Merge pull request gitpython-developers#467 from gitpython-developers/fix-dont-choke-on-invalid-unicode-paths

Byron · web-flow · commit d22c40b942cc · 2016-06-14T07:38:26.000+02:00
Don't choke on (legitimately) invalidly encoded Unicode paths
diff --git a/git/diff.py b/git/diff.py
@@ -404,15 +404,15 @@ def _index_from_patch_format(cls, repo, stream):
             a_mode = old_mode or deleted_file_mode or (a_path and (b_mode or new_mode or new_file_mode))
             b_mode = b_mode or new_mode or new_file_mode or (b_path and a_mode)
             index.append(Diff(repo,
-                              a_path and a_path.decode(defenc),
-                              b_path and b_path.decode(defenc),
+                              a_path and a_path.decode(defenc, 'replace'),
+                              b_path and b_path.decode(defenc, 'replace'),
                               a_blob_id and a_blob_id.decode(defenc),
                               b_blob_id and b_blob_id.decode(defenc),
                               a_mode and a_mode.decode(defenc),
                               b_mode and b_mode.decode(defenc),
                               new_file, deleted_file,
-                              rename_from and rename_from.decode(defenc),
-                              rename_to and rename_to.decode(defenc),
+                              rename_from and rename_from.decode(defenc, 'replace'),
+                              rename_to and rename_to.decode(defenc, 'replace'),
                               None))
 
             previous_header = header
diff --git a/git/test/fixtures/diff_patch_unsafe_paths b/git/test/fixtures/diff_patch_unsafe_paths
@@ -68,6 +68,13 @@ index 0000000000000000000000000000000000000000..eaf5f7510320b6a327fb308379de2f94
 +++ "b/path/\360\237\222\251.txt"
 @@ -0,0 +1 @@
 +dummy content
+diff --git "a/path/\200-invalid-unicode-path.txt" "b/path/\200-invalid-unicode-path.txt"
+new file mode 100644
+index 0000000000000000000000000000000000000000..eaf5f7510320b6a327fb308379de2f94d8859a54
+--- /dev/null
++++ "b/path/\200-invalid-unicode-path.txt"
+@@ -0,0 +1 @@
++dummy content
 diff --git a/a/with spaces b/b/with some spaces
 similarity index 100%
 rename from a/with spaces
diff --git a/git/test/test_diff.py b/git/test/test_diff.py
@@ -162,16 +162,17 @@ def test_diff_unsafe_paths(self):
         self.assertEqual(res[7].b_path, u'path/with-question-mark?')
         self.assertEqual(res[8].b_path, u'path/¯\\_(ツ)_|¯')
         self.assertEqual(res[9].b_path, u'path/💩.txt')
+        self.assertEqual(res[10].b_path, u'path/�-invalid-unicode-path.txt')
 
         # The "Moves"
         # NOTE: The path prefixes a/ and b/ here are legit!  We're actually
         # verifying that it's not "a/a/" that shows up, see the fixture data.
-        self.assertEqual(res[10].a_path, u'a/with spaces')       # NOTE: path a/ here legit!
-        self.assertEqual(res[10].b_path, u'b/with some spaces')  # NOTE: path b/ here legit!
-        self.assertEqual(res[11].a_path, u'a/ending in a space ')
-        self.assertEqual(res[11].b_path, u'b/ending with space ')
-        self.assertEqual(res[12].a_path, u'a/"with-quotes"')
-        self.assertEqual(res[12].b_path, u'b/"with even more quotes"')
+        self.assertEqual(res[11].a_path, u'a/with spaces')       # NOTE: path a/ here legit!
+        self.assertEqual(res[11].b_path, u'b/with some spaces')  # NOTE: path b/ here legit!
+        self.assertEqual(res[12].a_path, u'a/ending in a space ')
+        self.assertEqual(res[12].b_path, u'b/ending with space ')
+        self.assertEqual(res[13].a_path, u'a/"with-quotes"')
+        self.assertEqual(res[13].b_path, u'b/"with even more quotes"')
 
     def test_diff_patch_format(self):
         # test all of the 'old' format diffs for completness - it should at least
