videosite/DailyMotionGrabber.pm

   1 # Grabber for dailymotion.com
   2 #
   3 # (c) 2007 by Ralf Ertzinger <ralf@camperquake.de>
   4 # licensed under GNU GPL v2
   5
   6 package videosite::DailyMotionGrabber;
   7
   8 use videosite::GrabberBase;
   9 @ISA = qw(videosite::GrabberBase);
  10
  11 use LWP::Simple qw(!get);
  12 use HTML::Parser;
  13 use videosite::JSArrayParser;
  14 use Data::Dumper;
  15
  16 use strict;
  17
  18 sub new {
  19     my $class = shift;
  20     my $self = $class->SUPER::new();
  21
  22     $self->{'NAME'} = 'dailymotion';
  23     $self->{'PATTERNS'} = ['(http://(?:[-a-zA-Z0-9_.]+\.)*dailymotion.com/(?:[^/]+/)*video/([-a-zA-Z0-9_]+))'];
  24
  25     bless($self, $class);
  26
  27     $self->_prepare_parameters();
  28
  29     return $self;
  30 }
  31
  32 sub _parse {
  33     my $self = shift;
  34     my $url = shift;
  35     my $pattern = shift;
  36     my $content;
  37     my $metadata = {};
  38     my $p = HTML::Parser->new(api_version => 3);
  39     my @accum;
  40     my @text;
  41     my $e;
  42
  43     $url =~ m|$pattern|;
  44     $url = $1;
  45
  46     $metadata->{'URL'} = $url;
  47     $metadata->{'ID'} = $2;
  48     $metadata->{'TYPE'} = 'video';
  49     $metadata->{'SOURCE'} = $self->{'NAME'};
  50     $metadata->{'TITLE'} = undef;
  51     $metadata->{'DLURL'} = undef;
  52
  53     unless(defined($content = LWP::Simple::get(sprintf('http://www.dailymotion.com/video/%s', $2)))) {
  54         $self->error('Could not download %s', $url);
  55         return undef;
  56     }
  57
  58     $p->handler(start => \@accum, "tagname, attr");
  59     $p->handler(text => \@text, "text");
  60     $p->report_tags(qw(meta script));
  61     $p->utf8_mode(1);
  62     $p->parse($content);
  63
  64     # Look for the title in the meta tags
  65     foreach $e (@accum) {
  66         if ('meta' eq $e->[0]) {
  67             if ('title' eq $e->[1]->{'name'}) {
  68                 $metadata->{'TITLE'} = $e->[1]->{'content'};
  69                 $metadata->{'TITLE'} =~ s/^Dailymotion\s+-\s+//;
  70                 $metadata->{'TITLE'} =~ s/(?:\s+-\s+.*)?$//;
  71             }
  72         }
  73     }
  74
  75     # Look for the download URL
  76     foreach $e (@text) {
  77         if ($e->[0] =~ m|\.addVariable\("sequence",\s*"([^\"]+)"|) {
  78             my $sequence = $1;
  79             my $jsp = videosite::JSArrayParser->new();
  80             my $main;
  81             my $s;
  82
  83             $sequence =~ s/%(..)/chr(hex($1))/ge;
  84             $self->debug("Found sequence: %s", $sequence);
  85
  86             $self->debug("Using %s to parse", ref($jsp));
  87             $sequence = $jsp->parse($sequence);
  88             $self->debug(Dumper($sequence));
  89
  90             unless(defined($sequence)) {
  91                 $self->error("Found sequence, but could not parse");
  92                 return undef;
  93             } else {
  94                 $self->debug("Parsed sequence: %s", Dumper($sequence));
  95
  96                 foreach (@{$sequence}) {
  97                     if (exists($_->{'name'}) && ($_->{'name'} eq 'main')) {
  98                         # Found main section
  99                         $main = $_->{'layerList'};
 100                     }
 101                 }
 102                 unless(defined($main)) {
 103                     $self->error("Could not find layerList[main]");
 104                     return undef;
 105                 }
 106
 107                 foreach (@{$main}) {
 108                     if (exists($_->{'name'}) && ($_->{'name'} eq 'video')) {
 109                         # Found video section
 110                         if (exists($_->{'param'}->{'hdURL'})) {
 111                             $metadata->{'DLURL'} = $_->{'param'}->{'hdURL'};
 112                         } elsif (exists($_->{'param'}->{'hqURL'})) {
 113                             $metadata->{'DLURL'} = $_->{'param'}->{'hqURL'};
 114                         } elsif (exists($_->{'param'}->{'hqURL'})) {
 115                             $metadata->{'DLURL'} = $_->{'param'}->{'sdURL'};
 116                         } else {
 117                             $self->error("Video section found, but no URLs");
 118                             return undef;
 119                         }
 120                     }
 121                 }
 122             }
 123         }
 124     }
 125
 126     unless(defined($metadata->{'DLURL'}) && defined($metadata->{'TITLE'})) {
 127         $self->error('Could not determine download URL');
 128         return undef;
 129     }
 130
 131     return $metadata;
 132 }
 133
 134 1;