Fix DailyMotionGrabber, it now needs a JSON parser
authorRalf Ertzinger <ralf@skytale.net>
Sat, 17 Apr 2010 16:22:01 +0000 (18:22 +0200)
committerRalf Ertzinger <ralf@skytale.net>
Sat, 17 Apr 2010 16:22:01 +0000 (18:22 +0200)
videosite/DailyMotionGrabber.pm

index 903c610..981fef7 100644 (file)
@@ -10,6 +10,7 @@ use videosite::GrabberBase;
 
 use LWP::Simple qw(!get);
 use HTML::Parser;
+use videosite::JSArrayParser;
 use Data::Dumper;
 
 use strict;
@@ -65,19 +66,60 @@ sub _parse {
         if ('meta' eq $e->[0]) {
             if ('title' eq $e->[1]->{'name'}) {
                 $metadata->{'TITLE'} = $e->[1]->{'content'};
-                $metadata->{'TITLE'} =~ s/^Dailymotion\s*:\s*//;
+                $metadata->{'TITLE'} =~ s/^Dailymotion\s+-\s+//;
+                $metadata->{'TITLE'} =~ s/(?:\s+-\s+.*)?$//;
             }
         }
     }
 
     # Look for the download URL
     foreach $e (@text) {
-        if ($e->[0] =~ m|\.addVariable\("video", "([^\"]+)"|) {
-            $metadata->{'DLURL'} = $1;
-            $metadata->{'DLURL'} =~ s/%(..)/chr(hex($1))/ge;
-            $metadata->{'DLURL'} =~ s/\@\@spark.*//g;
-            $metadata->{'DLURL'} = 'http://www.dailymotion.com' . $metadata->{'DLURL'};
-
+        if ($e->[0] =~ m|\.addVariable\("sequence",\s*"([^\"]+)"|) {
+            my $sequence = $1;
+            my $jsp = videosite::JSArrayParser->new();
+            my $main;
+            my $s;
+
+            $sequence =~ s/%(..)/chr(hex($1))/ge;
+            $self->debug("Found sequence: %s", $sequence);
+
+            $self->debug("Using %s to parse", ref($jsp));
+            $sequence = $jsp->parse($sequence);
+            $self->debug(Dumper($sequence));
+
+            unless(defined($sequence)) {
+                $self->error("Found sequence, but could not parse");
+                return undef;
+            } else {
+                $self->debug("Parsed sequence: %s", Dumper($sequence));
+
+                foreach (@{$sequence}) {
+                    if (exists($_->{'name'}) && ($_->{'name'} eq 'main')) {
+                        # Found main section
+                        $main = $_->{'layerList'};
+                    }
+                }
+                unless(defined($main)) {
+                    $self->error("Could not find layerList[main]");
+                    return undef;
+                }
+
+                foreach (@{$main}) {
+                    if (exists($_->{'name'}) && ($_->{'name'} eq 'video')) {
+                        # Found video section
+                        if (exists($_->{'param'}->{'hdURL'})) {
+                            $metadata->{'DLURL'} = $_->{'param'}->{'hdURL'};
+                        } elsif (exists($_->{'param'}->{'hqURL'})) {
+                            $metadata->{'DLURL'} = $_->{'param'}->{'hqURL'};
+                        } elsif (exists($_->{'param'}->{'hqURL'})) {
+                            $metadata->{'DLURL'} = $_->{'param'}->{'sdURL'};
+                        } else {
+                            $self->error("Video section found, but no URLs");
+                            return undef;
+                        }
+                    }
+                }
+            }
         }
     }