yt_dlp/extractor/sportdeutschland.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     join_nonempty,
   4     strip_or_none,
   5     traverse_obj,
   6     unified_timestamp,
   7 )
   8
   9
  10 class SportDeutschlandIE(InfoExtractor):
  11     _VALID_URL = r'https?://sportdeutschland\.tv/(?P<id>(?:[^/]+/)?[^?#/&]+)'
  12     _TESTS = [{
  13         'url': 'https://sportdeutschland.tv/blauweissbuchholztanzsport/buchholzer-formationswochenende-2023-samstag-1-bundesliga-landesliga',
  14         'info_dict': {
  15             'id': '9839a5c7-0dbb-48a8-ab63-3b408adc7b54',
  16             'ext': 'mp4',
  17             'title': 'Buchholzer Formationswochenende 2023 - Samstag - 1. Bundesliga / Landesliga',
  18             'display_id': 'blauweissbuchholztanzsport/buchholzer-formationswochenende-2023-samstag-1-bundesliga-landesliga',
  19             'description': 'md5:a288c794a5ee69e200d8f12982f81a87',
  20             'live_status': 'was_live',
  21             'channel': 'Blau-Weiss Buchholz Tanzsport',
  22             'channel_url': 'https://sportdeutschland.tv/blauweissbuchholztanzsport',
  23             'channel_id': '93ec33c9-48be-43b6-b404-e016b64fdfa3',
  24             'duration': 32447,
  25             'upload_date': '20230114',
  26             'timestamp': 1673733618,
  27         }
  28     }, {
  29         'url': 'https://sportdeutschland.tv/deutscherbadmintonverband/bwf-tour-1-runde-feld-1-yonex-gainward-german-open-2022-0',
  30         'info_dict': {
  31             'id': '95c80c52-6b9a-4ae9-9197-984145adfced',
  32             'ext': 'mp4',
  33             'title': 'BWF Tour: 1. Runde Feld 1 - YONEX GAINWARD German Open 2022',
  34             'display_id': 'deutscherbadmintonverband/bwf-tour-1-runde-feld-1-yonex-gainward-german-open-2022-0',
  35             'description': 'md5:2afb5996ceb9ac0b2ac81f563d3a883e',
  36             'live_status': 'was_live',
  37             'channel': 'Deutscher Badminton Verband',
  38             'channel_url': 'https://sportdeutschland.tv/deutscherbadmintonverband',
  39             'channel_id': '93ca5866-2551-49fc-8424-6db35af58920',
  40             'duration': 41097,
  41             'upload_date': '20220309',
  42             'timestamp': 1646860727.0,
  43         }
  44     }, {
  45         'url': 'https://sportdeutschland.tv/ggcbremen/formationswochenende-latein-2023',
  46         'info_dict': {
  47             'id': '9889785e-55b0-4d97-a72a-ce9a9f157cce',
  48             'title': 'Formationswochenende Latein 2023 - Samstag',
  49             'display_id': 'ggcbremen/formationswochenende-latein-2023',
  50             'description': 'md5:6e4060d40ff6a8f8eeb471b51a8f08b2',
  51             'live_status': 'was_live',
  52             'channel': 'Grün-Gold-Club Bremen e.V.',
  53             'channel_id': '9888f04e-bb46-4c7f-be47-df960a4167bb',
  54             'channel_url': 'https://sportdeutschland.tv/ggcbremen',
  55         },
  56         'playlist_count': 3,
  57         'playlist': [{
  58             'info_dict': {
  59                 'id': '988e1fea-9d44-4fab-8c72-3085fb667547',
  60                 'ext': 'mp4',
  61                 'channel_url': 'https://sportdeutschland.tv/ggcbremen',
  62                 'channel_id': '9888f04e-bb46-4c7f-be47-df960a4167bb',
  63                 'channel': 'Grün-Gold-Club Bremen e.V.',
  64                 'duration': 86,
  65                 'title': 'Formationswochenende Latein 2023 - Samstag Part 1',
  66                 'upload_date': '20230225',
  67                 'timestamp': 1677349909,
  68                 'live_status': 'was_live',
  69             }
  70         }]
  71     }, {
  72         'url': 'https://sportdeutschland.tv/dtb/gymnastik-international-tag-1',
  73         'info_dict': {
  74             'id': '95d71b8a-370a-4b87-ad16-94680da18528',
  75             'ext': 'mp4',
  76             'title': r're:Gymnastik International - Tag 1 .+',
  77             'display_id': 'dtb/gymnastik-international-tag-1',
  78             'channel_id': '936ecef1-2f4a-4e08-be2f-68073cb7ecab',
  79             'channel': 'Deutscher Turner-Bund',
  80             'channel_url': 'https://sportdeutschland.tv/dtb',
  81             'description': 'md5:07a885dde5838a6f0796ee21dc3b0c52',
  82             'live_status': 'is_live',
  83         },
  84         'skip': 'live',
  85     }]
  86
  87     def _process_video(self, asset_id, video):
  88         is_live = video['type'] == 'mux_live'
  89         token = self._download_json(
  90             f'https://api.sportdeutschland.tv/api/frontend/asset-token/{asset_id}',
  91             video['id'], query={'type': video['type'], 'playback_id': video['src']})['token']
  92         formats, subtitles = self._extract_m3u8_formats_and_subtitles(
  93             f'https://stream.mux.com/{video["src"]}.m3u8?token={token}', video['id'], live=is_live)
  94
  95         return {
  96             'is_live': is_live,
  97             'formats': formats,
  98             'subtitles': subtitles,
  99             **traverse_obj(video, {
 100                 'id': 'id',
 101                 'duration': ('duration', {lambda x: float(x) > 0 and float(x)}),
 102                 'timestamp': ('created_at', {unified_timestamp})
 103             }),
 104         }
 105
 106     def _real_extract(self, url):
 107         display_id = self._match_id(url)
 108         meta = self._download_json(
 109             f'https://api.sportdeutschland.tv/api/stateless/frontend/assets/{display_id}',
 110             display_id, query={'access_token': 'true'})
 111
 112         info = {
 113             'display_id': display_id,
 114             **traverse_obj(meta, {
 115                 'id': (('id', 'uuid'), ),
 116                 'title': (('title', 'name'), {strip_or_none}),
 117                 'description': 'description',
 118                 'channel': ('profile', 'name'),
 119                 'channel_id': ('profile', 'id'),
 120                 'is_live': 'currently_live',
 121                 'was_live': 'was_live',
 122                 'channel_url': ('profile', 'slug', {lambda x: f'https://sportdeutschland.tv/{x}'}),
 123             }, get_all=False)
 124         }
 125
 126         parts = traverse_obj(meta, (('livestream', ('videos', ...)), ))
 127         entries = [{
 128             'title': join_nonempty(info.get('title'), f'Part {i}', delim=' '),
 129             **traverse_obj(info, {'channel': 'channel', 'channel_id': 'channel_id',
 130                                   'channel_url': 'channel_url', 'was_live': 'was_live'}),
 131             **self._process_video(info['id'], video),
 132         } for i, video in enumerate(parts, 1)]
 133         print(entries[0]['duration'])
 134         return {
 135             '_type': 'multi_video',
 136             **info,
 137             'entries': entries,
 138         } if len(entries) > 1 else {
 139             **info,
 140             **entries[0],
 141             'title': info.get('title'),
 142         }