# (c) 2009 by Ralf Ertzinger # licensed under GNU GPL v2 # # Grabber for wimp.com package videosite::WimpGrabber; use videosite::GrabberBase; @ISA = qw(videosite::GrabberBase); use MIME::Base64; use HTML::Parser; use Data::Dumper; use strict; sub new { my $class = shift; my $self = $class->SUPER::new( NAME => 'wimp', _SELFTESTURL => 'http://www.wimp.com/insanebuilding/', _SELFTESTTITLE => 'Insane building.', PATTERNS => ['(http://(?:[-a-zA-Z0-9_.]+\.)*wimp.com/([^/]+)/?)'], @_, ); return bless($self, $class); } 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.wimp.com/%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 $e->[0] and exists($e->[1]->{'name'})) { if ('description' eq $e->[1]->{'name'}) { $metadata->{'TITLE'} = $e->[1]->{'content'}; last; } } } # Look for the download URL # This is obfuscated a little bit. foreach $e (@text) { if ($e->[0] =~ m|var googleCode = '([^\x27]+)'|) { my $c = $1; $self->debug("Found obfuscated code: %s", $c); my $d = decode_base64($c); $self->debug("Decoded to: %s", $d); if ($d =~ m|\.addVariable\("file",\s*"([^\x22]+)"\)|) { $metadata->{'DLURL'} = $1; } } } unless(defined($metadata->{'DLURL'}) && defined($metadata->{'TITLE'})) { $self->error('Could not determine download URL'); return undef; } return $metadata; } 1;