X-Git-Url: https://git.camperquake.de/gitweb.cgi?a=blobdiff_plain;f=videosite%2FYouTubeGrabber.pm;h=199143e8b3d99e46f09035e45f182b5ce3c6baf9;hb=76a9c69c78eec62f96cacba4b5ca1ac67b030661;hp=d6232a90e9c918e599b9ba0139b8a5cc30e58860;hpb=2247c833229fb2809d6e49c5074771067a2125d2;p=videosite.git

diff --git a/videosite/YouTubeGrabber.pm b/videosite/YouTubeGrabber.pm
index d6232a9..199143e 100644
--- a/videosite/YouTubeGrabber.pm
+++ b/videosite/YouTubeGrabber.pm
@@ -14,8 +14,10 @@ use videosite::GrabberBase;
 use LWP::UserAgent;
 use HTTP::Cookies;
 use HTML::TokeParser;
+use HTML::Entities qw(decode_entities);
+use Encode;
 use Data::Dumper;
-use List::Util;
+use videosite::JSArrayParser;
 
 use strict;
 
@@ -56,11 +58,12 @@ 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;
 
     $url =~ m|$pattern|;
     $url = $1;
@@ -72,6 +75,8 @@ 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}));
 
@@ -95,40 +100,85 @@ sub _parse {
 
     $p = HTML::TokeParser->new(\$content);
 
-    while ($tag = $p->get_tag('div', 'meta', 'script')) {
+    SWF_ARGS: while ($tag = $p->get_tag('div', 'meta', 'script')) {
         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]) {
             $e = $p->get_text();
             $self->debug("Found script: %s", $e);
-            if ($e =~ m|\x22fmt_url_map\x22\s*:\s*\x22([^\x22]+)\x22|) {
-                $self->debug("Found fmt_url_map: %s", $1);
-                my $urls = $1;
-                my @urls;
+            if ($e =~ m|\x27SWF_ARGS\x27:\s+(.+),|) {
                 my %urls;
+                my $args = $1;
+
+                $self->debug("Found SWF_ARGS: %s", $args);
+                $jsp = videosite::JSArrayParser->new();
+                $self->debug("Using %s to parse", ref($jsp));
+                $r = $jsp->parse($args);
+
+                unless(defined($r)) {
+                    $self->error("Found information hash, but could not parse");
+                    return undef;
+                }
 
-                $urls =~ s/%(..)/chr(hex($1))/ge;
-                @urls = split(/,/, $urls);
-                foreach (@urls) {
-                    my ($tag, $url) = split(/\|/);
-                    if ($tag == 35) {
-                        $self->debug('Found flv,h264,large: %s', $url);
-                    } elsif ($tag == 34) {
-                        $self->debug('Found flv,h264: %s', $url);
-                    } elsif ($tag == 22) {
-                        $self->debug('Found mp4,h264,large: %s', $url);
-                    } elsif ($tag == 18) {
-                        $self->debug('Found mp4,h264: %s', $url);
-                    } elsif ($tag == 5) {
-                        $self->debug('Found flv,flv: %s', $url);
+                if (exists($r->{'fmt_url_map'}) and ($r->{'fmt_url_map'} ne '')) {
+                    my $urls =  $r->{'fmt_url_map'};
+
+                    $self->debug("Video has fmt_url_map: %s", $urls);
+
+                    $urls =~ s/%([[:xdigit:]]{2})/chr(hex($1))/ge;
+                    %urls = split(/[\|,]/, $urls);
+                } elsif (exists($r->{'t'}) and ($r->{'t'} ne '')) {
+                    my $thash = $r->{'t'};
+
+                    if (exists($r->{'fmt_map'}) && ($r->{'fmt_map'} ne '')) {
+                        my $fmt = $r->{'fmt_map'};
+                        my @fmt;
+
+                        $self->debug('Video has fmt_map');
+                        $fmt =~ s/%([[:xdigit:]]{2})/chr(hex($1))/ge;
+                        @fmt = split(/,/, $fmt);
+                        foreach (@fmt) {
+                            split(/\//);
+                            $urls{$_[0]} =  sprintf('http://www.youtube.com/get_video?video_id=%s&fmt=%d&t=%s', 
+                                $metadata->{'ID'},
+                                $_[0],
+                                $thash);
+                        }
+                    } else {
+                        $urls{5} = sprintf('http://www.youtube.com/get_video?video_id=%s&t=%s',
+                            $metadata->{'ID'},
+                            $thash);
+                    }
+                } else {
+                    $self->error('Neither fmt_url_map nor t found in video information hash');
+                    return undef;
+                }
+                $self->debug("Found quality levels [%s]", join(", ", keys(%urls)));
+
+                foreach (keys(%urls)) {
+                    if ($_ == 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', $urls{$_});
                     } else {
-                        $self->error('Unknown tag %s: %s', $tag, $url);
-                        next;
+                        $self->error('Unknown tag %s: %s', $_, $urls{$_});
                     }
-                    $urls{$tag} = $url;
                 }
 
                 foreach (@{$preflist}) {
@@ -140,9 +190,10 @@ sub _parse {
                 }
 
                 $self->debug('URL found: %s', $metadata->{'DLURL'});
+                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;