- New YouYube scheme, as some videos did not conform to
[videosite.git] / videosite / YouTubeGrabber.pm
index 33ca467..2ed52a7 100644 (file)
@@ -15,6 +15,7 @@ use LWP::UserAgent;
 use HTTP::Cookies;
 use HTML::TokeParser;
 use Data::Dumper;
+use videosite::JSArrayParser;
 
 use strict;
 
@@ -60,6 +61,7 @@ sub _parse {
         'high' => [34, 35, 18, 22, 6, 5],
         'normal' => [6, 5, 34, 35, 18, 22]);
     my $preflist;
+    my $jsp;
 
     $url =~ m|$pattern|;
     $url = $1;
@@ -94,7 +96,7 @@ 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'};
@@ -103,13 +105,52 @@ sub _parse {
         } 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 = $1;
+            if ($e =~ m|\x27SWF_ARGS\x27:\s+(.+),|) {
                 my %urls;
+                my $args = $1;
 
-                $urls =~ s/%([[:xdigit:]]{2})/chr(hex($1))/ge;
-                %urls = split(/[\|,]/, $urls);
+                $self->debug("Found SWF_ARGS: %s", $args);
+                $jsp = videosite::JSArrayParser->new();
+                $r = $jsp->parse($args);
+
+                unless(defined($r)) {
+                    $self->error("Found information hash, but could not parse");
+                    return undef;
+                }
+
+                if (exists($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)) {
@@ -137,6 +178,7 @@ 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'})) {