yt_dlp/extractor/bigo.py

   1 from .common import InfoExtractor
   2 from ..utils import ExtractorError, urlencode_postdata
   3
   4
   5 class BigoIE(InfoExtractor):
   6     _VALID_URL = r'https?://(?:www\.)?bigo\.tv/(?:[a-z]{2,}/)?(?P<id>[^/]+)'
   7
   8     _TESTS = [{
   9         'url': 'https://www.bigo.tv/ja/221338632',
  10         'info_dict': {
  11             'id': '6576287577575737440',
  12             'title': '土よ〜💁‍♂️ 休憩室/REST room',
  13             'thumbnail': r're:https?://.+',
  14             'uploader': '✨Shin💫',
  15             'uploader_id': '221338632',
  16             'is_live': True,
  17         },
  18         'skip': 'livestream',
  19     }, {
  20         'url': 'https://www.bigo.tv/th/Tarlerm1304',
  21         'only_matching': True,
  22     }, {
  23         'url': 'https://bigo.tv/115976881',
  24         'only_matching': True,
  25     }]
  26
  27     def _real_extract(self, url):
  28         user_id = self._match_id(url)
  29
  30         info_raw = self._download_json(
  31             'https://ta.bigo.tv/official_website/studio/getInternalStudioInfo',
  32             user_id, data=urlencode_postdata({'siteId': user_id}),
  33             headers={'Accept': 'application/json'})
  34
  35         if not isinstance(info_raw, dict):
  36             raise ExtractorError('Received invalid JSON data')
  37         if info_raw.get('code'):
  38             raise ExtractorError(
  39                 'Bigo says: {} (code {})'.format(info_raw.get('msg'), info_raw.get('code')), expected=True)
  40         info = info_raw.get('data') or {}
  41
  42         if not info.get('alive'):
  43             raise ExtractorError('This user is offline.', expected=True)
  44
  45         formats, subs = self._extract_m3u8_formats_and_subtitles(
  46             info.get('hls_src'), user_id, 'mp4', 'm3u8')
  47
  48         return {
  49             'id': info.get('roomId') or user_id,
  50             'title': info.get('roomTopic') or info.get('nick_name') or user_id,
  51             'formats': formats,
  52             'subtitles': subs,
  53             'thumbnail': info.get('snapshot'),
  54             'uploader': info.get('nick_name'),
  55             'uploader_id': user_id,
  56             'is_live': True,
  57         }