/*
 * The simplest AC3 encoder
 * Copyright (c) 2000 Gerard Lantau.
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program 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 General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 */
#include <stdlib.h>
#include <stdio.h>
#include <netinet/in.h>
#include <math.h>
#include "avcodec.h"

#include "ac3enc.h"
#include "ac3tab.h"

//#define DEBUG
//#define DEBUG_BITALLOC
#define NDEBUG
#include <assert.h>

#define MDCT_NBITS 9
#define N         (1 << MDCT_NBITS)
#define NB_BLOCKS 6 /* number of PCM blocks inside an AC3 frame */

/* new exponents are sent if their Norm 1 exceed this number */
#define EXP_DIFF_THRESHOLD 1000

/* exponent encoding strategy */
#define EXP_REUSE 0
#define EXP_NEW   1

#define EXP_D15   1
#define EXP_D25   2
#define EXP_D45   3

static void fft_init(int ln);
static void ac3_crc_init(void);

static inline INT16 fix15(float a)
{
    int v;
    v = (int)(a * (float)(1 << 15));
    if (v < -32767)
        v = -32767;
    else if (v > 32767) 
        v = 32767;
    return v;
}

static inline int calc_lowcomp1(int a, int b0, int b1)
{
    if ((b0 + 256) == b1) {
        a = 384 ;
    } else if (b0 > b1) { 
        a = a - 64;
        if (a < 0) a=0;
    }
    return a;
}

static inline int calc_lowcomp(int a, int b0, int b1, int bin)
{
    if (bin < 7) {
        if ((b0 + 256) == b1) {
            a = 384 ;
        } else if (b0 > b1) { 
            a = a - 64;
            if (a < 0) a=0;
        }
    } else if (bin < 20) {
        if ((b0 + 256) == b1) {
            a = 320 ;
        } else if (b0 > b1) {
            a= a - 64;
            if (a < 0) a=0;
        }
    } else {
        a = a - 128;
        if (a < 0) a=0;
    }
    return a;
}

/* AC3 bit allocation. The algorithm is the one described in the AC3
   spec with some optimizations because of our simplified encoding
   assumptions. */
void parametric_bit_allocation(AC3EncodeContext *s, UINT8 *bap,
                               INT8 *exp, int start, int end,
                               int snroffset, int fgain)
{
    int bin,i,j,k,end1,v,v1,bndstrt,bndend,lowcomp,begin;
    int fastleak,slowleak,address,tmp;
    INT16 psd[256]; /* scaled exponents */
    INT16 bndpsd[50]; /* interpolated exponents */
    INT16 excite[50]; /* excitation */
    INT16 mask[50];   /* masking value */

    /* exponent mapping to PSD */
    for(bin=start;bin<end;bin++) {
        psd[bin]=(3072 - (exp[bin] << 7));
    }

    /* PSD integration */
    j=start;
    k=masktab[start];
    do {
        v=psd[j];
        j++;
        end1=bndtab[k+1];
        if (end1 > end) end1=end;
        for(i=j;i<end1;i++) {
            int c,adr;
            /* logadd */
            v1=psd[j];
            c=v-v1;
            if (c >= 0) {
                adr=c >> 1;
                if (adr > 255) adr=255;
                v=v + latab[adr];
            } else {
                adr=(-c) >> 1;
                if (adr > 255) adr=255;
                v=v1 + latab[adr];
            }
            j++;
        }
        bndpsd[k]=v;
        k++;
    } while (end > bndtab[k]);

    /* excitation function */
    bndstrt = masktab[start];
    bndend = masktab[end-1] + 1;
    
    lowcomp = 0;
    lowcomp = calc_lowcomp1(lowcomp, bndpsd[0], bndpsd[1]) ;
    excite[0] = bndpsd[0] - fgain - lowcomp ;
    lowcomp = calc_lowcomp1(lowcomp, bndpsd[1], bndpsd[2]) ;
    excite[1] = bndpsd[1] - fgain - lowcomp ;
    begin = 7 ;
    for (bin = 2; bin < 7; bin++) {
        lowcomp = calc_lowcomp1(lowcomp, bndpsd[bin], bndpsd[bin+1]) ;
        fastleak = bndpsd[bin] - fgain ;
        slowleak = bndpsd[bin] - s->sgain ;
        excite[bin] = fastleak - lowcomp ;
        if (bndpsd[bin] <= bndpsd[bin+1]) {
            begin = bin + 1 ;
            break ;
        }
    }
    
    end1=bndend;
    if (end1 > 22) end1=22;
    
    for (bin = begin; bin < end1; bin++) {
        lowcomp = calc_lowcomp(lowcomp, bndpsd[bin], bndpsd[bin+1], bin) ;
        
        fastleak -= s->fdecay ;
        v = bndpsd[bin] - fgain;
        if (fastleak < v) fastleak = v;
        
        slowleak -= s->sdecay ;
        v = bndpsd[bin] - s->sgain;
        if (slowleak < v) slowleak = v;
        
        v=fastleak - lowcomp;
        if (slowleak > v) v=slowleak;
        
        excite[bin] = v;
    }

    for (bin = 22; bin < bndend; bin++) {
        fastleak -= s->fdecay ;
        v = bndpsd[bin] - fgain;
        if (fastleak < v) fastleak = v;
        slowleak -= s->sdecay ;
        v = bndpsd[bin] - s->sgain;
        if (slowleak < v) slowleak = v;

        v=fastleak;
        if (slowleak > v) v = slowleak;
        excite[bin] = v;
    }

    /* compute masking curve */

    for (bin = bndstrt; bin < bndend; bin++) {
        v1 = excite[bin];
        tmp = s->dbknee - bndpsd[bin];
        if (tmp > 0) {
            v1 += tmp >> 2;
        }
        v=hth[bin >> s->halfratecod][s->fscod];
        if (v1 > v) v=v1;
        mask[bin] = v;
    }

    /* compute bit allocation */
    
    i = start ;
    j = masktab[start] ;
    do {
        v=mask[j];
        v -= snroffset ;
        v -= s->floor ;
        if (v < 0) v = 0;
        v &= 0x1fe0 ;
        v += s->floor ;

        end1=bndtab[j] + bndsz[j];
        if (end1 > end) end1=end;

        for (k = i; k < end1; k++) {
            address = (psd[i] - v) >> 5 ;
            if (address < 0) address=0;
            else if (address > 63) address=63;
            bap[i] = baptab[address];
            i++;
        }
    } while (end > bndtab[j++]) ;
}

typedef struct IComplex {
    short re,im;
} IComplex;

static void fft_init(int ln)
{
    int i, j, m, n;
    float alpha;

    n = 1 << ln;

    for(i=0;i<(n/2);i++) {
        alpha = 2 * M_PI * (float)i / (float)n;
        costab[i] = fix15(cos(alpha));
        sintab[i] = fix15(sin(alpha));
    }

    for(i=0;i<n;i++) {
        m=0;
        for(j=0;j<ln;j++) {
            m |= ((i >> j) & 1) << (ln-j-1);
        }
        fft_rev[i]=m;
    }
}

/* butter fly op */
#define BF(pre, pim, qre, qim, pre1, pim1, qre1, qim1) \
{\
  int ax, ay, bx, by;\
  bx=pre1;\
  by=pim1;\
  ax=qre1;\
  ay=qim1;\
  pre = (bx + ax) >> 1;\
  pim = (by + ay) >> 1;\
  qre = (bx - ax) >> 1;\
  qim = (by - ay) >> 1;\
}

#define MUL16(a,b) ((a) * (b))

#define CMUL(pre, pim, are, aim, bre, bim) \
{\
   pre = (MUL16(are, bre) - MUL16(aim, bim)) >> 15;\
   pim = (MUL16(are, bim) + MUL16(bre, aim)) >> 15;\
}


/* do a 2^n point complex fft on 2^ln points. */
static void fft(IComplex *z, int ln)
{
    int	j, l, np, np2;
    int	nblocks, nloops;
    register IComplex *p,*q;
    int tmp_re, tmp_im;

    np = 1 << ln;

    /* reverse */
    for(j=0;j<np;j++) {
        int k;
        IComplex tmp;
        k = fft_rev[j];
        if (k < j) {
            tmp = z[k];
            z[k] = z[j];
            z[j] = tmp;
        }
    }

    /* pass 0 */

    p=&z[0];
    j=(np >> 1);
    do {
        BF(p[0].re, p[0].im, p[1].re, p[1].im, 
           p[0].re, p[0].im, p[1].re, p[1].im);
        p+=2;
    } while (--j != 0);

    /* pass 1 */

    p=&z[0];
    j=np >> 2;
    do {
        BF(p[0].re, p[0].im, p[2].re, p[2].im, 
           p[0].re, p[0].im, p[2].re, p[2].im);
        BF(p[1].re, p[1].im, p[3].re, p[3].im, 
           p[1].re, p[1].im, p[3].im, -p[3].re);
        p+=4;
    } while (--j != 0);

    /* pass 2 .. ln-1 */

    nblocks = np >> 3;
    nloops = 1 << 2;
    np2 = np >> 1;
    do {
        p = z;
        q = z + nloops;
        for (j = 0; j < nblocks; ++j) {

            BF(p->re, p->im, q->re, q->im,
               p->re, p->im, q->re, q->im);
            
            p++;
            q++;
            for(l = nblocks; l < np2; l += nblocks) {
                CMUL(tmp_re, tmp_im, costab[l], -sintab[l], q->re, q->im);
                BF(p->re, p->im, q->re, q->im,
                   p->re, p->im, tmp_re, tmp_im);
                p++;
                q++;
            }
            p += nloops;
            q += nloops;
        }
        nblocks = nblocks >> 1;
        nloops = nloops << 1;
    } while (nblocks != 0);
}

/* do a 512 point mdct */
static void mdct512(INT32 *out, INT16 *in)
{
    int i, re, im, re1, im1;
    INT16 rot[N]; 
    IComplex x[N/4];

    /* shift to simplify computations */
    for(i=0;i<N/4;i++)
        rot[i] = -in[i + 3*N/4];
    for(i=N/4;i<N;i++)
        rot[i] = in[i - N/4];
        
    /* pre rotation */
    for(i=0;i<N/4;i++) {
        re = ((int)rot[2*i] - (int)rot[N-1-2*i]) >> 1;
        im = -((int)rot[N/2+2*i] - (int)rot[N/2-1-2*i]) >> 1;
        CMUL(x[i].re, x[i].im, re, im, -xcos1[i], xsin1[i]);
    }

    fft(x, MDCT_NBITS - 2);
  
    /* post rotation */
    for(i=0;i<N/4;i++) {
        re = x[i].re;
        im = x[i].im;
        CMUL(re1, im1, re, im, xsin1[i], xcos1[i]);
        out[2*i] = im1;
        out[N/2-1-2*i] = re1;
    }
}

/* XXX: use another norm ? */
static int calc_exp_diff(UINT8 *exp1, UINT8 *exp2, int n)
{
    int sum, i;
    sum = 0;
    for(i=0;i<n;i++) {
        sum += abs(exp1[i] - exp2[i]);
    }
    return sum;
}

static void compute_exp_strategy(UINT8 exp_strategy[NB_BLOCKS][AC3_MAX_CHANNELS],
                                 UINT8 exp[NB_BLOCKS][AC3_MAX_CHANNELS][N/2],
                                 int ch)
{
    int i, j;
    int exp_diff;
    
    /* estimate if the exponent variation & decide if they should be
       reused in the next frame */
    exp_strategy[0][ch] = EXP_NEW;
    for(i=1;i<NB_BLOCKS;i++) {
        exp_diff = calc_exp_diff(exp[i][ch], exp[i-1][ch], N/2);
#ifdef DEBUG            
        printf("exp_diff=%d\n", exp_diff);
#endif
        if (exp_diff > EXP_DIFF_THRESHOLD)
            exp_strategy[i][ch] = EXP_NEW;
        else
            exp_strategy[i][ch] = EXP_REUSE;
    }
    /* now select the encoding strategy type : if exponents are often
       recoded, we use a coarse encoding */
    i = 0;
    while (i < NB_BLOCKS) {
        j = i + 1;
        while (j < NB_BLOCKS && exp_strategy[j][ch] == EXP_REUSE)
            j++;
        switch(j - i) {
        case 1:
            exp_strategy[i][ch] = EXP_D45;
            break;
        case 2:
        case 3:
            exp_strategy[i][ch] = EXP_D25;
            break;
        default:
            exp_strategy[i][ch] = EXP_D15;
            break;
        }
        i = j;
    }
}

/* set exp[i] to min(exp[i], exp1[i]) */
static void exponent_min(UINT8 exp[N/2], UINT8 exp1[N/2], int n)
{
    int i;

    for(i=0;i<n;i++) {
        if (exp1[i] < exp[i])
            exp[i] = exp1[i];
    }
}
                                 
/* update the exponents so that they are the ones the decoder will
   decode. Return the number of bits used to code the exponents */
static int encode_exp(UINT8 encoded_exp[N/2], 
                      UINT8 exp[N/2], 
                      int nb_exps,
                      int exp_strategy)
{
    int group_size, nb_groups, i, j, k, recurse, exp_min, delta;
    UINT8 exp1[N/2];

    switch(exp_strategy) {
    case EXP_D15:
        group_size = 1;
        break;
    case EXP_D25:
        group_size = 2;
        break;
    default:
    case EXP_D45:
        group_size = 4;
        break;
    }
    nb_groups = ((nb_exps + (group_size * 3) - 4) / (3 * group_size)) * 3;

    /* for each group, compute the minimum exponent */
    exp1[0] = exp[0]; /* DC exponent is handled separately */
    k = 1;
    for(i=1;i<=nb_groups;i++) {
        exp_min = exp[k];
        assert(exp_min >= 0 && exp_min <= 24);
        for(j=1;j<group_size;j++) {
            if (exp[k+j] < exp_min)
                exp_min = exp[k+j];