#include "CVideoDecoder.h"
#include <Windows.h>

extern "C"
{
#include "libavcodec/avcodec.h"
#include "libavformat/avformat.h"
#include "libswscale/swscale.h"
#include "libavdevice/avdevice.h"
#include "libavutil/opt.h"
#include "libavfilter/avfilter.h"
#include "libavfilter/buffersrc.h"
#include "libavfilter/buffersink.h"
#include "libavutil/imgutils.h"
#include "libavutil/frame.h"
};

#include "CVideoDataManager.h"
#include "../Include/libyuv.h"

#include "FusionAndTailor.cuh"


CVideoDecoder::CVideoDecoder(AVCodecContext* pDecoder, bool bHard, CVideoDataManager* pDataManager)
{
	m_pDecoder = pDecoder;
	m_pDataManager = pDataManager;
	m_bHard = bHard;

	m_pDataManager->SetDecodeDataSize(m_pDecoder->width* m_pDecoder->height*3/2);
	_beginthreadex(nullptr, 0, &CVideoDecoder::DecodeThread, this, 0, 0);
	
	CUstream Cur = ((CUstream)pCuStream);
	cudaError Err = cudaStreamCreate(&(Cur));
	pCuStream = Cur;
}

CVideoDecoder::~CVideoDecoder()
{
	if (m_pDecoder != nullptr)
	{
		avcodec_close(m_pDecoder);
		m_pDecoder = nullptr;
	}
}

unsigned __stdcall CVideoDecoder::DecodeThread(void* param)
{
	CVideoDecoder* pThis = (CVideoDecoder*)param;
	pThis->DecodeProcessor();
	return 0;
}

void CVideoDecoder::DecodeProcessor()
{
	while (true)
	{
		//if (m_pDataManager->GetSyncDataSize() > 0)
		{
			CPacketInfo* pPack = nullptr;
			if (m_pDataManager->GetSyncData(pPack))
			{
				uint8_t* pData = nullptr;
				DecodePacket(pPack->m_pPkt, pData);

				//DecodePacketGpu(pPack->m_pPkt, pData);

				m_pDataManager->AddDecoderData(pData, pPack->m_dPtzAngle);

				m_pDataManager->ReleaseSyncData(pPack);

			} 
		}
		Sleep(5);
	}
}

bool CVideoDecoder::DecodePacket(AVPacket* pkt, unsigned char* & pData)
{
	int nRet = 0;
	AVFrame* tmp_frame = nullptr;

	avcodec_send_packet(m_pDecoder, pkt);

	AVFrame* pFrame = av_frame_alloc();
	nRet = avcodec_receive_frame(m_pDecoder, pFrame);

	

	pData = new unsigned char[pFrame->linesize[0] * pFrame->height *3];
	if (m_bHard)
	{
		// 如果采用的硬件加速剂，则调用avcodec_receive_frame()函数后，解码后的数据还在GPU中，所以需要通过此函数
		// 将GPU中的数据转移到CPU中来
		tmp_frame = av_frame_alloc();
		if (av_hwframe_transfer_data(tmp_frame, pFrame, 0) < 0)
		{
			return false;
		}
		if (pFrame != nullptr)
		{
			av_frame_free(&pFrame);
			pFrame = nullptr;
		}

		pFrame = tmp_frame;
		
		//libyuv::NV12ToI420(
		//	pFrame->data[0], pFrame->linesize[0],
		//	pFrame->data[1], pFrame->linesize[1],
		//	pData, pFrame->width,
		//	pData + pFrame->width * pFrame->height, pFrame->width / 2,
		//	pData + pFrame->width * pFrame->height * 5 / 4, pFrame->width / 2,
		//	pFrame->width, pFrame->height
		//);

		libyuv::NV12ToRGB24(
			pFrame->data[0], pFrame->linesize[0],
			pFrame->data[1], pFrame->linesize[1],
			pData, pFrame->width * 3,
			pFrame->width, pFrame->height
		);

	}
	else
	{
		int dst_stride_y2 = pFrame->width;
		int dst_stride_u2 = dst_stride_y2 / 2;
		int dst_stride_v2 = dst_stride_y2 / 2;
		
		libyuv::I420Copy(pFrame->data[0], pFrame->linesize[0], pFrame->data[1], pFrame->linesize[1], pFrame->data[2], pFrame->linesize[2],
			pData, pFrame->width, pData + pFrame->width * pFrame->height, pFrame->width/2, pData + pFrame->width * pFrame->height * 5 / 4, pFrame->width / 2,
			pFrame->width, pFrame->height);
	}
	
	//cv::Mat mat(pFrame->height * 3 / 2, pFrame->width, CV_8UC1, pData);

	av_frame_free(&pFrame);
	pFrame = nullptr;

	return true;
}

//注意该函数返回的指针指向的是显存数据，
bool CVideoDecoder::DecodePacketGpu(AVPacket* pkt, unsigned char*& pData)
{
	int nRet = 0;
	AVFrame* tmp_frame = nullptr;

	avcodec_send_packet(m_pDecoder, pkt);

	AVFrame* pFrame = av_frame_alloc();
	nRet = avcodec_receive_frame(m_pDecoder, pFrame);

	pFrame->format = AV_PIX_FMT_NV12;

	CUstream CurStream = (CUstream)pCuStream;

	if (m_pI420 == nullptr)
	{
		cudaMalloc(&m_pI420, pFrame->width * pFrame->height * 3 / 2);
		YUVTailorAndBlender::setColorSpace2(0);
	}

	
	YUVTailorAndBlender::TurnCUDAFormatToI420(pFrame->data[0], pFrame->data[1], 
		pFrame->linesize[0], pFrame->linesize[1], m_pI420, pFrame->width, pFrame->height, &CurStream);

	cudaError err = cudaDeviceSynchronize();

	if (pData != nullptr)
		cudaFree(pData);
	cudaMalloc((void**)&pData, pFrame->width * pFrame->height * 3 / 2);
	cudaMemcpy(pData,m_pI420 , pFrame->width * pFrame->height * 3 / 2, cudaMemcpyDeviceToDevice);

	//下载
	//cudaMemcpy(pData, m_pI420, pFrame->width * pFrame->height * 3/2, cudaMemcpyDeviceToHost);


	//FILE* file = fopen("111.yuv", "wb"); //已只读方式打开

	//fwrite(pData, 1, pFrame->height * pFrame->width, file); //写入操作
	//fwrite(pData + pFrame->height* pFrame->width, 1, pFrame->height * pFrame->width/4, file); //写入操作
	//fwrite(pData + pFrame->height * pFrame->width*5/4, 1, pFrame->height * pFrame->width/4, file); //写入操作

	//fclose(file);

	av_frame_free(&pFrame);

	return false;
}