public static UriItemToken AttachBBK(Pullenti.Ner.Token t0)
{
StringBuilder txt = new StringBuilder();
Pullenti.Ner.Token t1 = t0;
int digs = 0;
for (Pullenti.Ner.Token t = t0; t != null; t = t.Next)
{
if (t.IsNewlineBefore && t != t0)
{
break;
}
if (t.IsTableControlChar)
{
break;
}
if (t is Pullenti.Ner.NumberToken)
{
Pullenti.Ner.NumberToken nt = t as Pullenti.Ner.NumberToken;
if (nt.Typ != Pullenti.Ner.NumberSpellingType.Digit || !nt.Morph.Class.IsUndefined)
{
break;
}
string d = nt.GetSourceText();
txt.Append(d);
digs += d.Length;
t1 = t;
continue;
}
Pullenti.Ner.TextToken tt = t as Pullenti.Ner.TextToken;
if (tt == null)
{
break;
}
if (tt.IsChar(','))
{
break;
}
if (tt.IsChar('('))
{
if (!(tt.Next is Pullenti.Ner.NumberToken))
{
break;
}
}
string s = tt.GetSourceText();
if (char.IsLetter(s[0]))
{
if (tt.IsWhitespaceBefore)
{
break;
}
}
txt.Append(s);
t1 = t;
}
if ((txt.Length < 3) || (digs < 2))
{
return(null);
}
if (txt[txt.Length - 1] == '.')
{
txt.Length--;
t1 = t1.Previous;
}
return(new UriItemToken(t0, t1)
{
Value = txt.ToString()
});
}
public static UnitToken TryParse(Pullenti.Ner.Token t, Pullenti.Ner.Core.TerminCollection addUnits, UnitToken prev, bool parseUnknownUnits = false)
{
if (t == null)
{
return(null);
}
Pullenti.Ner.Token t0 = t;
int pow = 1;
bool isNeg = false;
if ((t.IsCharOf("\\/") || t.IsValue("НА", null) || t.IsValue("OF", null)) || t.IsValue("PER", null))
{
isNeg = true;
t = t.Next;
}
else if (t.IsValue("В", null) && prev != null)
{
isNeg = true;
t = t.Next;
}
else if (MeasureHelper.IsMultChar(t))
{
t = t.Next;
}
Pullenti.Ner.TextToken tt = t as Pullenti.Ner.TextToken;
if (tt == null)
{
return(null);
}
if (tt.Term == "КВ" || tt.Term == "КВАДР" || tt.IsValue("КВАДРАТНЫЙ", null))
{
pow = 2;
tt = tt.Next as Pullenti.Ner.TextToken;
if (tt != null && tt.IsChar('.'))
{
tt = tt.Next as Pullenti.Ner.TextToken;
}
if (tt == null)
{
return(null);
}
}
else if (tt.Term == "КУБ" || tt.Term == "КУБИЧ" || tt.IsValue("КУБИЧЕСКИЙ", null))
{
pow = 3;
tt = tt.Next as Pullenti.Ner.TextToken;
if (tt != null && tt.IsChar('.'))
{
tt = tt.Next as Pullenti.Ner.TextToken;
}
if (tt == null)
{
return(null);
}
}
else if (tt.Term == "µ")
{
UnitToken res = TryParse(tt.Next, addUnits, prev, false);
if (res != null)
{
foreach (Unit u in UnitsHelper.Units)
{
if (u.Factor == UnitsFactors.Micro && string.Compare("мк" + u.NameCyr, res.Unit.NameCyr, true) == 0)
{
res.Unit = u;
res.BeginToken = tt;
res.Pow = pow;
if (isNeg)
{
res.Pow = -pow;
}
return(res);
}
}
}
}
List <Pullenti.Ner.Core.TerminToken> toks = UnitsHelper.Termins.TryParseAll(tt, Pullenti.Ner.Core.TerminParseAttr.No);
if (toks != null)
{
if ((prev != null && tt == t0 && toks.Count == 1) && t.IsWhitespaceBefore)
{
return(null);
}
if (toks[0].BeginToken == toks[0].EndToken && tt.Morph.Class.IsPreposition && (tt.WhitespacesAfterCount < 3))
{
if (Pullenti.Ner.Core.NounPhraseHelper.TryParse(tt, Pullenti.Ner.Core.NounPhraseParseAttr.ParsePreposition, 0, null) != null)
{
return(null);
}
if (tt.Next is Pullenti.Ner.NumberToken)
{
if ((tt.Next as Pullenti.Ner.NumberToken).Typ != Pullenti.Ner.NumberSpellingType.Digit)
{
return(null);
}
}
UnitToken nex = TryParse(tt.Next, addUnits, null, false);
if (nex != null)
{
return(null);
}
}
if (toks[0].BeginToken == toks[0].EndToken && ((toks[0].BeginToken.IsValue("М", null) || toks[0].BeginToken.IsValue("M", null))) && toks[0].BeginToken.Chars.IsAllLower)
{
if (prev != null && prev.Unit != null && prev.Unit.Kind == Pullenti.Ner.Measure.MeasureKind.Length)
{
UnitToken res = new UnitToken(t0, toks[0].EndToken)
{
Unit = UnitsHelper.uMinute
};
res.Pow = pow;
if (isNeg)
{
res.Pow = -pow;
}
return(res);
}
}
List <UnitToken> uts = new List <UnitToken>();
foreach (Pullenti.Ner.Core.TerminToken tok in toks)
{
UnitToken res = new UnitToken(t0, tok.EndToken)
{
Unit = tok.Termin.Tag as Unit
};
res.Pow = pow;
if (isNeg)
{
res.Pow = -pow;
}
if (res.Unit.BaseMultiplier == 1000000 && (t0 is Pullenti.Ner.TextToken) && char.IsLower((t0 as Pullenti.Ner.TextToken).GetSourceText()[0]))
{
foreach (Unit u in UnitsHelper.Units)
{
if (u.Factor == UnitsFactors.Milli && string.Compare(u.NameCyr, res.Unit.NameCyr, true) == 0)
{
res.Unit = u;
break;
}
}
}
res._correct();
res._checkDoubt();
uts.Add(res);
}
int max = 0;
UnitToken best = null;
foreach (UnitToken ut in uts)
{
if (ut.Keyword != null)
{
if (ut.Keyword.BeginChar >= max)
{
max = ut.Keyword.BeginChar;
best = ut;
}
}
}
if (best != null)
{
return(best);
}
foreach (UnitToken ut in uts)
{
if (!ut.IsDoubt)
{
return(ut);
}
}
return(uts[0]);
}
Pullenti.Ner.Token t1 = null;
if (t.IsCharOf("º°"))
{
t1 = t;
}
else if ((t.IsChar('<') && t.Next != null && t.Next.Next != null) && t.Next.Next.IsChar('>') && ((t.Next.IsValue("О", null) || t.Next.IsValue("O", null) || (((t.Next is Pullenti.Ner.NumberToken) && (t.Next as Pullenti.Ner.NumberToken).Value == "0")))))
{
t1 = t.Next.Next;
}
if (t1 != null)
{
UnitToken res = new UnitToken(t0, t1)
{
Unit = UnitsHelper.uGradus
};
res._checkDoubt();
t = t1.Next;
if (t != null && t.IsComma)
{
t = t.Next;
}
if (t != null && t.IsValue("ПО", null))
{
t = t.Next;
}
if (t is Pullenti.Ner.TextToken)
{
string vv = (t as Pullenti.Ner.TextToken).Term;
if (vv == "C" || vv == "С" || vv.StartsWith("ЦЕЛЬС"))
{
res.Unit = UnitsHelper.uGradusC;
res.IsDoubt = false;
res.EndToken = t;
}
if (vv == "F" || vv.StartsWith("ФАР"))
{
res.Unit = UnitsHelper.uGradusF;
res.IsDoubt = false;
res.EndToken = t;
}
}
return(res);
}
if ((t is Pullenti.Ner.TextToken) && ((t.IsValue("ОС", null) || t.IsValue("OC", null))))
{
string str = t.GetSourceText();
if (str == "оС" || str == "oC")
{
UnitToken res = new UnitToken(t, t)
{
Unit = UnitsHelper.uGradusC, IsDoubt = false
};
return(res);
}
}
if (t.IsChar('%'))
{
Pullenti.Ner.Token tt1 = t.Next;
if (tt1 != null && tt1.IsChar('('))
{
tt1 = tt1.Next;
}
if ((tt1 is Pullenti.Ner.TextToken) && (tt1 as Pullenti.Ner.TextToken).Term.StartsWith("ОБ"))
{
UnitToken re = new UnitToken(t, tt1)
{
Unit = UnitsHelper.uAlco
};
if (re.EndToken.Next != null && re.EndToken.Next.IsChar('.'))
{
re.EndToken = re.EndToken.Next;
}
if (re.EndToken.Next != null && re.EndToken.Next.IsChar(')') && t.Next.IsChar('('))
{
re.EndToken = re.EndToken.Next;
}
return(re);
}
return(new UnitToken(t, t)
{
Unit = UnitsHelper.uPercent
});
}
if (addUnits != null)
{
Pullenti.Ner.Core.TerminToken tok = addUnits.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok != null)
{
UnitToken res = new UnitToken(t0, tok.EndToken)
{
ExtOnto = tok.Termin.Tag as Pullenti.Ner.Measure.UnitReferent
};
if (tok.EndToken.Next != null && tok.EndToken.Next.IsChar('.'))
{
tok.EndToken = tok.EndToken.Next;
}
res.Pow = pow;
if (isNeg)
{
res.Pow = -pow;
}
res._correct();
return(res);
}
}
if (!parseUnknownUnits)
{
return(null);
}
if ((t.WhitespacesBeforeCount > 2 || !t.Chars.IsLetter || t.LengthChar > 5) || !(t is Pullenti.Ner.TextToken))
{
return(null);
}
if (Pullenti.Ner.Core.MiscHelper.CanBeStartOfSentence(t))
{
return(null);
}
t1 = t;
if (t.Next != null && t.Next.IsChar('.'))
{
t1 = t;
}
bool ok = false;
if (t1.Next == null || t1.WhitespacesAfterCount > 2)
{
ok = true;
}
else if (t1.Next.IsComma || t1.Next.IsCharOf("\\/") || t1.Next.IsTableControlChar)
{
ok = true;
}
else if (MeasureHelper.IsMultChar(t1.Next))
{
ok = true;
}
if (!ok)
{
return(null);
}
Pullenti.Morph.MorphClass mc = t.GetMorphClassInDictionary();
if (mc.IsUndefined)
{
}
else if (t.LengthChar > 7)
{
return(null);
}
UnitToken res1 = new UnitToken(t0, t1)
{
Pow = pow, IsDoubt = true
};
res1.UnknownName = (t as Pullenti.Ner.TextToken).GetSourceText();
res1._correct();
return(res1);
}
bool CalcRankAndValue(int minNewlinesCount)
{
Rank = 0;
if (BeginToken.Chars.IsAllLower)
{
Rank -= 30;
}
int words = 0;
int upWords = 0;
int notwords = 0;
int lineNumber = 0;
Pullenti.Ner.Token tstart = BeginToken;
Pullenti.Ner.Token tend = EndToken;
for (Pullenti.Ner.Token t = BeginToken; t != EndToken.Next && t != null && t.EndChar <= EndToken.EndChar; t = t.Next)
{
if (t.IsNewlineBefore)
{
}
TitleItemToken tit = TitleItemToken.TryAttach(t);
if (tit != null)
{
if (tit.Typ == TitleItemToken.Types.Theme || tit.Typ == TitleItemToken.Types.TypAndTheme)
{
if (t != BeginToken)
{
if (lineNumber > 0)
{
return(false);
}
words = (upWords = (notwords = 0));
tstart = tit.EndToken.Next;
}
t = tit.EndToken;
if (t.Next == null)
{
return(false);
}
if (t.Next.Chars.IsLetter && t.Next.Chars.IsAllLower)
{
Rank += 20;
}
else
{
Rank += 100;
}
tstart = t.Next;
if (tit.Typ == TitleItemToken.Types.TypAndTheme)
{
TypeValue = tit.Value;
}
continue;
}
if (tit.Typ == TitleItemToken.Types.Typ)
{
if (t == BeginToken)
{
if (tit.EndToken.IsNewlineAfter)
{
TypeValue = tit.Value;
Rank += 5;
tstart = tit.EndToken.Next;
}
}
t = tit.EndToken;
words++;
if (tit.BeginToken != tit.EndToken)
{
words++;
}
if (tit.Chars.IsAllUpper)
{
upWords++;
}
continue;
}
if (tit.Typ == TitleItemToken.Types.Dust || tit.Typ == TitleItemToken.Types.Speciality)
{
if (t == BeginToken)
{
return(false);
}
Rank -= 20;
if (tit.Typ == TitleItemToken.Types.Speciality)
{
Speciality = tit.Value;
}
t = tit.EndToken;
continue;
}
if (tit.Typ == TitleItemToken.Types.Consultant || tit.Typ == TitleItemToken.Types.Boss || tit.Typ == TitleItemToken.Types.Editor)
{
t = tit.EndToken;
if (t.Next != null && ((t.Next.IsCharOf(":") || t.Next.IsHiphen || t.WhitespacesAfterCount > 4)))
{
Rank -= 10;
}
else
{
Rank -= 2;
}
continue;
}
return(false);
}
Pullenti.Ner.Booklink.Internal.BookLinkToken blt = Pullenti.Ner.Booklink.Internal.BookLinkToken.TryParse(t, 0);
if (blt != null)
{
if (blt.Typ == Pullenti.Ner.Booklink.Internal.BookLinkTyp.Misc || blt.Typ == Pullenti.Ner.Booklink.Internal.BookLinkTyp.N || blt.Typ == Pullenti.Ner.Booklink.Internal.BookLinkTyp.Pages)
{
Rank -= 10;
}
else if (blt.Typ == Pullenti.Ner.Booklink.Internal.BookLinkTyp.N || blt.Typ == Pullenti.Ner.Booklink.Internal.BookLinkTyp.PageRange)
{
Rank -= 20;
}
}
if (t == BeginToken && Pullenti.Ner.Booklink.Internal.BookLinkToken.TryParseAuthor(t, Pullenti.Ner.Person.Internal.FioTemplateType.Undefined) != null)
{
Rank -= 20;
}
if (t.IsNewlineBefore && t != BeginToken)
{
lineNumber++;
if (lineNumber > 4)
{
return(false);
}
if (t.Chars.IsAllLower)
{
Rank += 10;
}
else if (t.Previous.IsChar('.'))
{
Rank -= 10;
}
else if (t.Previous.IsCharOf(",-"))
{
Rank += 10;
}
else
{
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t.Previous, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null && npt.EndChar >= t.EndChar)
{
Rank += 10;
}
}
}
if (t != BeginToken && t.NewlinesBeforeCount > minNewlinesCount)
{
Rank -= (t.NewlinesBeforeCount - minNewlinesCount);
}
Pullenti.Ner.Core.BracketSequenceToken bst = Pullenti.Ner.Core.BracketHelper.TryParse(t, Pullenti.Ner.Core.BracketParseAttr.No, 100);
if (bst != null && bst.IsQuoteType && bst.EndToken.EndChar <= EndToken.EndChar)
{
if (words == 0)
{
tstart = bst.BeginToken;
Rank += 10;
if (bst.EndToken == EndToken)
{
tend = EndToken;
Rank += 10;
}
}
}
List <Pullenti.Ner.Referent> rli = t.GetReferents();
if (rli != null)
{
foreach (Pullenti.Ner.Referent r in rli)
{
if (r is Pullenti.Ner.Org.OrganizationReferent)
{
if (t.IsNewlineBefore)
{
Rank -= 10;
}
else
{
Rank -= 4;
}
continue;
}
if ((r is Pullenti.Ner.Geo.GeoReferent) || (r is Pullenti.Ner.Person.PersonReferent))
{
if (t.IsNewlineBefore)
{
Rank -= 5;
if (t.IsNewlineAfter || t.Next == null)
{
Rank -= 20;
}
else if (t.Next.IsHiphen || (t.Next is Pullenti.Ner.NumberToken) || (t.Next.GetReferent() is Pullenti.Ner.Date.DateReferent))
{
Rank -= 20;
}
else if (t != BeginToken)
{
Rank -= 20;
}
}
continue;
}
if ((r is Pullenti.Ner.Geo.GeoReferent) || (r is Pullenti.Ner.Denomination.DenominationReferent))
{
continue;
}
if ((r is Pullenti.Ner.Uri.UriReferent) || (r is Pullenti.Ner.Phone.PhoneReferent))
{
return(false);
}
if (t.IsNewlineBefore)
{
Rank -= 4;
}
else
{
Rank -= 2;
}
if (t == BeginToken && (EndToken.GetReferent() is Pullenti.Ner.Person.PersonReferent))
{
Rank -= 10;
}
}
words++;
if (t.Chars.IsAllUpper)
{
upWords++;
}
if (t == BeginToken)
{
if (t.IsNewlineAfter)
{
Rank -= 10;
}
else if (t.Next != null && t.Next.IsChar('.') && t.Next.IsNewlineAfter)
{
Rank -= 10;
}
}
continue;
}
if (t is Pullenti.Ner.NumberToken)
{
if ((t as Pullenti.Ner.NumberToken).Typ == Pullenti.Ner.NumberSpellingType.Words)
{
words++;
if (t.Chars.IsAllUpper)
{
upWords++;
}
}
else
{
notwords++;
}
continue;
}
Pullenti.Ner.Person.Internal.PersonAttrToken pat = Pullenti.Ner.Person.Internal.PersonAttrToken.TryAttach(t, null, Pullenti.Ner.Person.Internal.PersonAttrToken.PersonAttrAttachAttrs.No);
if (pat != null)
{
if (t.IsNewlineBefore)
{
if (!pat.Morph.Case.IsUndefined && !pat.Morph.Case.IsNominative)
{
}
else if (pat.Chars.IsAllUpper)
{
}
else
{
Rank -= 20;
}
}
else if (t.Chars.IsAllLower)
{
Rank--;
}
for (; t != null; t = t.Next)
{
words++;
if (t.Chars.IsAllUpper)
{
upWords++;
}
if (t == pat.EndToken)
{
break;
}
}
continue;
}
Pullenti.Ner.Org.Internal.OrgItemTypeToken oitt = Pullenti.Ner.Org.Internal.OrgItemTypeToken.TryAttach(t, true, null);
if (oitt != null)
{
if (oitt.Morph.Number != Pullenti.Morph.MorphNumber.Plural && !oitt.IsDoubtRootWord)
{
if (!oitt.Morph.Case.IsUndefined && !oitt.Morph.Case.IsNominative)
{
words++;
if (t.Chars.IsAllUpper)
{
upWords++;
}
}
else
{
Rank -= 4;
if (t == BeginToken)
{
Rank -= 5;
}
}
}
else
{
words += 1;
if (t.Chars.IsAllUpper)
{
upWords++;
}
}
t = oitt.EndToken;
continue;
}
Pullenti.Ner.TextToken tt = t as Pullenti.Ner.TextToken;
if (tt != null)
{
if (tt.IsChar('©'))
{
Rank -= 10;
}
if (tt.IsChar('_'))
{
Rank--;
}
if (tt.Chars.IsLetter)
{
if (tt.LengthChar > 2)
{
words++;
if (t.Chars.IsAllUpper)
{
upWords++;
}
}
}
else if (!tt.IsChar(','))
{
notwords++;
}
if (tt.IsPureVerb)
{
{
Rank -= 30;
words--;
}
break;
}
if (tt == EndToken)
{
if (tt.Morph.Class.IsPreposition || tt.Morph.Class.IsConjunction)
{
Rank -= 10;
}
else if (tt.IsChar('.'))
{
Rank += 5;
}
}
else if (tt.IsCharOf("._"))
{
Rank -= 5;
}
}
}
Rank += words;
Rank -= notwords;
if ((words < 1) && (Rank < 50))
{
return(false);
}
if (tstart == null || tend == null)
{
return(false);
}
if (tstart.EndChar > tend.EndChar)
{
return(false);
}
TitleItemToken tit1 = TitleItemToken.TryAttach(EndToken.Next);
if (tit1 != null && ((tit1.Typ == TitleItemToken.Types.Typ || tit1.Typ == TitleItemToken.Types.Speciality)))
{
if (tit1.EndToken.IsNewlineAfter)
{
Rank += 15;
}
else
{
Rank += 10;
}
if (tit1.Typ == TitleItemToken.Types.Speciality)
{
Speciality = tit1.Value;
}
}
if (upWords > 4 && upWords > ((int)((0.8 * words))))
{
if (tstart.Previous != null && (tstart.Previous.GetReferent() is Pullenti.Ner.Person.PersonReferent))
{
Rank += (5 + upWords);
}
}
BeginNameToken = tstart;
EndNameToken = tend;
return(true);
}
public static TitleItemToken TryAttach(Pullenti.Ner.Token t)
{
Pullenti.Ner.TextToken tt = t as Pullenti.Ner.TextToken;
if (tt != null)
{
Pullenti.Ner.Token t1 = (Pullenti.Ner.Token)tt;
if (tt.Term == "ТЕМА")
{
TitleItemToken tit = TryAttach(tt.Next);
if (tit != null && tit.Typ == Types.Typ)
{
t1 = tit.EndToken;
if (t1.Next != null && t1.Next.IsChar(':'))
{
t1 = t1.Next;
}
return(new TitleItemToken(t, t1, Types.TypAndTheme)
{
Value = tit.Value
});
}
if (tt.Next != null && tt.Next.IsChar(':'))
{
t1 = tt.Next;
}
return(new TitleItemToken(tt, t1, Types.Theme));
}
if (tt.Term == "ПО" || tt.Term == "НА")
{
if (tt.Next != null && tt.Next.IsValue("ТЕМА", null))
{
t1 = tt.Next;
if (t1.Next != null && t1.Next.IsChar(':'))
{
t1 = t1.Next;
}
return(new TitleItemToken(tt, t1, Types.Theme));
}
}
if (tt.Term == "ПЕРЕВОД" || tt.Term == "ПЕР")
{
Pullenti.Ner.Token tt2 = tt.Next;
if (tt2 != null && tt2.IsChar('.'))
{
tt2 = tt2.Next;
}
if (tt2 is Pullenti.Ner.TextToken)
{
if ((tt2 as Pullenti.Ner.TextToken).Term == "C" || (tt2 as Pullenti.Ner.TextToken).Term == "С")
{
tt2 = tt2.Next;
if (tt2 is Pullenti.Ner.TextToken)
{
return(new TitleItemToken(t, tt2, Types.Translate));
}
}
}
}
if (tt.Term == "СЕКЦИЯ" || tt.Term == "SECTION" || tt.Term == "СЕКЦІЯ")
{
t1 = tt.Next;
if (t1 != null && t1.IsChar(':'))
{
t1 = t1.Next;
}
Pullenti.Ner.Core.BracketSequenceToken br = Pullenti.Ner.Core.BracketHelper.TryParse(t1, Pullenti.Ner.Core.BracketParseAttr.No, 100);
if (br != null)
{
t1 = br.EndToken;
}
else if (t1 != tt.Next)
{
for (; t1 != null; t1 = t1.Next)
{
if (t1.IsNewlineAfter)
{
break;
}
}
if (t1 == null)
{
return(null);
}
}
if (t1 != tt.Next)
{
return(new TitleItemToken(tt, t1, Types.Dust));
}
}
t1 = null;
if (tt.IsValue("СПЕЦИАЛЬНОСТЬ", "СПЕЦІАЛЬНІСТЬ"))
{
t1 = tt.Next;
}
else if (tt.Morph.Class.IsPreposition && tt.Next != null && tt.Next.IsValue("СПЕЦИАЛЬНОСТЬ", "СПЕЦІАЛЬНІСТЬ"))
{
t1 = tt.Next.Next;
}
else if (tt.IsChar('/') && tt.IsNewlineBefore)
{
t1 = tt.Next;
}
if (t1 != null)
{
if (t1.IsCharOf(":") || t1.IsHiphen)
{
t1 = t1.Next;
}
TitleItemToken spec = TryAttachSpeciality(t1, true);
if (spec != null)
{
spec.BeginToken = t;
return(spec);
}
}
}
TitleItemToken sss = TryAttachSpeciality(t, false);
if (sss != null)
{
return(sss);
}
if (t is Pullenti.Ner.ReferentToken)
{
return(null);
}
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null)
{
string s = npt.GetNormalCaseText(null, Pullenti.Morph.MorphNumber.Undefined, Pullenti.Morph.MorphGender.Undefined, false);
Pullenti.Ner.Core.TerminToken tok = m_Termins.TryParse(npt.EndToken, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok != null)
{
Types ty = (Types)tok.Termin.Tag;
if (ty == Types.Typ)
{
TitleItemToken tit = TryAttach(tok.EndToken.Next);
if (tit != null && tit.Typ == Types.Theme)
{
return new TitleItemToken(npt.BeginToken, tit.EndToken, Types.TypAndTheme)
{
Value = s
}
}
;
if (s == "РАБОТА" || s == "РОБОТА" || s == "ПРОЕКТ")
{
return(null);
}
Pullenti.Ner.Token t1 = tok.EndToken;
if (s == "ДИССЕРТАЦИЯ" || s == "ДИСЕРТАЦІЯ")
{
int err = 0;
for (Pullenti.Ner.Token ttt = t1.Next; ttt != null; ttt = ttt.Next)
{
if (ttt.Morph.Class.IsPreposition)
{
continue;
}
if (ttt.IsValue("СОИСКАНИЕ", ""))
{
continue;
}
Pullenti.Ner.Core.NounPhraseToken npt1 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(ttt, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt1 != null && npt1.Noun.IsValue("СТЕПЕНЬ", "СТУПІНЬ"))
{
t1 = (ttt = npt1.EndToken);
continue;
}
Pullenti.Ner.ReferentToken rt = t1.Kit.ProcessReferent("PERSON", ttt);
if (rt != null && (rt.Referent is Pullenti.Ner.Person.PersonPropertyReferent))
{
Pullenti.Ner.Person.PersonPropertyReferent ppr = rt.Referent as Pullenti.Ner.Person.PersonPropertyReferent;
if (ppr.Name == "доктор наук")
{
t1 = rt.EndToken;
s = "ДОКТОРСКАЯ ДИССЕРТАЦИЯ";
break;
}
else if (ppr.Name == "кандидат наук")
{
t1 = rt.EndToken;
s = "КАНДИДАТСКАЯ ДИССЕРТАЦИЯ";
break;
}
else if (ppr.Name == "магистр")
{
t1 = rt.EndToken;
s = "МАГИСТЕРСКАЯ ДИССЕРТАЦИЯ";
break;
}
}
if (ttt.IsValue("ДОКТОР", null) || ttt.IsValue("КАНДИДАТ", null) || ttt.IsValue("МАГИСТР", "МАГІСТР"))
{
t1 = ttt;
npt1 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(ttt.Next, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt1 != null && npt1.EndToken.IsValue("НАУК", null))
{
t1 = npt1.EndToken;
}
s = (ttt.IsValue("МАГИСТР", "МАГІСТР") ? "МАГИСТЕРСКАЯ ДИССЕРТАЦИЯ" : (ttt.IsValue("ДОКТОР", null) ? "ДОКТОРСКАЯ ДИССЕРТАЦИЯ" : "КАНДИДАТСКАЯ ДИССЕРТАЦИЯ"));
break;
}
if ((++err) > 3)
{
break;
}
}
}
if (t1.Next != null && t1.Next.IsChar('.'))
{
t1 = t1.Next;
}
if (s.EndsWith("ОТЧЕТ") && t1.Next != null && t1.Next.IsValue("О", null))
{
Pullenti.Ner.Core.NounPhraseToken npt1 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t1.Next, Pullenti.Ner.Core.NounPhraseParseAttr.ParsePreposition, 0, null);
if (npt1 != null && npt1.Morph.Case.IsPrepositional)
{
t1 = npt1.EndToken;
}
}
return(new TitleItemToken(npt.BeginToken, t1, ty)
{
Value = s
});
}
}
}
Pullenti.Ner.Core.TerminToken tok1 = m_Termins.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok1 != null)
{
Pullenti.Ner.Token t1 = tok1.EndToken;
TitleItemToken re = new TitleItemToken(tok1.BeginToken, t1, (Types)tok1.Termin.Tag);
return(re);
}
if (Pullenti.Ner.Core.BracketHelper.CanBeStartOfSequence(t, false, false))
{
tok1 = m_Termins.TryParse(t.Next, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok1 != null && Pullenti.Ner.Core.BracketHelper.CanBeEndOfSequence(tok1.EndToken.Next, false, null, false))
{
Pullenti.Ner.Token t1 = tok1.EndToken.Next;
return(new TitleItemToken(tok1.BeginToken, t1, (Types)tok1.Termin.Tag));
}
}
return(null);
}
static OrgItemNameToken _TryAttach(Pullenti.Ner.Token t, OrgItemNameToken prev, bool extOnto)
{
if (t == null)
{
return(null);
}
Pullenti.Ner.Referent r = t.GetReferent();
if (r != null)
{
if (r.TypeName == "DENOMINATION")
{
return new OrgItemNameToken(t, t)
{
Value = r.ToString(true, t.Kit.BaseLanguage, 0), IsDenomination = true
}
}
;
if ((r is Pullenti.Ner.Geo.GeoReferent) && t.Chars.IsLatinLetter)
{
OrgItemNameToken res2 = _TryAttach(t.Next, prev, extOnto);
if (res2 != null && res2.Chars.IsLatinLetter)
{
res2.BeginToken = t;
res2.Value = string.Format("{0} {1}", Pullenti.Ner.Core.MiscHelper.GetTextValueOfMetaToken(t as Pullenti.Ner.MetaToken, Pullenti.Ner.Core.GetTextAttr.No), res2.Value);
res2.IsInDictionary = false;
return(res2);
}
}
return(null);
}
Pullenti.Ner.TextToken tt = t as Pullenti.Ner.TextToken;
if (tt == null)
{
return(null);
}
OrgItemNameToken res = null;
Pullenti.Ner.Core.TerminToken tok = m_StdTails.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok == null && t.IsChar(','))
{
tok = m_StdTails.TryParse(t.Next, Pullenti.Ner.Core.TerminParseAttr.No);
}
if (tok != null)
{
return new OrgItemNameToken(t, tok.EndToken)
{
Value = tok.Termin.CanonicText, IsStdTail = tok.Termin.Tag == null, IsEmptyWord = tok.Termin.Tag != null, Morph = tok.Morph
}
}
;
if ((((tok = m_StdNames.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No)))) != null)
{
return new OrgItemNameToken(t, tok.EndToken)
{
Value = tok.Termin.CanonicText, IsStdName = true
}
}
;
OrgItemEngItem eng = OrgItemEngItem.TryAttach(t, false);
if (eng == null && t.IsChar(','))
{
eng = OrgItemEngItem.TryAttach(t.Next, false);
}
if (eng != null)
{
return new OrgItemNameToken(t, eng.EndToken)
{
Value = eng.FullValue, IsStdTail = true
}
}
;
if (tt.Chars.IsAllLower && prev != null)
{
if (!prev.Chars.IsAllLower && !prev.Chars.IsCapitalUpper)
{
return(null);
}
}
if (tt.IsChar(',') && prev != null)
{
Pullenti.Ner.Core.NounPhraseToken npt1 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t.Next, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt1 == null || npt1.Chars != prev.Chars || ((npt1.Morph.Case & prev.Morph.Case)).IsUndefined)
{
return(null);
}
OrgItemTypeToken ty = OrgItemTypeToken.TryAttach(t.Next, false, null);
if (ty != null)
{
return(null);
}
if (npt1.EndToken.Next == null || !npt1.EndToken.Next.IsValue("И", null))
{
return(null);
}
Pullenti.Ner.Token t1 = npt1.EndToken.Next;
Pullenti.Ner.Core.NounPhraseToken npt2 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t1.Next, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt2 == null || npt2.Chars != prev.Chars || ((npt2.Morph.Case & npt1.Morph.Case & prev.Morph.Case)).IsUndefined)
{
return(null);
}
ty = OrgItemTypeToken.TryAttach(t1.Next, false, null);
if (ty != null)
{
return(null);
}
res = new OrgItemNameToken(npt1.BeginToken, npt1.EndToken)
{
Morph = npt1.Morph, Value = npt1.GetNormalCaseText(null, Pullenti.Morph.MorphNumber.Undefined, Pullenti.Morph.MorphGender.Undefined, false)
};
res.IsNounPhrase = true;
res.IsAfterConjunction = true;
if (prev.Preposition != null)
{
res.Preposition = prev.Preposition;
}
return(res);
}
if (((tt.IsChar('&') || tt.IsValue("AND", null) || tt.IsValue("UND", null))) && prev != null)
{
if ((tt.Next is Pullenti.Ner.TextToken) && tt.LengthChar == 1 && tt.Next.Chars.IsLatinLetter)
{
res = new OrgItemNameToken(tt, tt.Next)
{
Chars = tt.Next.Chars
};
res.IsAfterConjunction = true;
res.Value = "& " + (tt.Next as Pullenti.Ner.TextToken).Term;
return(res);
}
res = OrgItemNameToken.TryAttach(tt.Next, null, extOnto, false);
if (res == null || res.Chars != prev.Chars)
{
return(null);
}
res.IsAfterConjunction = true;
res.Value = "& " + res.Value;
return(res);
}
if (!tt.Chars.IsLetter)
{
return(null);
}
List <Pullenti.Semantic.Utils.DerivateGroup> expinf = null;
if (prev != null && prev.EndToken.GetMorphClassInDictionary().IsNoun)
{
string wo = prev.EndToken.GetNormalCaseText(Pullenti.Morph.MorphClass.Noun, Pullenti.Morph.MorphNumber.Singular, Pullenti.Morph.MorphGender.Undefined, false);
expinf = Pullenti.Semantic.Utils.DerivateService.FindDerivates(wo, true, prev.EndToken.Morph.Language);
}
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null && npt.InternalNoun != null)
{
npt = null;
}
bool explOk = false;
if (npt != null && prev != null && prev.EndToken.GetMorphClassInDictionary().IsNoun)
{
Pullenti.Ner.Core.NounPhraseToken npt0 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(prev.EndToken, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt0 != null)
{
List <Pullenti.Semantic.Core.SemanticLink> links = Pullenti.Semantic.Core.SemanticHelper.TryCreateLinks(npt0, npt, null);
if (links.Count > 0)
{
explOk = true;
}
}
}
if (npt != null && ((explOk || npt.Morph.Case.IsGenitive || ((prev != null && !((prev.Morph.Case & npt.Morph.Case)).IsUndefined)))))
{
Pullenti.Morph.MorphClass mc = npt.BeginToken.GetMorphClassInDictionary();
if (mc.IsVerb || mc.IsPronoun)
{
return(null);
}
if (mc.IsAdverb)
{
if (npt.BeginToken.Next != null && npt.BeginToken.Next.IsHiphen)
{
}
else
{
return(null);
}
}
if (mc.IsPreposition)
{
return(null);
}
if (mc.IsNoun && npt.Chars.IsAllLower)
{
Pullenti.Morph.MorphCase ca = npt.Morph.Case;
if ((!ca.IsDative && !ca.IsGenitive && !ca.IsInstrumental) && !ca.IsPrepositional)
{
return(null);
}
}
res = new OrgItemNameToken(npt.BeginToken, npt.EndToken)
{
Morph = npt.Morph, Value = npt.GetNormalCaseText(null, Pullenti.Morph.MorphNumber.Undefined, Pullenti.Morph.MorphGender.Undefined, false)
};
res.IsNounPhrase = true;
if ((npt.EndToken.WhitespacesAfterCount < 2) && (npt.EndToken.Next is Pullenti.Ner.TextToken))
{
Pullenti.Ner.Core.NounPhraseToken npt2 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(npt.EndToken.Next, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt2 != null && npt2.Morph.Case.IsGenitive && npt2.Chars.IsAllLower)
{
OrgItemTypeToken typ = OrgItemTypeToken.TryAttach(npt.EndToken.Next, true, null);
OrgItemEponymToken epo = OrgItemEponymToken.TryAttach(npt.EndToken.Next, false);
Pullenti.Ner.ReferentToken rtt = t.Kit.ProcessReferent("PERSONPROPERTY", npt.EndToken.Next);
if (typ == null && epo == null && ((rtt == null || rtt.Morph.Number == Pullenti.Morph.MorphNumber.Plural)))
{
res.EndToken = npt2.EndToken;
res.Value = string.Format("{0} {1}", res.Value, Pullenti.Ner.Core.MiscHelper.GetTextValueOfMetaToken(npt2, Pullenti.Ner.Core.GetTextAttr.No));
}
}
else if (npt.EndToken.Next.IsComma && (npt.EndToken.Next.Next is Pullenti.Ner.TextToken))
{
Pullenti.Ner.Token tt2 = npt.EndToken.Next.Next;
Pullenti.Morph.MorphClass mv2 = tt2.GetMorphClassInDictionary();
if (mv2.IsAdjective && mv2.IsVerb)
{
Pullenti.Morph.MorphBaseInfo bi = new Pullenti.Morph.MorphBaseInfo()
{
Case = npt.Morph.Case, Gender = npt.Morph.Gender, Number = npt.Morph.Number
};
if (tt2.Morph.CheckAccord(bi, false, false))
{
npt2 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(tt2.Next, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt2 != null && ((npt2.Morph.Case.IsDative || npt2.Morph.Case.IsGenitive)) && npt2.Chars.IsAllLower)
{
res.EndToken = npt2.EndToken;
res.Value = string.Format("{0} {1}", res.Value, Pullenti.Ner.Core.MiscHelper.GetTextValue(npt.EndToken.Next, res.EndToken, Pullenti.Ner.Core.GetTextAttr.No));
}
}
}
}
}
if (explOk)
{
res.IsAfterConjunction = true;
}
}
else if (npt != null && ((((prev != null && prev.IsNounPhrase && npt.Morph.Case.IsInstrumental)) || extOnto)))
{
res = new OrgItemNameToken(npt.BeginToken, npt.EndToken)
{
Morph = npt.Morph, Value = npt.GetNormalCaseText(null, Pullenti.Morph.MorphNumber.Undefined, Pullenti.Morph.MorphGender.Undefined, false)
};
res.IsNounPhrase = true;
}
else if (tt.IsAnd)
{
res = TryAttach(tt.Next, prev, extOnto, false);
if (res == null || !res.IsNounPhrase || prev == null)
{
return(null);
}
if (((prev.Morph.Case & res.Morph.Case)).IsUndefined)
{
return(null);
}
if (prev.Morph.Number != Pullenti.Morph.MorphNumber.Undefined && res.Morph.Number != Pullenti.Morph.MorphNumber.Undefined)
{
if (((prev.Morph.Number & res.Morph.Number)) == Pullenti.Morph.MorphNumber.Undefined)
{
if (prev.Chars != res.Chars)
{
return(null);
}
OrgItemTypeToken ty = OrgItemTypeToken.TryAttach(res.EndToken.Next, false, null);
if (ty != null)
{
return(null);
}
}
}
Pullenti.Morph.CharsInfo ci = res.Chars;
res.Chars = ci;
res.IsAfterConjunction = true;
return(res);
}
else if (((tt.Term == "ПО" || tt.Term == "ПРИ" || tt.Term == "ЗА") || tt.Term == "С" || tt.Term == "В") || tt.Term == "НА")
{
npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t.Next, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null)
{
if (m_VervotWords.TryParse(npt.EndToken, Pullenti.Ner.Core.TerminParseAttr.No) != null)
{
return(null);
}
bool ok = false;
if (tt.Term == "ПО")
{
ok = npt.Morph.Case.IsDative;
}
else if (tt.Term == "С")
{
ok = npt.Morph.Case.IsInstrumental;
}
else if (tt.Term == "ЗА")
{
ok = npt.Morph.Case.IsGenitive | npt.Morph.Case.IsInstrumental;
}
else if (tt.Term == "НА")
{
ok = npt.Morph.Case.IsPrepositional;
}
else if (tt.Term == "В")
{
ok = npt.Morph.Case.IsDative | npt.Morph.Case.IsPrepositional;
if (ok)
{
ok = false;
if (t.Next.IsValue("СФЕРА", null) || t.Next.IsValue("ОБЛАСТЬ", null))
{
ok = true;
}
}
}
else if (tt.Term == "ПРИ")
{
ok = npt.Morph.Case.IsPrepositional;
if (ok)
{
if (OrgItemTypeToken.TryAttach(tt.Next, true, null) != null)
{
ok = false;
}
else
{
Pullenti.Ner.ReferentToken rt = tt.Kit.ProcessReferent(Pullenti.Ner.Org.OrganizationAnalyzer.ANALYZER_NAME, tt.Next);
if (rt != null)
{
ok = false;
}
}
}
string s = npt.Noun.GetNormalCaseText(null, Pullenti.Morph.MorphNumber.Undefined, Pullenti.Morph.MorphGender.Undefined, false);
if (s == "ПОДДЕРЖКА" || s == "УЧАСТИЕ")
{
ok = false;
}
}
else
{
ok = npt.Morph.Case.IsPrepositional;
}
if (ok)
{
res = new OrgItemNameToken(t, npt.EndToken)
{
Morph = npt.Morph, Value = npt.GetNormalCaseText(null, Pullenti.Morph.MorphNumber.Singular, Pullenti.Morph.MorphGender.Undefined, false), Chars = npt.Chars
};
res.IsNounPhrase = true;
res.Preposition = tt.Term;
if (((res.Value == "ДЕЛО" || res.Value == "ВОПРОС")) && !res.IsNewlineAfter)
{
OrgItemNameToken res2 = _TryAttach(res.EndToken.Next, res, extOnto);
if (res2 != null && res2.Morph.Case.IsGenitive)
{
res.Value = string.Format("{0} {1}", res.Value, res2.Value);
res.EndToken = res2.EndToken;
for (Pullenti.Ner.Token ttt = res2.EndToken.Next; ttt != null; ttt = ttt.Next)
{
if (!ttt.IsCommaAnd)
{
break;
}
OrgItemNameToken res3 = _TryAttach(ttt.Next, res2, extOnto);
if (res3 == null)
{
break;
}
res.Value = string.Format("{0} {1}", res.Value, res3.Value);
res.EndToken = res3.EndToken;
if (ttt.IsAnd)
{
break;
}
ttt = res.EndToken;
}
}
}
}
}
if (res == null)
{
return(null);
}
}
else if (tt.Term == "OF")
{
Pullenti.Ner.Token t1 = tt.Next;
if (t1 != null && Pullenti.Ner.Core.MiscHelper.IsEngArticle(t1))
{
t1 = t1.Next;
}
if (t1 != null && t1.Chars.IsLatinLetter && !t1.Chars.IsAllLower)
{
res = new OrgItemNameToken(t, t1)
{
Chars = t1.Chars, Morph = t1.Morph
};
for (Pullenti.Ner.Token ttt = t1.Next; ttt != null; ttt = ttt.Next)
{
if (ttt.WhitespacesBeforeCount > 2)
{
break;
}
if (Pullenti.Ner.Core.MiscHelper.IsEngAdjSuffix(ttt))
{
ttt = ttt.Next;
continue;
}
if (!ttt.Chars.IsLatinLetter)
{
break;
}
if (ttt.Morph.Class.IsPreposition)
{
break;
}
t1 = (res.EndToken = ttt);
}
res.Value = Pullenti.Ner.Core.MiscHelper.GetTextValue(t, t1, Pullenti.Ner.Core.GetTextAttr.IgnoreArticles);
res.Preposition = tt.Term;
return(res);
}
}
if (res == null)
{
if (tt.Chars.IsLatinLetter && tt.LengthChar == 1)
{
}
else if (tt.Chars.IsAllLower || (tt.LengthChar < 2))
{
if (!tt.Chars.IsLatinLetter || prev == null || !prev.Chars.IsLatinLetter)
{
return(null);
}
}
if (tt.Chars.IsCyrillicLetter)
{
Pullenti.Morph.MorphClass mc = tt.GetMorphClassInDictionary();
if (mc.IsVerb || mc.IsAdverb)
{
return(null);
}
}
else if (tt.Chars.IsLatinLetter && !tt.IsWhitespaceAfter)
{
if (!tt.IsWhitespaceAfter && (tt.LengthChar < 5))
{
if (tt.Next is Pullenti.Ner.NumberToken)
{
return(null);
}
}
}
res = new OrgItemNameToken(tt, tt)
{
Value = tt.Term, Morph = tt.Morph
};
for (t = tt.Next; t != null; t = t.Next)
{
if ((((t.IsHiphen || t.IsCharOf("\\/"))) && t.Next != null && (t.Next is Pullenti.Ner.TextToken)) && !t.IsWhitespaceBefore && !t.IsWhitespaceAfter)
{
t = t.Next;
res.EndToken = t;
res.Value = string.Format("{0}{1}{2}", res.Value, (t.Previous.IsChar('.') ? '.' : '-'), (t as Pullenti.Ner.TextToken).Term);
}
else if (t.IsChar('.'))
{
if (!t.IsWhitespaceAfter && !t.IsWhitespaceBefore && (t.Next is Pullenti.Ner.TextToken))
{
res.EndToken = t.Next;
t = t.Next;
res.Value = string.Format("{0}.{1}", res.Value, (t as Pullenti.Ner.TextToken).Term);
}
else if ((t.Next != null && !t.IsNewlineAfter && t.Next.Chars.IsLatinLetter) && tt.Chars.IsLatinLetter)
{
res.EndToken = t;
}
else
{
break;
}
}
else
{
break;
}
}
}
for (Pullenti.Ner.Token t0 = res.BeginToken; t0 != null; t0 = t0.Next)
{
if ((((tt = t0 as Pullenti.Ner.TextToken))) != null && tt.IsLetters)
{
if (!tt.Morph.Class.IsConjunction && !tt.Morph.Class.IsPreposition)
{
foreach (Pullenti.Morph.MorphBaseInfo mf in tt.Morph.Items)
{
if ((mf as Pullenti.Morph.MorphWordForm).IsInDictionary)
{
res.IsInDictionary = true;
}
}
}
}
if (t0 == res.EndToken)
{
break;
}
}
if (res.BeginToken == res.EndToken && res.BeginToken.Chars.IsAllUpper)
{
if (res.EndToken.Next != null && !res.EndToken.IsWhitespaceAfter)
{
Pullenti.Ner.Token t1 = res.EndToken.Next;
if (t1.Next != null && !t1.IsWhitespaceAfter && t1.IsHiphen)
{
t1 = t1.Next;
}
if (t1 is Pullenti.Ner.NumberToken)
{
res.Value += (t1 as Pullenti.Ner.NumberToken).Value;
res.EndToken = t1;
}
}
}
if (res.BeginToken == res.EndToken && res.BeginToken.Chars.IsLastLower)
{
string src = res.BeginToken.GetSourceText();
for (int i = src.Length - 1; i >= 0; i--)
{
if (char.IsUpper(src[i]))
{
res.Value = src.Substring(0, i + 1);
break;
}
}
}
return(res);
}