Youtube: Add video format 44 and 45 (also WebM)

[videosite.git] / videosite / BlipTVGrabber.pm

diff --git a/videosite/BlipTVGrabber.pm b/videosite/BlipTVGrabber.pm
index aeda2d9..5a53d2c 100644 (file)
--- a/videosite/BlipTVGrabber.pm
+++ b/videosite/BlipTVGrabber.pm
@@ -3,12 +3,15 @@
 #
 # Grabber for blip.tv
 
-package BlipTVGrabber;
+package videosite::BlipTVGrabber;
 
-use GrabberBase;
-@ISA = qw(GrabberBase);
+use videosite::GrabberBase;
+@ISA = qw(videosite::GrabberBase);
 
 use LWP::Simple qw(!get);
+use LWP::UserAgent;
+use URI;
+use URI::QueryParam;
 use HTML::Parser;
 use XML::Simple;
 use Data::Dumper;
@@ -20,7 +23,8 @@ sub new {
     my $self = $class->SUPER::new();
 
     $self->{'NAME'} = 'bliptv';
-    $self->{'PATTERNS'} = ['(http://(?:[-a-zA-Z0-9_.]+\.)*blip.tv/file/(\d+)(?:\?\S+)?)'];
+    $self->{'PATTERNS'} = ['(http://(?:[-a-zA-Z0-9_.]+\.)*blip.tv/file/(\d+)(?:\?\S+)?)',
+                           '(http://blip\.tv/play/(\w+)$)'];
 
     bless($self, $class);
     $self->_prepare_parameters();
@@ -36,8 +40,10 @@ sub _parse {
     my $metadata = {};
     my $p = HTML::Parser->new(api_version => 3);
     my @accum;
+    my @text;
     my $e;
     my $xml = undef;
+    my $ua;
 
     $url =~ m|$pattern|;
     $url = $1;
@@ -49,23 +55,53 @@ sub _parse {
     $metadata->{'TITLE'} = undef;
     $metadata->{'DLURL'} = undef;
 
-    unless(defined($content = LWP::Simple::get(sprintf('http://blip.tv/file/%s', $2)))) {
-        $self->error('Could not download page');
-        return undef;
-    }
+    if ($pattern eq $self->{'PATTERNS'}->[0]) {
+        # blip.tv/file pattern
+        unless(defined($content = LWP::Simple::get(sprintf('http://blip.tv/file/%s', $2)))) {
+            $self->error('Could not download page');
+            return undef;
+        }
 
-    $p->handler(start => \@accum, "tagname, attr");
-    $p->report_tags(qw(div));
-    $p->utf8_mode(1);
-    $p->parse($content);
+        $p->handler(start => \@accum, "tagname, attr");
+        $p->handler(text => \@text, "text");
+        $p->report_tags(qw(script));
+        $p->utf8_mode(1);
+        $p->parse($content);
 
-    # Look for the post id in the div tags
-    foreach $e (@accum) {
-        if (exists($e->[1]->{'pokkariparameters'})) {
-            if ($e->[1]->{'pokkariparameters'} =~ m|posts_id:"(\d+)"|) {
+        # Look for the post id in the javascript code
+        foreach $e (@text) {
+            if ($e->[0] =~ m|player.setPostsId\((\d+)\)|s) {
                 $xml = $1;
             }
         }
+    } elsif ($pattern eq $self->{'PATTERNS'}->[1]) {
+        my $r;
+        my $u;
+
+        $ua = LWP::UserAgent->new(max_redirect => 0);
+        $r = $ua->get(sprintf('http://blip.tv/play/%s', $2));
+
+        unless(defined($r)) {
+            $self->error('Could not download page');
+            return undef;
+        }
+        unless($r->is_redirect()) {
+            $self->error('Expected redirect, but got none');
+            return undef;
+        }
+        $u = URI->new($r->header('Location'));
+        $u = $u->query_param("file");
+
+        unless(defined($u)) {
+            $self->error('Did not find file parameter in URL');
+            return undef;
+        }
+        $u =~ m|^http://blip.tv/rss/flash/(\d+)$|;
+        $xml = $1;
+
+    } else {
+        $self->error("Don't know how to handle that pattern yet");
+        return undef;
     }
 
     unless(defined($xml)) { 
@@ -80,7 +116,7 @@ sub _parse {
     }
 
     $xml = XML::Simple::XMLin($content, KeepRoot => 1);
-    $metadata->{'DLURL'} = $xml->{'rss'}->{'channel'}->{'item'}->{'media:group'}->{'media:content'}->{'url'};
+    $metadata->{'DLURL'} = $xml->{'rss'}->{'channel'}->{'item'}->{'enclosure'}->{'url'};
     $metadata->{'TITLE'} = $xml->{'rss'}->{'channel'}->{'item'}->{'media:title'};
 
     unless(defined($metadata->{'DLURL'}) && defined($metadata->{'TITLE'})) {