yt_dlp/extractor/theholetv.py

   1 from .common import InfoExtractor
   2 from ..utils import extract_attributes, remove_end
   3
   4
   5 class TheHoleTvIE(InfoExtractor):
   6     _VALID_URL = r'https?://(?:www\.)?the-hole\.tv/episodes/(?P<id>[\w-]+)'
   7     _TESTS = [{
   8         'url': 'https://the-hole.tv/episodes/gromkii-vopros-sergey-orlov',
   9         'md5': 'fea6682f47786f3ae5a6cbd635ec4bf9',
  10         'info_dict': {
  11             'id': 'gromkii-vopros-sergey-orlov',
  12             'ext': 'mp4',
  13             'title': 'Сергей Орлов — Громкий вопрос',
  14             'thumbnail': 'https://assets-cdn.the-hole.tv/images/t8gan4n6zn627e7wni11b2uemqts',
  15             'description': 'md5:45741a9202331f995d9fb76996759379',
  16         },
  17     }]
  18
  19     def _real_extract(self, url):
  20         video_id = self._match_id(url)
  21         webpage = self._download_webpage(url, video_id)
  22
  23         player_attrs = extract_attributes(self._search_regex(
  24             r'(<div[^>]*\bdata-controller="player"[^>]*>)', webpage, 'video player'))
  25         formats, subtitles = self._extract_m3u8_formats_and_subtitles(
  26             player_attrs['data-player-source-value'], video_id, 'mp4')
  27
  28         return {
  29             'id': video_id,
  30             'title': remove_end(self._html_extract_title(webpage), ' — The Hole'),
  31             'description': self._og_search_description(webpage),
  32             'thumbnail': player_attrs.get('data-player-poster-value'),
  33             'formats': formats,
  34             'subtitles': subtitles,
  35         }