internal virtual void Deserialize(Stream stream, Pullenti.Ner.Core.AnalysisKit kit, int vers) { Kit = kit; m_BeginChar = Pullenti.Ner.Core.Internal.SerializerHelper.DeserializeInt(stream); m_EndChar = Pullenti.Ner.Core.Internal.SerializerHelper.DeserializeInt(stream); m_Attrs = (short)Pullenti.Ner.Core.Internal.SerializerHelper.DeserializeInt(stream); Chars = new Pullenti.Morph.CharsInfo() { Value = (short)Pullenti.Ner.Core.Internal.SerializerHelper.DeserializeInt(stream) }; m_Morph = new MorphCollection(); m_Morph.Deserialize(stream); }
static string corrChars(string str, Pullenti.Morph.CharsInfo ci, bool keepChars) { if (!keepChars) { return(str); } if (ci.IsAllLower) { return(str.ToLower()); } if (ci.IsCapitalUpper) { return(MiscHelper.ConvertFirstCharUpperAndOtherLower(str)); } return(str); }
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); }
public MetaToken(Token begin, Token end, Pullenti.Ner.Core.AnalysisKit kit = null) : base((kit != null ? kit : (begin != null ? begin.Kit : null)), (begin == null ? 0 : begin.BeginChar), (end == null ? 0 : end.EndChar)) { if (begin == this || end == this) { } m_BeginToken = begin; m_EndToken = end; if (begin == null || end == null) { return; } Chars = begin.Chars; if (begin != end) { for (Token t = begin.Next; t != null; t = t.Next) { if (t.Chars.IsLetter) { if (Chars.IsCapitalUpper && t.Chars.IsAllLower) { } else { Chars = new Pullenti.Morph.CharsInfo() { Value = (short)((Chars.Value & t.Chars.Value)) } }; } if (t == end) { break; } } } } void _RefreshCharsInfo() { if (m_BeginToken == null) { return; } Chars = m_BeginToken.Chars; int cou = 0; if (m_BeginToken != m_EndToken && m_EndToken != null) { for (Token t = m_BeginToken.Next; t != null; t = t.Next) { if ((++cou) > 100) { break; } if (t.EndChar > m_EndToken.EndChar) { break; } if (t.Chars.IsLetter) { Chars = new Pullenti.Morph.CharsInfo() { Value = (short)((Chars.Value & t.Chars.Value)) } } ; if (t == m_EndToken) { break; } } } }