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 = $class->SUPER::new(
58 _SELFTESTURL => 'http://www.youtube.com/watch?v=dMH0bHeiRNg',
59 _SELFTESTTITLE => 'Evolution of Dance - By Judson Laipply',
60 PATTERNS => ['(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/watch(?:_popup)?\?.*?v=([-a-zA-Z0-9_]+))',
61 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/watch\#\!v=([-a-zA-Z0-9_]+))',
62 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/v/([-a-zA-Z0-9_]+))',
63 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/embed/([-a-zA-Z0-9_]+))',
64 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/user/[[:alnum:]]+\?v=([-a-zA-Z0-9_]+))',
65 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/(?:user/)?[[:alnum:]]+#p/(?:\w+/)+\d+/([-a-zA-Z0-9_]+))',
66 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtu\.be/watch\?v=([-a-zA-Z0-9_]+))',
67 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtu\.be/([-a-zA-Z0-9_]+))',
68 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/user/\w+\?.*/([-a-zA-Z0-9_]+))'],
70 QUALITY => ['normal', 'Quality of the video to download.', {
71 normal => 'standard resolution flash video',
72 high => 'higher resolution flash video',
73 h264 => 'high resolution MPEG4 video',
74 hd => 'HD720 resolution'}],
75 USERNAME => ['', 'Username to use for YouTube login'],
76 PASSWORD => ['', 'Password to use for YouTube login'],
77 HTTPS => [1, 'Whether to use HTTPS (if available) to connect to YouTube']
82 return bless($self, $class);
96 $self->debug("Matched id %s from pattern %s", $id, $pattern);
98 $res = $self->_parse_by_video_info($url, $id);
103 $self->debug("_parse_by_video_info failed with status noretry");
107 $self->debug("_parse_by_video_info failed with status retry");
108 $res = $self->_parse_by_scrape($url, $id);
115 # Try to get video information by using the API call.
117 # Returns hashref on success
118 # Returns undef on retryable error (try to scrape the website)
119 # Returns 0 on non-retryable error
121 sub _parse_by_video_info {
125 my $quality = $self->_getval('QUALITY');
128 my $ua = $self->ua();
134 $metadata->{'URL'} = $url;
135 $metadata->{'ID'} = $id;
136 $metadata->{'TYPE'} = 'video';
137 $metadata->{'SOURCE'} = $self->{'NAME'};
138 $metadata->{'TITLE'} = undef;
139 $metadata->{'DLURL'} = undef;
141 $preflist = $preflist{$quality};
142 $self->debug("Quality: %s, preflist: [%s]", $quality, join(", ", @{$preflist}));
144 $videourl = sprintf('%s://www.youtube.com/get_video_info?video_id=%s&eurl=%s',
145 $self->_getval('HTTPS')?'https':'http', $id, 'http%3A%2F%2Fwww%2Eyoutube%2Ecom%2F');
146 $self->debug("Video info URL: %s", $videourl);
148 $r = $ua->get($videourl);
149 unless($r->is_success()) {
150 $self->debug('Could not download %s: %s', $videourl, $r->code());
154 $content = $r->content();
155 $self->debug('Content from get_video_info: %s', $content);
158 $content = $self->decode_querystring($content);
160 $self->debug("Decoded get_video_info: %s", Dumper($content));
162 if ($content->{'status'} ne 'ok') {
163 $self->debug("Non OK status code found: %s", $content->{'status'});
167 # Check if this is live content
168 if (exists($content->{ps}) and ($content->{ps} eq 'live')) {
169 $self->error("Video URL seems to point to a live stream, cannot save this");
173 if (exists($content->{'fmt_url_map'})) {
175 $urls = $self->decode_hexurl($content->{'fmt_url_map'});
176 $urls = { split /[\|,]/, $urls };
177 } elsif (exists($content->{'url_encoded_fmt_stream_map'})) {
178 $urls = $self->_decode_url_encoded_fmt_stream_map($content->{'url_encoded_fmt_stream_map'}, 1);
180 $self->debug("No URL data found");
184 unless(exists($content->{'title'})) {
185 $self->debug("No title found");
189 $self->__pick_url($urls, $preflist, $metadata);
191 $metadata->{'TITLE'} = $content->{'title'};
192 $metadata->{'TITLE'} =~ s/\+/ /g;
193 $metadata->{'TITLE'} = $self->decode_hexurl($metadata->{'TITLE'});
194 $metadata->{'TITLE'} = decode("utf8", $metadata->{'TITLE'});
196 $self->debug('Title found: %s', $metadata->{'TITLE'});
198 unless(defined($metadata->{'DLURL'}) && defined($metadata->{'TITLE'})) {
199 $self->error('Could not determine download URL');
206 sub _parse_by_scrape {
215 my $ua = $self->ua();
218 my $quality = $self->_getval('QUALITY');
222 $metadata->{'URL'} = $url;
223 $metadata->{'ID'} = $id;
224 $metadata->{'TYPE'} = 'video';
225 $metadata->{'SOURCE'} = $self->{'NAME'};
226 $metadata->{'TITLE'} = undef;
227 $metadata->{'DLURL'} = undef;
230 $preflist = $preflist{$quality};
231 $self->debug("Quality: %s, preflist: [%s]", $quality, join(", ", @{$preflist}));
233 $videourl = sprintf('%s://www.youtube.com/watch?v=%s', $self->_getval('HTTPS')?'https':'http', $id);
235 unless(defined($r = $ua->get($videourl))) {
236 $self->error('Could not download %s', $url);
240 if ($r->base->as_string() =~ m,/verify_age,) {
241 $self->debug('Video requires age verification');
242 my @logindata = $self->__login($videourl, $ua);
244 unless(defined($r)) {
245 $self->error('Could not log into YouTube');
249 $content = $r->content();
251 $p = HTML::TokeParser->new(\$content);
253 SWF_ARGS: while ($tag = $p->get_tag('div', 'meta', 'script')) {
254 if ('meta' eq $tag->[0]) {
255 if (exists($tag->[1]->{'name'}) and ('title' eq $tag->[1]->{'name'})) {
256 $metadata->{'TITLE'} = $tag->[1]->{'content'};
257 # Convert HTML entities in the title. This is a bit convoluted.
258 $metadata->{'TITLE'} = decode_entities(
259 decode("utf8", $metadata->{'TITLE'}));
261 $self->debug('Title found: %s', $metadata->{'TITLE'});
263 } elsif ('script' eq $tag->[0]) {
267 $self->debug("Found script: %s", $e);
269 if ($e =~ m|ytplayer\.config\s*=\s*(.+);$|) {
271 $self->debug("Found PLAYER_CONFIG: %s", $args);
273 $jsp = videosite::JSArrayParser->new();
274 $self->debug("Using %s to parse", ref($jsp));
275 $r = $jsp->parse($args);
277 unless(defined($r)) {
278 $self->error("Found information hash, but could not parse");
282 if (exists($r->{'args'}) and exists($r->{'args'}->{'ps'}) and ($r->{'args'}->{'ps'} eq 'live')) {
283 $self->error("Video URL seems to point to a live stream, cannot save this");
287 if (exists($r->{'args'}) and exists($r->{'args'}->{'fmt_url_map'}) and ($r->{'args'}->{'fmt_url_map'} ne '')) {
288 my $urls = $r->{'args'}->{'fmt_url_map'};
290 $self->debug("Video has fmt_url_map: %s", $urls);
292 %urls = split(/[\|,]/, $urls);
293 foreach (keys(%urls)) {
294 $urls{$_} = $self->decode_hexurl($urls{$_});
296 $self->debug("Pagetype: 2011 (PLAYER_CONFIG), fmt_url_map");
297 } elsif (exists($r->{'args'}) and exists($r->{'args'}->{'url_encoded_fmt_stream_map'}) and ($r->{'args'}->{'url_encoded_fmt_stream_map'} ne '')) {
298 %urls = %{$self->_decode_url_encoded_fmt_stream_map($r->{'args'}->{'url_encoded_fmt_stream_map'}, 0)};
300 $self->debug("Pagetype: 2011 (PLAYER_CONFIG), url_encoded_fmt_stream_map");
302 $self->error('fmt_url_map not found in PLAYER_CONFIG');
305 } elsif ($e =~ m|yt\.playerConfig\s*=\s*(.+);\n|) {
307 $self->debug("Found yt.playerConfig: %s", $args);
309 $jsp = videosite::JSArrayParser->new();
310 $self->debug("Using %s to parse", ref($jsp));
311 $r = $jsp->parse($args);
313 unless(defined($r)) {
314 $self->error("Found information hash, but could not parse");
318 if (exists($r->{'args'}) and exists($r->{'args'}->{'ps'}) and ($r->{'args'}->{'ps'} eq 'live')) {
319 $self->error("Video URL seems to point to a live stream, cannot save this");
323 if (exists($r->{'args'}) and exists($r->{'args'}->{'url_encoded_fmt_stream_map'}) and ($r->{'args'}->{'url_encoded_fmt_stream_map'} ne '')) {
324 %urls = %{$self->_decode_url_encoded_fmt_stream_map($r->{'args'}->{'url_encoded_fmt_stream_map'}, 0)};
326 $self->debug("Pagetype: 2012 (yt.playerConfig), url_encoded_fmt_stream_map");
328 $self->error('url_map not found in yt.playerConfig');
335 $self->__pick_url(\%urls, $preflist, $metadata);
338 } elsif ('div' eq $tag->[0]) {
339 if (exists($tag->[1]->{'id'}) and ('watch-player-unavailable-message-container' eq $tag->[1]->{'id'})) {
340 # Search forward to the next <div>
341 $tag = $p->get_tag('div');
342 $self->error("Could not get video data for youtube %s: %s",
343 $metadata->{'ID'}, $p->get_trimmed_text());
349 unless(defined($metadata->{'DLURL'}) && defined($metadata->{'TITLE'})) {
350 $self->error('Could not determine download URL');
359 my $videourl = shift;
361 my $user = $self->_getval('USERNAME');
362 my $pass = $self->_getval('PASSWORD');
374 $jar->scan(sub { $found = 1 if ( $key eq $_[1]) });
379 sub get_all_cookies {
384 $jar->scan(sub { $val .= "; " if !( $val eq "" ); $val .= "$_[1]=$_[2]" });
389 if (($user eq '') or ($pass eq '')) {
390 $self->error('No username or password defined for YouTube');
394 $self->debug('Logging in');
395 $r = $ua->get('https://www.google.com/accounts/ServiceLoginAuth?service=youtube');
396 unless($r->is_success()) {
397 $self->debug("Could not get login page (make sure your LWP supports HTTPS!)");
400 $c = $r->decoded_content();
401 $p = HTML::TokeParser->new(\$c);
402 while (my $tag = $p->get_tag('input')) {
403 $self->debug("%s", Dumper($tag));
404 if ($tag->[1]{name} eq 'GALX') {
405 $token = $tag->[1]{value};
409 $self->debug("GALX = %s", $token);
410 $r = $ua->post('https://www.google.com/accounts/ServiceLoginAuth?service=youtube', { 'service' => 'youtube', 'Email' => $user, 'Passwd' => $pass, 'GALX' => $token });
411 unless($r->is_success()) {
412 $self->debug("Could not get login page (make sure your LWP supports HTTPS!)");
415 $c = $r -> decoded_content();
416 $p = HTML::TokeParser->new(\$c);
417 while (my $tag = $p->get_tag('script')) {
418 if($p->get_text() =~ /location\.replace\("(.+)"\)/) {
420 $token =~ s/\\x([A-Fa-f0-9]{2})/pack('C', hex($1))/seg;
424 $r = $ua->get($token);
425 unless(check_cookie($ua->cookie_jar, 'LOGIN_INFO')) {
426 $self->error('Could not log into YouTube');
430 $self->debug("Got a cookie");
432 $r = $ua->get($videourl);
433 if ($r->base->as_string() =~ m,/verify_age,) {
434 $self->debug("Looking for session token...");
435 $c = $r->decoded_content();
436 $p = HTML::TokeParser->new(\$c);
437 while (my $tag = $p->get_tag('script')) {
438 if ($p->get_text() =~ /'XSRF_TOKEN': '(.+)'/) {
444 unless(defined($token)) {
445 $self->error("Could not find session token");
449 $self->debug('Authenticating session...');
450 $r = $ua->post($r->base->as_string, { 'next_url' => $r->base->path, 'action_confirm' => 'Confirm Birth Date', 'session_token' => $token });
453 # Apparently there is no longer a specific "is_adult" cookie
454 # or, by the looks of it, anything similar
456 # unless(check_cookie($ua->cookie_jar, 'is_adult')) {
457 # $self->error('Could not authenticate session');
461 my $cookie = get_all_cookies($ua->cookie_jar);
462 return ($ua->get($videourl), $cookie);
465 # Take an encoded url_encoded_fmt_stream_map and return a hash
466 # matching video IDs to download URLs
467 sub _decode_url_encoded_fmt_stream_map {
470 my $dataencoded = shift;
474 $data = $self->decode_hexurl($data) if $dataencoded;
476 # - Split the decoded string into segments (along ,)
477 # - Interpret each segment as a concatenated key-value list (key and value separated by =, pairs separated by &
478 # - URL-decode each key and value _again_
480 # @data will be an array of hash references
482 @data = map { { map { $self->decode_hexurl($_) } split /[&=]/ } } split /,/, $data;
483 $self->debug("_decode_url_encoded_fmt_stream_map() decoded %s", Dumper(\@data));
485 # From each array entry, pick the itag and the url values and return that
486 # as a hash reference
490 if (exists($_->{'sig'})) {
491 $h->{$_->{'itag'}} = sprintf('%s&signature=%s', $_->{'url'}, $_->{'sig'});
493 $h->{$_->{'itag'}} = $_->{'url'};
505 my $preflist = shift;
506 my $metadata = shift;
508 foreach (keys(%{$urls})) {
509 if (exists($videoformats{$_})) {
510 $self->debug('Found URL for format %s (%s): %s', $_, $videoformats{$_}, $urls->{$_});
512 $self->error('Unknown format %s: %s', $_, $urls->{$_});
516 foreach (@{$preflist}) {
517 if (exists($urls->{$_})) {
518 $self->debug("Selected URL with quality level %s", $_);
519 $metadata->{'DLURL'} = $urls->{$_};
520 if (exists($videoformats{$_})) {
521 $metadata->{'FORMAT'} = $videoformats{$_};
523 $metadata->{'FORMAT'} = 'unknown';
529 $self->debug('URL found: %s', $metadata->{'DLURL'});