You may also be interested in HTML::TreeBuilder. Consider:

use warnings;
use strict;
use HTML::TreeBuilder;

my $html = <<'HTML';
<a href="URL1"> text1.0 <img src="SRC1"> text1.1 <br>
                <garbage2> text1.2
   <a href="URL2"> text2.0 <img src="SRC2"> text2.1 <br> <garbage2>
                   <garbage3> text2.2
      <a href="URL3"> text3.0 <img src="SRC3"> text3.1 <br> <garbage3>
            text3.2 <x1 a="a" b="b"> text3.3
      </a>
      text2.3 <haha>
   </a>
   text1.3 <oho>
   <a href="URL4"> text4.0 <img src="SRC4"> text4.1 <br> <garbage4>
                    text4.2 <x1 a="a" b="b"> text4.3
   </a>
   text1.4
</a>
HTML

my $tree = HTML::TreeBuilder->new_from_content ($html);

for my $elt ($tree->look_down ('_tag', 'a')) {
    print "A " . $elt->attr ('href') . "\n\tTEXT: '";
    
    my @text_segs;
    
    for my $child ($elt->content_list ()) {
        next if ref $child and $child->{_tag} ne 'a';
        last if ref $child;
        push @text_segs, $child;
    }
    
    print "$_ " for @text_segs;
    print "\n";
}
[download]

Prints:

A URL1
    TEXT: ' text1.0   text1.1    text1.2  
A URL2
    TEXT: ' text2.0   text2.1     text2.2  
A URL3
    TEXT: ' text3.0   text3.1    text3.2  text3.3  
A URL4
    TEXT: ' text4.0   text4.1    text4.2  text4.3
[download]

Thank you for this one; I forgot mentioning HTML::Treebuilder, but I had it in mind too.

I hope to get the time to add the other variations too; once that done, I'll want to uniformize style about the different approaches and then comment a bit about pros or cons for each.

I'll probably avoid benchmarking (reasoning about No More Meaningless Benchmarks!)- not sure yet. In the end, I hope to have collected together a few code samples that might be a goot reading for all those that step into the html parsing task.

I'm still looking to find a good title ("Parsing HTML' ?) and a good way to place the whole thing in the end. I'm tempted to make it a set of linked nodes in 'Code catacombs', but I'm unsure yet.

Thanks again.

Well, about what the program does.. hmm, I really hoped that showing the input, saying that the program parses it and showing the output would say much more then I'd be able to say in words :)

So: I want to be able to extract links from HTML _and_ associate them with a reasonable link_text, even if the links are nested and the text that should pertain to a given link is interspersed with possible various other tags, even if the links are nested.

In order to do so, I wanted to explore the modules that would allow this sort of operation as easiest as possible - in order to get a good start the next time I'd have to do something similar, so that I'd know better the strenghts and weaknesses of the available tools.

HTML::Parser would be Way#1; having the same problem solved with alongside Way#2 (HTML::LinkExtractor), Way#3 (HTML::TokeParser), etc. would allow to get an easy overview/comparison on the available tools/modules.

Well, about what the program does.. hmm, I really hoped that showing the input, saying that the program parses it and showing the output would say much more then I'd be able to say in words :)

Well, it's too obvious I needed to ask. You want to share your effort, fine. But what peculiarity you expect people would find in the context of sharing? I'm not saying your effort is useless, not at all. Just that you asked in what part of PM your code should be posted.

So if you want to seek advices on what you've done, SoPW is the place and that's what you did. And I did suggested two other places. Again, my question was meant to help you clarify to yourself where to post.

---

Open source softwares? Share and enjoy. Make profit from them if you can. Yet, share and enjoy!