Merge branch 'master' of http://10.200.0.3/GIT/videosite

[videosite.git] / videosite / YouTubeGrabber.pm

diff --git a/videosite/YouTubeGrabber.pm b/videosite/YouTubeGrabber.pm
index 86b21bd..2a363da 100644 (file)
--- a/videosite/YouTubeGrabber.pm
+++ b/videosite/YouTubeGrabber.pm
@@ -26,11 +26,13 @@ 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)/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_]+))',
-                           '(http://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/user/\w+\?.*/([-a-zA-Z0-9_]+))'];
+    $self->{'PATTERNS'} = ['(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/watch\?(?:.+=.+&)*v=([-a-zA-Z0-9_]+))',
+                           '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/watch\#\!v=([-a-zA-Z0-9_]+))',
+                           '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/v/([-a-zA-Z0-9_]+))',
+                           '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/user/[[:alnum:]]+\?v=([-a-zA-Z0-9_]+))',
+                           '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtu\.be/watch\?v=([-a-zA-Z0-9_]+))',
+                           '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtu\.be/([-a-zA-Z0-9_]+))',
+                           '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/user/\w+\?.*/([-a-zA-Z0-9_]+))'];
     $self->{'_PARAMS'} = {
             'QUALITY' => ['normal', 'Quality of the video to download.', {
                     'normal' => 'standard resolution flash video',
@@ -61,10 +63,20 @@ sub _parse {
     my $videourl;
     my $quality = $self->_getval('QUALITY');
     my %preflist = (
-        '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]);
+        'insane' => [38, 37, 22, 35, 18, 34, 6, 5],
+        'hd' => [37, 22, 35, 18, 34, 6, 5, 38],
+        'h264' => [18, 34, 37, 22, 35, 6, 5, 38],
+        'high' => [34, 35, 18, 37, 22, 6, 5, 38],
+        'normal' => [6, 5, 34, 35, 18, 22, 37, 38]);
+    my %videoformats = (
+        38 => 'mp4,h264,4k',
+        37 => 'mp4,h264,1080p',
+        35 => 'flv,h264,large',
+        34 => 'flv,h264',
+        22 => 'mp4,h264,720p',
+        18 => 'mp4,h264',
+        5 => 'flv,flv',
+        );
     my $preflist;
     my $jsp;
 
@@ -84,7 +96,7 @@ sub _parse {
     $preflist = $preflist{$quality};
     $self->debug("Quality: %s, preflist: [%s]", $quality, join(", ", @{$preflist}));
 
-    $videourl = sprintf('http://www.youtube.com/watch?v=%s', $2);
+    $videourl = sprintf('https://www.youtube.com/watch?v=%s', $2);
 
     unless(defined($r = $ua->get($videourl))) {
         $self->error('Could not download %s', $url);
@@ -108,7 +120,7 @@ sub _parse {
 
     SWF_ARGS: while ($tag = $p->get_tag('div', 'meta', 'script')) {
         if ('meta' eq $tag->[0]) {
-            if ('title' eq $tag->[1]->{'name'}) {
+            if (exists($tag->[1]->{'name'}) and ('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",
@@ -185,20 +197,10 @@ sub _parse {
 
             if (%urls) {
                 foreach (keys(%urls)) {
-                    if ($_ == 35) {
-                        $self->debug('Found flv,h264,large: %s', $urls{$_});
-                    } elsif ($_ == 34) {
-                        $self->debug('Found flv,h264: %s', $urls{$_});
-                    } elsif ($_ == 22) {
-                        $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) {
-                        $self->debug('Found flv,flv: %s', $urls{$_});
+                    if (exists($videoformats{$_})) {
+                        $self->debug('Found URL for format %s (%s): %s', $_, $videoformats{$_}, $urls{$_});
                     } else {
-                        $self->error('Unknown tag %s: %s', $_, $urls{$_});
+                        $self->error('Unknown format %s: %s', $_, $urls{$_});
                     }
                 }
 
@@ -269,16 +271,25 @@ sub __login {
 
     $self->debug('Logging in');
     $r = $ua->get('https://www.google.com/accounts/ServiceLoginAuth?service=youtube');
+    unless($r->is_success()) {
+        $self->debug("Could not get login page (make sure your LWP supports HTTPS!)");
+        return undef;
+    }
     $c = $r->decoded_content();
     $p = HTML::TokeParser->new(\$c);
     while (my $tag = $p->get_tag('input')) {
+        $self->debug("%s", Dumper($tag));
         if ($tag->[1]{name} eq 'GALX') {
             $token = $tag->[1]{value};
             last;
         }
     }
-    $self->debug("GALX = .$token");
+    $self->debug("GALX = %s", $token);
     $r = $ua->post('https://www.google.com/accounts/ServiceLoginAuth?service=youtube', { 'service' => 'youtube', 'Email' => $user, 'Passwd' => $pass, 'GALX' => $token });
+    unless($r->is_success()) {
+        $self->debug("Could not get login page (make sure your LWP supports HTTPS!)");
+        return undef;
+    }
     $c = $r -> decoded_content();
     $p = HTML::TokeParser->new(\$c);
     while (my $tag = $p->get_tag('script')) {