void CMultiWordCreator::CollectArticleInfo(const SArtPointer& artP, EDicType dicType,
yvector<TGztArticle>& gztArticles,
yvector<SDictIndex>& auxArticles,
bool hasTextKeys) const
{
hasTextKeys = false;
const CDictsHolder& dicts = *GlobalDictsHolder;
const TGazetteer& gzt = *dicts.Gazetteer();
const CIndexedDictionary& auxDict = dicts.GetDict(dicType);
YASSERT(artP.IsValid());
if (artP.HasStrType()) {
NGzt::TArticlePtr art;
if (gzt.ArticlePool().FindCustomArticleByName(artP.GetStrType(), art))
gztArticles.push_back(TGztArticle(art));
else {
int iArt = auxDict.get_article_index(artP.GetStrType());
if (iArt != -1)
auxArticles.push_back(SDictIndex(dicType, iArt));
}
} else if (artP.GetKWType().GetDescriptor() == TGztArticle::AuxDicArticleDescriptor()) {
// do not search gzt articles by TAuxDictArticle kw-type
} else {
yhash<TKeyWordType, SKWTypeInfo>::const_iterator kwtype_info = auxDict.m_KWType2Articles.find(artP.GetKWType());
if (kwtype_info == auxDict.m_KWType2Articles.end() || kwtype_info->second.m_bOnlyTextContent)
hasTextKeys = true;
else
for (size_t i = 0; i < kwtype_info->second.m_Articles.size(); ++i)
auxArticles.push_back(SDictIndex(dicType, kwtype_info->second.m_Articles[i]));
const yset<ui32>* custom_offsets = gzt.ArticlePool().FindCustomArticleOffsetsByType(artP.GetKWType().GetDescriptor());
if (custom_offsets != NULL)
for (yset<ui32>::const_iterator it = custom_offsets->begin(); it != custom_offsets->end(); ++it)
gztArticles.push_back(TGztArticle(*it, gzt.ArticlePool()));
}
if (!hasTextKeys)
for (size_t i = 0; i < gztArticles.size(); ++i)
if (gztArticles[i].HasNonCustomKey()) {
hasTextKeys = true;
break;
}
}
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);
}
}
