public static UriItemToken AttachISOContent(Pullenti.Ner.Token t0, string specChars)
{
Pullenti.Ner.Token t = t0;
while (true)
{
if (t == null)
{
return(null);
}
if (t.IsCharOf(":/\\") || t.IsHiphen || t.IsValue("IEC", null))
{
t = t.Next;
continue;
}
break;
}
if (!(t is Pullenti.Ner.NumberToken))
{
return(null);
}
Pullenti.Ner.Token t1 = t;
char delim = (char)0;
StringBuilder txt = new StringBuilder();
for (; t != null; t = t.Next)
{
if (t.IsWhitespaceBefore && t != t1)
{
break;
}
if (t is Pullenti.Ner.NumberToken)
{
if (delim != ((char)0))
{
txt.Append(delim);
}
delim = (char)0;
t1 = t;
txt.Append(t.GetSourceText());
continue;
}
if (!(t is Pullenti.Ner.TextToken))
{
break;
}
if (!t.IsCharOf(specChars))
{
break;
}
delim = t.GetSourceText()[0];
}
if (txt.Length == 0)
{
return(null);
}
return(new UriItemToken(t0, t1)
{
Value = txt.ToString()
});
}
bool CanBeStartOfDenom(Pullenti.Ner.Token t)
{
if ((t == null || !t.Chars.IsLetter || t.Next == null) || t.IsNewlineAfter)
{
return(false);
}
if (!(t is Pullenti.Ner.TextToken))
{
return(false);
}
if (t.LengthChar > 4)
{
return(false);
}
t = t.Next;
if (t.Chars.IsLetter)
{
return(false);
}
if (t is Pullenti.Ner.NumberToken)
{
return(true);
}
if (t.IsCharOf("/\\") || t.IsHiphen)
{
return(t.Next is Pullenti.Ner.NumberToken);
}
if (t.IsCharOf("+*&^#@!_"))
{
return(true);
}
return(false);
}
static int _calcNewlineBetweenCoef(FragToken fr1, FragToken fr2)
{
if (fr1.NewlinesAfterCount > 1)
{
return(1);
}
for (Pullenti.Ner.Token tt = fr1.BeginToken; tt != null && tt.EndChar <= fr1.EndChar; tt = tt.Next)
{
if (Pullenti.Ner.Core.BracketHelper.CanBeStartOfSequence(tt, false, false))
{
Pullenti.Ner.Core.BracketSequenceToken br = Pullenti.Ner.Core.BracketHelper.TryParse(tt, Pullenti.Ner.Core.BracketParseAttr.CanBeManyLines, 100);
if (br != null && br.EndChar >= fr2.BeginChar)
{
return(-1);
}
}
}
Pullenti.Ner.Token t = fr1.EndToken;
if (t.IsCharOf(":;."))
{
return(1);
}
if ((t is Pullenti.Ner.TextToken) && ((t.Morph.Class.IsPreposition || t.Morph.Class.IsConjunction)))
{
return(-1);
}
Pullenti.Ner.Token t1 = fr2.BeginToken;
if (t1 is Pullenti.Ner.TextToken)
{
if (t1.Chars.IsAllLower)
{
return(-1);
}
if (Pullenti.Ner.Core.BracketHelper.CanBeStartOfSequence(t1, false, false))
{
if (t.Chars.IsAllLower)
{
return(-1);
}
}
}
else if (t1 is Pullenti.Ner.NumberToken)
{
if (t.Chars.IsAllLower)
{
return(-1);
}
}
if (t.Chars.IsAllLower)
{
if (fr2.EndToken.IsChar(';'))
{
return(-1);
}
}
return(0);
}
/// <summary>
/// Проверка токена, что он является скобкой или кавычкой
/// </summary>
/// <param name="t">проверяемый токен</param>
/// <param name="quotesOnly">должны быть именно кавычка, а не скобка</param>
/// <return>да-нет</return>
public static bool IsBracket(Pullenti.Ner.Token t, bool quotesOnly = false)
{
if (t == null)
{
return(false);
}
if (t.IsCharOf(m_OpenChars))
{
if (quotesOnly)
{
if (t is Pullenti.Ner.TextToken)
{
if (m_Quotes.IndexOf((t as Pullenti.Ner.TextToken).Term[0]) < 0)
{
return(false);
}
}
}
return(true);
}
if (t.IsCharOf(m_CloseChars))
{
if (quotesOnly)
{
if (t is Pullenti.Ner.TextToken)
{
if (m_Quotes.IndexOf((t as Pullenti.Ner.TextToken).Term[0]) < 0)
{
return(false);
}
}
}
return(true);
}
return(false);
}
internal void AddValue(Pullenti.Ner.Token begin, Pullenti.Ner.Token end)
{
StringBuilder tmp = new StringBuilder();
for (Pullenti.Ner.Token t = begin; t != null && t.Previous != end; t = t.Next)
{
if (t is Pullenti.Ner.NumberToken)
{
tmp.Append(t.GetSourceText());
continue;
}
if (t is Pullenti.Ner.TextToken)
{
string s = (t as Pullenti.Ner.TextToken).Term;
if (t.IsCharOf("-\\/"))
{
s = "-";
}
tmp.Append(s);
}
}
for (int i = 0; i < tmp.Length; i++)
{
if (tmp[i] == '-' && i > 0 && ((i + 1) < tmp.Length))
{
char ch0 = tmp[i - 1];
char ch1 = tmp[i + 1];
if (char.IsLetterOrDigit(ch0) && char.IsLetterOrDigit(ch1))
{
if (char.IsDigit(ch0) && !char.IsDigit(ch1))
{
tmp.Remove(i, 1);
}
else if (!char.IsDigit(ch0) && char.IsDigit(ch1))
{
tmp.Remove(i, 1);
}
}
}
}
this.AddSlot(ATTR_VALUE, tmp.ToString(), false, 0);
m_Names = null;
}
static void _correctExtTypes(Pullenti.Ner.Core.NumberExToken ex)
{
Pullenti.Ner.Token t = ex.EndToken.Next;
if (t == null)
{
return;
}
Pullenti.Ner.Core.NumberExType ty = ex.ExTyp;
Pullenti.Ner.Token tt = _corrExTyp2(t, ref ty);
if (tt != null)
{
ex.ExTyp = ty;
ex.EndToken = tt;
t = tt.Next;
}
if (t == null || t.Next == null)
{
return;
}
if (t.IsCharOf("/\\") || t.IsValue("НА", null))
{
}
else
{
return;
}
Pullenti.Ner.Core.TerminToken tok = m_Postfixes.TryParse(t.Next, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok != null && (((Pullenti.Ner.Core.NumberExType)tok.Termin.Tag) != Pullenti.Ner.Core.NumberExType.Money))
{
ex.ExTyp2 = (Pullenti.Ner.Core.NumberExType)tok.Termin.Tag;
ex.EndToken = tok.EndToken;
ty = ex.ExTyp2;
tt = _corrExTyp2(ex.EndToken.Next, ref ty);
if (tt != null)
{
ex.ExTyp2 = ty;
ex.EndToken = tt;
t = tt.Next;
}
}
}
static Pullenti.Ner.Core.NumberExToken _attachSpecPostfix(Pullenti.Ner.Token t)
{
if (t == null)
{
return(null);
}
if (t.IsCharOf("%"))
{
return(new Pullenti.Ner.Core.NumberExToken(t, t, "", Pullenti.Ner.NumberSpellingType.Digit, Pullenti.Ner.Core.NumberExType.Percent));
}
string money = Pullenti.Ner.Core.NumberHelper.IsMoneyChar(t);
if (money != null)
{
return new Pullenti.Ner.Core.NumberExToken(t, t, "", Pullenti.Ner.NumberSpellingType.Digit, Pullenti.Ner.Core.NumberExType.Money)
{
ExTypParam = money
}
}
;
return(null);
}
public static CanonicDecreeRefUri TryCreateCanonicDecreeRefUri(Pullenti.Ner.Token t)
{
if (!(t is Pullenti.Ner.ReferentToken))
{
return(null);
}
Pullenti.Ner.Decree.DecreeReferent dr = t.GetReferent() as Pullenti.Ner.Decree.DecreeReferent;
CanonicDecreeRefUri res;
if (dr != null)
{
if (dr.Kind == Pullenti.Ner.Decree.DecreeKind.Publisher)
{
return(null);
}
res = new CanonicDecreeRefUri(t.Kit.Sofa.Text)
{
Ref = dr, BeginChar = t.BeginChar, EndChar = t.EndChar
};
if ((t.Previous != null && t.Previous.IsChar('(') && t.Next != null) && t.Next.IsChar(')'))
{
return(res);
}
if ((t as Pullenti.Ner.ReferentToken).MiscAttrs != 0)
{
return(res);
}
Pullenti.Ner.ReferentToken rt = t as Pullenti.Ner.ReferentToken;
if (rt.BeginToken.IsChar('(') && rt.EndToken.IsChar(')'))
{
res = new CanonicDecreeRefUri(t.Kit.Sofa.Text)
{
Ref = dr, BeginChar = rt.BeginToken.Next.BeginChar, EndChar = rt.EndToken.Previous.EndChar
};
return(res);
}
List <DecreeToken> nextDecreeItems = null;
if ((t.Next != null && t.Next.IsCommaAnd && (t.Next.Next is Pullenti.Ner.ReferentToken)) && (t.Next.Next.GetReferent() is Pullenti.Ner.Decree.DecreeReferent))
{
nextDecreeItems = DecreeToken.TryAttachList((t.Next.Next as Pullenti.Ner.ReferentToken).BeginToken, null, 10, false);
if (nextDecreeItems != null && nextDecreeItems.Count > 1)
{
for (int i = 0; i < (nextDecreeItems.Count - 1); i++)
{
if (nextDecreeItems[i].IsNewlineAfter)
{
nextDecreeItems.RemoveRange(i + 1, nextDecreeItems.Count - i - 1);
break;
}
}
}
}
bool wasTyp = false;
bool wasNum = false;
for (Pullenti.Ner.Token tt = (t as Pullenti.Ner.MetaToken).BeginToken; tt != null && tt.EndChar <= t.EndChar; tt = tt.Next)
{
if (tt.BeginChar == t.BeginChar && tt.IsChar('(') && tt.Next != null)
{
res.BeginChar = tt.Next.BeginChar;
}
if (tt.IsChar('(') && tt.Next != null && tt.Next.IsValue("ДАЛЕЕ", null))
{
if (res.EndChar >= tt.BeginChar)
{
res.EndChar = tt.Previous.EndChar;
}
break;
}
if (tt.EndChar == t.EndChar && tt.IsChar(')'))
{
res.EndChar = tt.Previous.EndChar;
for (Pullenti.Ner.Token tt1 = tt.Previous; tt1 != null && tt1.BeginChar >= res.BeginChar; tt1 = tt1.Previous)
{
if (tt1.IsChar('(') && tt1.Previous != null)
{
if (res.BeginChar < tt1.Previous.BeginChar)
{
res.EndChar = tt1.Previous.EndChar;
}
}
}
}
List <DecreeToken> li = DecreeToken.TryAttachList(tt, null, 10, false);
if (li != null && li.Count > 0)
{
for (int ii = 0; ii < (li.Count - 1); ii++)
{
if (li[ii].Typ == DecreeToken.ItemType.Typ && li[ii + 1].Typ == DecreeToken.ItemType.Terr)
{
res.TypeWithGeo = Pullenti.Ner.Core.MiscHelper.GetTextValue(li[ii].BeginToken, li[ii + 1].EndToken, Pullenti.Ner.Core.GetTextAttr.FirstNounGroupToNominativeSingle);
}
}
if ((nextDecreeItems != null && nextDecreeItems.Count > 1 && (nextDecreeItems.Count < li.Count)) && nextDecreeItems[0].Typ != DecreeToken.ItemType.Typ)
{
int d = li.Count - nextDecreeItems.Count;
int j;
for (j = 0; j < nextDecreeItems.Count; j++)
{
if (nextDecreeItems[j].Typ != li[d + j].Typ)
{
break;
}
}
if (j >= nextDecreeItems.Count)
{
li.RemoveRange(0, d);
res.BeginChar = li[0].BeginChar;
}
}
else if ((nextDecreeItems != null && nextDecreeItems.Count == 1 && nextDecreeItems[0].Typ == DecreeToken.ItemType.Name) && li.Count == 2 && li[1].Typ == DecreeToken.ItemType.Name)
{
res.BeginChar = li[1].BeginChar;
res.EndChar = li[1].EndChar;
break;
}
else if ((nextDecreeItems != null && nextDecreeItems.Count == 1 && nextDecreeItems[0].Typ == DecreeToken.ItemType.Number) && li[li.Count - 1].Typ == DecreeToken.ItemType.Number)
{
res.BeginChar = li[li.Count - 1].BeginChar;
res.EndChar = li[li.Count - 1].EndChar;
}
for (int i = 0; i < li.Count; i++)
{
DecreeToken l = li[i];
if (l.BeginChar > t.EndChar)
{
li.RemoveRange(i, li.Count - i);
break;
}
if (l.Typ == DecreeToken.ItemType.Name)
{
if (!wasNum)
{
if (dr.Kind == Pullenti.Ner.Decree.DecreeKind.Contract)
{
continue;
}
if (((i + 1) < li.Count) && ((li[i + 1].Typ == DecreeToken.ItemType.Date || li[i + 1].Typ == DecreeToken.ItemType.Number)))
{
continue;
}
}
int ee = l.BeginToken.Previous.EndChar;
if (ee > res.BeginChar && (ee < res.EndChar))
{
res.EndChar = ee;
}
break;
}
if (l.Typ == DecreeToken.ItemType.Number)
{
wasNum = true;
}
if (i == 0)
{
if (l.Typ == DecreeToken.ItemType.Typ)
{
wasTyp = true;
}
else if (l.Typ == DecreeToken.ItemType.Owner || l.Typ == DecreeToken.ItemType.Org)
{
if (((i + 1) < li.Count) && ((li[1].Typ == DecreeToken.ItemType.Date || li[1].Typ == DecreeToken.ItemType.Number)))
{
wasTyp = true;
}
}
if (wasTyp)
{
Pullenti.Ner.Token tt0 = l.BeginToken.Previous;
if (tt0 != null && tt0.IsChar('.'))
{
tt0 = tt0.Previous;
}
if (tt0 != null && ((tt0.IsValue("УТВЕРЖДЕННЫЙ", null) || tt0.IsValue("УТВЕРДИТЬ", null) || tt0.IsValue("УТВ", null))))
{
if (l.BeginChar > res.BeginChar)
{
res.BeginChar = l.BeginChar;
if (res.EndChar < res.BeginChar)
{
res.EndChar = t.EndChar;
}
res.IsAdopted = true;
}
}
}
}
}
if (li.Count > 0)
{
tt = li[li.Count - 1].EndToken;
if (tt.IsChar(')'))
{
tt = tt.Previous;
}
continue;
}
}
if (wasTyp)
{
DecreeToken na = DecreeToken.TryAttachName(tt, dr.Typ0, true, false);
if (na != null && tt.BeginChar > t.BeginChar)
{
Pullenti.Ner.Token tt1 = na.EndToken.Next;
if (tt1 != null && tt1.IsCharOf(",()"))
{
tt1 = tt1.Next;
}
if (tt1 != null && (tt1.EndChar < t.EndChar))
{
if (tt1.IsValue("УТВЕРЖДЕННЫЙ", null) || tt1.IsValue("УТВЕРДИТЬ", null) || tt1.IsValue("УТВ", null))
{
tt = tt1;
continue;
}
}
if (tt.Previous != null && tt.Previous.IsChar(':') && na.EndChar <= res.EndChar)
{
res.BeginChar = tt.BeginChar;
break;
}
if (tt.Previous.EndChar > res.BeginChar)
{
res.EndChar = tt.Previous.EndChar;
break;
}
}
}
}
return(res);
}
Pullenti.Ner.Decree.DecreePartReferent dpr = t.GetReferent() as Pullenti.Ner.Decree.DecreePartReferent;
if (dpr == null)
{
return(null);
}
if ((t.Previous != null && t.Previous.IsHiphen && (t.Previous.Previous is Pullenti.Ner.ReferentToken)) && (t.Previous.Previous.GetReferent() is Pullenti.Ner.Decree.DecreePartReferent))
{
if (Pullenti.Ner.Decree.DecreePartReferent.CreateRangeReferent(t.Previous.Previous.GetReferent() as Pullenti.Ner.Decree.DecreePartReferent, dpr) != null)
{
return(null);
}
}
Pullenti.Ner.Token t1 = t;
bool hasDiap = false;
Pullenti.Ner.ReferentToken DiapRef = null;
if ((t.Next != null && t.Next.IsHiphen && (t.Next.Next is Pullenti.Ner.ReferentToken)) && (t.Next.Next.GetReferent() is Pullenti.Ner.Decree.DecreePartReferent))
{
Pullenti.Ner.Decree.DecreePartReferent diap = Pullenti.Ner.Decree.DecreePartReferent.CreateRangeReferent(dpr as Pullenti.Ner.Decree.DecreePartReferent, t.Next.Next.GetReferent() as Pullenti.Ner.Decree.DecreePartReferent);
if (diap != null)
{
dpr = diap;
hasDiap = true;
t1 = t.Next.Next;
DiapRef = t1 as Pullenti.Ner.ReferentToken;
}
}
res = new CanonicDecreeRefUri(t.Kit.Sofa.Text)
{
Ref = dpr, BeginChar = t.BeginChar, EndChar = t1.EndChar, IsDiap = hasDiap
};
if ((t.Previous != null && t.Previous.IsChar('(') && t1.Next != null) && t1.Next.IsChar(')'))
{
return(res);
}
for (Pullenti.Ner.Token tt = (t as Pullenti.Ner.MetaToken).BeginToken; tt != null && tt.EndChar <= t.EndChar; tt = tt.Next)
{
if (tt.GetReferent() is Pullenti.Ner.Decree.DecreeReferent)
{
if (tt.BeginChar > t.BeginChar)
{
res.EndChar = tt.Previous.EndChar;
if (tt.Previous.Morph.Class.IsPreposition && tt.Previous.Previous != null)
{
res.EndChar = tt.Previous.Previous.EndChar;
}
}
else if (tt.EndChar < t.EndChar)
{
res.BeginChar = tt.BeginChar;
}
break;
}
}
bool hasSameBefore = _hasSameDecree(t, dpr, true);
bool hasSameAfter = _hasSameDecree(t, dpr, false);
PartToken.ItemType ptmin = PartToken.ItemType.Prefix;
PartToken.ItemType ptmin2 = PartToken.ItemType.Prefix;
int max = 0;
int max2 = 00;
foreach (Pullenti.Ner.Slot s in dpr.Slots)
{
PartToken.ItemType pt = PartToken._getTypeByAttrName(s.TypeName);
if (pt == PartToken.ItemType.Prefix)
{
continue;
}
int co = PartToken._getRank(pt);
if (co < 1)
{
if (pt == PartToken.ItemType.Part && dpr.FindSlot(Pullenti.Ner.Decree.DecreePartReferent.ATTR_CLAUSE, null, true) != null)
{
co = PartToken._getRank(PartToken.ItemType.Paragraph);
}
else
{
continue;
}
}
if (co > max)
{
max2 = max;
ptmin2 = ptmin;
max = co;
ptmin = pt;
}
else if (co > max2)
{
max2 = co;
ptmin2 = pt;
}
}
if (ptmin != PartToken.ItemType.Prefix)
{
for (Pullenti.Ner.Token tt = (t as Pullenti.Ner.MetaToken).BeginToken; tt != null && tt.EndChar <= res.EndChar; tt = tt.Next)
{
if (tt.BeginChar >= res.BeginChar)
{
PartToken pt = PartToken.TryAttach(tt, null, false, false);
if (pt != null && pt.Typ == ptmin)
{
res.BeginChar = pt.BeginChar;
res.EndChar = pt.EndChar;
if (pt.Typ == PartToken.ItemType.Appendix && pt.EndToken.IsValue("К", null) && pt.BeginToken != pt.EndToken)
{
res.EndChar = pt.EndToken.Previous.EndChar;
}
if (pt.EndChar == t.EndChar)
{
if ((t.Next != null && t.Next.IsCommaAnd && (t.Next.Next is Pullenti.Ner.ReferentToken)) && (t.Next.Next.GetReferent() is Pullenti.Ner.Decree.DecreePartReferent))
{
Pullenti.Ner.Token tt1 = (t.Next.Next as Pullenti.Ner.ReferentToken).BeginToken;
bool ok = true;
if (tt1.Chars.IsLetter)
{
ok = false;
}
if (ok)
{
foreach (PartToken.PartValue v in pt.Values)
{
res.BeginChar = v.BeginChar;
res.EndChar = v.EndChar;
break;
}
}
}
}
if (!hasDiap)
{
return(res);
}
break;
}
}
}
if (hasDiap && DiapRef != null)
{
for (Pullenti.Ner.Token tt = DiapRef.BeginToken; tt != null && tt.EndChar <= DiapRef.EndChar; tt = tt.Next)
{
if (tt.IsChar(','))
{
break;
}
if (tt != DiapRef.BeginToken && tt.IsWhitespaceBefore)
{
break;
}
res.EndChar = tt.EndChar;
}
return(res);
}
}
if (((hasSameBefore || hasSameAfter)) && ptmin != PartToken.ItemType.Prefix)
{
for (Pullenti.Ner.Token tt = (t as Pullenti.Ner.MetaToken).BeginToken; tt != null && tt.EndChar <= res.EndChar; tt = tt.Next)
{
if (tt.BeginChar >= res.BeginChar)
{
PartToken pt = (!hasSameBefore ? PartToken.TryAttach(tt, null, false, false) : null);
if (pt != null)
{
if (pt.Typ == ptmin)
{
foreach (PartToken.PartValue v in pt.Values)
{
res.BeginChar = v.BeginChar;
res.EndChar = v.EndChar;
return(res);
}
}
tt = pt.EndToken;
continue;
}
if ((tt is Pullenti.Ner.NumberToken) && tt.BeginChar == res.BeginChar)
{
res.EndChar = tt.EndChar;
for (; tt != null && tt.Next != null;)
{
if (!tt.Next.IsChar('.') || tt.IsWhitespaceAfter || tt.Next.IsWhitespaceAfter)
{
break;
}
if (!(tt.Next.Next is Pullenti.Ner.NumberToken))
{
break;
}
tt = tt.Next.Next;
res.EndChar = tt.EndChar;
}
if (tt.Next != null && tt.Next.IsHiphen)
{
if (tt.Next.Next is Pullenti.Ner.NumberToken)
{
tt = tt.Next.Next;
res.EndChar = tt.EndChar;
for (; tt != null && tt.Next != null;)
{
if (!tt.Next.IsChar('.') || tt.IsWhitespaceAfter || tt.Next.IsWhitespaceAfter)
{
break;
}
if (!(tt.Next.Next is Pullenti.Ner.NumberToken))
{
break;
}
tt = tt.Next.Next;
res.EndChar = tt.EndChar;
}
}
else if (tt.Next.Next != null && (tt.Next.Next.GetReferent() is Pullenti.Ner.Decree.DecreePartReferent) && hasDiap)
{
res.EndChar = (tt.Next.Next as Pullenti.Ner.MetaToken).BeginToken.EndChar;
}
}
return(res);
}
if (Pullenti.Ner.Core.BracketHelper.CanBeStartOfSequence(tt, true, false) && tt.BeginChar == res.BeginChar && hasSameBefore)
{
Pullenti.Ner.Core.BracketSequenceToken br = Pullenti.Ner.Core.BracketHelper.TryParse(tt, Pullenti.Ner.Core.BracketParseAttr.No, 100);
if (br != null && br.EndToken.Previous == tt.Next)
{
res.EndChar = br.EndChar;
return(res);
}
}
}
}
return(res);
}
if (!hasSameBefore && !hasSameAfter && ptmin != PartToken.ItemType.Prefix)
{
for (Pullenti.Ner.Token tt = (t as Pullenti.Ner.MetaToken).BeginToken; tt != null && tt.EndChar <= res.EndChar; tt = tt.Next)
{
if (tt.BeginChar >= res.BeginChar)
{
List <PartToken> pts = PartToken.TryAttachList(tt, false, 40);
if (pts == null || pts.Count == 0)
{
break;
}
for (int i = 0; i < pts.Count; i++)
{
if (pts[i].Typ == ptmin)
{
res.BeginChar = pts[i].BeginChar;
res.EndChar = pts[i].EndChar;
tt = pts[i].EndToken;
if (tt.Next != null && tt.Next.IsHiphen)
{
if (tt.Next.Next is Pullenti.Ner.NumberToken)
{
res.EndChar = tt.Next.Next.EndChar;
}
else if (tt.Next.Next != null && (tt.Next.Next.GetReferent() is Pullenti.Ner.Decree.DecreePartReferent) && hasDiap)
{
res.EndChar = (tt.Next.Next as Pullenti.Ner.MetaToken).BeginToken.EndChar;
}
}
return(res);
}
}
}
}
}
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);
}
public static Pullenti.Ner.Core.NumberExToken TryParseNumberWithPostfix(Pullenti.Ner.Token t)
{
if (t == null)
{
return(null);
}
Pullenti.Ner.Token t0 = t;
string isDollar = null;
if (t.LengthChar == 1 && t.Next != null)
{
if ((((isDollar = Pullenti.Ner.Core.NumberHelper.IsMoneyChar(t)))) != null)
{
t = t.Next;
}
}
Pullenti.Ner.NumberToken nt = t as Pullenti.Ner.NumberToken;
if (nt == null)
{
if ((!(t.Previous is Pullenti.Ner.NumberToken) && t.IsChar('(') && (t.Next is Pullenti.Ner.NumberToken)) && t.Next.Next != null && t.Next.Next.IsChar(')'))
{
Pullenti.Ner.Core.TerminToken toks1 = m_Postfixes.TryParse(t.Next.Next.Next, Pullenti.Ner.Core.TerminParseAttr.No);
if (toks1 != null && ((Pullenti.Ner.Core.NumberExType)toks1.Termin.Tag) == Pullenti.Ner.Core.NumberExType.Money)
{
Pullenti.Ner.NumberToken nt0 = t.Next as Pullenti.Ner.NumberToken;
Pullenti.Ner.Core.NumberExToken res = new Pullenti.Ner.Core.NumberExToken(t, toks1.EndToken, nt0.Value, nt0.Typ, Pullenti.Ner.Core.NumberExType.Money)
{
AltRealValue = nt0.RealValue, Morph = toks1.BeginToken.Morph
};
return(_correctMoney(res, toks1.BeginToken));
}
}
Pullenti.Ner.TextToken tt = t as Pullenti.Ner.TextToken;
if (tt == null || !tt.Morph.Class.IsAdjective)
{
return(null);
}
string val = tt.Term;
for (int i = 4; i < (val.Length - 5); i++)
{
string v = val.Substring(0, i);
List <Pullenti.Ner.Core.Termin> li = Pullenti.Ner.Core.NumberHelper.m_Nums.FindTerminsByString(v, tt.Morph.Language);
if (li == null)
{
continue;
}
string vv = val.Substring(i);
List <Pullenti.Ner.Core.Termin> lii = m_Postfixes.FindTerminsByString(vv, tt.Morph.Language);
if (lii != null && lii.Count > 0)
{
Pullenti.Ner.Core.NumberExToken re = new Pullenti.Ner.Core.NumberExToken(t, t, ((int)li[0].Tag).ToString(), Pullenti.Ner.NumberSpellingType.Words, (Pullenti.Ner.Core.NumberExType)lii[0].Tag)
{
Morph = t.Morph
};
_correctExtTypes(re);
return(re);
}
break;
}
return(null);
}
if (t.Next == null && isDollar == null)
{
return(null);
}
double f = nt.RealValue;
if (double.IsNaN(f))
{
return(null);
}
Pullenti.Ner.Token t1 = nt.Next;
if (((t1 != null && t1.IsCharOf(",."))) || (((t1 is Pullenti.Ner.NumberToken) && (t1.WhitespacesBeforeCount < 3))))
{
double d;
Pullenti.Ner.NumberToken tt11 = Pullenti.Ner.Core.NumberHelper.TryParseRealNumber(nt, false, false);
if (tt11 != null)
{
t1 = tt11.EndToken.Next;
f = tt11.RealValue;
}
}
if (t1 == null)
{
if (isDollar == null)
{
return(null);
}
}
else if ((t1.Next != null && t1.Next.IsValue("С", "З") && t1.Next.Next != null) && t1.Next.Next.IsValue("ПОЛОВИНА", null))
{
f += 0.5;
t1 = t1.Next.Next;
}
if (t1 != null && t1.IsHiphen && t1.Next != null)
{
t1 = t1.Next;
}
bool det = false;
double altf = f;
if (((t1 is Pullenti.Ner.NumberToken) && t1.Previous != null && t1.Previous.IsHiphen) && (t1 as Pullenti.Ner.NumberToken).IntValue == 0 && t1.LengthChar == 2)
{
t1 = t1.Next;
}
if ((t1 != null && t1.Next != null && t1.IsChar('(')) && (((t1.Next is Pullenti.Ner.NumberToken) || t1.Next.IsValue("НОЛЬ", null))) && t1.Next.Next != null)
{
Pullenti.Ner.NumberToken nt1 = t1.Next as Pullenti.Ner.NumberToken;
double val = (double)0;
if (nt1 != null)
{
val = nt1.RealValue;
}
if (Math.Floor(f) == Math.Floor(val))
{
Pullenti.Ner.Token ttt = t1.Next.Next;
if (ttt.IsChar(')'))
{
t1 = ttt.Next;
det = true;
if ((t1 is Pullenti.Ner.NumberToken) && (t1 as Pullenti.Ner.NumberToken).IntValue != null && (t1 as Pullenti.Ner.NumberToken).IntValue.Value == 0)
{
t1 = t1.Next;
}
}
else if (((((ttt is Pullenti.Ner.NumberToken) && ((ttt as Pullenti.Ner.NumberToken).RealValue < 100) && ttt.Next != null) && ttt.Next.IsChar('/') && ttt.Next.Next != null) && ttt.Next.Next.GetSourceText() == "100" && ttt.Next.Next.Next != null) && ttt.Next.Next.Next.IsChar(')'))
{
int rest = GetDecimalRest100(f);
if ((ttt as Pullenti.Ner.NumberToken).IntValue != null && rest == (ttt as Pullenti.Ner.NumberToken).IntValue.Value)
{
t1 = ttt.Next.Next.Next.Next;
det = true;
}
}
else if ((ttt.IsValue("ЦЕЛЫХ", null) && (ttt.Next is Pullenti.Ner.NumberToken) && ttt.Next.Next != null) && ttt.Next.Next.Next != null && ttt.Next.Next.Next.IsChar(')'))
{
Pullenti.Ner.NumberToken num2 = ttt.Next as Pullenti.Ner.NumberToken;
altf = num2.RealValue;
if (ttt.Next.Next.IsValue("ДЕСЯТЫЙ", null))
{
altf /= 10;
}
else if (ttt.Next.Next.IsValue("СОТЫЙ", null))
{
altf /= 100;
}
else if (ttt.Next.Next.IsValue("ТЫСЯЧНЫЙ", null))
{
altf /= 1000;
}
else if (ttt.Next.Next.IsValue("ДЕСЯТИТЫСЯЧНЫЙ", null))
{
altf /= 10000;
}
else if (ttt.Next.Next.IsValue("СТОТЫСЯЧНЫЙ", null))
{
altf /= 100000;
}
else if (ttt.Next.Next.IsValue("МИЛЛИОННЫЙ", null))
{
altf /= 1000000;
}
if (altf < 1)
{
altf += val;
t1 = ttt.Next.Next.Next.Next;
det = true;
}
}
else
{
Pullenti.Ner.Core.TerminToken toks1 = m_Postfixes.TryParse(ttt, Pullenti.Ner.Core.TerminParseAttr.No);
if (toks1 != null)
{
if (((Pullenti.Ner.Core.NumberExType)toks1.Termin.Tag) == Pullenti.Ner.Core.NumberExType.Money)
{
if (toks1.EndToken.Next != null && toks1.EndToken.Next.IsChar(')'))
{
Pullenti.Ner.Core.NumberExToken res = new Pullenti.Ner.Core.NumberExToken(t, toks1.EndToken.Next, nt.Value, nt.Typ, Pullenti.Ner.Core.NumberExType.Money)
{
RealValue = f, AltRealValue = altf, Morph = toks1.BeginToken.Morph
};
return(_correctMoney(res, toks1.BeginToken));
}
}
}
Pullenti.Ner.Core.NumberExToken res2 = TryParseNumberWithPostfix(t1.Next);
if (res2 != null && res2.EndToken.Next != null && res2.EndToken.Next.IsChar(')'))
{
res2.BeginToken = t;
res2.EndToken = res2.EndToken.Next;
res2.AltRealValue = res2.RealValue;
res2.RealValue = f;
_correctExtTypes(res2);
if (res2.WhitespacesAfterCount < 2)
{
Pullenti.Ner.Core.TerminToken toks2 = m_Postfixes.TryParse(res2.EndToken.Next, Pullenti.Ner.Core.TerminParseAttr.No);
if (toks2 != null)
{
if (((Pullenti.Ner.Core.NumberExType)toks2.Termin.Tag) == Pullenti.Ner.Core.NumberExType.Money)
{
res2.EndToken = toks2.EndToken;
}
}
}
return(res2);
}
}
}
else if (nt1 != null && nt1.Typ == Pullenti.Ner.NumberSpellingType.Words && nt.Typ == Pullenti.Ner.NumberSpellingType.Digit)
{
altf = nt1.RealValue;
Pullenti.Ner.Token ttt = t1.Next.Next;
if (ttt.IsChar(')'))
{
t1 = ttt.Next;
det = true;
}
if (!det)
{
altf = f;
}
}
}
if ((t1 != null && t1.IsChar('(') && t1.Next != null) && t1.Next.IsValue("СУММА", null))
{
Pullenti.Ner.Core.BracketSequenceToken br = Pullenti.Ner.Core.BracketHelper.TryParse(t1, Pullenti.Ner.Core.BracketParseAttr.No, 100);
if (br != null)
{
t1 = br.EndToken.Next;
}
}
if (isDollar != null)
{
Pullenti.Ner.Token te = null;
if (t1 != null)
{
te = t1.Previous;
}
else
{
for (t1 = t0; t1 != null; t1 = t1.Next)
{
if (t1.Next == null)
{
te = t1;
}
}
}
if (te == null)
{
return(null);
}
if (te.IsHiphen && te.Next != null)
{
if (te.Next.IsValue("МИЛЛИОННЫЙ", null))
{
f *= 1000000;
altf *= 1000000;
te = te.Next;
}
else if (te.Next.IsValue("МИЛЛИАРДНЫЙ", null))
{
f *= 1000000000;
altf *= 1000000000;
te = te.Next;
}
}
if (!te.IsWhitespaceAfter && (te.Next is Pullenti.Ner.TextToken))
{
if (te.Next.IsValue("M", null))
{
f *= 1000000;
altf *= 1000000;
te = te.Next;
}
else if (te.Next.IsValue("BN", null))
{
f *= 1000000000;
altf *= 1000000000;
te = te.Next;
}
}
return(new Pullenti.Ner.Core.NumberExToken(t0, te, "", nt.Typ, Pullenti.Ner.Core.NumberExType.Money)
{
RealValue = f, AltRealValue = altf, ExTypParam = isDollar
});
}
if (t1 == null || ((t1.IsNewlineBefore && !det)))
{
return(null);
}
Pullenti.Ner.Core.TerminToken toks = m_Postfixes.TryParse(t1, Pullenti.Ner.Core.TerminParseAttr.No);
if ((toks == null && det && (t1 is Pullenti.Ner.NumberToken)) && (t1 as Pullenti.Ner.NumberToken).Value == "0")
{
toks = m_Postfixes.TryParse(t1.Next, Pullenti.Ner.Core.TerminParseAttr.No);
}
if (toks == null && t1.IsChar('р'))
{
int cou = 10;
for (Pullenti.Ner.Token ttt = t0.Previous; ttt != null && cou > 0; ttt = ttt.Previous, cou--)
{
if (ttt.IsValue("СУММА", null) || ttt.IsValue("НАЛИЧНЫЙ", null) || ttt.IsValue("БАЛАНС", null))
{
}
else if (ttt.GetReferent() != null && ttt.GetReferent().TypeName == "MONEY")
{
}
else
{
continue;
}
toks = new Pullenti.Ner.Core.TerminToken(t1, t1)
{
Termin = m_Postfixes.FindTerminsByCanonicText("RUB")[0]
};
if (t1.Next != null && t1.Next.IsChar('.'))
{
toks.EndToken = t1.Next;
}
Pullenti.Ner.Core.NumberExType ty = (Pullenti.Ner.Core.NumberExType)toks.Termin.Tag;
return(new Pullenti.Ner.Core.NumberExToken(t, toks.EndToken, nt.Value, nt.Typ, ty)
{
RealValue = f, AltRealValue = altf, Morph = toks.BeginToken.Morph, ExTypParam = "RUB"
});
}
}
if (toks != null)
{
t1 = toks.EndToken;
if (!t1.IsChar('.') && t1.Next != null && t1.Next.IsChar('.'))
{
if ((t1 is Pullenti.Ner.TextToken) && t1.IsValue(toks.Termin.Terms[0].CanonicalText, null))
{
}
else if (!t1.Chars.IsLetter)
{
}
else
{
t1 = t1.Next;
}
}
if (toks.Termin.CanonicText == "LTL")
{
return(null);
}
if (toks.BeginToken == t1)
{
if (t1.Morph.Class.IsPreposition || t1.Morph.Class.IsConjunction)
{
if (t1.IsWhitespaceBefore && t1.IsWhitespaceAfter)
{
return(null);
}
}
}
Pullenti.Ner.Core.NumberExType ty = (Pullenti.Ner.Core.NumberExType)toks.Termin.Tag;
Pullenti.Ner.Core.NumberExToken res = new Pullenti.Ner.Core.NumberExToken(t, t1, nt.Value, nt.Typ, ty)
{
RealValue = f, AltRealValue = altf, Morph = toks.BeginToken.Morph
};
if (ty != Pullenti.Ner.Core.NumberExType.Money)
{
_correctExtTypes(res);
return(res);
}
return(_correctMoney(res, toks.BeginToken));
}
Pullenti.Ner.Core.NumberExToken pfx = _attachSpecPostfix(t1);
if (pfx != null)
{
pfx.BeginToken = t;
pfx.Value = nt.Value;
pfx.Typ = nt.Typ;
pfx.RealValue = f;
pfx.AltRealValue = altf;
return(pfx);
}
if (t1.Next != null && ((t1.Morph.Class.IsPreposition || t1.Morph.Class.IsConjunction)))
{
if (t1.IsValue("НА", null))
{
}
else
{
Pullenti.Ner.Core.NumberExToken nn = TryParseNumberWithPostfix(t1.Next);
if (nn != null)
{
return new Pullenti.Ner.Core.NumberExToken(t, t, nt.Value, nt.Typ, nn.ExTyp)
{
RealValue = f, AltRealValue = altf, ExTyp2 = nn.ExTyp2, ExTypParam = nn.ExTypParam
}
}
;
}
}
if (!t1.IsWhitespaceAfter && (t1.Next is Pullenti.Ner.NumberToken) && (t1 is Pullenti.Ner.TextToken))
{
string term = (t1 as Pullenti.Ner.TextToken).Term;
Pullenti.Ner.Core.NumberExType ty = Pullenti.Ner.Core.NumberExType.Undefined;
if (term == "СМХ" || term == "CMX")
{
ty = Pullenti.Ner.Core.NumberExType.Santimeter;
}
else if (term == "MX" || term == "МХ")
{
ty = Pullenti.Ner.Core.NumberExType.Meter;
}
else if (term == "MMX" || term == "ММХ")
{
ty = Pullenti.Ner.Core.NumberExType.Millimeter;
}
if (ty != Pullenti.Ner.Core.NumberExType.Undefined)
{
return new Pullenti.Ner.Core.NumberExToken(t, t1, nt.Value, nt.Typ, ty)
{
RealValue = f, AltRealValue = altf, MultAfter = true
}
}
;
}
return(null);
}
public static PhoneItemToken TryAttachAlternate(Pullenti.Ner.Token t0, Pullenti.Ner.Phone.PhoneReferent ph0, List <PhoneItemToken> pli)
{
if (t0 == null)
{
return(null);
}
if (t0.IsCharOf("\\/") && (t0.Next is Pullenti.Ner.NumberToken) && (t0.Next.EndChar - t0.Next.BeginChar) <= 1)
{
List <PhoneItemToken> pli1 = PhoneItemToken.TryAttachAll(t0.Next, 15);
if (pli1 != null && pli1.Count > 1)
{
if (pli1[pli1.Count - 1].ItemType == PhoneItemType.Delim)
{
pli1.RemoveAt(pli1.Count - 1);
}
if (pli1.Count <= pli.Count)
{
int ii;
string num = "";
for (ii = 0; ii < pli1.Count; ii++)
{
PhoneItemToken p1 = pli1[ii];
PhoneItemToken p0 = pli[(pli.Count - pli1.Count) + ii];
if (p1.ItemType != p0.ItemType)
{
break;
}
if (p1.ItemType != PhoneItemType.Number && p1.ItemType != PhoneItemType.Delim)
{
break;
}
if (p1.ItemType == PhoneItemType.Number)
{
if (p1.LengthChar != p0.LengthChar)
{
break;
}
num += p1.Value;
}
}
if (ii >= pli1.Count)
{
return new PhoneItemToken(t0, pli1[pli1.Count - 1].EndToken)
{
ItemType = PhoneItemType.Alt, Value = num
}
}
;
}
}
return(new PhoneItemToken(t0, t0.Next)
{
ItemType = PhoneItemType.Alt, Value = t0.Next.GetSourceText()
});
}
if (t0.IsHiphen && (t0.Next is Pullenti.Ner.NumberToken) && (t0.Next.EndChar - t0.Next.BeginChar) <= 1)
{
Pullenti.Ner.Token t1 = t0.Next.Next;
bool ok = false;
if (t1 == null)
{
ok = true;
}
else if (t1.IsNewlineBefore || t1.IsCharOf(",."))
{
ok = true;
}
if (ok)
{
return new PhoneItemToken(t0, t0.Next)
{
ItemType = PhoneItemType.Alt, Value = t0.Next.GetSourceText()
}
}
;
}
if ((t0.IsChar('(') && (t0.Next is Pullenti.Ner.NumberToken) && (t0.Next.EndChar - t0.Next.BeginChar) == 1) && t0.Next.Next != null && t0.Next.Next.IsChar(')'))
{
return new PhoneItemToken(t0, t0.Next.Next)
{
ItemType = PhoneItemType.Alt, Value = t0.Next.GetSourceText()
}
}
;
if ((t0.IsCharOf("/-") && (t0.Next is Pullenti.Ner.NumberToken) && ph0.m_Template != null) && Pullenti.Morph.LanguageHelper.EndsWith(ph0.m_Template, (((t0.Next.EndChar - t0.Next.BeginChar) + 1)).ToString()))
{
return new PhoneItemToken(t0, t0.Next)
{
ItemType = PhoneItemType.Alt, Value = t0.Next.GetSourceText()
}
}
;
return(null);
}
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 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);
}
static TitleItemToken TryAttachSpeciality(Pullenti.Ner.Token t, bool keyWordBefore)
{
if (t == null)
{
return(null);
}
bool susp = false;
if (!keyWordBefore)
{
if (!t.IsNewlineBefore)
{
susp = true;
}
}
StringBuilder val = null;
Pullenti.Ner.Token t0 = t;
int digCount = 0;
for (int i = 0; i < 3; i++)
{
Pullenti.Ner.NumberToken nt = t as Pullenti.Ner.NumberToken;
if (nt == null)
{
break;
}
if (nt.Typ != Pullenti.Ner.NumberSpellingType.Digit || nt.Morph.Class.IsAdjective)
{
break;
}
if (val == null)
{
val = new StringBuilder();
}
if (susp && t.LengthChar != 2)
{
return(null);
}
string digs = nt.GetSourceText();
digCount += digs.Length;
val.Append(digs);
if (t.Next == null)
{
break;
}
t = t.Next;
if (t.IsCharOf(".,") || t.IsHiphen)
{
if (susp && (i < 2))
{
if (!t.IsChar('.') || t.IsWhitespaceAfter || t.IsWhitespaceBefore)
{
return(null);
}
}
if (t.Next != null)
{
t = t.Next;
}
}
}
if (val == null || (digCount < 5))
{
return(null);
}
if (digCount != 6)
{
if (!keyWordBefore)
{
return(null);
}
}
else
{
val.Insert(4, '.');
val.Insert(2, '.');
}
for (Pullenti.Ner.Token tt = t.Next; tt != null; tt = tt.Next)
{
if (tt.IsNewlineBefore)
{
break;
}
Pullenti.Ner.Core.BracketSequenceToken br = Pullenti.Ner.Core.BracketHelper.TryParse(tt, Pullenti.Ner.Core.BracketParseAttr.No, 100);
if (br != null)
{
t = (tt = br.EndToken);
continue;
}
t = tt;
}
return(new TitleItemToken(t0, t, Types.Speciality)
{
Value = val.ToString()
});
}
public static Pullenti.Ner.ReferentToken TryAttach(Pullenti.Ner.Token t)
{
if (t == null || !t.Chars.IsLetter)
{
return(null);
}
PersonIdToken noun = TryParse(t, null);
if (noun == null)
{
return(null);
}
List <PersonIdToken> li = new List <PersonIdToken>();
for (t = noun.EndToken.Next; t != null; t = t.Next)
{
if (t.IsTableControlChar)
{
break;
}
if (t.IsCharOf(",:"))
{
continue;
}
PersonIdToken idt = TryParse(t, (li.Count > 0 ? li[li.Count - 1] : noun));
if (idt == null)
{
if (t.IsValue("ОТДЕЛ", null) || t.IsValue("ОТДЕЛЕНИЕ", null))
{
continue;
}
break;
}
if (idt.Typ == Typs.Keyword)
{
break;
}
li.Add(idt);
t = idt.EndToken;
}
if (li.Count == 0)
{
return(null);
}
string num = null;
int i = 0;
if (li[0].Typ == Typs.Number)
{
if (li.Count > 1 && li[1].Typ == Typs.Number && li[1].HasPrefix)
{
num = li[0].Value + li[1].Value;
i = 2;
}
else
{
num = li[0].Value;
i = 1;
}
}
else if (li[0].Typ == Typs.Seria && li.Count > 1 && li[1].Typ == Typs.Number)
{
num = li[0].Value + li[1].Value;
i = 2;
}
else if (li[0].Typ == Typs.Seria && li[0].Value.Length > 5)
{
num = li[0].Value;
i = 1;
}
else
{
return(null);
}
Pullenti.Ner.Person.PersonIdentityReferent pid = new Pullenti.Ner.Person.PersonIdentityReferent();
pid.Typ = noun.Value.ToLower();
pid.Number = num;
if (noun.Referent is Pullenti.Ner.Geo.GeoReferent)
{
pid.State = noun.Referent;
}
for (; i < li.Count; i++)
{
if (li[i].Typ == Typs.Vidan || li[i].Typ == Typs.Code)
{
}
else if (li[i].Typ == Typs.Date && li[i].Referent != null)
{
if (pid.FindSlot(Pullenti.Ner.Person.PersonIdentityReferent.ATTR_DATE, null, true) != null)
{
break;
}
pid.AddSlot(Pullenti.Ner.Person.PersonIdentityReferent.ATTR_DATE, li[i].Referent, false, 0);
}
else if (li[i].Typ == Typs.Address && li[i].Referent != null)
{
if (pid.FindSlot(Pullenti.Ner.Person.PersonIdentityReferent.ATTR_ADDRESS, null, true) != null)
{
break;
}
pid.AddSlot(Pullenti.Ner.Person.PersonIdentityReferent.ATTR_ADDRESS, li[i].Referent, false, 0);
}
else if (li[i].Typ == Typs.Org && li[i].Referent != null)
{
if (pid.FindSlot(Pullenti.Ner.Person.PersonIdentityReferent.ATTR_ORG, null, true) != null)
{
break;
}
pid.AddSlot(Pullenti.Ner.Person.PersonIdentityReferent.ATTR_ORG, li[i].Referent, false, 0);
}
else
{
break;
}
}
return(new Pullenti.Ner.ReferentToken(pid, noun.BeginToken, li[i - 1].EndToken));
}
static PersonIdToken TryParse(Pullenti.Ner.Token t, PersonIdToken prev)
{
if (t.IsValue("СВИДЕТЕЛЬСТВО", null))
{
Pullenti.Ner.Token tt1 = t;
bool ip = false;
bool reg = false;
for (Pullenti.Ner.Token tt = t.Next; tt != null; tt = tt.Next)
{
if (tt.IsCommaAnd || tt.Morph.Class.IsPreposition)
{
continue;
}
if (tt.IsValue("РЕГИСТРАЦИЯ", null) || tt.IsValue("РЕЕСТР", null) || tt.IsValue("ЗАРЕГИСТРИРОВАТЬ", null))
{
reg = true;
tt1 = tt;
}
else if (tt.IsValue("ИНДИВИДУАЛЬНЫЙ", null) || tt.IsValue("ИП", null))
{
ip = true;
tt1 = tt;
}
else if ((tt.IsValue("ВНЕСЕНИЕ", null) || tt.IsValue("ГОСУДАРСТВЕННЫЙ", null) || tt.IsValue("ЕДИНЫЙ", null)) || tt.IsValue("ЗАПИСЬ", null) || tt.IsValue("ПРЕДПРИНИМАТЕЛЬ", null))
{
tt1 = tt;
}
else if (tt.GetReferent() != null && tt.GetReferent().TypeName == "DATERANGE")
{
tt1 = tt;
}
else
{
break;
}
}
if (reg && ip)
{
return new PersonIdToken(t, tt1)
{
Typ = Typs.Keyword, Value = "СВИДЕТЕЛЬСТВО О ГОСУДАРСТВЕННОЙ РЕГИСТРАЦИИ ФИЗИЧЕСКОГО ЛИЦА В КАЧЕСТВЕ ИНДИВИДУАЛЬНОГО ПРЕДПРИНИМАТЕЛЯ"
}
}
;
}
Pullenti.Ner.Core.TerminToken tok = m_Ontology.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok != null)
{
Typs ty = (Typs)tok.Termin.Tag;
PersonIdToken res = new PersonIdToken(tok.BeginToken, tok.EndToken)
{
Typ = ty, Value = tok.Termin.CanonicText
};
if (prev == null)
{
if (ty != Typs.Keyword)
{
return(null);
}
for (t = tok.EndToken.Next; t != null; t = t.Next)
{
Pullenti.Ner.Referent r = t.GetReferent();
if (r != null && (r is Pullenti.Ner.Geo.GeoReferent))
{
res.Referent = r;
res.EndToken = t;
continue;
}
if (t.IsValue("ГРАЖДАНИН", null) && t.Next != null && (t.Next.GetReferent() is Pullenti.Ner.Geo.GeoReferent))
{
res.Referent = t.Next.GetReferent();
t = (res.EndToken = t.Next);
continue;
}
if (r != null)
{
break;
}
PersonAttrToken ait = PersonAttrToken.TryAttach(t, null, PersonAttrToken.PersonAttrAttachAttrs.No);
if (ait != null)
{
if (ait.Referent != null)
{
foreach (Pullenti.Ner.Slot s in ait.Referent.Slots)
{
if (s.TypeName == Pullenti.Ner.Person.PersonPropertyReferent.ATTR_REF && (s.Value is Pullenti.Ner.Geo.GeoReferent))
{
res.Referent = s.Value as Pullenti.Ner.Referent;
}
}
}
res.EndToken = ait.EndToken;
break;
}
if (t.IsValue("ДАННЫЙ", null))
{
res.EndToken = t;
continue;
}
break;
}
if ((res.Referent is Pullenti.Ner.Geo.GeoReferent) && !(res.Referent as Pullenti.Ner.Geo.GeoReferent).IsState)
{
res.Referent = null;
}
return(res);
}
if (ty == Typs.Number)
{
StringBuilder tmp = new StringBuilder();
Pullenti.Ner.Token tt = tok.EndToken.Next;
if (tt != null && tt.IsChar(':'))
{
tt = tt.Next;
}
for (; tt != null; tt = tt.Next)
{
if (tt.IsNewlineBefore)
{
break;
}
if (!(tt is Pullenti.Ner.NumberToken))
{
break;
}
tmp.Append(tt.GetSourceText());
res.EndToken = tt;
}
if (tmp.Length < 1)
{
return(null);
}
res.Value = tmp.ToString();
res.HasPrefix = true;
return(res);
}
if (ty == Typs.Seria)
{
StringBuilder tmp = new StringBuilder();
Pullenti.Ner.Token tt = tok.EndToken.Next;
if (tt != null && tt.IsChar(':'))
{
tt = tt.Next;
}
bool nextNum = false;
for (; tt != null; tt = tt.Next)
{
if (tt.IsNewlineBefore)
{
break;
}
if (Pullenti.Ner.Core.MiscHelper.CheckNumberPrefix(tt) != null)
{
nextNum = true;
break;
}
if (!(tt is Pullenti.Ner.NumberToken))
{
if (!(tt is Pullenti.Ner.TextToken))
{
break;
}
if (!tt.Chars.IsAllUpper)
{
break;
}
Pullenti.Ner.NumberToken nu = Pullenti.Ner.Core.NumberHelper.TryParseRoman(tt);
if (nu != null)
{
tmp.Append(nu.GetSourceText());
tt = nu.EndToken;
}
else if (tt.LengthChar != 2)
{
break;
}
else
{
tmp.Append((tt as Pullenti.Ner.TextToken).Term);
res.EndToken = tt;
}
if (tt.Next != null && tt.Next.IsHiphen)
{
tt = tt.Next;
}
continue;
}
if (tmp.Length >= 4)
{
break;
}
tmp.Append(tt.GetSourceText());
res.EndToken = tt;
}
if (tmp.Length < 4)
{
if (tmp.Length < 2)
{
return(null);
}
Pullenti.Ner.Token tt1 = res.EndToken.Next;
if (tt1 != null && tt1.IsComma)
{
tt1 = tt1.Next;
}
PersonIdToken next = TryParse(tt1, res);
if (next != null && next.Typ == Typs.Number)
{
}
else
{
return(null);
}
}
res.Value = tmp.ToString();
res.HasPrefix = true;
return(res);
}
if (ty == Typs.Code)
{
for (Pullenti.Ner.Token tt = res.EndToken.Next; tt != null; tt = tt.Next)
{
if (tt.IsCharOf(":") || tt.IsHiphen)
{
continue;
}
if (tt is Pullenti.Ner.NumberToken)
{
res.EndToken = tt;
continue;
}
break;
}
}
if (ty == Typs.Address)
{
if (t.GetReferent() is Pullenti.Ner.Address.AddressReferent)
{
res.Referent = t.GetReferent();
res.EndToken = t;
return(res);
}
for (Pullenti.Ner.Token tt = res.EndToken.Next; tt != null; tt = tt.Next)
{
if (tt.IsCharOf(":") || tt.IsHiphen || tt.Morph.Class.IsPreposition)
{
continue;
}
if (tt.GetReferent() is Pullenti.Ner.Address.AddressReferent)
{
res.Referent = tt.GetReferent();
res.EndToken = tt;
}
break;
}
if (res.Referent == null)
{
return(null);
}
}
return(res);
}
else if (prev == null)
{
return(null);
}
Pullenti.Ner.Token t0 = t;
Pullenti.Ner.Token t1 = Pullenti.Ner.Core.MiscHelper.CheckNumberPrefix(t0);
if (t1 != null)
{
t = t1;
}
if (t is Pullenti.Ner.NumberToken)
{
StringBuilder tmp = new StringBuilder();
PersonIdToken res = new PersonIdToken(t0, t)
{
Typ = Typs.Number
};
for (Pullenti.Ner.Token tt = t; tt != null; tt = tt.Next)
{
if (tt.IsNewlineBefore || !(tt is Pullenti.Ner.NumberToken))
{
break;
}
tmp.Append(tt.GetSourceText());
res.EndToken = tt;
}
if (tmp.Length < 4)
{
if (tmp.Length < 2)
{
return(null);
}
if (prev == null || prev.Typ != Typs.Keyword)
{
return(null);
}
PersonIdToken ne = TryParse(res.EndToken.Next, prev);
if (ne != null && ne.Typ == Typs.Number)
{
res.Typ = Typs.Seria;
}
else
{
return(null);
}
}
res.Value = tmp.ToString();
if (t0 != t)
{
res.HasPrefix = true;
}
return(res);
}
if (t is Pullenti.Ner.ReferentToken)
{
Pullenti.Ner.Referent r = t.GetReferent();
if (r != null)
{
if (r.TypeName == "DATE")
{
return new PersonIdToken(t, t)
{
Typ = Typs.Date, Referent = r
}
}
;
if (r.TypeName == "ORGANIZATION")
{
return new PersonIdToken(t, t)
{
Typ = Typs.Org, Referent = r
}
}
;
if (r.TypeName == "ADDRESS")
{
return new PersonIdToken(t, t)
{
Typ = Typs.Address, Referent = r
}
}
;
}
}
if ((prev != null && prev.Typ == Typs.Keyword && (t is Pullenti.Ner.TextToken)) && !t.Chars.IsAllLower && t.Chars.IsLetter)
{
PersonIdToken rr = TryParse(t.Next, prev);
if (rr != null && rr.Typ == Typs.Number)
{
return new PersonIdToken(t, t)
{
Typ = Typs.Seria, Value = (t as Pullenti.Ner.TextToken).Term
}
}
;
}
if ((t != null && t.IsValue("ОТ", "ВІД") && (t.Next is Pullenti.Ner.ReferentToken)) && t.Next.GetReferent().TypeName == "DATE")
{
return new PersonIdToken(t, t.Next)
{
Typ = Typs.Date, Referent = t.Next.GetReferent()
}
}
;
return(null);
}
static BookLinkToken _tryParse(Pullenti.Ner.Token t, int lev)
{
if (t == null || lev > 3)
{
return(null);
}
if (t.IsChar('['))
{
BookLinkToken re = _tryParse(t.Next, lev + 1);
if (re != null && re.EndToken.Next != null && re.EndToken.Next.IsChar(']'))
{
re.BeginToken = t;
re.EndToken = re.EndToken.Next;
return(re);
}
if (re != null && re.EndToken.IsChar(']'))
{
re.BeginToken = t;
return(re);
}
if (re != null)
{
if (re.Typ == BookLinkTyp.Sostavitel || re.Typ == BookLinkTyp.Editors)
{
return(re);
}
}
Pullenti.Ner.Core.BracketSequenceToken br = Pullenti.Ner.Core.BracketHelper.TryParse(t, Pullenti.Ner.Core.BracketParseAttr.No, 100);
if (br != null)
{
if ((br.EndToken.Previous is Pullenti.Ner.NumberToken) && (br.LengthChar < 30))
{
return new BookLinkToken(t, br.EndToken)
{
Typ = BookLinkTyp.Number, Value = Pullenti.Ner.Core.MiscHelper.GetTextValue(br.BeginToken.Next, br.EndToken.Previous, Pullenti.Ner.Core.GetTextAttr.No)
}
}
;
}
}
Pullenti.Ner.Token t0 = t;
if (t is Pullenti.Ner.ReferentToken)
{
if (t.GetReferent() is Pullenti.Ner.Person.PersonReferent)
{
return(TryParseAuthor(t, Pullenti.Ner.Person.Internal.FioTemplateType.Undefined));
}
if (t.GetReferent() is Pullenti.Ner.Geo.GeoReferent)
{
return new BookLinkToken(t, t)
{
Typ = BookLinkTyp.Geo, Ref = t.GetReferent()
}
}
;
if (t.GetReferent() is Pullenti.Ner.Date.DateReferent)
{
Pullenti.Ner.Date.DateReferent dr = t.GetReferent() as Pullenti.Ner.Date.DateReferent;
if (dr.Slots.Count == 1 && dr.Year > 0)
{
return new BookLinkToken(t, t)
{
Typ = BookLinkTyp.Year, Value = dr.Year.ToString()
}
}
;
if (dr.Year > 0 && t.Previous != null && t.Previous.IsComma)
{
return new BookLinkToken(t, t)
{
Typ = BookLinkTyp.Year, Value = dr.Year.ToString()
}
}
;
}
if (t.GetReferent() is Pullenti.Ner.Org.OrganizationReferent)
{
Pullenti.Ner.Org.OrganizationReferent org = t.GetReferent() as Pullenti.Ner.Org.OrganizationReferent;
if (org.Kind == Pullenti.Ner.Org.OrganizationKind.Press)
{
return new BookLinkToken(t, t)
{
Typ = BookLinkTyp.Press, Ref = org
}
}
;
}
if (t.GetReferent() is Pullenti.Ner.Uri.UriReferent)
{
Pullenti.Ner.Uri.UriReferent uri = t.GetReferent() as Pullenti.Ner.Uri.UriReferent;
if ((uri.Scheme == "http" || uri.Scheme == "https" || uri.Scheme == "ftp") || uri.Scheme == null)
{
return new BookLinkToken(t, t)
{
Typ = BookLinkTyp.Url, Ref = uri
}
}
;
}
}
Pullenti.Ner.Core.TerminToken tok = m_Termins.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok != null)
{
BookLinkTyp typ = (BookLinkTyp)tok.Termin.Tag;
bool ok = true;
if (typ == BookLinkTyp.Type || typ == BookLinkTyp.NameTail || typ == BookLinkTyp.ElectronRes)
{
if (t.Previous != null && ((t.Previous.IsCharOf(".:[") || t.Previous.IsHiphen)))
{
}
else
{
ok = false;
}
}
if (ok)
{
return new BookLinkToken(t, tok.EndToken)
{
Typ = typ, Value = tok.Termin.CanonicText
}
}
;
if (typ == BookLinkTyp.ElectronRes)
{
for (Pullenti.Ner.Token tt = tok.EndToken.Next; tt != null; tt = tt.Next)
{
if ((tt is Pullenti.Ner.TextToken) && !tt.Chars.IsLetter)
{
continue;
}
if (tt.GetReferent() is Pullenti.Ner.Uri.UriReferent)
{
return new BookLinkToken(t, tt)
{
Typ = BookLinkTyp.ElectronRes, Ref = tt.GetReferent()
}
}
;
break;
}
}
}
if (t.IsChar('/'))
{
BookLinkToken res = new BookLinkToken(t, t)
{
Typ = BookLinkTyp.Delimeter, Value = "/"
};
if (t.Next != null && t.Next.IsChar('/'))
{
res.EndToken = t.Next;
res.Value = "//";
}
if (!t.IsWhitespaceBefore && !t.IsWhitespaceAfter)
{
int coo = 3;
bool no = true;
for (Pullenti.Ner.Token tt = t.Next; tt != null && coo > 0; tt = tt.Next, coo--)
{
BookLinkToken vvv = TryParse(tt, lev + 1);
if (vvv != null && vvv.Typ != BookLinkTyp.Number)
{
no = false;
break;
}
}
if (no)
{
return(null);
}
}
return(res);
}
if ((t is Pullenti.Ner.NumberToken) && (t as Pullenti.Ner.NumberToken).IntValue != null && (t as Pullenti.Ner.NumberToken).Typ == Pullenti.Ner.NumberSpellingType.Digit)
{
BookLinkToken res = new BookLinkToken(t, t)
{
Typ = BookLinkTyp.Number, Value = (t as Pullenti.Ner.NumberToken).Value.ToString()
};
int val = (t as Pullenti.Ner.NumberToken).IntValue.Value;
if (val >= 1930 && (val < 2030))
{
res.Typ = BookLinkTyp.Year;
}
if (t.Next != null && t.Next.IsChar('.'))
{
res.EndToken = t.Next;
}
else if ((t.Next != null && t.Next.LengthChar == 1 && !t.Next.Chars.IsLetter) && t.Next.IsWhitespaceAfter)
{
res.EndToken = t.Next;
}
else if (t.Next is Pullenti.Ner.TextToken)
{
string term = (t.Next as Pullenti.Ner.TextToken).Term;
if (((term == "СТР" || term == "C" || term == "С") || term == "P" || term == "S") || term == "PAGES")
{
res.EndToken = t.Next;
res.Typ = BookLinkTyp.Pages;
res.Value = (t as Pullenti.Ner.NumberToken).Value.ToString();
}
}
return(res);
}
if (t is Pullenti.Ner.TextToken)
{
string term = (t as Pullenti.Ner.TextToken).Term;
if (((((((term == "СТР" || term == "C" || term == "С") || term == "ТОМ" || term == "T") || term == "Т" || term == "P") || term == "PP" || term == "V") || term == "VOL" || term == "S") || term == "СТОР" || t.IsValue("PAGE", null)) || t.IsValue("СТРАНИЦА", "СТОРІНКА"))
{
Pullenti.Ner.Token tt = t.Next;
while (tt != null)
{
if (tt.IsCharOf(".:~"))
{
tt = tt.Next;
}
else
{
break;
}
}
if (tt is Pullenti.Ner.NumberToken)
{
BookLinkToken res = new BookLinkToken(t, tt)
{
Typ = BookLinkTyp.PageRange
};
Pullenti.Ner.Token tt0 = tt;
Pullenti.Ner.Token tt1 = tt;
for (tt = tt.Next; tt != null; tt = tt.Next)
{
if (tt.IsCharOf(",") || tt.IsHiphen)
{
if (tt.Next is Pullenti.Ner.NumberToken)
{
tt = tt.Next;
res.EndToken = tt;
tt1 = tt;
continue;
}
}
break;
}
res.Value = Pullenti.Ner.Core.MiscHelper.GetTextValue(tt0, tt1, Pullenti.Ner.Core.GetTextAttr.No);
return(res);
}
}
if ((term == "M" || term == "М" || term == "СПБ") || term == "K" || term == "К")
{
if (t.Next != null && t.Next.IsCharOf(":;"))
{
BookLinkToken re = new BookLinkToken(t, t.Next)
{
Typ = BookLinkTyp.Geo
};
return(re);
}
if (t.Next != null && t.Next.IsCharOf("."))
{
BookLinkToken res = new BookLinkToken(t, t.Next)
{
Typ = BookLinkTyp.Geo
};
if (t.Next.Next != null && t.Next.Next.IsCharOf(":;"))
{
res.EndToken = t.Next.Next;
}
else if (t.Next.Next != null && (t.Next.Next is Pullenti.Ner.NumberToken))
{
}
else if (t.Next.Next != null && t.Next.Next.IsComma && (t.Next.Next.Next is Pullenti.Ner.NumberToken))
{
}
else
{
return(null);
}
return(res);
}
}
if (term == "ПЕР" || term == "ПЕРЕВ" || term == "ПЕРЕВОД")
{
Pullenti.Ner.Token tt = t;
if (tt.Next != null && tt.Next.IsChar('.'))
{
tt = tt.Next;
}
if (tt.Next != null && ((tt.Next.IsValue("C", null) || tt.Next.IsValue("С", null))))
{
tt = tt.Next;
if (tt.Next == null || tt.WhitespacesAfterCount > 2)
{
return(null);
}
BookLinkToken re = new BookLinkToken(t, tt.Next)
{
Typ = BookLinkTyp.Translate
};
return(re);
}
}
if (term == "ТАМ" || term == "ТАМЖЕ")
{
BookLinkToken res = new BookLinkToken(t, t)
{
Typ = BookLinkTyp.Tamze
};
if (t.Next != null && t.Next.IsValue("ЖЕ", null))
{
res.EndToken = t.Next;
}
return(res);
}
if (((term == "СМ" || term == "CM" || term == "НАПР") || term == "НАПРИМЕР" || term == "SEE") || term == "ПОДРОБНЕЕ" || term == "ПОДРОБНО")
{
BookLinkToken res = new BookLinkToken(t, t)
{
Typ = BookLinkTyp.See
};
for (t = t.Next; t != null; t = t.Next)
{
if (t.IsCharOf(".:") || t.IsValue("ALSO", null))
{
res.EndToken = t;
continue;
}
if (t.IsValue("В", null) || t.IsValue("IN", null))
{
res.EndToken = t;
continue;
}
BookLinkToken vvv = _tryParse(t, lev + 1);
if (vvv != null && vvv.Typ == BookLinkTyp.See)
{
res.EndToken = vvv.EndToken;
break;
}
break;
}
return(res);
}
if (term == "БОЛЕЕ")
{
BookLinkToken vvv = _tryParse(t.Next, lev + 1);
if (vvv != null && vvv.Typ == BookLinkTyp.See)
{
vvv.BeginToken = t;
return(vvv);
}
}
Pullenti.Ner.Token no = Pullenti.Ner.Core.MiscHelper.CheckNumberPrefix(t);
if (no is Pullenti.Ner.NumberToken)
{
return new BookLinkToken(t, no)
{
Typ = BookLinkTyp.N
}
}
;
if (((term == "B" || term == "В")) && (t.Next is Pullenti.Ner.NumberToken) && (t.Next.Next is Pullenti.Ner.TextToken))
{
string term2 = (t.Next.Next as Pullenti.Ner.TextToken).Term;
if (((term2 == "Т" || term2 == "T" || term2.StartsWith("ТОМ")) || term2 == "TT" || term2 == "ТТ") || term2 == "КН" || term2.StartsWith("КНИГ"))
{
return new BookLinkToken(t, t.Next.Next)
{
Typ = BookLinkTyp.Volume
}
}
;
}
}
if (t.IsChar('('))
{
if (((t.Next is Pullenti.Ner.NumberToken) && (t.Next as Pullenti.Ner.NumberToken).IntValue != null && t.Next.Next != null) && t.Next.Next.IsChar(')'))
{
int num = (t.Next as Pullenti.Ner.NumberToken).IntValue.Value;
if (num > 1900 && num <= 2040)
{
if (num <= DateTime.Now.Year)
{
return new BookLinkToken(t, t.Next.Next)
{
Typ = BookLinkTyp.Year, Value = num.ToString()
}
}
;
}
}
if (((t.Next is Pullenti.Ner.ReferentToken) && (t.Next.GetReferent() is Pullenti.Ner.Date.DateReferent) && t.Next.Next != null) && t.Next.Next.IsChar(')'))
{
int num = (t.Next.GetReferent() as Pullenti.Ner.Date.DateReferent).Year;
if (num > 0)
{
return new BookLinkToken(t, t.Next.Next)
{
Typ = BookLinkTyp.Year, Value = num.ToString()
}
}
;
}
}
return(null);
}
Pullenti.Ner.ReferentToken TryAttach(Pullenti.Ner.Token t, bool keyWord)
{
if (t == null)
{
return(null);
}
Pullenti.Ner.Token t0 = t;
Pullenti.Ner.Token t1 = t;
List <string> urisKeys = null;
List <Pullenti.Ner.Uri.UriReferent> uris = null;
Pullenti.Ner.Referent org = null;
Pullenti.Ner.Referent corOrg = null;
bool orgIsBank = false;
int empty = 0;
Pullenti.Ner.Uri.UriReferent lastUri = null;
for (; t != null; t = t.Next)
{
if (t.IsTableControlChar && t != t0)
{
break;
}
if (t.IsComma || t.Morph.Class.IsPreposition || t.IsCharOf("/\\"))
{
continue;
}
bool bankKeyword = false;
if (t.IsValue("ПОЛНЫЙ", null) && t.Next != null && ((t.Next.IsValue("НАИМЕНОВАНИЕ", null) || t.Next.IsValue("НАЗВАНИЕ", null))))
{
t = t.Next.Next;
if (t == null)
{
break;
}
}
if (t.IsValue("БАНК", null))
{
if ((t is Pullenti.Ner.ReferentToken) && t.GetReferent().TypeName == "ORGANIZATION")
{
bankKeyword = true;
}
Pullenti.Ner.Token tt = t.Next;
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(tt, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null)
{
tt = npt.EndToken.Next;
}
if (tt != null && tt.IsChar(':'))
{
tt = tt.Next;
}
if (tt != null)
{
if (!bankKeyword)
{
t = tt;
bankKeyword = true;
}
else if (tt.GetReferent() != null && tt.GetReferent().TypeName == "ORGANIZATION")
{
t = tt;
}
}
}
Pullenti.Ner.Referent r = t.GetReferent();
if (r != null && r.TypeName == "ORGANIZATION")
{
bool isBank = false;
int kk = 0;
for (Pullenti.Ner.Referent rr = r; rr != null && (kk < 4); rr = rr.ParentReferent, kk++)
{
isBank = string.Compare(rr.GetStringValue("KIND") ?? "", "Bank", true) == 0;
if (isBank)
{
break;
}
}
if (!isBank && bankKeyword)
{
isBank = true;
}
if (!isBank && uris != null && urisKeys.Contains("ИНН"))
{
return(null);
}
if ((lastUri != null && lastUri.Scheme == "К/С" && t.Previous != null) && t.Previous.IsValue("В", null))
{
corOrg = r;
t1 = t;
}
else if (org == null || ((!orgIsBank && isBank)))
{
org = r;
t1 = t;
orgIsBank = isBank;
if (isBank)
{
continue;
}
}
if (uris == null && !keyWord)
{
return(null);
}
continue;
}
if (r is Pullenti.Ner.Uri.UriReferent)
{
Pullenti.Ner.Uri.UriReferent u = r as Pullenti.Ner.Uri.UriReferent;
if (uris == null)
{
if (!_isBankReq(u.Scheme))
{
return(null);
}
if (u.Scheme == "ИНН" && t.IsNewlineAfter)
{
return(null);
}
uris = new List <Pullenti.Ner.Uri.UriReferent>();
urisKeys = new List <string>();
}
else
{
if (!_isBankReq(u.Scheme))
{
break;
}
if (urisKeys.Contains(u.Scheme))
{
break;
}
if (u.Scheme == "ИНН")
{
if (empty > 0)
{
break;
}
}
}
urisKeys.Add(u.Scheme);
uris.Add(u);
lastUri = u;
t1 = t;
empty = 0;
continue;
}
else if (uris == null && !keyWord && !orgIsBank)
{
return(null);
}
if (r != null && ((r.TypeName == "GEO" || r.TypeName == "ADDRESS")))
{
empty++;
continue;
}
if (t is Pullenti.Ner.TextToken)
{
if (t.IsValue("ПОЛНЫЙ", null) || t.IsValue("НАИМЕНОВАНИЕ", null) || t.IsValue("НАЗВАНИЕ", null))
{
}
else if (t.Chars.IsLetter)
{
Pullenti.Ner.Core.TerminToken tok = m_Ontology.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok != null)
{
t = tok.EndToken;
empty = 0;
}
else
{
empty++;
if (t.IsNewlineBefore)
{
Pullenti.Ner.Core.NounPhraseToken nnn = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (nnn != null && nnn.EndToken.Next != null && nnn.EndToken.Next.IsChar(':'))
{
break;
}
}
}
if (uris == null)
{
break;
}
}
}
if (empty > 2)
{
break;
}
if (empty > 0 && t.IsChar(':') && t.IsNewlineAfter)
{
break;
}
if (((t is Pullenti.Ner.NumberToken) && t.IsNewlineBefore && t.Next != null) && !t.Next.Chars.IsLetter)
{
break;
}
}
if (uris == null)
{
return(null);
}
if (!urisKeys.Contains("Р/С") && !urisKeys.Contains("Л/С"))
{
return(null);
}
bool ok = false;
if ((uris.Count < 2) && org == null)
{
return(null);
}
BankDataReferent bdr = new BankDataReferent();
foreach (Pullenti.Ner.Uri.UriReferent u in uris)
{
bdr.AddSlot(BankDataReferent.ATTR_ITEM, u, false, 0);
}
if (org != null)
{
bdr.AddSlot(BankDataReferent.ATTR_BANK, org, false, 0);
}
if (corOrg != null)
{
bdr.AddSlot(BankDataReferent.ATTR_CORBANK, corOrg, false, 0);
}
Pullenti.Ner.Referent org0 = (t0.Previous == null ? null : t0.Previous.GetReferent());
if (org0 != null && org0.TypeName == "ORGANIZATION")
{
foreach (Pullenti.Ner.Slot s in org0.Slots)
{
if (s.Value is Pullenti.Ner.Uri.UriReferent)
{
Pullenti.Ner.Uri.UriReferent u = s.Value as Pullenti.Ner.Uri.UriReferent;
if (_isBankReq(u.Scheme))
{
if (!urisKeys.Contains(u.Scheme))
{
bdr.AddSlot(BankDataReferent.ATTR_ITEM, u, false, 0);
}
}
}
}
}
return(new Pullenti.Ner.ReferentToken(bdr, t0, t1));
}
internal static Pullenti.Ner.ReferentToken CreateReferentToken(Pullenti.Ner.Person.PersonReferent p, Pullenti.Ner.Token begin, Pullenti.Ner.Token end, Pullenti.Ner.MorphCollection morph, List <PersonAttrToken> attrs, Pullenti.Ner.Person.PersonAnalyzer.PersonAnalyzerData ad, bool forAttribute, bool afterBePredicate)
{
if (p == null)
{
return(null);
}
bool hasPrefix = false;
if (attrs != null)
{
foreach (PersonAttrToken a in attrs)
{
if (a.Typ == PersonAttrTerminType.BestRegards)
{
hasPrefix = true;
}
else
{
if (a.BeginChar < begin.BeginChar)
{
begin = a.BeginToken;
if ((a.EndToken.Next != null && a.EndToken.Next.IsChar(')') && begin.Previous != null) && begin.Previous.IsChar('('))
{
begin = begin.Previous;
}
}
if (a.Typ != PersonAttrTerminType.Prefix)
{
if (a.Age != null)
{
p.AddSlot(Pullenti.Ner.Person.PersonReferent.ATTR_AGE, a.Age, false, 0);
}
if (a.PropRef == null)
{
p.AddSlot(Pullenti.Ner.Person.PersonReferent.ATTR_ATTR, a.Value, false, 0);
}
else
{
p.AddSlot(Pullenti.Ner.Person.PersonReferent.ATTR_ATTR, a, false, 0);
}
}
else if (a.Gender == Pullenti.Morph.MorphGender.Feminie && !p.IsFemale)
{
p.IsFemale = true;
}
else if (a.Gender == Pullenti.Morph.MorphGender.Masculine && !p.IsMale)
{
p.IsMale = true;
}
}
}
}
else if ((begin.Previous is Pullenti.Ner.TextToken) && (begin.WhitespacesBeforeCount < 3))
{
if ((begin.Previous as Pullenti.Ner.TextToken).Term == "ИП")
{
PersonAttrToken a = new PersonAttrToken(begin.Previous, begin.Previous);
a.PropRef = new Pullenti.Ner.Person.PersonPropertyReferent();
a.PropRef.Name = "индивидуальный предприниматель";
p.AddSlot(Pullenti.Ner.Person.PersonReferent.ATTR_ATTR, a, false, 0);
begin = begin.Previous;
}
}
Pullenti.Ner.MorphCollection m0 = new Pullenti.Ner.MorphCollection();
foreach (Pullenti.Morph.MorphBaseInfo it in morph.Items)
{
Pullenti.Morph.MorphBaseInfo bi = new Pullenti.Morph.MorphBaseInfo();
bi.CopyFrom(it);
bi.Number = Pullenti.Morph.MorphNumber.Singular;
if (bi.Gender == Pullenti.Morph.MorphGender.Undefined)
{
if (p.IsMale && !p.IsFemale)
{
bi.Gender = Pullenti.Morph.MorphGender.Masculine;
}
if (!p.IsMale && p.IsFemale)
{
bi.Gender = Pullenti.Morph.MorphGender.Feminie;
}
}
m0.AddItem(bi);
}
morph = m0;
if ((attrs != null && attrs.Count > 0 && !attrs[0].Morph.Case.IsUndefined) && morph.Case.IsUndefined)
{
morph.Case = attrs[0].Morph.Case;
if (attrs[0].Morph.Number == Pullenti.Morph.MorphNumber.Singular)
{
morph.Number = Pullenti.Morph.MorphNumber.Singular;
}
if (p.IsMale && !p.IsFemale)
{
morph.Gender = Pullenti.Morph.MorphGender.Masculine;
}
else if (p.IsFemale)
{
morph.Gender = Pullenti.Morph.MorphGender.Feminie;
}
}
if (begin.Previous != null)
{
Pullenti.Ner.Token ttt = begin.Previous;
if (ttt.IsValue("ИМЕНИ", "ІМЕНІ"))
{
forAttribute = true;
}
else
{
if (ttt.IsChar('.') && ttt.Previous != null)
{
ttt = ttt.Previous;
}
if (ttt.WhitespacesAfterCount < 3)
{
if (ttt.IsValue("ИМ", "ІМ"))
{
forAttribute = true;
}
}
}
}
if (forAttribute)
{
return new Pullenti.Ner.ReferentToken(p, begin, end)
{
Morph = morph, MiscAttrs = (int)p.m_PersonIdentityTyp
}
}
;
if ((begin.Previous != null && begin.Previous.IsCommaAnd && (begin.Previous.Previous is Pullenti.Ner.ReferentToken)) && (begin.Previous.Previous.GetReferent() is Pullenti.Ner.Person.PersonReferent))
{
Pullenti.Ner.ReferentToken rt00 = begin.Previous.Previous as Pullenti.Ner.ReferentToken;
for (Pullenti.Ner.Token ttt = (Pullenti.Ner.Token)rt00; ttt != null;)
{
if (ttt.Previous == null || !(ttt.Previous.Previous is Pullenti.Ner.ReferentToken))
{
break;
}
if (!ttt.Previous.IsCommaAnd || !(ttt.Previous.Previous.GetReferent() is Pullenti.Ner.Person.PersonReferent))
{
break;
}
rt00 = ttt.Previous.Previous as Pullenti.Ner.ReferentToken;
ttt = rt00;
}
if (rt00.BeginToken.GetReferent() is Pullenti.Ner.Person.PersonPropertyReferent)
{
bool ok = false;
if ((rt00.BeginToken as Pullenti.Ner.ReferentToken).EndToken.Next != null && (rt00.BeginToken as Pullenti.Ner.ReferentToken).EndToken.Next.IsChar(':'))
{
ok = true;
}
else if (rt00.BeginToken.Morph.Number == Pullenti.Morph.MorphNumber.Plural)
{
ok = true;
}
if (ok)
{
p.AddSlot(Pullenti.Ner.Person.PersonReferent.ATTR_ATTR, rt00.BeginToken.GetReferent(), false, 0);
}
}
}
if (ad != null)
{
if (ad.OverflowLevel > 10)
{
return new Pullenti.Ner.ReferentToken(p, begin, end)
{
Morph = morph, MiscAttrs = (int)p.m_PersonIdentityTyp
}
}
;
ad.OverflowLevel++;
}
List <PersonAttrToken> attrs1 = null;
bool hasPosition = false;
bool openBr = false;
for (Pullenti.Ner.Token t = end.Next; t != null; t = t.Next)
{
if (t.IsTableControlChar)
{
break;
}
if (t.IsNewlineBefore)
{
if (t.NewlinesBeforeCount > 2)
{
break;
}
if (attrs1 != null && attrs1.Count > 0)
{
break;
}
Pullenti.Ner.Mail.Internal.MailLine ml = Pullenti.Ner.Mail.Internal.MailLine.Parse(t, 0, 0);
if (ml != null && ml.Typ == Pullenti.Ner.Mail.Internal.MailLine.Types.From)
{
break;
}
if (t.Chars.IsCapitalUpper)
{
PersonAttrToken attr1 = PersonAttrToken.TryAttach(t, (ad == null ? null : ad.LocalOntology), PersonAttrToken.PersonAttrAttachAttrs.No);
bool ok1 = false;
if (attr1 != null)
{
if (hasPrefix || attr1.IsNewlineAfter || ((attr1.EndToken.Next != null && attr1.EndToken.Next.IsTableControlChar)))
{
ok1 = true;
}
else
{
for (Pullenti.Ner.Token tt2 = t.Next; tt2 != null && tt2.EndChar <= attr1.EndChar; tt2 = tt2.Next)
{
if (tt2.IsWhitespaceBefore)
{
ok1 = true;
}
}
}
}
else
{
Pullenti.Ner.Token ttt = CorrectTailAttributes(p, t);
if (ttt != null && ttt != t)
{
end = (t = ttt);
continue;
}
}
if (!ok1)
{
break;
}
}
}
if (t.IsHiphen || t.IsCharOf("_>|"))
{
continue;
}
if (t.IsValue("МОДЕЛЬ", null))
{
break;
}
Pullenti.Ner.Token tt = CorrectTailAttributes(p, t);
if (tt != t && tt != null)
{
end = (t = tt);
continue;
}
bool isBe = false;
if (t.IsChar('(') && t == end.Next)
{
openBr = true;
t = t.Next;
if (t == null)
{
break;
}
PersonItemToken pit1 = PersonItemToken.TryAttach(t, null, PersonItemToken.ParseAttr.No, null);
if ((pit1 != null && t.Chars.IsCapitalUpper && pit1.EndToken.Next != null) && (t is Pullenti.Ner.TextToken) && pit1.EndToken.Next.IsChar(')'))
{
if (pit1.Lastname != null)
{
Pullenti.Morph.MorphBaseInfo inf = new Pullenti.Morph.MorphBaseInfo()
{
Case = Pullenti.Morph.MorphCase.Nominative
};
if (p.IsMale)
{
inf.Gender |= Pullenti.Morph.MorphGender.Masculine;
}
if (p.IsFemale)
{
inf.Gender |= Pullenti.Morph.MorphGender.Feminie;
}
PersonMorphCollection sur = PersonIdentityToken.CreateLastname(pit1, inf);
if (sur != null)
{
p.AddFioIdentity(sur, null, null);
end = (t = pit1.EndToken.Next);
continue;
}
}
}
if ((t is Pullenti.Ner.TextToken) && t.Chars.IsLatinLetter)
{
List <PersonItemToken> pits = PersonItemToken.TryAttachList(t, null, PersonItemToken.ParseAttr.CanBeLatin, 10);
if (((pits != null && pits.Count >= 2 && pits.Count <= 3) && pits[0].Chars.IsLatinLetter && pits[1].Chars.IsLatinLetter) && pits[pits.Count - 1].EndToken.Next != null && pits[pits.Count - 1].EndToken.Next.IsChar(')'))
{
Pullenti.Ner.Person.PersonReferent pr2 = new Pullenti.Ner.Person.PersonReferent();
int cou = 0;
foreach (PersonItemToken pi in pits)
{
foreach (Pullenti.Ner.Slot si in p.Slots)
{
if (si.TypeName == Pullenti.Ner.Person.PersonReferent.ATTR_FIRSTNAME || si.TypeName == Pullenti.Ner.Person.PersonReferent.ATTR_MIDDLENAME || si.TypeName == Pullenti.Ner.Person.PersonReferent.ATTR_LASTNAME)
{
if (Pullenti.Ner.Core.MiscHelper.CanBeEqualCyrAndLatSS(si.Value.ToString(), pi.Value))
{
cou++;
pr2.AddSlot(si.TypeName, pi.Value, false, 0);
break;
}
}
}
}
if (cou == pits.Count)
{
foreach (Pullenti.Ner.Slot si in pr2.Slots)
{
p.AddSlot(si.TypeName, si.Value, false, 0);
}
end = (t = pits[pits.Count - 1].EndToken.Next);
continue;
}
}
}
}
else if (t.IsComma)
{
t = t.Next;
if ((t is Pullenti.Ner.TextToken) && (t as Pullenti.Ner.TextToken).IsValue("WHO", null))
{
continue;
}
if ((t is Pullenti.Ner.TextToken) && t.Chars.IsLatinLetter)
{
List <PersonItemToken> pits = PersonItemToken.TryAttachList(t, null, PersonItemToken.ParseAttr.CanBeLatin, 10);
if ((pits != null && pits.Count >= 2 && pits.Count <= 3) && pits[0].Chars.IsLatinLetter && pits[1].Chars.IsLatinLetter)
{
Pullenti.Ner.Person.PersonReferent pr2 = new Pullenti.Ner.Person.PersonReferent();
int cou = 0;
foreach (PersonItemToken pi in pits)
{
foreach (Pullenti.Ner.Slot si in p.Slots)
{
if (si.TypeName == Pullenti.Ner.Person.PersonReferent.ATTR_FIRSTNAME || si.TypeName == Pullenti.Ner.Person.PersonReferent.ATTR_MIDDLENAME || si.TypeName == Pullenti.Ner.Person.PersonReferent.ATTR_LASTNAME)
{
if (Pullenti.Ner.Core.MiscHelper.CanBeEqualCyrAndLatSS(si.Value.ToString(), pi.Value))
{
cou++;
pr2.AddSlot(si.TypeName, pi.Value, false, 0);
break;
}
}
}
}
if (cou == pits.Count)
{
foreach (Pullenti.Ner.Slot si in pr2.Slots)
{
p.AddSlot(si.TypeName, si.Value, false, 0);
}
end = (t = pits[pits.Count - 1].EndToken);
continue;
}
}
}
}
else if ((t is Pullenti.Ner.TextToken) && (t as Pullenti.Ner.TextToken).IsVerbBe)
{
t = t.Next;
}
else if (t.IsAnd && t.IsWhitespaceAfter && !t.IsNewlineAfter)
{
if (t == end.Next)
{
break;
}
t = t.Next;
}
else if (t.IsHiphen && t == end.Next)
{
t = t.Next;
}
else if (t.IsChar('.') && t == end.Next && hasPrefix)
{
t = t.Next;
}
Pullenti.Ner.Token ttt2 = CreateNickname(p, t);
if (ttt2 != null)
{
t = (end = ttt2);
continue;
}
if (t == null)
{
break;
}
PersonAttrToken attr = null;
attr = PersonAttrToken.TryAttach(t, (ad == null ? null : ad.LocalOntology), PersonAttrToken.PersonAttrAttachAttrs.No);
if (attr == null)
{
if ((t != null && t.GetReferent() != null && t.GetReferent().TypeName == "GEO") && attrs1 != null && openBr)
{
continue;
}
if ((t.Chars.IsCapitalUpper && openBr && t.Next != null) && t.Next.IsChar(')'))
{
if (p.FindSlot(Pullenti.Ner.Person.PersonReferent.ATTR_LASTNAME, null, true) == null)
{
p.AddSlot(Pullenti.Ner.Person.PersonReferent.ATTR_LASTNAME, t.GetSourceText().ToUpper(), false, 0);
t = t.Next;
end = t;
}
}
if (t != null && t.IsValue("КОТОРЫЙ", null) && t.Morph.Number == Pullenti.Morph.MorphNumber.Singular)
{
if (!p.IsFemale && t.Morph.Gender == Pullenti.Morph.MorphGender.Feminie)
{
p.IsFemale = true;
p.CorrectData();
}
else if (!p.IsMale && t.Morph.Gender == Pullenti.Morph.MorphGender.Masculine)
{
p.IsMale = true;
p.CorrectData();
}
}
break;
}
if (attr.Morph.Number == Pullenti.Morph.MorphNumber.Plural)
{
break;
}
if (attr.Typ == PersonAttrTerminType.BestRegards)
{
break;
}
if (attr.IsDoubt)
{
if (hasPrefix)
{
}
else if (t.IsNewlineBefore && attr.IsNewlineAfter)
{
}
else if (t.Previous != null && ((t.Previous.IsHiphen || t.Previous.IsChar(':'))))
{
}
else
{
break;
}
}
if (!morph.Case.IsUndefined && !attr.Morph.Case.IsUndefined)
{
if (((morph.Case & attr.Morph.Case)).IsUndefined && !isBe)
{
break;
}
}
if (openBr)
{
if (Pullenti.Ner.Person.PersonAnalyzer.TryAttachPerson(t, ad, false, 0, true) != null)
{
break;
}
}
if (attrs1 == null)
{
if (t.Previous.IsComma && t.Previous == end.Next)
{
Pullenti.Ner.Token ttt = attr.EndToken.Next;
if (ttt != null)
{
if (ttt.Morph.Class.IsVerb)
{
if (Pullenti.Ner.Core.MiscHelper.CanBeStartOfSentence(begin))
{
}
else
{
break;
}
}
}
}
attrs1 = new List <PersonAttrToken>();
}
attrs1.Add(attr);
if (attr.Typ == PersonAttrTerminType.Position || attr.Typ == PersonAttrTerminType.King)
{
if (!isBe)
{
hasPosition = true;
}
}
else if (attr.Typ != PersonAttrTerminType.Prefix)
{
if (attr.Typ == PersonAttrTerminType.Other && attr.Age != null)
{
}
else
{
attrs1 = null;
break;
}
}
t = attr.EndToken;
}
if (attrs1 != null && hasPosition && attrs != null)
{
Pullenti.Ner.Token te1 = attrs[attrs.Count - 1].EndToken.Next;
Pullenti.Ner.Token te2 = attrs1[0].BeginToken;
if (te1.WhitespacesAfterCount > te2.WhitespacesBeforeCount && (te2.WhitespacesBeforeCount < 2))
{
}
else if (attrs1[0].Age != null)
{
}
else if (((te1.IsHiphen || te1.IsChar(':'))) && !attrs1[0].IsNewlineBefore && ((te2.Previous.IsComma || te2.Previous == end)))
{
}
else
{
foreach (PersonAttrToken a in attrs)
{
if (a.Typ == PersonAttrTerminType.Position)
{
Pullenti.Ner.Token te = attrs1[attrs1.Count - 1].EndToken;
if (te.Next != null)
{
if (!te.Next.IsChar('.'))
{
attrs1 = null;
break;
}
}
}
}
}
}
if (attrs1 != null && !hasPrefix)
{
PersonAttrToken attr = attrs1[attrs1.Count - 1];
bool ok = false;
if (attr.EndToken.Next != null && attr.EndToken.Next.Chars.IsCapitalUpper)
{
ok = true;
}
else
{
Pullenti.Ner.ReferentToken rt = Pullenti.Ner.Person.PersonAnalyzer.TryAttachPerson(attr.BeginToken, ad, false, -1, false);
if (rt != null && (rt.Referent is Pullenti.Ner.Person.PersonReferent))
{
ok = true;
}
}
if (ok)
{
if (attr.BeginToken.WhitespacesBeforeCount > attr.EndToken.WhitespacesAfterCount)
{
attrs1 = null;
}
else if (attr.BeginToken.WhitespacesBeforeCount == attr.EndToken.WhitespacesAfterCount)
{
Pullenti.Ner.ReferentToken rt1 = Pullenti.Ner.Person.PersonAnalyzer.TryAttachPerson(attr.BeginToken, ad, false, -1, false);
if (rt1 != null)
{
attrs1 = null;
}
}
}
}
if (attrs1 != null)
{
foreach (PersonAttrToken a in attrs1)
{
if (a.Typ != PersonAttrTerminType.Prefix)
{
if (a.Age != null)
{
p.AddSlot(Pullenti.Ner.Person.PersonReferent.ATTR_AGE, a.Age, true, 0);
}
else if (a.PropRef == null)
{
p.AddSlot(Pullenti.Ner.Person.PersonReferent.ATTR_ATTR, a.Value, false, 0);
}
else
{
p.AddSlot(Pullenti.Ner.Person.PersonReferent.ATTR_ATTR, a, false, 0);
}
end = a.EndToken;
if (a.Gender != Pullenti.Morph.MorphGender.Undefined && !p.IsFemale && !p.IsMale)
{
if (a.Gender == Pullenti.Morph.MorphGender.Masculine && !p.IsMale)
{
p.IsMale = true;
p.CorrectData();
}
else if (a.Gender == Pullenti.Morph.MorphGender.Feminie && !p.IsFemale)
{
p.IsFemale = true;
p.CorrectData();
}
}
}
}
if (openBr)
{
if (end.Next != null && end.Next.IsChar(')'))
{
end = end.Next;
}
}
}
int crlfCou = 0;
for (Pullenti.Ner.Token t = end.Next; t != null; t = t.Next)
{
if (t.IsTableControlChar)
{
break;
}
if (t.IsNewlineBefore)
{
Pullenti.Ner.Mail.Internal.MailLine ml = Pullenti.Ner.Mail.Internal.MailLine.Parse(t, 0, 0);
if (ml != null && ml.Typ == Pullenti.Ner.Mail.Internal.MailLine.Types.From)
{
break;
}
crlfCou++;
}
if (t.IsCharOf(":,(") || t.IsHiphen)
{
continue;
}
if (t.IsChar('.') && t == end.Next)
{
continue;
}
Pullenti.Ner.Referent r = t.GetReferent();
if (r != null)
{
if (r.TypeName == "PHONE" || r.TypeName == "URI" || r.TypeName == "ADDRESS")
{
string ty = r.GetStringValue("SCHEME");
if (r.TypeName == "URI")
{
if ((ty != "mailto" && ty != "skype" && ty != "ICQ") && ty != "http")
{
break;
}
}
p.AddContact(r);
end = t;
crlfCou = 0;
continue;
}
}
if (r is Pullenti.Ner.Person.PersonIdentityReferent)
{
p.AddSlot(Pullenti.Ner.Person.PersonReferent.ATTR_IDDOC, r, false, 0);
end = t;
crlfCou = 0;
continue;
}
if (r != null && r.TypeName == "ORGANIZATION")
{
if (t.Next != null && t.Next.Morph.Class.IsVerb)
{
break;
}
if (begin.Previous != null && begin.Previous.Morph.Class.IsVerb)
{
break;
}
if (t.WhitespacesAfterCount == 1)
{
break;
}
bool exist = false;
foreach (Pullenti.Ner.Slot s in p.Slots)
{
if (s.TypeName == Pullenti.Ner.Person.PersonReferent.ATTR_ATTR && (s.Value is Pullenti.Ner.Person.PersonPropertyReferent))
{
Pullenti.Ner.Person.PersonPropertyReferent pr = s.Value as Pullenti.Ner.Person.PersonPropertyReferent;
if (pr.FindSlot(Pullenti.Ner.Person.PersonPropertyReferent.ATTR_REF, r, true) != null)
{
exist = true;
break;
}
}
else if (s.TypeName == Pullenti.Ner.Person.PersonReferent.ATTR_ATTR && (s.Value is PersonAttrToken))
{
PersonAttrToken pr = s.Value as PersonAttrToken;
if (pr.Referent.FindSlot(Pullenti.Ner.Person.PersonPropertyReferent.ATTR_REF, r, true) != null)
{
exist = true;
break;
}
}
}
if (!exist)
{
PersonAttrToken pat = new PersonAttrToken(t, t);
pat.PropRef = new Pullenti.Ner.Person.PersonPropertyReferent()
{
Name = "сотрудник"
};
pat.PropRef.AddSlot(Pullenti.Ner.Person.PersonPropertyReferent.ATTR_REF, r, false, 0);
p.AddSlot(Pullenti.Ner.Person.PersonReferent.ATTR_ATTR, pat, false, 0);
}
continue;
}
if (r != null)
{
break;
}
if (!hasPrefix || crlfCou >= 2)
{
break;
}
Pullenti.Ner.ReferentToken rt = t.Kit.ProcessReferent("PERSON", t);
if (rt != null)
{
break;
}
}
if (ad != null)
{
ad.OverflowLevel--;
}
if (begin.IsValue("НА", null) && begin.Next != null && begin.Next.IsValue("ИМЯ", null))
{
Pullenti.Ner.Token t0 = begin.Previous;
if (t0 != null && t0.IsComma)
{
t0 = t0.Previous;
}
if (t0 != null && (t0.GetReferent() is Pullenti.Ner.Person.PersonIdentityReferent))
{
p.AddSlot(Pullenti.Ner.Person.PersonReferent.ATTR_IDDOC, t0.GetReferent(), false, 0);
}
}
return(new Pullenti.Ner.ReferentToken(p, begin, end)
{
Morph = morph, MiscAttrs = (int)p.m_PersonIdentityTyp
});
}
public override void Process(Pullenti.Ner.Core.AnalysisKit kit)
{
PhoneAnalizerData ad = kit.GetAnalyzerData(this) as PhoneAnalizerData;
for (Pullenti.Ner.Token t = kit.FirstToken; t != null; t = t.Next)
{
List<Pullenti.Ner.Phone.Internal.PhoneItemToken> pli = Pullenti.Ner.Phone.Internal.PhoneItemToken.TryAttachAll(t, 15);
if (pli == null || pli.Count == 0)
continue;
PhoneReferent prevPhone = null;
int kkk = 0;
for (Pullenti.Ner.Token tt = t.Previous; tt != null; tt = tt.Previous)
{
if (tt.GetReferent() is PhoneReferent)
{
prevPhone = tt.GetReferent() as PhoneReferent;
break;
}
else if (tt is Pullenti.Ner.ReferentToken)
{
}
else if (tt.IsChar(')'))
{
Pullenti.Ner.Token ttt = tt.Previous;
int cou = 0;
for (; ttt != null; ttt = ttt.Previous)
{
if (ttt.IsChar('('))
break;
else if ((++cou) > 100)
break;
}
if (ttt == null || !ttt.IsChar('('))
break;
tt = ttt;
}
else if (!tt.IsCharOf(",;/\\") && !tt.IsAnd)
{
if ((++kkk) > 5)
break;
if (tt.IsNewlineBefore || tt.IsNewlineAfter)
break;
}
}
int j = 0;
bool isPhoneBefore = false;
bool isPref = false;
PhoneKind ki = PhoneKind.Undefined;
while (j < pli.Count)
{
if (pli[j].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Prefix)
{
if (ki == PhoneKind.Undefined)
ki = pli[j].Kind;
isPref = true;
isPhoneBefore = true;
j++;
if ((j < pli.Count) && pli[j].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Delim)
j++;
}
else if (((j + 1) < pli.Count) && pli[j + 1].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Prefix && j == 0)
{
if (ki == PhoneKind.Undefined)
ki = pli[0].Kind;
isPref = true;
pli.RemoveAt(0);
}
else
break;
}
if (prevPhone != null)
isPhoneBefore = true;
if (pli.Count == 1 && pli[0].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Number)
{
Pullenti.Ner.Token tt = t.Previous;
if ((tt is Pullenti.Ner.TextToken) && !tt.Chars.IsLetter)
tt = tt.Previous;
if (tt is Pullenti.Ner.TextToken)
{
if (Pullenti.Ner.Uri.UriAnalyzer.m_Schemes.TryParse(tt, Pullenti.Ner.Core.TerminParseAttr.No) != null)
continue;
}
}
List<Pullenti.Ner.ReferentToken> rts = this.TryAttach(pli, j, isPhoneBefore, prevPhone);
if (rts == null)
{
for (j = 1; j < pli.Count; j++)
{
if (pli[j].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Prefix)
{
pli.RemoveRange(0, j);
rts = this.TryAttach(pli, 1, true, prevPhone);
break;
}
}
}
if (rts == null)
t = pli[pli.Count - 1].EndToken;
else
{
if ((ki == PhoneKind.Undefined && prevPhone != null && !isPref) && prevPhone.Kind != PhoneKind.Mobile && kkk == 0)
ki = prevPhone.Kind;
foreach (Pullenti.Ner.ReferentToken rt in rts)
{
PhoneReferent ph = rt.Referent as PhoneReferent;
if (ki != PhoneKind.Undefined)
ph.Kind = ki;
else
{
if (rt == rts[0] && (rt.WhitespacesBeforeCount < 3))
{
Pullenti.Ner.Token tt1 = rt.BeginToken.Previous;
if (tt1 != null && tt1.IsTableControlChar)
tt1 = tt1.Previous;
if ((tt1 is Pullenti.Ner.TextToken) && ((tt1.IsNewlineBefore || ((tt1.Previous != null && tt1.Previous.IsTableControlChar)))))
{
string term = (tt1 as Pullenti.Ner.TextToken).Term;
if (term == "T" || term == "Т")
rt.BeginToken = tt1;
else if (term == "Ф" || term == "F")
{
ph.Kind = (ki = PhoneKind.Fax);
rt.BeginToken = tt1;
}
else if (term == "M" || term == "М")
{
ph.Kind = (ki = PhoneKind.Mobile);
rt.BeginToken = tt1;
}
}
}
ph.Correct();
}
rt.Referent = ad.RegisterReferent(rt.Referent);
kit.EmbedToken(rt);
t = rt;
}
}
}
}
public static BlockLine Create(Pullenti.Ner.Token t, Pullenti.Ner.Core.TerminCollection names)
{
if (t == null)
{
return(null);
}
BlockLine res = new BlockLine(t, t);
for (Pullenti.Ner.Token tt = t; tt != null; tt = tt.Next)
{
if (tt != t && tt.IsNewlineBefore)
{
break;
}
else
{
res.EndToken = tt;
}
}
int nums = 0;
while (t != null && t.Next != null && t.EndChar <= res.EndChar)
{
if (t is Pullenti.Ner.NumberToken)
{
}
else
{
Pullenti.Ner.NumberToken rom = Pullenti.Ner.Core.NumberHelper.TryParseRoman(t);
if (rom != null && rom.EndToken.Next != null)
{
t = rom.EndToken;
}
else
{
break;
}
}
if (t.Next.IsChar('.'))
{
}
else if ((t.Next is Pullenti.Ner.TextToken) && !t.Next.Chars.IsAllLower)
{
}
else
{
break;
}
res.NumberEnd = t;
t = t.Next;
if (t.IsChar('.') && t.Next != null)
{
res.NumberEnd = t;
t = t.Next;
}
if (t.IsNewlineBefore)
{
return(res);
}
nums++;
}
Pullenti.Ner.Core.TerminToken tok = m_Ontology.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok == null)
{
Pullenti.Ner.Core.NounPhraseToken npt1 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt1 != null && npt1.EndToken != npt1.BeginToken)
{
tok = m_Ontology.TryParse(npt1.Noun.BeginToken, Pullenti.Ner.Core.TerminParseAttr.No);
}
}
if (tok != null)
{
if (t.Previous != null && t.Previous.IsChar(':'))
{
tok = null;
}
}
if (tok != null)
{
BlkTyps typ = (BlkTyps)tok.Termin.Tag;
if (typ == BlkTyps.Conslusion)
{
if (t.IsNewlineAfter)
{
}
else if (t.Next != null && t.Next.Morph.Class.IsPreposition && t.Next.Next != null)
{
Pullenti.Ner.Core.TerminToken tok2 = m_Ontology.TryParse(t.Next.Next, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok2 != null && ((BlkTyps)tok2.Termin.Tag) == BlkTyps.Chapter)
{
}
else
{
tok = null;
}
}
else
{
tok = null;
}
}
if (t.Kit.BaseLanguage != t.Morph.Language)
{
tok = null;
}
if (typ == BlkTyps.Index && !t.IsValue("ОГЛАВЛЕНИЕ", null))
{
if (!t.IsNewlineAfter && t.Next != null)
{
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t.Next, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null && npt.IsNewlineAfter && npt.Morph.Case.IsGenitive)
{
tok = null;
}
else if (npt == null)
{
tok = null;
}
}
}
if ((typ == BlkTyps.Intro && tok != null && !tok.IsNewlineAfter) && t.IsValue("ВВЕДЕНИЕ", null))
{
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t.Next, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null && npt.Morph.Case.IsGenitive)
{
tok = null;
}
}
if (tok != null)
{
if (res.NumberEnd == null)
{
res.NumberEnd = tok.EndToken;
if (res.NumberEnd.EndChar > res.EndChar)
{
res.EndToken = res.NumberEnd;
}
}
res.Typ = typ;
t = tok.EndToken;
if (t.Next != null && t.Next.IsCharOf(":."))
{
t = t.Next;
res.EndToken = t;
}
if (t.IsNewlineAfter || t.Next == null)
{
return(res);
}
t = t.Next;
}
}
if (t.IsChar('§') && (t.Next is Pullenti.Ner.NumberToken))
{
res.Typ = BlkTyps.Chapter;
res.NumberEnd = t;
t = t.Next;
}
if (names != null)
{
Pullenti.Ner.Core.TerminToken tok2 = names.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok2 != null && tok2.EndToken.IsNewlineAfter)
{
res.EndToken = tok2.EndToken;
res.IsExistName = true;
if (res.Typ == BlkTyps.Undefined)
{
BlockLine li2 = Create((res.NumberEnd == null ? null : res.NumberEnd.Next), null);
if (li2 != null && ((li2.Typ == BlkTyps.Literature || li2.Typ == BlkTyps.Intro || li2.Typ == BlkTyps.Conslusion)))
{
res.Typ = li2.Typ;
}
else
{
res.Typ = BlkTyps.Chapter;
}
}
return(res);
}
}
Pullenti.Ner.Token t1 = res.EndToken;
if ((((t1 is Pullenti.Ner.NumberToken) || t1.IsChar('.'))) && t1.Previous != null)
{
t1 = t1.Previous;
if (t1.IsChar('.'))
{
res.HasContentItemTail = true;
for (; t1 != null && t1.BeginChar > res.BeginChar; t1 = t1.Previous)
{
if (!t1.IsChar('.'))
{
break;
}
}
}
}
res.IsAllUpper = true;
for (; t != null && t.EndChar <= t1.EndChar; t = t.Next)
{
if (!(t is Pullenti.Ner.TextToken) || !t.Chars.IsLetter)
{
res.NotWords++;
}
else
{
Pullenti.Morph.MorphClass mc = t.GetMorphClassInDictionary();
if (mc.IsUndefined)
{
res.NotWords++;
}
else if (t.LengthChar > 2)
{
res.Words++;
}
if (!t.Chars.IsAllUpper)
{
res.IsAllUpper = false;
}
if ((t as Pullenti.Ner.TextToken).IsPureVerb)
{
if (!(t as Pullenti.Ner.TextToken).Term.EndsWith("ING"))
{
res.HasVerb = true;
}
}
}
}
if (res.Typ == BlkTyps.Undefined)
{
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse((res.NumberEnd == null ? res.BeginToken : res.NumberEnd.Next), Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt != null)
{
if (npt.Noun.IsValue("ХАРАКТЕРИСТИКА", null) || npt.Noun.IsValue("СОДЕРЖАНИЕ", "ЗМІСТ"))
{
bool ok = true;
for (Pullenti.Ner.Token tt = npt.EndToken.Next; tt != null && tt.EndChar <= res.EndChar; tt = tt.Next)
{
if (tt.IsChar('.'))
{
continue;
}
Pullenti.Ner.Core.NounPhraseToken npt2 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(tt, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt2 == null || !npt2.Morph.Case.IsGenitive)
{
ok = false;
break;
}
tt = npt2.EndToken;
if (tt.EndChar > res.EndChar)
{
res.EndToken = tt;
if (!tt.IsNewlineAfter)
{
for (; res.EndToken.Next != null; res.EndToken = res.EndToken.Next)
{
if (res.EndToken.IsNewlineAfter)
{
break;
}
}
}
}
}
if (ok)
{
res.Typ = BlkTyps.Intro;
res.IsExistName = true;
}
}
else if (npt.Noun.IsValue("ВЫВОД", "ВИСНОВОК") || npt.Noun.IsValue("РЕЗУЛЬТАТ", "ДОСЛІДЖЕННЯ"))
{
bool ok = true;
for (Pullenti.Ner.Token tt = npt.EndToken.Next; tt != null && tt.EndChar <= res.EndChar; tt = tt.Next)
{
if (tt.IsCharOf(",.") || tt.IsAnd)
{
continue;
}
Pullenti.Ner.Core.NounPhraseToken npt1 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(tt, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt1 != null)
{
if (npt1.Noun.IsValue("РЕЗУЛЬТАТ", "ДОСЛІДЖЕННЯ") || npt1.Noun.IsValue("РЕКОМЕНДАЦИЯ", "РЕКОМЕНДАЦІЯ") || npt1.Noun.IsValue("ИССЛЕДОВАНИЕ", "ДОСЛІДЖЕННЯ"))
{
tt = npt1.EndToken;
if (tt.EndChar > res.EndChar)
{
res.EndToken = tt;
if (!tt.IsNewlineAfter)
{
for (; res.EndToken.Next != null; res.EndToken = res.EndToken.Next)
{
if (res.EndToken.IsNewlineAfter)
{
break;
}
}
}
}
continue;
}
}
ok = false;
break;
}
if (ok)
{
res.Typ = BlkTyps.Conslusion;
res.IsExistName = true;
}
}
if (res.Typ == BlkTyps.Undefined && npt != null && npt.EndChar <= res.EndChar)
{
bool ok = false;
int publ = 0;
if (_isPub(npt))
{
ok = true;
publ = 1;
}
else if ((npt.Noun.IsValue("СПИСОК", null) || npt.Noun.IsValue("УКАЗАТЕЛЬ", "ПОКАЖЧИК") || npt.Noun.IsValue("ПОЛОЖЕНИЕ", "ПОЛОЖЕННЯ")) || npt.Noun.IsValue("ВЫВОД", "ВИСНОВОК") || npt.Noun.IsValue("РЕЗУЛЬТАТ", "ДОСЛІДЖЕННЯ"))
{
if (npt.EndChar == res.EndChar)
{
return(null);
}
ok = true;
}
if (ok)
{
if (npt.BeginToken == npt.EndToken && npt.Noun.IsValue("СПИСОК", null) && npt.EndChar == res.EndChar)
{
ok = false;
}
for (Pullenti.Ner.Token tt = npt.EndToken.Next; tt != null && tt.EndChar <= res.EndChar; tt = tt.Next)
{
if (tt.IsCharOf(",.:") || tt.IsAnd || tt.Morph.Class.IsPreposition)
{
continue;
}
if (tt.IsValue("ОТРАЖЕНЫ", "ВІДОБРАЖЕНІ"))
{
continue;
}
npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(tt, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt == null)
{
ok = false;
break;
}
if (((_isPub(npt) || npt.Noun.IsValue("РАБОТА", "РОБОТА") || npt.Noun.IsValue("ИССЛЕДОВАНИЕ", "ДОСЛІДЖЕННЯ")) || npt.Noun.IsValue("АВТОР", null) || npt.Noun.IsValue("ТРУД", "ПРАЦЯ")) || npt.Noun.IsValue("ТЕМА", null) || npt.Noun.IsValue("ДИССЕРТАЦИЯ", "ДИСЕРТАЦІЯ"))
{
tt = npt.EndToken;
if (_isPub(npt))
{
publ++;
}
if (tt.EndChar > res.EndChar)
{
res.EndToken = tt;
if (!tt.IsNewlineAfter)
{
for (; res.EndToken.Next != null; res.EndToken = res.EndToken.Next)
{
if (res.EndToken.IsNewlineAfter)
{
break;
}
}
}
}
continue;
}
ok = false;
break;
}
if (ok)
{
res.Typ = BlkTyps.Literature;
res.IsExistName = true;
if (publ == 0 && (res.EndChar < (((res.Kit.Sofa.Text.Length * 2) / 3))))
{
if (res.NumberEnd != null)
{
res.Typ = BlkTyps.Misc;
}
else
{
res.Typ = BlkTyps.Undefined;
}
}
}
}
}
}
}
return(res);
}
Pullenti.Ner.ReferentToken _TryAttach_(List<Pullenti.Ner.Phone.Internal.PhoneItemToken> pli, int ind, bool isPhoneBefore, PhoneReferent prevPhone, int lev = 0)
{
if (ind >= pli.Count || lev > 4)
return null;
string countryCode = null;
string cityCode = null;
int j = ind;
if (prevPhone != null && prevPhone.m_Template != null && pli[j].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Number)
{
StringBuilder tmp = new StringBuilder();
for (int jj = j; jj < pli.Count; jj++)
{
if (pli[jj].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Number)
tmp.Append(pli[jj].Value.Length);
else if (pli[jj].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Delim)
{
if (pli[jj].Value == " ")
break;
tmp.Append(pli[jj].Value);
continue;
}
else
break;
string templ0 = tmp.ToString();
if (templ0 == prevPhone.m_Template)
{
if ((jj + 1) < pli.Count)
{
if (pli[jj + 1].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Prefix && (jj + 2) == pli.Count)
{
}
else
pli.RemoveRange(jj + 1, pli.Count - jj - 1);
}
break;
}
}
}
if ((j < pli.Count) && pli[j].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.CountryCode)
{
countryCode = pli[j].Value;
if (countryCode != "8")
{
string cc = Pullenti.Ner.Phone.Internal.PhoneHelper.GetCountryPrefix(countryCode);
if (cc != null && (cc.Length < countryCode.Length))
{
cityCode = countryCode.Substring(cc.Length);
countryCode = cc;
}
}
j++;
}
else if ((j < pli.Count) && pli[j].CanBeCountryPrefix)
{
int k = j + 1;
if ((k < pli.Count) && pli[k].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Delim)
k++;
Pullenti.Ner.ReferentToken rrt = this._TryAttach_(pli, k, isPhoneBefore, null, lev + 1);
if (rrt != null)
{
if ((((isPhoneBefore && pli[j + 1].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Delim && pli[j + 1].BeginToken.IsHiphen) && pli[j].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Number && pli[j].Value.Length == 3) && ((j + 2) < pli.Count) && pli[j + 2].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Number) && pli[j + 2].Value.Length == 3)
{
}
else
{
countryCode = pli[j].Value;
j++;
}
}
}
if (((j < pli.Count) && pli[j].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Number && ((pli[j].Value[0] == '8' || pli[j].Value[0] == '7'))) && countryCode == null)
{
if (pli[j].Value.Length == 1)
{
countryCode = pli[j].Value;
j++;
}
else if (pli[j].Value.Length == 4)
{
countryCode = pli[j].Value.Substring(0, 1);
if (cityCode == null)
cityCode = pli[j].Value.Substring(1);
else
cityCode += pli[j].Value.Substring(1);
j++;
}
else if (pli[j].Value.Length == 11 && j == (pli.Count - 1) && isPhoneBefore)
{
PhoneReferent ph0 = new PhoneReferent();
if (pli[j].Value[0] != '8')
ph0.CountryCode = pli[j].Value.Substring(0, 1);
ph0.Number = pli[j].Value.Substring(1, 3) + pli[j].Value.Substring(4);
return new Pullenti.Ner.ReferentToken(ph0, pli[0].BeginToken, pli[j].EndToken);
}
else if (cityCode == null && pli[j].Value.Length > 3 && ((j + 1) < pli.Count))
{
int sum = 0;
foreach (Pullenti.Ner.Phone.Internal.PhoneItemToken it in pli)
{
if (it.ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Number)
sum += it.Value.Length;
}
if (sum == 11)
{
cityCode = pli[j].Value.Substring(1);
j++;
}
}
}
if ((j < pli.Count) && pli[j].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.CityCode)
{
if (cityCode == null)
cityCode = pli[j].Value;
else
cityCode += pli[j].Value;
j++;
}
if ((j < pli.Count) && pli[j].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Delim)
j++;
if ((countryCode == "8" && cityCode == null && ((j + 3) < pli.Count)) && pli[j].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Number)
{
if (pli[j].Value.Length == 3 || pli[j].Value.Length == 4)
{
cityCode = pli[j].Value;
j++;
if ((j < pli.Count) && pli[j].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Delim)
j++;
}
}
int normalNumLen = 0;
if (countryCode == "421")
normalNumLen = 9;
StringBuilder num = new StringBuilder();
StringBuilder templ = new StringBuilder();
List<int> partLength = new List<int>();
string delim = null;
bool ok = false;
string additional = null;
bool std = false;
if (countryCode != null && ((j + 4) < pli.Count) && j > 0)
{
if (((((pli[j - 1].Value == "-" || pli[j - 1].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.CountryCode)) && pli[j].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Number && pli[j + 1].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Delim) && pli[j + 2].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Number && pli[j + 3].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Delim) && pli[j + 4].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Number)
{
if ((((pli[j].Value.Length + pli[j + 2].Value.Length) == 6 || ((pli[j].Value.Length == 4 && pli[j + 2].Value.Length == 5)))) && ((pli[j + 4].Value.Length == 4 || pli[j + 4].Value.Length == 1)))
{
num.Append(pli[j].Value);
num.Append(pli[j + 2].Value);
num.Append(pli[j + 4].Value);
templ.AppendFormat("{0}{1}{2}{3}{4}", pli[j].Value.Length, pli[j + 1].Value, pli[j + 2].Value.Length, pli[j + 3].Value, pli[j + 4].Value.Length);
std = true;
ok = true;
j += 5;
}
}
}
for (; j < pli.Count; j++)
{
if (std)
break;
if (pli[j].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Delim)
{
if (pli[j].IsInBrackets)
continue;
if (j > 0 && pli[j - 1].IsInBrackets)
continue;
if (templ.Length > 0)
templ.Append(pli[j].Value);
if (delim == null)
delim = pli[j].Value;
else if (pli[j].Value != delim)
{
if ((partLength.Count == 2 && ((partLength[0] == 3 || partLength[0] == 4)) && cityCode == null) && partLength[1] == 3)
{
cityCode = num.ToString().Substring(0, partLength[0]);
num.Remove(0, partLength[0]);
partLength.RemoveAt(0);
delim = pli[j].Value;
continue;
}
if (isPhoneBefore && ((j + 1) < pli.Count) && pli[j + 1].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Number)
{
if (num.Length < 6)
continue;
if (normalNumLen > 0 && (num.Length + pli[j + 1].Value.Length) == normalNumLen)
continue;
}
break;
}
else
continue;
ok = false;
}
else if (pli[j].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Number)
{
if (num.Length == 0 && pli[j].BeginToken.Previous != null && pli[j].BeginToken.Previous.IsTableControlChar)
{
Pullenti.Ner.Token tt = pli[pli.Count - 1].EndToken.Next;
if (tt != null && tt.IsCharOf(",."))
tt = tt.Next;
if (tt is Pullenti.Ner.NumberToken)
return null;
}
if ((num.Length + pli[j].Value.Length) > 13)
{
if (j > 0 && pli[j - 1].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Delim)
j--;
ok = true;
break;
}
num.Append(pli[j].Value);
partLength.Add(pli[j].Value.Length);
templ.Append(pli[j].Value.Length);
ok = true;
if (num.Length > 10)
{
j++;
if ((j < pli.Count) && pli[j].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.AddNumber)
{
additional = pli[j].Value;
j++;
}
break;
}
}
else if (pli[j].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.AddNumber)
{
additional = pli[j].Value;
j++;
break;
}
else
break;
}
if ((j == (pli.Count - 1) && pli[j].IsInBrackets && ((pli[j].Value.Length == 3 || pli[j].Value.Length == 4))) && additional == null)
{
additional = pli[j].Value;
j++;
}
if ((j < pli.Count) && pli[j].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Prefix && pli[j].IsInBrackets)
{
isPhoneBefore = true;
j++;
}
if ((countryCode == null && cityCode != null && cityCode.Length > 3) && (num.Length < 8) && cityCode[0] != '8')
{
if ((cityCode.Length + num.Length) == 10)
{
}
else
{
string cc = Pullenti.Ner.Phone.Internal.PhoneHelper.GetCountryPrefix(cityCode);
if (cc != null)
{
if (cc.Length > 1 && (cityCode.Length - cc.Length) > 1)
{
countryCode = cc;
cityCode = cityCode.Substring(cc.Length);
}
}
}
}
if (countryCode == null && cityCode != null && cityCode.StartsWith("00"))
{
string cc = Pullenti.Ner.Phone.Internal.PhoneHelper.GetCountryPrefix(cityCode.Substring(2));
if (cc != null)
{
if (cityCode.Length > (cc.Length + 3))
{
countryCode = cc;
cityCode = cityCode.Substring(cc.Length + 2);
}
}
}
if (num.Length == 0 && cityCode != null)
{
if (cityCode.Length == 10)
{
num.Append(cityCode.Substring(3));
partLength.Add(num.Length);
cityCode = cityCode.Substring(0, 3);
ok = true;
}
else if (((cityCode.Length == 9 || cityCode.Length == 11 || cityCode.Length == 8)) && ((isPhoneBefore || countryCode != null)))
{
num.Append(cityCode);
partLength.Add(num.Length);
cityCode = null;
ok = true;
}
}
if (num.Length < 4)
ok = false;
if (num.Length < 7)
{
if (cityCode != null && (cityCode.Length + num.Length) > 7)
{
if (!isPhoneBefore && cityCode.Length == 3)
{
int ii;
for (ii = 0; ii < partLength.Count; ii++)
{
if (partLength[ii] == 3)
{
}
else if (partLength[ii] > 3)
break;
else if ((ii < (partLength.Count - 1)) || (partLength[ii] < 2))
break;
}
if (ii >= partLength.Count)
{
if (countryCode == "61")
{
}
else
ok = false;
}
}
}
else if (((num.Length == 6 || num.Length == 5)) && ((partLength.Count >= 1 && partLength.Count <= 3)) && isPhoneBefore)
{
if (pli[0].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Prefix && pli[0].Kind == PhoneKind.Home)
ok = false;
}
else if (prevPhone != null && prevPhone.Number != null && ((prevPhone.Number.Length == num.Length || prevPhone.Number.Length == (num.Length + 3) || prevPhone.Number.Length == (num.Length + 4))))
{
}
else if (num.Length > 4 && prevPhone != null && templ.ToString() == prevPhone.m_Template)
ok = true;
else
ok = false;
}
if (delim == "." && countryCode == null && cityCode == null)
ok = false;
if ((isPhoneBefore && countryCode == null && cityCode == null) && num.Length > 10)
{
string cc = Pullenti.Ner.Phone.Internal.PhoneHelper.GetCountryPrefix(num.ToString());
if (cc != null)
{
if ((num.Length - cc.Length) == 9)
{
countryCode = cc;
num.Remove(0, cc.Length);
ok = true;
}
}
}
if (ok)
{
if (std)
{
}
else if (prevPhone != null && prevPhone.Number != null && (((prevPhone.Number.Length == num.Length || prevPhone.Number.Length == (num.Length + 3) || prevPhone.Number.Length == (num.Length + 4)) || prevPhone.m_Template == templ.ToString())))
{
}
else if ((partLength.Count == 3 && partLength[0] == 3 && partLength[1] == 2) && partLength[2] == 2)
{
}
else if (partLength.Count == 3 && isPhoneBefore)
{
}
else if ((partLength.Count == 4 && ((partLength[0] + partLength[1]) == 3) && partLength[2] == 2) && partLength[3] == 2)
{
}
else if ((partLength.Count == 4 && partLength[0] == 3 && partLength[1] == 3) && partLength[2] == 2 && partLength[3] == 2)
{
}
else if (partLength.Count == 5 && (partLength[1] + partLength[2]) == 4 && (partLength[3] + partLength[4]) == 4)
{
}
else if (partLength.Count > 4)
ok = false;
else if (partLength.Count > 3 && cityCode != null)
ok = false;
else if ((isPhoneBefore || cityCode != null || countryCode != null) || additional != null)
ok = true;
else
{
ok = false;
if (((num.Length == 6 || num.Length == 7)) && (partLength.Count < 4) && j > 0)
{
PhoneReferent nextPh = this.GetNextPhone(pli[j - 1].EndToken.Next, lev + 1);
if (nextPh != null)
{
int d = nextPh.Number.Length - num.Length;
if (d == 0 || d == 3 || d == 4)
ok = true;
}
}
}
}
Pullenti.Ner.Token end = (j > 0 ? pli[j - 1].EndToken : null);
if (end == null)
ok = false;
if ((ok && cityCode == null && countryCode == null) && prevPhone == null && !isPhoneBefore)
{
if (!end.IsWhitespaceAfter && end.Next != null)
{
Pullenti.Ner.Token tt = end.Next;
if (tt.IsCharOf(".,)") && tt.Next != null)
tt = tt.Next;
if (!tt.IsWhitespaceBefore)
ok = false;
}
}
if (!ok)
return null;
if (templ.Length > 0 && !char.IsDigit(templ[templ.Length - 1]))
templ.Length--;
if ((countryCode == null && cityCode != null && cityCode.Length > 3) && num.Length > 6)
{
string cc = Pullenti.Ner.Phone.Internal.PhoneHelper.GetCountryPrefix(cityCode);
if (cc != null && ((cc.Length + 1) < cityCode.Length))
{
countryCode = cc;
cityCode = cityCode.Substring(cc.Length);
}
}
if (pli[0].BeginToken.Previous != null)
{
if (pli[0].BeginToken.Previous.IsValue("ГОСТ", null) || pli[0].BeginToken.Previous.IsValue("ТУ", null))
return null;
}
PhoneReferent ph = new PhoneReferent();
if (countryCode != null)
ph.CountryCode = countryCode;
string number = num.ToString();
if ((cityCode == null && num.Length > 7 && partLength.Count > 0) && (partLength[0] < 5))
{
cityCode = number.Substring(0, partLength[0]);
number = number.Substring(partLength[0]);
}
if (cityCode == null && num.Length == 11 && num[0] == '8')
{
cityCode = number.Substring(1, 3);
number = number.Substring(4);
}
if (cityCode == null && num.Length == 10)
{
cityCode = number.Substring(0, 3);
number = number.Substring(3);
}
if (cityCode != null)
number = cityCode + number;
else if (countryCode == null && prevPhone != null)
{
bool ok1 = false;
if (prevPhone.Number.Length >= (number.Length + 2))
ok1 = true;
else if (templ.Length > 0 && prevPhone.m_Template != null && Pullenti.Morph.LanguageHelper.EndsWith(prevPhone.m_Template, templ.ToString()))
ok1 = true;
if (ok1 && prevPhone.Number.Length > number.Length)
number = prevPhone.Number.Substring(0, prevPhone.Number.Length - number.Length) + number;
}
if (ph.CountryCode == null && prevPhone != null && prevPhone.CountryCode != null)
{
if (prevPhone.Number.Length == number.Length)
ph.CountryCode = prevPhone.CountryCode;
}
ok = false;
foreach (char d in number)
{
if (d != '0')
{
ok = true;
break;
}
}
if (!ok)
return null;
if (countryCode != null)
{
if (number.Length < 7)
return null;
}
else
{
string s = Pullenti.Ner.Phone.Internal.PhoneHelper.GetCountryPrefix(number);
if (s != null)
{
string num2 = number.Substring(s.Length);
if (num2.Length >= 10 && num2.Length <= 11)
{
number = num2;
if (s != "7")
ph.CountryCode = s;
}
}
if (number.Length == 8 && prevPhone == null)
return null;
}
if (number.Length > 11)
{
if ((number.Length < 14) && ((countryCode == "1" || countryCode == "43")))
{
}
else
return null;
}
ph.Number = number;
if (additional != null)
ph.AddSlot(PhoneReferent.ATTR_ADDNUMBER, additional, true, 0);
if (!isPhoneBefore && end.Next != null && !end.IsNewlineAfter)
{
if (end.Next.IsCharOf("+=") || end.Next.IsHiphen)
return null;
}
if (countryCode != null && countryCode == "7")
{
if (number.Length != 10)
return null;
}
ph.m_Template = templ.ToString();
if (j == (pli.Count - 1) && pli[j].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Prefix && !pli[j].IsNewlineBefore)
{
end = pli[j].EndToken;
if (pli[j].Kind != PhoneKind.Undefined)
ph.Kind = pli[j].Kind;
}
Pullenti.Ner.ReferentToken res = new Pullenti.Ner.ReferentToken(ph, pli[0].BeginToken, end);
if (pli[0].ItemType == Pullenti.Ner.Phone.Internal.PhoneItemToken.PhoneItemType.Prefix && pli[0].EndToken.Next.IsTableControlChar)
res.BeginToken = pli[1].BeginToken;
return res;
}
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 PhoneItemToken TryAttachAdditional(Pullenti.Ner.Token t0)
{
Pullenti.Ner.Token t = t0;
if (t == null)
{
return(null);
}
if (t.IsChar(','))
{
t = t.Next;
}
else if (t.IsCharOf("*#") && (t.Next is Pullenti.Ner.NumberToken))
{
string val0 = (t.Next as Pullenti.Ner.NumberToken).GetSourceText();
Pullenti.Ner.Token t1 = t.Next;
if ((t1.Next != null && t1.Next.IsHiphen && !t1.IsWhitespaceAfter) && (t1.Next.Next is Pullenti.Ner.NumberToken) && !t1.Next.IsWhitespaceAfter)
{
t1 = t1.Next.Next;
val0 += t1.GetSourceText();
}
if (val0.Length >= 3 && (val0.Length < 7))
{
return new PhoneItemToken(t, t1)
{
ItemType = PhoneItemType.AddNumber, Value = val0
}
}
;
}
bool br = false;
if (t != null && t.IsChar('('))
{
if (t.Previous != null && t.Previous.IsComma)
{
return(null);
}
br = true;
t = t.Next;
}
Pullenti.Ner.Core.TerminToken to = m_PhoneTermins.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No);
if (to == null)
{
if (!br)
{
return(null);
}
if (t0.WhitespacesBeforeCount > 1)
{
return(null);
}
}
else if (to.Termin.Tag == null)
{
return(null);
}
else
{
t = to.EndToken.Next;
}
if (t == null)
{
return(null);
}
if (((t.IsValue("НОМЕР", null) || t.IsValue("N", null) || t.IsValue("#", null)) || t.IsValue("№", null) || t.IsValue("NUMBER", null)) || ((t.IsChar('+') && br)))
{
t = t.Next;
}
else if (to == null && !br)
{
return(null);
}
else if (t.IsValue("НОМ", null) || t.IsValue("ТЕЛ", null))
{
t = t.Next;
if (t != null && t.IsChar('.'))
{
t = t.Next;
}
}
if (t != null && t.IsCharOf(":,") && !t.IsNewlineAfter)
{
t = t.Next;
}
if (!(t is Pullenti.Ner.NumberToken))
{
return(null);
}
string val = (t as Pullenti.Ner.NumberToken).GetSourceText();
if ((t.Next != null && t.Next.IsHiphen && !t.IsWhitespaceAfter) && (t.Next.Next is Pullenti.Ner.NumberToken))
{
val += t.Next.Next.GetSourceText();
t = t.Next.Next;
}
if ((val.Length < 2) || val.Length > 7)
{
return(null);
}
if (br)
{
if (t.Next == null || !t.Next.IsChar(')'))
{
return(null);
}
t = t.Next;
}
PhoneItemToken res = new PhoneItemToken(t0, t)
{
ItemType = PhoneItemType.AddNumber, Value = val
};
return(res);
}
static UriItemToken _AttachUriContent(Pullenti.Ner.Token t0, string chars, bool canBeWhitespaces = false)
{
StringBuilder txt = new StringBuilder();
Pullenti.Ner.Token t1 = t0;
UriItemToken dom = AttachDomainName(t0, true, canBeWhitespaces);
if (dom != null)
{
if (dom.Value.Length < 3)
{
return(null);
}
}
char openChar = (char)0;
Pullenti.Ner.Token t = t0;
if (dom != null)
{
t = dom.EndToken.Next;
}
for (; t != null; t = t.Next)
{
if (t != t0 && t.IsWhitespaceBefore)
{
if (t.IsNewlineBefore || !canBeWhitespaces)
{
break;
}
if (dom == null)
{
break;
}
if (t.Previous.IsHiphen)
{
}
else if (t.Previous.IsCharOf(",;"))
{
break;
}
else if (t.Previous.IsChar('.') && t.Chars.IsLetter && t.LengthChar == 2)
{
}
else
{
bool ok = false;
Pullenti.Ner.Token tt1 = t;
if (t.IsCharOf("\\/"))
{
tt1 = t.Next;
}
Pullenti.Ner.Token tt0 = tt1;
for (; tt1 != null; tt1 = tt1.Next)
{
if (tt1 != tt0 && tt1.IsWhitespaceBefore)
{
break;
}
if (tt1 is Pullenti.Ner.NumberToken)
{
continue;
}
if (!(tt1 is Pullenti.Ner.TextToken))
{
break;
}
string term1 = (tt1 as Pullenti.Ner.TextToken).Term;
if (((term1 == "HTM" || term1 == "HTML" || term1 == "SHTML") || term1 == "ASP" || term1 == "ASPX") || term1 == "JSP")
{
ok = true;
break;
}
if (!tt1.Chars.IsLetter)
{
if (tt1.IsCharOf("\\/"))
{
ok = true;
break;
}
if (!tt1.IsCharOf(chars))
{
break;
}
}
else if (!tt1.Chars.IsLatinLetter)
{
break;
}
}
if (!ok)
{
break;
}
}
}
if (t is Pullenti.Ner.NumberToken)
{
Pullenti.Ner.NumberToken nt = t as Pullenti.Ner.NumberToken;
txt.Append(nt.GetSourceText());
t1 = t;
continue;
}
Pullenti.Ner.TextToken tt = t as Pullenti.Ner.TextToken;
if (tt == null)
{
Pullenti.Ner.ReferentToken rt = t as Pullenti.Ner.ReferentToken;
if (rt != null && rt.BeginToken.IsValue("РФ", null))
{
if (txt.Length > 0 && txt[txt.Length - 1] == '.')
{
txt.Append(rt.BeginToken.GetSourceText());
t1 = t;
continue;
}
}
if (rt != null && rt.Chars.IsLatinLetter && rt.BeginToken == rt.EndToken)
{
txt.Append(rt.BeginToken.GetSourceText());
t1 = t;
continue;
}
break;
}
string src = tt.GetSourceText();
char ch = src[0];
if (!char.IsLetter(ch))
{
if (chars.IndexOf(ch) < 0)
{
break;
}
if (ch == '(' || ch == '[')
{
openChar = ch;
}
else if (ch == ')')
{
if (openChar != '(')
{
break;
}
openChar = (char)0;
}
else if (ch == ']')
{
if (openChar != '[')
{
break;
}
openChar = (char)0;
}
}
txt.Append(src);
t1 = t;
}
if (txt.Length == 0)
{
return(dom);
}
int i;
for (i = 0; i < txt.Length; i++)
{
if (char.IsLetterOrDigit(txt[i]))
{
break;
}
}
if (i >= txt.Length)
{
return(dom);
}
if (txt[txt.Length - 1] == '.' || txt[txt.Length - 1] == '/')
{
txt.Length--;
t1 = t1.Previous;
}
if (dom != null)
{
txt.Insert(0, dom.Value);
}
string tmp = txt.ToString();
if (tmp.StartsWith("\\\\"))
{
txt.Replace("\\\\", "//");
tmp = txt.ToString();
}
if (tmp.StartsWith("//"))
{
tmp = tmp.Substring(2);
}
if (string.Compare(tmp, "WWW", true) == 0)
{
return(null);
}
UriItemToken res = new UriItemToken(t0, t1)
{
Value = txt.ToString()
};
return(res);
}
public static MailLine Parse(Pullenti.Ner.Token t0, int lev, int maxCount = 0)
{
if (t0 == null)
{
return(null);
}
MailLine res = new MailLine(t0, t0);
bool pr = true;
int cou = 0;
for (Pullenti.Ner.Token t = t0; t != null; t = t.Next, cou++)
{
if (t.IsNewlineBefore && t0 != t)
{
break;
}
if (maxCount > 0 && cou > maxCount)
{
break;
}
res.EndToken = t;
if (t.IsTableControlChar || t.IsHiphen)
{
continue;
}
if (pr)
{
if ((t is Pullenti.Ner.TextToken) && t.IsCharOf(">|"))
{
res.Lev++;
}
else
{
pr = false;
Pullenti.Ner.Core.TerminToken tok = m_FromWords.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok != null && tok.EndToken.Next != null && tok.EndToken.Next.IsChar(':'))
{
res.Typ = Types.From;
t = tok.EndToken.Next;
continue;
}
}
}
if (t is Pullenti.Ner.ReferentToken)
{
Pullenti.Ner.Referent r = t.GetReferent();
if (r != null)
{
if ((((r is Pullenti.Ner.Person.PersonReferent) || (r is Pullenti.Ner.Geo.GeoReferent) || (r is Pullenti.Ner.Address.AddressReferent)) || r.TypeName == "PHONE" || r.TypeName == "URI") || (r is Pullenti.Ner.Person.PersonPropertyReferent) || r.TypeName == "ORGANIZATION")
{
res.Refs.Add(r);
}
}
}
}
if (res.Typ == Types.Undefined)
{
Pullenti.Ner.Token t = t0;
for (; t != null && (t.EndChar < res.EndChar); t = t.Next)
{
if (!t.IsHiphen && t.Chars.IsLetter)
{
break;
}
}
int ok = 0;
int nams = 0;
int oth = 0;
Pullenti.Ner.Token lastComma = null;
for (; t != null && (t.EndChar < res.EndChar); t = t.Next)
{
if (t.GetReferent() is Pullenti.Ner.Person.PersonReferent)
{
nams++;
continue;
}
if (t is Pullenti.Ner.TextToken)
{
if (!t.Chars.IsLetter)
{
lastComma = t;
continue;
}
Pullenti.Ner.Core.TerminToken tok = m_HelloWords.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok != null)
{
ok++;
t = tok.EndToken;
continue;
}
if (t.IsValue("ВСЕ", null) || t.IsValue("ALL", null) || t.IsValue("TEAM", null))
{
nams++;
continue;
}
Pullenti.Ner.Person.Internal.PersonItemToken pit = Pullenti.Ner.Person.Internal.PersonItemToken.TryAttach(t, null, Pullenti.Ner.Person.Internal.PersonItemToken.ParseAttr.No, null);
if (pit != null)
{
nams++;
t = pit.EndToken;
continue;
}
}
if ((++oth) > 3)
{
if (ok > 0 && lastComma != null)
{
res.EndToken = lastComma;
oth = 0;
}
break;
}
}
if ((oth < 3) && ok > 0)
{
res.Typ = Types.Hello;
}
}
if (res.Typ == Types.Undefined)
{
int okWords = 0;
if (t0.IsValue("HAVE", null))
{
}
for (Pullenti.Ner.Token t = t0; t != null && t.EndChar <= res.EndChar; t = t.Next)
{
if (!(t is Pullenti.Ner.TextToken))
{
continue;
}
if (t.IsChar('<'))
{
Pullenti.Ner.Core.BracketSequenceToken br = Pullenti.Ner.Core.BracketHelper.TryParse(t, Pullenti.Ner.Core.BracketParseAttr.No, 100);
if (br != null)
{
t = br.EndToken;
continue;
}
}
if (!t.IsLetters || t.IsTableControlChar)
{
continue;
}
Pullenti.Ner.Core.TerminToken tok = m_RegardWords.TryParse(t, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok != null)
{
okWords++;
for (; t != null && t.EndChar <= tok.EndChar; t = t.Next)
{
t.Tag = tok.Termin;
}
t = tok.EndToken;
if ((t.Next is Pullenti.Ner.TextToken) && t.Next.Morph.Case.IsGenitive)
{
for (t = t.Next; t.EndChar <= res.EndChar; t = t.Next)
{
if (t.Morph.Class.IsConjunction)
{
continue;
}
Pullenti.Ner.Core.NounPhraseToken npt1 = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt1 == null)
{
break;
}
if (!npt1.Morph.Case.IsGenitive)
{
break;
}
for (; t.EndChar < npt1.EndChar; t = t.Next)
{
t.Tag = t;
}
t.Tag = t;
}
}
continue;
}
if ((t.Morph.Class.IsPreposition || t.Morph.Class.IsConjunction || t.Morph.Class.IsMisc) || t.IsValue("C", null))
{
continue;
}
if ((okWords > 0 && t.Previous != null && t.Previous.IsComma) && t.Previous.BeginChar > t0.BeginChar && !t.Chars.IsAllLower)
{
res.EndToken = t.Previous;
break;
}
Pullenti.Ner.Core.NounPhraseToken npt = Pullenti.Ner.Core.NounPhraseHelper.TryParse(t, Pullenti.Ner.Core.NounPhraseParseAttr.No, 0, null);
if (npt == null)
{
if ((res.EndChar - t.EndChar) > 10)
{
okWords = 0;
}
break;
}
tok = m_RegardWords.TryParse(npt.EndToken, Pullenti.Ner.Core.TerminParseAttr.No);
if (tok != null && (npt.EndToken is Pullenti.Ner.TextToken))
{
string term = (npt.EndToken as Pullenti.Ner.TextToken).Term;
if (term == "ДЕЛ")
{
tok = null;
}
}
if (tok == null)
{
if (npt.Noun.IsValue("НАДЕЖДА", null))
{
t.Tag = t;
}
else if (okWords > 0 && t.IsValue("NICE", null) && ((res.EndChar - npt.EndChar) < 13))
{
t.Tag = t;
}
else
{
okWords = 0;
}
break;
}
okWords++;
for (; t != null && t.EndChar <= tok.EndChar; t = t.Next)
{
t.Tag = tok.Termin;
}
t = tok.EndToken;
}
if (okWords > 0)
{
res.Typ = Types.BestRegards;
}
}
if (res.Typ == Types.Undefined)
{
Pullenti.Ner.Token t = t0;
for (; t != null && (t.EndChar < res.EndChar); t = t.Next)
{
if (!(t is Pullenti.Ner.TextToken))
{
break;
}
else if (!t.IsHiphen && t.Chars.IsLetter)
{
break;
}
}
if (t != null)
{
if (t != t0)
{
}
if (((t.IsValue("ПЕРЕСЫЛАЕМОЕ", null) || t.IsValue("ПЕРЕАДРЕСОВАННОЕ", null))) && t.Next != null && t.Next.IsValue("СООБЩЕНИЕ", null))
{
res.Typ = Types.From;
res.MustBeFirstLine = true;
}
else if ((t.IsValue("НАЧАЛО", null) && t.Next != null && ((t.Next.IsValue("ПЕРЕСЫЛАЕМОЕ", null) || t.Next.IsValue("ПЕРЕАДРЕСОВАННОЕ", null)))) && t.Next.Next != null && t.Next.Next.IsValue("СООБЩЕНИЕ", null))
{
res.Typ = Types.From;
res.MustBeFirstLine = true;
}
else if (t.IsValue("ORIGINAL", null) && t.Next != null && ((t.Next.IsValue("MESSAGE", null) || t.Next.IsValue("APPOINTMENT", null))))
{
res.Typ = Types.From;
res.MustBeFirstLine = true;
}
else if (t.IsValue("ПЕРЕСЛАНО", null) && t.Next != null && t.Next.IsValue("ПОЛЬЗОВАТЕЛЕМ", null))
{
res.Typ = Types.From;
res.MustBeFirstLine = true;
}
else if (((t.GetReferent() != null && t.GetReferent().TypeName == "DATE")) || ((t.IsValue("IL", null) && t.Next != null && t.Next.IsValue("GIORNO", null))) || ((t.IsValue("ON", null) && (t.Next is Pullenti.Ner.ReferentToken) && t.Next.GetReferent().TypeName == "DATE")))
{
bool hasFrom = false;
bool hasDate = t.GetReferent() != null && t.GetReferent().TypeName == "DATE";
if (t.IsNewlineAfter && (lev < 5))
{
MailLine res1 = Parse(t.Next, lev + 1, 0);
if (res1 != null && res1.Typ == Types.Hello)
{
res.Typ = Types.From;
}
}
MailLine next = Parse(res.EndToken.Next, lev + 1, 0);
if (next != null)
{
if (next.Typ != Types.Undefined)
{
next = null;
}
}
int tmax = res.EndChar;
if (next != null)
{
tmax = next.EndChar;
}
Pullenti.Ner.Core.BracketSequenceToken br1 = null;
for (; t != null && t.EndChar <= tmax; t = t.Next)
{
if (t.IsValue("ОТ", null) || t.IsValue("FROM", null))
{
hasFrom = true;
}
else if (t.GetReferent() != null && ((t.GetReferent().TypeName == "URI" || (t.GetReferent() is Pullenti.Ner.Person.PersonReferent))))
{
if (t.GetReferent().TypeName == "URI" && hasDate)
{
if (br1 != null)
{
hasFrom = true;
next = null;
}
if (t.Previous.IsChar('<') && t.Next != null && t.Next.IsChar('>'))
{
t = t.Next;
if (t.Next != null && t.Next.IsChar(':'))
{
t = t.Next;
}
if (t.IsNewlineAfter)
{
hasFrom = true;
next = null;
}
}
}
for (t = t.Next; t != null && t.EndChar <= res.EndChar; t = t.Next)
{
if (t.IsValue("HA", null) && t.Next != null && t.Next.IsValue("SCRITTO", null))
{
hasFrom = true;
break;
}
else if (((t.IsValue("НАПИСАТЬ", null) || t.IsValue("WROTE", null))) && ((res.EndChar - t.EndChar) < 10))
{
hasFrom = true;
break;
}
}
if (hasFrom)
{
res.Typ = Types.From;
if (next != null && t.EndChar >= next.BeginChar)
{
res.EndToken = next.EndToken;
}
}
break;
}
else if (br1 == null && !t.IsChar('<') && Pullenti.Ner.Core.BracketHelper.CanBeStartOfSequence(t, true, false))
{
br1 = Pullenti.Ner.Core.BracketHelper.TryParse(t, Pullenti.Ner.Core.BracketParseAttr.No, 100);
if (br1 != null)
{
t = br1.EndToken;
}
}
}
}
else
{
bool hasUri = false;
for (; t != null && (t.EndChar < res.EndChar); t = t.Next)
{
if (t.GetReferent() != null && ((t.GetReferent().TypeName == "URI" || (t.GetReferent() is Pullenti.Ner.Person.PersonReferent))))
{
hasUri = true;
}
else if (t.IsValue("ПИСАТЬ", null) && hasUri)
{
if (t.Next != null && t.Next.IsChar('('))
{
if (hasUri)
{
res.Typ = Types.From;
}
break;
}
}
}
}
}
}
return(res);
}
public static List <UriItemToken> AttachMailUsers(Pullenti.Ner.Token t1)
{
if (t1 == null)
{
return(null);
}
if (t1.IsChar('}'))
{
List <UriItemToken> res0 = AttachMailUsers(t1.Previous);
if (res0 == null)
{
return(null);
}
t1 = res0[0].BeginToken.Previous;
for (; t1 != null; t1 = t1.Previous)
{
if (t1.IsChar('{'))
{
res0[0].BeginToken = t1;
return(res0);
}
if (t1.IsCharOf(";,"))
{
continue;
}
List <UriItemToken> res1 = AttachMailUsers(t1);
if (res1 == null)
{
return(null);
}
res0.Insert(0, res1[0]);
t1 = res1[0].BeginToken;
}
return(null);
}
StringBuilder txt = new StringBuilder();
Pullenti.Ner.Token t0 = t1;
for (Pullenti.Ner.Token t = t1; t != null; t = t.Previous)
{
if (t.IsWhitespaceAfter)
{
break;
}
if (t is Pullenti.Ner.NumberToken)
{
Pullenti.Ner.NumberToken nt = t as Pullenti.Ner.NumberToken;
txt.Insert(0, nt.GetSourceText());
t0 = t;
continue;
}
Pullenti.Ner.TextToken tt = t as Pullenti.Ner.TextToken;
if (tt == null)
{
break;
}
string src = tt.GetSourceText();
char ch = src[0];
if (!char.IsLetter(ch))
{
if (".-_".IndexOf(ch) < 0)
{
break;
}
}
txt.Insert(0, src);
t0 = t;
}
if (txt.Length == 0)
{
return(null);
}
List <UriItemToken> res = new List <UriItemToken>();
res.Add(new UriItemToken(t0, t1)
{
Value = txt.ToString().ToLower()
});
return(res);
}
public static Pullenti.Ner.ReferentToken TryAttachOrg(Pullenti.Ner.Token t, bool canBeCyr = false)
{
if (t == null)
{
return(null);
}
bool br = false;
if (t.IsChar('(') && t.Next != null)
{
t = t.Next;
br = true;
}
if (t is Pullenti.Ner.NumberToken)
{
if ((t as Pullenti.Ner.NumberToken).Typ == Pullenti.Ner.NumberSpellingType.Words && t.Morph.Class.IsAdjective && t.Chars.IsCapitalUpper)
{
}
else
{
return(null);
}
}
else
{
if (t.Chars.IsAllLower)
{
return(null);
}
if ((t.LengthChar < 3) && !t.Chars.IsLetter)
{
return(null);
}
if (!t.Chars.IsLatinLetter)
{
if (!canBeCyr || !t.Chars.IsCyrillicLetter)
{
return(null);
}
}
}
Pullenti.Ner.Token t0 = t;
Pullenti.Ner.Token t1 = t0;
int namWo = 0;
OrgItemEngItem tok = null;
Pullenti.Ner.Geo.GeoReferent geo = null;
OrgItemTypeToken addTyp = null;
for (; t != null; t = t.Next)
{
if (t != t0 && t.WhitespacesBeforeCount > 1)
{
break;
}
if (t.IsChar(')'))
{
break;
}
if (t.IsChar('(') && t.Next != null)
{
if ((t.Next.GetReferent() is Pullenti.Ner.Geo.GeoReferent) && t.Next.Next != null && t.Next.Next.IsChar(')'))
{
geo = t.Next.GetReferent() as Pullenti.Ner.Geo.GeoReferent;
t = t.Next.Next;
continue;
}
OrgItemTypeToken typ = OrgItemTypeToken.TryAttach(t.Next, true, null);
if ((typ != null && typ.EndToken.Next != null && typ.EndToken.Next.IsChar(')')) && typ.Chars.IsLatinLetter)
{
addTyp = typ;
t = typ.EndToken.Next;
continue;
}
if (((t.Next is Pullenti.Ner.TextToken) && t.Next.Next != null && t.Next.Next.IsChar(')')) && t.Next.Chars.IsCapitalUpper)
{
t1 = (t = t.Next.Next);
continue;
}
break;
}
tok = TryAttach(t, canBeCyr);
if (tok == null && t.IsCharOf(".,") && t.Next != null)
{
tok = TryAttach(t.Next, canBeCyr);
if (tok == null && t.Next.IsCharOf(",."))
{
tok = TryAttach(t.Next.Next, canBeCyr);
}
}
if (tok != null)
{
if (tok.LengthChar == 1 && t0.Chars.IsCyrillicLetter)
{
return(null);
}
break;
}
if (t.IsHiphen && !t.IsWhitespaceAfter && !t.IsWhitespaceBefore)
{
continue;
}
if (t.IsCharOf("&+") || t.IsAnd)
{
continue;
}
if (t.IsChar('.'))
{
if (t.Previous != null && t.Previous.LengthChar == 1)
{
continue;
}
else if (Pullenti.Ner.Core.MiscHelper.CanBeStartOfSentence(t.Next))
{
break;
}
}
if (!t.Chars.IsLatinLetter)
{
if (!canBeCyr || !t.Chars.IsCyrillicLetter)
{
break;
}
}
if (t.Chars.IsAllLower)
{
if (t.Morph.Class.IsPreposition || t.Morph.Class.IsConjunction)
{
continue;
}
if (br)
{
continue;
}
break;
}
Pullenti.Morph.MorphClass mc = t.GetMorphClassInDictionary();
if (mc.IsVerb)
{
if (t.Next != null && t.Next.Morph.Class.IsPreposition)
{
break;
}
}
if (t.Next != null && t.Next.IsValue("OF", null))
{
break;
}
if (t is Pullenti.Ner.TextToken)
{
namWo++;
}
t1 = t;
}
if (tok == null)
{
return(null);
}
if (t0 == tok.BeginToken)
{
Pullenti.Ner.Core.BracketSequenceToken br2 = Pullenti.Ner.Core.BracketHelper.TryParse(tok.EndToken.Next, Pullenti.Ner.Core.BracketParseAttr.No, 100);
if (br2 != null)
{
Pullenti.Ner.Org.OrganizationReferent org1 = new Pullenti.Ner.Org.OrganizationReferent();
if (tok.ShortValue != null)
{
org1.AddTypeStr(tok.ShortValue);
}
org1.AddTypeStr(tok.FullValue);
string nam1 = Pullenti.Ner.Core.MiscHelper.GetTextValue(br2.BeginToken, br2.EndToken, Pullenti.Ner.Core.GetTextAttr.No);
if (nam1 != null)
{
org1.AddName(nam1, true, null);
return(new Pullenti.Ner.ReferentToken(org1, t0, br2.EndToken));
}
}
return(null);
}
Pullenti.Ner.Org.OrganizationReferent org = new Pullenti.Ner.Org.OrganizationReferent();
Pullenti.Ner.Token te = tok.EndToken;
if (tok.IsBank)
{
t1 = tok.EndToken;
}
if (tok.FullValue == "company" && (tok.WhitespacesAfterCount < 3))
{
OrgItemEngItem tok1 = TryAttach(tok.EndToken.Next, canBeCyr);
if (tok1 != null)
{
t1 = tok.EndToken;
tok = tok1;
te = tok.EndToken;
}
}
if (tok.FullValue == "company")
{
if (namWo == 0)
{
return(null);
}
}
string nam = Pullenti.Ner.Core.MiscHelper.GetTextValue(t0, t1, Pullenti.Ner.Core.GetTextAttr.IgnoreArticles);
if (nam == "STOCK" && tok.FullValue == "company")
{
return(null);
}
string altNam = null;
if (string.IsNullOrEmpty(nam))
{
return(null);
}
if (nam.IndexOf('(') > 0)
{
int i1 = nam.IndexOf('(');
int i2 = nam.IndexOf(')');
if (i1 < i2)
{
altNam = nam;
string tai = null;
if ((i2 + 1) < nam.Length)
{
tai = nam.Substring(i2).Trim();
}
nam = nam.Substring(0, i1).Trim();
if (tai != null)
{
nam = string.Format("{0} {1}", nam, tai);
}
}
}
if (tok.IsBank)
{
org.AddTypeStr((tok.Kit.BaseLanguage.IsEn ? "bank" : "банк"));
org.AddProfile(Pullenti.Ner.Org.OrgProfile.Finance);
if ((t1.Next != null && t1.Next.IsValue("OF", null) && t1.Next.Next != null) && t1.Next.Next.Chars.IsLatinLetter)
{
OrgItemNameToken nam0 = OrgItemNameToken.TryAttach(t1.Next, null, false, false);
if (nam0 != null)
{
te = nam0.EndToken;
}
else
{
te = t1.Next.Next;
}
nam = Pullenti.Ner.Core.MiscHelper.GetTextValue(t0, te, Pullenti.Ner.Core.GetTextAttr.No);
if (te.GetReferent() is Pullenti.Ner.Geo.GeoReferent)
{
org.AddGeoObject(te.GetReferent() as Pullenti.Ner.Geo.GeoReferent);
}
}
else if (t0 == t1)
{
return(null);
}
}
else
{
if (tok.ShortValue != null)
{
org.AddTypeStr(tok.ShortValue);
}
org.AddTypeStr(tok.FullValue);
}
if (string.IsNullOrEmpty(nam))
{
return(null);
}
org.AddName(nam, true, null);
if (altNam != null)
{
org.AddName(altNam, true, null);
}
Pullenti.Ner.ReferentToken res = new Pullenti.Ner.ReferentToken(org, t0, te);
t = te;
while (t.Next != null)
{
if (t.Next.IsCharOf(",."))
{
t = t.Next;
}
else
{
break;
}
}
if (t.WhitespacesAfterCount < 2)
{
tok = TryAttach(t.Next, canBeCyr);
if (tok != null)
{
if (tok.ShortValue != null)
{
org.AddTypeStr(tok.ShortValue);
}
org.AddTypeStr(tok.FullValue);
res.EndToken = tok.EndToken;
}
}
if (geo != null)
{
org.AddGeoObject(geo);
}
if (addTyp != null)
{
org.AddType(addTyp, false);
}
if (!br)
{
return(res);
}
t = res.EndToken;
if (t.Next == null || t.Next.IsChar(')'))
{
res.EndToken = t.Next;
}
else
{
return(null);
}
return(res);
}
public Pullenti.Ner.ReferentToken TryAttach(Pullenti.Ner.Token t, bool forOntology = false)
{
if (t == null)
{
return(null);
}
Pullenti.Ner.ReferentToken rt0 = this.TryAttachSpec(t);
if (rt0 != null)
{
return(rt0);
}
if (t.Chars.IsAllLower)
{
if (!t.IsWhitespaceAfter && (t.Next is Pullenti.Ner.NumberToken))
{
if (t.Previous == null || t.IsWhitespaceBefore || t.Previous.IsCharOf(",:"))
{
}
else
{
return(null);
}
}
else
{
return(null);
}
}
StringBuilder tmp = new StringBuilder();
Pullenti.Ner.Token t1 = t;
bool hiph = false;
bool ok = true;
int nums = 0;
int chars = 0;
for (Pullenti.Ner.Token w = t1.Next; w != null; w = w.Next)
{
if (w.IsWhitespaceBefore && !forOntology)
{
break;
}
if (w.IsCharOf("/\\_") || w.IsHiphen)
{
hiph = true;
tmp.Append('-');
continue;
}
hiph = false;
Pullenti.Ner.NumberToken nt = w as Pullenti.Ner.NumberToken;
if (nt != null)
{
if (nt.Typ != Pullenti.Ner.NumberSpellingType.Digit)
{
break;
}
t1 = nt;
tmp.Append(nt.GetSourceText());
nums++;
continue;
}
Pullenti.Ner.TextToken tt = w as Pullenti.Ner.TextToken;
if (tt == null)
{
break;
}
if (tt.LengthChar > 3)
{
ok = false;
break;
}
if (!char.IsLetter(tt.Term[0]))
{
if (tt.IsCharOf(",:") || Pullenti.Ner.Core.BracketHelper.CanBeEndOfSequence(tt, false, null, false))
{
break;
}
if (!tt.IsCharOf("+*&^#@!"))
{
ok = false;
break;
}
chars++;
}
t1 = tt;
tmp.Append(tt.GetSourceText());
}
if (!forOntology)
{
if ((tmp.Length < 1) || !ok || hiph)
{
return(null);
}
if (tmp.Length > 12)
{
return(null);
}
char last = tmp[tmp.Length - 1];
if (last == '!')
{
return(null);
}
if ((nums + chars) == 0)
{
return(null);
}
if (!this.CheckAttach(t, t1))
{
return(null);
}
}
DenominationReferent newDr = new DenominationReferent();
newDr.AddValue(t, t1);
return(new Pullenti.Ner.ReferentToken(newDr, t, t1));
}
public static Pullenti.Ner.Token CreateNickname(Pullenti.Ner.Person.PersonReferent pr, Pullenti.Ner.Token t)
{
bool hasKeyw = false;
bool isBr = false;
for (; t != null; t = t.Next)
{
if (t.IsHiphen || t.IsComma || t.IsCharOf(".:;"))
{
continue;
}
if (t.Morph.Class.IsPreposition)
{
continue;
}
if (t.IsChar('('))
{
isBr = true;
continue;
}
if ((t.IsValue("ПРОЗВИЩЕ", "ПРІЗВИСЬКО") || t.IsValue("КЛИЧКА", null) || t.IsValue("ПСЕВДОНИМ", "ПСЕВДОНІМ")) || t.IsValue("ПСЕВДО", null) || t.IsValue("ПОЗЫВНОЙ", "ПОЗИВНИЙ"))
{
hasKeyw = true;
continue;
}
break;
}
if (!hasKeyw || t == null)
{
return(null);
}
if (Pullenti.Ner.Core.BracketHelper.IsBracket(t, true))
{
Pullenti.Ner.Core.BracketSequenceToken br = Pullenti.Ner.Core.BracketHelper.TryParse(t, Pullenti.Ner.Core.BracketParseAttr.No, 100);
if (br != null)
{
string ni = Pullenti.Ner.Core.MiscHelper.GetTextValue(br.BeginToken.Next, br.EndToken.Previous, Pullenti.Ner.Core.GetTextAttr.No);
if (ni != null)
{
pr.AddSlot(Pullenti.Ner.Person.PersonReferent.ATTR_NICKNAME, ni, false, 0);
t = br.EndToken;
for (Pullenti.Ner.Token tt = t.Next; tt != null; tt = tt.Next)
{
if (tt.IsCommaAnd)
{
continue;
}
if (!Pullenti.Ner.Core.BracketHelper.IsBracket(tt, true))
{
break;
}
br = Pullenti.Ner.Core.BracketHelper.TryParse(tt, Pullenti.Ner.Core.BracketParseAttr.No, 100);
if (br == null)
{
break;
}
ni = Pullenti.Ner.Core.MiscHelper.GetTextValue(br.BeginToken.Next, br.EndToken.Previous, Pullenti.Ner.Core.GetTextAttr.No);
if (ni != null)
{
pr.AddSlot(Pullenti.Ner.Person.PersonReferent.ATTR_NICKNAME, ni, false, 0);
}
t = (tt = br.EndToken);
}
if (isBr && t.Next != null && t.Next.IsChar(')'))
{
t = t.Next;
}
return(t);
}
}
}
else
{
Pullenti.Ner.Token ret = null;
for (; t != null; t = t.Next)
{
if (t.IsCommaAnd)
{
continue;
}
if (ret != null && t.Chars.IsAllLower)
{
break;
}
if (t.WhitespacesBeforeCount > 2)
{
break;
}
List <PersonItemToken> pli = PersonItemToken.TryAttachList(t, null, PersonItemToken.ParseAttr.No, 10);
if (pli != null && ((pli.Count == 1 || pli.Count == 2)))
{
string ni = Pullenti.Ner.Core.MiscHelper.GetTextValue(pli[0].BeginToken, pli[pli.Count - 1].EndToken, Pullenti.Ner.Core.GetTextAttr.No);
if (ni != null)
{
pr.AddSlot(Pullenti.Ner.Person.PersonReferent.ATTR_NICKNAME, ni, false, 0);
t = pli[pli.Count - 1].EndToken;
if (isBr && t.Next != null && t.Next.IsChar(')'))
{
t = t.Next;
}
ret = t;
continue;
}
}
if ((t is Pullenti.Ner.ReferentToken) && !t.Chars.IsAllLower && (t as Pullenti.Ner.ReferentToken).BeginToken == (t as Pullenti.Ner.ReferentToken).EndToken)
{
string val = Pullenti.Ner.Core.MiscHelper.GetTextValueOfMetaToken(t as Pullenti.Ner.ReferentToken, Pullenti.Ner.Core.GetTextAttr.No);
pr.AddSlot(Pullenti.Ner.Person.PersonReferent.ATTR_NICKNAME, val, false, 0);
if (isBr && t.Next != null && t.Next.IsChar(')'))
{
t = t.Next;
}
ret = t;
continue;
}
break;
}
return(ret);
}
return(null);
}