# (c) 2007 by Ralf Ertzinger # licensed under GNU GPL v2 # # Grabber for vimeo.com package videosite::VimeoGrabber; use videosite::GrabberBase; @ISA = qw(videosite::GrabberBase); use LWP::UserAgent; use XML::Simple; use Digest::MD5 qw(md5_hex); use Data::Dumper; use strict; sub new { my $class = shift; my $self = $class->SUPER::new(); $self->{'NAME'} = 'vimeo'; $self->{'PATTERNS'} = ['(http://(?:[-a-zA-Z0-9_.]+\.)*vimeo.com/(\d+))']; bless($self, $class); $self->_prepare_parameters(); return $self; } sub _parse { my $self = shift; my $url = shift; my $pattern = shift; my $content; my $metadata = {}; my $p = XML::Simple->new(); my $t; my $dlurl; my $hd; my $dlpath; my $timestamp; my $hash; my $ua = LWP::UserAgent->new(agent => 'Mozilla'); $url =~ m|$pattern|; $url = $1; $metadata->{'URL'} = $url; $metadata->{'ID'} = $2; $metadata->{'TYPE'} = 'video'; $metadata->{'SOURCE'} = $self->{'NAME'}; $metadata->{'TITLE'} = undef; $metadata->{'DLURL'} = undef; # Get the XML file containing the video metadata $content = $ua->get(sprintf('http://www.vimeo.com/moogaloop/load/clip:%s', $2)); unless ($content->is_success()) { $self->error('Could not download XML metadata'); return undef; } $content = $content->decoded_content(); unless(defined($t = $p->XMLin($content, KeepRoot => 1))) { $self->error('Could not parse XML metadata'); return undef; } if (exists($t->{'xml'}->{'video'}->{'isHD'}) and (0 != $t->{'xml'}->{'video'}->{'isHD'})) { $self->debug('Selecting HD video'); $hd = '/?q=hd'; } else { $self->debug('Selecting SD video'); $hd = ''; } $timestamp = $t->{'xml'}->{'request_signature_expires'}; $hash = $t->{'xml'}->{'request_signature'}; $dlurl = sprintf('http://vimeo.com/moogaloop/play/clip:%s/%s/%d%s', $metadata->{'ID'}, $hash, $timestamp, $hd); unless(defined($dlurl)) { $self->error('No dlurl found in XML'); return undef; } # # Vimeo appends a hash to the download URL, in order to thwart people like me. # # Unfortunately the algorithm isn't that complicated :) # if ($dlurl =~ m|http://bitcast.vimeo.com(.+)|) { # $dlpath = $1; # $timestamp += 1800; # $hash = md5_hex(sprintf('redFiretruck%s?e=%d', $dlpath, $timestamp)); # } else { # $self->error('Unknown dlurl scheme: %s', $dlurl); # return undef; # } # $metadata->{'DLURL'} = sprintf('%s?e=%d&h=%s', $dlurl, $timestamp, $hash); $metadata->{'DLURL'} = $dlurl; $metadata->{'TITLE'} = $t->{'xml'}->{'video'}->{'caption'}; unless(defined($metadata->{'DLURL'}) && defined($metadata->{'TITLE'})) { $self->error('Could not extract download URL and title'); return undef; } return $metadata; } 1;