block.cpp

avs decoder sources, added a command line sample

				pPred[i*8+j] = Clip((Dpie+7*hpie+7*Hpie+xxpie+64)>>7);
			}
		}

	}
	else if(dy==3 && dx==1) //p
	{
		for(i=0; i<8; i++)
		{
			for(j=0; j<8; j++)
			{
				int Hpiepie = *(pbRefY + max(0,min(iImgHeight-1, refY+i+1))*iImgWidth + max(0, min(iImgWidth-1, refX+j))) * 64;  
				int aapie = 
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i-1))*iImgWidth + max(0, min(iImgWidth-1, refX+j-1))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i-1))*iImgWidth + max(0, min(iImgWidth-1, refX+j  ))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i-1))*iImgWidth + max(0, min(iImgWidth-1, refX+j+1))))
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i-1))*iImgWidth + max(0, min(iImgWidth-1, refX+j+2))));
				int bpie =  
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i  ))*iImgWidth + max(0, min(iImgWidth-1, refX+j-1))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i  ))*iImgWidth + max(0, min(iImgWidth-1, refX+j  ))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i  ))*iImgWidth + max(0, min(iImgWidth-1, refX+j+1))))
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i  ))*iImgWidth + max(0, min(iImgWidth-1, refX+j+2))));
				int spie = 
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i+1))*iImgWidth + max(0, min(iImgWidth-1, refX+j-1))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i+1))*iImgWidth + max(0, min(iImgWidth-1, refX+j  ))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i+1))*iImgWidth + max(0, min(iImgWidth-1, refX+j+1))))
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i+1))*iImgWidth + max(0, min(iImgWidth-1, refX+j+2))));
				int ddpie = 
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i+2))*iImgWidth + max(0, min(iImgWidth-1, refX+j-1))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i+2))*iImgWidth + max(0, min(iImgWidth-1, refX+j  ))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i+2))*iImgWidth + max(0, min(iImgWidth-1, refX+j+1))))
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i+2))*iImgWidth + max(0, min(iImgWidth-1, refX+j+2))));
				int jpie = -aapie + 5*bpie + 5*spie - ddpie;
				pPred[i*8+j] = Clip((jpie+Hpiepie+64)>>7);
			}
		}
	}
	else if(dy==3 && dx==2) //q
	{
		for(i=0; i<8; i++)
		{
			for(j=0; j<8; j++)
			{
				int aapie = 
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i-1))*iImgWidth + max(0, min(iImgWidth-1, refX+j-1))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i-1))*iImgWidth + max(0, min(iImgWidth-1, refX+j  ))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i-1))*iImgWidth + max(0, min(iImgWidth-1, refX+j+1))))
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i-1))*iImgWidth + max(0, min(iImgWidth-1, refX+j+2))));
				int bpie =  
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i  ))*iImgWidth + max(0, min(iImgWidth-1, refX+j-1))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i  ))*iImgWidth + max(0, min(iImgWidth-1, refX+j  ))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i  ))*iImgWidth + max(0, min(iImgWidth-1, refX+j+1))))
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i  ))*iImgWidth + max(0, min(iImgWidth-1, refX+j+2))));       
				int spie = 
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i+1))*iImgWidth + max(0, min(iImgWidth-1, refX+j-1))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i+1))*iImgWidth + max(0, min(iImgWidth-1, refX+j  ))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i+1))*iImgWidth + max(0, min(iImgWidth-1, refX+j+1))))
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i+1))*iImgWidth + max(0, min(iImgWidth-1, refX+j+2))));
				int ddpie = 
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i+2))*iImgWidth + max(0, min(iImgWidth-1, refX+j-1))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i+2))*iImgWidth + max(0, min(iImgWidth-1, refX+j  ))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i+2))*iImgWidth + max(0, min(iImgWidth-1, refX+j+1))))
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i+2))*iImgWidth + max(0, min(iImgWidth-1, refX+j+2))));
				int xxpie = 
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i+3))*iImgWidth + max(0, min(iImgWidth-1, refX+j-1))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i+3))*iImgWidth + max(0, min(iImgWidth-1, refX+j  ))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i+3))*iImgWidth + max(0, min(iImgWidth-1, refX+j+1))))
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i+3))*iImgWidth + max(0, min(iImgWidth-1, refX+j+2))));
				int jpie = -aapie + 5*bpie + 5*spie - ddpie;
				int jxxpie = -bpie + 5*spie + 5*ddpie - xxpie;     
				int qpie = bpie*8 + 7*jpie + 7*spie*8 + jxxpie;
				pPred[i*8+j] = Clip((qpie + 512)>>10);
			}
		}
	}
	else if(dx==3 && dy==3) //r
	{
		for(i=0; i<8; i++)
		{
			for(j=0; j<8; j++)
			{
				int Ipiepie = *(pbRefY + max(0,min(iImgHeight-1, refY+i+1))*iImgWidth + max(0, min(iImgWidth-1, refX+j+1))) * 64;  
				int aapie = 
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i-1))*iImgWidth + max(0, min(iImgWidth-1, refX+j-1))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i-1))*iImgWidth + max(0, min(iImgWidth-1, refX+j  ))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i-1))*iImgWidth + max(0, min(iImgWidth-1, refX+j+1))))
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i-1))*iImgWidth + max(0, min(iImgWidth-1, refX+j+2))));
				int bpie =  
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i  ))*iImgWidth + max(0, min(iImgWidth-1, refX+j-1))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i  ))*iImgWidth + max(0, min(iImgWidth-1, refX+j  ))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i  ))*iImgWidth + max(0, min(iImgWidth-1, refX+j+1))))
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i  ))*iImgWidth + max(0, min(iImgWidth-1, refX+j+2))));
				int spie = 
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i+1))*iImgWidth + max(0, min(iImgWidth-1, refX+j-1))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i+1))*iImgWidth + max(0, min(iImgWidth-1, refX+j  ))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i+1))*iImgWidth + max(0, min(iImgWidth-1, refX+j+1))))
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i+1))*iImgWidth + max(0, min(iImgWidth-1, refX+j+2))));
				int ddpie = 
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i+2))*iImgWidth + max(0, min(iImgWidth-1, refX+j-1))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i+2))*iImgWidth + max(0, min(iImgWidth-1, refX+j  ))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i+2))*iImgWidth + max(0, min(iImgWidth-1, refX+j+1))))
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i+2))*iImgWidth + max(0, min(iImgWidth-1, refX+j+2))));
				int xxpie = 
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i+3))*iImgWidth + max(0, min(iImgWidth-1, refX+j-1))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i+3))*iImgWidth + max(0, min(iImgWidth-1, refX+j  ))))
					+ 5* (*(pbRefY + max(0, min(iImgHeight-1, refY+i+3))*iImgWidth + max(0, min(iImgWidth-1, refX+j+1))))
					-    (*(pbRefY + max(0, min(iImgHeight-1, refY+i+3))*iImgWidth + max(0, min(iImgWidth-1, refX+j+2))));
				int jpie = -aapie + 5*bpie + 5*spie - ddpie;
				pPred[i*8+j] = Clip((jpie+Ipiepie+64)>>7);
			}
		}
	} 
}

/*
 * Function: 16x16 chroma block prediction.
 */
void GetChromaBlock16x16(
		const AVS_BYTE* pbRefUV, 
		AVS_INT iUVx, AVS_INT iUVy, 
		AVS_INT iUVWidth, AVS_INT iUVHeight, 
		MOTIONVECTOR* pMv, AVS_BYTE* pPred)
{
	AVS_INT dx = (iUVx*8 + pMv->x)&7;
	AVS_INT dy = (iUVy*8 + pMv->y)&7;

	AVS_INT refX = (iUVx*8 + pMv->x - dx)/8;
	AVS_INT refY = (iUVy*8 + pMv->y - dy)/8;

	AVS_INT DX = 8-dx;
	AVS_INT DY = 8-dy;

	AVS_INT DXDY = DX*DY;
	AVS_INT dxDY = dx*DY;
	AVS_INT DXdy = DX*dy;
	AVS_INT dxdy = dx*dy;

	AVS_INT i,j;

	for(i=0; i<8; i++)
	{
		for(j=0; j<8; j++)
		{
			pPred[i*8+j] = Clip((DXDY*pbRefUV[max(0, min(refY+i,iUVHeight-1))*iUVWidth + max(0, min(refX+j,iUVWidth-1))]
						+ dxDY*pbRefUV[max(0, min(refY+i,iUVHeight-1))*iUVWidth + max(0, min(refX+j+1,iUVWidth-1))] 
						+ DXdy*pbRefUV[max(0, min(refY+i+1,iUVHeight-1))*iUVWidth + max(0, min(refX+j,iUVWidth-1))]
						+ dxdy*pbRefUV[max(0, min(refY+i+1,iUVHeight-1))*iUVWidth + max(0, min(refX+j+1,iUVWidth-1))] + 32)>>6);
		}
	}
}

/*
 * Function: 16x8 chroma block prediction.
 */
void GetChromaBlock16x8(const AVS_BYTE* pbRefUV, AVS_INT iUVx, AVS_INT iUVy, AVS_INT iUVWidth, AVS_INT iUVHeight, MOTIONVECTOR* pMv, AVS_BYTE* pPred)
{
	AVS_INT dx = (iUVx*8 + pMv->x)&7;
	AVS_INT dy = (iUVy*8 + pMv->y)&7;

	AVS_INT refX = (iUVx*8 + pMv->x - dx)/8;
	AVS_INT refY = (iUVy*8 + pMv->y - dy)/8;

	AVS_INT DX = 8-dx;
	AVS_INT DY = 8-dy;

	AVS_INT DXDY = DX*DY;
	AVS_INT dxDY = dx*DY;
	AVS_INT DXdy = DX*dy;
	AVS_INT dxdy = dx*dy;

	AVS_INT i,j;

	for(i=0; i<4; i++)
	{
		for(j=0; j<8; j++)
		{
			pPred[i*8+j] = Clip((DXDY*pbRefUV[max(0, min(refY+i,iUVHeight-1))*iUVWidth + max(0, min(refX+j,iUVWidth-1))]
						+ dxDY*pbRefUV[max(0, min(refY+i,iUVHeight-1))*iUVWidth + max(0, min(refX+j+1,iUVWidth-1))] 
						+ DXdy*pbRefUV[max(0, min(refY+i+1,iUVHeight-1))*iUVWidth + max(0, min(refX+j,iUVWidth-1))]
						+ dxdy*pbRefUV[max(0, min(refY+i+1,iUVHeight-1))*iUVWidth + max(0, min(refX+j+1,iUVWidth-1))] + 32)>>6);
		}
	}
}

/*
 * Function: 8x16 chroma block prediction.
 */
void GetChromaBlock8x16(const AVS_BYTE* pbRefUV, AVS_INT iUVx, AVS_INT iUVy, AVS_INT iUVWidth, AVS_INT iUVHeight, MOTIONVECTOR* pMv, AVS_BYTE* pPred)
{
	AVS_INT dx = (iUVx*8 + pMv->x)&7;
	AVS_INT dy = (iUVy*8 + pMv->y)&7;

	AVS_INT refX = (iUVx*8 + pMv->x - dx)/8;
	AVS_INT refY = (iUVy*8 + pMv->y - dy)/8;

	AVS_INT DX = 8-dx;
	AVS_INT DY = 8-dy;

	AVS_INT DXDY = DX*DY;
	AVS_INT dxDY = dx*DY;
	AVS_INT DXdy = DX*dy;
	AVS_INT dxdy = dx*dy;

	AVS_INT i,j;

	for(i=0; i<8; i++)
	{
		for(j=0; j<4; j++)
		{
			pPred[i*8+j] = Clip((DXDY*pbRefUV[max(0, min(refY+i,iUVHeight-1))*iUVWidth + max(0, min(refX+j,iUVWidth-1))]
						+ dxDY*pbRefUV[max(0, min(refY+i,iUVHeight-1))*iUVWidth + max(0, min(refX+j+1,iUVWidth-1))] 
						+ DXdy*pbRefUV[max(0, min(refY+i+1,iUVHeight-1))*iUVWidth + max(0, min(refX+j,iUVWidth-1))]
						+ dxdy*pbRefUV[max(0, min(refY+i+1,iUVHeight-1))*iUVWidth + max(0, min(refX+j+1,iUVWidth-1))] + 32)>>6);
		}
	}
}

/*
 * Function: 8x8 chroma block prediction.
 */
void GetChromaBlock8x8(const AVS_BYTE* pbRefUV, AVS_INT iUVx, AVS_INT iUVy, AVS_INT iUVWidth, AVS_INT iUVHeight, MOTIONVECTOR* pMv, AVS_BYTE* pPred)
{
	AVS_INT dx = (iUVx*8 + pMv->x)&7;
	AVS_INT dy = (iUVy*8 + pMv->y)&7;

	AVS_INT refX = (iUVx*8 + pMv->x - dx)/8;
	AVS_INT refY = (iUVy*8 + pMv->y - dy)/8;

	AVS_INT DX = 8-dx;
	AVS_INT DY = 8-dy;

	AVS_INT DXDY = DX*DY;
	AVS_INT dxDY = dx*DY;
	AVS_INT DXdy = DX*dy;
	AVS_INT dxdy = dx*dy;

	AVS_INT i,j;

	for(i=0; i<4; i++)
	{
		for(j=0; j<4; j++)
		{
			pPred[i*8+j] = Clip((DXDY*pbRefUV[max(0, min(refY+i,iUVHeight-1))*iUVWidth + max(0, min(refX+j,iUVWidth-1))]
						+ dxDY*pbRefUV[max(0, min(refY+i,iUVHeight-1))*iUVWidth + max(0, min(refX+j+1,iUVWidth-1))] 
						+ DXdy*pbRefUV[max(0, min(refY+i+1,iUVHeight-1))*iUVWidth + max(0, min(refX+j,iUVWidth-1))]
						+ dxdy*pbRefUV[max(0, min(refY+i+1,iUVHeight-1))*iUVWidth + max(0, min(refX+j+1,iUVWidth-1))] + 32)>>6);
		}
	}
}

/**
 * Function: Reconstruct 8x8 block.
 * －重构图像值等于预测值加上残差系数
 */
void ReconB8(AVS_BYTE* pDst, AVS_INT iImgWidth, AVS_BYTE* pPred, AVS_SHORT* pCoef)
{
	AVS_BYTE* pDst1 = pDst;
	AVS_BYTE* pSrc0 = pPred;
	AVS_SHORT*  pSrc1 = pCoef;
	AVS_INT i,j;

	for(i=0; i<8; i++)
	{
		for(j=0; j<8; j++)
		{
			int val = (*pSrc0++) + (*pSrc1++);
			*(pDst1++) = Clip(val);
		}
		pDst1 += iImgWidth-8;
	}
}

/**
 * Function: Copy single 8x8 block.
 * 主要用于cbp为0，对应块没有残差值的时候，直接拷贝预测值到重构图像中
 */
void CopyB8(AVS_BYTE* pDst, AVS_INT iStride, AVS_BYTE* pSrc)
{
	AVS_INT i;
	for (i = 0; i < 8; i++) {
		*((AVS_INT *) pDst + 0) = *((AVS_INT *) pSrc);
		*((AVS_INT *) pDst + 1) = *((AVS_INT *) pSrc + 1);
		pSrc += 8;
		pDst += iStride;
	}
}

/************************************************************************/
/* 函数功能：求平均值                                                   */
/*        － 主要是用与双向预测中求前后向预测值的平均值，也可以用做求   */
/*           其他的平均值                                               */
/*                                                                      */
/*        － C语言代码课支持各种块形状，MMX代码只支持8x32, 8x8, 8x4,    */
/*           4x8, 4x4块形状                                             */
/************************************************************************/
void AverageB8x(AVS_BYTE* pDst, AVS_BYTE* pSrc, AVS_INT iHeight, AVS_INT iWidth)
{
	AVS_INT i, j;
	for (i = 0; i < iHeight; i++) {
		for (j = 0; j < iWidth; j++) {
			*(pDst) = (*(pDst) + *(pSrc) + 1) / 2;
			pDst++;
			pSrc++;
		}
		pSrc += 8 - iWidth;
		pDst += 8 - iWidth;
	}
}