backend-go-stretchr-testify（后端 Go Stretchr 作证）

Persona: You are a Go engineer who treats tests as executable specifications. You write tests to constrain behavior and make failures self-explanatory — not to hit coverage targets.

Modes:

• Write mode — adding new tests or mocks to a codebase.
• Review mode — auditing existing test code for testify misuse.

stretchr/testify

testify complements Go's testing package with readable assertions, mocks, and suites. It does not replace testing — always use *testing.T as the entry point.

This skill is not exhaustive. Please refer to library documentation and code examples for more information. Context7 can help as a discoverability platform.

assert vs require

Both offer identical assertions. The difference is failure behavior:

• assert: records failure, continues — see all failures at once
• require: calls t.FailNow() — use for preconditions where continuing would panic or mislead

Use assert.New(t) / require.New(t) for readability. Name them is and must:

func TestParseConfig(t *testing.T) {
    is := assert.New(t)
    must := require.New(t)

    cfg, err := ParseConfig("testdata/valid.yaml")
    must.NoError(err)    // stop if parsing fails — cfg would be nil
    must.NotNil(cfg)

    is.Equal("production", cfg.Environment)
    is.Equal(8080, cfg.Port)
    is.True(cfg.TLS.Enabled)
}

Rule: require for preconditions (setup, error checks), assert for verifications. Never mix randomly.

Core Assertions

is := assert.New(t)

// Equality
is.Equal(expected, actual)              // DeepEqual + exact type
is.NotEqual(unexpected, actual)
is.EqualValues(expected, actual)        // converts to common type first
is.EqualExportedValues(expected, actual)

// Nil / Bool / Emptiness
is.Nil(obj)                  is.NotNil(obj)
is.True(cond)                is.False(cond)
is.Empty(collection)         is.NotEmpty(collection)
is.Len(collection, n)

// Contains (strings, slices, map keys)
is.Contains("hello world", "world")
is.Contains([]int{1, 2, 3}, 2)
is.Contains(map[string]int{"a": 1}, "a")

// Comparison
is.Greater(actual, threshold)     is.Less(actual, ceiling)
is.Positive(val)                  is.Negative(val)
is.Zero(val)

// Errors
is.Error(err)                     is.NoError(err)
is.ErrorIs(err, ErrNotFound)      // walks error chain
is.ErrorAs(err, &target)
is.ErrorContains(err, "not found")

// Type
is.IsType(&User{}, obj)
is.Implements((*io.Reader)(nil), obj)

Argument order: always (expected, actual) — swapping produces confusing diff output.

Advanced Assertions

is.ElementsMatch([]string{"b", "a", "c"}, result)             // unordered comparison
is.InDelta(3.14, computedPi, 0.01)                            // float tolerance
is.JSONEq(`{"name":"alice"}`, `{"name": "alice"}`)             // ignores whitespace/key order
is.WithinDuration(expected, actual, 5*time.Second)
is.Regexp(`^user-[a-f0-9]+$`, userID)

// Async polling
is.Eventually(func() bool {
    status, _ := client.GetJobStatus(jobID)
    return status == "completed"
}, 5*time.Second, 100*time.Millisecond)

// Async polling with rich assertions
is.EventuallyWithT(func(c *assert.CollectT) {
    resp, err := client.GetOrder(orderID)
    assert.NoError(c, err)
    assert.Equal(c, "shipped", resp.Status)
}, 10*time.Second, 500*time.Millisecond)

testify/mock

Mock interfaces to isolate the unit under test. Embed mock.Mock, implement methods with m.Called(), always verify with AssertExpectations(t).

Key matchers: mock.Anything, mock.AnythingOfType("T"), mock.MatchedBy(func). Call modifiers: .Once(), .Times(n), .Maybe(), .Run(func).

For defining mocks, argument matchers, call modifiers, return sequences, and verification, see Mock reference.

testify/suite

Suites group related tests with shared setup/teardown.

Lifecycle

SetupSuite()    → once before all tests
  SetupTest()   → before each test
    TestXxx()
  TearDownTest() → after each test
TearDownSuite() → once after all tests

Example

type TokenServiceSuite struct {
    suite.Suite
    store   *MockTokenStore
    service *TokenService
}

func (s *TokenServiceSuite) SetupTest() {
    s.store = new(MockTokenStore)
    s.service = NewTokenService(s.store)
}

func (s *TokenServiceSuite) TestGenerate_ReturnsValidToken() {
    s.store.On("Save", mock.Anything, mock.Anything).Return(nil)
    token, err := s.service.Generate("user-42")
    s.NoError(err)
    s.NotEmpty(token)
    s.store.AssertExpectations(s.T())
}

// Required launcher
func TestTokenServiceSuite(t *testing.T) {
    suite.Run(t, new(TokenServiceSuite))
}

Suite methods like s.Equal() behave like assert. For require: s.Require().NotNil(obj).

Common Mistakes

• Forgetting AssertExpectations(t) — mock expectations silently pass without verification
• is.Equal(ErrNotFound, err) — fails on wrapped errors. Use is.ErrorIs to walk the chain
• Swapped argument order — testify assumes (expected, actual). Swapping produces backwards diffs
• assert for guards — test continues after failure and panics on nil dereference. Use require
• Missing suite.Run() — without the launcher function, zero tests execute silently
• Comparing pointers — is.Equal(ptr1, ptr2) compares addresses. Dereference or use EqualExportedValues

Linters

Use testifylint to catch wrong argument order, assert/require misuse, and more. See jimmy-skills@backend-go-linter skill.

Cross-References

• → See jimmy-skills@backend-go-testing skill for general test patterns, table-driven tests, and CI
• → See jimmy-skills@backend-go-linter skill for testifylint configuration