From b7cc9f50268f41656746009b6a6e8f5d5bf1cb7a Mon Sep 17 00:00:00 2001 From: Philipp Hagemeister Date: Tue, 23 Jul 2013 18:35:52 +0200 Subject: [soundcloud] Support URLs with a slash at the end (Fixes #1104) --- youtube_dl/extractor/soundcloud.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'youtube_dl/extractor/soundcloud.py') diff --git a/youtube_dl/extractor/soundcloud.py b/youtube_dl/extractor/soundcloud.py index d47c49c03..8b8b5166d 100644 --- a/youtube_dl/extractor/soundcloud.py +++ b/youtube_dl/extractor/soundcloud.py @@ -19,7 +19,7 @@ class SoundcloudIE(InfoExtractor): of the stream token and uid """ - _VALID_URL = r'^(?:https?://)?(?:www\.)?soundcloud\.com/([\w\d-]+)/([\w\d-]+)(?:[?].*)?$' + _VALID_URL = r'^(?:https?://)?(?:www\.)?soundcloud\.com/([\w\d-]+)/([\w\d-]+)/?(?:[?].*)?$' IE_NAME = u'soundcloud' _TEST = { u'url': u'http://soundcloud.com/ethmusic/lostin-powers-she-so-heavy', -- cgit v1.2.3 From 7d2392691c18973c86820b796cbfe61cfad9ff11 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jaime=20Marqui=CC=81nez=20Ferra=CC=81ndiz?= Date: Wed, 24 Jul 2013 14:05:14 +0200 Subject: [soundcloud]: Some improvements Extract thumbnails. Make SoundcloudSetIE a subclass of SoundcloudIE to reuse some code. Directly extract the file url without downloading an extra page. --- youtube_dl/extractor/soundcloud.py | 90 ++++++++++++++------------------------ 1 file changed, 34 insertions(+), 56 deletions(-) (limited to 'youtube_dl/extractor/soundcloud.py') diff --git a/youtube_dl/extractor/soundcloud.py b/youtube_dl/extractor/soundcloud.py index 8b8b5166d..54ff8db12 100644 --- a/youtube_dl/extractor/soundcloud.py +++ b/youtube_dl/extractor/soundcloud.py @@ -33,10 +33,35 @@ class SoundcloudIE(InfoExtractor): } } + _CLIENT_ID = 'b45b1aa10f1ac2941910a7f0d10f8e28' + def report_resolve(self, video_id): """Report information extraction.""" self.to_screen(u'%s: Resolving id' % video_id) + @classmethod + 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): + video_id = info['id'] + name = full_title or video_id + self.report_extraction(name) + + thumbnail = info['artwork_url'] + if thumbnail is not None: + thumbnail = thumbnail.replace('-large', '-t500x500') + return { + 'id': info['id'], + 'url': info['stream_url'] + '?client_id=' + self._CLIENT_ID, + 'uploader': info['user']['username'], + 'upload_date': unified_strdate(info['created_at']), + 'title': info['title'], + 'ext': u'mp3', + 'description': info['description'], + 'thumbnail': thumbnail, + } + def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) if mobj is None: @@ -51,41 +76,13 @@ class SoundcloudIE(InfoExtractor): self.report_resolve(full_title) url = 'http://soundcloud.com/%s/%s' % (uploader, slug_title) - resolv_url = 'http://api.soundcloud.com/resolve.json?url=' + url + '&client_id=b45b1aa10f1ac2941910a7f0d10f8e28' + resolv_url = self._resolv_url(url) info_json = self._download_webpage(resolv_url, full_title, u'Downloading info JSON') info = json.loads(info_json) - video_id = info['id'] - self.report_extraction(full_title) - - streams_url = 'https://api.sndcdn.com/i1/tracks/' + str(video_id) + '/streams?client_id=b45b1aa10f1ac2941910a7f0d10f8e28' - stream_json = self._download_webpage(streams_url, full_title, - u'Downloading stream definitions', - u'unable to download stream definitions') - - streams = json.loads(stream_json) - mediaURL = streams['http_mp3_128_url'] - upload_date = unified_strdate(info['created_at']) - - return [{ - 'id': info['id'], - 'url': mediaURL, - 'uploader': info['user']['username'], - 'upload_date': upload_date, - 'title': info['title'], - 'ext': u'mp3', - 'description': info['description'], - }] - -class SoundcloudSetIE(InfoExtractor): - """Information extractor for soundcloud.com sets - To access the media, the uid of the song and a stream token - must be extracted from the page source and the script must make - a request to media.soundcloud.com/crossdomain.xml. Then - the media can be grabbed by requesting from an url composed - of the stream token and uid - """ + return self._extract_info_dict(info, full_title) +class SoundcloudSetIE(SoundcloudIE): _VALID_URL = r'^(?:https?://)?(?:www\.)?soundcloud\.com/([\w\d-]+)/sets/([\w\d-]+)(?:[?].*)?$' IE_NAME = u'soundcloud:set' _TEST = { @@ -153,10 +150,6 @@ class SoundcloudSetIE(InfoExtractor): ] } - def report_resolve(self, video_id): - """Report information extraction.""" - self.to_screen(u'%s: Resolving id' % video_id) - def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) if mobj is None: @@ -171,7 +164,7 @@ class SoundcloudSetIE(InfoExtractor): self.report_resolve(full_title) url = 'http://soundcloud.com/%s/sets/%s' % (uploader, slug_title) - resolv_url = 'http://api.soundcloud.com/resolve.json?url=' + url + '&client_id=b45b1aa10f1ac2941910a7f0d10f8e28' + resolv_url = self._resolv_url(url) info_json = self._download_webpage(resolv_url, full_title) videos = [] @@ -182,23 +175,8 @@ class SoundcloudSetIE(InfoExtractor): return self.report_extraction(full_title) - for track in info['tracks']: - video_id = track['id'] - - streams_url = 'https://api.sndcdn.com/i1/tracks/' + str(video_id) + '/streams?client_id=b45b1aa10f1ac2941910a7f0d10f8e28' - stream_json = self._download_webpage(streams_url, video_id, u'Downloading track info JSON') - - self.report_extraction(video_id) - streams = json.loads(stream_json) - mediaURL = streams['http_mp3_128_url'] - - videos.append({ - 'id': video_id, - 'url': mediaURL, - 'uploader': track['user']['username'], - 'upload_date': unified_strdate(track['created_at']), - 'title': track['title'], - 'ext': u'mp3', - 'description': track['description'], - }) - return videos + return {'_type': 'playlist', + 'entries': [self._extract_info_dict(track) for track in info['tracks']], + 'id': info['id'], + 'title': info['title'], + } -- cgit v1.2.3 From eb6a41ba0f5dbb836d5b48b9e38f406c3c46c0ec Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jaime=20Marqui=CC=81nez=20Ferra=CC=81ndiz?= Date: Wed, 24 Jul 2013 14:39:21 +0200 Subject: ExfmIE: extract Soundcloud songs using SoundcloudIE Now SouncloudIE accepts api urls. --- youtube_dl/extractor/soundcloud.py | 39 +++++++++++++++++++++++++------------- 1 file changed, 26 insertions(+), 13 deletions(-) (limited to 'youtube_dl/extractor/soundcloud.py') diff --git a/youtube_dl/extractor/soundcloud.py b/youtube_dl/extractor/soundcloud.py index 54ff8db12..7c9f1c6b6 100644 --- a/youtube_dl/extractor/soundcloud.py +++ b/youtube_dl/extractor/soundcloud.py @@ -19,7 +19,11 @@ class SoundcloudIE(InfoExtractor): 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\d+)) + ) + ''' IE_NAME = u'soundcloud' _TEST = { u'url': u'http://soundcloud.com/ethmusic/lostin-powers-she-so-heavy', @@ -35,6 +39,10 @@ class SoundcloudIE(InfoExtractor): _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) @@ -63,21 +71,26 @@ class SoundcloudIE(InfoExtractor): } 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 + 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) -- cgit v1.2.3 From 668de34c6bbe48f574f23ad898fa904a7c1ad84b Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jaime=20Marqui=CC=81nez=20Ferra=CC=81ndiz?= Date: Wed, 21 Aug 2013 17:06:37 +0200 Subject: [soundcloud] Support widget urls (fixes #1252) --- youtube_dl/extractor/soundcloud.py | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'youtube_dl/extractor/soundcloud.py') diff --git a/youtube_dl/extractor/soundcloud.py b/youtube_dl/extractor/soundcloud.py index 7c9f1c6b6..5f3a5540d 100644 --- a/youtube_dl/extractor/soundcloud.py +++ b/youtube_dl/extractor/soundcloud.py @@ -4,6 +4,7 @@ import re from .common import InfoExtractor from ..utils import ( compat_str, + compat_urlparse, ExtractorError, unified_strdate, @@ -22,6 +23,7 @@ class SoundcloudIE(InfoExtractor): _VALID_URL = r'''^(?:https?://)? (?:(?:(?:www\.)?soundcloud\.com/([\w\d-]+)/([\w\d-]+)/?(?:[?].*)?$) |(?:api\.soundcloud\.com/tracks/(?P\d+)) + |(?Pw.soundcloud.com/player/?.*?url=.*) ) ''' IE_NAME = u'soundcloud' @@ -79,6 +81,9 @@ class SoundcloudIE(InfoExtractor): 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) -- cgit v1.2.3