2nd/10_Wiki/Topics/Architecture/Mock Objects (가짜 객체).md

id, title, category, status, canonical_id, aliases, duplicate_of, source_trust_level, confidence_score, verification_status, tags, raw_sources, last_reinforced, github_commit, tech_stack

id	title	category	status	canonical_id	aliases	duplicate_of	source_trust_level	confidence_score	verification_status	tags	raw_sources	last_reinforced	github_commit	tech_stack
wiki-2026-0508-mock-objects-가짜-객체	Mock Objects (가짜 객체)	10_Wiki/Topics	verified	self	Mock Mocking Stubbing Test Doubles 테스트 대역	none	A	0.9	applied	testing tdd unit-test mock		2026-05-10	pending	language framework multi jest-pytest-mockito

Mock Objects (가짜 객체)

매 한 줄

"매 real collaborator 의 자리에 매 control 가능한 가짜를 끼워넣음.". Mock object 는 매 unit test 에서 매 dependency 를 isolate 하기 위한 매 test double. Gerard Meszaros 의 매 분류 (Dummy / Stub / Spy / Mock / Fake) 가 표준 — 매 잘못 쓰면 매 test 가 매 implementation 에 coupling 된다.

매 핵심

매 Test Double 5종 (Meszaros)

• Dummy: 매 인자 자리만 채움, 매 호출 안됨.
• Stub: 매 hardcoded response 반환 (state verification).
• Spy: 매 호출 record (verify 가능한 stub).
• Mock: 매 expected 호출 정의, 매 위반 시 fail (behavior verification).
• Fake: 매 실제 working 구현 (in-memory DB 등).

매 언제 mock

• 매 외부 service (network, DB, FS).
• 매 nondeterministic (time, random, env).
• 매 expensive (slow, paid API).
• 매 hard to trigger (error case, race condition).

매 언제 mock 하지 말 것

• 매 own code 의 internal collaborator — 매 over-mocking → 매 fragile test.
• 매 third-party library 의 직접 mock — 매 own wrapper 만들고 매 그걸 mock.
• 매 value object — 매 그냥 real 사용.

매 London vs Detroit (TDD school)

• London (mockist): 매 collaborator 다 mock, 매 interaction verify.
• Detroit (classical): 매 real object 사용, 매 state verify, 매 boundary 만 mock.
• 매 2026 trend: 매 Detroit 우세 (mockist 의 fragile test 비판) — 매 boundary 에서만 mock.

💻 패턴

Jest (JS) — module mock

// userService.test.js
import { fetchUser } from './userService';
import * as api from './api';

jest.mock('./api');

test('fetchUser returns user data', async () => {
  api.get.mockResolvedValue({ id: 1, name: 'Alice' });

  const user = await fetchUser(1);

  expect(user.name).toBe('Alice');
  expect(api.get).toHaveBeenCalledWith('/users/1');
  expect(api.get).toHaveBeenCalledTimes(1);
});

Jest — spy on real

import { logger } from './logger';
import { processOrder } from './orders';

test('logs order processing', async () => {
  const spy = jest.spyOn(logger, 'info');
  await processOrder({ id: 'o1' });
  expect(spy).toHaveBeenCalledWith('processing', { id: 'o1' });
  spy.mockRestore();
});

pytest — monkeypatch

def test_fetch_user(monkeypatch):
    def fake_get(url):
        return {"id": 1, "name": "Alice"}

    import api
    monkeypatch.setattr(api, "get", fake_get)

    from user_service import fetch_user
    user = fetch_user(1)
    assert user["name"] == "Alice"

pytest — unittest.mock

from unittest.mock import patch, MagicMock

@patch('user_service.api.get')
def test_fetch_user(mock_get):
    mock_get.return_value = {"id": 1, "name": "Alice"}

    from user_service import fetch_user
    user = fetch_user(1)

    mock_get.assert_called_once_with('/users/1')
    assert user["name"] == "Alice"

Mockito (Java)

import static org.mockito.Mockito.*;

@Test
void fetchUser_returnsUser() {
    ApiClient api = mock(ApiClient.class);
    when(api.get("/users/1")).thenReturn(new User(1, "Alice"));

    UserService svc = new UserService(api);
    User u = svc.fetchUser(1);

    assertEquals("Alice", u.getName());
    verify(api).get("/users/1");
    verify(api, times(1)).get(anyString());
}

Fake (in-memory DB)

// Fake user repo — real working impl, no DB
class InMemoryUserRepo implements UserRepo {
  private users = new Map<string, User>();
  async save(u: User) { this.users.set(u.id, u); }
  async find(id: string) { return this.users.get(id) ?? null; }
}

test('UserService.create persists user', async () => {
  const repo = new InMemoryUserRepo();
  const svc = new UserService(repo);
  await svc.create({ name: 'Alice' });
  const all = await repo.find('1');
  expect(all?.name).toBe('Alice');
});

MSW (Mock Service Worker) — HTTP-level mock

import { setupServer } from 'msw/node';
import { http, HttpResponse } from 'msw';

const server = setupServer(
  http.get('/api/users/:id', ({ params }) =>
    HttpResponse.json({ id: params.id, name: 'Alice' })
  )
);

beforeAll(() => server.listen());
afterEach(() => server.resetHandlers());
afterAll(() => server.close());

test('component shows user', async () => {
  render(<UserCard id="1" />);
  expect(await screen.findByText('Alice')).toBeInTheDocument();
});

Time mocking

// Jest fake timers
jest.useFakeTimers();
jest.setSystemTime(new Date('2026-05-10T00:00:00Z'));

const result = formatRelative(new Date('2026-05-10T00:00:30Z'));
expect(result).toBe('30 seconds ago');

jest.useRealTimers();

매 결정 기준

상황	Double type
외부 HTTP API	MSW (network-level) > module mock
DB	Fake (in-memory) > Mock
시간 / random	jest.useFakeTimers, freeze_time
로깅 / metrics	Spy (verify 호출)
단순 인자 채우기	Dummy
Behavior 검증 (rare)	Mock
State 검증 (default)	Stub / Fake

기본값: 매 boundary 만 mock, 매 internal 은 real. 매 state verify > behavior verify.

🔗 Graph

• 부모: Unit Testing · TDD
• 변형: Stub · Spy · Fake · Mockito · MSW
• 응용: Test Pyramid · Contract Testing
• Adjacent: Dependency Injection · Hexagonal Architecture · Mocking_Framework

🤖 LLM 활용

언제: unit test 작성 시 매 외부 dependency isolation, 매 deterministic test, 매 fast feedback. 언제 X: integration test — 매 real component 가 매 정답. 매 LLM 호출은 매 record/replay (VCR) 가 더 적합.

❌ 안티패턴

• Over-mocking: 매 internal 까지 mock → 매 refactor 시 매 test 다 깨짐.
• Mock 의 Mock: 매 mock 이 mock 을 반환 → 매 test 가 매 reality 와 무관.
• Implementation coupling: 매 verify(api).callMethod(exact_args) 의 남발 → 매 fragile.
• Real time / random: 매 flaky test 의 가장 흔한 원인.
• No teardown: 매 mock 이 매 다음 test 로 leak — 매 reset 필수.

🧪 검증 / 중복

• Verified (Meszaros "xUnit Test Patterns", Martin Fowler "Mocks Aren't Stubs", Jest/pytest/Mockito docs, MSW docs).
• 신뢰도 A.
• 매 Mocking and Stubbing (테스트 대역) 는 매 alias redirect.

🕓 Changelog

날짜	변경
2026-05-08	Phase 1
2026-05-10	Manual cleanup — Meszaros 5종 + Jest/pytest/Mockito/MSW 패턴