Testing Strategy

Testing Strategy

Testing approach: unit/integration/e2e split, what to test, test structure (AAA pattern), mocking strategies, and coverage targets.

Testing Pyramid

Follow the testing pyramid -- more unit tests, fewer integration tests, even fewer e2e tests:

        /  E2E  \           ~5%   -- Critical user journeys
       /----------\
      / Integration \       ~20%  -- Module boundaries, API contracts
     /----------------\
    /    Unit Tests     \   ~75%  -- Functions, components, utilities
   /____________________\

What to Test

Always Test

• Business logic and calculations
• Data transformations and mappings
• Validation rules
• Error handling paths
• Edge cases: empty inputs, boundary values, null/undefined
• Public API of modules (exported functions)
• State transitions (reducers, state machines)
• Critical user journeys (e2e)

Don't Test

• Third-party library internals
• Simple getters/setters with no logic
• Framework boilerplate (constructor, lifecycle method existence)
• Implementation details (private methods, internal state shape)
• Constants and configuration values
• CSS styling (use visual regression tools instead)

Test Structure -- AAA Pattern

Every test should follow Arrange, Act, Assert:

describe("OrderService", () => {
  describe("calculateTotal", () => {
    it("applies percentage discount to subtotal", () => {
      // Arrange
      const items = [
        { name: "Widget", price: 25.00, quantity: 2 },
        { name: "Gadget", price: 15.00, quantity: 1 },
      ];
      const discount = { type: "percentage", value: 10 };

      // Act
      const total = calculateTotal(items, discount);

      // Assert
      expect(total).toBe(58.50); // (50 + 15) * 0.90
    });
  });
});

Naming Rules

1. Describe blocks name the unit -- describe("OrderService"), describe("calculateTotal")
2. Test names describe the behavior -- it("applies percentage discount to subtotal")
3. Use the pattern: it("<expected behavior> when <condition>")
4. Don't start with "should" -- it("returns null for invalid input") not it("should return null...")

Structure Rules

1. One assertion per test (conceptual, not literal -- multiple expect calls are fine if testing one behavior)
2. No logic in tests -- no if, for, or switch in test code
3. No test interdependencies -- each test sets up and tears down its own state
4. Use beforeEach for shared setup, not beforeAll -- isolation matters more than speed
5. Group related tests with nested describe blocks

Mocking Strategies

When to Mock

Mocking Hierarchy

Prefer lighter mocking techniques when possible:

1. Stubs -- return a fixed value: jest.fn().mockReturnValue(42)
2. Spies -- observe calls without changing behavior: jest.spyOn(service, 'save')
3. Fakes -- lightweight in-memory implementation: new InMemoryUserRepository()
4. Mocks -- full behavior replacement: jest.mock('./database')

Rules

1. Mock at the boundary -- mock the database client, not the repository method
2. Don't mock what you don't own -- wrap third-party APIs in your own adapter, mock the adapter
3. Reset mocks between tests -- use afterEach(() => jest.restoreAllMocks())
4. Verify interactions sparingly -- prefer asserting on output over asserting mock was called

// Good -- mock at the boundary
const mockDb = { query: jest.fn().mockResolvedValue([{ id: 1, name: "Alice" }]) };
const repo = new UserRepository(mockDb);
const user = await repo.findById(1);
expect(user.name).toBe("Alice");

// Bad -- mocking the unit under test
jest.spyOn(repo, "findById").mockResolvedValue({ id: 1, name: "Alice" });

Test Doubles for Common Scenarios

// Fixed time
beforeEach(() => {
  jest.useFakeTimers();
  jest.setSystemTime(new Date("2025-06-15T12:00:00Z"));
});
afterEach(() => jest.useRealTimers());

// API responses
const mockFetch = jest.fn().mockResolvedValue({
  ok: true,
  json: async () => ({ data: { id: 1 } }),
});
global.fetch = mockFetch;

// Environment variables
const originalEnv = process.env;
beforeEach(() => {
  process.env = { ...originalEnv, API_KEY: "test-key" };
});
afterEach(() => {
  process.env = originalEnv;
});

Integration Tests

1. Test module boundaries -- service calls repository, repository calls database
2. Use a real (test) database -- SQLite in-memory or Docker containers
3. Test API endpoints end-to-end -- use supertest or similar
4. Seed data in beforeEach -- don't rely on database state from other tests
5. Test error paths -- connection failures, timeouts, constraint violations

describe("POST /api/users", () => {
  it("creates a user and returns 201", async () => {
    const response = await request(app)
      .post("/api/users")
      .send({ email: "test@example.com", name: "Test User" })
      .expect(201);

    expect(response.body.data).toMatchObject({
      email: "test@example.com",
      name: "Test User",
    });
    expect(response.body.data.id).toBeDefined();
  });

  it("returns 400 for invalid email", async () => {
    const response = await request(app)
      .post("/api/users")
      .send({ email: "not-an-email", name: "Test User" })
      .expect(400);

    expect(response.body.error.code).toBe("VALIDATION_ERROR");
  });
});

E2E Tests

1. Test critical user journeys only -- sign up, purchase, core workflow
2. Use realistic data -- not "test123" or "foo bar"
3. Use data-testid attributes -- data-testid="submit-button", not CSS selectors
4. Handle async operations explicitly -- wait for elements, not fixed timeouts
5. Run in CI against a staging environment
6. Keep under 10 minutes total

Coverage Targets

Rules

1. Coverage is a floor, not a ceiling -- don't write bad tests to hit a number
2. Track coverage trends -- it should go up or stay flat, never down
3. Enforce in CI -- fail the build if coverage drops below the threshold
4. Exclude generated code -- don't count auto-generated files, configs, or type definitions

Test File Organization

src/
  services/
    order.service.ts
    order.service.test.ts        # Unit tests colocated
  api/
    routes/
      users.route.ts
tests/
  integration/
    api/
      users.test.ts              # Integration tests separate
  e2e/
    flows/
      checkout.test.ts           # E2E tests separate
  fixtures/
    users.json                   # Shared test data
  helpers/
    test-db.ts                   # Shared test utilities

Anti-patterns

• Testing implementation details -- test behavior, not how it's implemented
• Snapshot overuse -- snapshots are brittle; use them for serializable output, not UI
• Flaky tests -- fix or delete them; a flaky test is worse than no test
• Test setup duplication -- extract to helper functions or fixtures
• Ignoring test failures -- a skipped test is a known bug you're choosing to keep
• 100% coverage obsession -- diminishing returns past 85%; focus on meaningful tests