/* Attempt to convert a CFG to a FSG. No heuristic simplifcation is performed
* here. The conversion will only take place if all expansion rule in the
* CFG takes one of the following form
*
* X -> w0 w1 ... wN Y
* X -> w0 w1 ... wN
* X -> Y
* X -> w0
* X -> nil
*
* where X, Y are non-terminals, and w0, ..., wN are terminals. If the
* conversion is not possible, _fsg is set to NULL and the function returns -1.
*/
void
s3_cfg_convert_to_fsg(s3_cfg_t *_cfg,s2_fsg_t **_fsg)
{
hash_table_t *item2state=NULL;
int num_states=1; /* let 0 be the end state */
int start_state=0;
int from_state;
int to_state;
int i=0,j=0;
s3_cfg_id_t id;
s3u_vector_t *items=NULL;
s3u_vector_t *rules=NULL;
s3_cfg_item_t *item=NULL;
s3_cfg_rule_t *rule=NULL;
s2_fsg_t *fsg=NULL;
char *word;
item2state=hash_new(S3_CFG_NAME_HASH_SIZE,HASH_CASE_YES);
if (item2state==NULL)
goto cleanup;
fsg=(s2_fsg_t*)ckd_calloc(1,sizeof(s2_fsg_t));
fsg->name=NULL;
fsg->trans_list=NULL;
items=&_cfg->item_info;
for (i=s3u_vec_count(items)-1;i>=0;i--) {
if ((item=s3u_vec_get(items,i))==NULL)
goto cleanup;
rules=&item->rules;
if (!s3_cfg_is_terminal(item->id) &&
(item->nil_rule!=NULL || (rules!=NULL && s3u_vec_count(rules)>0)))
hash_enter_bkey(item2state,&item->id,sizeof(s3_cfg_id_t),num_states++);
}
/* iterate through the CFG's expansion rules and convert them to FSG
* transitions. If at any point the conversion fails, do some cleanup
* and return.
*/
rules=&_cfg->rules;
for (i=s3u_vec_count(rules)-1;i>=0;i--) {
if ((rule=s3u_vec_get(rules,i))==NULL)
goto cleanup;
hash_lookup_bkey(item2state,&rule->src,sizeof(s3_cfg_id_t),&from_state);
/* a NULL production rule means we transition to the end state */
if (rule->len==0)
add_trans(fsg,from_state,0,rule->prob_score,NULL);
else if (rule->len==1) {
id=rule->products[0];
/* a single terminal means we output the terminal and transition to
* the end state
*/
if (s3_cfg_is_terminal(id)) {
word=((s3_cfg_item_t*)s3u_vec_get(items,s3_cfg_id2index(id)))->name;
add_trans(fsg,from_state,0,rule->prob_score,word);
}
/* a single non-terminal means we take an epsilon transition */
else {
hash_lookup_bkey(item2state,&id,sizeof(s3_cfg_id_t),&to_state);
add_trans(fsg,from_state,to_state,rule->prob_score,NULL);
}
}
else {
for (j=1;j<rule->len;j++) {
/* get the output for the transition */
id=rule->products[j-1];
if (!s3_cfg_is_terminal(id))
goto cleanup;
word=((s3_cfg_item_t*)s3u_vec_get(items,s3_cfg_id2index(id)))->name;
/* get the target state for the transition */
id=rule->products[j];
if (s3_cfg_is_terminal(id))
to_state=num_states++;
else
hash_lookup_bkey(item2state,&id,sizeof(s3_cfg_id_t),&to_state);
add_trans(fsg,from_state,to_state,j==1?rule->prob_score:1.0,word);
from_state=to_state;
}
}
}
*_fsg=fsg;
return 0;
cleanup:
if (fsg!=NULL)
free_fsg(fsg);
return -1;
}
/*
* Convert the glist of ssids to a composite sseq id. Return the composite ID.
*/
static s3ssid_t ssidlist2comsseq (glist_t g, mdef_t *mdef, dict2pid_t *dict2pid,
hash_table_t *hs, /* For composite states */
hash_table_t *hp) /* For composite senone seq */
{
int32 i, j, n, s, ssid;
s3senid_t **sen;
s3senid_t *comsenid;
gnode_t *gn;
n = glist_count (g);
if (n <= 0)
E_FATAL("Panic: length(ssidlist)= %d\n", n);
/* Space for list of senones for each state, derived from the given glist */
sen = (s3senid_t **) ckd_calloc (mdef_n_emit_state (mdef), sizeof(s3senid_t *));
for (i = 0; i < mdef_n_emit_state (mdef); i++) {
sen[i] = (s3senid_t *) ckd_calloc (n+1, sizeof(s3senid_t));
sen[i][0] = BAD_S3SENID; /* Sentinel */
}
/* Space for composite senone ID for each state position */
comsenid = (s3senid_t *) ckd_calloc (mdef_n_emit_state (mdef), sizeof(s3senid_t));
for (gn = g; gn; gn = gnode_next(gn)) {
ssid = gnode_int32 (gn);
/* Expand ssid into individual states (senones); insert in sen[][] if not present */
for (i = 0; i < mdef_n_emit_state (mdef); i++) {
s = mdef->sseq[ssid][i];
for (j = 0; (IS_S3SENID(sen[i][j])) && (sen[i][j] != s); j++);
if (NOT_S3SENID(sen[i][j])) {
sen[i][j] = s;
sen[i][j+1] = BAD_S3SENID;
}
}
}
/* Convert senones list for each state position into composite state */
for (i = 0; i < mdef_n_emit_state (mdef); i++) {
for (j = 0; IS_S3SENID(sen[i][j]); j++);
assert (j > 0);
j = hash_enter_bkey (hs, (char *)(sen[i]), j*sizeof(s3senid_t), dict2pid->n_comstate);
if (j == dict2pid->n_comstate)
dict2pid->n_comstate++; /* New composite state */
else
ckd_free ((void *) sen[i]);
comsenid[i] = j;
}
ckd_free (sen);
/* Convert sequence of composite senids to composite sseq ID */
j = hash_enter_bkey (hp, (char *)comsenid, mdef->n_emit_state * sizeof(s3senid_t),
dict2pid->n_comsseq);
if (j == dict2pid->n_comsseq) {
dict2pid->n_comsseq++;
if (dict2pid->n_comsseq >= MAX_S3SENID)
E_FATAL("#Composite sseq limit(%d) reached; increase MAX_S3SENID\n",
dict2pid->n_comsseq);
} else
ckd_free ((void *) comsenid);
return ((s3ssid_t)j);
}