/;chr(-||-|5+lengthSq)`"S|oS2"`map{chr |+ord
}map{substrSq`S_+|`|}3E|-|`7**2-3:)=~y+S|`+$1,++print+eval$q,q,a,
[download]

Hi, Thank you very much for the solution - I've been programming in Perl for some years and I must confess that I wasn't aware of this important detail in regexs that you kindly provided. Kind regards, Kepler

You can also store the matching parts in an array in a (?{}) expression:

#!/usr/bin/perl
use warnings;
use strict;
use feature qw{ say };

for my $string (qw( ab1b2b3d
                    ab1b2x
              )) {
    my @two;
    my @matches = $string =~ /(a)
                              (b.(?{ push @two, $2 if defined $2 }))+
                              (d)/x;
    say for @matches, '---', @two, '======';
}
[download]

($q=q:Sq=~/;[c](.)(.)/;chr(-||-|5+lengthSq)`"S|oS2"`map{chr |+ord
}map{substrSq`S_+|`|}3E|-|`7**2-3:)=~y+S|`+$1,++print+eval$q,q,a,
[download]

  ((?:regex2)+)

HTH :)

Just for grins, here's another approach, although I would still recommend the two-step approach outlined above:

c:\@Work\Perl>perl -wMstrict -le
"use Data::Dump qw(pp);
 ;;
 print qq{Perl version: $]};
 ;;
 my $ra = qr{ a  }xms;
 my $rb = qr{ b. }xms;
 my $rc = qr{ c  }xms;
 for my $string (qw(ab1b2b3b4c  ab5b6b7c  ab8b9c  abxc  ac  b0)) {
   my @all = $string =~ m{
     \G (?: $ra (?= $rb)
             |
            $rb (?= $rb | $rc)
             |
            $rc (?= \z)
        )
     }xmsg;
   print qq{'$string' -> }, pp \@all;
   }
"
Perl version: 5.008009
'ab1b2b3b4c' -> ["a", "b1", "b2", "b3", "b4", "c"]
'ab5b6b7c' -> ["a", "b5", "b6", "b7", "c"]
'ab8b9c' -> ["a", "b8", "b9", "c"]
'abxc' -> ["a", "bx", "c"]
'ac' -> []
'b0' -> []
[download]

Please use code tags when providing source here   <code>

then please note that this regex is matching more cases because order constrains are lost!

Cheers Rolf
_{(addicted to the Perl Programming Language and ☆☆☆☆ :)
Je suis Charlie!}

Interesting this ideone if it really lets you run actual perl code, anyway, its best in the future if you also post the code here in code tags, here is the output on my machine

#!/usr/bin/perl
use strict; use warnings;
use Data::Dumper;

my $text = 'XYYYZ';
my @source;
push @{$source[$1 ? 0 : $2 ? 1 : 2]}, $& while $text =~ /(X)|(Y)|(Z)/g
+;
print Dumper \@source;
__END__
$VAR1 = [
          [
            'X'
          ],
          [
            'Y',
            'Y',
            'Y'
          ],
          [
            'Z'
          ]
        ];
[download]

update: yeah wrong window, pasted the wrong ideone, fixed

#!/usr/bin/env perl
use strict;
use warnings;
use English qw( -no_match_vars );
use Data::Dumper;

my $string = 'foobarbaaaarbaz';
my @second;
if (my @matches = $string =~ m{\A
      (fo*)
      (?: (?<BAR>\s*ba+r) (?{push @second, $^N}))+
      (\s*ba*z) \z}mxs
   ) {
   $matches[1] = \@second;
   print Dumper \@matches;
}
# $VAR1 = [
#           'foo',
#           [
#             'bar',
#             'baaaar'
#           ],
#           'baz'
#         ];
[download]

Update: as noted below in a comment, the capture BAR is not needed. I added it while playing with %- and %+, and forgot to remove it before posting the example... I'm leaving it though, so that the comment below can still make sense. Anyway, always remember the old adage about regular expressions...

(?: (?<BAR>\s*ba+r) (?{push @second, $^N}))+


What is the advantage of using the  (?<BAR>\s*ba+r) named capture group rather than an ordinary capture group? The name is never referred to anywhere.

---

Give a man a fish:  <%-{-{-{-<