Mq B Lab2qrcode Modern Cpp Refactor

Modern C++ Refactor - Code Modernization Intelligence

Automated detection and refactoring of legacy C/C++98/11 code to modern C++17/20 standards using smart pointers, ranges, std::format, and contemporary best practices.

Prerequisites

Check if Python is installed:

python3 --version || python --version

If Python is not installed, install it based on user's OS:

macOS:

brew install python3

Ubuntu/Debian:

sudo apt update && sudo apt install python3

Windows:

winget install Python.Python.3.12

How to Use This Skill

When user requests C++ code refactoring, modernization, or asks to write new C++ code, follow this workflow:

Step 0: Check Project Coding Standards (CRITICAL)

ALWAYS check for .clang-format file first:

# Check if .clang-format exists in project
ls .clang-format || find . -name ".clang-format" -type f

If .clang-format exists:

• STRICTLY FOLLOW the formatting rules defined in it
• Do NOT deviate from the project's established style
• Use the same brace style, indentation, spacing as configured
• After refactoring, format code with: clang-format -i <file>

If .clang-format does NOT exist:

• Follow Modern C++ best practices from this guide
• Consider creating a .clang-format for consistency

Also check for other project standards:

• CMakeLists.txt or Makefile for C++ standard version (C++17/20/23)
• CONTRIBUTING.md or STYLE.md for coding guidelines
• Existing code patterns in the codebase

Step 1: Scan for Legacy Patterns

Use the scanner script to detect legacy C++ patterns:

python3 .claude/skills/modern-cpp-refactor/scripts/scan.py <file_or_directory> [--detailed]

This will identify:

• Raw pointers and manual memory management
• C-style arrays and strings
• Old-style casts
• Manual loops vs ranges
• Printf/sprintf vs std::format
• NULL vs nullptr
• typedef vs using

Step 2: Analyze Code Context

Before refactoring, understand:

• C++ Standard: What standard is currently used? (Check CMakeLists.txt, Makefile, compiler flags)
• Dependencies: Are there third-party libraries that require specific patterns?
• Ownership: Who owns the memory? (unique_ptr, shared_ptr, or value semantics?)
• Performance: Is this hot path code? (Consider move semantics, perfect forwarding)

Step 3: Apply Modern C++ Transformations

Follow the modernization rules below to refactor code systematically.

Modern C++ Transformation Rules

Memory Management

Example:

// Legacy
int* data = new int[100];
// ... use data ...
delete[] data;

// Modern C++17/20
auto data = std::vector<int>(100);
// ... use data ... (automatic cleanup)

// Or for fixed size:
auto data = std::array<int, 100>{};

Raw Pointers to Smart Pointers

Example:

// Legacy
class Manager {
    Widget* widget;
public:
    Manager() : widget(new Widget()) {}
    ~Manager() { delete widget; }
};

// Modern C++17/20
class Manager {
    std::unique_ptr<Widget> widget;
public:
    Manager() : widget(std::make_unique<Widget>()) {}
    // No destructor needed - automatic cleanup
};

Containers and Arrays

Example:

// Legacy
char buffer[256];
sprintf(buffer, "Value: %d", value);

// Modern C++20
auto buffer = std::format("Value: {}", value);

// Modern C++17 (without format)
auto buffer = std::string("Value: ") + std::to_string(value);

Loops and Algorithms

Example:

// Legacy
std::vector<int> vec = {1, 2, 3, 4, 5};
for (size_t i = 0; i < vec.size(); i++) {
    vec[i] *= 2;
}

// Modern C++17
for (auto& val : vec) {
    val *= 2;
}

// Modern C++20 (ranges)
#include <ranges>
auto doubled = vec | std::views::transform([](int x) { return x * 2; });

Type Declarations

Example:

// Legacy
std::map<std::string, int>::iterator it = myMap.find("key");
if (it != myMap.end()) {
    int value = it->second;
}

// Modern C++17
if (auto it = myMap.find("key"); it != myMap.end()) {
    auto [key, value] = *it;
    // Use key and value directly
}

// Modern C++17 (structured binding)
for (const auto& [key, value] : myMap) {
    // Use key and value directly
}

Error Handling

Example:

// Legacy
Widget* findWidget(int id) {
    // ... search ...
    if (not_found) return nullptr;
    return widget;
}

// Modern C++17
std::optional<Widget> findWidget(int id) {
    // ... search ...
    if (not_found) return std::nullopt;
    return widget;
}

// Usage:
if (auto widget = findWidget(42)) {
    // Use *widget
}

String Formatting

Example:

// Legacy
char buffer[256];
sprintf(buffer, "Name: %s, Age: %d", name.c_str(), age);

// Modern C++20
auto result = std::format("Name: {}, Age: {}", name, age);

// Alternative (C++17 with fmt library):
auto result = fmt::format("Name: {}, Age: {}", name, age);

Casts

Initialization

Example:

// Legacy
std::vector<int> vec = std::vector<int>(10, 5);

// Modern C++17
auto vec = std::vector<int>(10, 5);
// Or even better:
std::vector<int> vec(10, 5);

Const Correctness

Example:

// Legacy
int getValue() { return value; }

// Modern C++17
int getValue() const { return value; }
constexpr int getMax() const { return 100; }

Move Semantics

Example:

// Modern C++17
class Widget {
    std::vector<int> data;
public:
    // Move constructor
    Widget(Widget&& other) noexcept
        : data(std::move(other.data)) {}

    // Move assignment
    Widget& operator=(Widget&& other) noexcept {
        data = std::move(other.data);
        return *this;
    }
};

Concepts (C++20)

Example:

// Legacy
template<typename T>
typename std::enable_if<std::is_integral<T>::value, T>::type
add(T a, T b) { return a + b; }

// Modern C++20
template<std::integral T>
T add(T a, T b) { return a + b; }

// Or with concept:
std::integral auto add(std::integral auto a, std::integral auto b) {
    return a + b;
}

Code Scanning Categories

The scanner will check for these categories:

1. Memory Safety Issues

• Raw new / delete
• Manual array allocation
• Missing delete (potential leaks)
• malloc / free in C++ code

2. Type Safety Issues

• C-style casts
• NULL instead of nullptr
• Missing const qualifiers
• Implicit conversions

3. Modernization Opportunities

• typedef instead of using
• Manual loops vs ranges
• std::bind vs lambdas
• Raw function pointers vs std::function

4. String and I/O Issues

• C-style strings (char*, char[])
• sprintf family
• printf vs std::format

5. Concurrency (if applicable)

• Raw std::thread without RAII wrapper
• Manual locking vs std::lock_guard
• Missing std::atomic for shared data

Pre-Refactor Checklist

Before modernizing code, verify:

• Identify the target C++ standard (C++17 or C++20)
• Check if compiler supports required features
• Understand existing ownership semantics
• Identify performance-critical sections
• Check for third-party library constraints
• Ensure comprehensive test coverage exists
• Plan incremental refactoring (not all at once)

Post-Refactor Checklist

After modernizing code, verify:

• All tests still pass
• No memory leaks (use valgrind/sanitizers)
• Performance is acceptable (profile if needed)
• Code compiles with -Wall -Wextra -Werror
• Smart pointers have correct ownership semantics
• Move semantics applied where beneficial
• const correctness maintained
• No raw new / delete remaining (except special cases)

Example Workflow

User request: "Modernize this C++ class to use C++17/20 features"

AI should:

1. Scan the code:

python3 .claude/skills/modern-cpp-refactor/scripts/scan.py src/MyClass.cpp --detailed

2. Analyze findings:

   • Identify legacy patterns (raw pointers, manual loops, etc.)
   • Understand ownership semantics
   • Check for performance implications

3. Refactor systematically:

   • Replace raw pointers with smart pointers
   • Convert manual loops to range-based or algorithms
   • Use auto for obvious types
   • Apply structured bindings where applicable
   • Replace sprintf with std::format (C++20) or alternatives
   • Add constexpr and const where appropriate

4. Verify:

   • Compile with modern C++ flags
   • Run tests
   • Check for memory leaks
   • Review performance

Common Anti-Patterns to Avoid

Don't Over-Use Smart Pointers

// Bad: Using unique_ptr for simple value
std::unique_ptr<int> value = std::make_unique<int>(42);

// Good: Just use a value
int value = 42;

Don't Use Shared Ptr Everywhere

// Bad: Shared ownership without need
std::shared_ptr<Widget> widget = std::make_shared<Widget>();

// Good: Unique ownership when possible
auto widget = std::make_unique<Widget>();

Don't Ignore Move Semantics

// Bad: Unnecessary copy
std::vector<int> getData() {
    std::vector<int> result = computeData();
    return result; // Copy
}

// Good: Move semantics (or RVO)
std::vector<int> getData() {
    return computeData(); // Move or RVO
}

Quick Reference: Modern C++ Features by Standard

C++17 Must-Use Features

• Structured bindings: auto [a, b] = pair;
• if / switch with initializer: if (auto x = foo(); x > 0)
• std::optional<T> for maybe-values
• std::variant<T1, T2> for type-safe unions
• std::string_view for non-owning string refs
• Fold expressions in templates
• Class template argument deduction (CTAD)

C++20 Must-Use Features

• std::format for string formatting
• Ranges and views
• Concepts for template constraints
• std::span<T> for array views
• Three-way comparison operator <=>
• Designated initializers: Point{.x=1, .y=2}
• consteval for compile-time evaluation

When NOT to Modernize

Sometimes legacy patterns are necessary:

1. Interfacing with C APIs: May require raw pointers
2. Performance-critical code: Profile before/after
3. Third-party library constraints: API may require specific types
4. Incremental migration: Don't refactor everything at once
5. Platform limitations: Embedded systems may lack std library

Additional Resources

• C++ Core Guidelines
• Modern C++ Features
• C++17 in Detail
• C++20 Ranges

Default Behavior

When user asks you to write NEW C++ code (not just refactor):

STEP 0 - ALWAYS CHECK FIRST:

• Check for .clang-format file - If it exists, STRICTLY follow its formatting rules

THEN follow these modern C++ standards:

1. Always use C++17/20 standards by default
2. Never use raw new / delete - use smart pointers or RAII containers
3. Prefer std::format (C++20) or fmt library over sprintf
4. Use auto for obvious types
5. Use range-based for instead of manual index loops
6. Use std::optional for maybe-values
7. Use structured bindings for pairs/tuples/structs
8. Mark functions const and constexpr when applicable
9. Use nullptr never NULL
10. Use using never typedef
11. Respect project's .clang-format for all code formatting decisions

This ensures all generated code follows both modern C++ best practices AND the project's established coding standards.

Converted and distributed by TomeVault — claim your Tome and manage your conversions.