TDD Mastery

Core Cycle: Red-Green-Refactor

1. Red - Write a failing test that defines the desired behavior
2. Green - Write the minimum code to make the test pass
3. Refactor - Clean up while keeping tests green

Never write production code without a failing test first. Each cycle should take 2-10 minutes.

Test Structure

Use the Arrange-Act-Assert pattern consistently:

Arrange: Set up test data and dependencies
Act:     Execute the behavior under test
Assert:  Verify the expected outcome

Name tests as test_<unit>_<scenario>_<expected_result> or it("should <behavior> when <condition>").

Jest / Vitest Patterns

describe("OrderService", () => {
  it("should apply discount when order exceeds threshold", () => {
    const order = createOrder({ items: [{ price: 150, qty: 1 }] });
    const result = applyDiscount(order, { threshold: 100, percent: 10 });
    expect(result.total).toBe(135);
  });

  it("should throw when applying discount to empty order", () => {
    const order = createOrder({ items: [] });
    expect(() => applyDiscount(order, defaultDiscount)).toThrow(EmptyOrderError);
  });
});

Use vi.fn() / jest.fn() for mocks. Prefer dependency injection over module mocking. Use beforeEach for shared setup, never share mutable state between tests.

pytest Patterns

@pytest.fixture
def db_session():
    session = create_test_session()
    yield session
    session.rollback()

def test_create_user_stores_hashed_password(db_session):
    user = UserService(db_session).create(email="a@b.com", password="secret")
    assert user.password_hash != "secret"
    assert verify_password("secret", user.password_hash)

@pytest.mark.parametrize("input,expected", [
    ("", False),
    ("short", False),
    ("ValidPass1!", True),
])
def test_password_validation(input, expected):
    assert validate_password(input) == expected

Use pytest.raises for exceptions. Use conftest.py for shared fixtures. Mark slow tests with @pytest.mark.slow.

Go Testing Patterns

func TestParseConfig(t *testing.T) {
    tests := []struct {
        name    string
        input   string
        want    Config
        wantErr bool
    }{
        {"valid yaml", "port: 8080", Config{Port: 8080}, false},
        {"empty input", "", Config{}, true},
        {"invalid port", "port: -1", Config{}, true},
    }
    for _, tt := range tests {
        t.Run(tt.name, func(t *testing.T) {
            got, err := ParseConfig([]byte(tt.input))
            if (err != nil) != tt.wantErr {
                t.Errorf("ParseConfig() error = %v, wantErr %v", err, tt.wantErr)
                return
            }
            if !tt.wantErr && got != tt.want {
                t.Errorf("ParseConfig() = %v, want %v", got, tt.want)
            }
        })
    }
}

Use table-driven tests by default. Use t.Helper() in test utility functions. Use testify/assert only if the team already uses it.

Test Levels

Level	Scope	Speed	Dependencies
Unit	Single function/class	<100ms	None (mock all)
Integration	Module boundaries	<5s	Real DB, real FS
E2E	Full user flow	<30s	Full stack

Ratio target: 70% unit, 20% integration, 10% e2e.

Coverage Rules

• Enforce 80% line coverage minimum in CI
• Track branch coverage, not just line coverage
• Exclude generated code, type definitions, and config files
• Never write tests just to hit coverage numbers; test behavior

# Jest/Vitest
vitest run --coverage --coverage.thresholds.lines=80 --coverage.thresholds.branches=75

# pytest
pytest --cov=src --cov-fail-under=80 --cov-branch

# Go
go test -coverprofile=cover.out -coverpkg=./... ./...
go tool cover -func=cover.out

Mocking Guidelines

• Mock at boundaries: HTTP clients, databases, file systems, clocks
• Never mock the unit under test
• Prefer fakes (in-memory implementations) over mocks for repositories
• Assert on behavior, not on mock call counts
• Use t.Cleanup / afterEach to reset shared mocks

Anti-Patterns to Avoid

• Testing implementation details instead of behavior
• Tests that pass when code is deleted (tautological tests)
• Shared mutable state between test cases
• Ignoring flaky tests instead of fixing them
• Testing private methods directly
• Giant test setup that obscures intent