4 from .common
import InfoExtractor
5 from ..compat
import compat_str
14 class YandexMusicBaseIE(InfoExtractor
):
15 _VALID_URL_BASE
= r
'https?://music\.yandex\.(?P<tld>ru|kz|ua|by|com)'
18 def _handle_error(response
):
19 if isinstance(response
, dict):
20 error
= response
.get('error')
22 raise ExtractorError(error
, expected
=True)
23 if response
.get('type') == 'captcha' or 'captcha' in response
:
24 YandexMusicBaseIE
._raise
_captcha
()
29 'YandexMusic has considered yt-dlp requests automated and '
30 'asks you to solve a CAPTCHA. You can either wait for some '
31 'time until unblocked and optionally use --sleep-interval '
32 'in future or alternatively you can go to https://music.yandex.ru/ '
33 'solve CAPTCHA, then export cookies and pass cookie file to '
34 'yt-dlp with --cookies',
37 def _download_webpage_handle(self
, *args
, **kwargs
):
38 webpage
= super(YandexMusicBaseIE
, self
)._download
_webpage
_handle
(*args
, **kwargs
)
39 if 'Нам очень жаль, но запросы, поступившие с вашего IP-адреса, похожи на автоматические.' in webpage
:
43 def _download_json(self
, *args
, **kwargs
):
44 response
= super(YandexMusicBaseIE
, self
)._download
_json
(*args
, **kwargs
)
45 self
._handle
_error
(response
)
48 def _call_api(self
, ep
, tld
, url
, item_id
, note
, query
):
49 return self
._download
_json
(
50 'https://music.yandex.%s/handlers/%s.jsx' % (tld
, ep
),
55 'X-Requested-With': 'XMLHttpRequest',
61 class YandexMusicTrackIE(YandexMusicBaseIE
):
62 IE_NAME
= 'yandexmusic:track'
63 IE_DESC
= 'Яндекс.Музыка - Трек'
64 _VALID_URL
= r
'%s/album/(?P<album_id>\d+)/track/(?P<id>\d+)' % YandexMusicBaseIE
._VALID
_URL
_BASE
67 'url': 'http://music.yandex.ru/album/540508/track/4878838',
68 'md5': 'dec8b661f12027ceaba33318787fff76',
72 'title': 'md5:c63e19341fdbe84e43425a30bc777856',
75 'track': 'md5:210508c6ffdfd67a493a6c378f22c3ff',
76 'album': 'md5:cd04fb13c4efeafdfa0a6a6aca36d01a',
77 'album_artist': 'md5:5f54c35462c07952df33d97cfb5fc200',
78 'artist': 'md5:e6fd86621825f14dc0b25db3acd68160',
81 # 'skip': 'Travis CI servers blocked by YandexMusic',
84 'url': 'http://music.yandex.ru/album/3840501/track/705105',
85 'md5': '82a54e9e787301dd45aba093cf6e58c0',
89 'title': 'md5:f86d4a9188279860a83000277024c1a6',
92 'track': 'md5:40f887f0666ba1aa10b835aca44807d1',
93 'album': 'md5:624f5224b14f5c88a8e812fd7fbf1873',
94 'album_artist': 'md5:dd35f2af4e8927100cbe6f5e62e1fb12',
95 'artist': 'md5:dd35f2af4e8927100cbe6f5e62e1fb12',
101 # 'skip': 'Travis CI servers blocked by YandexMusic',
103 'url': 'http://music.yandex.com/album/540508/track/4878838',
104 'only_matching': True,
107 def _real_extract(self
, url
):
108 mobj
= self
._match
_valid
_url
(url
)
109 tld
, album_id
, track_id
= mobj
.group('tld'), mobj
.group('album_id'), mobj
.group('id')
111 track
= self
._call
_api
(
112 'track', tld
, url
, track_id
, 'Downloading track JSON',
113 {'track': '%s:%s' % (track_id, album_id)}
)['track']
114 track_title
= track
['title']
116 download_data
= self
._download
_json
(
117 'https://music.yandex.ru/api/v2.1/handlers/track/%s:%s/web-album_track-track-track-main/download/m' % (track_id
, album_id
),
118 track_id
, 'Downloading track location url JSON', query
={'hq': 1}
, headers
={'X-Retpath-Y': url}
)
120 fd_data
= self
._download
_json
(
121 download_data
['src'], track_id
,
122 'Downloading track location JSON',
123 query
={'format': 'json'}
)
124 key
= hashlib
.md5(('XGRlBW9FXlekgbPrRHuSiA' + fd_data
['path'][1:] + fd_data
['s']).encode('utf-8')).hexdigest()
125 f_url
= 'http://%s/get-mp3/%s/%s?track-id=%s ' % (fd_data
['host'], key
, fd_data
['ts'] + fd_data
['path'], track
['id'])
128 cover_uri
= track
.get('albums', [{}])[0].get('coverUri')
130 thumbnail
= cover_uri
.replace('%%', 'orig')
131 if not thumbnail
.startswith('http'):
132 thumbnail
= 'http://' + thumbnail
138 'filesize': int_or_none(track
.get('fileSize')),
139 'duration': float_or_none(track
.get('durationMs'), 1000),
140 'thumbnail': thumbnail
,
141 'track': track_title
,
142 'acodec': download_data
.get('codec'),
143 'abr': int_or_none(download_data
.get('bitrate')),
146 def extract_artist_name(artist
):
147 decomposed
= artist
.get('decomposed')
148 if not isinstance(decomposed
, list):
149 return artist
['name']
150 parts
= [artist
['name']]
151 for element
in decomposed
:
152 if isinstance(element
, dict) and element
.get('name'):
153 parts
.append(element
['name'])
154 elif isinstance(element
, compat_str
):
155 parts
.append(element
)
156 return ''.join(parts
)
158 def extract_artist(artist_list
):
159 if artist_list
and isinstance(artist_list
, list):
160 artists_names
= [extract_artist_name(a
) for a
in artist_list
if a
.get('name')]
162 return ', '.join(artists_names
)
164 albums
= track
.get('albums')
165 if albums
and isinstance(albums
, list):
167 if isinstance(album
, dict):
168 year
= album
.get('year')
169 disc_number
= int_or_none(try_get(
170 album
, lambda x
: x
['trackPosition']['volume']))
171 track_number
= int_or_none(try_get(
172 album
, lambda x
: x
['trackPosition']['index']))
174 'album': album
.get('title'),
175 'album_artist': extract_artist(album
.get('artists')),
176 'release_year': int_or_none(year
),
177 'genre': album
.get('genre'),
178 'disc_number': disc_number
,
179 'track_number': track_number
,
182 track_artist
= extract_artist(track
.get('artists'))
185 'artist': track_artist
,
186 'title': '%s - %s' % (track_artist
, track_title
),
189 track_info
['title'] = track_title
194 class YandexMusicPlaylistBaseIE(YandexMusicBaseIE
):
195 def _extract_tracks(self
, source
, item_id
, url
, tld
):
196 tracks
= source
['tracks']
197 track_ids
= [compat_str(track_id
) for track_id
in source
['trackIds']]
199 # tracks dictionary shipped with playlist.jsx API is limited to 150 tracks,
200 # missing tracks should be retrieved manually.
201 if len(tracks
) < len(track_ids
):
202 present_track_ids
= set([
203 compat_str(track
['id'])
204 for track
in tracks
if track
.get('id')])
205 missing_track_ids
= [
206 track_id
for track_id
in track_ids
207 if track_id
not in present_track_ids
]
208 # Request missing tracks in chunks to avoid exceeding max HTTP header size,
209 # see https://github.com/ytdl-org/youtube-dl/issues/27355
210 _TRACKS_PER_CHUNK
= 250
211 for chunk_num
in itertools
.count(0):
212 start
= chunk_num
* _TRACKS_PER_CHUNK
213 end
= start
+ _TRACKS_PER_CHUNK
214 missing_track_ids_req
= missing_track_ids
[start
:end
]
215 assert missing_track_ids_req
216 missing_tracks
= self
._call
_api
(
217 'track-entries', tld
, url
, item_id
,
218 'Downloading missing tracks JSON chunk %d' % (chunk_num
+ 1), {
219 'entries': ','.join(missing_track_ids_req
),
221 'external-domain': 'music.yandex.%s' % tld
,
222 'overembed': 'false',
226 tracks
.extend(missing_tracks
)
227 if end
>= len(missing_track_ids
):
232 def _build_playlist(self
, tracks
):
235 track_id
= track
.get('id') or track
.get('realId')
238 albums
= track
.get('albums')
239 if not albums
or not isinstance(albums
, list):
242 if not isinstance(album
, dict):
244 album_id
= album
.get('id')
247 entries
.append(self
.url_result(
248 'http://music.yandex.ru/album/%s/track/%s' % (album_id
, track_id
),
249 ie
=YandexMusicTrackIE
.ie_key(), video_id
=track_id
))
253 class YandexMusicAlbumIE(YandexMusicPlaylistBaseIE
):
254 IE_NAME
= 'yandexmusic:album'
255 IE_DESC
= 'Яндекс.Музыка - Альбом'
256 _VALID_URL
= r
'%s/album/(?P<id>\d+)' % YandexMusicBaseIE
._VALID
_URL
_BASE
259 'url': 'http://music.yandex.ru/album/540508',
262 'title': 'md5:7ed1c3567f28d14be9f61179116f5571',
264 'playlist_count': 50,
265 # 'skip': 'Travis CI servers blocked by YandexMusic',
267 'url': 'https://music.yandex.ru/album/3840501',
270 'title': 'md5:36733472cdaa7dcb1fd9473f7da8e50f',
272 'playlist_count': 33,
273 # 'skip': 'Travis CI servers blocked by YandexMusic',
276 'url': 'https://music.yandex.ru/album/9091882',
279 'title': 'ТЕД на русском',
281 'playlist_count': 187,
285 def suitable(cls
, url
):
286 return False if YandexMusicTrackIE
.suitable(url
) else super(YandexMusicAlbumIE
, cls
).suitable(url
)
288 def _real_extract(self
, url
):
289 mobj
= self
._match
_valid
_url
(url
)
290 tld
= mobj
.group('tld')
291 album_id
= mobj
.group('id')
293 album
= self
._call
_api
(
294 'album', tld
, url
, album_id
, 'Downloading album JSON',
297 entries
= self
._build
_playlist
([track
for volume
in album
['volumes'] for track
in volume
])
299 title
= album
['title']
300 artist
= try_get(album
, lambda x
: x
['artists'][0]['name'], compat_str
)
302 title
= '%s - %s' % (artist
, title
)
303 year
= album
.get('year')
305 title
+= ' (%s)' % year
307 return self
.playlist_result(entries
, compat_str(album
['id']), title
)
310 class YandexMusicPlaylistIE(YandexMusicPlaylistBaseIE
):
311 IE_NAME
= 'yandexmusic:playlist'
312 IE_DESC
= 'Яндекс.Музыка - Плейлист'
313 _VALID_URL
= r
'%s/users/(?P<user>[^/]+)/playlists/(?P<id>\d+)' % YandexMusicBaseIE
._VALID
_URL
_BASE
316 'url': 'http://music.yandex.ru/users/music.partners/playlists/1245',
319 'title': 'md5:841559b3fe2b998eca88d0d2e22a3097',
320 'description': 'md5:3b9f27b0efbe53f2ee1e844d07155cc9',
323 # 'skip': 'Travis CI servers blocked by YandexMusic',
325 'url': 'https://music.yandex.ru/users/ya.playlist/playlists/1036',
326 'only_matching': True,
328 # playlist exceeding the limit of 150 tracks (see
329 # https://github.com/ytdl-org/youtube-dl/issues/6666)
330 'url': 'https://music.yandex.ru/users/mesiaz/playlists/1364',
333 'title': 'md5:b3b400f997d3f878a13ae0699653f7db',
335 'playlist_mincount': 437,
336 # 'skip': 'Travis CI servers blocked by YandexMusic',
339 def _real_extract(self
, url
):
340 mobj
= self
._match
_valid
_url
(url
)
341 tld
= mobj
.group('tld')
342 user
= mobj
.group('user')
343 playlist_id
= mobj
.group('id')
345 playlist
= self
._call
_api
(
346 'playlist', tld
, url
, playlist_id
, 'Downloading playlist JSON', {
348 'kinds': playlist_id
,
351 'external-domain': 'music.yandex.%s' % tld
,
352 'overembed': 'false',
355 tracks
= self
._extract
_tracks
(playlist
, playlist_id
, url
, tld
)
357 return self
.playlist_result(
358 self
._build
_playlist
(tracks
),
359 compat_str(playlist_id
),
360 playlist
.get('title'), playlist
.get('description'))
363 class YandexMusicArtistBaseIE(YandexMusicPlaylistBaseIE
):
364 def _call_artist(self
, tld
, url
, artist_id
):
365 return self
._call
_api
(
366 'artist', tld
, url
, artist_id
,
367 'Downloading artist %s JSON' % self
._ARTIST
_WHAT
, {
369 'what': self
._ARTIST
_WHAT
,
370 'sort': self
._ARTIST
_SORT
or '',
374 'external-domain': 'music.yandex.%s' % tld
,
375 'overembed': 'false',
378 def _real_extract(self
, url
):
379 mobj
= self
._match
_valid
_url
(url
)
380 tld
= mobj
.group('tld')
381 artist_id
= mobj
.group('id')
382 data
= self
._call
_artist
(tld
, url
, artist_id
)
383 tracks
= self
._extract
_tracks
(data
, artist_id
, url
, tld
)
384 title
= try_get(data
, lambda x
: x
['artist']['name'], compat_str
)
385 return self
.playlist_result(
386 self
._build
_playlist
(tracks
), artist_id
, title
)
389 class YandexMusicArtistTracksIE(YandexMusicArtistBaseIE
):
390 IE_NAME
= 'yandexmusic:artist:tracks'
391 IE_DESC
= 'Яндекс.Музыка - Артист - Треки'
392 _VALID_URL
= r
'%s/artist/(?P<id>\d+)/tracks' % YandexMusicBaseIE
._VALID
_URL
_BASE
395 'url': 'https://music.yandex.ru/artist/617526/tracks',
398 'title': 'md5:131aef29d45fd5a965ca613e708c040b',
400 'playlist_count': 507,
401 # 'skip': 'Travis CI servers blocked by YandexMusic',
405 _ARTIST_WHAT
= 'tracks'
407 def _real_extract(self
, url
):
408 mobj
= self
._match
_valid
_url
(url
)
409 tld
= mobj
.group('tld')
410 artist_id
= mobj
.group('id')
411 data
= self
._call
_artist
(tld
, url
, artist_id
)
412 tracks
= self
._extract
_tracks
(data
, artist_id
, url
, tld
)
413 artist
= try_get(data
, lambda x
: x
['artist']['name'], compat_str
)
414 title
= '%s - %s' % (artist
or artist_id
, 'Треки')
415 return self
.playlist_result(
416 self
._build
_playlist
(tracks
), artist_id
, title
)
419 class YandexMusicArtistAlbumsIE(YandexMusicArtistBaseIE
):
420 IE_NAME
= 'yandexmusic:artist:albums'
421 IE_DESC
= 'Яндекс.Музыка - Артист - Альбомы'
422 _VALID_URL
= r
'%s/artist/(?P<id>\d+)/albums' % YandexMusicBaseIE
._VALID
_URL
_BASE
425 'url': 'https://music.yandex.ru/artist/617526/albums',
428 'title': 'md5:55dc58d5c85699b7fb41ee926700236c',
431 # 'skip': 'Travis CI servers blocked by YandexMusic',
434 _ARTIST_SORT
= 'year'
435 _ARTIST_WHAT
= 'albums'
437 def _real_extract(self
, url
):
438 mobj
= self
._match
_valid
_url
(url
)
439 tld
= mobj
.group('tld')
440 artist_id
= mobj
.group('id')
441 data
= self
._call
_artist
(tld
, url
, artist_id
)
443 for album
in data
['albums']:
444 if not isinstance(album
, dict):
446 album_id
= album
.get('id')
449 entries
.append(self
.url_result(
450 'http://music.yandex.ru/album/%s' % album_id
,
451 ie
=YandexMusicAlbumIE
.ie_key(), video_id
=album_id
))
452 artist
= try_get(data
, lambda x
: x
['artist']['name'], compat_str
)
453 title
= '%s - %s' % (artist
or artist_id
, 'Альбомы')
454 return self
.playlist_result(entries
, artist_id
, title
)