[douyutv] Improve and extract all formats
[youtube-dl] / youtube_dl / extractor / douyutv.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 from .common import InfoExtractor
5 from ..utils import ExtractorError
6
7
8 class DouyuTVIE(InfoExtractor):
9     _VALID_URL = r'http://(?:www\.)?douyutv\.com/(?P<id>[A-Za-z0-9]+)'
10     _TEST = {
11         'url': 'http://www.douyutv.com/iseven',
12         'info_dict': {
13             'id': 'iseven',
14             'ext': 'flv',
15             'title': 're:^清晨醒脑!T-ara根本停不下来! [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
16             'description': 'md5:9e525642c25a0a24302869937cf69d17',
17             'thumbnail': 're:^https?://.*\.jpg$',
18             'uploader': '7师傅',
19             'uploader_id': '431925',
20             'is_live': True,
21         },
22         'params': {
23             'skip_download': True,
24         }
25     }
26
27     def _real_extract(self, url):
28         video_id = self._match_id(url)
29
30         config = self._download_json(
31             'http://www.douyutv.com/api/client/room/%s' % video_id, video_id)
32
33         data = config['data']
34
35         error_code = config.get('error', 0)
36         show_status = data.get('show_status')
37         if error_code is not 0:
38             raise ExtractorError(
39                 'Server reported error %i' % error_code, expected=True)
40
41         # 1 = live, 2 = offline
42         if show_status == '2':
43             raise ExtractorError(
44                 'Live stream is offline', expected=True)
45
46         base_url = data['rtmp_url']
47         live_path = data['rtmp_live']
48
49         title = self._live_title(data['room_name'])
50         description = data.get('show_details')
51         thumbnail = data.get('room_src')
52
53         uploader = data.get('nickname')
54         uploader_id = data.get('owner_uid')
55
56         multi_formats = data.get('rtmp_multi_bitrate')
57         if not isinstance(multi_formats, dict):
58             multi_formats = {}
59         multi_formats['live'] = live_path
60
61         formats = [{
62             'url': '%s/%s' % (base_url, format_path),
63             'format_id': format_id,
64             'preference': 1 if format_id == 'live' else 0,
65         } for format_id, format_path in multi_formats.items()]
66         self._sort_formats(formats)
67
68         return {
69             'id': video_id,
70             'title': title,
71             'description': description,
72             'thumbnail': thumbnail,
73             'uploader': uploader,
74             'uploader_id': uploader_id,
75             'formats': formats,
76             'is_live': True,
77         }