yt_dlp/extractor/fifa.py

   1 from .common import InfoExtractor
   2
   3 from ..utils import (
   4     int_or_none,
   5     traverse_obj,
   6     unified_timestamp,
   7 )
   8
   9
  10 class FifaIE(InfoExtractor):
  11     _VALID_URL = r'https?://www.fifa.com/fifaplus/(?P<locale>\w{2})/watch/([^#?]+/)?(?P<id>\w+)'
  12     _TESTS = [{
  13         'url': 'https://www.fifa.com/fifaplus/en/watch/7on10qPcnyLajDDU3ntg6y',
  14         'info_dict': {
  15             'id': '7on10qPcnyLajDDU3ntg6y',
  16             'title': 'Italy v France | Final | 2006 FIFA World Cup Germany™ | Full Match Replay',
  17             'description': 'md5:f4520d0ee80529c8ba4134a7d692ff8b',
  18             'ext': 'mp4',
  19             'categories': ['FIFA Tournaments', 'Replay'],
  20             'thumbnail': 'https://digitalhub.fifa.com/transform/fa6f0b3e-a2e9-4cf7-9f32-53c57bcb7360/2006_Final_ITA_FRA',
  21             'duration': 8164,
  22         },
  23         'params': {'skip_download': 'm3u8'},
  24     }, {
  25         'url': 'https://www.fifa.com/fifaplus/pt/watch/1cg5r5Qt6Qt12ilkDgb1sV',
  26         'info_dict': {
  27             'id': '1cg5r5Qt6Qt12ilkDgb1sV',
  28             'title': 'Brasil x Alemanha | Semifinais | Copa do Mundo FIFA Brasil 2014 | Compacto',
  29             'description': 'md5:ba4ffcc084802b062beffc3b4c4b19d6',
  30             'ext': 'mp4',
  31             'categories': ['FIFA Tournaments', 'Highlights'],
  32             'thumbnail': 'https://digitalhub.fifa.com/transform/d8fe6f61-276d-4a73-a7fe-6878a35fd082/FIFAPLS_100EXTHL_2014BRAvGER_TMB',
  33             'duration': 901,
  34             'release_timestamp': 1404777600,
  35             'release_date': '20140708',
  36         },
  37         'params': {'skip_download': 'm3u8'},
  38     }, {
  39         'url': 'https://www.fifa.com/fifaplus/fr/watch/3C6gQH9C2DLwzNx7BMRQdp',
  40         'info_dict': {
  41             'id': '3C6gQH9C2DLwzNx7BMRQdp',
  42             'title': 'Le but de Josimar contre le Irlande du Nord | Buts classiques',
  43             'description': 'md5:16f9f789f09960bfe7220fe67af31f34',
  44             'ext': 'mp4',
  45             'categories': ['FIFA Tournaments', 'Goal'],
  46             'duration': 28,
  47             'thumbnail': 'https://digitalhub.fifa.com/transform/f9301391-f8d9-48b5-823e-c093ac5e3e11/CG_MEN_1986_JOSIMAR',
  48         },
  49         'params': {'skip_download': 'm3u8'},
  50     }]
  51
  52     def _real_extract(self, url):
  53         video_id, locale = self._match_valid_url(url).group('id', 'locale')
  54         webpage = self._download_webpage(url, video_id)
  55
  56         preconnect_link = self._search_regex(
  57             r'<link[^>]+rel\s*=\s*"preconnect"[^>]+href\s*=\s*"([^"]+)"', webpage, 'Preconnect Link')
  58
  59         json_data = self._download_json(
  60             f'{preconnect_link}/video/GetVideoPlayerData/{video_id}', video_id,
  61             'Downloading Video Player Data', query={'includeIdents': True, 'locale': locale})
  62
  63         video_details = self._download_json(
  64             f'{preconnect_link}/sections/videoDetails/{video_id}', video_id, 'Downloading Video Details', fatal=False)
  65
  66         preplay_parameters = self._download_json(
  67             f'{preconnect_link}/video/GetVerizonPreplayParameters', video_id, 'Downloading Preplay Parameters', query={
  68                 'entryId': video_id,
  69                 'assetId': json_data['verizonAssetId'],
  70                 'useExternalId': False,
  71                 'requiresToken': json_data['requiresToken'],
  72                 'adConfig': 'fifaplusvideo',
  73                 'prerollAds': True,
  74                 'adVideoId': json_data['externalVerizonAssetId'],
  75                 'preIdentId': json_data['preIdentId'],
  76                 'postIdentId': json_data['postIdentId'],
  77             })
  78
  79         cid = f'{json_data["preIdentId"]},{json_data["verizonAssetId"]},{json_data["postIdentId"]}'
  80         content_data = self._download_json(
  81             f'https://content.uplynk.com/preplay/{cid}/multiple.json', video_id, 'Downloading Content Data', query={
  82                 'v': preplay_parameters['preplayAPIVersion'],
  83                 'tc': preplay_parameters['tokenCheckAlgorithmVersion'],
  84                 'rn': preplay_parameters['randomNumber'],
  85                 'exp': preplay_parameters['tokenExpirationDate'],
  86                 'ct': preplay_parameters['contentType'],
  87                 'cid': cid,
  88                 'mbtracks': preplay_parameters['tracksAssetNumber'],
  89                 'ad': preplay_parameters['adConfiguration'],
  90                 'ad.preroll': int(preplay_parameters['adPreroll']),
  91                 'ad.cmsid': preplay_parameters['adCMSSourceId'],
  92                 'ad.vid': preplay_parameters['adSourceVideoID'],
  93                 'sig': preplay_parameters['signature'],
  94             })
  95
  96         formats, subtitles = self._extract_m3u8_formats_and_subtitles(content_data['playURL'], video_id)
  97         self._sort_formats(formats)
  98
  99         return {
 100             'id': video_id,
 101             'title': json_data.get('title'),
 102             'description': json_data.get('description'),
 103             'duration': int_or_none(json_data.get('duration')),
 104             'release_timestamp': unified_timestamp(video_details.get('dateOfRelease')),
 105             'categories': traverse_obj(video_details, (('videoCategory', 'videoSubcategory'),)),
 106             'thumbnail': traverse_obj(video_details, ('backgroundImage', 'src')),
 107             'formats': formats,
 108             'subtitles': subtitles,
 109         }