yt_dlp/extractor/lci.py

   1 from .common import InfoExtractor
   2 from .wat import WatIE
   3 from ..utils import ExtractorError, int_or_none
   4 from ..utils.traversal import traverse_obj
   5
   6
   7 class LCIIE(InfoExtractor):
   8     _VALID_URL = r'https?://(?:www\.)?(?:lci|tf1info)\.fr/(?:[^/?#]+/)+[\w-]+-(?P<id>\d+)\.html'
   9     _TESTS = [{
  10         'url': 'https://www.tf1info.fr/replay-lci/videos/video-24h-pujadas-du-vendredi-24-mai-6708-2300831.html',
  11         'info_dict': {
  12             'id': '14113788',
  13             'ext': 'mp4',
  14             'title': '24H Pujadas du vendredi 24 mai 2024',
  15             'thumbnail': 'https://photos.tf1.fr/1280/720/24h-pujadas-du-24-mai-2024-55bf2d-0@1x.jpg',
  16             'upload_date': '20240524',
  17             'duration': 6158,
  18         },
  19         'params': {
  20             'skip_download': True,
  21         },
  22     }, {
  23         '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',
  24         'info_dict': {
  25             'id': '13875948',
  26             'ext': 'mp4',
  27             'title': 'md5:660df5481fd418bc3bbb0d070e6fdb5a',
  28             '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',
  29             'upload_date': '20220422',
  30             'duration': 33,
  31         },
  32         'params': {
  33             'skip_download': True,
  34         },
  35     }, {
  36         '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',
  37         'only_matching': True,
  38     }]
  39
  40     def _real_extract(self, url):
  41         video_id = self._match_id(url)
  42         webpage = self._download_webpage(url, video_id)
  43         next_data = self._search_nextjs_data(webpage, video_id)
  44         wat_id = traverse_obj(next_data, (
  45             'props', 'pageProps', 'page', 'tms', 'videos', {dict.keys}, ..., {int_or_none}, any))
  46         if wat_id is None:
  47             raise ExtractorError('Could not find wat_id')
  48
  49         return self.url_result(f'wat:{wat_id}', WatIE, str(wat_id))