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();
}
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;
}
CFSArray<CFSWString> do_utterances(CFSWString s) {
CFSWString res = empty_str;
CFSArray<CFSWString> res_array;
if (s.GetLength() == 1)
res_array.AddItem(s);
else
for (INTPTR i = 0; i < s.GetLength(); i++) {
CFSWString c = s.GetAt(i);
CFSWString pc = res.GetAt(res.GetLength() - 1);
CFSWString nc = s.GetAt(i + 1);
CFSWString nnc = s.GetAt(i + 2);
if (is_ending(c) && is_whitespace(nc) && is_upper(nnc)) {
res.Trim();
res_array.AddItem(res);
res = empty_str;
} else
if (is_tab(c)) {
if (res.GetLength() > 0) {
res.Trim();
res_array.AddItem(res);
res = empty_str;
}
} else
res += c;
}
res.Trim();
if (res.GetLength() > 0) {
while (is_ending(res.GetAt(res.GetLength() - 1))) {
res.Delete(res.GetLength() - 1, 1);
}
res_array.AddItem(res);
}
for (INTPTR i=0; i < res_array.GetSize(); i++) {
if (is_ending(res_array[i].GetAt(res_array[i].GetLength()-1)))
res_array[i].Delete( res_array[i].GetLength()-1, 1 );
}
return res_array;
}
bool VOTAFAILIST::RidaTrimmitud(CFSWString& rida)
{
while (Rida(rida) == true)
{
rida.Trim();
if (rida.GetLength() > 0)
return true;
}
return false;
}
void push_ph_res(utterance_struct &u, phrase_struct &p, CFSWString &res) {
res.Trim();
p.s = res;
u.phr_vector.AddItem(p);
res = empty_str;
} //do_phrases osa