- Add broadcaster.com grabber
authorRalf Ertzinger <sun@lain.camperquake.de>
Fri, 18 Apr 2008 20:13:38 +0000 (22:13 +0200)
committerRalf Ertzinger <sun@lain.camperquake.de>
Fri, 18 Apr 2008 20:13:38 +0000 (22:13 +0200)
videosite/BroadcasterGrabber.pm [new file with mode: 0644]

diff --git a/videosite/BroadcasterGrabber.pm b/videosite/BroadcasterGrabber.pm
new file mode 100644 (file)
index 0000000..8d0cda9
--- /dev/null
@@ -0,0 +1,86 @@
+# (c) 2007 by Ralf Ertzinger <ralf@camperquake.de>
+# licensed under GNU GPL v2
+#
+# Grabber for broadcaster.com
+
+package BroadcasterGrabber;
+
+use GrabberBase;
+@ISA = qw(GrabberBase);
+
+use LWP::Simple qw(!get);
+use HTML::Parser;
+use Data::Dumper;
+
+use strict;
+
+sub new {
+    my $class = shift;
+    my $self = $class->SUPER::new();
+
+    $self->{'NAME'} = 'broadcaster';
+    $self->{'PATTERNS'} = ['(http://(?:[-a-zA-Z0-9_.]+\.)*broadcaster\.com/clip/(\d+))'];
+
+    bless($self, $class);
+    $self->_prepare_parameters();
+
+    return $self;
+}
+
+sub _parse {
+    my $self = shift;
+    my $url = shift;
+    my $pattern = shift;
+    my $content;
+    my $metadata = {};
+    my $p = HTML::Parser->new(api_version => 3);
+    my @accum;
+    my @text;
+    my $e;
+
+    $url =~ m|$pattern|;
+    $url = $1;
+
+    $metadata->{'URL'} = $url;
+    $metadata->{'ID'} = $2;
+    $metadata->{'TYPE'} = 'video';
+    $metadata->{'SOURCE'} = $self->{'NAME'};
+    $metadata->{'TITLE'} = undef;
+    $metadata->{'DLURL'} = undef;
+
+    unless(defined($content = LWP::Simple::get(sprintf('http://www.broadcaster.com/clip/%s', $2)))) {
+        $self->error('Could not download %s', $url);
+        return undef;
+    }
+
+    $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 title
+    foreach $e (@text) {
+        if ($e->[0] =~ m|\&page_title=([^\x22]+)\x22|s) {
+            $metadata->{'TITLE'} = $1;
+            last;
+        }
+    }
+
+    # Look for the download URL
+    foreach $e (@text) {
+        if ($e->[0] =~ m|\&clip_loc=.+?cache.broadcaster.com.+?escape\(\x22([^\x22]+)\x22\)|s) {
+            $metadata->{'DLURL'} = 'http://cache.broadcaster.com/peoplecaster/' . $1;
+            last;
+        }
+    }
+
+    unless(defined($metadata->{'DLURL'}) && defined($metadata->{'TITLE'})) {
+        $self->error('Could not determine download URL');
+        return undef;
+    }
+
+    return $metadata;
+}
+
+1;