yt_dlp/extractor/swearnet.py

   1 from .vidyard import VidyardBaseIE
   2 from ..utils import ExtractorError, int_or_none, make_archive_id
   3
   4
   5 class SwearnetEpisodeIE(VidyardBaseIE):
   6     _VALID_URL = r'https?://www\.swearnet\.com/shows/(?P<id>[\w-]+)/seasons/(?P<season_num>\d+)/episodes/(?P<episode_num>\d+)'
   7     _TESTS = [{
   8         'url': 'https://www.swearnet.com/shows/gettin-learnt-with-ricky/seasons/1/episodes/1',
   9         'info_dict': {
  10             'id': 'wicK2EOzjOdxkUXGDIgcPw',
  11             'display_id': '232819',
  12             'ext': 'mp4',
  13             'episode_number': 1,
  14             'episode': 'Episode 1',
  15             'duration': 719,
  16             'description': r're:Are you drunk and high and craving a grilled cheese sandwich.+',
  17             'season': 'Season 1',
  18             'title': 'Episode 1 - Grilled Cheese Sammich',
  19             'season_number': 1,
  20             'thumbnail': 'https://cdn.vidyard.com/thumbnails/custom/0dd74f9b-388a-452e-b570-b407fb64435b_small.jpg',
  21             'tags': ['Getting Learnt with Ricky', 'drunk', 'grilled cheese', 'high'],
  22             '_old_archive_ids': ['swearnetepisode 232819'],
  23         },
  24     }]
  25
  26     def _real_extract(self, url):
  27         slug, season_number, episode_number = self._match_valid_url(url).group('id', 'season_num', 'episode_num')
  28         webpage = self._download_webpage(url, slug)
  29
  30         try:
  31             external_id = self._search_regex(r'externalid\s*=\s*"([^"]+)', webpage, 'externalid')
  32         except ExtractorError:
  33             if 'Upgrade Now' in webpage:
  34                 self.raise_login_required()
  35             raise
  36
  37         info = self._process_video_json(self._fetch_video_json(external_id)['chapters'][0], external_id)
  38         if info.get('display_id'):
  39             info['_old_archive_ids'] = [make_archive_id(self, info['display_id'])]
  40
  41         return {
  42             **info,
  43             'season_number': int_or_none(season_number),
  44             'episode_number': int_or_none(episode_number),
  45         }