yt_dlp/extractor/cbssports.py

   1 # from .cbs import CBSBaseIE
   2 from .common import InfoExtractor
   3 from ..utils import (
   4     int_or_none,
   5     try_get,
   6 )
   7
   8
   9 # class CBSSportsEmbedIE(CBSBaseIE):
  10 class CBSSportsEmbedIE(InfoExtractor):
  11     IE_NAME = 'cbssports:embed'
  12     _VALID_URL = r'''(?ix)https?://(?:(?:www\.)?cbs|embed\.247)sports\.com/player/embed.+?
  13         (?:
  14             ids%3D(?P<id>[\da-f]{8}-(?:[\da-f]{4}-){3}[\da-f]{12})|
  15             pcid%3D(?P<pcid>\d+)
  16         )'''
  17     _TESTS = [{
  18         'url': 'https://www.cbssports.com/player/embed/?args=player_id%3Db56c03a6-231a-4bbe-9c55-af3c8a8e9636%26ids%3Db56c03a6-231a-4bbe-9c55-af3c8a8e9636%26resizable%3D1%26autoplay%3Dtrue%26domain%3Dcbssports.com%26comp_ads_enabled%3Dfalse%26watchAndRead%3D0%26startTime%3D0%26env%3Dprod',
  19         'only_matching': True,
  20     }, {
  21         'url': 'https://embed.247sports.com/player/embed/?args=%3fplayer_id%3d1827823171591%26channel%3dcollege-football-recruiting%26pcid%3d1827823171591%26width%3d640%26height%3d360%26autoplay%3dTrue%26comp_ads_enabled%3dFalse%26uvpc%3dhttps%253a%252f%252fwww.cbssports.com%252fapi%252fcontent%252fvideo%252fconfig%252f%253fcfg%253duvp_247sports_v4%2526partner%253d247%26uvpc_m%3dhttps%253a%252f%252fwww.cbssports.com%252fapi%252fcontent%252fvideo%252fconfig%252f%253fcfg%253duvp_247sports_m_v4%2526partner_m%253d247_mobile%26utag%3d247sportssite%26resizable%3dTrue',
  22         'only_matching': True,
  23     }]
  24
  25     # def _extract_video_info(self, filter_query, video_id):
  26     #     return self._extract_feed_info('dJ5BDC', 'VxxJg8Ymh8sE', filter_query, video_id)
  27
  28     def _real_extract(self, url):
  29         uuid, pcid = self._match_valid_url(url).groups()
  30         query = {'id': uuid} if uuid else {'pcid': pcid}
  31         video = self._download_json(
  32             'https://www.cbssports.com/api/content/video/',
  33             uuid or pcid, query=query)[0]
  34         video_id = video['id']
  35         title = video['title']
  36         metadata = video.get('metaData') or {}
  37         # return self._extract_video_info('byId=%d' % metadata['mpxOutletId'], video_id)
  38         # return self._extract_video_info('byGuid=' + metadata['mpxRefId'], video_id)
  39
  40         formats = self._extract_m3u8_formats(
  41             metadata['files'][0]['url'], video_id, 'mp4',
  42             'm3u8_native', m3u8_id='hls', fatal=False)
  43         self._sort_formats(formats)
  44
  45         image = video.get('image')
  46         thumbnails = None
  47         if image:
  48             image_path = image.get('path')
  49             if image_path:
  50                 thumbnails = [{
  51                     'url': image_path,
  52                     'width': int_or_none(image.get('width')),
  53                     'height': int_or_none(image.get('height')),
  54                     'filesize': int_or_none(image.get('size')),
  55                 }]
  56
  57         return {
  58             'id': video_id,
  59             'title': title,
  60             'formats': formats,
  61             'thumbnails': thumbnails,
  62             'description': video.get('description'),
  63             'timestamp': int_or_none(try_get(video, lambda x: x['dateCreated']['epoch'])),
  64             'duration': int_or_none(metadata.get('duration')),
  65         }
  66
  67
  68 class CBSSportsBaseIE(InfoExtractor):
  69     def _real_extract(self, url):
  70         display_id = self._match_id(url)
  71         webpage = self._download_webpage(url, display_id)
  72         iframe_url = self._search_regex(
  73             r'<iframe[^>]+(?:data-)?src="(https?://[^/]+/player/embed[^"]+)"',
  74             webpage, 'embed url')
  75         return self.url_result(iframe_url, CBSSportsEmbedIE.ie_key())
  76
  77
  78 class CBSSportsIE(CBSSportsBaseIE):
  79     IE_NAME = 'cbssports'
  80     _VALID_URL = r'https?://(?:www\.)?cbssports\.com/[^/]+/video/(?P<id>[^/?#&]+)'
  81     _TESTS = [{
  82         'url': 'https://www.cbssports.com/college-football/video/cover-3-stanford-spring-gleaning/',
  83         'info_dict': {
  84             'id': 'b56c03a6-231a-4bbe-9c55-af3c8a8e9636',
  85             'ext': 'mp4',
  86             'title': 'Cover 3: Stanford Spring Gleaning',
  87             'description': 'The Cover 3 crew break down everything you need to know about the Stanford Cardinal this spring.',
  88             'timestamp': 1617218398,
  89             'upload_date': '20210331',
  90             'duration': 502,
  91         },
  92     }]
  93
  94
  95 class TwentyFourSevenSportsIE(CBSSportsBaseIE):
  96     IE_NAME = '247sports'
  97     _VALID_URL = r'https?://(?:www\.)?247sports\.com/Video/(?:[^/?#&]+-)?(?P<id>\d+)'
  98     _TESTS = [{
  99         'url': 'https://247sports.com/Video/2021-QB-Jake-Garcia-senior-highlights-through-five-games-10084854/',
 100         'info_dict': {
 101             'id': '4f1265cb-c3b5-44a8-bb1d-1914119a0ccc',
 102             'ext': 'mp4',
 103             'title': '2021 QB Jake Garcia senior highlights through five games',
 104             'description': 'md5:8cb67ebed48e2e6adac1701e0ff6e45b',
 105             'timestamp': 1607114223,
 106             'upload_date': '20201204',
 107             'duration': 208,
 108         },
 109     }]