1 # (c) 2007 by Ralf Ertzinger <ralf@camperquake.de>
2 # licensed under GNU GPL v2
4 # Grabber for youtube.com/de/...
6 # download strategy revised using
7 # http://www.kde-apps.org/content/show.php?content=41456
9 package videosite::YouTubeGrabber;
11 use videosite::GrabberBase;
12 @ISA = qw(videosite::GrabberBase);
23 my $self = $class->SUPER::new();
25 $self->{'NAME'} = 'youtube';
26 $self->{'PATTERNS'} = ['(http://(?:[-a-zA-Z0-9_.]+\.)*youtube.(?:com|de|co.uk)/watch\?(?:.+=.+&)*v=([-a-zA-Z0-9_]+))',
27 '(http://(?:[-a-zA-Z0-9_.]+\.)*youtube.(?:com|de|co.uk)/v/([-a-zA-Z0-9_]+))'];
28 $self->{'_PARAMS'} = {
29 'QUALITY' => ['normal', 'Quality of the video to download.', {
30 'normal' => 'standard resolution flash video',
31 'high' => 'higher resolution flash video',
32 'h264' => 'high resolution MPEG4 video',
33 'hd' => 'HD720 resolution'}],
34 'USERNAME' => ['', 'Username to use for YouTube login'],
35 'PASSWORD' => ['', 'Password to use for YouTube login']};
38 $self->_prepare_parameters();
52 my $jar = HTTP::Cookies->new;
53 my $ua = LWP::UserAgent->new('agent' => 'Mozilla/5.0');
56 my $quality = $self->_getval('QUALITY');
58 'hd' => [22, 35, 18, 34, 6, 5],
59 'h264' => [18, 34, 22, 35, 6, 5],
60 'high' => [34, 35, 18, 22, 6, 5],
61 'normal' => [6, 5, 34, 35, 18, 22]);
67 $metadata->{'URL'} = $url;
68 $metadata->{'ID'} = $2;
69 $metadata->{'TYPE'} = 'video';
70 $metadata->{'SOURCE'} = $self->{'NAME'};
71 $metadata->{'TITLE'} = undef;
72 $metadata->{'DLURL'} = undef;
74 $preflist = $preflist{$quality};
75 $self->debug("Quality: %s, preflist: [%s]", $quality, join(", ", @{$preflist}));
77 $videourl = sprintf('http://www.youtube.com/watch?v=%s', $2);
79 unless(defined($r = $ua->get($videourl))) {
80 $self->error('Could not download %s', $url);
84 if ($r->base->as_string() =~ m,/verify_age,) {
85 $self->debug('Video requires age verification');
86 $ua->cookie_jar($jar);
87 $r = $self->__login($videourl, $ua);
89 $self->error('Could not log into YouTube');
93 $content = $r->content();
95 $p = HTML::TokeParser->new(\$content);
97 while ($tag = $p->get_tag('div', 'meta', 'script')) {
98 if ('meta' eq $tag->[0]) {
99 if ('title' eq $tag->[1]->{'name'}) {
100 $metadata->{'TITLE'} = $tag->[1]->{'content'};
101 $self->debug('Title found: %s', $metadata->{'TITLE'});
103 } elsif ('script' eq $tag->[0]) {
105 $self->debug("Found script: %s", $e);
106 if ($e =~ m|\x22fmt_url_map\x22\s*:\s*\x22([^\x22]+)\x22|) {
107 $self->debug("Found fmt_url_map: %s", $1);
111 $urls =~ s/%([[:xdigit:]]{2})/chr(hex($1))/ge;
112 %urls = split(/[\|,]/, $urls);
113 $self->debug("Found quality levels [%s]", join(", ", keys(%urls)));
115 foreach (keys(%urls)) {
117 $self->debug('Found flv,h264,large: %s', $urls{$_});
119 $self->debug('Found flv,h264: %s', $urls{$_});
121 $self->debug('Found mp4,h264,large: %s', $urls{$_});
123 $self->debug('Found mp4,h264: %s', $urls{$_});
125 $self->debug('Found flv,flv: %s', $urls{$_});
127 $self->error('Unknown tag %s: %s', $_, $urls{$_});
131 foreach (@{$preflist}) {
132 if (exists($urls{$_})) {
133 $self->debug("Selected URL with quality level %s", $_);
134 $metadata->{'DLURL'} = $urls{$_};
139 $self->debug('URL found: %s', $metadata->{'DLURL'});
141 } elsif ('div' eq $tag->[0]) {
142 if (exists($tag->[1]->{'class'}) and ('errorBox' eq $tag->[1]->{'class'})) {
143 $self->error("Could not get video data for youtube %s: %s",
144 $metadata->{'ID'}, $p->get_trimmed_text());
150 unless(defined($metadata->{'DLURL'}) && defined($metadata->{'TITLE'})) {
151 $self->error('Could not determine download URL');
160 my $videourl = shift;
162 my $user = $self->_getval('USERNAME');
163 my $pass = $self->_getval('PASSWORD');
175 $jar->scan(sub { $found = 1 if ( $key eq $_[1]) });
180 if (($user eq '') or ($pass eq '')) {
181 $self->error('No username or password defined for YouTube');
185 $self->debug('Logging in');
186 $r = $ua->post('http://www.youtube.com/signup', { 'next' => '/', 'current_form' => 'login', 'action_login' => '1', 'username' => $user, 'password' => $pass });
188 unless(check_cookie($ua->cookie_jar, 'LOGIN_INFO')) {
189 $self->error('Could not log into YouTube');
193 $r = $ua->get($videourl);
194 if ($r->base->as_string() =~ m,/verify_age,) {
195 $self->debug("Looking for session token...");
196 $c = $r->decoded_content();
197 $p = HTML::TokeParser->new(\$c);
198 while (my $tag = $p->get_tag('script')) {
199 if ($p->get_text() =~ /gXSRF_token = '(.+)'/) {
205 unless(defined($token)) {
206 $self->error("Could not find session token");
210 $self->debug('Authenticating session...');
211 $r = $ua->post($r->base->as_string, { 'next_url' => $r->base->path, 'action_confirm' => 'Confirm Birth Date', 'session_token' => $token });
214 unless(check_cookie($ua->cookie_jar, 'is_adult')) {
215 $self->error('Could not authenticate session');
219 return $ua->get($videourl);