<data>
  <r>
     <rec1>data</rec1>
     <something>else</something>
  </r>
</data>

##</code><code>##

    while ($html =~ m/<(\/?r)>/g) { 
        $r_pos = pos($html) if $1 eq 'r'; 
        if ($1 eq '/r') {
            $r = substr($html, $r_pos, pos($html) - $r_pos - length('</r>'));
            _extract_tags($r);
            $rec_count++;
        }
    }

sub _extract_tags {

    my $html = $_[0];
    my $tag_pos;
    my $curr_tag;
    my %data;
    
    while ($html =~ m/<(\/?[\w-]+)>/g) {

        my $tag = $1;
        my $n_tag = substr($tag, 1); # a 'naked' slashless tag

        if ($tag =~ /^\// && $n_tag eq $curr_tag) {
                
            # get the text inside the tag
            my $text = substr($html, $tag_pos, pos($html) - $tag_pos - length($tag) - 2); 

            $text =~ s/\r\n/\n/g;        
            $data{$n_tag} = $text;
        } else {

            # only get tags specified
            if (defined $columns{$tag}) {
                $tag_pos = pos($html);  
                $curr_tag = $tag;
            }
            
        }
    
    }

    # pass off to subroutine
    $handle_data->(\%data);

}