]> jfr.im git - yt-dlp.git/blame - yt_dlp/extractor/vk.py
[extractors] Use new framework for existing embeds (#4307)
[yt-dlp.git] / yt_dlp / extractor / vk.py
CommitLineData
51815886 1import collections
59f63c8f 2import hashlib
60d142aa 3import re
60d142aa
JMF
4
5from .common import InfoExtractor
59f63c8f
M
6from .dailymotion import DailymotionIE
7from .odnoklassniki import OdnoklassnikiIE
8from .pladform import PladformIE
9from .vimeo import VimeoIE
10from .youtube import YoutubeIE
059cd768 11from ..compat import compat_urlparse
60d142aa 12from ..utils import (
9032dc28 13 ExtractorError,
59f63c8f 14 clean_html,
2d19fb50 15 get_element_by_class,
bf4b3b6b 16 int_or_none,
1cc79574 17 orderedSet,
ad1bc71a 18 str_or_none,
8117df4c 19 str_to_int,
60d142aa 20 unescapeHTML,
a7ee8a00 21 unified_timestamp,
59f63c8f 22 update_url_query,
3052a30d 23 url_or_none,
6e6bc8da 24 urlencode_postdata,
1cc79574 25)
60d142aa
JMF
26
27
2d19fb50
S
28class VKBaseIE(InfoExtractor):
29 _NETRC_MACHINE = 'vk'
30
59f63c8f
M
31 def _download_webpage_handle(self, url_or_request, video_id, *args, fatal=True, **kwargs):
32 response = super()._download_webpage_handle(url_or_request, video_id, *args, fatal=fatal, **kwargs)
33 challenge_url, cookie = response[1].geturl() if response else '', None
34 if challenge_url.startswith('https://vk.com/429.html?'):
35 cookie = self._get_cookies(challenge_url).get('hash429')
36 if not cookie:
37 return response
38
39 hash429 = hashlib.md5(cookie.value.encode('ascii')).hexdigest()
40 self._request_webpage(
41 update_url_query(challenge_url, {'key': hash429}), video_id, fatal=fatal,
42 note='Resolving WAF challenge', errnote='Failed to bypass WAF challenge')
43 return super()._download_webpage_handle(url_or_request, video_id, *args, fatal=True, **kwargs)
44
52efa4b3 45 def _perform_login(self, username, password):
2d19fb50
S
46 login_page, url_handle = self._download_webpage_handle(
47 'https://vk.com', None, 'Downloading login page')
48
49 login_form = self._hidden_inputs(login_page)
50
51 login_form.update({
52 'email': username.encode('cp1251'),
53 'pass': password.encode('cp1251'),
54 })
55
e3c1266f
S
56 # vk serves two same remixlhk cookies in Set-Cookie header and expects
57 # first one to be actually set
58 self._apply_first_set_cookie_header(url_handle, 'remixlhk')
2d19fb50
S
59
60 login_page = self._download_webpage(
f0ffaa16 61 'https://vk.com/login', None,
e4d95865 62 note='Logging in',
2d19fb50
S
63 data=urlencode_postdata(login_form))
64
65 if re.search(r'onLoginFailed', login_page):
66 raise ExtractorError(
67 'Unable to login, incorrect username and/or password', expected=True)
68
3c989818 69 def _download_payload(self, path, video_id, data, fatal=True):
59f63c8f 70 endpoint = f'https://vk.com/{path}.php'
3c989818
RA
71 data['al'] = 1
72 code, payload = self._download_json(
59f63c8f
M
73 endpoint, video_id, data=urlencode_postdata(data), fatal=fatal,
74 headers={
75 'Referer': endpoint,
76 'X-Requested-With': 'XMLHttpRequest',
77 })['payload']
3c989818
RA
78 if code == '3':
79 self.raise_login_required()
80 elif code == '8':
81 raise ExtractorError(clean_html(payload[0][1:-1]), expected=True)
82 return payload
83
2d19fb50
S
84
85class VKIE(VKBaseIE):
1ecb5d1d
S
86 IE_NAME = 'vk'
87 IE_DESC = 'VK'
bfd973ec 88 _EMBED_REGEX = [r'<iframe[^>]+?src=(["\'])(?P<url>https?://vk\.com/video_ext\.php.+?)\1']
cf9cf7dd
S
89 _VALID_URL = r'''(?x)
90 https?://
91 (?:
04e88ca2 92 (?:
bdafd88d 93 (?:(?:m|new)\.)?vk\.com/video_|
04e88ca2 94 (?:www\.)?daxab.com/
95 )
96 ext\.php\?(?P<embed_query>.*?\boid=(?P<oid>-?\d+).*?\bid=(?P<id>\d+).*)|
cf9cf7dd 97 (?:
21df2117 98 (?:(?:m|new)\.)?vk\.com/(?:.+?\?.*?z=)?(?:video|clip)|
04e88ca2 99 (?:www\.)?daxab.com/embed/
cf9cf7dd 100 )
af3cbd87 101 (?P<videoid>-?\d+_\d+)(?:.*\blist=(?P<list_id>([\da-f]+)|(ln-[\da-zA-Z]+)))?
cf9cf7dd
S
102 )
103 '''
bfd973ec 104 # https://help.sibnet.ru/?sibnet_video_embed
105 _EMBED_REGEX = [r'<iframe\b[^>]+\bsrc=(["\'])(?P<url>(?:https?:)?//video\.sibnet\.ru/shell\.php\?.*?\bvideoid=\d+.*?)\1']
9032dc28
S
106 _TESTS = [
107 {
108 'url': 'http://vk.com/videos-77521?z=video-77521_162222515%2Fclub77521',
9032dc28 109 'info_dict': {
220828f2 110 'id': '-77521_162222515',
09f934b0 111 'ext': 'mp4',
9032dc28 112 'title': 'ProtivoGunz - Хуёвая песня',
36300346 113 'uploader': 're:(?:Noize MC|Alexander Ilyashenko).*',
59f63c8f 114 'uploader_id': '39545378',
9032dc28 115 'duration': 195,
ad1bc71a 116 'timestamp': 1329049880,
42e1ff86 117 'upload_date': '20120212',
59f63c8f
M
118 'comment_count': int,
119 'like_count': int,
120 'thumbnail': r're:https?://.+\.jpg$',
9032dc28 121 },
59f63c8f 122 'params': {'skip_download': 'm3u8'},
60d142aa 123 },
9032dc28 124 {
c52331f3 125 'url': 'http://vk.com/video205387401_165548505',
9032dc28 126 'info_dict': {
220828f2 127 'id': '205387401_165548505',
9032dc28 128 'ext': 'mp4',
c52331f3 129 'title': 'No name',
ad1bc71a
RA
130 'uploader': 'Tom Cruise',
131 'uploader_id': '205387401',
c52331f3 132 'duration': 9,
ad1bc71a
RA
133 'timestamp': 1374364108,
134 'upload_date': '20130720',
59f63c8f
M
135 'comment_count': int,
136 'like_count': int,
137 'thumbnail': r're:https?://.+\.jpg$',
9032dc28
S
138 }
139 },
ca97a56e
S
140 {
141 'note': 'Embedded video',
3c989818 142 'url': 'https://vk.com/video_ext.php?oid=-77521&id=162222515&hash=87b046504ccd8bfa',
ca97a56e 143 'info_dict': {
3c989818 144 'id': '-77521_162222515',
ca97a56e 145 'ext': 'mp4',
3c989818
RA
146 'uploader': 're:(?:Noize MC|Alexander Ilyashenko).*',
147 'title': 'ProtivoGunz - Хуёвая песня',
148 'duration': 195,
149 'upload_date': '20120212',
150 'timestamp': 1329049880,
59f63c8f
M
151 'uploader_id': '39545378',
152 'thumbnail': r're:https?://.+\.jpg$',
04e88ca2 153 },
59f63c8f 154 'params': {'skip_download': 'm3u8'},
ca97a56e 155 },
9032dc28 156 {
c52331f3
WS
157 # VIDEO NOW REMOVED
158 # please update if you find a video whose URL follows the same pattern
9032dc28
S
159 'url': 'http://vk.com/video-8871596_164049491',
160 'md5': 'a590bcaf3d543576c9bd162812387666',
161 'note': 'Only available for registered users',
162 'info_dict': {
220828f2 163 'id': '-8871596_164049491',
9032dc28
S
164 'ext': 'mp4',
165 'uploader': 'Триллеры',
57bdc730 166 'title': '► Бойцовский клуб / Fight Club 1999 [HD 720]',
9032dc28 167 'duration': 8352,
8117df4c
S
168 'upload_date': '20121218',
169 'view_count': int,
9032dc28 170 },
3c989818 171 'skip': 'Removed',
ca97a56e 172 },
57bdc730
S
173 {
174 'url': 'http://vk.com/hd_kino_mania?z=video-43215063_168067957%2F15c66b9b533119788d',
57bdc730 175 'info_dict': {
220828f2 176 'id': '-43215063_168067957',
57bdc730 177 'ext': 'mp4',
3c989818 178 'uploader': 'Bro Mazter',
57bdc730
S
179 'title': ' ',
180 'duration': 7291,
42e1ff86 181 'upload_date': '20140328',
3c989818
RA
182 'uploader_id': '223413403',
183 'timestamp': 1396018030,
57bdc730
S
184 },
185 'skip': 'Requires vk account credentials',
186 },
849086a1
S
187 {
188 'url': 'http://m.vk.com/video-43215063_169084319?list=125c627d1aa1cebb83&from=wall-43215063_2566540',
189 'md5': '0c45586baa71b7cb1d0784ee3f4e00a6',
190 'note': 'ivi.ru embed',
191 'info_dict': {
220828f2 192 'id': '-43215063_169084319',
849086a1
S
193 'ext': 'mp4',
194 'title': 'Книга Илая',
195 'duration': 6771,
42e1ff86 196 'upload_date': '20140626',
8117df4c 197 'view_count': int,
849086a1 198 },
3c989818 199 'skip': 'Removed',
849086a1 200 },
af3cbd87 201 {
202 'url': 'https://vk.com/video-93049196_456239755?list=ln-cBjJ7S4jYYx3ADnmDT',
203 'info_dict': {
204 'id': '-93049196_456239755',
205 'ext': 'mp4',
206 'title': '8 серия (озвучка)',
207 'duration': 8383,
59f63c8f
M
208 'comment_count': int,
209 'uploader': 'Dizi2021',
210 'like_count': int,
211 'timestamp': 1640162189,
af3cbd87 212 'upload_date': '20211222',
59f63c8f
M
213 'uploader_id': '-93049196',
214 'thumbnail': r're:https?://.+\.jpg$',
af3cbd87 215 },
216 },
79913fde
S
217 {
218 # video (removed?) only available with list id
219 'url': 'https://vk.com/video30481095_171201961?list=8764ae2d21f14088d4',
220 'md5': '091287af5402239a1051c37ec7b92913',
221 'info_dict': {
220828f2 222 'id': '30481095_171201961',
79913fde
S
223 'ext': 'mp4',
224 'title': 'ТюменцевВВ_09.07.2015',
225 'uploader': 'Anton Ivanov',
226 'duration': 109,
227 'upload_date': '20150709',
228 'view_count': int,
229 },
a7ee8a00 230 'skip': 'Removed',
79913fde 231 },
9281f6d2
S
232 {
233 # youtube embed
234 'url': 'https://vk.com/video276849682_170681728',
235 'info_dict': {
236 'id': 'V3K4mi0SYkc',
220828f2 237 'ext': 'mp4',
9281f6d2 238 'title': "DSWD Awards 'Children's Joy Foundation, Inc.' Certificate of Registration and License to Operate",
ad1bc71a 239 'description': 'md5:bf9c26cfa4acdfb146362682edd3827a',
220828f2 240 'duration': 178,
59f63c8f 241 'upload_date': '20130117',
ad1bc71a 242 'uploader': "Children's Joy Foundation Inc.",
9281f6d2
S
243 'uploader_id': 'thecjf',
244 'view_count': int,
59f63c8f
M
245 'channel_id': 'UCgzCNQ11TmR9V97ECnhi3gw',
246 'availability': 'public',
247 'like_count': int,
248 'live_status': 'not_live',
249 'playable_in_embed': True,
250 'channel': 'Children\'s Joy Foundation Inc.',
251 'uploader_url': 'http://www.youtube.com/user/thecjf',
252 'thumbnail': r're:https?://.+\.jpg$',
253 'tags': 'count:27',
254 'start_time': 0.0,
255 'categories': ['Nonprofits & Activism'],
256 'channel_url': 'https://www.youtube.com/channel/UCgzCNQ11TmR9V97ECnhi3gw',
257 'age_limit': 0,
9281f6d2
S
258 },
259 },
e3845525
KM
260 {
261 # dailymotion embed
262 'url': 'https://vk.com/video-37468416_456239855',
263 'info_dict': {
264 'id': 'k3lz2cmXyRuJQSjGHUv',
265 'ext': 'mp4',
266 'title': 'md5:d52606645c20b0ddbb21655adaa4f56f',
5ef62fc4 267 'description': 'md5:424b8e88cc873217f520e582ba28bb36',
e3845525
KM
268 'uploader': 'AniLibria.Tv',
269 'upload_date': '20160914',
270 'uploader_id': 'x1p5vl5',
271 'timestamp': 1473877246,
272 },
59f63c8f 273 'skip': 'Removed'
e3845525 274 },
bf4b3b6b
S
275 {
276 # video key is extra_data not url\d+
277 'url': 'http://vk.com/video-110305615_171782105',
278 'md5': 'e13fcda136f99764872e739d13fac1d1',
279 'info_dict': {
220828f2 280 'id': '-110305615_171782105',
bf4b3b6b
S
281 'ext': 'mp4',
282 'title': 'S-Dance, репетиции к The way show',
283 'uploader': 'THE WAY SHOW | 17 апреля',
ad1bc71a
RA
284 'uploader_id': '-110305615',
285 'timestamp': 1454859345,
bf4b3b6b 286 'upload_date': '20160207',
ad1bc71a 287 },
59f63c8f 288 'skip': 'Removed',
bf4b3b6b 289 },
93aa0b63 290 {
424ed37e 291 # finished live stream, postlive_mp4
93aa0b63 292 'url': 'https://vk.com/videos-387766?z=video-387766_456242764%2Fpl_-387766_-2',
93aa0b63 293 'info_dict': {
220828f2 294 'id': '-387766_456242764',
93aa0b63 295 'ext': 'mp4',
220828f2 296 'title': 'ИгроМир 2016 День 1 — Игромания Утром',
93aa0b63
S
297 'uploader': 'Игромания',
298 'duration': 5239,
220828f2
RA
299 'upload_date': '20160929',
300 'uploader_id': '-387766',
301 'timestamp': 1475137527,
59f63c8f
M
302 'thumbnail': r're:https?://.+\.jpg$',
303 'comment_count': int,
304 'like_count': int,
93aa0b63 305 },
3c989818
RA
306 'params': {
307 'skip_download': True,
308 },
93aa0b63 309 },
475f8a45 310 {
424ed37e 311 # live stream, hls and rtmp links, most likely already finished live
475f8a45
S
312 # stream by the time you are reading this comment
313 'url': 'https://vk.com/video-140332_456239111',
314 'only_matching': True,
315 },
a8363f3a
PH
316 {
317 # removed video, just testing that we match the pattern
318 'url': 'http://vk.com/feed?z=video-43215063_166094326%2Fbb50cacd3177146d7a',
319 'only_matching': True,
320 },
e58066e2
S
321 {
322 # age restricted video, requires vk account credentials
323 'url': 'https://vk.com/video205387401_164765225',
324 'only_matching': True,
325 },
a5e52a1f
S
326 {
327 # pladform embed
328 'url': 'https://vk.com/video-76116461_171554880',
329 'only_matching': True,
bdafd88d
S
330 },
331 {
332 'url': 'http://new.vk.com/video205387401_165548505',
333 'only_matching': True,
643dc0fc
CP
334 },
335 {
336 # This video is no longer available, because its author has been blocked.
337 'url': 'https://vk.com/video-10639516_456240611',
338 'only_matching': True,
a640c4d2 339 },
340 {
341 # The video is not available in your region.
342 'url': 'https://vk.com/video-51812607_171445436',
343 'only_matching': True,
21df2117 344 },
345 {
346 'url': 'https://vk.com/clip30014565_456240946',
347 'only_matching': True,
a640c4d2 348 }]
9032dc28 349
60d142aa 350 def _real_extract(self, url):
5ad28e7f 351 mobj = self._match_valid_url(url)
ca97a56e
S
352 video_id = mobj.group('videoid')
353
3c989818 354 mv_data = {}
04e88ca2 355 if video_id:
3c989818 356 data = {
59f63c8f 357 'act': 'show',
3c989818
RA
358 'video': video_id,
359 }
04e88ca2 360 # Some videos (removed?) can only be downloaded with list id specified
361 list_id = mobj.group('list_id')
362 if list_id:
3c989818
RA
363 data['list'] = list_id
364
365 payload = self._download_payload('al_video', video_id, data)
366 info_page = payload[1]
367 opts = payload[-1]
368 mv_data = opts.get('mvData') or {}
369 player = opts.get('player') or {}
04e88ca2 370 else:
ca97a56e 371 video_id = '%s_%s' % (mobj.group('oid'), mobj.group('id'))
9032dc28 372
3c989818
RA
373 info_page = self._download_webpage(
374 'http://vk.com/video_ext.php?' + mobj.group('embed_query'), video_id)
9032dc28 375
3c989818
RA
376 error_message = self._html_search_regex(
377 [r'(?s)<!><div[^>]+class="video_layer_message"[^>]*>(.+?)</div>',
378 r'(?s)<div[^>]+id="video_ext_msg"[^>]*>(.+?)</div>'],
379 info_page, 'error message', default=None)
380 if error_message:
381 raise ExtractorError(error_message, expected=True)
ee48b6a8 382
3c989818
RA
383 if re.search(r'<!>/login\.php\?.*\bact=security_check', info_page):
384 raise ExtractorError(
385 'You are trying to log in from an unusual location. You should confirm ownership at vk.com to log in with this IP.',
386 expected=True)
7f220b2f 387
3c989818 388 ERROR_COPYRIGHT = 'Video %s has been removed from public access due to rightholder complaint.'
1d1d60f6 389
3c989818
RA
390 ERRORS = {
391 r'>Видеозапись .*? была изъята из публичного доступа в связи с обращением правообладателя.<':
392 ERROR_COPYRIGHT,
1d1d60f6 393
3c989818
RA
394 r'>The video .*? was removed from public access by request of the copyright holder.<':
395 ERROR_COPYRIGHT,
3d36cea4 396
3c989818
RA
397 r'<!>Please log in or <':
398 'Video %s is only available for registered users, '
399 'use --username and --password options to provide account credentials.',
3d36cea4 400
3c989818
RA
401 r'<!>Unknown error':
402 'Video %s does not exist.',
1aa5172f 403
3c989818
RA
404 r'<!>Видео временно недоступно':
405 'Video %s is temporarily unavailable.',
d919fa33 406
3c989818
RA
407 r'<!>Access denied':
408 'Access denied to video %s.',
643dc0fc 409
3c989818
RA
410 r'<!>Видеозапись недоступна, так как её автор был заблокирован.':
411 'Video %s is no longer available, because its author has been blocked.',
643dc0fc 412
3c989818
RA
413 r'<!>This video is no longer available, because its author has been blocked.':
414 'Video %s is no longer available, because its author has been blocked.',
ad1bc71a 415
3c989818
RA
416 r'<!>This video is no longer available, because it has been deleted.':
417 'Video %s is no longer available, because it has been deleted.',
a640c4d2 418
3c989818
RA
419 r'<!>The video .+? is not available in your region.':
420 'Video %s is not available in your region.',
421 }
422
423 for error_re, error_msg in ERRORS.items():
424 if re.search(error_re, info_page):
425 raise ExtractorError(error_msg % video_id, expected=True)
9032dc28 426
3c989818
RA
427 player = self._parse_json(self._search_regex(
428 r'var\s+playerParams\s*=\s*({.+?})\s*;\s*\n',
429 info_page, 'player params'), video_id)
9334f8f1 430
5113b691 431 youtube_url = YoutubeIE._extract_url(info_page)
46478456 432 if youtube_url:
3c989818 433 return self.url_result(youtube_url, YoutubeIE.ie_key())
849086a1 434
09b9c45e 435 vimeo_url = VimeoIE._extract_url(url, info_page)
84663361 436 if vimeo_url is not None:
3c989818 437 return self.url_result(vimeo_url, VimeoIE.ie_key())
84663361 438
c4737bea
S
439 pladform_url = PladformIE._extract_url(info_page)
440 if pladform_url:
3c989818 441 return self.url_result(pladform_url, PladformIE.ie_key())
c4737bea 442
7a1818c9 443 m_rutube = re.search(
35972ba1 444 r'\ssrc="((?:https?:)?//rutube\.ru\\?/(?:video|play)\\?/embed(?:.*?))\\?"', info_page)
7a1818c9 445 if m_rutube is not None:
7a1818c9
PH
446 rutube_url = self._proto_relative_url(
447 m_rutube.group(1).replace('\\', ''))
448 return self.url_result(rutube_url)
449
bfd973ec 450 dailymotion_urls = DailymotionIE._extract_embed_urls(url, info_page)
e3845525
KM
451 if dailymotion_urls:
452 return self.url_result(dailymotion_urls[0], DailymotionIE.ie_key())
453
3c989818
RA
454 odnoklassniki_url = OdnoklassnikiIE._extract_url(info_page)
455 if odnoklassniki_url:
456 return self.url_result(odnoklassniki_url, OdnoklassnikiIE.ie_key())
457
bfd973ec 458 sibnet_urls = self._extract_embed_urls(url, info_page)
b73612a2 459 if sibnet_urls:
460 return self.url_result(sibnet_urls[0])
461
054932f4 462 m_opts = re.search(r'(?s)var\s+opts\s*=\s*({.+?});', info_page)
849086a1 463 if m_opts:
054932f4 464 m_opts_url = re.search(r"url\s*:\s*'((?!/\b)[^']+)", m_opts.group(1))
849086a1
S
465 if m_opts_url:
466 opts_url = m_opts_url.group(1)
467 if opts_url.startswith('//'):
468 opts_url = 'http:' + opts_url
469 return self.url_result(opts_url)
470
3c989818 471 data = player['params'][0]
475f8a45
S
472 title = unescapeHTML(data['md_title'])
473
424ed37e
S
474 # 2 = live
475 # 3 = post live (finished live)
9cdb0a33 476 is_live = data.get('live') == 2
475f8a45 477
a7ee8a00 478 timestamp = unified_timestamp(self._html_search_regex(
70d7b323 479 r'class=["\']mv_info_date[^>]+>([^<]+)(?:<|from)', info_page,
ad1bc71a 480 'upload date', default=None)) or int_or_none(data.get('date'))
3aa3953d 481
70d7b323
S
482 view_count = str_to_int(self._search_regex(
483 r'class=["\']mv_views_count[^>]+>\s*([\d,.]+)',
498a8a4c 484 info_page, 'view count', default=None))
8117df4c 485
bf4b3b6b 486 formats = []
475f8a45 487 for format_id, format_url in data.items():
3052a30d
S
488 format_url = url_or_none(format_url)
489 if not format_url or not format_url.startswith(('http', '//', 'rtmp')):
bf4b3b6b 490 continue
3089bc74
S
491 if (format_id.startswith(('url', 'cache'))
492 or format_id in ('extra_data', 'live_mp4', 'postlive_mp4')):
475f8a45
S
493 height = int_or_none(self._search_regex(
494 r'^(?:url|cache)(\d+)', format_id, 'height', default=None))
495 formats.append({
496 'format_id': format_id,
497 'url': format_url,
498 'height': height,
499 })
500 elif format_id == 'hls':
501 formats.extend(self._extract_m3u8_formats(
fb4fc449 502 format_url, video_id, 'mp4', 'm3u8_native',
9cdb0a33 503 m3u8_id=format_id, fatal=False, live=is_live))
475f8a45
S
504 elif format_id == 'rtmp':
505 formats.append({
506 'format_id': format_id,
507 'url': format_url,
508 'ext': 'flv',
509 })
913f3292
PH
510 self._sort_formats(formats)
511
5b6cb562 512 subtitles = {}
513 for sub in data.get('subs') or {}:
514 subtitles.setdefault(sub.get('lang', 'en'), []).append({
515 'ext': sub.get('title', '.srt').split('.')[-1],
516 'url': url_or_none(sub.get('url')),
517 })
518
60d142aa 519 return {
220828f2 520 'id': video_id,
913f3292 521 'formats': formats,
475f8a45 522 'title': title,
913f3292
PH
523 'thumbnail': data.get('jpg'),
524 'uploader': data.get('md_author'),
3c989818
RA
525 'uploader_id': str_or_none(data.get('author_id') or mv_data.get('authorId')),
526 'duration': int_or_none(data.get('duration') or mv_data.get('duration')),
a7ee8a00 527 'timestamp': timestamp,
8117df4c 528 'view_count': view_count,
3c989818
RA
529 'like_count': int_or_none(mv_data.get('likes')),
530 'comment_count': int_or_none(mv_data.get('commcount')),
9cdb0a33 531 'is_live': is_live,
5b6cb562 532 'subtitles': subtitles,
60d142aa 533 }
469d4c89
WS
534
535
2d19fb50 536class VKUserVideosIE(VKBaseIE):
1ecb5d1d
S
537 IE_NAME = 'vk:uservideos'
538 IE_DESC = "VK - User's Videos"
a70b71e8 539 _VALID_URL = r'https?://(?:(?:m|new)\.)?vk\.com/video/@(?P<id>[^?$#/&]+)(?!\?.*\bz=video)(?:[/?#&](?:.*?\bsection=(?P<section>\w+))?|$)'
469d4c89 540 _TEMPLATE_URL = 'https://vk.com/videos'
dc786d3d 541 _TESTS = [{
a70b71e8 542 'url': 'https://vk.com/video/@mobidevices',
0e6ec3ca 543 'info_dict': {
a70b71e8 544 'id': '-17892518_all',
0e6ec3ca 545 },
a70b71e8 546 'playlist_mincount': 1355,
0e6ec3ca 547 }, {
a70b71e8 548 'url': 'https://vk.com/video/@mobidevices?section=uploaded',
15ec6693 549 'info_dict': {
a70b71e8 550 'id': '-17892518_uploaded',
15ec6693 551 },
a70b71e8 552 'playlist_mincount': 182,
dc786d3d 553 }]
0e6ec3ca 554 _VIDEO = collections.namedtuple('Video', ['owner_id', 'id'])
dc786d3d 555
a70b71e8
AG
556 def _entries(self, page_id, section):
557 video_list_json = self._download_payload('al_video', page_id, {
3c989818 558 'act': 'load_videos_silent',
a70b71e8 559 'offset': 0,
3c989818 560 'oid': page_id,
0e6ec3ca 561 'section': section,
a70b71e8
AG
562 })[0][section]
563 count = video_list_json['count']
564 total = video_list_json['total']
565 video_list = video_list_json['list']
566
567 while True:
568 for video in video_list:
569 v = self._VIDEO._make(video[:2])
570 video_id = '%d_%d' % (v.owner_id, v.id)
571 yield self.url_result(
572 'http://vk.com/video' + video_id, VKIE.ie_key(), video_id)
573 if count >= total:
574 break
575 video_list_json = self._download_payload('al_video', page_id, {
576 'act': 'load_videos_silent',
577 'offset': count,
578 'oid': page_id,
579 'section': section,
580 })[0][section]
581 count += video_list_json['count']
582 video_list = video_list_json['list']
0e6ec3ca
RA
583
584 def _real_extract(self, url):
a70b71e8
AG
585 u_id, section = self._match_valid_url(url).groups()
586 webpage = self._download_webpage(url, u_id)
587 page_id = self._search_regex(r'data-owner-id\s?=\s?"([^"]+)"', webpage, 'page_id')
0e6ec3ca
RA
588 if not section:
589 section = 'all'
590
a70b71e8 591 return self.playlist_result(self._entries(page_id, section), '%s_%s' % (page_id, section))
2d19fb50
S
592
593
594class VKWallPostIE(VKBaseIE):
595 IE_NAME = 'vk:wallpost'
596 _VALID_URL = r'https?://(?:(?:(?:(?:m|new)\.)?vk\.com/(?:[^?]+\?.*\bw=)?wall(?P<id>-?\d+_\d+)))'
597 _TESTS = [{
598 # public page URL, audio playlist
599 'url': 'https://vk.com/bs.official?w=wall-23538238_35',
600 'info_dict': {
3c989818
RA
601 'id': '-23538238_35',
602 'title': 'Black Shadow - Wall post -23538238_35',
2d19fb50
S
603 'description': 'md5:3f84b9c4f9ef499731cf1ced9998cc0c',
604 },
605 'playlist': [{
606 'md5': '5ba93864ec5b85f7ce19a9af4af080f6',
607 'info_dict': {
608 'id': '135220665_111806521',
3c989818 609 'ext': 'mp4',
2d19fb50
S
610 'title': 'Black Shadow - Слепое Верование',
611 'duration': 370,
612 'uploader': 'Black Shadow',
613 'artist': 'Black Shadow',
614 'track': 'Слепое Верование',
615 },
616 }, {
617 'md5': '4cc7e804579122b17ea95af7834c9233',
618 'info_dict': {
619 'id': '135220665_111802303',
3c989818 620 'ext': 'mp4',
2d19fb50
S
621 'title': 'Black Shadow - Война - Негасимое Бездны Пламя!',
622 'duration': 423,
623 'uploader': 'Black Shadow',
624 'artist': 'Black Shadow',
625 'track': 'Война - Негасимое Бездны Пламя!',
626 },
2d19fb50 627 }],
51815886 628 'params': {
3c989818 629 'skip_download': True,
51815886 630 },
2d19fb50
S
631 'skip': 'Requires vk account credentials',
632 }, {
633 # single YouTube embed, no leading -
634 'url': 'https://vk.com/wall85155021_6319',
635 'info_dict': {
636 'id': '85155021_6319',
3c989818 637 'title': 'Сергей Горбунов - Wall post 85155021_6319',
2d19fb50
S
638 },
639 'playlist_count': 1,
640 'skip': 'Requires vk account credentials',
641 }, {
642 # wall page URL
643 'url': 'https://vk.com/wall-23538238_35',
644 'only_matching': True,
645 }, {
646 # mobile wall page URL
647 'url': 'https://m.vk.com/wall-23538238_35',
648 'only_matching': True,
649 }]
3c989818 650 _BASE64_CHARS = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMN0PQRSTUVWXYZO123456789+/='
0e6ec3ca 651 _AUDIO = collections.namedtuple('Audio', ['id', 'owner_id', 'url', 'title', 'performer', 'duration', 'album_id', 'unk', 'author_link', 'lyrics', 'flags', 'context', 'extra', 'hashes', 'cover_url', 'ads'])
3c989818
RA
652
653 def _decode(self, enc):
654 dec = ''
655 e = n = 0
656 for c in enc:
657 r = self._BASE64_CHARS.index(c)
658 cond = n % 4
659 e = 64 * e + r if cond else r
660 n += 1
661 if cond:
662 dec += chr(255 & e >> (-2 * n & 6))
663 return dec
664
665 def _unmask_url(self, mask_url, vk_id):
666 if 'audio_api_unavailable' in mask_url:
667 extra = mask_url.split('?extra=')[1].split('#')
668 func, base = self._decode(extra[1]).split(chr(11))
3c989818
RA
669 mask_url = list(self._decode(extra[0]))
670 url_len = len(mask_url)
671 indexes = [None] * url_len
672 index = int(base) ^ vk_id
673 for n in range(url_len - 1, -1, -1):
674 index = (url_len * (n + 1) ^ index + n) % url_len
675 indexes[n] = index
676 for n in range(1, url_len):
677 c = mask_url[n]
678 index = indexes[url_len - 1 - n]
679 mask_url[n] = mask_url[index]
680 mask_url[index] = c
681 mask_url = ''.join(mask_url)
682 return mask_url
2d19fb50
S
683
684 def _real_extract(self, url):
685 post_id = self._match_id(url)
686
3c989818
RA
687 webpage = self._download_payload('wkview', post_id, {
688 'act': 'show',
689 'w': 'wall' + post_id,
690 })[1]
2d19fb50
S
691
692 description = clean_html(get_element_by_class('wall_post_text', webpage))
51815886 693 uploader = clean_html(get_element_by_class('author', webpage))
2d19fb50
S
694
695 entries = []
696
3c989818
RA
697 for audio in re.findall(r'data-audio="([^"]+)', webpage):
698 audio = self._parse_json(unescapeHTML(audio), post_id)
0e6ec3ca 699 a = self._AUDIO._make(audio[:16])
3c989818
RA
700 if not a.url:
701 continue
702 title = unescapeHTML(a.title)
7e70620a 703 performer = unescapeHTML(a.performer)
3c989818
RA
704 entries.append({
705 'id': '%s_%s' % (a.owner_id, a.id),
706 'url': self._unmask_url(a.url, a.ads['vk_id']),
7e70620a
RA
707 'title': '%s - %s' % (performer, title) if performer else title,
708 'thumbnails': [{'url': c_url} for c_url in a.cover_url.split(',')] if a.cover_url else None,
709 'duration': int_or_none(a.duration),
3c989818 710 'uploader': uploader,
7e70620a 711 'artist': performer,
3c989818
RA
712 'track': title,
713 'ext': 'mp4',
079a7cfc 714 'protocol': 'm3u8_native',
3c989818 715 })
2d19fb50
S
716
717 for video in re.finditer(
718 r'<a[^>]+href=(["\'])(?P<url>/video(?:-?[\d_]+).*?)\1', webpage):
719 entries.append(self.url_result(
720 compat_urlparse.urljoin(url, video.group('url')), VKIE.ie_key()))
721
722 title = 'Wall post %s' % post_id
723
724 return self.playlist_result(
725 orderedSet(entries), post_id,
726 '%s - %s' % (uploader, title) if uploader else title,
727 description)