2 from __future__ import unicode_literals
7 from .common import InfoExtractor
8 from ..compat import compat_str
18 class YandexMusicBaseIE(InfoExtractor):
20 def _handle_error(response):
21 if isinstance(response, dict):
22 error = response.get('error')
24 raise ExtractorError(error, expected=True)
26 def _download_webpage(self, *args, **kwargs):
27 webpage = super(YandexMusicBaseIE, self)._download_webpage(*args, **kwargs)
28 if 'Нам очень жаль, но запросы, поступившие с вашего IP-адреса, похожи на автоматические.' in webpage:
29 raise ExtractorError('Blocked by YandexMusic', expected=True)
32 def _download_json(self, *args, **kwargs):
33 response = super(YandexMusicBaseIE, self)._download_json(*args, **kwargs)
34 self._handle_error(response)
38 class YandexMusicTrackIE(YandexMusicBaseIE):
39 IE_NAME = 'yandexmusic:track'
40 IE_DESC = 'Яндекс.Музыка - Трек'
41 _VALID_URL = r'https?://music\.yandex\.(?:ru|kz|ua|by)/album/(?P<album_id>\d+)/track/(?P<id>\d+)'
44 'url': 'http://music.yandex.ru/album/540508/track/4878838',
45 'md5': 'f496818aa2f60b6c0062980d2e00dc20',
49 'title': 'Carlo Ambrosio & Fabio Di Bari, Carlo Ambrosio - Gypsy Eyes 1',
52 'track': 'Gypsy Eyes 1',
53 'album': 'Gypsy Soul',
54 'album_artist': 'Carlo Ambrosio',
55 'artist': 'Carlo Ambrosio & Fabio Di Bari, Carlo Ambrosio',
56 'release_year': '2009',
58 'skip': 'Travis CI servers blocked by YandexMusic',
61 def _get_track_url(self, storage_dir, track_id):
62 data = self._download_json(
63 'http://music.yandex.ru/api/v1.5/handlers/api-jsonp.jsx?action=getTrackSrc&p=download-info/%s'
65 track_id, 'Downloading track location JSON')
67 key = hashlib.md5(('XGRlBW9FXlekgbPrRHuSiA' + data['path'][1:] + data['s']).encode('utf-8')).hexdigest()
68 storage = storage_dir.split('.')
70 return ('http://%s/get-mp3/%s/%s?track-id=%s&from=service-10-track&similarities-experiment=default'
71 % (data['host'], key, data['ts'] + data['path'], storage[1]))
73 def _get_track_info(self, track):
75 cover_uri = track.get('albums', [{}])[0].get('coverUri')
77 thumbnail = cover_uri.replace('%%', 'orig')
78 if not thumbnail.startswith('http'):
79 thumbnail = 'http://' + thumbnail
81 track_title = track['title']
85 'url': self._get_track_url(track['storageDir'], track['id']),
86 'filesize': int_or_none(track.get('fileSize')),
87 'duration': float_or_none(track.get('durationMs'), 1000),
88 'thumbnail': thumbnail,
92 def extract_artist(artist_list):
93 if artist_list and isinstance(artist_list, list):
94 artists_names = [a['name'] for a in artist_list if a.get('name')]
96 return ', '.join(artists_names)
98 albums = track.get('albums')
99 if albums and isinstance(albums, list):
101 if isinstance(album, dict):
102 year = album.get('year')
104 'album': album.get('title'),
105 'album_artist': extract_artist(album.get('artists')),
106 'release_year': compat_str(year) if year else None,
109 track_artist = extract_artist(track.get('artists'))
112 'artist': track_artist,
113 'title': '%s - %s' % (track_artist, track_title),
116 track_info['title'] = track_title
119 def _real_extract(self, url):
120 mobj = re.match(self._VALID_URL, url)
121 album_id, track_id = mobj.group('album_id'), mobj.group('id')
123 track = self._download_json(
124 'http://music.yandex.ru/handlers/track.jsx?track=%s:%s' % (track_id, album_id),
125 track_id, 'Downloading track JSON')['track']
127 return self._get_track_info(track)
130 class YandexMusicPlaylistBaseIE(YandexMusicBaseIE):
131 def _build_playlist(self, tracks):
134 'http://music.yandex.ru/album/%s/track/%s' % (track['albums'][0]['id'], track['id']))
135 for track in tracks if track.get('albums') and isinstance(track.get('albums'), list)]
138 class YandexMusicAlbumIE(YandexMusicPlaylistBaseIE):
139 IE_NAME = 'yandexmusic:album'
140 IE_DESC = 'Яндекс.Музыка - Альбом'
141 _VALID_URL = r'https?://music\.yandex\.(?:ru|kz|ua|by)/album/(?P<id>\d+)/?(\?|$)'
144 'url': 'http://music.yandex.ru/album/540508',
147 'title': 'Carlo Ambrosio - Gypsy Soul (2009)',
149 'playlist_count': 50,
150 'skip': 'Travis CI servers blocked by YandexMusic',
153 def _real_extract(self, url):
154 album_id = self._match_id(url)
156 album = self._download_json(
157 'http://music.yandex.ru/handlers/album.jsx?album=%s' % album_id,
158 album_id, 'Downloading album JSON')
160 entries = self._build_playlist(album['volumes'][0])
162 title = '%s - %s' % (album['artists'][0]['name'], album['title'])
163 year = album.get('year')
165 title += ' (%s)' % year
167 return self.playlist_result(entries, compat_str(album['id']), title)
170 class YandexMusicPlaylistIE(YandexMusicPlaylistBaseIE):
171 IE_NAME = 'yandexmusic:playlist'
172 IE_DESC = 'Яндекс.Музыка - Плейлист'
173 _VALID_URL = r'https?://music\.yandex\.(?:ru|kz|ua|by)/users/[^/]+/playlists/(?P<id>\d+)'
176 'url': 'http://music.yandex.ru/users/music.partners/playlists/1245',
179 'title': 'Что слушают Enter Shikari',
180 'description': 'md5:3b9f27b0efbe53f2ee1e844d07155cc9',
183 'skip': 'Travis CI servers blocked by YandexMusic',
185 # playlist exceeding the limit of 150 tracks shipped with webpage (see
186 # https://github.com/rg3/youtube-dl/issues/6666)
187 'url': 'https://music.yandex.ru/users/ya.playlist/playlists/1036',
190 'title': 'Музыка 90-х',
192 'playlist_count': 310,
193 'skip': 'Travis CI servers blocked by YandexMusic',
196 def _real_extract(self, url):
197 playlist_id = self._match_id(url)
199 webpage = self._download_webpage(url, playlist_id)
201 mu = self._parse_json(
203 r'var\s+Mu\s*=\s*({.+?});\s*</script>', webpage, 'player'),
206 playlist = mu['pageData']['playlist']
207 tracks, track_ids = playlist['tracks'], playlist['trackIds']
209 # tracks dictionary shipped with webpage is limited to 150 tracks,
210 # missing tracks should be retrieved manually.
211 if len(tracks) < len(track_ids):
212 present_track_ids = set([compat_str(track['id']) for track in tracks if track.get('id')])
213 missing_track_ids = set(map(compat_str, track_ids)) - set(present_track_ids)
214 request = sanitized_Request(
215 'https://music.yandex.ru/handlers/track-entries.jsx',
217 'entries': ','.join(missing_track_ids),
218 'lang': mu.get('settings', {}).get('lang', 'en'),
219 'external-domain': 'music.yandex.ru',
220 'overembed': 'false',
221 'sign': mu.get('authData', {}).get('user', {}).get('sign'),
224 request.add_header('Referer', url)
225 request.add_header('X-Requested-With', 'XMLHttpRequest')
227 missing_tracks = self._download_json(
228 request, playlist_id, 'Downloading missing tracks JSON', fatal=False)
230 tracks.extend(missing_tracks)
232 return self.playlist_result(
233 self._build_playlist(tracks),
234 compat_str(playlist_id),
235 playlist['title'], playlist.get('description'))