2 from __future__ import unicode_literals
7 from .common import InfoExtractor
10 compat_urllib_request,
17 class VKIE(InfoExtractor):
19 _VALID_URL = r'https?://vk\.com/(?:video_ext\.php\?.*?\boid=(?P<oid>-?\d+).*?\bid=(?P<id>\d+)|(?:.+?\?.*?z=)?video(?P<videoid>.*?)(?:\?|%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.*',
35 'url': 'http://vk.com/video4643923_163339118',
36 'md5': 'f79bccb5cd182b1f43502ca5685b2b36',
40 'uploader': 'Elya Iskhakova',
41 'title': 'Dream Theater - Hollow Years Live at Budokan 720*',
46 'note': 'Embedded video',
47 'url': 'http://vk.com/video_ext.php?oid=32194266&id=162925554&hash=7d8c2e0d5e05aeaa&hd=1',
48 'md5': 'c7ce8f1f87bec05b3de07fdeafe21a0a',
52 'uploader': 'Vladimir Gavrin',
58 'url': 'http://vk.com/video-8871596_164049491',
59 'md5': 'a590bcaf3d543576c9bd162812387666',
60 'note': 'Only available for registered users',
64 'uploader': 'Триллеры',
65 'title': '► Бойцовский клуб / Fight Club 1999 [HD 720]',
68 'skip': 'Requires vk account credentials',
71 'url': 'http://vk.com/feed?z=video-43215063_166094326%2Fbb50cacd3177146d7a',
72 'md5': 'd82c22e449f036282d1d3f7f4d276869',
76 'uploader': 'Киномания - лучшее из мира кино',
77 'title': 'Запах женщины (1992)',
80 'skip': 'Requires vk account credentials',
83 'url': 'http://vk.com/hd_kino_mania?z=video-43215063_168067957%2F15c66b9b533119788d',
84 'md5': '4d7a5ef8cf114dfa09577e57b2993202',
88 'uploader': 'Киномания - лучшее из мира кино',
92 'skip': 'Requires vk account credentials',
97 (username, password) = self._get_login_info()
109 request = compat_urllib_request.Request('https://login.vk.com/?act=login',
110 compat_urllib_parse.urlencode(login_form).encode('utf-8'))
111 login_page = self._download_webpage(request, None, note='Logging in as %s' % username)
113 if re.search(r'onLoginFailed', login_page):
114 raise ExtractorError('Unable to login, incorrect username and/or password', expected=True)
116 def _real_initialize(self):
119 def _real_extract(self, url):
120 mobj = re.match(self._VALID_URL, url)
121 video_id = mobj.group('videoid')
124 video_id = '%s_%s' % (mobj.group('oid'), mobj.group('id'))
126 info_url = 'http://vk.com/al_video.php?act=show&al=1&video=%s' % video_id
127 info_page = self._download_webpage(info_url, video_id)
129 if re.search(r'<!>Please log in or <', info_page):
130 raise ExtractorError('This video is only available for registered users, '
131 'use --username and --password options to provide account credentials.', expected=True)
133 m_yt = re.search(r'src="(http://www.youtube.com/.*?)"', info_page)
135 self.to_screen('Youtube video detected')
136 return self.url_result(m_yt.group(1), 'Youtube')
137 data_json = self._search_regex(r'var vars = ({.*?});', info_page, 'vars')
138 data = json.loads(data_json)
143 'width': int(k[len('url'):]),
144 } for k, v in data.items()
145 if k.startswith('url')]
146 self._sort_formats(formats)
149 'id': compat_str(data['vid']),
151 'title': unescapeHTML(data['md_title']),
152 'thumbnail': data.get('jpg'),
153 'uploader': data.get('md_author'),
154 'duration': data.get('duration')