And to elaborate slightly on why the given solutions work, parens ( ) in a regexp capture their contents to the special variables $1, $2, ....

These are numbered according to the order of the capture groups (as ordered by their left hand paren).

You can use them (as shown) in the right-hand-side of a search and replace, but they're also valid perl variables which you can make use of after a substitution or ordinary match:

  my $line = "furrfu";
  $line =~ /(r+)/;
  print "$line contained ", length($1), " repeated 'r's\n";
[download]

Since regexp could in principle overwrite these global special variables, you should make use of them immediately after the regexp and/or save off their values into more appropriateley named vars. (e.g. if you call a subroutine, you don't know if that sub has done a regex and overwritten them).

Since parens are used for purposes other than grouping (e.g. to have alternating choices (foo|bar|baz), you can use a "non-capturing group", with (?:).

Search for 'captur' in perlre for more gory details.