add youtu.be short links

[videosite.git] / videosite / YouTubeGrabber.pm

diff --git a/videosite/YouTubeGrabber.pm b/videosite/YouTubeGrabber.pm
index 644584a..4449d1f 100644 (file)
--- a/videosite/YouTubeGrabber.pm
+++ b/videosite/YouTubeGrabber.pm
@@ -14,6 +14,8 @@ use videosite::GrabberBase;
 use LWP::UserAgent;
 use HTTP::Cookies;
 use HTML::TokeParser;
+use HTML::Entities qw(decode_entities);
+use Encode;
 use Data::Dumper;
 use videosite::JSArrayParser;
 
@@ -24,8 +26,9 @@ sub new {
     my $self = $class->SUPER::new();
 
     $self->{'NAME'} = 'youtube';
-    $self->{'PATTERNS'} = ['(http://(?:[-a-zA-Z0-9_.]+\.)*youtube.(?:com|de|co.uk)/watch\?(?:.+=.+&)*v=([-a-zA-Z0-9_]+))',
-                           '(http://(?:[-a-zA-Z0-9_.]+\.)*youtube.(?:com|de|co.uk)/v/([-a-zA-Z0-9_]+))'];
+    $self->{'PATTERNS'} = ['(http://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/watch\?(?:.+=.+&)*v=([-a-zA-Z0-9_]+))',
+                           '(http://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/v/([-a-zA-Z0-9_]+))',
+                          '(http://(?:[-a-zA-Z0-9_.]+\.)*youtu\.be/([-a-zA-Z0-9_]+))'];
     $self->{'_PARAMS'} = {
             'QUALITY' => ['normal', 'Quality of the video to download.', {
                     'normal' => 'standard resolution flash video',
@@ -56,10 +59,10 @@ sub _parse {
     my $videourl;
     my $quality = $self->_getval('QUALITY');
     my %preflist = (
-        'hd' => [22, 35, 18, 34, 6, 5],
-        'h264' => [18, 34, 22, 35, 6, 5],
-        'high' => [34, 35, 18, 22, 6, 5],
-        'normal' => [6, 5, 34, 35, 18, 22]);
+        'hd' => [37, 22, 35, 18, 34, 6, 5],
+        'h264' => [18, 34, 37, 22, 35, 6, 5],
+        'high' => [34, 35, 18, 37, 22, 6, 5],
+        'normal' => [6, 5, 34, 35, 18, 22, 37]);
     my $preflist;
     my $jsp;
 
@@ -73,6 +76,8 @@ sub _parse {
     $metadata->{'TITLE'} = undef;
     $metadata->{'DLURL'} = undef;
 
+    $self->debug("Matched id %s from pattern %s", $2, $pattern);
+
     $preflist = $preflist{$quality};
     $self->debug("Quality: %s, preflist: [%s]", $quality, join(", ", @{$preflist}));
 
@@ -100,6 +105,11 @@ sub _parse {
         if ('meta' eq $tag->[0]) {
             if ('title' eq $tag->[1]->{'name'}) {
                 $metadata->{'TITLE'} = $tag->[1]->{'content'};
+                # Convert HTML entities in the title. This is a bit convoluted.
+                $metadata->{'TITLE'} = encode("utf8",
+                                         decode_entities(
+                                           decode("utf8", $metadata->{'TITLE'})));
+                    
                 $self->debug('Title found: %s', $metadata->{'TITLE'});
             }
         } elsif ('script' eq $tag->[0]) {
@@ -111,7 +121,6 @@ sub _parse {
 
                 $self->debug("Found SWF_ARGS: %s", $args);
                 $jsp = videosite::JSArrayParser->new();
-                print ref($jsp);
                 $self->debug("Using %s to parse", ref($jsp));
                 $r = $jsp->parse($args);
 
@@ -161,7 +170,9 @@ sub _parse {
                     } elsif ($_ == 34) {
                         $self->debug('Found flv,h264: %s', $urls{$_});
                     } elsif ($_ == 22) {
-                        $self->debug('Found mp4,h264,large: %s', $urls{$_});
+                        $self->debug('Found mp4,h264,720p: %s', $urls{$_});
+                    } elsif ($_ == 37) {
+                        $self->debug('Found mp4,h264,1080p: %s', $urls{$_});
                     } elsif ($_ == 18) {
                         $self->debug('Found mp4,h264: %s', $urls{$_});
                     } elsif ($_ == 5) {
@@ -183,7 +194,7 @@ sub _parse {
                 last SWF_ARGS;
             }
         } elsif ('div' eq $tag->[0]) {
-            if (exists($tag->[1]->{'class'}) and ('errorBox' eq $tag->[1]->{'class'})) {
+            if (exists($tag->[1]->{'class'}) and ('yt-alert-content' eq $tag->[1]->{'class'})) {
                 $self->error("Could not get video data for youtube %s: %s",
                         $metadata->{'ID'}, $p->get_trimmed_text());
                 return undef;