[drtv] Improve _VALID_URL (#4814)
[youtube-dl] / youtube_dl / extractor / drtv.py
1 from __future__ import unicode_literals
2
3 from .subtitles import SubtitlesInfoExtractor
4 from .common import ExtractorError
5 from ..utils import parse_iso8601
6
7
8 class DRTVIE(SubtitlesInfoExtractor):
9     _VALID_URL = r'https?://(?:www\.)?dr\.dk/tv/se/(?:[^/]+/)*(?P<id>[\da-z-]+)(?:[/#?]|$)'
10
11     _TEST = {
12         'url': 'http://www.dr.dk/tv/se/partiets-mand/partiets-mand-7-8',
13         'md5': '4a7e1dd65cdb2643500a3f753c942f25',
14         'info_dict': {
15             'id': 'partiets-mand-7-8',
16             'ext': 'mp4',
17             'title': 'Partiets mand (7:8)',
18             'description': 'md5:a684b90a8f9336cd4aab94b7647d7862',
19             'timestamp': 1403047940,
20             'upload_date': '20140617',
21             'duration': 1299.040,
22         },
23     }
24
25     def _real_extract(self, url):
26         video_id = self._match_id(url)
27
28         programcard = self._download_json(
29             'http://www.dr.dk/mu/programcard/expanded/%s' % video_id, video_id, 'Downloading video JSON')
30
31         data = programcard['Data'][0]
32
33         title = data['Title']
34         description = data['Description']
35         timestamp = parse_iso8601(data['CreatedTime'])
36
37         thumbnail = None
38         duration = None
39
40         restricted_to_denmark = False
41
42         formats = []
43         subtitles = {}
44
45         for asset in data['Assets']:
46             if asset['Kind'] == 'Image':
47                 thumbnail = asset['Uri']
48             elif asset['Kind'] == 'VideoResource':
49                 duration = asset['DurationInMilliseconds'] / 1000.0
50                 restricted_to_denmark = asset['RestrictedToDenmark']
51                 spoken_subtitles = asset['Target'] == 'SpokenSubtitles'
52                 for link in asset['Links']:
53                     target = link['Target']
54                     uri = link['Uri']
55                     format_id = target
56                     preference = -1 if target == 'HDS' else -2
57                     if spoken_subtitles:
58                         preference -= 2
59                         format_id += '-spoken-subtitles'
60                     formats.append({
61                         'url': uri + '?hdcore=3.3.0&plugin=aasp-3.3.0.99.43' if target == 'HDS' else uri,
62                         'format_id': format_id,
63                         'ext': link['FileFormat'],
64                         'preference': preference,
65                     })
66                 subtitles_list = asset.get('SubtitlesList')
67                 if isinstance(subtitles_list, list):
68                     LANGS = {
69                         'Danish': 'dk',
70                     }
71                     for subs in subtitles_list:
72                         lang = subs['Language']
73                         subtitles[LANGS.get(lang, lang)] = subs['Uri']
74
75         if not formats and restricted_to_denmark:
76             raise ExtractorError(
77                 'Unfortunately, DR is not allowed to show this program outside Denmark.', expected=True)
78
79         self._sort_formats(formats)
80
81         if self._downloader.params.get('listsubtitles', False):
82             self._list_available_subtitles(video_id, subtitles)
83             return
84
85         return {
86             'id': video_id,
87             'title': title,
88             'description': description,
89             'thumbnail': thumbnail,
90             'timestamp': timestamp,
91             'duration': duration,
92             'formats': formats,
93             'subtitles': self.extract_subtitles(video_id, subtitles),
94         }