Fix unicode_escape (Fixes #2695)

diff --git a/test/test_utils.py b/test/test_utils.py
index 2348c04159e54cb5fe293fcd300c2f25116588a6..51eb0b6b936c7ea5d21cfef9bdc0b70f2ee7663a 100644 (file)
--- a/test/test_utils.py
+++ b/test/test_utils.py
@@ -38,6 +38,7 @@ from youtube_dl.utils import (
     xpath_with_ns,
     parse_iso8601,
     strip_jsonp,
+    uppercase_escape,
 )
 
 if sys.version_info < (3, 0):
@@ -279,6 +280,9 @@ class TestUtil(unittest.TestCase):
         d = json.loads(stripped)
         self.assertEqual(d, [{"id": "532cb", "x": 3}])
 
+    def test_uppercase_escpae(self):
+        self.assertEqual(uppercase_escape(u'aä'), u'aä')
+        self.assertEqual(uppercase_escape(u'\\U0001d550'), u'𝕐')
 
 if __name__ == '__main__':
     unittest.main()

diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py
index 5f1f664c88f609868839ebf3f1db043a1c69d1d9..92fee966fce8c5432268e33e90f893a8ca322606 100644 (file)
--- a/youtube_dl/utils.py
+++ b/youtube_dl/utils.py
@@ -2,6 +2,7 @@
 # -*- coding: utf-8 -*-
 
 import calendar
+import codecs
 import contextlib
 import ctypes
 import datetime
@@ -1263,9 +1264,11 @@ class PagedList(object):
 
 
 def uppercase_escape(s):
+    unicode_escape = codecs.getdecoder('unicode_escape')
     return re.sub(
         r'\\U[0-9a-fA-F]{8}',
-        lambda m: m.group(0).decode('unicode-escape'), s)
+        lambda m: unicode_escape(m.group(0))[0],
+        s)
 
 try:
     struct.pack(u'!I', 0)