2 from __future__ import unicode_literals
7 from .common import InfoExtractor
10 compat_urllib_request,
17 class VKIE(InfoExtractor):
19 _VALID_URL = r'https?://(?:m\.)?vk\.com/(?:video_ext\.php\?.*?\boid=(?P<oid>-?\d+).*?\bid=(?P<id>\d+)|(?:.+?\?.*?z=)?video(?P<videoid>[^s].*?)(?:\?|%2F|$))'
24 'url': 'http://vk.com/videos-77521?z=video-77521_162222515%2Fclub77521',
25 'md5': '0deae91935c54e00003c2a00646315f0',
29 'title': 'ProtivoGunz - Хуёвая песня',
30 'uploader': 're:Noize MC.*',
32 'upload_date': '20120212',
36 'url': 'http://vk.com/video205387401_165548505',
37 'md5': '6c0aeb2e90396ba97035b9cbde548700',
41 'uploader': 'Tom Cruise',
44 'upload_date': '20130721'
48 'note': 'Embedded video',
49 'url': 'http://vk.com/video_ext.php?oid=32194266&id=162925554&hash=7d8c2e0d5e05aeaa&hd=1',
50 'md5': 'c7ce8f1f87bec05b3de07fdeafe21a0a',
54 'uploader': 'Vladimir Gavrin',
57 'upload_date': '20120730',
62 # please update if you find a video whose URL follows the same pattern
63 'url': 'http://vk.com/video-8871596_164049491',
64 'md5': 'a590bcaf3d543576c9bd162812387666',
65 'note': 'Only available for registered users',
69 'uploader': 'Триллеры',
70 'title': '► Бойцовский клуб / Fight Club 1999 [HD 720]',
72 'upload_date': '20121218'
74 'skip': 'Requires vk account credentials',
78 # please update if you find a video whose URL follows the same pattern
79 'url': 'http://vk.com/feed?z=video-43215063_166094326%2Fbb50cacd3177146d7a',
80 'md5': 'd82c22e449f036282d1d3f7f4d276869',
84 'uploader': 'Киномания - лучшее из мира кино',
85 'title': 'Запах женщины (1992)',
87 'upload_date': '20130914'
89 'skip': 'Requires vk account credentials',
92 'url': 'http://vk.com/hd_kino_mania?z=video-43215063_168067957%2F15c66b9b533119788d',
93 'md5': '4d7a5ef8cf114dfa09577e57b2993202',
97 'uploader': 'Киномания - лучшее из мира кино',
100 'upload_date': '20140328',
102 'skip': 'Requires vk account credentials',
105 'url': 'http://m.vk.com/video-43215063_169084319?list=125c627d1aa1cebb83&from=wall-43215063_2566540',
106 'md5': '0c45586baa71b7cb1d0784ee3f4e00a6',
107 'note': 'ivi.ru embed',
111 'title': 'Книга Илая',
113 'upload_date': '20140626',
115 'skip': 'Only works from Russia',
120 (username, password) = self._get_login_info()
132 request = compat_urllib_request.Request('https://login.vk.com/?act=login',
133 compat_urllib_parse.urlencode(login_form).encode('utf-8'))
134 login_page = self._download_webpage(request, None, note='Logging in as %s' % username)
136 if re.search(r'onLoginFailed', login_page):
137 raise ExtractorError('Unable to login, incorrect username and/or password', expected=True)
139 def _real_initialize(self):
142 def _real_extract(self, url):
143 mobj = re.match(self._VALID_URL, url)
144 video_id = mobj.group('videoid')
147 video_id = '%s_%s' % (mobj.group('oid'), mobj.group('id'))
149 info_url = 'http://vk.com/al_video.php?act=show&al=1&video=%s' % video_id
150 info_page = self._download_webpage(info_url, video_id)
153 r'>Видеозапись .*? была изъята из публичного доступа в связи с обращением правообладателя.<':
154 'Video %s has been removed from public access due to rightholder complaint.',
155 r'<!>Please log in or <':
156 'Video %s is only available for registered users, '
157 'use --username and --password options to provide account credentials.',
159 'Video %s does not exist.'
162 for error_re, error_msg in ERRORS.items():
163 if re.search(error_re, info_page):
164 raise ExtractorError(error_msg % video_id, expected=True)
166 m_yt = re.search(r'src="(http://www.youtube.com/.*?)"', info_page)
168 self.to_screen('Youtube video detected')
169 return self.url_result(m_yt.group(1), 'Youtube')
171 m_opts = re.search(r'(?s)var\s+opts\s*=\s*({.*?});', info_page)
173 m_opts_url = re.search(r"url\s*:\s*'([^']+)", m_opts.group(1))
175 opts_url = m_opts_url.group(1)
176 if opts_url.startswith('//'):
177 opts_url = 'http:' + opts_url
178 return self.url_result(opts_url)
180 data_json = self._search_regex(r'var vars = ({.*?});', info_page, 'vars')
181 data = json.loads(data_json)
183 # Extract upload date
185 mobj = re.search(r'id="mv_date_wrap".*?Added ([a-zA-Z]+ [0-9]+), ([0-9]+) at', info_page)
187 x = mobj.group(1) + ' ' + mobj.group(2)
188 upload_date = unified_strdate(mobj.group(1) + ' ' + mobj.group(2))
193 'width': int(k[len('url'):]),
194 } for k, v in data.items()
195 if k.startswith('url')]
196 self._sort_formats(formats)
199 'id': compat_str(data['vid']),
201 'title': unescapeHTML(data['md_title']),
202 'thumbnail': data.get('jpg'),
203 'uploader': data.get('md_author'),
204 'duration': data.get('duration'),
205 'upload_date': upload_date,
209 class VKUserVideosIE(InfoExtractor):
210 IE_NAME = 'vk.com:user-videos'
211 IE_DESC = 'vk.com:All of a user\'s videos'
212 _VALID_URL = r'https?://vk\.com/videos(?P<id>[0-9]+)(?:m\?.*)?'
213 _TEMPLATE_URL = 'https://vk.com/videos'
215 'url': 'http://vk.com/videos205387401',
216 'playlist_mincount': 4,
219 def extract_videos_from_page(self, page):
221 for mobj in re.finditer(r'href="/video([0-9_]+)"', page):
222 if mobj.group(1) not in ids_in_page:
223 ids_in_page.append(mobj.group(1))
226 def _real_extract(self, url):
228 mobj = re.match(self._VALID_URL, url)
230 raise ExtractorError('Invalid URL: %s' % url)
232 # Download page and get video ids
233 page_id = mobj.group(1)
234 page = self._download_webpage(url, page_id)
235 video_ids = self.extract_videos_from_page(page)
237 self._downloader.to_screen('[vk] User videos %s: Found %i videos' % (page_id, len(video_ids)))
239 url_entries = [self.url_result('http://vk.com/video' + video_id, 'VK', video_id=video_id)
240 for video_id in video_ids]
241 return self.playlist_result(url_entries, page_id)