]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/soundcloud.py
10 from .. compat
import compat_str
11 from .. networking
import HEADRequest
, Request
12 from .. networking
. exceptions
import HTTPError
30 class SoundcloudEmbedIE ( InfoExtractor
):
31 _VALID_URL
= r
'https?://(?:w|player|p)\.soundcloud\.com/player/?.*?\burl=(?P<id>.+)'
32 _EMBED_REGEX
= [ r
'<iframe[^>]+src=(["\' ])( ?P
< url
>( ?
: https?
://) ?
( ?
: w\
.) ?soundcloud\
. com
/ player
.+ ?
) \
1 ']
34 # from https://www.soundi.fi/uutiset/ennakkokuuntelussa-timo-kaukolammen-station-to-station-to-station-julkaisua-juhlitaan-tanaan-g-livelabissa/
35 ' url
': ' https
:// w
. soundcloud
. com
/ player
/ ?visual
= true
& url
= https
% 3 A
% 2 F
% 2 Fapi
. soundcloud
. com
% 2 Fplaylists
% 2 F922213810
& show_artwork
= true
& maxwidth
= 640 & maxheight
= 960 & dnt
= 1 & secret_token
= s
- ziYey
',
36 ' only_matching
': True,
39 def _real_extract(self, url):
41 api_url = query[' url
'][0]
42 secret_token = query.get(' secret_token
')
44 api_url = update_url_query(api_url, {'secret_token': secret_token[0]} )
45 return self.url_result(api_url)
48 class SoundcloudBaseIE(InfoExtractor):
49 _NETRC_MACHINE = ' soundcloud
'
51 _API_V2_BASE = ' https
:// api
- v2
. soundcloud
. com
/ '
52 _BASE_URL = ' https
:// soundcloud
. com
/ '
53 _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 '
54 _API_AUTH_QUERY_TEMPLATE = ' ?client_id
= %s '
55 _API_AUTH_URL_PW = ' https
:// api
- auth
. soundcloud
. com
/ web
- auth
/ sign
- in / password
%s '
56 _API_VERIFY_AUTH_TOKEN = ' https
:// api
- auth
. soundcloud
. com
/ connect
/ session
%s '
60 _IMAGE_REPL_RE = r' -([ 0 - 9 a
- z
]+) \
. jpg
'
75 def _store_client_id(self, client_id):
76 self.cache.store(' soundcloud
', ' client_id
', client_id)
78 def _update_client_id(self):
79 webpage = self._download_webpage(' https
:// soundcloud
. com
/ ', None)
80 for src in reversed(re.findall(r' < script
[ ^
>]+ src
= "([^" ]+) "', webpage)):
81 script = self._download_webpage(src, None, fatal=False)
83 client_id = self._search_regex(
84 r'client_id\s*:\s*" ([ 0 - 9 a
- zA
- Z
] {32}
) "',
85 script, 'client id', default=None)
87 self._CLIENT_ID = client_id
88 self._store_client_id(client_id)
90 raise ExtractorError('Unable to extract client id')
92 def _download_json(self, *args, **kwargs):
93 non_fatal = kwargs.get('fatal') is False
96 query = kwargs.get('query', {}).copy()
98 query['client_id'] = self._CLIENT_ID
99 kwargs['query'] = query
101 return super()._download_json(*args, **kwargs)
102 except ExtractorError as e:
103 if isinstance(e.cause, HTTPError) and e.cause.status in (401, 403):
104 self._store_client_id(None)
105 self._update_client_id()
108 self.report_warning(error_to_compat_str(e))
112 def _initialize_pre_login(self):
113 self._CLIENT_ID = self.cache.load('soundcloud', 'client_id') or 'a3e059563d7fd3372b49b37f00a00bcf'
115 def _perform_login(self, username, password):
116 if username != 'oauth':
118 'Login using username and password is not currently supported. '
119 'Use " -- username oauth
-- password
< oauth_token
> " to login using an oauth token')
120 self._access_token = password
121 query = self._API_AUTH_QUERY_TEMPLATE % self._CLIENT_ID
122 payload = {'session': {'access_token': self._access_token} }
123 token_verification = Request(self._API_VERIFY_AUTH_TOKEN % query, json.dumps(payload).encode('utf-8'))
124 response = self._download_json(token_verification, None, note='Verifying login token...', fatal=False)
125 if response is not False:
126 self._HEADERS = {'Authorization': 'OAuth ' + self._access_token}
129 self.report_warning('Provided authorization token seems to be invalid. Continue as guest')
134 return ''.join([str(random.randrange(10)) for i in range(6)])
135 return '-'.join([genNumBlock() for i in range(4)])
138 'client_id': self._CLIENT_ID,
139 'recaptcha_pubkey': 'null',
140 'recaptcha_response': 'null',
142 'identifier': username,
145 'signature': self.sign(username, password, self._CLIENT_ID),
146 'device_id': genDevId(),
147 'user_agent': self._USER_AGENT
150 query = self._API_AUTH_QUERY_TEMPLATE % self._CLIENT_ID
151 login = sanitized_Request(self._API_AUTH_URL_PW % query, json.dumps(payload).encode('utf-8'))
152 response = self._download_json(login, None)
153 self._access_token = response.get('session').get('access_token')
154 if not self._access_token:
155 self.report_warning('Unable to get access token, login may has failed')
157 self._HEADERS = {'Authorization': 'OAuth ' + self._access_token}
160 # signature generation
161 def sign(self, user, pw, clid):
171 n = '0763ed7314c69015fd4a0dc16bbf4b90' # _KEY
173 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
175 t = clid # _CLIENT_ID
177 d = '-'.join([str(mInt) for mInt in [a, i, s, w, u, l, b, k]])
178 p = n + y + d + r + e + t + d + n
184 for f in range(f, len(h)):
185 m = (m >> 1) + ((1 & m) << 23)
189 # c is not even needed
190 out = str(y) + ':' + str(d) + ':' + format(m, 'x') + ':' + str(c)
194 def _extract_info_dict(self, info, full_title=None, secret_token=None, extract_flat=False):
195 track_id = compat_str(info['id'])
196 title = info['title']
200 query = {'client_id': self._CLIENT_ID}
202 query['secret_token'] = secret_token
204 if not extract_flat and info.get('downloadable') and info.get('has_downloads_left'):
205 download_url = update_url_query(
206 self._API_V2_BASE + 'tracks/' + track_id + '/download', query)
207 redirect_url = (self._download_json(download_url, track_id, fatal=False) or {}).get('redirectUri')
209 urlh = self._request_webpage(
210 HEADRequest(redirect_url), track_id, fatal=False)
212 format_url = urlh.url
213 format_urls.add(format_url)
215 'format_id': 'download',
216 'ext': urlhandle_detect_ext(urlh) or 'mp3',
217 'filesize': int_or_none(urlh.headers.get('Content-Length')),
222 def invalid_url(url):
223 return not url or url in format_urls
225 def add_format(f, protocol, is_preview=False):
226 mobj = re.search(r'\.(?P<abr>\d+)\.(?P<ext>[0-9a-z] {3,4} )(?=[/?])', stream_url)
228 for k, v in mobj.groupdict().items():
233 format_id_list.append(protocol)
237 for k in ('ext', 'abr'):
240 format_id_list.append(v)
241 preview = is_preview or re.search(r'/(?:preview|playlist)/0/30/', f['url'])
243 format_id_list.append('preview')
247 if protocol == 'hls':
248 protocol = 'm3u8' if ext == 'aac' else 'm3u8_native'
252 'format_id': '_'.join(format_id_list),
253 'protocol': protocol,
254 'preference': -10 if preview else None,
259 transcodings = try_get(
260 info, lambda x: x['media']['transcodings'], list) or []
261 for t in transcodings:
262 if not isinstance(t, dict):
264 format_url = url_or_none(t.get('url'))
267 stream = None if extract_flat else self._download_json(
268 format_url, track_id, query=query, fatal=False, headers=self._HEADERS)
269 if not isinstance(stream, dict):
271 stream_url = url_or_none(stream.get('url'))
272 if invalid_url(stream_url):
274 format_urls.add(stream_url)
275 stream_format = t.get('format') or {}
276 protocol = stream_format.get('protocol')
277 if protocol != 'hls' and '/hls' in format_url:
280 preset = str_or_none(t.get('preset'))
282 ext = preset.split('_')[0]
283 if ext not in KNOWN_EXTENSIONS:
284 ext = mimetype2ext(stream_format.get('mime_type'))
288 }, 'http' if protocol == 'progressive' else protocol,
289 t.get('snipped') or '/preview/' in format_url)
294 if not formats and info.get('policy') == 'BLOCK':
295 self.raise_geo_restricted(metadata_available=True)
297 user = info.get('user') or {}
300 artwork_url = info.get('artwork_url')
301 thumbnail = artwork_url or user.get('avatar_url')
302 if isinstance(thumbnail, compat_str):
303 if re.search(self._IMAGE_REPL_RE, thumbnail):
304 for image_id, size in self._ARTWORK_MAP.items():
307 'url': re.sub(self._IMAGE_REPL_RE, '- %s .jpg' % image_id, thumbnail),
309 if image_id == 'tiny' and not artwork_url:
311 elif image_id == 'original':
320 thumbnails = [ {'url': thumbnail} ]
322 def extract_count(key):
323 return int_or_none(info.get(' %s _count' % key))
327 'uploader': user.get('username'),
328 'uploader_id': str_or_none(user.get('id')) or user.get('permalink'),
329 'uploader_url': user.get('permalink_url'),
330 'timestamp': unified_timestamp(info.get('created_at')),
332 'description': info.get('description'),
333 'thumbnails': thumbnails,
334 'duration': float_or_none(info.get('duration'), 1000),
335 'webpage_url': info.get('permalink_url'),
336 'license': info.get('license'),
337 'view_count': extract_count('playback'),
338 'like_count': extract_count('favoritings') or extract_count('likes'),
339 'comment_count': extract_count('comment'),
340 'repost_count': extract_count('reposts'),
341 'genre': info.get('genre'),
342 'formats': formats if not extract_flat else None
346 def _resolv_url(cls, url):
347 return cls._API_V2_BASE + 'resolve?url=' + url
350 class SoundcloudIE(SoundcloudBaseIE):
351 """Information extractor for soundcloud.com
352 To access the media, the uid of the song and a stream token
353 must be extracted from the page source and the script must make
354 a request to media.soundcloud.com/crossdomain.xml. Then
355 the media can be grabbed by requesting from an url composed
356 of the stream token and uid
359 _VALID_URL = r'''(?x)^(?:https?://)?
360 (?:(?:(?:www\.|m\.)?soundcloud\.com/
362 (?P<uploader>[\w\d-]+)/
363 (?!(?:tracks|albums|sets(?:/.+?)?|reposts|likes|spotlight)/?(?:$|[?#]))
365 (?:/(?P<token>(?!(?:albums|sets|recommended))[^?]+?))?
367 |(?:api(?:-v2)?\.soundcloud\.com/tracks/(?P<track_id>\d+)
368 (?:/? \? secret_token=(?P<secret_token>[^&]+))?)
371 IE_NAME = 'soundcloud'
374 'url': 'http://soundcloud.com/ethmusic/lostin-powers-she-so-heavy',
375 'md5': 'ebef0a451b909710ed1d7787dddbf0d7',
379 'title': 'Lostin Powers - She so Heavy (SneakPreview) Adrian Ackers Blueprint 1',
380 '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',
381 'uploader': 'E.T. ExTerrestrial Music',
382 'uploader_id': '1571244',
383 'timestamp': 1349920598,
384 'upload_date': '20121011',
386 'license': 'all-rights-reserved',
389 'comment_count': int,
395 'url': 'https://soundcloud.com/the-concept-band/goldrushed-mastered?in=the-concept-band/sets/the-royal-concept-ep',
399 'title': 'Goldrushed',
400 'description': 'From Stockholm Sweden \r\n Povel / Magnus / Filip / David \r\n www.theroyalconcept.com',
401 'uploader': 'The Royal Concept',
402 'uploader_id': '9615865',
403 'timestamp': 1337635207,
404 'upload_date': '20120521',
406 'license': 'all-rights-reserved',
409 'comment_count': int,
415 'url': 'https://soundcloud.com/jaimemf/youtube-dl-test-video-a-y-baw/s-8Pjrp',
416 'md5': 'aa0dd32bfea9b0c5ef4f02aacd080604',
420 'title': 'Youtube - Dl Test Video \'\' Äâ†',
421 'description': 'test chars: \"\' / \\ äâ†',
422 'uploader': 'jaimeMF',
423 'uploader_id': '69767071',
424 'timestamp': 1386604920,
425 'upload_date': '20131209',
427 'license': 'all-rights-reserved',
430 'comment_count': int,
434 # private link (alt format)
436 'url': 'https://api.soundcloud.com/tracks/123998367?secret_token=s-8Pjrp',
437 'md5': 'aa0dd32bfea9b0c5ef4f02aacd080604',
441 'title': 'Youtube - Dl Test Video \'\' Äâ†',
442 'description': 'test chars: \"\' / \\ äâ†',
443 'uploader': 'jaimeMF',
444 'uploader_id': '69767071',
445 'timestamp': 1386604920,
446 'upload_date': '20131209',
448 'license': 'all-rights-reserved',
451 'comment_count': int,
457 'url': 'https://soundcloud.com/the80m/the-following',
458 'md5': '9ffcddb08c87d74fb5808a3c183a1d04',
464 # private link, downloadable format
466 'url': 'https://soundcloud.com/oriuplift/uponly-238-no-talking-wav/s-AyZUd',
467 'md5': '64a60b16e617d41d0bef032b7f55441e',
471 'title': 'Uplifting Only 238 [No Talking] (incl. Alex Feed Guestmix) (Aug 31, 2017) [wav]',
472 'description': 'md5:fa20ee0fca76a3d6df8c7e57f3715366',
473 'uploader': 'Ori Uplift Music',
474 'uploader_id': '12563093',
475 'timestamp': 1504206263,
476 'upload_date': '20170831',
477 'duration': 7449.096,
478 'license': 'all-rights-reserved',
481 'comment_count': int,
485 # no album art, use avatar pic for thumbnail
487 'url': 'https://soundcloud.com/garyvee/sideways-prod-mad-real',
488 'md5': '59c7872bc44e5d99b7211891664760c2',
492 'title': 'Sideways (Prod. Mad Real)',
493 'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
494 'uploader': 'garyvee',
495 'uploader_id': '2366352',
496 'timestamp': 1488152409,
497 'upload_date': '20170226',
499 'thumbnail': r're:https?://.*\.jpg',
500 'license': 'all-rights-reserved',
503 'comment_count': int,
507 'skip_download': True,
511 'url': 'https://soundcloud.com/giovannisarani/mezzo-valzer',
512 'md5': 'e22aecd2bc88e0e4e432d7dcc0a1abf7',
516 'title': 'Mezzo Valzer',
517 'description': 'md5:4138d582f81866a530317bae316e8b61',
518 'uploader': 'Micronie',
519 'uploader_id': '3352531',
520 'timestamp': 1551394171,
521 'upload_date': '20190228',
523 'thumbnail': r're:https?://.*\.jpg',
524 'license': 'all-rights-reserved',
527 'comment_count': int,
532 # AAC HQ format available (account with active subscription needed)
533 'url': 'https://soundcloud.com/wandw/the-chainsmokers-ft-daya-dont-let-me-down-ww-remix-1',
534 'only_matching': True,
537 # Go+ (account with active subscription needed)
538 'url': 'https://soundcloud.com/taylorswiftofficial/look-what-you-made-me-do',
539 'only_matching': True,
543 def _real_extract(self, url):
544 mobj = self._match_valid_url(url)
546 track_id = mobj.group('track_id')
550 info_json_url = self._API_V2_BASE + 'tracks/' + track_id
551 full_title = track_id
552 token = mobj.group('secret_token')
554 query['secret_token'] = token
556 full_title = resolve_title = ' %s / %s ' % mobj.group('uploader', 'title')
557 token = mobj.group('token')
559 resolve_title += '/ %s ' % token
560 info_json_url = self._resolv_url(self._BASE_URL + resolve_title)
562 info = self._download_json(
563 info_json_url, full_title, 'Downloading info JSON', query=query, headers=self._HEADERS)
565 return self._extract_info_dict(info, full_title, token)
568 class SoundcloudPlaylistBaseIE(SoundcloudBaseIE):
569 def _extract_set(self, playlist, token=None):
570 playlist_id = compat_str(playlist['id'])
571 tracks = playlist.get('tracks') or []
572 if not all([t.get('permalink_url') for t in tracks]) and token:
573 tracks = self._download_json(
574 self._API_V2_BASE + 'tracks', playlist_id,
575 'Downloading tracks', query={
576 'ids': ','.join([compat_str(t['id']) for t in tracks]),
577 'playlistId': playlist_id,
578 'playlistSecretToken': token,
579 }, headers=self._HEADERS)
582 track_id = str_or_none(track.get('id'))
583 url = track.get('permalink_url')
587 url = self._API_V2_BASE + 'tracks/' + track_id
589 url += '?secret_token=' + token
590 entries.append(self.url_result(
591 url, SoundcloudIE.ie_key(), track_id))
592 return self.playlist_result(
593 entries, playlist_id,
594 playlist.get('title'),
595 playlist.get('description'))
598 class SoundcloudSetIE(SoundcloudPlaylistBaseIE):
599 _VALID_URL = r'https?://(?:(?:www|m)\.)?soundcloud\.com/(?P<uploader>[\w\d-]+)/sets/(?P<slug_title>[:\w\d-]+)(?:/(?P<token>[^?/]+))?'
600 IE_NAME = 'soundcloud:set'
602 'url': 'https://soundcloud.com/the-concept-band/sets/the-royal-concept-ep',
605 'title': 'The Royal Concept EP',
606 'description': 'md5:71d07087c7a449e8941a70a29e34671e',
608 'playlist_mincount': 5,
610 'url': 'https://soundcloud.com/the-concept-band/sets/the-royal-concept-ep/token',
611 'only_matching': True,
613 'url': 'https://soundcloud.com/discover/sets/weekly::flacmatic',
614 'only_matching': True,
616 'url': 'https://soundcloud.com/discover/sets/charts-top:all-music:de',
617 'only_matching': True,
619 'url': 'https://soundcloud.com/discover/sets/charts-top:hiphoprap:kr',
620 'only_matching': True,
623 def _real_extract(self, url):
624 mobj = self._match_valid_url(url)
626 full_title = ' %s /sets/ %s ' % mobj.group('uploader', 'slug_title')
627 token = mobj.group('token')
629 full_title += '/' + token
631 info = self._download_json(self._resolv_url(
632 self._BASE_URL + full_title), full_title, headers=self._HEADERS)
635 msgs = (compat_str(err['error_message']) for err in info['errors'])
636 raise ExtractorError('unable to download video webpage: %s ' % ','.join(msgs))
638 return self._extract_set(info, token)
641 class SoundcloudPagedPlaylistBaseIE(SoundcloudBaseIE):
642 def _extract_playlist(self, base_url, playlist_id, playlist_title):
646 'title': playlist_title,
647 'entries': self._entries(base_url, playlist_id),
650 def _entries(self, url, playlist_id):
651 # Per the SoundCloud documentation, the maximum limit for a linked partitioning query is 200.
652 # https://developers.soundcloud.com/blog/offset-pagination-deprecated
655 'linked_partitioning': '1',
659 for i in itertools.count():
660 for retry in self.RetryManager():
662 response = self._download_json(
663 url, playlist_id, query=query, headers=self._HEADERS,
664 note=f'Downloading track page {i + 1} ')
666 except ExtractorError as e:
667 # Downloading page may result in intermittent 502 HTTP error
668 # See https://github.com/yt-dlp/yt-dlp/issues/872
669 if not isinstance(e.cause, HTTPError) or e.cause.status != 502:
674 def resolve_entry(*candidates):
675 for cand in candidates:
676 if not isinstance(cand, dict):
678 permalink_url = url_or_none(cand.get('permalink_url'))
680 return self.url_result(
682 SoundcloudIE.ie_key() if SoundcloudIE.suitable(permalink_url) else None,
683 str_or_none(cand.get('id')), cand.get('title'))
685 for e in response['collection'] or []:
686 yield resolve_entry(e, e.get('track'), e.get('playlist'))
688 url = response.get('next_href')
691 query.pop('offset', None)
694 class SoundcloudUserIE(SoundcloudPagedPlaylistBaseIE):
695 _VALID_URL = r'''(?x)
697 (?:(?:www|m)\.)?soundcloud\.com/
700 (?P<rsrc>tracks|albums|sets|reposts|likes|spotlight)
704 IE_NAME = 'soundcloud:user'
706 'url': 'https://soundcloud.com/soft-cell-official',
709 'title': 'Soft Cell (All)',
711 'playlist_mincount': 28,
713 'url': 'https://soundcloud.com/soft-cell-official/tracks',
716 'title': 'Soft Cell (Tracks)',
718 'playlist_mincount': 27,
720 'url': 'https://soundcloud.com/soft-cell-official/albums',
723 'title': 'Soft Cell (Albums)',
725 'playlist_mincount': 1,
727 'url': 'https://soundcloud.com/jcv246/sets',
730 'title': 'Jordi / cv (Sets)',
732 'playlist_mincount': 2,
734 'url': 'https://soundcloud.com/jcv246/reposts',
737 'title': 'Jordi / cv (Reposts)',
739 'playlist_mincount': 6,
741 'url': 'https://soundcloud.com/clalberg/likes',
744 'title': 'clalberg (Likes)',
746 'playlist_mincount': 5,
748 'url': 'https://soundcloud.com/grynpyret/spotlight',
751 'title': 'Grynpyret (Spotlight)',
753 'playlist_mincount': 1,
757 'all': 'stream/users/ %s ',
758 'tracks': 'users/ %s /tracks',
759 'albums': 'users/ %s /albums',
760 'sets': 'users/ %s /playlists',
761 'reposts': 'stream/users/ %s /reposts',
762 'likes': 'users/ %s /likes',
763 'spotlight': 'users/ %s /spotlight',
766 def _real_extract(self, url):
767 mobj = self._match_valid_url(url)
768 uploader = mobj.group('user')
770 user = self._download_json(
771 self._resolv_url(self._BASE_URL + uploader),
772 uploader, 'Downloading user info', headers=self._HEADERS)
774 resource = mobj.group('rsrc') or 'all'
776 return self._extract_playlist(
777 self._API_V2_BASE + self._BASE_URL_MAP[resource] % user['id'],
778 str_or_none(user.get('id')),
779 ' %s ( %s )' % (user['username'], resource.capitalize()))
782 class SoundcloudUserPermalinkIE(SoundcloudPagedPlaylistBaseIE):
783 _VALID_URL = r'https?://api\.soundcloud\.com/users/(?P<id>\d+)'
784 IE_NAME = 'soundcloud:user:permalink'
786 'url': 'https://api.soundcloud.com/users/30909869',
791 'playlist_mincount': 23,
794 def _real_extract(self, url):
795 user_id = self._match_id(url)
796 user = self._download_json(
797 self._resolv_url(url), user_id, 'Downloading user info', headers=self._HEADERS)
799 return self._extract_playlist(
800 f' {self._API_V2_BASE} stream/users/ {user["id"]} ', str(user['id']), user.get('username'))
803 class SoundcloudTrackStationIE(SoundcloudPagedPlaylistBaseIE):
804 _VALID_URL = r'https?://(?:(?:www|m)\.)?soundcloud\.com/stations/track/[^/]+/(?P<id>[^/?#&]+)'
805 IE_NAME = 'soundcloud:trackstation'
807 'url': 'https://soundcloud.com/stations/track/officialsundial/your-text',
810 'title': 'Track station: your text',
812 'playlist_mincount': 47,
815 def _real_extract(self, url):
816 track_name = self._match_id(url)
818 track = self._download_json(self._resolv_url(url), track_name, headers=self._HEADERS)
819 track_id = self._search_regex(
820 r'soundcloud:track-stations:(\d+)', track['id'], 'track id')
822 return self._extract_playlist(
823 self._API_V2_BASE + 'stations/ %s /tracks' % track['id'],
824 track_id, 'Track station: %s ' % track['title'])
827 class SoundcloudRelatedIE(SoundcloudPagedPlaylistBaseIE):
828 _VALID_URL = r'https?://(?:(?:www|m)\.)?soundcloud\.com/(?P<slug>[\w\d-]+/[\w\d-]+)/(?P<relation>albums|sets|recommended)'
829 IE_NAME = 'soundcloud:related'
831 'url': 'https://soundcloud.com/wajang/sexapil-pingers-5/recommended',
834 'title': 'Sexapil - Pingers 5 (Recommended)',
836 'playlist_mincount': 50,
838 'url': 'https://soundcloud.com/wajang/sexapil-pingers-5/albums',
841 'title': 'Sexapil - Pingers 5 (Albums)',
843 'playlist_mincount': 1,
845 'url': 'https://soundcloud.com/wajang/sexapil-pingers-5/sets',
848 'title': 'Sexapil - Pingers 5 (Sets)',
850 'playlist_mincount': 4,
854 'albums': 'tracks/ %s /albums',
855 'sets': 'tracks/ %s /playlists_without_albums',
856 'recommended': 'tracks/ %s /related',
859 def _real_extract(self, url):
860 slug, relation = self._match_valid_url(url).group('slug', 'relation')
862 track = self._download_json(
863 self._resolv_url(self._BASE_URL + slug),
864 slug, 'Downloading track info', headers=self._HEADERS)
866 if track.get('errors'):
867 raise ExtractorError(f' {self.IE_NAME} said: %s ' % ','.join(
868 str(err['error_message']) for err in track['errors']), expected=True)
870 return self._extract_playlist(
871 self._API_V2_BASE + self._BASE_URL_MAP[relation] % track['id'], str(track['id']),
872 ' %s ( %s )' % (track.get('title') or slug, relation.capitalize()))
875 class SoundcloudPlaylistIE(SoundcloudPlaylistBaseIE):
876 _VALID_URL = r'https?://api(?:-v2)?\.soundcloud\.com/playlists/(?P<id>[0-9]+)(?:/? \? secret_token=(?P<token>[^&]+?))?$'
877 IE_NAME = 'soundcloud:playlist'
879 'url': 'https://api.soundcloud.com/playlists/4110309',
882 'title': 'TILT Brass - Bowery Poetry Club, August \' 03 [Non-Site SCR 02]',
883 'description': 're:.*?TILT Brass - Bowery Poetry Club',
888 def _real_extract(self, url):
889 mobj = self._match_valid_url(url)
890 playlist_id = mobj.group('id')
893 token = mobj.group('token')
895 query['secret_token'] = token
897 data = self._download_json(
898 self._API_V2_BASE + 'playlists/' + playlist_id,
899 playlist_id, 'Downloading playlist', query=query, headers=self._HEADERS)
901 return self._extract_set(data, token)
904 class SoundcloudSearchIE(SoundcloudBaseIE, SearchInfoExtractor):
905 IE_NAME = 'soundcloud:search'
906 IE_DESC = 'Soundcloud search'
907 _SEARCH_KEY = 'scsearch'
909 'url': 'scsearch15:post-avant jazzcore',
911 'id': 'post-avant jazzcore',
912 'title': 'post-avant jazzcore',
914 'playlist_count': 15,
917 _MAX_RESULTS_PER_PAGE = 200
918 _DEFAULT_RESULTS_PER_PAGE = 50
920 def _get_collection(self, endpoint, collection_id, **query):
922 query.get('limit', self._DEFAULT_RESULTS_PER_PAGE),
923 self._MAX_RESULTS_PER_PAGE)
926 'linked_partitioning': 1,
929 next_url = update_url_query(self._API_V2_BASE + endpoint, query)
931 for i in itertools.count(1):
932 response = self._download_json(
933 next_url, collection_id, f'Downloading page {i} ',
934 'Unable to download API page', headers=self._HEADERS)
936 for item in response.get('collection') or []:
938 yield self.url_result(
939 item['uri'], SoundcloudIE.ie_key(), **self._extract_info_dict(item, extract_flat=True))
941 next_url = response.get('next_href')
945 def _get_n_results(self, query, n):
946 return self.playlist_result(itertools.islice(
947 self._get_collection('search/tracks', query, limit=n, q=query),
948 0, None if n == float('inf') else n), query, query)