2 from __future__ import unicode_literals
11 from ..compat import (
14 compat_urllib_parse_urlencode,
23 class SoundcloudIE(InfoExtractor):
24 """Information extractor for soundcloud.com
25 To access the media, the uid of the song and a stream token
26 must be extracted from the page source and the script must make
27 a request to media.soundcloud.com/crossdomain.xml. Then
28 the media can be grabbed by requesting from an url composed
29 of the stream token and uid
32 _VALID_URL = r'''(?x)^(?:https?://)?
33 (?:(?:(?:www\.|m\.)?soundcloud\.com/
34 (?P<uploader>[\w\d-]+)/
35 (?!(?:tracks|sets(?:/[^/?#]+)?|reposts|likes|spotlight)/?(?:$|[?#]))
37 (?P<token>[^?]+?)?(?:[?].*)?$)
38 |(?:api\.soundcloud\.com/tracks/(?P<track_id>\d+)
39 (?:/?\?secret_token=(?P<secret_token>[^&]+))?)
40 |(?P<player>(?:w|player|p.)\.soundcloud\.com/player/?.*?url=.*)
43 IE_NAME = 'soundcloud'
46 'url': 'http://soundcloud.com/ethmusic/lostin-powers-she-so-heavy',
47 'md5': 'ebef0a451b909710ed1d7787dddbf0d7',
51 'upload_date': '20121011',
52 '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',
53 'uploader': 'E.T. ExTerrestrial Music',
54 'title': 'Lostin Powers - She so Heavy (SneakPreview) Adrian Ackers Blueprint 1',
60 'url': 'https://soundcloud.com/the-concept-band/goldrushed-mastered?in=the-concept-band/sets/the-royal-concept-ep',
64 'title': 'Goldrushed',
65 'description': 'From Stockholm Sweden\r\nPovel / Magnus / Filip / David\r\nwww.theroyalconcept.com',
66 'uploader': 'The Royal Concept',
67 'upload_date': '20120521',
72 'skip_download': True,
77 'url': 'https://soundcloud.com/jaimemf/youtube-dl-test-video-a-y-baw/s-8Pjrp',
78 'md5': 'aa0dd32bfea9b0c5ef4f02aacd080604',
82 'title': 'Youtube - Dl Test Video \'\' Ä↭',
83 'uploader': 'jaimeMF',
84 'description': 'test chars: \"\'/\\ä↭',
85 'upload_date': '20131209',
89 # private link (alt format)
91 'url': 'https://api.soundcloud.com/tracks/123998367?secret_token=s-8Pjrp',
92 'md5': 'aa0dd32bfea9b0c5ef4f02aacd080604',
96 'title': 'Youtube - Dl Test Video \'\' Ä↭',
97 'uploader': 'jaimeMF',
98 'description': 'test chars: \"\'/\\ä↭',
99 'upload_date': '20131209',
105 'url': 'https://soundcloud.com/oddsamples/bus-brakes',
106 'md5': '7624f2351f8a3b2e7cd51522496e7631',
110 'title': 'Bus Brakes',
111 'description': 'md5:0053ca6396e8d2fd7b7e1595ef12ab66',
112 'uploader': 'oddsamples',
113 'upload_date': '20140109',
119 _CLIENT_ID = '02gUJC0hH2ct1EGOcYXQIzRFU91c72Ea'
120 _IPHONE_CLIENT_ID = '376f225bf427445fc4bfb6b99b72e0bf'
122 def report_resolve(self, video_id):
123 """Report information extraction."""
124 self.to_screen('%s: Resolving id' % video_id)
127 def _resolv_url(cls, url):
128 return 'http://api.soundcloud.com/resolve.json?url=' + url + '&client_id=' + cls._CLIENT_ID
130 def _extract_info_dict(self, info, full_title=None, quiet=False, secret_token=None):
131 track_id = compat_str(info['id'])
132 name = full_title or track_id
134 self.report_extraction(name)
136 thumbnail = info['artwork_url']
137 if thumbnail is not None:
138 thumbnail = thumbnail.replace('-large', '-t500x500')
142 'uploader': info['user']['username'],
143 'upload_date': unified_strdate(info['created_at']),
144 'title': info['title'],
145 'description': info['description'],
146 'thumbnail': thumbnail,
147 'duration': int_or_none(info.get('duration'), 1000),
148 'webpage_url': info.get('permalink_url'),
151 if info.get('downloadable', False):
152 # We can build a direct link to the song
154 'https://api.soundcloud.com/tracks/{0}/download?client_id={1}'.format(
155 track_id, self._CLIENT_ID))
157 'format_id': 'download',
158 'ext': info.get('original_format', 'mp3'),
164 # We have to retrieve the url
165 streams_url = ('http://api.soundcloud.com/i1/tracks/{0}/streams?'
166 'client_id={1}&secret_token={2}'.format(track_id, self._IPHONE_CLIENT_ID, secret_token))
167 format_dict = self._download_json(
169 track_id, 'Downloading track url')
171 for key, stream_url in format_dict.items():
172 if key.startswith('http'):
179 elif key.startswith('rtmp'):
180 # The url doesn't have an rtmp app, we have to extract the playpath
181 url, path = stream_url.split('mp3:', 1)
185 'play_path': 'mp3:' + path,
191 # We fallback to the stream_url in the original info, this
192 # cannot be always used, sometimes it can give an HTTP 404 error
194 'format_id': 'fallback',
195 'url': info['stream_url'] + '?client_id=' + self._CLIENT_ID,
201 if f['format_id'].startswith('http'):
202 f['protocol'] = 'http'
203 if f['format_id'].startswith('rtmp'):
204 f['protocol'] = 'rtmp'
206 self._check_formats(formats, track_id)
207 self._sort_formats(formats)
208 result['formats'] = formats
212 def _real_extract(self, url):
213 mobj = re.match(self._VALID_URL, url, flags=re.VERBOSE)
215 raise ExtractorError('Invalid URL: %s' % url)
217 track_id = mobj.group('track_id')
219 if track_id is not None:
220 info_json_url = 'http://api.soundcloud.com/tracks/' + track_id + '.json?client_id=' + self._CLIENT_ID
221 full_title = track_id
222 token = mobj.group('secret_token')
224 info_json_url += '&secret_token=' + token
225 elif mobj.group('player'):
226 query = compat_urlparse.parse_qs(compat_urlparse.urlparse(url).query)
227 real_url = query['url'][0]
228 # If the token is in the query of the original url we have to
230 if 'secret_token' in query:
231 real_url += '?secret_token=' + query['secret_token'][0]
232 return self.url_result(real_url)
234 # extract uploader (which is in the url)
235 uploader = mobj.group('uploader')
236 # extract simple title (uploader + slug of song title)
237 slug_title = mobj.group('title')
238 token = mobj.group('token')
239 full_title = resolve_title = '%s/%s' % (uploader, slug_title)
241 resolve_title += '/%s' % token
243 self.report_resolve(full_title)
245 url = 'http://soundcloud.com/%s' % resolve_title
246 info_json_url = self._resolv_url(url)
247 info = self._download_json(info_json_url, full_title, 'Downloading info JSON')
249 return self._extract_info_dict(info, full_title, secret_token=token)
252 class SoundcloudSetIE(SoundcloudIE):
253 _VALID_URL = r'https?://(?:(?:www|m)\.)?soundcloud\.com/(?P<uploader>[\w\d-]+)/sets/(?P<slug_title>[\w\d-]+)(?:/(?P<token>[^?/]+))?'
254 IE_NAME = 'soundcloud:set'
256 'url': 'https://soundcloud.com/the-concept-band/sets/the-royal-concept-ep',
259 'title': 'The Royal Concept EP',
261 'playlist_mincount': 6,
264 def _real_extract(self, url):
265 mobj = re.match(self._VALID_URL, url)
267 # extract uploader (which is in the url)
268 uploader = mobj.group('uploader')
269 # extract simple title (uploader + slug of song title)
270 slug_title = mobj.group('slug_title')
271 full_title = '%s/sets/%s' % (uploader, slug_title)
272 url = 'http://soundcloud.com/%s/sets/%s' % (uploader, slug_title)
274 token = mobj.group('token')
276 full_title += '/' + token
279 self.report_resolve(full_title)
281 resolv_url = self._resolv_url(url)
282 info = self._download_json(resolv_url, full_title)
285 msgs = (compat_str(err['error_message']) for err in info['errors'])
286 raise ExtractorError('unable to download video webpage: %s' % ','.join(msgs))
288 entries = [self.url_result(track['permalink_url'], 'Soundcloud') for track in info['tracks']]
293 'id': '%s' % info['id'],
294 'title': info['title'],
298 class SoundcloudUserIE(SoundcloudIE):
299 _VALID_URL = r'''(?x)
301 (?:(?:www|m)\.)?soundcloud\.com/
304 (?P<rsrc>tracks|sets|reposts|likes|spotlight)
308 IE_NAME = 'soundcloud:user'
310 'url': 'https://soundcloud.com/the-akashic-chronicler',
313 'title': 'The Akashic Chronicler (All)',
315 'playlist_mincount': 111,
317 'url': 'https://soundcloud.com/the-akashic-chronicler/tracks',
320 'title': 'The Akashic Chronicler (Tracks)',
322 'playlist_mincount': 50,
324 'url': 'https://soundcloud.com/the-akashic-chronicler/sets',
327 'title': 'The Akashic Chronicler (Playlists)',
329 'playlist_mincount': 3,
331 'url': 'https://soundcloud.com/the-akashic-chronicler/reposts',
334 'title': 'The Akashic Chronicler (Reposts)',
336 'playlist_mincount': 7,
338 'url': 'https://soundcloud.com/the-akashic-chronicler/likes',
341 'title': 'The Akashic Chronicler (Likes)',
343 'playlist_mincount': 321,
345 'url': 'https://soundcloud.com/grynpyret/spotlight',
348 'title': 'Grynpyret (Spotlight)',
350 'playlist_mincount': 1,
353 _API_BASE = 'https://api.soundcloud.com'
354 _API_V2_BASE = 'https://api-v2.soundcloud.com'
357 'all': '%s/profile/soundcloud:users:%%s' % _API_V2_BASE,
358 'tracks': '%s/users/%%s/tracks' % _API_BASE,
359 'sets': '%s/users/%%s/playlists' % _API_V2_BASE,
360 'reposts': '%s/profile/soundcloud:users:%%s/reposts' % _API_V2_BASE,
361 'likes': '%s/users/%%s/likes' % _API_V2_BASE,
362 'spotlight': '%s/users/%%s/spotlight' % _API_V2_BASE,
369 'reposts': 'Reposts',
371 'spotlight': 'Spotlight',
374 def _real_extract(self, url):
375 mobj = re.match(self._VALID_URL, url)
376 uploader = mobj.group('user')
378 url = 'http://soundcloud.com/%s/' % uploader
379 resolv_url = self._resolv_url(url)
380 user = self._download_json(
381 resolv_url, uploader, 'Downloading user info')
383 resource = mobj.group('rsrc') or 'all'
384 base_url = self._BASE_URL_MAP[resource] % user['id']
388 'client_id': self._CLIENT_ID,
389 'linked_partitioning': '1',
392 query = COMMON_QUERY.copy()
395 next_href = base_url + '?' + compat_urllib_parse_urlencode(query)
398 for i in itertools.count():
399 response = self._download_json(
400 next_href, uploader, 'Downloading track page %s' % (i + 1))
402 collection = response['collection']
406 def resolve_permalink_url(candidates):
407 for cand in candidates:
408 if isinstance(cand, dict):
409 permalink_url = cand.get('permalink_url')
410 if permalink_url and permalink_url.startswith('http'):
414 permalink_url = resolve_permalink_url((e, e.get('track'), e.get('playlist')))
416 entries.append(self.url_result(permalink_url))
418 next_href = response.get('next_href')
422 parsed_next_href = compat_urlparse.urlparse(response['next_href'])
423 qs = compat_urlparse.parse_qs(parsed_next_href.query)
424 qs.update(COMMON_QUERY)
425 next_href = compat_urlparse.urlunparse(
426 parsed_next_href._replace(query=compat_urllib_parse_urlencode(qs, True)))
430 'id': compat_str(user['id']),
431 'title': '%s (%s)' % (user['username'], self._TITLE_MAP[resource]),
436 class SoundcloudPlaylistIE(SoundcloudIE):
437 _VALID_URL = r'https?://api\.soundcloud\.com/playlists/(?P<id>[0-9]+)(?:/?\?secret_token=(?P<token>[^&]+?))?$'
438 IE_NAME = 'soundcloud:playlist'
440 'url': 'http://api.soundcloud.com/playlists/4110309',
443 'title': 'TILT Brass - Bowery Poetry Club, August \'03 [Non-Site SCR 02]',
444 'description': 're:.*?TILT Brass - Bowery Poetry Club',
449 def _real_extract(self, url):
450 mobj = re.match(self._VALID_URL, url)
451 playlist_id = mobj.group('id')
452 base_url = '%s//api.soundcloud.com/playlists/%s.json?' % (self.http_scheme(), playlist_id)
455 'client_id': self._CLIENT_ID,
457 token = mobj.group('token')
460 data_dict['secret_token'] = token
462 data = compat_urllib_parse_urlencode(data_dict)
463 data = self._download_json(
464 base_url + data, playlist_id, 'Downloading playlist')
466 entries = [self.url_result(track['permalink_url'], 'Soundcloud') for track in data['tracks']]
471 'title': data.get('title'),
472 'description': data.get('description'),
477 class SoundcloudSearchIE(SearchInfoExtractor, SoundcloudIE):
478 IE_NAME = 'soundcloud:search'
479 IE_DESC = 'Soundcloud search'
480 _MAX_RESULTS = float('inf')
482 'url': 'scsearch15:post-avant jazzcore',
484 'title': 'post-avant jazzcore',
486 'playlist_count': 15,
489 _SEARCH_KEY = 'scsearch'
490 _MAX_RESULTS_PER_PAGE = 200
491 _DEFAULT_RESULTS_PER_PAGE = 50
492 _API_V2_BASE = 'https://api-v2.soundcloud.com'
494 def _get_collection(self, endpoint, collection_id, **query):
496 query.get('limit', self._DEFAULT_RESULTS_PER_PAGE),
497 self._MAX_RESULTS_PER_PAGE)
498 query['limit'] = limit
499 query['client_id'] = self._CLIENT_ID
500 query['linked_partitioning'] = '1'
502 data = compat_urllib_parse_urlencode(query)
503 next_url = '{0}{1}?{2}'.format(self._API_V2_BASE, endpoint, data)
505 collected_results = 0
507 for i in itertools.count(1):
508 response = self._download_json(
509 next_url, collection_id, 'Downloading page {0}'.format(i),
510 'Unable to download API page')
512 collection = response.get('collection', [])
516 collection = list(filter(bool, collection))
517 collected_results += len(collection)
519 for item in collection:
520 yield self.url_result(item['uri'], SoundcloudIE.ie_key())
522 if not collection or collected_results >= limit:
525 next_url = response.get('next_href')
529 def _get_n_results(self, query, n):
530 tracks = self._get_collection('/search/tracks', query, limit=n, q=query)
531 return self.playlist_result(tracks, playlist_title=query)