2 from __future__ import unicode_literals
7 from .common import InfoExtractor
8 from ..compat import compat_str
16 class YandexMusicBaseIE(InfoExtractor):
18 def _handle_error(response):
19 if isinstance(response, dict):
20 error = response.get('error')
22 raise ExtractorError(error, expected=True)
24 def _download_webpage(self, *args, **kwargs):
25 webpage = super(YandexMusicBaseIE, self)._download_webpage(*args, **kwargs)
26 if 'Нам очень жаль, но запросы, поступившие с вашего IP-адреса, похожи на автоматические.' in webpage:
28 'YandexMusic has considered youtube-dl requests automated and '
29 'asks you to solve a CAPTCHA. You can either wait for some '
30 'time until unblocked and optionally use --sleep-interval '
31 'in future or alternatively you can go to https://music.yandex.ru/ '
32 'solve CAPTCHA, then export cookies and pass cookie file to '
33 'youtube-dl with --cookies',
37 def _download_json(self, *args, **kwargs):
38 response = super(YandexMusicBaseIE, self)._download_json(*args, **kwargs)
39 self._handle_error(response)
43 class YandexMusicTrackIE(YandexMusicBaseIE):
44 IE_NAME = 'yandexmusic:track'
45 IE_DESC = 'Яндекс.Музыка - Трек'
46 _VALID_URL = r'https?://music\.yandex\.(?:ru|kz|ua|by)/album/(?P<album_id>\d+)/track/(?P<id>\d+)'
49 'url': 'http://music.yandex.ru/album/540508/track/4878838',
50 'md5': 'f496818aa2f60b6c0062980d2e00dc20',
54 'title': 'Carlo Ambrosio & Fabio Di Bari, Carlo Ambrosio - Gypsy Eyes 1',
57 'track': 'Gypsy Eyes 1',
58 'album': 'Gypsy Soul',
59 'album_artist': 'Carlo Ambrosio',
60 'artist': 'Carlo Ambrosio & Fabio Di Bari, Carlo Ambrosio',
61 'release_year': '2009',
63 'skip': 'Travis CI servers blocked by YandexMusic',
66 def _get_track_url(self, storage_dir, track_id):
67 data = self._download_json(
68 'http://music.yandex.ru/api/v1.5/handlers/api-jsonp.jsx?action=getTrackSrc&p=download-info/%s'
70 track_id, 'Downloading track location JSON')
72 key = hashlib.md5(('XGRlBW9FXlekgbPrRHuSiA' + data['path'][1:] + data['s']).encode('utf-8')).hexdigest()
73 storage = storage_dir.split('.')
75 return ('http://%s/get-mp3/%s/%s?track-id=%s&from=service-10-track&similarities-experiment=default'
76 % (data['host'], key, data['ts'] + data['path'], storage[1]))
78 def _get_track_info(self, track):
80 cover_uri = track.get('albums', [{}])[0].get('coverUri')
82 thumbnail = cover_uri.replace('%%', 'orig')
83 if not thumbnail.startswith('http'):
84 thumbnail = 'http://' + thumbnail
86 track_title = track['title']
90 'url': self._get_track_url(track['storageDir'], track['id']),
91 'filesize': int_or_none(track.get('fileSize')),
92 'duration': float_or_none(track.get('durationMs'), 1000),
93 'thumbnail': thumbnail,
97 def extract_artist(artist_list):
98 if artist_list and isinstance(artist_list, list):
99 artists_names = [a['name'] for a in artist_list if a.get('name')]
101 return ', '.join(artists_names)
103 albums = track.get('albums')
104 if albums and isinstance(albums, list):
106 if isinstance(album, dict):
107 year = album.get('year')
109 'album': album.get('title'),
110 'album_artist': extract_artist(album.get('artists')),
111 'release_year': compat_str(year) if year else None,
114 track_artist = extract_artist(track.get('artists'))
117 'artist': track_artist,
118 'title': '%s - %s' % (track_artist, track_title),
121 track_info['title'] = track_title
124 def _real_extract(self, url):
125 mobj = re.match(self._VALID_URL, url)
126 album_id, track_id = mobj.group('album_id'), mobj.group('id')
128 track = self._download_json(
129 'http://music.yandex.ru/handlers/track.jsx?track=%s:%s' % (track_id, album_id),
130 track_id, 'Downloading track JSON')['track']
132 return self._get_track_info(track)
135 class YandexMusicPlaylistBaseIE(YandexMusicBaseIE):
136 def _build_playlist(self, tracks):
139 'http://music.yandex.ru/album/%s/track/%s' % (track['albums'][0]['id'], track['id']))
140 for track in tracks if track.get('albums') and isinstance(track.get('albums'), list)]
143 class YandexMusicAlbumIE(YandexMusicPlaylistBaseIE):
144 IE_NAME = 'yandexmusic:album'
145 IE_DESC = 'Яндекс.Музыка - Альбом'
146 _VALID_URL = r'https?://music\.yandex\.(?:ru|kz|ua|by)/album/(?P<id>\d+)/?(\?|$)'
149 'url': 'http://music.yandex.ru/album/540508',
152 'title': 'Carlo Ambrosio - Gypsy Soul (2009)',
154 'playlist_count': 50,
155 'skip': 'Travis CI servers blocked by YandexMusic',
158 def _real_extract(self, url):
159 album_id = self._match_id(url)
161 album = self._download_json(
162 'http://music.yandex.ru/handlers/album.jsx?album=%s' % album_id,
163 album_id, 'Downloading album JSON')
165 entries = self._build_playlist(album['volumes'][0])
167 title = '%s - %s' % (album['artists'][0]['name'], album['title'])
168 year = album.get('year')
170 title += ' (%s)' % year
172 return self.playlist_result(entries, compat_str(album['id']), title)
175 class YandexMusicPlaylistIE(YandexMusicPlaylistBaseIE):
176 IE_NAME = 'yandexmusic:playlist'
177 IE_DESC = 'Яндекс.Музыка - Плейлист'
178 _VALID_URL = r'https?://music\.yandex\.(?P<tld>ru|kz|ua|by)/users/(?P<user>[^/]+)/playlists/(?P<id>\d+)'
181 'url': 'http://music.yandex.ru/users/music.partners/playlists/1245',
184 'title': 'Что слушают Enter Shikari',
185 'description': 'md5:3b9f27b0efbe53f2ee1e844d07155cc9',
188 'skip': 'Travis CI servers blocked by YandexMusic',
190 # playlist exceeding the limit of 150 tracks shipped with webpage (see
191 # https://github.com/rg3/youtube-dl/issues/6666)
192 'url': 'https://music.yandex.ru/users/ya.playlist/playlists/1036',
195 'title': 'Музыка 90-х',
197 'playlist_mincount': 300,
198 'skip': 'Travis CI servers blocked by YandexMusic',
201 def _real_extract(self, url):
202 mobj = re.match(self._VALID_URL, url)
203 tld = mobj.group('tld')
204 user = mobj.group('user')
205 playlist_id = mobj.group('id')
207 playlist = self._download_json(
208 'https://music.yandex.%s/handlers/playlist.jsx' % tld,
209 playlist_id, 'Downloading missing tracks JSON',
213 'X-Requested-With': 'XMLHttpRequest',
218 'kinds': playlist_id,
221 'external-domain': 'music.yandex.%s' % tld,
222 'overembed': 'false',
225 tracks, track_ids = playlist['tracks'], map(compat_str, playlist['trackIds'])
227 # tracks dictionary shipped with playlist.jsx API is limited to 150 tracks,
228 # missing tracks should be retrieved manually.
229 if len(tracks) < len(track_ids):
230 present_track_ids = set([
231 compat_str(track['id'])
232 for track in tracks if track.get('id')])
233 missing_track_ids = [
234 track_id for track_id in track_ids
235 if track_id not in present_track_ids]
236 missing_tracks = self._download_json(
237 'https://music.yandex.%s/handlers/track-entries.jsx' % tld,
238 playlist_id, 'Downloading missing tracks JSON',
242 'X-Requested-With': 'XMLHttpRequest',
245 'entries': ','.join(missing_track_ids),
247 'external-domain': 'music.yandex.%s' % tld,
248 'overembed': 'false',
252 tracks.extend(missing_tracks)
254 return self.playlist_result(
255 self._build_playlist(tracks),
256 compat_str(playlist_id),
257 playlist.get('title'), playlist.get('description'))