- Completely reworked the youtube parser to removed videos better

[videosite.git] / videosite / YouTubeGrabber.pm

diff --git a/videosite/YouTubeGrabber.pm b/videosite/YouTubeGrabber.pm
index bf7640c..dc6bba7 100644 (file)
--- a/videosite/YouTubeGrabber.pm
+++ b/videosite/YouTubeGrabber.pm
@@ -1,10 +1,18 @@
+# (c) 2007 by Ralf Ertzinger <ralf@camperquake.de>
+# licensed under GNU GPL v2
+#
+# Grabber for youtube.com/de/...
+#
+# download strategy revised using
+# http://www.kde-apps.org/content/show.php?content=41456
+
 package YouTubeGrabber;
 
 use GrabberBase;
 @ISA = qw(GrabberBase);
 
 use LWP::Simple qw(!get);
-use HTML::Parser;
+use HTML::TokeParser;
 use Data::Dumper;
 
 use strict;
@@ -29,17 +37,17 @@ sub _parse {
     my $pattern = shift;
     my $content;
     my $metadata = {};
-    my $p = HTML::Parser->new(api_version => 3);
-    my @accum;
-    my @text;
+    my $p;
     my $e;
+    my $tag;
 
     $url =~ m|$pattern|;
     $url = $1;
 
     $metadata->{'URL'} = $url;
     $metadata->{'ID'} = $2;
-    $metadata->{'TYPE'} = 'youtube';
+    $metadata->{'TYPE'} = 'video';
+    $metadata->{'SOURCE'} = $self->{'NAME'};
     $metadata->{'TITLE'} = undef;
     $metadata->{'DLURL'} = undef;
 
@@ -48,25 +56,28 @@ sub _parse {
         return undef;
     }
 
-    $p->handler(start => \@accum, "tagname, attr");
-    $p->handler(text => \@text, "text");
-    $p->report_tags(qw(meta script));
-    $p->utf8_mode(1);
-    $p->parse($content);
-
-    # Look for the title in the meta tags
-    foreach $e (@accum) {
-        if ('meta' eq $e->[0]) {
-            if ('title' eq $e->[1]->{'name'}) {
-                $metadata->{'TITLE'} = $e->[1]->{'content'};
-            }
-        }
-    }
+    $p = HTML::TokeParser->new(\$content);
 
-    # Look for the download URL
-    foreach $e (@text) {
-        if ($e->[0] =~ m|/watch_fullscreen\?(.*)\&fs|) {
-            $metadata->{'DLURL'} = 'http://www.youtube.com/get_video.php?' . $1;
+    while ($tag = $p->get_tag('div', 'meta', 'script')) {
+        if ('meta' eq $tag->[0]) {
+            if ('title' eq $tag->[1]->{'name'}) {
+                $metadata->{'TITLE'} = $tag->[1]->{'content'};
+                $self->debug('Title found: %s', $metadata->{'TITLE'});
+            }
+        } elsif ('script' eq $tag->[0]) {
+            $e = $p->get_text();
+            if ($e =~ m|/watch_fullscreen\?(.+)\x27|) {
+                my %args = map { split(/=/, $_, 2); } split(/&/, $1);
+                $metadata->{'DLURL'} = sprintf('http://www.youtube.com/get_video.php?video_id=%s&t=%s',
+                        $metadata->{'ID'}, $args{'t'});
+                $self->debug('URL found: %s', $metadata->{'DLURL'});
+            }
+        } elsif ('div' eq $tag->[0]) {
+            if ('errorBox' eq $tag->[1]->{'class'}) {
+                $self->error("Could not get video data for youtube %s: %s",
+                        $metadata->{'ID'}, $p->get_trimmed_text());
+                return undef;
+            }
         }
     }