yt_dlp/extractor/digg.py

   1 from .common import InfoExtractor
   2 from ..utils import js_to_json
   3
   4
   5 class DiggIE(InfoExtractor):
   6     _VALID_URL = r'https?://(?:www\.)?digg\.com/video/(?P<id>[^/?#&]+)'
   7     _TESTS = [{
   8         # JWPlatform via provider
   9         'url': 'http://digg.com/video/sci-fi-short-jonah-daniel-kaluuya-get-out',
  10         'info_dict': {
  11             'id': 'LcqvmS0b',
  12             'ext': 'mp4',
  13             'title': "'Get Out' Star Daniel Kaluuya Goes On 'Moby Dick'-Like Journey In Sci-Fi Short 'Jonah'",
  14             'description': 'md5:541bb847648b6ee3d6514bc84b82efda',
  15             'upload_date': '20180109',
  16             'timestamp': 1515530551,
  17         },
  18         'params': {
  19             'skip_download': True,
  20         },
  21     }, {
  22         # Youtube via provider
  23         'url': 'http://digg.com/video/dog-boat-seal-play',
  24         'only_matching': True,
  25     }, {
  26         # vimeo as regular embed
  27         'url': 'http://digg.com/video/dream-girl-short-film',
  28         'only_matching': True,
  29     }]
  30
  31     def _real_extract(self, url):
  32         display_id = self._match_id(url)
  33
  34         webpage = self._download_webpage(url, display_id)
  35
  36         info = self._parse_json(
  37             self._search_regex(
  38                 r'(?s)video_info\s*=\s*({.+?});\n', webpage, 'video info',
  39                 default='{}'), display_id, transform_source=js_to_json,
  40             fatal=False)
  41
  42         video_id = info.get('video_id')
  43
  44         if video_id:
  45             provider = info.get('provider_name')
  46             if provider == 'youtube':
  47                 return self.url_result(
  48                     video_id, ie='Youtube', video_id=video_id)
  49             elif provider == 'jwplayer':
  50                 return self.url_result(
  51                     'jwplatform:%s' % video_id, ie='JWPlatform',
  52                     video_id=video_id)
  53
  54         return self.url_result(url, 'Generic')