]> jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/anchorfm.py
[misc] Add `hatch`, `ruff`, `pre-commit` and improve dev docs (#7409)
[yt-dlp.git] / yt_dlp / extractor / anchorfm.py
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
11
12 class AnchorFMEpisodeIE(InfoExtractor):
13 _VALID_URL = r'https?://anchor\.fm/(?P<channel_name>\w+)/(?:embed/)?episodes/[\w-]+-(?P<episode_id>\w+)'
14 _EMBED_REGEX = [rf'<iframe[^>]+\bsrc=[\'"](?P<url>{_VALID_URL})']
15 _TESTS = [{
16 'url': 'https://anchor.fm/lovelyti/episodes/Chrisean-Rock-takes-to-twitter-to-announce-shes-pregnant--Blueface-denies-he-is-the-father-e1tpt3d',
17 'info_dict': {
18 'id': 'e1tpt3d',
19 'ext': 'mp3',
20 'title': ' Chrisean Rock takes to twitter to announce she\'s pregnant, Blueface denies he is the father!',
21 'description': 'md5:207d167de3e28ceb4ddc1ebf5a30044c',
22 'thumbnail': 'https://s3-us-west-2.amazonaws.com/anchor-generated-image-bank/production/podcast_uploaded_nologo/1034827/1034827-1658438968460-5f3bfdf3601e8.jpg',
23 'duration': 624.718,
24 'uploader': 'Lovelyti ',
25 'uploader_id': '991541',
26 'channel': 'lovelyti',
27 'modified_date': '20230121',
28 'modified_timestamp': 1674285178,
29 'release_date': '20230121',
30 'release_timestamp': 1674285179,
31 'episode_id': 'e1tpt3d',
32 }
33 }, {
34 # embed url
35 'url': 'https://anchor.fm/apakatatempo/embed/episodes/S2E75-Perang-Bintang-di-Balik-Kasus-Ferdy-Sambo-dan-Ismail-Bolong-e1shjqd',
36 'info_dict': {
37 'id': 'e1shjqd',
38 'ext': 'mp3',
39 'title': 'S2E75 Perang Bintang di Balik Kasus Ferdy Sambo dan Ismail Bolong',
40 'description': 'md5:9e95ad9293bf00178bf8d33e9cb92c41',
41 'duration': 1042.008,
42 'thumbnail': 'https://s3-us-west-2.amazonaws.com/anchor-generated-image-bank/production/podcast_uploaded_episode400/2627805/2627805-1671590688729-4db3882ac9e4b.jpg',
43 'release_date': '20221221',
44 'release_timestamp': 1671595916,
45 'modified_date': '20221221',
46 'modified_timestamp': 1671590834,
47 'channel': 'apakatatempo',
48 'uploader': 'Podcast Tempo',
49 'uploader_id': '2585461',
50 'season': 'Season 2',
51 'season_number': 2,
52 'episode_id': 'e1shjqd',
53 }
54 }]
55
56 _WEBPAGE_TESTS = [{
57 'url': 'https://podcast.tempo.co/podcast/192/perang-bintang-di-balik-kasus-ferdy-sambo-dan-ismail-bolong',
58 'info_dict': {
59 'id': 'e1shjqd',
60 'ext': 'mp3',
61 'release_date': '20221221',
62 'duration': 1042.008,
63 'season': 'Season 2',
64 'modified_timestamp': 1671590834,
65 'uploader_id': '2585461',
66 'modified_date': '20221221',
67 'description': 'md5:9e95ad9293bf00178bf8d33e9cb92c41',
68 'season_number': 2,
69 'title': 'S2E75 Perang Bintang di Balik Kasus Ferdy Sambo dan Ismail Bolong',
70 'release_timestamp': 1671595916,
71 'episode_id': 'e1shjqd',
72 'thumbnail': 'https://s3-us-west-2.amazonaws.com/anchor-generated-image-bank/production/podcast_uploaded_episode400/2627805/2627805-1671590688729-4db3882ac9e4b.jpg',
73 'uploader': 'Podcast Tempo',
74 'channel': 'apakatatempo',
75 }
76 }]
77
78 def _real_extract(self, url):
79 channel_name, episode_id = self._match_valid_url(url).group('channel_name', 'episode_id')
80 api_data = self._download_json(f'https://anchor.fm/api/v3/episodes/{episode_id}', episode_id)
81
82 return {
83 'id': episode_id,
84 'title': traverse_obj(api_data, ('episode', 'title')),
85 'url': traverse_obj(api_data, ('episode', 'episodeEnclosureUrl'), ('episodeAudios', 0, 'url')),
86 'ext': 'mp3',
87 'vcodec': 'none',
88 'thumbnail': traverse_obj(api_data, ('episode', 'episodeImage')),
89 'description': clean_html(traverse_obj(api_data, ('episode', ('description', 'descriptionPreview')), get_all=False)),
90 'duration': float_or_none(traverse_obj(api_data, ('episode', 'duration')), 1000),
91 'modified_timestamp': unified_timestamp(traverse_obj(api_data, ('episode', 'modified'))),
92 'release_timestamp': int_or_none(traverse_obj(api_data, ('episode', 'publishOnUnixTimestamp'))),
93 'episode_id': episode_id,
94 'uploader': traverse_obj(api_data, ('creator', 'name')),
95 'uploader_id': str_or_none(traverse_obj(api_data, ('creator', 'userId'))),
96 'season_number': int_or_none(traverse_obj(api_data, ('episode', 'podcastSeasonNumber'))),
97 'channel': channel_name or traverse_obj(api_data, ('creator', 'vanitySlug')),
98 }