yt_dlp/extractor/onenewsnz.py

   1 from .brightcove import BrightcoveNewIE
   2 from .common import InfoExtractor
   3
   4 from ..utils import (
   5     ExtractorError,
   6     traverse_obj
   7 )
   8
   9
  10 class OneNewsNZIE(InfoExtractor):
  11     IE_NAME = '1News'
  12     IE_DESC = '1news.co.nz article videos'
  13     _VALID_URL = r'https?://(?:www\.)?(?:1|one)news\.co\.nz/\d+/\d+/\d+/(?P<id>[^/?#&]+)'
  14     _TESTS = [
  15         {   # Brightcove video
  16             'url': 'https://www.1news.co.nz/2022/09/29/cows-painted-green-on-parliament-lawn-in-climate-protest/',
  17             'info_dict': {
  18                 'id': 'cows-painted-green-on-parliament-lawn-in-climate-protest',
  19                 'title': '\'Cows\' painted green on Parliament lawn in climate protest',
  20             },
  21             'playlist': [{
  22                 'info_dict': {
  23                     'id': '6312993358112',
  24                     'title': 'Activists dressed as cows painted green outside Parliament in climate protest',
  25                     'ext': 'mp4',
  26                     'tags': 'count:6',
  27                     'uploader_id': '963482464001',
  28                     'timestamp': 1664416255,
  29                     'upload_date': '20220929',
  30                     'duration': 38.272,
  31                     'thumbnail': r're:^https?://.*\.jpg$',
  32                     'description': 'Greenpeace accused the Government of "greenwashing" instead of taking climate action.',
  33                 }
  34             }]
  35         }, {
  36             # YouTube video
  37             'url': 'https://www.1news.co.nz/2022/09/30/now-is-the-time-to-care-about-womens-rugby/',
  38             'info_dict': {
  39                 'id': 'now-is-the-time-to-care-about-womens-rugby',
  40                 'title': 'Now is the time to care about women\'s rugby',
  41             },
  42             'playlist': [{
  43                 'info_dict': {
  44                     'id': 's4wEB9neTfU',
  45                     'title': 'Why I love women’s rugby: Black Fern Ruahei Demant',
  46                     'ext': 'mp4',
  47                     'channel_follower_count': int,
  48                     'channel_url': 'https://www.youtube.com/channel/UC2BQ3U9IxoYIJyulv0bN5PQ',
  49                     'tags': 'count:12',
  50                     'uploader': 'Re: News',
  51                     'upload_date': '20211215',
  52                     'uploader_id': 'UC2BQ3U9IxoYIJyulv0bN5PQ',
  53                     'uploader_url': 'http://www.youtube.com/channel/UC2BQ3U9IxoYIJyulv0bN5PQ',
  54                     'channel_id': 'UC2BQ3U9IxoYIJyulv0bN5PQ',
  55                     'channel': 'Re: News',
  56                     'like_count': int,
  57                     'thumbnail': 'https://i.ytimg.com/vi/s4wEB9neTfU/maxresdefault.jpg',
  58                     'age_limit': 0,
  59                     'view_count': int,
  60                     'categories': ['Sports'],
  61                     'duration': 222,
  62                     'description': 'md5:8874410e5740ed1d8fd0df839f849813',
  63                     'availability': 'public',
  64                     'playable_in_embed': True,
  65                     'live_status': 'not_live',
  66                 }
  67             }]
  68         }, {
  69             # 2 Brightcove videos
  70             'url': 'https://www.1news.co.nz/2022/09/29/raw-videos-capture-hurricane-ians-fury-as-it-slams-florida/',
  71             'info_dict': {
  72                 'id': 'raw-videos-capture-hurricane-ians-fury-as-it-slams-florida',
  73                 'title': 'Raw videos capture Hurricane Ian\'s fury as it slams Florida',
  74             },
  75             'playlist_mincount': 2,
  76         }, {
  77             'url': 'https://www.onenews.co.nz/2022/09/29/cows-painted-green-on-parliament-lawn-in-climate-protest/',
  78             'only_matching': True,
  79         }]
  80
  81     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/%s/0xpHIR6IB_default/index.html?videoId=%s'
  82
  83     def _real_extract(self, url):
  84         display_id = self._match_id(url)
  85         webpage = self._download_webpage(url, display_id)
  86
  87         fusion_metadata = self._search_json(r'Fusion\.globalContent\s*=', webpage, 'fusion metadata', display_id)
  88
  89         entries = []
  90         for item in traverse_obj(fusion_metadata, 'content_elements') or []:
  91             item_type = traverse_obj(item, 'subtype')
  92             if item_type == 'video':
  93                 brightcove_config = traverse_obj(item, ('embed', 'config'))
  94                 brightcove_url = self.BRIGHTCOVE_URL_TEMPLATE % (
  95                     traverse_obj(brightcove_config, 'brightcoveAccount') or '963482464001',
  96                     traverse_obj(brightcove_config, 'brightcoveVideoId')
  97                 )
  98                 entries.append(self.url_result(brightcove_url, BrightcoveNewIE))
  99             elif item_type == 'youtube':
 100                 video_id_or_url = traverse_obj(item, ('referent', 'id'), ('raw_oembed', '_id'))
 101                 if video_id_or_url:
 102                     entries.append(self.url_result(video_id_or_url, ie='Youtube'))
 103
 104         if not entries:
 105             raise ExtractorError('This article does not have a video.', expected=True)
 106
 107         playlist_title = (
 108             traverse_obj(fusion_metadata, ('headlines', 'basic'))
 109             or self._generic_title('', webpage)
 110         )
 111         return self.playlist_result(entries, display_id, playlist_title)