#include "NvInfer.h"
#include "cuda_runtime_api.h"
#include "logging.h"
#include <fstream>
#include <iostream>
#include <map>
#include <sstream>
#include <vector>
#include <chrono>
#include <cmath>

#define CHECK(status) \
    do\
    {\
        auto ret = (status);\
        if(ret != 0)\
        {\
            std::cerr << "Cuda failure: " << ret << std::endl;\
            abort();\
        }\
    } while(0)

// stuff we know about the network and the input/output blobs 
static const int INPUT_H = 224;
static const int INPUT_W = 224;
static const int OUTPUT_SIZE = 1000;

const char* INPUT_BLOB_NAME = "data";
const char* OUTPUT_BLOB_NAME = "prob";

using namespace nvinfer1;

static Logger gLogger;

// Load weights from files shared with TensorRT samples.
// TensorRT weigths files have a simple space delimited format:
// [tpyt] [size] <data x size in hex>
std::map<std::string, Weights> loadWeights(const std::string file)
{
    std::cout << "Loading weights: " << file << std::endl;
    std::map<std::string, Weights>weightMap;

    // Open weights file
    std::ifstream input(file);
    assert(input.is_open() && "Unable to load weight file.");

    // Read number of weight blobs
    int32_t count;
    input >> count;
    assert(count > 0 && "Invalis weight map file");

    while (count--)
    {
        Weights wt{ DataType::kFLOAT, nullptr,0 };
        uint32_t size;

        // Read name and type of blob
        std::string name;
        input >> name >> std::dec >> size;
        wt.type = DataType::kFLOAT;

        // Load blob
        uint32_t* val = reinterpret_cast<uint32_t*>(malloc(sizeof(val)* size));
        for (uint32_t x = 0, y = size; x < y; ++x)
        {
            input >> std::hex >> val[x];
        }
        wt.values = val;

        wt.count = size;
        weightMap[name] = wt;
    }

    return weightMap;

}

IScaleLayer* addBatchNorm2d(INetworkDefinition *network, std::map<std::string, Weights>& weightMap, ITensor& input, std::string lname, float eps) {
    float *gamma = (float*)weightMap[lname + ".weight"].values;
    float *bata = (float*)weightMap[lname + ".bias"].values;
    float *mean = (float*)weightMap[lname + ".running_mean"].values;
    float *var = (float*)weightMap[lname + ".running_var"].values;
    int len = weightMap[lname + ".running_var"].count;
    std::cout << "len " << len << std::endl;

    float *scval = reinterpret_cast<float*>(malloc(sizeof(float) * len));
    for(int i=0; i < len; i++){
        scval[i] = gamma[i] / sqrt(var[i] + eps);
    }
    Weights scale{ DataType::kFLOAT,scval,len };

    float *shval = reinterpret_cast<float*>(malloc(sizeof(float) * len));
    for (int i = 0; i < len; i++) {
        shval[i] = bata[i] - mean[i] * gamma[i] / sqrt(var[i] + eps);
    }
    Weights shift{ DataType::kFLOAT, shval, len };

    float *pval = reinterpret_cast<float*>(malloc(sizeof(float) * len));
    for (int i = 0; i < len; i++) {
        pval[i] = 1.0;
    }
    Weights power{ DataType::kFLOAT, pval, len };

    weightMap[lname + ".scale"] = scale;
    weightMap[lname + ".shift"] = shift;
    weightMap[lname + ".power"] = power;
    IScaleLayer* scale_1 = network->addScale(input, ScaleMode::kCHANNEL, shift, scale, power);
    assert(scale_1);
    return scale_1;
}

IActivationLayer* basicBlock(INetworkDefinition* network, std::map<std::string, Weights>& weightMap, ITensor& input, int inch, int outch, int stride, std::string lname) {
    Weights emptywts{ DataType::kFLOAT, nullptr, 0 };

    IConvolutionLayer* conv1 = network->addConvolutionNd(input, outch, DimsHW{ 3,3 }, weightMap[lname + "conv1.weight"], emptywts);
    assert(conv1);
    conv1->setStrideNd(DimsHW{ stride,stride });
    conv1->setPaddingNd(DimsHW{ 1,1 });

    IScaleLayer* bn1 = addBatchNorm2d(network, weightMap, *conv1->getOutput(0), lname + "bn1", 1e-5);

    IActivationLayer* relu1 = network->addActivation(*bn1->getOutput(0), ActivationType::kRELU);
    assert(relu1);

    IConvolutionLayer* conv2 = network->addConvolutionNd(*relu1->getOutput(0), outch, DimsHW{ 3,3 }, weightMap[lname + "conv2.weight"], emptywts);
    assert(conv2);
    conv2->setPaddingNd(DimsHW{ 1,1 });

    IScaleLayer* bn2 = addBatchNorm2d(network, weightMap, *conv2->getOutput(0), lname + "bn2", 1e-5);

    IElementWiseLayer* ew1;
    if (inch != outch) {
        IConvolutionLayer* conv3 = network->addConvolutionNd(input, outch, DimsHW{ 1,1 }, weightMap[lname + "downsample.0.weight"], emptywts);
        assert(conv3);
        conv3->setStrideNd(DimsHW{ stride, stride });
        IScaleLayer* bn3 = addBatchNorm2d(network, weightMap, *conv3->getOutput(0), lname + "downsample.1", 1e-5);
        ew1 = network->addElementWise(*bn3->getOutput(0), *bn2->getOutput(0), ElementWiseOperation::kSUM);

    }else {
        ew1 = network->addElementWise(input, *bn2->getOutput(0),
            ElementWiseOperation::kSUM);

    }
    IActivationLayer* relu2 = network->addActivation(*ew1->getOutput(0), ActivationType::kRELU);
    assert(relu2);
    return relu2;
}

// Create the engine using only the API and not any parser.
ICudaEngine* createEngine(unsigned int maxBatchSize, IBuilder* builder, IBuilderConfig* config, DataType dt) {
    INetworkDefinition* network = builder->createNetworkV2(0U);

    // Create input tensor of shpae { 3, INPUT_H INPPUT_W} with name INPUT_BLOB_NAME
    ITensor* data = network->addInput(INPUT_BLOB_NAME, dt, Dims3{ 3,INPUT_H,INPUT_W });
    assert(data);

    std::map<std::string, Weights> weightMap = loadWeights("../resnet34.wts");
    Weights emptywts{ DataType::kFLOAT,nullptr,0 };

    IConvolutionLayer* conv1 = network->addConvolutionNd(*data, 64, DimsHW{ 7,7 }, weightMap["conv1.weight"], emptywts);
    assert(conv1);
    conv1->setStrideNd(DimsHW{ 2,2 });
    conv1->setPaddingNd(DimsHW{ 3,3 });

    IScaleLayer* bn1 = addBatchNorm2d(network, weightMap, *conv1->getOutput(0), "bn1", 1e-5);
    IActivationLayer* relu1 = network->addActivation(*bn1->getOutput(0), ActivationType::kRELU);
    assert(relu1);

    IPoolingLayer* pool1 = network->addPoolingNd(*relu1->getOutput(0), PoolingType::kMAX, DimsHW{ 3,3 });
    assert(pool1);
    pool1->setStrideNd(DimsHW{ 2,2 });
    pool1->setPaddingNd(DimsHW{ 1,1 });

    IActivationLayer* relu2 = basicBlock(network, weightMap, *pool1->getOutput(0), 64, 64, 1, "layer1.0.");
    IActivationLayer* relu3 = basicBlock(network, weightMap, *relu2->getOutput(0), 64, 64, 1, "layer1.1.");
    IActivationLayer* relu4 = basicBlock(network, weightMap, *relu3->getOutput(0), 64, 64, 1, "layer1.2.");
    IActivationLayer* relu5 = basicBlock(network, weightMap, *relu4->getOutput(0), 64, 128, 2, "layer2.0.");
    IActivationLayer* relu6 = basicBlock(network, weightMap, *relu5->getOutput(0), 128, 128, 1, "layer2.1.");
    IActivationLayer* relu7 = basicBlock(network, weightMap, *relu6->getOutput(0), 128, 128, 1, "layer2.2.");
    IActivationLayer* relu8 = basicBlock(network, weightMap, *relu7->getOutput(0), 128, 128, 1, "layer2.3.");
    IActivationLayer* relu9 = basicBlock(network, weightMap, *relu8->getOutput(0), 128, 256, 2, "layer3.0.");
    IActivationLayer* relu10 = basicBlock(network, weightMap, *relu9->getOutput(0), 256, 256, 1, "layer3.1.");
    IActivationLayer* relu11 = basicBlock(network, weightMap, *relu10->getOutput(0), 256, 256, 1, "layer3.2.");
    IActivationLayer* relu12 = basicBlock(network, weightMap, *relu11->getOutput(0), 256, 256, 1, "layer3.3.");
    IActivationLayer* relu13 = basicBlock(network, weightMap, *relu12->getOutput(0), 256, 256, 1, "layer3.4.");
    IActivationLayer* relu14 = basicBlock(network, weightMap, *relu13->getOutput(0), 256, 256, 1, "layer3.5.");
    IActivationLayer* relu15 = basicBlock(network, weightMap, *relu14->getOutput(0), 256, 512, 2, "layer4.0.");
    IActivationLayer* relu16 = basicBlock(network, weightMap, *relu15->getOutput(0), 512, 512, 1, "layer4.1.");
    IActivationLayer* relu17 = basicBlock(network, weightMap, *relu16->getOutput(0), 512, 512, 1, "layer4.2.");
    IPoolingLayer* pool2 = network->addPoolingNd(*relu17->getOutput(0), PoolingType::kAVERAGE, DimsHW{ 7,7 });
    assert(pool2);
    pool2->setStrideNd(DimsHW{ 1,1 });
    IFullyConnectedLayer* fc1 = network->addFullyConnected(*pool2->getOutput(0), 1000, weightMap["fc.weight"], weightMap["fc.bias"]);
    assert(fc1);

    fc1->getOutput(0)->setName(OUTPUT_BLOB_NAME);
    std::cout << "set name out" << std::endl;
    network->markOutput(*fc1->getOutput(0));

    // Build engine
    builder->setMaxBatchSize(maxBatchSize);
    config->setMaxWorkspaceSize(1 << 20);
    ICudaEngine* engine = builder->buildEngineWithConfig(*network, *config);
    std::cout << "build out" << std::endl;

    // Don't need the network any more
    network->destroy();

    // Release host memory
    for (auto& mem : weightMap) {
        free((void*)(mem.second.values));
    }

    return engine;
}

void APIToModel(unsigned int maxBatchSize, IHostMemory** modelStream)
{
    // Create builder
    IBuilder* builder = createInferBuilder(gLogger);
    IBuilderConfig* config = builder->createBuilderConfig();

    // Create model to populate the network, then set outputs and create an engine
    ICudaEngine* engine = createEngine(maxBatchSize, builder, config, DataType::kFLOAT);
    assert(engine != nullptr);

    // Serialize the engine
    (*modelStream) = engine->serialize();

    // Close everything down
    engine->destroy();
    builder->destroy();
    config->destroy();
}

void doInference(IExecutionContext& context, float* input, float* output, int batchSize)
{
    const ICudaEngine& engine = context.getEngine();

    // Pointers to input and output device buffers to pass to engine.
    // Engine requires exactly IEngine::getNbBingdings() number of buffers.
    assert(engine.getNbBindings() == 2);
    void* buffers[2];

    // In order to bind the buffers, we need to konow the names of the input and output tensors.
    // Note that indices are guaranteed to be less than IEngine::getNbBindings()
    const int inputIndex = engine.getBindingIndex(INPUT_BLOB_NAME);
    const int outputIndex = engine.getBindingIndex(OUTPUT_BLOB_NAME);

    // Create GPU buffers on device
    CHECK(cudaMalloc(&buffers[inputIndex], batchSize * 3 * INPUT_H* INPUT_W * sizeof(float)));
    CHECK(cudaMalloc(&buffers[outputIndex], batchSize * OUTPUT_SIZE * sizeof(float)));

    // Create stream
    cudaStream_t stream;
    CHECK(cudaStreamCreate(&stream));

    // DMA input batch data to device, infer on the batch asynchronously, and DMA output back to host
    CHECK(cudaMemcpyAsync(buffers[inputIndex], input, batchSize * 3 * INPUT_H * INPUT_W * sizeof(float), cudaMemcpyHostToDevice, stream));
    context.enqueue(batchSize, buffers, stream, nullptr);
    CHECK(cudaMemcpyAsync(output, buffers[outputIndex], batchSize * OUTPUT_SIZE * sizeof(float), cudaMemcpyDeviceToHost, stream));
    cudaStreamSynchronize(stream);

    // Release stream and buffers
    cudaStreamDestroy(stream);
    CHECK(cudaFree(buffers[inputIndex]));
    CHECK(cudaFree(buffers[outputIndex]));

}

int main(int argc, char** argv)
{
    if (argc != 2) {
        std::cerr << "arguments not right!" << std::endl;
        std::cerr << "./resnet34 -s   // serialize model to plan file" << std::endl;
        std::cerr << "./resnet34 -d    // desrialize plan file and run inference" << std::endl;
        return -1;
    }

    // Create a model using the API directly and serialize it to a stream
    char *trtModelStream{ nullptr };
    size_t size(0);

    if (std::string(argv[1]) == "-s") {
        IHostMemory* modelStream{ nullptr };
        APIToModel(1, &modelStream);
        assert(modelStream != nullptr);

        std::ofstream p("resnet34.engine", std::ios::binary);
        if (!p) {
            std::cerr << "could not open plan output file" << std::endl;
            return -1;
        }
        p.write(reinterpret_cast<const char*>(modelStream->data()), modelStream->size());
        modelStream->destroy();
        return 1;
    }else if (std::string(argv[1]) == "-d") {
        std::ifstream file("resnet34.engine", std::ios::binary);
        if (file.good()) {
            file.seekg(0, file.end);
            size = file.tellg();
            file.seekg(0, file.beg);
            trtModelStream = new char[size];
            assert(trtModelStream);
            file.read(trtModelStream, size);
            file.close();
        }
    }else {
        return -1;

    }

    // Subtract mean from image
    static float data[3 * INPUT_H * INPUT_W];
    for (int i = 0; i < 3 * INPUT_H * INPUT_W; i++) {
        data[i] = 1.0;
    }

    IRuntime* runtime = createInferRuntime(gLogger);
    assert(runtime != nullptr);
    ICudaEngine* engine = runtime->deserializeCudaEngine(trtModelStream, size, nullptr);
    assert(engine != nullptr);
    IExecutionContext* context = engine->createExecutionContext();
    assert(context != nullptr);
    delete[] trtModelStream;

    // Run inference
    static float prob[OUTPUT_SIZE];
    for (int i = 0; i < 100; i++) {
        auto start = std::chrono::system_clock::now();
        doInference(*context, data, prob, 1);
        auto end = std::chrono::system_clock::now();
        std::cout << std::chrono::duration_cast<std::chrono::milliseconds>(end - start).count() << "ms" << std::endl;
    }

    // Destroy the engine
    context->destroy();
    engine->destroy();
    runtime->destroy();

    // Print historgram of the output distribution
    std::cout << "\nOutput:\n\n";
    for (unsigned int i = 0; i < 10; i++)
    {
        std::cout << prob[i] << ",";
    }
    std::cout << std::endl;
    for (unsigned int i = 0; i < 10; i++)
    {
    std::cout << prob[OUTPUT_SIZE - 10 + i] << ",";
    }
    std::cout << std::endl;

    return 0;
}