From 7eb7ead38020bb17683dc3e7c5e609b150771dad Mon Sep 17 00:00:00 2001
From: MaxwellGoblin <lispmachine@outlook.com>
Date: Sat, 29 Jul 2017 19:59:29 +0800
Subject: [PATCH] [soundcloud]update client id and new api

---
 src/you_get/common.py                |  5 ++++
 src/you_get/extractors/soundcloud.py | 40 ++++++++++++++++++++--------
 2 files changed, 34 insertions(+), 11 deletions(-)

diff --git a/src/you_get/common.py b/src/you_get/common.py
index 4e8e168f..33c8acf6 100755
--- a/src/you_get/common.py
+++ b/src/you_get/common.py
@@ -459,6 +459,9 @@ def url_info(url, faker = False, headers = {}):
         'video/x-ms-asf': 'asf',
         'audio/mp4': 'mp4',
         'audio/mpeg': 'mp3',
+        'audio/wav': 'wav',
+        'audio/x-wav': 'wav',
+        'audio/wave': 'wav',
         'image/jpeg': 'jpg',
         'image/png': 'png',
         'image/gif': 'gif',
@@ -1085,6 +1088,8 @@ def print_info(site_info, title, type, size):
         type_info = "MPEG-4 audio (%s)" % type
     elif type in ['audio/mpeg']:
         type_info = "MP3 (%s)" % type
+    elif type in ['audio/wav', 'audio/wave', 'audio/x-wav']:
+        type_info = 'Waveform Audio File Format ({})'.format(type)
 
     elif type in ['image/jpeg']:
         type_info = "JPEG Image (%s)" % type
diff --git a/src/you_get/extractors/soundcloud.py b/src/you_get/extractors/soundcloud.py
index 97d96012..6115041c 100644
--- a/src/you_get/extractors/soundcloud.py
+++ b/src/you_get/extractors/soundcloud.py
@@ -3,28 +3,46 @@
 __all__ = ['soundcloud_download', 'soundcloud_download_by_id']
 
 from ..common import *
+import json
+import urllib.error
 
-def soundcloud_download_by_id(id, title = None, output_dir = '.', merge = True, info_only = False):
+client_id = 'JlZIsxg2hY5WnBgtn3jfS0UYCl0K8DOg'
+
+def soundcloud_download_by_id(id, title=None, output_dir='.', merge=True, info_only=False):
     assert title
+    url = 'https://api.soundcloud.com/tracks/{}/{}?client_id={}'.format(id, 'stream', client_id)
     
-    #if info["downloadable"]:
-    #   url = 'https://api.soundcloud.com/tracks/' + id + '/download?client_id=b45b1aa10f1ac2941910a7f0d10f8e28'
-    url = 'https://api.soundcloud.com/tracks/' + id + '/stream?client_id=02gUJC0hH2ct1EGOcYXQIzRFU91c72Ea'
-    assert url
     type, ext, size = url_info(url)
     
     print_info(site_info, title, type, size)
+
     if not info_only:
         download_urls([url], title, ext, size, output_dir, merge = merge)
 
-def soundcloud_download(url, output_dir = '.', merge = True, info_only = False, **kwargs):
-    metadata = get_html('https://api.soundcloud.com/resolve.json?url=' + url + '&client_id=02gUJC0hH2ct1EGOcYXQIzRFU91c72Ea')
-    import json
+def soundcloud_i1_api(track_id):
+    url = 'https://api.soundcloud.com/i1/tracks/{}/streams?client_id={}'.format(track_id, client_id)
+    return json.loads(get_content(url))['http_mp3_128_url']
+
+def soundcloud_download(url, output_dir='.', merge=True, info_only=False, **kwargs):
+    url = 'https://api.soundcloud.com/resolve.json?url={}&client_id={}'.format(url, client_id)
+    metadata = get_content(url)
     info = json.loads(metadata)
     title = info["title"]
-    id = str(info["id"])
-    
-    soundcloud_download_by_id(id, title, output_dir, merge = merge, info_only = info_only)
+    real_url = info.get('download_url')
+    if real_url is None:
+        real_url = info.get('steram_url')
+    if real_url is None:
+        raise Exception('Cannot get media URI for {}'.format(url))
+    real_url = '{}?client_id={}'.format(real_url, client_id)
+    try:
+        mime, ext, size = url_info(real_url)
+    except urllib.error.HTTPError as e:
+        if 401 == e.status:
+            real_url = soundcloud_i1_api(info['id'])
+            mime, ext, size = url_info(real_url)
+    print_info(site_info, title, mime, size)
+    if not info_only:
+        download_urls([real_url], title, ext, size, output_dir, merge=merge)
 
 site_info = "SoundCloud.com"
 download = soundcloud_download