Perfect processing, 35% faster than pathspec. #24 cpburnz/python-pa…

…thspec#38
excitoon · Aug 28, 2022 · 7b9266a · 7b9266a
1 parent 14f1214
commit 7b9266a
Show file tree

Hide file tree

Showing 2 changed files with 20 additions and 6 deletions.
diff --git a/gitignorefile/__init__.py b/gitignorefile/__init__.py
@@ -41,18 +41,22 @@ def __call__(self, path, is_dir=None):
 
             parent_gitignore = parent.join(".gitignore")
             if parent_gitignore.isfile():
-                p = parse(str(parent_gitignore), base_path=parent)
-                add_to_children[parent] = (p, plain_paths)
+                matches = parse(str(parent_gitignore), base_path=parent)
+                add_to_children[parent] = (matches, plain_paths)
                 plain_paths = []
 
             else:
                 plain_paths.append(parent)
 
         else:
             for plain_path in plain_paths:
-                self.__gitignores[plain_path] = []
+                assert plain_path.parts not in self.__gitignores
+                self.__gitignores[plain_path.parts] = []
 
-            if not add_to_children:
+            if add_to_children:
+                plain_paths.clear()
+
+            else:
                 return False
 
         for parent, (_, parent_plain_paths) in reversed(list(add_to_children.items())):
@@ -64,8 +68,13 @@ def __call__(self, path, is_dir=None):
 
             self.__gitignores[parent.parts].reverse()
             for plain_path in parent_plain_paths:
+                assert plain_path.parts not in self.__gitignores
                 self.__gitignores[plain_path.parts] = self.__gitignores[parent.parts]
 
+        for plain_path in plain_paths:
+            assert plain_path.parts not in self.__gitignores
+            self.__gitignores[plain_path.parts] = self.__gitignores[parent.parts]
+
         return any(
             (m(path, is_dir=is_dir) for m in self.__gitignores[parent.parts])
         )  # This parent comes either from first or second loop.
@@ -114,7 +123,7 @@ def isdir(self):
 
     def __str__(self):
         if self.__joined is None:
-            self.__joined = "/".join(self.__parts)
+            self.__joined = "/".join(self.__parts) if self.__parts != ("",) else "/"
         return self.__joined
 
 

diff --git a/tests/test_cache.py b/tests/test_cache.py
@@ -43,6 +43,8 @@ def __call__(self, path):
                 "/",
             ],
             [
+                "/home/vladimir/project/directory/subdirectory/subdirectory/file.txt",
+                "/home/vladimir/project/directory/subdirectory/subdirectory/file2.txt",
                 "/home/vladimir/project/directory/subdirectory/file.txt",
                 "/home/vladimir/project/directory/subdirectory/file2.txt",
                 "/home/vladimir/project/directory/.gitignore",
@@ -75,12 +77,15 @@ def mock_stat(path):
                 matches = gitignorefile.Cache()
                 self.assertTrue(matches("/home/vladimir/project/directory/subdirectory/file.txt"))
                 self.assertTrue(matches("/home/vladimir/project/directory/subdirectory/file2.txt"))
+                self.assertTrue(matches("/home/vladimir/project/directory/subdirectory/subdirectory/file.txt"))
+                self.assertTrue(matches("/home/vladimir/project/directory/subdirectory/subdirectory/file2.txt"))
+                self.assertFalse(matches("/home/vladimir/project/directory/subdirectory/subdirectory/file3.txt"))
                 self.assertTrue(matches("/home/vladimir/project/directory/file.txt"))
                 self.assertTrue(matches("/home/vladimir/project/directory/file2.txt"))
                 self.assertFalse(matches("/home/vladimir/project/file.txt"))
 
         self.assertEqual(statistics["open"], 2)
-        self.assertEqual(statistics["stat"], 6 + 5)
+        self.assertEqual(statistics["stat"], 7 + 8)
 
     def test_wrong_symlink(self):
         with tempfile.TemporaryDirectory() as d: