/**
* Increase the number of disjuncts associated to each word in the
* sentence by working with word-clusters. Return true if the number
* of disjuncts were expanded, else return false.
*/
bool lg_expand_disjunct_list(Sentence sent)
{
size_t w;
Cluster *clu = lg_cluster_new();
bool expanded = false;
for (w = 0; w < sent->length; w++)
{
X_node * x;
Disjunct * d = sent->word[w].d;
for (x = sent->word[w].x; x != NULL; x = x->next)
{
Disjunct *dx = build_expansion_disjuncts(clu, x);
if (dx)
{
unsigned int cnt = count_disjuncts(d);
d = catenate_disjuncts(dx, d);
d = eliminate_duplicate_disjuncts(d);
if (cnt < count_disjuncts(d)) expanded = true;
}
}
sent->word[w].d = d;
}
lg_cluster_delete(clu);
return expanded;
}
/**
* Takes the list of disjuncts pointed to by d, eliminates all
* duplicates, and returns a pointer to a new list.
* It frees the disjuncts that are eliminated.
*/
Disjunct * eliminate_duplicate_disjuncts(Disjunct * d)
{
int i, h, count;
Disjunct *dn, *dx, *dxn, *front;
count = 0;
disjunct_dup_table *dt;
dt = disjunct_dup_table_new(next_power_of_two_up(2 * count_disjuncts(d)));
for (;d!=NULL; d = dn)
{
dn = d->next;
h = hash_disjunct(d);
front = NULL;
for (dx = dt->dup_table[h]; dx != NULL; dx = dxn)
{
dxn = dx->next;
if (disjunct_matches_alam(dx,d))
{
/* we know that d should be killed */
d->next = NULL;
free_disjuncts(d);
count++;
front = catenate_disjuncts(front, dx);
break;
} else if (disjunct_matches_alam(d,dx)) {
/* we know that dx should be killed off */
dx->next = NULL;
free_disjuncts(dx);
count++;
} else {
/* neither should be killed off */
dx->next = front;
front = dx;
}
}
if (dx == NULL) {
/* we put d in the table */
d->next = front;
front = d;
}
dt->dup_table[h] = front;
}
/* d is now NULL */
for (i = 0; i < dt->dup_table_size; i++)
{
for (dx = dt->dup_table[i]; dx != NULL; dx = dxn)
{
dxn = dx->next;
dx->next = d;
d = dx;
}
}
if ((verbosity > 2) && (count != 0)) printf("killed %d duplicates\n", count);
disjunct_dup_table_delete(dt);
return d;
}
