yt_dlp/extractor/ntvcojp.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     ExtractorError,
   7     smuggle_url,
   8     traverse_obj,
   9 )
  10
  11
  12 class NTVCoJpCUIE(InfoExtractor):
  13     IE_NAME = 'cu.ntv.co.jp'
  14     IE_DESC = 'Nippon Television Network'
  15     _VALID_URL = r'https?://cu\.ntv\.co\.jp/(?!program)(?P<id>[^/?&#]+)'
  16     _TEST = {
  17         'url': 'https://cu.ntv.co.jp/televiva-chill-gohan_181031/',
  18         'info_dict': {
  19             'id': '5978891207001',
  20             'ext': 'mp4',
  21             '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',
  22             'upload_date': '20181213',
  23             'description': 'md5:1985b51a9abc285df0104d982a325f2a',
  24             'uploader_id': '3855502814001',
  25             'timestamp': 1544669941,
  26         },
  27         'params': {
  28             # m3u8 download
  29             'skip_download': True,
  30         },
  31     }
  32
  33     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/%s/default_default/index.html?videoId=%s'
  34
  35     def _real_extract(self, url):
  36         display_id = self._match_id(url)
  37         webpage = self._download_webpage(url, display_id)
  38         player_config = self._search_nuxt_data(webpage, display_id)
  39         video_id = traverse_obj(player_config, ('movie', 'video_id'))
  40         if not video_id:
  41             raise ExtractorError('Failed to extract video ID for Brightcove')
  42         account_id = traverse_obj(player_config, ('player', 'account')) or '3855502814001'
  43         title = traverse_obj(player_config, ('movie', 'name'))
  44         if not title:
  45             og_title = self._og_search_title(webpage, fatal=False) or traverse_obj(player_config, ('player', 'title'))
  46             if og_title:
  47                 title = og_title.split('(', 1)[0].strip()
  48         description = (traverse_obj(player_config, ('movie', 'description'))
  49                        or self._html_search_meta(['description', 'og:description'], webpage))
  50         return {
  51             '_type': 'url_transparent',
  52             'id': video_id,
  53             'display_id': display_id,
  54             'title': title,
  55             'description': description,
  56             'url': smuggle_url(self.BRIGHTCOVE_URL_TEMPLATE % (account_id, video_id), {'geo_countries': ['JP']}),
  57             'ie_key': 'BrightcoveNew',
  58         }