static i64 psi12_psi(CSA *csa, i64 i)
{
i64 j,k;
i64 x;
i64 k2,p,n;
i64 L;
i64 b,d,sp;
unsigned short *B;
psi1 *ps;
int runlen;
ps = (psi1 *)csa->psi_struc;
L = ps->L;
if (ps->id & ID_COMPPTR) {
x = SPARSEARRAY_select(ps->sx, (i/L)+1) % (csa->n+1);
sp = SPARSEARRAY_select(ps->sb, (i/L)+1);
} else {
x = getuint(ps->R,(i / L)*2,ps->k);
sp = getuint(ps->R,(i / L)*2+1,ps->k);
}
b = 0;
j = i % L;
n = ps->n;
B = ps->B;
b = 0;
d = getbit(B+sp,b+1);
if (d == 1) runlen = 1; else runlen = 0;
b++;
k = 1;
while (k<=j) {
if (runlen == 1) {
b += DECODENUM(B+sp,b,&d);
// printf("1. k = %d j = %d x = %ld d = %d runlen = %d\n",k,j,x,d,runlen);
if (k+d-1 > j) {
x += j-k+1;
x %= (n+1);
break;
}
x += d-1;
k += d-1;
// printf("11. k = %d j = %d x = %ld d = %d runlen = %d\n",k,j,x,d,runlen);
}
if (k > j) break;
b += DECODENUM(B+sp,b,&d);
// printf("2. k = %d j = %d x = %ld d = %d runlen = %d\n",k,j,x,d,runlen);
x += d+1;
x %= (n+1);
k++;
// printf("22. k = %d j = %d x = %ld d = %d runlen = %d\n",k,j,x,d,runlen);
runlen = 1;
}
return x;
}
void csa_decodeall(unsigned char *p,CSA *SA) {
int *I;
int i,n,pos;
int x,b,d;
unsigned short *B;
n = SA->n;
I =(int *)malloc((n+1)*sizeof(*I));
if (I == NULL) {perror("decodeall"); exit(1);}
B = SA->B;
x = -1; b = 0;
for (i=1; i<=n; i++) {
b += DECODENUM(B,b,&d);
x += d;
if (x > n) {
x = -1; i--;
}
else {
I[i] = x;
}
}
pos = csa_inverse(SA,1);
for (i=1; i<=n; i++) {
if (pos < 1 || pos > n) {
printf("i %d pos %d\n",i,pos);
}
*p++ = csa_T(SA,pos);
pos = I[pos];
}
}
static void psi1_iterator_readpage(psi1_iterator *pi, i64 page)
{
i64 L,i,j,k,id;
i64 x,sp,n,b,d;
unsigned short *B;
i64 maxrun;
psi1 *ps;
n = pi->n;
ps = pi->ps;
L = ps->L;
id = ps->id;
maxrun = L;
B = ps->B;
j = L;
if (page*L + j > n) j = n - page*L;
x = getuint(ps->R,page*2,ps->k);
sp = getuint(ps->R,page*2+1,ps->k);
pi->buf[0] = x;
b = 0;
for (k=1; k<j; k++) {
b += DECODENUM(B+sp,b,&d);
if (id == ID_DIFF_GAMMA) {
x += d;
x %= n;
pi->buf[k] = x;
} else if (id == ID_DIFF_GAMMA_RL) {
if (d <= maxrun*2) {
if (d % 2 == 0) {
for (i=0; i<d/2; i++) {
x += 1;
x %= n;
if (k+i >= L) {
printf("readpage: error k=%ld i=%ld l=%ld\n",k,i,L);
}
pi->buf[k+i] = x;
}
k += (d/2)-1;
} else {
x += (d+3)/2;
x %= n;
pi->buf[k] = x;
}
} else {
x += d-maxrun+1;
x %= n;
pi->buf[k] = x;
}
} else {
printf("??? id = %ld\n",id);
}
}
pi->page = page;
}
static void mkdecodetable(void)
{
unsigned short B[256];
i64 i,j,b,b2,d,x;
for (i=0; i<256; i++) B[i] = 0xffff;
for (i = 0; i < TBLSIZE; i++) {
B[0] = i;
b = 0; j = 0; x = 0;
while (1) {
b2 = DECODENUM(B,b,&d);
if (b+b2 > DD) break;
b += b2;
x += d;
j++;
}
R5n[i] = j; R5b[i] = b; R5x[i] = x;
}
}
static i64 psi1_pred(CSA *csa, i64 pl, i64 l, i64 r)
{
i64 m,ll,rr,j;
i64 x;
i64 sp,L,d,n;
uchar *R;
unsigned short *B, *buf, offset;
i64 maxrun;
int k;
psi1 *ps;
ps = (psi1 *)csa->psi_struc;
R = ps->R; B = ps->B; L = ps->L; k = ps->k;
n = ps->n;
maxrun = L;
ll = (l+L-1) / L;
// ll = l / L + 1;
rr = r / L;
while (ll <= rr) {
m = (ll + rr) / 2;
x = getuint(R,m*2,k);
if (x >= pl) rr = m-1; else ll = m+1;
}
m = rr*L;
x = getuint(R,(m / L)*2,k);
sp = getuint(R,(m / L)*2+1,k);
// r = (rr+1)*L; if (r > n) r = n+1;
if ((rr+1)*L <= r) r = (rr+1)*L; else r = r+1;
if (r > n) r = n+1;
buf = B + sp;
offset = 0;
switch (ps->id & 0x3f) {
case ID_DIFF_GAMMA:
while (m < r) {
if (m >= l && x >= pl) break;
offset += DECODENUM(buf,offset,&d);
x += d;
x %= (n+1);
m++;
}
return m;
break;
case ID_DIFF_GAMMA_RL:
while (m < r) {
if (m >= l && x >= pl) break;
offset += DECODENUM(buf,offset,&d);
if (d <= maxrun*2) {
if (d % 2 == 0) {
for (j=0; j<d/2 && m<r; j++) {
if (m >= l && x >= pl) goto end;
x += 1;
x %= (n+1);
m++;
}
} else {
x += (d+3)/2;
x %= (n+1);
m++;
}
} else {
x += d-maxrun+1;
x %= (n+1);
m++;
}
}
end: return m;
break;
}
return -1;
}
static i64 psi1_psi(CSA *csa, i64 i)
{
i64 j,k;
i64 x;
i64 k2,p,n;
i64 L;
i64 b,d,sp;
i64 maxrun;
unsigned short *B;
psi1 *ps;
ps = (psi1 *)csa->psi_struc;
// printf("psi1_psi[%ld] (no run)\n",i);
#ifdef DEBUG
if (i > csa->n || i < 1) {
printf("error csa2_psi i=%u n=%u\n",i,csa->n);
exit(1);
}
#endif
L = ps->L;
if (ps->id & ID_COMPPTR) {
x = SPARSEARRAY_select(ps->sx, (i/L)+1) % (csa->n+1);
sp = SPARSEARRAY_select(ps->sb, (i/L)+1);
} else {
x = getuint(ps->R,(i / L)*2,ps->k);
sp = getuint(ps->R,(i / L)*2+1,ps->k);
}
maxrun = L;
b = 0;
j = i % L;
n = ps->n;
B = ps->B;
switch (ps->id & 0x3f) {
case ID_DIFF_GAMMA:
case ID_DIFF_GAMMA_SPARSE:
k = 0;
while (k < j) {
p = getbitD(B+sp,1+b);
k2 = R5n[p];
if (k2 == 0) {
b += DECODENUM(B+sp,b,&d);
x += d;
x %= (n+1);
k++;
} else {
if (k+k2 > j) break;
k += k2;
b += R5b[p];
x += R5x[p];
x %= (n+1);
}
}
for (; k<j; k++) {
b += DECODENUM(B+sp,b,&d);
x += d;
x %= (n+1);
}
break;
case ID_DIFF_GAMMA_RL:
case ID_DIFF_GAMMA_RL_SPARSE:
// psi1_decbuf[0] = x;
b = 0;
for (k=1; k<=j; k++) {
b += DECODENUM(B+sp,b,&d);
if (d <= maxrun*2) {
if (d % 2 == 0) {
#if 0
for (i=0; i<d/2; i++) {
x += 1;
x %= (n+1);
if (k+i >= L) {
printf("psi1_psi: error k=%d i=%d l=%d\n",k,i,L);
}
// psi1_decbuf[k+i] = x;
if (k+i == j) break;
}
#else
if (k+d/2-1 >= j) {
x += j-k+1;
x %= (n+1);
break;
}
x += d/2;
#endif
k += (d/2)-1;
} else {
x += (d+3)/2;
x %= (n+1);
// psi1_decbuf[k] = x;
}
} else {
x += d-maxrun+1;
x %= (n+1);
// psi1_decbuf[k] = x;
}
}
// x = psi1_decbuf[j];
break;
}
#ifdef DEBUG
if (x < 0 || x > csa->n) {
printf("error csa2_psi(%u) %u\n",i,x);
}
#endif
// printf("psi1_psi: psi[%ld] = %ld\n",i,x);
return x;
}
/* backward search */
int csa_bsearch(unsigned char *key,int keylen,CSA *SA,int *li,int *ri) {
int c,h,l,r,m,ll,rr,pl,pr;
int x,b,w,d,n,*R;
unsigned short *B;
int len;
c = key[keylen-1];
r = SA->C[c]; if (c>0) l = SA->C[c-1]+1; else l = 1;
len = 0;
if (l > r) goto end;
len++;
for (h = keylen-2; h >= 0; h--) {
pl = l; pr = r;
c = key[h];
r = SA->C[c]; if (c>0) l = SA->C[c-1]+1; else l = 1;
if (l > r) goto end;
#if 0
while (1) { // find maximum r such that Psi[r] <= pr
j = csa_psi(SA,r);
if (j <= pr) break;
r--;
//if (l > r) goto end;
}
#else
#if 0
ll = l; rr = r;
while (ll <= rr) {
m = (ll + rr) / 2;
if (csa_psi(SA,m) <= pr) ll = m+1; else rr = m-1;
}
r = ll-1;
#else
R = SA->R; B = SA->B; w = SA->l; n = SA->n;
ll = l / w + 1;
rr = r / w;
while (ll <= rr) {
m = (ll + rr) / 2;
if (R[m*2] <= pr) ll = m+1; else rr = m-1;
}
m = (ll-1)*w;
x = R[(m / w)*2];
b = R[(m / w)*2+1];
#if 1
while (m < l) {
b += DECODENUM(B,b,&d);
x += d;
//if (x > n) printf("??? \n");
if (x > n) {x = -1; m--;}
m++;
}
#endif
while (x <= pr && m <= r) {
b += DECODENUM(B,b,&d);
x += d;
//if (x > n) printf("??? \n");
m++;
}
r = m-1;
#endif
#endif
#if 0
while (1) { // find minimum l such that Psi[l] >= pl
j = csa_psi(SA,l);
if (j >= pl) break;
l++;
//if (l > r) goto end;
}
#else
#if 0
ll = l; rr = r;
while (ll <= rr) {
m = (ll + rr) / 2;
if (csa_psi(SA,m) >= pl) rr = m-1; else ll = m+1;
}
l = rr+1;
#else
//ll = l / w + 1;
ll = l / w;
rr = r / w;
while (ll <= rr) {
m = (ll + rr) / 2;
if (R[m*2] >= pl) rr = m-1; else ll = m+1;
}
m = rr*w;
x = R[(m / w)*2];
b = R[(m / w)*2+1];
while (m < l) {
b += DECODENUM(B,b,&d);
x += d;
if (x > n) {x = -1; m--;}
m++;
}
while (x < pl && m <= r) {
b += DECODENUM(B,b,&d);
x += d;
m++;
}
l = m;
#endif
#endif
if (l > r) goto end;
len++;
}
end:
*li = l; *ri = r;
return len;
}
int *csa_batchlookup3(CSA *SA,int l, int r,int len) {
int *I; /* z */
int *P; /* r i z */
int v; /* [ */
int m; /* SA */
int q;
int i,j;
int two;
int *sa;
int k,b,d,x,n,w;
unsigned short *B;
n = SA->n;
B = SA->B;
two = SA->two;
sa = SA->SA;
w = SA->l;
I =(int *) malloc((r-l+1+1)*sizeof(*I));
P =(int *) malloc((r-l+1+1)*sizeof(*I));
#if 1
x = SA->R[(l / w)*2];
b = SA->R[(l / w)*2+1];
j = l % w;
for (k=0; k<j; k++) {
b += DECODENUM(B,b,&d);
x += d;
if (x > n) {x = -1; k--;}
}
for (m = 0, q = 0, i = l; i <= r; i++) {
if (i % two == 0) {
I[1+m] = sa[i / two];
m++;
}
else {
P[q++] = x;
}
b += DECODENUM(B,b,&d);
x += d;
if (x > n) {
x = -1;
b += DECODENUM(B,b,&d);
x += d;
}
}
v = 1;
#else
for (q = 0, i = l; i <= r; i++) {
P[q++] = i;
}
v = 0;
m = 0;
#endif
while (q > 0 && v <= len) {
for (k = 0, j = 0; j < q; j++) {
i = P[j];
if (i % two == 0) {
I[1+m] = sa[i / two] - v;
m++;
}
else {
P[k++] = csa_psi(SA,i);
}
}
q = k;
v++;
}
for (j = 0; j < q; j++) {
I[1+m] = csa_lookup(SA,P[j])-v;
m++;
}
qsort(I+1, r-l+1, sizeof(int), intcompare);
I[0] = r-l+1;
free(P);
return I;
}
inline
int csa_psi(CSA *SA, int i) {
int j,k,b,d,x;
int k2,p,n;
int l;
unsigned short *B;
#ifdef DEBUG
if (i > SA->n || i < 1) {
printf("error csa2_psi i=%d n=%d\n",i,SA->n);
exit(1);
}
#endif
l = SA->l;
x = SA->R[(i / l)*2];
b = SA->R[(i / l)*2+1];
j = i % l;
n = SA->n;
B = SA->B;
#if 0
for (k=0; k<j; k++) {
b += DECODENUM(B,b,&d);
x += d;
if (x > n) {
//printf("i %d k %d d %d x %d n %d\n",i,k,d,x,n);
x = -1;
k--;
}
//printf("k %d j %d b %d \n",k,j,b);
}
#else
k = 0;
while (k < j) {
p = getbitD(B,1+b);
k2 = R5n[p];
if (k2 == 0) {
b += DECODENUM(B,b,&d);
x += d;
k++;
if (x > n) {
x = -1;
k--;
}
}
else {
if (k+k2 > j) break;
k += k2;
b += R5b[p];
x += R5x[p];
}
}
for (; k<j; k++) {
b += DECODENUM(B,b,&d);
x += d;
if (x > n) {
x = -1;
k--;
}
}
#endif
#ifdef DEBUG
if (x < 0 || x > SA->n) {
printf("error csa2_psi(%d) %d\n",i,x);
}
#endif
return x;
}