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);
17 use HTML::Entities qw(decode_entities);
20 use videosite::JSArrayParser;
26 my $self = $class->SUPER::new();
28 $self->{'NAME'} = 'youtube';
29 $self->{'PATTERNS'} = ['(http://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/watch\?(?:.+=.+&)*v=([-a-zA-Z0-9_]+))',
30 '(http://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/v/([-a-zA-Z0-9_]+))',
31 '(http://(?:[-a-zA-Z0-9_.]+\.)*youtu\.be/([-a-zA-Z0-9_]+))'];
32 $self->{'_PARAMS'} = {
33 'QUALITY' => ['normal', 'Quality of the video to download.', {
34 'normal' => 'standard resolution flash video',
35 'high' => 'higher resolution flash video',
36 'h264' => 'high resolution MPEG4 video',
37 'hd' => 'HD720 resolution'}],
38 'USERNAME' => ['', 'Username to use for YouTube login'],
39 'PASSWORD' => ['', 'Password to use for YouTube login']};
42 $self->_prepare_parameters();
56 my $jar = HTTP::Cookies->new;
57 my $ua = LWP::UserAgent->new('agent' => 'Mozilla/5.0');
60 my $quality = $self->_getval('QUALITY');
62 'hd' => [37, 22, 35, 18, 34, 6, 5],
63 'h264' => [18, 34, 37, 22, 35, 6, 5],
64 'high' => [34, 35, 18, 37, 22, 6, 5],
65 'normal' => [6, 5, 34, 35, 18, 22, 37]);
72 $metadata->{'URL'} = $url;
73 $metadata->{'ID'} = $2;
74 $metadata->{'TYPE'} = 'video';
75 $metadata->{'SOURCE'} = $self->{'NAME'};
76 $metadata->{'TITLE'} = undef;
77 $metadata->{'DLURL'} = undef;
79 $self->debug("Matched id %s from pattern %s", $2, $pattern);
81 $preflist = $preflist{$quality};
82 $self->debug("Quality: %s, preflist: [%s]", $quality, join(", ", @{$preflist}));
84 $videourl = sprintf('http://www.youtube.com/watch?v=%s', $2);
86 unless(defined($r = $ua->get($videourl))) {
87 $self->error('Could not download %s', $url);
91 if ($r->base->as_string() =~ m,/verify_age,) {
92 $self->debug('Video requires age verification');
93 $ua->cookie_jar($jar);
94 $r = $self->__login($videourl, $ua);
96 $self->error('Could not log into YouTube');
100 $content = $r->content();
102 $p = HTML::TokeParser->new(\$content);
104 SWF_ARGS: while ($tag = $p->get_tag('div', 'meta', 'script')) {
105 if ('meta' eq $tag->[0]) {
106 if ('title' eq $tag->[1]->{'name'}) {
107 $metadata->{'TITLE'} = $tag->[1]->{'content'};
108 # Convert HTML entities in the title. This is a bit convoluted.
109 $metadata->{'TITLE'} = encode("utf8",
111 decode("utf8", $metadata->{'TITLE'})));
113 $self->debug('Title found: %s', $metadata->{'TITLE'});
115 } elsif ('script' eq $tag->[0]) {
117 $self->debug("Found script: %s", $e);
118 if ($e =~ m|\x27SWF_ARGS\x27:\s+(.+),|) {
122 $self->debug("Found SWF_ARGS: %s", $args);
123 $jsp = videosite::JSArrayParser->new();
124 $self->debug("Using %s to parse", ref($jsp));
125 $r = $jsp->parse($args);
127 unless(defined($r)) {
128 $self->error("Found information hash, but could not parse");
132 if (exists($r->{'fmt_url_map'}) and ($r->{'fmt_url_map'} ne '')) {
133 my $urls = $r->{'fmt_url_map'};
135 $self->debug("Video has fmt_url_map: %s", $urls);
137 $urls =~ s/%([[:xdigit:]]{2})/chr(hex($1))/ge;
138 %urls = split(/[\|,]/, $urls);
139 } elsif (exists($r->{'t'}) and ($r->{'t'} ne '')) {
140 my $thash = $r->{'t'};
142 if (exists($r->{'fmt_map'}) && ($r->{'fmt_map'} ne '')) {
143 my $fmt = $r->{'fmt_map'};
146 $self->debug('Video has fmt_map');
147 $fmt =~ s/%([[:xdigit:]]{2})/chr(hex($1))/ge;
148 @fmt = split(/,/, $fmt);
151 $urls{$_[0]} = sprintf('http://www.youtube.com/get_video?video_id=%s&fmt=%d&t=%s',
157 $urls{5} = sprintf('http://www.youtube.com/get_video?video_id=%s&t=%s',
162 $self->error('Neither fmt_url_map nor t found in video information hash');
165 $self->debug("Found quality levels [%s]", join(", ", keys(%urls)));
167 foreach (keys(%urls)) {
169 $self->debug('Found flv,h264,large: %s', $urls{$_});
171 $self->debug('Found flv,h264: %s', $urls{$_});
173 $self->debug('Found mp4,h264,720p: %s', $urls{$_});
175 $self->debug('Found mp4,h264,1080p: %s', $urls{$_});
177 $self->debug('Found mp4,h264: %s', $urls{$_});
179 $self->debug('Found flv,flv: %s', $urls{$_});
181 $self->error('Unknown tag %s: %s', $_, $urls{$_});
185 foreach (@{$preflist}) {
186 if (exists($urls{$_})) {
187 $self->debug("Selected URL with quality level %s", $_);
188 $metadata->{'DLURL'} = $urls{$_};
193 $self->debug('URL found: %s', $metadata->{'DLURL'});
196 } elsif ('div' eq $tag->[0]) {
197 if (exists($tag->[1]->{'class'}) and ('yt-alert-content' eq $tag->[1]->{'class'})) {
198 $self->error("Could not get video data for youtube %s: %s",
199 $metadata->{'ID'}, $p->get_trimmed_text());
205 unless(defined($metadata->{'DLURL'}) && defined($metadata->{'TITLE'})) {
206 $self->error('Could not determine download URL');
215 my $videourl = shift;
217 my $user = $self->_getval('USERNAME');
218 my $pass = $self->_getval('PASSWORD');
230 $jar->scan(sub { $found = 1 if ( $key eq $_[1]) });
235 if (($user eq '') or ($pass eq '')) {
236 $self->error('No username or password defined for YouTube');
240 $self->debug('Logging in');
241 $r = $ua->post('http://www.youtube.com/signup', { 'next' => '/', 'current_form' => 'login', 'action_login' => '1', 'username' => $user, 'password' => $pass });
243 unless(check_cookie($ua->cookie_jar, 'LOGIN_INFO')) {
244 $self->error('Could not log into YouTube');
248 $r = $ua->get($videourl);
249 if ($r->base->as_string() =~ m,/verify_age,) {
250 $self->debug("Looking for session token...");
251 $c = $r->decoded_content();
252 $p = HTML::TokeParser->new(\$c);
253 while (my $tag = $p->get_tag('script')) {
254 if ($p->get_text() =~ /gXSRF_token = '(.+)'/) {
260 unless(defined($token)) {
261 $self->error("Could not find session token");
265 $self->debug('Authenticating session...');
266 $r = $ua->post($r->base->as_string, { 'next_url' => $r->base->path, 'action_confirm' => 'Confirm Birth Date', 'session_token' => $token });
269 unless(check_cookie($ua->cookie_jar, 'is_adult')) {
270 $self->error('Could not authenticate session');
274 return $ua->get($videourl);