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;
}
CFSWString chars_to_phones_part_I(CFSWString &s) {
CFSWString res;
for (INTPTR i = 0; i < s.GetLength(); i++) {
CFSWString c = s.GetAt(i);
if (c == L']') {
CFSWString t = CFSWString(s.GetAt(i - 1)).ToUpper();
res.SetAt(res.GetLength() - 1, t.GetAt(0));
//vaatab taha; pole kindel, et kas on vajalik
t = CFSWString(s.GetAt(i - 2)).ToUpper();
if (can_palat(t)) {
res.SetAt(res.GetLength() - 2, t.GetAt(0));
t = CFSWString(s.GetAt(i - 3)).ToUpper();
if (can_palat(t)) {
res.SetAt(res.GetLength() - 2, t.GetAt(0));
}
}
//vaatab ette
t = CFSWString(s.GetAt(i + 1)).ToUpper();
if (can_palat(t)) {
s.SetAt(i + 1, t.GetAt(0));
t = CFSWString(s.GetAt(i + 2)).ToUpper();
if (can_palat(t)) {
s.SetAt(i + 2, t.GetAt(0));
}
}
} else
if (c == L'<') {
CFSWString t = CFSWString(s.GetAt(i + 1)).ToUpper();
s.SetAt(i + 1, t.GetAt(0));
} else
if (c == L'?') {
}//Ebanormaalne rõhk. Pärast vaatab, mis teha
else
if (c == L'x') res += L"ks";
else
if (c == L'y') res += L"i";
else
if (c == L'w') res += L"v";
else
if (c == L'z') res += L"ts";
else
if (c == L'c') {
res += L"k";
} else
if (c == L'ü' && is_vowel(s.GetAt(i + 1)) && s.GetAt(i - 1) == L'ü')
res += L"i";
else
res += c;
}
return res;
}
bool is_word(CFSWString s) {
if ((has_vowel(s) == true) &&
(s.GetLength() == make_char_string(s).GetLength()) &&
(s.GetLength() > 1))
return true;
return false;
}
bool is_stressed_syl(CFSWString syl) {
bool res = false;
for (INTPTR i = 0; i < syl.GetLength(); i++) {
if ((syl.GetAt(i) == colon) || ((is_vowel(syl.GetAt(i))) && (is_vowel(syl.GetAt(i + 1)))))
res = true;
}
return res;
}
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();
}
bool VOTAFAILIST::RidaTrimmitud(CFSWString& rida)
{
while (Rida(rida) == true)
{
rida.Trim();
if (rida.GetLength() > 0)
return true;
}
return false;
}
CFSWString simplify_pattern(CFSWString s) {
CFSWString res;
for (INTPTR i = 0; i < (s.GetLength()); i++) {
CFSWString c = s.GetAt(i);
if (c.FindOneOf(L"jhvsLQ") > -1) res += L"C";
else
res += c;
}
return res;
}
void CSuggestor::MultiReplace(const CFSWString &szWord, INTPTR ipStartPos)
{
if (ipStartPos>0) CheckAndAdd(szWord);
INTPTR ipLength=szWord.GetLength();
for (; ipStartPos<ipLength; ipStartPos++){
for (INTPTR ip=0; ip<(INTPTR)(sizeof(ChangeStringsMultiple)/sizeof(__CChangeStrings)); ip++){
if (szWord.ContainsAt(ipStartPos, ChangeStringsMultiple[ip].m_lpszFrom)){
MultiReplace(szWord.Left(ipStartPos)+ChangeStringsMultiple[ip].m_lpszTo+szWord.Mid(ipStartPos+FSStrLen(ChangeStringsMultiple[ip].m_lpszFrom)), ipStartPos+FSStrLen(ChangeStringsMultiple[ip].m_lpszTo));
}
}
}
}
CFSWString syllabify2(CFSWString s) {
CFSWString res;
for (INTPTR i = 0; i < s.GetLength(); i++) {
CFSWString c = s.GetAt(i);
if (is_consonant(c) && is_vowel(s.GetAt(i - 1)) && is_vowel(s.GetAt(i + 1)))
res += d;
if (is_vowel(c) && is_vowel(s.GetAt(i - 1)) && is_vowel(s.GetAt(i + 1)) && c.ToLower() == s.GetAt(i + 1))
res += d;
if (is_consonant(c) && is_consonant(s.GetAt(i - 1)) && is_vowel(s.GetAt(i + 1)) && has_vowel(res)) //küsitav
res += d;
res += c;
}
return res;
}
void PTWSplitBuffer(const CFSWString &szBuffer, CPTWordArray &Words)
{
Words.Cleanup();
INTPTR ipStartPos=0;
INTPTR ipPos;
for (ipPos=0; ipPos<szBuffer.GetLength(); ipPos++) {
if (FSIsSpace(szBuffer[ipPos])) {
if (ipPos>ipStartPos) {
Words.AddItem(CPTWord(szBuffer.Mid(ipStartPos, ipPos-ipStartPos), ipStartPos));
}
ipStartPos=ipPos+1;
}
}
if (ipPos>ipStartPos) {
Words.AddItem(CPTWord(szBuffer.Mid(ipStartPos, ipPos-ipStartPos), ipStartPos));
}
}
CFSWString palat_vru (CFSWString s) {
CFSWString res;
bool m = false;
if (s == L"är'") res = L"ärq";
else
if (s == L"ar'") res = L"arq";
else
if (s == L"jäl'") res = L"jälq";
else
if (s == L"jal'") res = L"jalq";
else
if (s == L"kül'") res = L"külq";
else
if (s == L"pan'") res = L"panq";
else
if (s == L"tul'") res = L"tulq";
else
if (s == L"ol'") res = L"olq";
else
for (INTPTR i = s.GetLength()-1; i >= 0; i--) {
CFSWString c = s.GetAt(i);
if (c == L"'") {
m = true;
} else
if (must(c)) {
res = c + res;
m = true;
}
else
if (m) {
if (can_palat_vr(c)) {
c = c.ToUpper();
res = c + res;
} else {
res = c + res;
m = false;
}
} else {
res = c + res;
}
}
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;
}
void print_u(utterance_struct u) {
for (INTPTR i = 0; i < u.phr_vector.GetSize(); i++)
for (INTPTR i1 = 0; i1 < u.phr_vector[i].word_vector.GetSize(); i1++) {
fprintf(stderr, "%s\n\n", ccstr(u.phr_vector[i].word_vector[i1].mi.m_szRoot));
for (INTPTR i2 = 0; i2 < u.phr_vector[i].word_vector[i1].syl_vector.GetSize(); i2++) {
fprintf(stderr, "\t%s\n", ccstr(u.phr_vector[i].word_vector[i1].syl_vector[i2].syl));
for (INTPTR i3 = 0; i3 < u.phr_vector[i].word_vector[i1].syl_vector[i2].phone_vector.GetSize(); i3++) {
CFSWString w = u.phr_vector[i].word_vector[i1].syl_vector[i2].phone_vector[i3].phone;
while (w.GetLength() < 6) w += sp;
fprintf(stderr, "\t\t%s", ccstr(w));
wprintf(L"%i %i %i %i\t", u.phra_c, u.word_c, u.syl_c, u.phone_c);
wprintf(L"%i %i %i [%i]\t", u.phr_vector[i].word_c, u.phr_vector[i].syl_c, u.phr_vector[i].phone_c, u.phr_vector[i].utt_p);
wprintf(L"%i %i [%i %i]\t",
u.phr_vector[i].word_vector[i1].syl_c,
u.phr_vector[i].word_vector[i1].phone_c,
u.phr_vector[i].word_vector[i1].utt_p,
u.phr_vector[i].word_vector[i1].phr_p
);
wprintf(L"%i [%i %i %i]\t",
u.phr_vector[i].word_vector[i1].syl_vector[i2].phone_c,
u.phr_vector[i].word_vector[i1].syl_vector[i2].utt_p,
u.phr_vector[i].word_vector[i1].syl_vector[i2].phr_p,
u.phr_vector[i].word_vector[i1].syl_vector[i2].word_p
);
phone_struct p = u.phr_vector[i].word_vector[i1].syl_vector[i2].phone_vector[i3];
wprintf(L"[%i %i %i %i]\t", p.utt_p, p.phr_p, p.word_p, p.syl_p);
wprintf(L"{ %i }", u.phr_vector[i].word_vector[i1].syl_vector[i2].stress);
wprintf(L"\n");
}
}
}
wprintf(L"\n");
}
std::string to_stdstring(CFSWString s) {
std::string res = "";
for (INTPTR i = 0; i < s.GetLength(); i++)
res += s.GetAt(i);
return res;
}
CFSWString chars_to_phones_part_I(CFSWString &s) {
/* müüa -> müia siia?
Kuna vältemärgi nihutamise reeglid on ehitatud selliselt, et kohati kasutatakse
foneeme ja kohati ei, siis tuleb täht->foneem teisendus teha kahes jaos.
Palataliseerimine põhineb ideel, et palataliseeritud foneemi ümbruses ei saa
olla palataliseeruvaid mittepalataliseeritud foneeme :D
*/
CFSWString res;
for (INTPTR i = 0; i < s.GetLength(); i++) {
CFSWString c = s.GetAt(i);
if (c == L']') {
CFSWString t = CFSWString(s.GetAt(i - 1)).ToUpper();
res.SetAt(res.GetLength() - 1, t.GetAt(0));
//vaatab taha; pole kindel, et kas on vajalik
t = CFSWString(s.GetAt(i - 2)).ToUpper();
if (can_palat(t)) {
res.SetAt(res.GetLength() - 2, t.GetAt(0));
t = CFSWString(s.GetAt(i - 3)).ToUpper();
if (can_palat(t)) {
res.SetAt(res.GetLength() - 2, t.GetAt(0));
}
}
//vaatab ette
t = CFSWString(s.GetAt(i + 1)).ToUpper();
if (can_palat(t)) {
s.SetAt(i + 1, t.GetAt(0));
t = CFSWString(s.GetAt(i + 2)).ToUpper();
if (can_palat(t)) {
s.SetAt(i + 2, t.GetAt(0));
}
}
} else
if (c == L'<') {
CFSWString t = CFSWString(s.GetAt(i + 1)).ToUpper();
s.SetAt(i + 1, t.GetAt(0));
} else
if (c == L'?') {
}//Ebanormaalne rõhk. Pärast vaatab, mis teha
else
if (c == L'x') res += L"ks";
else
if (c == L'y') res += L"i";
else
if (c == L'w') res += L"v";
else
if (c == L'z') res += L"ts";
else
if (c == L'c') {
res += L"k";
} else
if (c == L'ü' && is_vowel(s.GetAt(i + 1)) && s.GetAt(i - 1) == L'ü')
res += L"i";
else
if (c == L'q') {
if (is_vowel(s.GetAt(i + 1))) {
res += L"k";
s.SetAt(i + 1, L'v');
} else
res += L"k";
} else
res += c;
}
return res;
}
CFSWString the_shift(CFSWString s) {
/*
On mingi võimalus, et lihtsustus tuleb teha kahes astmes. LQ-ta ja LQ-ga (vt shift_pattern). Kõik
seotud sellega, et pole vältenihutusreeglitest lõpuni aru saanud. Eksisteerib Mihkla versioon ja
ametlik versioon. Tänud Mihklale, kes kala asemel annab tattninale õnge, see õpetab ujuma.
Maadlesin õngega pikalt.
*/
CFSWString res;
CFSWString code;
INTPTR pos;
INTPTR i = 0;
INTPTR x;
while (s.GetLength() > 0) {
CFSWString c = s.GetAt(0);
s.Delete(0, 1);
if (is_uvowel(c)) {
c = c.ToLower();
code += shift_pattern(c);
res += c;
pos = i;
} else
if (c == d && code.GetLength() > 0) {
res += c;
code += c;
CFSWString t_code = code;
t_code += shift_pattern(s.GetAt(0));
x = pattern_lookup(t_code); //orig üle silbipiiri
if (x > -1) {
x += pos;
if (x > res.GetLength()) { // kui kargab järgmisse silpi
x = x - res.GetLength();
s.Insert(x, colon);
} else
res.Insert(x, colon);
i++;
} else {
t_code = simplify_pattern(t_code);
x = pattern_lookup(t_code); //liht üle silbipiiri
if (x > -1) {
x += pos;
if (x > res.GetLength()) { // kui kargab järgmisse silpi
x = x - res.GetLength();
s.Insert(x, colon);
} else
res.Insert(x, colon);
i++;
} else {
x = pattern_lookup(code); //orig
if (x > -1) {
x += pos;
res.Insert(x, colon);
i++;
} else {
code = simplify_pattern(code);
x = pattern_lookup(code); //liht
if (x > -1) {
x += pos;
res.Insert(x, colon);
i++;
}
}
}
}
code = empty_str;
} else {
res += c;
if (code.GetLength() > 0) {
code += shift_pattern(c);
}
}
i++;
} //while
// sõna lõpus
if (code.GetLength() > 0) {
code += L"#";
//imelik koht ainult "lonksu" pärast
if ((code.Left(3) == L"VLQ") && ((code.GetAt(3) == L's') || (code.GetAt(3) == L'h') || (code.GetAt(3) == L'v') || (code.GetAt(3) == L'j'))) {
code = L"VLQC#";
}
INTPTR x = pattern_lookup(code);
if (x > -1) {
x += pos;
res.Insert(x, colon);
} else {
code = simplify_pattern(code);
x = pattern_lookup(code);
if (x > -1) {
x += pos;
res.Insert(x, colon);
}
}
code = empty_str;
}
return res;
}