[mixcloud] fix cloudcast data extraction(closes #22821)

[youtube-dl] / youtube_dl / extractor / mixcloud.py

diff --git a/youtube_dl/extractor/mixcloud.py b/youtube_dl/extractor/mixcloud.py
index 785b99bc3e976b06d7ff67291dc78478cfff8350..e5f631506e70c4be6b884abf4d8cadd756a71617 100644 (file)
--- a/youtube_dl/extractor/mixcloud.py
+++ b/youtube_dl/extractor/mixcloud.py
@@ -1,6 +1,5 @@
 from __future__ import unicode_literals
 
-import base64
 import functools
 import itertools
 import re
@@ -87,9 +86,10 @@ class MixcloudIE(InfoExtractor):
                 r'<script id="relay-data" type="text/x-mixcloud">([^<]+)</script>',
                 webpage, 'play info'), 'play info')
             for item in full_info_json:
-                item_data = try_get(
-                    item, lambda x: x['cloudcast']['data']['cloudcastLookup'],
-                    dict)
+                item_data = try_get(item, [
+                    lambda x: x['cloudcast']['data']['cloudcastLookup'],
+                    lambda x: x['cloudcastLookup']['data']['cloudcastLookup'],
+                ], dict)
                 if try_get(item_data, lambda x: x['streamInfo']['url']):
                     info_json = item_data
                     break
@@ -162,11 +162,17 @@ class MixcloudIE(InfoExtractor):
             stream_info = info_json['streamInfo']
             formats = []
 
+            def decrypt_url(f_url):
+                for k in (key, 'IFYOUWANTTHEARTISTSTOGETPAIDDONOTDOWNLOADFROMMIXCLOUD'):
+                    decrypted_url = self._decrypt_xor_cipher(k, f_url)
+                    if re.search(r'^https?://[0-9A-Za-z.]+/[0-9A-Za-z/.?=&_-]+$', decrypted_url):
+                        return decrypted_url
+
             for url_key in ('url', 'hlsUrl', 'dashUrl'):
                 format_url = stream_info.get(url_key)
                 if not format_url:
                     continue
-                decrypted = self._decrypt_xor_cipher(key, compat_b64decode(format_url))
+                decrypted = decrypt_url(compat_b64decode(format_url))
                 if not decrypted:
                     continue
                 if url_key == 'hlsUrl':
@@ -180,6 +186,10 @@ class MixcloudIE(InfoExtractor):
                     formats.append({
                         'format_id': 'http',
                         'url': decrypted,
+                        'downloader_options': {
+                            # Mixcloud starts throttling at >~5M
+                            'http_chunk_size': 5242880,
+                        },
                     })
             self._sort_formats(formats)