]>
Commit | Line | Data |
---|---|---|
60d142aa | 1 | # encoding: utf-8 |
94a23d2a PH |
2 | from __future__ import unicode_literals |
3 | ||
60d142aa JMF |
4 | import re |
5 | import json | |
6 | ||
7 | from .common import InfoExtractor | |
1cc79574 PH |
8 | from ..compat import ( |
9 | compat_str, | |
10 | compat_urllib_parse, | |
11 | compat_urllib_request, | |
12 | ) | |
60d142aa | 13 | from ..utils import ( |
9032dc28 | 14 | ExtractorError, |
1cc79574 | 15 | orderedSet, |
8117df4c | 16 | str_to_int, |
60d142aa | 17 | unescapeHTML, |
e1e8b689 | 18 | unified_strdate, |
1cc79574 | 19 | ) |
84663361 | 20 | from .vimeo import VimeoIE |
60d142aa JMF |
21 | |
22 | ||
23 | class VKIE(InfoExtractor): | |
1ecb5d1d S |
24 | IE_NAME = 'vk' |
25 | IE_DESC = 'VK' | |
cf9cf7dd S |
26 | _VALID_URL = r'''(?x) |
27 | https?:// | |
28 | (?: | |
29 | (?:m\.)?vk\.com/video_ext\.php\?.*?\boid=(?P<oid>-?\d+).*?\bid=(?P<id>\d+)| | |
30 | (?: | |
31 | (?:m\.)?vk\.com/(?:.+?\?.*?z=)?video| | |
32 | (?:www\.)?biqle\.ru/watch/ | |
33 | ) | |
79913fde | 34 | (?P<videoid>[^s].*?)(?:\?(?:.*\blist=(?P<list_id>[\da-f]+))?|%2F|$) |
cf9cf7dd S |
35 | ) |
36 | ''' | |
5544e038 | 37 | _NETRC_MACHINE = 'vk' |
60d142aa | 38 | |
9032dc28 S |
39 | _TESTS = [ |
40 | { | |
41 | 'url': 'http://vk.com/videos-77521?z=video-77521_162222515%2Fclub77521', | |
42 | 'md5': '0deae91935c54e00003c2a00646315f0', | |
43 | 'info_dict': { | |
44 | 'id': '162222515', | |
45 | 'ext': 'flv', | |
46 | 'title': 'ProtivoGunz - Хуёвая песня', | |
36300346 | 47 | 'uploader': 're:(?:Noize MC|Alexander Ilyashenko).*', |
9032dc28 | 48 | 'duration': 195, |
42e1ff86 | 49 | 'upload_date': '20120212', |
8117df4c | 50 | 'view_count': int, |
9032dc28 | 51 | }, |
60d142aa | 52 | }, |
9032dc28 | 53 | { |
c52331f3 WS |
54 | 'url': 'http://vk.com/video205387401_165548505', |
55 | 'md5': '6c0aeb2e90396ba97035b9cbde548700', | |
9032dc28 | 56 | 'info_dict': { |
c52331f3 | 57 | 'id': '165548505', |
9032dc28 | 58 | 'ext': 'mp4', |
c52331f3 WS |
59 | 'uploader': 'Tom Cruise', |
60 | 'title': 'No name', | |
61 | 'duration': 9, | |
8117df4c S |
62 | 'upload_date': '20130721', |
63 | 'view_count': int, | |
9032dc28 S |
64 | } |
65 | }, | |
ca97a56e S |
66 | { |
67 | 'note': 'Embedded video', | |
68 | 'url': 'http://vk.com/video_ext.php?oid=32194266&id=162925554&hash=7d8c2e0d5e05aeaa&hd=1', | |
69 | 'md5': 'c7ce8f1f87bec05b3de07fdeafe21a0a', | |
70 | 'info_dict': { | |
71 | 'id': '162925554', | |
72 | 'ext': 'mp4', | |
73 | 'uploader': 'Vladimir Gavrin', | |
74 | 'title': 'Lin Dan', | |
75 | 'duration': 101, | |
42e1ff86 | 76 | 'upload_date': '20120730', |
8117df4c | 77 | 'view_count': int, |
ca97a56e S |
78 | } |
79 | }, | |
9032dc28 | 80 | { |
c52331f3 WS |
81 | # VIDEO NOW REMOVED |
82 | # please update if you find a video whose URL follows the same pattern | |
9032dc28 S |
83 | 'url': 'http://vk.com/video-8871596_164049491', |
84 | 'md5': 'a590bcaf3d543576c9bd162812387666', | |
85 | 'note': 'Only available for registered users', | |
86 | 'info_dict': { | |
87 | 'id': '164049491', | |
88 | 'ext': 'mp4', | |
89 | 'uploader': 'Триллеры', | |
57bdc730 | 90 | 'title': '► Бойцовский клуб / Fight Club 1999 [HD 720]', |
9032dc28 | 91 | 'duration': 8352, |
8117df4c S |
92 | 'upload_date': '20121218', |
93 | 'view_count': int, | |
9032dc28 S |
94 | }, |
95 | 'skip': 'Requires vk account credentials', | |
ca97a56e | 96 | }, |
57bdc730 S |
97 | { |
98 | 'url': 'http://vk.com/hd_kino_mania?z=video-43215063_168067957%2F15c66b9b533119788d', | |
99 | 'md5': '4d7a5ef8cf114dfa09577e57b2993202', | |
100 | 'info_dict': { | |
101 | 'id': '168067957', | |
102 | 'ext': 'mp4', | |
103 | 'uploader': 'Киномания - лучшее из мира кино', | |
104 | 'title': ' ', | |
105 | 'duration': 7291, | |
42e1ff86 | 106 | 'upload_date': '20140328', |
57bdc730 S |
107 | }, |
108 | 'skip': 'Requires vk account credentials', | |
109 | }, | |
849086a1 S |
110 | { |
111 | 'url': 'http://m.vk.com/video-43215063_169084319?list=125c627d1aa1cebb83&from=wall-43215063_2566540', | |
112 | 'md5': '0c45586baa71b7cb1d0784ee3f4e00a6', | |
113 | 'note': 'ivi.ru embed', | |
114 | 'info_dict': { | |
115 | 'id': '60690', | |
116 | 'ext': 'mp4', | |
117 | 'title': 'Книга Илая', | |
118 | 'duration': 6771, | |
42e1ff86 | 119 | 'upload_date': '20140626', |
8117df4c | 120 | 'view_count': int, |
849086a1 | 121 | }, |
d518d06e | 122 | 'skip': 'Only works from Russia', |
849086a1 | 123 | }, |
79913fde S |
124 | { |
125 | # video (removed?) only available with list id | |
126 | 'url': 'https://vk.com/video30481095_171201961?list=8764ae2d21f14088d4', | |
127 | 'md5': '091287af5402239a1051c37ec7b92913', | |
128 | 'info_dict': { | |
129 | 'id': '171201961', | |
130 | 'ext': 'mp4', | |
131 | 'title': 'ТюменцевВВ_09.07.2015', | |
132 | 'uploader': 'Anton Ivanov', | |
133 | 'duration': 109, | |
134 | 'upload_date': '20150709', | |
135 | 'view_count': int, | |
136 | }, | |
137 | }, | |
9281f6d2 S |
138 | { |
139 | # youtube embed | |
140 | 'url': 'https://vk.com/video276849682_170681728', | |
141 | 'info_dict': { | |
142 | 'id': 'V3K4mi0SYkc', | |
143 | 'ext': 'mp4', | |
144 | 'title': "DSWD Awards 'Children's Joy Foundation, Inc.' Certificate of Registration and License to Operate", | |
145 | 'description': 'md5:bf9c26cfa4acdfb146362682edd3827a', | |
146 | 'duration': 179, | |
147 | 'upload_date': '20130116', | |
148 | 'uploader': "Children's Joy Foundation", | |
149 | 'uploader_id': 'thecjf', | |
150 | 'view_count': int, | |
151 | }, | |
152 | }, | |
a8363f3a PH |
153 | { |
154 | # removed video, just testing that we match the pattern | |
155 | 'url': 'http://vk.com/feed?z=video-43215063_166094326%2Fbb50cacd3177146d7a', | |
156 | 'only_matching': True, | |
157 | }, | |
e58066e2 S |
158 | { |
159 | # age restricted video, requires vk account credentials | |
160 | 'url': 'https://vk.com/video205387401_164765225', | |
161 | 'only_matching': True, | |
162 | }, | |
cf9cf7dd S |
163 | { |
164 | # vk wrapper | |
165 | 'url': 'http://www.biqle.ru/watch/847655_160197695', | |
166 | 'only_matching': True, | |
167 | } | |
9032dc28 S |
168 | ] |
169 | ||
170 | def _login(self): | |
171 | (username, password) = self._get_login_info() | |
172 | if username is None: | |
173 | return | |
174 | ||
226b886c S |
175 | login_page = self._download_webpage( |
176 | 'https://vk.com', None, 'Downloading login page') | |
177 | ||
f8da79f8 | 178 | login_form = self._hidden_inputs(login_page) |
226b886c S |
179 | |
180 | login_form.update({ | |
4f3bf679 S |
181 | 'email': username.encode('cp1251'), |
182 | 'pass': password.encode('cp1251'), | |
226b886c | 183 | }) |
9032dc28 | 184 | |
226b886c S |
185 | request = compat_urllib_request.Request( |
186 | 'https://login.vk.com/?act=login', | |
187 | compat_urllib_parse.urlencode(login_form).encode('utf-8')) | |
188 | login_page = self._download_webpage( | |
189 | request, None, note='Logging in as %s' % username) | |
9032dc28 S |
190 | |
191 | if re.search(r'onLoginFailed', login_page): | |
226b886c S |
192 | raise ExtractorError( |
193 | 'Unable to login, incorrect username and/or password', expected=True) | |
9032dc28 S |
194 | |
195 | def _real_initialize(self): | |
196 | self._login() | |
60d142aa JMF |
197 | |
198 | def _real_extract(self, url): | |
199 | mobj = re.match(self._VALID_URL, url) | |
ca97a56e S |
200 | video_id = mobj.group('videoid') |
201 | ||
202 | if not video_id: | |
203 | video_id = '%s_%s' % (mobj.group('oid'), mobj.group('id')) | |
9032dc28 | 204 | |
275c0423 | 205 | info_url = 'https://vk.com/al_video.php?act=show&al=1&module=video&video=%s' % video_id |
79913fde S |
206 | |
207 | # Some videos (removed?) can only be downloaded with list id specified | |
208 | list_id = mobj.group('list_id') | |
209 | if list_id: | |
210 | info_url += '&list=%s' % list_id | |
211 | ||
60d142aa | 212 | info_page = self._download_webpage(info_url, video_id) |
9032dc28 | 213 | |
ee48b6a8 S |
214 | error_message = self._html_search_regex( |
215 | r'(?s)<!><div[^>]+class="video_layer_message"[^>]*>(.+?)</div>', | |
216 | info_page, 'error message', default=None) | |
217 | if error_message: | |
218 | raise ExtractorError(error_message, expected=True) | |
219 | ||
7f220b2f S |
220 | if re.search(r'<!>/login\.php\?.*\bact=security_check', info_page): |
221 | raise ExtractorError( | |
222 | 'You are trying to log in from an unusual location. You should confirm ownership at vk.com to log in with this IP.', | |
223 | expected=True) | |
224 | ||
e0c51cda S |
225 | ERRORS = { |
226 | r'>Видеозапись .*? была изъята из публичного доступа в связи с обращением правообладателя.<': | |
3d36cea4 PH |
227 | 'Video %s has been removed from public access due to rightholder complaint.', |
228 | ||
e0c51cda | 229 | r'<!>Please log in or <': |
3d36cea4 PH |
230 | 'Video %s is only available for registered users, ' |
231 | 'use --username and --password options to provide account credentials.', | |
232 | ||
233 | r'<!>Unknown error': | |
1aa5172f S |
234 | 'Video %s does not exist.', |
235 | ||
236 | r'<!>Видео временно недоступно': | |
237 | 'Video %s is temporarily unavailable.', | |
d919fa33 S |
238 | |
239 | r'<!>Access denied': | |
240 | 'Access denied to video %s.', | |
e0c51cda | 241 | } |
9032dc28 | 242 | |
e0c51cda S |
243 | for error_re, error_msg in ERRORS.items(): |
244 | if re.search(error_re, info_page): | |
245 | raise ExtractorError(error_msg % video_id, expected=True) | |
9334f8f1 | 246 | |
46478456 S |
247 | youtube_url = self._search_regex( |
248 | r'<iframe[^>]+src="((?:https?:)?//www.youtube.com/embed/[^"]+)"', | |
249 | info_page, 'youtube iframe', default=None) | |
250 | if youtube_url: | |
9281f6d2 | 251 | return self.url_result(youtube_url, 'Youtube') |
849086a1 | 252 | |
84663361 S |
253 | vimeo_url = VimeoIE._extract_vimeo_url(url, info_page) |
254 | if vimeo_url is not None: | |
255 | return self.url_result(vimeo_url) | |
256 | ||
7a1818c9 PH |
257 | m_rutube = re.search( |
258 | r'\ssrc="((?:https?:)?//rutube\.ru\\?/video\\?/embed(?:.*?))\\?"', info_page) | |
7a1818c9 PH |
259 | if m_rutube is not None: |
260 | self.to_screen('rutube video detected') | |
261 | rutube_url = self._proto_relative_url( | |
262 | m_rutube.group(1).replace('\\', '')) | |
263 | return self.url_result(rutube_url) | |
264 | ||
054932f4 | 265 | m_opts = re.search(r'(?s)var\s+opts\s*=\s*({.+?});', info_page) |
849086a1 | 266 | if m_opts: |
054932f4 | 267 | m_opts_url = re.search(r"url\s*:\s*'((?!/\b)[^']+)", m_opts.group(1)) |
849086a1 S |
268 | if m_opts_url: |
269 | opts_url = m_opts_url.group(1) | |
270 | if opts_url.startswith('//'): | |
271 | opts_url = 'http:' + opts_url | |
272 | return self.url_result(opts_url) | |
273 | ||
054932f4 | 274 | data_json = self._search_regex(r'var\s+vars\s*=\s*({.+?});', info_page, 'vars') |
608bf698 | 275 | data = json.loads(data_json) |
60d142aa | 276 | |
02a12f9f WS |
277 | # Extract upload date |
278 | upload_date = None | |
7c7dd9dc | 279 | mobj = re.search(r'id="mv_date(?:_views)?_wrap"[^>]*>([a-zA-Z]+ [0-9]+), ([0-9]+) at', info_page) |
02a12f9f | 280 | if mobj is not None: |
9e1a5b84 | 281 | mobj.group(1) + ' ' + mobj.group(2) |
02a12f9f WS |
282 | upload_date = unified_strdate(mobj.group(1) + ' ' + mobj.group(2)) |
283 | ||
8117df4c S |
284 | view_count = str_to_int(self._search_regex( |
285 | r'"mv_views_count_number"[^>]*>([\d,.]+) views<', | |
286 | info_page, 'view count', fatal=False)) | |
287 | ||
913f3292 PH |
288 | formats = [{ |
289 | 'format_id': k, | |
290 | 'url': v, | |
291 | 'width': int(k[len('url'):]), | |
292 | } for k, v in data.items() | |
293 | if k.startswith('url')] | |
294 | self._sort_formats(formats) | |
295 | ||
60d142aa | 296 | return { |
608bf698 | 297 | 'id': compat_str(data['vid']), |
913f3292 | 298 | 'formats': formats, |
608bf698 | 299 | 'title': unescapeHTML(data['md_title']), |
913f3292 PH |
300 | 'thumbnail': data.get('jpg'), |
301 | 'uploader': data.get('md_author'), | |
02a12f9f WS |
302 | 'duration': data.get('duration'), |
303 | 'upload_date': upload_date, | |
8117df4c | 304 | 'view_count': view_count, |
60d142aa | 305 | } |
469d4c89 WS |
306 | |
307 | ||
308 | class VKUserVideosIE(InfoExtractor): | |
1ecb5d1d S |
309 | IE_NAME = 'vk:uservideos' |
310 | IE_DESC = "VK - User's Videos" | |
dc786d3d | 311 | _VALID_URL = r'https?://vk\.com/videos(?P<id>-?[0-9]+)$' |
469d4c89 | 312 | _TEMPLATE_URL = 'https://vk.com/videos' |
dc786d3d | 313 | _TESTS = [{ |
469d4c89 | 314 | 'url': 'http://vk.com/videos205387401', |
15ec6693 PH |
315 | 'info_dict': { |
316 | 'id': '205387401', | |
dc786d3d | 317 | 'title': "Tom Cruise's Videos", |
15ec6693 | 318 | }, |
469d4c89 | 319 | 'playlist_mincount': 4, |
dc786d3d S |
320 | }, { |
321 | 'url': 'http://vk.com/videos-77521', | |
322 | 'only_matching': True, | |
323 | }] | |
469d4c89 | 324 | |
469d4c89 | 325 | def _real_extract(self, url): |
021a0db8 | 326 | page_id = self._match_id(url) |
dc786d3d S |
327 | |
328 | webpage = self._download_webpage(url, page_id) | |
329 | ||
330 | entries = [ | |
d16abf43 PH |
331 | self.url_result( |
332 | 'http://vk.com/video' + video_id, 'VK', video_id=video_id) | |
fec73daa | 333 | for video_id in orderedSet(re.findall(r'href="/video(-?[0-9_]+)"', webpage))] |
dc786d3d S |
334 | |
335 | title = unescapeHTML(self._search_regex( | |
336 | r'<title>\s*([^<]+?)\s+\|\s+\d+\s+videos', | |
337 | webpage, 'title', default=page_id)) | |
338 | ||
339 | return self.playlist_result(entries, page_id, title) |