Re: How to get ($1, $2, ...)?

[almut]

Presuming I'm understanding your problem correctly :) I'd do a two step matching process. For this, you'd need additional outer parentheses around each regex for the m//mgc step (in scalar context). Then you could do something like

my $text = <<TEXT;
    A B
    1 2
    a b c
TEXT

my @rxs = (
    qr/(\s*(\w)\s+(\w)\s*$)/,
    qr/(\s*(\d)\s+(\d)\s*$)/,
    qr/(\s*([a-z])\s+([a-z])\s+([a-z])\s*$)/,
);

my @result;

for my $rx (@rxs) {
    if ($text =~ /$rx/mgc) {
        my $match = $1;
        # match again to extract submatches
        my @grps = $match =~ $rx;
        shift @grps;  # remove $1 (outer parens)
        push @result, [ @grps ];
    }
}

use Data::Dumper;
print Dumper \@result;

Prints:

$VAR1 = [
          [
            'A',
            'B'
          ],
          [
            '1',
            '2'
          ],
          [
            'a',
            'b',
            'c'
          ]
        ];

Maybe not the best approach performance-wise, but at least reasonably easy to code and maintain...

Alternatively, you could also simply use $& (update: look for the first WARNING: in perldoc perlre for why you might want to avoid $&, $`, $'), i.e.

...
my @rxs = (
    qr/\s*(\w)\s+(\w)\s*$/,
    qr/\s*(\d)\s+(\d)\s*$/,
    qr/\s*([a-z])\s+([a-z])\s+([a-z])\s*$/,
);

my @result;

for my $rx (@rxs) {
    if ($text =~ /$rx/mgc) {
        my $match = $&;
        # match again to extract submatches
        my @grps = $match =~ $rx;
        push @result, [ @grps ];
    }
}