[yandexmusic] Defer link resolve till actual download time to prevent link expiry...
[youtube-dl] / youtube_dl / extractor / yandexmusic.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 import re
5 import hashlib
6
7 from .common import InfoExtractor
8 from ..compat import compat_str
9 from ..utils import (
10     int_or_none,
11     float_or_none,
12 )
13
14
15 class YandexMusicBaseIE(InfoExtractor):
16     def _get_track_url(self, storage_dir, track_id):
17         data = self._download_json(
18             'http://music.yandex.ru/api/v1.5/handlers/api-jsonp.jsx?action=getTrackSrc&p=download-info/%s'
19             % storage_dir,
20             track_id, 'Downloading track location JSON')
21
22         key = hashlib.md5(('XGRlBW9FXlekgbPrRHuSiA' + data['path'][1:] + data['s']).encode('utf-8')).hexdigest()
23         storage = storage_dir.split('.')
24
25         return ('http://%s/get-mp3/%s/%s?track-id=%s&from=service-10-track&similarities-experiment=default'
26                 % (data['host'], key, data['ts'] + data['path'], storage[1]))
27
28     def _get_track_info(self, track):
29         return {
30             'id': track['id'],
31             'ext': 'mp3',
32             'url': self._get_track_url(track['storageDir'], track['id']),
33             'title': '%s - %s' % (track['artists'][0]['name'], track['title']),
34             'filesize': int_or_none(track.get('fileSize')),
35             'duration': float_or_none(track.get('durationMs'), 1000),
36         }
37
38
39 class YandexMusicTrackIE(YandexMusicBaseIE):
40     IE_NAME = 'yandexmusic:track'
41     IE_DESC = 'Яндекс.Музыка - Трек'
42     _VALID_URL = r'https?://music\.yandex\.(?:ru|kz|ua|by)/album/(?P<album_id>\d+)/track/(?P<id>\d+)'
43
44     _TEST = {
45         'url': 'http://music.yandex.ru/album/540508/track/4878838',
46         'md5': 'f496818aa2f60b6c0062980d2e00dc20',
47         'info_dict': {
48             'id': '4878838',
49             'ext': 'mp3',
50             'title': 'Carlo Ambrosio - Gypsy Eyes 1',
51             'filesize': 4628061,
52             'duration': 193.04,
53         }
54     }
55
56     def _real_extract(self, url):
57         mobj = re.match(self._VALID_URL, url)
58         album_id, track_id = mobj.group('album_id'), mobj.group('id')
59
60         track = self._download_json(
61             'http://music.yandex.ru/handlers/track.jsx?track=%s:%s' % (track_id, album_id),
62             track_id, 'Downloading track JSON')['track']
63
64         return self._get_track_info(track)
65
66
67 class YandexMusicPlaylistBaseIE(InfoExtractor):
68     def _build_playlist(self, tracks):
69         return [
70             self.url_result(
71                 'http://music.yandex.ru/album/%s/track/%s' % (track['albums'][0]['id'], track['id']))
72             for track in tracks]
73
74
75 class YandexMusicAlbumIE(YandexMusicPlaylistBaseIE):
76     IE_NAME = 'yandexmusic:album'
77     IE_DESC = 'Яндекс.Музыка - Альбом'
78     _VALID_URL = r'https?://music\.yandex\.(?:ru|kz|ua|by)/album/(?P<id>\d+)/?(\?|$)'
79
80     _TEST = {
81         'url': 'http://music.yandex.ru/album/540508',
82         'info_dict': {
83             'id': '540508',
84             'title': 'Carlo Ambrosio - Gypsy Soul (2009)',
85         },
86         'playlist_count': 50,
87     }
88
89     def _real_extract(self, url):
90         album_id = self._match_id(url)
91
92         album = self._download_json(
93             'http://music.yandex.ru/handlers/album.jsx?album=%s' % album_id,
94             album_id, 'Downloading album JSON')
95
96         entries = self._build_playlist(album['volumes'][0])
97
98         title = '%s - %s' % (album['artists'][0]['name'], album['title'])
99         year = album.get('year')
100         if year:
101             title += ' (%s)' % year
102
103         return self.playlist_result(entries, compat_str(album['id']), title)
104
105
106 class YandexMusicPlaylistIE(YandexMusicPlaylistBaseIE):
107     IE_NAME = 'yandexmusic:playlist'
108     IE_DESC = 'Яндекс.Музыка - Плейлист'
109     _VALID_URL = r'https?://music\.yandex\.(?:ru|kz|ua|by)/users/[^/]+/playlists/(?P<id>\d+)'
110
111     _TEST = {
112         'url': 'http://music.yandex.ru/users/music.partners/playlists/1245',
113         'info_dict': {
114             'id': '1245',
115             'title': 'Что слушают Enter Shikari',
116             'description': 'md5:3b9f27b0efbe53f2ee1e844d07155cc9',
117         },
118         'playlist_count': 6,
119     }
120
121     def _real_extract(self, url):
122         playlist_id = self._match_id(url)
123
124         webpage = self._download_webpage(url, playlist_id)
125
126         playlist = self._parse_json(
127             self._search_regex(
128                 r'var\s+Mu\s*=\s*({.+?});\s*</script>', webpage, 'player'),
129             playlist_id)['pageData']['playlist']
130
131         return self.playlist_result(
132             self._build_playlist(playlist['tracks']),
133             compat_str(playlist_id),
134             playlist['title'], playlist.get('description'))