]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/soundcloud.py
10 from .. compat
import (
34 class SoundcloudEmbedIE ( InfoExtractor
):
35 _VALID_URL
= r
'https?://(?:w|player|p)\.soundcloud\.com/player/?.*?\burl=(?P<id>.+)'
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%3A%2F%2Fapi.soundcloud.com%2Fplaylists%2F922213810&show_artwork=true&maxwidth=640&maxheight=960&dnt=1&secret_token=s-ziYey' ,
39 'only_matching' : True ,
43 def _extract_urls ( webpage
):
44 return [ m
. group ( 'url' ) for m
in re
. finditer (
45 r
'<iframe[^>]+src=(["\' ])( ?P
< url
>( ?
: https?
://) ?
( ?
: w\
.) ?soundcloud\
. com
/ player
.+ ?
) \
1 ',
48 def _real_extract(self, url):
50 api_url = query[' url
'][0]
51 secret_token = query.get(' secret_token
')
53 api_url = update_url_query(api_url, {'secret_token': secret_token[0]} )
54 return self.url_result(api_url)
57 class SoundcloudBaseIE(InfoExtractor):
58 _NETRC_MACHINE = ' soundcloud
'
60 _API_V2_BASE = ' https
:// api
- v2
. soundcloud
. com
/ '
61 _BASE_URL = ' https
:// soundcloud
. com
/ '
62 _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 '
63 _API_AUTH_QUERY_TEMPLATE = ' ?client_id
= %s '
64 _API_AUTH_URL_PW = ' https
:// api
- auth
. soundcloud
. com
/ web
- auth
/ sign
- in / password
%s '
65 _API_VERIFY_AUTH_TOKEN = ' https
:// api
- auth
. soundcloud
. com
/ connect
/ session
%s '
69 def _store_client_id(self, client_id):
70 self._downloader.cache.store(' soundcloud
', ' client_id
', client_id)
72 def _update_client_id(self):
73 webpage = self._download_webpage(' https
:// soundcloud
. com
/ ', None)
74 for src in reversed(re.findall(r' < script
[ ^
>]+ src
= "([^" ]+) "', webpage)):
75 script = self._download_webpage(src, None, fatal=False)
77 client_id = self._search_regex(
78 r'client_id\s*:\s*" ([ 0 - 9 a
- zA
- Z
] {32}
) "',
79 script, 'client id', default=None)
81 self._CLIENT_ID = client_id
82 self._store_client_id(client_id)
84 raise ExtractorError('Unable to extract client id')
86 def _download_json(self, *args, **kwargs):
87 non_fatal = kwargs.get('fatal') is False
90 query = kwargs.get('query', {}).copy()
92 query['client_id'] = self._CLIENT_ID
93 kwargs['query'] = query
95 return super()._download_json(*args, **kwargs)
96 except ExtractorError as e:
97 if isinstance(e.cause, compat_HTTPError) and e.cause.code in (401, 403):
98 self._store_client_id(None)
99 self._update_client_id()
102 self.report_warning(error_to_compat_str(e))
106 def _initialize_pre_login(self):
107 self._CLIENT_ID = self._downloader.cache.load('soundcloud', 'client_id') or 'a3e059563d7fd3372b49b37f00a00bcf'
109 def _perform_login(self, username, password):
110 if username != 'oauth':
112 'Login using username and password is not currently supported. '
113 'Use " -- username oauth
-- password
< oauth_token
> " to login using an oauth token')
114 self._access_token = password
115 query = self._API_AUTH_QUERY_TEMPLATE % self._CLIENT_ID
116 payload = {'session': {'access_token': self._access_token} }
117 token_verification = sanitized_Request(self._API_VERIFY_AUTH_TOKEN % query, json.dumps(payload).encode('utf-8'))
118 response = self._download_json(token_verification, None, note='Verifying login token...', fatal=False)
119 if response is not False:
120 self._HEADERS = {'Authorization': 'OAuth ' + self._access_token}
123 self.report_warning('Provided authorization token seems to be invalid. Continue as guest')
128 return ''.join([str(random.randrange(10)) for i in range(6)])
129 return '-'.join([genNumBlock() for i in range(4)])
132 'client_id': self._CLIENT_ID,
133 'recaptcha_pubkey': 'null',
134 'recaptcha_response': 'null',
136 'identifier': username,
139 'signature': self.sign(username, password, self._CLIENT_ID),
140 'device_id': genDevId(),
141 'user_agent': self._USER_AGENT
144 query = self._API_AUTH_QUERY_TEMPLATE % self._CLIENT_ID
145 login = sanitized_Request(self._API_AUTH_URL_PW % query, json.dumps(payload).encode('utf-8'))
146 response = self._download_json(login, None)
147 self._access_token = response.get('session').get('access_token')
148 if not self._access_token:
149 self.report_warning('Unable to get access token, login may has failed')
151 self._HEADERS = {'Authorization': 'OAuth ' + self._access_token}
154 # signature generation
155 def sign(self, user, pw, clid):
165 n = '0763ed7314c69015fd4a0dc16bbf4b90' # _KEY
167 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
169 t = clid # _CLIENT_ID
171 d = '-'.join([str(mInt) for mInt in [a, i, s, w, u, l, b, k]])
172 p = n + y + d + r + e + t + d + n
178 for f in range(f, len(h)):
179 m = (m >> 1) + ((1 & m) << 23)
183 # c is not even needed
184 out = str(y) + ':' + str(d) + ':' + format(m, 'x') + ':' + str(c)
189 def _resolv_url(cls, url):
190 return cls._API_V2_BASE + 'resolve?url=' + url
193 class SoundcloudIE(SoundcloudBaseIE):
194 """Information extractor for soundcloud.com
195 To access the media, the uid of the song and a stream token
196 must be extracted from the page source and the script must make
197 a request to media.soundcloud.com/crossdomain.xml. Then
198 the media can be grabbed by requesting from an url composed
199 of the stream token and uid
202 _VALID_URL = r'''(?x)^(?:https?://)?
203 (?:(?:(?:www\.|m\.)?soundcloud\.com/
205 (?P<uploader>[\w\d-]+)/
206 (?!(?:tracks|albums|sets(?:/.+?)?|reposts|likes|spotlight)/?(?:$|[?#]))
208 (?:/(?P<token>(?!(?:albums|sets|recommended))[^?]+?))?
210 |(?:api(?:-v2)?\.soundcloud\.com/tracks/(?P<track_id>\d+)
211 (?:/? \? secret_token=(?P<secret_token>[^&]+))?)
214 IE_NAME = 'soundcloud'
217 'url': 'http://soundcloud.com/ethmusic/lostin-powers-she-so-heavy',
218 'md5': 'ebef0a451b909710ed1d7787dddbf0d7',
222 'title': 'Lostin Powers - She so Heavy (SneakPreview) Adrian Ackers Blueprint 1',
223 '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',
224 'uploader': 'E.T. ExTerrestrial Music',
225 'uploader_id': '1571244',
226 'timestamp': 1349920598,
227 'upload_date': '20121011',
229 'license': 'all-rights-reserved',
232 'comment_count': int,
238 'url': 'https://soundcloud.com/the-concept-band/goldrushed-mastered?in=the-concept-band/sets/the-royal-concept-ep',
242 'title': 'Goldrushed',
243 'description': 'From Stockholm Sweden \r\n Povel / Magnus / Filip / David \r\n www.theroyalconcept.com',
244 'uploader': 'The Royal Concept',
245 'uploader_id': '9615865',
246 'timestamp': 1337635207,
247 'upload_date': '20120521',
249 'license': 'all-rights-reserved',
252 'comment_count': int,
258 'url': 'https://soundcloud.com/jaimemf/youtube-dl-test-video-a-y-baw/s-8Pjrp',
259 'md5': 'aa0dd32bfea9b0c5ef4f02aacd080604',
263 'title': 'Youtube - Dl Test Video \'\' Äâ†',
264 'description': 'test chars: \"\' / \\ äâ†',
265 'uploader': 'jaimeMF',
266 'uploader_id': '69767071',
267 'timestamp': 1386604920,
268 'upload_date': '20131209',
270 'license': 'all-rights-reserved',
273 'comment_count': int,
277 # private link (alt format)
279 'url': 'https://api.soundcloud.com/tracks/123998367?secret_token=s-8Pjrp',
280 'md5': 'aa0dd32bfea9b0c5ef4f02aacd080604',
284 'title': 'Youtube - Dl Test Video \'\' Äâ†',
285 'description': 'test chars: \"\' / \\ äâ†',
286 'uploader': 'jaimeMF',
287 'uploader_id': '69767071',
288 'timestamp': 1386604920,
289 'upload_date': '20131209',
291 'license': 'all-rights-reserved',
294 'comment_count': int,
300 'url': 'https://soundcloud.com/the80m/the-following',
301 'md5': '9ffcddb08c87d74fb5808a3c183a1d04',
307 # private link, downloadable format
309 'url': 'https://soundcloud.com/oriuplift/uponly-238-no-talking-wav/s-AyZUd',
310 'md5': '64a60b16e617d41d0bef032b7f55441e',
314 'title': 'Uplifting Only 238 [No Talking] (incl. Alex Feed Guestmix) (Aug 31, 2017) [wav]',
315 'description': 'md5:fa20ee0fca76a3d6df8c7e57f3715366',
316 'uploader': 'Ori Uplift Music',
317 'uploader_id': '12563093',
318 'timestamp': 1504206263,
319 'upload_date': '20170831',
320 'duration': 7449.096,
321 'license': 'all-rights-reserved',
324 'comment_count': int,
328 # no album art, use avatar pic for thumbnail
330 'url': 'https://soundcloud.com/garyvee/sideways-prod-mad-real',
331 'md5': '59c7872bc44e5d99b7211891664760c2',
335 'title': 'Sideways (Prod. Mad Real)',
336 'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
337 'uploader': 'garyvee',
338 'uploader_id': '2366352',
339 'timestamp': 1488152409,
340 'upload_date': '20170226',
342 'thumbnail': r're:https?://.*\.jpg',
343 'license': 'all-rights-reserved',
346 'comment_count': int,
350 'skip_download': True,
354 'url': 'https://soundcloud.com/giovannisarani/mezzo-valzer',
355 'md5': 'e22aecd2bc88e0e4e432d7dcc0a1abf7',
359 'title': 'Mezzo Valzer',
360 'description': 'md5:4138d582f81866a530317bae316e8b61',
361 'uploader': 'Micronie',
362 'uploader_id': '3352531',
363 'timestamp': 1551394171,
364 'upload_date': '20190228',
366 'thumbnail': r're:https?://.*\.jpg',
367 'license': 'all-rights-reserved',
370 'comment_count': int,
375 # AAC HQ format available (account with active subscription needed)
376 'url': 'https://soundcloud.com/wandw/the-chainsmokers-ft-daya-dont-let-me-down-ww-remix-1',
377 'only_matching': True,
380 # Go+ (account with active subscription needed)
381 'url': 'https://soundcloud.com/taylorswiftofficial/look-what-you-made-me-do',
382 'only_matching': True,
386 _IMAGE_REPL_RE = r'-([0-9a-z]+)\.jpg'
401 def _extract_info_dict(self, info, full_title=None, secret_token=None):
402 track_id = compat_str(info['id'])
403 title = info['title']
407 query = {'client_id': self._CLIENT_ID}
409 query['secret_token'] = secret_token
411 if info.get('downloadable') and info.get('has_downloads_left'):
412 download_url = update_url_query(
413 self._API_V2_BASE + 'tracks/' + track_id + '/download', query)
414 redirect_url = (self._download_json(download_url, track_id, fatal=False) or {}).get('redirectUri')
416 urlh = self._request_webpage(
417 HEADRequest(redirect_url), track_id, fatal=False)
419 format_url = urlh.geturl()
420 format_urls.add(format_url)
422 'format_id': 'download',
423 'ext': urlhandle_detect_ext(urlh) or 'mp3',
424 'filesize': int_or_none(urlh.headers.get('Content-Length')),
429 def invalid_url(url):
430 return not url or url in format_urls
432 def add_format(f, protocol, is_preview=False):
433 mobj = re.search(r'\.(?P<abr>\d+)\.(?P<ext>[0-9a-z] {3,4} )(?=[/?])', stream_url)
435 for k, v in mobj.groupdict().items():
440 format_id_list.append(protocol)
444 for k in ('ext', 'abr'):
447 format_id_list.append(v)
448 preview = is_preview or re.search(r'/(?:preview|playlist)/0/30/', f['url'])
450 format_id_list.append('preview')
454 if protocol == 'hls':
455 protocol = 'm3u8' if ext == 'aac' else 'm3u8_native'
459 'format_id': '_'.join(format_id_list),
460 'protocol': protocol,
461 'preference': -10 if preview else None,
466 transcodings = try_get(
467 info, lambda x: x['media']['transcodings'], list) or []
468 for t in transcodings:
469 if not isinstance(t, dict):
471 format_url = url_or_none(t.get('url'))
474 stream = self._download_json(
475 format_url, track_id, query=query, fatal=False, headers=self._HEADERS)
476 if not isinstance(stream, dict):
478 stream_url = url_or_none(stream.get('url'))
479 if invalid_url(stream_url):
481 format_urls.add(stream_url)
482 stream_format = t.get('format') or {}
483 protocol = stream_format.get('protocol')
484 if protocol != 'hls' and '/hls' in format_url:
487 preset = str_or_none(t.get('preset'))
489 ext = preset.split('_')[0]
490 if ext not in KNOWN_EXTENSIONS:
491 ext = mimetype2ext(stream_format.get('mime_type'))
495 }, 'http' if protocol == 'progressive' else protocol,
496 t.get('snipped') or '/preview/' in format_url)
501 if not formats and info.get('policy') == 'BLOCK':
502 self.raise_geo_restricted(metadata_available=True)
503 self._sort_formats(formats)
505 user = info.get('user') or {}
508 artwork_url = info.get('artwork_url')
509 thumbnail = artwork_url or user.get('avatar_url')
510 if isinstance(thumbnail, compat_str):
511 if re.search(self._IMAGE_REPL_RE, thumbnail):
512 for image_id, size in self._ARTWORK_MAP.items():
515 'url': re.sub(self._IMAGE_REPL_RE, '- %s .jpg' % image_id, thumbnail),
517 if image_id == 'tiny' and not artwork_url:
519 elif image_id == 'original':
528 thumbnails = [ {'url': thumbnail} ]
530 def extract_count(key):
531 return int_or_none(info.get(' %s _count' % key))
535 'uploader': user.get('username'),
536 'uploader_id': str_or_none(user.get('id')) or user.get('permalink'),
537 'uploader_url': user.get('permalink_url'),
538 'timestamp': unified_timestamp(info.get('created_at')),
540 'description': info.get('description'),
541 'thumbnails': thumbnails,
542 'duration': float_or_none(info.get('duration'), 1000),
543 'webpage_url': info.get('permalink_url'),
544 'license': info.get('license'),
545 'view_count': extract_count('playback'),
546 'like_count': extract_count('favoritings') or extract_count('likes'),
547 'comment_count': extract_count('comment'),
548 'repost_count': extract_count('reposts'),
549 'genre': info.get('genre'),
553 def _real_extract(self, url):
554 mobj = self._match_valid_url(url)
556 track_id = mobj.group('track_id')
560 info_json_url = self._API_V2_BASE + 'tracks/' + track_id
561 full_title = track_id
562 token = mobj.group('secret_token')
564 query['secret_token'] = token
566 full_title = resolve_title = ' %s / %s ' % mobj.group('uploader', 'title')
567 token = mobj.group('token')
569 resolve_title += '/ %s ' % token
570 info_json_url = self._resolv_url(self._BASE_URL + resolve_title)
572 info = self._download_json(
573 info_json_url, full_title, 'Downloading info JSON', query=query, headers=self._HEADERS)
575 return self._extract_info_dict(info, full_title, token)
578 class SoundcloudPlaylistBaseIE(SoundcloudBaseIE):
579 def _extract_set(self, playlist, token=None):
580 playlist_id = compat_str(playlist['id'])
581 tracks = playlist.get('tracks') or []
582 if not all([t.get('permalink_url') for t in tracks]) and token:
583 tracks = self._download_json(
584 self._API_V2_BASE + 'tracks', playlist_id,
585 'Downloading tracks', query={
586 'ids': ','.join([compat_str(t['id']) for t in tracks]),
587 'playlistId': playlist_id,
588 'playlistSecretToken': token,
589 }, headers=self._HEADERS)
592 track_id = str_or_none(track.get('id'))
593 url = track.get('permalink_url')
597 url = self._API_V2_BASE + 'tracks/' + track_id
599 url += '?secret_token=' + token
600 entries.append(self.url_result(
601 url, SoundcloudIE.ie_key(), track_id))
602 return self.playlist_result(
603 entries, playlist_id,
604 playlist.get('title'),
605 playlist.get('description'))
608 class SoundcloudSetIE(SoundcloudPlaylistBaseIE):
609 _VALID_URL = r'https?://(?:(?:www|m)\.)?soundcloud\.com/(?P<uploader>[\w\d-]+)/sets/(?P<slug_title>[:\w\d-]+)(?:/(?P<token>[^?/]+))?'
610 IE_NAME = 'soundcloud:set'
612 'url': 'https://soundcloud.com/the-concept-band/sets/the-royal-concept-ep',
615 'title': 'The Royal Concept EP',
616 'description': 'md5:71d07087c7a449e8941a70a29e34671e',
618 'playlist_mincount': 5,
620 'url': 'https://soundcloud.com/the-concept-band/sets/the-royal-concept-ep/token',
621 'only_matching': True,
623 'url': 'https://soundcloud.com/discover/sets/weekly::flacmatic',
624 'only_matching': True,
626 'url': 'https://soundcloud.com/discover/sets/charts-top:all-music:de',
627 'only_matching': True,
629 'url': 'https://soundcloud.com/discover/sets/charts-top:hiphoprap:kr',
630 'only_matching': True,
633 def _real_extract(self, url):
634 mobj = self._match_valid_url(url)
636 full_title = ' %s /sets/ %s ' % mobj.group('uploader', 'slug_title')
637 token = mobj.group('token')
639 full_title += '/' + token
641 info = self._download_json(self._resolv_url(
642 self._BASE_URL + full_title), full_title, headers=self._HEADERS)
645 msgs = (compat_str(err['error_message']) for err in info['errors'])
646 raise ExtractorError('unable to download video webpage: %s ' % ','.join(msgs))
648 return self._extract_set(info, token)
651 class SoundcloudPagedPlaylistBaseIE(SoundcloudBaseIE):
652 def _extract_playlist(self, base_url, playlist_id, playlist_title):
656 'title': playlist_title,
657 'entries': self._entries(base_url, playlist_id),
660 def _entries(self, url, playlist_id):
661 # Per the SoundCloud documentation, the maximum limit for a linked partitioning query is 200.
662 # https://developers.soundcloud.com/blog/offset-pagination-deprecated
665 'linked_partitioning': '1',
669 retries = self.get_param('extractor_retries', 3)
671 for i in itertools.count():
672 attempt, last_error = -1, None
673 while attempt < retries:
676 self.report_warning(' %s . Retrying ...' % remove_end(last_error, '.'), playlist_id)
678 response = self._download_json(
679 url, playlist_id, query=query, headers=self._HEADERS,
680 note='Downloading track page %s%s ' % (i + 1, f' (retry # {attempt} )' if attempt else ''))
682 except ExtractorError as e:
683 # Downloading page may result in intermittent 502 HTTP error
684 # See https://github.com/yt-dlp/yt-dlp/issues/872
685 if attempt >= retries or not isinstance(e.cause, compat_HTTPError) or e.cause.code != 502:
687 last_error = str(e.cause or e.msg)
689 def resolve_entry(*candidates):
690 for cand in candidates:
691 if not isinstance(cand, dict):
693 permalink_url = url_or_none(cand.get('permalink_url'))
695 return self.url_result(
697 SoundcloudIE.ie_key() if SoundcloudIE.suitable(permalink_url) else None,
698 str_or_none(cand.get('id')), cand.get('title'))
700 for e in response['collection'] or []:
701 yield resolve_entry(e, e.get('track'), e.get('playlist'))
703 url = response.get('next_href')
706 query.pop('offset', None)
709 class SoundcloudUserIE(SoundcloudPagedPlaylistBaseIE):
710 _VALID_URL = r'''(?x)
712 (?:(?:www|m)\.)?soundcloud\.com/
715 (?P<rsrc>tracks|albums|sets|reposts|likes|spotlight)
719 IE_NAME = 'soundcloud:user'
721 'url': 'https://soundcloud.com/soft-cell-official',
724 'title': 'Soft Cell (All)',
726 'playlist_mincount': 28,
728 'url': 'https://soundcloud.com/soft-cell-official/tracks',
731 'title': 'Soft Cell (Tracks)',
733 'playlist_mincount': 27,
735 'url': 'https://soundcloud.com/soft-cell-official/albums',
738 'title': 'Soft Cell (Albums)',
740 'playlist_mincount': 1,
742 'url': 'https://soundcloud.com/jcv246/sets',
745 'title': 'Jordi / cv (Sets)',
747 'playlist_mincount': 2,
749 'url': 'https://soundcloud.com/jcv246/reposts',
752 'title': 'Jordi / cv (Reposts)',
754 'playlist_mincount': 6,
756 'url': 'https://soundcloud.com/clalberg/likes',
759 'title': 'clalberg (Likes)',
761 'playlist_mincount': 5,
763 'url': 'https://soundcloud.com/grynpyret/spotlight',
766 'title': 'Grynpyret (Spotlight)',
768 'playlist_mincount': 1,
772 'all': 'stream/users/ %s ',
773 'tracks': 'users/ %s /tracks',
774 'albums': 'users/ %s /albums',
775 'sets': 'users/ %s /playlists',
776 'reposts': 'stream/users/ %s /reposts',
777 'likes': 'users/ %s /likes',
778 'spotlight': 'users/ %s /spotlight',
781 def _real_extract(self, url):
782 mobj = self._match_valid_url(url)
783 uploader = mobj.group('user')
785 user = self._download_json(
786 self._resolv_url(self._BASE_URL + uploader),
787 uploader, 'Downloading user info', headers=self._HEADERS)
789 resource = mobj.group('rsrc') or 'all'
791 return self._extract_playlist(
792 self._API_V2_BASE + self._BASE_URL_MAP[resource] % user['id'],
793 str_or_none(user.get('id')),
794 ' %s ( %s )' % (user['username'], resource.capitalize()))
797 class SoundcloudTrackStationIE(SoundcloudPagedPlaylistBaseIE):
798 _VALID_URL = r'https?://(?:(?:www|m)\.)?soundcloud\.com/stations/track/[^/]+/(?P<id>[^/?#&]+)'
799 IE_NAME = 'soundcloud:trackstation'
801 'url': 'https://soundcloud.com/stations/track/officialsundial/your-text',
804 'title': 'Track station: your text',
806 'playlist_mincount': 47,
809 def _real_extract(self, url):
810 track_name = self._match_id(url)
812 track = self._download_json(self._resolv_url(url), track_name, headers=self._HEADERS)
813 track_id = self._search_regex(
814 r'soundcloud:track-stations:(\d+)', track['id'], 'track id')
816 return self._extract_playlist(
817 self._API_V2_BASE + 'stations/ %s /tracks' % track['id'],
818 track_id, 'Track station: %s ' % track['title'])
821 class SoundcloudRelatedIE(SoundcloudPagedPlaylistBaseIE):
822 _VALID_URL = r'https?://(?:(?:www|m)\.)?soundcloud\.com/(?P<slug>[\w\d-]+/[\w\d-]+)/(?P<relation>albums|sets|recommended)'
823 IE_NAME = 'soundcloud:related'
825 'url': 'https://soundcloud.com/wajang/sexapil-pingers-5/recommended',
828 'title': 'Sexapil - Pingers 5 (Recommended)',
830 'playlist_mincount': 50,
832 'url': 'https://soundcloud.com/wajang/sexapil-pingers-5/albums',
835 'title': 'Sexapil - Pingers 5 (Albums)',
837 'playlist_mincount': 1,
839 'url': 'https://soundcloud.com/wajang/sexapil-pingers-5/sets',
842 'title': 'Sexapil - Pingers 5 (Sets)',
844 'playlist_mincount': 4,
848 'albums': 'tracks/ %s /albums',
849 'sets': 'tracks/ %s /playlists_without_albums',
850 'recommended': 'tracks/ %s /related',
853 def _real_extract(self, url):
854 slug, relation = self._match_valid_url(url).group('slug', 'relation')
856 track = self._download_json(
857 self._resolv_url(self._BASE_URL + slug),
858 slug, 'Downloading track info', headers=self._HEADERS)
860 if track.get('errors'):
861 raise ExtractorError(f' {self.IE_NAME} said: %s ' % ','.join(
862 str(err['error_message']) for err in track['errors']), expected=True)
864 return self._extract_playlist(
865 self._API_V2_BASE + self._BASE_URL_MAP[relation] % track['id'], str(track['id']),
866 ' %s ( %s )' % (track.get('title') or slug, relation.capitalize()))
869 class SoundcloudPlaylistIE(SoundcloudPlaylistBaseIE):
870 _VALID_URL = r'https?://api(?:-v2)?\.soundcloud\.com/playlists/(?P<id>[0-9]+)(?:/? \? secret_token=(?P<token>[^&]+?))?$'
871 IE_NAME = 'soundcloud:playlist'
873 'url': 'https://api.soundcloud.com/playlists/4110309',
876 'title': 'TILT Brass - Bowery Poetry Club, August \' 03 [Non-Site SCR 02]',
877 'description': 're:.*?TILT Brass - Bowery Poetry Club',
882 def _real_extract(self, url):
883 mobj = self._match_valid_url(url)
884 playlist_id = mobj.group('id')
887 token = mobj.group('token')
889 query['secret_token'] = token
891 data = self._download_json(
892 self._API_V2_BASE + 'playlists/' + playlist_id,
893 playlist_id, 'Downloading playlist', query=query, headers=self._HEADERS)
895 return self._extract_set(data, token)
898 class SoundcloudSearchIE(SoundcloudBaseIE, SearchInfoExtractor):
899 IE_NAME = 'soundcloud:search'
900 IE_DESC = 'Soundcloud search'
901 _SEARCH_KEY = 'scsearch'
903 'url': 'scsearch15:post-avant jazzcore',
905 'title': 'post-avant jazzcore',
907 'playlist_count': 15,
910 _MAX_RESULTS_PER_PAGE = 200
911 _DEFAULT_RESULTS_PER_PAGE = 50
913 def _get_collection(self, endpoint, collection_id, **query):
915 query.get('limit', self._DEFAULT_RESULTS_PER_PAGE),
916 self._MAX_RESULTS_PER_PAGE)
919 'linked_partitioning': 1,
922 next_url = update_url_query(self._API_V2_BASE + endpoint, query)
924 for i in itertools.count(1):
925 response = self._download_json(
926 next_url, collection_id, f'Downloading page {i} ',
927 'Unable to download API page', headers=self._HEADERS)
929 for item in response.get('collection') or []:
931 yield self.url_result(item['uri'], SoundcloudIE.ie_key())
933 next_url = response.get('next_href')
937 def _get_n_results(self, query, n):
938 return self.playlist_result(itertools.islice(
939 self._get_collection('search/tracks', query, limit=n, q=query),
940 0, None if n == float('inf') else n), query, query)