# Grabber for dailymotion.com # # (c) 2007 by Ralf Ertzinger # licensed under GNU GPL v2 package videosite::DailyMotionGrabber; use videosite::GrabberBase; @ISA = qw(videosite::GrabberBase); use HTML::Parser; use videosite::JSArrayParser; use Data::Dumper; use strict; sub new { my $class = shift; my $self = $class->SUPER::new(); $self->{'NAME'} = 'dailymotion'; $self->{'PATTERNS'} = ['(http://(?:[-a-zA-Z0-9_.]+\.)*dailymotion.com/(?:[^/]+/)*video/([-a-zA-Z0-9_]+))']; bless($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 = $self->simple_get(sprintf('http://www.dailymotion.com/video/%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(meta script)); $p->utf8_mode(1); $p->parse($content); # Look for the title in the meta tags foreach $e (@accum) { if ('meta' eq $e->[0]) { if ('title' eq $e->[1]->{'name'}) { $metadata->{'TITLE'} = $e->[1]->{'content'}; $metadata->{'TITLE'} =~ s/^Dailymotion\s+-\s+//; $metadata->{'TITLE'} =~ s/(?:\s+-\s+.*)?$//; } } } # Look for the download URL foreach $e (@text) { if ($e->[0] =~ m|\.addVariable\("sequence",\s*"([^\"]+)"|) { my $sequence = $1; my $jsp = videosite::JSArrayParser->new(); my $l; 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)); $l = $self->_fetch_layer($sequence, "root/layerList", "background/sequenceList", "main/layerList", "video/param"); unless(defined($l)) { $self->error("Could not find video layer"); return undef; } # Found video section if (exists($l->{'videoPluginParameters'}->{'hdURL'})) { $metadata->{'DLURL'} = $l->{'videoPluginParameters'}->{'hdURL'}; } elsif (exists($l->{'videoPluginParameters'}->{'hqURL'})) { $metadata->{'DLURL'} = $l->{'videoPluginParameters'}->{'hqURL'}; } elsif (exists($l->{'videoPluginParameters'}->{'hqURL'})) { $metadata->{'DLURL'} = $l->{'videoPluginParameters'}->{'sdURL'}; } else { $self->error("Video section found, but no URLs"); return undef; } } } } unless(defined($metadata->{'DLURL'}) && defined($metadata->{'TITLE'})) { $self->error('Could not determine download URL'); return undef; } return $metadata; } sub _fetch_layer { my $self = shift; my $sequence = shift; my $point = shift; my $next; my @points = @_; my $l; $self->debug("Looking for %s in %s", $point, Dumper($sequence)); unless(defined($point)) { $self->debug("Reached last point"); return $sequence; } ($point, $next) = split(/\//, $point, 2); foreach (@{$sequence}) { if (exists($_->{'name'}) and ($_->{'name'} eq $point)) { if (exists($_->{$next})) { $self->debug("Using %s in %s", $next, $point); return $self->_fetch_layer($_->{$next}, @points); } else { $self->debug("%s found, but no %s", $point, $next); return undef; } } } $self->debug("Could not find entry named %s", $point); return undef; } 1;