package Sources::HtmlParser;

use strict;
use warnings;

use base 'HTML::Parser';

sub parseText
{
   my($self, $text) = @_;

   $self->{mytext} = "";
   $self->parse($text);
}


sub text
{
   my($self, $origtext, $is_cdata) = @_;

   $self->{mytext} .= $origtext;

   1;
}

sub get_text
{
   my($self) = @_;

   return $self->{mytext};
}

1;

my $hp = Sources::HtmlParser->new();
$hp->parseText($httpbody);
$httpbody = HTML::Entities::decode_entities($hp->get_text);
$httpbody =~ s/<!--.+?-->//gms;