gistfile1.pl

use strict;
 
my $content = '=toc<br />
=toc 2-4<br />
<p>=toc 2</p>
<div id="stff">=toc -3<br />
<p>=toc 2-</p>
';
 
my ($toc_h_min, $toc_h_max);
my $toc_params_RE = qr/\s+ (\d*)(?{$toc_h_min = $^N}) \s* - \s* (\d*)(?{$toc_h_max = $^N})/x;
while (
    # replace the =toc markup tag if it's on a line of its own (<p>), or not (followed by <br />)
    $content =~ s{
        <p>=toc(?:$toc_params_RE)? \s* </p>
      | (?:<=>)?=toc(?:$toc_params_RE)? \s* (?=<br)
    }{<!--mojomojoTOCwillgohere-->}x) {
    
    $toc_h_min ||= 1, $toc_h_max ||= 12;
    # FIXME: Perl 5.10 has a regexp branch reset operator which simplifies this
    #if ($1) {
    # # matched the <p> branch
    # $toc_h_min = $2 || 1;
    # $toc_h_max = $4 || 12;
    #} else {
    # # matched the <br branch
    # $toc_h_min = $4 || 1;
    # $toc_h_max = $5 || 12; # 6 is the HTML max, but hey, knock yourself out
    #}
    print "$toc_h_min - $toc_h_max\n";
}