/// <summary>
/// Попытаться выделить предлог с указанного токена
/// </summary>
/// <param name="t">начальный токен</param>
/// <return>результат или null</return>
public static PrepositionToken TryParse(Pullenti.Ner.Token t)
{
if (!(t is Pullenti.Ner.TextToken))
{
return(null);
}
TerminToken tok = m_Ontology.TryParse(t, TerminParseAttr.No);
if (tok != null)
{
return new PrepositionToken(t, tok.EndToken)
{
Normal = tok.Termin.CanonicText, NextCase = (Pullenti.Morph.MorphCase)tok.Termin.Tag
}
}
;
Pullenti.Morph.MorphClass mc = t.GetMorphClassInDictionary();
if (!mc.IsPreposition)
{
return(null);
}
PrepositionToken res = new PrepositionToken(t, t);
res.Normal = t.GetNormalCaseText(Pullenti.Morph.MorphClass.Preposition, Pullenti.Morph.MorphNumber.Undefined, Pullenti.Morph.MorphGender.Undefined, false);
res.NextCase = Pullenti.Morph.LanguageHelper.GetCaseAfterPreposition(res.Normal);
if ((t.Next != null && t.Next.IsHiphen && !t.IsWhitespaceAfter) && (t.Next.Next is Pullenti.Ner.TextToken) && t.Next.Next.GetMorphClassInDictionary().IsPreposition)
{
res.EndToken = t.Next.Next;
}
return(res);
}
// Привязка с точностью до похожести
// simD - параметр "похожесть (0.05..1)"
public List <TerminToken> TryParseAllSim(Pullenti.Ner.Token token, double simD)
{
if (simD >= 1 || (simD < 0.05))
{
return(this.TryParseAll(token, TerminParseAttr.No));
}
if (Termins.Count == 0 || token == null)
{
return(null);
}
Pullenti.Ner.TextToken tt = token as Pullenti.Ner.TextToken;
if (tt == null && (token is Pullenti.Ner.ReferentToken))
{
tt = (token as Pullenti.Ner.ReferentToken).BeginToken as Pullenti.Ner.TextToken;
}
List <TerminToken> res = null;
foreach (Termin t in Termins)
{
if (!t.Lang.IsUndefined)
{
if (!token.Morph.Language.IsUndefined)
{
if (((token.Morph.Language & t.Lang)).IsUndefined)
{
continue;
}
}
}
TerminToken ar = t.TryParseSim(tt, simD, TerminParseAttr.No);
if (ar == null)
{
continue;
}
ar.Termin = t;
if (res == null || ar.TokensCount > res[0].TokensCount)
{
res = new List <TerminToken>();
res.Add(ar);
}
else if (ar.TokensCount == res[0].TokensCount)
{
res.Add(ar);
}
}
return(res);
}
/// <summary>
/// Попытка привязать все возможные термины
/// </summary>
/// <param name="token">начальный токен</param>
/// <param name="attrs">атрибуты выделения</param>
/// <return>список из всех подходящих привязок TerminToken или null</return>
public List <TerminToken> TryParseAll(Pullenti.Ner.Token token, TerminParseAttr attrs = TerminParseAttr.No)
{
if (token == null)
{
return(null);
}
List <TerminToken> re = this._TryAttachAll_(token, attrs, false);
if (re == null && token.Morph.Language.IsUa)
{
re = this._TryAttachAll_(token, attrs, true);
}
if (re == null && Synonyms != null)
{
TerminToken re0 = Synonyms.TryParse(token, TerminParseAttr.No);
if (re0 != null && (re0.Termin.Tag is List <string>))
{
Termin term = this.Find(re0.Termin.CanonicText);
foreach (string syn in re0.Termin.Tag as List <string> )
{
if (term != null)
{
break;
}
term = this.Find(syn);
}
if (term != null)
{
re0.Termin = term;
List <TerminToken> res1 = new List <TerminToken>();
res1.Add(re0);
return(res1);
}
}
}
return(re);
}
bool _manageVar(Pullenti.Ner.Token token, TerminParseAttr pars, string v, CharNode nod, int i0, ref List <TerminToken> res)
{
for (int i = i0; i < v.Length; i++)
{
short ch = (short)v[i];
if (nod.Children == null)
{
return(false);
}
CharNode nn;
if (!nod.Children.TryGetValue(ch, out nn))
{
return(false);
}
nod = nn;
}
List <Termin> vars = nod.Termins;
if (vars == null || vars.Count == 0)
{
return(false);
}
foreach (Termin t in vars)
{
TerminToken ar = t.TryParse(token, pars);
if (ar != null)
{
ar.Termin = t;
if (res == null)
{
res = new List <TerminToken>();
res.Add(ar);
}
else if (ar.TokensCount > res[0].TokensCount)
{
res.Clear();
res.Add(ar);
}
else if (ar.TokensCount == res[0].TokensCount)
{
int j;
for (j = 0; j < res.Count; j++)
{
if (res[j].Termin == ar.Termin)
{
break;
}
}
if (j >= res.Count)
{
res.Add(ar);
}
}
}
if (t.AdditionalVars != null)
{
foreach (Termin av in t.AdditionalVars)
{
ar = av.TryParse(token, pars);
if (ar == null)
{
continue;
}
ar.Termin = t;
if (res == null)
{
res = new List <TerminToken>();
res.Add(ar);
}
else if (ar.TokensCount > res[0].TokensCount)
{
res.Clear();
res.Add(ar);
}
else if (ar.TokensCount == res[0].TokensCount)
{
int j;
for (j = 0; j < res.Count; j++)
{
if (res[j].Termin == ar.Termin)
{
break;
}
}
if (j >= res.Count)
{
res.Add(ar);
}
}
}
}
}
return(v.Length > 1);
}
List <TerminToken> _TryAttachAll_(Pullenti.Ner.Token token, TerminParseAttr pars = TerminParseAttr.No, bool mainRoot = false)
{
if (Termins.Count == 0 || token == null)
{
return(null);
}
string s = null;
Pullenti.Ner.TextToken tt = token as Pullenti.Ner.TextToken;
if (tt == null && (token is Pullenti.Ner.ReferentToken))
{
tt = (token as Pullenti.Ner.ReferentToken).BeginToken as Pullenti.Ner.TextToken;
}
List <TerminToken> res = null;
bool wasVars = false;
CharNode root = (mainRoot ? m_Root : this._getRoot(token.Morph.Language, token.Chars.IsLatinLetter));
if (tt != null)
{
s = tt.Term;
CharNode nod = root;
bool noVars = false;
int len0 = 0;
if (((pars & TerminParseAttr.TermOnly)) != TerminParseAttr.No)
{
}
else if (tt.InvariantPrefixLengthOfMorphVars <= s.Length)
{
len0 = tt.InvariantPrefixLengthOfMorphVars;
for (int i = 0; i < tt.InvariantPrefixLengthOfMorphVars; i++)
{
short ch = (short)s[i];
if (nod.Children == null)
{
noVars = true;
break;
}
CharNode nn;
if (!nod.Children.TryGetValue(ch, out nn))
{
noVars = true;
break;
}
nod = nn;
}
}
if (!noVars)
{
if (this._manageVar(token, pars, s, nod, len0, ref res))
{
wasVars = true;
}
for (int i = 0; i < tt.Morph.ItemsCount; i++)
{
if (((pars & TerminParseAttr.TermOnly)) != TerminParseAttr.No)
{
continue;
}
Pullenti.Morph.MorphWordForm wf = tt.Morph[i] as Pullenti.Morph.MorphWordForm;
if (wf == null)
{
continue;
}
if (((pars & TerminParseAttr.InDictionaryOnly)) != TerminParseAttr.No)
{
if (!wf.IsInDictionary)
{
continue;
}
}
int j;
bool ok = true;
if (wf.NormalCase == null || wf.NormalCase == s)
{
ok = false;
}
else
{
for (j = 0; j < i; j++)
{
Pullenti.Morph.MorphWordForm wf2 = tt.Morph[j] as Pullenti.Morph.MorphWordForm;
if (wf2 != null)
{
if (wf2.NormalCase == wf.NormalCase || wf2.NormalFull == wf.NormalCase)
{
break;
}
}
}
if (j < i)
{
ok = false;
}
}
if (ok)
{
if (this._manageVar(token, pars, wf.NormalCase, nod, tt.InvariantPrefixLengthOfMorphVars, ref res))
{
wasVars = true;
}
}
if (wf.NormalFull == null || wf.NormalFull == wf.NormalCase || wf.NormalFull == s)
{
continue;
}
for (j = 0; j < i; j++)
{
Pullenti.Morph.MorphWordForm wf2 = tt.Morph[j] as Pullenti.Morph.MorphWordForm;
if (wf2 != null && wf2.NormalFull == wf.NormalFull)
{
break;
}
}
if (j < i)
{
continue;
}
if (this._manageVar(token, pars, wf.NormalFull, nod, tt.InvariantPrefixLengthOfMorphVars, ref res))
{
wasVars = true;
}
}
}
}
else if (token is Pullenti.Ner.NumberToken)
{
if (this._manageVar(token, pars, (token as Pullenti.Ner.NumberToken).Value.ToString(), root, 0, ref res))
{
wasVars = true;
}
}
else
{
return(null);
}
if (!wasVars && s != null && s.Length == 1)
{
List <Termin> vars;
if (m_Hash1.TryGetValue((short)s[0], out vars))
{
foreach (Termin t in vars)
{
if (!t.Lang.IsUndefined)
{
if (!token.Morph.Language.IsUndefined)
{
if (((token.Morph.Language & t.Lang)).IsUndefined)
{
continue;
}
}
}
TerminToken ar = t.TryParse(tt, TerminParseAttr.No);
if (ar == null)
{
continue;
}
ar.Termin = t;
if (res == null)
{
res = new List <TerminToken>();
res.Add(ar);
}
else if (ar.TokensCount > res[0].TokensCount)
{
res.Clear();
res.Add(ar);
}
else if (ar.TokensCount == res[0].TokensCount)
{
res.Add(ar);
}
}
}
}
if (res != null)
{
int ii = 0;
int max = 0;
for (int i = 0; i < res.Count; i++)
{
if (res[i].LengthChar > max)
{
max = res[i].LengthChar;
ii = i;
}
}
if (ii > 0)
{
TerminToken v = res[ii];
res.RemoveAt(ii);
res.Insert(0, v);
}
}
return(res);
}
/// <summary>
/// Попытаться выделить союз с указанного токена.
/// </summary>
/// <param name="t">начальный токен</param>
/// <return>результат или null</return>
public static ConjunctionToken TryParse(Pullenti.Ner.Token t)
{
if (!(t is Pullenti.Ner.TextToken))
{
return(null);
}
if (t.IsComma)
{
ConjunctionToken ne = TryParse(t.Next);
if (ne != null)
{
ne.BeginToken = t;
ne.IsSimple = false;
return(ne);
}
return(new ConjunctionToken(t, t)
{
Typ = ConjunctionType.Comma, IsSimple = true, Normal = ","
});
}
TerminToken tok = m_Ontology.TryParse(t, TerminParseAttr.No);
if (tok != null)
{
if (t.IsValue("ТО", null))
{
NounPhraseToken npt = NounPhraseHelper.TryParse(t, NounPhraseParseAttr.ParseAdverbs, 0, null);
if (npt != null && npt.EndChar > tok.EndToken.EndChar)
{
return(null);
}
}
if (tok.Termin.Tag2 != null)
{
if (!(tok.EndToken is Pullenti.Ner.TextToken))
{
return(null);
}
if (tok.EndToken.GetMorphClassInDictionary().IsVerb)
{
if (!(tok.EndToken as Pullenti.Ner.TextToken).Term.EndsWith("АЯ"))
{
return(null);
}
}
}
return(new ConjunctionToken(t, tok.EndToken)
{
Normal = tok.Termin.CanonicText, Typ = (ConjunctionType)tok.Termin.Tag
});
}
if (!t.GetMorphClassInDictionary().IsConjunction)
{
return(null);
}
if (t.IsAnd || t.IsOr)
{
ConjunctionToken res = new ConjunctionToken(t, t)
{
Normal = (t as Pullenti.Ner.TextToken).Term, IsSimple = true, Typ = (t.IsOr ? ConjunctionType.Or : ConjunctionType.And)
};
if (((t.Next != null && t.Next.IsChar('(') && (t.Next.Next is Pullenti.Ner.TextToken)) && t.Next.Next.IsOr && t.Next.Next.Next != null) && t.Next.Next.Next.IsChar(')'))
{
res.EndToken = t.Next.Next.Next;
}
else if ((t.Next != null && t.Next.IsCharOf("\\/") && (t.Next.Next is Pullenti.Ner.TextToken)) && t.Next.Next.IsOr)
{
res.EndToken = t.Next.Next;
}
return(res);
}
string term = (t as Pullenti.Ner.TextToken).Term;
if (term == "НИ")
{
return new ConjunctionToken(t, t)
{
Normal = term, Typ = ConjunctionType.Not
}
}
;
if ((term == "А" || term == "НО" || term == "ЗАТО") || term == "ОДНАКО")
{
return new ConjunctionToken(t, t)
{
Normal = term, Typ = ConjunctionType.But
}
}
;
return(null);
}
