2 from __future__
import unicode_literals
13 from ..compat
import (
37 class SoundcloudEmbedIE(InfoExtractor
):
38 _VALID_URL
= r
'https?://(?:w|player|p)\.soundcloud\.com/player/?.*?\burl=(?P<id>.+)'
40 # from https://www.soundi.fi/uutiset/ennakkokuuntelussa-timo-kaukolammen-station-to-station-to-station-julkaisua-juhlitaan-tanaan-g-livelabissa/
41 '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',
42 'only_matching': True,
46 def _extract_urls(webpage
):
47 return [m
.group('url') for m
in re
.finditer(
48 r
'<iframe[^>]+src=(["\'])(?P
<url
>(?
:https?
://)?
(?
:w\
.)?soundcloud\
.com
/player
.+?
)\
1',
51 def _real_extract(self, url):
52 query = compat_urlparse.parse_qs(
53 compat_urlparse.urlparse(url).query)
54 api_url = query['url
'][0]
55 secret_token = query.get('secret_token
')
57 api_url = update_url_query(api_url, {'secret_token': secret_token[0]})
58 return self.url_result(api_url)
61 class SoundcloudIE(InfoExtractor):
62 """Information extractor for soundcloud.com
63 To access the media, the uid of the song and a stream token
64 must be extracted from the page source and the script must make
65 a request to media.soundcloud.com/crossdomain.xml. Then
66 the media can be grabbed by requesting from an url composed
67 of the stream token and uid
70 _VALID_URL = r'''(?x)^(?:https?://)?
71 (?:(?:(?:www\.|m\.)?soundcloud\.com/
73 (?P<uploader>[\w\d-]+)/
74 (?!(?:tracks|albums|sets(?:/.+?)?|reposts|likes|spotlight)/?(?:$|[?#]))
76 (?P<token>[^?]+?)?(?:[?].*)?$)
77 |(?:api(?:-v2)?\.soundcloud\.com/tracks/(?P<track_id>\d+)
78 (?:/?\?secret_token=(?P<secret_token>[^&]+))?)
81 IE_NAME = 'soundcloud
'
84 'url
': 'http
://soundcloud
.com
/ethmusic
/lostin
-powers
-she
-so
-heavy
',
85 'md5
': 'ebef0a451b909710ed1d7787dddbf0d7
',
89 'title
': 'Lostin Powers
- She so
Heavy (SneakPreview
) Adrian Ackers Blueprint
1',
90 '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
',
91 'uploader
': 'E
.T
. ExTerrestrial Music
',
92 'uploader_id
': '1571244',
93 'timestamp
': 1349920598,
94 'upload_date
': '20121011',
96 'license
': 'all
-rights
-reserved
',
105 'url
': 'https
://soundcloud
.com
/the
-concept
-band
/goldrushed
-mastered?
in=the
-concept
-band
/sets
/the
-royal
-concept
-ep
',
109 'title
': 'Goldrushed
',
110 'description
': 'From Stockholm Sweden
\r\nPovel
/ Magnus
/ Filip
/ David
\r\nwww
.theroyalconcept
.com
',
111 'uploader
': 'The Royal Concept
',
112 'uploader_id
': '9615865',
113 'timestamp
': 1337635207,
114 'upload_date
': '20120521',
116 'license
': 'all
-rights
-reserved
',
119 'comment_count
': int,
125 'url
': 'https
://soundcloud
.com
/jaimemf
/youtube
-dlc
-test
-video
-a
-y
-baw
/s
-8Pjrp
',
126 'md5
': 'aa0dd32bfea9b0c5ef4f02aacd080604
',
130 'title
': 'Youtube
- Dl Test Video
\'\' Ä↭
',
131 'description
': 'test chars
: \"\'/\\ä↭
',
132 'uploader
': 'jaimeMF
',
133 'uploader_id
': '69767071',
134 'timestamp
': 1386604920,
135 'upload_date
': '20131209',
137 'license
': 'all
-rights
-reserved
',
140 'comment_count
': int,
144 # private link (alt format)
146 'url
': 'https
://api
.soundcloud
.com
/tracks
/123998367?secret_token
=s
-8Pjrp
',
147 'md5
': 'aa0dd32bfea9b0c5ef4f02aacd080604
',
151 'title
': 'Youtube
- Dl Test Video
\'\' Ä↭
',
152 'description
': 'test chars
: \"\'/\\ä↭
',
153 'uploader
': 'jaimeMF
',
154 'uploader_id
': '69767071',
155 'timestamp
': 1386604920,
156 'upload_date
': '20131209',
158 'license
': 'all
-rights
-reserved
',
161 'comment_count
': int,
167 'url
': 'https
://soundcloud
.com
/oddsamples
/bus
-brakes
',
168 'md5
': '7624f2351f8a3b2e7cd51522496e7631
',
172 'title
': 'Bus Brakes
',
173 'description
': 'md5
:0053ca6396e8d2fd7b7e1595ef12ab66
',
174 'uploader
': 'oddsamples
',
175 'uploader_id
': '73680509',
176 'timestamp
': 1389232924,
177 'upload_date
': '20140109',
179 'license
': 'cc
-by
-sa
',
182 'comment_count
': int,
186 # private link, downloadable format
188 'url
': 'https
://soundcloud
.com
/oriuplift
/uponly
-238-no
-talking
-wav
/s
-AyZUd
',
189 'md5
': '64a60b16e617d41d0bef032b7f55441e
',
193 'title
': 'Uplifting Only
238 [No Talking
] (incl
. Alex Feed Guestmix
) (Aug
31, 2017) [wav
]',
194 'description
': 'md5
:fa20ee0fca76a3d6df8c7e57f3715366
',
195 'uploader
': 'Ori Uplift Music
',
196 'uploader_id
': '12563093',
197 'timestamp
': 1504206263,
198 'upload_date
': '20170831',
199 'duration
': 7449.096,
200 'license
': 'all
-rights
-reserved
',
203 'comment_count
': int,
207 # no album art, use avatar pic for thumbnail
209 'url
': 'https
://soundcloud
.com
/garyvee
/sideways
-prod
-mad
-real
',
210 'md5
': '59c7872bc44e5d99b7211891664760c2
',
214 'title
': 'Sideways (Prod
. Mad Real
)',
215 'description
': 'md5
:d41d8cd98f00b204e9800998ecf8427e
',
216 'uploader
': 'garyvee
',
217 'uploader_id
': '2366352',
218 'timestamp
': 1488152409,
219 'upload_date
': '20170226',
221 'thumbnail
': r're
:https?
://.*\
.jpg
',
222 'license
': 'all
-rights
-reserved
',
225 'comment_count
': int,
229 'skip_download
': True,
233 'url
': 'https
://soundcloud
.com
/giovannisarani
/mezzo
-valzer
',
234 'md5
': 'e22aecd2bc88e0e4e432d7dcc0a1abf7
',
238 'title
': 'Mezzo Valzer
',
239 'description
': 'md5
:4138d582f81866a530317bae316e8b61
',
240 'uploader
': 'Micronie
',
241 'uploader_id
': '3352531',
242 'timestamp
': 1551394171,
243 'upload_date
': '20190228',
245 'thumbnail
': r're
:https?
://.*\
.jpg
',
246 'license
': 'all
-rights
-reserved
',
249 'comment_count
': int,
254 # AAC HQ format available (account with active subscription needed)
255 'url
': 'https
://soundcloud
.com
/wandw
/the
-chainsmokers
-ft
-daya
-dont
-let
-me
-down
-ww
-remix
-1',
256 'only_matching
': True,
259 # Go+ (account with active subscription needed)
260 'url
': 'https
://soundcloud
.com
/taylorswiftofficial
/look
-what
-you
-made
-me
-do
',
261 'only_matching
': True,
265 _API_V2_BASE = 'https
://api
-v2
.soundcloud
.com
/'
266 _BASE_URL = 'https
://soundcloud
.com
/'
267 _IMAGE_REPL_RE = r'-([0-9a
-z
]+)\
.jpg
'
282 def _store_client_id(self, client_id):
283 self._downloader.cache.store('soundcloud
', 'client_id
', client_id)
285 def _update_client_id(self):
286 webpage = self._download_webpage('https
://soundcloud
.com
/', None)
287 for src in reversed(re.findall(r'<script
[^
>]+src
="([^"]+)"', webpage)):
288 script = self._download_webpage(src, None, fatal=False)
290 client_id = self._search_regex(
291 r'client_id\s*:\s*"([0-9a
-zA
-Z
]{32}
)"',
292 script, 'client id', default=None)
294 self._CLIENT_ID = client_id
295 self._store_client_id(client_id)
297 raise ExtractorError('Unable to extract client id')
299 def _download_json(self, *args, **kwargs):
300 non_fatal = kwargs.get('fatal') is False
303 query = kwargs.get('query', {}).copy()
305 query['client_id'] = self._CLIENT_ID
306 kwargs['query'] = query
308 return super(SoundcloudIE, self)._download_json(*args, **compat_kwargs(kwargs))
309 except ExtractorError as e:
310 if isinstance(e.cause, compat_HTTPError) and e.cause.code == 401:
311 self._store_client_id(None)
312 self._update_client_id()
315 self._downloader.report_warning(error_to_compat_str(e))
319 def _real_initialize(self):
320 self._CLIENT_ID = self._downloader.cache.load('soundcloud', 'client_id') or "T5R4kgWS2PRf6lzLyIravUMnKlbIxQag
" # 'EXLwg5lHTO2dslU5EePe3xkw0m1h86Cd' # 'YUKXoArFcqrlQn9tfNHvvyfnDISj04zk'
323 _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"
324 _API_AUTH_QUERY_TEMPLATE = '?client_id=%s'
325 _API_AUTH_URL_PW = 'https://api-auth.soundcloud.com/web-auth/sign-in/password%s'
328 _NETRC_MACHINE = 'soundcloud'
331 username, password = self._get_login_info()
337 return ''.join([str(random.randrange(10)) for i in range(6)])
338 return '-'.join([genNumBlock() for i in range(4)])
341 'client_id': self._CLIENT_ID,
342 'recaptcha_pubkey': 'null',
343 'recaptcha_response': 'null',
345 'identifier': username,
348 'signature': self.sign(username, password, self._CLIENT_ID),
349 'device_id': genDevId(),
350 'user_agent': self._USER_AGENT
353 query = self._API_AUTH_QUERY_TEMPLATE % self._CLIENT_ID
354 login = sanitized_Request(self._API_AUTH_URL_PW % query, json.dumps(payload).encode('utf-8'))
355 response = self._download_json(login, None)
356 self._access_token = response.get('session').get('access_token')
357 if not self._access_token:
358 self.report_warning('Unable to get access token, login may has failed')
360 self._HEADERS = {'Authorization': 'OAuth ' + self._access_token}
362 # signature generation
363 def sign(self, user, pw, clid):
373 n = "0763ed7314c69015fd4a0dc16bbf4b90
" # _KEY
375 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
377 t = clid # _CLIENT_ID
379 d = '-'.join([str(mInt) for mInt in [a, i, s, w, u, l, b, k]])
380 p = n + y + d + r + e + t + d + n
386 for f in range(f, len(h)):
387 m = (m >> 1) + ((1 & m) << 23)
391 # c is not even needed
392 out = str(y) + ':' + str(d) + ':' + format(m, 'x') + ':' + str(c)
397 def _resolv_url(cls, url):
398 return SoundcloudIE._API_V2_BASE + 'resolve?url=' + url
400 def _extract_info_dict(self, info, full_title=None, secret_token=None):
401 track_id = compat_str(info['id'])
402 title = info['title']
406 query = {'client_id': self._CLIENT_ID}
408 query['secret_token'] = secret_token
410 if info.get('downloadable') and info.get('has_downloads_left'):
411 download_url = update_url_query(
412 self._API_V2_BASE + 'tracks/' + track_id + '/download', query)
413 redirect_url = (self._download_json(download_url, track_id, fatal=False) or {}).get('redirectUri')
415 urlh = self._request_webpage(
416 HEADRequest(redirect_url), track_id, fatal=False)
418 format_url = urlh.geturl()
419 format_urls.add(format_url)
421 'format_id': 'download',
422 'ext': urlhandle_detect_ext(urlh) or 'mp3',
423 'filesize': int_or_none(urlh.headers.get('Content-Length')),
428 def invalid_url(url):
429 return not url or url in format_urls
431 def add_format(f, protocol, is_preview=False):
432 mobj = re.search(r'\.(?P<abr>\d+)\.(?P<ext>[0-9a-z]{3,4})(?=[/?])', stream_url)
434 for k, v in mobj.groupdict().items():
439 format_id_list.append(protocol)
443 for k in ('ext', 'abr'):
446 format_id_list.append(v)
447 preview = is_preview or re.search(r'/(?:preview|playlist)/0/30/', f['url'])
449 format_id_list.append('preview')
453 if protocol == 'hls':
454 protocol = 'm3u8' if ext == 'aac' else 'm3u8_native'
458 'format_id': '_'.join(format_id_list),
459 'protocol': protocol,
460 'preference': -10 if preview else None,
465 transcodings = try_get(
466 info, lambda x: x['media']['transcodings'], list) or []
467 for t in transcodings:
468 if not isinstance(t, dict):
470 format_url = url_or_none(t.get('url'))
473 stream = self._download_json(
474 format_url, track_id, query=query, fatal=False, headers=self._HEADERS)
475 if not isinstance(stream, dict):
477 stream_url = url_or_none(stream.get('url'))
478 if invalid_url(stream_url):
480 format_urls.add(stream_url)
481 stream_format = t.get('format') or {}
482 protocol = stream_format.get('protocol')
483 if protocol != 'hls' and '/hls' in format_url:
486 preset = str_or_none(t.get('preset'))
488 ext = preset.split('_')[0]
489 if ext not in KNOWN_EXTENSIONS:
490 ext = mimetype2ext(stream_format.get('mime_type'))
494 }, 'http' if protocol == 'progressive' else protocol,
495 t.get('snipped') or '/preview/' in format_url)
500 if not formats and info.get('policy') == 'BLOCK':
501 self.raise_geo_restricted()
502 self._sort_formats(formats)
504 user = info.get('user') or {}
507 artwork_url = info.get('artwork_url')
508 thumbnail = artwork_url or user.get('avatar_url')
509 if isinstance(thumbnail, compat_str):
510 if re.search(self._IMAGE_REPL_RE, thumbnail):
511 for image_id, size in self._ARTWORK_MAP.items():
514 'url': re.sub(self._IMAGE_REPL_RE, '-%s.jpg' % image_id, thumbnail),
516 if image_id == 'tiny' and not artwork_url:
518 elif image_id == 'original':
527 thumbnails = [{'url': thumbnail}]
529 def extract_count(key):
530 return int_or_none(info.get('%s_count' % key))
534 'uploader': user.get('username'),
535 'uploader_id': str_or_none(user.get('id')) or user.get('permalink'),
536 'uploader_url': user.get('permalink_url'),
537 'timestamp': unified_timestamp(info.get('created_at')),
539 'description': info.get('description'),
540 'thumbnails': thumbnails,
541 'duration': float_or_none(info.get('duration'), 1000),
542 'webpage_url': info.get('permalink_url'),
543 'license': info.get('license'),
544 'view_count': extract_count('playback'),
545 'like_count': extract_count('favoritings') or extract_count('likes'),
546 'comment_count': extract_count('comment'),
547 'repost_count': extract_count('reposts'),
548 'genre': info.get('genre'),
552 def _real_extract(self, url):
553 mobj = re.match(self._VALID_URL, url)
555 track_id = mobj.group('track_id')
559 info_json_url = self._API_V2_BASE + 'tracks/' + track_id
560 full_title = track_id
561 token = mobj.group('secret_token')
563 query['secret_token'] = token
565 full_title = resolve_title = '%s/%s' % mobj.group('uploader', 'title')
566 token = mobj.group('token')
568 resolve_title += '/%s' % token
569 info_json_url = self._resolv_url(self._BASE_URL + resolve_title)
571 info = self._download_json(
572 info_json_url, full_title, 'Downloading info JSON', query=query, headers=self._HEADERS)
574 return self._extract_info_dict(info, full_title, token)
577 class SoundcloudPlaylistBaseIE(SoundcloudIE):
578 def _extract_set(self, playlist, token=None):
579 playlist_id = compat_str(playlist['id'])
580 tracks = playlist.get('tracks') or []
581 if not all([t.get('permalink_url') for t in tracks]) and token:
582 tracks = self._download_json(
583 self._API_V2_BASE + 'tracks', playlist_id,
584 'Downloading tracks', query={
585 'ids': ','.join([compat_str(t['id']) for t in tracks]),
586 'playlistId': playlist_id,
587 'playlistSecretToken': token,
588 }, headers=self._HEADERS)
591 track_id = str_or_none(track.get('id'))
592 url = track.get('permalink_url')
596 url = self._API_V2_BASE + 'tracks/' + track_id
598 url += '?secret_token=' + token
599 entries.append(self.url_result(
600 url, SoundcloudIE.ie_key(), track_id))
601 return self.playlist_result(
602 entries, playlist_id,
603 playlist.get('title'),
604 playlist.get('description'))
607 class SoundcloudSetIE(SoundcloudPlaylistBaseIE):
608 _VALID_URL = r'https?://(?:(?:www|m)\.)?soundcloud\.com/(?P<uploader>[\w\d-]+)/sets/(?P<slug_title>[:\w\d-]+)(?:/(?P<token>[^?/]+))?'
609 IE_NAME = 'soundcloud:set'
611 'url': 'https://soundcloud.com/the-concept-band/sets/the-royal-concept-ep',
614 'title': 'The Royal Concept EP',
615 'description': 'md5:71d07087c7a449e8941a70a29e34671e',
617 'playlist_mincount': 5,
619 'url': 'https://soundcloud.com/the-concept-band/sets/the-royal-concept-ep/token',
620 'only_matching': True,
622 'url': 'https://soundcloud.com/discover/sets/weekly::flacmatic',
623 'only_matching': True,
625 'url': 'https://soundcloud.com/discover/sets/charts-top:all-music:de',
626 'only_matching': True,
628 'url': 'https://soundcloud.com/discover/sets/charts-top:hiphoprap:kr',
629 'only_matching': True,
632 def _real_extract(self, url):
633 mobj = re.match(self._VALID_URL, url)
635 full_title = '%s/sets/%s' % mobj.group('uploader', 'slug_title')
636 token = mobj.group('token')
638 full_title += '/' + token
640 info = self._download_json(self._resolv_url(
641 self._BASE_URL + full_title), full_title, headers=self._HEADERS)
644 msgs = (compat_str(err['error_message']) for err in info['errors'])
645 raise ExtractorError('unable to download video webpage: %s' % ','.join(msgs))
647 return self._extract_set(info, token)
650 class SoundcloudPagedPlaylistBaseIE(SoundcloudIE):
651 def _extract_playlist(self, base_url, playlist_id, playlist_title):
654 'linked_partitioning': '1',
657 query = COMMON_QUERY.copy()
663 for i in itertools.count():
664 response = self._download_json(
665 next_href, playlist_id,
666 'Downloading track page %s' % (i + 1), query=query, headers=self._HEADERS)
668 collection = response['collection']
670 if not isinstance(collection, list):
673 # Empty collection may be returned, in this case we proceed
674 # straight to next_href
676 def resolve_entry(candidates):
677 for cand in candidates:
678 if not isinstance(cand, dict):
680 permalink_url = url_or_none(cand.get('permalink_url'))
681 if not 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'))
689 entry = resolve_entry((e, e.get('track'), e.get('playlist')))
691 entries.append(entry)
693 next_href = response.get('next_href')
697 next_href = response['next_href']
698 parsed_next_href = compat_urlparse.urlparse(next_href)
699 query = compat_urlparse.parse_qs(parsed_next_href.query)
700 query.update(COMMON_QUERY)
705 'title': playlist_title,
710 class SoundcloudUserIE(SoundcloudPagedPlaylistBaseIE):
711 _VALID_URL = r'''(?x)
713 (?:(?:www|m)\.)?soundcloud\.com/
716 (?P<rsrc>tracks|albums|sets|reposts|likes|spotlight)
720 IE_NAME = 'soundcloud:user'
722 'url': 'https://soundcloud.com/soft-cell-official',
725 'title': 'Soft Cell (All)',
727 'playlist_mincount': 28,
729 'url': 'https://soundcloud.com/soft-cell-official/tracks',
732 'title': 'Soft Cell (Tracks)',
734 'playlist_mincount': 27,
736 'url': 'https://soundcloud.com/soft-cell-official/albums',
739 'title': 'Soft Cell (Albums)',
741 'playlist_mincount': 1,
743 'url': 'https://soundcloud.com/jcv246/sets',
746 'title': 'Jordi / cv (Sets)',
748 'playlist_mincount': 2,
750 'url': 'https://soundcloud.com/jcv246/reposts',
753 'title': 'Jordi / cv (Reposts)',
755 'playlist_mincount': 6,
757 'url': 'https://soundcloud.com/clalberg/likes',
760 'title': 'clalberg (Likes)',
762 'playlist_mincount': 5,
764 'url': 'https://soundcloud.com/grynpyret/spotlight',
767 'title': 'Grynpyret (Spotlight)',
769 'playlist_mincount': 1,
773 'all': 'stream/users/%s',
774 'tracks': 'users/%s/tracks',
775 'albums': 'users/%s/albums',
776 'sets': 'users/%s/playlists',
777 'reposts': 'stream/users/%s/reposts',
778 'likes': 'users/%s/likes',
779 'spotlight': 'users/%s/spotlight',
782 def _real_extract(self, url):
783 mobj = re.match(self._VALID_URL, url)
784 uploader = mobj.group('user')
786 user = self._download_json(
787 self._resolv_url(self._BASE_URL + uploader),
788 uploader, 'Downloading user info', headers=self._HEADERS)
790 resource = mobj.group('rsrc') or 'all'
792 return self._extract_playlist(
793 self._API_V2_BASE + self._BASE_URL_MAP[resource] % user['id'],
794 str_or_none(user.get('id')),
795 '%s (%s)' % (user['username'], resource.capitalize()))
798 class SoundcloudTrackStationIE(SoundcloudPagedPlaylistBaseIE):
799 _VALID_URL = r'https?://(?:(?:www|m)\.)?soundcloud\.com/stations/track/[^/]+/(?P<id>[^/?#&]+)'
800 IE_NAME = 'soundcloud:trackstation'
802 'url': 'https://soundcloud.com/stations/track/officialsundial/your-text',
805 'title': 'Track station: your text',
807 'playlist_mincount': 47,
810 def _real_extract(self, url):
811 track_name = self._match_id(url)
813 track = self._download_json(self._resolv_url(url), track_name, headers=self._HEADERS)
814 track_id = self._search_regex(
815 r'soundcloud:track-stations:(\d+)', track['id'], 'track id')
817 return self._extract_playlist(
818 self._API_V2_BASE + 'stations/%s/tracks' % track['id'],
819 track_id, 'Track station: %s' % track['title'])
822 class SoundcloudPlaylistIE(SoundcloudPlaylistBaseIE):
823 _VALID_URL = r'https?://api(?:-v2)?\.soundcloud\.com/playlists/(?P<id>[0-9]+)(?:/?\?secret_token=(?P<token>[^&]+?))?$'
824 IE_NAME = 'soundcloud:playlist'
826 'url': 'https://api.soundcloud.com/playlists/4110309',
829 'title': 'TILT Brass - Bowery Poetry Club, August \'03 [Non-Site SCR 02]',
830 'description': 're:.*?TILT Brass - Bowery Poetry Club',
835 def _real_extract(self, url):
836 mobj = re.match(self._VALID_URL, url)
837 playlist_id = mobj.group('id')
840 token = mobj.group('token')
842 query['secret_token'] = token
844 data = self._download_json(
845 self._API_V2_BASE + 'playlists/' + playlist_id,
846 playlist_id, 'Downloading playlist', query=query, headers=self._HEADERS)
848 return self._extract_set(data, token)
851 class SoundcloudSearchIE(SearchInfoExtractor, SoundcloudIE):
852 IE_NAME = 'soundcloud:search'
853 IE_DESC = 'Soundcloud search'
854 _MAX_RESULTS = float('inf')
856 'url': 'scsearch15:post-avant jazzcore',
858 'title': 'post-avant jazzcore',
860 'playlist_count': 15,
863 _SEARCH_KEY = 'scsearch'
864 _MAX_RESULTS_PER_PAGE = 200
865 _DEFAULT_RESULTS_PER_PAGE = 50
867 def _get_collection(self, endpoint, collection_id, **query):
869 query.get('limit', self._DEFAULT_RESULTS_PER_PAGE),
870 self._MAX_RESULTS_PER_PAGE)
873 'linked_partitioning': 1,
876 next_url = update_url_query(self._API_V2_BASE + endpoint, query)
878 collected_results = 0
880 for i in itertools.count(1):
881 response = self._download_json(
882 next_url, collection_id, 'Downloading page {0}'.format(i),
883 'Unable to download API page', headers=self._HEADERS)
885 collection = response.get('collection', [])
889 collection = list(filter(bool, collection))
890 collected_results += len(collection)
892 for item in collection:
893 yield self.url_result(item['uri'], SoundcloudIE.ie_key())
895 if not collection or collected_results >= limit:
898 next_url = response.get('next_href')
902 def _get_n_results(self, query, n):
903 tracks = self._get_collection('search/tracks', query, limit=n, q=query)
904 return self.playlist_result(tracks, playlist_title=query)