static Pullenti.Ner.ReferentToken _tryNounName(List <CityItemToken> li, out Pullenti.Ner.Core.IntOntologyItem oi, bool always)
{
oi = null;
if (li == null || (li.Count < 2) || ((li[0].Typ != CityItemToken.ItemType.Noun && li[0].Typ != CityItemToken.ItemType.Misc)))
{
return(null);
}
bool ok = !li[0].Doubtful;
if (ok && li[0].Typ == CityItemToken.ItemType.Misc)
{
ok = false;
}
string typ = (li[0].Typ == CityItemToken.ItemType.Misc ? null : li[0].Value);
string typ2 = (li[0].Typ == CityItemToken.ItemType.Misc ? null : li[0].AltValue);
string probAdj = null;
int i1 = 1;
Pullenti.Ner.Referent org = null;
if ((typ != null && li[i1].Typ == CityItemToken.ItemType.Noun && ((i1 + 1) < li.Count)) && li[0].WhitespacesAfterCount <= 1 && (((Pullenti.Morph.LanguageHelper.EndsWith(typ, "ПОСЕЛОК") || Pullenti.Morph.LanguageHelper.EndsWith(typ, "СЕЛИЩЕ") || typ == "ДЕРЕВНЯ") || typ == "СЕЛО")))
{
if (li[i1].BeginToken == li[i1].EndToken)
{
Pullenti.Ner.Address.Internal.AddressItemToken ooo = Pullenti.Ner.Address.Internal.AddressItemToken.TryAttachOrg(li[i1].BeginToken);
if (ooo != null && ooo.RefToken != null)
{
return(null);
}
}
typ2 = li[i1].Value;
if (typ2 == "СТАНЦИЯ" && li[i1].BeginToken.IsValue("СТ", null) && ((i1 + 1) < li.Count))
{
Pullenti.Ner.MorphCollection m = li[i1 + 1].Morph;
if (m.Number == Pullenti.Morph.MorphNumber.Plural)
{
probAdj = "СТАРЫЕ";
}
else if (m.Gender == Pullenti.Morph.MorphGender.Feminie)
{
probAdj = "СТАРАЯ";
}
else if (m.Gender == Pullenti.Morph.MorphGender.Masculine)
{
probAdj = "СТАРЫЙ";
}
else
{
probAdj = "СТАРОЕ";
}
}
i1++;
}
string name = li[i1].Value ?? ((li[i1].OntoItem == null ? null : li[i1].OntoItem.CanonicText));
string altName = li[i1].AltValue;
if (name == null)
{
return(null);
}
Pullenti.Ner.MorphCollection mc = li[0].Morph;
if (i1 == 1 && li[i1].Typ == CityItemToken.ItemType.City && ((li[0].Value == "ГОРОД" || li[0].Value == "МІСТО" || li[0].Typ == CityItemToken.ItemType.Misc)))
{
if (typ == null && ((i1 + 1) < li.Count) && li[i1 + 1].Typ == CityItemToken.ItemType.Noun)
{
return(null);
}
oi = li[i1].OntoItem;
if (oi != null)
{
name = oi.CanonicText;
}
if (name.Length > 2 || oi.MiscAttr != null)
{
if (!li[1].Doubtful || ((oi != null && oi.MiscAttr != null)))
{
ok = true;
}
else if (!ok && !li[1].IsNewlineBefore)
{
if (li[0].GeoObjectBefore || li[1].GeoObjectAfter)
{
ok = true;
}
else if (Pullenti.Ner.Address.Internal.StreetDefineHelper.CheckStreetAfter(li[1].EndToken.Next))
{
ok = true;
}
else if (li[1].EndToken.Next != null && (li[1].EndToken.Next.GetReferent() is Pullenti.Ner.Date.DateReferent))
{
ok = true;
}
else if ((li[1].WhitespacesBeforeCount < 2) && li[1].OntoItem != null)
{
if (li[1].IsNewlineAfter)
{
ok = true;
}
else
{
ok = true;
}
}
}
if (li[1].Doubtful && li[1].EndToken.Next != null && li[1].EndToken.Chars == li[1].EndToken.Next.Chars)
{
ok = false;
}
if (li[0].BeginToken.Previous != null && li[0].BeginToken.Previous.IsValue("В", null))
{
ok = true;
}
}
if (!ok)
{
ok = CheckYearAfter(li[1].EndToken.Next);
}
if (!ok)
{
ok = CheckCityAfter(li[1].EndToken.Next);
}
}
else if ((li[i1].Typ == CityItemToken.ItemType.ProperName || li[i1].Typ == CityItemToken.ItemType.City))
{
if (((li[0].Value == "АДМИНИСТРАЦИЯ" || li[0].Value == "АДМІНІСТРАЦІЯ")) && i1 == 1)
{
return(null);
}
if (li[i1].IsNewlineBefore)
{
if (li.Count != 2)
{
return(null);
}
}
if (!li[0].Doubtful)
{
ok = true;
if (name.Length < 2)
{
ok = false;
}
else if ((name.Length < 3) && li[0].Morph.Number != Pullenti.Morph.MorphNumber.Singular)
{
ok = false;
}
if (li[i1].Doubtful && !li[i1].GeoObjectAfter && !li[0].GeoObjectBefore)
{
if (li[i1].Morph.Case.IsGenitive)
{
if (li[i1].EndToken.Next == null || MiscLocationHelper.CheckGeoObjectAfter(li[i1].EndToken.Next, false) || Pullenti.Ner.Address.Internal.AddressItemToken.CheckHouseAfter(li[i1].EndToken.Next, false, true))
{
}
else if (li[0].BeginToken.Previous == null || MiscLocationHelper.CheckGeoObjectBefore(li[0].BeginToken))
{
}
else
{
ok = false;
}
}
if (ok)
{
Pullenti.Ner.ReferentToken rt0 = li[i1].Kit.ProcessReferent("PERSONPROPERTY", li[0].BeginToken.Previous);
if (rt0 != null)
{
Pullenti.Ner.ReferentToken rt1 = li[i1].Kit.ProcessReferent("PERSON", li[i1].BeginToken);
if (rt1 != null)
{
ok = false;
}
}
}
}
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(li[i1].BeginToken, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null)
{
if (npt.EndToken.EndChar > li[i1].EndChar && npt.Adjectives.Count > 0 && !npt.Adjectives[0].EndToken.Next.IsComma)
{
ok = false;
}
else if (TerrItemToken.m_UnknownRegions.TryParse(npt.EndToken, Pullenti.Ner.Core.TerminParseAttr.FullwordsOnly) != null)
{
bool ok1 = false;
if (li[0].BeginToken.Previous != null)
{
Pullenti.Ner.Token ttt = li[0].BeginToken.Previous;
if (ttt.IsComma && ttt.Previous != null)
{
ttt = ttt.Previous;
}
Pullenti.Ner.Geo.GeoReferent geo = ttt.GetReferent() as Pullenti.Ner.Geo.GeoReferent;
if (geo != null && !geo.IsCity)
{
ok1 = true;
}
}
if (npt.EndToken.Next != null)
{
Pullenti.Ner.Token ttt = npt.EndToken.Next;
if (ttt.IsComma && ttt.Next != null)
{
ttt = ttt.Next;
}
Pullenti.Ner.Geo.GeoReferent geo = ttt.GetReferent() as Pullenti.Ner.Geo.GeoReferent;
if (geo != null && !geo.IsCity)
{
ok1 = true;
}
}
if (!ok1)
{
return(null);
}
}
}
if (li[0].Value == "ПОРТ")
{
if (li[i1].Chars.IsAllUpper || li[i1].Chars.IsLatinLetter)
{
return(null);
}
}
}
else if (li[0].GeoObjectBefore)
{
ok = true;
}
else if (li[i1].GeoObjectAfter && !li[i1].IsNewlineAfter)
{
ok = true;
}
else
{
ok = CheckYearAfter(li[i1].EndToken.Next);
}
if (!ok)
{
ok = CheckStreetAfter(li[i1].EndToken.Next);
}
if (!ok && li[0].BeginToken.Previous != null && li[0].BeginToken.Previous.IsValue("В", null))
{
ok = true;
}
}
else
{
return(null);
}
if (!ok && !always)
{
if (MiscLocationHelper.CheckNearBefore(li[0].BeginToken.Previous) == null)
{
return(null);
}
}
if (li.Count > (i1 + 1))
{
li.RemoveRange(i1 + 1, li.Count - i1 - 1);
}
Pullenti.Ner.Geo.GeoReferent city = new Pullenti.Ner.Geo.GeoReferent();
if (oi != null && oi.Referent != null)
{
city = oi.Referent.Clone() as Pullenti.Ner.Geo.GeoReferent;
city.Occurrence.Clear();
}
if (!li[0].Morph.Case.IsUndefined && li[0].Morph.Gender != Pullenti.Morph.MorphGender.Undefined)
{
if (li[i1].EndToken.Morph.Class.IsAdjective && li[i1].BeginToken == li[i1].EndToken)
{
string nam = Pullenti.Ner.Core.ProperNameHelper.GetNameEx(li[i1].BeginToken, li[i1].EndToken, Pullenti.Morph.MorphClass.Adjective, li[0].Morph.Case, li[0].Morph.Gender, false, false);
if (nam != null && nam != name)
{
name = nam;
}
}
}
if (li[0].Morph.Case.IsNominative)
{
if (altName != null)
{
city.AddName(altName);
}
altName = null;
}
city.AddName(name);
if (probAdj != null)
{
city.AddName(probAdj + " " + name);
}
if (altName != null)
{
city.AddName(altName);
if (probAdj != null)
{
city.AddName(probAdj + " " + altName);
}
}
if (typ != null)
{
city.AddTyp(typ);
}
else if (!city.IsCity)
{
city.AddTypCity(li[0].Kit.BaseLanguage);
}
if (typ2 != null)
{
city.AddTyp(typ2.ToLower());
}
if (li[0].HigherGeo != null && GeoOwnerHelper.CanBeHigher(li[0].HigherGeo, city))
{
city.Higher = li[0].HigherGeo;
}
if (li[0].Typ == CityItemToken.ItemType.Misc)
{
li.RemoveAt(0);
}
Pullenti.Ner.ReferentToken res = new Pullenti.Ner.ReferentToken(city, li[0].BeginToken, li[li.Count - 1].EndToken)
{
Morph = mc
};
if (res.EndToken.Next != null && res.EndToken.Next.IsHiphen && (res.EndToken.Next.Next is Pullenti.Ner.NumberToken))
{
Pullenti.Ner.NumberToken num = res.EndToken.Next.Next as Pullenti.Ner.NumberToken;
if ((num.Typ == Pullenti.Ner.NumberSpellingType.Digit && !num.Morph.Class.IsAdjective && num.IntValue != null) && (num.IntValue.Value < 50))
{
foreach (Pullenti.Ner.Slot s in city.Slots)
{
if (s.TypeName == Pullenti.Ner.Geo.GeoReferent.ATTR_NAME)
{
city.UploadSlot(s, string.Format("{0}-{1}", s.Value, num.Value));
}
}
res.EndToken = num;
}
}
if (li[0].BeginToken == li[0].EndToken && li[0].BeginToken.IsValue("ГОРОДОК", null))
{
if (Pullenti.Ner.Address.Internal.AddressItemToken.CheckHouseAfter(res.EndToken.Next, true, false))
{
return(null);
}
}
return(res);
}
static Pullenti.Ner.ReferentToken Try1(List <CityItemToken> li, out Pullenti.Ner.Core.IntOntologyItem oi, Pullenti.Ner.Core.AnalyzerDataWithOntology ad)
{
oi = null;
if (li == null || (li.Count < 1))
{
return(null);
}
else if (li[0].Typ != CityItemToken.ItemType.City)
{
if (li.Count != 2 || li[0].Typ != CityItemToken.ItemType.ProperName || li[1].Typ != CityItemToken.ItemType.Noun)
{
return(null);
}
}
int i = 1;
oi = li[0].OntoItem;
bool ok = !li[0].Doubtful;
if ((ok && li[0].OntoItem != null && li[0].OntoItem.MiscAttr == null) && ad != null)
{
if (li[0].OntoItem.Owner != ad.LocalOntology && !li[0].OntoItem.Owner.IsExtOntology)
{
if (li[0].BeginToken.Previous != null && li[0].BeginToken.Previous.IsValue("В", null))
{
}
else
{
ok = false;
}
}
}
if (li.Count == 1 && li[0].BeginToken.Morph.Class.IsAdjective)
{
List <Pullenti.Ner.Address.Internal.StreetItemToken> sits = Pullenti.Ner.Address.Internal.StreetItemToken.TryParseList(li[0].BeginToken, null, 3);
if (sits != null && sits.Count == 2 && sits[1].Typ == Pullenti.Ner.Address.Internal.StreetItemType.Noun)
{
return(null);
}
}
string typ = null;
string alttyp = null;
Pullenti.Ner.MorphCollection mc = li[0].Morph;
if (i < li.Count)
{
if (li[i].Typ == CityItemToken.ItemType.Noun)
{
Pullenti.Ner.Address.Internal.AddressItemToken at = null;
if (!li[i].Chars.IsAllLower && (li[i].WhitespacesAfterCount < 2))
{
Pullenti.Ner.Address.Internal.StreetItemToken sit = Pullenti.Ner.Address.Internal.StreetItemToken.TryParse(li[i].EndToken.Next, null, false, null, false);
if (sit != null && sit.Typ == Pullenti.Ner.Address.Internal.StreetItemType.Noun)
{
at = Pullenti.Ner.Address.Internal.AddressItemToken.TryParse(li[i].BeginToken, null, false, false, null);
if (at != null)
{
Pullenti.Ner.Address.Internal.AddressItemToken at2 = Pullenti.Ner.Address.Internal.AddressItemToken.TryParse(li[i].EndToken.Next, null, false, false, null);
if (at2 != null && at2.Typ == Pullenti.Ner.Address.Internal.AddressItemToken.ItemType.Street)
{
at = null;
}
}
}
}
if (at == null)
{
typ = li[i].Value;
alttyp = li[i].AltValue;
if (li[i].BeginToken.IsValue("СТ", null) && li[i].BeginToken.Chars.IsAllUpper)
{
return(null);
}
if ((i + 1) == li.Count)
{
ok = true;
if (!li[i].Morph.Case.IsUndefined)
{
mc = li[i].Morph;
}
i++;
}
else if (ok)
{
i++;
}
else
{
Pullenti.Ner.Token tt0 = li[0].BeginToken.Previous;
if ((tt0 is Pullenti.Ner.TextToken) && (tt0.WhitespacesAfterCount < 3))
{
if (tt0.IsValue("МЭР", "МЕР") || tt0.IsValue("ГЛАВА", null) || tt0.IsValue("ГРАДОНАЧАЛЬНИК", null))
{
ok = true;
i++;
}
}
}
}
}
}
if (!ok && oi != null && (oi.CanonicText.Length < 4))
{
return(null);
}
if (!ok && li[0].BeginToken.Morph.Class.IsProperName)
{
return(null);
}
if (!ok)
{
if (!Pullenti.Ner.Core.MiscHelper.IsExistsInDictionary(li[0].BeginToken, li[0].EndToken, Pullenti.Morph.MorphClass.Adjective | Pullenti.Morph.MorphClass.Noun | Pullenti.Morph.MorphClass.Pronoun))
{
ok = li[0].GeoObjectBefore || li[i - 1].GeoObjectAfter;
if (ok && li[0].BeginToken == li[0].EndToken)
{
Pullenti.Morph.MorphClass mcc = li[0].BeginToken.GetMorphClassInDictionary();
if (mcc.IsProperName || mcc.IsProperSurname)
{
ok = false;
}
else if (li[0].GeoObjectBefore && (li[0].WhitespacesAfterCount < 2))
{
Pullenti.Ner.Address.Internal.AddressItemToken ad1 = Pullenti.Ner.Address.Internal.AddressItemToken.TryParse(li[0].BeginToken, null, false, false, null);
if (ad1 != null && ad1.Typ == Pullenti.Ner.Address.Internal.AddressItemToken.ItemType.Street)
{
Pullenti.Ner.Address.Internal.AddressItemToken ad2 = Pullenti.Ner.Address.Internal.AddressItemToken.TryParse(li[0].EndToken.Next, null, false, false, null);
if (ad2 == null || ad2.Typ != Pullenti.Ner.Address.Internal.AddressItemToken.ItemType.Street)
{
ok = false;
}
}
else if (Pullenti.Ner.Address.Internal.AddressItemToken.TryAttachOrg(li[0].BeginToken) != null)
{
ok = false;
}
}
}
}
if (ok)
{
if (li[0].Kit.ProcessReferent("PERSON", li[0].BeginToken) != null)
{
ok = false;
}
}
}
if (!ok)
{
ok = CheckYearAfter(li[0].EndToken.Next);
}
if (!ok && ((!li[0].BeginToken.Morph.Class.IsAdjective || li[0].BeginToken != li[0].EndToken)))
{
ok = CheckCityAfter(li[0].EndToken.Next);
}
if (!ok)
{
return(null);
}
if (i < li.Count)
{
li.RemoveRange(i, li.Count - i);
}
Pullenti.Ner.ReferentToken rt = null;
if (oi == null)
{
if (li[0].Value != null && li[0].HigherGeo != null)
{
Pullenti.Ner.Geo.GeoReferent cap = new Pullenti.Ner.Geo.GeoReferent();
cap.AddName(li[0].Value);
cap.AddTypCity(li[0].Kit.BaseLanguage);
cap.Higher = li[0].HigherGeo;
if (typ != null)
{
cap.AddTyp(typ);
}
if (alttyp != null)
{
cap.AddTyp(alttyp);
}
rt = new Pullenti.Ner.ReferentToken(cap, li[0].BeginToken, li[0].EndToken);
}
else
{
if (li[0].Value == null)
{
return(null);
}
if (typ == null)
{
if ((li.Count == 1 && li[0].BeginToken.Previous != null && li[0].BeginToken.Previous.IsHiphen) && (li[0].BeginToken.Previous.Previous is Pullenti.Ner.ReferentToken) && (li[0].BeginToken.Previous.Previous.GetReferent() is Pullenti.Ner.Geo.GeoReferent))
{
}
else
{
return(null);
}
}
else
{
if (!Pullenti.Morph.LanguageHelper.EndsWithEx(typ, "ПУНКТ", "ПОСЕЛЕНИЕ", "ПОСЕЛЕННЯ", "ПОСЕЛОК"))
{
if (!Pullenti.Morph.LanguageHelper.EndsWith(typ, "CITY"))
{
if (typ == "СТАНЦИЯ" && (MiscLocationHelper.CheckGeoObjectBefore(li[0].BeginToken)))
{
}
else if (li.Count > 1 && li[1].Typ == CityItemToken.ItemType.Noun && li[0].Typ == CityItemToken.ItemType.City)
{
}
else if ((li.Count == 2 && li[1].Typ == CityItemToken.ItemType.Noun && li[0].Typ == CityItemToken.ItemType.ProperName) && ((li[0].GeoObjectBefore || li[1].GeoObjectAfter)))
{
}
else
{
return(null);
}
}
}
if (li[0].BeginToken.Morph.Class.IsAdjective)
{
li[0].Value = Pullenti.Ner.Core.ProperNameHelper.GetNameEx(li[0].BeginToken, li[0].EndToken, Pullenti.Morph.MorphClass.Adjective, li[1].Morph.Case, li[1].Morph.Gender, false, false);
}
}
}
}
else if (oi.Referent is Pullenti.Ner.Geo.GeoReferent)
{
Pullenti.Ner.Geo.GeoReferent city = oi.Referent.Clone() as Pullenti.Ner.Geo.GeoReferent;
city.Occurrence.Clear();
rt = new Pullenti.Ner.ReferentToken(city, li[0].BeginToken, li[li.Count - 1].EndToken)
{
Morph = mc
};
}
else if (typ == null)
{
typ = oi.Typ;
}
if (rt == null)
{
Pullenti.Ner.Geo.GeoReferent city = new Pullenti.Ner.Geo.GeoReferent();
city.AddName((oi == null ? li[0].Value : oi.CanonicText));
if (typ != null)
{
city.AddTyp(typ);
}
else
{
city.AddTypCity(li[0].Kit.BaseLanguage);
}
if (alttyp != null)
{
city.AddTyp(alttyp);
}
rt = new Pullenti.Ner.ReferentToken(city, li[0].BeginToken, li[li.Count - 1].EndToken)
{
Morph = mc
};
}
if ((rt.Referent is Pullenti.Ner.Geo.GeoReferent) && li.Count == 1 && (rt.Referent as Pullenti.Ner.Geo.GeoReferent).IsCity)
{
if (rt.BeginToken.Previous != null && rt.BeginToken.Previous.IsValue("Г", null))
{
rt.BeginToken = rt.BeginToken.Previous;
}
else if ((rt.BeginToken.Previous != null && rt.BeginToken.Previous.IsChar('.') && rt.BeginToken.Previous.Previous != null) && rt.BeginToken.Previous.Previous.IsValue("Г", null))
{
rt.BeginToken = rt.BeginToken.Previous.Previous;
}
else if (rt.EndToken.Next != null && (rt.WhitespacesAfterCount < 2) && rt.EndToken.Next.IsValue("Г", null))
{
rt.EndToken = rt.EndToken.Next;
if (rt.EndToken.Next != null && rt.EndToken.Next.IsChar('.'))
{
rt.EndToken = rt.EndToken.Next;
}
}
}
return(rt);
}
public static Pullenti.Ner.ReferentToken TryAttachTerritory(List <TerrItemToken> li, Pullenti.Ner.Core.AnalyzerData ad, bool attachAlways = false, List <CityItemToken> cits = null, List <Pullenti.Ner.Geo.GeoReferent> exists = null)
{
if (li == null || li.Count == 0)
{
return(null);
}
TerrItemToken exObj = null;
TerrItemToken newName = null;
List <TerrItemToken> adjList = new List <TerrItemToken>();
TerrItemToken noun = null;
TerrItemToken addNoun = null;
Pullenti.Ner.ReferentToken rt = _tryAttachMoscowAO(li, ad);
if (rt != null)
{
return(rt);
}
if (li[0].TerminItem != null && li[0].TerminItem.CanonicText == "ТЕРРИТОРИЯ")
{
Pullenti.Ner.ReferentToken res2 = _tryAttachPureTerr(li, ad);
return(res2);
}
if (li.Count == 2)
{
if (li[0].Rzd != null && li[1].RzdDir != null)
{
Pullenti.Ner.Geo.GeoReferent rzd = new Pullenti.Ner.Geo.GeoReferent();
rzd.AddName(li[1].RzdDir);
rzd.AddTypTer(li[0].Kit.BaseLanguage);
rzd.AddSlot(Pullenti.Ner.Geo.GeoReferent.ATTR_REF, li[0].Rzd.Referent, false, 0);
rzd.AddExtReferent(li[0].Rzd);
return(new Pullenti.Ner.ReferentToken(rzd, li[0].BeginToken, li[1].EndToken));
}
if (li[1].Rzd != null && li[0].RzdDir != null)
{
Pullenti.Ner.Geo.GeoReferent rzd = new Pullenti.Ner.Geo.GeoReferent();
rzd.AddName(li[0].RzdDir);
rzd.AddTypTer(li[0].Kit.BaseLanguage);
rzd.AddSlot(Pullenti.Ner.Geo.GeoReferent.ATTR_REF, li[1].Rzd.Referent, false, 0);
rzd.AddExtReferent(li[1].Rzd);
return(new Pullenti.Ner.ReferentToken(rzd, li[0].BeginToken, li[1].EndToken));
}
}
bool canBeCityBefore = false;
bool adjTerrBefore = false;
if (cits != null)
{
if (cits[0].Typ == CityItemToken.ItemType.City)
{
canBeCityBefore = true;
}
else if (cits[0].Typ == CityItemToken.ItemType.Noun && cits.Count > 1)
{
canBeCityBefore = true;
}
}
int k;
for (k = 0; k < li.Count; k++)
{
if (li[k].OntoItem != null)
{
if (exObj != null || newName != null)
{
break;
}
if (noun != null)
{
if (k == 1)
{
if (noun.TerminItem.CanonicText == "РАЙОН" || noun.TerminItem.CanonicText == "ОБЛАСТЬ" || noun.TerminItem.CanonicText == "СОЮЗ")
{
if (li[k].OntoItem.Referent is Pullenti.Ner.Geo.GeoReferent)
{
if ((li[k].OntoItem.Referent as Pullenti.Ner.Geo.GeoReferent).IsState)
{
break;
}
}
bool ok = false;
Pullenti.Ner.Token tt = li[k].EndToken.Next;
if (tt == null)
{
ok = true;
}
else if (tt.IsCharOf(",."))
{
ok = true;
}
if (!ok)
{
ok = MiscLocationHelper.CheckGeoObjectBefore(li[0].BeginToken);
}
if (!ok)
{
Pullenti.Ner.Address.Internal.AddressItemToken adr = Pullenti.Ner.Address.Internal.AddressItemToken.TryParse(tt, null, false, false, null);
if (adr != null)
{
if (adr.Typ == Pullenti.Ner.Address.Internal.AddressItemToken.ItemType.Street)
{
ok = true;
}
}
}
if (!ok)
{
break;
}
}
if (li[k].OntoItem != null)
{
if (noun.BeginToken.IsValue("МО", null) || noun.BeginToken.IsValue("ЛО", null))
{
return(null);
}
}
}
}
exObj = li[k];
}
else if (li[k].TerminItem != null)
{
if (noun != null)
{
break;
}
if (li[k].TerminItem.IsAlwaysPrefix && k > 0)
{
break;
}
if (k > 0 && li[k].IsDoubt)
{
if (li[k].BeginToken == li[k].EndToken && li[k].BeginToken.IsValue("ЗАО", null))
{
break;
}
}
if (li[k].TerminItem.IsAdjective || li[k].IsGeoInDictionary)
{
adjList.Add(li[k]);
}
else
{
if (exObj != null)
{
Pullenti.Ner.Geo.GeoReferent geo = exObj.OntoItem.Referent as Pullenti.Ner.Geo.GeoReferent;
if (geo == null)
{
break;
}
if (exObj.IsAdjective && ((li[k].TerminItem.CanonicText == "СОЮЗ" || li[k].TerminItem.CanonicText == "ФЕДЕРАЦИЯ")))
{
string str = exObj.OntoItem.ToString();
if (!str.Contains(li[k].TerminItem.CanonicText))
{
return(null);
}
}
if (li[k].TerminItem.CanonicText == "РАЙОН" || li[k].TerminItem.CanonicText == "ОКРУГ" || li[k].TerminItem.CanonicText == "КРАЙ")
{
StringBuilder tmp = new StringBuilder();
foreach (Pullenti.Ner.Slot s in geo.Slots)
{
if (s.TypeName == Pullenti.Ner.Geo.GeoReferent.ATTR_TYPE)
{
tmp.AppendFormat("{0};", s.Value);
}
}
if (!tmp.ToString().ToUpper().Contains(li[k].TerminItem.CanonicText))
{
if (k != 1 || newName != null)
{
break;
}
newName = li[0];
newName.IsAdjective = true;
newName.OntoItem = null;
exObj = null;
}
}
}
noun = li[k];
if (k == 0)
{
TerrItemToken tt = TerrItemToken.TryParse(li[k].BeginToken.Previous, null, true, false, null);
if (tt != null && tt.Morph.Class.IsAdjective)
{
adjTerrBefore = true;
}
}
}
}
else
{
if (exObj != null)
{
break;
}
if (newName != null)
{
break;
}
newName = li[k];
}
}
string name = null;
string altName = null;
string fullName = null;
Pullenti.Ner.MorphCollection morph = null;
if (exObj != null)
{
if (exObj.IsAdjective && !exObj.Morph.Language.IsEn && noun == null)
{
if (attachAlways && exObj.EndToken.Next != null)
{
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(exObj.BeginToken, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (exObj.EndToken.Next.IsCommaAnd)
{
}
else if (npt == null)
{
}
else
{
Pullenti.Ner.Address.Internal.StreetItemToken str = Pullenti.Ner.Address.Internal.StreetItemToken.TryParse(exObj.EndToken.Next, null, false, null, false);
if (str != null)
{
if (str.Typ == Pullenti.Ner.Address.Internal.StreetItemType.Noun && str.EndToken == npt.EndToken)
{
return(null);
}
}
}
}
else
{
CityItemToken cit = CityItemToken.TryParse(exObj.EndToken.Next, null, false, null);
if (cit != null && ((cit.Typ == CityItemToken.ItemType.Noun || cit.Typ == CityItemToken.ItemType.City)))
{
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(exObj.BeginToken, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null && npt.EndToken == cit.EndToken)
{
}
else
{
return(null);
}
}
else if (exObj.BeginToken.IsValue("ПОДНЕБЕСНЫЙ", null))
{
}
else
{
return(null);
}
}
}
if (noun == null && exObj.CanBeCity)
{
CityItemToken cit0 = CityItemToken.TryParseBack(exObj.BeginToken.Previous);
if (cit0 != null && cit0.Typ != CityItemToken.ItemType.ProperName)
{
return(null);
}
}
if (exObj.IsDoubt && noun == null)
{
bool ok2 = false;
if (_canBeGeoAfter(exObj.EndToken.Next))
{
ok2 = true;
}
else if (!exObj.CanBeSurname && !exObj.CanBeCity)
{
if ((exObj.EndToken.Next != null && exObj.EndToken.Next.IsChar(')') && exObj.BeginToken.Previous != null) && exObj.BeginToken.Previous.IsChar('('))
{
ok2 = true;
}
else if (exObj.Chars.IsLatinLetter && exObj.BeginToken.Previous != null)
{
if (exObj.BeginToken.Previous.IsValue("IN", null))
{
ok2 = true;
}
else if (exObj.BeginToken.Previous.IsValue("THE", null) && exObj.BeginToken.Previous.Previous != null && exObj.BeginToken.Previous.Previous.IsValue("IN", null))
{
ok2 = true;
}
}
}
if (!ok2)
{
CityItemToken cit0 = CityItemToken.TryParseBack(exObj.BeginToken.Previous);
if (cit0 != null && cit0.Typ != CityItemToken.ItemType.ProperName)
{
}
else if (MiscLocationHelper.CheckGeoObjectBefore(exObj.BeginToken.Previous))
{
}
else
{
return(null);
}
}
}
name = exObj.OntoItem.CanonicText;
morph = exObj.Morph;
}
else if (newName != null)
{
if (noun == null)
{
return(null);
}
for (int j = 1; j < k; j++)
{
if (li[j].IsNewlineBefore && !li[0].IsNewlineBefore)
{
if (Pullenti.Ner.Core.BracketHelper.CanBeStartOfSequence(li[j].BeginToken, false, false))
{
}
else
{
return(null);
}
}
}
morph = noun.Morph;
if (newName.IsAdjective)
{
if (noun.TerminItem.Acronym == "АО")
{
if (noun.BeginToken != noun.EndToken)
{
return(null);
}
if (newName.Morph.Gender != Pullenti.Morph.MorphGender.Feminie)
{
return(null);
}
}
Pullenti.Ner.Geo.GeoReferent geoBefore = null;
Pullenti.Ner.Token tt0 = li[0].BeginToken.Previous;
if (tt0 != null && tt0.IsCommaAnd)
{
tt0 = tt0.Previous;
}
if (!li[0].IsNewlineBefore && tt0 != null)
{
geoBefore = tt0.GetReferent() as Pullenti.Ner.Geo.GeoReferent;
}
if (li.IndexOf(noun) < li.IndexOf(newName))
{
if (noun.TerminItem.IsState)
{
return(null);
}
if (newName.CanBeSurname && geoBefore == null)
{
if (((noun.Morph.Case & newName.Morph.Case)).IsUndefined)
{
return(null);
}
}
if (Pullenti.Ner.Core.MiscHelper.IsExistsInDictionary(newName.BeginToken, newName.EndToken, Pullenti.Morph.MorphClass.Adjective | Pullenti.Morph.MorphClass.Pronoun | Pullenti.Morph.MorphClass.Verb))
{
if (noun.BeginToken != newName.BeginToken)
{
if (geoBefore == null)
{
if (li.Count == 2 && _canBeGeoAfter(li[1].EndToken.Next))
{
}
else if (li.Count == 3 && li[2].TerminItem != null && _canBeGeoAfter(li[2].EndToken.Next))
{
}
else if (newName.IsGeoInDictionary)
{
}
else if (newName.EndToken.IsNewlineAfter)
{
}
else
{
return(null);
}
}
}
}
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(newName.EndToken, Pullenti.Ner.Core.NounPhraseParseAttr.ParsePronouns, 0, null);
if (npt != null && npt.EndToken != newName.EndToken)
{
if (li.Count >= 3 && li[2].TerminItem != null && npt.EndToken == li[2].EndToken)
{
addNoun = li[2];
}
else
{
return(null);
}
}
Pullenti.Ner.ReferentToken rtp = newName.Kit.ProcessReferent("PERSON", newName.BeginToken);
if (rtp != null)
{
return(null);
}
name = Pullenti.Ner.Core.ProperNameHelper.GetNameEx(newName.BeginToken, newName.EndToken, Pullenti.Morph.MorphClass.Adjective, Pullenti.Morph.MorphCase.Undefined, noun.TerminItem.Gender, false, false);
}
else
{
bool ok = false;
if (((k + 1) < li.Count) && li[k].TerminItem == null && li[k + 1].TerminItem != null)
{
ok = true;
}
else if ((k < li.Count) && li[k].OntoItem != null)
{
ok = true;
}
else if (k == li.Count && !newName.IsAdjInDictionary)
{
ok = true;
}
else if (MiscLocationHelper.CheckGeoObjectBefore(li[0].BeginToken) || canBeCityBefore)
{
ok = true;
}
else if (MiscLocationHelper.CheckGeoObjectAfter(li[k - 1].EndToken, false))
{
ok = true;
}
else if (li.Count == 3 && k == 2)
{
CityItemToken cit = CityItemToken.TryParse(li[2].BeginToken, null, false, null);
if (cit != null)
{
if (cit.Typ == CityItemToken.ItemType.City || cit.Typ == CityItemToken.ItemType.Noun)
{
ok = true;
}
}
}
else if (li.Count == 2)
{
ok = _canBeGeoAfter(li[li.Count - 1].EndToken.Next);
}
if (!ok && !li[0].IsNewlineBefore && !li[0].Chars.IsAllLower)
{
Pullenti.Ner.ReferentToken rt00 = li[0].Kit.ProcessReferent("PERSONPROPERTY", li[0].BeginToken.Previous);
if (rt00 != null)
{
ok = true;
}
}
if (noun.TerminItem != null && noun.TerminItem.IsStrong && newName.IsAdjective)
{
ok = true;
}
if (noun.IsDoubt && adjList.Count == 0 && geoBefore == null)
{
return(null);
}
name = Pullenti.Ner.Core.ProperNameHelper.GetNameEx(newName.BeginToken, newName.EndToken, Pullenti.Morph.MorphClass.Adjective, Pullenti.Morph.MorphCase.Undefined, noun.TerminItem.Gender, false, false);
if (!ok && !attachAlways)
{
if (Pullenti.Ner.Core.MiscHelper.IsExistsInDictionary(newName.BeginToken, newName.EndToken, Pullenti.Morph.MorphClass.Adjective | Pullenti.Morph.MorphClass.Pronoun | Pullenti.Morph.MorphClass.Verb))
{
if (exists != null)
{
foreach (Pullenti.Ner.Geo.GeoReferent e in exists)
{
if (e.FindSlot(Pullenti.Ner.Geo.GeoReferent.ATTR_NAME, name, true) != null)
{
ok = true;
break;
}
}
}
if (!ok)
{
return(null);
}
}
}
fullName = string.Format("{0} {1}", Pullenti.Ner.Core.ProperNameHelper.GetNameEx(li[0].BeginToken, noun.BeginToken.Previous, Pullenti.Morph.MorphClass.Adjective, Pullenti.Morph.MorphCase.Undefined, noun.TerminItem.Gender, false, false), noun.TerminItem.CanonicText);
}
}
else
{
if (!attachAlways || ((noun.TerminItem != null && noun.TerminItem.CanonicText == "ФЕДЕРАЦИЯ")))
{
bool isLatin = noun.Chars.IsLatinLetter && newName.Chars.IsLatinLetter;
if (li.IndexOf(noun) > li.IndexOf(newName))
{
if (!isLatin)
{
return(null);
}
}
if (!newName.IsDistrictName && !Pullenti.Ner.Core.BracketHelper.CanBeStartOfSequence(newName.BeginToken, false, false))
{
if (adjList.Count == 0 && Pullenti.Ner.Core.MiscHelper.IsExistsInDictionary(newName.BeginToken, newName.EndToken, Pullenti.Morph.MorphClass.Noun | Pullenti.Morph.MorphClass.Pronoun))
{
if (li.Count == 2 && noun.IsCityRegion && (noun.WhitespacesAfterCount < 2))
{
}
else
{
return(null);
}
}
if (!isLatin)
{
if ((noun.TerminItem.IsRegion && !attachAlways && ((!adjTerrBefore || newName.IsDoubt))) && !noun.IsCityRegion && !noun.TerminItem.IsSpecificPrefix)
{
if (!MiscLocationHelper.CheckGeoObjectBefore(noun.BeginToken))
{
if (!noun.IsDoubt && noun.BeginToken != noun.EndToken)
{
}
else if ((noun.TerminItem.IsAlwaysPrefix && li.Count == 2 && li[0] == noun) && li[1] == newName)
{
}
else
{
return(null);
}
}
}
if (noun.IsDoubt && adjList.Count == 0)
{
if (noun.TerminItem.Acronym == "МО" || noun.TerminItem.Acronym == "ЛО")
{
if (k == (li.Count - 1) && li[k].TerminItem != null)
{
addNoun = li[k];
k++;
}
else if (li.Count == 2 && noun == li[0] && newName.ToString().EndsWith("совет"))
{
}
else
{
return(null);
}
}
else
{
return(null);
}
}
Pullenti.Ner.ReferentToken pers = newName.Kit.ProcessReferent("PERSON", newName.BeginToken);
if (pers != null)
{
return(null);
}
}
}
}
name = Pullenti.Ner.Core.MiscHelper.GetTextValue(newName.BeginToken, newName.EndToken, Pullenti.Ner.Core.GetTextAttr.No);
if (newName.BeginToken != newName.EndToken)
{
for (Pullenti.Ner.Token ttt = newName.BeginToken.Next; ttt != null && ttt.EndChar <= newName.EndChar; ttt = ttt.Next)
{
if (ttt.Chars.IsLetter)
{
TerrItemToken ty = TerrItemToken.TryParse(ttt, null, false, false, null);
if ((ty != null && ty.TerminItem != null && noun != null) && ((ty.TerminItem.CanonicText.Contains(noun.TerminItem.CanonicText) || noun.TerminItem.CanonicText.Contains(ty.TerminItem.CanonicText))))
{
name = Pullenti.Ner.Core.MiscHelper.GetTextValue(newName.BeginToken, ttt.Previous, Pullenti.Ner.Core.GetTextAttr.No);
break;
}
}
}
}
if (adjList.Count > 0)
{
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(adjList[0].BeginToken, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null && npt.EndToken == noun.EndToken)
{
altName = string.Format("{0} {1}", npt.GetNormalCaseText(null, Pullenti.Morph.MorphNumber.Undefined, Pullenti.Morph.MorphGender.Undefined, false), name);
}
}
}
}
else
{
if ((li.Count == 1 && noun != null && noun.EndToken.Next != null) && (noun.EndToken.Next.GetReferent() is Pullenti.Ner.Geo.GeoReferent))
{
Pullenti.Ner.Geo.GeoReferent g = noun.EndToken.Next.GetReferent() as Pullenti.Ner.Geo.GeoReferent;
if (noun.TerminItem != null)
{
string tyy = noun.TerminItem.CanonicText.ToLower();
bool ooo = false;
if (g.FindSlot(Pullenti.Ner.Geo.GeoReferent.ATTR_TYPE, tyy, true) != null)
{
ooo = true;
}
else if (tyy.EndsWith("район") && g.FindSlot(Pullenti.Ner.Geo.GeoReferent.ATTR_TYPE, "район", true) != null)
{
ooo = true;
}
if (ooo)
{
return new Pullenti.Ner.ReferentToken(g, noun.BeginToken, noun.EndToken.Next)
{
Morph = noun.BeginToken.Morph
}
}
;
}
}
if ((li.Count == 1 && noun == li[0] && li[0].TerminItem != null) && TerrItemToken.TryParse(li[0].EndToken.Next, null, true, false, null) == null && TerrItemToken.TryParse(li[0].BeginToken.Previous, null, true, false, null) == null)
{
if (li[0].Morph.Number == Pullenti.Morph.MorphNumber.Plural)
{
return(null);
}
int cou = 0;
string str = li[0].TerminItem.CanonicText.ToLower();
for (Pullenti.Ner.Token tt = li[0].BeginToken.Previous; tt != null; tt = tt.Previous)
{
if (tt.IsNewlineAfter)
{
cou += 10;
}
else
{
cou++;
}
if (cou > 500)
{
break;
}
Pullenti.Ner.Geo.GeoReferent g = tt.GetReferent() as Pullenti.Ner.Geo.GeoReferent;
if (g == null)
{
continue;
}
bool ok = true;
cou = 0;
for (tt = li[0].EndToken.Next; tt != null; tt = tt.Next)
{
if (tt.IsNewlineBefore)
{
cou += 10;
}
else
{
cou++;
}
if (cou > 500)
{
break;
}
TerrItemToken tee = TerrItemToken.TryParse(tt, null, true, false, null);
if (tee == null)
{
continue;
}
ok = false;
break;
}
if (ok)
{
for (int ii = 0; g != null && (ii < 3); g = g.Higher, ii++)
{
if (g.FindSlot(Pullenti.Ner.Geo.GeoReferent.ATTR_TYPE, str, true) != null)
{
return new Pullenti.Ner.ReferentToken(g, li[0].BeginToken, li[0].EndToken)
{
Morph = noun.BeginToken.Morph
}
}
;
}
}
break;
}
}
return(null);
}
Pullenti.Ner.Geo.GeoReferent ter = null;
if (exObj != null && (exObj.Tag is Pullenti.Ner.Geo.GeoReferent))
{
ter = exObj.Tag as Pullenti.Ner.Geo.GeoReferent;
}
else
{
ter = new Pullenti.Ner.Geo.GeoReferent();
if (exObj != null)
{
Pullenti.Ner.Geo.GeoReferent geo = exObj.OntoItem.Referent as Pullenti.Ner.Geo.GeoReferent;
if (geo != null && !geo.IsCity)
{
ter.MergeSlots2(geo, li[0].Kit.BaseLanguage);
}
else
{
ter.AddName(name);
}
if (noun == null && exObj.CanBeCity)
{
ter.AddTypCity(li[0].Kit.BaseLanguage);
}
else
{
}
}
else if (newName != null)
{
ter.AddName(name);
if (altName != null)
{
ter.AddName(altName);
}
}
if (noun != null)
{
if (noun.TerminItem.CanonicText == "АО")
{
ter.AddTyp((li[0].Kit.BaseLanguage.IsUa ? "АВТОНОМНИЙ ОКРУГ" : "АВТОНОМНЫЙ ОКРУГ"));
}
else if (noun.TerminItem.CanonicText == "МУНИЦИПАЛЬНОЕ СОБРАНИЕ" || noun.TerminItem.CanonicText == "МУНІЦИПАЛЬНЕ ЗБОРИ")
{
ter.AddTyp((li[0].Kit.BaseLanguage.IsUa ? "МУНІЦИПАЛЬНЕ УТВОРЕННЯ" : "МУНИЦИПАЛЬНОЕ ОБРАЗОВАНИЕ"));
}
else if (noun.TerminItem.Acronym == "МО" && addNoun != null)
{
ter.AddTyp(addNoun.TerminItem.CanonicText);
}
else
{
if (noun.TerminItem.CanonicText == "СОЮЗ" && exObj != null && exObj.EndChar > noun.EndChar)
{
return new Pullenti.Ner.ReferentToken(ter, exObj.BeginToken, exObj.EndToken)
{
Morph = exObj.Morph
}
}
;
ter.AddTyp(noun.TerminItem.CanonicText);
if (noun.TerminItem.IsRegion && ter.IsState)
{
ter.AddTypReg(li[0].Kit.BaseLanguage);
}
}
}
if (ter.IsState && ter.IsRegion)
{
foreach (TerrItemToken a in adjList)
{
if (a.TerminItem.IsRegion)
{
ter.AddTypReg(li[0].Kit.BaseLanguage);
break;
}
}
}
if (ter.IsState)
{
if (fullName != null)
{
ter.AddName(fullName);
}
}
}
Pullenti.Ner.ReferentToken res = new Pullenti.Ner.ReferentToken(ter, li[0].BeginToken, li[k - 1].EndToken);
if (noun != null && noun.Morph.Class.IsNoun)
{
res.Morph = noun.Morph;
}
else
{
res.Morph = new Pullenti.Ner.MorphCollection();
for (int ii = 0; ii < k; ii++)
{
foreach (Pullenti.Morph.MorphBaseInfo v in li[ii].Morph.Items)
{
Pullenti.Morph.MorphBaseInfo bi = new Pullenti.Morph.MorphBaseInfo();
bi.CopyFrom(v);
if (noun != null)
{
if (bi.Class.IsAdjective)
{
bi.Class = Pullenti.Morph.MorphClass.Noun;
}
}
res.Morph.AddItem(bi);
}
}
}
if (li[0].TerminItem != null && li[0].TerminItem.IsSpecificPrefix)
{
res.BeginToken = li[0].EndToken.Next;
}
if (addNoun != null && addNoun.EndChar > res.EndChar)
{
res.EndToken = addNoun.EndToken;
}
if ((res.BeginToken.Previous is Pullenti.Ner.TextToken) && (res.WhitespacesBeforeCount < 2))
{
Pullenti.Ner.TextToken tt = res.BeginToken.Previous as Pullenti.Ner.TextToken;
if (tt.Term == "АР")
{
foreach (string ty in ter.Typs)
{
if (ty.Contains("республика") || ty.Contains("республіка"))
{
res.BeginToken = tt;
break;
}
}
}
}
return(res);
}