Cope with non-utf8 characters in commit messages.

3 files changed, 21 insertions, 1 deletions

diff --git a/NEWS b/NEWS
index 8bdbd84..6f38e42 100644
--- a/NEWS
+++ b/NEWS
@@ -36,6 +36,8 @@ Bug fixes
 * Strip leading slashes when exporting Subversion repositories.
   (Jelmer Vernooij, #477869)
 
+* Cope with non-utf8 characters in commit messages. (Jelmer Vernooij, #444896)
+
 0.9 28-Feb-2010
 ===============
 
diff --git a/bzr_commit_handler.py b/bzr_commit_handler.py
index d81b271..72c7807 100644
--- a/bzr_commit_handler.py
+++ b/bzr_commit_handler.py
@@ -276,10 +276,11 @@ class GenericCommitHandler(processor.CommitHandler):
         who = self._format_name_email("committer", committer[0], committer[1])
         try:
             message = self.command.message.decode("utf-8")
+
         except UnicodeDecodeError:
             self.warning(
                 "commit message not in utf8 - replacing unknown characters")
-            message = message.decode('utf-8', 'replace')
+            message = self.command.message.decode('utf-8', 'replace')
         if not _serializer_handles_escaping:
             # We need to assume the bad ol' days
             message = helpers.escape_commit_message(message)
diff --git a/tests/test_generic_processor.py b/tests/test_generic_processor.py
index 41f846e..111fd56 100644
--- a/tests/test_generic_processor.py
+++ b/tests/test_generic_processor.py
@@ -1910,3 +1910,20 @@ class TestModifyRevertInBranch(TestCaseForGenericProcessor):
         self.assertEqual(rev_b, rtree_b.inventory[foo_id].revision)
         self.assertEqual(rev_c, rtree_c.inventory[foo_id].revision)
         self.assertEqual(rev_c, rtree_d.inventory[foo_id].revision)
+
+
+class TestCommitCommands(TestCaseForGenericProcessor):
+
+    def test_non_utf8_commit_message(self):
+        handler, branch = self.get_handler()
+        def files_one():
+            yield commands.FileModifyCommand('a',
+                kind_to_mode('file', False), None, "data")
+        def command_list():
+            committer = ['', 'elmer@a.com', time.time(), time.timezone]
+            yield commands.CommitCommand('head', '1', None,
+                committer, 'This is a funky character: \x83', None, [],
+                files_one)
+        handler.process(command_list)
+        rev = branch.repository.get_revision(branch.last_revision())
+        self.assertEquals(u"This is a funky character: \ufffd", rev.message)