import json
import re
+import itertools
from .common import InfoExtractor
from ..utils import (
compat_str,
+ compat_urlparse,
+ compat_urllib_parse,
ExtractorError,
unified_strdate,
of the stream token and uid
"""
- _VALID_URL = r'^(?:https?://)?(?:www\.)?soundcloud\.com/([\w\d-]+)/([\w\d-]+)/?(?:[?].*)?$'
+ _VALID_URL = r'''^(?:https?://)?
+ (?:(?:(?:www\.)?soundcloud\.com/([\w\d-]+)/([\w\d-]+)/?(?:[?].*)?$)
+ |(?:api\.soundcloud\.com/tracks/(?P<track_id>\d+))
+ |(?P<widget>w.soundcloud.com/player/?.*?url=.*)
+ )
+ '''
IE_NAME = u'soundcloud'
_TEST = {
u'url': u'http://soundcloud.com/ethmusic/lostin-powers-she-so-heavy',
_CLIENT_ID = 'b45b1aa10f1ac2941910a7f0d10f8e28'
+ @classmethod
+ def suitable(cls, url):
+ return re.match(cls._VALID_URL, url, flags=re.VERBOSE) is not None
+
def report_resolve(self, video_id):
"""Report information extraction."""
self.to_screen(u'%s: Resolving id' % video_id)
def _resolv_url(cls, url):
return 'http://api.soundcloud.com/resolve.json?url=' + url + '&client_id=' + cls._CLIENT_ID
- def _extract_info_dict(self, info, full_title=None):
+ def _extract_info_dict(self, info, full_title=None, quiet=False):
video_id = info['id']
name = full_title or video_id
- self.report_extraction(name)
+ if quiet == False:
+ self.report_extraction(name)
thumbnail = info['artwork_url']
if thumbnail is not None:
}
def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url)
+ mobj = re.match(self._VALID_URL, url, flags=re.VERBOSE)
if mobj is None:
raise ExtractorError(u'Invalid URL: %s' % url)
- # extract uploader (which is in the url)
- uploader = mobj.group(1)
- # extract simple title (uploader + slug of song title)
- slug_title = mobj.group(2)
- full_title = '%s/%s' % (uploader, slug_title)
-
- self.report_resolve(full_title)
-
- url = 'http://soundcloud.com/%s/%s' % (uploader, slug_title)
- resolv_url = self._resolv_url(url)
- info_json = self._download_webpage(resolv_url, full_title, u'Downloading info JSON')
+ track_id = mobj.group('track_id')
+ if track_id is not None:
+ info_json_url = 'http://api.soundcloud.com/tracks/' + track_id + '.json?client_id=' + self._CLIENT_ID
+ full_title = track_id
+ elif mobj.group('widget'):
+ query = compat_urlparse.parse_qs(compat_urlparse.urlparse(url).query)
+ return self.url_result(query['url'][0], ie='Soundcloud')
+ else:
+ # extract uploader (which is in the url)
+ uploader = mobj.group(1)
+ # extract simple title (uploader + slug of song title)
+ slug_title = mobj.group(2)
+ full_title = '%s/%s' % (uploader, slug_title)
+
+ self.report_resolve(full_title)
+
+ url = 'http://soundcloud.com/%s/%s' % (uploader, slug_title)
+ info_json_url = self._resolv_url(url)
+ info_json = self._download_webpage(info_json_url, full_title, u'Downloading info JSON')
info = json.loads(info_json)
return self._extract_info_dict(info, full_title)
'id': info['id'],
'title': info['title'],
}
+
+
+class SoundcloudUserIE(SoundcloudIE):
+ _VALID_URL = r'https?://(www\.)?soundcloud.com/(?P<user>[^/]+)(/?(tracks/)?)?(\?.*)?$'
+ IE_NAME = u'soundcloud:user'
+
+ # it's in tests/test_playlists.py
+ _TEST = None
+
+ def _real_extract(self, url):
+ mobj = re.match(self._VALID_URL, url)
+ uploader = mobj.group('user')
+
+ url = 'http://soundcloud.com/%s/' % uploader
+ resolv_url = self._resolv_url(url)
+ user_json = self._download_webpage(resolv_url, uploader,
+ u'Downloading user info')
+ user = json.loads(user_json)
+
+ tracks = []
+ for i in itertools.count():
+ data = compat_urllib_parse.urlencode({'offset': i*50,
+ 'client_id': self._CLIENT_ID,
+ })
+ tracks_url = 'http://api.soundcloud.com/users/%s/tracks.json?' % user['id'] + data
+ response = self._download_webpage(tracks_url, uploader,
+ u'Downloading tracks page %s' % (i+1))
+ new_tracks = json.loads(response)
+ tracks.extend(self._extract_info_dict(track, quiet=True) for track in new_tracks)
+ if len(new_tracks) < 50:
+ break
+
+ return {
+ '_type': 'playlist',
+ 'id': compat_str(user['id']),
+ 'title': user['username'],
+ 'entries': tracks,
+ }