void
write_grammar(FILE *fp, grammar g, si_t si)
{
sihashbrsit bhit;
sihashursit uhit;
size_t i;
for (bhit=sihashbrsit_init(g.brs); sihashbrsit_ok(bhit); bhit=sihashbrsit_next(bhit))
for (i=0; i<bhit.value.n; i++)
fprintf(fp, "%g %s " REWRITES " %s %s\n", (double) bhit.value.e[i]->prob,
si_index_string(si, bhit.value.e[i]->parent),
si_index_string(si, bhit.value.e[i]->left),
si_index_string(si, bhit.value.e[i]->right));
for (uhit=sihashursit_init(g.urs); sihashursit_ok(uhit); uhit=sihashursit_next(uhit))
for (i=0; i<uhit.value.n; i++)
fprintf(fp, "%g %s " REWRITES " %s\n", (double) uhit.value.e[i]->prob,
si_index_string(si, uhit.value.e[i]->parent),
si_index_string(si, uhit.value.e[i]->child));
}
static void
apply_unary(sihashcc chart_entry, grammar g, int right_pos,
sihashcc left_vertex)
{
sihashursit ursit;
size_t i;
for (ursit=sihashursit_init(g.parent_urs); sihashursit_ok(ursit);
ursit = sihashursit_next(ursit)) {
/* look up the rule's child category */
chart_cell c = sihashcc_ref(chart_entry, ursit.key);
if (c) /* such categories exist in this cell */
for (i=0; i<ursit.value.n; i++) {
chart_cell cc = add_edge(chart_entry, ursit.value.e[i]->parent,
&c->tree, NULL,
c->prob * ursit.value.e[i]->prob,
right_pos, left_vertex);
if (cc)
follow_unary(cc, chart_entry, g, right_pos, left_vertex);
}}}
grammar
read_grammar(FILE *fp, si_t si)
{
sihashbrs left_brules_ht = make_sihashbrs(NLABELS);
sihashurs child_urules_ht = make_sihashurs(NLABELS);
sihashf parent_weight_ht = make_sihashf(NLABELS);
brihashbr brihtbr = make_brihashbr(NLABELS);
int n;
double weight;
urule ur;
sihashbrsit bhit;
sihashursit uhit;
size_t root_label = 0, lhs, cat, rhs[MAXRHS];
while ((n = fscanf(fp, " %lg ", &weight)) == 1) { /* read the count */
lhs = read_cat(fp, si);
assert(weight > 0);
assert(lhs);
if (!root_label)
root_label = lhs;
fscanf(fp, " " REWRITES); /* read the rewrites symbol */
for (n=0; n<MAXRHS; n++) { /* read the rhs, n is length of rhs */
cat = read_cat(fp, si);
if (!cat)
break;
rhs[n] = cat;
}
if (n >= MAXRHS) {
fprintf(stderr, "read_grammar() in grammar.c: rule rhs too long\n");
exit(EXIT_FAILURE);
}
switch (n) {
case 0:
fprintf(stderr, "read_grammar() in grammar.c: rule with empty rhs\n");
exit(EXIT_FAILURE);
break;
case 1:
ur = make_urule(weight, lhs, rhs[0]);
push_urule(child_urules_ht, ur->child, ur);
sihashf_inc(parent_weight_ht, ur->parent, weight);
break;
case 2:
add_brule(left_brules_ht, brihtbr, weight, lhs, rhs[0], rhs[1]);
sihashf_inc(parent_weight_ht, lhs, weight);
break;
default:
{ int start, i, j;
char bcat[MAXBLABELLEN], *s;
si_index bparent, left, right;
right = rhs[n-1]; /* rightmost category */
for (start=n-2; start>=1; start--) {
i = 0; /* i is index into bcat[] */
for (j=start; j<n; j++) { /* j is index into rhs[] */
if (j!=start) {
bcat[i++] = BINSEP;
assert(i < MAXBLABELLEN);
}
s = si_index_string(si, rhs[j]);
while (*s) {
bcat[i++] = *s++;
assert(i < MAXBLABELLEN);
}}
bcat[i] = '\0';
bparent = si_string_index(si, bcat);
left = rhs[start];
add_brule(left_brules_ht, brihtbr, weight, bparent, left, right);
sihashf_inc(parent_weight_ht, bparent, weight);
right = bparent;
}
add_brule(left_brules_ht, brihtbr, weight, lhs, rhs[0], right);
sihashf_inc(parent_weight_ht, lhs, weight);
}}}
free_brihashbr(brihtbr); /* free brindex hash table */
{
int i; /* normalize grammar rules */
for (bhit = sihashbrsit_init(left_brules_ht); sihashbrsit_ok(bhit); bhit = sihashbrsit_next(bhit))
for (i=0; i<bhit.value.n; i++)
bhit.value.e[i]->prob /= sihashf_ref(parent_weight_ht, bhit.value.e[i]->parent);
for (uhit = sihashursit_init(child_urules_ht); sihashursit_ok(uhit); uhit = sihashursit_next(uhit))
for (i=0; i<uhit.value.n; i++)
uhit.value.e[i]->prob /= sihashf_ref(parent_weight_ht, uhit.value.e[i]->parent);
}
free_sihashf(parent_weight_ht);
{
grammar g;
g.urs = child_urules_ht;
g.brs = left_brules_ht;
g.root_label = root_label;
return g;
}
}