yt_dlp/extractor/onefootball.py

   1 from .common import InfoExtractor
   2 from .jwplatform import JWPlatformIE
   3 from ..utils import make_archive_id
   4
   5
   6 class OneFootballIE(InfoExtractor):
   7     _VALID_URL = r'https?://(?:www\.)?onefootball\.com/[a-z]{2}/video/[^/&?#]+-(?P<id>\d+)'
   8
   9     _TESTS = [{
  10         'url': 'https://onefootball.com/en/video/highlights-fc-zuerich-3-3-fc-basel-34012334',
  11         'info_dict': {
  12             'id': 'Y2VtcWAT',
  13             'ext': 'mp4',
  14             'title': 'Highlights: FC Zürich 3-3 FC Basel',
  15             'description': 'md5:33d9855cb790702c4fe42a513700aba8',
  16             'thumbnail': 'https://cdn.jwplayer.com/v2/media/Y2VtcWAT/poster.jpg?width=720',
  17             'timestamp': 1635874895,
  18             'upload_date': '20211102',
  19             'duration': 375.0,
  20             'tags': ['Football', 'Soccer', 'OneFootball'],
  21             '_old_archive_ids': ['onefootball 34012334'],
  22         },
  23         'params': {'skip_download': True},
  24         'expected_warnings': ['Failed to download m3u8 information'],
  25     }, {
  26         'url': 'https://onefootball.com/en/video/klopp-fumes-at-var-decisions-in-west-ham-defeat-34041020',
  27         'info_dict': {
  28             'id': 'leVJrMho',
  29             'ext': 'mp4',
  30             'title': 'Klopp fumes at VAR decisions in West Ham defeat',
  31             'description': 'md5:9c50371095a01ad3f63311c73d8f51a5',
  32             'thumbnail': 'https://cdn.jwplayer.com/v2/media/leVJrMho/poster.jpg?width=720',
  33             'timestamp': 1636315232,
  34             'upload_date': '20211107',
  35             'duration': 93.0,
  36             'tags': ['Football', 'Soccer', 'OneFootball'],
  37             '_old_archive_ids': ['onefootball 34041020'],
  38         },
  39         'params': {'skip_download': True},
  40     }]
  41
  42     def _real_extract(self, url):
  43         video_id = self._match_id(url)
  44         webpage = self._download_webpage(url, video_id)
  45         data_json = self._search_json_ld(webpage, video_id, fatal=False)
  46         data_json.pop('url', None)
  47         m3u8_url = self._html_search_regex(r'(https://cdn\.jwplayer\.com/manifests/\w+\.m3u8)', webpage, 'm3u8_url')
  48
  49         return self.url_result(
  50             m3u8_url, JWPlatformIE, video_id, _old_archive_ids=[make_archive_id(self, video_id)],
  51             **data_json, url_transparent=True)