]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/soundcloud.py
10 from .. compat
import (
33 class SoundcloudEmbedIE ( InfoExtractor
):
34 _VALID_URL
= r
'https?://(?:w|player|p)\.soundcloud\.com/player/?.*?\burl=(?P<id>.+)'
35 _EMBED_REGEX
= [ r
'<iframe[^>]+src=(["\' ])( ?P
< url
>( ?
: https?
://) ?
( ?
: w\
.) ?soundcloud\
. com
/ player
.+ ?
) \
1 ']
37 # from https://www.soundi.fi/uutiset/ennakkokuuntelussa-timo-kaukolammen-station-to-station-to-station-julkaisua-juhlitaan-tanaan-g-livelabissa/
38 ' 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
',
39 ' only_matching
': True,
42 def _real_extract(self, url):
44 api_url = query[' url
'][0]
45 secret_token = query.get(' secret_token
')
47 api_url = update_url_query(api_url, {'secret_token': secret_token[0]} )
48 return self.url_result(api_url)
51 class SoundcloudBaseIE(InfoExtractor):
52 _NETRC_MACHINE = ' soundcloud
'
54 _API_V2_BASE = ' https
:// api
- v2
. soundcloud
. com
/ '
55 _BASE_URL = ' https
:// soundcloud
. com
/ '
56 _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 '
57 _API_AUTH_QUERY_TEMPLATE = ' ?client_id
= %s '
58 _API_AUTH_URL_PW = ' https
:// api
- auth
. soundcloud
. com
/ web
- auth
/ sign
- in / password
%s '
59 _API_VERIFY_AUTH_TOKEN = ' https
:// api
- auth
. soundcloud
. com
/ connect
/ session
%s '
63 _IMAGE_REPL_RE = r' -([ 0 - 9 a
- z
]+) \
. jpg
'
78 def _store_client_id(self, client_id):
79 self.cache.store(' soundcloud
', ' client_id
', client_id)
81 def _update_client_id(self):
82 webpage = self._download_webpage(' https
:// soundcloud
. com
/ ', None)
83 for src in reversed(re.findall(r' < script
[ ^
>]+ src
= "([^" ]+) "', webpage)):
84 script = self._download_webpage(src, None, fatal=False)
86 client_id = self._search_regex(
87 r'client_id\s*:\s*" ([ 0 - 9 a
- zA
- Z
] {32}
) "',
88 script, 'client id', default=None)
90 self._CLIENT_ID = client_id
91 self._store_client_id(client_id)
93 raise ExtractorError('Unable to extract client id')
95 def _download_json(self, *args, **kwargs):
96 non_fatal = kwargs.get('fatal') is False
99 query = kwargs.get('query', {}).copy()
101 query['client_id'] = self._CLIENT_ID
102 kwargs['query'] = query
104 return super()._download_json(*args, **kwargs)
105 except ExtractorError as e:
106 if isinstance(e.cause, compat_HTTPError) and e.cause.code in (401, 403):
107 self._store_client_id(None)
108 self._update_client_id()
111 self.report_warning(error_to_compat_str(e))
115 def _initialize_pre_login(self):
116 self._CLIENT_ID = self.cache.load('soundcloud', 'client_id') or 'a3e059563d7fd3372b49b37f00a00bcf'
118 def _perform_login(self, username, password):
119 if username != 'oauth':
121 'Login using username and password is not currently supported. '
122 'Use " -- username oauth
-- password
< oauth_token
> " to login using an oauth token')
123 self._access_token = password
124 query = self._API_AUTH_QUERY_TEMPLATE % self._CLIENT_ID
125 payload = {'session': {'access_token': self._access_token} }
126 token_verification = sanitized_Request(self._API_VERIFY_AUTH_TOKEN % query, json.dumps(payload).encode('utf-8'))
127 response = self._download_json(token_verification, None, note='Verifying login token...', fatal=False)
128 if response is not False:
129 self._HEADERS = {'Authorization': 'OAuth ' + self._access_token}
132 self.report_warning('Provided authorization token seems to be invalid. Continue as guest')
137 return ''.join([str(random.randrange(10)) for i in range(6)])
138 return '-'.join([genNumBlock() for i in range(4)])
141 'client_id': self._CLIENT_ID,
142 'recaptcha_pubkey': 'null',
143 'recaptcha_response': 'null',
145 'identifier': username,
148 'signature': self.sign(username, password, self._CLIENT_ID),
149 'device_id': genDevId(),
150 'user_agent': self._USER_AGENT
153 query = self._API_AUTH_QUERY_TEMPLATE % self._CLIENT_ID
154 login = sanitized_Request(self._API_AUTH_URL_PW % query, json.dumps(payload).encode('utf-8'))
155 response = self._download_json(login, None)
156 self._access_token = response.get('session').get('access_token')
157 if not self._access_token:
158 self.report_warning('Unable to get access token, login may has failed')
160 self._HEADERS = {'Authorization': 'OAuth ' + self._access_token}
163 # signature generation
164 def sign(self, user, pw, clid):
174 n = '0763ed7314c69015fd4a0dc16bbf4b90' # _KEY
176 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
178 t = clid # _CLIENT_ID
180 d = '-'.join([str(mInt) for mInt in [a, i, s, w, u, l, b, k]])
181 p = n + y + d + r + e + t + d + n
187 for f in range(f, len(h)):
188 m = (m >> 1) + ((1 & m) << 23)
192 # c is not even needed
193 out = str(y) + ':' + str(d) + ':' + format(m, 'x') + ':' + str(c)
197 def _extract_info_dict(self, info, full_title=None, secret_token=None, extract_flat=False):
198 track_id = compat_str(info['id'])
199 title = info['title']
203 query = {'client_id': self._CLIENT_ID}
205 query['secret_token'] = secret_token
207 if not extract_flat and info.get('downloadable') and info.get('has_downloads_left'):
208 download_url = update_url_query(
209 self._API_V2_BASE + 'tracks/' + track_id + '/download', query)
210 redirect_url = (self._download_json(download_url, track_id, fatal=False) or {}).get('redirectUri')
212 urlh = self._request_webpage(
213 HEADRequest(redirect_url), track_id, fatal=False)
215 format_url = urlh.geturl()
216 format_urls.add(format_url)
218 'format_id': 'download',
219 'ext': urlhandle_detect_ext(urlh) or 'mp3',
220 'filesize': int_or_none(urlh.headers.get('Content-Length')),
225 def invalid_url(url):
226 return not url or url in format_urls
228 def add_format(f, protocol, is_preview=False):
229 mobj = re.search(r'\.(?P<abr>\d+)\.(?P<ext>[0-9a-z] {3,4} )(?=[/?])', stream_url)
231 for k, v in mobj.groupdict().items():
236 format_id_list.append(protocol)
240 for k in ('ext', 'abr'):
243 format_id_list.append(v)
244 preview = is_preview or re.search(r'/(?:preview|playlist)/0/30/', f['url'])
246 format_id_list.append('preview')
250 if protocol == 'hls':
251 protocol = 'm3u8' if ext == 'aac' else 'm3u8_native'
255 'format_id': '_'.join(format_id_list),
256 'protocol': protocol,
257 'preference': -10 if preview else None,
262 transcodings = try_get(
263 info, lambda x: x['media']['transcodings'], list) or []
264 for t in transcodings:
265 if not isinstance(t, dict):
267 format_url = url_or_none(t.get('url'))
270 stream = None if extract_flat else self._download_json(
271 format_url, track_id, query=query, fatal=False, headers=self._HEADERS)
272 if not isinstance(stream, dict):
274 stream_url = url_or_none(stream.get('url'))
275 if invalid_url(stream_url):
277 format_urls.add(stream_url)
278 stream_format = t.get('format') or {}
279 protocol = stream_format.get('protocol')
280 if protocol != 'hls' and '/hls' in format_url:
283 preset = str_or_none(t.get('preset'))
285 ext = preset.split('_')[0]
286 if ext not in KNOWN_EXTENSIONS:
287 ext = mimetype2ext(stream_format.get('mime_type'))
291 }, 'http' if protocol == 'progressive' else protocol,
292 t.get('snipped') or '/preview/' in format_url)
297 if not formats and info.get('policy') == 'BLOCK':
298 self.raise_geo_restricted(metadata_available=True)
300 user = info.get('user') or {}
303 artwork_url = info.get('artwork_url')
304 thumbnail = artwork_url or user.get('avatar_url')
305 if isinstance(thumbnail, compat_str):
306 if re.search(self._IMAGE_REPL_RE, thumbnail):
307 for image_id, size in self._ARTWORK_MAP.items():
310 'url': re.sub(self._IMAGE_REPL_RE, '- %s .jpg' % image_id, thumbnail),
312 if image_id == 'tiny' and not artwork_url:
314 elif image_id == 'original':
323 thumbnails = [ {'url': thumbnail} ]
325 def extract_count(key):
326 return int_or_none(info.get(' %s _count' % key))
330 'uploader': user.get('username'),
331 'uploader_id': str_or_none(user.get('id')) or user.get('permalink'),
332 'uploader_url': user.get('permalink_url'),
333 'timestamp': unified_timestamp(info.get('created_at')),
335 'description': info.get('description'),
336 'thumbnails': thumbnails,
337 'duration': float_or_none(info.get('duration'), 1000),
338 'webpage_url': info.get('permalink_url'),
339 'license': info.get('license'),
340 'view_count': extract_count('playback'),
341 'like_count': extract_count('favoritings') or extract_count('likes'),
342 'comment_count': extract_count('comment'),
343 'repost_count': extract_count('reposts'),
344 'genre': info.get('genre'),
345 'formats': formats if not extract_flat else None
349 def _resolv_url(cls, url):
350 return cls._API_V2_BASE + 'resolve?url=' + url
353 class SoundcloudIE(SoundcloudBaseIE):
354 """Information extractor for soundcloud.com
355 To access the media, the uid of the song and a stream token
356 must be extracted from the page source and the script must make
357 a request to media.soundcloud.com/crossdomain.xml. Then
358 the media can be grabbed by requesting from an url composed
359 of the stream token and uid
362 _VALID_URL = r'''(?x)^(?:https?://)?
363 (?:(?:(?:www\.|m\.)?soundcloud\.com/
365 (?P<uploader>[\w\d-]+)/
366 (?!(?:tracks|albums|sets(?:/.+?)?|reposts|likes|spotlight)/?(?:$|[?#]))
368 (?:/(?P<token>(?!(?:albums|sets|recommended))[^?]+?))?
370 |(?:api(?:-v2)?\.soundcloud\.com/tracks/(?P<track_id>\d+)
371 (?:/? \? secret_token=(?P<secret_token>[^&]+))?)
374 IE_NAME = 'soundcloud'
377 'url': 'http://soundcloud.com/ethmusic/lostin-powers-she-so-heavy',
378 'md5': 'ebef0a451b909710ed1d7787dddbf0d7',
382 'title': 'Lostin Powers - She so Heavy (SneakPreview) Adrian Ackers Blueprint 1',
383 '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',
384 'uploader': 'E.T. ExTerrestrial Music',
385 'uploader_id': '1571244',
386 'timestamp': 1349920598,
387 'upload_date': '20121011',
389 'license': 'all-rights-reserved',
392 'comment_count': int,
398 'url': 'https://soundcloud.com/the-concept-band/goldrushed-mastered?in=the-concept-band/sets/the-royal-concept-ep',
402 'title': 'Goldrushed',
403 'description': 'From Stockholm Sweden \r\n Povel / Magnus / Filip / David \r\n www.theroyalconcept.com',
404 'uploader': 'The Royal Concept',
405 'uploader_id': '9615865',
406 'timestamp': 1337635207,
407 'upload_date': '20120521',
409 'license': 'all-rights-reserved',
412 'comment_count': int,
418 'url': 'https://soundcloud.com/jaimemf/youtube-dl-test-video-a-y-baw/s-8Pjrp',
419 'md5': 'aa0dd32bfea9b0c5ef4f02aacd080604',
423 'title': 'Youtube - Dl Test Video \'\' Äâ†',
424 'description': 'test chars: \"\' / \\ äâ†',
425 'uploader': 'jaimeMF',
426 'uploader_id': '69767071',
427 'timestamp': 1386604920,
428 'upload_date': '20131209',
430 'license': 'all-rights-reserved',
433 'comment_count': int,
437 # private link (alt format)
439 'url': 'https://api.soundcloud.com/tracks/123998367?secret_token=s-8Pjrp',
440 'md5': 'aa0dd32bfea9b0c5ef4f02aacd080604',
444 'title': 'Youtube - Dl Test Video \'\' Äâ†',
445 'description': 'test chars: \"\' / \\ äâ†',
446 'uploader': 'jaimeMF',
447 'uploader_id': '69767071',
448 'timestamp': 1386604920,
449 'upload_date': '20131209',
451 'license': 'all-rights-reserved',
454 'comment_count': int,
460 'url': 'https://soundcloud.com/the80m/the-following',
461 'md5': '9ffcddb08c87d74fb5808a3c183a1d04',
467 # private link, downloadable format
469 'url': 'https://soundcloud.com/oriuplift/uponly-238-no-talking-wav/s-AyZUd',
470 'md5': '64a60b16e617d41d0bef032b7f55441e',
474 'title': 'Uplifting Only 238 [No Talking] (incl. Alex Feed Guestmix) (Aug 31, 2017) [wav]',
475 'description': 'md5:fa20ee0fca76a3d6df8c7e57f3715366',
476 'uploader': 'Ori Uplift Music',
477 'uploader_id': '12563093',
478 'timestamp': 1504206263,
479 'upload_date': '20170831',
480 'duration': 7449.096,
481 'license': 'all-rights-reserved',
484 'comment_count': int,
488 # no album art, use avatar pic for thumbnail
490 'url': 'https://soundcloud.com/garyvee/sideways-prod-mad-real',
491 'md5': '59c7872bc44e5d99b7211891664760c2',
495 'title': 'Sideways (Prod. Mad Real)',
496 'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
497 'uploader': 'garyvee',
498 'uploader_id': '2366352',
499 'timestamp': 1488152409,
500 'upload_date': '20170226',
502 'thumbnail': r're:https?://.*\.jpg',
503 'license': 'all-rights-reserved',
506 'comment_count': int,
510 'skip_download': True,
514 'url': 'https://soundcloud.com/giovannisarani/mezzo-valzer',
515 'md5': 'e22aecd2bc88e0e4e432d7dcc0a1abf7',
519 'title': 'Mezzo Valzer',
520 'description': 'md5:4138d582f81866a530317bae316e8b61',
521 'uploader': 'Micronie',
522 'uploader_id': '3352531',
523 'timestamp': 1551394171,
524 'upload_date': '20190228',
526 'thumbnail': r're:https?://.*\.jpg',
527 'license': 'all-rights-reserved',
530 'comment_count': int,
535 # AAC HQ format available (account with active subscription needed)
536 'url': 'https://soundcloud.com/wandw/the-chainsmokers-ft-daya-dont-let-me-down-ww-remix-1',
537 'only_matching': True,
540 # Go+ (account with active subscription needed)
541 'url': 'https://soundcloud.com/taylorswiftofficial/look-what-you-made-me-do',
542 'only_matching': True,
546 def _real_extract(self, url):
547 mobj = self._match_valid_url(url)
549 track_id = mobj.group('track_id')
553 info_json_url = self._API_V2_BASE + 'tracks/' + track_id
554 full_title = track_id
555 token = mobj.group('secret_token')
557 query['secret_token'] = token
559 full_title = resolve_title = ' %s / %s ' % mobj.group('uploader', 'title')
560 token = mobj.group('token')
562 resolve_title += '/ %s ' % token
563 info_json_url = self._resolv_url(self._BASE_URL + resolve_title)
565 info = self._download_json(
566 info_json_url, full_title, 'Downloading info JSON', query=query, headers=self._HEADERS)
568 return self._extract_info_dict(info, full_title, token)
571 class SoundcloudPlaylistBaseIE(SoundcloudBaseIE):
572 def _extract_set(self, playlist, token=None):
573 playlist_id = compat_str(playlist['id'])
574 tracks = playlist.get('tracks') or []
575 if not all([t.get('permalink_url') for t in tracks]) and token:
576 tracks = self._download_json(
577 self._API_V2_BASE + 'tracks', playlist_id,
578 'Downloading tracks', query={
579 'ids': ','.join([compat_str(t['id']) for t in tracks]),
580 'playlistId': playlist_id,
581 'playlistSecretToken': token,
582 }, headers=self._HEADERS)
585 track_id = str_or_none(track.get('id'))
586 url = track.get('permalink_url')
590 url = self._API_V2_BASE + 'tracks/' + track_id
592 url += '?secret_token=' + token
593 entries.append(self.url_result(
594 url, SoundcloudIE.ie_key(), track_id))
595 return self.playlist_result(
596 entries, playlist_id,
597 playlist.get('title'),
598 playlist.get('description'))
601 class SoundcloudSetIE(SoundcloudPlaylistBaseIE):
602 _VALID_URL = r'https?://(?:(?:www|m)\.)?soundcloud\.com/(?P<uploader>[\w\d-]+)/sets/(?P<slug_title>[:\w\d-]+)(?:/(?P<token>[^?/]+))?'
603 IE_NAME = 'soundcloud:set'
605 'url': 'https://soundcloud.com/the-concept-band/sets/the-royal-concept-ep',
608 'title': 'The Royal Concept EP',
609 'description': 'md5:71d07087c7a449e8941a70a29e34671e',
611 'playlist_mincount': 5,
613 'url': 'https://soundcloud.com/the-concept-band/sets/the-royal-concept-ep/token',
614 'only_matching': True,
616 'url': 'https://soundcloud.com/discover/sets/weekly::flacmatic',
617 'only_matching': True,
619 'url': 'https://soundcloud.com/discover/sets/charts-top:all-music:de',
620 'only_matching': True,
622 'url': 'https://soundcloud.com/discover/sets/charts-top:hiphoprap:kr',
623 'only_matching': True,
626 def _real_extract(self, url):
627 mobj = self._match_valid_url(url)
629 full_title = ' %s /sets/ %s ' % mobj.group('uploader', 'slug_title')
630 token = mobj.group('token')
632 full_title += '/' + token
634 info = self._download_json(self._resolv_url(
635 self._BASE_URL + full_title), full_title, headers=self._HEADERS)
638 msgs = (compat_str(err['error_message']) for err in info['errors'])
639 raise ExtractorError('unable to download video webpage: %s ' % ','.join(msgs))
641 return self._extract_set(info, token)
644 class SoundcloudPagedPlaylistBaseIE(SoundcloudBaseIE):
645 def _extract_playlist(self, base_url, playlist_id, playlist_title):
649 'title': playlist_title,
650 'entries': self._entries(base_url, playlist_id),
653 def _entries(self, url, playlist_id):
654 # Per the SoundCloud documentation, the maximum limit for a linked partitioning query is 200.
655 # https://developers.soundcloud.com/blog/offset-pagination-deprecated
658 'linked_partitioning': '1',
662 for i in itertools.count():
663 for retry in self.RetryManager():
665 response = self._download_json(
666 url, playlist_id, query=query, headers=self._HEADERS,
667 note=f'Downloading track page {i + 1} ')
669 except ExtractorError as e:
670 # Downloading page may result in intermittent 502 HTTP error
671 # See https://github.com/yt-dlp/yt-dlp/issues/872
672 if not isinstance(e.cause, compat_HTTPError) or e.cause.code != 502:
677 def resolve_entry(*candidates):
678 for cand in candidates:
679 if not isinstance(cand, dict):
681 permalink_url = url_or_none(cand.get('permalink_url'))
683 return self.url_result(
685 SoundcloudIE.ie_key() if SoundcloudIE.suitable(permalink_url) else None,
686 str_or_none(cand.get('id')), cand.get('title'))
688 for e in response['collection'] or []:
689 yield resolve_entry(e, e.get('track'), e.get('playlist'))
691 url = response.get('next_href')
694 query.pop('offset', None)
697 class SoundcloudUserIE(SoundcloudPagedPlaylistBaseIE):
698 _VALID_URL = r'''(?x)
700 (?:(?:www|m)\.)?soundcloud\.com/
703 (?P<rsrc>tracks|albums|sets|reposts|likes|spotlight)
707 IE_NAME = 'soundcloud:user'
709 'url': 'https://soundcloud.com/soft-cell-official',
712 'title': 'Soft Cell (All)',
714 'playlist_mincount': 28,
716 'url': 'https://soundcloud.com/soft-cell-official/tracks',
719 'title': 'Soft Cell (Tracks)',
721 'playlist_mincount': 27,
723 'url': 'https://soundcloud.com/soft-cell-official/albums',
726 'title': 'Soft Cell (Albums)',
728 'playlist_mincount': 1,
730 'url': 'https://soundcloud.com/jcv246/sets',
733 'title': 'Jordi / cv (Sets)',
735 'playlist_mincount': 2,
737 'url': 'https://soundcloud.com/jcv246/reposts',
740 'title': 'Jordi / cv (Reposts)',
742 'playlist_mincount': 6,
744 'url': 'https://soundcloud.com/clalberg/likes',
747 'title': 'clalberg (Likes)',
749 'playlist_mincount': 5,
751 'url': 'https://soundcloud.com/grynpyret/spotlight',
754 'title': 'Grynpyret (Spotlight)',
756 'playlist_mincount': 1,
760 'all': 'stream/users/ %s ',
761 'tracks': 'users/ %s /tracks',
762 'albums': 'users/ %s /albums',
763 'sets': 'users/ %s /playlists',
764 'reposts': 'stream/users/ %s /reposts',
765 'likes': 'users/ %s /likes',
766 'spotlight': 'users/ %s /spotlight',
769 def _real_extract(self, url):
770 mobj = self._match_valid_url(url)
771 uploader = mobj.group('user')
773 user = self._download_json(
774 self._resolv_url(self._BASE_URL + uploader),
775 uploader, 'Downloading user info', headers=self._HEADERS)
777 resource = mobj.group('rsrc') or 'all'
779 return self._extract_playlist(
780 self._API_V2_BASE + self._BASE_URL_MAP[resource] % user['id'],
781 str_or_none(user.get('id')),
782 ' %s ( %s )' % (user['username'], resource.capitalize()))
785 class SoundcloudUserPermalinkIE(SoundcloudPagedPlaylistBaseIE):
786 _VALID_URL = r'https?://api\.soundcloud\.com/users/(?P<id>\d+)'
787 IE_NAME = 'soundcloud:user:permalink'
789 'url': 'https://api.soundcloud.com/users/30909869',
794 'playlist_mincount': 23,
797 def _real_extract(self, url):
798 user_id = self._match_id(url)
799 user = self._download_json(
800 self._resolv_url(url), user_id, 'Downloading user info', headers=self._HEADERS)
802 return self._extract_playlist(
803 f' {self._API_V2_BASE} stream/users/ {user["id"]} ', str(user['id']), user.get('username'))
806 class SoundcloudTrackStationIE(SoundcloudPagedPlaylistBaseIE):
807 _VALID_URL = r'https?://(?:(?:www|m)\.)?soundcloud\.com/stations/track/[^/]+/(?P<id>[^/?#&]+)'
808 IE_NAME = 'soundcloud:trackstation'
810 'url': 'https://soundcloud.com/stations/track/officialsundial/your-text',
813 'title': 'Track station: your text',
815 'playlist_mincount': 47,
818 def _real_extract(self, url):
819 track_name = self._match_id(url)
821 track = self._download_json(self._resolv_url(url), track_name, headers=self._HEADERS)
822 track_id = self._search_regex(
823 r'soundcloud:track-stations:(\d+)', track['id'], 'track id')
825 return self._extract_playlist(
826 self._API_V2_BASE + 'stations/ %s /tracks' % track['id'],
827 track_id, 'Track station: %s ' % track['title'])
830 class SoundcloudRelatedIE(SoundcloudPagedPlaylistBaseIE):
831 _VALID_URL = r'https?://(?:(?:www|m)\.)?soundcloud\.com/(?P<slug>[\w\d-]+/[\w\d-]+)/(?P<relation>albums|sets|recommended)'
832 IE_NAME = 'soundcloud:related'
834 'url': 'https://soundcloud.com/wajang/sexapil-pingers-5/recommended',
837 'title': 'Sexapil - Pingers 5 (Recommended)',
839 'playlist_mincount': 50,
841 'url': 'https://soundcloud.com/wajang/sexapil-pingers-5/albums',
844 'title': 'Sexapil - Pingers 5 (Albums)',
846 'playlist_mincount': 1,
848 'url': 'https://soundcloud.com/wajang/sexapil-pingers-5/sets',
851 'title': 'Sexapil - Pingers 5 (Sets)',
853 'playlist_mincount': 4,
857 'albums': 'tracks/ %s /albums',
858 'sets': 'tracks/ %s /playlists_without_albums',
859 'recommended': 'tracks/ %s /related',
862 def _real_extract(self, url):
863 slug, relation = self._match_valid_url(url).group('slug', 'relation')
865 track = self._download_json(
866 self._resolv_url(self._BASE_URL + slug),
867 slug, 'Downloading track info', headers=self._HEADERS)
869 if track.get('errors'):
870 raise ExtractorError(f' {self.IE_NAME} said: %s ' % ','.join(
871 str(err['error_message']) for err in track['errors']), expected=True)
873 return self._extract_playlist(
874 self._API_V2_BASE + self._BASE_URL_MAP[relation] % track['id'], str(track['id']),
875 ' %s ( %s )' % (track.get('title') or slug, relation.capitalize()))
878 class SoundcloudPlaylistIE(SoundcloudPlaylistBaseIE):
879 _VALID_URL = r'https?://api(?:-v2)?\.soundcloud\.com/playlists/(?P<id>[0-9]+)(?:/? \? secret_token=(?P<token>[^&]+?))?$'
880 IE_NAME = 'soundcloud:playlist'
882 'url': 'https://api.soundcloud.com/playlists/4110309',
885 'title': 'TILT Brass - Bowery Poetry Club, August \' 03 [Non-Site SCR 02]',
886 'description': 're:.*?TILT Brass - Bowery Poetry Club',
891 def _real_extract(self, url):
892 mobj = self._match_valid_url(url)
893 playlist_id = mobj.group('id')
896 token = mobj.group('token')
898 query['secret_token'] = token
900 data = self._download_json(
901 self._API_V2_BASE + 'playlists/' + playlist_id,
902 playlist_id, 'Downloading playlist', query=query, headers=self._HEADERS)
904 return self._extract_set(data, token)
907 class SoundcloudSearchIE(SoundcloudBaseIE, SearchInfoExtractor):
908 IE_NAME = 'soundcloud:search'
909 IE_DESC = 'Soundcloud search'
910 _SEARCH_KEY = 'scsearch'
912 'url': 'scsearch15:post-avant jazzcore',
914 'id': 'post-avant jazzcore',
915 'title': 'post-avant jazzcore',
917 'playlist_count': 15,
920 _MAX_RESULTS_PER_PAGE = 200
921 _DEFAULT_RESULTS_PER_PAGE = 50
923 def _get_collection(self, endpoint, collection_id, **query):
925 query.get('limit', self._DEFAULT_RESULTS_PER_PAGE),
926 self._MAX_RESULTS_PER_PAGE)
929 'linked_partitioning': 1,
932 next_url = update_url_query(self._API_V2_BASE + endpoint, query)
934 for i in itertools.count(1):
935 response = self._download_json(
936 next_url, collection_id, f'Downloading page {i} ',
937 'Unable to download API page', headers=self._HEADERS)
939 for item in response.get('collection') or []:
941 yield self.url_result(
942 item['uri'], SoundcloudIE.ie_key(), **self._extract_info_dict(item, extract_flat=True))
944 next_url = response.get('next_href')
948 def _get_n_results(self, query, n):
949 return self.playlist_result(itertools.islice(
950 self._get_collection('search/tracks', query, limit=n, q=query),
951 0, None if n == float('inf') else n), query, query)