yt_dlp/extractor/acast.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     clean_html,
   4     clean_podcast_url,
   5     int_or_none,
   6     parse_iso8601,
   7 )
   8
   9
  10 class ACastBaseIE(InfoExtractor):
  11     def _extract_episode(self, episode, show_info):
  12         title = episode['title']
  13         info = {
  14             'id': episode['id'],
  15             'display_id': episode.get('episodeUrl'),
  16             'url': clean_podcast_url(episode['url']),
  17             'title': title,
  18             'description': clean_html(episode.get('description') or episode.get('summary')),
  19             'thumbnail': episode.get('image'),
  20             'timestamp': parse_iso8601(episode.get('publishDate')),
  21             'duration': int_or_none(episode.get('duration')),
  22             'filesize': int_or_none(episode.get('contentLength')),
  23             'season_number': int_or_none(episode.get('season')),
  24             'episode': title,
  25             'episode_number': int_or_none(episode.get('episode')),
  26         }
  27         info.update(show_info)
  28         return info
  29
  30     def _extract_show_info(self, show):
  31         return {
  32             'creator': show.get('author'),
  33             'series': show.get('title'),
  34         }
  35
  36     def _call_api(self, path, video_id, query=None):
  37         return self._download_json(
  38             'https://feeder.acast.com/api/v1/shows/' + path, video_id, query=query)
  39
  40
  41 class ACastIE(ACastBaseIE):
  42     IE_NAME = 'acast'
  43     _VALID_URL = r'''(?x)
  44                     https?://
  45                         (?:
  46                             (?:(?:embed|www)\.)?acast\.com/|
  47                             play\.acast\.com/s/
  48                         )
  49                         (?P<channel>[^/]+)/(?P<id>[^/#?]+)
  50                     '''
  51     _TESTS = [{
  52         'url': 'https://www.acast.com/sparpodcast/2.raggarmordet-rosterurdetforflutna',
  53         'md5': 'f5598f3ad1e4776fed12ec1407153e4b',
  54         'info_dict': {
  55             'id': '2a92b283-1a75-4ad8-8396-499c641de0d9',
  56             'ext': 'mp3',
  57             'title': '2. Raggarmordet - Röster ur det förflutna',
  58             'description': 'md5:a992ae67f4d98f1c0141598f7bebbf67',
  59             'timestamp': 1477346700,
  60             'upload_date': '20161024',
  61             'duration': 2766,
  62             'creator': 'Anton Berg & Martin Johnson',
  63             'series': 'Spår',
  64             'episode': '2. Raggarmordet - Röster ur det förflutna',
  65         }
  66     }, {
  67         'url': 'http://embed.acast.com/adambuxton/ep.12-adam-joeschristmaspodcast2015',
  68         'only_matching': True,
  69     }, {
  70         'url': 'https://play.acast.com/s/rattegangspodden/s04e09styckmordetihelenelund-del2-2',
  71         'only_matching': True,
  72     }, {
  73         'url': 'https://play.acast.com/s/sparpodcast/2a92b283-1a75-4ad8-8396-499c641de0d9',
  74         'only_matching': True,
  75     }]
  76
  77     def _real_extract(self, url):
  78         channel, display_id = self._match_valid_url(url).groups()
  79         episode = self._call_api(
  80             '%s/episodes/%s' % (channel, display_id),
  81             display_id, {'showInfo': 'true'})
  82         return self._extract_episode(
  83             episode, self._extract_show_info(episode.get('show') or {}))
  84
  85
  86 class ACastChannelIE(ACastBaseIE):
  87     IE_NAME = 'acast:channel'
  88     _VALID_URL = r'''(?x)
  89                     https?://
  90                         (?:
  91                             (?:www\.)?acast\.com/|
  92                             play\.acast\.com/s/
  93                         )
  94                         (?P<id>[^/#?]+)
  95                     '''
  96     _TESTS = [{
  97         'url': 'https://www.acast.com/todayinfocus',
  98         'info_dict': {
  99             'id': '4efc5294-5385-4847-98bd-519799ce5786',
 100             'title': 'Today in Focus',
 101             'description': 'md5:c09ce28c91002ce4ffce71d6504abaae',
 102         },
 103         'playlist_mincount': 200,
 104     }, {
 105         'url': 'http://play.acast.com/s/ft-banking-weekly',
 106         'only_matching': True,
 107     }]
 108
 109     @classmethod
 110     def suitable(cls, url):
 111         return False if ACastIE.suitable(url) else super(ACastChannelIE, cls).suitable(url)
 112
 113     def _real_extract(self, url):
 114         show_slug = self._match_id(url)
 115         show = self._call_api(show_slug, show_slug)
 116         show_info = self._extract_show_info(show)
 117         entries = []
 118         for episode in (show.get('episodes') or []):
 119             entries.append(self._extract_episode(episode, show_info))
 120         return self.playlist_result(
 121             entries, show.get('id'), show.get('title'), show.get('description'))