yt_dlp/extractor/boxcast.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     js_to_json,
   4     traverse_obj,
   5     unified_timestamp
   6 )
   7
   8
   9 class BoxCastVideoIE(InfoExtractor):
  10     _VALID_URL = r'''(?x)
  11         https?://boxcast\.tv/(?:
  12             view-embed/|
  13             channel/\w+\?(?:[^#]+&)?b=|
  14             video-portal/(?:\w+/){2}
  15         )(?P<id>[\w-]+)'''
  16     _EMBED_REGEX = [r'<iframe[^>]+src=["\'](?P<url>https?://boxcast\.tv/view-embed/[\w-]+)']
  17     _TESTS = [{
  18         'url': 'https://boxcast.tv/view-embed/in-the-midst-of-darkness-light-prevails-an-interdisciplinary-symposium-ozmq5eclj50ujl4bmpwx',
  19         'info_dict': {
  20             'id': 'da1eqqgkacngd5djlqld',
  21             'ext': 'mp4',
  22             'thumbnail': r're:https?://uploads\.boxcast\.com/(?:[\w+-]+/){3}.+\.png$',
  23             'title': 'In the Midst of Darkness Light Prevails: An Interdisciplinary Symposium',
  24             'release_timestamp': 1670686812,
  25             'release_date': '20221210',
  26             'uploader_id': 're8w0v8hohhvpqtbskpe',
  27             'uploader': 'Children\'s Health Defense',
  28         }
  29     }, {
  30         'url': 'https://boxcast.tv/video-portal/vctwevwntun3o0ikq7af/rvyblnn0fxbfjx5nwxhl/otbpltj2kzkveo2qz3ad',
  31         'info_dict': {
  32             'id': 'otbpltj2kzkveo2qz3ad',
  33             'ext': 'mp4',
  34             'uploader_id': 'vctwevwntun3o0ikq7af',
  35             'uploader': 'Legacy Christian Church',
  36             'title': 'The Quest | 1: Beginner\'s Bay | Jamie Schools',
  37             'thumbnail': r're:https?://uploads.boxcast.com/(?:[\w-]+/){3}.+\.jpg'
  38         }
  39     }, {
  40         'url': 'https://boxcast.tv/channel/z03fqwaeaby5lnaawox2?b=ssihlw5gvfij2by8tkev',
  41         'info_dict': {
  42             'id': 'ssihlw5gvfij2by8tkev',
  43             'ext': 'mp4',
  44             'thumbnail': r're:https?://uploads.boxcast.com/(?:[\w-]+/){3}.+\.jpg$',
  45             'release_date': '20230101',
  46             'uploader_id': 'ds25vaazhlu4ygcvffid',
  47             'release_timestamp': 1672543201,
  48             'uploader': 'Lighthouse Ministries International  - Beltsville, Maryland',
  49             'description': 'md5:ac23e3d01b0b0be592e8f7fe0ec3a340',
  50             'title': 'New Year\'s Eve CROSSOVER Service at LHMI | December 31, 2022',
  51         }
  52     }]
  53     _WEBPAGE_TESTS = [{
  54         'url': 'https://childrenshealthdefense.eu/live-stream/',
  55         'info_dict': {
  56             'id': 'da1eqqgkacngd5djlqld',
  57             'ext': 'mp4',
  58             'thumbnail': r're:https?://uploads\.boxcast\.com/(?:[\w+-]+/){3}.+\.png$',
  59             'title': 'In the Midst of Darkness Light Prevails: An Interdisciplinary Symposium',
  60             'release_timestamp': 1670686812,
  61             'release_date': '20221210',
  62             'uploader_id': 're8w0v8hohhvpqtbskpe',
  63             'uploader': 'Children\'s Health Defense',
  64         }
  65     }]
  66
  67     def _real_extract(self, url):
  68         display_id = self._match_id(url)
  69         webpage = self._download_webpage(url, display_id)
  70         webpage_json_data = self._search_json(
  71             r'var\s*BOXCAST_PRELOAD\s*=', webpage, 'broadcast data', display_id,
  72             transform_source=js_to_json, default={})
  73
  74         # Ref: https://support.boxcast.com/en/articles/4235158-build-a-custom-viewer-experience-with-boxcast-api
  75         broadcast_json_data = (
  76             traverse_obj(webpage_json_data, ('broadcast', 'data'))
  77             or self._download_json(f'https://api.boxcast.com/broadcasts/{display_id}', display_id))
  78         view_json_data = (
  79             traverse_obj(webpage_json_data, ('view', 'data'))
  80             or self._download_json(f'https://api.boxcast.com/broadcasts/{display_id}/view',
  81                                    display_id, fatal=False) or {})
  82
  83         formats, subtitles = [], {}
  84         if view_json_data.get('status') == 'recorded':
  85             formats, subtitles = self._extract_m3u8_formats_and_subtitles(
  86                 view_json_data['playlist'], display_id)
  87
  88         return {
  89             'id': str(broadcast_json_data['id']),
  90             'title': (broadcast_json_data.get('name')
  91                       or self._html_search_meta(['og:title', 'twitter:title'], webpage)),
  92             'description': (broadcast_json_data.get('description')
  93                             or self._html_search_meta(['og:description', 'twitter:description'], webpage)
  94                             or None),
  95             'thumbnail': (broadcast_json_data.get('preview')
  96                           or self._html_search_meta(['og:image', 'twitter:image'], webpage)),
  97             'formats': formats,
  98             'subtitles': subtitles,
  99             'release_timestamp': unified_timestamp(broadcast_json_data.get('streamed_at')),
 100             'uploader': broadcast_json_data.get('account_name'),
 101             'uploader_id': broadcast_json_data.get('account_id'),
 102         }