Expose git submodule file mode via is_submodule (#147)

README.md

@@ -157,7 +157,7 @@ True
 ('there was a fix\n', 2)
 ```
 
-## Git file modes, symlinks and line numbers
+## Git file modes, symlinks, submodules and line numbers
 
 For git diffs, the file mode is exposed through the `source_mode` and
 `target_mode` attributes (e.g. `'100644'`, `'100755'`, `'120000'`), or `None`
@@ -178,6 +178,21 @@ True
 True
 ```
 
+Likewise, `is_submodule` detects git submodule (gitlink) entries, mode
+`160000`:
+
+```python
+>>> from unidiff import PatchSet
+>>> patch = PatchSet.from_filename('tests/samples/git_submodule.diff')
+>>> patched_file = patch[0]
+>>> patched_file.path
+'tests/diffs/submodule'
+>>> patched_file.target_mode
+'160000'
+>>> patched_file.is_submodule
+True
+```
+
 Each `PatchedFile` also exposes `diff_line_no`, the 1-based line number in the
 diff where its entry starts. This is useful to locate files that have no hunks,
 such as binary changes:

tests/samples/git_submodule.diff

@@ -0,0 +1,7 @@
+diff --git a/tests/diffs/submodule b/tests/diffs/submodule
+new file mode 160000
+index 0000000..b399108
+--- /dev/null
++++ b/tests/diffs/submodule
+@@ -0,0 +1 @@
++Subproject commit b399108e316b17e4e6eed616d112c33796289533

tests/test_parser.py

@@ -615,6 +615,57 @@ def test_added_symlink_file_mode(self):
         self.assertEqual(res[0].target_mode, '120000')
         self.assertTrue(res[0].is_symlink)
 
+    def test_added_submodule_file_mode(self):
+        # issue #147: a new git submodule (gitlink) has mode 160000
+        filename = os.path.join(self.samples_dir, 'samples/git_submodule.diff')
+        with open(filename) as f:
+            res = PatchSet(f)
+
+        self.assertEqual(len(res), 1)
+        self.assertTrue(res[0].is_added_file)
+        self.assertIsNone(res[0].source_mode)
+        self.assertEqual(res[0].target_mode, '160000')
+        self.assertTrue(res[0].is_submodule)
+        self.assertFalse(res[0].is_symlink)
+
+    def test_deleted_submodule_file_mode(self):
+        # issue #147: a removed submodule only carries the old mode
+        diff = (
+            'diff --git a/submodule b/submodule\n'
+            'deleted file mode 160000\n'
+            'index b399108..0000000\n'
+            '--- a/submodule\n'
+            '+++ /dev/null\n'
+            '@@ -1 +0,0 @@\n'
+            '-Subproject commit b399108e316b17e4e6eed616d112c33796289533\n'
+        )
+        res = PatchSet(diff)
+
+        self.assertTrue(res[0].is_removed_file)
+        self.assertEqual(res[0].source_mode, '160000')
+        self.assertIsNone(res[0].target_mode)
+        self.assertTrue(res[0].is_submodule)
+
+    def test_updated_submodule_file_mode(self):
+        # issue #147: a submodule pointer update carries the mode on the
+        # index line
+        diff = (
+            'diff --git a/submodule b/submodule\n'
+            'index b399108..c0ffee1 160000\n'
+            '--- a/submodule\n'
+            '+++ b/submodule\n'
+            '@@ -1 +1 @@\n'
+            '-Subproject commit b399108e316b17e4e6eed616d112c33796289533\n'
+            '+Subproject commit c0ffee1e316b17e4e6eed616d112c33796289533\n'
+        )
+        res = PatchSet(diff)
+
+        self.assertEqual(res[0].source_mode, '160000')
+        self.assertEqual(res[0].target_mode, '160000')
+        self.assertTrue(res[0].is_submodule)
+        self.assertFalse(res[0].is_symlink)
+        self.assertTrue(res[0].is_modified_file)
+
     def test_new_file_mode(self):
         # issue #125: a regular new file carries `new file mode 100644`
         filename = os.path.join(self.samples_dir, 'samples/git_quoted_filename.diff')

unidiff/constants.py

@@ -91,6 +91,9 @@
 # git file mode for a symbolic link
 SYMLINK_FILE_MODE = '120000'
 
+# git file mode for a submodule (gitlink) entry
+SUBMODULE_FILE_MODE = '160000'
+
 LINE_TYPE_ADDED = '+'
 LINE_TYPE_REMOVED = '-'
 LINE_TYPE_CONTEXT = ' '

unidiff/patch.py

@@ -54,6 +54,7 @@
     RE_NO_NEWLINE_MARKER,
     RE_BINARY_DIFF,
     RE_PATCH_FILE_PREFIX,
+    SUBMODULE_FILE_MODE,
     SYMLINK_FILE_MODE,
 )
 from unidiff.errors import UnidiffParseError
@@ -421,12 +422,21 @@ def is_modified_file(self) -> bool:
         return not (self.is_added_file or self.is_removed_file)
 
     @property
-    def is_symlink(self) -> bool:
-        """Return True if the patched file is a symbolic link."""
+    def _file_mode(self) -> Optional[str]:
+        """Return the relevant git file mode, if known."""
         # prefer the target mode; fall back to the source mode (e.g. a
         # removed symlink only carries the old mode)
-        mode = self.target_mode if self.target_mode is not None else self.source_mode
-        return mode == SYMLINK_FILE_MODE
+        return self.target_mode if self.target_mode is not None else self.source_mode
+
+    @property
+    def is_symlink(self) -> bool:
+        """Return True if the patched file is a symbolic link."""
+        return self._file_mode == SYMLINK_FILE_MODE
+
+    @property
+    def is_submodule(self) -> bool:
+        """Return True if the patched file is a git submodule (gitlink)."""
+        return self._file_mode == SUBMODULE_FILE_MODE
 
 
 class PatchSet(list[PatchedFile]):