2 from __future__ import unicode_literals
4 from .common import InfoExtractor
6 compat_etree_fromstring,
8 compat_urllib_parse_unquote,
9 compat_urllib_parse_urlparse,
21 class OdnoklassnikiIE(InfoExtractor):
22 _VALID_URL = r'https?://(?:(?:www|m|mobile)\.)?(?:odnoklassniki|ok)\.ru/(?:video(?:embed)?/|web-api/video/moviePlayer/|live/|dk\?.*?st\.mvId=)(?P<id>[\d-]+)'
25 'url': 'http://ok.ru/video/20079905452',
26 'md5': '0b62089b479e06681abaaca9d204f152',
30 'title': 'Культура меняет нас (прекрасный ролик!))',
32 'upload_date': '20141207',
33 'uploader_id': '330537914540',
34 'uploader': 'Виталий Добровольский',
40 'url': 'http://ok.ru/video/63567059965189-0?fromTime=5',
41 'md5': '6ff470ea2dd51d5d18c295a355b0b6bc',
43 'id': '63567059965189-0',
45 'title': 'Девушка без комплексов ...',
47 'upload_date': '20150518',
48 'uploader_id': '534380003155',
49 'uploader': '☭ Андрей Мещанинов ☭',
55 # YouTube embed (metadataUrl, provider == USER_YOUTUBE)
56 'url': 'http://ok.ru/video/64211978996595-1',
57 'md5': '2f206894ffb5dbfcce2c5a14b909eea5',
61 'title': 'Космическая среда от 26 августа 2015',
62 'description': 'md5:848eb8b85e5e3471a3a803dae1343ed0',
64 'upload_date': '20150826',
65 'uploader_id': 'tvroscosmos',
66 'uploader': 'Телестудия Роскосмоса',
70 # YouTube embed (metadata, provider == USER_YOUTUBE, no metadata.movie.title field)
71 'url': 'http://ok.ru/video/62036049272859-0',
73 'id': '62036049272859-0',
75 'title': 'МУЗЫКА ДОЖДЯ .',
76 'description': 'md5:6f1867132bd96e33bf53eda1091e8ed0',
77 'upload_date': '20120106',
78 'uploader_id': '473534735899',
79 'uploader': 'МARINA D',
83 'skip_download': True,
85 'skip': 'Video has not been found',
87 'url': 'http://ok.ru/web-api/video/moviePlayer/20079905452',
88 'only_matching': True,
90 'url': 'http://www.ok.ru/video/20648036891',
91 'only_matching': True,
93 'url': 'http://www.ok.ru/videoembed/20648036891',
94 'only_matching': True,
96 'url': 'http://m.ok.ru/video/20079905452',
97 'only_matching': True,
99 'url': 'http://mobile.ok.ru/video/20079905452',
100 'only_matching': True,
102 'url': 'https://www.ok.ru/live/484531969818',
103 'only_matching': True,
105 'url': 'https://m.ok.ru/dk?st.cmd=movieLayer&st.discId=863789452017&st.retLoc=friend&st.rtu=%2Fdk%3Fst.cmd%3DfriendMovies%26st.mode%3Down%26st.mrkId%3D%257B%2522uploadedMovieMarker%2522%253A%257B%2522marker%2522%253A%25221519410114503%2522%252C%2522hasMore%2522%253Atrue%257D%252C%2522sharedMovieMarker%2522%253A%257B%2522marker%2522%253Anull%252C%2522hasMore%2522%253Afalse%257D%257D%26st.friendId%3D561722190321%26st.frwd%3Don%26_prevCmd%3DfriendMovies%26tkn%3D7257&st.discType=MOVIE&st.mvId=863789452017&_prevCmd=friendMovies&tkn=3648#lst#',
106 'only_matching': True,
109 def _real_extract(self, url):
110 start_time = int_or_none(compat_parse_qs(
111 compat_urllib_parse_urlparse(url).query).get('fromTime', [None])[0])
113 video_id = self._match_id(url)
115 webpage = self._download_webpage(
116 'http://ok.ru/video/%s' % video_id, video_id)
118 error = self._search_regex(
119 r'[^>]+class="vp_video_stub_txt"[^>]*>([^<]+)<',
120 webpage, 'error', default=None)
122 raise ExtractorError(error, expected=True)
124 player = self._parse_json(
125 unescapeHTML(self._search_regex(
126 r'data-options=(?P<quote>["\'])(?P<player>{.+?%s.+?})(?P=quote)' % video_id,
127 webpage, 'player', group='player')),
130 flashvars = player['flashvars']
132 metadata = flashvars.get('metadata')
134 metadata = self._parse_json(metadata, video_id)
137 st_location = flashvars.get('location')
139 data['st.location'] = st_location
140 metadata = self._download_json(
141 compat_urllib_parse_unquote(flashvars['metadataUrl']),
142 video_id, 'Downloading metadata JSON',
143 data=urlencode_postdata(data))
145 movie = metadata['movie']
147 # Some embedded videos may not contain title in movie dict (e.g.
148 # http://ok.ru/video/62036049272859-0) thus we allow missing title
149 # here and it's going to be extracted later by an extractor that
150 # will process the actual embed.
151 provider = metadata.get('provider')
152 title = movie['title'] if provider == 'UPLOADED_ODKL' else movie.get('title')
154 thumbnail = movie.get('poster')
155 duration = int_or_none(movie.get('duration'))
157 author = metadata.get('author', {})
158 uploader_id = author.get('id')
159 uploader = author.get('name')
161 upload_date = unified_strdate(self._html_search_meta(
162 'ya:ovs:upload_date', webpage, 'upload date', default=None))
165 adult = self._html_search_meta(
166 'ya:ovs:adult', webpage, 'age limit', default=None)
168 age_limit = 18 if adult == 'true' else 0
170 like_count = int_or_none(metadata.get('likeCount'))
175 'thumbnail': thumbnail,
176 'duration': duration,
177 'upload_date': upload_date,
178 'uploader': uploader,
179 'uploader_id': uploader_id,
180 'like_count': like_count,
181 'age_limit': age_limit,
182 'start_time': start_time,
185 if provider == 'USER_YOUTUBE':
187 '_type': 'url_transparent',
188 'url': movie['contentId'],
193 if provider == 'LIVE_TV_APP':
194 info['title'] = self._live_title(title)
196 quality = qualities(('4', '0', '1', '2', '3', '5'))
201 'format_id': f['name'],
202 } for f in metadata['videos']]
204 m3u8_url = metadata.get('hlsManifestUrl')
206 formats.extend(self._extract_m3u8_formats(
207 m3u8_url, video_id, 'mp4', 'm3u8_native',
208 m3u8_id='hls', fatal=False))
210 dash_manifest = metadata.get('metadataEmbedded')
212 formats.extend(self._parse_mpd_formats(
213 compat_etree_fromstring(dash_manifest), 'mpd'))
216 fmt_type = self._search_regex(
217 r'\btype[/=](\d)', fmt['url'],
218 'format type', default=None)
220 fmt['quality'] = quality(fmt_type)
223 m3u8_url = metadata.get('hlsMasterPlaylistUrl')
225 formats.extend(self._extract_m3u8_formats(
226 m3u8_url, video_id, 'mp4', entry_protocol='m3u8',
227 m3u8_id='hls', fatal=False))
228 rtmp_url = metadata.get('rtmpUrl')
236 self._sort_formats(formats)
238 info['formats'] = formats