yt_dlp/extractor/jiosaavn.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     js_to_json,
   4     url_or_none,
   5     urlencode_postdata,
   6     urljoin,
   7 )
   8 from ..utils.traversal import traverse_obj
   9
  10
  11 class JioSaavnBaseIE(InfoExtractor):
  12     def _extract_initial_data(self, url, audio_id):
  13         webpage = self._download_webpage(url, audio_id)
  14         return self._search_json(
  15             r'window\.__INITIAL_DATA__\s*=', webpage,
  16             'init json', audio_id, transform_source=js_to_json)
  17
  18
  19 class JioSaavnSongIE(JioSaavnBaseIE):
  20     _VALID_URL = r'https?://(?:www\.)?(?:jiosaavn\.com/song/[^/?#]+/|saavn\.com/s/song/(?:[^/?#]+/){3})(?P<id>[^/?#]+)'
  21     _TESTS = [{
  22         'url': 'https://www.jiosaavn.com/song/leja-re/OQsEfQFVUXk',
  23         'md5': '7b1f70de088ede3a152ea34aece4df42',
  24         'info_dict': {
  25             'id': 'OQsEfQFVUXk',
  26             'ext': 'mp3',
  27             'title': 'Leja Re',
  28             'album': 'Leja Re',
  29             'thumbnail': 'https://c.saavncdn.com/258/Leja-Re-Hindi-2018-20181124024539-500x500.jpg',
  30         },
  31     }, {
  32         'url': 'https://www.saavn.com/s/song/hindi/Saathiya/O-Humdum-Suniyo-Re/KAMiazoCblU',
  33         'only_matching': True,
  34     }]
  35
  36     def _real_extract(self, url):
  37         audio_id = self._match_id(url)
  38         song_data = self._extract_initial_data(url, audio_id)['song']['song']
  39         media_data = self._download_json(
  40             'https://www.jiosaavn.com/api.php', audio_id, data=urlencode_postdata({
  41                 '__call': 'song.generateAuthToken',
  42                 '_format': 'json',
  43                 'bitrate': '128',
  44                 'url': song_data['encrypted_media_url'],
  45             }))
  46
  47         return {
  48             'id': audio_id,
  49             'url': media_data['auth_url'],
  50             'ext': media_data.get('type'),
  51             'vcodec': 'none',
  52             **traverse_obj(song_data, {
  53                 'title': ('title', 'text'),
  54                 'album': ('album', 'text'),
  55                 'thumbnail': ('image', 0, {url_or_none}),
  56             }),
  57         }
  58
  59
  60 class JioSaavnAlbumIE(JioSaavnBaseIE):
  61     _VALID_URL = r'https?://(?:www\.)?(?:jio)?saavn\.com/album/[^/?#]+/(?P<id>[^/?#]+)'
  62     _TESTS = [{
  63         'url': 'https://www.jiosaavn.com/album/96/buIOjYZDrNA_',
  64         'info_dict': {
  65             'id': 'buIOjYZDrNA_',
  66             'title': '96',
  67         },
  68         'playlist_count': 10,
  69     }]
  70
  71     def _real_extract(self, url):
  72         album_id = self._match_id(url)
  73         album_view = self._extract_initial_data(url, album_id)['albumView']
  74
  75         return self.playlist_from_matches(
  76             traverse_obj(album_view, (
  77                 'modules', lambda _, x: x['key'] == 'list', 'data', ..., 'title', 'action', {str})),
  78             album_id, traverse_obj(album_view, ('album', 'title', 'text', {str})), ie=JioSaavnSongIE,
  79             getter=lambda x: urljoin('https://www.jiosaavn.com/', x))