CFSWString DealWithText(CFSWString text) {
/* Proovin kogu sõnniku minema loopida */
CFSWString res;
text.Trim();
text.Replace(L"\n\n", L"\n", 1);
text.Replace(L"‘", L"'", 1);
text.Replace(L"`", L"'", 1);
text.Replace(L"´", L"'", 1);
text.Replace(L"’", L"'", 1);
for (INTPTR i = 0; i < text.GetLength(); i++) {
CFSWString c = text.GetAt(i);
CFSWString pc = res.GetAt(res.GetLength() - 1);
CFSWString nc = text.GetAt(i + 1);
if (c == L"'") {
if (is_vowel(pc))
res += L"q";
else
res += c;
}
else
if (is_char(c)) res += c;
else
if (is_digit(c)) res += c;
else
if (is_hyphen(c) && is_char(pc) && is_char(nc)) res += sp;
else
if (is_symbol(c)) res += c;
else
if (is_colon(c) && !is_colon(pc)) res += c;
else
if (is_bbracket(c) && !is_bbracket(pc)) res += c;
else
if (is_ebracket(c) && is_ending(nc)) res += L"";
else
if (is_ebracket(c) && !is_ebracket(pc)) res += c;
else
if (is_comma(c) && !is_comma(pc)) res += c;
else
if (is_fchar(c)) res += replace_fchar(c);
else
if (is_space(c) && !is_whitespace(pc)) res += c;
else
if (is_break(c) && !is_break(pc)) {
res += c;
} //kahtlane
else
if (is_tab(c) && !is_whitespace(pc)) res += c;
else
if (is_ending(c) && !is_ending(pc) && !is_whitespace(pc)) res += c;
}
res.Trim();
return res;
}
void do_phones(syl_struct &ss) {
CFSArray<phone_struct> pv;
phone_struct p;
p.utt_p = 0;
p.phr_p = 0;
p.word_p = 0;
p.syl_p = 0;
INTPTR phone_syl_p = 1;
for (INTPTR i = 0; i < ss.syl.GetLength(); i++) {
CFSWString c = ss.syl.GetAt(i);
if ((c == L'š') || (c == L'ž')) c = L"sh";
if (c == L'õ') c = L"q";
if (c == L'ä') c = L"x";
if (c == L'ö') c = L"c";
if (c == L'ü') c = L"y";
if (is_colon(c)) {
// 1. nihutusvigade kaitseks (vt "piirkonda")
// 2-3. kolmandas vältes v ja j on kõnebaasis sedavõrd haruldased,
// et väljundis kuuleb nende asemel mingit r-i laadset hääikut.
// Kellel on parem baas, kommenteerigu 2. ja 3. tingimus välja.
if ((i > 0) && (pv[pv.GetSize() - 1].phone != L"j") && (pv[pv.GetSize() - 1].phone != L"v"))
pv[pv.GetSize() - 1].phone += doq;
} else {
p.phone = c;
p.syl_p = phone_syl_p++;
pv.AddItem(p);
}
}
ss.phone_vector = pv;
}
void do_phones(syl_struct &ss) {
CFSArray<phone_struct> pv;
phone_struct p;
p.utt_p = 0;
p.phr_p = 0;
p.word_p = 0;
p.syl_p = 0;
INTPTR phone_syl_p = 1;
for (INTPTR i = 0; i < ss.syl.GetLength(); i++) {
CFSWString c = ss.syl.GetAt(i);
if ((c == L'š') || (c == L'ž')) c = L"sh";
else
if (c == L'q') c = L"kw";
else
if (c == L'õ') c = L"q";
else
if (c == L'ä') c = L"x";
else
if (c == L'ö') c = L"c";
else
if (c == L'ü') c = L"y";
if (is_colon(c)) {
if (i > 0) // nihutusvigade kaitseks (vt "piirkonda")
pv[pv.GetSize() - 1].phone += doq;
} else {
p.phone = c;
p.syl_p = phone_syl_p++;
pv.AddItem(p);
}
}
ss.phone_vector = pv;
}
INTPTR do_phrases(utterance_struct &u) {
phrase_struct p;
CFSWString res;
p.phone_c = 0;
p.syl_c = 0;
p.word_c = 0;
for (INTPTR i = 0; i < u.s.GetLength(); i++) {
CFSWString c = u.s.GetAt(i);
CFSWString pc = res.GetAt(res.GetLength() - 1);
CFSWString nc = u.s.GetAt(i + 1);
CFSWString nnc = u.s.GetAt(i + 2);
if ((is_comma(c) || is_colon(c) || is_semicolon(c)) && is_space(nc) && is_char(nnc)) {
res.Trim();
if (res.GetLength() > 0) {
push_ph_res(u, p, res);
}
} else
if (is_bbracket(c)) {
res.Trim();
if (res.GetLength() > 0) {
push_ph_res(u, p, res);
}
p.s = L"sulgudes";
u.phr_vector.AddItem(p);
} else
if (is_ebracket(c)) {
res.Trim();
if (res.GetLength() > 0) {
push_ph_res(u, p, res);
}
} else
if (is_space(c)) { // komatud sidesõnad
CFSWString tempm = u.s.Mid(i + 1, -1);
res.Trim();
if (is_conju(tempm.Left(tempm.Find(sp))) && res.GetLength() > 0) {
push_ph_res(u, p, res);
} else
res += c;
} else
if (is_bhyphen(c)) {
res.Trim();
if (res.GetLength() > 0 && ((is_char(pc) && is_space(nc)) || (is_space(nc) && is_char(nnc)) || (is_space(pc) && is_char(nc)))) {
push_ph_res(u, p, res);
} else
res += c;
} else
res += c;
}
if (res.GetLength() > 0) {
// if (is_ending(res.GetAt(res.GetLength() - 1))) {
// res.Delete(res.GetLength() - 1, 1);
// }
push_ph_res(u, p, res);
}
return u.phr_vector.GetSize();
}
// time := hour colon minute [ colon second ] .
void time()
{
hour();
colon();
minute();
if (is_colon(scan.get())) {
colon();
second();
}
}
