yt_dlp/extractor/eurosport.py

   1 from .common import InfoExtractor
   2 from ..utils import traverse_obj
   3
   4
   5 class EurosportIE(InfoExtractor):
   6     _VALID_URL = r'''(?x)
   7         https?://(?:
   8             (?:(?:www|espanol)\.)?eurosport\.(?:com(?:\.tr)?|de|dk|es|fr|hu|it|nl|no|ro)|
   9             eurosport\.tvn24\.pl
  10         )/[\w-]+/(?:[\w-]+/[\d-]+/)?[\w.-]+_(?P<id>vid\d+)
  11     '''
  12     _TESTS = [{
  13         'url': 'https://www.eurosport.com/tennis/roland-garros/2022/highlights-rafael-nadal-brushes-aside-caper-ruud-to-win-record-extending-14th-french-open-title_vid1694147/video.shtml',
  14         'info_dict': {
  15             'id': '2480939',
  16             'ext': 'mp4',
  17             'title': 'Highlights: Rafael Nadal brushes aside Caper Ruud to win record-extending 14th French Open title',
  18             'description': 'md5:b564db73ecfe4b14ebbd8e62a3692c76',
  19             'thumbnail': 'https://imgresizer.eurosport.com/unsafe/1280x960/smart/filters:format(jpeg)/origin-imgresizer.eurosport.com/2022/06/05/3388285-69245968-2560-1440.png',
  20             'duration': 195.0,
  21             'display_id': 'vid1694147',
  22             'timestamp': 1654446698,
  23             'upload_date': '20220605',
  24         },
  25     }, {
  26         'url': 'https://www.eurosport.com/tennis/roland-garros/2022/watch-the-top-five-shots-from-men-s-final-as-rafael-nadal-beats-casper-ruud-to-seal-14th-french-open_vid1694283/video.shtml',
  27         'info_dict': {
  28             'id': '2481254',
  29             'ext': 'mp4',
  30             'title': 'md5:149dcc5dfb38ab7352acc008cc9fb071',
  31             'duration': 130.0,
  32             'thumbnail': 'https://imgresizer.eurosport.com/unsafe/1280x960/smart/filters:format(jpeg)/origin-imgresizer.eurosport.com/2022/06/05/3388422-69248708-2560-1440.png',
  33             'description': 'md5:a0c8a7f6b285e48ae8ddbe7aa85cfee6',
  34             'display_id': 'vid1694283',
  35             'timestamp': 1654456090,
  36             'upload_date': '20220605',
  37         },
  38     }, {
  39         # geo-fence but can bypassed by xff
  40         'url': 'https://www.eurosport.com/cycling/tour-de-france-femmes/2022/incredible-ride-marlen-reusser-storms-to-stage-4-win-at-tour-de-france-femmes_vid1722221/video.shtml',
  41         'info_dict': {
  42             'id': '2582552',
  43             'ext': 'mp4',
  44             'title': '‘Incredible ride!’ - Marlen Reusser storms to Stage 4 win at Tour de France Femmes',
  45             'duration': 188.0,
  46             'display_id': 'vid1722221',
  47             'timestamp': 1658936167,
  48             'thumbnail': 'https://imgresizer.eurosport.com/unsafe/1280x960/smart/filters:format(jpeg)/origin-imgresizer.eurosport.com/2022/07/27/3423347-69852108-2560-1440.jpg',
  49             'description': 'md5:32bbe3a773ac132c57fb1e8cca4b7c71',
  50             'upload_date': '20220727',
  51         },
  52     }, {
  53         'url': 'https://www.eurosport.com/football/champions-league/2022-2023/pep-guardiola-emotionally-destroyed-after-manchester-city-win-over-bayern-munich-in-champions-league_vid1896254/video.shtml',
  54         'info_dict': {
  55             'id': '3096477',
  56             'ext': 'mp4',
  57             'title': 'md5:82edc17370124c7a19b3cf518517583b',
  58             'duration': 84.0,
  59             'description': 'md5:b3f44ef7f5b5b95b24a273b163083feb',
  60             'thumbnail': 'https://imgresizer.eurosport.com/unsafe/1280x960/smart/filters:format(jpeg)/origin-imgresizer.eurosport.com/2023/04/12/3682873-74947393-2560-1440.jpg',
  61             'timestamp': 1681292028,
  62             'upload_date': '20230412',
  63             'display_id': 'vid1896254',
  64         },
  65     }, {
  66         'url': 'https://www.eurosport.com/football/last-year-s-semi-final-pain-was-still-there-pep-guardiola-after-man-city-reach-cl-final_vid1914115/video.shtml',
  67         'info_dict': {
  68             'id': '3149108',
  69             'ext': 'mp4',
  70             'title': '\'Last year\'s semi-final pain was still there\' - Pep Guardiola after Man City reach CL final',
  71             'description': 'md5:89ef142fe0170a66abab77fac2955d8e',
  72             'display_id': 'vid1914115',
  73             'timestamp': 1684403618,
  74             'thumbnail': 'https://imgresizer.eurosport.com/unsafe/1280x960/smart/filters:format(jpeg)/origin-imgresizer.eurosport.com/2023/05/18/3707254-75435008-2560-1440.jpg',
  75             'duration': 105.0,
  76             'upload_date': '20230518',
  77         },
  78     }, {
  79         'url': 'https://www.eurosport.de/radsport/vuelta-a-espana/2024/vuelta-a-espana-2024-wout-van-aert-und-co.-verzweifeln-an-mcnulty-zeitfahr-krimi-in-lissabon_vid2219478/video.shtml',
  80         'only_matching': True,
  81     }, {
  82         'url': 'https://www.eurosport.dk/speedway/mikkel-michelsen-misser-finalen-i-cardiff-se-danskeren-i-semifinalen-her_vid2219363/video.shtml',
  83         'only_matching': True,
  84     }, {
  85         'url': 'https://www.eurosport.nl/mixed-martial-arts/ufc/2022/ufc-305-respect-tussen-adesanya-en-du-plessis_vid2219650/video.shtml',
  86         'only_matching': True,
  87     }, {
  88         'url': 'https://www.eurosport.es/ciclismo/la-vuelta-2024-carlos-rodriguez-olvida-la-crono-y-ya-espera-que-llegue-la-montana-no-me-encontre-nada-comodo_vid2219682/video.shtml',
  89         'only_matching': True,
  90     }, {
  91         'url': 'https://www.eurosport.fr/football/supercoupe-d-europe/2024-2025/kylian-mbappe-vinicius-junior-eduardo-camavinga-touche.-extraits-de-l-entrainement-du-real-madrid-en-video_vid2216993/video.shtml',
  92         'only_matching': True,
  93     }, {
  94         'url': 'https://www.eurosport.it/calcio/serie-a/2024-2025/samardzic-a-bergamo-per-le-visite-mediche-con-l-atalanta_vid2219680/video.shtml',
  95         'only_matching': True,
  96     }, {
  97         'url': 'https://www.eurosport.hu/kerekpar/vuelta-a-espana/2024/dramai-harc-a-masodpercekert-meglepetesgyoztes-a-vuelta-nyitoszakaszan_vid2219481/video.shtml',
  98         'only_matching': True,
  99     }, {
 100         'url': 'https://www.eurosport.no/golf/fedex-st-jude-championship/2024/ligger-pa-andreplass-sa-skjer-dette-drama_vid30000618/video.shtml',
 101         'only_matching': True,
 102     }, {
 103         'url': 'https://www.eurosport.no/golf/fedex-st-jude-championship/2024/ligger-pa-andreplass-sa-skjer-dette-drama_vid2219531/video.shtml',
 104         'only_matching': True,
 105     }, {
 106         'url': 'https://www.eurosport.ro/tenis/western-southern-open-2/2024/rezumatul-partidei-dintre-zverev-si-shelton-de-la-cincinnati_vid2219657/video.shtml',
 107         'only_matching': True,
 108     }, {
 109         'url': 'https://www.eurosport.com.tr/hentbol/olympic-games-paris-2024/2024/paris-2024-denmark-ile-germany-olimpiyatlarin-onemli-anlari_vid2215836/video.shtml',
 110         'only_matching': True,
 111     }, {
 112         'url': 'https://eurosport.tvn24.pl/kolarstwo/tour-de-france-kobiet/2024/kasia-niewiadoma-przed-ostatnim-8.-etapem-tour-de-france-kobiet_vid2219765/video.shtml',
 113         'only_matching': True,
 114     }]
 115
 116     _TOKEN = None
 117
 118     # actually defined in https://netsport.eurosport.io/?variables={"databaseId":<databaseId>,"playoutType":"VDP"}&extensions={"persistedQuery":{"version":1 ..
 119     # but this method require to get sha256 hash
 120     _GEO_COUNTRIES = ['DE', 'NL', 'EU', 'IT', 'FR']  # Not complete list but it should work
 121     _GEO_BYPASS = False
 122
 123     def _real_initialize(self):
 124         if EurosportIE._TOKEN is None:
 125             EurosportIE._TOKEN = self._download_json(
 126                 'https://eu3-prod-direct.eurosport.com/token?realm=eurosport', None,
 127                 'Trying to get token')['data']['attributes']['token']
 128
 129     def _real_extract(self, url):
 130         display_id = self._match_id(url)
 131         webpage = self._download_webpage(url, display_id)
 132
 133         json_data = self._download_json(
 134             f'https://eu3-prod-direct.eurosport.com/playback/v2/videoPlaybackInfo/sourceSystemId/eurosport-{display_id}',
 135             display_id, query={'usePreAuth': True}, headers={'Authorization': f'Bearer {EurosportIE._TOKEN}'})['data']
 136
 137         json_ld_data = self._search_json_ld(webpage, display_id)
 138
 139         formats, subtitles = [], {}
 140         for stream_type in json_data['attributes']['streaming']:
 141             if stream_type == 'hls':
 142                 fmts, subs = self._extract_m3u8_formats_and_subtitles(
 143                     traverse_obj(json_data, ('attributes', 'streaming', stream_type, 'url')), display_id, ext='mp4', fatal=False)
 144             elif stream_type == 'dash':
 145                 fmts, subs = self._extract_mpd_formats_and_subtitles(
 146                     traverse_obj(json_data, ('attributes', 'streaming', stream_type, 'url')), display_id, fatal=False)
 147             elif stream_type == 'mss':
 148                 fmts, subs = self._extract_ism_formats_and_subtitles(
 149                     traverse_obj(json_data, ('attributes', 'streaming', stream_type, 'url')), display_id, fatal=False)
 150
 151             formats.extend(fmts)
 152             self._merge_subtitles(subs, target=subtitles)
 153
 154         return {
 155             'id': json_data['id'],
 156             'title': json_ld_data.get('title') or self._og_search_title(webpage),
 157             'display_id': display_id,
 158             'formats': formats,
 159             'subtitles': subtitles,
 160             'thumbnails': json_ld_data.get('thumbnails'),
 161             'description': (json_ld_data.get('description')
 162                             or self._html_search_meta(['og:description', 'description'], webpage)),
 163             'duration': json_ld_data.get('duration'),
 164             'timestamp': json_ld_data.get('timestamp'),
 165         }