void bw_to_psi(FILE *out, CSA *csa, char *fbw, char *flst, int *k)
{
FILE *in;
i64 last,i,j;
i64 C2[SIGMA];
i64 c;
in = fopen(flst,"r");
if (in == NULL) {
perror("bw_to_psi:"); exit(1);
}
fscanf(in,"%ld",&last);
fclose(in);
for (c=0; c<SIGMA; c++) {
csa->C[c] = 0;
}
in = fopen(fbw,"r");
if (in == NULL) {
perror("bw_to_psi:"); exit(1);
}
csa->n = 0;
while (1) {
display_progressbar("reading ",csa->n,0L);
c = fgetc(in);
if (c == EOF) break;
csa->C[c]++;
csa->n++;
}
rewind(in);
printf("n = %ld last = %ld\n",csa->n,last);
*k = (blog(csa->n+1)+1+8-1)/8;
for (c=0; c<SIGMA; c++) {
psi[c] = open_diskbuf(out,*k);
}
for (j=1,c=0; c<SIGMA; c++) {
C2[c] = j;
j += csa->C[c];
}
for (i = 0; i<=csa->n; i++) {
display_progressbar("computing psi ",i,csa->n);
if (i == last) {
setint_diskbuf(psi[0], 0, i);
} else {
c = fgetc(in);
setint_diskbuf(psi[c], C2[c]++, i);
}
}
fclose(in);
for (c=0; c<SIGMA; c++) {
close_diskbuf(psi[c]);
}
}
void bw_to_psi(FILE *out, CSA *csa, char *fbw, char *flst, int *k)
{
FILE *in;
i64 last,i,j;
i64 *C2;
i64 c;
diskbuf **psi;
int sigma;
int k2;
in = fopen(flst,"r");
if (in == NULL) {
perror("bw_to_psi:"); exit(1);
}
fscanf(in,"%ld",&last);
fclose(in);
sigma = csa->sigma;
k2 = csa->k2;
csa->C = mymalloc(sizeof(*csa->C)*sigma);
C2 = mymalloc(sizeof(*C2)*sigma);
for (c=0; c<sigma; c++) {
csa->C[c] = 0;
}
in = fopen(fbw,"r");
if (in == NULL) {
perror("bw_to_psi:"); exit(1);
}
csa->n = 0;
while (1) {
display_progressbar("reading ",csa->n,0L);
// c = fgetc(in);
c = readint(k2,in);
if (c == EOF) break;
if (c >= sigma) {
printf("bw_to_psi: c = %d sigma = %d\n", c, sigma);
exit(1);
}
csa->C[c]++;
csa->n++;
}
rewind(in);
printf("n = %ld last = %ld\n",csa->n,last);
psi = mymalloc(sizeof(diskbuf)*sigma);
*k = (blog(csa->n+1)+1+8-1)/8;
for (c=0; c<sigma; c++) {
psi[c] = open_diskbuf(out,*k);
}
for (j=1,c=0; c<sigma; c++) {
C2[c] = j;
j += csa->C[c];
}
for (i = 0; i<=csa->n; i++) {
display_progressbar("computing psi ",i,csa->n);
if (i == last) {
setint_diskbuf(psi[0], 0, i);
} else {
// c = fgetc(in);
c = readint(k2,in);
setint_diskbuf(psi[c], C2[c]++, i);
}
}
fclose(in);
for (c=0; c<sigma; c++) {
close_diskbuf(psi[c]);
}
free(psi);
free(C2);
}
i64 psi12_makeindex(CSA *csa, char *fname)
{
i64 psize,psize1,psize2;
i64 b, b2;
i64 i,j,x,xx;
i64 y,d,w;
int k;
FILE *f1,*f2;
char *fpsi, *fpsd;
//psi1_iterator *pi;
i64 runlen;
i64 maxrun;
i64 n,L;
psi1 *ps;
int id,id2;
FILE *out;
diskbuf *psi;
char *fbw, *flst;
SPARSEARRAY sx, sb;
int mm;
ps = (psi1 *)csa->psi_struc;
id = ps->id;
id2 = id & 0x3f;
k = strlen(fname);
fbw = mymalloc(k+5);
flst = mymalloc(k+5);
fpsi = mymalloc(k+5);
fpsd = mymalloc(k+5);
sprintf(fbw,"%s.bw",fname);
sprintf(flst,"%s.lst",fname);
switch (id2) {
case ID_DIFF_GAMMA_RR:
sprintf(fpsi,"%s.pxi",fname);
sprintf(fpsd,"%s.pxd",fname);
break;
}
out = create_tmp(0);
bw_to_psi(out,csa,fbw,flst,&k);
psi = open_diskbuf(out,k);
ps->last = getint_diskbuf(psi,0);
printf("last = %ld\n",ps->last);
n = csa->n;
L = ps->L;
if (L >= n) {
printf("L=%ld >= n=%ld\n",L,n);
exit(0);
}
maxrun = L;
mkdecodetable();
f1 = fopen(fpsi,"wb");
psize1 = 0;
f2 = fopen(fpsd,"wb");
psize2 = 0;
ps->k = k = (blog(n+1)+1+8-1)/8;
// pi = psi1_iterator_new(ps,0);
writeint(1,ID_PSI,f2);
writeint(1,k,f2); /* #bytes of integer */
writeint(k,n,f2);
writeint(k,L,f2);
psize2 += 1+1+2*k;
writeint(1,id,f2);
psize2 += 1;
if (id & ID_COMPPTR) {
mm = 0;
for (i=0; i<SIGMA; i++) {
if (csa->C[i] > 0) mm++;
}
SPARSEARRAY_construct_init(&sx, (mm+1)*(n+1), n/L+1);
SPARSEARRAY_construct_init(&sb, n, n/L+1);
}
b = b2 = 0;
mm = 0; xx = 0;
for (j=0; j<=n/L; j++) {
// display_progressbar("writing psi ",j,n/L);
if (j % 100000 == 0) {
printf("%ld %1.3f bpc\r",j,(double)psize2*8/(j+1)/L); fflush(stdout);
}
y = getint_diskbuf(psi,j*L);
if (id & ID_COMPPTR) {
if (y <= xx) {
mm++;
}
SPARSEARRAY_construct_set(&sx, j, mm*(n+1) + y);
SPARSEARRAY_construct_set(&sb, j, b);
xx = y;
} else {
// printf("%ld x=%ld sp=%ld\n",j,y,b);
writeint(k,y,f2);
writeint(k,b,f2);
psize2 += 2*k;
}
x = y;
runlen = 0;
b2 = 0;
for (i=j*L+1; i<(j+1)*L && i <= n; i++) { /* psi[j*L] are not encoded */
y = getint_diskbuf(psi,i);
d = y - x;
if (d <= 0) {
d += n+1;
}
if (i == j*L+1) {
if (d == 1) {setbit(Btmp,b2+1,1); runlen = 1;}
else {setbit(Btmp,b2+1,0); runlen = 0;}
b2++;
}
if (d > 1) {
if (runlen>0) {
// printf("d=%ld encode runlen=%ld\n",d,runlen);
w = ENCODENUM(Btmp,b2,runlen);
b2 += w;
runlen = 0;
}
// printf("encode d-1=%ld\n",d-1);
w = ENCODENUM(Btmp,b2,d-1);
b2 += w;
runlen = 1;
} else {
runlen++;
}
x = y;
}
if (runlen>0) {
// printf("encode runlen=%ld\n",runlen);
w = ENCODENUM(Btmp,b2,runlen);
b2 += w;
runlen = 0;
}
fwrite(Btmp,(b2+15) / 16,sizeof(short),f1);
psize1 += (b2+15)/16*sizeof(short);
b += (b2+15) / 16;
b2 = 0;
}
if (b2 > 0) {
fwrite(Btmp,(b2+15) / 16,sizeof(short),f1);
psize1 += (b2+15)/16*sizeof(short);
}
fwrite(Btmp,1,sizeof(short),f1); // getbitDで１ワード余計に読むため
psize1 += 1*sizeof(short);
if (id & ID_COMPPTR) {
SPARSEARRAY_construct_end(&sx, SDARRAY_SELECT1);
SPARSEARRAY_construct_end(&sb, SDARRAY_SELECT1);
SPARSEARRAY_write(&sx, f2);
SPARSEARRAY_write(&sb, f2);
}
psize = psize1 + psize2;
printf("size %ld (%1.3f bpc)\n",psize,(double)psize*8 / n);
fclose(f1);
fclose(f2);
// psi1_iterator_remove(pi);
close_diskbuf(psi);
fclose(out);
remove_tmp(0);
psi1_read(csa, fpsd);
free(fpsi);
free(fpsd);
free(fbw);
free(flst);
return psize;
}
i64 psi2_makeindex(CSA *csa, char *fname)
{
i64 psize;
i64 b, b2;
i64 i,j,x;
i64 y,d,w;
int k;
FILE *f1,*f2;
char *fpsi;
//psi1_iterator *pi;
i64 n,m;
i64 nn, mm;
psi2 *ps;
int id;
FILE *out;
diskbuf *psi;
char *fbw, *flst;
sparsearray4 sa;
ps = (psi2 *)csa->psi_struc;
id = csa->id;
k = strlen(fname);
fbw = (char *) mymalloc(k+5);
flst = (char *) mymalloc(k+5);
fpsi = (char *) mymalloc(k+5);
sprintf(fbw,"%s.bw",fname);
sprintf(flst,"%s.lst",fname);
sprintf(fpsi,"%s.psa",fname);
out = create_tmp(0);
bw_to_psi(out,csa,fbw,flst,&k);
m = 0;
for (i=0; i<SIGMA; i++) {
if (csa->C[i] > 0) m++;
}
psi = open_diskbuf(out,k);
ps->last = getint_diskbuf(psi,0);
printf("last = %ld\n",ps->last);
n = csa->n;
mm = n+1; nn = (m+1) * (n+1);
sparsearray4_construct_init(&sa, nn, mm);
y = 0; d = 0;
for (j=0; j<=n; j++) {
display_progressbar("compressing psi ",j,n);
x = getint_diskbuf(psi,j);
if (x <= y) d++;
sparsearray4_construct_set(&sa, j, d*(n+1)+x);
y = x;
}
sparsearray4_construct_end(&sa,0, SDARRAY_SELECT1);
f2 = fopen(fpsi,"wb");
psize = 0;
ps->k = k = (blog(n+1)+1+8-1)/8;
writeint(1,ID_PSI,f2);
writeint(1,k,f2); /* #bytes of integer */
writeint(k,n,f2);
psize += 1+1+k;
writeint(1,id,f2);
psize += 1;
psize += sparsearray4_write(&sa, f2);
printf("size %ld (%1.3f bpc)\n",psize,(double)psize*8 / n);
fclose(f2);
// psi1_iterator_remove(pi);
close_diskbuf(psi);
fclose(out);
remove_tmp(0);
psi2_read(csa, fpsi);
free(fpsi);
free(fbw);
free(flst);
return psize;
}
