Merge lp:~jelmer/brz/lookup-error into lp:brz

[Jelmer Vernooij]

Raise custom error for unknown encodings in Git commits.

Raise custom error for unknown encodings in Git commits.

[Jelmer Vernooij (jelmer)]

review: Approve

[The Breezy Bot (the-breezy-bot)]

Running landing tests failed
https://ci.breezy-vcs.org/job/brz/job/brz-land/1028/

=== modified file 'breezy/git/mapping.py'
--- breezy/git/mapping.py	2022-05-15 21:32:05 +0000
+++ breezy/git/mapping.py	2022-08-26 09:17:46 +0000
@@ -89,6 +89,14 @@
         self.fields = b",".join(fields)
 
 
+class UnknownCommitEncoding(errors.BzrError):
+    _fmt = "Unknown commit encoding: %(encoding)s"
+
+    def __init__(self, encoding):
+        errors.BzrError.__init__(self)
+        self.encoding = encoding
+
+
 def escape_file_id(file_id):
     file_id = file_id.replace(b'_', b'__')
     file_id = file_id.replace(b' ', b'_s')
@@ -409,9 +417,15 @@
         rev.git_metadata = None
 
         def decode_using_encoding(rev, commit, encoding):
-            rev.committer = commit.committer.decode(encoding)
-            if commit.committer != commit.author:
-                rev.properties[u'author'] = commit.author.decode(encoding)
+            try:
+                rev.committer = commit.committer.decode(encoding)
+            except LookupError:
+                raise UnknownCommitEncoding(encoding)
+            try:
+                if commit.committer != commit.author:
+                    rev.properties[u'author'] = commit.author.decode(encoding)
+            except LookupError:
+                raise UnknownCommitEncoding(encoding)
             rev.message, rev.git_metadata = self._decode_commit_message(
                 rev, commit.message, encoding)
 
@@ -443,7 +457,8 @@
                 'utf-8', 'surrogateescape')
         if commit.mergetag:
             for i, tag in enumerate(commit.mergetag):
-                rev.properties[u'git-mergetag-%d' % i] = tag.as_raw_string().decode('utf-8', 'surrogateescape')
+                rev.properties[u'git-mergetag-%d' % i] = tag.as_raw_string().decode(
+                    'utf-8', 'surrogateescape')
         rev.timestamp = commit.commit_time
         rev.timezone = commit.commit_timezone
         rev.parent_ids = None
@@ -514,7 +529,10 @@
         message = self._extract_hg_metadata(rev, message)
         message = self._extract_git_svn_metadata(rev, message)
         message, metadata = self._extract_bzr_metadata(rev, message)
-        return message.decode(encoding), metadata
+        try:
+            return message.decode(encoding), metadata
+        except LookupError:
+            raise UnknownCommitEncoding(encoding)
 
     def _encode_commit_message(self, rev, message, encoding):
         ret = message.encode(encoding)

=== modified file 'breezy/git/tests/test_mapping.py'
--- breezy/git/tests/test_mapping.py	2022-07-04 08:43:18 +0000
+++ breezy/git/tests/test_mapping.py	2022-08-26 09:17:46 +0000
@@ -38,6 +38,7 @@
     fix_person_identifier,
     unescape_file_id,
     UnknownCommitExtra,
+    UnknownCommitEncoding,
     UnknownMercurialCommitExtra,
     )
 
@@ -120,6 +121,23 @@
         self.assertEqual("180", rev.properties[u'author-timezone'])
         self.assertEqual(b"git-v1:" + c.id, rev.revision_id)
 
+    def test_unknown_encoding(self):
+        c = Commit()
+        c.tree = b"cc9462f7f8263ef5adfbeff2fb936bb36b504cba"
+        c.message = b"Some message"
+        c.committer = b"Committer"
+        c.commit_time = 4
+        c.author_time = 5
+        c.commit_timezone = 60 * 5
+        c.author_timezone = 60 * 3
+        c.author = u"Authér".encode("iso8859-1")
+        c.encoding = b"Unknown"
+        mapping = BzrGitMappingv1()
+        e = self.assertRaises(
+            UnknownCommitEncoding, mapping.import_commit,
+            c, mapping.revision_id_foreign_to_bzr)
+        self.assertEquals(e.encoding, "Unknown")
+
     def test_explicit_encoding(self):
         c = Commit()
         c.tree = b"cc9462f7f8263ef5adfbeff2fb936bb36b504cba"