yt_dlp/extractor/ninenews.py

   1 from .brightcove import BrightcoveNewIE
   2 from .common import InfoExtractor
   3 from ..utils import ExtractorError
   4 from ..utils.traversal import traverse_obj
   5
   6
   7 class NineNewsIE(InfoExtractor):
   8     IE_NAME = '9News'
   9     _VALID_URL = r'https?://(?:www\.)?9news\.com\.au/(?:[\w-]+/){2,3}(?P<id>[\w-]+)/?(?:$|[?#])'
  10     _TESTS = [{
  11         'url': 'https://www.9news.com.au/videos/national/fair-trading-pulls-dozens-of-toys-from-shelves/clqgc7dvj000y0jnvfism0w5m',
  12         'md5': 'd1a65b2e9d126e5feb9bc5cb96e62c80',
  13         'info_dict': {
  14             'id': '6343717246112',
  15             'ext': 'mp4',
  16             'title': 'Fair Trading pulls dozens of toys from shelves',
  17             'description': 'Fair Trading Australia have been forced to pull dozens of toys from shelves over hazard fears.',
  18             'thumbnail': 'md5:bdbe44294e2323b762d97acf8843f66c',
  19             'duration': 93.44,
  20             'timestamp': 1703231748,
  21             'upload_date': '20231222',
  22             'uploader_id': '664969388001',
  23             'tags': ['networkclip', 'aunews_aunationalninenews', 'christmas presents', 'toys', 'fair trading', 'au_news'],
  24         },
  25     }, {
  26         'url': 'https://www.9news.com.au/world/tape-reveals-donald-trump-pressured-michigan-officials-not-to-certify-2020-vote-a-new-report-says/0b8b880e-7d3c-41b9-b2bd-55bc7e492259',
  27         'md5': 'a885c44d20898c3e70e9a53e8188cea1',
  28         'info_dict': {
  29             'id': '6343587450112',
  30             'ext': 'mp4',
  31             'title': 'Trump found ineligible to run for president by state court',
  32             'description': 'md5:40e6e7db7a4ac6be0e960569a5af6066',
  33             'thumbnail': 'md5:3e132c48c186039fd06c10787de9bff2',
  34             'duration': 104.64,
  35             'timestamp': 1703058034,
  36             'upload_date': '20231220',
  37             'uploader_id': '664969388001',
  38             'tags': ['networkclip', 'aunews_aunationalninenews', 'ineligible', 'presidential candidate', 'donald trump', 'au_news'],
  39         },
  40     }, {
  41         'url': 'https://www.9news.com.au/national/outrage-as-parents-banned-from-giving-gifts-to-kindergarten-teachers/e19b49d4-a1a4-4533-9089-6e10e2d9386a',
  42         'info_dict': {
  43             'id': '6343716797112',
  44             'ext': 'mp4',
  45             'title': 'Outrage as parents banned from giving gifts to kindergarten teachers',
  46             'description': 'md5:7a8b0ed2f9e08875fd9a3e86e462bc46',
  47             'thumbnail': 'md5:5ee4d66717bdd0dee9fc9a705ef041b8',
  48             'duration': 91.307,
  49             'timestamp': 1703229584,
  50             'upload_date': '20231222',
  51             'uploader_id': '664969388001',
  52             'tags': ['networkclip', 'aunews_aunationalninenews', 'presents', 'teachers', 'kindergarten', 'au_news'],
  53         },
  54     }]
  55
  56     def _real_extract(self, url):
  57         article_id = self._match_id(url)
  58         webpage = self._download_webpage(url, article_id)
  59         initial_state = self._search_json(
  60             r'var\s+__INITIAL_STATE__\s*=', webpage, 'initial state', article_id)
  61         video_id = traverse_obj(
  62             initial_state, ('videoIndex', 'currentVideo', 'brightcoveId', {str}),
  63             ('article', ..., 'media', lambda _, v: v['type'] == 'video', 'urn', {str}), get_all=False)
  64         account = traverse_obj(initial_state, (
  65             'videoIndex', 'config', (None, 'video'), 'account', {str}), get_all=False)
  66
  67         if not video_id or not account:
  68             raise ExtractorError('Unable to get the required video data')
  69
  70         return self.url_result(
  71             f'https://players.brightcove.net/{account}/default_default/index.html?videoId={video_id}',
  72             BrightcoveNewIE, video_id)