X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fsoundcloud.py;h=e22ff9c387ab0e01c1e6fcb1da793af877f37a5c;hb=eadaf08c1670d2fc9a2078faca95fbc49d6a5af0;hp=4717fbb77e0ec21a26147b25d9cc3be2f83d9f94;hpb=6c5ad80cdcd3f51b61a9d21c55e21d51e6b2f39a;p=youtube-dl diff --git a/youtube_dl/extractor/soundcloud.py b/youtube_dl/extractor/soundcloud.py index 4717fbb77..e22ff9c38 100644 --- a/youtube_dl/extractor/soundcloud.py +++ b/youtube_dl/extractor/soundcloud.py @@ -1,3 +1,4 @@ +# encoding: utf-8 import json import re import itertools @@ -23,9 +24,12 @@ class SoundcloudIE(InfoExtractor): """ _VALID_URL = r'''^(?:https?://)? - (?:(?:(?:www\.)?soundcloud\.com/([\w\d-]+)/([\w\d-]+)/?(?:[?].*)?$) + (?:(?:(?:www\.|m\.)?soundcloud\.com/ + (?P[\w\d-]+)/ + (?!sets/)(?P[\w\d-]+)/? + (?P<token>[^?]+?)?(?:[?].*)?$) |(?:api\.soundcloud\.com/tracks/(?P<track_id>\d+)) - |(?P<widget>w.soundcloud.com/player/?.*?url=.*) + |(?P<widget>w\.soundcloud\.com/player/?.*?url=.*) ) ''' IE_NAME = u'soundcloud' @@ -56,9 +60,36 @@ class SoundcloudIE(InfoExtractor): u'skip_download': True, }, }, + # private link + { + u'url': u'https://soundcloud.com/jaimemf/youtube-dl-test-video-a-y-baw/s-8Pjrp', + u'md5': u'aa0dd32bfea9b0c5ef4f02aacd080604', + u'info_dict': { + u'id': u'123998367', + u'ext': u'mp3', + u'title': u'Youtube - Dl Test Video \'\' Ä↭', + u'uploader': u'jaimeMF', + u'description': u'test chars: \"\'/\\ä↭', + u'upload_date': u'20131209', + }, + }, + # downloadable song + { + u'url': u'https://soundcloud.com/simgretina/just-your-problem-baby-1', + u'md5': u'56a8b69568acaa967b4c49f9d1d52d19', + u'info_dict': { + u'id': u'105614606', + u'ext': u'wav', + u'title': u'Just Your Problem Baby (Acapella)', + u'description': u'Vocals', + u'uploader': u'Sim Gretina', + u'upload_date': u'20130815', + }, + }, ] _CLIENT_ID = 'b45b1aa10f1ac2941910a7f0d10f8e28' + _IPHONE_CLIENT_ID = '376f225bf427445fc4bfb6b99b72e0bf' @classmethod def suitable(cls, url): @@ -72,39 +103,84 @@ class SoundcloudIE(InfoExtractor): 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, quiet=False): + def _extract_info_dict(self, info, full_title=None, quiet=False, secret_token=None): track_id = compat_str(info['id']) name = full_title or track_id - if quiet == False: + if quiet: self.report_extraction(name) thumbnail = info['artwork_url'] if thumbnail is not None: thumbnail = thumbnail.replace('-large', '-t500x500') + ext = u'mp3' result = { - 'id': track_id, - 'url': info['stream_url'] + '?client_id=' + self._CLIENT_ID, + 'id': track_id, 'uploader': info['user']['username'], 'upload_date': unified_strdate(info['created_at']), - 'title': info['title'], - 'ext': u'mp3', + 'title': info['title'], 'description': info['description'], 'thumbnail': thumbnail, } if info.get('downloadable', False): - result['url'] = 'https://api.soundcloud.com/tracks/{0}/download?client_id={1}'.format(track_id, self._CLIENT_ID) - if not info.get('streamable', False): - # We have to get the rtmp url + # We can build a direct link to the song + format_url = ( + u'https://api.soundcloud.com/tracks/{0}/download?client_id={1}'.format( + track_id, self._CLIENT_ID)) + result['formats'] = [{ + 'format_id': 'download', + 'ext': info.get('original_format', u'mp3'), + 'url': format_url, + 'vcodec': 'none', + }] + else: + # We have to retrieve the url + streams_url = ('http://api.soundcloud.com/i1/tracks/{0}/streams?' + 'client_id={1}&secret_token={2}'.format(track_id, self._IPHONE_CLIENT_ID, secret_token)) stream_json = self._download_webpage( - 'http://api.soundcloud.com/i1/tracks/{0}/streams?client_id={1}'.format(track_id, self._CLIENT_ID), + streams_url, track_id, u'Downloading track url') - rtmp_url = json.loads(stream_json)['rtmp_mp3_128_url'] - # The url doesn't have an rtmp app, we have to extract the playpath - url, path = rtmp_url.split('mp3:', 1) - result.update({ - 'url': url, - 'play_path': 'mp3:' + path, - }) + + formats = [] + format_dict = json.loads(stream_json) + for key, stream_url in format_dict.items(): + if key.startswith(u'http'): + formats.append({ + 'format_id': key, + 'ext': ext, + 'url': stream_url, + 'vcodec': 'none', + }) + elif key.startswith(u'rtmp'): + # The url doesn't have an rtmp app, we have to extract the playpath + url, path = stream_url.split('mp3:', 1) + formats.append({ + 'format_id': key, + 'url': url, + 'play_path': 'mp3:' + path, + 'ext': ext, + 'vcodec': 'none', + }) + + if not formats: + # We fallback to the stream_url in the original info, this + # cannot be always used, sometimes it can give an HTTP 404 error + formats.append({ + 'format_id': u'fallback', + 'url': info['stream_url'] + '?client_id=' + self._CLIENT_ID, + 'ext': ext, + 'vcodec': 'none', + }) + + def format_pref(f): + if f['format_id'].startswith('http'): + return 2 + if f['format_id'].startswith('rtmp'): + return 1 + return 0 + + formats.sort(key=format_pref) + result['formats'] = formats + return result def _real_extract(self, url): @@ -113,6 +189,7 @@ class SoundcloudIE(InfoExtractor): raise ExtractorError(u'Invalid URL: %s' % url) track_id = mobj.group('track_id') + token = None 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 @@ -121,19 +198,22 @@ class SoundcloudIE(InfoExtractor): return self.url_result(query['url'][0], ie='Soundcloud') else: # extract uploader (which is in the url) - uploader = mobj.group(1) + uploader = mobj.group('uploader') # extract simple title (uploader + slug of song title) - slug_title = mobj.group(2) - full_title = '%s/%s' % (uploader, slug_title) + slug_title = mobj.group('title') + token = mobj.group('token') + full_title = resolve_title = '%s/%s' % (uploader, slug_title) + if token: + resolve_title += '/%s' % token self.report_resolve(full_title) - url = 'http://soundcloud.com/%s/%s' % (uploader, slug_title) + url = 'http://soundcloud.com/%s' % resolve_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) + return self._extract_info_dict(info, full_title, secret_token=token) class SoundcloudSetIE(SoundcloudIE): _VALID_URL = r'^(?:https?://)?(?:www\.)?soundcloud\.com/([\w\d-]+)/sets/([\w\d-]+)(?:[?].*)?$' @@ -158,7 +238,6 @@ class SoundcloudSetIE(SoundcloudIE): resolv_url = self._resolv_url(url) info_json = self._download_webpage(resolv_url, full_title) - videos = [] info = json.loads(info_json) if 'errors' in info: for err in info['errors']: @@ -174,7 +253,7 @@ class SoundcloudSetIE(SoundcloudIE): class SoundcloudUserIE(SoundcloudIE): - _VALID_URL = r'https?://(www\.)?soundcloud.com/(?P<user>[^/]+)(/?(tracks/)?)?(\?.*)?$' + _VALID_URL = r'https?://(www\.)?soundcloud\.com/(?P<user>[^/]+)(/?(tracks/)?)?(\?.*)?$' IE_NAME = u'soundcloud:user' # it's in tests/test_playlists.py