TGramBitSet THomonymInflector::SelectBestHomoform(const CHomonym& hom, const TGramBitSet& desired) {
if (!hom.Grammems.HasForms())
return TGramBitSet();
TGramBitSet res;
size_t bestCommon = 0;
bool bestSingular = 0;
size_t bestNormal = 0;
// select a form with the maximum of @desired grammems
for (THomonymGrammems::TFormIter it = hom.Grammems.IterForms(); it.Ok(); ++it) {
size_t curCommon = (*it & desired).count();
bool curSingular = it->Has(gSingular);
size_t curNormal = NInfl::DefaultFeatures().NormalMutable(*it).count();
bool better = false;
if (bestCommon < curCommon)
better = true;
else if (bestCommon == curCommon) {
if (curSingular && !bestSingular)
better = true;
else if (curSingular == bestSingular && bestNormal < curNormal)
better = true;
}
if (better) {
res = *it;
bestCommon = curCommon;
bestSingular = curSingular;
bestNormal = curNormal;
}
}
return res;
}
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);
}
bool THomonymInflector::FindInForms(const THomonymGrammems& forms, const TGramBitSet& grammems, TGramBitSet& resgram) {
using NInfl::DefaultFeatures;
using NInfl::TFeature;
if (forms.HasForms()) {
for (THomonymGrammems::TFormIter it = forms.IterForms(); it.Ok(); ++it)
if (it->HasAll(grammems)) {
resgram = *it;
return true;
}
} else if (forms.IsIndeclinable() && DefaultFeatures().BitSet(TFeature::Case, TFeature::Number).HasAll(grammems)) {
resgram = grammems;
return true;
}
return false;
}
void CAnalyticFormBuilder::ChangeGrammemsAsAnalyticForm(CHomonym& H, const CHomonym& VerbHomonym)
{
THomonymGrammems old_grammems;
H.Grammems.Swap(old_grammems);
for (THomonymGrammems::TFormIter old = old_grammems.IterForms(); old.Ok(); ++old)
for (THomonymGrammems::TFormIter verb = VerbHomonym.Grammems.IterForms(); verb.Ok(); ++verb) {
Stroka strPos;
// auxiliary verb grammems
const TGramBitSet& VerbGrammems = *verb;
// meaningful part grammems
TGramBitSet MainGrammems = *old;
// final grammems to set
TGramBitSet ResultedGrammems;
if (MainGrammems.Has(gInfinitive)) {
ResultedGrammems = MainGrammems & ~TMorph::AllPOS();
if (VerbGrammems.Has(gImperative)) {
// analytical form for imperatives in singular number does not exist
if (VerbGrammems.Has(gSingular))
continue;
ResultedGrammems.Set(gImperative); // "будем же жить!"
} else
ResultedGrammems |= VerbGrammems & NSpike::AllTimes; // "я стал пилить" или "стану писать"
ResultedGrammems |= VerbGrammems & NSpike::AllPersons;
ResultedGrammems |= VerbGrammems & NSpike::AllNumbers;
ResultedGrammems |= VerbGrammems & NSpike::AllGenders;
//copy all POS grammems from verb to main
ResultedGrammems |= VerbGrammems & TMorph::AllPOS();
H.PutAuxArticle(SDictIndex(TEMPLATE_DICT, VerbHomonym.GetAuxArticleIndex(TEMPLATE_DICT)));
strPos = "Г";
} else if (TMorph::IsShortParticiple(MainGrammems)) {
// "*будем же взяты!"
if (VerbGrammems.Has(gImperative))
continue;
ResultedGrammems = MainGrammems & ~TMorph::AllPOS();
// remove any time grammems from participle
ResultedGrammems &= ~NSpike::AllTimes;
ResultedGrammems |= VerbGrammems & NSpike::AllPersons;
ResultedGrammems |= VerbGrammems & NSpike::AllTimes;
if (VerbGrammems.Has(gImperative)) // ??? the same check second time, always false?
ResultedGrammems.Set(gImperative);
strPos = "ПРИЧАСТИЕ";
ResultedGrammems |= TGramBitSet(gParticiple, gShort);
} else if (TMorph::IsShortAdjective(MainGrammems)) {
if (VerbGrammems.Has(gImperative))
continue; // будем cчитать, что "будем же красивы!" - это плохо!
// на самом деле, просто не хочется вводить ее кучу кодов.
ResultedGrammems = VerbGrammems;
ResultedGrammems |= MainGrammems & (NSpike::AllNumbers | NSpike::AllGenders | TGramBitSet(gAnimated, gInanimated));
ResultedGrammems &= ~TMorph::AllPOS();
if (ResultedGrammems.Has(gActive))
ResultedGrammems &= ~TGramBitSet(gActive);
ResultedGrammems |= TGramBitSet(gAdjective, gShort);
strPos = "П";
} else if (MainGrammems.Has(gPraedic)) // "мне было больно"
{
ResultedGrammems = VerbGrammems;
ResultedGrammems |= NSpike::AllCases & MainGrammems; //copied from PronounPredk code (commented below) - preserve cases if any
ResultedGrammems &= ~TMorph::AllPOS();
if (ResultedGrammems.Has(gActive))
ResultedGrammems.Reset(gActive);
strPos = "ПРЕДК";
ResultedGrammems |= MainGrammems & TMorph::AllPOS();
} else if (MainGrammems.Has(gComparative)) // он был больше тебя
{
ResultedGrammems = (VerbGrammems & ~TMorph::AllPOS()) | TGramBitSet(gComparative);
if (ResultedGrammems.Has(gActive))
ResultedGrammems.Reset(gActive);
strPos = "П";
ResultedGrammems |= MainGrammems & TMorph::AllPOS();
} else if (TMorph::IsFullAdjective(MainGrammems))
// resolve disambiguity of homonyms, because analytical forms with full adjectives do not exist.
continue;
// "стал писать" "стану писать" "стать писать" - совершенный вид
if (VerbHomonym.Lemma == kStat)
ResultedGrammems.Reset(gImperfect).Set(gPerfect);
// if the auxiliary verb was an infinitive then it is all an infinitive
// "быть лучше" или "должно быть принесено"
if (VerbHomonym.HasGrammem(gInfinitive)) {
ResultedGrammems &= ~TMorph::AllPOS();
ResultedGrammems.Set(gInfinitive);
strPos = "ИНФИНИТИВ";
} else if (VerbHomonym.HasGrammem(gGerund)) // "будучи лишней"
{
ResultedGrammems &= ~TMorph::AllPOS();
ResultedGrammems.Set(gGerund);
strPos = "ДЕЕПРИЧАСТИЕ";
}
if (strPos.empty())
continue;
/* do some corrections (code taken from RusGramTab.ProcessPOSAndGrammems) */
if (ResultedGrammems.HasAll(NSpike::AllCases | TGramBitSet(gAdjPronoun)))
ResultedGrammems |= NSpike::AllGenders | NSpike::AllNumbers;
if (ResultedGrammems.Has(gMasFem))
ResultedGrammems |= TGramBitSet(gMasculine, gFeminine);
if (!ResultedGrammems.Has(gPraedic) && ResultedGrammems.HasAll(NSpike::AllCases) && !ResultedGrammems.Has(gSingular))
ResultedGrammems |= TGramBitSet(gSingular, gPlural);
H.Grammems.AddForm(ResultedGrammems);
}
}