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 37 => 'MP4/H264/AAC/1080p', # High
39 38 => 'MP4/H264/AAC/3072p', # High
40 43 => 'WebM/VP8/Vorbis/360p',
41 44 => 'WebM/VP8/Vorbis/480p',
42 45 => 'WebM/VP8/Vorbis/720p',
43 46 => 'WebM/VP8/Vorbis/1080p/3D', # effective 540p
44 82 => 'MP4/H264/AAC/360p/3D', # isomavc1mp42
45 83 => 'MP4/H264/AAC/240p/3D', # isomavc1mp42
46 84 => 'MP4/H264/AAC/720p/3D', # isomavc1mp42
47 85 => 'MP4/H264/AAC/1080p/3D', # isomavc1mp42, effective 540p
48 100 => 'WebM/VP8/Vorbis/360p/3D',
49 101 => 'WebM/VP8/Vorbis/480p/3D',
50 102 => 'WebM/VP8/Vorbis/720p/3D',
55 my $self = $class->SUPER::new();
57 $self->{'NAME'} = 'youtube';
58 $self->{_SELFTESTURL} = 'http://www.youtube.com/watch?v=dMH0bHeiRNg';
59 $self->{_SELFTESTTITLE} = 'Evolution of Dance - By Judson Laipply';
60 $self->{'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)/user/[[:alnum:]]+\?v=([-a-zA-Z0-9_]+))',
64 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/(?:user/)?[[:alnum:]]+#p/(?:\w+/)+\d+/([-a-zA-Z0-9_]+))',
65 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtu\.be/watch\?v=([-a-zA-Z0-9_]+))',
66 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtu\.be/([-a-zA-Z0-9_]+))',
67 '(https?://(?:[-a-zA-Z0-9_.]+\.)*youtube\.(?:com|de|co.uk)/user/\w+\?.*/([-a-zA-Z0-9_]+))'];
68 $self->{'_PARAMS'} = {
69 'QUALITY' => ['normal', 'Quality of the video to download.', {
70 'normal' => 'standard resolution flash video',
71 'high' => 'higher resolution flash video',
72 'h264' => 'high resolution MPEG4 video',
73 'hd' => 'HD720 resolution'}],
74 'USERNAME' => ['', 'Username to use for YouTube login'],
75 'PASSWORD' => ['', 'Password to use for YouTube login'],
76 'HTTPS' => [1, 'Whether to use HTTPS (if available) to connect to YouTube']};
79 $self->_prepare_parameters();
95 $self->debug("Matched id %s from pattern %s", $id, $pattern);
97 $res = $self->_parse_by_video_info($url, $id);
98 if (defined($res) && ref($res)) {
101 $res = $self->_parse_by_scrape($url, $id);
107 sub _parse_by_video_info {
111 my $quality = $self->_getval('QUALITY');
114 my $ua = $self->ua();
120 $metadata->{'URL'} = $url;
121 $metadata->{'ID'} = $id;
122 $metadata->{'TYPE'} = 'video';
123 $metadata->{'SOURCE'} = $self->{'NAME'};
124 $metadata->{'TITLE'} = undef;
125 $metadata->{'DLURL'} = undef;
126 $metadata->{'COOKIE'} = undef;
128 $preflist = $preflist{$quality};
129 $self->debug("Quality: %s, preflist: [%s]", $quality, join(", ", @{$preflist}));
131 $videourl = sprintf('%s://www.youtube.com/get_video_info?video_id=%s&eurl=%s',
132 $self->_getval('HTTPS')?'https':'http', $id, 'http%3A%2F%2Fwww%2Eyoutube%2Ecom%2F');
133 $self->debug("Video info URL: %s", $videourl);
135 $r = $ua->get($videourl);
136 unless($r->is_success()) {
137 $self->debug('Could not download %s: %s', $videourl, $r->code());
141 $content = $r->content();
142 $self->debug('Content from get_video_info: %s', $content);
145 $content = $self->decode_querystring($content);
147 if ($content->{'status'} ne 'ok') {
148 $self->debug("Non OK status code found: %s", $content->{'status'});
152 if (exists($content->{'fmt_url_map'})) {
154 $urls = $self->decode_hexurl($content->{'fmt_url_map'});
155 $urls = { split /[\|,]/, $urls };
156 } elsif (exists($content->{'url_encoded_fmt_stream_map'})) {
157 $urls = $self->_decode_url_encoded_fmt_stream_map($content->{'url_encoded_fmt_stream_map'}, 1);
159 $self->debug("No URL data found");
163 unless(exists($content->{'title'})) {
164 $self->debug("No title found");
168 $self->__pick_url($urls, $preflist, $metadata);
170 $metadata->{'TITLE'} = $content->{'title'};
171 $metadata->{'TITLE'} =~ s/\+/ /g;
172 $metadata->{'TITLE'} = $self->decode_hexurl($metadata->{'TITLE'});
173 $metadata->{'TITLE'} = decode("utf8", $metadata->{'TITLE'});
175 $self->debug('Title found: %s', $metadata->{'TITLE'});
177 unless(defined($metadata->{'DLURL'}) && defined($metadata->{'TITLE'})) {
178 $self->error('Could not determine download URL');
185 sub _parse_by_scrape {
194 my $ua = $self->ua();
197 my $quality = $self->_getval('QUALITY');
201 $metadata->{'URL'} = $url;
202 $metadata->{'ID'} = $id;
203 $metadata->{'TYPE'} = 'video';
204 $metadata->{'SOURCE'} = $self->{'NAME'};
205 $metadata->{'TITLE'} = undef;
206 $metadata->{'DLURL'} = undef;
207 $metadata->{'COOKIE'} = 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 $metadata->{'COOKIE'} = $logindata[1];
225 unless(defined($r)) {
226 $self->error('Could not log into YouTube');
230 $content = $r->content();
232 $p = HTML::TokeParser->new(\$content);
234 SWF_ARGS: while ($tag = $p->get_tag('div', 'meta', 'script')) {
235 if ('meta' eq $tag->[0]) {
236 if (exists($tag->[1]->{'name'}) and ('title' eq $tag->[1]->{'name'})) {
237 $metadata->{'TITLE'} = $tag->[1]->{'content'};
238 # Convert HTML entities in the title. This is a bit convoluted.
239 $metadata->{'TITLE'} = decode_entities(
240 decode("utf8", $metadata->{'TITLE'}));
242 $self->debug('Title found: %s', $metadata->{'TITLE'});
244 } elsif ('script' eq $tag->[0]) {
248 $self->debug("Found script: %s", $e);
250 if ($e =~ m|\x27SWF_ARGS\x27:\s+(.+),|) {
253 $self->debug("Found SWF_ARGS: %s", $args);
254 $jsp = videosite::JSArrayParser->new();
255 $self->debug("Using %s to parse", ref($jsp));
256 $r = $jsp->parse($args);
258 unless(defined($r)) {
259 $self->error("Found information hash, but could not parse");
263 if (exists($r->{'fmt_url_map'}) and ($r->{'fmt_url_map'} ne '')) {
264 my $urls = $r->{'fmt_url_map'};
266 $self->debug("Video has fmt_url_map: %s", $urls);
268 $urls = $self->decode_hexurl($urls);
269 %urls = split(/[\|,]/, $urls);
270 $self->debug("Pagetype: old (SWF_ARGS), fmt_url_map");
272 } elsif (exists($r->{'t'}) and ($r->{'t'} ne '')) {
273 my $thash = $r->{'t'};
275 if (exists($r->{'fmt_map'}) && ($r->{'fmt_map'} ne '')) {
276 my $fmt = $r->{'fmt_map'};
279 $self->debug('Video has fmt_map');
280 $fmt = $self->decode_hexurl($fmt);
281 @fmt = split(/,/, $fmt);
284 $urls{$_[0]} = sprintf('http://www.youtube.com/get_video?video_id=%s&fmt=%d&t=%s',
289 $self->debug("Pagetype: 2009 (SWF_ARGS), t with fmt_map");
292 $urls{5} = sprintf('http://www.youtube.com/get_video?video_id=%s&t=%s',
295 $self->debug("Pagetype: 2009 (SWF_ARGS), t without fmt_map");
298 $self->error('Neither fmt_url_map nor t found in video information hash');
301 } elsif ($e =~ m|var swfHTML = .*fmt_url_map=([^\&]+)\&|) {
303 $self->debug("Video has fmt_url_map: %s", $urls);
305 $urls = $self->decode_hexurl($urls);
306 %urls = split(/[\|,]/, $urls);
307 $self->debug("Pagetype: 2010 (swfHTML), fmt_url_map");
308 } elsif ($e =~ m|\x27PLAYER_CONFIG\x27:\s+(.+)(?:\}\);)?|) {
310 $self->debug("Found PLAYER_CONFIG: %s", $args);
312 $jsp = videosite::JSArrayParser->new();
313 $self->debug("Using %s to parse", ref($jsp));
314 $r = $jsp->parse($args);
316 unless(defined($r)) {
317 $self->error("Found information hash, but could not parse");
321 if (exists($r->{'args'}) and exists($r->{'args'}->{'ps'}) and ($r->{'args'}->{'ps'} eq 'live')) {
322 $self->error("Video URL seems to point to a live stream, cannot save this");
326 if (exists($r->{'args'}) and exists($r->{'args'}->{'fmt_url_map'}) and ($r->{'args'}->{'fmt_url_map'} ne '')) {
327 my $urls = $r->{'args'}->{'fmt_url_map'};
329 $self->debug("Video has fmt_url_map: %s", $urls);
331 %urls = split(/[\|,]/, $urls);
332 foreach (keys(%urls)) {
333 $urls{$_} = $self->decode_hexurl($urls{$_});
335 $self->debug("Pagetype: 2011 (PLAYER_CONFIG), fmt_url_map");
336 } elsif (exists($r->{'args'}) and exists($r->{'args'}->{'url_encoded_fmt_stream_map'}) and ($r->{'args'}->{'url_encoded_fmt_stream_map'} ne '')) {
337 %urls = %{$self->_decode_url_encoded_fmt_stream_map($r->{'args'}->{'url_encoded_fmt_stream_map'}, 0)};
339 $self->debug("Pagetype: 2011 (PLAYER_CONFIG), url_encoded_fmt_stream_map");
341 $self->error('fmt_url_map not found in PLAYER_CONFIG');
347 $self->__pick_url(\%urls, $preflist, $metadata);
350 } elsif ('div' eq $tag->[0]) {
351 if (exists($tag->[1]->{'id'}) and ('watch-player-unavailable-message-container' eq $tag->[1]->{'id'})) {
352 # Search forward to the next <div>
353 $tag = $p->get_tag('div');
354 $self->error("Could not get video data for youtube %s: %s",
355 $metadata->{'ID'}, $p->get_trimmed_text());
361 unless(defined($metadata->{'DLURL'}) && defined($metadata->{'TITLE'})) {
362 $self->error('Could not determine download URL');
371 my $videourl = shift;
373 my $user = $self->_getval('USERNAME');
374 my $pass = $self->_getval('PASSWORD');
386 $jar->scan(sub { $found = 1 if ( $key eq $_[1]) });
391 sub get_all_cookies {
396 $jar->scan(sub { $val .= "; " if !( $val eq "" ); $val .= "$_[1]=$_[2]" });
401 if (($user eq '') or ($pass eq '')) {
402 $self->error('No username or password defined for YouTube');
406 $self->debug('Logging in');
407 $r = $ua->get('https://www.google.com/accounts/ServiceLoginAuth?service=youtube');
408 unless($r->is_success()) {
409 $self->debug("Could not get login page (make sure your LWP supports HTTPS!)");
412 $c = $r->decoded_content();
413 $p = HTML::TokeParser->new(\$c);
414 while (my $tag = $p->get_tag('input')) {
415 $self->debug("%s", Dumper($tag));
416 if ($tag->[1]{name} eq 'GALX') {
417 $token = $tag->[1]{value};
421 $self->debug("GALX = %s", $token);
422 $r = $ua->post('https://www.google.com/accounts/ServiceLoginAuth?service=youtube', { 'service' => 'youtube', 'Email' => $user, 'Passwd' => $pass, 'GALX' => $token });
423 unless($r->is_success()) {
424 $self->debug("Could not get login page (make sure your LWP supports HTTPS!)");
427 $c = $r -> decoded_content();
428 $p = HTML::TokeParser->new(\$c);
429 while (my $tag = $p->get_tag('script')) {
430 if($p->get_text() =~ /location\.replace\("(.+)"\)/) {
432 $token =~ s/\\x([A-Fa-f0-9]{2})/pack('C', hex($1))/seg;
436 $r = $ua->get($token);
437 unless(check_cookie($ua->cookie_jar, 'LOGIN_INFO')) {
438 $self->error('Could not log into YouTube');
442 $self->debug("Got a cookie");
444 $r = $ua->get($videourl);
445 if ($r->base->as_string() =~ m,/verify_age,) {
446 $self->debug("Looking for session token...");
447 $c = $r->decoded_content();
448 $p = HTML::TokeParser->new(\$c);
449 while (my $tag = $p->get_tag('script')) {
450 if ($p->get_text() =~ /'XSRF_TOKEN': '(.+)'/) {
456 unless(defined($token)) {
457 $self->error("Could not find session token");
461 $self->debug('Authenticating session...');
462 $r = $ua->post($r->base->as_string, { 'next_url' => $r->base->path, 'action_confirm' => 'Confirm Birth Date', 'session_token' => $token });
465 # Apparently there is no longer a specific "is_adult" cookie
466 # or, by the looks of it, anything similar
468 # unless(check_cookie($ua->cookie_jar, 'is_adult')) {
469 # $self->error('Could not authenticate session');
473 my $cookie = get_all_cookies($ua->cookie_jar);
474 return ($ua->get($videourl), $cookie);
477 # Take an encoded url_encoded_fmt_stream_map and return a hash
478 # matching video IDs to download URLs
479 sub _decode_url_encoded_fmt_stream_map {
482 my $dataencoded = shift;
485 $data = $self->decode_hexurl($data) if $dataencoded;
487 # - Split the decoded string into segments (along ,)
488 # - Interpret each segment as a concatenated key-value list (key and value separated by =, pairs separated by &
489 # - URL-decode each key and value _again_
491 # @data will be an array of hash references
493 @data = map { { map { $self->decode_hexurl($_) } split /[&=]/ } } split /,/, $data;
494 $self->debug("_decode_url_encoded_fmt_stream_map() decoded %s", Dumper(\@data));
496 # From each array entry, pick the itag and the url values and return that
497 # as a hash reference
499 return { map { $_->{'itag'}, $_->{'url'} } @data };
507 my $preflist = shift;
508 my $metadata = shift;
510 foreach (keys(%{$urls})) {
511 if (exists($videoformats{$_})) {
512 $self->debug('Found URL for format %s (%s): %s', $_, $videoformats{$_}, $urls->{$_});
514 $self->error('Unknown format %s: %s', $_, $urls->{$_});
518 foreach (@{$preflist}) {
519 if (exists($urls->{$_})) {
520 $self->debug("Selected URL with quality level %s", $_);
521 $metadata->{'DLURL'} = $urls->{$_};
522 if (exists($videoformats{$_})) {
523 $metadata->{'FORMAT'} = $videoformats{$_};
525 $metadata->{'FORMAT'} = 'unknown';
531 $self->debug('URL found: %s', $metadata->{'DLURL'});