lorairo-test-generator

name	lorairo-test-generator
version	1.0.0
description	Generate pytest unit, integration, and GUI tests for LoRAIro with fixtures, mocks, 75%+ coverage, and pytest-qt for PySide6. Use when creating test suites or ensuring code quality.
metadata	{"short-description":"LoRAIro向けpytest/pytest-qtテスト生成（fixtures、mocks、カバレッジ重視）。"}
allowed-tools	["Grep","Grep","Grep","Grep","Read","Write","Bash"]

Test Generation for LoRAIro

pytest+pytest-qt test generation with fixtures, mocks, and 75%+ coverage for LoRAIro project.

When to Use

Use this skill when:

• Creating tests: After implementing new features
• Improving coverage: Increasing existing test coverage
• Regression testing: After refactoring code
• GUI testing: Implementing PySide6 widget tests

Test Categories

pytest Markers

Three test levels:

# Unit tests: Single function/class, business logic
@pytest.mark.unit
def test_calculate_score():
    assert calculate_score(10, 20) == 0.5

# Integration tests: Multiple components
@pytest.mark.integration
def test_repository_service_integration():
    service = ImageProcessingService(repository)
    result = service.process_batch(images)
    assert len(result) > 0

# GUI tests: PySide6 widgets
@pytest.mark.gui
def test_widget_interaction(qtbot):
    widget = ThumbnailWidget()
    qtbot.addWidget(widget)
    assert widget.isVisible()

Running Tests

# All tests
uv run pytest

# By category
uv run pytest -m unit
uv run pytest -m integration
uv run pytest -m gui

# With coverage
uv run pytest --cov=src --cov-report=html

Native Dependency Smoke Tests

LoRAIro shares one canonical .venv at /workspaces/LoRAIro/.venv, including when commands run from .agents/worktree/ checkouts. Treat default-sync uv run, uv sync, dependency upgrades, and torch/NVIDIA wheel repairs as shared environment mutations; do not run them concurrently across workers.

Run a torch/CUDA smoke test after dependency updates, after failures involving local annotators, or when errors mention missing native libraries such as libcudnn.so.9, libcusparseLt.so.0, libtorch_cuda.so, or triton:

uv run --no-sync python -c "import torch, torchvision; print(torch.__version__, torchvision.__version__, torch.cuda.is_available())"
find .venv/lib/python*/site-packages -name 'libcudnn.so*' -o -name 'libcusparseLt.so*'
uv pip check

Known failure mode: uv can leave NVIDIA/PyTorch packages in an inconsistent state where package metadata and RECORD say nvidia-cudnn-cu13 or nvidia-cusparselt-cu13 is installed, but the actual .so files are missing. uv pip check may not detect this because dependencies are still installed from the resolver's perspective.

Repair order:

1. Prefer non-mutating checks first (uv run --no-sync ..., find, uv pip check).
2. If only specific NVIDIA shared objects are missing, force-reinstall the narrow package(s), for example uv pip install --force-reinstall nvidia-cudnn-cu13 nvidia-cusparselt-cu13.
3. Reinstall torch / torchvision from the configured PyTorch CUDA index only when the narrow repair fails.
4. If repeated narrow repairs expose more missing native libraries, or installed metadata repeatedly disagrees with actual files, stop repairing piecemeal and tell the user that the shared .venv likely needs to be rebuilt.
5. Do not remove or recreate the shared .venv from an agent session unless the user explicitly asks for shared environment maintenance. The user is the authority on whether other sessions are still using the shared .venv.

Core Patterns

1. Unit Tests

Repository test example:

@pytest.fixture
def test_repository(test_db_engine):
    session_factory = scoped_session(sessionmaker(bind=test_db_engine))
    repo = ImageRepository(session_factory)
    yield repo
    session_factory.remove()

@pytest.mark.unit
def test_add_image(test_repository):
    """Test image addition."""
    image = Image(path="/test/image.jpg", phash="abc123")
    result = test_repository.add(image)

    assert result.id is not None
    assert result.path == "/test/image.jpg"

Service test with mocks:

@pytest.fixture
def mock_repository():
    repo = Mock(spec=ImageRepository)
    repo.get_all.return_value = [Image(id=1, path="/img1.jpg")]
    return repo

@pytest.mark.unit
def test_process_batch(mock_repository):
    service = ImageProcessingService(mock_repository)
    result = service.process_batch(["/img1.jpg"])
    assert len(result) == 1

2. Integration Tests

Full workflow test:

@pytest.mark.integration
def test_full_workflow(test_repository):
    """Test complete workflow."""
    # Arrange
    images = [Image(path=f"/img{i}.jpg") for i in range(5)]

    # Act & Assert
    added = test_repository.batch_add(images)
    assert len(added) == 5

    results = test_repository.search(SearchCriteria(min_score=0.5))
    assert isinstance(results, list)

3. GUI Tests (pytest-qt)

Widget test:

@pytest.fixture
def thumbnail_widget(qtbot):
    widget = ThumbnailWidget()
    qtbot.addWidget(widget)
    return widget

@pytest.mark.gui
def test_signal_emission(qtbot, thumbnail_widget):
    """Test signal emission."""
    with qtbot.waitSignal(thumbnail_widget.image_selected, timeout=1000) as blocker:
        thumbnail_widget.select_image(0)

    assert blocker.args[0] == "/path/to/image.jpg"

@pytest.mark.gui
def test_button_click(qtbot, thumbnail_widget):
    """Test button interaction."""
    qtbot.mouseClick(thumbnail_widget._ui.loadButton, Qt.LeftButton)
    assert thumbnail_widget._images_loaded is True

4. Fixtures

Common fixtures:

@pytest.fixture(scope="session")
def test_data_dir():
    return Path(__file__).parent / "resources"

@pytest.fixture
def sample_image(test_data_dir):
    return test_data_dir / "sample.jpg"

@pytest.fixture(params=[1, 5, 10])
def batch_size(request):
    """Parameterized fixture."""
    return request.param

Best Practices

DO:

• Use AAA pattern (Arrange, Act, Assert)
• Single assertion per test
• Use fixtures for setup/teardown
• Apply appropriate pytest markers
• Maintain 75%+ code coverage

DON'T:

• Create dependencies between tests
• Call external APIs (use mocks)
• Hardcode paths (use tests/resources/)
• Use print statements (use logger or assert messages)
• Write slow unit tests (keep under 1 second)

Coverage Requirements

# Generate coverage report
uv run pytest --cov=src --cov-report=html

# View in browser: htmlcov/index.html

# Requirements:
# - Minimum: 75%
# - Target: Unit 90%+, Integration 80%+, GUI 70%+

Project Structure

tests/
├── conftest.py              # Shared fixtures
├── resources/               # Test data
│   ├── sample.jpg
│   └── test_config.toml
├── database/                # Database tests
│   └── test_db_repository.py
├── services/                # Service tests
│   └── test_image_processing_service.py
└── gui/                     # GUI tests
    ├── widgets/
    │   └── test_thumbnail_widget.py
    └── window/
        └── test_main_window.py

Test Sync (diff-driven, 旧 /test sync)

コード変更後に、テストの追加・修正・削除を差分から判定して同期する。新規実装直後だけでなく、 リファクタやシグネチャ変更の後にも使う。

手順

1. 変更検出: 変更されたソースを特定し、種別を分類する。

   git diff --name-status HEAD~1..HEAD -- 'src/**/*.py'
   

   • A (Added) → 対応テストを追加
   • M (Modified) → シグネチャ/振る舞い変更ならテストを修正（内部実装のみなら実行で確認、変更不要のことが多い）
   • D (Deleted) → 対応テストを削除（削除前にユーザー確認）
2. 影響範囲の特定: investigation agent または Grep で変更シンボルの参照元を追い、波及するテストを洗う。
3. テストファイル対応表:
   • src/lorairo/services/foo.py → tests/unit/services/test_foo.py
   • src/lorairo/gui/widgets/bar.py → tests/unit/gui/widgets/test_bar.py
4. 同期アクション実行: 追加は本 skill の Core Patterns に従い生成。修正は変更 API に合わせ更新。削除は確認後に実施。
5. 検証: 新規/修正テストと回帰テストを uv run pytest で確認（CI-equivalent filter は .claude/rules/testing.md）。

クイック品質チェック（Ruff/mypy/pytest）は make format / make mypy / uv run pytest で直接実行する（専用コマンドは不要）。 エラー診断は superpowers systematic-debugging + build-error-resolver agent に委ねる。

Before Writing Tests

類似のテストパターン・既存 fixture を確認する（tests/conftest.py、近接する test_*.py）。再利用価値のあるテスト戦略・モック方針は docs/lessons-learned.md に記録する。

Examples

See examples.md for detailed test implementation scenarios.

Reference

See reference.md for complete pytest and pytest-qt API reference.