bool CWord::RightPartIsSurname(int& iH, THomonymGrammems& grammems, Wtroka& strLemma)
{
iH = HasMorphNounWithGrammems_i(TGramBitSet(gSurname));
if (iH != -1) {
CHomonym& h = GetRusHomonym(iH);
grammems = h.Grammems;
strLemma = h.GetLemma();
size_t ii = strLemma.find('-');
YASSERT(ii != Wtroka::npos);
strLemma = strLemma.substr(ii + 1);
return true;
}
//if this word is in morphology - do not try to predict
if (IsDictionary())
return false;
size_t ii = m_txt.find('-');
if (ii == Wtroka::npos)
return false;
Wtroka strRightPart = m_txt.substr(ii + 1);
TMorph::ToLower(strRightPart);
yvector<TSurnamePredictor::TPredictedSurname> out;
if (!TMorph::PredictSurname(strRightPart, out))
return false;
TGrammarBunch newForms;
NSpike::ToGrammarBunch(out[0].StemGrammar, out[0].FlexGrammars, newForms);
grammems.Reset(newForms);
strLemma = out[0].Lemma;
return true;
}
static void MergeGrammems(THomonymGrammems& dst, const TGramBitSet& art_grammems, const TGramBitSet& newPos)
{
// first, reset Part Of Speech if any
if (newPos.HasAny(TMorph::AllPOS()))
dst.SetPOS(newPos);
else if (art_grammems.HasAny(TMorph::AllPOS()))
dst.SetPOS(art_grammems);
// take other grammems from @art_grammems if any
TGramBitSet other = art_grammems & ~TMorph::AllPOS();
if (other.any()) {
// if there is a form with such grammems - just leave it alone and drop the rest ones
bool found = false;
for (THomonymGrammems::TFormIter it = dst.IterForms(); it.Ok(); ++it)
if (it->HasAll(other)) {
dst.ResetSingleForm(*it);
found = true;
}
if (!found) {
// otherwise merge all forms and replace grammems by classes
TGramBitSet newForm = dst.All();
newForm.ReplaceByMaskIfAny(art_grammems, NSpike::AllCases);
newForm.ReplaceByMaskIfAny(art_grammems, NSpike::AllGenders);
newForm.ReplaceByMaskIfAny(art_grammems, NSpike::AllNumbers);
const TGramBitSet anim(gAnimated, gInanimated);
newForm.ReplaceByMaskIfAny(art_grammems, anim);
newForm.ReplaceByMaskIfAny(art_grammems, NSpike::AllTimes);
newForm.ReplaceByMaskIfAny(art_grammems, NSpike::AllPersons);
// just add the rest non-classified grammems
static const TGramBitSet nonclassified = ~(NSpike::AllCases | NSpike::AllGenders | NSpike::AllNumbers |
anim | NSpike::AllTimes | NSpike::AllPersons);
newForm |= art_grammems & nonclassified;
dst.Reset(newForm);
}
}
// if we still do not known POS, apply some workarounds:
if (dst.GetPOS().none()) {
dst.SetPOS(TGramBitSet(gSubstantive));
if (!dst.HasAny(NSpike::AllCases))
dst.Add(NSpike::AllCases);
if (!dst.HasAny(NSpike::AllGenders))
dst.Add(NSpike::AllGenders);
if (!dst.HasAny(NSpike::AllNumbers))
dst.Add(NSpike::AllNumbers);
}
// set a noun or adj without additional grammem as indeclinable
if (!dst.HasAny(~TMorph::AllPOS()) &&
(art_grammems.Has(gSubstantive) || TMorph::IsFullAdjective(art_grammems)))
dst.Add(NSpike::AllCases | NSpike::AllGenders | NSpike::AllNumbers);
}