youtube_dl/extractor/drtv.py

   1 from __future__ import unicode_literals
   2
   3 import re
   4
   5 from .subtitles import SubtitlesInfoExtractor
   6 from .common import ExtractorError
   7 from ..utils import parse_iso8601
   8
   9
  10 class DRTVIE(SubtitlesInfoExtractor):
  11     _VALID_URL = r'http://(?:www\.)?dr\.dk/tv/se/(?:[^/]+/)+(?P<id>[\da-z-]+)(?:[/#?]|$)'
  12
  13     _TEST = {
  14         'url': 'http://www.dr.dk/tv/se/partiets-mand/partiets-mand-7-8',
  15         'md5': '4a7e1dd65cdb2643500a3f753c942f25',
  16         'info_dict': {
  17             'id': 'partiets-mand-7-8',
  18             'ext': 'mp4',
  19             'title': 'Partiets mand (7:8)',
  20             'description': 'md5:a684b90a8f9336cd4aab94b7647d7862',
  21             'timestamp': 1403047940,
  22             'upload_date': '20140617',
  23             'duration': 1299.040,
  24         },
  25     }
  26
  27     def _real_extract(self, url):
  28         mobj = re.match(self._VALID_URL, url)
  29         video_id = mobj.group('id')
  30
  31         programcard = self._download_json(
  32             'http://www.dr.dk/mu/programcard/expanded/%s' % video_id, video_id, 'Downloading video JSON')
  33
  34         data = programcard['Data'][0]
  35
  36         title = data['Title']
  37         description = data['Description']
  38         timestamp = parse_iso8601(data['CreatedTime'][:-5])
  39
  40         thumbnail = None
  41         duration = None
  42
  43         restricted_to_denmark = False
  44
  45         formats = []
  46         subtitles = {}
  47
  48         for asset in data['Assets']:
  49             if asset['Kind'] == 'Image':
  50                 thumbnail = asset['Uri']
  51             elif asset['Kind'] == 'VideoResource':
  52                 duration = asset['DurationInMilliseconds'] / 1000.0
  53                 restricted_to_denmark = asset['RestrictedToDenmark']
  54                 for link in asset['Links']:
  55                     target = link['Target']
  56                     uri = link['Uri']
  57                     formats.append({
  58                         'url': uri + '?hdcore=3.3.0&plugin=aasp-3.3.0.99.43' if target == 'HDS' else uri,
  59                         'format_id': target,
  60                         'ext': link['FileFormat'],
  61                         'preference': -1 if target == 'HDS' else -2,
  62                     })
  63                 subtitles_list = asset.get('SubtitlesList')
  64                 if isinstance(subtitles_list, list):
  65                     LANGS = {
  66                         'Danish': 'dk',
  67                     }
  68                     for subs in subtitles_list:
  69                         lang = subs['Language']
  70                         subtitles[LANGS.get(lang, lang)] = subs['Uri']
  71
  72         if not formats and restricted_to_denmark:
  73             raise ExtractorError(
  74                 'Unfortunately, DR is not allowed to show this program outside Denmark.', expected=True)
  75
  76         self._sort_formats(formats)
  77
  78         if self._downloader.params.get('listsubtitles', False):
  79             self._list_available_subtitles(video_id, subtitles)
  80             return
  81
  82         return {
  83             'id': video_id,
  84             'title': title,
  85             'description': description,
  86             'thumbnail': thumbnail,
  87             'timestamp': timestamp,
  88             'duration': duration,
  89             'formats': formats,
  90             'subtitles': self.extract_subtitles(video_id, subtitles),
  91         }