#pragma once

#include <atomic>
#include <memory>
#include <vector>
#include <queue>
#include <thread>
#include <mutex>
#include <condition_variable>
#include <functional>
#include <future>

namespace EngineManager {

// thread communication
template<typename T>
class LockFreeQueue {
private:
    struct Node {
        std::atomic<T*> data{nullptr};
        std::atomic<Node*> next{nullptr};
    };
    
    std::atomic<Node*> head_{nullptr};
    std::atomic<Node*> tail_{nullptr};
    
public:
    LockFreeQueue() {
        Node* dummy = new Node;
        head_.store(dummy);
        tail_.store(dummy);
    }
    
    ~LockFreeQueue() {
        while (Node* const old_head = head_.load()) {
            head_.store(old_head->next);
            delete old_head->data.load();
            delete old_head;
        }
    }
    
    void push(T item) {
        Node* new_node = new Node;
        T* data = new T(std::move(item));
        new_node->data.store(data);
        
        Node* prev_tail = tail_.exchange(new_node);
        prev_tail->next.store(new_node);
    }
    
    bool try_pop(T& result) {
        Node* head = head_.load();
        Node* next = head->next.load();
        
        if (next == nullptr) return false;
        
        T* data = next->data.load();
        if (data == nullptr) return false;
        
        result = *data;
        delete data;
        head_.store(next);
        delete head;
        return true;
    }
    
    bool empty() const {
        Node* head = head_.load();
        return (head->next.load() == nullptr);
    }
};

// thread pool stealing
class ThreadPool {
private:
    std::vector<std::thread> workers_;
    std::queue<std::function<void()>> tasks_;
    std::mutex queue_mutex_;
    std::condition_variable condition_;
    std::atomic<bool> stop_{false};
    
public:
    explicit ThreadPool(size_t num_threads = std::thread::hardware_concurrency()) {
        for (size_t i = 0; i < num_threads; ++i) {
            workers_.emplace_back([this] {
                for (;;) {
                    std::function<void()> task;
                    
                    {
                        std::unique_lock<std::mutex> lock(queue_mutex_);
                        condition_.wait(lock, [this] { return stop_ || !tasks_.empty(); });
                        
                        if (stop_ && tasks_.empty()) return;
                        
                        task = std::move(tasks_.front());
                        tasks_.pop();
                    }
                    
                    task();
                }
            });
        }
    }
    
    template<class F, class... Args>
    auto enqueue(F&& f, Args&&... args) -> std::future<typename std::invoke_result<F, Args...>::type> {
        using return_type = typename std::invoke_result<F, Args...>::type;
        
        auto task = std::make_shared<std::packaged_task<return_type()>>(
            std::bind(std::forward<F>(f), std::forward<Args>(args)...)
        );
        
        std::future<return_type> res = task->get_future();
        
        {
            std::unique_lock<std::mutex> lock(queue_mutex_);
            if (stop_) {
                throw std::runtime_error("ThreadPool: enqueue stopped");
            }
            
            tasks_.emplace([task](){ (*task)(); });
        }
        
        condition_.notify_one();
        return res;
    }
    
    size_t pending_tasks() const {
        std::lock_guard<std::mutex> lock(const_cast<std::mutex&>(queue_mutex_));
        return tasks_.size();
    }
    
    ~ThreadPool() {
        stop_ = true;
        condition_.notify_all();
        
        for (std::thread& worker : workers_) {
            worker.join();
        }
    }
};

class AsyncEngine {
private:
    static std::unique_ptr<AsyncEngine> instance_;
    static std::once_flag initialized_;
    
    std::unique_ptr<ThreadPool> main_pool_;
    std::unique_ptr<ThreadPool> io_pool_;
    LockFreeQueue<std::function<void()>> event_queue_;
    
    AsyncEngine() {
        unsigned int hw_threads = std::thread::hardware_concurrency();
        size_t main_threads = hw_threads > 4 ? hw_threads / 2 : 2;
        size_t io_threads = hw_threads > 8 ? 4 : 2;
        
        main_pool_ = std::make_unique<ThreadPool>(main_threads);
        io_pool_ = std::make_unique<ThreadPool>(io_threads);
    }
    
public:
    static AsyncEngine& instance() {
        std::call_once(initialized_, []() {
            instance_ = std::unique_ptr<AsyncEngine>(new AsyncEngine());
        });
        return *instance_;
    }
    
    // main
    template<class F, class... Args>
    auto execute(F&& f, Args&&... args) -> std::future<typename std::invoke_result<F, Args...>::type> {
        return main_pool_->enqueue(std::forward<F>(f), std::forward<Args>(args)...);
    }
    
    template<class F, class... Args>
    auto execute_io(F&& f, Args&&... args) -> std::future<typename std::invoke_result<F, Args...>::type> {
        return io_pool_->enqueue(std::forward<F>(f), std::forward<Args>(args)...);
    }
    
    void push_event(std::function<void()> event) {
        event_queue_.push(std::move(event));
    }
    
    void process_events() {
        std::function<void()> event;
        while (event_queue_.try_pop(event)) {
            event();
        }
    }
    
    size_t pending_main_tasks() const { return main_pool_->pending_tasks(); }
    size_t pending_io_tasks() const { return io_pool_->pending_tasks(); }
    
    static void shutdown() {
        instance_.reset();
    }
};

}