6 from .common
import InfoExtractor
, SearchInfoExtractor
7 from ..networking
import HEADRequest
8 from ..networking
.exceptions
import HTTPError
24 from ..utils
.traversal
import traverse_obj
27 class SoundcloudEmbedIE(InfoExtractor
):
28 _VALID_URL
= r
'https?://(?:w|player|p)\.soundcloud\.com/player/?.*?\burl=(?P<id>.+)'
29 _EMBED_REGEX
= [r
'<iframe[^>]+src=(["\'])(?P
<url
>(?
:https?
://)?
(?
:w\
.)?soundcloud\
.com
/player
.+?
)\
1']
31 # from https://www.soundi.fi/uutiset/ennakkokuuntelussa-timo-kaukolammen-station-to-station-to-station-julkaisua-juhlitaan-tanaan-g-livelabissa/
32 'url
': 'https
://w
.soundcloud
.com
/player
/?visual
=true
&url
=https
%3A
%2F
%2Fapi
.soundcloud
.com
%2Fplaylists
%2F922213810
&show_artwork
=true
&maxwidth
=640&maxheight
=960&dnt
=1&secret_token
=s
-ziYey
',
33 'only_matching
': True,
36 def _real_extract(self, url):
38 api_url = query['url
'][0]
39 secret_token = query.get('secret_token
')
41 api_url = update_url_query(api_url, {'secret_token': secret_token[0]})
42 return self.url_result(api_url)
45 class SoundcloudBaseIE(InfoExtractor):
46 _NETRC_MACHINE = 'soundcloud
'
48 _API_V2_BASE = 'https
://api
-v2
.soundcloud
.com
/'
49 _BASE_URL = 'https
://soundcloud
.com
/'
50 _USER_AGENT = 'Mozilla
/5.0 (Windows NT
10.0; Win64
; x64
) AppleWebKit
/537.36 (KHTML
, like Gecko
) Chrome
/84.0.4147.105 Safari
/537.36'
51 _API_AUTH_QUERY_TEMPLATE = '?client_id
=%s'
52 _API_AUTH_URL_PW = 'https
://api
-auth
.soundcloud
.com
/web
-auth
/sign
-in/password
%s'
53 _API_VERIFY_AUTH_TOKEN = 'https
://api
-auth
.soundcloud
.com
/connect
/session
%s'
56 _IMAGE_REPL_RE = r'-([0-9a
-z
]+)\
.jpg
'
71 _DEFAULT_FORMATS = ['http_aac
', 'hls_aac
', 'http_opus
', 'hls_opus
', 'http_mp3
', 'hls_mp3
']
73 @functools.cached_property
74 def _is_requested(self):
75 return re.compile(r'|
'.join(set(
76 re.escape(pattern).replace(r'\
*', r'.*') if pattern != 'default
'
77 else '|
'.join(map(re.escape, self._DEFAULT_FORMATS))
78 for pattern in self._configuration_arg('formats
', ['default
'], ie_key=SoundcloudIE)
81 def _store_client_id(self, client_id):
82 self.cache.store('soundcloud
', 'client_id
', client_id)
84 def _update_client_id(self):
85 webpage = self._download_webpage('https
://soundcloud
.com
/', None)
86 for src in reversed(re.findall(r'<script
[^
>]+src
="([^"]+)"', webpage)):
87 script = self._download_webpage(src, None, fatal=False)
89 client_id = self._search_regex(
90 r'client_id\s*:\s*"([0-9a
-zA
-Z
]{32}
)"',
91 script, 'client id', default=None)
93 self._CLIENT_ID = client_id
94 self._store_client_id(client_id)
96 raise ExtractorError('Unable to extract client id')
98 def _call_api(self, *args, **kwargs):
99 non_fatal = kwargs.get('fatal') is False
102 query = kwargs.get('query', {}).copy()
104 query['client_id'] = self._CLIENT_ID
105 kwargs['query'] = query
107 return self._download_json(*args, **kwargs)
108 except ExtractorError as e:
109 if isinstance(e.cause, HTTPError) and e.cause.status in (401, 403):
110 self._store_client_id(None)
111 self._update_client_id()
114 self.report_warning(str(e))
118 def _initialize_pre_login(self):
119 self._CLIENT_ID = self.cache.load('soundcloud', 'client_id') or 'a3e059563d7fd3372b49b37f00a00bcf'
121 def _verify_oauth_token(self, token):
122 if self._request_webpage(
123 self._API_VERIFY_AUTH_TOKEN % (self._API_AUTH_QUERY_TEMPLATE % self._CLIENT_ID),
124 None, note='Verifying login token...', fatal=False,
125 data=json.dumps({'session': {'access_token': token}}).encode()):
126 self._HEADERS['Authorization'] = f'OAuth {token}'
129 self.report_warning('Provided authorization token is invalid. Continuing as guest')
131 def _real_initialize(self):
134 if token := try_call(lambda: self._get_cookies(self._BASE_URL)['oauth_token'].value):
135 self._verify_oauth_token(token)
137 def _perform_login(self, username, password):
138 if username != 'oauth':
139 raise ExtractorError(
140 'Login using username and password is not currently supported. '
141 'Use "--username oauth
--password
<oauth_token
>" to login using an oauth token, '
142 f'or else {self._login_hint(method="cookies")}', expected=True)
145 self._verify_oauth_token(password)
150 return ''.join([str(random.randrange(10)) for i in range(6)])
151 return '-'.join([genNumBlock() for i in range(4)])
154 'client_id': self._CLIENT_ID,
155 'recaptcha_pubkey': 'null',
156 'recaptcha_response': 'null',
158 'identifier': username,
161 'signature': self.sign(username, password, self._CLIENT_ID),
162 'device_id': genDevId(),
163 'user_agent': self._USER_AGENT
166 response = self._call_api(
167 self._API_AUTH_URL_PW % (self._API_AUTH_QUERY_TEMPLATE % self._CLIENT_ID),
168 None, note='Verifying login token...', fatal=False,
169 data=json.dumps(payload).encode())
171 if token := traverse_obj(response, ('session', 'access_token', {str})):
172 self._HEADERS['Authorization'] = f'OAuth {token}'
176 raise ExtractorError('Unable to get access token, login may have failed', expected=True)
179 # signature generation
180 def sign(self, user, pw, clid):
190 n = '0763ed7314c69015fd4a0dc16bbf4b90' # _KEY
192 r = 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/84.0.4147.105 Safari/537.36' # _USER_AGENT
194 t = clid # _CLIENT_ID
196 d = '-'.join([str(mInt) for mInt in [a, i, s, w, u, l, b, k]])
197 h = n + y + d + r + e + t + d + n
201 for f in range(len(h)):
202 m = (m >> 1) + ((1 & m) << 23)
206 # c is not even needed
207 return f'{y}:{d}:{m:x}:{c}'
209 def _extract_info_dict(self, info, full_title=None, secret_token=None, extract_flat=False):
210 track_id = str(info['id'])
211 title = info['title']
215 query = {'client_id': self._CLIENT_ID}
217 query['secret_token'] = secret_token
219 if not extract_flat and info.get('downloadable') and info.get('has_downloads_left'):
221 # Do not use _call_api(); HTTP Error codes have different meanings for this request
222 download_data = self._download_json(
223 f'{self._API_V2_BASE}tracks/{track_id}/download', track_id,
224 'Downloading original download format info JSON', query=query, headers=self._HEADERS)
225 except ExtractorError as e:
226 if isinstance(e.cause, HTTPError) and e.cause.status == 401:
228 'Original download format is only available '
229 f'for registered users. {self._login_hint()}')
230 elif isinstance(e.cause, HTTPError) and e.cause.status == 403:
231 self.write_debug('Original download format is not available for this client')
233 self.report_warning(e.msg)
236 if redirect_url := traverse_obj(download_data, ('redirectUri', {url_or_none})):
237 urlh = self._request_webpage(
238 HEADRequest(redirect_url), track_id, 'Checking original download format availability',
239 'Original download format is not available', fatal=False)
241 format_url = urlh.url
242 format_urls.add(format_url)
244 'format_id': 'download',
245 'ext': urlhandle_detect_ext(urlh) or 'mp3',
246 'filesize': int_or_none(urlh.headers.get('Content-Length')),
249 'format_note': 'Original',
252 def invalid_url(url):
253 return not url or url in format_urls
255 def add_format(f, protocol, is_preview=False):
256 mobj = re.search(r'\.(?P<abr>\d+)\.(?P<ext>[0-9a-z]{3,4})(?=[/?])', stream_url)
258 for k, v in mobj.groupdict().items():
263 format_id_list.append(protocol)
269 'format_note': 'Premium',
271 for k in ('ext', 'abr'):
272 v = str_or_none(f.get(k))
274 format_id_list.append(v)
275 preview = is_preview or re.search(r'/(?:preview|playlist)/0/30/', f['url'])
277 format_id_list.append('preview')
281 if protocol in ('hls', 'hls-aes'):
282 protocol = 'm3u8' if ext == 'aac' else 'm3u8_native'
286 'format_id': '_'.join(format_id_list),
287 'protocol': protocol,
288 'preference': -10 if preview else None,
293 for t in traverse_obj(info, ('media', 'transcodings', lambda _, v: url_or_none(v['url']))):
296 format_url = t['url']
298 protocol = traverse_obj(t, ('format', 'protocol', {str}))
299 if protocol == 'progressive':
301 if protocol != 'hls' and '/hls' in format_url:
303 if protocol == 'encrypted-hls' or '/encrypted-hls' in format_url:
307 if preset := traverse_obj(t, ('preset', {str_or_none})):
308 ext = preset.split('_')[0]
309 if ext not in KNOWN_EXTENSIONS:
310 ext = mimetype2ext(traverse_obj(t, ('format', 'mime_type', {str})))
312 identifier = join_nonempty(protocol, ext, delim='_')
313 if not self._is_requested(identifier):
314 self.write_debug(f'"{identifier}
" is not a requested format, skipping')
318 for retry in self.RetryManager(fatal=False):
320 stream = self._call_api(
321 format_url, track_id, f'Downloading {identifier} format info JSON',
322 query=query, headers=self._HEADERS)
323 except ExtractorError as e:
324 if isinstance(e.cause, HTTPError) and e.cause.status == 429:
326 'You have reached the API rate limit, which is ~600 requests per '
327 '10 minutes. Use the --extractor-retries and --retry-sleep options '
328 'to configure an appropriate retry count and wait time', only_once=True)
329 retry.error = e.cause
331 self.report_warning(e.msg)
333 stream_url = traverse_obj(stream, ('url', {url_or_none}))
334 if invalid_url(stream_url):
336 format_urls.add(stream_url)
340 }, protocol, t.get('snipped') or '/preview/' in format_url)
345 if not formats and info.get('policy') == 'BLOCK':
346 self.raise_geo_restricted(metadata_available=True)
348 user = info.get('user') or {}
351 artwork_url = info.get('artwork_url')
352 thumbnail = artwork_url or user.get('avatar_url')
353 if isinstance(thumbnail, str):
354 if re.search(self._IMAGE_REPL_RE, thumbnail):
355 for image_id, size in self._ARTWORK_MAP.items():
358 'url': re.sub(self._IMAGE_REPL_RE, f'-{image_id}.jpg', thumbnail),
360 if image_id == 'tiny' and not artwork_url:
362 elif image_id == 'original':
371 thumbnails = [{'url': thumbnail}]
373 def extract_count(key):
374 return int_or_none(info.get(f'{key}_count'))
378 'uploader': user.get('username'),
379 'uploader_id': str_or_none(user.get('id')) or user.get('permalink'),
380 'uploader_url': user.get('permalink_url'),
381 'timestamp': unified_timestamp(info.get('created_at')),
383 'description': info.get('description'),
384 'thumbnails': thumbnails,
385 'duration': float_or_none(info.get('duration'), 1000),
386 'webpage_url': info.get('permalink_url'),
387 'license': info.get('license'),
388 'view_count': extract_count('playback'),
389 'like_count': extract_count('favoritings') or extract_count('likes'),
390 'comment_count': extract_count('comment'),
391 'repost_count': extract_count('reposts'),
392 'genres': traverse_obj(info, ('genre', {str}, {lambda x: x or None}, all)),
393 'formats': formats if not extract_flat else None,
397 def _resolv_url(cls, url):
398 return cls._API_V2_BASE + 'resolve?url=' + url
401 class SoundcloudIE(SoundcloudBaseIE):
402 """Information extractor for soundcloud.com
403 To access the media, the uid of the song and a stream token
404 must be extracted from the page source and the script must make
405 a request to media.soundcloud.com/crossdomain.xml. Then
406 the media can be grabbed by requesting from an url composed
407 of the stream token and uid
410 _VALID_URL = r'''(?x)^(?:https?://)?
411 (?:(?:(?:www\.|m\.)?soundcloud\.com/
413 (?P<uploader>[\w\d-]+)/
414 (?!(?:tracks|albums|sets(?:/.+?)?|reposts|likes|spotlight)/?(?:$|[?#]))
416 (?:/(?P<token>(?!(?:albums|sets|recommended))[^?]+?))?
418 |(?:api(?:-v2)?\.soundcloud\.com/tracks/(?P<track_id>\d+)
419 (?:/?\?secret_token=(?P<secret_token>[^&]+))?)
422 IE_NAME = 'soundcloud'
425 'url': 'http://soundcloud.com/ethmusic/lostin-powers-she-so-heavy',
426 'md5': 'de9bac153e7427a7333b4b0c1b6a18d2',
430 'title': 'Lostin Powers - She so Heavy (SneakPreview) Adrian Ackers Blueprint 1',
431 'description': 'No Downloads untill we record the finished version this weekend, i was too pumped n i had to post it , earl is prolly gonna b hella p.o\'d',
432 'uploader': 'E.T. ExTerrestrial Music',
433 'uploader_id': '1571244',
434 'timestamp': 1349920598,
435 'upload_date': '20121011',
437 'license': 'all-rights-reserved',
440 'comment_count': int,
442 'thumbnail': 'https://i1.sndcdn.com/artworks-000031955188-rwb18x-original.jpg',
443 'uploader_url': 'https://soundcloud.com/ethmusic',
449 'url': 'https://soundcloud.com/the-concept-band/goldrushed-mastered?in=the-concept-band/sets/the-royal-concept-ep',
453 'title': 'Goldrushed',
454 'description': 'From Stockholm Sweden\r\nPovel / Magnus / Filip / David\r\nwww.theroyalconcept.com',
455 'uploader': 'The Royal Concept',
456 'uploader_id': '9615865',
457 'timestamp': 1337635207,
458 'upload_date': '20120521',
460 'license': 'all-rights-reserved',
463 'comment_count': int,
465 'uploader_url': 'https://soundcloud.com/the-concept-band',
466 'thumbnail': 'https://i1.sndcdn.com/artworks-v8bFHhXm7Au6-0-original.jpg',
467 'genres': ['Alternative'],
472 'url': 'https://soundcloud.com/jaimemf/youtube-dl-test-video-a-y-baw/s-8Pjrp',
473 'md5': 'aa0dd32bfea9b0c5ef4f02aacd080604',
477 'title': 'Youtube - Dl Test Video \'\' Äâ†',
478 'description': 'test chars: "\'/\\äâ†
',
479 'uploader
': 'jaimeMF
',
480 'uploader_id
': '69767071',
481 'timestamp
': 1386604920,
482 'upload_date
': '20131209',
484 'license
': 'all
-rights
-reserved
',
487 'comment_count
': int,
489 'uploader_url
': 'https
://soundcloud
.com
/jaimemf
',
490 'thumbnail
': 'https
://a1
.sndcdn
.com
/images
/default_avatar_large
.png
',
491 'genres
': ['youtubedl
'],
494 # private link (alt format)
496 'url
': 'https
://api
.soundcloud
.com
/tracks
/123998367?secret_token
=s
-8Pjrp
',
497 'md5
': 'aa0dd32bfea9b0c5ef4f02aacd080604
',
501 'title
': 'Youtube
- Dl Test Video
\'\' Äâ†
',
502 'description
': 'test chars
: "\'/\\äâ†',
503 'uploader': 'jaimeMF',
504 'uploader_id': '69767071',
505 'timestamp': 1386604920,
506 'upload_date': '20131209',
508 'license': 'all-rights-reserved',
511 'comment_count': int,
513 'uploader_url': 'https://soundcloud.com/jaimemf',
514 'thumbnail': 'https://a1.sndcdn.com/images/default_avatar_large.png',
515 'genres': ['youtubedl'],
520 'url': 'https://soundcloud.com/the80m/the-following',
521 'md5': '9ffcddb08c87d74fb5808a3c183a1d04',
525 'title': 'The Following',
528 'uploader_id': '312384765',
529 'uploader_url': 'https://soundcloud.com/the80m',
530 'upload_date': '20170922',
531 'timestamp': 1506120436,
533 'thumbnail': 'https://i1.sndcdn.com/artworks-000243916348-ktoo7d-original.jpg',
534 'license': 'all-rights-reserved',
536 'comment_count': int,
539 'genres': ['Dance & EDM'],
542 # private link, downloadable format
544 'url': 'https://soundcloud.com/oriuplift/uponly-238-no-talking-wav/s-AyZUd',
545 'md5': '64a60b16e617d41d0bef032b7f55441e',
549 'title': 'Uplifting Only 238 [No Talking] (incl. Alex Feed Guestmix) (Aug 31, 2017) [wav]',
550 'description': 'md5:fa20ee0fca76a3d6df8c7e57f3715366',
551 'uploader': 'Ori Uplift Music',
552 'uploader_id': '12563093',
553 'timestamp': 1504206263,
554 'upload_date': '20170831',
555 'duration': 7449.096,
556 'license': 'all-rights-reserved',
559 'comment_count': int,
561 'thumbnail': 'https://i1.sndcdn.com/artworks-000240712245-kedn4p-original.jpg',
562 'uploader_url': 'https://soundcloud.com/oriuplift',
563 'genres': ['Trance'],
566 # no album art, use avatar pic for thumbnail
568 'url': 'https://soundcloud.com/garyvee/sideways-prod-mad-real',
569 'md5': '59c7872bc44e5d99b7211891664760c2',
573 'title': 'Sideways (Prod. Mad Real)',
574 'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
575 'uploader': 'garyvee',
576 'uploader_id': '2366352',
577 'timestamp': 1488152409,
578 'upload_date': '20170226',
580 'thumbnail': r're:https?://.*\.jpg',
581 'license': 'all-rights-reserved',
584 'comment_count': int,
586 'uploader_url': 'https://soundcloud.com/garyvee',
590 'skip_download': True,
594 'url': 'https://soundcloud.com/giovannisarani/mezzo-valzer',
595 'md5': '8227c3473a4264df6b02ad7e5b7527ac',
599 'title': 'Mezzo Valzer',
600 'description': 'md5:f4d5f39d52e0ccc2b4f665326428901a',
601 'uploader': 'Giovanni Sarani',
602 'uploader_id': '3352531',
603 'timestamp': 1551394171,
604 'upload_date': '20190228',
606 'thumbnail': r're:https?://.*\.jpg',
607 'license': 'all-rights-reserved',
610 'comment_count': int,
613 'uploader_url': 'https://soundcloud.com/giovannisarani',
617 # AAC HQ format available (account with active subscription needed)
618 'url': 'https://soundcloud.com/wandw/the-chainsmokers-ft-daya-dont-let-me-down-ww-remix-1',
619 'only_matching': True,
622 # Go+ (account with active subscription needed)
623 'url': 'https://soundcloud.com/taylorswiftofficial/look-what-you-made-me-do',
624 'only_matching': True,
628 def _real_extract(self, url):
629 mobj = self._match_valid_url(url)
631 track_id = mobj.group('track_id')
635 info_json_url = self._API_V2_BASE + 'tracks/' + track_id
636 full_title = track_id
637 token = mobj.group('secret_token')
639 query['secret_token'] = token
641 full_title = resolve_title = '{}/{}'.format(*mobj.group('uploader', 'title'))
642 token = mobj.group('token')
644 resolve_title += f'/{token}'
645 info_json_url = self._resolv_url(self._BASE_URL + resolve_title)
647 info = self._call_api(
648 info_json_url, full_title, 'Downloading info JSON', query=query, headers=self._HEADERS)
650 return self._extract_info_dict(info, full_title, token)
653 class SoundcloudPlaylistBaseIE(SoundcloudBaseIE):
654 def _extract_set(self, playlist, token=None):
655 playlist_id = str(playlist['id'])
656 tracks = playlist.get('tracks') or []
657 if not all(t.get('permalink_url') for t in tracks) and token:
658 tracks = self._call_api(
659 self._API_V2_BASE + 'tracks', playlist_id,
660 'Downloading tracks', query={
661 'ids': ','.join([str(t['id']) for t in tracks]),
662 'playlistId': playlist_id,
663 'playlistSecretToken': token,
664 }, headers=self._HEADERS)
667 track_id = str_or_none(track.get('id'))
668 url = track.get('permalink_url')
672 url = self._API_V2_BASE + 'tracks/' + track_id
674 url += '?secret_token=' + token
675 entries.append(self.url_result(
676 url, SoundcloudIE.ie_key(), track_id))
677 return self.playlist_result(
678 entries, playlist_id,
679 playlist.get('title'),
680 playlist.get('description'))
683 class SoundcloudSetIE(SoundcloudPlaylistBaseIE):
684 _VALID_URL = r'https?://(?:(?:www|m)\.)?soundcloud\.com/(?P<uploader>[\w\d-]+)/sets/(?P<slug_title>[:\w\d-]+)(?:/(?P<token>[^?/]+))?'
685 IE_NAME = 'soundcloud:set'
687 'url': 'https://soundcloud.com/the-concept-band/sets/the-royal-concept-ep',
690 'title': 'The Royal Concept EP',
691 'description': 'md5:71d07087c7a449e8941a70a29e34671e',
693 'playlist_mincount': 5,
695 'url': 'https://soundcloud.com/the-concept-band/sets/the-royal-concept-ep/token',
696 'only_matching': True,
698 'url': 'https://soundcloud.com/discover/sets/weekly::flacmatic',
699 'only_matching': True,
701 'url': 'https://soundcloud.com/discover/sets/charts-top:all-music:de',
702 'only_matching': True,
704 'url': 'https://soundcloud.com/discover/sets/charts-top:hiphoprap:kr',
705 'only_matching': True,
708 def _real_extract(self, url):
709 mobj = self._match_valid_url(url)
711 full_title = '{}/sets/{}'.format(*mobj.group('uploader', 'slug_title'))
712 token = mobj.group('token')
714 full_title += '/' + token
716 info = self._call_api(self._resolv_url(
717 self._BASE_URL + full_title), full_title, headers=self._HEADERS)
720 msgs = (str(err['error_message']) for err in info['errors'])
721 raise ExtractorError('unable to download video webpage: {}'.format(','.join(msgs)))
723 return self._extract_set(info, token)
726 class SoundcloudPagedPlaylistBaseIE(SoundcloudBaseIE):
727 def _extract_playlist(self, base_url, playlist_id, playlist_title):
731 'title': playlist_title,
732 'entries': self._entries(base_url, playlist_id),
735 def _entries(self, url, playlist_id):
736 # Per the SoundCloud documentation, the maximum limit for a linked partitioning query is 200.
737 # https://developers.soundcloud.com/blog/offset-pagination-deprecated
740 'linked_partitioning': '1',
744 for i in itertools.count():
745 for retry in self.RetryManager():
747 response = self._call_api(
748 url, playlist_id, query=query, headers=self._HEADERS,
749 note=f'Downloading track page {i + 1}')
751 except ExtractorError as e:
752 # Downloading page may result in intermittent 502 HTTP error
753 # See https://github.com/yt-dlp/yt-dlp/issues/872
754 if not isinstance(e.cause, HTTPError) or e.cause.status != 502:
759 def resolve_entry(*candidates):
760 for cand in candidates:
761 if not isinstance(cand, dict):
763 permalink_url = url_or_none(cand.get('permalink_url'))
765 return self.url_result(
767 SoundcloudIE.ie_key() if SoundcloudIE.suitable(permalink_url) else None,
768 str_or_none(cand.get('id')), cand.get('title'))
770 for e in response['collection'] or []:
771 yield resolve_entry(e, e.get('track'), e.get('playlist'))
773 url = response.get('next_href')
776 query.pop('offset', None)
779 class SoundcloudUserIE(SoundcloudPagedPlaylistBaseIE):
780 _VALID_URL = r'''(?x)
782 (?:(?:www|m)\.)?soundcloud\.com/
785 (?P<rsrc>tracks|albums|sets|reposts|likes|spotlight)
789 IE_NAME = 'soundcloud:user'
791 'url': 'https://soundcloud.com/soft-cell-official',
794 'title': 'Soft Cell (All)',
796 'playlist_mincount': 28,
798 'url': 'https://soundcloud.com/soft-cell-official/tracks',
801 'title': 'Soft Cell (Tracks)',
803 'playlist_mincount': 27,
805 'url': 'https://soundcloud.com/soft-cell-official/albums',
808 'title': 'Soft Cell (Albums)',
810 'playlist_mincount': 1,
812 'url': 'https://soundcloud.com/jcv246/sets',
815 'title': 'Jordi / cv (Sets)',
817 'playlist_mincount': 2,
819 'url': 'https://soundcloud.com/jcv246/reposts',
822 'title': 'Jordi / cv (Reposts)',
824 'playlist_mincount': 6,
826 'url': 'https://soundcloud.com/clalberg/likes',
829 'title': 'clalberg (Likes)',
831 'playlist_mincount': 5,
833 'url': 'https://soundcloud.com/grynpyret/spotlight',
836 'title': 'Grynpyret (Spotlight)',
838 'playlist_mincount': 1,
842 'all': 'stream/users/%s',
843 'tracks': 'users/%s/tracks',
844 'albums': 'users/%s/albums',
845 'sets': 'users/%s/playlists',
846 'reposts': 'stream/users/%s/reposts',
847 'likes': 'users/%s/likes',
848 'spotlight': 'users/%s/spotlight',
851 def _real_extract(self, url):
852 mobj = self._match_valid_url(url)
853 uploader = mobj.group('user')
855 user = self._call_api(
856 self._resolv_url(self._BASE_URL + uploader),
857 uploader, 'Downloading user info', headers=self._HEADERS)
859 resource = mobj.group('rsrc') or 'all'
861 return self._extract_playlist(
862 self._API_V2_BASE + self._BASE_URL_MAP[resource] % user['id'],
863 str_or_none(user.get('id')),
864 '{} ({})'.format(user['username'], resource.capitalize()))
867 class SoundcloudUserPermalinkIE(SoundcloudPagedPlaylistBaseIE):
868 _VALID_URL = r'https?://api\.soundcloud\.com/users/(?P<id>\d+)'
869 IE_NAME = 'soundcloud:user:permalink'
871 'url': 'https://api.soundcloud.com/users/30909869',
876 'playlist_mincount': 23,
879 def _real_extract(self, url):
880 user_id = self._match_id(url)
881 user = self._call_api(
882 self._resolv_url(url), user_id, 'Downloading user info', headers=self._HEADERS)
884 return self._extract_playlist(
885 f'{self._API_V2_BASE}stream/users/{user["id"]}', str(user['id']), user.get('username'))
888 class SoundcloudTrackStationIE(SoundcloudPagedPlaylistBaseIE):
889 _VALID_URL = r'https?://(?:(?:www|m)\.)?soundcloud\.com/stations/track/[^/]+/(?P<id>[^/?#&]+)'
890 IE_NAME = 'soundcloud:trackstation'
892 'url': 'https://soundcloud.com/stations/track/officialsundial/your-text',
895 'title': 'Track station: your text',
897 'playlist_mincount': 47,
900 def _real_extract(self, url):
901 track_name = self._match_id(url)
903 track = self._call_api(self._resolv_url(url), track_name, headers=self._HEADERS)
904 track_id = self._search_regex(
905 r'soundcloud:track-stations:(\d+)', track['id'], 'track id')
907 return self._extract_playlist(
908 self._API_V2_BASE + 'stations/{}/tracks'.format(track['id']),
909 track_id, 'Track station: {}'.format(track['title']))
912 class SoundcloudRelatedIE(SoundcloudPagedPlaylistBaseIE):
913 _VALID_URL = r'https?://(?:(?:www|m)\.)?soundcloud\.com/(?P<slug>[\w\d-]+/[\w\d-]+)/(?P<relation>albums|sets|recommended)'
914 IE_NAME = 'soundcloud:related'
916 'url': 'https://soundcloud.com/wajang/sexapil-pingers-5/recommended',
919 'title': 'Sexapil - Pingers 5 (Recommended)',
921 'playlist_mincount': 50,
923 'url': 'https://soundcloud.com/wajang/sexapil-pingers-5/albums',
926 'title': 'Sexapil - Pingers 5 (Albums)',
928 'playlist_mincount': 1,
930 'url': 'https://soundcloud.com/wajang/sexapil-pingers-5/sets',
933 'title': 'Sexapil - Pingers 5 (Sets)',
935 'playlist_mincount': 4,
939 'albums': 'tracks/%s/albums',
940 'sets': 'tracks/%s/playlists_without_albums',
941 'recommended': 'tracks/%s/related',
944 def _real_extract(self, url):
945 slug, relation = self._match_valid_url(url).group('slug', 'relation')
947 track = self._call_api(
948 self._resolv_url(self._BASE_URL + slug),
949 slug, 'Downloading track info', headers=self._HEADERS)
951 if track.get('errors'):
952 raise ExtractorError(f'{self.IE_NAME} said: %s' % ','.join(
953 str(err['error_message']) for err in track['errors']), expected=True)
955 return self._extract_playlist(
956 self._API_V2_BASE + self._BASE_URL_MAP[relation] % track['id'], str(track['id']),
957 '{} ({})'.format(track.get('title') or slug, relation.capitalize()))
960 class SoundcloudPlaylistIE(SoundcloudPlaylistBaseIE):
961 _VALID_URL = r'https?://api(?:-v2)?\.soundcloud\.com/playlists/(?P<id>[0-9]+)(?:/?\?secret_token=(?P<token>[^&]+?))?$'
962 IE_NAME = 'soundcloud:playlist'
964 'url': 'https://api.soundcloud.com/playlists/4110309',
967 'title': 'TILT Brass - Bowery Poetry Club, August \'03 [Non-Site SCR 02]',
968 'description': 're:.*?TILT Brass - Bowery Poetry Club',
973 def _real_extract(self, url):
974 mobj = self._match_valid_url(url)
975 playlist_id = mobj.group('id')
978 token = mobj.group('token')
980 query['secret_token'] = token
982 data = self._call_api(
983 self._API_V2_BASE + 'playlists/' + playlist_id,
984 playlist_id, 'Downloading playlist', query=query, headers=self._HEADERS)
986 return self._extract_set(data, token)
989 class SoundcloudSearchIE(SoundcloudBaseIE, SearchInfoExtractor):
990 IE_NAME = 'soundcloud:search'
991 IE_DESC = 'Soundcloud search'
992 _SEARCH_KEY = 'scsearch'
994 'url': 'scsearch15:post-avant jazzcore',
996 'id': 'post-avant jazzcore',
997 'title': 'post-avant jazzcore',
999 'playlist_count': 15,
1002 _MAX_RESULTS_PER_PAGE = 200
1003 _DEFAULT_RESULTS_PER_PAGE = 50
1005 def _get_collection(self, endpoint, collection_id, **query):
1007 query.get('limit', self._DEFAULT_RESULTS_PER_PAGE),
1008 self._MAX_RESULTS_PER_PAGE)
1011 'linked_partitioning': 1,
1014 next_url = update_url_query(self._API_V2_BASE + endpoint, query)
1016 for i in itertools.count(1):
1017 response = self._call_api(
1018 next_url, collection_id, f'Downloading page {i}',
1019 'Unable to download API page', headers=self._HEADERS)
1021 for item in response.get('collection') or []:
1023 yield self.url_result(
1024 item['uri'], SoundcloudIE.ie_key(), **self._extract_info_dict(item, extract_flat=True))
1026 next_url = response.get('next_href')
1030 def _get_n_results(self, query, n):
1031 return self.playlist_result(itertools.islice(
1032 self._get_collection('search/tracks', query, limit=n, q=query),
1033 0, None if n == float('inf') else n), query, query)