# encoding: utf-8
from __future__ import unicode_literals
-import json
import re
import itertools
_VALID_URL = r'''(?x)^(?:https?://)?
(?:(?:(?:www\.|m\.)?soundcloud\.com/
(?P<uploader>[\w\d-]+)/
- (?!sets/)(?P<title>[\w\d-]+)/?
+ (?!sets/|likes/?(?:$|[?#]))
+ (?P<title>[\w\d-]+)/?
(?P<token>[^?]+?)?(?:[?].*)?$)
|(?:api\.soundcloud\.com/tracks/(?P<track_id>\d+))
|(?P<player>(?:w|player|p.)\.soundcloud\.com/player/?.*?url=.*)
},
# downloadable song
{
- 'url': 'https://soundcloud.com/simgretina/just-your-problem-baby-1',
- 'md5': '56a8b69568acaa967b4c49f9d1d52d19',
+ 'url': 'https://soundcloud.com/oddsamples/bus-brakes',
+ 'md5': '7624f2351f8a3b2e7cd51522496e7631',
'info_dict': {
- 'id': '105614606',
- 'ext': 'wav',
- 'title': 'Just Your Problem Baby (Acapella)',
- 'description': 'Vocals',
- 'uploader': 'Sim Gretina',
- 'upload_date': '20130815',
- #'duration': 42,
+ 'id': '128590877',
+ 'ext': 'mp3',
+ 'title': 'Bus Brakes',
+ 'description': 'md5:0170be75dd395c96025d210d261c784e',
+ 'uploader': 'oddsamples',
+ 'upload_date': '20140109',
+ 'duration': 17,
},
},
]
class SoundcloudSetIE(SoundcloudIE):
_VALID_URL = r'https?://(?:www\.)?soundcloud\.com/([\w\d-]+)/sets/([\w\d-]+)'
IE_NAME = 'soundcloud:set'
- # it's in tests/test_playlists.py
- _TESTS = []
+ _TESTS = [{
+ 'url': 'https://soundcloud.com/the-concept-band/sets/the-royal-concept-ep',
+ 'info_dict': {
+ 'title': 'The Royal Concept EP',
+ },
+ 'playlist_mincount': 6,
+ }]
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
- if mobj is None:
- raise ExtractorError('Invalid URL: %s' % url)
# extract uploader (which is in the url)
uploader = mobj.group(1)
self._downloader.report_error('unable to download video webpage: %s' % compat_str(err['error_message']))
return
- self.report_extraction(full_title)
- return {'_type': 'playlist',
- 'entries': [self._extract_info_dict(track) for track in info['tracks']],
- 'id': info['id'],
- 'title': info['title'],
- }
+ return {
+ '_type': 'playlist',
+ 'entries': [self._extract_info_dict(track) for track in info['tracks']],
+ 'id': info['id'],
+ 'title': info['title'],
+ }
class SoundcloudUserIE(SoundcloudIE):
- _VALID_URL = r'https?://(www\.)?soundcloud\.com/(?P<user>[^/]+)(/?(tracks/)?)?(\?.*)?$'
+ _VALID_URL = r'https?://(www\.)?soundcloud\.com/(?P<user>[^/]+)/?((?P<rsrc>tracks|likes)/?)?(\?.*)?$'
IE_NAME = 'soundcloud:user'
-
- # it's in tests/test_playlists.py
- _TESTS = []
+ _TESTS = [{
+ 'url': 'https://soundcloud.com/the-concept-band',
+ 'info_dict': {
+ 'id': '9615865',
+ 'title': 'The Royal Concept',
+ },
+ 'playlist_mincount': 12
+ }, {
+ 'url': 'https://soundcloud.com/the-concept-band/likes',
+ 'info_dict': {
+ 'id': '9615865',
+ 'title': 'The Royal Concept',
+ },
+ 'playlist_mincount': 1,
+ }]
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
uploader = mobj.group('user')
+ resource = mobj.group('rsrc')
+ if resource is None:
+ resource = 'tracks'
+ elif resource == 'likes':
+ resource = 'favorites'
url = 'http://soundcloud.com/%s/' % uploader
resolv_url = self._resolv_url(url)
user = self._download_json(
resolv_url, uploader, 'Downloading user info')
- base_url = 'http://api.soundcloud.com/users/%s/tracks.json?' % uploader
+ base_url = 'http://api.soundcloud.com/users/%s/%s.json?' % (uploader, resource)
entries = []
for i in itertools.count():
data = compat_urllib_parse.urlencode({
'offset': i * 50,
+ 'limit': 50,
'client_id': self._CLIENT_ID,
})
new_entries = self._download_json(
base_url + data, uploader, 'Downloading track page %s' % (i + 1))
- entries.extend(self._extract_info_dict(e, quiet=True) for e in new_entries)
- if len(new_entries) < 50:
+ if len(new_entries) == 0:
+ self.to_screen('%s: End page received' % uploader)
break
+ entries.extend(self._extract_info_dict(e, quiet=True) for e in new_entries)
return {
'_type': 'playlist',
class SoundcloudPlaylistIE(SoundcloudIE):
_VALID_URL = r'https?://api\.soundcloud\.com/playlists/(?P<id>[0-9]+)'
IE_NAME = 'soundcloud:playlist'
+ _TESTS = [
- # it's in tests/test_playlists.py
- _TESTS = []
+ {
+ 'url': 'http://api.soundcloud.com/playlists/4110309',
+ 'info_dict': {
+ 'id': '4110309',
+ 'title': 'TILT Brass - Bowery Poetry Club, August \'03 [Non-Site SCR 02]',
+ 'description': 're:.*?TILT Brass - Bowery Poetry Club',
+ },
+ 'playlist_count': 6,
+ }
+ ]
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)