yt_dlp/extractor/dbtv.py

   1 import re
   2
   3 from .common import InfoExtractor
   4
   5
   6 class DBTVIE(InfoExtractor):
   7     _VALID_URL = r'https?://(?:www\.)?dagbladet\.no/video/(?:(?:embed|(?P<display_id>[^/]+))/)?(?P<id>[0-9A-Za-z_-]{11}|[a-zA-Z0-9]{8})'
   8     _TESTS = [{
   9         'url': 'https://www.dagbladet.no/video/PynxJnNWChE/',
  10         'md5': 'b8f850ba1860adbda668d367f9b77699',
  11         'info_dict': {
  12             'id': 'PynxJnNWChE',
  13             'ext': 'mp4',
  14             'title': 'Skulle teste ut fornøyelsespark, men kollegaen var bare opptatt av bikinikroppen',
  15             'description': 'md5:49cc8370e7d66e8a2ef15c3b4631fd3f',
  16             'thumbnail': r're:https?://.*\.jpg',
  17             'upload_date': '20160916',
  18             'duration': 69,
  19             'uploader_id': 'UCk5pvsyZJoYJBd7_oFPTlRQ',
  20             'uploader': 'Dagbladet',
  21         },
  22         'add_ie': ['Youtube']
  23     }, {
  24         'url': 'https://www.dagbladet.no/video/embed/xlGmyIeN9Jo/?autoplay=false',
  25         'only_matching': True,
  26     }, {
  27         'url': 'https://www.dagbladet.no/video/truer-iran-bor-passe-dere/PalfB2Cw',
  28         'only_matching': True,
  29     }]
  30
  31     @staticmethod
  32     def _extract_urls(webpage):
  33         return [url for _, url in re.findall(
  34             r'<iframe[^>]+src=(["\'])((?:https?:)?//(?:www\.)?dagbladet\.no/video/embed/(?:[0-9A-Za-z_-]{11}|[a-zA-Z0-9]{8}).*?)\1',
  35             webpage)]
  36
  37     def _real_extract(self, url):
  38         display_id, video_id = self._match_valid_url(url).groups()
  39         info = {
  40             '_type': 'url_transparent',
  41             'id': video_id,
  42             'display_id': display_id,
  43         }
  44         if len(video_id) == 11:
  45             info.update({
  46                 'url': video_id,
  47                 'ie_key': 'Youtube',
  48             })
  49         else:
  50             info.update({
  51                 'url': 'jwplatform:' + video_id,
  52                 'ie_key': 'JWPlatform',
  53             })
  54         return info