yt_dlp/extractor/ntvcojp.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     ExtractorError,
   4     smuggle_url,
   5     traverse_obj,
   6 )
   7
   8
   9 class NTVCoJpCUIE(InfoExtractor):
  10     IE_NAME = 'cu.ntv.co.jp'
  11     IE_DESC = 'Nippon Television Network'
  12     _VALID_URL = r'https?://cu\.ntv\.co\.jp/(?!program)(?P<id>[^/?&#]+)'
  13     _TEST = {
  14         'url': 'https://cu.ntv.co.jp/televiva-chill-gohan_181031/',
  15         'info_dict': {
  16             'id': '5978891207001',
  17             'ext': 'mp4',
  18             'title': '桜エビと炒り卵がポイント！ 「中華風 エビチリおにぎり」──『美虎』五十嵐美幸',
  19             'upload_date': '20181213',
  20             'description': 'md5:1985b51a9abc285df0104d982a325f2a',
  21             'uploader_id': '3855502814001',
  22             'timestamp': 1544669941,
  23         },
  24         'params': {
  25             # m3u8 download
  26             'skip_download': True,
  27         },
  28     }
  29
  30     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/%s/default_default/index.html?videoId=%s'
  31
  32     def _real_extract(self, url):
  33         display_id = self._match_id(url)
  34         webpage = self._download_webpage(url, display_id)
  35         player_config = self._search_nuxt_data(webpage, display_id)
  36         video_id = traverse_obj(player_config, ('movie', 'video_id'))
  37         if not video_id:
  38             raise ExtractorError('Failed to extract video ID for Brightcove')
  39         account_id = traverse_obj(player_config, ('player', 'account')) or '3855502814001'
  40         title = traverse_obj(player_config, ('movie', 'name'))
  41         if not title:
  42             og_title = self._og_search_title(webpage, fatal=False) or traverse_obj(player_config, ('player', 'title'))
  43             if og_title:
  44                 title = og_title.split('(', 1)[0].strip()
  45         description = (traverse_obj(player_config, ('movie', 'description'))
  46                        or self._html_search_meta(['description', 'og:description'], webpage))
  47         return {
  48             '_type': 'url_transparent',
  49             'id': video_id,
  50             'display_id': display_id,
  51             'title': title,
  52             'description': description,
  53             'url': smuggle_url(self.BRIGHTCOVE_URL_TEMPLATE % (account_id, video_id), {'geo_countries': ['JP']}),
  54             'ie_key': 'BrightcoveNew',
  55         }