yt_dlp/extractor/drtalks.py

   1 from .brightcove import BrightcoveNewIE
   2 from .common import InfoExtractor
   3 from ..utils import url_or_none
   4 from ..utils.traversal import traverse_obj
   5
   6
   7 class DrTalksIE(InfoExtractor):
   8     _VALID_URL = r'https?://(?:www\.)?drtalks\.com/videos/(?P<id>[\w-]+)'
   9     _TESTS = [{
  10         'url': 'https://drtalks.com/videos/six-pillars-of-resilience-tools-for-managing-stress-and-flourishing/',
  11         'info_dict': {
  12             'id': '6366193757112',
  13             'ext': 'mp4',
  14             'uploader_id': '6314452011001',
  15             'tags': ['resilience'],
  16             'description': 'md5:9c6805aee237ee6de8052461855b9dda',
  17             'timestamp': 1734546659,
  18             'thumbnail': 'https://drtalks.com/wp-content/uploads/2024/12/Episode-82-Eva-Selhub-DrTalks-Thumbs.jpg',
  19             'title': 'Six Pillars of Resilience: Tools for Managing Stress and Flourishing',
  20             'duration': 2800.682,
  21             'upload_date': '20241218',
  22         },
  23     }, {
  24         'url': 'https://drtalks.com/videos/the-pcos-puzzle-mastering-metabolic-health-with-marcelle-pick/',
  25         'info_dict': {
  26             'id': '6364699891112',
  27             'ext': 'mp4',
  28             'title': 'The PCOS Puzzle: Mastering Metabolic Health with Marcelle Pick',
  29             'description': 'md5:e87cbe00ca50135d5702787fc4043aaa',
  30             'thumbnail': 'https://drtalks.com/wp-content/uploads/2024/11/Episode-34-Marcelle-Pick-OBGYN-NP-DrTalks.jpg',
  31             'duration': 3515.2,
  32             'tags': ['pcos'],
  33             'upload_date': '20241114',
  34             'timestamp': 1731592119,
  35             'uploader_id': '6314452011001',
  36         },
  37     }]
  38
  39     def _real_extract(self, url):
  40         video_id = self._match_id(url)
  41         webpage = self._download_webpage(url, video_id)
  42         next_data = self._search_nextjs_data(webpage, video_id)['props']['pageProps']['data']['video']
  43
  44         return self.url_result(
  45             next_data['videos']['brightcoveVideoLink'], BrightcoveNewIE, video_id,
  46             url_transparent=True,
  47             **traverse_obj(next_data, {
  48                 'title': ('title', {str}),
  49                 'description': ('videos', 'summury', {str}),
  50                 'thumbnail': ('featuredImage', 'node', 'sourceUrl', {url_or_none}),
  51             }))