gk_i2cc2i_t *gk_i2cc2i_create_common(char *alphabet)
{
int nsymbols;
gk_idx_t i;
gk_i2cc2i_t *t;
nsymbols = strlen(alphabet);
t = gk_malloc(sizeof(gk_i2cc2i_t),"gk_i2c_create_common");
t->n = nsymbols;
t->i2c = gk_cmalloc(256, "gk_i2c_create_common");
t->c2i = gk_imalloc(256, "gk_i2c_create_common");
gk_cset(256, -1, t->i2c);
gk_iset(256, -1, t->c2i);
for(i=0;i<nsymbols;i++){
t->i2c[i] = alphabet[i];
t->c2i[(int)alphabet[i]] = i;
}
return t;
}
void gk_array2csr(size_t n, size_t range, int *array, int *ptr, int *ind)
{
size_t i;
gk_iset(range+1, 0, ptr);
for (i=0; i<n; i++)
ptr[array[i]]++;
/* Compute the ptr, ind structure */
MAKECSR(i, range, ptr);
for (i=0; i<n; i++)
ind[ptr[array[i]]++] = i;
SHIFTCSR(i, range, ptr);
}
gk_csr_t *itemsets_project_matrix(isparams_t *params, gk_csr_t *mat, int cid)
{
ssize_t i, j, k, ii, pnnz;
int nrows, ncols, pnrows, pncols;
ssize_t *colptr, *pcolptr;
int *colind, *colids, *pcolind, *pcolids, *rmarker;
gk_csr_t *pmat;
gk_ikv_t *cand;
nrows = mat->nrows;
ncols = mat->ncols;
colptr = mat->colptr;
colind = mat->colind;
colids = mat->colids;
rmarker = params->rmarker;
cand = params->cand;
/* Allocate space for the projected matrix based on what you know thus far */
pmat = gk_csr_Create();
pmat->nrows = pnrows = (cid == -1 ? nrows : colptr[cid+1]-colptr[cid]);
/* Mark the rows that will be kept and determine the prowids */
if (cid == -1) { /* Initial projection */
gk_iset(nrows, 1, rmarker);
}
else { /* The other projections */
for (i=colptr[cid]; i<colptr[cid+1]; i++)
rmarker[colind[i]] = 1;
}
/* Determine the length of each column that will be left in the projected matrix */
for (pncols=0, pnnz=0, i=cid+1; i<ncols; i++) {
for (k=0, j=colptr[i]; j<colptr[i+1]; j++) {
k += rmarker[colind[j]];
}
if (k >= params->minfreq && k <= params->maxfreq) {
cand[pncols].val = i;
cand[pncols++].key = k;
pnnz += k;
}
}
/* Sort the columns in increasing order */
gk_ikvsorti(pncols, cand);
/* Allocate space for the remaining fields of the projected matrix */
pmat->ncols = pncols;
pmat->colids = pcolids = gk_imalloc(pncols, "itemsets_project_matrix: pcolids");
pmat->colptr = pcolptr = gk_zmalloc(pncols+1, "itemsets_project_matrix: pcolptr");
pmat->colind = pcolind = gk_imalloc(pnnz, "itemsets_project_matrix: pcolind");
/* Populate the projected matrix */
pcolptr[0] = 0;
for (pnnz=0, ii=0; ii<pncols; ii++) {
i = cand[ii].val;
for (j=colptr[i]; j<colptr[i+1]; j++) {
if (rmarker[colind[j]])
pcolind[pnnz++] = colind[j];
}
pcolids[ii] = colids[i];
pcolptr[ii+1] = pnnz;
}
/* Reset the rmarker array */
if (cid == -1) { /* Initial projection */
gk_iset(nrows, 0, rmarker);
}
else { /* The other projections */
for (i=colptr[cid]; i<colptr[cid+1]; i++)
rmarker[colind[i]] = 0;
}
return pmat;
}
