yt_dlp/extractor/ketnet.py

   1 from .canvas import CanvasIE
   2 from .common import InfoExtractor
   3 from ..compat import compat_urllib_parse_unquote
   4 from ..utils import (
   5     int_or_none,
   6     parse_iso8601,
   7 )
   8
   9
  10 class KetnetIE(InfoExtractor):
  11     _VALID_URL = r'https?://(?:www\.)?ketnet\.be/(?P<id>(?:[^/]+/)*[^/?#&]+)'
  12     _TESTS = [{
  13         'url': 'https://www.ketnet.be/kijken/n/nachtwacht/3/nachtwacht-s3a1-de-greystook',
  14         'md5': '37b2b7bb9b3dcaa05b67058dc3a714a9',
  15         'info_dict': {
  16             'id': 'pbs-pub-aef8b526-115e-4006-aa24-e59ff6c6ef6f$vid-ddb815bf-c8e7-467b-8879-6bad7a32cebd',
  17             'ext': 'mp4',
  18             'title': 'Nachtwacht - Reeks 3: Aflevering 1',
  19             'description': 'De Nachtwacht krijgt te maken met een parasiet',
  20             'thumbnail': r're:^https?://.*\.jpg$',
  21             'duration': 1468.02,
  22             'timestamp': 1609225200,
  23             'upload_date': '20201229',
  24             'series': 'Nachtwacht',
  25             'season': 'Reeks 3',
  26             'episode': 'De Greystook',
  27             'episode_number': 1,
  28         },
  29         'expected_warnings': ['is not a supported codec', 'Unknown MIME type'],
  30     }, {
  31         'url': 'https://www.ketnet.be/themas/karrewiet/jaaroverzicht-20200/karrewiet-het-jaar-van-black-mamba',
  32         'only_matching': True,
  33     }]
  34
  35     def _real_extract(self, url):
  36         display_id = self._match_id(url)
  37
  38         video = self._download_json(
  39             'https://senior-bff.ketnet.be/graphql', display_id, query={
  40                 'query': '''{
  41   video(id: "content/ketnet/nl/%s.model.json") {
  42     description
  43     episodeNr
  44     imageUrl
  45     mediaReference
  46     programTitle
  47     publicationDate
  48     seasonTitle
  49     subtitleVideodetail
  50     titleVideodetail
  51   }
  52 }''' % display_id,
  53             })['data']['video']
  54
  55         mz_id = compat_urllib_parse_unquote(video['mediaReference'])
  56
  57         return {
  58             '_type': 'url_transparent',
  59             'id': mz_id,
  60             'title': video['titleVideodetail'],
  61             'url': 'https://mediazone.vrt.be/api/v1/ketnet/assets/' + mz_id,
  62             'thumbnail': video.get('imageUrl'),
  63             'description': video.get('description'),
  64             'timestamp': parse_iso8601(video.get('publicationDate')),
  65             'series': video.get('programTitle'),
  66             'season': video.get('seasonTitle'),
  67             'episode': video.get('subtitleVideodetail'),
  68             'episode_number': int_or_none(video.get('episodeNr')),
  69             'ie_key': CanvasIE.ie_key(),
  70         }