yt_dlp/extractor/onenewsnz.py

   1 from .brightcove import BrightcoveNewIE
   2 from .common import InfoExtractor
   3 from ..utils import ExtractorError, traverse_obj
   4
   5
   6 class OneNewsNZIE(InfoExtractor):
   7     IE_NAME = '1News'
   8     IE_DESC = '1news.co.nz article videos'
   9     _VALID_URL = r'https?://(?:www\.)?(?:1|one)news\.co\.nz/\d+/\d+/\d+/(?P<id>[^/?#&]+)'
  10     _TESTS = [
  11         {   # Brightcove video
  12             'url': 'https://www.1news.co.nz/2022/09/29/cows-painted-green-on-parliament-lawn-in-climate-protest/',
  13             'info_dict': {
  14                 'id': 'cows-painted-green-on-parliament-lawn-in-climate-protest',
  15                 'title': '\'Cows\' painted green on Parliament lawn in climate protest',
  16             },
  17             'playlist': [{
  18                 'info_dict': {
  19                     'id': '6312993358112',
  20                     'title': 'Activists dressed as cows painted green outside Parliament in climate protest',
  21                     'ext': 'mp4',
  22                     'tags': 'count:6',
  23                     'uploader_id': '963482464001',
  24                     'timestamp': 1664416255,
  25                     'upload_date': '20220929',
  26                     'duration': 38.272,
  27                     'thumbnail': r're:^https?://.*\.jpg$',
  28                     'description': 'Greenpeace accused the Government of "greenwashing" instead of taking climate action.',
  29                 },
  30             }],
  31         }, {
  32             # YouTube video
  33             'url': 'https://www.1news.co.nz/2022/09/30/now-is-the-time-to-care-about-womens-rugby/',
  34             'info_dict': {
  35                 'id': 'now-is-the-time-to-care-about-womens-rugby',
  36                 'title': 'Now is the time to care about women\'s rugby',
  37             },
  38             'playlist': [{
  39                 'info_dict': {
  40                     'id': 's4wEB9neTfU',
  41                     'title': 'Why I love women’s rugby: Black Fern Ruahei Demant',
  42                     'ext': 'mp4',
  43                     'channel_follower_count': int,
  44                     'channel_url': 'https://www.youtube.com/channel/UC2BQ3U9IxoYIJyulv0bN5PQ',
  45                     'tags': 'count:12',
  46                     'uploader': 'Re: News',
  47                     'upload_date': '20211215',
  48                     'uploader_id': 'UC2BQ3U9IxoYIJyulv0bN5PQ',
  49                     'uploader_url': 'http://www.youtube.com/channel/UC2BQ3U9IxoYIJyulv0bN5PQ',
  50                     'channel_id': 'UC2BQ3U9IxoYIJyulv0bN5PQ',
  51                     'channel': 'Re: News',
  52                     'like_count': int,
  53                     'thumbnail': 'https://i.ytimg.com/vi/s4wEB9neTfU/maxresdefault.jpg',
  54                     'age_limit': 0,
  55                     'view_count': int,
  56                     'categories': ['Sports'],
  57                     'duration': 222,
  58                     'description': 'md5:8874410e5740ed1d8fd0df839f849813',
  59                     'availability': 'public',
  60                     'playable_in_embed': True,
  61                     'live_status': 'not_live',
  62                 },
  63             }],
  64         }, {
  65             # 2 Brightcove videos
  66             'url': 'https://www.1news.co.nz/2022/09/29/raw-videos-capture-hurricane-ians-fury-as-it-slams-florida/',
  67             'info_dict': {
  68                 'id': 'raw-videos-capture-hurricane-ians-fury-as-it-slams-florida',
  69                 'title': 'Raw videos capture Hurricane Ian\'s fury as it slams Florida',
  70             },
  71             'playlist_mincount': 2,
  72         }, {
  73             'url': 'https://www.onenews.co.nz/2022/09/29/cows-painted-green-on-parliament-lawn-in-climate-protest/',
  74             'only_matching': True,
  75         }]
  76
  77     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/%s/0xpHIR6IB_default/index.html?videoId=%s'
  78
  79     def _real_extract(self, url):
  80         display_id = self._match_id(url)
  81         webpage = self._download_webpage(url, display_id)
  82
  83         fusion_metadata = self._search_json(r'Fusion\.globalContent\s*=', webpage, 'fusion metadata', display_id)
  84
  85         entries = []
  86         for item in traverse_obj(fusion_metadata, 'content_elements') or []:
  87             item_type = traverse_obj(item, 'subtype')
  88             if item_type == 'video':
  89                 brightcove_config = traverse_obj(item, ('embed', 'config'))
  90                 brightcove_url = self.BRIGHTCOVE_URL_TEMPLATE % (
  91                     traverse_obj(brightcove_config, 'brightcoveAccount') or '963482464001',
  92                     traverse_obj(brightcove_config, 'brightcoveVideoId'),
  93                 )
  94                 entries.append(self.url_result(brightcove_url, BrightcoveNewIE))
  95             elif item_type == 'youtube':
  96                 video_id_or_url = traverse_obj(item, ('referent', 'id'), ('raw_oembed', '_id'))
  97                 if video_id_or_url:
  98                     entries.append(self.url_result(video_id_or_url, ie='Youtube'))
  99
 100         if not entries:
 101             raise ExtractorError('This article does not have a video.', expected=True)
 102
 103         playlist_title = (
 104             traverse_obj(fusion_metadata, ('headlines', 'basic'))
 105             or self._generic_title('', webpage)
 106         )
 107         return self.playlist_result(entries, display_id, playlist_title)