# (c) 2009 by Ralf Ertzinger # licensed under GNU GPL v2 # # Grabber for veoh.com package videosite::VeohGrabber; use videosite::GrabberBase; @ISA = qw(videosite::GrabberBase); use HTML::TokeParser; use Data::Dumper; use strict; sub new { my $class = shift; my $self = $class->SUPER::new( NAME => 'veoh', _SELFTESTURL => 'http://www.veoh.com/watch/v18348952fyn2twbe', _SELFTESTTITLE => '518_2 kureyon shinchan', PATTERNS => ['(http://(?:[-a-zA-Z0-9_.]+\.)*veoh.com/+watch/(\w+)\??)'], @_, ); return bless($self, $class); } sub _parse { my $self = shift; my $url = shift; my $pattern = shift; my $content; my $metadata = {}; my $ua = $self->ua(); my $p; my $t; my @text; my @accum; my $dlurl; my $r; $url =~ m|$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.veoh.com/watch/%s', $2), $ua))) { $self->error('Could not download'); return undef; } $p = HTML::TokeParser->new(\$content); while ($t = $p->get_tag('script')) { if ($t->[0] eq 'script') { my $e = $p->get_text(); my $jsp; if ($e =~ m|__watch.videoDetailsJSON = '([^\x27]+)|) { $self->debug("Found JSON: %s", $1); $jsp = videosite::JSArrayParser->new(); $r = $jsp->parse($1); unless(defined($r)) { $self->error("Found information hash, but could not parse"); return undef; } $self->debug("Parsed JSON: %s", Dumper($r)); unless(exists($r->{previewUrl})) { $self->error("previewUrl not found in information hash"); return undef; } $metadata->{'DLURL'} = $r->{previewUrl}; $metadata->{'TITLE'} = $r->{title}; } } } unless(defined($metadata->{'DLURL'}) && defined($metadata->{'TITLE'})) { $self->error('Could not extract download URL and title'); return undef; } return $metadata; } 1;