Not a TokeParser solution, but using HTML::TreeBuilder I'd use $t->look_down( _tag => "div", class => "full" ) to get a list of the divs you're interested and then call $div->look_down( _tag => 'a' ) on each of those. Sometimes the tree solution's just conceptually easier to get your brane around.

Thanks, I know HTML::TreeBuilder - used it some time ago. But I'd like to use only one parser module, if it is possible, I decided for HTML::TokeParser. I hope exists some solution for this module, also I dont want use hack with regular expressions or so. Maybe I have to use $p->unget_token( @tokens ) to get desired links.

Of about the last 10 "I need to parse this HTML/XML structure" questions asked here nine of the answers were trivial using ::TreeParser (there are XML and HTML versions) and the other was trivial using XML::Twig.

Personally I use TreeParser more often in an HTML context and XML::Twig for XHTML and XML. XML::Twig is very powerful for editing, TreeBuilder is very good at looking stuff up.

At the end of the day the more modules you know a little bit about the more quickly and reliably you get stuff done. Don't be afraid to read documentation! Sometimes a quick question in the CB can save a huge amount of time, if you have a general idea where you are headed in the first place.

Limiting yourself to a single module is ... limiting! There is no one tool that does every job, not even computers.

---

DWIM is Perl's answer to Gödel

If you're dead set on tokeing it, build a state machine:

• start in looking_for_full until you see a div with class full, when you transition to looking_for_content
• when you see a div with class content in state looking_for_content, transition to looking_for_anchors
• when you see an anchor in looking_for_anchors, save the href attribute
• when you see a </div> in looking_for_anchors, go back to looking_for_full

Additional: Little note on implementation: you'd have a $state variable which keeps track of which state you're in (start with my $state = 'looking_for_full';). You'd then have a while( my $t = $stream->get_token ) { ... } loop, inside of which you'd implement the above behaviors. Any non-interesting token for the current state would be ignored (e.g. just next back to fetch the next token).

(assumes all links within class="full")

#!/usr/bin/perl

use strict;
use warnings;
use HTML::TokeParser::Simple;

my $html = do {local $/; <DATA>};
my $p = HTML::TokeParser::Simple->new(\$html)
  or die "can't parse: $!";
  
my ($in_full, @href);

while (my $t = $p->get_token){
  
  next if  
    $t->is_start_tag('div') 
      and 
    $t->get_attr('class') 
      and 
    $t->get_attr('class') eq 'content';
    
  $in_full++, next if 
    $t->is_start_tag('div')
      and
    $t->get_attr('class') eq 'full';
     
  $in_full = 0, next if
    $t->is_start_tag('div')
      and
    $t->get_attr('class') ne 'full';
        
  next unless $in_full;
  
  push @href, $t->get_attr('href') if 
    $t->is_start_tag('a');
  
}

print "$_\n" for @href;

__DATA__
<!-- some other divs and so here -->
<div class="full">
    <div class="content">
        <ul class="topics">
            <-- I want extract these links div class "full" only -->
            <li><a href="foobar">foobar</a></li>
            <li><a href="foobr2">fobar2</a></li>
            <li><a href="fobar3">foobr3</a></li>
        </ul>
    </div>
</div>
<div class="otherclass">
    <div class="content">
        <ul class="topics">
            <-- I DO NOT WANT these links -->
            <li><a href="fbaor">fbaor</a></li>
            <li><a href="fabar2">fabar2</a></li>
            <li><a href="fbar3">fbar3</a></li>
        </ul>
    </div>
</div>
<!-- some other divs and so here -->
<div class="full">
    <div class="content">
        <ul class="topics">
            <-- I want extract these links div class "full" only -->
            <li><a href="foobar4">foobar4</a></li>
            <li><a href="foobr5">fobar5</a></li>
            <li><a href="fobar6">foobr6</a></li>
        </ul>
    </div>
</div>
<div class="otherclass">
    <div class="content">
        <ul class="topics">
            <-- I DO NOT WANT these links -->
            <li><a href="fbaor">fbaor</a></li>
            <li><a href="fabar2">fabar2</a></li>
            <li><a href="fbar3">fbar3</a></li>
        </ul>
    </div>
</div>
<!-- some other divs and so here -->
[download]

#!/usr/bin/perl

use strict;
use warnings;

use HTML::TokeParser;

my $doc = do { local $/; <DATA> };

my $p = HTML::TokeParser->new( \$doc ); 

while ( my $outer = $p->get_tag("div") ) {
    
    next unless $outer->[1]{class} eq "full";
    my $nested_div = 0;

    while ( my $inner = $p->get_tag ) {
    
        # keep count of nested divs
        $nested_div++ if $inner->[0] eq "div";
        $nested_div-- if $inner->[0] eq "/div";

        # "full" div has closed
        last if $nested_div == -1;

        print $p->get_text, "\n" if $inner->[0] eq "a";
    }
}

__DATA__
<!-- some other divs and so here -->
<div class="full">
    <div class="content">
        <ul class="topics">
            <-- I want extract these links div class "full" only -->
            <li><a href="foobar">foobar</a></li>
            <li><a href="foobr2">fobar2</a></li>
            <li><a href="fobar3">foobr3</a></li>
        </ul>
    </div>
</div>
<div class="otherclass">
    <div class="content">
        <ul class="topics">
            <-- I DO NOT WANT these links -->
            <li><a href="fbaor">fbaor</a></li>
            <li><a href="fabar2">fabar2</a></li>
            <li><a href="fbar3">fbar3</a></li>
        </ul>
    </div>
</div>
<!-- some other divs and so here -->
[download]

Thanks! Very very nice solution, easy to understand and it works like it should. I give you ++