yt_dlp/extractor/boxcast.py

   1 from .common import InfoExtractor
   2 from ..utils import js_to_json, traverse_obj, unified_timestamp
   3
   4
   5 class BoxCastVideoIE(InfoExtractor):
   6     _VALID_URL = r'''(?x)
   7         https?://boxcast\.tv/(?:
   8             view-embed/|
   9             channel/\w+\?(?:[^#]+&)?b=|
  10             video-portal/(?:\w+/){2}
  11         )(?P<id>[\w-]+)'''
  12     _EMBED_REGEX = [r'<iframe[^>]+src=["\'](?P<url>https?://boxcast\.tv/view-embed/[\w-]+)']
  13     _TESTS = [{
  14         'url': 'https://boxcast.tv/view-embed/in-the-midst-of-darkness-light-prevails-an-interdisciplinary-symposium-ozmq5eclj50ujl4bmpwx',
  15         'info_dict': {
  16             'id': 'da1eqqgkacngd5djlqld',
  17             'ext': 'mp4',
  18             'thumbnail': r're:https?://uploads\.boxcast\.com/(?:[\w+-]+/){3}.+\.png$',
  19             'title': 'In the Midst of Darkness Light Prevails: An Interdisciplinary Symposium',
  20             'release_timestamp': 1670686812,
  21             'release_date': '20221210',
  22             'uploader_id': 're8w0v8hohhvpqtbskpe',
  23             'uploader': 'Children\'s Health Defense',
  24         }
  25     }, {
  26         'url': 'https://boxcast.tv/video-portal/vctwevwntun3o0ikq7af/rvyblnn0fxbfjx5nwxhl/otbpltj2kzkveo2qz3ad',
  27         'info_dict': {
  28             'id': 'otbpltj2kzkveo2qz3ad',
  29             'ext': 'mp4',
  30             'uploader_id': 'vctwevwntun3o0ikq7af',
  31             'uploader': 'Legacy Christian Church',
  32             'title': 'The Quest | 1: Beginner\'s Bay | Jamie Schools',
  33             'thumbnail': r're:https?://uploads.boxcast.com/(?:[\w-]+/){3}.+\.jpg'
  34         }
  35     }, {
  36         'url': 'https://boxcast.tv/channel/z03fqwaeaby5lnaawox2?b=ssihlw5gvfij2by8tkev',
  37         'info_dict': {
  38             'id': 'ssihlw5gvfij2by8tkev',
  39             'ext': 'mp4',
  40             'thumbnail': r're:https?://uploads.boxcast.com/(?:[\w-]+/){3}.+\.jpg$',
  41             'release_date': '20230101',
  42             'uploader_id': 'ds25vaazhlu4ygcvffid',
  43             'release_timestamp': 1672543201,
  44             'uploader': 'Lighthouse Ministries International  - Beltsville, Maryland',
  45             'description': 'md5:ac23e3d01b0b0be592e8f7fe0ec3a340',
  46             'title': 'New Year\'s Eve CROSSOVER Service at LHMI | December 31, 2022',
  47         }
  48     }]
  49     _WEBPAGE_TESTS = [{
  50         'url': 'https://childrenshealthdefense.eu/live-stream/',
  51         'info_dict': {
  52             'id': 'da1eqqgkacngd5djlqld',
  53             'ext': 'mp4',
  54             'thumbnail': r're:https?://uploads\.boxcast\.com/(?:[\w+-]+/){3}.+\.png$',
  55             'title': 'In the Midst of Darkness Light Prevails: An Interdisciplinary Symposium',
  56             'release_timestamp': 1670686812,
  57             'release_date': '20221210',
  58             'uploader_id': 're8w0v8hohhvpqtbskpe',
  59             'uploader': 'Children\'s Health Defense',
  60         }
  61     }]
  62
  63     def _real_extract(self, url):
  64         display_id = self._match_id(url)
  65         webpage = self._download_webpage(url, display_id)
  66         webpage_json_data = self._search_json(
  67             r'var\s*BOXCAST_PRELOAD\s*=', webpage, 'broadcast data', display_id,
  68             transform_source=js_to_json, default={})
  69
  70         # Ref: https://support.boxcast.com/en/articles/4235158-build-a-custom-viewer-experience-with-boxcast-api
  71         broadcast_json_data = (
  72             traverse_obj(webpage_json_data, ('broadcast', 'data'))
  73             or self._download_json(f'https://api.boxcast.com/broadcasts/{display_id}', display_id))
  74         view_json_data = (
  75             traverse_obj(webpage_json_data, ('view', 'data'))
  76             or self._download_json(f'https://api.boxcast.com/broadcasts/{display_id}/view',
  77                                    display_id, fatal=False) or {})
  78
  79         formats, subtitles = [], {}
  80         if view_json_data.get('status') == 'recorded':
  81             formats, subtitles = self._extract_m3u8_formats_and_subtitles(
  82                 view_json_data['playlist'], display_id)
  83
  84         return {
  85             'id': str(broadcast_json_data['id']),
  86             'title': (broadcast_json_data.get('name')
  87                       or self._html_search_meta(['og:title', 'twitter:title'], webpage)),
  88             'description': (broadcast_json_data.get('description')
  89                             or self._html_search_meta(['og:description', 'twitter:description'], webpage)
  90                             or None),
  91             'thumbnail': (broadcast_json_data.get('preview')
  92                           or self._html_search_meta(['og:image', 'twitter:image'], webpage)),
  93             'formats': formats,
  94             'subtitles': subtitles,
  95             'release_timestamp': unified_timestamp(broadcast_json_data.get('streamed_at')),
  96             'uploader': broadcast_json_data.get('account_name'),
  97             'uploader_id': broadcast_json_data.get('account_id'),
  98         }