public override void Process(Pullenti.Ner.Core.AnalysisKit kit)
{
Pullenti.Ner.Core.AnalyzerDataWithOntology ad = kit.GetAnalyzerData(this) as Pullenti.Ner.Core.AnalyzerDataWithOntology;
for (Pullenti.Ner.Token t = kit.FirstToken; t != null; t = t.Next)
{
List <Pullenti.Ner.Named.Internal.NamedItemToken> li = Pullenti.Ner.Named.Internal.NamedItemToken.TryParseList(t, ad.LocalOntology);
if (li == null || li.Count == 0)
{
continue;
}
Pullenti.Ner.ReferentToken rt = _tryAttach(li);
if (rt != null)
{
rt.Referent = ad.RegisterReferent(rt.Referent);
kit.EmbedToken(rt);
t = rt;
continue;
}
}
}
// Основная функция выделения объектов
public override void Process(Pullenti.Ner.Core.AnalysisKit kit)
{
Pullenti.Ner.Core.AnalyzerDataWithOntology ad = kit.GetAnalyzerData(this) as Pullenti.Ner.Core.AnalyzerDataWithOntology;
for (int k = 0; k < 2; k++)
{
bool detectNewDenoms = false;
DateTime dt = DateTime.Now;
for (Pullenti.Ner.Token t = kit.FirstToken; t != null; t = t.Next)
{
if (t.IsWhitespaceBefore)
{
}
else if (t.Previous != null && ((t.Previous.IsCharOf(",") || Pullenti.Ner.Core.BracketHelper.CanBeStartOfSequence(t.Previous, false, false))))
{
}
else
{
continue;
}
Pullenti.Ner.ReferentToken rt0 = this.TryAttachSpec(t);
if (rt0 != null)
{
rt0.Referent = ad.RegisterReferent(rt0.Referent);
kit.EmbedToken(rt0);
t = rt0;
continue;
}
if (!t.Chars.IsLetter)
{
continue;
}
if (!this.CanBeStartOfDenom(t))
{
continue;
}
if (((DateTime.Now - dt)).TotalMinutes > 1)
{
break;
}
List <Pullenti.Ner.Core.IntOntologyToken> ot = null;
ot = ad.LocalOntology.TryAttach(t, null, false);
if (ot != null && (ot[0].Item.Referent is DenominationReferent))
{
if (this.CheckAttach(ot[0].BeginToken, ot[0].EndToken))
{
DenominationReferent cl = ot[0].Item.Referent.Clone() as DenominationReferent;
cl.Occurrence.Clear();
Pullenti.Ner.ReferentToken rt = new Pullenti.Ner.ReferentToken(cl, ot[0].BeginToken, ot[0].EndToken);
kit.EmbedToken(rt);
t = rt;
continue;
}
}
if (k > 0)
{
continue;
}
if (t != null && t.Kit.Ontology != null)
{
if ((((ot = t.Kit.Ontology.AttachToken(DenominationReferent.OBJ_TYPENAME, t)))) != null)
{
if (this.CheckAttach(ot[0].BeginToken, ot[0].EndToken))
{
DenominationReferent dr = new DenominationReferent();
dr.MergeSlots(ot[0].Item.Referent, true);
Pullenti.Ner.ReferentToken rt = new Pullenti.Ner.ReferentToken(ad.RegisterReferent(dr), ot[0].BeginToken, ot[0].EndToken);
kit.EmbedToken(rt);
t = rt;
continue;
}
}
}
rt0 = this.TryAttach(t, false);
if (rt0 != null)
{
rt0.Referent = ad.RegisterReferent(rt0.Referent);
kit.EmbedToken(rt0);
detectNewDenoms = true;
t = rt0;
if (ad.LocalOntology.Items.Count > 1000)
{
break;
}
}
}
if (!detectNewDenoms)
{
break;
}
}
}
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 TryAttachCity(List <CityItemToken> li, Pullenti.Ner.Core.AnalyzerDataWithOntology ad, bool always = false)
{
if (li == null)
{
return(null);
}
Pullenti.Ner.Core.IntOntologyItem oi;
if (li.Count > 2 && li[0].Typ == CityItemToken.ItemType.Misc && li[1].Typ == CityItemToken.ItemType.Noun)
{
li[1].Doubtful = false;
li.RemoveAt(0);
}
Pullenti.Ner.ReferentToken res = null;
if (res == null && li.Count > 1)
{
res = Try4(li);
if (res != null && res.EndChar <= li[1].EndChar)
{
res = null;
}
}
if (res == null)
{
res = Try1(li, out oi, ad);
}
if (res == null)
{
res = _tryNounName(li, out oi, false);
}
if (res == null)
{
res = _tryNameExist(li, out oi, false);
}
if (res == null)
{
res = Try4(li);
}
if (res == null && always)
{
res = _tryNounName(li, out oi, true);
}
if (res == null && always)
{
if (Pullenti.Ner.Address.Internal.AddressItemToken.TryAttachOrg(li[0].BeginToken) != null)
{
}
else
{
res = _tryNameExist(li, out oi, true);
}
}
if (res == null)
{
return(null);
}
if (res != null && res.Morph != null)
{
}
if (res.BeginToken.Previous != null)
{
if (res.BeginToken.Previous.IsValue("ТЕРРИТОРИЯ", null))
{
res.BeginToken = res.BeginToken.Previous;
res.Morph = res.BeginToken.Morph;
}
if ((Pullenti.Ner.Core.BracketHelper.CanBeStartOfSequence(res.BeginToken.Previous, false, false) && Pullenti.Ner.Core.BracketHelper.CanBeEndOfSequence(res.EndToken.Next, false, null, false) && res.BeginToken.Previous.Previous != null) && res.BeginToken.Previous.Previous.IsValue("ТЕРРИТОРИЯ", null))
{
res.BeginToken = res.BeginToken.Previous.Previous;
res.Morph = res.BeginToken.Morph;
res.EndToken = res.EndToken.Next;
}
}
return(res);
}