/*
 * H.26L/H.264/AVC/JVT/14496-10/... encoder/decoder
 * Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at>
 *
 * This file is part of FFmpeg.
 *
 * FFmpeg is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * FFmpeg is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with FFmpeg; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

/**
 * @file h264pred.c
 * H.264 / AVC / MPEG4 part10 prediction functions.
 * @author Michael Niedermayer <michaelni@gmx.at>
 */

#include <stdio.h>
#include <assert.h>
#include "mem.h"
#include "common.h"
#include "bswap.h"
#include "intreadwrite.h"
#include "eval.h"
#include "rational.h"
#include "log.h"
#include "avutil.h"
#include "avcodec.h"
#include "dsputil.h"
#include "ratecontrol.h"
#include "parser.h"
#include "bitstream.h"
#include "rl.h"
#include "mpegvideo.h"
#include "h264pred.h"

static void pred4x4_vertical_c(uint8_t *src, uint8_t *topright, int stride){
    const uint32_t a= ((uint32_t*)(src-stride))[0];
    ((uint32_t*)(src+0*stride))[0]= a;
    ((uint32_t*)(src+1*stride))[0]= a;
    ((uint32_t*)(src+2*stride))[0]= a;
    ((uint32_t*)(src+3*stride))[0]= a;
}

static void pred4x4_horizontal_c(uint8_t *src, uint8_t *topright, int stride){
    ((uint32_t*)(src+0*stride))[0]= src[-1+0*stride]*0x01010101;
    ((uint32_t*)(src+1*stride))[0]= src[-1+1*stride]*0x01010101;
    ((uint32_t*)(src+2*stride))[0]= src[-1+2*stride]*0x01010101;
    ((uint32_t*)(src+3*stride))[0]= src[-1+3*stride]*0x01010101;
}

static void pred4x4_dc_c(uint8_t *src, uint8_t *topright, int stride){
    const int dc= (  src[-stride] + src[1-stride] + src[2-stride] + src[3-stride]
                   + src[-1+0*stride] + src[-1+1*stride] + src[-1+2*stride] + src[-1+3*stride] + 4) >>3;

    ((uint32_t*)(src+0*stride))[0]=
    ((uint32_t*)(src+1*stride))[0]=
    ((uint32_t*)(src+2*stride))[0]=
    ((uint32_t*)(src+3*stride))[0]= dc* 0x01010101;
}

static void pred4x4_left_dc_c(uint8_t *src, uint8_t *topright, int stride){
    const int dc= (  src[-1+0*stride] + src[-1+1*stride] + src[-1+2*stride] + src[-1+3*stride] + 2) >>2;

    ((uint32_t*)(src+0*stride))[0]=
    ((uint32_t*)(src+1*stride))[0]=
    ((uint32_t*)(src+2*stride))[0]=
    ((uint32_t*)(src+3*stride))[0]= dc* 0x01010101;
}

static void pred4x4_top_dc_c(uint8_t *src, uint8_t *topright, int stride){
    const int dc= (  src[-stride] + src[1-stride] + src[2-stride] + src[3-stride] + 2) >>2;

    ((uint32_t*)(src+0*stride))[0]=
    ((uint32_t*)(src+1*stride))[0]=
    ((uint32_t*)(src+2*stride))[0]=
    ((uint32_t*)(src+3*stride))[0]= dc* 0x01010101;
}

static void pred4x4_128_dc_c(uint8_t *src, uint8_t *topright, int stride){
    ((uint32_t*)(src+0*stride))[0]=
    ((uint32_t*)(src+1*stride))[0]=
    ((uint32_t*)(src+2*stride))[0]=
    ((uint32_t*)(src+3*stride))[0]= 128U*0x01010101U;
}


#define LOAD_TOP_RIGHT_EDGE\
    const int av_unused t4= topright[0];\
    const int av_unused t5= topright[1];\
    const int av_unused t6= topright[2];\
    const int av_unused t7= topright[3];\

#define LOAD_DOWN_LEFT_EDGE\
    const int av_unused l4= src[-1+4*stride];\
    const int av_unused l5= src[-1+5*stride];\
    const int av_unused l6= src[-1+6*stride];\
    const int av_unused l7= src[-1+7*stride];\

#define LOAD_LEFT_EDGE\
    const int av_unused l0= src[-1+0*stride];\
    const int av_unused l1= src[-1+1*stride];\
    const int av_unused l2= src[-1+2*stride];\
    const int av_unused l3= src[-1+3*stride];\

#define LOAD_TOP_EDGE\
    const int av_unused t0= src[ 0-1*stride];\
    const int av_unused t1= src[ 1-1*stride];\
    const int av_unused t2= src[ 2-1*stride];\
    const int av_unused t3= src[ 3-1*stride];\

static void pred4x4_down_right_c(uint8_t *src, uint8_t *topright, int stride){
    const int lt= src[-1-1*stride];
    LOAD_TOP_EDGE
    LOAD_LEFT_EDGE

    src[0+3*stride]=(l3 + 2*l2 + l1 + 2)>>2;
    src[0+2*stride]=
    src[1+3*stride]=(l2 + 2*l1 + l0 + 2)>>2;
    src[0+1*stride]=
    src[1+2*stride]=
    src[2+3*stride]=(l1 + 2*l0 + lt + 2)>>2;
    src[0+0*stride]=
    src[1+1*stride]=
    src[2+2*stride]=
    src[3+3*stride]=(l0 + 2*lt + t0 + 2)>>2;
    src[1+0*stride]=
    src[2+1*stride]=
    src[3+2*stride]=(lt + 2*t0 + t1 + 2)>>2;
    src[2+0*stride]=
    src[3+1*stride]=(t0 + 2*t1 + t2 + 2)>>2;
    src[3+0*stride]=(t1 + 2*t2 + t3 + 2)>>2;
}

static void pred4x4_down_left_c(uint8_t *src, uint8_t *topright, int stride){
    LOAD_TOP_EDGE
    LOAD_TOP_RIGHT_EDGE
//    LOAD_LEFT_EDGE

    src[0+0*stride]=(t0 + t2 + 2*t1 + 2)>>2;
    src[1+0*stride]=
    src[0+1*stride]=(t1 + t3 + 2*t2 + 2)>>2;
    src[2+0*stride]=
    src[1+1*stride]=
    src[0+2*stride]=(t2 + t4 + 2*t3 + 2)>>2;
    src[3+0*stride]=
    src[2+1*stride]=
    src[1+2*stride]=
    src[0+3*stride]=(t3 + t5 + 2*t4 + 2)>>2;
    src[3+1*stride]=
    src[2+2*stride]=
    src[1+3*stride]=(t4 + t6 + 2*t5 + 2)>>2;
    src[3+2*stride]=
    src[2+3*stride]=(t5 + t7 + 2*t6 + 2)>>2;
    src[3+3*stride]=(t6 + 3*t7 + 2)>>2;
}

static void pred4x4_down_left_svq3_c(uint8_t *src, uint8_t *topright, int stride){
    LOAD_TOP_EDGE
    LOAD_LEFT_EDGE
    //const av_unused int unu0= t0;
    //const av_unused int unu1= l0;

    src[0+0*stride]=(l1 + t1)>>1;
    src[1+0*stride]=
    src[0+1*stride]=(l2 + t2)>>1;
    src[2+0*stride]=
    src[1+1*stride]=
    src[0+2*stride]=
    src[3+0*stride]=
    src[2+1*stride]=
    src[1+2*stride]=
    src[0+3*stride]=
    src[3+1*stride]=
    src[2+2*stride]=
    src[1+3*stride]=
    src[3+2*stride]=
    src[2+3*stride]=
    src[3+3*stride]=(l3 + t3)>>1;
}

static void pred4x4_down_left_rv40_c(uint8_t *src, uint8_t *topright, int stride){
    LOAD_TOP_EDGE
    LOAD_TOP_RIGHT_EDGE
    LOAD_LEFT_EDGE
    LOAD_DOWN_LEFT_EDGE

    src[0+0*stride]=(t0 + t2 + 2*t1 + 2 + l0 + l2 + 2*l1 + 2)>>3;
    src[1+0*stride]=
    src[0+1*stride]=(t1 + t3 + 2*t2 + 2 + l1 + l3 + 2*l2 + 2)>>3;
    src[2+0*stride]=
    src[1+1*stride]=
    src[0+2*stride]=(t2 + t4 + 2*t3 + 2 + l2 + l4 + 2*l3 + 2)>>3;
    src[3+0*stride]=
    src[2+1*stride]=
    src[1+2*stride]=
    src[0+3*stride]=(t3 + t5 + 2*t4 + 2 + l3 + l5 + 2*l4 + 2)>>3;
    src[3+1*stride]=
    src[2+2*stride]=
    src[1+3*stride]=(t4 + t6 + 2*t5 + 2 + l4 + l6 + 2*l5 + 2)>>3;
    src[3+2*stride]=
    src[2+3*stride]=(t5 + t7 + 2*t6 + 2 + l5 + l7 + 2*l6 + 2)>>3;
    src[3+3*stride]=(t6 + t7 + 1 + l6 + l7 + 1)>>2;
}

/*static void pred4x4_down_left_rv40_notop_c(uint8_t *src, uint8_t *topright, int stride){
    LOAD_LEFT_EDGE
    LOAD_DOWN_LEFT_EDGE

    src[0+0*stride]=(l0 + l2 + 2*l1 + 2)>>2;
    src[1+0*stride]=
    src[0+1*stride]=(l1 + l3 + 2*l2 + 2)>>2;
    src[2+0*stride]=
    src[1+1*stride]=
    src[0+2*stride]=(l2 + l4 + 2*l3 + 2)>>2;
    src[3+0*stride]=
    src[2+1*stride]=
    src[1+2*stride]=
    src[0+3*stride]=(l3 + l5 + 2*l4 + 2)>>2;
    src[3+1*stride]=
    src[2+2*stride]=
    src[1+3*stride]=(l4 + l6 + 2*l5 + 2)>>2;
    src[3+2*stride]=
    src[2+3*stride]=(l5 + l7 + 2*l6 + 2)>>2;
    src[3+3*stride]=(l6 + l7 + 1)>>1;
}*/

static void pred4x4_down_left_rv40_nodown_c(uint8_t *src, uint8_t *topright, int stride){
    LOAD_TOP_EDGE
    LOAD_TOP_RIGHT_EDGE
    LOAD_LEFT_EDGE

    src[0+0*stride]=(t0 + t2 + 2*t1 + 2 + l0 + l2 + 2*l1 + 2)>>3;
    src[1+0*stride]=
    src[0+1*stride]=(t1 + t3 + 2*t2 + 2 + l1 + l3 + 2*l2 + 2)>>3;
    src[2+0*stride]=
    src[1+1*stride]=
    src[0+2*stride]=(t2 + t4 + 2*t3 + 2 + l2 + 3*l3 + 2)>>3;
    src[3+0*stride]=
    src[2+1*stride]=
    src[1+2*stride]=
    src[0+3*stride]=(t3 + t5 + 2*t4 + 2 + l3*4 + 2)>>3;
    src[3+1*stride]=
    src[2+2*stride]=
    src[1+3*stride]=(t4 + t6 + 2*t5 + 2 + l3*4 + 2)>>3;
    src[3+2*stride]=
    src[2+3*stride]=(t5 + t7 + 2*t6 + 2 + l3*4 + 2)>>3;
    src[3+3*stride]=(t6 + t7 + 1 + 2*l3 + 1)>>2;
}

static void pred4x4_vertical_right_c(uint8_t *src, uint8_t *topright, int stride){
    const int lt= src[-1-1*stride];
    LOAD_TOP_EDGE
    LOAD_LEFT_EDGE

    src[0+0*stride]=
    src[1+2*stride]=(lt + t0 + 1)>>1;
    src[1+0*stride]=
    src[2+2*stride]=(t0 + t1 + 1)>>1;
    src[2+0*stride]=
    src[3+2*stride]=(t1 + t2 + 1)>>1;
    src[3+0*stride]=(t2 + t3 + 1)>>1;
    src[0+1*stride]=
    src[1+3*stride]=(l0 + 2*lt + t0 + 2)>>2;
    src[1+1*stride]=
    src[2+3*stride]=(lt + 2*t0 + t1 + 2)>>2;
    src[2+1*stride]=
    src[3+3*stride]=(t0 + 2*t1 + t2 + 2)>>2;
    src[3+1*stride]=(t1 + 2*t2 + t3 + 2)>>2;
    src[0+2*stride]=(lt + 2*l0 + l1 + 2)>>2;
    src[0+3*stride]=(l0 + 2*l1 + l2 + 2)>>2;
}

static void pred4x4_vertical_left_c(uint8_t *src, uint8_t *topright, int stride){
    LOAD_TOP_EDGE
    LOAD_TOP_RIGHT_EDGE

    src[0+0*stride]=(t0 + t1 + 1)>>1;
    src[1+0*stride]=
    src[0+2*stride]=(t1 + t2 + 1)>>1;
    src[2+0*stride]=
    src[1+2*stride]=(t2 + t3 + 1)>>1;
    src[3+0*stride]=
    src[2+2*stride]=(t3 + t4+ 1)>>1;
    src[3+2*stride]=(t4 + t5+ 1)>>1;
    src[0+1*stride]=(t0 + 2*t1 + t2 + 2)>>2;
    src[1+1*stride]=
    src[0+3*stride]=(t1 + 2*t2 + t3 + 2)>>2;
    src[2+1*stride]=
    src[1+3*stride]=(t2 + 2*t3 + t4 + 2)>>2;
    src[3+1*stride]=
    src[2+3*stride]=(t3 + 2*t4 + t5 + 2)>>2;
    src[3+3*stride]=(t4 + 2*t5 + t6 + 2)>>2;
}

static void pred4x4_vertical_left_rv40(uint8_t *src, uint8_t *topright, int stride,
                                      const int l0, const int l1, const int l2, const int l3, const int l4){
    LOAD_TOP_EDGE
    LOAD_TOP_RIGHT_EDGE

    src[0+0*stride]=(2*t0 + 2*t1 + l1 + 2*l2 + l3 + 4)>>3;
    src[1+0*stride]=
    src[0+2*stride]=(t1 + t2 + 1)>>1;
    src[2+0*stride]=
    src[1+2*stride]=(t2 + t3 + 1)>>1;
    src[3+0*stride]=
    src[2+2*stride]=(t3 + t4+ 1)>>1;
    src[3+2*stride]=(t4 + t5+ 1)>>1;
    src[0+1*stride]=(t0 + 2*t1 + t2 + l2 + 2*l3 + l4 + 4)>>3;
    src[1+1*stride]=
    src[0+3*stride]=(t1 + 2*t2 + t3 + 2)>>2;
    src[2+1*stride]=
    src[1+3*stride]=(t2 + 2*t3 + t4 + 2)>>2;
    src[3+1*stride]=
    src[2+3*stride]=(t3 + 2*t4 + t5 + 2)>>2;
    src[3+3*stride]=(t4 + 2*t5 + t6 + 2)>>2;
}

static void pred4x4_vertical_left_rv40_c(uint8_t *src, uint8_t *topright, int stride){
    LOAD_LEFT_EDGE
    LOAD_DOWN_LEFT_EDGE

    pred4x4_vertical_left_rv40(src, topright, stride, l0, l1, l2, l3, l4);
}

static void pred4x4_vertical_left_rv40_nodown_c(uint8_t *src, uint8_t *topright, int stride){
    LOAD_LEFT_EDGE

    pred4x4_vertical_left_rv40(src, topright, stride, l0, l1, l2, l3, l3);
}

static void pred4x4_horizontal_up_c(uint8_t *src, uint8_t *topright, int stride){
    LOAD_LEFT_EDGE

    src[0+0*stride]=(l0 + l1 + 1)>>1;
    src[1+0*stride]=(l0 + 2*l1 + l2 + 2)>>2;
    src[2+0*stride]=
    src[0+1*stride]=(l1 + l2 + 1)>>1;
    src[3+0*stride]=
    src[1+1*stride]=(l1 + 2*l2 + l3 + 2)>>2;
    src[2+1*stride]=
    src[0+2*stride]=(l2 + l3 + 1)>>1;
    src[3+1*stride]=
    src[1+2*stride]=(l2 + 2*l3 + l3 + 2)>>2;
    src[3+2*stride]=
    src[1+3*stride]=
    src[0+3*stride]=
    src[2+2*stride]=
    src[2+3*stride]=
    src[3+3*stride]=l3;
}

static void pred4x4_horizontal_up_rv40_c(uint8_t *src, uint8_t *topright, int stride){