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 error = response.get('error')
23 raise ExtractorError(error, expected=True)
25 def _download_webpage(self, *args, **kwargs):
26 webpage = super(YandexMusicBaseIE, self)._download_webpage(*args, **kwargs)
27 if 'Нам очень жаль, но запросы, поступившие с вашего IP-адреса, похожи на автоматические.' in webpage:
28 raise ExtractorError('Blocked by YandexMusic', expected=True)
31 def _download_json(self, *args, **kwargs):
32 response = super(YandexMusicBaseIE, self)._download_json(*args, **kwargs)
33 self._handle_error(response)
37 class YandexMusicTrackIE(YandexMusicBaseIE):
38 IE_NAME = 'yandexmusic:track'
39 IE_DESC = 'Яндекс.Музыка - Трек'
40 _VALID_URL = r'https?://music\.yandex\.(?:ru|kz|ua|by)/album/(?P<album_id>\d+)/track/(?P<id>\d+)'
43 'url': 'http://music.yandex.ru/album/540508/track/4878838',
44 'md5': 'f496818aa2f60b6c0062980d2e00dc20',
48 'title': 'Carlo Ambrosio & Fabio Di Bari, Carlo Ambrosio - Gypsy Eyes 1',
51 'track': 'Gypsy Eyes 1',
52 'album': 'Gypsy Soul',
53 'album_artist': 'Carlo Ambrosio',
54 'artist': 'Carlo Ambrosio & Fabio Di Bari, Carlo Ambrosio',
55 'release_year': '2009',
57 'skip': 'Travis CI servers blocked by YandexMusic',
60 def _get_track_url(self, storage_dir, track_id):
61 data = self._download_json(
62 'http://music.yandex.ru/api/v1.5/handlers/api-jsonp.jsx?action=getTrackSrc&p=download-info/%s'
64 track_id, 'Downloading track location JSON')
66 key = hashlib.md5(('XGRlBW9FXlekgbPrRHuSiA' + data['path'][1:] + data['s']).encode('utf-8')).hexdigest()
67 storage = storage_dir.split('.')
69 return ('http://%s/get-mp3/%s/%s?track-id=%s&from=service-10-track&similarities-experiment=default'
70 % (data['host'], key, data['ts'] + data['path'], storage[1]))
72 def _get_track_info(self, track):
74 cover_uri = track.get('albums', [{}])[0].get('coverUri')
76 thumbnail = cover_uri.replace('%%', 'orig')
77 if not thumbnail.startswith('http'):
78 thumbnail = 'http://' + thumbnail
80 track_title = track['title']
84 'url': self._get_track_url(track['storageDir'], track['id']),
85 'filesize': int_or_none(track.get('fileSize')),
86 'duration': float_or_none(track.get('durationMs'), 1000),
87 'thumbnail': thumbnail,
91 def extract_artist(artist_list):
92 if artist_list and isinstance(artist_list, list):
93 artists_names = [a['name'] for a in artist_list if a.get('name')]
95 return ', '.join(artists_names)
97 albums = track.get('albums')
98 if albums and isinstance(albums, list):
100 if isinstance(album, dict):
101 year = album.get('year')
103 'album': album.get('title'),
104 'album_artist': extract_artist(album.get('artists')),
105 'release_year': compat_str(year) if year else None,
108 track_artist = extract_artist(track.get('artists'))
111 'artist': track_artist,
112 'title': '%s - %s' % (track_artist, track_title),
115 track_info['title'] = track_title
118 def _real_extract(self, url):
119 mobj = re.match(self._VALID_URL, url)
120 album_id, track_id = mobj.group('album_id'), mobj.group('id')
122 track = self._download_json(
123 'http://music.yandex.ru/handlers/track.jsx?track=%s:%s' % (track_id, album_id),
124 track_id, 'Downloading track JSON')['track']
126 return self._get_track_info(track)
129 class YandexMusicPlaylistBaseIE(YandexMusicBaseIE):
130 def _build_playlist(self, tracks):
133 'http://music.yandex.ru/album/%s/track/%s' % (track['albums'][0]['id'], track['id']))
134 for track in tracks if track.get('albums') and isinstance(track.get('albums'), list)]
137 class YandexMusicAlbumIE(YandexMusicPlaylistBaseIE):
138 IE_NAME = 'yandexmusic:album'
139 IE_DESC = 'Яндекс.Музыка - Альбом'
140 _VALID_URL = r'https?://music\.yandex\.(?:ru|kz|ua|by)/album/(?P<id>\d+)/?(\?|$)'
143 'url': 'http://music.yandex.ru/album/540508',
146 'title': 'Carlo Ambrosio - Gypsy Soul (2009)',
148 'playlist_count': 50,
149 'skip': 'Travis CI servers blocked by YandexMusic',
152 def _real_extract(self, url):
153 album_id = self._match_id(url)
155 album = self._download_json(
156 'http://music.yandex.ru/handlers/album.jsx?album=%s' % album_id,
157 album_id, 'Downloading album JSON')
159 entries = self._build_playlist(album['volumes'][0])
161 title = '%s - %s' % (album['artists'][0]['name'], album['title'])
162 year = album.get('year')
164 title += ' (%s)' % year
166 return self.playlist_result(entries, compat_str(album['id']), title)
169 class YandexMusicPlaylistIE(YandexMusicPlaylistBaseIE):
170 IE_NAME = 'yandexmusic:playlist'
171 IE_DESC = 'Яндекс.Музыка - Плейлист'
172 _VALID_URL = r'https?://music\.yandex\.(?:ru|kz|ua|by)/users/[^/]+/playlists/(?P<id>\d+)'
175 'url': 'http://music.yandex.ru/users/music.partners/playlists/1245',
178 'title': 'Что слушают Enter Shikari',
179 'description': 'md5:3b9f27b0efbe53f2ee1e844d07155cc9',
182 'skip': 'Travis CI servers blocked by YandexMusic',
184 # playlist exceeding the limit of 150 tracks shipped with webpage (see
185 # https://github.com/rg3/youtube-dl/issues/6666)
186 'url': 'https://music.yandex.ru/users/ya.playlist/playlists/1036',
189 'title': 'Музыка 90-х',
191 'playlist_count': 310,
192 'skip': 'Travis CI servers blocked by YandexMusic',
195 def _real_extract(self, url):
196 playlist_id = self._match_id(url)
198 webpage = self._download_webpage(url, playlist_id)
200 mu = self._parse_json(
202 r'var\s+Mu\s*=\s*({.+?});\s*</script>', webpage, 'player'),
205 playlist = mu['pageData']['playlist']
206 tracks, track_ids = playlist['tracks'], playlist['trackIds']
208 # tracks dictionary shipped with webpage is limited to 150 tracks,
209 # missing tracks should be retrieved manually.
210 if len(tracks) < len(track_ids):
211 present_track_ids = set([compat_str(track['id']) for track in tracks if track.get('id')])
212 missing_track_ids = set(map(compat_str, track_ids)) - set(present_track_ids)
213 request = sanitized_Request(
214 'https://music.yandex.ru/handlers/track-entries.jsx',
216 'entries': ','.join(missing_track_ids),
217 'lang': mu.get('settings', {}).get('lang', 'en'),
218 'external-domain': 'music.yandex.ru',
219 'overembed': 'false',
220 'sign': mu.get('authData', {}).get('user', {}).get('sign'),
223 request.add_header('Referer', url)
224 request.add_header('X-Requested-With', 'XMLHttpRequest')
226 missing_tracks = self._download_json(
227 request, playlist_id, 'Downloading missing tracks JSON', fatal=False)
229 tracks.extend(missing_tracks)
231 return self.playlist_result(
232 self._build_playlist(tracks),
233 compat_str(playlist_id),
234 playlist['title'], playlist.get('description'))