Python Test Updater

Python Test Updater

Update Python tests to work correctly with new code versions.

Workflow

1. Understand the Inputs

Gather required inputs:

• Old version of Python code (file or content)
• New version of Python code (file or content)
• Old test module or test functions (file or content)

Verify inputs:

• Code files are valid Python
• Test files use pytest or unittest
• Files are readable

2. Analyze Code Changes

Automated analysis:

python scripts/analyze_code_diff.py <old_file> <new_file>

Manual analysis:

• Read both old and new code versions
• Identify what changed
• Understand the nature of changes

Identify change types:

Function signature changes:

• Added parameters
• Removed parameters
• Renamed parameters
• Changed parameter types
• Changed default values

Return value changes:

• Changed return type
• Changed return structure
• Added return fields
• Removed return fields

Behavior changes:

• Modified logic
• Changed validation rules
• Updated error handling
• Altered workflows

Class changes:

• Added/removed methods
• Modified constructor
• Changed inheritance
• Updated attributes

Async changes:

• Sync to async conversion
• Async to sync conversion

See test-update-patterns.md for detailed patterns.

3. Analyze Test Structure

Read the old tests:

• Identify test functions
• Understand test setup (fixtures, mocks)
• Note test assertions
• Check test organization

Identify test components:

• Test functions/methods
• Fixtures
• Parametrized tests
• Mocked dependencies
• Assertions

Map tests to code:

• Which tests cover which functions/classes
• What behavior each test verifies
• What assertions check what conditions

4. Determine Required Updates

For each code change, identify test impact:

Signature changes → Update function calls

# Old code: function(arg1, arg2)
# New code: function(arg1, arg2, arg3=default)

# Old test
result = function(value1, value2)

# Updated test
result = function(value1, value2)  # Works with default
# OR
result = function(value1, value2, value3)  # Explicit value

Return value changes → Update assertions

# Old code: return value
# New code: return {"result": value, "status": "ok"}

# Old test
assert result == expected_value

# Updated test
assert result["result"] == expected_value
assert result["status"] == "ok"

Behavior changes → Update expected values

# Old code: validates length >= 6
# New code: validates length >= 8 and has digit

# Old test
assert validate("abc123") == True

# Updated test
assert validate("abc12345") == True  # Updated
assert validate("abc123") == False   # Now fails

New functionality → Add new tests

# New code: added get_display_name() method

# Add new test
def test_get_display_name():
    obj = MyClass("value")
    assert obj.get_display_name() == "Value: value"

5. Update Test Code

Apply updates systematically:

Step 1: Update imports if needed

# If new exceptions or classes added
from module import NewException, NewClass

Step 2: Update function/method calls

• Add new required parameters
• Remove obsolete parameters
• Rename parameters if changed
• Update keyword arguments

Step 3: Update assertions

• Change expected values if behavior changed
• Update assertion structure if return type changed
• Add new assertions for new fields
• Remove assertions for removed fields

Step 4: Update exception handling

# Old
with pytest.raises(OldException):
    function()

# New
with pytest.raises(NewException):
    function()

Step 5: Update async/await if needed

# Old
def test_function():
    result = function()

# New (if function became async)
@pytest.mark.asyncio
async def test_function():
    result = await function()

Step 6: Add new test cases

• Test new functionality
• Test new parameters
• Test new behavior
• Test edge cases

6. Run Tests

Execute the updated tests:

# Run all tests
pytest test_file.py

# Run specific test
pytest test_file.py::test_function

# Run with verbose output
pytest -v test_file.py

Check results:

• All tests should pass
• No import errors
• No syntax errors
• No assertion failures

7. Fix Remaining Failures

If tests still fail:

Analyze error messages:

• Read the error carefully
• Identify what's failing
• Understand why it's failing

Common failure types:

1. AssertionError

AssertionError: assert 10 == 15

→ Expected value changed, update assertion

2. TypeError

TypeError: function() missing 1 required positional argument: 'new_param'

→ Add missing parameter to function call

3. AttributeError

AttributeError: 'dict' object has no attribute 'field'

→ Return type changed, update how result is accessed

4. ImportError

ImportError: cannot import name 'OldClass'

→ Class renamed or removed, update import

Fix each failure:

• Identify the root cause
• Apply appropriate fix
• Re-run tests
• Verify fix works

8. Verify and Refine

Final verification:

• All tests pass
• Test coverage maintained
• Test intent preserved
• Code quality good

Refine if needed:

• Improve test names
• Add docstrings
• Use fixtures for common setup
• Parametrize similar tests

Example refinement:

# Before
def test_function_case1():
    assert function(5) == 10

def test_function_case2():
    assert function(10) == 20

# After (parametrized)
@pytest.mark.parametrize("input,expected", [
    (5, 10),
    (10, 20),
])
def test_function(input, expected):
    assert function(input) == expected

Common Update Patterns

Pattern 1: Add Parameter with Default

Code change:

# Old
def function(a, b):
    return a + b

# New
def function(a, b, c=0):
    return a + b + c

Test update:

# Old test (still works)
def test_function():
    assert function(1, 2) == 3

# Add new test for new parameter
def test_function_with_c():
    assert function(1, 2, 3) == 6

Pattern 2: Change Return Type

Code change:

# Old
def get_data():
    return [1, 2, 3]

# New
def get_data():
    return {"data": [1, 2, 3], "count": 3}

Test update:

# Old
def test_get_data():
    data = get_data()
    assert len(data) == 3

# New
def test_get_data():
    result = get_data()
    assert len(result["data"]) == 3
    assert result["count"] == 3

Pattern 3: Change Validation Logic

Code change:

# Old
def validate(value):
    return len(value) >= 6

# New
def validate(value):
    return len(value) >= 8 and any(c.isdigit() for c in value)

Test update:

# Old
def test_validate():
    assert validate("abc123") == True
    assert validate("abc") == False

# New
def test_validate():
    assert validate("abc12345") == True  # Updated
    assert validate("abc123") == False   # Now fails validation
    assert validate("abcdefgh") == False # No digit
    assert validate("abc") == False

Pattern 4: Sync to Async

Code change:

# Old
def fetch():
    return data

# New
async def fetch():
    return await async_data()

Test update:

# Old
def test_fetch():
    result = fetch()
    assert result is not None

# New
@pytest.mark.asyncio
async def test_fetch():
    result = await fetch()
    assert result is not None

Best Practices

Preserve Test Intent

• Keep testing the same functionality
• Don't change what's being verified
• Only update how it's tested

Maintain Coverage

• Don't remove tests unless functionality removed
• Add tests for new functionality
• Keep edge case tests

Update Systematically

• Fix one type of issue at a time
• Run tests after each change
• Verify fixes don't break other tests

Improve While Updating

• Use fixtures for common setup
• Parametrize similar tests
• Improve test names and docs

Verify Thoroughly

• Run full test suite
• Check for flaky tests
• Verify test independence

Troubleshooting

Issue: Tests pass but don't test new behavior

Solution:

• Add new test cases for new functionality
• Update existing tests to cover new parameters
• Verify test coverage

Issue: Can't determine what changed

Solution:

• Use code diff analyzer script
• Compare function signatures manually
• Run old tests against new code to see failures
• Analyze error messages

Issue: Too many test failures

Solution:

• Fix one test at a time
• Group similar failures
• Fix systematic issues first (imports, signatures)
• Then fix assertion issues

Issue: Tests pass but behavior seems wrong

Solution:

• Verify test assertions are correct
• Check if test is actually testing new behavior
• Add more specific assertions
• Test edge cases

Example Workflow

Scenario: Function signature changed

Old code:

def calculate_price(quantity, unit_price):
    return quantity * unit_price

New code:

def calculate_price(quantity, unit_price, discount=0.0):
    subtotal = quantity * unit_price
    return subtotal * (1 - discount)

Old test:

def test_calculate_price():
    price = calculate_price(5, 10.0)
    assert price == 50.0

Analysis:

• Added parameter: discount with default value 0.0
• Behavior unchanged when discount not provided
• New behavior when discount provided

Updated test:

def test_calculate_price():
    # Test without discount (original behavior)
    price = calculate_price(5, 10.0)
    assert price == 50.0

def test_calculate_price_with_discount():
    # Test with discount (new behavior)
    price = calculate_price(5, 10.0, 0.1)
    assert price == 45.0  # 50 * 0.9

Verification:

pytest test_file.py -v
# Both tests should pass

Output Format

Provide updated test code with:

• Summary of changes:
• What was updated
• Why it was updated
• New tests added

• Updated test code:
• Complete updated test file
• All necessary imports
• All test functions

• Verification notes:
• How to run tests
• Expected results
• Any caveats or notes