[drtv] Check for unavailable videos
[youtube-dl] / 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         if '>Programmet er ikke længere tilgængeligt' in webpage:
30             raise ExtractorError(
31                 'Video %s is not available' % video_id, expected=True)
32
33         video_id = self._search_regex(
34             r'data-(?:material-identifier|episode-slug)="([^"]+)"',
35             webpage, 'video id')
36
37         programcard = self._download_json(
38             'http://www.dr.dk/mu/programcard/expanded/%s' % video_id,
39             video_id, 'Downloading video JSON')
40         data = programcard['Data'][0]
41
42         title = data['Title']
43         description = data['Description']
44         timestamp = parse_iso8601(data['CreatedTime'])
45
46         thumbnail = None
47         duration = None
48
49         restricted_to_denmark = False
50
51         formats = []
52         subtitles = {}
53
54         for asset in data['Assets']:
55             if asset['Kind'] == 'Image':
56                 thumbnail = asset['Uri']
57             elif asset['Kind'] == 'VideoResource':
58                 duration = asset['DurationInMilliseconds'] / 1000.0
59                 restricted_to_denmark = asset['RestrictedToDenmark']
60                 spoken_subtitles = asset['Target'] == 'SpokenSubtitles'
61                 for link in asset['Links']:
62                     target = link['Target']
63                     uri = link['Uri']
64                     format_id = target
65                     preference = -1 if target == 'HDS' else -2
66                     if spoken_subtitles:
67                         preference -= 2
68                         format_id += '-spoken-subtitles'
69                     formats.append({
70                         'url': uri + '?hdcore=3.3.0&plugin=aasp-3.3.0.99.43' if target == 'HDS' else uri,
71                         'format_id': format_id,
72                         'ext': link['FileFormat'],
73                         'preference': preference,
74                     })
75                 subtitles_list = asset.get('SubtitlesList')
76                 if isinstance(subtitles_list, list):
77                     LANGS = {
78                         'Danish': 'dk',
79                     }
80                     for subs in subtitles_list:
81                         lang = subs['Language']
82                         subtitles[LANGS.get(lang, lang)] = [{'url': subs['Uri'], 'ext': 'vtt'}]
83
84         if not formats and restricted_to_denmark:
85             raise ExtractorError(
86                 'Unfortunately, DR is not allowed to show this program outside Denmark.', expected=True)
87
88         self._sort_formats(formats)
89
90         return {
91             'id': video_id,
92             'title': title,
93             'description': description,
94             'thumbnail': thumbnail,
95             'timestamp': timestamp,
96             'duration': duration,
97             'formats': formats,
98             'subtitles': subtitles,
99         }