static void _optimizeTokens(Pullenti.Semantic.SemObject o)
{
for (int i = 0; i < o.Tokens.Count; i++)
{
bool ch = false;
for (int j = 0; j < (o.Tokens.Count - 1); j++)
{
if (_compareToks(o.Tokens[j], o.Tokens[j + 1]) > 0)
{
Pullenti.Ner.MetaToken t = o.Tokens[j];
o.Tokens[j] = o.Tokens[j + 1];
o.Tokens[j + 1] = t;
ch = true;
}
}
if (!ch)
{
break;
}
}
for (int i = 0; i < (o.Tokens.Count - 1); i++)
{
if (o.Tokens[i].EndToken.Next == o.Tokens[i + 1].BeginToken)
{
o.Tokens[i] = new Pullenti.Ner.MetaToken(o.Tokens[i].BeginToken, o.Tokens[i + 1].EndToken);
o.Tokens.RemoveAt(i + 1);
i--;
}
}
}
public static Pullenti.Ner.MetaToken TryAttachNordWest(Pullenti.Ner.Token t)
{
if (!(t is Pullenti.Ner.TextToken))
{
return(null);
}
Pullenti.Ner.Core.TerminToken tok = m_Nords.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok == null)
{
return(null);
}
Pullenti.Ner.MetaToken res = new Pullenti.Ner.MetaToken(t, t)
{
Morph = t.Morph
};
Pullenti.Ner.Token t1 = null;
if ((t.Next != null && t.Next.IsHiphen && !t.IsWhitespaceAfter) && !t.IsWhitespaceAfter)
{
t1 = t.Next.Next;
}
else if (t.Morph.Class.IsAdjective && (t.WhitespacesAfterCount < 2))
{
t1 = t.Next;
}
if (t1 != null)
{
if ((((tok = m_Nords.TryParse(t1, Pullenti.Ner.Core.TerminParseAttr.No)))) != null)
{
res.EndToken = tok.EndToken;
res.Morph = tok.Morph;
}
}
return(res);
}
public static Pullenti.Semantic.Utils.DerivateWord FindWordInGroup(Pullenti.Ner.MetaToken mt, Pullenti.Semantic.Utils.DerivateGroup gr)
{
if (gr == null || mt == null)
{
return(null);
}
Pullenti.Ner.Token t = null;
if (mt is Pullenti.Ner.Core.NounPhraseToken)
{
t = (mt as Pullenti.Ner.Core.NounPhraseToken).Noun.EndToken;
}
else if ((mt is SemanticAbstractSlave) && ((mt as SemanticAbstractSlave).Source is Pullenti.Ner.Core.NounPhraseToken))
{
t = ((mt as SemanticAbstractSlave).Source as Pullenti.Ner.Core.NounPhraseToken).Noun.EndToken;
}
else
{
t = mt.EndToken;
}
foreach (Pullenti.Semantic.Utils.DerivateWord w in gr.Words)
{
if (w.Class != null && w.Class.IsNoun && w.Lang.IsRu)
{
if (t.IsValue(w.Spelling, null))
{
return(w);
}
}
}
return(null);
}
public SentItem(Pullenti.Ner.MetaToken mt)
{
Source = mt;
if (mt is Pullenti.Ner.Core.NounPhraseToken)
{
Pullenti.Ner.Core.NounPhraseToken npt = mt as Pullenti.Ner.Core.NounPhraseToken;
if (npt.Preposition != null)
{
Prep = npt.Preposition.Normal;
}
else
{
Prep = "";
}
Typ = SentItemType.Noun;
string Normal = npt.Noun.GetNormalCaseText(Pullenti.Morph.MorphClass.Noun, Pullenti.Morph.MorphNumber.Singular, Pullenti.Morph.MorphGender.Masculine, false);
if (Normal != null)
{
DrGroups = Pullenti.Semantic.Utils.DerivateService.FindDerivates(Normal, true, null);
}
}
else if ((mt is Pullenti.Ner.ReferentToken) || (mt is Pullenti.Ner.Measure.Internal.NumbersWithUnitToken))
{
Typ = SentItemType.Noun;
}
else if (mt is AdverbToken)
{
Typ = SentItemType.Adverb;
}
else if (mt is Pullenti.Ner.Core.ConjunctionToken)
{
Typ = SentItemType.Conj;
}
else if (mt is DelimToken)
{
Typ = SentItemType.Delim;
}
else if (mt is Pullenti.Ner.Core.VerbPhraseToken)
{
Pullenti.Ner.Core.VerbPhraseToken vpt = mt as Pullenti.Ner.Core.VerbPhraseToken;
string Normal = (vpt.FirstVerb.VerbMorph == null ? null : vpt.FirstVerb.VerbMorph.NormalFull ?? vpt.FirstVerb.VerbMorph.NormalCase);
if (Normal != null)
{
DrGroups = Pullenti.Semantic.Utils.DerivateService.FindDerivates(Normal, true, null);
}
if (vpt.FirstVerb != vpt.LastVerb)
{
Normal = (vpt.LastVerb.VerbMorph == null ? vpt.GetNormalCaseText(null, Pullenti.Morph.MorphNumber.Undefined, Pullenti.Morph.MorphGender.Undefined, false) : vpt.LastVerb.VerbMorph.NormalFull ?? vpt.LastVerb.VerbMorph.NormalCase);
DrGroups2 = Pullenti.Semantic.Utils.DerivateService.FindDerivates(Normal, true, null);
}
else
{
DrGroups2 = DrGroups;
}
Prep = (vpt.Preposition == null ? "" : vpt.Preposition.Normal);
Typ = SentItemType.Verb;
}
}
static void _corrPrevNext(Pullenti.Ner.MetaToken mt, Pullenti.Ner.Token prev, Pullenti.Ner.Token next)
{
mt.BeginToken.m_Previous = prev;
mt.EndToken.m_Next = next;
for (Pullenti.Ner.Token t = mt.BeginToken; t != null && t.EndChar <= mt.EndChar; t = t.Next)
{
if (t is Pullenti.Ner.MetaToken)
{
_corrPrevNext(t as Pullenti.Ner.MetaToken, t.Previous, t.Next);
}
}
}
public static string GetKeyword(Pullenti.Ner.MetaToken mt)
{
Pullenti.Ner.Core.VerbPhraseToken vpt = mt as Pullenti.Ner.Core.VerbPhraseToken;
if (vpt != null)
{
return(vpt.LastVerb.VerbMorph.NormalFull ?? vpt.LastVerb.VerbMorph.NormalCase);
}
Pullenti.Ner.Core.NounPhraseToken npt = mt as Pullenti.Ner.Core.NounPhraseToken;
if (npt != null)
{
return(npt.Noun.EndToken.GetNormalCaseText(Pullenti.Morph.MorphClass.Noun, Pullenti.Morph.MorphNumber.Singular, Pullenti.Morph.MorphGender.Undefined, false));
}
return(null);
}
static Pullenti.Semantic.Utils.ControlModelItem FindControlItem(Pullenti.Ner.MetaToken mt, Pullenti.Semantic.Utils.DerivateGroup gr)
{
if (gr == null)
{
return(null);
}
if (mt is Pullenti.Ner.Core.NounPhraseToken)
{
Pullenti.Ner.Token t = (mt as Pullenti.Ner.Core.NounPhraseToken).Noun.EndToken;
foreach (Pullenti.Semantic.Utils.ControlModelItem m in gr.Model.Items)
{
if (m.Word != null)
{
if (t.IsValue(m.Word, null))
{
return(m);
}
}
}
foreach (Pullenti.Semantic.Utils.DerivateWord w in gr.Words)
{
if (w.Attrs.IsVerbNoun)
{
if (t.IsValue(w.Spelling, null))
{
return(gr.Model.FindItemByTyp(Pullenti.Semantic.Utils.ControlModelItemType.Noun));
}
}
}
return(null);
}
if (mt is Pullenti.Ner.Core.VerbPhraseItemToken)
{
Pullenti.Ner.Core.VerbPhraseItemToken ti = mt as Pullenti.Ner.Core.VerbPhraseItemToken;
bool rev = ti.IsVerbReversive || ti.IsVerbPassive;
foreach (Pullenti.Semantic.Utils.ControlModelItem it in gr.Model.Items)
{
if (rev && it.Typ == Pullenti.Semantic.Utils.ControlModelItemType.Reflexive)
{
return(it);
}
else if (!rev && it.Typ == Pullenti.Semantic.Utils.ControlModelItemType.Verb)
{
return(it);
}
}
}
return(null);
}
public override string ToString()
{
if (BeginNameToken == null || EndNameToken == null)
{
return("?");
}
Pullenti.Ner.MetaToken mt = new Pullenti.Ner.MetaToken(BeginNameToken, EndNameToken);
if (TypeValue == null)
{
return(string.Format("{0}: {1}", Rank, mt.ToString()));
}
else
{
return(string.Format("{0}: {1} ({2})", Rank, mt.ToString(), TypeValue));
}
}
static bool _containsTableChar(Pullenti.Ner.MetaToken mt)
{
for (Pullenti.Ner.Token tt = mt.BeginToken; tt != null && tt.EndChar <= mt.EndChar; tt = tt.Next)
{
if (tt is Pullenti.Ner.MetaToken)
{
if (_containsTableChar(tt as Pullenti.Ner.MetaToken))
{
return(true);
}
}
else if (((tt.IsTableControlChar && tt.Previous != null && !tt.Previous.IsTableControlChar) && tt.Next != null && !tt.Next.IsTableControlChar) && tt.Previous.BeginChar >= mt.BeginChar && tt.Next.EndChar <= mt.EndChar)
{
return(true);
}
}
return(false);
}
/// <summary>
/// Убрать метатокен из цепочки, восстановив исходное
/// </summary>
/// <param name="t">удаляемый из цепочки метатокен</param>
/// <return>первый токен удалённого метатокена</return>
public Pullenti.Ner.Token DebedToken(Pullenti.Ner.Token t)
{
Pullenti.Ner.Referent r = t.GetReferent();
if (r != null)
{
foreach (Pullenti.Ner.TextAnnotation o in r.Occurrence)
{
if (o.BeginChar == t.BeginChar && o.EndChar == t.EndChar)
{
r.Occurrence.Remove(o);
break;
}
}
}
Pullenti.Ner.MetaToken mt = t as Pullenti.Ner.MetaToken;
if (mt == null)
{
return(t);
}
if (t.Next != null)
{
t.Next.Previous = mt.EndToken;
}
if (t.Previous != null)
{
t.Previous.Next = mt.BeginToken;
}
if (mt == FirstToken)
{
FirstToken = mt.BeginToken;
}
if (r != null && r.Occurrence.Count == 0)
{
foreach (AnalyzerData d in m_Datas.Values)
{
if (d.Referents.Contains(r))
{
d.RemoveReferent(r);
break;
}
}
}
return(mt.BeginToken);
}
static bool _compareListItemTails(Pullenti.Ner.MetaToken mt1, Pullenti.Ner.MetaToken mt2)
{
Pullenti.Ner.TextToken t1 = mt1.EndToken as Pullenti.Ner.TextToken;
Pullenti.Ner.TextToken t2 = mt2.EndToken as Pullenti.Ner.TextToken;
if (t1 == null || t2 == null)
{
return(true);
}
int k = 0;
int i1 = t1.Term.Length - 1;
int i2 = t2.Term.Length - 1;
for (; i1 > 0 && i2 > 0; i1--, i2--, k++)
{
if (t1.Term[i1] != t2.Term[i2])
{
break;
}
}
if (k >= 2)
{
return(true);
}
string nn = t2.GetNormalCaseText(null, Pullenti.Morph.MorphNumber.Singular, Pullenti.Morph.MorphGender.Undefined, false);
if (t1.IsValue(nn, null))
{
return(true);
}
if (((t1.Morph.Number & t2.Morph.Number)) == Pullenti.Morph.MorphNumber.Undefined)
{
return(false);
}
if (((t1.Morph.Case & t2.Morph.Case)).IsUndefined)
{
return(false);
}
if (t1.Morph.Class.IsVerb != t2.Morph.Class.IsVerb && t1.Morph.Class.IsAdjective != t2.Morph.Class.IsAdjective)
{
return(false);
}
return(true);
}
static List <SemanticLink> _tryCreateInf(Pullenti.Ner.MetaToken master, Pullenti.Ner.Core.VerbPhraseToken vpt2, Pullenti.Semantic.Utils.DerivateGroup gr)
{
Pullenti.Semantic.Utils.ControlModelItem cit = FindControlItem(master, gr);
List <SemanticLink> res = new List <SemanticLink>();
SemanticRole? rol = null;
if (cit != null && cit.Links.ContainsKey(Pullenti.Semantic.Utils.ControlModelQuestion.ToDo))
{
rol = cit.Links[Pullenti.Semantic.Utils.ControlModelQuestion.ToDo];
}
if (rol != null)
{
res.Add(new SemanticLink()
{
Rank = (rol.Value != SemanticRole.Common ? 2 : 1), Question = Pullenti.Semantic.Utils.ControlModelQuestion.ToDo
});
}
return(res);
}
static Pullenti.Ner.Token DeserializeToken(Stream stream, Pullenti.Ner.Core.AnalysisKit kit, int vers)
{
short typ = DeserializeShort(stream);
if (typ == 0)
{
return(null);
}
Pullenti.Ner.Token t = null;
if (typ == 1)
{
t = new Pullenti.Ner.TextToken(null, kit);
}
else if (typ == 2)
{
t = new Pullenti.Ner.NumberToken(null, null, null, Pullenti.Ner.NumberSpellingType.Digit, kit);
}
else if (typ == 3)
{
t = new Pullenti.Ner.ReferentToken(null, null, null, kit);
}
else
{
t = new Pullenti.Ner.MetaToken(null, null, kit);
}
t.Deserialize(stream, kit, vers);
if (t is Pullenti.Ner.MetaToken)
{
Pullenti.Ner.Token tt = DeserializeTokens(stream, kit, vers);
if (tt != null)
{
(t as Pullenti.Ner.MetaToken).m_BeginToken = tt;
for (; tt != null; tt = tt.Next)
{
(t as Pullenti.Ner.MetaToken).m_EndToken = tt;
}
}
}
return(t);
}
/// <summary>
/// Встроить токен в основную цепочку токенов
/// </summary>
/// <param name="mt">встраиваемый метатокен</param>
public void EmbedToken(Pullenti.Ner.MetaToken mt)
{
if (mt == null)
{
return;
}
if (mt.BeginChar > mt.EndChar)
{
Pullenti.Ner.Token bg = mt.BeginToken;
mt.BeginToken = mt.EndToken;
mt.EndToken = bg;
}
if (mt.BeginChar > mt.EndChar)
{
return;
}
if (mt.BeginToken == FirstToken)
{
FirstToken = mt;
}
else
{
Pullenti.Ner.Token tp = mt.BeginToken.Previous;
mt.Previous = tp;
}
Pullenti.Ner.Token tn = mt.EndToken.Next;
mt.Next = tn;
if (mt is Pullenti.Ner.ReferentToken)
{
if ((mt as Pullenti.Ner.ReferentToken).Referent != null)
{
(mt as Pullenti.Ner.ReferentToken).Referent.AddOccurence(new Pullenti.Ner.TextAnnotation()
{
Sofa = Sofa, BeginChar = mt.BeginChar, EndChar = mt.EndChar
});
}
}
}
static Pullenti.Ner.ReferentToken TryParseThesis(Pullenti.Ner.Token t)
{
if (t == null)
{
return(null);
}
Pullenti.Ner.Token t0 = t;
Pullenti.Ner.Token tt = t;
Pullenti.Morph.MorphClass mc = tt.GetMorphClassInDictionary();
Pullenti.Ner.MetaToken preamb = null;
if (mc.IsConjunction)
{
return(null);
}
if (t.IsValue("LET", null))
{
return(null);
}
if (mc.IsPreposition || mc.IsMisc || mc.IsAdverb)
{
if (!Pullenti.Ner.Core.MiscHelper.IsEngArticle(tt))
{
for (tt = tt.Next; tt != null; tt = tt.Next)
{
if (tt.IsComma)
{
break;
}
if (tt.IsChar('('))
{
Pullenti.Ner.Core.BracketSequenceToken br = Pullenti.Ner.Core.BracketHelper.TryParse(tt, Pullenti.Ner.Core.BracketParseAttr.No, 100);
if (br != null)
{
tt = br.EndToken;
continue;
}
}
if (Pullenti.Ner.Core.MiscHelper.CanBeStartOfSentence(tt))
{
break;
}
Pullenti.Ner.Core.NounPhraseToken npt0 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(tt, Pullenti.Ner.Core.NounPhraseParseAttr.ParseNumericAsAdjective | Pullenti.Ner.Core.NounPhraseParseAttr.ReferentCanBeNoun, 0, null);
if (npt0 != null)
{
tt = npt0.EndToken;
continue;
}
if (tt.GetMorphClassInDictionary().IsVerb)
{
break;
}
}
if (tt == null || !tt.IsComma || tt.Next == null)
{
return(null);
}
preamb = new Pullenti.Ner.MetaToken(t0, tt.Previous);
tt = tt.Next;
}
}
Pullenti.Ner.Token t1 = tt;
mc = tt.GetMorphClassInDictionary();
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(tt, Pullenti.Ner.Core.NounPhraseParseAttr.ParseNumericAsAdjective | Pullenti.Ner.Core.NounPhraseParseAttr.ReferentCanBeNoun | Pullenti.Ner.Core.NounPhraseParseAttr.ParseAdverbs, 0, null);
if (npt == null && (tt is Pullenti.Ner.TextToken))
{
if (tt.Chars.IsAllUpper)
{
npt = new Pullenti.Ner.Core.NounPhraseToken(tt, tt);
}
else if (!tt.Chars.IsAllLower)
{
if (mc.IsProper || preamb != null)
{
npt = new Pullenti.Ner.Core.NounPhraseToken(tt, tt);
}
}
}
if (npt == null)
{
return(null);
}
if (mc.IsPersonalPronoun)
{
return(null);
}
Pullenti.Ner.Token t2 = npt.EndToken.Next;
if (t2 == null || Pullenti.Ner.Core.MiscHelper.CanBeStartOfSentence(t2) || !(t2 is Pullenti.Ner.TextToken))
{
return(null);
}
if (!t2.GetMorphClassInDictionary().IsVerb)
{
return(null);
}
Pullenti.Ner.Token t3 = t2;
for (tt = t2.Next; tt != null; tt = tt.Next)
{
if (!tt.GetMorphClassInDictionary().IsVerb)
{
break;
}
}
for (; tt != null; tt = tt.Next)
{
if (tt.Next == null)
{
t3 = tt;
break;
}
if (tt.IsCharOf(".;!?"))
{
if (Pullenti.Ner.Core.MiscHelper.CanBeStartOfSentence(tt.Next))
{
t3 = tt;
break;
}
}
if (!(tt is Pullenti.Ner.TextToken))
{
continue;
}
if (Pullenti.Ner.Core.BracketHelper.CanBeStartOfSequence(tt, false, false))
{
Pullenti.Ner.Core.BracketSequenceToken br = Pullenti.Ner.Core.BracketHelper.TryParse(tt, Pullenti.Ner.Core.BracketParseAttr.No, 100);
if (br != null)
{
tt = br.EndToken;
continue;
}
}
}
tt = t3;
if (t3.IsCharOf(";.!?"))
{
tt = tt.Previous;
}
string txt = Pullenti.Ner.Core.MiscHelper.GetTextValue(t2, tt, Pullenti.Ner.Core.GetTextAttr.KeepRegister | Pullenti.Ner.Core.GetTextAttr.KeepQuotes);
if (txt == null || (txt.Length < 15))
{
return(null);
}
if (t0 != t1)
{
tt = t1.Previous;
if (tt.IsComma)
{
tt = tt.Previous;
}
string txt0 = Pullenti.Ner.Core.MiscHelper.GetTextValue(t0, tt, Pullenti.Ner.Core.GetTextAttr.KeepRegister | Pullenti.Ner.Core.GetTextAttr.KeepQuotes);
if (txt0 != null && txt0.Length > 10)
{
if (t0.Chars.IsCapitalUpper)
{
txt0 = char.ToLower(txt0[0]) + txt0.Substring(1);
}
txt = string.Format("{0}, {1}", txt, txt0);
}
}
tt = t1;
if (Pullenti.Ner.Core.MiscHelper.IsEngArticle(tt))
{
tt = tt.Next;
}
string nam = Pullenti.Ner.Core.MiscHelper.GetTextValue(tt, t2.Previous, Pullenti.Ner.Core.GetTextAttr.KeepQuotes);
if (nam.StartsWith("SO-CALLED"))
{
nam = nam.Substring(9).Trim();
}
Pullenti.Ner.Definition.DefinitionReferent dr = new Pullenti.Ner.Definition.DefinitionReferent();
dr.Kind = Pullenti.Ner.Definition.DefinitionKind.Assertation;
dr.AddSlot(Pullenti.Ner.Definition.DefinitionReferent.ATTR_TERMIN, nam, false, 0);
dr.AddSlot(Pullenti.Ner.Definition.DefinitionReferent.ATTR_VALUE, txt, false, 0);
return(new Pullenti.Ner.ReferentToken(dr, t0, t3));
}
public static NamedItemToken TryParse(Pullenti.Ner.Token t, Pullenti.Ner.Core.IntOntologyCollection locOnto)
{
if (t == null)
{
return(null);
}
if (t is Pullenti.Ner.ReferentToken)
{
Pullenti.Ner.Referent r = t.GetReferent();
if ((r.TypeName == "PERSON" || r.TypeName == "PERSONPROPERTY" || (r is Pullenti.Ner.Geo.GeoReferent)) || r.TypeName == "ORGANIZATION")
{
return new NamedItemToken(t, t)
{
Ref = r, Morph = t.Morph
}
}
;
return(null);
}
Pullenti.Ner.Core.TerminToken typ = m_Types.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No);
Pullenti.Ner.Core.TerminToken nam = m_Names.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No);
if (typ != null)
{
if (!(t is Pullenti.Ner.TextToken))
{
return(null);
}
NamedItemToken res = new NamedItemToken(typ.BeginToken, typ.EndToken)
{
Morph = typ.Morph, Chars = typ.Chars
};
res.Kind = (Pullenti.Ner.Named.NamedEntityKind)typ.Termin.Tag;
res.TypeValue = typ.Termin.CanonicText;
if ((nam != null && nam.EndToken == typ.EndToken && !t.Chars.IsAllLower) && ((Pullenti.Ner.Named.NamedEntityKind)nam.Termin.Tag) == res.Kind)
{
res.NameValue = nam.Termin.CanonicText;
res.IsWellknown = true;
}
return(res);
}
if (nam != null)
{
if (nam.BeginToken.Chars.IsAllLower)
{
return(null);
}
NamedItemToken res = new NamedItemToken(nam.BeginToken, nam.EndToken)
{
Morph = nam.Morph, Chars = nam.Chars
};
res.Kind = (Pullenti.Ner.Named.NamedEntityKind)nam.Termin.Tag;
res.NameValue = nam.Termin.CanonicText;
bool ok = true;
if (!t.IsWhitespaceBefore && t.Previous != null)
{
ok = false;
}
else if (!t.IsWhitespaceAfter && t.Next != null)
{
if (t.Next.IsCharOf(",.;!?") && t.Next.IsWhitespaceAfter)
{
}
else
{
ok = false;
}
}
if (ok)
{
res.IsWellknown = true;
res.TypeValue = nam.Termin.Tag2 as string;
}
return(res);
}
Pullenti.Ner.MetaToken adj = Pullenti.Ner.Geo.Internal.MiscLocationHelper.TryAttachNordWest(t);
if (adj != null)
{
if (adj.Morph.Class.IsNoun)
{
if (adj.EndToken.IsValue("ВОСТОК", null))
{
if (adj.BeginToken == adj.EndToken)
{
return(null);
}
NamedItemToken re = new NamedItemToken(t, adj.EndToken)
{
Morph = adj.Morph
};
re.Kind = Pullenti.Ner.Named.NamedEntityKind.Location;
re.NameValue = Pullenti.Ner.Core.MiscHelper.GetTextValue(t, adj.EndToken, Pullenti.Ner.Core.GetTextAttr.FirstNounGroupToNominative);
re.IsWellknown = true;
return(re);
}
return(null);
}
if (adj.WhitespacesAfterCount > 2)
{
return(null);
}
if ((adj.EndToken.Next is Pullenti.Ner.ReferentToken) && (adj.EndToken.Next.GetReferent() is Pullenti.Ner.Geo.GeoReferent))
{
NamedItemToken re = new NamedItemToken(t, adj.EndToken.Next)
{
Morph = adj.EndToken.Next.Morph
};
re.Kind = Pullenti.Ner.Named.NamedEntityKind.Location;
re.NameValue = Pullenti.Ner.Core.MiscHelper.GetTextValue(t, adj.EndToken.Next, Pullenti.Ner.Core.GetTextAttr.FirstNounGroupToNominative);
re.IsWellknown = true;
re.Ref = adj.EndToken.Next.GetReferent();
return(re);
}
NamedItemToken res = TryParse(adj.EndToken.Next, locOnto);
if (res != null && res.Kind == Pullenti.Ner.Named.NamedEntityKind.Location)
{
string s = adj.GetNormalCaseText(Pullenti.Morph.MorphClass.Adjective, Pullenti.Morph.MorphNumber.Singular, res.Morph.Gender, false);
if (s != null)
{
if (res.NameValue == null)
{
res.NameValue = s.ToUpper();
}
else
{
res.NameValue = string.Format("{0} {1}", s.ToUpper(), res.NameValue);
res.TypeValue = null;
}
res.BeginToken = t;
res.Chars = t.Chars;
res.IsWellknown = true;
return(res);
}
}
}
if (t.Chars.IsCapitalUpper && !Pullenti.Ner.Core.MiscHelper.CanBeStartOfSentence(t))
{
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null && npt.Adjectives.Count > 0)
{
NamedItemToken test = TryParse(npt.Noun.BeginToken, locOnto);
if (test != null && test.EndToken == npt.EndToken && test.TypeValue != null)
{
test.BeginToken = t;
StringBuilder tmp = new StringBuilder();
foreach (Pullenti.Ner.MetaToken a in npt.Adjectives)
{
string s = a.GetNormalCaseText(Pullenti.Morph.MorphClass.Adjective, Pullenti.Morph.MorphNumber.Singular, test.Morph.Gender, false);
if (tmp.Length > 0)
{
tmp.Append(' ');
}
tmp.Append(s);
}
test.NameValue = tmp.ToString();
test.Chars = t.Chars;
if (test.Kind == Pullenti.Ner.Named.NamedEntityKind.Location)
{
test.IsWellknown = true;
}
return(test);
}
}
}
if ((Pullenti.Ner.Core.BracketHelper.IsBracket(t, true) && t.Next != null && t.Next.Chars.IsLetter) && !t.Next.Chars.IsAllLower)
{
Pullenti.Ner.Core.BracketSequenceToken br = Pullenti.Ner.Core.BracketHelper.TryParse(t, Pullenti.Ner.Core.BracketParseAttr.No, 100);
if (br != null)
{
NamedItemToken res = new NamedItemToken(t, br.EndToken);
res.IsInBracket = true;
res.NameValue = Pullenti.Ner.Core.MiscHelper.GetTextValue(t, br.EndToken, Pullenti.Ner.Core.GetTextAttr.No);
nam = m_Names.TryParse(t.Next, Pullenti.Ner.Core.TerminParseAttr.No);
if (nam != null && nam.EndToken == br.EndToken.Previous)
{
res.Kind = (Pullenti.Ner.Named.NamedEntityKind)nam.Termin.Tag;
res.IsWellknown = true;
res.NameValue = nam.Termin.CanonicText;
}
return(res);
}
}
if (((t is Pullenti.Ner.TextToken) && t.Chars.IsLetter && !t.Chars.IsAllLower) && t.LengthChar > 2)
{
NamedItemToken res = new NamedItemToken(t, t)
{
Morph = t.Morph
};
string str = (t as Pullenti.Ner.TextToken).Term;
if (str.EndsWith("О") || str.EndsWith("И") || str.EndsWith("Ы"))
{
res.NameValue = str;
}
else
{
res.NameValue = t.GetNormalCaseText(null, Pullenti.Morph.MorphNumber.Undefined, Pullenti.Morph.MorphGender.Undefined, false);
}
res.Chars = t.Chars;
if (((!t.IsWhitespaceAfter && t.Next != null && t.Next.IsHiphen) && (t.Next.Next is Pullenti.Ner.TextToken) && !t.Next.Next.IsWhitespaceAfter) && t.Chars.IsCyrillicLetter == t.Next.Next.Chars.IsCyrillicLetter)
{
t = (res.EndToken = t.Next.Next);
res.NameValue = string.Format("{0}-{1}", res.NameValue, t.GetNormalCaseText(null, Pullenti.Morph.MorphNumber.Undefined, Pullenti.Morph.MorphGender.Undefined, false));
}
return(res);
}
return(null);
}
static List <SemanticLink> _tryCreateNoun(Pullenti.Ner.Core.NounPhraseToken npt1, Pullenti.Ner.MetaToken slave, Pullenti.Semantic.Utils.DerivateGroup gr)
{
if (npt1 == null || slave == null)
{
return(null);
}
if (slave is Pullenti.Ner.Core.VerbPhraseToken)
{
return(_tryCreateInf(npt1, slave as Pullenti.Ner.Core.VerbPhraseToken, gr));
}
SemanticAbstractSlave sla2 = slave as SemanticAbstractSlave;
List <SemanticLink> res = new List <SemanticLink>();
if (sla2 == null)
{
return(res);
}
Pullenti.Semantic.Utils.ControlModelItem cit = FindControlItem(npt1, gr);
_createRoles(cit, sla2.Preposition, sla2.Morph.Case, res, false, false);
if (res.Count == 1 && res[0].Role == SemanticRole.Agent && res[0].Question == Pullenti.Semantic.Utils.ControlModelQuestion.BaseInstrumental)
{
if (gr.Model.Items.Count > 0 && gr.Model.Items[0].Typ == Pullenti.Semantic.Utils.ControlModelItemType.Verb && gr.Model.Items[0].Links.ContainsKey(Pullenti.Semantic.Utils.ControlModelQuestion.BaseInstrumental))
{
res[0].Role = gr.Model.Items[0].Links[Pullenti.Semantic.Utils.ControlModelQuestion.BaseInstrumental];
}
}
bool ok = false;
Pullenti.Semantic.Utils.DerivateWord w = FindWordInGroup(npt1, gr);
if (w != null && w.NextWords != null && w.NextWords.Count > 0)
{
foreach (string n in w.NextWords)
{
if (sla2.Source != null)
{
if (sla2.Source.EndToken.IsValue(n, null))
{
ok = true;
break;
}
}
}
}
if (gr != null && gr.Model.Pacients.Count > 0)
{
foreach (string n in gr.Model.Pacients)
{
if (sla2.Source != null)
{
if (sla2.Source.EndToken.IsValue(n, null))
{
ok = true;
break;
}
}
}
}
if (ok)
{
if (res.Count == 0)
{
res.Add(new SemanticLink()
{
Question = Pullenti.Semantic.Utils.ControlModelQuestion.BaseGenetive, Role = SemanticRole.Pacient, Idiom = true
});
}
foreach (SemanticLink r in res)
{
r.Rank += 4;
if (r.Role == SemanticRole.Common)
{
r.Role = SemanticRole.Strong;
}
if (npt1.EndToken.Next == sla2.BeginToken)
{
r.Rank += 2;
}
r.Idiom = true;
}
}
return(res);
}
/// <summary>
/// Попробовать создать семантическую связь между элементами.
/// Элементом м.б. именная (NounPhraseToken) или глагольная группа (VerbPhraseToken).
/// </summary>
/// <param name="master">основной элемент</param>
/// <param name="slave">стыкуемый элемент (также м.б. SemanticAbstractSlave)</param>
/// <param name="onto">дополнительный онтологический словарь</param>
/// <return>список вариантов (возможно, пустой)</return>
public static List <SemanticLink> TryCreateLinks(Pullenti.Ner.MetaToken master, Pullenti.Ner.MetaToken slave, ISemanticOnto onto = null)
{
List <SemanticLink> res = new List <SemanticLink>();
Pullenti.Ner.Core.VerbPhraseToken vpt1 = master as Pullenti.Ner.Core.VerbPhraseToken;
Pullenti.Ner.Core.VerbPhraseToken vpt2 = slave as Pullenti.Ner.Core.VerbPhraseToken;
Pullenti.Ner.Core.NounPhraseToken npt1 = master as Pullenti.Ner.Core.NounPhraseToken;
if (slave is Pullenti.Ner.Core.NounPhraseToken)
{
slave = SemanticAbstractSlave.CreateFromNoun(slave as Pullenti.Ner.Core.NounPhraseToken);
}
SemanticAbstractSlave sla2 = slave as SemanticAbstractSlave;
if (vpt2 != null)
{
if (!vpt2.FirstVerb.IsVerbInfinitive || !vpt2.LastVerb.IsVerbInfinitive)
{
return(res);
}
}
List <Pullenti.Semantic.Utils.DerivateGroup> grs = FindDerivates(master);
if (grs == null || grs.Count == 0)
{
List <SemanticLink> rl = (vpt1 != null ? _tryCreateVerb(vpt1, slave, null) : _tryCreateNoun(npt1, slave, null));
if (rl != null)
{
res.AddRange(rl);
}
}
else
{
foreach (Pullenti.Semantic.Utils.DerivateGroup gr in grs)
{
List <SemanticLink> rl = (vpt1 != null ? _tryCreateVerb(vpt1, slave, gr) : _tryCreateNoun(npt1, slave, gr));
if (rl == null || rl.Count == 0)
{
continue;
}
res.AddRange(rl);
}
}
if ((npt1 != null && sla2 != null && sla2.Morph.Case.IsGenitive) && sla2.Preposition == null)
{
if (npt1.Noun.BeginToken.GetMorphClassInDictionary().IsPersonalPronoun)
{
}
else
{
bool hasGen = false;
foreach (SemanticLink r in res)
{
if (r.Question == Pullenti.Semantic.Utils.ControlModelQuestion.BaseGenetive)
{
hasGen = true;
break;
}
}
if (!hasGen)
{
res.Add(new SemanticLink()
{
Modelled = true, Master = npt1, Slave = sla2, Rank = 0.5, Question = Pullenti.Semantic.Utils.ControlModelQuestion.BaseGenetive
});
}
}
}
if (onto != null)
{
string str1 = GetKeyword(master);
string str2 = GetKeyword(slave);
if (str2 != null)
{
if (onto.CheckLink(str1, str2))
{
if (res.Count > 0)
{
foreach (SemanticLink r in res)
{
r.Rank += 3;
if (r.Role == SemanticRole.Common)
{
r.Role = SemanticRole.Strong;
}
}
}
else
{
res.Add(new SemanticLink()
{
Role = SemanticRole.Strong, Master = master, Slave = slave, Rank = 3
});
}
}
}
}
if (npt1 != null)
{
if (((npt1.Adjectives.Count > 0 && npt1.Adjectives[0].BeginToken.Morph.Class.IsPronoun)) || npt1.Anafor != null)
{
foreach (SemanticLink r in res)
{
if (r.Question == Pullenti.Semantic.Utils.ControlModelQuestion.BaseGenetive)
{
r.Rank -= 0.5;
if (r.Role == SemanticRole.Strong)
{
r.Role = SemanticRole.Common;
}
}
}
}
}
foreach (SemanticLink r in res)
{
if (r.Role == SemanticRole.Strong)
{
foreach (SemanticLink rr in res)
{
if (rr != r && rr.Role != SemanticRole.Strong)
{
rr.Rank /= 2;
}
}
}
}
for (int i = 0; i < res.Count; i++)
{
for (int j = 0; j < (res.Count - 1); j++)
{
if (res[j].CompareTo(res[j + 1]) > 0)
{
SemanticLink r = res[j];
res[j] = res[j + 1];
res[j + 1] = r;
}
}
}
foreach (SemanticLink r in res)
{
r.Master = master;
r.Slave = slave;
}
return(res);
}
public static MeasureToken TryParse(Pullenti.Ner.Token t, Pullenti.Ner.Core.TerminCollection addUnits, bool canBeSet = true, bool canUnitsAbsent = false, bool isResctriction = false, bool isSubval = false)
{
if (!(t is Pullenti.Ner.TextToken))
{
return(null);
}
if (t.IsTableControlChar)
{
return(null);
}
Pullenti.Ner.Token t0 = t;
Pullenti.Ner.MetaToken whd = null;
int minmax = 0;
Pullenti.Ner.Token tt = NumbersWithUnitToken._isMinOrMax(t0, ref minmax);
if (tt != null)
{
t = tt.Next;
}
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t, Pullenti.Ner.Core.NounPhraseParseAttr.ParsePreposition | Pullenti.Ner.Core.NounPhraseParseAttr.IgnoreBrackets, 0, null);
if (npt == null)
{
whd = NumbersWithUnitToken._tryParseWHL(t);
if (whd != null)
{
npt = new Pullenti.Ner.Core.NounPhraseToken(t0, whd.EndToken);
}
else if (t0.IsValue("КПД", null))
{
npt = new Pullenti.Ner.Core.NounPhraseToken(t0, t0);
}
else if ((t0 is Pullenti.Ner.TextToken) && t0.LengthChar > 3 && t0.GetMorphClassInDictionary().IsUndefined)
{
npt = new Pullenti.Ner.Core.NounPhraseToken(t0, t0);
}
else if (t0.IsValue("T", null) && t0.Chars.IsAllLower)
{
npt = new Pullenti.Ner.Core.NounPhraseToken(t0, t0);
t = t0;
if (t.Next != null && t.Next.IsChar('='))
{
npt.EndToken = t.Next;
}
}
else if ((t0 is Pullenti.Ner.TextToken) && t0.Chars.IsLetter && isSubval)
{
if (NumbersWithUnitToken.TryParse(t, addUnits, false, false, false, false) != null)
{
return(null);
}
npt = new Pullenti.Ner.Core.NounPhraseToken(t0, t0);
for (t = t0.Next; t != null; t = t.Next)
{
if (t.WhitespacesBeforeCount > 2)
{
break;
}
else if (!(t is Pullenti.Ner.TextToken))
{
break;
}
else if (!t.Chars.IsLetter)
{
Pullenti.Ner.Core.BracketSequenceToken br = Pullenti.Ner.Core.BracketHelper.TryParse(t, Pullenti.Ner.Core.BracketParseAttr.No, 100);
if (br != null)
{
npt.EndToken = (t = br.EndToken);
}
else
{
break;
}
}
else if (NumbersWithUnitToken.TryParse(t, addUnits, false, false, false, false) != null)
{
break;
}
else
{
npt.EndToken = t;
}
}
}
else
{
return(null);
}
}
else if (Pullenti.Ner.Core.NumberHelper.TryParseRealNumber(t, true, false) != null)
{
return(null);
}
else
{
Pullenti.Ner.Date.Internal.DateItemToken dtok = Pullenti.Ner.Date.Internal.DateItemToken.TryAttach(t, null, false);
if (dtok != null)
{
return(null);
}
}
Pullenti.Ner.Token t1 = npt.EndToken;
t = npt.EndToken;
Pullenti.Ner.MetaToken name = new Pullenti.Ner.MetaToken(npt.BeginToken, npt.EndToken)
{
Morph = npt.Morph
};
List <UnitToken> units = null;
List <UnitToken> units2 = null;
List <MeasureToken> internals = new List <MeasureToken>();
bool not = false;
for (tt = t1.Next; tt != null; tt = tt.Next)
{
if (tt.IsNewlineBefore)
{
break;
}
if (tt.IsTableControlChar)
{
break;
}
Pullenti.Ner.Token tt2 = NumbersWithUnitToken._isMinOrMax(tt, ref minmax);
if (tt2 != null)
{
t1 = (t = (tt = tt2));
continue;
}
if ((tt.IsValue("БЫТЬ", null) || tt.IsValue("ДОЛЖЕН", null) || tt.IsValue("ДОЛЖНЫЙ", null)) || tt.IsValue("МОЖЕТ", null) || ((tt.IsValue("СОСТАВЛЯТЬ", null) && !tt.GetMorphClassInDictionary().IsAdjective)))
{
t1 = (t = tt);
if (tt.Previous.IsValue("НЕ", null))
{
not = true;
}
continue;
}
Pullenti.Ner.MetaToken www = NumbersWithUnitToken._tryParseWHL(tt);
if (www != null)
{
whd = www;
t1 = (t = (tt = www.EndToken));
continue;
}
if (tt.IsValue("ПРИ", null))
{
MeasureToken mt1 = TryParse(tt.Next, addUnits, false, false, true, false);
if (mt1 != null)
{
internals.Add(mt1);
t1 = (t = (tt = mt1.EndToken));
continue;
}
NumbersWithUnitToken n1 = NumbersWithUnitToken.TryParse(tt.Next, addUnits, false, false, false, false);
if (n1 != null && n1.Units.Count > 0)
{
mt1 = new MeasureToken(n1.BeginToken, n1.EndToken)
{
Nums = n1
};
internals.Add(mt1);
t1 = (t = (tt = mt1.EndToken));
continue;
}
}
if (tt.IsValue("ПО", null) && tt.Next != null && tt.Next.IsValue("U", null))
{
t1 = (t = (tt = tt.Next));
continue;
}
if (internals.Count > 0)
{
if (tt.IsChar(':'))
{
break;
}
MeasureToken mt1 = TryParse(tt.Next, addUnits, false, false, true, false);
if (mt1 != null && mt1.Reliable)
{
internals.Add(mt1);
t1 = (t = (tt = mt1.EndToken));
continue;
}
}
if ((tt is Pullenti.Ner.NumberToken) && (tt as Pullenti.Ner.NumberToken).Typ == Pullenti.Ner.NumberSpellingType.Words)
{
Pullenti.Ner.Core.NounPhraseToken npt3 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(tt, Pullenti.Ner.Core.NounPhraseParseAttr.ParseNumericAsAdjective, 0, null);
if (npt3 != null)
{
t1 = (tt = npt3.EndToken);
if (internals.Count == 0)
{
name.EndToken = t1;
}
continue;
}
}
if (((tt.IsHiphen && !tt.IsWhitespaceBefore && !tt.IsWhitespaceAfter) && (tt.Next is Pullenti.Ner.NumberToken) && (tt.Previous is Pullenti.Ner.TextToken)) && tt.Previous.Chars.IsAllUpper)
{
t1 = (tt = (t = tt.Next));
if (internals.Count == 0)
{
name.EndToken = t1;
}
continue;
}
if (((tt is Pullenti.Ner.NumberToken) && !tt.IsWhitespaceBefore && (tt.Previous is Pullenti.Ner.TextToken)) && tt.Previous.Chars.IsAllUpper)
{
t1 = (t = tt);
if (internals.Count == 0)
{
name.EndToken = t1;
}
continue;
}
if ((((tt is Pullenti.Ner.NumberToken) && !tt.IsWhitespaceAfter && tt.Next.IsHiphen) && !tt.Next.IsWhitespaceAfter && (tt.Next.Next is Pullenti.Ner.TextToken)) && tt.Next.Next.LengthChar > 2)
{
t1 = (t = (tt = tt.Next.Next));
Pullenti.Ner.Core.NounPhraseToken npt1 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(tt, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt1 != null && npt1.EndChar > tt.EndChar)
{
t1 = (t = (tt = npt1.EndToken));
}
if (internals.Count == 0)
{
name.EndToken = t1;
}
continue;
}
if ((tt is Pullenti.Ner.NumberToken) && tt.Previous != null)
{
if (tt.Previous.IsValue("USB", null))
{
t1 = (t = tt);
if (internals.Count == 0)
{
name.EndToken = t1;
}
for (Pullenti.Ner.Token ttt = tt.Next; ttt != null; ttt = ttt.Next)
{
if (ttt.IsWhitespaceBefore)
{
break;
}
if (ttt.IsCharOf(",:"))
{
break;
}
t1 = (t = (tt = ttt));
if (internals.Count == 0)
{
name.EndToken = t1;
}
}
continue;
}
}
NumbersWithUnitToken mt0 = NumbersWithUnitToken.TryParse(tt, addUnits, false, false, false, false);
if (mt0 != null)
{
Pullenti.Ner.Core.NounPhraseToken npt1 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(tt, Pullenti.Ner.Core.NounPhraseParseAttr.ParseNumericAsAdjective | Pullenti.Ner.Core.NounPhraseParseAttr.ParsePreposition, 0, null);
if (npt1 != null && npt1.EndChar > mt0.EndChar)
{
t1 = (t = (tt = npt1.EndToken));
if (internals.Count == 0)
{
name.EndToken = t1;
}
continue;
}
break;
}
if (((tt.IsComma || tt.IsChar('('))) && tt.Next != null)
{
www = NumbersWithUnitToken._tryParseWHL(tt.Next);
if (www != null)
{
whd = www;
t1 = (t = (tt = www.EndToken));
if (tt.Next != null && tt.Next.IsComma)
{
t1 = (tt = tt.Next);
}
if (tt.Next != null && tt.Next.IsChar(')'))
{
t1 = (tt = tt.Next);
continue;
}
}
List <UnitToken> uu = UnitToken.TryParseList(tt.Next, addUnits, false);
if (uu != null)
{
t1 = (t = uu[uu.Count - 1].EndToken);
units = uu;
if (tt.IsChar('(') && t1.Next != null && t1.Next.IsChar(')'))
{
t1 = (t = (tt = t1.Next));
continue;
}
else if (t1.Next != null && t1.Next.IsChar('('))
{
uu = UnitToken.TryParseList(t1.Next.Next, addUnits, false);
if (uu != null && uu[uu.Count - 1].EndToken.Next != null && uu[uu.Count - 1].EndToken.Next.IsChar(')'))
{
units2 = uu;
t1 = (t = (tt = uu[uu.Count - 1].EndToken.Next));
continue;
}
www = NumbersWithUnitToken._tryParseWHL(t1.Next);
if (www != null)
{
whd = www;
t1 = (t = (tt = www.EndToken));
continue;
}
}
if (uu != null && uu.Count > 0 && !uu[0].IsDoubt)
{
break;
}
if (t1.Next != null)
{
if (t1.Next.IsTableControlChar || t1.IsNewlineAfter)
{
break;
}
}
units = null;
}
}
if (Pullenti.Ner.Core.BracketHelper.CanBeStartOfSequence(tt, false, false) && !(tt.Next is Pullenti.Ner.NumberToken))
{
Pullenti.Ner.Core.BracketSequenceToken br = Pullenti.Ner.Core.BracketHelper.TryParse(tt, Pullenti.Ner.Core.BracketParseAttr.No, 100);
if (br != null)
{
t1 = (t = (tt = br.EndToken));
continue;
}
}
if (tt.IsValue("НЕ", null) && tt.Next != null)
{
Pullenti.Morph.MorphClass mc = tt.Next.GetMorphClassInDictionary();
if (mc.IsAdverb || mc.IsMisc)
{
break;
}
continue;
}
if (tt.IsValue("ЯМЗ", null))
{
}
Pullenti.Ner.Core.NounPhraseToken npt2 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(tt, Pullenti.Ner.Core.NounPhraseParseAttr.ParsePreposition | Pullenti.Ner.Core.NounPhraseParseAttr.IgnoreBrackets | Pullenti.Ner.Core.NounPhraseParseAttr.ParsePronouns, 0, null);
if (npt2 == null)
{
if (tt.Morph.Class.IsPreposition || tt.Morph.Class.IsConjunction)
{
Pullenti.Ner.Core.TerminToken to = NumbersWithUnitToken.m_Termins.TryParse(tt, Pullenti.Ner.Core.TerminParseAttr.No);
if (to != null)
{
if ((to.EndToken.Next is Pullenti.Ner.TextToken) && to.EndToken.Next.IsLetters)
{
}
else
{
break;
}
}
t1 = tt;
continue;
}
Pullenti.Morph.MorphClass mc = tt.GetMorphClassInDictionary();
if (((tt is Pullenti.Ner.TextToken) && tt.Chars.IsLetter && tt.LengthChar > 1) && (((tt.Chars.IsAllUpper || mc.IsAdverb || mc.IsUndefined) || mc.IsAdjective)))
{
List <UnitToken> uu = UnitToken.TryParseList(tt, addUnits, false);
if (uu != null)
{
if (uu[0].LengthChar > 1 || uu.Count > 1)
{
units = uu;
t1 = (t = uu[uu.Count - 1].EndToken);
break;
}
}
t1 = (t = tt);
if (internals.Count == 0)
{
name.EndToken = tt;
}
continue;
}
if (tt.IsComma)
{
continue;
}
if (tt.IsChar('.'))
{
if (!Pullenti.Ner.Core.MiscHelper.CanBeStartOfSentence(tt.Next))
{
continue;
}
List <UnitToken> uu = UnitToken.TryParseList(tt.Next, addUnits, false);
if (uu != null)
{
if (uu[0].LengthChar > 2 || uu.Count > 1)
{
units = uu;
t1 = (t = uu[uu.Count - 1].EndToken);
break;
}
}
}
break;
}
t1 = (t = (tt = npt2.EndToken));
if (internals.Count > 0)
{
}
else if (t.IsValue("ПРЕДЕЛ", null) || t.IsValue("ГРАНИЦА", null) || t.IsValue("ДИАПАЗОН", null))
{
}
else if (t.Chars.IsLetter)
{
name.EndToken = t1;
}
}
Pullenti.Ner.Token t11 = t1;
for (t1 = t1.Next; t1 != null; t1 = t1.Next)
{
if (t1.IsTableControlChar)
{
}
else if (t1.IsCharOf(":,_"))
{
if (isResctriction)
{
return(null);
}
Pullenti.Ner.MetaToken www = NumbersWithUnitToken._tryParseWHL(t1.Next);
if (www != null)
{
whd = www;
t1 = (t = www.EndToken);
continue;
}
List <UnitToken> uu = UnitToken.TryParseList(t1.Next, addUnits, false);
if (uu != null)
{
if (uu[0].LengthChar > 1 || uu.Count > 1)
{
units = uu;
t1 = (t = uu[uu.Count - 1].EndToken);
continue;
}
}
if (t1.IsChar(':'))
{
List <MeasureToken> li = new List <MeasureToken>();
for (Pullenti.Ner.Token ttt = t1.Next; ttt != null; ttt = ttt.Next)
{
if (ttt.IsHiphen || ttt.IsTableControlChar)
{
continue;
}
if ((ttt is Pullenti.Ner.TextToken) && !ttt.Chars.IsLetter)
{
continue;
}
MeasureToken mt1 = TryParse(ttt, addUnits, true, true, false, true);
if (mt1 == null)
{
break;
}
li.Add(mt1);
ttt = mt1.EndToken;
if (ttt.Next != null && ttt.Next.IsChar(';'))
{
ttt = ttt.Next;
}
if (ttt.IsChar(';'))
{
}
else if (ttt.IsNewlineAfter && mt1.IsNewlineBefore)
{
}
else
{
break;
}
}
if (li.Count > 1)
{
MeasureToken res0 = new MeasureToken(t0, li[li.Count - 1].EndToken)
{
Internals = li, IsEmpty = true
};
if (internals != null && internals.Count > 0)
{
res0.InternalEx = internals[0];
}
string nam = Pullenti.Ner.Core.MiscHelper.GetTextValueOfMetaToken(name, Pullenti.Ner.Core.GetTextAttr.FirstNounGroupToNominative);
li[0].BeginToken = t0;
foreach (MeasureToken v in li)
{
v.Name = string.Format("{0} ({1})", nam, v.Name ?? "").Trim();
if (v.Nums != null && v.Nums.Units.Count == 0 && units != null)
{
v.Nums.Units = units;
}
}
return(res0);
}
}
}
else if (t1.IsHiphen && t1.IsWhitespaceAfter && t1.IsWhitespaceBefore)
{
}
else if (t1.IsHiphen && t1.Next != null && t1.Next.IsChar('('))
{
}
else
{
break;
}
}
if (t1 == null)
{
return(null);
}
List <NumbersWithUnitToken> mts = NumbersWithUnitToken.TryParseMulti(t1, addUnits, false, not, true, isResctriction);
if (mts == null)
{
if (units != null && units.Count > 0)
{
if (t1 == null || t1.Previous.IsChar(':'))
{
mts = new List <NumbersWithUnitToken>();
if (t1 == null)
{
for (t1 = t11; t1 != null && t1.Next != null; t1 = t1.Next)
{
}
}
else
{
t1 = t1.Previous;
}
mts.Add(new NumbersWithUnitToken(t0, t1)
{
SingleVal = double.NaN
});
}
}
if (mts == null)
{
return(null);
}
}
NumbersWithUnitToken mt = mts[0];
if (mt.BeginToken == mt.EndToken && !(mt.BeginToken is Pullenti.Ner.NumberToken))
{
return(null);
}
if (!isSubval && name.BeginToken.Morph.Class.IsPreposition)
{
name.BeginToken = name.BeginToken.Next;
}
if (mt.WHL != null)
{
whd = mt.WHL;
}
for (int kk = 0; kk < 10; kk++)
{
if (whd != null && whd.EndToken == name.EndToken)
{
name.EndToken = whd.BeginToken.Previous;
continue;
}
if (units != null)
{
if (units[units.Count - 1].EndToken == name.EndToken)
{
name.EndToken = units[0].BeginToken.Previous;
continue;
}
}
break;
}
if (mts.Count > 1 && internals.Count == 0)
{
if (mt.Units.Count == 0)
{
if (units != null)
{
foreach (NumbersWithUnitToken m in mts)
{
m.Units = units;
}
}
}
MeasureToken res1 = new MeasureToken(t0, mts[mts.Count - 1].EndToken)
{
Morph = name.Morph, Reliable = true
};
res1.Name = Pullenti.Ner.Core.MiscHelper.GetTextValueOfMetaToken(name, Pullenti.Ner.Core.GetTextAttr.FirstNounGroupToNominative);
for (int k = 0; k < mts.Count; k++)
{
MeasureToken ttt = new MeasureToken(mts[k].BeginToken, mts[k].EndToken)
{
Nums = mts[k]
};
if (whd != null)
{
List <string> nams = whd.Tag as List <string>;
if (k < nams.Count)
{
ttt.Name = nams[k];
}
}
res1.Internals.Add(ttt);
}
Pullenti.Ner.Token tt1 = res1.EndToken.Next;
if (tt1 != null && tt1.IsChar('±'))
{
NumbersWithUnitToken nn = NumbersWithUnitToken._tryParse(tt1, addUnits, true, false, false);
if (nn != null && nn.PlusMinusPercent)
{
res1.EndToken = nn.EndToken;
res1.Nums = nn;
if (nn.Units.Count > 0 && units == null && mt.Units.Count == 0)
{
foreach (NumbersWithUnitToken m in mts)
{
m.Units = nn.Units;
}
}
}
}
return(res1);
}
if (!mt.IsWhitespaceBefore)
{
if (mt.BeginToken.Previous == null)
{
return(null);
}
if (mt.BeginToken.Previous.IsCharOf(":),") || mt.BeginToken.Previous.IsTableControlChar || mt.BeginToken.Previous.IsValue("IP", null))
{
}
else if (mt.BeginToken.IsHiphen && mt.Units.Count > 0 && !mt.Units[0].IsDoubt)
{
}
else
{
return(null);
}
}
if (mt.Units.Count == 0 && units != null)
{
mt.Units = units;
if (mt.DivNum != null && units.Count > 1 && mt.DivNum.Units.Count == 0)
{
for (int i = 1; i < units.Count; i++)
{
if (units[i].Pow == -1)
{
for (int j = i; j < units.Count; j++)
{
mt.DivNum.Units.Add(units[j]);
units[j].Pow = -units[j].Pow;
}
mt.Units.RemoveRange(i, units.Count - i);
break;
}
}
}
}
if ((minmax < 0) && mt.SingleVal != null)
{
mt.FromVal = mt.SingleVal;
mt.FromInclude = true;
mt.SingleVal = null;
}
if (minmax > 0 && mt.SingleVal != null)
{
mt.ToVal = mt.SingleVal;
mt.ToInclude = true;
mt.SingleVal = null;
}
if (mt.Units.Count == 0)
{
units = UnitToken.TryParseList(mt.EndToken.Next, addUnits, true);
if (units == null)
{
if (canUnitsAbsent)
{
}
else
{
return(null);
}
}
else
{
mt.Units = units;
}
}
MeasureToken res = new MeasureToken(t0, mt.EndToken)
{
Morph = name.Morph, Internals = internals
};
if (((!t0.IsWhitespaceBefore && t0.Previous != null && t0 == name.BeginToken) && t0.Previous.IsHiphen && !t0.Previous.IsWhitespaceBefore) && (t0.Previous.Previous is Pullenti.Ner.TextToken))
{
name.BeginToken = (res.BeginToken = name.BeginToken.Previous.Previous);
}
res.Name = Pullenti.Ner.Core.MiscHelper.GetTextValueOfMetaToken(name, (!isSubval ? Pullenti.Ner.Core.GetTextAttr.FirstNounGroupToNominative : Pullenti.Ner.Core.GetTextAttr.No));
res.Nums = mt;
foreach (UnitToken u in res.Nums.Units)
{
if (u.Keyword != null)
{
if (u.Keyword.BeginChar >= res.BeginChar)
{
res.Reliable = true;
}
}
}
res._parseInternals(addUnits);
if (res.Internals.Count > 0 || !canBeSet)
{
return(res);
}
t1 = res.EndToken.Next;
if (t1 != null && t1.IsCommaAnd)
{
t1 = t1.Next;
}
List <NumbersWithUnitToken> mts1 = NumbersWithUnitToken.TryParseMulti(t1, addUnits, false, false, false, false);
if ((mts1 != null && mts1.Count == 1 && (t1.WhitespacesBeforeCount < 3)) && mts1[0].Units.Count > 0 && !UnitToken.CanBeEquals(mts[0].Units, mts1[0].Units))
{
res.IsSet = true;
res.Nums = null;
res.Internals.Add(new MeasureToken(mt.BeginToken, mt.EndToken)
{
Nums = mt
});
res.Internals.Add(new MeasureToken(mts1[0].BeginToken, mts1[0].EndToken)
{
Nums = mts1[0]
});
res.EndToken = mts1[0].EndToken;
}
return(res);
}
/// <summary>
/// Попробовать создать именную группу с указанного токена
/// </summary>
/// <param name="t">начальный токен</param>
/// <param name="attrs">атрибуты (можно битовую маску)</param>
/// <param name="maxCharPos">максимальная позиция в тексте, до которой выделять (если 0, то без ограничений)</param>
/// <param name="noun">это если нужно выделить только прилагательные для ранее выделенного существительного (из другой группы)</param>
/// <return>именная группа или null</return>
public static NounPhraseToken TryParse(Pullenti.Ner.Token t, NounPhraseParseAttr attrs = NounPhraseParseAttr.No, int maxCharPos = 0, Pullenti.Ner.MetaToken noun = null)
{
NounPhraseToken res = _NounPraseHelperInt.TryParse(t, attrs, maxCharPos, noun as Pullenti.Ner.Core.Internal.NounPhraseItem);
if (res != null)
{
if (((attrs & NounPhraseParseAttr.ParsePreposition)) != NounPhraseParseAttr.No)
{
if (res.BeginToken == res.EndToken && t.Morph.Class.IsPreposition)
{
PrepositionToken prep = PrepositionHelper.TryParse(t);
if (prep != null)
{
NounPhraseToken res2 = _NounPraseHelperInt.TryParse(t.Next, attrs, maxCharPos, noun as Pullenti.Ner.Core.Internal.NounPhraseItem);
if (res2 != null)
{
if (!((prep.NextCase & res2.Morph.Case)).IsUndefined)
{
res2.Morph.RemoveItems(prep.NextCase);
res2.Preposition = prep;
res2.BeginToken = t;
return(res2);
}
}
}
}
}
return(res);
}
if (((attrs & NounPhraseParseAttr.ParsePreposition)) != NounPhraseParseAttr.No)
{
PrepositionToken prep = PrepositionHelper.TryParse(t);
if (prep != null && (prep.NewlinesAfterCount < 2))
{
res = _NounPraseHelperInt.TryParse(prep.EndToken.Next, attrs, maxCharPos, noun as Pullenti.Ner.Core.Internal.NounPhraseItem);
if (res != null)
{
res.Preposition = prep;
res.BeginToken = t;
if (!((prep.NextCase & res.Morph.Case)).IsUndefined)
{
res.Morph.RemoveItems(prep.NextCase);
}
else if (t.Morph.Class.IsAdverb)
{
return(null);
}
return(res);
}
}
}
return(null);
}
internal static NumbersWithUnitToken _tryParse(Pullenti.Ner.Token t, Pullenti.Ner.Core.TerminCollection addUnits, bool second, bool canOmitNumber, bool canBeNan)
{
if (t == null)
{
return(null);
}
while (t != null)
{
if (t.IsCommaAnd || t.IsValue("НО", null))
{
t = t.Next;
}
else
{
break;
}
}
Pullenti.Ner.Token t0 = t;
bool about = false;
bool hasKeyw = false;
bool isDiapKeyw = false;
int minMax = 0;
Pullenti.Ner.Token ttt = _isMinOrMax(t, ref minMax);
if (ttt != null)
{
t = ttt.Next;
if (t == null)
{
return(null);
}
}
if (t == null)
{
return(null);
}
if (t.IsChar('~') || t.IsValue("ОКОЛО", null) || t.IsValue("ПРИМЕРНО", null))
{
t = t.Next;
about = true;
hasKeyw = true;
if (t == null)
{
return(null);
}
}
if (t.IsValue("В", null) && t.Next != null)
{
if (t.Next.IsValue("ПРЕДЕЛ", null) || t.IsValue("ДИАПАЗОН", null))
{
t = t.Next.Next;
if (t == null)
{
return(null);
}
isDiapKeyw = true;
}
}
if (t0.IsChar('('))
{
NumbersWithUnitToken mt0 = _tryParse(t.Next, addUnits, false, false, false);
if (mt0 != null && mt0.EndToken.Next != null && mt0.EndToken.Next.IsChar(')'))
{
if (second)
{
if (mt0.FromVal != null && mt0.ToVal != null && mt0.FromVal.Value == (-mt0.ToVal.Value))
{
}
else
{
return(null);
}
}
mt0.BeginToken = t0;
mt0.EndToken = mt0.EndToken.Next;
List <UnitToken> uu = UnitToken.TryParseList(mt0.EndToken.Next, addUnits, false);
if (uu != null && mt0.Units.Count == 0)
{
mt0.Units = uu;
mt0.EndToken = uu[uu.Count - 1].EndToken;
}
return(mt0);
}
}
bool plusminus = false;
bool unitBefore = false;
bool isAge = false;
DiapTyp dty = DiapTyp.Undefined;
Pullenti.Ner.MetaToken whd = null;
List <UnitToken> uni = null;
Pullenti.Ner.Core.TerminToken tok = (m_Termins == null ? null : m_Termins.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No));
if (tok != null)
{
if (tok.EndToken.IsValue("СТАРШЕ", null) || tok.EndToken.IsValue("МЛАДШЕ", null))
{
isAge = true;
}
t = tok.EndToken.Next;
dty = (DiapTyp)tok.Termin.Tag;
hasKeyw = true;
if (!tok.IsWhitespaceAfter)
{
if (t == null)
{
return(null);
}
if (t is Pullenti.Ner.NumberToken)
{
if (tok.BeginToken == tok.EndToken && !tok.Chars.IsAllLower)
{
return(null);
}
}
else if (t.IsComma && t.Next != null && t.Next.IsValue("ЧЕМ", null))
{
t = t.Next.Next;
if (t != null && t.Morph.Class.IsPreposition)
{
t = t.Next;
}
}
else if (t.IsCharOf(":,(") || t.IsTableControlChar)
{
}
else
{
return(null);
}
}
if (t != null && t.IsChar('('))
{
uni = UnitToken.TryParseList(t.Next, addUnits, false);
if (uni != null)
{
t = uni[uni.Count - 1].EndToken.Next;
while (t != null)
{
if (t.IsCharOf("):"))
{
t = t.Next;
}
else
{
break;
}
}
NumbersWithUnitToken mt0 = _tryParse(t, addUnits, false, canOmitNumber, false);
if (mt0 != null && mt0.Units.Count == 0)
{
mt0.BeginToken = t0;
mt0.Units = uni;
return(mt0);
}
}
whd = _tryParseWHL(t);
if (whd != null)
{
t = whd.EndToken.Next;
}
}
else if (t != null && t.IsValue("IP", null))
{
uni = UnitToken.TryParseList(t, addUnits, false);
if (uni != null)
{
t = uni[uni.Count - 1].EndToken.Next;
}
}
if ((t != null && t.IsHiphen && t.IsWhitespaceBefore) && t.IsWhitespaceAfter)
{
t = t.Next;
}
}
else if (t.IsChar('<'))
{
dty = DiapTyp.Ls;
t = t.Next;
hasKeyw = true;
if (t != null && t.IsChar('='))
{
t = t.Next;
dty = DiapTyp.Le;
}
}
else if (t.IsChar('>'))
{
dty = DiapTyp.Gt;
t = t.Next;
hasKeyw = true;
if (t != null && t.IsChar('='))
{
t = t.Next;
dty = DiapTyp.Ge;
}
}
else if (t.IsChar('≤'))
{
dty = DiapTyp.Le;
hasKeyw = true;
t = t.Next;
}
else if (t.IsChar('≥'))
{
dty = DiapTyp.Ge;
hasKeyw = true;
t = t.Next;
}
else if (t.IsValue("IP", null))
{
uni = UnitToken.TryParseList(t, addUnits, false);
if (uni != null)
{
t = uni[uni.Count - 1].EndToken.Next;
}
}
else if (t.IsValue("ЗА", null) && (t.Next is Pullenti.Ner.NumberToken))
{
dty = DiapTyp.Ge;
t = t.Next;
}
while (t != null && ((t.IsCharOf(":,") || t.IsValue("ЧЕМ", null) || t.IsTableControlChar)))
{
t = t.Next;
}
if (t != null)
{
if (t.IsChar('+') || t.IsValue("ПЛЮС", null))
{
t = t.Next;
if (t != null && !t.IsWhitespaceBefore)
{
if (t.IsHiphen)
{
t = t.Next;
plusminus = true;
}
else if ((t.IsCharOf("\\/") && t.Next != null && !t.IsNewlineAfter) && t.Next.IsHiphen)
{
t = t.Next.Next;
plusminus = true;
}
}
}
else if (second && (t.IsCharOf("\\/÷…~")))
{
t = t.Next;
}
else if ((t.IsHiphen && t == t0 && !second) && m_Termins.TryParse(t.Next, Pullenti.Ner.Core.TerminParseAttr.No) != null)
{
tok = m_Termins.TryParse(t.Next, Pullenti.Ner.Core.TerminParseAttr.No);
t = tok.EndToken.Next;
dty = (DiapTyp)tok.Termin.Tag;
}
else if (t.IsHiphen && t == t0 && ((t.IsWhitespaceAfter || second)))
{
t = t.Next;
}
else if (t.IsChar('±'))
{
t = t.Next;
plusminus = true;
hasKeyw = true;
}
else if ((second && t.IsChar('.') && t.Next != null) && t.Next.IsChar('.'))
{
t = t.Next.Next;
if (t != null && t.IsChar('.'))
{
t = t.Next;
}
}
}
Pullenti.Ner.NumberToken num = Pullenti.Ner.Core.NumberHelper.TryParseRealNumber(t, true, false);
if (num == null)
{
uni = UnitToken.TryParseList(t, addUnits, false);
if (uni != null)
{
unitBefore = true;
t = uni[uni.Count - 1].EndToken.Next;
bool delim = false;
while (t != null)
{
if (t.IsCharOf(":,"))
{
delim = true;
t = t.Next;
}
else if (t.IsHiphen && t.IsWhitespaceAfter)
{
delim = true;
t = t.Next;
}
else
{
break;
}
}
if (!delim)
{
if (t == null)
{
if (hasKeyw && canBeNan)
{
}
else
{
return(null);
}
}
else if (!t.IsWhitespaceBefore)
{
return(null);
}
if (t.Next != null && t.IsHiphen && t.IsWhitespaceAfter)
{
delim = true;
t = t.Next;
}
}
num = Pullenti.Ner.Core.NumberHelper.TryParseRealNumber(t, true, false);
}
}
NumbersWithUnitToken res = null;
double rval = (double)0;
if (num == null)
{
Pullenti.Ner.Core.TerminToken tt = m_Spec.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No);
if (tt != null)
{
rval = (double)tt.Termin.Tag;
string unam = (string)tt.Termin.Tag2;
foreach (Unit u in UnitsHelper.Units)
{
if (u.FullnameCyr == unam)
{
uni = new List <UnitToken>();
uni.Add(new UnitToken(t, t)
{
Unit = u
});
break;
}
}
if (uni == null)
{
return(null);
}
res = new NumbersWithUnitToken(t0, tt.EndToken)
{
About = about
};
t = tt.EndToken.Next;
}
else
{
if (!canOmitNumber && !hasKeyw && !canBeNan)
{
return(null);
}
if ((uni != null && uni.Count == 1 && uni[0].BeginToken == uni[0].EndToken) && uni[0].LengthChar > 3)
{
rval = 1;
res = new NumbersWithUnitToken(t0, uni[uni.Count - 1].EndToken)
{
About = about
};
t = res.EndToken.Next;
}
else if (hasKeyw && canBeNan)
{
rval = double.NaN;
res = new NumbersWithUnitToken(t0, t0)
{
About = about
};
if (t != null)
{
res.EndToken = t.Previous;
}
else
{
for (t = t0; t != null; t = t.Next)
{
res.EndToken = t;
}
}
}
else
{
return(null);
}
}
}
else
{
if ((t == t0 && t0.IsHiphen && !t.IsWhitespaceBefore) && !t.IsWhitespaceAfter && (num.RealValue < 0))
{
num = Pullenti.Ner.Core.NumberHelper.TryParseRealNumber(t.Next, true, false);
if (num == null)
{
return(null);
}
}
if (t == t0 && (t is Pullenti.Ner.NumberToken) && t.Morph.Class.IsAdjective)
{
Pullenti.Ner.TextToken nn = (t as Pullenti.Ner.NumberToken).EndToken as Pullenti.Ner.TextToken;
if (nn == null)
{
return(null);
}
string norm = nn.GetNormalCaseText(Pullenti.Morph.MorphClass.Adjective, Pullenti.Morph.MorphNumber.Singular, Pullenti.Morph.MorphGender.Undefined, false);
if ((norm.EndsWith("Ь") || norm == "ЧЕТЫРЕ" || norm == "ТРИ") || norm == "ДВА")
{
}
else
{
Pullenti.Morph.MorphWordForm mi = Pullenti.Morph.MorphologyService.GetWordBaseInfo("КОКО" + nn.Term, null, false, false);
if (mi.Class.IsAdjective)
{
return(null);
}
}
}
t = num.EndToken.Next;
res = new NumbersWithUnitToken(t0, num.EndToken)
{
About = about
};
rval = num.RealValue;
}
if (uni == null)
{
uni = UnitToken.TryParseList(t, addUnits, false);
if (uni != null)
{
if ((plusminus && second && uni.Count >= 1) && uni[0].Unit == UnitsHelper.uPercent)
{
res.EndToken = uni[0].EndToken;
res.PlusMinusPercent = true;
Pullenti.Ner.Token tt1 = uni[0].EndToken.Next;
uni = UnitToken.TryParseList(tt1, addUnits, false);
if (uni != null)
{
res.Units = uni;
res.EndToken = uni[uni.Count - 1].EndToken;
}
}
else
{
res.Units = uni;
res.EndToken = uni[uni.Count - 1].EndToken;
}
t = res.EndToken.Next;
}
}
else
{
res.Units = uni;
if (uni.Count > 1)
{
List <UnitToken> uni1 = UnitToken.TryParseList(t, addUnits, false);
if (((uni1 != null && uni1[0].Unit == uni[0].Unit && (uni1.Count < uni.Count)) && uni[uni1.Count].Pow == -1 && uni1[uni1.Count - 1].EndToken.Next != null) && uni1[uni1.Count - 1].EndToken.Next.IsCharOf("/\\"))
{
NumbersWithUnitToken num2 = _tryParse(uni1[uni1.Count - 1].EndToken.Next.Next, addUnits, false, false, false);
if (num2 != null && num2.Units != null && num2.Units[0].Unit == uni[uni1.Count].Unit)
{
res.Units = uni1;
res.DivNum = num2;
res.EndToken = num2.EndToken;
}
}
}
}
res.WHL = whd;
if (dty != DiapTyp.Undefined)
{
if (dty == DiapTyp.Ge || dty == DiapTyp.From)
{
res.FromInclude = true;
res.FromVal = rval;
}
else if (dty == DiapTyp.Gt)
{
res.FromInclude = false;
res.FromVal = rval;
}
else if (dty == DiapTyp.Le || dty == DiapTyp.To)
{
res.ToInclude = true;
res.ToVal = rval;
}
else if (dty == DiapTyp.Ls)
{
res.ToInclude = false;
res.ToVal = rval;
}
}
bool isSecondMax = false;
if (!second)
{
int iii = 0;
ttt = _isMinOrMax(t, ref iii);
if (ttt != null && iii > 0)
{
isSecondMax = true;
t = ttt.Next;
}
}
NumbersWithUnitToken next = (second || plusminus || ((t != null && ((t.IsTableControlChar || t.IsNewlineBefore)))) ? null : _tryParse(t, addUnits, true, false, canBeNan));
if (next != null && (t.Previous is Pullenti.Ner.NumberToken))
{
if (MeasureHelper.IsMultChar((t.Previous as Pullenti.Ner.NumberToken).EndToken))
{
next = null;
}
}
if (next != null && ((next.ToVal != null || next.SingleVal != null)) && next.FromVal == null)
{
if ((((next.BeginToken.IsChar('+') && next.SingleVal != null && !double.IsNaN(next.SingleVal.Value)) && next.EndToken.Next != null && next.EndToken.Next.IsCharOf("\\/")) && next.EndToken.Next.Next != null && next.EndToken.Next.Next.IsHiphen) && !hasKeyw && !double.IsNaN(rval))
{
NumbersWithUnitToken next2 = _tryParse(next.EndToken.Next.Next.Next, addUnits, true, false, false);
if (next2 != null && next2.SingleVal != null && !double.IsNaN(next2.SingleVal.Value))
{
res.FromVal = rval - next2.SingleVal.Value;
res.FromInclude = true;
res.ToVal = rval + next.SingleVal.Value;
res.ToInclude = true;
if (next2.Units != null && res.Units.Count == 0)
{
res.Units = next2.Units;
}
res.EndToken = next2.EndToken;
return(res);
}
}
if (next.Units.Count > 0)
{
if (res.Units.Count == 0)
{
res.Units = next.Units;
}
else if (!UnitToken.CanBeEquals(res.Units, next.Units))
{
next = null;
}
}
else if (res.Units.Count > 0 && !unitBefore && !next.PlusMinusPercent)
{
next = null;
}
if (next != null)
{
res.EndToken = next.EndToken;
}
if (next != null && next.ToVal != null)
{
res.ToVal = next.ToVal;
res.ToInclude = next.ToInclude;
}
else if (next != null && next.SingleVal != null)
{
if (next.BeginToken.IsCharOf("/\\"))
{
res.DivNum = next;
res.SingleVal = rval;
return(res);
}
else if (next.PlusMinusPercent)
{
res.SingleVal = rval;
res.PlusMinus = next.SingleVal;
res.PlusMinusPercent = true;
res.ToInclude = true;
}
else
{
res.ToVal = next.SingleVal;
res.ToInclude = true;
}
}
if (next != null)
{
if (res.FromVal == null)
{
res.FromVal = rval;
res.FromInclude = true;
}
return(res);
}
}
else if ((next != null && next.FromVal != null && next.ToVal != null) && next.ToVal.Value == (-next.FromVal.Value))
{
if (next.Units.Count == 1 && next.Units[0].Unit == UnitsHelper.uPercent && res.Units.Count > 0)
{
res.SingleVal = rval;
res.PlusMinus = next.ToVal.Value;
res.PlusMinusPercent = true;
res.EndToken = next.EndToken;
return(res);
}
if (next.Units.Count == 0)
{
res.SingleVal = rval;
res.PlusMinus = next.ToVal.Value;
res.EndToken = next.EndToken;
return(res);
}
res.FromVal = next.FromVal + rval;
res.FromInclude = true;
res.ToVal = next.ToVal + rval;
res.ToInclude = true;
res.EndToken = next.EndToken;
if (next.Units.Count > 0)
{
res.Units = next.Units;
}
return(res);
}
if (dty == DiapTyp.Undefined)
{
if (plusminus && ((!res.PlusMinusPercent || !second)))
{
res.FromInclude = true;
res.FromVal = -rval;
res.ToInclude = true;
res.ToVal = rval;
}
else
{
res.SingleVal = rval;
res.PlusMinusPercent = plusminus;
}
}
if (isAge)
{
res.IsAge = true;
}
return(res);
}
public static List <NumbersWithUnitToken> TryParseMulti(Pullenti.Ner.Token t, Pullenti.Ner.Core.TerminCollection addUnits, bool canOmitNumber = false, bool not = false, bool canBeNon = false, bool isResctriction = false)
{
if (t == null || (t is Pullenti.Ner.ReferentToken))
{
return(null);
}
Pullenti.Ner.Token tt0 = t;
if (tt0.IsChar('('))
{
Pullenti.Ner.MetaToken whd = _tryParseWHL(tt0);
if (whd != null)
{
tt0 = whd.EndToken;
}
List <NumbersWithUnitToken> res0 = TryParseMulti(tt0.Next, addUnits, false, canOmitNumber, canBeNon, false);
if (res0 != null)
{
res0[0].WHL = whd;
Pullenti.Ner.Token tt2 = res0[res0.Count - 1].EndToken.Next;
if (tt2 != null && tt2.IsCharOf(","))
{
tt2 = tt2.Next;
}
if (whd != null)
{
return(res0);
}
if (tt2 != null && tt2.IsChar(')'))
{
res0[res0.Count - 1].EndToken = tt2;
return(res0);
}
}
}
NumbersWithUnitToken mt = TryParse(t, addUnits, canOmitNumber, not, canBeNon, isResctriction);
if (mt == null)
{
return(null);
}
List <NumbersWithUnitToken> res = new List <NumbersWithUnitToken>();
Pullenti.Ner.Token nnn = null;
if (mt.WhitespacesAfterCount < 2)
{
if (MeasureHelper.IsMultChar(mt.EndToken.Next))
{
nnn = mt.EndToken.Next.Next;
}
else if ((mt.EndToken is Pullenti.Ner.NumberToken) && MeasureHelper.IsMultChar((mt.EndToken as Pullenti.Ner.NumberToken).EndToken))
{
nnn = mt.EndToken.Next;
}
}
if (nnn != null)
{
NumbersWithUnitToken mt2 = NumbersWithUnitToken.TryParse(nnn, addUnits, not, false, false, false);
if (mt2 != null)
{
NumbersWithUnitToken mt3 = null;
nnn = null;
if (mt2.WhitespacesAfterCount < 2)
{
if (MeasureHelper.IsMultChar(mt2.EndToken.Next))
{
nnn = mt2.EndToken.Next.Next;
}
else if ((mt2.EndToken is Pullenti.Ner.NumberToken) && MeasureHelper.IsMultChar((mt2.EndToken as Pullenti.Ner.NumberToken).EndToken))
{
nnn = mt2.EndToken.Next;
}
}
if (nnn != null)
{
mt3 = NumbersWithUnitToken.TryParse(nnn, addUnits, false, false, false, false);
}
if (mt3 == null)
{
Pullenti.Ner.Token tt2 = mt2.EndToken.Next;
if (tt2 != null && !tt2.IsWhitespaceBefore)
{
if (!tt2.IsCharOf(",.;"))
{
return(null);
}
}
}
if (mt3 != null && mt3.Units.Count > 0)
{
if (mt2.Units.Count == 0)
{
mt2.Units = mt3.Units;
}
}
res.Add(mt);
if (mt2 != null)
{
if (mt2.Units.Count > 0 && mt.Units.Count == 0)
{
mt.Units = mt2.Units;
}
res.Add(mt2);
if (mt3 != null)
{
res.Add(mt3);
}
}
return(res);
}
}
if ((!mt.IsWhitespaceAfter && MeasureHelper.IsMultCharEnd(mt.EndToken.Next) && (mt.EndToken.Next.Next is Pullenti.Ner.NumberToken)) && mt.Units.Count == 0)
{
string utxt = (mt.EndToken.Next as Pullenti.Ner.TextToken).Term;
utxt = utxt.Substring(0, utxt.Length - 1);
List <Pullenti.Ner.Core.Termin> terms = UnitsHelper.Termins.FindTerminsByString(utxt, null);
if (terms != null && terms.Count > 0)
{
mt.Units.Add(new UnitToken(mt.EndToken.Next, mt.EndToken.Next)
{
Unit = terms[0].Tag as Unit
});
mt.EndToken = mt.EndToken.Next;
List <NumbersWithUnitToken> res1 = TryParseMulti(mt.EndToken.Next, addUnits, false, false, false, false);
if (res1 != null)
{
res1.Insert(0, mt);
return(res1);
}
}
}
res.Add(mt);
return(res);
}
public static Pullenti.Ner.MetaToken _tryParseWHL(Pullenti.Ner.Token t)
{
if (!(t is Pullenti.Ner.TextToken))
{
return(null);
}
if (t.IsCharOf(":-"))
{
Pullenti.Ner.MetaToken re0 = _tryParseWHL(t.Next);
if (re0 != null)
{
return(re0);
}
}
if (t.IsCharOf("("))
{
Pullenti.Ner.MetaToken re0 = _tryParseWHL(t.Next);
if (re0 != null)
{
if (re0.EndToken.Next != null && re0.EndToken.Next.IsChar(')'))
{
re0.BeginToken = t;
re0.EndToken = re0.EndToken.Next;
return(re0);
}
}
}
string txt = (t as Pullenti.Ner.TextToken).Term;
List <string> nams = null;
if (txt.Length == 5 && ((txt[1] == 'Х' || txt[1] == 'X')) && ((txt[3] == 'Х' || txt[3] == 'X')))
{
nams = new List <string>();
for (int i = 0; i < 3; i++)
{
char ch = txt[i * 2];
if (ch == 'Г')
{
nams.Add("ГЛУБИНА");
}
else if (ch == 'В' || ch == 'H' || ch == 'Н')
{
nams.Add("ВЫСОТА");
}
else if (ch == 'Ш' || ch == 'B' || ch == 'W')
{
nams.Add("ШИРИНА");
}
else if (ch == 'Д' || ch == 'L')
{
nams.Add("ДЛИНА");
}
else if (ch == 'D')
{
nams.Add("ДИАМЕТР");
}
else
{
return(null);
}
}
return(new Pullenti.Ner.MetaToken(t, t)
{
Tag = nams
});
}
Pullenti.Ner.Token t0 = t;
Pullenti.Ner.Token t1 = t;
for (; t != null; t = t.Next)
{
if (!(t is Pullenti.Ner.TextToken) || ((t.WhitespacesBeforeCount > 1 && t != t0)))
{
break;
}
string term = (t as Pullenti.Ner.TextToken).Term;
if (term.EndsWith("X") || term.EndsWith("Х"))
{
term = term.Substring(0, term.Length - 1);
}
string nam = null;
if (((t.IsValue("ДЛИНА", null) || t.IsValue("ДЛИННА", null) || term == "Д") || term == "ДЛ" || term == "ДЛИН") || term == "L")
{
nam = "ДЛИНА";
}
else if (((t.IsValue("ШИРИНА", null) || t.IsValue("ШИРОТА", null) || term == "Ш") || term == "ШИР" || term == "ШИРИН") || term == "W" || term == "B")
{
nam = "ШИРИНА";
}
else if ((t.IsValue("ГЛУБИНА", null) || term == "Г" || term == "ГЛ") || term == "ГЛУБ")
{
nam = "ГЛУБИНА";
}
else if ((t.IsValue("ВЫСОТА", null) || term == "В" || term == "ВЫС") || term == "H" || term == "Н")
{
nam = "ВЫСОТА";
}
else if (t.IsValue("ДИАМЕТР", null) || term == "D" || term == "ДИАМ")
{
nam = "ДИАМЕТР";
}
else
{
break;
}
if (nams == null)
{
nams = new List <string>();
}
nams.Add(nam);
t1 = t;
if (t.Next != null && t.Next.IsChar('.'))
{
t1 = (t = t.Next);
}
if (t.Next == null)
{
break;
}
if (MeasureHelper.IsMultChar(t.Next) || t.Next.IsComma || t.Next.IsCharOf("\\/"))
{
t = t.Next;
}
}
if (nams == null || (nams.Count < 2))
{
return(null);
}
return(new Pullenti.Ner.MetaToken(t0, t1)
{
Tag = nams
});
}
/// <summary>
/// Попробовать восстановить последовательность, обрамляемую кавычками или скобками. Поддерживается
/// вложенность, возможность отсутствия закрывающего элемента и др.
/// </summary>
/// <param name="t">начальный токен</param>
/// <param name="attrs">параметры выделения</param>
/// <param name="maxTokens">максимально токенов (вдруг забыли закрывающую кавычку)</param>
/// <return>метатокен BracketSequenceToken</return>
public static BracketSequenceToken TryParse(Pullenti.Ner.Token t, BracketParseAttr attrs = BracketParseAttr.No, int maxTokens = 100)
{
Pullenti.Ner.Token t0 = t;
int cou = 0;
if (!CanBeStartOfSequence(t0, false, false))
{
return(null);
}
List <Bracket> brList = new List <Bracket>();
brList.Add(new Bracket(t0));
cou = 0;
int crlf = 0;
Pullenti.Ner.Token last = null;
int lev = 1;
bool isAssim = brList[0].Char != '«' && m_AssymOPenChars.IndexOf(brList[0].Char) >= 0;
bool genCase = false;
for (t = t0.Next; t != null; t = t.Next)
{
if (t.IsTableControlChar)
{
break;
}
last = t;
if (t.IsCharOf(m_OpenChars) || t.IsCharOf(m_CloseChars))
{
if (t.IsNewlineBefore && ((attrs & BracketParseAttr.CanBeManyLines)) == BracketParseAttr.No)
{
if (t.WhitespacesBeforeCount > 10 || CanBeStartOfSequence(t, false, false))
{
if (t.IsChar('(') && !t0.IsChar('('))
{
}
else
{
last = t.Previous;
break;
}
}
}
Bracket bb = new Bracket(t);
brList.Add(bb);
if (brList.Count > 20)
{
break;
}
if ((brList.Count == 3 && brList[1].CanBeOpen && bb.CanBeClose) && MustBeCloseChar(bb.Char, brList[1].Char) && MustBeCloseChar(bb.Char, brList[0].Char))
{
bool ok = false;
for (Pullenti.Ner.Token tt = t.Next; tt != null; tt = tt.Next)
{
if (tt.IsNewlineBefore)
{
break;
}
if (tt.IsChar(','))
{
break;
}
if (tt.IsChar('.'))
{
for (tt = tt.Next; tt != null; tt = tt.Next)
{
if (tt.IsNewlineBefore)
{
break;
}
else if (tt.IsCharOf(m_OpenChars) || tt.IsCharOf(m_CloseChars))
{
Bracket bb2 = new Bracket(tt);
if (BracketHelper.CanBeEndOfSequence(tt, false, null, false) && CanBeCloseChar(bb2.Char, brList[0].Char))
{
ok = true;
}
break;
}
}
break;
}
if (t.IsCharOf(m_OpenChars) || t.IsCharOf(m_CloseChars))
{
ok = true;
break;
}
}
if (!ok)
{
break;
}
}
if (isAssim)
{
if (bb.CanBeOpen && !bb.CanBeClose && bb.Char == brList[0].Char)
{
lev++;
}
else if (bb.CanBeClose && !bb.CanBeOpen && m_OpenChars.IndexOf(brList[0].Char) == m_CloseChars.IndexOf(bb.Char))
{
lev--;
if (lev == 0)
{
break;
}
}
}
}
else
{
if ((++cou) > maxTokens)
{
break;
}
if (((attrs & BracketParseAttr.CanContainsVerbs)) == BracketParseAttr.No)
{
if (t.Morph.Language.IsCyrillic)
{
if (t.GetMorphClassInDictionary() == Pullenti.Morph.MorphClass.Verb)
{
if (!t.Morph.Class.IsAdjective && !t.Morph.ContainsAttr("страд.з.", null))
{
if (t.Chars.IsAllLower)
{
string norm = t.GetNormalCaseText(null, Pullenti.Morph.MorphNumber.Undefined, Pullenti.Morph.MorphGender.Undefined, false);
if (!Pullenti.Morph.LanguageHelper.EndsWith(norm, "СЯ"))
{
if (brList.Count > 1)
{
break;
}
if (brList[0].Char != '(')
{
break;
}
}
}
}
}
}
else if (t.Morph.Language.IsEn)
{
if (t.Morph.Class == Pullenti.Morph.MorphClass.Verb && t.Chars.IsAllLower)
{
break;
}
}
Pullenti.Ner.Referent r = t.GetReferent();
if (r != null && r.TypeName == "ADDRESS")
{
if (!t0.IsChar('('))
{
break;
}
}
}
}
if (((attrs & BracketParseAttr.CanBeManyLines)) != BracketParseAttr.No)
{
if (t.IsNewlineBefore)
{
if (t.NewlinesBeforeCount > 1)
{
break;
}
crlf++;
}
continue;
}
if (t.IsNewlineBefore)
{
if (t.WhitespacesBeforeCount > 15)
{
last = t.Previous;
break;
}
crlf++;
if (!t.Chars.IsAllLower)
{
if (MiscHelper.CanBeStartOfSentence(t))
{
bool has = false;
for (Pullenti.Ner.Token tt = t.Next; tt != null; tt = tt.Next)
{
if (tt.IsNewlineBefore)
{
break;
}
else if (tt.LengthChar == 1 && tt.IsCharOf(m_OpenChars) && tt.IsWhitespaceBefore)
{
break;
}
else if (tt.LengthChar == 1 && tt.IsCharOf(m_CloseChars) && !tt.IsWhitespaceBefore)
{
has = true;
break;
}
}
if (!has)
{
last = t.Previous;
break;
}
}
}
if ((t.Previous is Pullenti.Ner.MetaToken) && CanBeEndOfSequence((t.Previous as Pullenti.Ner.MetaToken).EndToken, false, null, false))
{
last = t.Previous;
break;
}
}
if (crlf > 1)
{
if (brList.Count > 1)
{
break;
}
if (crlf > 10)
{
break;
}
}
if (t.IsChar(';') && t.IsNewlineAfter)
{
break;
}
NounPhraseToken npt = NounPhraseHelper.TryParse(t, NounPhraseParseAttr.No, 0, null);
if (npt != null)
{
if (t.IsNewlineBefore)
{
genCase = npt.Morph.Case.IsGenitive;
}
last = (t = npt.EndToken);
}
}
if ((brList.Count == 1 && brList[0].CanBeOpen && (last is Pullenti.Ner.MetaToken)) && last.IsNewlineAfter)
{
if (BracketHelper.CanBeEndOfSequence((last as Pullenti.Ner.MetaToken).EndToken, false, null, false))
{
return(new BracketSequenceToken(t0, last));
}
}
if ((brList.Count == 1 && brList[0].CanBeOpen && genCase) && last.IsNewlineAfter && crlf <= 2)
{
return(new BracketSequenceToken(t0, last));
}
if (brList.Count < 1)
{
return(null);
}
for (int i = 1; i < (brList.Count - 1); i++)
{
if (brList[i].Char == '<' && brList[i + 1].Char == '>')
{
brList[i].CanBeOpen = true;
brList[i + 1].CanBeClose = true;
}
}
List <BracketSequenceToken> internals = null;
while (brList.Count > 3)
{
int i = brList.Count - 1;
if ((brList[i].CanBeClose && brList[i - 1].CanBeOpen && !CanBeCloseChar(brList[i].Char, brList[0].Char)) && CanBeCloseChar(brList[i].Char, brList[i - 1].Char))
{
brList.RemoveRange(brList.Count - 2, 2);
continue;
}
break;
}
while (brList.Count >= 4)
{
bool changed = false;
for (int i = 1; i < (brList.Count - 2); i++)
{
if ((brList[i].CanBeOpen && !brList[i].CanBeClose && brList[i + 1].CanBeClose) && !brList[i + 1].CanBeOpen)
{
bool ok = false;
if (MustBeCloseChar(brList[i + 1].Char, brList[i].Char) || brList[i].Char != brList[0].Char)
{
ok = true;
if ((i == 1 && ((i + 2) < brList.Count) && brList[i + 2].Char == ')') && brList[i + 1].Char != ')' && CanBeCloseChar(brList[i + 1].Char, brList[i - 1].Char))
{
brList[i + 2] = brList[i + 1];
}
}
else if (i > 1 && ((i + 2) < brList.Count) && MustBeCloseChar(brList[i + 2].Char, brList[i - 1].Char))
{
ok = true;
}
if (ok)
{
if (internals == null)
{
internals = new List <BracketSequenceToken>();
}
internals.Add(new BracketSequenceToken(brList[i].Source, brList[i + 1].Source));
brList.RemoveRange(i, 2);
changed = true;
break;
}
}
}
if (!changed)
{
break;
}
}
BracketSequenceToken res = null;
if ((brList.Count >= 4 && brList[1].CanBeOpen && brList[2].CanBeClose) && brList[3].CanBeClose && !brList[3].CanBeOpen)
{
if (CanBeCloseChar(brList[3].Char, brList[0].Char))
{
res = new BracketSequenceToken(brList[0].Source, brList[3].Source);
if (brList[0].Source.Next != brList[1].Source || brList[2].Source.Next != brList[3].Source)
{
res.Internal.Add(new BracketSequenceToken(brList[1].Source, brList[2].Source));
}
if (internals != null)
{
res.Internal.AddRange(internals);
}
}
}
if ((res == null && brList.Count >= 3 && brList[2].CanBeClose) && !brList[2].CanBeOpen)
{
if (((attrs & BracketParseAttr.NearCloseBracket)) != BracketParseAttr.No)
{
if (CanBeCloseChar(brList[1].Char, brList[0].Char))
{
return(new BracketSequenceToken(brList[0].Source, brList[1].Source));
}
}
bool ok = true;
if (CanBeCloseChar(brList[2].Char, brList[0].Char) && CanBeCloseChar(brList[1].Char, brList[0].Char) && brList[1].CanBeClose)
{
for (t = brList[1].Source; t != brList[2].Source && t != null; t = t.Next)
{
if (t.IsNewlineBefore)
{
ok = false;
break;
}
if (t.Chars.IsLetter && t.Chars.IsAllLower)
{
ok = false;
break;
}
NounPhraseToken npt = NounPhraseHelper.TryParse(t, NounPhraseParseAttr.No, 0, null);
if (npt != null)
{
t = npt.EndToken;
}
}
if (ok)
{
for (t = brList[0].Source.Next; t != brList[1].Source && t != null; t = t.Next)
{
if (t.IsNewlineBefore)
{
return(new BracketSequenceToken(brList[0].Source, t.Previous));
}
}
}
int lev1 = 0;
for (Pullenti.Ner.Token tt = brList[0].Source.Previous; tt != null; tt = tt.Previous)
{
if (tt.IsNewlineAfter || tt.IsTableControlChar)
{
break;
}
if (!(tt is Pullenti.Ner.TextToken))
{
continue;
}
if (tt.Chars.IsLetter || tt.LengthChar > 1)
{
continue;
}
char ch = (tt as Pullenti.Ner.TextToken).Term[0];
if (CanBeCloseChar(ch, brList[0].Char))
{
lev1++;
}
else if (CanBeCloseChar(brList[1].Char, ch))
{
lev1--;
if (lev1 < 0)
{
return(new BracketSequenceToken(brList[0].Source, brList[1].Source));
}
}
}
}
if (ok && CanBeCloseChar(brList[2].Char, brList[0].Char))
{
BracketSequenceToken intern = new BracketSequenceToken(brList[1].Source, brList[2].Source);
res = new BracketSequenceToken(brList[0].Source, brList[2].Source);
res.Internal.Add(intern);
}
else if (ok && CanBeCloseChar(brList[2].Char, brList[1].Char) && brList[0].CanBeOpen)
{
if (CanBeCloseChar(brList[2].Char, brList[0].Char))
{
BracketSequenceToken intern = new BracketSequenceToken(brList[1].Source, brList[2].Source);
res = new BracketSequenceToken(brList[0].Source, brList[2].Source);
res.Internal.Add(intern);
}
else if (brList.Count == 3)
{
return(null);
}
}
}
if (res == null && brList.Count > 1 && brList[1].CanBeClose)
{
res = new BracketSequenceToken(brList[0].Source, brList[1].Source);
}
if (res == null && brList.Count > 1 && CanBeCloseChar(brList[1].Char, brList[0].Char))
{
res = new BracketSequenceToken(brList[0].Source, brList[1].Source);
}
if (res == null && brList.Count == 2 && brList[0].Char == brList[1].Char)
{
res = new BracketSequenceToken(brList[0].Source, brList[1].Source);
}
if (res != null && internals != null)
{
foreach (BracketSequenceToken i in internals)
{
if (i.BeginChar < res.EndChar)
{
res.Internal.Add(i);
}
}
}
if (res == null)
{
cou = 0;
for (Pullenti.Ner.Token tt = t0.Next; tt != null; tt = tt.Next, cou++)
{
if (tt.IsTableControlChar)
{
break;
}
if (MiscHelper.CanBeStartOfSentence(tt))
{
break;
}
if (maxTokens > 0 && cou > maxTokens)
{
break;
}
Pullenti.Ner.MetaToken mt = tt as Pullenti.Ner.MetaToken;
if (mt == null)
{
continue;
}
if (mt.EndToken is Pullenti.Ner.TextToken)
{
if ((mt.EndToken as Pullenti.Ner.TextToken).IsCharOf(m_CloseChars))
{
Bracket bb = new Bracket(mt.EndToken as Pullenti.Ner.TextToken);
if (bb.CanBeClose && CanBeCloseChar(bb.Char, brList[0].Char))
{
return(new BracketSequenceToken(t0, tt));
}
}
}
}
}
return(res);
}
public static Pullenti.Semantic.SemObject CreateNptAdj(Pullenti.Semantic.SemGraph gr, Pullenti.Ner.Core.NounPhraseToken npt, Pullenti.Ner.MetaToken a)
{
if (a.Morph.Class.IsPronoun)
{
Pullenti.Semantic.SemObject asem = new Pullenti.Semantic.SemObject(gr);
gr.Objects.Add(asem);
asem.Tokens.Add(a);
asem.Typ = (a.BeginToken.Morph.Class.IsPersonalPronoun ? Pullenti.Semantic.SemObjectType.PersonalPronoun : Pullenti.Semantic.SemObjectType.Pronoun);
foreach (Pullenti.Morph.MorphBaseInfo it in a.BeginToken.Morph.Items)
{
Pullenti.Morph.MorphWordForm wf = it as Pullenti.Morph.MorphWordForm;
if (wf == null)
{
continue;
}
if (!npt.Morph.Case.IsUndefined)
{
if (((npt.Morph.Case & wf.Case)).IsUndefined)
{
continue;
}
}
_setMorph(asem, wf);
if (asem.Morph.NormalFull == "КАКОВ")
{
asem.Morph.NormalFull = "КАКОЙ";
}
break;
}
if (asem.Morph.NormalFull == null)
{
asem.Morph.NormalFull = (asem.Morph.NormalCase = a.GetNormalCaseText(null, Pullenti.Morph.MorphNumber.Undefined, Pullenti.Morph.MorphGender.Undefined, false));
}
return(asem);
}
if (!a.Morph.Class.IsVerb)
{
Pullenti.Semantic.SemObject asem = new Pullenti.Semantic.SemObject(gr);
gr.Objects.Add(asem);
asem.Tokens.Add(a);
asem.Typ = Pullenti.Semantic.SemObjectType.Adjective;
foreach (Pullenti.Morph.MorphBaseInfo wf in a.BeginToken.Morph.Items)
{
if (wf.CheckAccord(npt.Morph, false, false) && wf.Class.IsAdjective && (wf is Pullenti.Morph.MorphWordForm))
{
_setMorph(asem, wf as Pullenti.Morph.MorphWordForm);
break;
}
}
if (asem.Morph.NormalCase == null)
{
asem.Morph.NormalCase = a.GetNormalCaseText(Pullenti.Morph.MorphClass.Adjective, Pullenti.Morph.MorphNumber.Undefined, Pullenti.Morph.MorphGender.Undefined, false);
asem.Morph.NormalFull = a.GetNormalCaseText(Pullenti.Morph.MorphClass.Adjective, Pullenti.Morph.MorphNumber.Singular, Pullenti.Morph.MorphGender.Masculine, false);
_setMorph0(asem, a.BeginToken.Morph);
}
List <Pullenti.Semantic.Utils.DerivateGroup> grs = Pullenti.Semantic.Utils.DerivateService.FindDerivates(asem.Morph.NormalFull, true, null);
if (grs != null && grs.Count > 0)
{
asem.Concept = grs[0];
}
return(asem);
}
return(null);
}
static List <SemanticLink> _tryCreateVerb(Pullenti.Ner.Core.VerbPhraseToken vpt1, Pullenti.Ner.MetaToken slave, Pullenti.Semantic.Utils.DerivateGroup gr)
{
if (slave is Pullenti.Ner.Core.VerbPhraseToken)
{
return(_tryCreateInf(vpt1, slave as Pullenti.Ner.Core.VerbPhraseToken, gr));
}
SemanticAbstractSlave sla2 = slave as SemanticAbstractSlave;
List <SemanticLink> res = new List <SemanticLink>();
if (sla2 == null)
{
return(res);
}
Pullenti.Semantic.Utils.ControlModelItem cit = FindControlItem(vpt1.LastVerb, gr);
string prep = sla2.Preposition;
Pullenti.Morph.MorphBaseInfo morph = (Pullenti.Morph.MorphBaseInfo)sla2.Morph;
bool isRev1 = vpt1.LastVerb.IsVerbReversive || vpt1.LastVerb.IsVerbPassive;
bool noNomin = false;
bool noInstr = false;
if (prep == null && morph.Case.IsNominative && !vpt1.FirstVerb.IsParticiple)
{
bool ok = true;
bool err = false;
Pullenti.Morph.MorphWordForm vm = vpt1.FirstVerb.VerbMorph;
if (vm == null)
{
return(res);
}
if (vm.Number == Pullenti.Morph.MorphNumber.Singular)
{
if (morph.Number == Pullenti.Morph.MorphNumber.Plural)
{
if (!vpt1.FirstVerb.IsVerbInfinitive)
{
ok = false;
}
}
}
if (!CheckMorphAccord(morph, false, vm, false))
{
if (!err && !vpt1.FirstVerb.IsVerbInfinitive)
{
ok = false;
}
}
else if (vm.Misc.Person != Pullenti.Morph.MorphPerson.Undefined)
{
if (((vm.Misc.Person & Pullenti.Morph.MorphPerson.Third)) == Pullenti.Morph.MorphPerson.Undefined)
{
if (((vm.Misc.Person & Pullenti.Morph.MorphPerson.First)) == Pullenti.Morph.MorphPerson.First)
{
if (!morph.ContainsAttr("1 л.", null))
{
ok = false;
}
}
if (((vm.Misc.Person & Pullenti.Morph.MorphPerson.Second)) == Pullenti.Morph.MorphPerson.Second)
{
if (!morph.ContainsAttr("2 л.", null))
{
ok = false;
}
}
}
}
noNomin = true;
if (ok)
{
Pullenti.Semantic.Utils.ControlModelItem cit00 = cit;
bool isRev0 = isRev1;
if (vpt1.FirstVerb != vpt1.LastVerb && ((vpt1.FirstVerb.IsVerbReversive || vpt1.FirstVerb.IsVerbPassive || vpt1.FirstVerb.Normal == "ИМЕТЬ")))
{
cit00 = null;
isRev0 = true;
List <Pullenti.Semantic.Utils.DerivateGroup> grs = FindDerivates(vpt1.FirstVerb);
if (grs != null)
{
foreach (Pullenti.Semantic.Utils.DerivateGroup gg in grs)
{
if ((((cit00 = FindControlItem(vpt1.FirstVerb, gg)))) != null)
{
break;
}
}
}
}
SemanticLink sl = null;
bool addagent = false;
if (cit00 == null)
{
sl = new SemanticLink()
{
Modelled = true, Role = (isRev0 ? SemanticRole.Pacient : SemanticRole.Agent), Rank = 1, Question = Pullenti.Semantic.Utils.ControlModelQuestion.BaseNominative, IsPassive = isRev0
}
}
;
else
{
foreach (KeyValuePair <Pullenti.Semantic.Utils.ControlModelQuestion, SemanticRole> kp in cit00.Links)
{
Pullenti.Semantic.Utils.ControlModelQuestion q = kp.Key;
if (q.Check(null, Pullenti.Morph.MorphCase.Nominative))
{
sl = new SemanticLink()
{
Role = kp.Value, Rank = 2, Question = q, IsPassive = isRev0
};
if (sl.Role == SemanticRole.Agent)
{
sl.IsPassive = false;
}
else if (sl.Role == SemanticRole.Pacient && cit00.NominativeCanBeAgentAndPacient && vpt1.LastVerb.IsVerbReversive)
{
addagent = true;
}
break;
}
}
}
if (sl != null)
{
if (cit00 == null && morph.Case.IsInstrumental && isRev0)
{
sl.Rank -= 0.5;
}
if (morph.Case.IsAccusative)
{
sl.Rank -= 0.5;
}
if (sla2.BeginChar > vpt1.BeginChar)
{
sl.Rank -= 0.5;
}
if (err)
{
sl.Rank -= 0.5;
}
res.Add(sl);
if (addagent)
{
res.Add(new SemanticLink()
{
Role = SemanticRole.Agent, Rank = sl.Rank, Question = sl.Question
});
}
}
}
}
if (prep == null && isRev1 && morph.Case.IsInstrumental)
{
noInstr = true;
Pullenti.Semantic.Utils.ControlModelItem cit00 = cit;
SemanticLink sl = null;
if (cit00 == null)
{
sl = new SemanticLink()
{
Modelled = true, Role = SemanticRole.Agent, Rank = 1, Question = Pullenti.Semantic.Utils.ControlModelQuestion.BaseInstrumental, IsPassive = true
}
}
;
else
{
foreach (KeyValuePair <Pullenti.Semantic.Utils.ControlModelQuestion, SemanticRole> kp in cit00.Links)
{
Pullenti.Semantic.Utils.ControlModelQuestion q = kp.Key;
if (q.Check(null, Pullenti.Morph.MorphCase.Instrumental))
{
sl = new SemanticLink()
{
Role = kp.Value, Rank = 2, Question = q
};
if (sl.Role == SemanticRole.Agent)
{
sl.IsPassive = true;
}
break;
}
}
}
if (sl != null)
{
if (cit00 == null && morph.Case.IsNominative)
{
sl.Rank -= 0.5;
}
if (morph.Case.IsAccusative)
{
sl.Rank -= 0.5;
}
if (sla2.BeginChar < vpt1.BeginChar)
{
sl.Rank -= 0.5;
}
res.Add(sl);
if ((gr != null && gr.Model.Items.Count > 0 && gr.Model.Items[0].Typ == Pullenti.Semantic.Utils.ControlModelItemType.Verb) && gr.Model.Items[0].Links.ContainsKey(Pullenti.Semantic.Utils.ControlModelQuestion.BaseInstrumental))
{
sl.Rank = 0;
SemanticLink sl0 = new SemanticLink()
{
Question = sl.Question, Rank = 1, Role = gr.Model.Items[0].Links[Pullenti.Semantic.Utils.ControlModelQuestion.BaseInstrumental]
};
res.Insert(0, sl0);
}
}
}
if (prep == null && morph.Case.IsDative && ((cit == null || !cit.Links.ContainsKey(Pullenti.Semantic.Utils.ControlModelQuestion.BaseDative))))
{
SemanticLink sl = new SemanticLink()
{
Modelled = cit == null, Role = SemanticRole.Strong, Rank = 1, Question = Pullenti.Semantic.Utils.ControlModelQuestion.BaseDative
};
if (morph.Case.IsAccusative || morph.Case.IsNominative)
{
sl.Rank -= 0.5;
}
if (vpt1.EndToken.Next != sla2.BeginToken)
{
sl.Rank -= 0.5;
}
if (cit != null)
{
sl.Rank -= 0.5;
}
res.Add(sl);
}
_createRoles(cit, prep, morph.Case, res, noNomin, noInstr);
if (gr != null && gr.Model.Pacients.Count > 0)
{
bool ok = false;
foreach (string n in gr.Model.Pacients)
{
if (sla2.Source != null)
{
if (sla2.Source.EndToken.IsValue(n, null))
{
ok = true;
break;
}
}
else if (sla2.EndToken.IsValue(n, null))
{
ok = true;
break;
}
}
if (ok)
{
if (res.Count == 0)
{
ok = false;
if (prep == null && isRev1 && morph.Case.IsNominative)
{
ok = true;
}
else if (prep == null && !isRev1 && morph.Case.IsAccusative)
{
ok = true;
}
if (ok)
{
res.Add(new SemanticLink()
{
Role = SemanticRole.Pacient, Question = (isRev1 ? Pullenti.Semantic.Utils.ControlModelQuestion.BaseNominative : Pullenti.Semantic.Utils.ControlModelQuestion.BaseAccusative), Idiom = true
});
}
}
else
{
foreach (SemanticLink r in res)
{
r.Rank += 4;
if (r.Role == SemanticRole.Common)
{
r.Role = SemanticRole.Strong;
}
if (vpt1.EndToken.Next == sla2.BeginToken)
{
r.Rank += 2;
}
r.Idiom = true;
}
}
}
}
return(res);
}
public SemAttributeEx(Pullenti.Ner.MetaToken mt)
{
Token = mt;
}
static int _analizeListItems(List <FragToken> chi, int ind)
{
if (ind >= chi.Count)
{
return(-1);
}
FragToken res = chi[ind];
Pullenti.Ner.Instrument.InstrumentKind ki = res.Kind;
if (((ki == Pullenti.Ner.Instrument.InstrumentKind.Chapter || ki == Pullenti.Ner.Instrument.InstrumentKind.Clause || ki == Pullenti.Ner.Instrument.InstrumentKind.Content) || ki == Pullenti.Ner.Instrument.InstrumentKind.Item || ki == Pullenti.Ner.Instrument.InstrumentKind.Subitem) || ki == Pullenti.Ner.Instrument.InstrumentKind.ClausePart || ki == Pullenti.Ner.Instrument.InstrumentKind.Indention)
{
}
else
{
return(-1);
}
if (res.HasChanges && res.MultilineChangesValue != null)
{
Pullenti.Ner.MetaToken ci = res.MultilineChangesValue;
FragToken cit = new FragToken(ci.BeginToken, ci.EndToken)
{
Kind = Pullenti.Ner.Instrument.InstrumentKind.Citation
};
res.Children.Add(cit);
if (Pullenti.Ner.Core.BracketHelper.IsBracket(cit.BeginToken.Previous, true))
{
cit.BeginToken = cit.BeginToken.Previous;
}
if (Pullenti.Ner.Core.BracketHelper.IsBracket(cit.EndToken.Next, true))
{
cit.EndToken = cit.EndToken.Next;
if (cit.EndToken.Next != null && cit.EndToken.Next.IsCharOf(";."))
{
cit.EndToken = cit.EndToken.Next;
}
}
res.FillByContentChildren();
if (res.Children[0].HasChanges)
{
}
Pullenti.Ner.Instrument.InstrumentKind citKind = Pullenti.Ner.Instrument.InstrumentKind.Undefined;
if (ci.Tag is Pullenti.Ner.Decree.DecreeChangeReferent)
{
Pullenti.Ner.Decree.DecreeChangeReferent dcr = ci.Tag as Pullenti.Ner.Decree.DecreeChangeReferent;
if (dcr.Value != null && dcr.Value.NewItems.Count > 0)
{
string mnem = dcr.Value.NewItems[0];
int i;
if ((((i = mnem.IndexOf(' ')))) > 0)
{
mnem = mnem.Substring(0, i);
}
citKind = Pullenti.Ner.Decree.Internal.PartToken._getInstrKindByTyp(Pullenti.Ner.Decree.Internal.PartToken._getTypeByAttrName(mnem));
}
else if (dcr.Owners.Count > 0 && (dcr.Owners[0] is Pullenti.Ner.Decree.DecreePartReferent) && dcr.Kind == Pullenti.Ner.Decree.DecreeChangeKind.New)
{
Pullenti.Ner.Decree.DecreePartReferent pat = dcr.Owners[0] as Pullenti.Ner.Decree.DecreePartReferent;
int min = 0;
foreach (Pullenti.Ner.Slot s in pat.Slots)
{
Pullenti.Ner.Decree.Internal.PartToken.ItemType ty = Pullenti.Ner.Decree.Internal.PartToken._getTypeByAttrName(s.TypeName);
if (ty == Pullenti.Ner.Decree.Internal.PartToken.ItemType.Undefined)
{
continue;
}
int l = Pullenti.Ner.Decree.Internal.PartToken._getRank(ty);
if (l == 0)
{
continue;
}
if (l > min || min == 0)
{
min = l;
citKind = Pullenti.Ner.Decree.Internal.PartToken._getInstrKindByTyp(ty);
}
}
}
}
FragToken sub = null;
if (citKind != Pullenti.Ner.Instrument.InstrumentKind.Undefined && citKind != Pullenti.Ner.Instrument.InstrumentKind.Appendix)
{
sub = new FragToken(ci.BeginToken, ci.EndToken);
ContentAnalyzeWhapper wr = new ContentAnalyzeWhapper();
wr.Analyze(sub, null, true, citKind);
sub.Kind = Pullenti.Ner.Instrument.InstrumentKind.Content;
}
else
{
sub = FragToken.CreateDocument(ci.BeginToken, ci.EndChar, citKind);
}
if (sub == null || sub.Children.Count == 0)
{
}
else if ((sub.Kind == Pullenti.Ner.Instrument.InstrumentKind.Content && sub.Children.Count > 0 && sub.Children[0].BeginToken == sub.BeginToken) && sub.Children[sub.Children.Count - 1].EndToken == sub.EndToken)
{
cit.Children.AddRange(sub.Children);
}
else
{
cit.Children.Add(sub);
}
return(1);
}
int endChar = res.EndChar;
if (res.Itok == null)
{
res.Itok = InstrToken1.Parse(res.BeginToken, true, null, 0, null, false, res.EndChar, false, false);
}
List <LineToken> lines = LineToken.ParseList(res.BeginToken, endChar, null);
if (lines == null || (lines.Count < 1))
{
return(-1);
}
int ret = 1;
if (res.Kind == Pullenti.Ner.Instrument.InstrumentKind.Content)
{
for (int j = ind + 1; j < chi.Count; j++)
{
if (chi[j].Kind == Pullenti.Ner.Instrument.InstrumentKind.Content)
{
List <LineToken> lines2 = LineToken.ParseList(chi[j].BeginToken, chi[j].EndChar, lines[lines.Count - 1]);
if (lines2 == null || (lines2.Count < 1))
{
break;
}
if (!lines2[0].IsListItem)
{
if ((lines2.Count > 1 && lines2[1].IsListItem && lines2[0].EndToken.IsCharOf(":")) && !lines2[0].BeginToken.Chars.IsCapitalUpper)
{
lines2[0].IsListItem = true;
}
else
{
break;
}
}
lines.AddRange(lines2);
ret = (j - ind) + 1;
}
else if (chi[j].Kind != Pullenti.Ner.Instrument.InstrumentKind.Editions && chi[j].Kind != Pullenti.Ner.Instrument.InstrumentKind.Comment)
{
break;
}
}
}
if (lines.Count < 2)
{
return(-1);
}
if ((lines.Count > 1 && lines[0].IsListItem && lines[1].IsListItem) && lines[0].Number != 1)
{
if (lines.Count == 2 || !lines[2].IsListItem)
{
lines[0].IsListItem = (lines[1].IsListItem = false);
}
}
for (int i = 0; i < lines.Count; i++)
{
if (lines[i].IsListItem)
{
if (i > 0 && lines[i - 1].IsListItem)
{
continue;
}
if (((i + 1) < lines.Count) && lines[i + 1].IsListItem)
{
}
else
{
lines[i].IsListItem = false;
continue;
}
int j;
bool newLine = false;
for (j = i + 1; j < lines.Count; j++)
{
if (!lines[j].IsListItem)
{
break;
}
else if (lines[j].IsNewlineBefore)
{
newLine = true;
}
}
if (newLine)
{
continue;
}
if (i > 0 && lines[i - 1].EndToken.IsChar(':'))
{
continue;
}
for (j = i; j < lines.Count; j++)
{
if (!lines[j].IsListItem)
{
break;
}
else
{
lines[j].IsListItem = false;
}
}
}
}
if (lines.Count > 2)
{
LineToken last = lines[lines.Count - 1];
LineToken last2 = lines[lines.Count - 2];
if ((!last.IsListItem && last.EndToken.IsChar('.') && last2.IsListItem) && last2.EndToken.IsChar(';'))
{
if ((last.LengthChar < (last2.LengthChar * 2)) || last.BeginToken.Chars.IsAllLower)
{
last.IsListItem = true;
}
}
}
for (int i = 0; i < (lines.Count - 1); i++)
{
if (!lines[i].IsListItem && !lines[i + 1].IsListItem)
{
if (((i + 2) < lines.Count) && lines[i + 2].IsListItem && lines[i + 1].EndToken.IsChar(':'))
{
}
else
{
lines[i].EndToken = lines[i + 1].EndToken;
lines.RemoveAt(i + 1);
i--;
}
}
}
for (int i = 0; i < (lines.Count - 1); i++)
{
if (lines[i].IsListItem)
{
if (lines[i].Number == 1)
{
bool ok = true;
int num = 1;
int nonum = 0;
for (int j = i + 1; j < lines.Count; j++)
{
if (!lines[j].IsListItem)
{
ok = false;
break;
}
else if (lines[j].Number > 0)
{
num++;
if (lines[j].Number != num)
{
ok = false;
break;
}
}
else
{
nonum++;
}
}
if (!ok || nonum == 0 || (num < 2))
{
break;
}
LineToken lt = lines[i];
for (int j = i + 1; j < lines.Count; j++)
{
if (lines[j].Number > 0)
{
lt = lines[j];
}
else
{
List <LineToken> chli = lt.Tag as List <LineToken>;
if (chli == null)
{
lt.Tag = (chli = new List <LineToken>());
}
lt.EndToken = lines[j].EndToken;
chli.Add(lines[j]);
lines.RemoveAt(j);
j--;
}
}
}
}
}
int cou = 0;
foreach (LineToken li in lines)
{
if (li.IsListItem)
{
cou++;
}
}
if (cou < 2)
{
return(-1);
}
for (int i = 0; i < lines.Count; i++)
{
if (lines[i].IsListItem)
{
int i0 = i;
bool ok = true;
cou = 1;
for (; i < lines.Count; i++, cou++)
{
if (!lines[i].IsListItem)
{
break;
}
else if (lines[i].Number != cou)
{
ok = false;
}
}
if (!ok)
{
for (i = i0; i < lines.Count; i++)
{
if (!lines[i].IsListItem)
{
break;
}
else
{
lines[i].Number = 0;
}
}
}
if (cou > 3 && lines[i0].BeginToken.GetSourceText() != lines[i0 + 1].BeginToken.GetSourceText() && lines[i0 + 1].BeginToken.GetSourceText() == lines[i0 + 2].BeginToken.GetSourceText())
{
string pref = lines[i0 + 1].BeginToken.GetSourceText();
ok = true;
for (int j = i0 + 2; j < i; j++)
{
if (pref != lines[j].BeginToken.GetSourceText())
{
ok = false;
break;
}
}
if (!ok)
{
continue;
}
Pullenti.Ner.Token tt = null;
ok = false;
for (tt = lines[i0].EndToken.Previous; tt != null && tt != lines[i0].BeginToken; tt = tt.Previous)
{
if (tt.GetSourceText() == pref)
{
ok = true;
break;
}
}
if (ok)
{
LineToken li0 = new LineToken(lines[i0].BeginToken, tt.Previous);
lines[i0].BeginToken = tt;
lines.Insert(i0, li0);
i++;
}
}
}
}
foreach (LineToken li in lines)
{
li.CorrectBeginToken();
FragToken ch = new FragToken(li.BeginToken, li.EndToken)
{
Kind = (li.IsListItem ? Pullenti.Ner.Instrument.InstrumentKind.ListItem : Pullenti.Ner.Instrument.InstrumentKind.Content), Number = li.Number
};
if (ch.Kind == Pullenti.Ner.Instrument.InstrumentKind.Content && ch.EndToken.IsChar(':'))
{
ch.Kind = Pullenti.Ner.Instrument.InstrumentKind.ListHead;
}
res.Children.Add(ch);
List <LineToken> chli = li.Tag as List <LineToken>;
if (chli != null)
{
foreach (LineToken lt in chli)
{
ch.Children.Add(new FragToken(lt.BeginToken, lt.EndToken)
{
Kind = Pullenti.Ner.Instrument.InstrumentKind.ListItem
});
}
if (ch.BeginChar < ch.Children[0].BeginChar)
{
ch.Children.Insert(0, new FragToken(ch.BeginToken, ch.Children[0].BeginToken.Previous)
{
Kind = Pullenti.Ner.Instrument.InstrumentKind.Content
});
}
}
}
return(ret);
}
