Merge branch 'oppetarkiv' of https://github.com/thc202/youtube-dl into thc202-oppetarkiv
[youtube-dl] / youtube_dl / extractor / oppetarkiv.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 from .common import InfoExtractor
5 from ..utils import (
6     determine_ext,
7 )
8
9
10 class OppetArkivIE(InfoExtractor):
11     _VALID_URL = r'https?://(?:www\.)?oppetarkiv.se/video/(?P<id>[0-9]+)'
12     _TEST = {
13         'url': 'http://www.oppetarkiv.se/video/1058509/rederiet-sasong-1-avsnitt-1-av-318',
14         'md5': '7b95ca9bedeead63012b2d7c3992c28f',
15         'info_dict': {
16             'id': '1058509',
17             'ext': 'mp4',
18             'title': 'Farlig kryssning',
19             'duration': 2566,
20             'thumbnail': 're:^https?://.*[\.-]jpg$',
21         },
22     }
23
24     def _real_extract(self, url):
25         video_id = self._match_id(url)
26         info = self._download_json(
27             'http://www.oppetarkiv.se/video/%s?output=json' % video_id, video_id)
28
29         title = info['context']['title']
30         thumbnail = info['context'].get('thumbnailImage')
31
32         video_info = info['video']
33         formats = []
34         for vr in video_info['videoReferences']:
35             vurl = vr['url']
36             if determine_ext(vurl) == 'm3u8':
37                 formats.extend(self._extract_m3u8_formats(
38                     vurl, video_id,
39                     ext='mp4', entry_protocol='m3u8_native',
40                     m3u8_id=vr.get('playerType')))
41             else:
42                 formats.append({
43                     'format_id': vr.get('playerType'),
44                     'url': vurl,
45                 })
46         self._sort_formats(formats)
47
48         duration = video_info.get('materialLength')
49
50         return {
51             'id': video_id,
52             'title': title,
53             'formats': formats,
54             'thumbnail': thumbnail,
55             'duration': duration,
56         }