Support git mnemonic source/target prefixes (#81, #101)

tests/test_parser.py

@@ -340,6 +340,29 @@ def test_parse_diff_git_no_prefix(self):
         self.assertTrue(res[2].is_added_file)
         self.assertEqual(res[2].path, 'file3')
 
+    def test_parse_diff_git_mnemonic_prefix(self):
+        # issue #81: git diff with diff.mnemonicPrefix set uses i/ w/ (etc.)
+        # instead of a/ b/; path should still resolve to the plain filename.
+        diff = (
+            'diff --git i/foo/bar.py w/foo/bar.py\n'
+            'index abc1234..def5678 100644\n'
+            '--- i/foo/bar.py\n'
+            '+++ w/foo/bar.py\n'
+            '@@ -1,2 +1,2 @@\n'
+            ' a\n'
+            '-b\n'
+            '+c\n'
+        )
+        res = PatchSet(diff)
+
+        self.assertEqual(len(res), 1)
+        self.assertEqual(res[0].source_file, 'i/foo/bar.py')
+        self.assertEqual(res[0].target_file, 'w/foo/bar.py')
+        self.assertEqual(res[0].path, 'foo/bar.py')
+        self.assertFalse(res[0].is_rename)
+        self.assertTrue(res[0].is_modified_file)
+        self.assertEqual(str(res), diff)
+
     def test_parse_diff_with_empty_filenames(self):
         # regression test for issue #115: difflib.unified_diff() without
         # fromfile/tofile emits bare "--- " / "+++ " headers (empty

tests/test_patchedfile.py

@@ -50,3 +50,20 @@ def test_is_modified_file(self):
         hunk = Hunk(src_start=1, src_len=10, tgt_start=1, tgt_len=8)
         self.patched_file.append(hunk)
         self.assertTrue(self.patched_file.is_modified_file)
+
+    def test_default_file_prefix(self):
+        patched_file = PatchedFile(source="a/foo/bar", target="b/foo/bar")
+        self.assertEqual(patched_file.path, "foo/bar")
+
+    def test_git_mnemonic_file_prefix(self):
+        # mnemonic prefixes used when diff.mnemonicPrefix is set (c/ i/ o/ w/)
+        # and the 1/ 2/ pair used by `git diff --no-index`
+        for prefix in ('c', 'i', 'o', 'w', '1', '2'):
+            patched_file = PatchedFile(source="%s/foo/bar" % prefix,
+                                       target="%s/foo/bar" % prefix)
+            self.assertEqual(patched_file.path, "foo/bar")
+
+    def test_no_file_prefix(self):
+        # a leading slash is not a prefix and must be preserved
+        patched_file = PatchedFile(source="/foo/bar", target="/foo/bar")
+        self.assertEqual(patched_file.path, "/foo/bar")

unidiff/constants.py

@@ -79,6 +79,11 @@
     r'(?P<source_filename>[^\t]+?)(?:\t(?P<source_timestamp>[\s0-9:\+-]+))?'
     r'(?: and (?P<target_filename>[^\t]+?)(?:\t(?P<target_timestamp>[\s0-9:\+-]+))?)? (differ|has changed)')
 
+# git source/target filename prefixes: the standard "a/" and "b/", plus the
+# mnemonic prefixes used when diff.mnemonicPrefix is set (c/ i/ o/ w/) and the
+# 1/ 2/ pair used by `git diff --no-index`
+RE_PATCH_FILE_PREFIX = re.compile(r'^[abciow12]/')
+
 DEFAULT_ENCODING = 'UTF-8'
 
 DEV_NULL = '/dev/null'

unidiff/patch.py

@@ -53,6 +53,7 @@
     RE_TARGET_FILENAME,
     RE_NO_NEWLINE_MARKER,
     RE_BINARY_DIFF,
+    RE_PATCH_FILE_PREFIX,
     SYMLINK_FILE_MODE,
 )
 from unidiff.errors import UnidiffParseError
@@ -366,7 +367,7 @@ def path(self) -> str:
         if quoted:
             filepath = filepath[1:-1]
 
-        if filepath.startswith('a/') or filepath.startswith('b/'):
+        if RE_PATCH_FILE_PREFIX.match(filepath):
             filepath = filepath[2:]
 
         if quoted: