# 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::TokeParser; use videosite::JSArrayParser; use Data::Dumper; use strict; sub new { my $class = shift; my $self = $class->SUPER::new( NAME => 'dailymotion', _SELFTESTURL => 'http://www.dailymotion.com/video/xylv6u_moon-duo-sleepwalker_music', _SELFTESTTITLE => 'Moon Duo - Sleepwalker', PATTERNS => ['(http://(?:[-a-zA-Z0-9_.]+\.)*dailymotion.com/(?:[^/]+/)*video/([-a-zA-Z0-9_]+))'], @_, ); return bless($self, $class); } sub _parse { my $self = shift; my $url = shift; my $pattern = shift; my $content; my $metadata = {}; my $p; 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; } $p = HTML::TokeParser->new(\$content); # Look for the title in the meta tags while ($e = $p->get_tag('meta', 'script')) { if ('meta' eq $e->[0]) { if (exists($e->[1]->{'property'}) && ('og:title' eq $e->[1]->{'property'})) { $metadata->{'TITLE'} = $e->[1]->{'content'}; } } elsif ('script' eq $e->[0]) { my $c = $p->get_text(); $self->debug("Found script: %s", $c); if ($c =~ m|flashvars = (.+);$|m) { my $flashvars = $1; my $jsp = videosite::JSArrayParser->new(); my $l; my $s; my $sequence; $self->debug("Found flashvars: %s", $flashvars); $self->debug("Using %s to parse", ref($jsp)); $flashvars = $jsp->parse($flashvars); $self->debug("Parsed flashvars: %s", Dumper($flashvars)); $sequence = $flashvars->{'sequence'}; $sequence =~ s/%(..)/chr(hex($1))/ge; $self->debug("Decoded sequence: %s", $sequence); $sequence = $jsp->parse($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->{'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->{'hd1080URL'})) { $metadata->{'DLURL'} = $l->{'hd1080URL'}; } elsif (exists($l->{'hd720URL'})) { $metadata->{'DLURL'} = $l->{'hd720URL'}; } elsif (exists($l->{'hqURL'})) { $metadata->{'DLURL'} = $l->{'hqURL'}; } elsif (exists($l->{'sdURL'})) { $metadata->{'DLURL'} = $l->{'sdURL'}; } elsif (exists($l->{'ldURL'})) { $metadata->{'DLURL'} = $l->{'ldURL'}; } 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;