Print number of excludes

.github/workflows/check.yaml

@@ -9,7 +9,6 @@ on:
 jobs:
   check:
     runs-on: ubuntu-latest
-
     name: test with ${{ matrix.py }}
     strategy:
       fail-fast: false
@@ -31,3 +30,17 @@ jobs:
         run: python -m pip install -r .devcontainer/requirements.txt
       - name: Run test suite
         run: tox -e py
+
+  show-excluded-files-count:
+    runs-on: ubuntu-latest
+    name: show excluded files count
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4
+      - name: Setup python
+        uses: actions/setup-python@a26af69be951a213d495a4c3e4e4022e16d87065 # v5
+        with:
+          python-version: "3.13"
+      - name: Install pre-commit
+        run: python -m pip install pre-commit
+      - name: Run pre-commit to show excluded files
+        run: pre-commit run check-non-existing-and-duplicate-excludes --verbose

.pre-commit-config.yaml

@@ -14,6 +14,13 @@ repos:
       - id: check-useless-excludes
   - repo: local
     hooks:
+      - id: check-non-existing-and-duplicate-excludes
+        name: Check non-existing and duplicate excludes in pre-commit-config
+        entry: dev_tools/check_useless_exclude_paths_hooks.py
+        pass_filenames: false
+        always_run: true
+        language: python
+        additional_dependencies: ["pre-commit >= 3.5.0"]
       - id: generate-hook-docs
         name: generate hook docs
         description: Generate markdown documentation for pre-commit hooks in README.md

.pre-commit-hooks.yaml

@@ -103,7 +103,7 @@
     - svg
 - id: check-non-existing-and-duplicate-excludes
   name: Check non-existing and duplicate excludes in pre-commit-config
-  description: "Check for non existing and duplicate paths in `.pre-commit-config.yaml`. Background: In a big codebase, the exclude lists can be quite long and it's easy to make a typo or forget to remove an entry when it's no longer needed."
+  description: "Check for non existing and duplicate paths in `.pre-commit-config.yaml`. Background: In a big codebase, the exclude lists can be quite long and it's easy to make a typo or forget to remove an entry when it's no longer needed. If you run this hook with `--verbose` it will also print the number of excluded files for each hook."
   entry: check-useless-exclude-paths-hooks
   pass_filenames: false
   always_run: true

README.md

@@ -128,7 +128,7 @@ Check that all TODO comments follow the same pattern and link a Jira ticket: `TO
 
 ### `check-non-existing-and-duplicate-excludes`
 
-Check for non existing and duplicate paths in `.pre-commit-config.yaml`. Background: In a big codebase, the exclude lists can be quite long and it's easy to make a typo or forget to remove an entry when it's no longer needed.
+Check for non existing and duplicate paths in `.pre-commit-config.yaml`. Background: In a big codebase, the exclude lists can be quite long and it's easy to make a typo or forget to remove an entry when it's no longer needed. If you run this hook with `--verbose` it will also print the number of excluded files for each hook.
 
 ### `sync-vscode-config`
 

dev_tools/check_useless_exclude_paths_hooks.py

@@ -1,3 +1,5 @@
+#!/usr/bin/env python
+
 # Copyright (c) Luminar Technologies, Inc. All rights reserved.
 # Licensed under the MIT License.
 
@@ -58,6 +60,15 @@ def has_duplicates(self) -> bool:
     def has_non_existing_paths(self) -> bool:
         return bool(self.find_non_existing_paths())
 
+    def count_excluded_files(self) -> int:
+        existing_paths = [path for path in self.exclude_paths if path.exists()]
+        total_file_count = sum(1 for path in existing_paths if path.is_file())
+        excluded_dirs = [path for path in existing_paths if path.is_dir()]
+        total_dir_count = sum(
+            sum(1 for file in excluded_dir.rglob("*") if file.is_file()) for excluded_dir in excluded_dirs
+        )
+        return total_file_count + total_dir_count
+
 
 def is_regex_pattern(exclude: str) -> bool:
     return any(regex_key in exclude for regex_key in ["*", "$", "^"])
@@ -101,10 +112,22 @@ def have_non_existent_paths_or_duplicates(hooks_list: list[Any]) -> bool:
     return bool(non_existing_paths or duplicates)
 
 
+def print_excluded_files_count(hooks_list: list[Any]) -> None:
+    total_excluded = sum(hook.count_excluded_files() for hook in hooks_list)
+    print(f"Total number of excluded files: {total_excluded}")
+
+    for hook in hooks_list:
+        excluded_count = hook.count_excluded_files()
+        if excluded_count > 0:
+            print(f"{excluded_count} files excluded from hook {hook.id}")
+
+
 def main() -> int:
     repo_root = Path.cwd()
     pre_commit_config = repo_root / CONFIG_FILE
-    return 1 if have_non_existent_paths_or_duplicates(load_hooks(repo_root, pre_commit_config)) else 0
+    hooks_list = load_hooks(repo_root, pre_commit_config)
+    print_excluded_files_count(hooks_list)
+    return 1 if have_non_existent_paths_or_duplicates(hooks_list) else 0
 
 
 if __name__ == "__main__":

pyproject.toml

@@ -83,6 +83,11 @@ ignore = [
   "T201"
 ]
 
+[tool.ruff.lint.per-file-ignores]
+"tests/**" = [
+  "PLR2004" # Magic values in tests are ok since they are often the expected values
+]
+
 [tool.vulture]
 min_confidence = 100
 ignore_decorators = ["@pytest.fixture"]

tests/test_check_useless_exclude_paths_hooks.py

@@ -202,3 +202,63 @@ def test_have_non_existent_paths_or_duplicates_for_duplicate_paths(
     assert "duplicates" in output
     assert "existing_path1" in output
     assert "existing_path2" in output
+
+
+def test_count_excluded_files_for_single_file(fs: FakeFilesystem) -> None:
+    fs.create_file(Path("Repo/test_file.txt"))
+    hook_instance = Hook("test_id", [Path("Repo/test_file.txt")])
+
+    assert hook_instance.count_excluded_files() == 1
+
+
+def test_count_excluded_files_for_multiple_files(fs: FakeFilesystem) -> None:
+    fs.create_file(Path("Repo/file1.txt"))
+    fs.create_file(Path("Repo/file2.txt"))
+    fs.create_file(Path("Repo/file3.txt"))
+    hook_instance = Hook("test_id", [Path("Repo/file1.txt"), Path("Repo/file2.txt"), Path("Repo/file3.txt")])
+
+    assert hook_instance.count_excluded_files() == 3
+
+
+def test_count_excluded_files_for_directory(fs: FakeFilesystem) -> None:
+    fs.create_dir(Path("Repo/test_dir"))
+    fs.create_file(Path("Repo/test_dir/file1.txt"))
+    fs.create_file(Path("Repo/test_dir/file2.txt"))
+    fs.create_file(Path("Repo/test_dir/subdir/file3.txt"))
+    hook_instance = Hook("test_id", [Path("Repo/test_dir")])
+
+    assert hook_instance.count_excluded_files() == 3
+
+
+def test_count_excluded_files_for_mixed_paths(fs: FakeFilesystem) -> None:
+    fs.create_file(Path("Repo/single_file.txt"))
+    fs.create_dir(Path("Repo/test_dir"))
+    fs.create_file(Path("Repo/test_dir/file1.txt"))
+    fs.create_file(Path("Repo/test_dir/file2.txt"))
+    hook_instance = Hook("test_id", [Path("Repo/single_file.txt"), Path("Repo/test_dir")])
+
+    assert hook_instance.count_excluded_files() == 3
+
+
+def test_count_excluded_files_for_non_existing_paths() -> None:
+    hook_instance = Hook("test_id", [Path("Repo/non_existing_file.txt"), Path("Repo/non_existing_dir")])
+
+    assert hook_instance.count_excluded_files() == 0
+
+
+def test_count_excluded_files_for_mixed_existing_and_non_existing_paths(fs: FakeFilesystem) -> None:
+    fs.create_file(Path("Repo/existing_file.txt"))
+    fs.create_dir(Path("Repo/existing_dir"))
+    fs.create_file(Path("Repo/existing_dir/file.txt"))
+    hook_instance = Hook(
+        "test_id",
+        [Path("Repo/existing_file.txt"), Path("Repo/existing_dir"), Path("Repo/non_existing_file.txt")],
+    )
+
+    assert hook_instance.count_excluded_files() == 2
+
+
+def test_count_excluded_files_for_empty_exclude_paths() -> None:
+    hook_instance = Hook("test_id", [])
+
+    assert hook_instance.count_excluded_files() == 0