yt_dlp/extractor/videomore.py

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