I found HTML::TokeParser (part of the HTML::Parser distribution) to be easier to use but in this case I used HTML::TreeBuilder.

This example reads a 2x2 table.

#!/usr/local/bin/perl

use Data::Dumper;
use HTML::TreeBuilder;
use strict;

die "must input filename" unless @ARGV;

foreach my $file_name (@ARGV) {
  my $tree = HTML::TreeBuilder->new; # empty tree
  $tree->parse_file($file_name);
  print "Hey, here's a dump of the parse tree of $file_name:\n";

  #  $tree->dump; # a method we inherit from HTML::Element
  # Now that we're done with it, we must destroy it.

  my %table;
  (
   $table{root},
   $table{cond},
   $table{'cond-alternatives'},
   $table{action},
   $table{'action-entries'}
  )
    = $tree->find_by_tag_name('table');

  my %td;

  map { $td{$_} = [ $table{$_}->find_by_tag_name('td') ] } (keys %tabl
+e);


  my %x;
  map {
    my $field = $_;
    map { push @{$x{$field}}, $_->content_array_ref } @{$td{$_}}
  } (keys %td);

  printf "cond-alt has %s", Dumper $x{'cond-alternatives'};

  $tree = $tree->delete;
}
[download]