yt_dlp/extractor/mediaite.py

   1 from .common import InfoExtractor
   2
   3
   4 class MediaiteIE(InfoExtractor):
   5     _VALID_URL = r'https?://(?:www\.)?mediaite\.com(?!/category)(?:/[\w-]+){2}'
   6     _TESTS = [{
   7         'url': 'https://www.mediaite.com/sports/bill-burr-roasts-nfl-for-promoting-black-lives-matter-while-scheduling-more-games-after-all-the-sht-they-know-about-cte/',
   8         'info_dict': {
   9             'id': 'vPHKITzy',
  10             'ext': 'm4a',
  11             'title': 'Bill Burr On NFL And Black Lives Matter',
  12             'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
  13             'thumbnail': 'https://cdn.jwplayer.com/v2/media/vPHKITzy/poster.jpg?width=720',
  14             'duration': 55,
  15             'timestamp': 1631630185,
  16             'upload_date': '20210914',
  17         },
  18         'params': {'skip_download': True},
  19     }, {
  20         'url': 'https://www.mediaite.com/tv/joe-scarborough-goes-off-on-tax-breaks-for-super-wealthy-largest-income-redistribution-scam-in-american-history/',
  21         'info_dict': {
  22             'id': 'eeFcK4Xm',
  23             'ext': 'mp4',
  24             'title': 'Morning Joe-6_16_52 am - 6_21_10 am-2021-09-14.mp4',
  25             'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
  26             'thumbnail': 'https://cdn.jwplayer.com/v2/media/eeFcK4Xm/poster.jpg?width=720',
  27             'duration': 258,
  28             'timestamp': 1631618057,
  29             'upload_date': '20210914',
  30         },
  31         'params': {'skip_download': True},
  32     }, {
  33         'url': 'https://www.mediaite.com/politics/watch-rudy-giuliani-impersonates-queen-elizabeth-calls-mark-milley-an-asshle-in-bizarre-9-11-speech/',
  34         'info_dict': {
  35             'id': 'EiyiXKcr',
  36             'ext': 'mp4',
  37             'title': 'Giuliani 1',
  38             'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
  39             'thumbnail': 'https://cdn.jwplayer.com/v2/media/EiyiXKcr/poster.jpg?width=720',
  40             'duration': 39,
  41             'timestamp': 1631536476,
  42             'upload_date': '20210913',
  43         },
  44         'params': {'skip_download': True},
  45     }, {
  46         'url': 'https://www.mediaite.com/podcasts/clarissa-ward-says-she-decided-to-become-a-journalist-on-9-11/',
  47         'info_dict': {
  48             'id': 'TxavoRTx',
  49             'ext': 'mp4',
  50             'title': 'clarissa-ward-3.mp4',
  51             'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
  52             'thumbnail': 'https://cdn.jwplayer.com/v2/media/TxavoRTx/poster.jpg?width=720',
  53             'duration': 83,
  54             'timestamp': 1631311188,
  55             'upload_date': '20210910',
  56         },
  57         'params': {'skip_download': True},
  58     }, {
  59         'url': 'https://www.mediaite.com/opinion/mainstream-media-ignores-rose-mcgowans-bombshell-allegation-that-newsoms-wife-tried-to-silence-her-on-weinstein/',
  60         'info_dict': {
  61             'id': 'sEIWvKR7',
  62             'ext': 'mp4',
  63             'title': 'KTTV_09-13-2021_05.34.21',
  64             'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
  65             'thumbnail': 'https://cdn.jwplayer.com/v2/media/sEIWvKR7/poster.jpg?width=720',
  66             'duration': 52,
  67             'timestamp': 1631553328,
  68             'upload_date': '20210913',
  69         },
  70         'params': {'skip_download': True},
  71     }, {
  72         'url': 'https://www.mediaite.com/news/watch-cnbcs-jim-cramer-says-nobody-wants-to-die-getting-infected-by-unvaccinated-coworker-even-for-22-an-hour/',
  73         'info_dict': {
  74             'id': 'nwpt1elX',
  75             'ext': 'mp4',
  76             'title': "CNBC's Jim Cramer Says Nobody Wants to Die Getting Infected by Unvaccinated Coworker 'Even for $22 an Hour'.mp4",
  77             'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
  78             'thumbnail': 'https://cdn.jwplayer.com/v2/media/nwpt1elX/poster.jpg?width=720',
  79             'duration': 60,
  80             'timestamp': 1633014214,
  81             'upload_date': '20210930',
  82         },
  83         'params': {'skip_download': True},
  84     }, {
  85         'url': 'https://www.mediaite.com/politics/i-cant-read-it-fast-enough-while-defending-trump-larry-kudlow-overwhelmed-by-volume-of-ex-presidents-legal-troubles/',
  86         'info_dict': {
  87             'id': 'E6EhDX5z',
  88             'ext': 'mp4',
  89             'title': 'Fox Business Network - 4:00 PM - 5:00 PM - 1:39:42 pm - 1:42:20 pm',
  90             'description': '',
  91             'thumbnail': 'https://cdn.jwplayer.com/v2/media/E6EhDX5z/poster.jpg?width=720',
  92             'duration': 157,
  93             'timestamp': 1691015535,
  94             'upload_date': '20230802',
  95         },
  96         'params': {'skip_download': True},
  97     }]
  98
  99     def _real_extract(self, url):
 100         webpage = self._download_webpage(url, None)
 101         video_id = self._search_regex(
 102             [r'"https://cdn\.jwplayer\.com/players/(\w+)', r'data-video-id\s*=\s*\"([^\"]+)\"'], webpage, 'id')
 103         data_json = self._download_json(f'https://cdn.jwplayer.com/v2/media/{video_id}', video_id)
 104         return self._parse_jwplayer_data(data_json)