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);
15 use HTML::Entities qw(decode_entities);
18 use videosite::JSArrayParser;
23 'insane' => [38, 37, 22, 35, 18, 34, 6, 5, 43],
24 'hd' => [37, 22, 35, 18, 34, 6, 5, 38, 43],
25 'h264' => [18, 34, 37, 22, 35, 6, 5, 38, 43],
26 'high' => [34, 35, 18, 37, 22, 6, 5, 38, 43],
27 'normal' => [6, 5, 34, 35, 18, 22, 37, 38, 43]);
29 # Container/Video codec/Audio codec/Resolution
30 5 => 'FLV/Sorenson/MP3/240p',
31 6 => 'FLV/Sorenson/MP3/270p',
32 13 => '3GP/MPEG4-Visual/144p', # 0.5MBit
33 17 => '3GP/MPEG4-Visual/144p', # 2MBit
34 18 => 'MP4/H264/AAC/360p', # isommp42, Baseline
35 22 => 'MP4/H264/AAC/720p', # isommp42, High
36 34 => 'FLV/H264/AAC/360p', # Main
37 35 => 'FLV/H264/AAC/480p', # Main
38 36 => '3GP/MPEG4-Visual/240p',
39 37 => 'MP4/H264/AAC/1080p', # High
40 38 => 'MP4/H264/AAC/3072p', # High
41 43 => 'WebM/VP8/Vorbis/360p',
42 44 => 'WebM/VP8/Vorbis/480p',
43 45 => 'WebM/VP8/Vorbis/720p',
44 46 => 'WebM/VP8/Vorbis/1080p/3D', # effective 540p
45 82 => 'MP4/H264/AAC/360p/3D', # isomavc1mp42
46 83 => 'MP4/H264/AAC/240p/3D', # isomavc1mp42
47 84 => 'MP4/H264/AAC/720p/3D', # isomavc1mp42
48 85 => 'MP4/H264/AAC/1080p/3D', # isomavc1mp42, effective 540p
49 100 => 'WebM/VP8/Vorbis/360p/3D',
50 101 => 'WebM/VP8/Vorbis/480p/3D',
51 102 => 'WebM/VP8/Vorbis/720p/3D',
56 my $self = $class->SUPER::new();
58 $self->{'NAME'} = 'youtube';
59 $self->{_SELFTESTURL} = 'http://www.youtube.com/watch?v=dMH0bHeiRNg';
60 $self->{_SELFTESTTITLE} = 'Evolution of Dance - By Judson Laipply';
61 $self->{'PATTERNS'} = ['(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/watch(?:_popup)?\?.*?v=([-a-zA-Z0-9_]+))',
62 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/watch\#\!v=([-a-zA-Z0-9_]+))',
63 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/v/([-a-zA-Z0-9_]+))',
64 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/embed/([-a-zA-Z0-9_]+))',
65 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/user/[[:alnum:]]+\?v=([-a-zA-Z0-9_]+))',
66 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/(?:user/)?[[:alnum:]]+#p/(?:\w+/)+\d+/([-a-zA-Z0-9_]+))',
67 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtu\.be/watch\?v=([-a-zA-Z0-9_]+))',
68 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtu\.be/([-a-zA-Z0-9_]+))',
69 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/user/\w+\?.*/([-a-zA-Z0-9_]+))'];
70 $self->{'_PARAMS'} = {
71 'QUALITY' => ['normal', 'Quality of the video to download.', {
72 'normal' => 'standard resolution flash video',
73 'high' => 'higher resolution flash video',
74 'h264' => 'high resolution MPEG4 video',
75 'hd' => 'HD720 resolution'}],
76 'USERNAME' => ['', 'Username to use for YouTube login'],
77 'PASSWORD' => ['', 'Password to use for YouTube login'],
78 'HTTPS' => [1, 'Whether to use HTTPS (if available) to connect to YouTube']};
81 $self->_prepare_parameters();
97 $self->debug("Matched id %s from pattern %s", $id, $pattern);
99 $res = $self->_parse_by_video_info($url, $id);
100 if (defined($res) && ref($res)) {
103 $res = $self->_parse_by_scrape($url, $id);
109 sub _parse_by_video_info {
113 my $quality = $self->_getval('QUALITY');
116 my $ua = $self->ua();
122 $metadata->{'URL'} = $url;
123 $metadata->{'ID'} = $id;
124 $metadata->{'TYPE'} = 'video';
125 $metadata->{'SOURCE'} = $self->{'NAME'};
126 $metadata->{'TITLE'} = undef;
127 $metadata->{'DLURL'} = undef;
129 $preflist = $preflist{$quality};
130 $self->debug("Quality: %s, preflist: [%s]", $quality, join(", ", @{$preflist}));
132 $videourl = sprintf('%s://www.youtube.com/get_video_info?video_id=%s&eurl=%s',
133 $self->_getval('HTTPS')?'https':'http', $id, 'http%3A%2F%2Fwww%2Eyoutube%2Ecom%2F');
134 $self->debug("Video info URL: %s", $videourl);
136 $r = $ua->get($videourl);
137 unless($r->is_success()) {
138 $self->debug('Could not download %s: %s', $videourl, $r->code());
142 $content = $r->content();
143 $self->debug('Content from get_video_info: %s', $content);
146 $content = $self->decode_querystring($content);
148 if ($content->{'status'} ne 'ok') {
149 $self->debug("Non OK status code found: %s", $content->{'status'});
153 if (exists($content->{'fmt_url_map'})) {
155 $urls = $self->decode_hexurl($content->{'fmt_url_map'});
156 $urls = { split /[\|,]/, $urls };
157 } elsif (exists($content->{'url_encoded_fmt_stream_map'})) {
158 $urls = $self->_decode_url_encoded_fmt_stream_map($content->{'url_encoded_fmt_stream_map'}, 1);
160 $self->debug("No URL data found");
164 unless(exists($content->{'title'})) {
165 $self->debug("No title found");
169 $self->__pick_url($urls, $preflist, $metadata);
171 $metadata->{'TITLE'} = $content->{'title'};
172 $metadata->{'TITLE'} =~ s/\+/ /g;
173 $metadata->{'TITLE'} = $self->decode_hexurl($metadata->{'TITLE'});
174 $metadata->{'TITLE'} = decode("utf8", $metadata->{'TITLE'});
176 $self->debug('Title found: %s', $metadata->{'TITLE'});
178 unless(defined($metadata->{'DLURL'}) && defined($metadata->{'TITLE'})) {
179 $self->error('Could not determine download URL');
186 sub _parse_by_scrape {
195 my $ua = $self->ua();
198 my $quality = $self->_getval('QUALITY');
202 $metadata->{'URL'} = $url;
203 $metadata->{'ID'} = $id;
204 $metadata->{'TYPE'} = 'video';
205 $metadata->{'SOURCE'} = $self->{'NAME'};
206 $metadata->{'TITLE'} = undef;
207 $metadata->{'DLURL'} = undef;
210 $preflist = $preflist{$quality};
211 $self->debug("Quality: %s, preflist: [%s]", $quality, join(", ", @{$preflist}));
213 $videourl = sprintf('%s://www.youtube.com/watch?v=%s', $self->_getval('HTTPS')?'https':'http', $id);
215 unless(defined($r = $ua->get($videourl))) {
216 $self->error('Could not download %s', $url);
220 if ($r->base->as_string() =~ m,/verify_age,) {
221 $self->debug('Video requires age verification');
222 my @logindata = $self->__login($videourl, $ua);
224 unless(defined($r)) {
225 $self->error('Could not log into YouTube');
229 $content = $r->content();
231 $p = HTML::TokeParser->new(\$content);
233 SWF_ARGS: while ($tag = $p->get_tag('div', 'meta', 'script')) {
234 if ('meta' eq $tag->[0]) {
235 if (exists($tag->[1]->{'name'}) and ('title' eq $tag->[1]->{'name'})) {
236 $metadata->{'TITLE'} = $tag->[1]->{'content'};
237 # Convert HTML entities in the title. This is a bit convoluted.
238 $metadata->{'TITLE'} = decode_entities(
239 decode("utf8", $metadata->{'TITLE'}));
241 $self->debug('Title found: %s', $metadata->{'TITLE'});
243 } elsif ('script' eq $tag->[0]) {
247 $self->debug("Found script: %s", $e);
249 if ($e =~ m|ytplayer\.config\s*=\s*(.+)(?:;)$|) {
251 $self->debug("Found PLAYER_CONFIG: %s", $args);
253 $jsp = videosite::JSArrayParser->new();
254 $self->debug("Using %s to parse", ref($jsp));
255 $r = $jsp->parse($args);
257 unless(defined($r)) {
258 $self->error("Found information hash, but could not parse");
262 if (exists($r->{'args'}) and exists($r->{'args'}->{'ps'}) and ($r->{'args'}->{'ps'} eq 'live')) {
263 $self->error("Video URL seems to point to a live stream, cannot save this");
267 if (exists($r->{'args'}) and exists($r->{'args'}->{'fmt_url_map'}) and ($r->{'args'}->{'fmt_url_map'} ne '')) {
268 my $urls = $r->{'args'}->{'fmt_url_map'};
270 $self->debug("Video has fmt_url_map: %s", $urls);
272 %urls = split(/[\|,]/, $urls);
273 foreach (keys(%urls)) {
274 $urls{$_} = $self->decode_hexurl($urls{$_});
276 $self->debug("Pagetype: 2011 (PLAYER_CONFIG), fmt_url_map");
277 } elsif (exists($r->{'args'}) and exists($r->{'args'}->{'url_encoded_fmt_stream_map'}) and ($r->{'args'}->{'url_encoded_fmt_stream_map'} ne '')) {
278 %urls = %{$self->_decode_url_encoded_fmt_stream_map($r->{'args'}->{'url_encoded_fmt_stream_map'}, 0)};
280 $self->debug("Pagetype: 2011 (PLAYER_CONFIG), url_encoded_fmt_stream_map");
282 $self->error('fmt_url_map not found in PLAYER_CONFIG');
285 } elsif ($e =~ m|yt\.playerConfig\s*=\s*(.+);\n|) {
287 $self->debug("Found yt.playerConfig: %s", $args);
289 $jsp = videosite::JSArrayParser->new();
290 $self->debug("Using %s to parse", ref($jsp));
291 $r = $jsp->parse($args);
293 unless(defined($r)) {
294 $self->error("Found information hash, but could not parse");
298 if (exists($r->{'args'}) and exists($r->{'args'}->{'ps'}) and ($r->{'args'}->{'ps'} eq 'live')) {
299 $self->error("Video URL seems to point to a live stream, cannot save this");
303 if (exists($r->{'args'}) and exists($r->{'args'}->{'url_encoded_fmt_stream_map'}) and ($r->{'args'}->{'url_encoded_fmt_stream_map'} ne '')) {
304 %urls = %{$self->_decode_url_encoded_fmt_stream_map($r->{'args'}->{'url_encoded_fmt_stream_map'}, 0)};
306 $self->debug("Pagetype: 2012 (yt.playerConfig), url_encoded_fmt_stream_map");
308 $self->error('url_map not found in yt.playerConfig');
315 $self->__pick_url(\%urls, $preflist, $metadata);
318 } elsif ('div' eq $tag->[0]) {
319 if (exists($tag->[1]->{'id'}) and ('watch-player-unavailable-message-container' eq $tag->[1]->{'id'})) {
320 # Search forward to the next <div>
321 $tag = $p->get_tag('div');
322 $self->error("Could not get video data for youtube %s: %s",
323 $metadata->{'ID'}, $p->get_trimmed_text());
329 unless(defined($metadata->{'DLURL'}) && defined($metadata->{'TITLE'})) {
330 $self->error('Could not determine download URL');
339 my $videourl = shift;
341 my $user = $self->_getval('USERNAME');
342 my $pass = $self->_getval('PASSWORD');
354 $jar->scan(sub { $found = 1 if ( $key eq $_[1]) });
359 sub get_all_cookies {
364 $jar->scan(sub { $val .= "; " if !( $val eq "" ); $val .= "$_[1]=$_[2]" });
369 if (($user eq '') or ($pass eq '')) {
370 $self->error('No username or password defined for YouTube');
374 $self->debug('Logging in');
375 $r = $ua->get('https://www.google.com/accounts/ServiceLoginAuth?service=youtube');
376 unless($r->is_success()) {
377 $self->debug("Could not get login page (make sure your LWP supports HTTPS!)");
380 $c = $r->decoded_content();
381 $p = HTML::TokeParser->new(\$c);
382 while (my $tag = $p->get_tag('input')) {
383 $self->debug("%s", Dumper($tag));
384 if ($tag->[1]{name} eq 'GALX') {
385 $token = $tag->[1]{value};
389 $self->debug("GALX = %s", $token);
390 $r = $ua->post('https://www.google.com/accounts/ServiceLoginAuth?service=youtube', { 'service' => 'youtube', 'Email' => $user, 'Passwd' => $pass, 'GALX' => $token });
391 unless($r->is_success()) {
392 $self->debug("Could not get login page (make sure your LWP supports HTTPS!)");
395 $c = $r -> decoded_content();
396 $p = HTML::TokeParser->new(\$c);
397 while (my $tag = $p->get_tag('script')) {
398 if($p->get_text() =~ /location\.replace\("(.+)"\)/) {
400 $token =~ s/\\x([A-Fa-f0-9]{2})/pack('C', hex($1))/seg;
404 $r = $ua->get($token);
405 unless(check_cookie($ua->cookie_jar, 'LOGIN_INFO')) {
406 $self->error('Could not log into YouTube');
410 $self->debug("Got a cookie");
412 $r = $ua->get($videourl);
413 if ($r->base->as_string() =~ m,/verify_age,) {
414 $self->debug("Looking for session token...");
415 $c = $r->decoded_content();
416 $p = HTML::TokeParser->new(\$c);
417 while (my $tag = $p->get_tag('script')) {
418 if ($p->get_text() =~ /'XSRF_TOKEN': '(.+)'/) {
424 unless(defined($token)) {
425 $self->error("Could not find session token");
429 $self->debug('Authenticating session...');
430 $r = $ua->post($r->base->as_string, { 'next_url' => $r->base->path, 'action_confirm' => 'Confirm Birth Date', 'session_token' => $token });
433 # Apparently there is no longer a specific "is_adult" cookie
434 # or, by the looks of it, anything similar
436 # unless(check_cookie($ua->cookie_jar, 'is_adult')) {
437 # $self->error('Could not authenticate session');
441 my $cookie = get_all_cookies($ua->cookie_jar);
442 return ($ua->get($videourl), $cookie);
445 # Take an encoded url_encoded_fmt_stream_map and return a hash
446 # matching video IDs to download URLs
447 sub _decode_url_encoded_fmt_stream_map {
450 my $dataencoded = shift;
454 $data = $self->decode_hexurl($data) if $dataencoded;
456 # - Split the decoded string into segments (along ,)
457 # - Interpret each segment as a concatenated key-value list (key and value separated by =, pairs separated by &
458 # - URL-decode each key and value _again_
460 # @data will be an array of hash references
462 @data = map { { map { $self->decode_hexurl($_) } split /[&=]/ } } split /,/, $data;
463 $self->debug("_decode_url_encoded_fmt_stream_map() decoded %s", Dumper(\@data));
465 # From each array entry, pick the itag and the url values and return that
466 # as a hash reference
470 if (exists($_->{'sig'})) {
471 $h->{$_->{'itag'}} = sprintf('%s&signature=%s', $_->{'url'}, $_->{'sig'});
473 $h->{$_->{'itag'}} = $_->{'url'};
485 my $preflist = shift;
486 my $metadata = shift;
488 foreach (keys(%{$urls})) {
489 if (exists($videoformats{$_})) {
490 $self->debug('Found URL for format %s (%s): %s', $_, $videoformats{$_}, $urls->{$_});
492 $self->error('Unknown format %s: %s', $_, $urls->{$_});
496 foreach (@{$preflist}) {
497 if (exists($urls->{$_})) {
498 $self->debug("Selected URL with quality level %s", $_);
499 $metadata->{'DLURL'} = $urls->{$_};
500 if (exists($videoformats{$_})) {
501 $metadata->{'FORMAT'} = $videoformats{$_};
503 $metadata->{'FORMAT'} = 'unknown';
509 $self->debug('URL found: %s', $metadata->{'DLURL'});