yt_dlp/extractor/lci.py

   1 from .common import InfoExtractor
   2
   3
   4 class LCIIE(InfoExtractor):
   5     _VALID_URL = r'https?://(?:www\.)?(?:lci|tf1info)\.fr/[^/]+/[\w-]+-(?P<id>\d+)\.html'
   6     _TESTS = [{
   7         'url': 'https://www.tf1info.fr/politique/election-presidentielle-2022-second-tour-j-2-marine-le-pen-et-emmanuel-macron-en-interview-de-lci-vendredi-soir-2217486.html',
   8         'info_dict': {
   9             'id': '13875948',
  10             'ext': 'mp4',
  11             'title': 'md5:660df5481fd418bc3bbb0d070e6fdb5a',
  12             'thumbnail': 'https://photos.tf1.fr/1280/720/presidentielle-2022-marine-le-pen-et-emmanuel-macron-invites-de-lci-ce-vendredi-9c0e73-e1a036-0@1x.jpg',
  13             'upload_date': '20220422',
  14             'duration': 33,
  15         },
  16         'params': {
  17             'skip_download': True,
  18         },
  19     }, {
  20         'url': 'https://www.lci.fr/politique/election-presidentielle-2022-second-tour-j-2-marine-le-pen-et-emmanuel-macron-en-interview-de-lci-vendredi-soir-2217486.html',
  21         'only_matching': True,
  22     }]
  23
  24     def _real_extract(self, url):
  25         video_id = self._match_id(url)
  26         webpage = self._download_webpage(url, video_id)
  27         wat_id = self._search_regex(r'watId["\']?\s*:\s*["\']?(\d+)', webpage, 'wat id')
  28         return self.url_result('wat:' + wat_id, 'Wat', wat_id)