public static OrgItemEponymToken TryAttach(Pullenti.Ner.Token t, bool mustHasPrefix = false)
{
Pullenti.Ner.TextToken tt = t as Pullenti.Ner.TextToken;
if (tt == null)
{
if (t == null)
{
return(null);
}
Pullenti.Ner.Referent r1 = t.GetReferent();
if (r1 != null && r1.TypeName == "DATE")
{
string str = r1.ToString().ToUpper();
if ((str == "1 МАЯ" || str == "7 ОКТЯБРЯ" || str == "9 МАЯ") || str == "8 МАРТА")
{
OrgItemEponymToken dt = new OrgItemEponymToken(t, t)
{
Eponyms = new List <string>()
};
dt.Eponyms.Add(str);
return(dt);
}
}
Pullenti.Ner.NumberToken age = Pullenti.Ner.Core.NumberHelper.TryParseAge(t);
if ((age != null && (((age.EndToken.Next is Pullenti.Ner.TextToken) || (age.EndToken.Next is Pullenti.Ner.ReferentToken))) && (age.WhitespacesAfterCount < 3)) && !age.EndToken.Next.Chars.IsAllLower && age.EndToken.Next.Chars.IsCyrillicLetter)
{
OrgItemEponymToken dt = new OrgItemEponymToken(t, age.EndToken.Next)
{
Eponyms = new List <string>()
};
dt.Eponyms.Add(string.Format("{0} {1}", age.Value, dt.EndToken.GetSourceText().ToUpper()));
return(dt);
}
return(null);
}
Pullenti.Ner.Token t1 = null;
bool full = false;
bool hasName = false;
if (tt.Term == "ИМЕНИ" || tt.Term == "ІМЕНІ")
{
t1 = t.Next;
full = true;
hasName = true;
}
else if (((tt.Term == "ИМ" || tt.Term == "ІМ")) && tt.Next != null)
{
if (tt.Next.IsChar('.'))
{
t1 = tt.Next.Next;
full = true;
}
else if ((tt.Next is Pullenti.Ner.TextToken) && tt.Chars.IsAllLower && !tt.Next.Chars.IsAllLower)
{
t1 = tt.Next;
}
hasName = true;
}
else if (tt.Previous != null && ((tt.Previous.IsValue("ФОНД", null) || tt.Previous.IsValue("ХРАМ", null) || tt.Previous.IsValue("ЦЕРКОВЬ", "ЦЕРКВА"))))
{
if ((!tt.Chars.IsCyrillicLetter || tt.Morph.Class.IsPreposition || tt.Morph.Class.IsConjunction) || !tt.Chars.IsLetter)
{
return(null);
}
if (tt.WhitespacesBeforeCount != 1)
{
return(null);
}
if (tt.Chars.IsAllLower)
{
return(null);
}
if (tt.Morph.Class.IsAdjective)
{
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(tt, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null && npt.BeginToken != npt.EndToken)
{
return(null);
}
}
OrgItemNameToken na = OrgItemNameToken.TryAttach(tt, null, false, true);
if (na != null)
{
if (na.IsEmptyWord || na.IsStdName || na.IsStdTail)
{
return(null);
}
}
t1 = tt;
}
if (t1 == null || ((t1.IsNewlineBefore && !full)))
{
return(null);
}
if (tt.Previous != null && tt.Previous.Morph.Class.IsPreposition)
{
return(null);
}
if (mustHasPrefix && !hasName)
{
return(null);
}
Pullenti.Ner.Referent r = t1.GetReferent();
if ((r != null && r.TypeName == "DATE" && full) && r.FindSlot("DAY", null, true) != null && r.FindSlot("YEAR", null, true) == null)
{
OrgItemEponymToken dt = new OrgItemEponymToken(t, t1)
{
Eponyms = new List <string>()
};
dt.Eponyms.Add(r.ToString().ToUpper());
return(dt);
}
bool holy = false;
if ((t1.IsValue("СВЯТОЙ", null) || t1.IsValue("СВЯТИЙ", null) || t1.IsValue("СВ", null)) || t1.IsValue("СВЯТ", null))
{
t1 = t1.Next;
holy = true;
if (t1 != null && t1.IsChar('.'))
{
t1 = t1.Next;
}
}
if (t1 == null)
{
return(null);
}
Pullenti.Morph.MorphClass cl = t1.GetMorphClassInDictionary();
if (cl.IsNoun || cl.IsAdjective)
{
Pullenti.Ner.ReferentToken rt = t1.Kit.ProcessReferent("PERSON", t1);
if (rt != null && rt.Referent.TypeName == "PERSON" && rt.BeginToken != rt.EndToken)
{
string e = rt.Referent.GetStringValue("LASTNAME");
if (e != null)
{
if (rt.EndToken.IsValue(e, null))
{
OrgItemEponymToken re = new OrgItemEponymToken(t, rt.EndToken);
re.Eponyms.Add(rt.EndToken.GetSourceText());
return(re);
}
}
}
}
Pullenti.Ner.NumberToken nt = Pullenti.Ner.Core.NumberHelper.TryParseAnniversary(t1);
if (nt != null && nt.Typ == Pullenti.Ner.NumberSpellingType.Age)
{
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(nt.EndToken.Next, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null)
{
string s = string.Format("{0}-{1} {2}", nt.Value, (t.Kit.BaseLanguage.IsUa ? "РОКІВ" : "ЛЕТ"), Pullenti.Ner.Core.MiscHelper.GetTextValue(npt.BeginToken, npt.EndToken, Pullenti.Ner.Core.GetTextAttr.No));
OrgItemEponymToken res = new OrgItemEponymToken(t, npt.EndToken);
res.Eponyms.Add(s);
return(res);
}
}
List <PersonItemToken> its = PersonItemToken.TryAttach(t1);
if (its == null)
{
if ((t1 is Pullenti.Ner.ReferentToken) && (t1.GetReferent() is Pullenti.Ner.Geo.GeoReferent))
{
string s = Pullenti.Ner.Core.MiscHelper.GetTextValue(t1, t1, Pullenti.Ner.Core.GetTextAttr.No);
OrgItemEponymToken re = new OrgItemEponymToken(t, t1);
re.Eponyms.Add(s);
return(re);
}
return(null);
}
List <string> eponims = new List <string>();
int i = 0;
int j;
if (its[i].Typ == PersonItemType.LocaseWord)
{
i++;
}
if (i >= its.Count)
{
return(null);
}
if (!full)
{
if (its[i].BeginToken.Morph.Class.IsAdjective && !its[i].BeginToken.Morph.Class.IsProperSurname)
{
return(null);
}
}
if (its[i].Typ == PersonItemType.Initial)
{
i++;
while (true)
{
if ((i < its.Count) && its[i].Typ == PersonItemType.Initial)
{
i++;
}
if (i >= its.Count || ((its[i].Typ != PersonItemType.Surname && its[i].Typ != PersonItemType.Name)))
{
break;
}
eponims.Add(its[i].Value);
t1 = its[i].EndToken;
if ((i + 2) >= its.Count || its[i + 1].Typ != PersonItemType.And || its[i + 2].Typ != PersonItemType.Initial)
{
break;
}
i += 3;
}
}
else if (((i + 1) < its.Count) && its[i].Typ == PersonItemType.Name && its[i + 1].Typ == PersonItemType.Surname)
{
eponims.Add(its[i + 1].Value);
t1 = its[i + 1].EndToken;
i += 2;
if ((((i + 2) < its.Count) && its[i].Typ == PersonItemType.And && its[i + 1].Typ == PersonItemType.Name) && its[i + 2].Typ == PersonItemType.Surname)
{
eponims.Add(its[i + 2].Value);
t1 = its[i + 2].EndToken;
}
}
else if (its[i].Typ == PersonItemType.Surname)
{
if (its.Count == (i + 2) && its[i].Chars == its[i + 1].Chars)
{
its[i].Value += (" " + its[i + 1].Value);
its[i].EndToken = its[i + 1].EndToken;
its.RemoveAt(i + 1);
}
eponims.Add(its[i].Value);
if (((i + 1) < its.Count) && its[i + 1].Typ == PersonItemType.Name)
{
if ((i + 2) == its.Count)
{
i++;
}
else if (its[i + 2].Typ != PersonItemType.Surname)
{
i++;
}
}
else if (((i + 1) < its.Count) && its[i + 1].Typ == PersonItemType.Initial)
{
if ((i + 2) == its.Count)
{
i++;
}
else if (its[i + 2].Typ == PersonItemType.Initial && (i + 3) == its.Count)
{
i += 2;
}
}
else if (((i + 2) < its.Count) && its[i + 1].Typ == PersonItemType.And && its[i + 2].Typ == PersonItemType.Surname)
{
bool ok = true;
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(its[i + 2].BeginToken, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null && !npt.Morph.Case.IsGenitive && !npt.Morph.Case.IsUndefined)
{
ok = false;
}
if (ok)
{
eponims.Add(its[i + 2].Value);
i += 2;
}
}
t1 = its[i].EndToken;
}
else if (its[i].Typ == PersonItemType.Name && holy)
{
t1 = its[i].EndToken;
bool sec = false;
if (((i + 1) < its.Count) && its[i].Chars == its[i + 1].Chars && its[i + 1].Typ != PersonItemType.Initial)
{
sec = true;
t1 = its[i + 1].EndToken;
}
if (sec)
{
eponims.Add(string.Format("СВЯТ.{0} {1}", its[i].Value, its[i + 1].Value));
}
else
{
eponims.Add(string.Format("СВЯТ.{0}", its[i].Value));
}
}
else if (full && (i + 1) == its.Count && ((its[i].Typ == PersonItemType.Name || its[i].Typ == PersonItemType.Surname)))
{
t1 = its[i].EndToken;
eponims.Add(its[i].Value);
}
else if ((its[i].Typ == PersonItemType.Name && its.Count == 3 && its[i + 1].Typ == PersonItemType.Name) && its[i + 2].Typ == PersonItemType.Surname)
{
t1 = its[i + 2].EndToken;
eponims.Add(string.Format("{0} {1} {2}", its[i].Value, its[i + 1].Value, its[i + 2].Value));
i += 2;
}
if (eponims.Count == 0)
{
return(null);
}
return(new OrgItemEponymToken(t, t1)
{
Eponyms = eponims
});
}
public static OrgItemNameToken TryAttach(Pullenti.Ner.Token t, OrgItemNameToken prev, bool extOnto, bool first)
{
if (t == null)
{
return(null);
}
if (t.IsValue("ОРДЕНА", null) && t.Next != null)
{
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t.Next, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null)
{
Pullenti.Ner.Token t1 = npt.EndToken;
if (((t1.IsValue("ЗНАК", null) || t1.IsValue("ДРУЖБА", null))) && (t1.WhitespacesAfterCount < 2))
{
npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t1.Next, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null)
{
t1 = npt.EndToken;
}
}
return(new OrgItemNameToken(t, t1)
{
IsIgnoredPart = true
});
}
if (t.Next.GetMorphClassInDictionary().IsProperSurname)
{
return new OrgItemNameToken(t, t.Next)
{
IsIgnoredPart = true
}
}
;
Pullenti.Ner.ReferentToken ppp = t.Kit.ProcessReferent("PERSON", t.Next);
if (ppp != null)
{
return new OrgItemNameToken(t, ppp.EndToken)
{
IsIgnoredPart = true
}
}
;
if ((t.WhitespacesAfterCount < 2) && Pullenti.Ner.Core.BracketHelper.CanBeStartOfSequence(t.Next, true, false))
{
Pullenti.Ner.Core.BracketSequenceToken br = Pullenti.Ner.Core.BracketHelper.TryParse(t.Next, Pullenti.Ner.Core.BracketParseAttr.NearCloseBracket, 10);
if (br != null && (br.LengthChar < 40))
{
return new OrgItemNameToken(t, br.EndToken)
{
IsIgnoredPart = true
}
}
;
}
}
if (first && t.Chars.IsCyrillicLetter && t.Morph.Class.IsPreposition)
{
if (!t.IsValue("ПО", null) && !t.IsValue("ПРИ", null))
{
return(null);
}
}
OrgItemNameToken res = _TryAttach(t, prev, extOnto);
if (res == null)
{
if (extOnto)
{
if ((t.GetReferent() is Pullenti.Ner.Geo.GeoReferent) || (((t is Pullenti.Ner.TextToken) && !t.IsChar(';'))))
{
return new OrgItemNameToken(t, t)
{
Value = t.GetSourceText()
}
}
;
}
return(null);
}
if (prev == null && !extOnto)
{
if (t.Kit.Ontology != null)
{
Pullenti.Ner.Org.OrganizationAnalyzer.OrgAnalyzerData ad = t.Kit.Ontology._getAnalyzerData(Pullenti.Ner.Org.OrganizationAnalyzer.ANALYZER_NAME) as Pullenti.Ner.Org.OrganizationAnalyzer.OrgAnalyzerData;
if (ad != null)
{
Pullenti.Ner.Core.TerminToken tok = ad.OrgPureNames.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok != null && tok.EndChar > res.EndChar)
{
res.EndToken = tok.EndToken;
}
}
}
}
if (prev != null && !extOnto)
{
if ((prev.Chars.IsAllLower && !res.Chars.IsAllLower && !res.IsStdTail) && !res.IsStdName)
{
if (prev.Chars.IsLatinLetter && res.Chars.IsLatinLetter)
{
}
else if (m_StdNouns.TryParse(res.BeginToken, Pullenti.Ner.Core.TerminParseAttr.No) != null)
{
}
else
{
return(null);
}
}
}
if ((res.EndToken.Next != null && !res.EndToken.IsWhitespaceAfter && res.EndToken.Next.IsHiphen) && !res.EndToken.Next.IsWhitespaceAfter)
{
Pullenti.Ner.TextToken tt = res.EndToken.Next.Next as Pullenti.Ner.TextToken;
if (tt != null)
{
if (tt.Chars == res.Chars || tt.Chars.IsAllUpper)
{
res.EndToken = tt;
res.Value = string.Format("{0}-{1}", res.Value, tt.Term);
}
}
}
if ((res.EndToken.Next != null && res.EndToken.Next.IsAnd && res.EndToken.WhitespacesAfterCount == 1) && res.EndToken.Next.WhitespacesAfterCount == 1)
{
OrgItemNameToken res1 = _TryAttach(res.EndToken.Next.Next, prev, extOnto);
if (res1 != null && res1.Chars == res.Chars && OrgItemTypeToken.TryAttach(res.EndToken.Next.Next, false, null) == null)
{
if (!((res1.Morph.Case & res.Morph.Case)).IsUndefined)
{
res.EndToken = res1.EndToken;
res.Value = string.Format("{0} {1} {2}", res.Value, (res.Kit.BaseLanguage.IsUa ? "ТА" : "И"), res1.Value);
}
}
}
for (Pullenti.Ner.Token tt = res.BeginToken; tt != null && tt.EndChar <= res.EndChar; tt = tt.Next)
{
if (m_StdNouns.TryParse(tt, Pullenti.Ner.Core.TerminParseAttr.No) != null)
{
res.StdOrgNameNouns++;
}
}
if (m_StdNouns.TryParse(res.EndToken, Pullenti.Ner.Core.TerminParseAttr.No) != null)
{
int cou = 1;
bool non = false;
Pullenti.Ner.Token et = res.EndToken;
if (!_isNotTermNoun(res.EndToken))
{
non = true;
}
bool br = false;
for (Pullenti.Ner.Token tt = res.EndToken.Next; tt != null; tt = tt.Next)
{
if (tt.IsTableControlChar)
{
break;
}
if (tt.IsChar('('))
{
if (!non)
{
break;
}
br = true;
continue;
}
if (tt.IsChar(')'))
{
br = false;
et = tt;
break;
}
if (!(tt is Pullenti.Ner.TextToken))
{
break;
}
if (tt.WhitespacesBeforeCount > 1)
{
if (tt.NewlinesBeforeCount > 1)
{
break;
}
if (tt.Chars != res.EndToken.Chars)
{
break;
}
}
if (tt.Morph.Class.IsPreposition || tt.IsCommaAnd)
{
continue;
}
Pullenti.Morph.MorphClass dd = tt.GetMorphClassInDictionary();
if (!dd.IsNoun && !dd.IsAdjective)
{
break;
}
Pullenti.Ner.Core.NounPhraseToken npt2 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(tt, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt2 == null)
{
if (dd == Pullenti.Morph.MorphClass.Adjective)
{
continue;
}
break;
}
if (m_StdNouns.TryParse(npt2.EndToken, Pullenti.Ner.Core.TerminParseAttr.No) == null)
{
break;
}
if (npt2.EndToken.Chars != res.EndToken.Chars)
{
break;
}
if ((npt2.EndToken.IsValue("УПРАВЛЕНИЕ", null) || npt2.EndToken.IsValue("ИНСТИТУТ", null) || npt2.EndToken.IsValue("УПРАВЛІННЯ", null)) || npt2.EndToken.IsValue("ІНСТИТУТ", null) || tt.Previous.IsValue("ПРИ", null))
{
Pullenti.Ner.ReferentToken rt = tt.Kit.ProcessReferent(Pullenti.Ner.Org.OrganizationAnalyzer.ANALYZER_NAME, tt);
if (rt != null)
{
break;
}
}
cou++;
tt = npt2.EndToken;
if (!_isNotTermNoun(tt))
{
non = true;
et = tt;
}
}
if (non && !br)
{
res.StdOrgNameNouns += cou;
res.EndToken = et;
}
}
return(res);
}
public static Pullenti.Ner.ReferentToken TryAttachOrg(Pullenti.Ner.Token t, bool canBeCyr = false)
{
if (t == null)
{
return(null);
}
bool br = false;
if (t.IsChar('(') && t.Next != null)
{
t = t.Next;
br = true;
}
if (t is Pullenti.Ner.NumberToken)
{
if ((t as Pullenti.Ner.NumberToken).Typ == Pullenti.Ner.NumberSpellingType.Words && t.Morph.Class.IsAdjective && t.Chars.IsCapitalUpper)
{
}
else
{
return(null);
}
}
else
{
if (t.Chars.IsAllLower)
{
return(null);
}
if ((t.LengthChar < 3) && !t.Chars.IsLetter)
{
return(null);
}
if (!t.Chars.IsLatinLetter)
{
if (!canBeCyr || !t.Chars.IsCyrillicLetter)
{
return(null);
}
}
}
Pullenti.Ner.Token t0 = t;
Pullenti.Ner.Token t1 = t0;
int namWo = 0;
OrgItemEngItem tok = null;
Pullenti.Ner.Geo.GeoReferent geo = null;
OrgItemTypeToken addTyp = null;
for (; t != null; t = t.Next)
{
if (t != t0 && t.WhitespacesBeforeCount > 1)
{
break;
}
if (t.IsChar(')'))
{
break;
}
if (t.IsChar('(') && t.Next != null)
{
if ((t.Next.GetReferent() is Pullenti.Ner.Geo.GeoReferent) && t.Next.Next != null && t.Next.Next.IsChar(')'))
{
geo = t.Next.GetReferent() as Pullenti.Ner.Geo.GeoReferent;
t = t.Next.Next;
continue;
}
OrgItemTypeToken typ = OrgItemTypeToken.TryAttach(t.Next, true, null);
if ((typ != null && typ.EndToken.Next != null && typ.EndToken.Next.IsChar(')')) && typ.Chars.IsLatinLetter)
{
addTyp = typ;
t = typ.EndToken.Next;
continue;
}
if (((t.Next is Pullenti.Ner.TextToken) && t.Next.Next != null && t.Next.Next.IsChar(')')) && t.Next.Chars.IsCapitalUpper)
{
t1 = (t = t.Next.Next);
continue;
}
break;
}
tok = TryAttach(t, canBeCyr);
if (tok == null && t.IsCharOf(".,") && t.Next != null)
{
tok = TryAttach(t.Next, canBeCyr);
if (tok == null && t.Next.IsCharOf(",."))
{
tok = TryAttach(t.Next.Next, canBeCyr);
}
}
if (tok != null)
{
if (tok.LengthChar == 1 && t0.Chars.IsCyrillicLetter)
{
return(null);
}
break;
}
if (t.IsHiphen && !t.IsWhitespaceAfter && !t.IsWhitespaceBefore)
{
continue;
}
if (t.IsCharOf("&+") || t.IsAnd)
{
continue;
}
if (t.IsChar('.'))
{
if (t.Previous != null && t.Previous.LengthChar == 1)
{
continue;
}
else if (Pullenti.Ner.Core.MiscHelper.CanBeStartOfSentence(t.Next))
{
break;
}
}
if (!t.Chars.IsLatinLetter)
{
if (!canBeCyr || !t.Chars.IsCyrillicLetter)
{
break;
}
}
if (t.Chars.IsAllLower)
{
if (t.Morph.Class.IsPreposition || t.Morph.Class.IsConjunction)
{
continue;
}
if (br)
{
continue;
}
break;
}
Pullenti.Morph.MorphClass mc = t.GetMorphClassInDictionary();
if (mc.IsVerb)
{
if (t.Next != null && t.Next.Morph.Class.IsPreposition)
{
break;
}
}
if (t.Next != null && t.Next.IsValue("OF", null))
{
break;
}
if (t is Pullenti.Ner.TextToken)
{
namWo++;
}
t1 = t;
}
if (tok == null)
{
return(null);
}
if (t0 == tok.BeginToken)
{
Pullenti.Ner.Core.BracketSequenceToken br2 = Pullenti.Ner.Core.BracketHelper.TryParse(tok.EndToken.Next, Pullenti.Ner.Core.BracketParseAttr.No, 100);
if (br2 != null)
{
Pullenti.Ner.Org.OrganizationReferent org1 = new Pullenti.Ner.Org.OrganizationReferent();
if (tok.ShortValue != null)
{
org1.AddTypeStr(tok.ShortValue);
}
org1.AddTypeStr(tok.FullValue);
string nam1 = Pullenti.Ner.Core.MiscHelper.GetTextValue(br2.BeginToken, br2.EndToken, Pullenti.Ner.Core.GetTextAttr.No);
if (nam1 != null)
{
org1.AddName(nam1, true, null);
return(new Pullenti.Ner.ReferentToken(org1, t0, br2.EndToken));
}
}
return(null);
}
Pullenti.Ner.Org.OrganizationReferent org = new Pullenti.Ner.Org.OrganizationReferent();
Pullenti.Ner.Token te = tok.EndToken;
if (tok.IsBank)
{
t1 = tok.EndToken;
}
if (tok.FullValue == "company" && (tok.WhitespacesAfterCount < 3))
{
OrgItemEngItem tok1 = TryAttach(tok.EndToken.Next, canBeCyr);
if (tok1 != null)
{
t1 = tok.EndToken;
tok = tok1;
te = tok.EndToken;
}
}
if (tok.FullValue == "company")
{
if (namWo == 0)
{
return(null);
}
}
string nam = Pullenti.Ner.Core.MiscHelper.GetTextValue(t0, t1, Pullenti.Ner.Core.GetTextAttr.IgnoreArticles);
if (nam == "STOCK" && tok.FullValue == "company")
{
return(null);
}
string altNam = null;
if (string.IsNullOrEmpty(nam))
{
return(null);
}
if (nam.IndexOf('(') > 0)
{
int i1 = nam.IndexOf('(');
int i2 = nam.IndexOf(')');
if (i1 < i2)
{
altNam = nam;
string tai = null;
if ((i2 + 1) < nam.Length)
{
tai = nam.Substring(i2).Trim();
}
nam = nam.Substring(0, i1).Trim();
if (tai != null)
{
nam = string.Format("{0} {1}", nam, tai);
}
}
}
if (tok.IsBank)
{
org.AddTypeStr((tok.Kit.BaseLanguage.IsEn ? "bank" : "банк"));
org.AddProfile(Pullenti.Ner.Org.OrgProfile.Finance);
if ((t1.Next != null && t1.Next.IsValue("OF", null) && t1.Next.Next != null) && t1.Next.Next.Chars.IsLatinLetter)
{
OrgItemNameToken nam0 = OrgItemNameToken.TryAttach(t1.Next, null, false, false);
if (nam0 != null)
{
te = nam0.EndToken;
}
else
{
te = t1.Next.Next;
}
nam = Pullenti.Ner.Core.MiscHelper.GetTextValue(t0, te, Pullenti.Ner.Core.GetTextAttr.No);
if (te.GetReferent() is Pullenti.Ner.Geo.GeoReferent)
{
org.AddGeoObject(te.GetReferent() as Pullenti.Ner.Geo.GeoReferent);
}
}
else if (t0 == t1)
{
return(null);
}
}
else
{
if (tok.ShortValue != null)
{
org.AddTypeStr(tok.ShortValue);
}
org.AddTypeStr(tok.FullValue);
}
if (string.IsNullOrEmpty(nam))
{
return(null);
}
org.AddName(nam, true, null);
if (altNam != null)
{
org.AddName(altNam, true, null);
}
Pullenti.Ner.ReferentToken res = new Pullenti.Ner.ReferentToken(org, t0, te);
t = te;
while (t.Next != null)
{
if (t.Next.IsCharOf(",."))
{
t = t.Next;
}
else
{
break;
}
}
if (t.WhitespacesAfterCount < 2)
{
tok = TryAttach(t.Next, canBeCyr);
if (tok != null)
{
if (tok.ShortValue != null)
{
org.AddTypeStr(tok.ShortValue);
}
org.AddTypeStr(tok.FullValue);
res.EndToken = tok.EndToken;
}
}
if (geo != null)
{
org.AddGeoObject(geo);
}
if (addTyp != null)
{
org.AddType(addTyp, false);
}
if (!br)
{
return(res);
}
t = res.EndToken;
if (t.Next == null || t.Next.IsChar(')'))
{
res.EndToken = t.Next;
}
else
{
return(null);
}
return(res);
}
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);
}