unicode handling in messages and trees was improved. Messages are now written according to the encoding of the commit object, and decoded using that information as well. Trees will encode and decode their names with utf8

Author: Byron

lib/git/ext/gitdb

@@ -31,6 +31,7 @@
 					altzone
 				)
 import os
+import sys
 
 __all__ = ('Commit', )
 
@@ -381,7 +382,13 @@ def _serialize(self, stream):
 			write("encoding %s\n" % self.encoding)
 
 		write("\n")
-		write(self.message)
+		
+		# write plain bytes, be sure its encoded according to our encoding
+		if isinstance(self.message, unicode):
+			write(self.message.encode(self.encoding))
+		else:
+			write(self.message)
+		# END handle encoding
 		return self
 
 	def _deserialize(self, stream):
@@ -421,6 +428,11 @@ def _deserialize(self, stream):
 		# a stream from our data simply gives us the plain message
 		# The end of our message stream is marked with a newline that we strip
 		self.message = stream.read()
+		try:
+			self.message = self.message.decode(self.encoding)
+		except Exception:
+			print >> sys.stderr, "Failed to decode message: %s" % self.message
+		# END exception handling 
 		return self
 
 	#} END serializable implementation

lib/git/objects/commit.py

@@ -66,7 +66,14 @@ def tree_entries_from_data(data):
 		while data[i] != '\0':
 			i += 1
 		# END while not reached NULL
+		
+		# default encoding for strings in git is utf8
+		# Only use the respective unicode object if the byte stream was encoded
 		name = data[ns:i]
+		name_enc = name.decode("utf-8")
+		if len(name) > len(name_enc):
+			name = name_enc
+		# END handle encoding
 
 		# byte is NULL, get next 20
 		i += 1