yt_dlp/extractor/crooksandliars.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     int_or_none,
   4     qualities,
   5 )
   6
   7
   8 class CrooksAndLiarsIE(InfoExtractor):
   9     _VALID_URL = r'https?://embed\.crooksandliars\.com/(?:embed|v)/(?P<id>[A-Za-z0-9]+)'
  10     _EMBED_REGEX = [r'<(?:iframe[^>]+src|param[^>]+value)=(["\'])(?P<url>(?:https?:)?//embed\.crooksandliars\.com/(?:embed|v)/.+?)\1']
  11
  12     _TESTS = [{
  13         'url': 'https://embed.crooksandliars.com/embed/8RUoRhRi',
  14         'info_dict': {
  15             'id': '8RUoRhRi',
  16             'ext': 'mp4',
  17             'title': 'Fox & Friends Says Protecting Atheists From Discrimination Is Anti-Christian!',
  18             'description': 'md5:e1a46ad1650e3a5ec7196d432799127f',
  19             'thumbnail': r're:^https?://.*\.jpg',
  20             'timestamp': 1428207000,
  21             'upload_date': '20150405',
  22             'uploader': 'Heather',
  23             'duration': 236,
  24         },
  25     }, {
  26         'url': 'http://embed.crooksandliars.com/v/MTE3MjUtMzQ2MzA',
  27         'only_matching': True,
  28     }]
  29
  30     def _real_extract(self, url):
  31         video_id = self._match_id(url)
  32
  33         webpage = self._download_webpage(
  34             f'http://embed.crooksandliars.com/embed/{video_id}', video_id)
  35
  36         manifest = self._search_json(r'var\s+manifest\s*=', webpage, 'manifest JSON', video_id)
  37
  38         quality = qualities(('webm_low', 'mp4_low', 'webm_high', 'mp4_high'))
  39
  40         formats = [{
  41             'url': item['url'],
  42             'format_id': item['type'],
  43             'quality': quality(item['type']),
  44         } for item in manifest['flavors'] if item['mime'].startswith('video/')]
  45
  46         return {
  47             'url': url,
  48             'id': video_id,
  49             'title': manifest['title'],
  50             'description': manifest.get('description'),
  51             'thumbnail': self._proto_relative_url(manifest.get('poster')),
  52             'timestamp': int_or_none(manifest.get('created')),
  53             'uploader': manifest.get('author'),
  54             'duration': int_or_none(manifest.get('duration')),
  55             'formats': formats,
  56         }