Issue #5170: Fixed Unicode output bug in logging and added test case. This is...

Issue #5170: Fixed Unicode output bug in logging and added test case. This is a regression which did not occur in 2.5.

Lib/logging/__init__.py

@@ -748,17 +748,19 @@ class StreamHandler(Handler):
         """
         try:
             msg = self.format(record)
+            stream = self.stream
             fs = "%s\n"
             if not hasattr(types, "UnicodeType"): #if no unicode support...
-                self.stream.write(fs % msg)
+                stream.write(fs % msg)
             else:
                 try:
-                    if getattr(self.stream, 'encoding', None) is not None:
-                        self.stream.write(fs % msg.encode(self.stream.encoding))
+                    if (isinstance(msg, unicode) or
+                        getattr(stream, 'encoding', None) is None):
+                        stream.write(fs % msg)
                     else:
-                        self.stream.write(fs % msg)
+                        stream.write(fs % msg.encode(stream.encoding))
                 except UnicodeError:
-                    self.stream.write(fs % msg.encode("UTF-8"))
+                    stream.write(fs % msg.encode("UTF-8"))
             self.flush()
         except (KeyboardInterrupt, SystemExit):
             raise

Lib/test/test_logging.py

@@ -859,6 +859,7 @@ class MemoryTest(BaseTest):
             ('foo', 'DEBUG', '3'),
         ])
 
+
 class EncodingTest(BaseTest):
     def test_encoding_plain_file(self):
         # In Python 2.x, a plain file object is treated as having no encoding.
@@ -885,6 +886,27 @@ class EncodingTest(BaseTest):
             if os.path.isfile(fn):
                 os.remove(fn)
 
+    def test_encoding_cyrillic_unicode(self):
+        log = logging.getLogger("test")
+        #Get a message in Unicode: Do svidanya in Cyrillic (meaning goodbye)
+        message = u'\u0434\u043e \u0441\u0432\u0438\u0434\u0430\u043d\u0438\u044f'
+        #Ensure it's written in a Cyrillic encoding
+        writer_class = codecs.getwriter('cp1251')
+        stream = cStringIO.StringIO()
+        writer = writer_class(stream, 'strict')
+        handler = logging.StreamHandler(writer)
+        log.addHandler(handler)
+        try:
+            log.warning(message)
+        finally:
+            log.removeHandler(handler)
+            handler.close()
+        # check we wrote exactly those bytes, ignoring trailing \n etc
+        s = stream.getvalue()
+        #Compare against what the data should be when encoded in CP-1251
+        self.assertEqual(s, '\xe4\xee \xf1\xe2\xe8\xe4\xe0\xed\xe8\xff\n')
+
+
 # Set the locale to the platform-dependent default.  I have no idea
 # why the test does this, but in any case we save the current locale
 # first and restore it at the end.

Misc/NEWS

@@ -80,11 +80,14 @@ Core and Builtins
 Library
 -------
 
+- Issue #5170: Fixed Unicode output bug in logging and added test case.
+  This is a regression which did not occur in 2.5.
+
 - Partial fix to issue #1731706: memory leak in Tkapp_Call when calling
   from a thread different than the one that created the Tcl interpreter.
   Patch by Robert Hancock.
 
-- Issue #5132: Fixed trouble building extensions under Solaris with 
+- Issue #5132: Fixed trouble building extensions under Solaris with
   --enabled-shared activated. Initial patch by Dave Peterson.
 
 - Issue #1581476: Always use the Tcl global namespace when calling into Tcl.