yt_dlp/extractor/tonline.py

   1 from .common import InfoExtractor
   2 from ..utils import int_or_none, join_nonempty
   3
   4
   5 class TOnlineIE(InfoExtractor):
   6     _WORKING = False
   7     _ENABLED = None  # XXX: pass through to GenericIE
   8     IE_NAME = 't-online.de'
   9     _VALID_URL = r'https?://(?:www\.)?t-online\.de/tv/(?:[^/]+/)*id_(?P<id>\d+)'
  10     _TEST = {
  11         'url': 'http://www.t-online.de/tv/sport/fussball/id_79166266/drittes-remis-zidane-es-muss-etwas-passieren-.html',
  12         'md5': '7d94dbdde5f9d77c5accc73c39632c29',
  13         'info_dict': {
  14             'id': '79166266',
  15             'ext': 'mp4',
  16             'title': 'Drittes Remis! Zidane: "Es muss etwas passieren"',
  17             'description': 'Es läuft nicht rund bei Real Madrid. Das 1:1 gegen den SD Eibar war das dritte Unentschieden in Folge in der Liga.',
  18         }
  19     }
  20
  21     def _real_extract(self, url):
  22         video_id = self._match_id(url)
  23         video_data = self._download_json(
  24             'http://www.t-online.de/tv/id_%s/tid_json_video' % video_id, video_id)
  25         title = video_data['subtitle']
  26
  27         formats = []
  28         for asset in video_data.get('assets', []):
  29             asset_source = asset.get('source') or asset.get('source2')
  30             if not asset_source:
  31                 continue
  32             formats.append({
  33                 'format_id': join_nonempty('type', 'profile', from_dict=asset),
  34                 'url': asset_source,
  35             })
  36
  37         thumbnails = []
  38         for image in video_data.get('images', []):
  39             image_source = image.get('source')
  40             if not image_source:
  41                 continue
  42             thumbnails.append({
  43                 'url': image_source,
  44             })
  45
  46         return {
  47             'id': video_id,
  48             'title': title,
  49             'description': video_data.get('description'),
  50             'duration': int_or_none(video_data.get('duration')),
  51             'thumbnails': thumbnails,
  52             'formats': formats,
  53         }