yt_dlp/extractor/skynewsau.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     try_get,
   4     unified_strdate,
   5 )
   6
   7
   8 class SkyNewsAUIE(InfoExtractor):
   9     _VALID_URL = r'https?://(?:www\.)?skynews\.com\.au/[^/]+/[^/]+/[^/]+/video/(?P<id>[a-z0-9]+)'
  10
  11     _TESTS = [{
  12         'url': 'https://www.skynews.com.au/world-news/united-states/incredible-vision-shows-lava-overflowing-from-spains-la-palma-volcano/video/0f4c6243d6903502c01251f228b91a71',
  13         'info_dict': {
  14             'id': '6277184925001',
  15             'ext': 'mp4',
  16             'title': 'md5:60594f1ea6d5ae93e292900f4d34e9ae',
  17             'description': 'md5:60594f1ea6d5ae93e292900f4d34e9ae',
  18             'thumbnail': r're:^https?://.*\.jpg',
  19             'duration': 76.394,
  20             'timestamp': 1634271300,
  21             'uploader_id': '5348771529001',
  22             'tags': ['fblink', 'msn', 'usa', 'world', 'yt'],
  23             'upload_date': '20211015',
  24         },
  25         'params': {'skip_download': True, 'format': 'bv'}
  26     }]
  27
  28     _API_KEY = '6krsj3w249nk779d8fukqx9f'
  29
  30     def _real_extract(self, url):
  31         id = self._match_id(url)
  32         webpage = self._download_webpage(url, id)
  33         embedcode = self._search_regex(r'embedcode\s?=\s?\"([^\"]+)\"', webpage, 'embedcode')
  34         data_json = self._download_json(
  35             f'https://content.api.news/v3/videos/brightcove/{embedcode}?api_key={self._API_KEY}', id)['content']
  36         return {
  37             'id': id,
  38             '_type': 'url_transparent',
  39             'url': 'https://players.brightcove.net/%s/default_default/index.html?videoId=%s' % tuple(embedcode.split('-')),
  40             'ie_key': 'BrightcoveNew',
  41             'title': data_json.get('caption'),
  42             'upload_date': unified_strdate(try_get(data_json, lambda x: x['date']['created'])),
  43         }