yt_dlp/extractor/podomatic.py

   1 from __future__ import unicode_literals
   2
   3 import json
   4
   5 from .common import InfoExtractor
   6 from ..utils import int_or_none
   7
   8
   9 class PodomaticIE(InfoExtractor):
  10     IE_NAME = 'podomatic'
  11     _VALID_URL = r'''(?x)
  12                     (?P<proto>https?)://
  13                         (?:
  14                             (?P<channel>[^.]+)\.podomatic\.com/entry|
  15                             (?:www\.)?podomatic\.com/podcasts/(?P<channel_2>[^/]+)/episodes
  16                         )/
  17                         (?P<id>[^/?#&]+)
  18                 '''
  19
  20     _TESTS = [{
  21         'url': 'http://scienceteachingtips.podomatic.com/entry/2009-01-02T16_03_35-08_00',
  22         'md5': '84bb855fcf3429e6bf72460e1eed782d',
  23         'info_dict': {
  24             'id': '2009-01-02T16_03_35-08_00',
  25             'ext': 'mp3',
  26             'uploader': 'Science Teaching Tips',
  27             'uploader_id': 'scienceteachingtips',
  28             'title': '64.  When the Moon Hits Your Eye',
  29             'duration': 446,
  30         }
  31     }, {
  32         'url': 'http://ostbahnhof.podomatic.com/entry/2013-11-15T16_31_21-08_00',
  33         'md5': 'd2cf443931b6148e27638650e2638297',
  34         'info_dict': {
  35             'id': '2013-11-15T16_31_21-08_00',
  36             'ext': 'mp3',
  37             'uploader': 'Ostbahnhof / Techno Mix',
  38             'uploader_id': 'ostbahnhof',
  39             'title': 'Einunddreizig',
  40             'duration': 3799,
  41         }
  42     }, {
  43         'url': 'https://www.podomatic.com/podcasts/scienceteachingtips/episodes/2009-01-02T16_03_35-08_00',
  44         'only_matching': True,
  45     }]
  46
  47     def _real_extract(self, url):
  48         mobj = self._match_valid_url(url)
  49         video_id = mobj.group('id')
  50         channel = mobj.group('channel') or mobj.group('channel_2')
  51
  52         json_url = (('%s://%s.podomatic.com/entry/embed_params/%s'
  53                      + '?permalink=true&rtmp=0') %
  54                     (mobj.group('proto'), channel, video_id))
  55         data_json = self._download_webpage(
  56             json_url, video_id, 'Downloading video info')
  57         data = json.loads(data_json)
  58
  59         video_url = data['downloadLink']
  60         if not video_url:
  61             video_url = '%s/%s' % (data['streamer'].replace('rtmp', 'http'), data['mediaLocation'])
  62         uploader = data['podcast']
  63         title = data['title']
  64         thumbnail = data['imageLocation']
  65         duration = int_or_none(data.get('length'), 1000)
  66
  67         return {
  68             'id': video_id,
  69             'url': video_url,
  70             'title': title,
  71             'uploader': uploader,
  72             'uploader_id': channel,
  73             'thumbnail': thumbnail,
  74             'duration': duration,
  75         }