yt_dlp/extractor/videomore.py

   1 from .common import InfoExtractor
   2 from ..compat import (
   3     compat_str,
   4 )
   5 from ..utils import (
   6     int_or_none,
   7     parse_qs,
   8 )
   9
  10
  11 class VideomoreBaseIE(InfoExtractor):
  12     _API_BASE_URL = 'https://more.tv/api/v3/web/'
  13     _VALID_URL_BASE = r'https?://(?:videomore\.ru|more\.tv)/'
  14
  15     def _download_page_data(self, display_id):
  16         return self._download_json(
  17             self._API_BASE_URL + 'PageData', display_id, query={
  18                 'url': '/' + display_id,
  19             })['attributes']['response']['data']
  20
  21     def _track_url_result(self, track):
  22         track_vod = track['trackVod']
  23         video_url = track_vod.get('playerLink') or track_vod['link']
  24         return self.url_result(
  25             video_url, VideomoreIE.ie_key(), track_vod.get('hubId'))
  26
  27
  28 class VideomoreIE(InfoExtractor):
  29     IE_NAME = 'videomore'
  30     _VALID_URL = r'''(?x)
  31                     videomore:(?P<sid>\d+)$|
  32                     https?://
  33                         (?:
  34                             videomore\.ru/
  35                             (?:
  36                                 embed|
  37                                 [^/]+/[^/]+
  38                             )/|
  39                             (?:
  40                                 (?:player\.)?videomore\.ru|
  41                                 siren\.more\.tv/player
  42                             )/[^/]*\?.*?\btrack_id=|
  43                             odysseus\.more.tv/player/(?P<partner_id>\d+)/
  44                         )
  45                         (?P<id>\d+)
  46                         (?:[/?#&]|\.(?:xml|json)|$)
  47                     '''
  48     _EMBED_REGEX = [r'''(?x)
  49         (?:
  50             <iframe[^>]+src=([\'"])|
  51             <object[^>]+data=(["\'])https?://videomore\.ru/player\.swf\?.*config=
  52         )(?P<url>https?://videomore\.ru/[^?#"']+/\d+(?:\.xml)?)
  53     ''']
  54     _TESTS = [{
  55         'url': 'http://videomore.ru/kino_v_detalayah/5_sezon/367617',
  56         'md5': '44455a346edc0d509ac5b5a5b531dc35',
  57         'info_dict': {
  58             'id': '367617',
  59             'ext': 'flv',
  60             'title': 'Кино в деталях 5 сезон В гостях Алексей Чумаков и Юлия Ковальчук',
  61             'series': 'Кино в деталях',
  62             'episode': 'В гостях Алексей Чумаков и Юлия Ковальчук',
  63             'thumbnail': r're:^https?://.*\.jpg',
  64             'duration': 2910,
  65             'view_count': int,
  66             'comment_count': int,
  67             'age_limit': 16,
  68         },
  69         'skip': 'The video is not available for viewing.',
  70     }, {
  71         'url': 'http://videomore.ru/embed/259974',
  72         'info_dict': {
  73             'id': '259974',
  74             'ext': 'mp4',
  75             'title': 'Молодежка 2 сезон 40 серия',
  76             'series': 'Молодежка',
  77             'season': '2 сезон',
  78             'episode': '40 серия',
  79             'thumbnail': r're:^https?://.*\.jpg',
  80             'duration': 2789,
  81             'view_count': int,
  82             'age_limit': 16,
  83         },
  84         'params': {
  85             'skip_download': True,
  86         },
  87     }, {
  88         'url': 'http://videomore.ru/molodezhka/sezon_promo/341073',
  89         'info_dict': {
  90             'id': '341073',
  91             'ext': 'flv',
  92             'title': 'Промо Команда проиграла из-за Бакина?',
  93             'episode': 'Команда проиграла из-за Бакина?',
  94             'thumbnail': r're:^https?://.*\.jpg',
  95             'duration': 29,
  96             'age_limit': 16,
  97             'view_count': int,
  98         },
  99         'params': {
 100             'skip_download': True,
 101         },
 102         'skip': 'The video is not available for viewing.',
 103     }, {
 104         'url': 'http://videomore.ru/elki_3?track_id=364623',
 105         'only_matching': True,
 106     }, {
 107         'url': 'http://videomore.ru/embed/364623',
 108         'only_matching': True,
 109     }, {
 110         'url': 'http://videomore.ru/video/tracks/364623.xml',
 111         'only_matching': True,
 112     }, {
 113         'url': 'http://videomore.ru/video/tracks/364623.json',
 114         'only_matching': True,
 115     }, {
 116         'url': 'http://videomore.ru/video/tracks/158031/quotes/33248',
 117         'only_matching': True,
 118     }, {
 119         'url': 'videomore:367617',
 120         'only_matching': True,
 121     }, {
 122         'url': 'https://player.videomore.ru/?partner_id=97&track_id=736234&autoplay=0&userToken=',
 123         'only_matching': True,
 124     }, {
 125         'url': 'https://odysseus.more.tv/player/1788/352317',
 126         'only_matching': True,
 127     }, {
 128         'url': 'https://siren.more.tv/player/config?track_id=352317&partner_id=1788&user_token=',
 129         'only_matching': True,
 130     }]
 131     _GEO_BYPASS = False
 132
 133     def _real_extract(self, url):
 134         mobj = self._match_valid_url(url)
 135         video_id = mobj.group('sid') or mobj.group('id')
 136         partner_id = mobj.group('partner_id') or parse_qs(url).get('partner_id', [None])[0] or '97'
 137
 138         item = self._download_json(
 139             'https://siren.more.tv/player/config', video_id, query={
 140                 'partner_id': partner_id,
 141                 'track_id': video_id,
 142             })['data']['playlist']['items'][0]
 143
 144         title = item.get('title')
 145         series = item.get('project_name')
 146         season = item.get('season_name')
 147         episode = item.get('episode_name')
 148         if not title:
 149             title = []
 150             for v in (series, season, episode):
 151                 if v:
 152                     title.append(v)
 153             title = ' '.join(title)
 154
 155         streams = item.get('streams') or []
 156         for protocol in ('DASH', 'HLS'):
 157             stream_url = item.get(protocol.lower() + '_url')
 158             if stream_url:
 159                 streams.append({'protocol': protocol, 'url': stream_url})
 160
 161         formats = []
 162         for stream in streams:
 163             stream_url = stream.get('url')
 164             if not stream_url:
 165                 continue
 166             protocol = stream.get('protocol')
 167             if protocol == 'DASH':
 168                 formats.extend(self._extract_mpd_formats(
 169                     stream_url, video_id, mpd_id='dash', fatal=False))
 170             elif protocol == 'HLS':
 171                 formats.extend(self._extract_m3u8_formats(
 172                     stream_url, video_id, 'mp4', 'm3u8_native',
 173                     m3u8_id='hls', fatal=False))
 174             elif protocol == 'MSS':
 175                 formats.extend(self._extract_ism_formats(
 176                     stream_url, video_id, ism_id='mss', fatal=False))
 177
 178         if not formats:
 179             error = item.get('error')
 180             if error:
 181                 if error in ('Данное видео недоступно для просмотра на территории этой страны', 'Данное видео доступно для просмотра только на территории России'):
 182                     self.raise_geo_restricted(countries=['RU'], metadata_available=True)
 183                 self.raise_no_formats(error, expected=True)
 184         self._sort_formats(formats)
 185
 186         return {
 187             'id': video_id,
 188             'title': title,
 189             'series': series,
 190             'season': season,
 191             'episode': episode,
 192             'thumbnail': item.get('thumbnail_url'),
 193             'duration': int_or_none(item.get('duration')),
 194             'view_count': int_or_none(item.get('views')),
 195             'age_limit': int_or_none(item.get('min_age')),
 196             'formats': formats,
 197         }
 198
 199
 200 class VideomoreVideoIE(VideomoreBaseIE):
 201     IE_NAME = 'videomore:video'
 202     _VALID_URL = VideomoreBaseIE._VALID_URL_BASE + r'(?P<id>(?:(?:[^/]+/){2})?[^/?#&]+)(?:/*|[?#&].*?)$'
 203     _TESTS = [{
 204         # single video with og:video:iframe
 205         'url': 'http://videomore.ru/elki_3',
 206         'info_dict': {
 207             'id': '364623',
 208             'ext': 'flv',
 209             'title': 'Ёлки 3',
 210             'description': '',
 211             'thumbnail': r're:^https?://.*\.jpg',
 212             'duration': 5579,
 213             'age_limit': 6,
 214             'view_count': int,
 215         },
 216         'params': {
 217             'skip_download': True,
 218         },
 219         'skip': 'Requires logging in',
 220     }, {
 221         # season single series with og:video:iframe
 222         'url': 'http://videomore.ru/poslednii_ment/1_sezon/14_seriya',
 223         'info_dict': {
 224             'id': '352317',
 225             'ext': 'mp4',
 226             'title': 'Последний мент 1 сезон 14 серия',
 227             'series': 'Последний мент',
 228             'season': '1 сезон',
 229             'episode': '14 серия',
 230             'thumbnail': r're:^https?://.*\.jpg',
 231             'duration': 2464,
 232             'age_limit': 16,
 233             'view_count': int,
 234         },
 235         'params': {
 236             'skip_download': True,
 237         },
 238     }, {
 239         'url': 'http://videomore.ru/sejchas_v_seti/serii_221-240/226_vypusk',
 240         'only_matching': True,
 241     }, {
 242         # single video without og:video:iframe
 243         'url': 'http://videomore.ru/marin_i_ego_druzya',
 244         'info_dict': {
 245             'id': '359073',
 246             'ext': 'flv',
 247             'title': '1 серия. Здравствуй, Аквавилль!',
 248             'description': 'md5:c6003179538b5d353e7bcd5b1372b2d7',
 249             'thumbnail': r're:^https?://.*\.jpg',
 250             'duration': 754,
 251             'age_limit': 6,
 252             'view_count': int,
 253         },
 254         'params': {
 255             'skip_download': True,
 256         },
 257         'skip': 'redirects to https://more.tv/'
 258     }, {
 259         'url': 'https://videomore.ru/molodezhka/6_sezon/29_seriya?utm_so',
 260         'only_matching': True,
 261     }, {
 262         'url': 'https://more.tv/poslednii_ment/1_sezon/14_seriya',
 263         'only_matching': True,
 264     }]
 265
 266     @classmethod
 267     def suitable(cls, url):
 268         return False if VideomoreIE.suitable(url) else super(VideomoreVideoIE, cls).suitable(url)
 269
 270     def _real_extract(self, url):
 271         display_id = self._match_id(url)
 272         return self._track_url_result(self._download_page_data(display_id))
 273
 274
 275 class VideomoreSeasonIE(VideomoreBaseIE):
 276     IE_NAME = 'videomore:season'
 277     _VALID_URL = VideomoreBaseIE._VALID_URL_BASE + r'(?!embed)(?P<id>[^/]+/[^/?#&]+)(?:/*|[?#&].*?)$'
 278     _TESTS = [{
 279         'url': 'http://videomore.ru/molodezhka/film_o_filme',
 280         'info_dict': {
 281             'id': 'molodezhka/film_o_filme',
 282             'title': 'Фильм о фильме',
 283         },
 284         'playlist_mincount': 3,
 285     }, {
 286         'url': 'http://videomore.ru/molodezhka/sezon_promo?utm_so',
 287         'only_matching': True,
 288     }, {
 289         'url': 'https://more.tv/molodezhka/film_o_filme',
 290         'only_matching': True,
 291     }]
 292
 293     @classmethod
 294     def suitable(cls, url):
 295         return (False if (VideomoreIE.suitable(url) or VideomoreVideoIE.suitable(url))
 296                 else super(VideomoreSeasonIE, cls).suitable(url))
 297
 298     def _real_extract(self, url):
 299         display_id = self._match_id(url)
 300         season = self._download_page_data(display_id)
 301         season_id = compat_str(season['id'])
 302         tracks = self._download_json(
 303             self._API_BASE_URL + 'seasons/%s/tracks' % season_id,
 304             season_id)['data']
 305         entries = []
 306         for track in tracks:
 307             entries.append(self._track_url_result(track))
 308         return self.playlist_result(entries, display_id, season.get('title'))