yt_dlp/extractor/podomatic.py

   1 import json
   2
   3 from .common import InfoExtractor
   4 from ..utils import int_or_none
   5
   6
   7 class PodomaticIE(InfoExtractor):
   8     _WORKING = False
   9     IE_NAME = 'podomatic'
  10     _VALID_URL = r'''(?x)
  11                     (?P<proto>https?)://
  12                         (?:
  13                             (?P<channel>[^.]+)\.podomatic\.com/entry|
  14                             (?:www\.)?podomatic\.com/podcasts/(?P<channel_2>[^/]+)/episodes
  15                         )/
  16                         (?P<id>[^/?#&]+)
  17                 '''
  18
  19     _TESTS = [{
  20         'url': 'http://scienceteachingtips.podomatic.com/entry/2009-01-02T16_03_35-08_00',
  21         'md5': '84bb855fcf3429e6bf72460e1eed782d',
  22         'info_dict': {
  23             'id': '2009-01-02T16_03_35-08_00',
  24             'ext': 'mp3',
  25             'uploader': 'Science Teaching Tips',
  26             'uploader_id': 'scienceteachingtips',
  27             'title': '64.  When the Moon Hits Your Eye',
  28             'duration': 446,
  29         }
  30     }, {
  31         'url': 'http://ostbahnhof.podomatic.com/entry/2013-11-15T16_31_21-08_00',
  32         'md5': 'd2cf443931b6148e27638650e2638297',
  33         'info_dict': {
  34             'id': '2013-11-15T16_31_21-08_00',
  35             'ext': 'mp3',
  36             'uploader': 'Ostbahnhof / Techno Mix',
  37             'uploader_id': 'ostbahnhof',
  38             'title': 'Einunddreizig',
  39             'duration': 3799,
  40         }
  41     }, {
  42         'url': 'https://www.podomatic.com/podcasts/scienceteachingtips/episodes/2009-01-02T16_03_35-08_00',
  43         'only_matching': True,
  44     }]
  45
  46     def _real_extract(self, url):
  47         mobj = self._match_valid_url(url)
  48         video_id = mobj.group('id')
  49         channel = mobj.group('channel') or mobj.group('channel_2')
  50
  51         json_url = (('%s://%s.podomatic.com/entry/embed_params/%s'
  52                      + '?permalink=true&rtmp=0') %
  53                     (mobj.group('proto'), channel, video_id))
  54         data_json = self._download_webpage(
  55             json_url, video_id, 'Downloading video info')
  56         data = json.loads(data_json)
  57
  58         video_url = data['downloadLink']
  59         if not video_url:
  60             video_url = '%s/%s' % (data['streamer'].replace('rtmp', 'http'), data['mediaLocation'])
  61         uploader = data['podcast']
  62         title = data['title']
  63         thumbnail = data['imageLocation']
  64         duration = int_or_none(data.get('length'), 1000)
  65
  66         return {
  67             'id': video_id,
  68             'url': video_url,
  69             'title': title,
  70             'uploader': uploader,
  71             'uploader_id': channel,
  72             'thumbnail': thumbnail,
  73             'duration': duration,
  74         }