Run tests

Basic commands

# Run all tests with coverage
uv run pytest

# Verbose output on failure
uv run pytest --tb=short -v

# Run a specific directory
uv run pytest tests/example/note/

# Generate an HTML coverage report
uv run pytest --cov=src --cov-report=html
# → open htmlcov/index.html in your browser

Test layout

tests/
  nene2/              Framework core unit tests
    use_case/         UseCaseProtocol compliance
    auth/             Auth middleware and verifiers
    database/         TransactionManager tests
    mcp/              McpHttpClient tests
    middleware/       Each middleware in isolation
  example/
    note/             Note domain tests
      test_list_notes.py           UseCase unit tests
      test_note_repository.py      Repository contract tests
      test_async_note_use_case.py  Async UseCase tests
    comment/
      test_comment_use_case.py     UseCase unit tests (no DB)
      test_comment_repository.py   InMemory + SQLAlchemy contract tests
      test_comment_http.py         HTTP integration tests (TestClient)

Test types

UseCase unit tests

No database, no HTTP — use InMemory repositories. Fastest.

def test_create_note() -> None:
    repo = InMemoryNoteRepository()
    note = CreateNoteUseCase(repo).execute(CreateNoteInput(title="t", body="b"))
    assert note.title == "t"

Repository contract tests

@pytest.fixture(params=["inmemory", "sqlalchemy"]) runs the same assertions against both implementations.

@pytest.fixture(params=["inmemory", "sqlalchemy"])
def repo(request): ...

def test_save_and_find(repo) -> None:
    note = repo.save("title", "body")
    assert repo.find_by_id(note.id) == note

HTTP integration tests

Use FastAPI's TestClient. Tests the full stack from HTTP to repository.

def test_create_note_returns_201() -> None:
    client = TestClient(create_app(AppSettings(throttle_enabled=False)))
    response = client.post("/notes", json={"title": "t", "body": "b"})
    assert response.status_code == 201

Async tests

asyncio_mode = "auto" is set in pyproject.toml, so async def test_* works directly.

async def test_async_list_notes() -> None:
    repo = InMemoryNoteRepository()
    result = await AsyncListNotesUseCase(repo).execute(ListNotesInput(limit=10, offset=0))
    assert result.total == 0

In-memory SQLite for integration tests

When using SqlAlchemyQueryExecutor or SqlAlchemyTransactionManager with an in-memory SQLite database, always pass poolclass=StaticPool. Without it, SQLAlchemy may open a new physical connection that sees an empty database.

from sqlalchemy import create_engine
from sqlalchemy.pool import StaticPool

engine = create_engine(
    "sqlite:///:memory:",
    connect_args={"check_same_thread": False},
    poolclass=StaticPool,
)

StaticPool guarantees all logical connections share the same underlying SQLite connection, so tables created in one operation are visible to the next.

SQLite foreign-key enforcement: SQLite disables foreign-key constraints by default. Enable them with PRAGMA foreign_keys=ON right after the engine is created:

from sqlalchemy import text

with engine.begin() as conn:
    conn.execute(text("PRAGMA foreign_keys=ON"))

With StaticPool, one call applies to the single shared connection, so all subsequent operations see FK constraints enforced.

Capturing structlog output with caplog

Call configure_for_testing() at module level in conftest.py to route structlog through stdlib logging so pytest's caplog fixture can capture it.

# conftest.py
from nene2.log import configure_for_testing
configure_for_testing()

Then assert on message strings in tests:

def test_handler_logs(caplog: pytest.LogCaptureFixture) -> None:
    client = TestClient(create_app())
    client.post("/api/echo", json={"message": "hello"})
    assert any("processing echo" in r.message for r in caplog.records)

Note: caplog.records returns stdlib LogRecord objects. Fields bound with structlog.contextvars.bind_contextvars() (such as request_id) are not directly accessible as record.request_id — they appear as part of the formatted message string instead.

TestClient HTTP メソッドと json パラメーター

TestClient の .get(), .post(), .put(), .patch() は json= パラメーターを受け付けるが、 .delete() は受け付けない（TypeError）。DELETE にボディを付ける場合は .request() を使う。

# ✅ GET/POST/PUT/PATCH は json= が使える
r = client.post("/items", json={"name": "Alice"})
r = client.put("/items/1", json={"name": "Bob"})

# ❌ DELETE は json= が使えない
r = client.delete("/items/bulk", json={"ids": [1, 2]})  # TypeError

# ✅ DELETE + ボディは request() を使う
r = client.request("DELETE", "/items/bulk", json={"ids": [1, 2]})

設計上の注意: DELETE にリクエストボディを持たせることは RFC 9110 では「推奨されない」（サーバーによっては無視される）。一括削除の代替として POST /items/bulk-delete パターンも検討する。

---

Coverage requirements

Scope	Target
Overall	≥ 80% (CI enforced)
UseCase / Domain	≥ 90% (goal)

Static analysis

uv run mypy src/          # Type checking (strict)
uv run ruff check src/    # Lint
uv run ruff format --check src/ tests/  # Format check
uv run pip-audit          # Dependency vulnerability scan