yuv2yuv.c

linux下的一款播放器

            {
                /* process top line (if chroma is not pairable): */
                if (dest_y & 1) {
                    AltiVec_lineI420toYUY2 (sy, su, sv, d, dest_dx);
                    sy += yPitch;
                    su += uPitch;
                    sv += vPitch;
                    d  += dest_pitch;
                    dest_dy --;
                }
                /* the main loop (processes two lines a time): */
                for (i = 0; i < dest_dy/2; i ++) {
                    AltiVec_lineI420toYUY2 (sy, su, sv, d, dest_dx);
                    sy += yPitch;
                    d  += dest_pitch;
                    AltiVec_lineI420toYUY2 (sy, su, sv, d, dest_dx);
                    sy += yPitch;
                    su += uPitch;
                    sv += vPitch;
                    d  += dest_pitch;
                }
                /* process the last line (if dy is odd): */
                if (dest_dy & 1) {
                    AltiVec_lineI420toYUY2 (sy, su, sv, d, dest_dx);
                }
                /* end of MMX code */
            } else
#endif
            /* a generic "C" version: */
            {
                /* process top line (if chroma is not pairable): */
                if (dest_y & 1) {
                    for (j = 0; j < dest_dx/2; j ++)
                        *(unsigned int *)(d + j*4) = COMBINE(sy[j*2], su[j], sy[j*2+1], sv[j]);
                    sy += yPitch;
                    su += uPitch;
                    sv += vPitch;
                    d  += dest_pitch;
                    dest_dy --;
                }
                /* the main loop (processes two lines a time): */
                for (i = 0; i < dest_dy/2; i ++) {
                    for (j = 0; j < dest_dx/2; j ++)
                        *(unsigned int *)(d + j*4) = COMBINE(sy[j*2], su[j], sy[j*2+1], sv[j]);
                    sy += yPitch;
                    d  += dest_pitch;
                    for (j = 0; j < dest_dx/2; j ++)
                        *(unsigned int *)(d + j*4) = COMBINE(sy[j*2], su[j], sy[j*2+1], sv[j]);
                    sy += yPitch;
                    su += uPitch;
                    sv += vPitch;
                    d  += dest_pitch;
                }
                /* process the last line (if dy is odd): */
                if (dest_dy & 1) {
                    for (j = 0; j < dest_dx/2; j ++)
                        *(unsigned int *)(d + j*4) = COMBINE(sy[j*2], su[j], sy[j*2+1], sv[j]);
                }
            }

        } else

        /* check if no hue adjustment: */
        if (!is_alpha) {

            /* no chroma rotation: */
            unsigned char *sy, *sv, *su, *d;
            register int i, j;

            /* get pointers: */
            sy = pY + src_x + src_y * yPitch;           /* luma offset */
            su = pU + (src_x/2 + src_y/2 * uPitch);
            sv = pV + (src_x/2 + src_y/2 * vPitch);
            d  = dest_ptr + dest_x * 2 + dest_y * dest_pitch;   /* 2 bytes/pixel */

            /* process top line (if chroma is not pairable): */
            if (dest_y & 1) {
                for (j = 0; j < dest_dx/2; j ++)
                    *(unsigned int *)(d + j*4) = COMBINE(_yytab[sy[j*2]], _uutab[su[j]], _yytab[sy[j*2+1]], _vvtab[sv[j]]);
                sy += yPitch;
                su += uPitch;
                sv += vPitch;
                d  += dest_pitch;
                dest_dy --;
            }
            /* the main loop (processes two lines a time): */
            for (i = 0; i < dest_dy/2; i ++) {
                for (j = 0; j < dest_dx/2; j ++)
                    *(unsigned int *)(d + j*4) = COMBINE(_yytab[sy[j*2]], _uutab[su[j]], _yytab[sy[j*2+1]], _vvtab[sv[j]]);
                sy += yPitch;
                d  += dest_pitch;
                for (j = 0; j < dest_dx/2; j ++)
                    *(unsigned int *)(d + j*4) = COMBINE(_yytab[sy[j*2]], _uutab[su[j]], _yytab[sy[j*2+1]], _vvtab[sv[j]]);
                sy += yPitch;
                su += uPitch;
                sv += vPitch;
                d  += dest_pitch;
            }
            /* process the last line (if dy is odd): */
            if (dest_dy & 1) {
                for (j = 0; j < dest_dx/2; j ++)
                    *(unsigned int *)(d + j*4) = COMBINE(_yytab[sy[j*2]], _uutab[su[j]], _yytab[sy[j*2+1]], _vvtab[sv[j]]);
            }

        } else {

            /* the most complex case (w. hue adjustment): */
            unsigned char *sy, *sv, *su, *d;
            register int i, j, u, v;

            /* get pointers: */
            sy = pY + src_x + src_y * yPitch;           /* luma offset */
            su = pU + (src_x/2 + src_y/2 * uPitch);
            sv = pV + (src_x/2 + src_y/2 * vPitch);
            d  = dest_ptr + dest_x * 2 + dest_y * dest_pitch;   /* 2 bytes/pixel */

            /* process top line (if chroma is not pairable): */
            if (dest_y & 1) {
                for (j = 0; j < dest_dx/2; j ++) {
                    v = _CLIP(8,_vvtab[sv[j]] + _vutab[su[j]]);
                    u = _CLIP(8,_uutab[su[j]] + _uvtab[sv[j]]);
                    *(unsigned int *)(d + j*4) = COMBINE(_yytab[sy[j*2]], u, _yytab[sy[j*2+1]], v);
                }
                sy += yPitch;
                su += uPitch;
                sv += vPitch;
                d  += dest_pitch;
                dest_dy --;
            }
            /* the main loop (processes two lines a time): */
            for (i = 0; i < dest_dy/2; i ++) {
                for (j = 0; j < dest_dx/2; j ++) {
                    v = _CLIP(8,_vvtab[sv[j]] + _vutab[su[j]]);
                    u = _CLIP(8,_uutab[su[j]] + _uvtab[sv[j]]);
                    *(unsigned int *)(d + j*4) = COMBINE(_yytab[sy[j*2]], u, _yytab[sy[j*2+1]], v);
                    *(unsigned int *)(d + j*4 + dest_pitch) = COMBINE(_yytab[sy[j*2 + yPitch]], u, _yytab[sy[j*2+1+yPitch]], v);
                }
                sy += yPitch*2;
                su += uPitch;
                sv += vPitch;
                d  += dest_pitch*2;
            }
            /* process the last line (if dy is odd): */
            if (dest_dy & 1) {
                for (j = 0; j < dest_dx/2; j ++) {
                    v = _CLIP(8,_vvtab[sv[j]] + _vutab[su[j]]);
                    u = _CLIP(8,_uutab[su[j]] + _uvtab[sv[j]]);
                    *(unsigned int *)(d + j*4) = COMBINE(_yytab[sy[j*2]], u, _yytab[sy[j*2+1]], v);
                }
            }
        }
        return 0;
    }

    /* conversion is not supported */
    return -1;
}

/*
 * I420toYUY2() converter:
 */
int I420toYUY2 (unsigned char *dest_ptr, int dest_width, int dest_height,
    int dest_pitch, int dest_x, int dest_y, int dest_dx, int dest_dy,
    unsigned char *src_ptr, int src_width, int src_height, int src_pitch,
    int src_x, int src_y, int src_dx, int src_dy)
{
#if 1
    unsigned char *pU = src_ptr+src_height*src_pitch,
                  *pV = pU + src_height*src_pitch/4;
    return I420toYUY2x(dest_ptr, dest_width, dest_height, dest_pitch, dest_x, dest_y, dest_dx, dest_dy,
                       src_ptr, pU, pV, 
                       src_width, src_height, src_pitch, src_pitch/2, src_pitch/2,
                       src_x, src_y, src_dx, src_dy);
                    

#else

    /* scale factors: */
    int scale_x, scale_y;

    /* check arguments: */
    if (!chk_args (dest_ptr, dest_width, dest_height, dest_pitch,
        dest_x, dest_y, dest_dx, dest_dy, src_ptr, src_width, src_height,
        src_pitch, src_x, src_y, src_dx, src_dy, &scale_x, &scale_y))
        return -1;

    /* remove odd destination columns: */
    if (!adjust_range (&dest_x, &dest_dx, &src_x, &src_dx, scale_x))
        return 0;

    /* check if we have misaligned input: */
    if (src_x & 1)
        return -1;                          /* can't shift chromas */

    /* check if bottop-up images: */
    if (dest_pitch < 0) dest_ptr -= (dest_height-1) * dest_pitch;
    if (src_pitch <= 0) return -1;          /* not supported */

    /* check if 1:1 scale: */
    if (scale_x == 1 && scale_y == 1) {

        /* check if no color adjustments: */
        if (!(is_alpha | is_beta | is_gamma | is_kappa)) {

            /* no color adjustments: */
            unsigned char *sy, *sv, *su, *d;
            register int i, j;

            /* get pointers: */
            sy = src_ptr + src_x + src_y * src_pitch;           /* luma offset */
            su = src_ptr + src_height * src_pitch + (src_x/2 + src_y/2 * src_pitch);
            sv = su + src_height * src_pitch / 4;
            d  = dest_ptr + dest_x * 2 + dest_y * dest_pitch;   /* 2 bytes/pixel */

#if (defined(_M_IX86) || defined(_USE_MMX_BLENDERS)) && !defined(WINCE_EMULATOR)
            /* check if we can use MMX-optimized code here: */
            if (_x86_MMX_Available)
            {
                /* process top line (if chroma is not pairable): */
                if (dest_y & 1) {
                    _MMX_lineI420toYUY2 (sy, su, sv, d, dest_dx);
                    sy += src_pitch;
                    su += src_pitch/2;
                    sv += src_pitch/2;
                    d  += dest_pitch;
                    dest_dy --;
                }
                /* the main loop (processes two lines a time): */
                for (i = 0; i < dest_dy/2; i ++) {
                    _MMX_lineI420toYUY2 (sy, su, sv, d, dest_dx);
                    sy += src_pitch;
                    d  += dest_pitch;
                    _MMX_lineI420toYUY2 (sy, su, sv, d, dest_dx);
                    sy += src_pitch;
                    su += src_pitch/2;
                    sv += src_pitch/2;
                    d  += dest_pitch;
                }
                /* process the last line (if dy is odd): */
                if (dest_dy & 1) {
                    _MMX_lineI420toYUY2 (sy, su, sv, d, dest_dx);
                }
                /* end of MMX code */
#if defined(_M_IX86)  && !defined(WINCE_EMULATOR)                
                __asm emms
#elif defined(_USE_MMX_BLENDERS) && !defined(WINCE_EMULATOR)
                __asm__ __volatile__ ( "emms" );
#endif                    

            } else
#endif
            /* a generic "C" version: */
            {
                /* process top line (if chroma is not pairable): */
                if (dest_y & 1) {
                    for (j = 0; j < dest_dx/2; j ++)
                        *(unsigned int *)(d + j*4) = COMBINE(sy[j*2], su[j], sy[j*2+1], sv[j]);
                    sy += src_pitch;
                    su += src_pitch/2;
                    sv += src_pitch/2;
                    d  += dest_pitch;
                    dest_dy --;
                }
                /* the main loop (processes two lines a time): */
                for (i = 0; i < dest_dy/2; i ++) {
                    for (j = 0; j < dest_dx/2; j ++)
                        *(unsigned int *)(d + j*4) = COMBINE(sy[j*2], su[j], sy[j*2+1], sv[j]);
                    sy += src_pitch;
                    d  += dest_pitch;
                    for (j = 0; j < dest_dx/2; j ++)
                        *(unsigned int *)(d + j*4) = COMBINE(sy[j*2], su[j], sy[j*2+1], sv[j]);
                    sy += src_pitch;
                    su += src_pitch/2;
                    sv += src_pitch/2;
                    d  += dest_pitch;
                }
                /* process the last line (if dy is odd): */
                if (dest_dy & 1) {
                    for (j = 0; j < dest_dx/2; j ++)
                        *(unsigned int *)(d + j*4) = COMBINE(sy[j*2], su[j], sy[j*2+1], sv[j]);
                }
            }

        } else

        /* check if no hue adjustment: */
        if (!is_alpha) {

            /* no chroma rotation: */
            unsigned char *sy, *sv, *su, *d;
            register int i, j;

            /* get pointers: */
            sy = src_ptr + src_x + src_y * src_pitch;           /* luma offset */
            su = src_ptr + src_height * src_pitch + (src_x/2 + src_y/2 * src_pitch);
            sv = su + src_height * src_pitch / 4;
            d  = dest_ptr + dest_x * 2 + dest_y * dest_pitch;   /* 2 bytes/pixel */

            /* process top line (if chroma is not pairable): */
            if (dest_y & 1) {
                for (j = 0; j < dest_dx/2; j ++)
                    *(unsigned int *)(d + j*4) = COMBINE(_yytab[sy[j*2]], _uutab[su[j]], _yytab[sy[j*2+1]], _vvtab[sv[j]]);
                sy += src_pitch;
                su += src_pitch/2;
                sv += src_pitch/2;
                d  += dest_pitch;
                dest_dy --;
            }
            /* the main loop (processes two lines a time): */
            for (i = 0; i < dest_dy/2; i ++) {
                for (j = 0; j < dest_dx/2; j ++)
                    *(unsigned int *)(d + j*4) = COMBINE(_yytab[sy[j*2]], _uutab[su[j]], _yytab[sy[j*2+1]], _vvtab[sv[j]]);
                sy += src_pitch;
                d  += dest_pitch;
                for (j = 0; j < dest_dx/2; j ++)
                    *(unsigned int *)(d + j*4) = COMBINE(_yytab[sy[j*2]], _uutab[su[j]], _yytab[sy[j*2+1]], _vvtab[sv[j]]);
                sy += src_pitch;
                su += src_pitch/2;
                sv += src_pitch/2;
                d  += dest_pitch;
            }
            /* process the last line (if dy is odd): */
            if (dest_dy & 1) {
                for (j = 0; j < dest_dx/2; j ++)
                    *(unsigned int *)(d + j*4) = COMBINE(_yytab[sy[j*2]], _uutab[su[j]], _yytab[sy[j*2+1]], _vvtab[sv[j]]);
            }

        } else {

            /* the most complex case (w. hue adjustment): */
            unsigned char *sy, *sv, *su, *d;
            register int i, j, u, v;

            /* get pointers: */
            sy = src_ptr + src_x + src_y * src_pitch;           /* luma offset */
            su = src_ptr + src_height * src_pitch + (src_x/2 + src_y/2 * src_pitch/2);
            sv = su + src_height * src_pitch / 4;
            d  = dest_ptr + dest_x * 2 + dest_y * dest_pitch;   /* 2 bytes/pixel */

            /* process top line (if chroma is not pairable): */
            if (dest_y & 1) {
                for (j = 0; j < dest_dx/2; j ++) {
                    v = _CLIP(8,_vvtab[sv[j]] + _vutab[su[j]]);