anchorfm.py 4.7 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798
  1. from .common import InfoExtractor
  2. from ..utils import (
  3. clean_html,
  4. float_or_none,
  5. int_or_none,
  6. str_or_none,
  7. traverse_obj,
  8. unified_timestamp,
  9. )
  10. class AnchorFMEpisodeIE(InfoExtractor):
  11. _VALID_URL = r'https?://anchor\.fm/(?P<channel_name>\w+)/(?:embed/)?episodes/[\w-]+-(?P<episode_id>\w+)'
  12. _EMBED_REGEX = [rf'<iframe[^>]+\bsrc=[\'"](?P<url>{_VALID_URL})']
  13. _TESTS = [{
  14. 'url': 'https://anchor.fm/lovelyti/episodes/Chrisean-Rock-takes-to-twitter-to-announce-shes-pregnant--Blueface-denies-he-is-the-father-e1tpt3d',
  15. 'info_dict': {
  16. 'id': 'e1tpt3d',
  17. 'ext': 'mp3',
  18. 'title': ' Chrisean Rock takes to twitter to announce she\'s pregnant, Blueface denies he is the father!',
  19. 'description': 'md5:207d167de3e28ceb4ddc1ebf5a30044c',
  20. 'thumbnail': 'https://s3-us-west-2.amazonaws.com/anchor-generated-image-bank/production/podcast_uploaded_nologo/1034827/1034827-1658438968460-5f3bfdf3601e8.jpg',
  21. 'duration': 624.718,
  22. 'uploader': 'Lovelyti ',
  23. 'uploader_id': '991541',
  24. 'channel': 'lovelyti',
  25. 'modified_date': '20230121',
  26. 'modified_timestamp': 1674285178,
  27. 'release_date': '20230121',
  28. 'release_timestamp': 1674285179,
  29. 'episode_id': 'e1tpt3d',
  30. },
  31. }, {
  32. # embed url
  33. 'url': 'https://anchor.fm/apakatatempo/embed/episodes/S2E75-Perang-Bintang-di-Balik-Kasus-Ferdy-Sambo-dan-Ismail-Bolong-e1shjqd',
  34. 'info_dict': {
  35. 'id': 'e1shjqd',
  36. 'ext': 'mp3',
  37. 'title': 'S2E75 Perang Bintang di Balik Kasus Ferdy Sambo dan Ismail Bolong',
  38. 'description': 'md5:9e95ad9293bf00178bf8d33e9cb92c41',
  39. 'duration': 1042.008,
  40. 'thumbnail': 'https://s3-us-west-2.amazonaws.com/anchor-generated-image-bank/production/podcast_uploaded_episode400/2627805/2627805-1671590688729-4db3882ac9e4b.jpg',
  41. 'release_date': '20221221',
  42. 'release_timestamp': 1671595916,
  43. 'modified_date': '20221221',
  44. 'modified_timestamp': 1671590834,
  45. 'channel': 'apakatatempo',
  46. 'uploader': 'Podcast Tempo',
  47. 'uploader_id': '2585461',
  48. 'season': 'Season 2',
  49. 'season_number': 2,
  50. 'episode_id': 'e1shjqd',
  51. },
  52. }]
  53. _WEBPAGE_TESTS = [{
  54. 'url': 'https://podcast.tempo.co/podcast/192/perang-bintang-di-balik-kasus-ferdy-sambo-dan-ismail-bolong',
  55. 'info_dict': {
  56. 'id': 'e1shjqd',
  57. 'ext': 'mp3',
  58. 'release_date': '20221221',
  59. 'duration': 1042.008,
  60. 'season': 'Season 2',
  61. 'modified_timestamp': 1671590834,
  62. 'uploader_id': '2585461',
  63. 'modified_date': '20221221',
  64. 'description': 'md5:9e95ad9293bf00178bf8d33e9cb92c41',
  65. 'season_number': 2,
  66. 'title': 'S2E75 Perang Bintang di Balik Kasus Ferdy Sambo dan Ismail Bolong',
  67. 'release_timestamp': 1671595916,
  68. 'episode_id': 'e1shjqd',
  69. 'thumbnail': 'https://s3-us-west-2.amazonaws.com/anchor-generated-image-bank/production/podcast_uploaded_episode400/2627805/2627805-1671590688729-4db3882ac9e4b.jpg',
  70. 'uploader': 'Podcast Tempo',
  71. 'channel': 'apakatatempo',
  72. },
  73. }]
  74. def _real_extract(self, url):
  75. channel_name, episode_id = self._match_valid_url(url).group('channel_name', 'episode_id')
  76. api_data = self._download_json(f'https://anchor.fm/api/v3/episodes/{episode_id}', episode_id)
  77. return {
  78. 'id': episode_id,
  79. 'title': traverse_obj(api_data, ('episode', 'title')),
  80. 'url': traverse_obj(api_data, ('episode', 'episodeEnclosureUrl'), ('episodeAudios', 0, 'url')),
  81. 'ext': 'mp3',
  82. 'vcodec': 'none',
  83. 'thumbnail': traverse_obj(api_data, ('episode', 'episodeImage')),
  84. 'description': clean_html(traverse_obj(api_data, ('episode', ('description', 'descriptionPreview')), get_all=False)),
  85. 'duration': float_or_none(traverse_obj(api_data, ('episode', 'duration')), 1000),
  86. 'modified_timestamp': unified_timestamp(traverse_obj(api_data, ('episode', 'modified'))),
  87. 'release_timestamp': int_or_none(traverse_obj(api_data, ('episode', 'publishOnUnixTimestamp'))),
  88. 'episode_id': episode_id,
  89. 'uploader': traverse_obj(api_data, ('creator', 'name')),
  90. 'uploader_id': str_or_none(traverse_obj(api_data, ('creator', 'userId'))),
  91. 'season_number': int_or_none(traverse_obj(api_data, ('episode', 'podcastSeasonNumber'))),
  92. 'channel': channel_name or traverse_obj(api_data, ('creator', 'vanitySlug')),
  93. }