Merge lp:~jelmer/brz/ignore-unicode-thingy into lp:brz

[Jelmer Vernooij]

Cope with utf-8 invalid characters in commit messages when scanning for revision ids.

=== modified file 'breezy/git/mapping.py'
--- breezy/git/mapping.py	2019-02-11 19:40:51 +0000
+++ breezy/git/mapping.py	2019-04-07 00:36:55 +0000
@@ -383,10 +383,14 @@
             encoding = commit.encoding.decode('ascii')
         else:
             encoding = 'utf-8'
-        message, metadata = self._decode_commit_message(
-            None, commit.message, encoding)
-        if metadata.revision_id:
-            return metadata.revision_id
+        try:
+            message, metadata = self._decode_commit_message(
+                None, commit.message, encoding)
+        except UnicodeDecodeError:
+            pass
+        else:
+            if metadata.revision_id:
+                return metadata.revision_id
         return self.revision_id_foreign_to_bzr(commit.id)
 
     def import_commit(self, commit, lookup_parent_revid):

=== modified file 'breezy/git/tests/test_mapping.py'
--- breezy/git/tests/test_mapping.py	2019-02-11 19:40:51 +0000
+++ breezy/git/tests/test_mapping.py	2019-04-07 00:36:55 +0000
@@ -239,6 +239,21 @@
             mapping.revision_id_foreign_to_bzr(c.id),
             mapping.get_revision_id(c))
 
+    def test_invalid_utf8(self):
+        c = Commit()
+        c.tree = b"cc9462f7f8263ef5adfbeff2fb936bb36b504cba"
+        c.message = b"Some message \xc1"
+        c.committer = b"Committer"
+        c.commit_time = 4
+        c.author_time = 5
+        c.commit_timezone = 60 * 5
+        c.author_timezone = 60 * 3
+        c.author = b"Author"
+        mapping = BzrGitMappingv1()
+        self.assertEqual(
+            mapping.revision_id_foreign_to_bzr(c.id),
+            mapping.get_revision_id(c))
+
 
 class RoundtripRevisionsFromBazaar(tests.TestCase):