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

   1 from __future__ import unicode_literals
   2
   3 from .common import InfoExtractor, ExtractorError
   4 from ..utils import parse_iso8601
   5
   6
   7 class DRTVIE(InfoExtractor):
   8     _VALID_URL = r'https?://(?:www\.)?dr\.dk/tv/se/(?:[^/]+/)*(?P<id>[\da-z-]+)(?:[/#?]|$)'
   9
  10     _TEST = {
  11         'url': 'http://www.dr.dk/tv/se/partiets-mand/partiets-mand-7-8',
  12         'md5': '4a7e1dd65cdb2643500a3f753c942f25',
  13         'info_dict': {
  14             'id': 'partiets-mand-7-8',
  15             'ext': 'mp4',
  16             'title': 'Partiets mand (7:8)',
  17             'description': 'md5:a684b90a8f9336cd4aab94b7647d7862',
  18             'timestamp': 1403047940,
  19             'upload_date': '20140617',
  20             'duration': 1299.040,
  21         },
  22     }
  23
  24     def _real_extract(self, url):
  25         video_id = self._match_id(url)
  26
  27         webpage = self._download_webpage(url, video_id)
  28
  29         video_id = self._search_regex(
  30             r'data-(?:material-identifier|episode-slug)="([^"]+)"',
  31             webpage, 'video id')
  32
  33         programcard = self._download_json(
  34             'http://www.dr.dk/mu/programcard/expanded/%s' % video_id,
  35             video_id, 'Downloading video JSON')
  36         data = programcard['Data'][0]
  37
  38         title = data['Title']
  39         description = data['Description']
  40         timestamp = parse_iso8601(data['CreatedTime'])
  41
  42         thumbnail = None
  43         duration = None
  44
  45         restricted_to_denmark = False
  46
  47         formats = []
  48         subtitles = {}
  49
  50         for asset in data['Assets']:
  51             if asset['Kind'] == 'Image':
  52                 thumbnail = asset['Uri']
  53             elif asset['Kind'] == 'VideoResource':
  54                 duration = asset['DurationInMilliseconds'] / 1000.0
  55                 restricted_to_denmark = asset['RestrictedToDenmark']
  56                 spoken_subtitles = asset['Target'] == 'SpokenSubtitles'
  57                 for link in asset['Links']:
  58                     target = link['Target']
  59                     uri = link['Uri']
  60                     format_id = target
  61                     preference = -1 if target == 'HDS' else -2
  62                     if spoken_subtitles:
  63                         preference -= 2
  64                         format_id += '-spoken-subtitles'
  65                     formats.append({
  66                         'url': uri + '?hdcore=3.3.0&plugin=aasp-3.3.0.99.43' if target == 'HDS' else uri,
  67                         'format_id': format_id,
  68                         'ext': link['FileFormat'],
  69                         'preference': preference,
  70                     })
  71                 subtitles_list = asset.get('SubtitlesList')
  72                 if isinstance(subtitles_list, list):
  73                     LANGS = {
  74                         'Danish': 'dk',
  75                     }
  76                     for subs in subtitles_list:
  77                         lang = subs['Language']
  78                         subtitles[LANGS.get(lang, lang)] = [{'url': subs['Uri'], 'ext': 'vtt'}]
  79
  80         if not formats and restricted_to_denmark:
  81             raise ExtractorError(
  82                 'Unfortunately, DR is not allowed to show this program outside Denmark.', expected=True)
  83
  84         self._sort_formats(formats)
  85
  86         return {
  87             'id': video_id,
  88             'title': title,
  89             'description': description,
  90             'thumbnail': thumbnail,
  91             'timestamp': timestamp,
  92             'duration': duration,
  93             'formats': formats,
  94             'subtitles': subtitles,
  95         }