#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <unistd.h>

#include "../../context.h"

int cpu_num = 1;
int length = 100;
int gpu_num = 0;
int CPU_ANY = -1;
int CPU_CUDA = -1;

void *start_taskManager(struct Context *context) {
    goto initDataGears(context, Gearef(context, LoopCounter), Gearef(context, TaskManager));
    return 0;
}

#ifdef USE_CUDAWorker
#ifdef USE_CUDA_MAIN_THREAD
extern volatile int cuda_initialized;
#endif
#endif

__code initDataGears(struct LoopCounter* loopCounter, struct TaskManager* taskManager) {
    // loopCounter->tree = createRedBlackTree(context);
    loopCounter->i = 0;
    taskManager->taskManager = (union Data*)createTaskManagerImpl(context, cpu_num, gpu_num, 0);
#ifdef USE_CUDAWorker
#ifdef USE_CUDA_MAIN_THREAD
    while(! cuda_initialized) {};
#endif
#endif
    goto meta(context, C_code1);
}

__code initDataGears_stub(struct Context* context) {
    struct TaskManager* taskManager =  Gearef(context, TaskManager);
    taskManager->taskManager = 0;
#if (! defined(USE_CUDAWorker) || ! defined(USE_CUDA_MAIN_THREAD))
    struct LoopCounter* loopCounter = Gearef(context, LoopCounter);
    goto initDataGears(context, loopCounter, taskManager);
#else
    cuda_initialized = 0;
    pthread_t thread;
    pthread_create(&thread, NULL, (void*)&start_taskManager, context);
    while (taskManager->taskManager == 0);
    TaskManager *t = (TaskManager*)taskManager->taskManager;
    TaskManagerImpl *im = (TaskManagerImpl*)t->taskManager;
    struct Queue *q = (Queue *)im->workers[0];
    createCUDAWorker(context,0,q, im);
    pthread_join(thread,0);
    exit(0);
#endif
}

__code code1(struct Time* time) {
    printf("cpus:\t\t%d\n", cpu_num);
    printf("gpus:\t\t%d\n", gpu_num);
    printf("length:\t\t%d\n", length);
    /* puts("queue"); */
    /* print_queue(context->data[ActiveQueue]->queue.first); */
    /* puts("tree"); */
    /* print_tree(context->data[Tree]->tree.root); */
    /* puts("result"); */

    time->time = (union Data*)createTimeImpl(context);
    time->next = C_createTask1;
    goto meta(context, time->time->Time.start);
}

//__code createTask1(struct LoopCounter* loopCounter, struct TaskManager* taskManager) {
//    int waitFlag = 0;
//    struct SortArray* inputSortArray = &ALLOCATE_DATA_GEAR(context, sortArray)->sortArray;
//    inputSortArray->array = array;
//    for (int i=2; i <= length; i=2*i) {
//        int first = 1;
//        lastOne = (length <= i*2) ? true : false;
//        for (int j=i>>1; 0 < j; j=j>>1) {
//            struct SortArray* outputSortArray = &ALLOCATE_DATA_GEAR(context, sortArray)->sortArray;
//            outputSortArray->array = array;
//            for(int k=0; 0 < length/2; k++) {
//                struct Context* task = NEW(struct Context);
//                struct Integer* integer1 = &ALLOCATE_DATA_GEAR(context, Integer)->Integer;
//                struct Integer* integer2 = &ALLOCATE_DATA_GEAR(context, Integer)->Integer;
//                struct Integer* integer3 = &ALLOCATE_DATA_GEAR(context, Integer)->Integer;
//                integer1->value = j;
//                integer2->value = first;
//                integer3->value = k;
//
//                task->next = C_swap;
//                if (waitFlag) {
//                    task->idgCount = length/2;
//                } else {
//                    task->idgCount = 0;
//                }
//                task->idg = task->dataNum;
//                task->data[task->idg] = (union Data*)inputSortArray;
//                task->data[task->idg+1] = (union Data*)integer1;
//                task->data[task->idg+2] = (union Data*)integer2;
//                task->data[task->idg+3] = (union Data*)integer3;
//                task->maxIdg = task->idg + 4;
//                task->odg = task->maxIdg;
//                task->data[task->odg] = outputSortArray;
//                task->maxOdg = task->odg + 1;
//                first = 0;
//            }
//            inputSortArray = outputSortArray;
//        }
//        waitFlag = 1;
//    }
//    goto meta(context, taskManager->taskManager->TaskManager.shutdown);
//}

__code createTask1(struct LoopCounter* loopCounter, struct TaskManager* taskManager, struct Time* time) {
    int i = loopCounter->i;

    if (i < length) {
        loopCounter->i++;
        goto meta(context, C_createTask2);
    }

    loopCounter->i = 0;
    taskManager->next = C_exit_code;
    goto meta(context, taskManager->taskManager->TaskManager.shutdown);
}

__code createTask2(struct LoopCounter* loopCounter, struct TaskManager* taskManager) {
    struct Context** tasks = (struct Context**)ALLOC_ARRAY(context, Context, 1);
    struct Context* task = NEW(struct Context);
    task->iterator = createOneDimIterator(context, 10);
    task->next = C_printIterator;
    task->idgCount = 0;
    task->idg = task->dataNum;
    task->maxIdg = task->idg;
    task->odg = task->maxIdg;
    task->maxOdg = task->odg;
    tasks[0] = task;
    taskManager->contexts = tasks;
    taskManager->next1 = C_createTask1;
    goto meta(context, taskManager->taskManager->TaskManager.spawnTasks);
}

void init(int argc, char** argv) {
    for (int i = 1; argv[i]; ++i) {
        if (strcmp(argv[i], "-cpu") == 0)
            cpu_num = (int)atoi(argv[i+1]);
        else if (strcmp(argv[i], "-l") == 0)
            length = (int)atoi(argv[i+1]);
        else if (strcmp(argv[i], "-cuda") == 0) {
            gpu_num = 1;
            CPU_CUDA = 0;
        }
    }
}

int main(int argc, char** argv) {
    init(argc, argv);
    struct Context* main_context = NEW(struct Context);
    initContext(main_context);
    main_context->next = C_initDataGears;

    goto start_code(main_context);
}