Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/urplay.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import unified_timestamp
   6
   7
   8 class URPlayIE(InfoExtractor):
   9     _VALID_URL = r'https?://(?:www\.)?ur(?:play|skola)\.se/(?:program|Produkter)/(?P<id>[0-9]+)'
  10     _TESTS = [{
  11         'url': 'https://urplay.se/program/203704-ur-samtiden-livet-universum-och-rymdens-markliga-musik-om-vetenskap-kritiskt-tankande-och-motstand',
  12         'md5': 'ff5b0c89928f8083c74bbd5099c9292d',
  13         'info_dict': {
  14             'id': '203704',
  15             'ext': 'mp4',
  16             'title': 'UR Samtiden - Livet, universum och rymdens märkliga musik : Om vetenskap, kritiskt tänkande och motstånd',
  17             'description': 'md5:5344508a52aa78c1ced6c1b8b9e44e9a',
  18             'timestamp': 1513512768,
  19             'upload_date': '20171217',
  20         },
  21     }, {
  22         'url': 'https://urskola.se/Produkter/190031-Tripp-Trapp-Trad-Sovkudde',
  23         'info_dict': {
  24             'id': '190031',
  25             'ext': 'mp4',
  26             'title': 'Tripp, Trapp, Träd : Sovkudde',
  27             'description': 'md5:b86bffdae04a7e9379d1d7e5947df1d1',
  28             'timestamp': 1440093600,
  29             'upload_date': '20150820',
  30         },
  31     }, {
  32         'url': 'http://urskola.se/Produkter/155794-Smasagor-meankieli-Grodan-i-vida-varlden',
  33         'only_matching': True,
  34     }]
  35
  36     def _real_extract(self, url):
  37         video_id = self._match_id(url)
  38
  39         webpage = self._download_webpage(url, video_id)
  40         urplayer_data = self._parse_json(self._search_regex(
  41             r'urPlayer\.init\(({.+?})\);', webpage, 'urplayer data'), video_id)
  42         host = self._download_json('http://streaming-loadbalancer.ur.se/loadbalancer.json', video_id)['redirect']
  43
  44         formats = []
  45         for quality_attr, quality, preference in (('', 'sd', 0), ('_hd', 'hd', 1)):
  46             file_http = urplayer_data.get('file_http' + quality_attr) or urplayer_data.get('file_http_sub' + quality_attr)
  47             if file_http:
  48                 formats.extend(self._extract_wowza_formats(
  49                     'http://%s/%splaylist.m3u8' % (host, file_http), video_id, skip_protocols=['rtmp', 'rtsp']))
  50         self._sort_formats(formats)
  51
  52         subtitles = {}
  53         for subtitle in urplayer_data.get('subtitles', []):
  54             subtitle_url = subtitle.get('file')
  55             kind = subtitle.get('kind')
  56             if not subtitle_url or (kind and kind != 'captions'):
  57                 continue
  58             subtitles.setdefault(subtitle.get('label', 'Svenska'), []).append({
  59                 'url': subtitle_url,
  60             })
  61
  62         return {
  63             'id': video_id,
  64             'title': urplayer_data['title'],
  65             'description': self._og_search_description(webpage),
  66             'thumbnail': urplayer_data.get('image'),
  67             'timestamp': unified_timestamp(self._html_search_meta(('uploadDate', 'schema:uploadDate'), webpage, 'timestamp')),
  68             'series': urplayer_data.get('series_title'),
  69             'subtitles': subtitles,
  70             'formats': formats,
  71         }