extract href from html

Question

I am building a Perl module which I am attempting to use as few non-core dependencies as possible. Should my following code fail, I would add a dependency on HTML::LinkExtor and be done with it, but I want to try. All I want is to extract the href= attributes from <a> tags. I do it using Text::Balanced which is core as of modern Perls and is installable for others. So yes, I know I should use a HTML library. That said, is this passably ok?

#!/usr/bin/env perl
use strict;
use warnings;
use Data::Dumper;
use Text::Balanced qw/extract_bracketed extract_delimited extract_multiple/;
my $html = q#Some <a href=link>link text</a> stuff. And a little <A HREF="link2">different link text</a>.#;
my @tags = find_anchor_targets($html);
print Dumper \@tags;
sub find_anchor_targets {
 my $html = shift;
 my @tags = extract_multiple( 
 $html, 
 [ sub { extract_bracketed($_[0], '<>') } ],
 undef, 1
 );
 @tags = 
 map { extract_href($_) } # find related href=
 grep { /^<a/i } # only anchor begin tags
 @tags;
 return @tags;
}
sub extract_href {
 my $tag = shift;
 if($tag =~ /href=(?='|")/gci) {
 my $text = scalar extract_delimited( $tag, q{'"} );
 my $delim = substr $text, 0, 1;
 $text =~ s/^$delim//;
 $text =~ s/$delim$//;
 return $text;
 } elsif ($tag =~ /href=(.*?)(?:\s|\n|>)/) {
 return 1ドル;
 } else {
 return ();
 }
}

This dumps

$VAR1 = [
 'link',
 'link2'
 ];

which is what one would expect.

Answer 1

<script language="javascript">
var a='<a href="1" title="Passably ok, yes, why not. Perfect, no.">'
document.write('<a href="2" title="Real-world HTML is just really complicated.">')
</script>
<style type="text/css">
p { font-family: "<a href='3' title='...in so many ways'>" }
</style>

• Joel Berger: Thanks, point taken. In the days since I posted this I reworked my code to use HTML::LinkExtor if available and my code elsewise. Further, once the links are extracted, they are filtered for a certain file name pattern (source files of C projects), so there isn't too much chance of false posititives. Thanks for the good examples though!