Merge branch 'theintercept' of https://github.com/bit/youtube-dl into bit-theintercept
[youtube-dl] / youtube_dl / extractor / periscope.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 from .common import InfoExtractor
5 from ..utils import parse_iso8601
6
7
8 class PeriscopeIE(InfoExtractor):
9     IE_DESC = 'Periscope'
10     _VALID_URL = r'https?://(?:www\.)?periscope\.tv/[^/]+/(?P<id>[^/?#]+)'
11     # Alive example URLs can be found here http://onperiscope.com/
12     _TESTS = [{
13         'url': 'https://www.periscope.tv/w/aJUQnjY3MjA3ODF8NTYxMDIyMDl2zCg2pECBgwTqRpQuQD352EMPTKQjT4uqlM3cgWFA-g==',
14         'md5': '65b57957972e503fcbbaeed8f4fa04ca',
15         'info_dict': {
16             'id': '56102209',
17             'ext': 'mp4',
18             'title': 'Bec Boop - ๐Ÿš โœˆ๏ธ๐Ÿ‡ฌ๐Ÿ‡ง Fly above #London in Emirates Air Line cable car at night ๐Ÿ‡ฌ๐Ÿ‡งโœˆ๏ธ๐Ÿš  #BoopScope ๐ŸŽ€๐Ÿ’—',
19             'timestamp': 1438978559,
20             'upload_date': '20150807',
21             'uploader': 'Bec Boop',
22             'uploader_id': '1465763',
23         },
24         'skip': 'Expires in 24 hours',
25     }, {
26         'url': 'https://www.periscope.tv/w/1ZkKzPbMVggJv',
27         'only_matching': True,
28     }, {
29         'url': 'https://www.periscope.tv/bastaakanoggano/1OdKrlkZZjOJX',
30         'only_matching': True,
31     }]
32
33     def _call_api(self, method, value):
34         return self._download_json(
35             'https://api.periscope.tv/api/v2/%s?broadcast_id=%s' % (method, value), value)
36
37     def _real_extract(self, url):
38         token = self._match_id(url)
39
40         broadcast_data = self._call_api('getBroadcastPublic', token)
41         broadcast = broadcast_data['broadcast']
42         status = broadcast['status']
43
44         uploader = broadcast.get('user_display_name') or broadcast_data.get('user', {}).get('display_name')
45         uploader_id = broadcast.get('user_id') or broadcast_data.get('user', {}).get('id')
46
47         title = '%s - %s' % (uploader, status) if uploader else status
48         state = broadcast.get('state').lower()
49         if state == 'running':
50             title = self._live_title(title)
51         timestamp = parse_iso8601(broadcast.get('created_at'))
52
53         thumbnails = [{
54             'url': broadcast[image],
55         } for image in ('image_url', 'image_url_small') if broadcast.get(image)]
56
57         stream = self._call_api('getAccessPublic', token)
58
59         formats = []
60         for format_id in ('replay', 'rtmp', 'hls', 'https_hls'):
61             video_url = stream.get(format_id + '_url')
62             if not video_url:
63                 continue
64             f = {
65                 'url': video_url,
66                 'ext': 'flv' if format_id == 'rtmp' else 'mp4',
67             }
68             if format_id != 'rtmp':
69                 f['protocol'] = 'm3u8_native' if state == 'ended' else 'm3u8'
70             formats.append(f)
71         self._sort_formats(formats)
72
73         return {
74             'id': broadcast.get('id') or token,
75             'title': title,
76             'timestamp': timestamp,
77             'uploader': uploader,
78             'uploader_id': uploader_id,
79             'thumbnails': thumbnails,
80             'formats': formats,
81         }