yt_dlp/extractor/mediaite.py

   1 from __future__ import unicode_literals
   2
   3
   4 from .common import InfoExtractor
   5
   6
   7 class MediaiteIE(InfoExtractor):
   8     _VALID_URL = r'https?://(?:www\.)?mediaite.com(?!/category)(?:/[\w-]+){2}'
   9     _TESTS = [{
  10         '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/',
  11         'info_dict': {
  12             'id': 'vPHKITzy',
  13             'ext': 'm4a',
  14             'title': 'Bill Burr On NFL And Black Lives Matter',
  15             'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
  16             'thumbnail': 'https://cdn.jwplayer.com/v2/media/vPHKITzy/poster.jpg?width=720',
  17             'duration': 55,
  18             'timestamp': 1631630185,
  19             'upload_date': '20210914',
  20         },
  21         'params': {'skip_download': True}
  22     }, {
  23         'url': 'https://www.mediaite.com/tv/joe-scarborough-goes-off-on-tax-breaks-for-super-wealthy-largest-income-redistribution-scam-in-american-history/',
  24         'info_dict': {
  25             'id': 'eeFcK4Xm',
  26             'ext': 'mp4',
  27             'title': 'Morning Joe-6_16_52 am - 6_21_10 am-2021-09-14.mp4',
  28             'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
  29             'thumbnail': 'https://cdn.jwplayer.com/v2/media/eeFcK4Xm/poster.jpg?width=720',
  30             'duration': 258,
  31             'timestamp': 1631618057,
  32             'upload_date': '20210914',
  33         },
  34         'params': {'skip_download': True}
  35     }, {
  36         'url': 'https://www.mediaite.com/politics/watch-rudy-giuliani-impersonates-queen-elizabeth-calls-mark-milley-an-asshle-in-bizarre-9-11-speech/',
  37         'info_dict': {
  38             'id': 'EiyiXKcr',
  39             'ext': 'mp4',
  40             'title': 'Giuliani 1',
  41             'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
  42             'thumbnail': 'https://cdn.jwplayer.com/v2/media/EiyiXKcr/poster.jpg?width=720',
  43             'duration': 39,
  44             'timestamp': 1631536476,
  45             'upload_date': '20210913',
  46         },
  47         'params': {'skip_download': True}
  48     }, {
  49         'url': 'https://www.mediaite.com/podcasts/clarissa-ward-says-she-decided-to-become-a-journalist-on-9-11/',
  50         'info_dict': {
  51             'id': 'TxavoRTx',
  52             'ext': 'mp4',
  53             'title': 'clarissa-ward-3.mp4',
  54             'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
  55             'thumbnail': 'https://cdn.jwplayer.com/v2/media/TxavoRTx/poster.jpg?width=720',
  56             'duration': 83,
  57             'timestamp': 1631311188,
  58             'upload_date': '20210910',
  59         },
  60         'params': {'skip_download': True}
  61     }, {
  62         'url': 'https://www.mediaite.com/opinion/mainstream-media-ignores-rose-mcgowans-bombshell-allegation-that-newsoms-wife-tried-to-silence-her-on-weinstein/',
  63         'info_dict': {
  64             'id': 'sEIWvKR7',
  65             'ext': 'mp4',
  66             'title': 'KTTV_09-13-2021_05.34.21',
  67             'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
  68             'thumbnail': 'https://cdn.jwplayer.com/v2/media/sEIWvKR7/poster.jpg?width=720',
  69             'duration': 52,
  70             'timestamp': 1631553328,
  71             'upload_date': '20210913',
  72         },
  73         'params': {'skip_download': True}
  74     }, {
  75         '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/',
  76         'info_dict': {
  77             'id': 'nwpt1elX',
  78             'ext': 'mp4',
  79             'title': "CNBC's Jim Cramer Says Nobody Wants to Die Getting Infected by Unvaccinated Coworker 'Even for $22 an Hour'.mp4",
  80             'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
  81             'thumbnail': 'https://cdn.jwplayer.com/v2/media/nwpt1elX/poster.jpg?width=720',
  82             'duration': 60,
  83             'timestamp': 1633014214,
  84             'upload_date': '20210930',
  85         },
  86         'params': {'skip_download': True}
  87     }]
  88
  89     def _real_extract(self, url):
  90         webpage = self._download_webpage(url, None)
  91         id = self._search_regex(r'data-video-id\s?=\s?\"([^\"]+)\"', webpage, 'id')
  92         data_json = self._download_json(f'https://cdn.jwplayer.com/v2/media/{id}', id)
  93         return self._parse_jwplayer_data(data_json)