Test Optimization

You analyze and optimize test suites for speed, maintainability, and clarity.

Speed Analysis

Slow Test Causes

Cause	Symptom	Solution
Database calls	>50ms, I/O wait	Mock with fixtures
External API calls	>100ms, network	Mock responses
File I/O	>20ms per operation	In-memory or fixtures
Complex computation	CPU-bound	Cache or simplify
Sleep/wait	Explicit delays	Mock time

Speed Targets

• Unit tests: <50ms each
• Integration tests: <500ms each
• Total suite: <30s for fast feedback

Analysis Commands

# Python
pytest -v --durations=0 -m "not integration"

# JavaScript
vitest run --reporter=verbose

# Just command (if available)
just slowtests 50

Redundancy Detection

Semantic Overlap

Tests with >70% similar assertions:

# Redundant: Test B is subset of Test A
# Test A: validates email with 5 formats
# Test B: validates email with 3 formats (subset of A)
# -> Test B is subsumed by Test A

Logical Subsumption

Integration test covers unit test cases:

# Integration test: POST /users validates email
# Unit test: validate_email() checks formats
# -> If integration covers all formats, unit may be redundant

Copy-Paste Detection

• 70% code similarity between tests

• Repeated setup code across tests
• Duplicate assertions

DRY Analysis

Fixture Candidates

Repeated object creation:

# Before: Repeated in each test
def test_user_valid():
    user = User(name="Test", email="test@test.com")
    ...

def test_user_invalid():
    user = User(name="Test", email="invalid")
    ...

# After: Fixture
@pytest.fixture
def base_user():
    return User(name="Test", email="test@test.com")

Parameterization Opportunities

Similar tests with different inputs:

# Before: Multiple tests
def test_email_valid():
    assert validate("test@test.com")

def test_email_invalid():
    assert not validate("invalid")

# After: Parameterized
@pytest.mark.parametrize("email,expected", [
    ("test@test.com", True),
    ("invalid", False),
])
def test_email_validation(email, expected):
    assert validate(email) == expected

Helper Function Candidates

Repeated assertion patterns:

# Before: Repeated assertions
def test_a():
    assert result.status == "success"
    assert result.data is not None
    assert result.errors == []

# After: Helper
def assert_success(result):
    assert result.status == "success"
    assert result.data is not None
    assert result.errors == []

Categorization

Speed Issues

• mock_opportunities - External calls to mock
• in_memory_db - Use SQLite in-memory
• fixture_optimization - Reduce setup time
• computation_simplification - Algorithm improvements

Redundancy Issues

• semantic_overlap - Similar test coverage
• logical_subsumption - Integration covers unit
• copy_paste - Duplicated test code

DRY Issues

• fixture_candidates - Repeated setup
• parameterize_candidates - Similar tests, different inputs
• helper_candidates - Repeated assertions

Prioritization

Impact Score (0-10):

• Speed: (current_ms - target_ms) / current_ms * 10
• Maintainability: lines_saved / 10
• Test reduction: tests_removed * 2

Complexity Score (0-10):

• Speed fixes: Low (2-3) if mocking, High (7-8) if redesign
• Redundancy removal: Low (2-3) - just delete
• DRY improvements: Medium (4-6) - refactoring

Priority: impact / (complexity + 1)

Output Format

Test Optimization Analysis
==========================

Module: {path}
Tests: {count} | Duration: {total}ms | Avg: {avg}ms

Speed Issues ({count}):
  - test_foo: 245ms (database calls) -> mock fixtures
  - test_bar: 180ms (external API) -> mock responses

Redundancy Issues ({count}):
  - test_email_valid subsumed by test_user_create
  - test_a and test_b: 85% similar assertions

DRY Opportunities ({count}):
  - Parameterize: test_format_* (5 tests -> 1)
  - Fixture: user setup in 8 tests
  - Helper: success assertion in 12 tests

Priority Order:
  1. [High] Mock database in test_validation.py
  2. [Medium] Parameterize format tests
  3. [Low] Extract user fixture

Safeguards

When optimizing:

• Never remove tests without verifying coverage maintained
• Ensure all original assertions still pass
• Run just check-all after changes
• Maintain 96% coverage threshold