Merge pull request #9400 from inondle/master
[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     IE_NAME = 'periscope'
11     _VALID_URL = r'https?://(?:www\.)?periscope\.tv/[^/]+/(?P<id>[^/?#]+)'
12     # Alive example URLs can be found here http://onperiscope.com/
13     _TESTS = [{
14         'url': 'https://www.periscope.tv/w/aJUQnjY3MjA3ODF8NTYxMDIyMDl2zCg2pECBgwTqRpQuQD352EMPTKQjT4uqlM3cgWFA-g==',
15         'md5': '65b57957972e503fcbbaeed8f4fa04ca',
16         'info_dict': {
17             'id': '56102209',
18             'ext': 'mp4',
19             '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’—',
20             'timestamp': 1438978559,
21             'upload_date': '20150807',
22             'uploader': 'Bec Boop',
23             'uploader_id': '1465763',
24         },
25         'skip': 'Expires in 24 hours',
26     }, {
27         'url': 'https://www.periscope.tv/w/1ZkKzPbMVggJv',
28         'only_matching': True,
29     }, {
30         'url': 'https://www.periscope.tv/bastaakanoggano/1OdKrlkZZjOJX',
31         'only_matching': True,
32     }]
33
34     def _call_api(self, method, value):
35         return self._download_json(
36             'https://api.periscope.tv/api/v2/%s?broadcast_id=%s' % (method, value), value)
37
38     def _real_extract(self, url):
39         token = self._match_id(url)
40
41         broadcast_data = self._call_api('getBroadcastPublic', token)
42         broadcast = broadcast_data['broadcast']
43         status = broadcast['status']
44
45         uploader = broadcast.get('user_display_name') or broadcast_data.get('user', {}).get('display_name')
46         uploader_id = broadcast.get('user_id') or broadcast_data.get('user', {}).get('id')
47
48         title = '%s - %s' % (uploader, status) if uploader else status
49         state = broadcast.get('state').lower()
50         if state == 'running':
51             title = self._live_title(title)
52         timestamp = parse_iso8601(broadcast.get('created_at'))
53
54         thumbnails = [{
55             'url': broadcast[image],
56         } for image in ('image_url', 'image_url_small') if broadcast.get(image)]
57
58         stream = self._call_api('getAccessPublic', token)
59
60         formats = []
61         for format_id in ('replay', 'rtmp', 'hls', 'https_hls'):
62             video_url = stream.get(format_id + '_url')
63             if not video_url:
64                 continue
65             f = {
66                 'url': video_url,
67                 'ext': 'flv' if format_id == 'rtmp' else 'mp4',
68             }
69             if format_id != 'rtmp':
70                 f['protocol'] = 'm3u8_native' if state == 'ended' else 'm3u8'
71             formats.append(f)
72         self._sort_formats(formats)
73
74         return {
75             'id': broadcast.get('id') or token,
76             'title': title,
77             'timestamp': timestamp,
78             'uploader': uploader,
79             'uploader_id': uploader_id,
80             'thumbnails': thumbnails,
81             'formats': formats,
82         }
83
84
85 class PeriscopeUserIE(InfoExtractor):
86     _VALID_URL = r'https?://www\.periscope\.tv/(?P<id>[^/]+)/?$'
87     IE_DESC = 'Periscope user videos'
88     IE_NAME = 'periscope:user'
89
90     _TEST = {
91         'url': 'https://www.periscope.tv/LularoeHusbandMike/',
92         'info_dict': {
93             'id': 'LularoeHusbandMike',
94             'title': 'LULAROE HUSBAND MIKE',
95         },
96         # Periscope only shows videos in the last 24 hours, so it's possible to
97         # get 0 videos
98         'playlist_mincount': 0,
99     }
100
101     def _real_extract(self, url):
102         user_id = self._match_id(url)
103
104         webpage = self._download_webpage(url, user_id)
105
106         broadcast_data = self._parse_json(self._html_search_meta(
107             'broadcast-data', webpage, default='{}'), user_id)
108         username = broadcast_data.get('user', {}).get('display_name')
109         user_broadcasts = self._parse_json(
110             self._html_search_meta('user-broadcasts', webpage, default='{}'),
111             user_id)
112
113         entries = [
114             self.url_result(
115                 'https://www.periscope.tv/%s/%s' % (user_id, broadcast['id']))
116             for broadcast in user_broadcasts.get('broadcasts', [])]
117
118         return self.playlist_result(entries, user_id, username)