youtube_dl/extractor/ketnet.py

   1 from __future__ import unicode_literals
   2
   3 from .canvas import CanvasIE
   4 from .common import InfoExtractor
   5 from ..compat import compat_urllib_parse_unquote
   6 from ..utils import (
   7     int_or_none,
   8     parse_iso8601,
   9 )
  10
  11
  12 class KetnetIE(InfoExtractor):
  13     _VALID_URL = r'https?://(?:www\.)?ketnet\.be/(?P<id>(?:[^/]+/)*[^/?#&]+)'
  14     _TESTS = [{
  15         'url': 'https://www.ketnet.be/kijken/n/nachtwacht/3/nachtwacht-s3a1-de-greystook',
  16         'md5': '37b2b7bb9b3dcaa05b67058dc3a714a9',
  17         'info_dict': {
  18             'id': 'pbs-pub-aef8b526-115e-4006-aa24-e59ff6c6ef6f$vid-ddb815bf-c8e7-467b-8879-6bad7a32cebd',
  19             'ext': 'mp4',
  20             'title': 'Nachtwacht - Reeks 3: Aflevering 1',
  21             'description': 'De Nachtwacht krijgt te maken met een parasiet',
  22             'thumbnail': r're:^https?://.*\.jpg$',
  23             'duration': 1468.02,
  24             'timestamp': 1609225200,
  25             'upload_date': '20201229',
  26             'series': 'Nachtwacht',
  27             'season': 'Reeks 3',
  28             'episode': 'De Greystook',
  29             'episode_number': 1,
  30         },
  31         'expected_warnings': ['is not a supported codec', 'Unknown MIME type'],
  32     }, {
  33         'url': 'https://www.ketnet.be/themas/karrewiet/jaaroverzicht-20200/karrewiet-het-jaar-van-black-mamba',
  34         'only_matching': True,
  35     }]
  36
  37     def _real_extract(self, url):
  38         display_id = self._match_id(url)
  39
  40         video = self._download_json(
  41             'https://senior-bff.ketnet.be/graphql', display_id, query={
  42                 'query': '''{
  43   video(id: "content/ketnet/nl/%s.model.json") {
  44     description
  45     episodeNr
  46     imageUrl
  47     mediaReference
  48     programTitle
  49     publicationDate
  50     seasonTitle
  51     subtitleVideodetail
  52     titleVideodetail
  53   }
  54 }''' % display_id,
  55             })['data']['video']
  56
  57         mz_id = compat_urllib_parse_unquote(video['mediaReference'])
  58
  59         return {
  60             '_type': 'url_transparent',
  61             'id': mz_id,
  62             'title': video['titleVideodetail'],
  63             'url': 'https://mediazone.vrt.be/api/v1/ketnet/assets/' + mz_id,
  64             'thumbnail': video.get('imageUrl'),
  65             'description': video.get('description'),
  66             'timestamp': parse_iso8601(video.get('publicationDate')),
  67             'series': video.get('programTitle'),
  68             'season': video.get('seasonTitle'),
  69             'episode': video.get('subtitleVideodetail'),
  70             'episode_number': int_or_none(video.get('episodeNr')),
  71             'ie_key': CanvasIE.ie_key(),
  72         }