public static List <PersonItemToken> TryAttach(Pullenti.Ner.Token t)
{
List <PersonItemToken> res = new List <PersonItemToken>();
for (; t != null; t = t.Next)
{
if (t.IsNewlineBefore && res.Count > 0)
{
break;
}
Pullenti.Ner.TextToken tt = t as Pullenti.Ner.TextToken;
if (tt == null)
{
break;
}
string s = tt.Term;
if (!char.IsLetter(s[0]))
{
break;
}
if (((s.Length == 1 || s == "ДЖ")) && !tt.Chars.IsAllLower)
{
Pullenti.Ner.Token t1 = t;
if (t1.Next != null && t1.Next.IsChar('.'))
{
t1 = t1.Next;
}
res.Add(new PersonItemToken(t, t1)
{
Typ = Pullenti.Ner.Org.Internal.OrgItemEponymToken.PersonItemType.Initial, Value = s
});
t = t1;
continue;
}
if (tt.IsAnd)
{
res.Add(new PersonItemToken(t, t)
{
Typ = Pullenti.Ner.Org.Internal.OrgItemEponymToken.PersonItemType.And
});
continue;
}
if (tt.Morph.Class.IsPronoun || tt.Morph.Class.IsPersonalPronoun)
{
break;
}
if (tt.Chars.IsAllLower)
{
Pullenti.Morph.MorphClass mc = tt.GetMorphClassInDictionary();
if (mc.IsPreposition || mc.IsVerb || mc.IsAdverb)
{
break;
}
Pullenti.Ner.Token t1 = t;
if (t1.Next != null && !t1.IsWhitespaceAfter && t1.Next.IsChar('.'))
{
t1 = t1.Next;
}
res.Add(new PersonItemToken(t, t1)
{
Typ = Pullenti.Ner.Org.Internal.OrgItemEponymToken.PersonItemType.LocaseWord, Value = s
});
t = t1;
continue;
}
if (tt.Morph.Class.IsProperName)
{
res.Add(new PersonItemToken(t, t)
{
Typ = Pullenti.Ner.Org.Internal.OrgItemEponymToken.PersonItemType.Name, Value = s
});
}
else if ((t.Next != null && t.Next.IsHiphen && (t.Next.Next is Pullenti.Ner.TextToken)) && !t.Next.IsWhitespaceAfter)
{
res.Add(new PersonItemToken(t, t.Next.Next)
{
Typ = Pullenti.Ner.Org.Internal.OrgItemEponymToken.PersonItemType.Surname, Value = string.Format("{0}-{1}", s, (t.Next.Next as Pullenti.Ner.TextToken).Term)
});
t = t.Next.Next;
}
else
{
res.Add(new PersonItemToken(t, t)
{
Typ = Pullenti.Ner.Org.Internal.OrgItemEponymToken.PersonItemType.Surname, Value = s
});
}
}
return(res.Count > 0 ? res : null);
}
void CorrectWordsByMorph(Pullenti.Morph.MorphLang lang)
{
for (Pullenti.Ner.Token tt = FirstToken; tt != null; tt = tt.Next)
{
if (!(tt is Pullenti.Ner.TextToken))
{
continue;
}
if (tt.Morph.ContainsAttr("прдктв.", null))
{
continue;
}
Pullenti.Morph.MorphClass dd = tt.GetMorphClassInDictionary();
if (!dd.IsUndefined || (tt.LengthChar < 4))
{
continue;
}
if (tt.Morph.Class.IsProperSurname && !tt.Chars.IsAllLower)
{
continue;
}
if (tt.Chars.IsAllUpper)
{
continue;
}
string corw = Pullenti.Morph.MorphologyService.CorrectWord((tt as Pullenti.Ner.TextToken).Term, (tt.Morph.Language.IsUndefined ? lang : tt.Morph.Language));
if (corw == null)
{
continue;
}
List <Pullenti.Morph.MorphToken> ccc = Pullenti.Morph.MorphologyService.Process(corw, lang, null);
if (ccc == null || ccc.Count != 1)
{
continue;
}
Pullenti.Ner.TextToken tt1 = new Pullenti.Ner.TextToken(ccc[0], this, tt.BeginChar, tt.EndChar)
{
Chars = tt.Chars, Term0 = (tt as Pullenti.Ner.TextToken).Term
};
Pullenti.Morph.MorphClass mc = tt1.GetMorphClassInDictionary();
if (mc.IsProperSurname)
{
continue;
}
if (tt == FirstToken)
{
FirstToken = tt1;
}
else
{
tt.Previous.Next = tt1;
}
tt1.Next = tt.Next;
tt = tt1;
if (CorrectedTokens == null)
{
CorrectedTokens = new Dictionary <Pullenti.Ner.Token, string>();
}
CorrectedTokens.Add(tt, tt.GetSourceText());
}
}
static VerbPhraseToken TryParseRu(Pullenti.Ner.Token t, bool canBePartition, bool canBeAdjPartition, bool forceParse)
{
VerbPhraseToken res = null;
Pullenti.Ner.Token t0 = t;
Pullenti.Ner.Token not = null;
bool hasVerb = false;
bool verbBeBefore = false;
PrepositionToken prep = null;
for (; t != null; t = t.Next)
{
if (!(t is Pullenti.Ner.TextToken))
{
break;
}
Pullenti.Ner.TextToken tt = t as Pullenti.Ner.TextToken;
bool isParticiple = false;
if (tt.Term == "НЕ")
{
not = t;
continue;
}
int ty = 0;
string norm = null;
Pullenti.Morph.MorphClass mc = tt.GetMorphClassInDictionary();
if (tt.Term == "НЕТ")
{
if (hasVerb)
{
break;
}
ty = 1;
}
else if (tt.Term == "ДОПУСТИМО")
{
ty = 3;
}
else if (mc.IsAdverb && !mc.IsVerb)
{
ty = 2;
}
else if (tt.IsPureVerb || tt.IsVerbBe)
{
ty = 1;
if (hasVerb)
{
if (!tt.Morph.ContainsAttr("инф.", null))
{
if (verbBeBefore)
{
}
else
{
break;
}
}
}
}
else if (mc.IsVerb)
{
if (mc.IsPreposition || mc.IsMisc || mc.IsPronoun)
{
}
else if (mc.IsNoun)
{
if (tt.Term == "СТАЛИ" || tt.Term == "СТЕКЛО" || tt.Term == "БЫЛИ")
{
ty = 1;
}
else if (!tt.Chars.IsAllLower && !MiscHelper.CanBeStartOfSentence(tt))
{
ty = 1;
}
else if (mc.IsAdjective && canBePartition)
{
ty = 1;
}
else if (forceParse)
{
ty = 1;
}
}
else if (mc.IsProper)
{
if (tt.Chars.IsAllLower)
{
ty = 1;
}
}
else
{
ty = 1;
}
if (mc.IsAdjective)
{
isParticiple = true;
}
if (!tt.Morph.Case.IsUndefined)
{
isParticiple = true;
}
if (!canBePartition && isParticiple)
{
break;
}
if (hasVerb)
{
if (tt.Morph.ContainsAttr("инф.", null))
{
}
else if (!isParticiple)
{
}
else
{
break;
}
}
}
else if ((mc.IsAdjective && tt.Morph.ContainsAttr("к.ф.", null) && tt.Term.EndsWith("О")) && NounPhraseHelper.TryParse(tt, NounPhraseParseAttr.No, 0, null) == null)
{
ty = 2;
}
else if (mc.IsAdjective && ((canBePartition || canBeAdjPartition)))
{
if (tt.Morph.ContainsAttr("к.ф.", null) && !canBeAdjPartition)
{
break;
}
norm = tt.GetNormalCaseText(Pullenti.Morph.MorphClass.Adjective, Pullenti.Morph.MorphNumber.Singular, Pullenti.Morph.MorphGender.Masculine, false);
if (norm.EndsWith("ЙШИЙ"))
{
}
else
{
List <Pullenti.Semantic.Utils.DerivateGroup> grs = Pullenti.Semantic.Utils.DerivateService.FindDerivates(norm, true, null);
if (grs != null && grs.Count > 0)
{
bool hVerb = false;
bool hPart = false;
foreach (Pullenti.Semantic.Utils.DerivateGroup gr in grs)
{
foreach (Pullenti.Semantic.Utils.DerivateWord w in gr.Words)
{
if (w.Class.IsAdjective && w.Class.IsVerb)
{
if (w.Spelling == norm)
{
hPart = true;
}
}
else if (w.Class.IsVerb)
{
hVerb = true;
}
}
}
if (hPart && hVerb)
{
ty = 3;
}
else if (canBeAdjPartition)
{
ty = 3;
}
if (ty != 3 && !string.IsNullOrEmpty(grs[0].Prefix) && norm.StartsWith(grs[0].Prefix))
{
hVerb = false;
hPart = false;
string norm1 = norm.Substring(grs[0].Prefix.Length);
grs = Pullenti.Semantic.Utils.DerivateService.FindDerivates(norm1, true, null);
if (grs != null && grs.Count > 0)
{
foreach (Pullenti.Semantic.Utils.DerivateGroup gr in grs)
{
foreach (Pullenti.Semantic.Utils.DerivateWord w in gr.Words)
{
if (w.Class.IsAdjective && w.Class.IsVerb)
{
if (w.Spelling == norm1)
{
hPart = true;
}
}
else if (w.Class.IsVerb)
{
hVerb = true;
}
}
}
}
if (hPart && hVerb)
{
ty = 3;
}
}
}
}
}
if (ty == 0 && t == t0 && canBePartition)
{
prep = PrepositionHelper.TryParse(t);
if (prep != null)
{
t = prep.EndToken;
continue;
}
}
if (ty == 0)
{
break;
}
if (res == null)
{
res = new VerbPhraseToken(t0, t);
}
res.EndToken = t;
VerbPhraseItemToken it = new VerbPhraseItemToken(t, t)
{
Morph = new Pullenti.Ner.MorphCollection(t.Morph)
};
if (not != null)
{
it.BeginToken = not;
it.Not = true;
not = null;
}
it.IsAdverb = ty == 2;
if (prep != null && !t.Morph.Case.IsUndefined && res.Items.Count == 0)
{
if (((prep.NextCase & t.Morph.Case)).IsUndefined)
{
return(null);
}
it.Morph.RemoveItems(prep.NextCase);
res.Preposition = prep;
}
if (norm == null)
{
norm = t.GetNormalCaseText((ty == 3 ? Pullenti.Morph.MorphClass.Adjective : (ty == 2 ? Pullenti.Morph.MorphClass.Adverb : Pullenti.Morph.MorphClass.Verb)), Pullenti.Morph.MorphNumber.Singular, Pullenti.Morph.MorphGender.Masculine, false);
if (ty == 1 && !tt.Morph.Case.IsUndefined)
{
Pullenti.Morph.MorphWordForm mi = new Pullenti.Morph.MorphWordForm()
{
Case = Pullenti.Morph.MorphCase.Nominative, Number = Pullenti.Morph.MorphNumber.Singular, Gender = Pullenti.Morph.MorphGender.Masculine
};
foreach (Pullenti.Morph.MorphBaseInfo mit in tt.Morph.Items)
{
if (mit is Pullenti.Morph.MorphWordForm)
{
mi.Misc = (mit as Pullenti.Morph.MorphWordForm).Misc;
break;
}
}
string nnn = Pullenti.Morph.MorphologyService.GetWordform("КК" + (t as Pullenti.Ner.TextToken).Term, mi);
if (nnn != null)
{
norm = nnn.Substring(2);
}
}
}
it.Normal = norm;
res.Items.Add(it);
if (!hasVerb && ((ty == 1 || ty == 3)))
{
res.Morph = it.Morph;
hasVerb = true;
}
if (ty == 1 || ty == 3)
{
if (ty == 1 && tt.IsVerbBe)
{
verbBeBefore = true;
}
else
{
verbBeBefore = false;
}
}
}
if (!hasVerb)
{
return(null);
}
for (int i = res.Items.Count - 1; i > 0; i--)
{
if (res.Items[i].IsAdverb)
{
res.Items.RemoveAt(i);
res.EndToken = res.Items[i - 1].EndToken;
}
else
{
break;
}
}
return(res);
}
void CorrectWordsByMerging(Pullenti.Morph.MorphLang lang)
{
for (Pullenti.Ner.Token t = FirstToken; t != null && t.Next != null; t = t.Next)
{
if (!t.Chars.IsLetter || (t.LengthChar < 2))
{
continue;
}
Pullenti.Morph.MorphClass mc0 = t.GetMorphClassInDictionary();
if (t.Morph.ContainsAttr("прдктв.", null))
{
continue;
}
Pullenti.Ner.Token t1 = t.Next;
if (t1.IsHiphen && t1.Next != null && !t1.IsNewlineAfter)
{
t1 = t1.Next;
}
if (t1.LengthChar == 1)
{
continue;
}
if (!t1.Chars.IsLetter || !t.Chars.IsLetter || t1.Chars.IsLatinLetter != t.Chars.IsLatinLetter)
{
continue;
}
if (t1.Chars.IsAllUpper && !t.Chars.IsAllUpper)
{
continue;
}
else if (!t1.Chars.IsAllLower)
{
continue;
}
else if (t.Chars.IsAllUpper)
{
continue;
}
if (t1.Morph.ContainsAttr("прдктв.", null))
{
continue;
}
Pullenti.Morph.MorphClass mc1 = t1.GetMorphClassInDictionary();
if (!mc1.IsUndefined && !mc0.IsUndefined)
{
continue;
}
if (((t as Pullenti.Ner.TextToken).Term.Length + (t1 as Pullenti.Ner.TextToken).Term.Length) < 6)
{
continue;
}
string corw = (t as Pullenti.Ner.TextToken).Term + (t1 as Pullenti.Ner.TextToken).Term;
List <Pullenti.Morph.MorphToken> ccc = Pullenti.Morph.MorphologyService.Process(corw, lang, null);
if (ccc == null || ccc.Count != 1)
{
continue;
}
if (corw == "ПОСТ" || corw == "ВРЕД")
{
continue;
}
Pullenti.Ner.TextToken tt = new Pullenti.Ner.TextToken(ccc[0], this, t.BeginChar, t1.EndChar);
if (tt.GetMorphClassInDictionary().IsUndefined)
{
continue;
}
tt.Chars = t.Chars;
if (t == FirstToken)
{
FirstToken = tt;
}
else
{
t.Previous.Next = tt;
}
if (t1.Next != null)
{
tt.Next = t1.Next;
}
t = tt;
}
}
static OrgItemNameToken _TryAttach(Pullenti.Ner.Token t, OrgItemNameToken prev, bool extOnto)
{
if (t == null)
{
return(null);
}
Pullenti.Ner.Referent r = t.GetReferent();
if (r != null)
{
if (r.TypeName == "DENOMINATION")
{
return new OrgItemNameToken(t, t)
{
Value = r.ToString(true, t.Kit.BaseLanguage, 0), IsDenomination = true
}
}
;
if ((r is Pullenti.Ner.Geo.GeoReferent) && t.Chars.IsLatinLetter)
{
OrgItemNameToken res2 = _TryAttach(t.Next, prev, extOnto);
if (res2 != null && res2.Chars.IsLatinLetter)
{
res2.BeginToken = t;
res2.Value = string.Format("{0} {1}", Pullenti.Ner.Core.MiscHelper.GetTextValueOfMetaToken(t as Pullenti.Ner.MetaToken, Pullenti.Ner.Core.GetTextAttr.No), res2.Value);
res2.IsInDictionary = false;
return(res2);
}
}
return(null);
}
Pullenti.Ner.TextToken tt = t as Pullenti.Ner.TextToken;
if (tt == null)
{
return(null);
}
OrgItemNameToken res = null;
Pullenti.Ner.Core.TerminToken tok = m_StdTails.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok == null && t.IsChar(','))
{
tok = m_StdTails.TryParse(t.Next, Pullenti.Ner.Core.TerminParseAttr.No);
}
if (tok != null)
{
return new OrgItemNameToken(t, tok.EndToken)
{
Value = tok.Termin.CanonicText, IsStdTail = tok.Termin.Tag == null, IsEmptyWord = tok.Termin.Tag != null, Morph = tok.Morph
}
}
;
if ((((tok = m_StdNames.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No)))) != null)
{
return new OrgItemNameToken(t, tok.EndToken)
{
Value = tok.Termin.CanonicText, IsStdName = true
}
}
;
OrgItemEngItem eng = OrgItemEngItem.TryAttach(t, false);
if (eng == null && t.IsChar(','))
{
eng = OrgItemEngItem.TryAttach(t.Next, false);
}
if (eng != null)
{
return new OrgItemNameToken(t, eng.EndToken)
{
Value = eng.FullValue, IsStdTail = true
}
}
;
if (tt.Chars.IsAllLower && prev != null)
{
if (!prev.Chars.IsAllLower && !prev.Chars.IsCapitalUpper)
{
return(null);
}
}
if (tt.IsChar(',') && prev != null)
{
Pullenti.Ner.Core.NounPhraseToken npt1 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t.Next, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt1 == null || npt1.Chars != prev.Chars || ((npt1.Morph.Case & prev.Morph.Case)).IsUndefined)
{
return(null);
}
OrgItemTypeToken ty = OrgItemTypeToken.TryAttach(t.Next, false, null);
if (ty != null)
{
return(null);
}
if (npt1.EndToken.Next == null || !npt1.EndToken.Next.IsValue("И", null))
{
return(null);
}
Pullenti.Ner.Token t1 = npt1.EndToken.Next;
Pullenti.Ner.Core.NounPhraseToken npt2 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t1.Next, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt2 == null || npt2.Chars != prev.Chars || ((npt2.Morph.Case & npt1.Morph.Case & prev.Morph.Case)).IsUndefined)
{
return(null);
}
ty = OrgItemTypeToken.TryAttach(t1.Next, false, null);
if (ty != null)
{
return(null);
}
res = new OrgItemNameToken(npt1.BeginToken, npt1.EndToken)
{
Morph = npt1.Morph, Value = npt1.GetNormalCaseText(null, Pullenti.Morph.MorphNumber.Undefined, Pullenti.Morph.MorphGender.Undefined, false)
};
res.IsNounPhrase = true;
res.IsAfterConjunction = true;
if (prev.Preposition != null)
{
res.Preposition = prev.Preposition;
}
return(res);
}
if (((tt.IsChar('&') || tt.IsValue("AND", null) || tt.IsValue("UND", null))) && prev != null)
{
if ((tt.Next is Pullenti.Ner.TextToken) && tt.LengthChar == 1 && tt.Next.Chars.IsLatinLetter)
{
res = new OrgItemNameToken(tt, tt.Next)
{
Chars = tt.Next.Chars
};
res.IsAfterConjunction = true;
res.Value = "& " + (tt.Next as Pullenti.Ner.TextToken).Term;
return(res);
}
res = OrgItemNameToken.TryAttach(tt.Next, null, extOnto, false);
if (res == null || res.Chars != prev.Chars)
{
return(null);
}
res.IsAfterConjunction = true;
res.Value = "& " + res.Value;
return(res);
}
if (!tt.Chars.IsLetter)
{
return(null);
}
List <Pullenti.Semantic.Utils.DerivateGroup> expinf = null;
if (prev != null && prev.EndToken.GetMorphClassInDictionary().IsNoun)
{
string wo = prev.EndToken.GetNormalCaseText(Pullenti.Morph.MorphClass.Noun, Pullenti.Morph.MorphNumber.Singular, Pullenti.Morph.MorphGender.Undefined, false);
expinf = Pullenti.Semantic.Utils.DerivateService.FindDerivates(wo, true, prev.EndToken.Morph.Language);
}
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null && npt.InternalNoun != null)
{
npt = null;
}
bool explOk = false;
if (npt != null && prev != null && prev.EndToken.GetMorphClassInDictionary().IsNoun)
{
Pullenti.Ner.Core.NounPhraseToken npt0 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(prev.EndToken, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt0 != null)
{
List <Pullenti.Semantic.Core.SemanticLink> links = Pullenti.Semantic.Core.SemanticHelper.TryCreateLinks(npt0, npt, null);
if (links.Count > 0)
{
explOk = true;
}
}
}
if (npt != null && ((explOk || npt.Morph.Case.IsGenitive || ((prev != null && !((prev.Morph.Case & npt.Morph.Case)).IsUndefined)))))
{
Pullenti.Morph.MorphClass mc = npt.BeginToken.GetMorphClassInDictionary();
if (mc.IsVerb || mc.IsPronoun)
{
return(null);
}
if (mc.IsAdverb)
{
if (npt.BeginToken.Next != null && npt.BeginToken.Next.IsHiphen)
{
}
else
{
return(null);
}
}
if (mc.IsPreposition)
{
return(null);
}
if (mc.IsNoun && npt.Chars.IsAllLower)
{
Pullenti.Morph.MorphCase ca = npt.Morph.Case;
if ((!ca.IsDative && !ca.IsGenitive && !ca.IsInstrumental) && !ca.IsPrepositional)
{
return(null);
}
}
res = new OrgItemNameToken(npt.BeginToken, npt.EndToken)
{
Morph = npt.Morph, Value = npt.GetNormalCaseText(null, Pullenti.Morph.MorphNumber.Undefined, Pullenti.Morph.MorphGender.Undefined, false)
};
res.IsNounPhrase = true;
if ((npt.EndToken.WhitespacesAfterCount < 2) && (npt.EndToken.Next is Pullenti.Ner.TextToken))
{
Pullenti.Ner.Core.NounPhraseToken npt2 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(npt.EndToken.Next, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt2 != null && npt2.Morph.Case.IsGenitive && npt2.Chars.IsAllLower)
{
OrgItemTypeToken typ = OrgItemTypeToken.TryAttach(npt.EndToken.Next, true, null);
OrgItemEponymToken epo = OrgItemEponymToken.TryAttach(npt.EndToken.Next, false);
Pullenti.Ner.ReferentToken rtt = t.Kit.ProcessReferent("PERSONPROPERTY", npt.EndToken.Next);
if (typ == null && epo == null && ((rtt == null || rtt.Morph.Number == Pullenti.Morph.MorphNumber.Plural)))
{
res.EndToken = npt2.EndToken;
res.Value = string.Format("{0} {1}", res.Value, Pullenti.Ner.Core.MiscHelper.GetTextValueOfMetaToken(npt2, Pullenti.Ner.Core.GetTextAttr.No));
}
}
else if (npt.EndToken.Next.IsComma && (npt.EndToken.Next.Next is Pullenti.Ner.TextToken))
{
Pullenti.Ner.Token tt2 = npt.EndToken.Next.Next;
Pullenti.Morph.MorphClass mv2 = tt2.GetMorphClassInDictionary();
if (mv2.IsAdjective && mv2.IsVerb)
{
Pullenti.Morph.MorphBaseInfo bi = new Pullenti.Morph.MorphBaseInfo()
{
Case = npt.Morph.Case, Gender = npt.Morph.Gender, Number = npt.Morph.Number
};
if (tt2.Morph.CheckAccord(bi, false, false))
{
npt2 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(tt2.Next, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt2 != null && ((npt2.Morph.Case.IsDative || npt2.Morph.Case.IsGenitive)) && npt2.Chars.IsAllLower)
{
res.EndToken = npt2.EndToken;
res.Value = string.Format("{0} {1}", res.Value, Pullenti.Ner.Core.MiscHelper.GetTextValue(npt.EndToken.Next, res.EndToken, Pullenti.Ner.Core.GetTextAttr.No));
}
}
}
}
}
if (explOk)
{
res.IsAfterConjunction = true;
}
}
else if (npt != null && ((((prev != null && prev.IsNounPhrase && npt.Morph.Case.IsInstrumental)) || extOnto)))
{
res = new OrgItemNameToken(npt.BeginToken, npt.EndToken)
{
Morph = npt.Morph, Value = npt.GetNormalCaseText(null, Pullenti.Morph.MorphNumber.Undefined, Pullenti.Morph.MorphGender.Undefined, false)
};
res.IsNounPhrase = true;
}
else if (tt.IsAnd)
{
res = TryAttach(tt.Next, prev, extOnto, false);
if (res == null || !res.IsNounPhrase || prev == null)
{
return(null);
}
if (((prev.Morph.Case & res.Morph.Case)).IsUndefined)
{
return(null);
}
if (prev.Morph.Number != Pullenti.Morph.MorphNumber.Undefined && res.Morph.Number != Pullenti.Morph.MorphNumber.Undefined)
{
if (((prev.Morph.Number & res.Morph.Number)) == Pullenti.Morph.MorphNumber.Undefined)
{
if (prev.Chars != res.Chars)
{
return(null);
}
OrgItemTypeToken ty = OrgItemTypeToken.TryAttach(res.EndToken.Next, false, null);
if (ty != null)
{
return(null);
}
}
}
Pullenti.Morph.CharsInfo ci = res.Chars;
res.Chars = ci;
res.IsAfterConjunction = true;
return(res);
}
else if (((tt.Term == "ПО" || tt.Term == "ПРИ" || tt.Term == "ЗА") || tt.Term == "С" || tt.Term == "В") || tt.Term == "НА")
{
npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t.Next, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null)
{
if (m_VervotWords.TryParse(npt.EndToken, Pullenti.Ner.Core.TerminParseAttr.No) != null)
{
return(null);
}
bool ok = false;
if (tt.Term == "ПО")
{
ok = npt.Morph.Case.IsDative;
}
else if (tt.Term == "С")
{
ok = npt.Morph.Case.IsInstrumental;
}
else if (tt.Term == "ЗА")
{
ok = npt.Morph.Case.IsGenitive | npt.Morph.Case.IsInstrumental;
}
else if (tt.Term == "НА")
{
ok = npt.Morph.Case.IsPrepositional;
}
else if (tt.Term == "В")
{
ok = npt.Morph.Case.IsDative | npt.Morph.Case.IsPrepositional;
if (ok)
{
ok = false;
if (t.Next.IsValue("СФЕРА", null) || t.Next.IsValue("ОБЛАСТЬ", null))
{
ok = true;
}
}
}
else if (tt.Term == "ПРИ")
{
ok = npt.Morph.Case.IsPrepositional;
if (ok)
{
if (OrgItemTypeToken.TryAttach(tt.Next, true, null) != null)
{
ok = false;
}
else
{
Pullenti.Ner.ReferentToken rt = tt.Kit.ProcessReferent(Pullenti.Ner.Org.OrganizationAnalyzer.ANALYZER_NAME, tt.Next);
if (rt != null)
{
ok = false;
}
}
}
string s = npt.Noun.GetNormalCaseText(null, Pullenti.Morph.MorphNumber.Undefined, Pullenti.Morph.MorphGender.Undefined, false);
if (s == "ПОДДЕРЖКА" || s == "УЧАСТИЕ")
{
ok = false;
}
}
else
{
ok = npt.Morph.Case.IsPrepositional;
}
if (ok)
{
res = new OrgItemNameToken(t, npt.EndToken)
{
Morph = npt.Morph, Value = npt.GetNormalCaseText(null, Pullenti.Morph.MorphNumber.Singular, Pullenti.Morph.MorphGender.Undefined, false), Chars = npt.Chars
};
res.IsNounPhrase = true;
res.Preposition = tt.Term;
if (((res.Value == "ДЕЛО" || res.Value == "ВОПРОС")) && !res.IsNewlineAfter)
{
OrgItemNameToken res2 = _TryAttach(res.EndToken.Next, res, extOnto);
if (res2 != null && res2.Morph.Case.IsGenitive)
{
res.Value = string.Format("{0} {1}", res.Value, res2.Value);
res.EndToken = res2.EndToken;
for (Pullenti.Ner.Token ttt = res2.EndToken.Next; ttt != null; ttt = ttt.Next)
{
if (!ttt.IsCommaAnd)
{
break;
}
OrgItemNameToken res3 = _TryAttach(ttt.Next, res2, extOnto);
if (res3 == null)
{
break;
}
res.Value = string.Format("{0} {1}", res.Value, res3.Value);
res.EndToken = res3.EndToken;
if (ttt.IsAnd)
{
break;
}
ttt = res.EndToken;
}
}
}
}
}
if (res == null)
{
return(null);
}
}
else if (tt.Term == "OF")
{
Pullenti.Ner.Token t1 = tt.Next;
if (t1 != null && Pullenti.Ner.Core.MiscHelper.IsEngArticle(t1))
{
t1 = t1.Next;
}
if (t1 != null && t1.Chars.IsLatinLetter && !t1.Chars.IsAllLower)
{
res = new OrgItemNameToken(t, t1)
{
Chars = t1.Chars, Morph = t1.Morph
};
for (Pullenti.Ner.Token ttt = t1.Next; ttt != null; ttt = ttt.Next)
{
if (ttt.WhitespacesBeforeCount > 2)
{
break;
}
if (Pullenti.Ner.Core.MiscHelper.IsEngAdjSuffix(ttt))
{
ttt = ttt.Next;
continue;
}
if (!ttt.Chars.IsLatinLetter)
{
break;
}
if (ttt.Morph.Class.IsPreposition)
{
break;
}
t1 = (res.EndToken = ttt);
}
res.Value = Pullenti.Ner.Core.MiscHelper.GetTextValue(t, t1, Pullenti.Ner.Core.GetTextAttr.IgnoreArticles);
res.Preposition = tt.Term;
return(res);
}
}
if (res == null)
{
if (tt.Chars.IsLatinLetter && tt.LengthChar == 1)
{
}
else if (tt.Chars.IsAllLower || (tt.LengthChar < 2))
{
if (!tt.Chars.IsLatinLetter || prev == null || !prev.Chars.IsLatinLetter)
{
return(null);
}
}
if (tt.Chars.IsCyrillicLetter)
{
Pullenti.Morph.MorphClass mc = tt.GetMorphClassInDictionary();
if (mc.IsVerb || mc.IsAdverb)
{
return(null);
}
}
else if (tt.Chars.IsLatinLetter && !tt.IsWhitespaceAfter)
{
if (!tt.IsWhitespaceAfter && (tt.LengthChar < 5))
{
if (tt.Next is Pullenti.Ner.NumberToken)
{
return(null);
}
}
}
res = new OrgItemNameToken(tt, tt)
{
Value = tt.Term, Morph = tt.Morph
};
for (t = tt.Next; t != null; t = t.Next)
{
if ((((t.IsHiphen || t.IsCharOf("\\/"))) && t.Next != null && (t.Next is Pullenti.Ner.TextToken)) && !t.IsWhitespaceBefore && !t.IsWhitespaceAfter)
{
t = t.Next;
res.EndToken = t;
res.Value = string.Format("{0}{1}{2}", res.Value, (t.Previous.IsChar('.') ? '.' : '-'), (t as Pullenti.Ner.TextToken).Term);
}
else if (t.IsChar('.'))
{
if (!t.IsWhitespaceAfter && !t.IsWhitespaceBefore && (t.Next is Pullenti.Ner.TextToken))
{
res.EndToken = t.Next;
t = t.Next;
res.Value = string.Format("{0}.{1}", res.Value, (t as Pullenti.Ner.TextToken).Term);
}
else if ((t.Next != null && !t.IsNewlineAfter && t.Next.Chars.IsLatinLetter) && tt.Chars.IsLatinLetter)
{
res.EndToken = t;
}
else
{
break;
}
}
else
{
break;
}
}
}
for (Pullenti.Ner.Token t0 = res.BeginToken; t0 != null; t0 = t0.Next)
{
if ((((tt = t0 as Pullenti.Ner.TextToken))) != null && tt.IsLetters)
{
if (!tt.Morph.Class.IsConjunction && !tt.Morph.Class.IsPreposition)
{
foreach (Pullenti.Morph.MorphBaseInfo mf in tt.Morph.Items)
{
if ((mf as Pullenti.Morph.MorphWordForm).IsInDictionary)
{
res.IsInDictionary = true;
}
}
}
}
if (t0 == res.EndToken)
{
break;
}
}
if (res.BeginToken == res.EndToken && res.BeginToken.Chars.IsAllUpper)
{
if (res.EndToken.Next != null && !res.EndToken.IsWhitespaceAfter)
{
Pullenti.Ner.Token t1 = res.EndToken.Next;
if (t1.Next != null && !t1.IsWhitespaceAfter && t1.IsHiphen)
{
t1 = t1.Next;
}
if (t1 is Pullenti.Ner.NumberToken)
{
res.Value += (t1 as Pullenti.Ner.NumberToken).Value;
res.EndToken = t1;
}
}
}
if (res.BeginToken == res.EndToken && res.BeginToken.Chars.IsLastLower)
{
string src = res.BeginToken.GetSourceText();
for (int i = src.Length - 1; i >= 0; i--)
{
if (char.IsUpper(src[i]))
{
res.Value = src.Substring(0, i + 1);
break;
}
}
}
return(res);
}