Safe Refactoring Techniques

This skill provides systematic approaches to refactoring code safely, improving structure and maintainability while preserving behavior.

When to use this skill

Use this skill when you need to:

• Improve code readability and maintainability
• Reduce technical debt
• Prepare code for new features
• Apply design patterns
• Simplify complex code

Prerequisites

• Comprehensive test coverage for the code to refactor
• Understanding of the codebase and dependencies
• Version control with clean history
• Time to verify behavior hasn't changed

Guidelines

Refactoring Principles

Golden Rule

• Never refactor without tests
• Make small, incremental changes
• Run tests after each change
• Commit frequently

When to Refactor

• Rule of Three: Similar code appearing 3+ times
• Adding a feature is harder than it should be
• Understanding code takes too long
• Code violates SOLID principles
• Performance optimization needed

Refactoring Workflow

Preparation

1. Ensure tests exist and pass
2. Review code to understand current behavior
3. Identify specific issues to address
4. Plan refactoring steps

Execution

1. Make one small change at a time
2. Run tests after each change
3. Commit working states frequently
4. Use IDE refactoring tools when possible

Verification

1. All tests pass
2. Code review by peer
3. Verify in staging environment
4. Monitor production after deployment

Common Refactorings

Extract Method

# Before
def process_order(order):
    print(f"Processing order {order.id}")
    if order.total > 1000:
        print("High value order - requires approval")
        send_notification(order.manager, "Approval needed")
    update_inventory(order.items)
    send_confirmation(order.customer)

# After
def process_order(order):
    log_order_processing(order)
    if requires_approval(order):
        request_approval(order)
    fulfill_order(order)

def log_order_processing(order):
    print(f"Processing order {order.id}")

def requires_approval(order):
    return order.total > 1000

def request_approval(order):
    print("High value order - requires approval")
    send_notification(order.manager, "Approval needed")

def fulfill_order(order):
    update_inventory(order.items)
    send_confirmation(order.customer)

Introduce Parameter Object

# Before
def create_user(name, email, phone, address, city, zip_code):
    pass

# After
@dataclass
class UserInfo:
    name: str
    email: str
    phone: str
    address: str
    city: str
    zip_code: str

def create_user(user_info: UserInfo):
    pass

Replace Conditional with Polymorphism

# Before
def calculate_salary(employee):
    if employee.type == "fulltime":
        return employee.base_salary
    elif employee.type == "contractor":
        return employee.hourly_rate * employee.hours
    elif employee.type == "intern":
        return 0

# After
class FullTimeEmployee(Employee):
    def calculate_salary(self):
        return self.base_salary

class Contractor(Employee):
    def calculate_salary(self):
        return self.hourly_rate * self.hours

class Intern(Employee):
    def calculate_salary(self):
        return 0

Code Smells to Address

Bloaters

• Long methods → Extract methods
• Large classes → Extract classes
• Long parameter lists → Introduce parameter objects
• Data clumps → Extract classes

Object-Orientation Abusers

• Switch statements → Polymorphism
• Temporary fields → Extract classes
• Refused bequests → Replace inheritance with delegation
• Alternative classes → Unify interfaces

Change Preventers

• Divergent change → Split responsibilities
• Shotgun surgery → Move methods/fields
• Parallel inheritance → Bridge pattern

Dispensables

• Duplicate code → Extract common code
• Lazy classes → Inline or remove
• Data classes → Add behavior
• Dead code → Delete

Safety Techniques

Characterization Tests When tests don't exist, write tests that capture current behavior before refactoring:

def test_current_behavior():
    """Document current behavior before refactoring."""
    result = legacy_function(input_data)
    assert result == expected_output  # Record actual output

Parallel Implementation Keep old implementation while building new:

def process_data(data, use_new_impl=False):
    if use_new_impl:
        return new_implementation(data)
    return old_implementation(data)

Feature Flags Enable gradual rollout of refactored code:

if feature_flags.enabled("new-payment-flow"):
    process_payment_v2(order)
else:
    process_payment_v1(order)

Examples

See the examples/ directory for:

• extract-method-examples/ - Method extraction patterns
• design-pattern-refactorings/ - Applying design patterns
• legacy-code-techniques.md - Working with untested code
• before-after-comparisons.md - Real refactoring examples

References

• Refactoring by Martin Fowler
• Refactoring Guru
• Working Effectively with Legacy Code
• Clean Code by Robert Martin