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);
}
public static List <Line> Parse(Pullenti.Ner.Token t0, int maxLines, int maxChars, int maxEndChar)
{
List <Line> res = new List <Line>();
int totalChars = 0;
for (Pullenti.Ner.Token t = t0; t != null; t = t.Next)
{
if (maxEndChar > 0)
{
if (t.BeginChar > maxEndChar)
{
break;
}
}
Pullenti.Ner.Token t1;
for (t1 = t; t1 != null && t1.Next != null; t1 = t1.Next)
{
if (t1.IsNewlineAfter)
{
if (t1.Next == null || Pullenti.Ner.Core.MiscHelper.CanBeStartOfSentence(t1.Next))
{
break;
}
}
if (t1 == t && t.IsNewlineBefore && (t.GetReferent() is Pullenti.Ner.Person.PersonReferent))
{
if (t1.Next == null)
{
continue;
}
if ((t1.Next is Pullenti.Ner.TextToken) && t1.Next.Chars.IsLetter && !t1.Next.Chars.IsAllLower)
{
break;
}
}
}
if (t1 == null)
{
t1 = t;
}
TitleItemToken tit = TitleItemToken.TryAttach(t);
if (tit != null)
{
if (tit.Typ == TitleItemToken.Types.Keywords)
{
break;
}
}
Pullenti.Ner.Core.Internal.BlockTitleToken bl = Pullenti.Ner.Core.Internal.BlockTitleToken.TryAttach(t, false, null);
if (bl != null)
{
if (bl.Typ != Pullenti.Ner.Core.Internal.BlkTyps.Undefined)
{
break;
}
}
Line l = new Line(t, t1);
res.Add(l);
totalChars += l.CharsCount;
if (res.Count >= maxLines || totalChars >= maxChars)
{
break;
}
t = t1;
}
return(res);
}