]>
Commit | Line | Data |
---|---|---|
dcdb292f | 1 | # coding: utf-8 |
94a23d2a PH |
2 | from __future__ import unicode_literals |
3 | ||
51815886 | 4 | import collections |
60d142aa | 5 | import re |
75ca6bce | 6 | import sys |
60d142aa JMF |
7 | |
8 | from .common import InfoExtractor | |
2d19fb50 S |
9 | from ..compat import ( |
10 | compat_str, | |
11 | compat_urlparse, | |
12 | ) | |
60d142aa | 13 | from ..utils import ( |
2d19fb50 | 14 | clean_html, |
9032dc28 | 15 | ExtractorError, |
2d19fb50 | 16 | get_element_by_class, |
bf4b3b6b | 17 | int_or_none, |
1cc79574 | 18 | orderedSet, |
2d19fb50 | 19 | remove_start, |
8117df4c | 20 | str_to_int, |
60d142aa | 21 | unescapeHTML, |
a7ee8a00 | 22 | unified_timestamp, |
6e6bc8da | 23 | urlencode_postdata, |
1cc79574 | 24 | ) |
e3845525 | 25 | from .dailymotion import DailymotionIE |
c4737bea | 26 | from .pladform import PladformIE |
e3845525 | 27 | from .vimeo import VimeoIE |
60d142aa JMF |
28 | |
29 | ||
2d19fb50 S |
30 | class VKBaseIE(InfoExtractor): |
31 | _NETRC_MACHINE = 'vk' | |
32 | ||
33 | def _login(self): | |
34 | (username, password) = self._get_login_info() | |
35 | if username is None: | |
36 | return | |
37 | ||
38 | login_page, url_handle = self._download_webpage_handle( | |
39 | 'https://vk.com', None, 'Downloading login page') | |
40 | ||
41 | login_form = self._hidden_inputs(login_page) | |
42 | ||
43 | login_form.update({ | |
44 | 'email': username.encode('cp1251'), | |
45 | 'pass': password.encode('cp1251'), | |
46 | }) | |
47 | ||
48 | # https://new.vk.com/ serves two same remixlhk cookies in Set-Cookie header | |
49 | # and expects the first one to be set rather than second (see | |
50 | # https://github.com/rg3/youtube-dl/issues/9841#issuecomment-227871201). | |
51 | # As of RFC6265 the newer one cookie should be set into cookie store | |
52 | # what actually happens. | |
53 | # We will workaround this VK issue by resetting the remixlhk cookie to | |
54 | # the first one manually. | |
08a42f9c S |
55 | for header, cookies in url_handle.headers.items(): |
56 | if header.lower() != 'set-cookie': | |
57 | continue | |
5f5a9d61 S |
58 | if sys.version_info[0] >= 3: |
59 | cookies = cookies.encode('iso-8859-1') | |
60 | cookies = cookies.decode('utf-8') | |
61 | remixlhk = re.search(r'remixlhk=(.+?);.*?\bdomain=(.+?)(?:[,;]|$)', cookies) | |
62 | if remixlhk: | |
63 | value, domain = remixlhk.groups() | |
64 | self._set_cookie(domain, 'remixlhk', value) | |
08a42f9c | 65 | break |
2d19fb50 S |
66 | |
67 | login_page = self._download_webpage( | |
68 | 'https://login.vk.com/?act=login', None, | |
69 | note='Logging in as %s' % username, | |
70 | data=urlencode_postdata(login_form)) | |
71 | ||
72 | if re.search(r'onLoginFailed', login_page): | |
73 | raise ExtractorError( | |
74 | 'Unable to login, incorrect username and/or password', expected=True) | |
75 | ||
76 | def _real_initialize(self): | |
77 | self._login() | |
78 | ||
79 | ||
80 | class VKIE(VKBaseIE): | |
1ecb5d1d S |
81 | IE_NAME = 'vk' |
82 | IE_DESC = 'VK' | |
cf9cf7dd S |
83 | _VALID_URL = r'''(?x) |
84 | https?:// | |
85 | (?: | |
04e88ca2 | 86 | (?: |
bdafd88d | 87 | (?:(?:m|new)\.)?vk\.com/video_| |
04e88ca2 | 88 | (?:www\.)?daxab.com/ |
89 | ) | |
90 | ext\.php\?(?P<embed_query>.*?\boid=(?P<oid>-?\d+).*?\bid=(?P<id>\d+).*)| | |
cf9cf7dd | 91 | (?: |
bdafd88d | 92 | (?:(?:m|new)\.)?vk\.com/(?:.+?\?.*?z=)?video| |
04e88ca2 | 93 | (?:www\.)?daxab.com/embed/ |
cf9cf7dd | 94 | ) |
04e88ca2 | 95 | (?P<videoid>-?\d+_\d+)(?:.*\blist=(?P<list_id>[\da-f]+))? |
cf9cf7dd S |
96 | ) |
97 | ''' | |
9032dc28 S |
98 | _TESTS = [ |
99 | { | |
100 | 'url': 'http://vk.com/videos-77521?z=video-77521_162222515%2Fclub77521', | |
101 | 'md5': '0deae91935c54e00003c2a00646315f0', | |
102 | 'info_dict': { | |
103 | 'id': '162222515', | |
104 | 'ext': 'flv', | |
105 | 'title': 'ProtivoGunz - Хуёвая песня', | |
36300346 | 106 | 'uploader': 're:(?:Noize MC|Alexander Ilyashenko).*', |
9032dc28 | 107 | 'duration': 195, |
a7ee8a00 | 108 | 'timestamp': 1329060660, |
42e1ff86 | 109 | 'upload_date': '20120212', |
8117df4c | 110 | 'view_count': int, |
9032dc28 | 111 | }, |
60d142aa | 112 | }, |
9032dc28 | 113 | { |
c52331f3 WS |
114 | 'url': 'http://vk.com/video205387401_165548505', |
115 | 'md5': '6c0aeb2e90396ba97035b9cbde548700', | |
9032dc28 | 116 | 'info_dict': { |
c52331f3 | 117 | 'id': '165548505', |
9032dc28 | 118 | 'ext': 'mp4', |
c52331f3 WS |
119 | 'uploader': 'Tom Cruise', |
120 | 'title': 'No name', | |
121 | 'duration': 9, | |
a7ee8a00 | 122 | 'timestamp': 1374374880, |
8117df4c S |
123 | 'upload_date': '20130721', |
124 | 'view_count': int, | |
9032dc28 S |
125 | } |
126 | }, | |
ca97a56e S |
127 | { |
128 | 'note': 'Embedded video', | |
129 | 'url': 'http://vk.com/video_ext.php?oid=32194266&id=162925554&hash=7d8c2e0d5e05aeaa&hd=1', | |
130 | 'md5': 'c7ce8f1f87bec05b3de07fdeafe21a0a', | |
131 | 'info_dict': { | |
132 | 'id': '162925554', | |
133 | 'ext': 'mp4', | |
134 | 'uploader': 'Vladimir Gavrin', | |
135 | 'title': 'Lin Dan', | |
136 | 'duration': 101, | |
42e1ff86 | 137 | 'upload_date': '20120730', |
8117df4c | 138 | 'view_count': int, |
04e88ca2 | 139 | }, |
140 | 'skip': 'This video has been removed from public access.', | |
ca97a56e | 141 | }, |
9032dc28 | 142 | { |
c52331f3 WS |
143 | # VIDEO NOW REMOVED |
144 | # please update if you find a video whose URL follows the same pattern | |
9032dc28 S |
145 | 'url': 'http://vk.com/video-8871596_164049491', |
146 | 'md5': 'a590bcaf3d543576c9bd162812387666', | |
147 | 'note': 'Only available for registered users', | |
148 | 'info_dict': { | |
149 | 'id': '164049491', | |
150 | 'ext': 'mp4', | |
151 | 'uploader': 'Триллеры', | |
57bdc730 | 152 | 'title': '► Бойцовский клуб / Fight Club 1999 [HD 720]', |
9032dc28 | 153 | 'duration': 8352, |
8117df4c S |
154 | 'upload_date': '20121218', |
155 | 'view_count': int, | |
9032dc28 S |
156 | }, |
157 | 'skip': 'Requires vk account credentials', | |
ca97a56e | 158 | }, |
57bdc730 S |
159 | { |
160 | 'url': 'http://vk.com/hd_kino_mania?z=video-43215063_168067957%2F15c66b9b533119788d', | |
161 | 'md5': '4d7a5ef8cf114dfa09577e57b2993202', | |
162 | 'info_dict': { | |
163 | 'id': '168067957', | |
164 | 'ext': 'mp4', | |
165 | 'uploader': 'Киномания - лучшее из мира кино', | |
166 | 'title': ' ', | |
167 | 'duration': 7291, | |
42e1ff86 | 168 | 'upload_date': '20140328', |
57bdc730 S |
169 | }, |
170 | 'skip': 'Requires vk account credentials', | |
171 | }, | |
849086a1 S |
172 | { |
173 | 'url': 'http://m.vk.com/video-43215063_169084319?list=125c627d1aa1cebb83&from=wall-43215063_2566540', | |
174 | 'md5': '0c45586baa71b7cb1d0784ee3f4e00a6', | |
175 | 'note': 'ivi.ru embed', | |
176 | 'info_dict': { | |
177 | 'id': '60690', | |
178 | 'ext': 'mp4', | |
179 | 'title': 'Книга Илая', | |
180 | 'duration': 6771, | |
42e1ff86 | 181 | 'upload_date': '20140626', |
8117df4c | 182 | 'view_count': int, |
849086a1 | 183 | }, |
d518d06e | 184 | 'skip': 'Only works from Russia', |
849086a1 | 185 | }, |
79913fde S |
186 | { |
187 | # video (removed?) only available with list id | |
188 | 'url': 'https://vk.com/video30481095_171201961?list=8764ae2d21f14088d4', | |
189 | 'md5': '091287af5402239a1051c37ec7b92913', | |
190 | 'info_dict': { | |
191 | 'id': '171201961', | |
192 | 'ext': 'mp4', | |
193 | 'title': 'ТюменцевВВ_09.07.2015', | |
194 | 'uploader': 'Anton Ivanov', | |
195 | 'duration': 109, | |
196 | 'upload_date': '20150709', | |
197 | 'view_count': int, | |
198 | }, | |
a7ee8a00 | 199 | 'skip': 'Removed', |
79913fde | 200 | }, |
9281f6d2 S |
201 | { |
202 | # youtube embed | |
203 | 'url': 'https://vk.com/video276849682_170681728', | |
204 | 'info_dict': { | |
205 | 'id': 'V3K4mi0SYkc', | |
f22ba4bd | 206 | 'ext': 'webm', |
9281f6d2 | 207 | 'title': "DSWD Awards 'Children's Joy Foundation, Inc.' Certificate of Registration and License to Operate", |
04e88ca2 | 208 | 'description': 'md5:d9903938abdc74c738af77f527ca0596', |
f22ba4bd | 209 | 'duration': 178, |
9281f6d2 S |
210 | 'upload_date': '20130116', |
211 | 'uploader': "Children's Joy Foundation", | |
212 | 'uploader_id': 'thecjf', | |
213 | 'view_count': int, | |
214 | }, | |
215 | }, | |
e3845525 KM |
216 | { |
217 | # dailymotion embed | |
218 | 'url': 'https://vk.com/video-37468416_456239855', | |
219 | 'info_dict': { | |
220 | 'id': 'k3lz2cmXyRuJQSjGHUv', | |
221 | 'ext': 'mp4', | |
222 | 'title': 'md5:d52606645c20b0ddbb21655adaa4f56f', | |
223 | 'description': 'md5:c651358f03c56f1150b555c26d90a0fd', | |
224 | 'uploader': 'AniLibria.Tv', | |
225 | 'upload_date': '20160914', | |
226 | 'uploader_id': 'x1p5vl5', | |
227 | 'timestamp': 1473877246, | |
228 | }, | |
229 | 'params': { | |
230 | 'skip_download': True, | |
93aa0b63 | 231 | }, |
e3845525 | 232 | }, |
bf4b3b6b S |
233 | { |
234 | # video key is extra_data not url\d+ | |
235 | 'url': 'http://vk.com/video-110305615_171782105', | |
236 | 'md5': 'e13fcda136f99764872e739d13fac1d1', | |
237 | 'info_dict': { | |
238 | 'id': '171782105', | |
239 | 'ext': 'mp4', | |
240 | 'title': 'S-Dance, репетиции к The way show', | |
241 | 'uploader': 'THE WAY SHOW | 17 апреля', | |
a7ee8a00 | 242 | 'timestamp': 1454870100, |
bf4b3b6b S |
243 | 'upload_date': '20160207', |
244 | 'view_count': int, | |
245 | }, | |
246 | }, | |
93aa0b63 S |
247 | { |
248 | # finished live stream, live_mp4 | |
249 | 'url': 'https://vk.com/videos-387766?z=video-387766_456242764%2Fpl_-387766_-2', | |
250 | 'md5': '90d22d051fccbbe9becfccc615be6791', | |
251 | 'info_dict': { | |
252 | 'id': '456242764', | |
253 | 'ext': 'mp4', | |
254 | 'title': 'ИгроМир 2016 — день 1', | |
255 | 'uploader': 'Игромания', | |
256 | 'duration': 5239, | |
70d7b323 | 257 | 'view_count': int, |
93aa0b63 S |
258 | }, |
259 | }, | |
475f8a45 S |
260 | { |
261 | # live stream, hls and rtmp links,most likely already finished live | |
262 | # stream by the time you are reading this comment | |
263 | 'url': 'https://vk.com/video-140332_456239111', | |
264 | 'only_matching': True, | |
265 | }, | |
a8363f3a PH |
266 | { |
267 | # removed video, just testing that we match the pattern | |
268 | 'url': 'http://vk.com/feed?z=video-43215063_166094326%2Fbb50cacd3177146d7a', | |
269 | 'only_matching': True, | |
270 | }, | |
e58066e2 S |
271 | { |
272 | # age restricted video, requires vk account credentials | |
273 | 'url': 'https://vk.com/video205387401_164765225', | |
274 | 'only_matching': True, | |
275 | }, | |
a5e52a1f S |
276 | { |
277 | # pladform embed | |
278 | 'url': 'https://vk.com/video-76116461_171554880', | |
279 | 'only_matching': True, | |
bdafd88d S |
280 | }, |
281 | { | |
282 | 'url': 'http://new.vk.com/video205387401_165548505', | |
283 | 'only_matching': True, | |
cf9cf7dd | 284 | } |
9032dc28 S |
285 | ] |
286 | ||
60d142aa JMF |
287 | def _real_extract(self, url): |
288 | mobj = re.match(self._VALID_URL, url) | |
ca97a56e S |
289 | video_id = mobj.group('videoid') |
290 | ||
04e88ca2 | 291 | if video_id: |
292 | info_url = 'https://vk.com/al_video.php?act=show&al=1&module=video&video=%s' % video_id | |
293 | # Some videos (removed?) can only be downloaded with list id specified | |
294 | list_id = mobj.group('list_id') | |
295 | if list_id: | |
296 | info_url += '&list=%s' % list_id | |
297 | else: | |
298 | info_url = 'http://vk.com/video_ext.php?' + mobj.group('embed_query') | |
ca97a56e | 299 | video_id = '%s_%s' % (mobj.group('oid'), mobj.group('id')) |
9032dc28 | 300 | |
60d142aa | 301 | info_page = self._download_webpage(info_url, video_id) |
9032dc28 | 302 | |
ee48b6a8 | 303 | error_message = self._html_search_regex( |
04e88ca2 | 304 | [r'(?s)<!><div[^>]+class="video_layer_message"[^>]*>(.+?)</div>', |
305 | r'(?s)<div[^>]+id="video_ext_msg"[^>]*>(.+?)</div>'], | |
ee48b6a8 S |
306 | info_page, 'error message', default=None) |
307 | if error_message: | |
308 | raise ExtractorError(error_message, expected=True) | |
309 | ||
7f220b2f S |
310 | if re.search(r'<!>/login\.php\?.*\bact=security_check', info_page): |
311 | raise ExtractorError( | |
312 | 'You are trying to log in from an unusual location. You should confirm ownership at vk.com to log in with this IP.', | |
313 | expected=True) | |
314 | ||
e0c51cda S |
315 | ERRORS = { |
316 | r'>Видеозапись .*? была изъята из публичного доступа в связи с обращением правообладателя.<': | |
3d36cea4 PH |
317 | 'Video %s has been removed from public access due to rightholder complaint.', |
318 | ||
e0c51cda | 319 | r'<!>Please log in or <': |
3d36cea4 PH |
320 | 'Video %s is only available for registered users, ' |
321 | 'use --username and --password options to provide account credentials.', | |
322 | ||
323 | r'<!>Unknown error': | |
1aa5172f S |
324 | 'Video %s does not exist.', |
325 | ||
326 | r'<!>Видео временно недоступно': | |
327 | 'Video %s is temporarily unavailable.', | |
d919fa33 S |
328 | |
329 | r'<!>Access denied': | |
330 | 'Access denied to video %s.', | |
e0c51cda | 331 | } |
9032dc28 | 332 | |
e0c51cda S |
333 | for error_re, error_msg in ERRORS.items(): |
334 | if re.search(error_re, info_page): | |
335 | raise ExtractorError(error_msg % video_id, expected=True) | |
9334f8f1 | 336 | |
46478456 S |
337 | youtube_url = self._search_regex( |
338 | r'<iframe[^>]+src="((?:https?:)?//www.youtube.com/embed/[^"]+)"', | |
339 | info_page, 'youtube iframe', default=None) | |
340 | if youtube_url: | |
9281f6d2 | 341 | return self.url_result(youtube_url, 'Youtube') |
849086a1 | 342 | |
09b9c45e | 343 | vimeo_url = VimeoIE._extract_url(url, info_page) |
84663361 S |
344 | if vimeo_url is not None: |
345 | return self.url_result(vimeo_url) | |
346 | ||
c4737bea S |
347 | pladform_url = PladformIE._extract_url(info_page) |
348 | if pladform_url: | |
349 | return self.url_result(pladform_url) | |
350 | ||
7a1818c9 | 351 | m_rutube = re.search( |
35972ba1 | 352 | r'\ssrc="((?:https?:)?//rutube\.ru\\?/(?:video|play)\\?/embed(?:.*?))\\?"', info_page) |
7a1818c9 | 353 | if m_rutube is not None: |
7a1818c9 PH |
354 | rutube_url = self._proto_relative_url( |
355 | m_rutube.group(1).replace('\\', '')) | |
356 | return self.url_result(rutube_url) | |
357 | ||
e3845525 KM |
358 | dailymotion_urls = DailymotionIE._extract_urls(info_page) |
359 | if dailymotion_urls: | |
360 | return self.url_result(dailymotion_urls[0], DailymotionIE.ie_key()) | |
361 | ||
054932f4 | 362 | m_opts = re.search(r'(?s)var\s+opts\s*=\s*({.+?});', info_page) |
849086a1 | 363 | if m_opts: |
054932f4 | 364 | m_opts_url = re.search(r"url\s*:\s*'((?!/\b)[^']+)", m_opts.group(1)) |
849086a1 S |
365 | if m_opts_url: |
366 | opts_url = m_opts_url.group(1) | |
367 | if opts_url.startswith('//'): | |
368 | opts_url = 'http:' + opts_url | |
369 | return self.url_result(opts_url) | |
370 | ||
9305a0dc S |
371 | # vars does not look to be served anymore since 24.10.2016 |
372 | data = self._parse_json( | |
373 | self._search_regex( | |
374 | r'var\s+vars\s*=\s*({.+?});', info_page, 'vars', default='{}'), | |
375 | video_id, fatal=False) | |
376 | ||
377 | # <!json> is served instead | |
378 | if not data: | |
379 | data = self._parse_json( | |
380 | self._search_regex( | |
381 | r'<!json>\s*({.+?})\s*<!>', info_page, 'json'), | |
382 | video_id)['player']['params'][0] | |
60d142aa | 383 | |
475f8a45 S |
384 | title = unescapeHTML(data['md_title']) |
385 | ||
386 | if data.get('live') == 2: | |
387 | title = self._live_title(title) | |
388 | ||
a7ee8a00 | 389 | timestamp = unified_timestamp(self._html_search_regex( |
70d7b323 | 390 | r'class=["\']mv_info_date[^>]+>([^<]+)(?:<|from)', info_page, |
a7ee8a00 | 391 | 'upload date', fatal=False)) |
3aa3953d | 392 | |
70d7b323 S |
393 | view_count = str_to_int(self._search_regex( |
394 | r'class=["\']mv_views_count[^>]+>\s*([\d,.]+)', | |
395 | info_page, 'view count', fatal=False)) | |
8117df4c | 396 | |
bf4b3b6b | 397 | formats = [] |
475f8a45 S |
398 | for format_id, format_url in data.items(): |
399 | if not isinstance(format_url, compat_str) or not format_url.startswith(('http', '//', 'rtmp')): | |
bf4b3b6b | 400 | continue |
475f8a45 S |
401 | if format_id.startswith(('url', 'cache')) or format_id in ('extra_data', 'live_mp4'): |
402 | height = int_or_none(self._search_regex( | |
403 | r'^(?:url|cache)(\d+)', format_id, 'height', default=None)) | |
404 | formats.append({ | |
405 | 'format_id': format_id, | |
406 | 'url': format_url, | |
407 | 'height': height, | |
408 | }) | |
409 | elif format_id == 'hls': | |
410 | formats.extend(self._extract_m3u8_formats( | |
411 | format_url, video_id, 'mp4', m3u8_id=format_id, | |
412 | fatal=False, live=True)) | |
413 | elif format_id == 'rtmp': | |
414 | formats.append({ | |
415 | 'format_id': format_id, | |
416 | 'url': format_url, | |
417 | 'ext': 'flv', | |
418 | }) | |
913f3292 PH |
419 | self._sort_formats(formats) |
420 | ||
60d142aa | 421 | return { |
475f8a45 | 422 | 'id': compat_str(data.get('vid') or video_id), |
913f3292 | 423 | 'formats': formats, |
475f8a45 | 424 | 'title': title, |
913f3292 PH |
425 | 'thumbnail': data.get('jpg'), |
426 | 'uploader': data.get('md_author'), | |
02a12f9f | 427 | 'duration': data.get('duration'), |
a7ee8a00 | 428 | 'timestamp': timestamp, |
8117df4c | 429 | 'view_count': view_count, |
60d142aa | 430 | } |
469d4c89 WS |
431 | |
432 | ||
2d19fb50 | 433 | class VKUserVideosIE(VKBaseIE): |
1ecb5d1d S |
434 | IE_NAME = 'vk:uservideos' |
435 | IE_DESC = "VK - User's Videos" | |
bdafd88d | 436 | _VALID_URL = r'https?://(?:(?:m|new)\.)?vk\.com/videos(?P<id>-?[0-9]+)(?!\?.*\bz=video)(?:[/?#&]|$)' |
469d4c89 | 437 | _TEMPLATE_URL = 'https://vk.com/videos' |
dc786d3d | 438 | _TESTS = [{ |
469d4c89 | 439 | 'url': 'http://vk.com/videos205387401', |
15ec6693 PH |
440 | 'info_dict': { |
441 | 'id': '205387401', | |
dc786d3d | 442 | 'title': "Tom Cruise's Videos", |
15ec6693 | 443 | }, |
469d4c89 | 444 | 'playlist_mincount': 4, |
dc786d3d S |
445 | }, { |
446 | 'url': 'http://vk.com/videos-77521', | |
447 | 'only_matching': True, | |
0436157b S |
448 | }, { |
449 | 'url': 'http://vk.com/videos-97664626?section=all', | |
450 | 'only_matching': True, | |
bdafd88d S |
451 | }, { |
452 | 'url': 'http://m.vk.com/videos205387401', | |
453 | 'only_matching': True, | |
454 | }, { | |
455 | 'url': 'http://new.vk.com/videos205387401', | |
456 | 'only_matching': True, | |
dc786d3d | 457 | }] |
469d4c89 | 458 | |
469d4c89 | 459 | def _real_extract(self, url): |
021a0db8 | 460 | page_id = self._match_id(url) |
dc786d3d S |
461 | |
462 | webpage = self._download_webpage(url, page_id) | |
463 | ||
464 | entries = [ | |
d16abf43 PH |
465 | self.url_result( |
466 | 'http://vk.com/video' + video_id, 'VK', video_id=video_id) | |
fec73daa | 467 | for video_id in orderedSet(re.findall(r'href="/video(-?[0-9_]+)"', webpage))] |
dc786d3d S |
468 | |
469 | title = unescapeHTML(self._search_regex( | |
470 | r'<title>\s*([^<]+?)\s+\|\s+\d+\s+videos', | |
471 | webpage, 'title', default=page_id)) | |
472 | ||
473 | return self.playlist_result(entries, page_id, title) | |
2d19fb50 S |
474 | |
475 | ||
476 | class VKWallPostIE(VKBaseIE): | |
477 | IE_NAME = 'vk:wallpost' | |
478 | _VALID_URL = r'https?://(?:(?:(?:(?:m|new)\.)?vk\.com/(?:[^?]+\?.*\bw=)?wall(?P<id>-?\d+_\d+)))' | |
479 | _TESTS = [{ | |
480 | # public page URL, audio playlist | |
481 | 'url': 'https://vk.com/bs.official?w=wall-23538238_35', | |
482 | 'info_dict': { | |
483 | 'id': '23538238_35', | |
484 | 'title': 'Black Shadow - Wall post 23538238_35', | |
485 | 'description': 'md5:3f84b9c4f9ef499731cf1ced9998cc0c', | |
486 | }, | |
487 | 'playlist': [{ | |
488 | 'md5': '5ba93864ec5b85f7ce19a9af4af080f6', | |
489 | 'info_dict': { | |
490 | 'id': '135220665_111806521', | |
491 | 'ext': 'mp3', | |
492 | 'title': 'Black Shadow - Слепое Верование', | |
493 | 'duration': 370, | |
494 | 'uploader': 'Black Shadow', | |
495 | 'artist': 'Black Shadow', | |
496 | 'track': 'Слепое Верование', | |
497 | }, | |
498 | }, { | |
499 | 'md5': '4cc7e804579122b17ea95af7834c9233', | |
500 | 'info_dict': { | |
501 | 'id': '135220665_111802303', | |
502 | 'ext': 'mp3', | |
503 | 'title': 'Black Shadow - Война - Негасимое Бездны Пламя!', | |
504 | 'duration': 423, | |
505 | 'uploader': 'Black Shadow', | |
506 | 'artist': 'Black Shadow', | |
507 | 'track': 'Война - Негасимое Бездны Пламя!', | |
508 | }, | |
509 | 'params': { | |
510 | 'skip_download': True, | |
511 | }, | |
512 | }], | |
51815886 S |
513 | 'params': { |
514 | 'usenetrc': True, | |
515 | }, | |
2d19fb50 S |
516 | 'skip': 'Requires vk account credentials', |
517 | }, { | |
518 | # single YouTube embed, no leading - | |
519 | 'url': 'https://vk.com/wall85155021_6319', | |
520 | 'info_dict': { | |
521 | 'id': '85155021_6319', | |
522 | 'title': 'Sergey Gorbunov - Wall post 85155021_6319', | |
523 | }, | |
524 | 'playlist_count': 1, | |
51815886 S |
525 | 'params': { |
526 | 'usenetrc': True, | |
527 | }, | |
2d19fb50 S |
528 | 'skip': 'Requires vk account credentials', |
529 | }, { | |
530 | # wall page URL | |
531 | 'url': 'https://vk.com/wall-23538238_35', | |
532 | 'only_matching': True, | |
533 | }, { | |
534 | # mobile wall page URL | |
535 | 'url': 'https://m.vk.com/wall-23538238_35', | |
536 | 'only_matching': True, | |
537 | }] | |
538 | ||
539 | def _real_extract(self, url): | |
540 | post_id = self._match_id(url) | |
541 | ||
542 | wall_url = 'https://vk.com/wall%s' % post_id | |
543 | ||
544 | post_id = remove_start(post_id, '-') | |
545 | ||
546 | webpage = self._download_webpage(wall_url, post_id) | |
547 | ||
548 | error = self._html_search_regex( | |
549 | r'>Error</div>\s*<div[^>]+class=["\']body["\'][^>]*>([^<]+)', | |
550 | webpage, 'error', default=None) | |
551 | if error: | |
552 | raise ExtractorError('VK said: %s' % error, expected=True) | |
553 | ||
554 | description = clean_html(get_element_by_class('wall_post_text', webpage)) | |
51815886 | 555 | uploader = clean_html(get_element_by_class('author', webpage)) |
2d19fb50 S |
556 | thumbnail = self._og_search_thumbnail(webpage) |
557 | ||
558 | entries = [] | |
559 | ||
51815886 S |
560 | audio_ids = re.findall(r'data-full-id=["\'](\d+_\d+)', webpage) |
561 | if audio_ids: | |
562 | al_audio = self._download_webpage( | |
563 | 'https://vk.com/al_audio.php', post_id, | |
564 | note='Downloading audio info', fatal=False, | |
565 | data=urlencode_postdata({ | |
566 | 'act': 'reload_audio', | |
567 | 'al': '1', | |
568 | 'ids': ','.join(audio_ids) | |
569 | })) | |
570 | if al_audio: | |
571 | Audio = collections.namedtuple( | |
572 | 'Audio', ['id', 'user_id', 'url', 'track', 'artist', 'duration']) | |
573 | audios = self._parse_json( | |
574 | self._search_regex( | |
575 | r'<!json>(.+?)<!>', al_audio, 'audios', default='[]'), | |
576 | post_id, fatal=False, transform_source=unescapeHTML) | |
577 | if isinstance(audios, list): | |
578 | for audio in audios: | |
579 | a = Audio._make(audio[:6]) | |
580 | entries.append({ | |
581 | 'id': '%s_%s' % (a.user_id, a.id), | |
582 | 'url': a.url, | |
583 | 'title': '%s - %s' % (a.artist, a.track) if a.artist and a.track else a.id, | |
584 | 'thumbnail': thumbnail, | |
585 | 'duration': a.duration, | |
586 | 'uploader': uploader, | |
587 | 'artist': a.artist, | |
588 | 'track': a.track, | |
589 | }) | |
2d19fb50 S |
590 | |
591 | for video in re.finditer( | |
592 | r'<a[^>]+href=(["\'])(?P<url>/video(?:-?[\d_]+).*?)\1', webpage): | |
593 | entries.append(self.url_result( | |
594 | compat_urlparse.urljoin(url, video.group('url')), VKIE.ie_key())) | |
595 | ||
596 | title = 'Wall post %s' % post_id | |
597 | ||
598 | return self.playlist_result( | |
599 | orderedSet(entries), post_id, | |
600 | '%s - %s' % (uploader, title) if uploader else title, | |
601 | description) |