generating-test-doubles

自动生成各类测试替身以实现依赖隔离,包括模拟对象、桩对象、间谍对象和伪对象,支持在单元测试和集成测试中快速构建独立可控的测试环境。

快捷安装

在终端运行此命令,即可一键安装该 Skill 到您的 Claude 中

npx skills add jeremylongshore/claude-code-plugins-plus-skills --skill "generating-test-doubles"

Test Doubles Generator

Overview

Generate mocks, stubs, spies, and fakes to isolate units under test from external dependencies. Supports Jest mocks, Sinon.js stubs, Python unittest.mock, Go interfaces, and testdouble.js patterns.

Prerequisites

  • Testing framework installed (Jest, Vitest, Mocha, pytest, JUnit 5, or Go testing)
  • Mocking library available (Sinon.js, testdouble.js, unittest.mock, Mockito, or gomock)
  • TypeScript strict mode enabled for type-safe mocks (if applicable)
  • Source code with clear interface or class boundaries for dependency injection

Instructions

  1. Scan the codebase with Glob and Grep to identify modules with external dependencies (database clients, HTTP clients, file system access, third-party SDKs).
  2. Read each module under test and catalog its dependency interfaces — list every method signature, return type, and side effect.
  3. Determine the appropriate test double type for each dependency:
    • Stub: Returns canned data, no behavior verification (use for database queries returning fixed datasets).
    • Mock: Verifies interactions — call count, argument matching, call order (use for email senders, event emitters).
    • Spy: Wraps real implementation while recording calls (use when partial behavior is needed).
    • Fake: Lightweight working implementation (use for in-memory repositories replacing real databases).
  4. Generate test double files following the project’s existing test directory structure (e.g., __mocks__/, test/doubles/, testutil/).
  5. For each test double, implement:
    • Factory function or class matching the dependency interface exactly.
    • Configurable return values via builder pattern or method chaining.
    • Call recording for assertion (arguments, call count, call order).
    • Reset/restore mechanism for cleanup between tests.
  6. Wire test doubles into existing test files using the framework’s dependency injection pattern (jest.mock(), @patch, constructor injection, or Go interface substitution).
  7. Validate all test doubles compile and pass type checks by running tsc --noEmit or equivalent.

Output

  • Test double source files (one per dependency) placed in the project’s mock directory
  • Factory functions with TypeScript generics or equivalent type safety
  • Jest __mocks__ auto-mock modules where applicable
  • Updated test files wired to use generated doubles instead of real dependencies
  • Summary listing each double, its type (mock/stub/spy/fake), and the interface it replaces

Error Handling

ErrorCauseSolution
TypeError: X is not a functionMock missing a method from the real interfaceRegenerate the double from the current interface definition; add the missing method
Mock leaking between testsShared mock state not reset in afterEachAdd jest.restoreAllMocks() or sinon.restore() in teardown hooks
Type mismatch on mock return valueReturn type does not match interface contractUse as ReturnType<typeof fn> or update the mock factory to return the correct type
Spy not recording callsSpy created after the function was already boundCreate spies before the module under test imports the dependency
Over-mocking hides real bugsToo many layers replaced with fakesLimit mocks to true external boundaries (I/O, network); let pure logic run unmocked

Examples

Jest mock factory for a UserRepository:

// __mocks__/userRepository.ts
export const createMockUserRepo = () => ({
  findById: jest.fn().mockResolvedValue({ id: '1', name: 'Test User' }),
  save: jest.fn().mockResolvedValue(undefined),
  delete: jest.fn().mockResolvedValue(true),
});

Python unittest.mock patch for an HTTP client:

from unittest.mock import patch, MagicMock

@patch('myapp.client.requests.get')
def test_fetch_data(mock_get):
    mock_get.return_value = MagicMock(status_code=200, json=lambda: {"key": "value"})  # HTTP 200 OK
    result = fetch_data("https://api.example.com/data")
    assert result == {"key": "value"}
    mock_get.assert_called_once()

Go interface-based fake:

type FakeStore struct {
    data map[string]string
}
func (f *FakeStore) Get(key string) (string, error) {
    v, ok := f.data[key]
    if !ok { return "", ErrNotFound }
    return v, nil
}

Resources