Youtube: Disable the scrap handlers for the 2009 and 2010 page formats, they should...

[videosite.git] / videosite / BreakGrabber.pm

diff --git a/videosite/BreakGrabber.pm b/videosite/BreakGrabber.pm
index 4870da8..0896078 100644 (file)
--- a/videosite/BreakGrabber.pm
+++ b/videosite/BreakGrabber.pm
@@ -3,13 +3,12 @@
 # (c) 2007 by Ralf Ertzinger <ralf@camperquake.de>
 # licensed under GNU GPL v2
 
-package BreakGrabber;
+package videosite::BreakGrabber;
 
-use GrabberBase;
-@ISA = qw(GrabberBase);
+use videosite::GrabberBase;
+@ISA = qw(videosite::GrabberBase);
 
-use LWP::Simple qw(!get);
-use HTML::Parser;
+use videosite::HTMLHelper;
 use Data::Dumper;
 
 use strict;
@@ -34,10 +33,8 @@ sub _parse {
     my $pattern = shift;
     my $content;
     my $metadata = {};
-    my $p = HTML::Parser->new(api_version => 3);
-    my @accum;
-    my @text;
-    my $e;
+    my $p = videosite::HTMLHelper->new();
+    my $n;
 
     $url =~ m|$pattern|;
     $url = $1;
@@ -49,28 +46,20 @@ sub _parse {
     $metadata->{'TITLE'} = undef;
     $metadata->{'DLURL'} = undef;
 
-    unless(defined($content = LWP::Simple::get($1))) {
+    unless(defined($content = $p->load($1))) {
         $self->error('Could not download %s', $url);
         return undef;
     }
 
-    $p->handler(start => \@accum, "tagname, attr");
-    $p->report_tags(qw(meta));
-    $p->utf8_mode(1);
-    $p->parse($content);
-
-    # Look for the title in the meta tags
-    foreach $e (@accum) {
-        if ('meta' eq $e->[0]) {
-            if ('embed_video_title' eq $e->[1]->{'name'}) {
-                $metadata->{'TITLE'} = $e->[1]->{'content'};
-            }
-
-            if ('embed_video_thumb_url' eq $e->[1]->{'name'}) {
-                $metadata->{'DLURL'} = $e->[1]->{'content'};
-                $metadata->{'DLURL'} =~ s/\.jpg$/\.flv/;
-            }
-        }
+    $n = $p->findnodes('meta[@name="embed_video_title"]');
+    if (defined($n)) {
+        $metadata->{'TITLE'} = $n->{'content'};
+    }
+
+    $n = $p->findnodes('meta[@name="embed_video_thumb_url"]');
+    if (defined($n)) {
+        $metadata->{'DLURL'} = $n->{'content'};
+        $metadata->{'DLURL'} =~ s/\.jpg$/\.flv/;
     }
 
     unless(defined($metadata->{'DLURL'}) && defined($metadata->{'TITLE'})) {