2 from __future__ import unicode_literals
4 from .common import InfoExtractor
5 from ..compat import compat_urllib_parse_unquote
15 class OdnoklassnikiIE(InfoExtractor):
16 _VALID_URL = r'https?://(?:(?:www|m|mobile)\.)?(?:odnoklassniki|ok)\.ru/(?:video(?:embed)?|web-api/video/moviePlayer)/(?P<id>[\d-]+)'
19 'url': 'http://ok.ru/video/20079905452',
20 'md5': '6ba728d85d60aa2e6dd37c9e70fdc6bc',
24 'title': 'Культура меняет нас (прекрасный ролик!))',
26 'upload_date': '20141207',
27 'uploader_id': '330537914540',
28 'uploader': 'Виталий Добровольский',
32 'skip': 'Video has been blocked',
35 'url': 'http://ok.ru/video/63567059965189-0',
36 'md5': '9676cf86eff5391d35dea675d224e131',
38 'id': '63567059965189-0',
40 'title': 'Девушка без комплексов ...',
42 'upload_date': '20150518',
43 'uploader_id': '534380003155',
44 'uploader': '☭ Андрей Мещанинов ☭',
49 # YouTube embed (metadataUrl, provider == USER_YOUTUBE)
50 'url': 'http://ok.ru/video/64211978996595-1',
51 'md5': '5d7475d428845cd2e13bae6f1a992278',
53 'id': '64211978996595-1',
55 'title': 'Космическая среда от 26 августа 2015',
56 'description': 'md5:848eb8b85e5e3471a3a803dae1343ed0',
58 'upload_date': '20150826',
59 'uploader_id': '750099571',
60 'uploader': 'Алина П',
64 # YouTube embed (metadata, provider == USER_YOUTUBE, no metadata.movie.title field)
65 'url': 'http://ok.ru/video/62036049272859-0',
67 'id': '62036049272859-0',
69 'title': 'МУЗЫКА ДОЖДЯ .',
70 'description': 'md5:6f1867132bd96e33bf53eda1091e8ed0',
71 'upload_date': '20120106',
72 'uploader_id': '473534735899',
73 'uploader': 'МARINA D',
77 'skip_download': True,
80 'url': 'http://ok.ru/web-api/video/moviePlayer/20079905452',
81 'only_matching': True,
83 'url': 'http://www.ok.ru/video/20648036891',
84 'only_matching': True,
86 'url': 'http://www.ok.ru/videoembed/20648036891',
87 'only_matching': True,
89 'url': 'http://m.ok.ru/video/20079905452',
90 'only_matching': True,
92 'url': 'http://mobile.ok.ru/video/20079905452',
93 'only_matching': True,
96 def _real_extract(self, url):
97 video_id = self._match_id(url)
99 webpage = self._download_webpage(
100 'http://ok.ru/video/%s' % video_id, video_id)
102 error = self._search_regex(
103 r'[^>]+class="vp_video_stub_txt"[^>]*>([^<]+)<',
104 webpage, 'error', default=None)
106 raise ExtractorError(error, expected=True)
108 player = self._parse_json(
109 unescapeHTML(self._search_regex(
110 r'data-options=(?P<quote>["\'])(?P<player>{.+?%s.+?})(?P=quote)' % video_id,
111 webpage, 'player', group='player')),
114 flashvars = player['flashvars']
116 metadata = flashvars.get('metadata')
118 metadata = self._parse_json(metadata, video_id)
120 metadata = self._download_json(
121 compat_urllib_parse_unquote(flashvars['metadataUrl']),
122 video_id, 'Downloading metadata JSON')
124 movie = metadata['movie']
126 # Some embedded videos may not contain title in movie dict (e.g.
127 # http://ok.ru/video/62036049272859-0) thus we allow missing title
128 # here and it's going to be extracted later by an extractor that
129 # will process the actual embed.
130 provider = metadata.get('provider')
131 title = movie['title'] if provider == 'UPLOADED_ODKL' else movie.get('title')
133 thumbnail = movie.get('poster')
134 duration = int_or_none(movie.get('duration'))
136 author = metadata.get('author', {})
137 uploader_id = author.get('id')
138 uploader = author.get('name')
140 upload_date = unified_strdate(self._html_search_meta(
141 'ya:ovs:upload_date', webpage, 'upload date', default=None))
144 adult = self._html_search_meta(
145 'ya:ovs:adult', webpage, 'age limit', default=None)
147 age_limit = 18 if adult == 'true' else 0
149 like_count = int_or_none(metadata.get('likeCount'))
154 'thumbnail': thumbnail,
155 'duration': duration,
156 'upload_date': upload_date,
157 'uploader': uploader,
158 'uploader_id': uploader_id,
159 'like_count': like_count,
160 'age_limit': age_limit,
163 if provider == 'USER_YOUTUBE':
165 '_type': 'url_transparent',
166 'url': movie['contentId'],
170 quality = qualities(('mobile', 'lowest', 'low', 'sd', 'hd'))
175 'format_id': f['name'],
176 'quality': quality(f['name']),
177 } for f in metadata['videos']]
178 self._sort_formats(formats)
180 info['formats'] = formats