// add morphological analysis to CFSArray containing the sentence
void addAnalysis(CLinguistic& linguistic, CDisambiguator& disambiguator, CFSArray<CFSVar>& words, const bool disambiguate) {
//CFSVar &words=Data["words"];
CFSArray<CPTWord> PTWords;
for (INTPTR ip=0; ip<words.GetSize(); ip++) {
PTWords.AddItem(words[ip]["text"].GetWString());
}
// perform analysis and optional disambiguation
CFSArray<CMorphInfos> MorphResults=linguistic.AnalyzeSentense(PTWords);
if (disambiguate) {
MorphResults=disambiguator.Disambiguate(MorphResults);
}
// collect the analysis results
ASSERT(PTWords.GetSize()==MorphResults.GetSize());
for (INTPTR ip=0; ip<words.GetSize(); ip++) {
const CFSArray<CMorphInfo> &Analysis=MorphResults[ip].m_MorphInfo;
CFSVar VarAnalysis;
VarAnalysis.Cast(CFSVar::VAR_ARRAY);
for (INTPTR ipRes=0; ipRes<Analysis.GetSize(); ipRes++) {
const CMorphInfo &Analysis1=Analysis[ipRes];
CFSVar VarAnalysis1;
VarAnalysis1["root"]=Analysis1.m_szRoot;
VarAnalysis1["ending"]=Analysis1.m_szEnding;
VarAnalysis1["clitic"]=Analysis1.m_szClitic;
VarAnalysis1["partofspeech"]=CFSWString(Analysis1.m_cPOS);
VarAnalysis1["form"]=Analysis1.m_szForm;
VarAnalysis[ipRes]=VarAnalysis1;
}
words[ip]["analysis"]=VarAnalysis;
}
}
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;
}
CFSArray<CFSWString> do_all(CFSWString utt, bool print_label, bool print_utt) {
CFSArray<CFSWString> res;
CFSArray<CPTWord> PTW;
utterance_struct u;
u.s = utt;
u.syl_c = 0;
u.phone_c = 0;
u.phra_c = do_phrases(u);
INTPTR word_count = 0;
if (print_utt) fprintf(stderr, "%s\n", ccstr(utt));
for (INTPTR i = 0; i < u.phr_vector.GetSize(); i++) {
u.phr_vector[i].utt_p = i;
phrase2words(u.phr_vector[i], PTW);
word_count += u.phr_vector[i].word_c;
}
CFSArray<CMorphInfos> MRs = Disambiguator.Disambiguate(Linguistic.AnalyzeSentense(PTW));
CFSArray<CMorphInfo> words;
for (INTPTR i = 0; i < MRs.GetSize(); i++)
//for (INTPTR i1 = 0; i1 < MRs[i].m_MorphInfo.GetSize(); i1++)
words.AddItem(MRs[i].m_MorphInfo[0]); //Ühestamistulemuse ühestamise koht
u.word_c = words.GetSize();
word_struct w;
INTPTR utt_phone_c = 1;
INTPTR syl_utt_p = 1;
INTPTR phone_utt_p = 1;
for (INTPTR i = 0; i < u.phr_vector.GetSize(); i++) {
u.phr_vector[i].utt_p = i + 1;
INTPTR syl_phr_p = 1;
INTPTR phone_phr_p = 1;
INTPTR phrase_pho_c = 1;
for (INTPTR i1 = 0; i1 < u.phr_vector[i].word_c; i1++) {
w.utt_p = utt_phone_c++;
w.phr_p = i1 + 1;
w.mi = words[0];
w.mi.m_szRoot += make_char_string(w.mi.m_szEnding) + w.mi.m_szClitic;
w.mi.m_szRoot = w.mi.m_szRoot.ToLower();
// sidesõnad + ei välteta
if ((CFSWString(w.mi.m_cPOS) == L"J") || (w.mi.m_szRoot == L"<ei")) w.mi.m_szRoot.Replace(L"<", L"", 1);
do_syls(w);
u.phr_vector[i].word_vector.AddItem(w);
INTPTR phone_word_p = 1;
INTPTR word_pho_c = 1;
for (INTPTR i2 = 0; i2 < u.phr_vector[i].word_vector[i1].syl_vector.GetSize(); i2++) {
u.syl_c++;
u.phr_vector[i].syl_c++;
u.phr_vector[i].word_vector[i1].syl_c++;
u.phr_vector[i].word_vector[i1].syl_vector[i2].phr_p = syl_phr_p++;
u.phr_vector[i].word_vector[i1].syl_vector[i2].utt_p = syl_utt_p++;
INTPTR syl_phone_c = 1;
do_phones(u.phr_vector[i].word_vector[i1].syl_vector[i2]);
for (INTPTR i3 = 0; i3 < u.phr_vector[i].word_vector[i1].syl_vector[i2].phone_vector.GetSize(); i3++) {
u.phone_c++;
u.phr_vector[i].phone_c = phrase_pho_c++;
u.phr_vector[i].word_vector[i1].phone_c = word_pho_c++;
u.phr_vector[i].word_vector[i1].syl_vector[i2].phone_c = syl_phone_c++;
phone_struct p = u.phr_vector[i].word_vector[i1].syl_vector[i2].phone_vector[i3];
p.utt_p = phone_utt_p++;
p.phr_p = phone_phr_p++;
p.word_p = phone_word_p++;
u.phr_vector[i].word_vector[i1].syl_vector[i2].phone_vector[i3] = p;
}
}
words.RemoveItem(0, 1);
}
}
if (print_label) print_u(u);
res = do_label(u);
return res;
}
int CSuggestor::Suggest(const CFSWString &szWord, bool bStartSentence){
m_TimeStart=CFSTime::Now();
m_Items.Cleanup();
m_Cap.SetCap(szWord);
if (bStartSentence && m_Cap.GetCapMode()==CFSStrCap<CFSWString>::CAP_LOWER) {
m_Cap.SetCapMode(CFSStrCap<CFSWString>::CAP_INITIAL);
}
CFSWString szWordHigh=szWord.ToUpper();
INTPTR ipWordLength=szWordHigh.GetLength();
CFSWString szTemp;
INTPTR i, j;
long lLevel=100;
SetLevel(lLevel);
// Case problems & change list
i=SpellWord(szWordHigh, szTemp, &lLevel);
if ((i==SPL_NOERROR || i==SPL_CHANGEONCE) && !szTemp.IsEmpty()){
SetLevel(GetLevelGroup(lLevel));
m_Items.AddItem(CSuggestorItem(szTemp, lLevel));
}
else SetLevel(5);
// Abbrevations
// !!! Unimplemented
// Quotes
/* if (ipWordLength>=2 &&
(szAllQuot.Find(szWordHigh[0])>=0 || szAllQuot.Find(szWordHigh[ipWordLength-1])>=0))
{
szTemp=szWordHigh;
int iPos;
if (szAllQuot.Find(szTemp[0])>=0){
if (szQuotLeft.Find(szTemp[0])>=0) { }
else if ((iPos=szQuotRight.Find(szTemp[0]))>=0) { szTemp[0]=szQuotLeft[iPos]; }
else if (szDQuotLeft.Find(szTemp[0])>=0) { }
else if ((iPos=szDQuotRight.Find(szTemp[0]))>=0) { szTemp[0]=szDQuotLeft[iPos]; }
if (szAllQuot.Find(szTemp[ipWordLength-1])>=0) { szTemp[ipWordLength-1]=(szQuotRight+szDQuotRight)[(szQuotLeft+szDQuotLeft).Find(szTemp[0])];
}
else{
if (szQuotRight.Find(szTemp[ipWordLength-1])>=0) { }
else if ((iPos=szQuotLeft.Find(szTemp[ipWordLength-1]))>=0) { szTemp[ipWordLength-1]=szQuotRight[iPos]; }
else if (szDQuotRight.Find(szTemp[ipWordLength-1])>=0) { }
else if ((iPos=szDQuotLeft.Find(szTemp[ipWordLength-1]))>=0) { szTemp[ipWordLength-1]=szDQuotRight[iPos]; }
}
CheckAndAdd(szTemp);
}*/
// Add space
for (i=1; i<ipWordLength-1; i++){
static CFSWString szPunktuation=FSWSTR(".:,;!?");
if (szPunktuation.Find(szWord[i])>=0){
long lLevel1, lLevel2;
CFSWString szTemp1, szTemp2;
if (SpellWord(szWord.Left(i+1), szTemp1, &lLevel1)==SPL_NOERROR &&
SpellWord(szWord.Mid(i+1), szTemp2, &lLevel2)==SPL_NOERROR)
{
m_Items.AddItem(CSuggestorItem(szWord.Left(i+1)+L' '+szWord.Mid(i+1), FSMAX(lLevel1, lLevel2)));
}
}
}
// Delete following blocks: le[nnu][nnu]jaam
for (i=2; i<=3; i++){
for (j=0; j<ipWordLength-i-i; j++){
if (memcmp((const FSWCHAR *)szWordHigh+j, (const FSWCHAR *)szWordHigh+j+i, i*sizeof(FSWCHAR))==0){
szTemp=szWordHigh.Left(j)+szWordHigh.Mid(j+i);
CheckAndAdd(szTemp);
}
}
}
// Change following letters: abb -> aab & aab -> abb
for (i=1; i<ipWordLength-1; i++){
if (szWordHigh[i]==szWordHigh[i+1]){
szTemp=szWordHigh;
szTemp[i]=szTemp[i-1];
if (FSIsLetterEst(szTemp[i])) CheckAndAdd(szTemp);
}
else if (szWordHigh[i]==szWordHigh[i-1]){
szTemp=szWordHigh;
szTemp[i]=szTemp[i+1];
if (FSIsLetterEst(szTemp[i])) CheckAndAdd(szTemp);
}
}
// Exchange letters: van[na]ema -> van[an]ema
szTemp=szWordHigh;
for (i=1; i<ipWordLength; i++){
if (szTemp[i]!=szTemp[i-1]){
FSWCHAR ch=szTemp[i];
szTemp[i]=szTemp[i-1];
szTemp[i-1]=ch;
CheckAndAdd(szTemp);
szTemp[i-1]=szTemp[i];
szTemp[i]=ch;
}
}
// Change blocks
for (i=0; i<ipWordLength; i++){
for (j=0; j<(INTPTR)(sizeof(ChangeStrings)/sizeof(__CChangeStrings)); j++){
if (szWordHigh.ContainsAt(i, ChangeStrings[j].m_lpszFrom)){
szTemp=szWordHigh.Left(i)+ChangeStrings[j].m_lpszTo+szWordHigh.Mid(i+FSStrLen(ChangeStrings[j].m_lpszFrom));
CheckAndAdd(szTemp);
}
}
}
// Change end blocks
for (i=0; i<(INTPTR)(sizeof(ChangeStringsEnd)/sizeof(__CChangeStrings)); i++){
if (szWordHigh.EndsWith(ChangeStringsEnd[i].m_lpszFrom)){
szTemp=szWordHigh.Left(ipWordLength-FSStrLen(ChangeStringsEnd[i].m_lpszFrom))+ChangeStringsEnd[i].m_lpszTo;
CheckAndAdd(szTemp);
}
}
// Po~o~sas
MultiReplace(szWordHigh, 0);
// gi/ki: Kylli[gi]le -> Kyllile[gi]
for (i=3; i<=6; i++){
if (i>ipWordLength) break;
if (memcmp((const FSWCHAR *)szWordHigh+ipWordLength-i, FSWSTR("GI"), 2*sizeof(FSWCHAR))==0){
szTemp=szWordHigh.Left(ipWordLength-i)+szWordHigh.Mid(ipWordLength-i+2)+FSWSTR("GI");
CheckAndAdd(szTemp);
szTemp=szWordHigh.Left(ipWordLength-i)+szWordHigh.Mid(ipWordLength-i+2)+FSWSTR("KI");
CheckAndAdd(szTemp);
}
}
// Delete letters: van[n]aema -> vanaema
szTemp=szWordHigh.Mid(1);
CheckAndAdd(szTemp);
for (i=0; i<ipWordLength-1; i++){
if (szTemp[i]!=szWordHigh[i]){
szTemp[i]=szWordHigh[i];
CheckAndAdd(szTemp);
}
}
// Change letters from list
for (i=0; i<ipWordLength; i++){
const FSWCHAR *lpszTo=__SuggestChangeLetters(szWordHigh[i]);
if (!lpszTo) continue;
szTemp=szWordHigh;
for (; lpszTo[0]; lpszTo++){
szTemp[i]=lpszTo[0];
CheckAndAdd(szTemp);
}
}
// Insert letters to word body
for (i=1; i<ipWordLength; i++){
szTemp=szWordHigh.Left(i)+FSWSTR(' ')+szWordHigh.Mid(i);
for (j=0; szInsertLetters[j]; j++){
szTemp[i]=szInsertLetters[j];
CheckAndAdd(szTemp);
}
}
// Insert letters to the beginning
szTemp=CFSWString(FSWSTR(" "))+szWordHigh;
for (i=0; szInsertLettersBeg[i]; i++){
if (szTemp[1]==szInsertLettersBeg[i]) continue;
szTemp[0]=szInsertLettersBeg[i];
CheckAndAdd(szTemp);
}
// Try apostrophe for names
if (szWord[0]!=szWordHigh[0] && szWordHigh.Find('\'')<0){
for (i=0; i<5; i++){
if (i>=ipWordLength) break;
szTemp=szWordHigh.Left(ipWordLength-i)+L'\''+szWordHigh.Mid(ipWordLength-i);
CheckAndAdd(szTemp);
}
}
Order();
RemoveImmoderate();
RemoveDuplicates();
return 0;
}
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;
}
void OnValReadEnd(const CFSAString &szKey, CFSVar &Data) {
if (szKey.IsEmpty()) {
SubKeys("paragraphs", Data);
m_Writer.ObjectEnd();
} else if (szKey=="/paragraphs") {
m_Writer.ArrayEnd();
m_iCollectData++;
} else if (KeyMatch(szKey, "/paragraphs/%d")) {
SubKeys("sentences", Data);
m_Writer.ObjectEnd();
m_iCollectData--;
} else if (KeyMatch(szKey, "/paragraphs/%d/sentences")) {
m_Writer.ArrayEnd();
m_iCollectData++;
} else if (KeyMatch(szKey, "/paragraphs/%d/sentences/%d")) {
if (Data.KeyExist("words")) {
CFSVar &Words=Data["words"];
CFSArray<CMorphInfos> WordsAnalysis;
for (INTPTR ip=0; ip<Words.GetSize(); ip++) {
const CFSVar &Word=Words[ip];
CMorphInfos Analysis;
Analysis.m_szWord=Word["text"].GetWString();
const CFSVar &VarAnalysis=Word["analysis"];
for (INTPTR ip2=0; ip2<VarAnalysis.GetSize(); ip2++) {
const CFSVar &VarAnalysis1=VarAnalysis[ip2];
CMorphInfo Analysis1;
Analysis1.m_szRoot=VarAnalysis1["root"].GetWString();
Analysis1.m_szEnding=VarAnalysis1["ending"].GetWString();
Analysis1.m_szClitic=VarAnalysis1["clitic"].GetWString();
Analysis1.m_cPOS=VarAnalysis1["partofspeech"].GetWString()[0];
Analysis1.m_szForm=VarAnalysis1["form"].GetWString();
Analysis.m_MorphInfo.AddItem(Analysis1);
}
WordsAnalysis.AddItem(Analysis);
}
WordsAnalysis=m_Disambiguator.Disambiguate(WordsAnalysis);
RT_ASSERT(Words.GetSize()==WordsAnalysis.GetSize());
for (INTPTR ip=0; ip<Words.GetSize(); ip++) {
const CMorphInfos &Analysis=WordsAnalysis[ip];
CFSVar VarAnalysis;
VarAnalysis.Cast(CFSVar::VAR_ARRAY);
for (INTPTR ipRes=0; ipRes<Analysis.m_MorphInfo.GetSize(); ipRes++) {
const CMorphInfo &Analysis1=Analysis.m_MorphInfo[ipRes];
CFSVar VarAnalysis1;
VarAnalysis1["root"]=Analysis1.m_szRoot;
VarAnalysis1["ending"]=Analysis1.m_szEnding;
VarAnalysis1["clitic"]=Analysis1.m_szClitic;
VarAnalysis1["partofspeech"]=CFSWString(Analysis1.m_cPOS);
VarAnalysis1["form"]=Analysis1.m_szForm;
VarAnalysis[ipRes]=VarAnalysis1;
}
Words[ip]["analysis"]=VarAnalysis;
}
}
m_Writer.Val(Data);
m_iCollectData--;
}
}
