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