cst_item *item_last_daughter(const cst_item *i)
{
cst_item *p;
for (p=item_daughter(i); item_next(p); p=item_next(p));
return p;
}
cst_item *item_nth_daughter(const cst_item *i,int n)
{
int d;
cst_item *p;
for (d=0,p=item_daughter(i); p && (d < n); p=item_next(p),d++);
return p;
}
static const cst_val *word_numsyls(const cst_item *word)
{
cst_item *d;
int c;
for (c=0,d=item_daughter(item_as(word,"SylStructure"));
d;
d=item_next(d),c++);
return val_int_n(c);
}
static const cst_val *syl_onsetsize(const cst_item *syl)
{
cst_item *d;
int c;
for (c=0,d=item_daughter(item_as(syl,"SylStructure"));
d;
d=item_next(d),c++)
{
if (cst_streq("+",val_string(ph_vc(d))))
break;
}
return val_string_n(c);
}
cst_utterance *russian_postlex_function(cst_utterance *u)
{
const cst_item *word,*seg;
const char *answer,*name,*pair;
for(word=relation_head(utt_relation(u,"Transcription"));word;word=item_next(word))
{
if(item_feat_present(word,"no_pl")||item_feat_present(word,"no_vr"))
continue;
for(seg=item_daughter(word);seg;seg=item_next(seg))
{
name=item_feat_string(seg,"name");
if(cst_member_string(name,unstressed_vowels))
{
answer=val_string(cart_interpret(item_as(seg,"Segment"),&ru_vowel_reduction_cart));
if(!cst_streq(answer,"N"))
item_set_string(seg,"name",answer);
}
else
{
if(cst_streq(name,"ii")&&
cst_streq(ffeature_string(seg,"R:Segment.p.ph_csoft"),"-")&&
!(cst_streq(item_feat_string(word,"name"),"и")&&
cst_streq(ffeature_string(word,"gpos"),"content")))
{
item_set_string(seg,"name","yy");
}
}
}
}
for(word=relation_tail(utt_relation(u,"Transcription"));word;word=item_prev(word))
{
if(item_feat_present(word,"no_pl"))
continue;
for(seg=item_last_daughter(word);seg;seg=item_prev(seg))
{
name=item_feat_string(seg,"name");
pair=russian_vpair(name);
if(pair!=NULL)
{
answer=val_string(cart_interpret(item_as(seg,"Segment"),&ru_vpair_cart));
if(cst_streq(answer,"Y"))
item_set_string(seg,"name",pair);
}
}
}
return u;
}
static const cst_val *seg_onset_ctype(const cst_item *seg, const char *ctype)
{
const cst_item *s;
const cst_phoneset *ps = item_phoneset(seg);
for (s=item_daughter(item_parent(item_as(seg,"SylStructure")));
s;
s=item_next(s))
{
if (cst_streq("+",phone_feature_string(ps,item_feat_string(s,"name"),
"vc")))
return VAL_STRING_0;
if (cst_streq(ctype,phone_feature_string(ps,item_feat_string(s,"name"),
"ctype")))
return VAL_STRING_1;
}
return VAL_STRING_0;
}
cst_utterance *russian_lexical_insertion(cst_utterance *u)
{
cst_item *word;
cst_relation *sylstructure,*seg,*syl,*sylvowel,*transcription;
const cst_val *p;
const char *phone_name;
cst_val *phones;
cst_item *ssword, *sssyl, *segitem, *sylitem, *seg_in_syl, *svsyl, *vowel_in_syl, *tword, *seg_in_word;
cst_item *i,*tmp;
int num_segs;
int total_num_segs=0;
syl = utt_relation_create(u,"Syllable");
sylstructure = utt_relation_create(u,"SylStructure");
seg = utt_relation_create(u,"Segment");
sylvowel = utt_relation_create(u,"SylVowel");
transcription = utt_relation_create(u,"Transcription");
for (word=relation_head(utt_relation(u,"Word"));word;word=item_next(word))
{
phones=word_to_phones(word);
if(!phones)
continue;
num_segs=val_length(phones);
if((total_num_segs+num_segs)>max_num_segs)
{
delete_val(phones);
break;
}
ssword = relation_append(sylstructure,word);
tword = relation_append(transcription,word);
for (sssyl=NULL,sylitem=NULL,p=phones; p; p=val_cdr(p))
{
if (sylitem == NULL)
{
sylitem = relation_append(syl,NULL);
sssyl = item_add_daughter(ssword,sylitem);
}
segitem = relation_append(seg,NULL);
phone_name = val_string(val_car(p));
item_set_string(segitem,"name",phone_name);
seg_in_syl = item_add_daughter(sssyl,segitem);
seg_in_word = item_add_daughter(tword,segitem);
if(is_vowel(phone_name))
{
svsyl=relation_append(sylvowel,sylitem);
vowel_in_syl=item_add_daughter(svsyl,segitem);
}
if (ru_syl_boundary(seg_in_syl,val_cdr(p)))
{
sylitem = NULL;
if (sssyl)
item_set_string(sssyl,"stress","0");
}
}
assign_stress(word);
delete_val(phones);
total_num_segs+=num_segs;
}
i=relation_head(utt_relation(u,"Word"));
while(i)
{
tmp=item_next(i);
if(item_as(i,"Transcription")==NULL)
{
delete_item(item_as(i,"Token"));
delete_item(item_as(i,"Phrase"));
delete_item(i);
}
i=tmp;
}
i=relation_head(utt_relation(u,"Phrase"));
while(i)
{
tmp=item_next(i);
if(item_daughter(i)==NULL)
delete_item(i);
i=tmp;
}
return u;
}
static void assign_stress(cst_item *word)
{
int numsyls=ffeature_int(word,"word_numsyls");
const char *gpos=ffeature_string(word,"gpos");
const cst_item *word_in_phrase=item_as(word,"Phrase");
const cst_item *syls=item_as(word,"SylStructure");
const cst_item *syl=item_daughter(syls);
const cst_item *transcription=item_as(word,"Transcription");
int stressed=FALSE;
int n=item_feat_present(word,"stressed_syl_num")?item_feat_int(word,"stressed_syl_num"):0;
const char *name=item_name(word);
const char *pname=ffeature_string(word,"R:Phrase.p.name");
const char *nname=ffeature_string(word,"R:Phrase.n.name");
if(cst_streq(ffeature_string(word,"R:Token.p.name"),"по")&&
(cst_streq(name,"моему")||cst_streq(name,"своему")||cst_streq(name,"твоему")))
{
item_set_string(syl,"stress","1");
return;
}
else if((cst_streq(name,"не")||cst_streq(name,"ни"))&&
(cst_streq(nname,"был")||cst_streq(nname,"были")||cst_streq(nname,"было")))
{
item_set_string(syl,"stress","1");
return;
}
else if((cst_streq(name,"был")||cst_streq(name,"были")||cst_streq(name,"было"))&&
(cst_streq(pname,"не")||cst_streq(pname,"ни")))
return;
else if(cst_streq(gpos,"enc")&&item_prev(word_in_phrase))
return;
else if(cst_streq(gpos,"proc")&&item_next(word_in_phrase))
return;
if(!vowel_seg_between(item_daughter(transcription),item_last_daughter(transcription)))
return;
if(numsyls==1)
{
item_set_string(syl,"stress","1");
return;
}
for(;syl;syl=item_next(syl))
{
if(is_stressed_vowel(item_feat_string(item_daughter(item_as(syl,"SylVowel")),"name")))
{
item_set_string(syl,"stress","1");
stressed=TRUE;
}
}
if(n==0)
{
if(stressed)
return;
n=val_int(cart_interpret(word,&ru_stress_cart));
if((numsyls+n) < 0)
{
if(numsyls <= 4)
n=-2;
else
if(numsyls <= 6)
n=-3;
else
n=-4;
}
}
item_set_string(item_nth_daughter(syls,(numsyls+n)),"stress","1");
}
