yt_dlp/extractor/magentamusik.py

   1 from .common import InfoExtractor
   2 from ..utils import ExtractorError, int_or_none, join_nonempty, url_or_none
   3 from ..utils.traversal import traverse_obj
   4
   5
   6 class MagentaMusikIE(InfoExtractor):
   7     _VALID_URL = r'https?://(?:www\.)?magentamusik\.de/(?P<id>[^/?#]+)'
   8
   9     _TESTS = [{
  10         'url': 'https://www.magentamusik.de/marty-friedman-woa-2023-9208205928595409235',
  11         'md5': 'd82dd4748f55fc91957094546aaf8584',
  12         'info_dict': {
  13             'id': '9208205928595409235',
  14             'display_id': 'marty-friedman-woa-2023-9208205928595409235',
  15             'ext': 'mp4',
  16             'title': 'Marty Friedman: W:O:A 2023',
  17             'alt_title': 'Konzert vom: 05.08.2023 13:00',
  18             'duration': 2760,
  19             'categories': ['Musikkonzert'],
  20             'release_year': 2023,
  21             'location': 'Deutschland',
  22         }
  23     }]
  24
  25     def _real_extract(self, url):
  26         display_id = self._match_id(url)
  27         webpage = self._download_webpage(url, display_id)
  28         player_config = self._search_json(
  29             r'data-js-element="o-video-player__config">', webpage, 'player config', display_id, fatal=False)
  30         if not player_config:
  31             raise ExtractorError('No video found', expected=True)
  32
  33         asset_id = player_config['assetId']
  34         asset_details = self._download_json(
  35             f'https://wcps.t-online.de/cvss/magentamusic/vodclient/v2/assetdetails/58938/{asset_id}',
  36             display_id, note='Downloading asset details')
  37
  38         video_id = traverse_obj(
  39             asset_details, ('content', 'partnerInformation', ..., 'reference', {str}), get_all=False)
  40         if not video_id:
  41             raise ExtractorError('Unable to extract video id')
  42
  43         vod_data = self._download_json(
  44             f'https://wcps.t-online.de/cvss/magentamusic/vodclient/v2/player/58935/{video_id}/Main%20Movie', video_id)
  45         smil_url = traverse_obj(
  46             vod_data, ('content', 'feature', 'representations', ...,
  47                        'contentPackages', ..., 'media', 'href', {url_or_none}), get_all=False)
  48
  49         return {
  50             'id': video_id,
  51             'display_id': display_id,
  52             'formats': self._extract_smil_formats(smil_url, video_id),
  53             **traverse_obj(vod_data, ('content', 'feature', 'metadata', {
  54                 'title': 'title',
  55                 'alt_title': 'originalTitle',
  56                 'description': 'longDescription',
  57                 'duration': ('runtimeInSeconds', {int_or_none}),
  58                 'location': ('countriesOfProduction', {list}, {lambda x: join_nonempty(*x, delim=', ')}),
  59                 'release_year': ('yearOfProduction', {int_or_none}),
  60                 'categories': ('mainGenre', {str}, {lambda x: x and [x]}),
  61             })),
  62         }