/* Constructs the suffix array by using the sorted order of type B* suffixes. */
void
construct_SA(const uint8_t *T, int32_t *SA,
int32_t *bucket_A, int32_t *bucket_B,
int32_t n, int32_t m) {
int32_t *i, *j, *k;
int32_t s;
int32_t c0, c1, c2;
if (0 < m) {
/* Construct the sorted order of type B suffixes by using
the sorted order of type B* suffixes. */
for (c1 = ALPHABET_SIZE - 2; 0 <= c1; --c1) {
/* Scan the suffix array from right to left. */
for (i = SA + BUCKET_BSTAR(c1, c1 + 1),
j = SA + BUCKET_A(c1 + 1) - 1, k = NULL, c2 = -1;
i <= j;
--j) {
if (0 < (s = *j)) {
*j = ~s;
c0 = T[--s];
if ((0 < s) && (T[s - 1] > c0)) {
s = ~s;
}
if (c0 != c2) {
if (0 <= c2) {
BUCKET_B(c2, c1) = k - SA;
}
k = SA + BUCKET_B(c2 = c0, c1);
}
*k-- = s;
} else {
*j = ~s;
}
}
}
}
/* Construct the suffix array by using
the sorted order of type B suffixes. */
k = SA + BUCKET_A(c2 = T[n - 1]);
*k++ = (T[n - 2] < c2) ? ~(n - 1) : (n - 1);
/* Scan the suffix array from left to right. */
for (i = SA, j = SA + n; i < j; ++i) {
if (0 < (s = *i)) {
c0 = T[--s];
if ((s == 0) || (T[s - 1] < c0)) {
s = ~s;
}
if (c0 != c2) {
BUCKET_A(c2) = k - SA;
k = SA + BUCKET_A(c2 = c0);
}
*k++ = s;
} else {
*i = ~s;
}
}
}
/* Sorts suffixes of type B*. */
int32_t
sort_typeBstar(const uint8_t *T, int32_t *SA,
int32_t *bucket_A, int32_t *bucket_B,
int32_t n) {
int32_t *PAb;
int32_t i, j, k, t, m, bufsize;
int32_t c0, c1;
memset(bucket_A, 0, BUCKET_A_SIZE * sizeof (int32_t));
memset(bucket_B, 0, BUCKET_B_SIZE * sizeof (int32_t));
/* Count the number of occurrences of the first one or two characters of each
type A, B and B* suffix. Moreover, store the beginning position of all
type B* suffixes into the array SA. */
for (i = n - 1, m = n, c0 = T[n - 1]; 0 <= i;) {
/* type A suffix. */
do {
++BUCKET_A(c1 = c0);
} while ((0 <= --i) && ((c0 = T[i]) >= c1));
if (0 <= i) {
/* type B* suffix. */
++BUCKET_BSTAR(c0, c1);
SA[--m] = i;
/* type B suffix. */
for (--i, c1 = c0; (0 <= i) && ((c0 = T[i]) <= c1); --i, c1 = c0) {
++BUCKET_B(c0, c1);
}
}
}
m = n - m;
/*
note:
A type B* suffix is lexicographically smaller than a type B suffix that
begins with the same first two characters.
*/
/* Calculate the index of start/end point of each bucket. */
for (c0 = 0, i = 0, j = 0; c0 < ALPHABET_SIZE; ++c0) {
t = i + BUCKET_A(c0);
BUCKET_A(c0) = i + j; /* start point */
i = t + BUCKET_B(c0, c0);
for (c1 = c0 + 1; c1 < ALPHABET_SIZE; ++c1) {
j += BUCKET_BSTAR(c0, c1);
BUCKET_BSTAR(c0, c1) = j; /* end point */
i += BUCKET_B(c0, c1);
}
}
if (0 < m) {
int32_t *ISAb, *buf;
/* Sort the type B* suffixes by their first two characters. */
PAb = SA + n - m;
ISAb = SA + m;
for (i = m - 2; 0 <= i; --i) {
t = PAb[i], c0 = T[t], c1 = T[t + 1];
SA[--BUCKET_BSTAR(c0, c1)] = i;
}
t = PAb[m - 1], c0 = T[t], c1 = T[t + 1];
SA[--BUCKET_BSTAR(c0, c1)] = m - 1;
/* Sort the type B* substrings using sssort. */
buf = SA + m, bufsize = n - (2 * m);
for (c0 = ALPHABET_SIZE - 2, j = m; 0 < j; --c0) {
for (c1 = ALPHABET_SIZE - 1; c0 < c1; j = i, --c1) {
i = BUCKET_BSTAR(c0, c1);
if (1 < (j - i)) {
sssort(T, PAb, SA + i, SA + j,
buf, bufsize, 2, n, *(SA + i) == (m - 1));
}
}
}
/* Compute ranks of type B* substrings. */
for (i = m - 1; 0 <= i; --i) {
if (0 <= SA[i]) {
j = i;
do {
ISAb[SA[i]] = i;
} while ((0 <= --i) && (0 <= SA[i]));
SA[i + 1] = i - j;
if (i <= 0) {
break;
}
}
j = i;
do {
ISAb[SA[i] = ~SA[i]] = j;
} while (SA[--i] < 0);
ISAb[SA[i]] = j;
}
/* Construct the inverse suffix array of type B* suffixes using trsort. */
trsort(ISAb, SA, m, 1);
/* Set the sorted order of tyoe B* suffixes. */
for (i = n - 1, j = m, c0 = T[n - 1]; 0 <= i;) {
for (--i, c1 = c0; (0 <= i) && ((c0 = T[i]) >= c1); --i, c1 = c0) {
}
if (0 <= i) {
t = i;
for (--i, c1 = c0; (0 <= i) && ((c0 = T[i]) <= c1); --i, c1 = c0) {
}
SA[ISAb[--j]] = ((t == 0) || (1 < (t - i))) ? t : ~t;
}
}
/* Calculate the index of start/end point of each bucket. */
BUCKET_B(ALPHABET_SIZE - 1, ALPHABET_SIZE - 1) = n; /* end point */
for (c0 = ALPHABET_SIZE - 2, k = m - 1; 0 <= c0; --c0) {
i = BUCKET_A(c0 + 1) - 1;
for (c1 = ALPHABET_SIZE - 1; c0 < c1; --c1) {
t = i - BUCKET_B(c0, c1);
BUCKET_B(c0, c1) = i; /* end point */
/* Move all type B* suffixes to the correct position. */
for (i = t, j = BUCKET_BSTAR(c0, c1);
j <= k;
--i, --k) {
SA[i] = SA[k];
}
}
BUCKET_BSTAR(c0, c0 + 1) = i - BUCKET_B(c0, c0) + 1; /* start point */
BUCKET_B(c0, c0) = i; /* end point */
}
}
return m;
}
/* Constructs the burrows-wheeler transformed string directly
by using the sorted order of type B* suffixes. */
static
saidx_t
construct_BWT(const sauchar_t *T, saidx_t *SA,
saidx_t *bucket_A, saidx_t *bucket_B,
saidx_t n, saidx_t m) {
saidx_t *i, *j, *k, *orig;
saidx_t s;
saint_t c0, c1, c2;
if(0 < m) {
/* Construct the sorted order of type B suffixes by using
the sorted order of type B* suffixes. */
for(c1 = ALPHABET_SIZE - 2; 0 <= c1; --c1) {
/* Scan the suffix array from right to left. */
for(i = SA + BUCKET_BSTAR(c1, c1 + 1),
j = SA + BUCKET_A(c1 + 1) - 1, k = NULL, c2 = -1;
i <= j;
--j) {
if(0 < (s = *j)) {
assert(T[s] == c1);
assert(((s + 1) < n) && (T[s] <= T[s + 1]));
assert(T[s - 1] <= T[s]);
c0 = T[--s];
*j = ~((saidx_t)c0);
if((0 < s) && (T[s - 1] > c0)) { s = ~s; }
if(c0 != c2) {
if(0 <= c2) { BUCKET_B(c2, c1) = k - SA; }
k = SA + BUCKET_B(c2 = c0, c1);
}
assert(k < j);
*k-- = s;
} else if(s != 0) {
*j = ~s;
#ifndef NDEBUG
} else {
assert(T[s] == c1);
#endif
}
}
}
}
/* Construct the BWTed string by using
the sorted order of type B suffixes. */
k = SA + BUCKET_A(c2 = T[n - 1]);
*k++ = (T[n - 2] < c2) ? ~((saidx_t)T[n - 2]) : (n - 1);
/* Scan the suffix array from left to right. */
for(i = SA, j = SA + n, orig = SA; i < j; ++i) {
if(0 < (s = *i)) {
assert(T[s - 1] >= T[s]);
c0 = T[--s];
*i = c0;
if((0 < s) && (T[s - 1] < c0)) { s = ~((saidx_t)T[s - 1]); }
if(c0 != c2) {
BUCKET_A(c2) = k - SA;
k = SA + BUCKET_A(c2 = c0);
}
assert(i < k);
*k++ = s;
} else if(s != 0) {
*i = ~s;
} else {
orig = i;
}
}
return orig - SA;
}
/* Sorts suffixes of type B*. */
static
saidx_t
sort_typeBstar(const sauchar_t *T, saidx_t *SA,
saidx_t *bucket_A, saidx_t *bucket_B,
saidx_t n) {
saidx_t *PAb, *ISAb, *buf;
#ifdef _OPENMP
saidx_t *curbuf;
saidx_t l;
#endif
saidx_t i, j, k, t, m, bufsize;
saint_t c0, c1;
#ifdef _OPENMP
saint_t d0, d1;
int tmp;
#endif
/* Initialize bucket arrays. */
for(i = 0; i < BUCKET_A_SIZE; ++i) { bucket_A[i] = 0; }
for(i = 0; i < BUCKET_B_SIZE; ++i) { bucket_B[i] = 0; }
/* Count the number of occurrences of the first one or two characters of each
type A, B and B* suffix. Moreover, store the beginning position of all
type B* suffixes into the array SA. */
for(i = n - 1, m = n, c0 = T[n - 1]; 0 <= i;) {
/* type A suffix. */
do { ++BUCKET_A(c1 = c0); } while((0 <= --i) && ((c0 = T[i]) >= c1));
if(0 <= i) {
/* type B* suffix. */
++BUCKET_BSTAR(c0, c1);
SA[--m] = i;
/* type B suffix. */
for(--i, c1 = c0; (0 <= i) && ((c0 = T[i]) <= c1); --i, c1 = c0) {
++BUCKET_B(c0, c1);
}
}
}
m = n - m;
/*
note:
A type B* suffix is lexicographically smaller than a type B suffix that
begins with the same first two characters.
*/
/* Calculate the index of start/end point of each bucket. */
for(c0 = 0, i = 0, j = 0; c0 < ALPHABET_SIZE; ++c0) {
t = i + BUCKET_A(c0);
BUCKET_A(c0) = i + j; /* start point */
i = t + BUCKET_B(c0, c0);
for(c1 = c0 + 1; c1 < ALPHABET_SIZE; ++c1) {
j += BUCKET_BSTAR(c0, c1);
BUCKET_BSTAR(c0, c1) = j; /* end point */
i += BUCKET_B(c0, c1);
}
}
if(0 < m) {
/* Sort the type B* suffixes by their first two characters. */
PAb = SA + n - m; ISAb = SA + m;
for(i = m - 2; 0 <= i; --i) {
t = PAb[i], c0 = T[t], c1 = T[t + 1];
SA[--BUCKET_BSTAR(c0, c1)] = i;
}
t = PAb[m - 1], c0 = T[t], c1 = T[t + 1];
SA[--BUCKET_BSTAR(c0, c1)] = m - 1;
/* Sort the type B* substrings using sssort. */
#ifdef _OPENMP
tmp = omp_get_max_threads();
buf = SA + m, bufsize = (n - (2 * m)) / tmp;
c0 = ALPHABET_SIZE - 2, c1 = ALPHABET_SIZE - 1, j = m;
#pragma omp parallel default(shared) private(curbuf, k, l, d0, d1, tmp)
{
tmp = omp_get_thread_num();
curbuf = buf + tmp * bufsize;
k = 0;
for(;;) {
#pragma omp critical(sssort_lock)
{
if(0 < (l = j)) {
d0 = c0, d1 = c1;
do {
k = BUCKET_BSTAR(d0, d1);
if(--d1 <= d0) {
d1 = ALPHABET_SIZE - 1;
if(--d0 < 0) { break; }
}
} while(((l - k) <= 1) && (0 < (l = k)));
c0 = d0, c1 = d1, j = k;
}
}
if(l == 0) { break; }
sssort(T, PAb, SA + k, SA + l,
curbuf, bufsize, 2, n, *(SA + k) == (m - 1));
}
}
#else
buf = SA + m, bufsize = n - (2 * m);
for(c0 = ALPHABET_SIZE - 2, j = m; 0 < j; --c0) {
for(c1 = ALPHABET_SIZE - 1; c0 < c1; j = i, --c1) {
i = BUCKET_BSTAR(c0, c1);
if(1 < (j - i)) {
sssort(T, PAb, SA + i, SA + j,
buf, bufsize, 2, n, *(SA + i) == (m - 1));
}
}
}
#endif
/* Compute ranks of type B* substrings. */
for(i = m - 1; 0 <= i; --i) {
if(0 <= SA[i]) {
j = i;
do { ISAb[SA[i]] = i; } while((0 <= --i) && (0 <= SA[i]));
SA[i + 1] = i - j;
if(i <= 0) { break; }
}
j = i;
do { ISAb[SA[i] = ~SA[i]] = j; } while(SA[--i] < 0);
ISAb[SA[i]] = j;
}
/* Construct the inverse suffix array of type B* suffixes using trsort. */
trsort(ISAb, SA, m, 1);
/* Set the sorted order of tyoe B* suffixes. */
for(i = n - 1, j = m, c0 = T[n - 1]; 0 <= i;) {
for(--i, c1 = c0; (0 <= i) && ((c0 = T[i]) >= c1); --i, c1 = c0) { }
if(0 <= i) {
t = i;
for(--i, c1 = c0; (0 <= i) && ((c0 = T[i]) <= c1); --i, c1 = c0) { }
SA[ISAb[--j]] = ((t == 0) || (1 < (t - i))) ? t : ~t;
}
}
/* Calculate the index of start/end point of each bucket. */
BUCKET_B(ALPHABET_SIZE - 1, ALPHABET_SIZE - 1) = n; /* end point */
for(c0 = ALPHABET_SIZE - 2, k = m - 1; 0 <= c0; --c0) {
i = BUCKET_A(c0 + 1) - 1;
for(c1 = ALPHABET_SIZE - 1; c0 < c1; --c1) {
t = i - BUCKET_B(c0, c1);
BUCKET_B(c0, c1) = i; /* end point */
/* Move all type B* suffixes to the correct position. */
for(i = t, j = BUCKET_BSTAR(c0, c1);
j <= k;
--i, --k) { SA[i] = SA[k]; }
}
BUCKET_BSTAR(c0, c0 + 1) = i - BUCKET_B(c0, c0) + 1; /* start point */
BUCKET_B(c0, c0) = i; /* end point */
}
}
return m;
}
/* Constructs the suffix array by using the sorted order of type B* suffixes. */
static
void
construct_SA(const sauchar_t *T, saidx_t *SA,
saidx_t *bucket_A, saidx_t *bucket_B,
saidx_t n, saidx_t m) {
saidx_t *i, *j, *k;
saidx_t s;
saint_t c0, c1, c2;
if(0 < m) {
/* Construct the sorted order of type B suffixes by using
the sorted order of type B* suffixes. */
for(c1 = ALPHABET_SIZE - 2; 0 <= c1; --c1) {
/* Scan the suffix array from right to left. */
for(i = SA + BUCKET_BSTAR(c1, c1 + 1),
j = SA + BUCKET_A(c1 + 1) - 1, k = NULL, c2 = -1;
i <= j;
--j) {
if(0 < (s = *j)) {
assert(T[s] == c1);
assert(((s + 1) < n) && (T[s] <= T[s + 1]));
assert(T[s - 1] <= T[s]);
*j = ~s;
c0 = T[--s];
if((0 < s) && (T[s - 1] > c0)) { s = ~s; }
if(c0 != c2) {
if(0 <= c2) { BUCKET_B(c2, c1) = k - SA; }
k = SA + BUCKET_B(c2 = c0, c1);
}
assert(k < j);
*k-- = s;
} else {
assert(((s == 0) && (T[s] == c1)) || (s < 0));
*j = ~s;
}
}
}
}
/* Construct the suffix array by using
the sorted order of type B suffixes. */
k = SA + BUCKET_A(c2 = T[n - 1]);
*k++ = (T[n - 2] < c2) ? ~(n - 1) : (n - 1);
/* Scan the suffix array from left to right. */
for(i = SA, j = SA + n; i < j; ++i) {
if(0 < (s = *i)) {
assert(T[s - 1] >= T[s]);
c0 = T[--s];
if((s == 0) || (T[s - 1] < c0)) { s = ~s; }
if(c0 != c2) {
BUCKET_A(c2) = k - SA;
k = SA + BUCKET_A(c2 = c0);
}
assert(i < k);
*k++ = s;
} else {
assert(s < 0);
*i = ~s;
}
}
}