yt_dlp/extractor/vtm.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     int_or_none,
   4     parse_iso8601,
   5     try_get,
   6 )
   7
   8
   9 class VTMIE(InfoExtractor):
  10     _WORKING = False
  11     _VALID_URL = r'https?://(?:www\.)?vtm\.be/([^/?&#]+)~v(?P<id>[0-9a-f]{8}(?:-[0-9a-f]{4}){3}-[0-9a-f]{12})'
  12     _TEST = {
  13         'url': 'https://vtm.be/gast-vernielt-genkse-hotelkamer~ve7534523-279f-4b4d-a5c9-a33ffdbe23e1',
  14         'md5': '37dca85fbc3a33f2de28ceb834b071f8',
  15         'info_dict': {
  16             'id': '192445',
  17             'ext': 'mp4',
  18             'title': 'Gast vernielt Genkse hotelkamer',
  19             'timestamp': 1611060180,
  20             'upload_date': '20210119',
  21             'duration': 74,
  22             # TODO: fix url _type result processing
  23             # 'series': 'Op Interventie',
  24         },
  25     }
  26
  27     def _real_extract(self, url):
  28         uuid = self._match_id(url)
  29         video = self._download_json(
  30             'https://omc4vm23offuhaxx6hekxtzspi.appsync-api.eu-west-1.amazonaws.com/graphql',
  31             uuid, query={
  32                 'query': '''{
  33   getComponent(type: Video, uuid: "%s") {
  34     ... on Video {
  35       description
  36       duration
  37       myChannelsVideo
  38       program {
  39         title
  40       }
  41       publishedAt
  42       title
  43     }
  44   }
  45 }''' % uuid,  # noqa: UP031
  46             }, headers={
  47                 'x-api-key': 'da2-lz2cab4tfnah3mve6wiye4n77e',
  48             })['data']['getComponent']
  49
  50         return {
  51             '_type': 'url',
  52             'id': uuid,
  53             'title': video.get('title'),
  54             'url': 'http://mychannels.video/embed/%d' % video['myChannelsVideo'],
  55             'description': video.get('description'),
  56             'timestamp': parse_iso8601(video.get('publishedAt')),
  57             'duration': int_or_none(video.get('duration')),
  58             'series': try_get(video, lambda x: x['program']['title']),
  59             'ie_key': 'Medialaan',
  60         }