I believe this is what you're looking for.

Code:

#! /usr/bin/perl

use strict ;
use warnings ;
use Data::Dumper ;

my $str = qq~This <b>contains</b> both text
and <a href="http://www.w3c.org">html</a>.~ ;

print $str, "\n\n" ;

# Do the replacement...
my @tags  = () ;
my $index = -1 ;
$str =~ s|<([^>]+)>(?{ push @tags, $1 ; $index++ })|<$index>|gs ;

# Show the replaced text & the stored tags.
print "-----\n", Dumper( \@tags ), "\n\n", $str, "\n\n" ;

# Sub the tags back in.
$index = -1 ;
$str =~ s|<([^>]+)>(?{ $index++ })|<$tags[$index]>|gs ;

# Show the string with the HTML put back in.
print "-----\n", $str, "\n\n" ;
[download]

Output:

This <b>contains</b> both text
and <a href="http://www.w3c.org">html</a>.

-----
$VAR1 = [
          'b',
          '/b',
          'a href="http://www.w3c.org"',
          '/a'
        ];


This <0>contains<1> both text
and <2>html<3>.

-----
This <b>contains</b> both text
and <a href="http://www.w3c.org">html</a>.
[download]

Update: Argh. I was downtown about an hour after I posted this, and it suddenly occurred to me that the second substitution makes much more sense as:

# Sub the tags back in.
$str =~ s|<(\d+)>|<$tags[$1]>|gs ;
[download]