#include <catch.hpp>
#include <pls/internal/base/thread.h>
#include <pls/internal/base/spin_lock.h>
#include <pls/internal/base/aligned_stack.h>
#include <pls/internal/base/system_details.h>

#include <vector>
#include <mutex>

using namespace pls::internal::base;
using namespace std;

static bool base_tests_visited;
static int base_tests_local_value_one;
static vector<int> base_tests_local_value_two;

TEST_CASE( "thread creation and joining", "[internal/base/thread.h]") {
    base_tests_visited = false;
    auto t1 = start_thread([]() { base_tests_visited = true; });
    t1.join();

    REQUIRE(base_tests_visited);
}

TEST_CASE( "thread state", "[internal/base/thread.h]") {
    int state_one = 1;
    vector<int> state_two{1, 2};

    auto t1 = start_thread([]() { base_tests_local_value_one = *this_thread::state<int>(); }, &state_one);
    auto t2 = start_thread([]() { base_tests_local_value_two = *this_thread::state<vector<int>>(); }, &state_two);
    t1.join();
    t2.join();

    REQUIRE(base_tests_local_value_one == 1);
    REQUIRE(base_tests_local_value_two == vector<int>{1, 2});
}

int base_tests_shared_counter;

TEST_CASE( "spinlock protects concurrent counter", "[internal/base/spinlock.h]") {
    constexpr int num_iterations = 1000000;
    base_tests_shared_counter = 0;
    spin_lock lock{};

    SECTION( "lock can be used by itself" ) {
        auto t1 = start_thread([&]() {
            for (int i = 0; i < num_iterations; i++) {
                lock.lock();
                base_tests_shared_counter++;
                lock.unlock();
            }
        });
        auto t2 = start_thread([&]() {
            for (int i = 0; i < num_iterations; i++) {
                lock.lock();
                base_tests_shared_counter--;
                lock.unlock();
            }
        });

        t1.join();
        t2.join();

        REQUIRE(base_tests_shared_counter == 0);
    }

    SECTION( "lock can be used with std::lock_guard" ) {
        auto t1 = start_thread([&]() {
            for (int i = 0; i < num_iterations; i++) {
                std::lock_guard<spin_lock> my_lock{lock};
                base_tests_shared_counter++;
            }
        });
        auto t2 = start_thread([&]() {
            for (int i = 0; i < num_iterations; i++) {
                std::lock_guard<spin_lock> my_lock{lock};
                base_tests_shared_counter--;
            }
        });

        t1.join();
        t2.join();

        REQUIRE(base_tests_shared_counter == 0);
    }
}

TEST_CASE( "aligned stack stores objects correctly", "[internal/base/aligned_stack.h]") {
    constexpr long data_size = 1024;
    char data[data_size];
    aligned_stack stack{data, data_size};

    SECTION( "stack correctly pushes sub linesize objects" ) {
        std::array<char, 5> small_data_one{'a', 'b', 'c', 'd', 'e'};
        std::array<char, 64> small_data_two{};
        std::array<char, 1> small_data_three{'A'};

        auto pointer_one = stack.push(small_data_one);
        auto pointer_two = stack.push(small_data_two);
        auto pointer_three = stack.push(small_data_three);

        REQUIRE(reinterpret_cast<std::uintptr_t>(pointer_one) % CACHE_LINE_SIZE == 0);
        REQUIRE(reinterpret_cast<std::uintptr_t>(pointer_two) % CACHE_LINE_SIZE == 0);
        REQUIRE(reinterpret_cast<std::uintptr_t>(pointer_three) % CACHE_LINE_SIZE == 0);
    }

    SECTION( "stack correctly pushes above linesize objects" ) {
        std::array<char, 5> small_data_one{'a', 'b', 'c', 'd', 'e'};
        std::array<char, CACHE_LINE_SIZE + 10> big_data_one{};

        auto big_pointer_one = stack.push(big_data_one);
        auto small_pointer_one = stack.push(small_data_one);

        REQUIRE(reinterpret_cast<std::uintptr_t>(big_pointer_one) % CACHE_LINE_SIZE == 0);
        REQUIRE(reinterpret_cast<std::uintptr_t>(small_pointer_one) % CACHE_LINE_SIZE == 0);
    }

    SECTION( "stack correctly stores and retrieves objects" ) {
        std::array<char, 5> data_one{'a', 'b', 'c', 'd', 'e'};

        stack.push(data_one);
        auto retrieved_data = stack.pop<std::array<char, 5>>();

        REQUIRE(retrieved_data == std::array<char, 5>{'a', 'b', 'c', 'd', 'e'});
    }

    SECTION( "stack can push and pop multiple times with correct alignment" ) {
        std::array<char, 5> small_data_one{'a', 'b', 'c', 'd', 'e'};
        std::array<char, 64> small_data_two{};
        std::array<char, 1> small_data_three{'A'};

        auto pointer_one = stack.push(small_data_one);
        auto pointer_two = stack.push(small_data_two);
        auto pointer_three = stack.push(small_data_three);
        stack.pop<typeof(small_data_three)>();
        stack.pop<typeof(small_data_two)>();
        auto pointer_four = stack.push(small_data_two);
        auto pointer_five = stack.push(small_data_three);

        REQUIRE(reinterpret_cast<std::uintptr_t>(pointer_one) % CACHE_LINE_SIZE == 0);
        REQUIRE(reinterpret_cast<std::uintptr_t>(pointer_two) % CACHE_LINE_SIZE == 0);
        REQUIRE(reinterpret_cast<std::uintptr_t>(pointer_three) % CACHE_LINE_SIZE == 0);
        REQUIRE(reinterpret_cast<std::uintptr_t>(pointer_four) % CACHE_LINE_SIZE == 0);
        REQUIRE(reinterpret_cast<std::uintptr_t>(pointer_five) % CACHE_LINE_SIZE == 0);

        REQUIRE(pointer_four == pointer_two);
        REQUIRE(pointer_five == pointer_three);
    }
}