yt_dlp/extractor/cultureunplugged.py

   1 from .common import InfoExtractor
   2 from ..utils import int_or_none
   3
   4
   5 class CultureUnpluggedIE(InfoExtractor):
   6     _VALID_URL = r'https?://(?:www\.)?cultureunplugged\.com/documentary/watch-online/play/(?P<id>\d+)(?:/(?P<display_id>[^/]+))?'
   7     _TESTS = [{
   8         'url': 'http://www.cultureunplugged.com/documentary/watch-online/play/53662/The-Next--Best-West',
   9         'md5': 'ac6c093b089f7d05e79934dcb3d228fc',
  10         'info_dict': {
  11             'id': '53662',
  12             'display_id': 'The-Next--Best-West',
  13             'ext': 'mp4',
  14             'title': 'The Next, Best West',
  15             'description': 'md5:0423cd00833dea1519cf014e9d0903b1',
  16             'thumbnail': r're:^https?://.*\.jpg$',
  17             'creator': 'Coldstream Creative',
  18             'duration': 2203,
  19             'view_count': int,
  20         },
  21     }, {
  22         'url': 'http://www.cultureunplugged.com/documentary/watch-online/play/53662',
  23         'only_matching': True,
  24     }]
  25
  26     def _real_extract(self, url):
  27         mobj = self._match_valid_url(url)
  28         video_id = mobj.group('id')
  29         display_id = mobj.group('display_id') or video_id
  30
  31         movie_data = self._download_json(
  32             f'http://www.cultureunplugged.com/movie-data/cu-{video_id}.json', display_id)
  33
  34         video_url = movie_data['url']
  35         title = movie_data['title']
  36
  37         description = movie_data.get('synopsis')
  38         creator = movie_data.get('producer')
  39         duration = int_or_none(movie_data.get('duration'))
  40         view_count = int_or_none(movie_data.get('views'))
  41
  42         thumbnails = [{
  43             'url': movie_data[f'{size}_thumb'],
  44             'id': size,
  45             'preference': preference,
  46         } for preference, size in enumerate((
  47             'small', 'large')) if movie_data.get(f'{size}_thumb')]
  48
  49         return {
  50             'id': video_id,
  51             'display_id': display_id,
  52             'url': video_url,
  53             'title': title,
  54             'description': description,
  55             'creator': creator,
  56             'duration': duration,
  57             'view_count': view_count,
  58             'thumbnails': thumbnails,
  59         }