public void SetNextPrev(word_t next, NerOutputType nerOutputType)
{
nerNext = next;
next.nerPrev = this;
//System.Diagnostics.Debug.Assert( next != this );
this.nerOutputType = next.nerOutputType = nerOutputType;
}
private void TryCreateWordAndPut2List()
{
const int THAT_LENGTH = 2;
const int DUSH_LENGTH = 1;
const int PARTICLE_THAT_LENGTH = DUSH_LENGTH + THAT_LENGTH;
const string THAT_UPPER = "ТО";
const string THAT_LOWER = "то";
const string DUSH = "-";
const char T_RU = 'Т';
const char O_RU = 'О';
if (_Length != 0)
{
#region [.to upper invariant & pos-tagger-list & etc.]
var startPtr = _BASE + _StartIndex;
if (_WordToUpperBufferSize < _Length)
{
ReAllocWordToUpperBuffer(_Length);
}
for (int i = 0; i < _Length; i++)
{
*(_WordToUpperBufferPtrBase + i) = *(_UIM + *(startPtr + i));
}
var valueUpper = new string( _WordToUpperBufferPtrBase, 0, _Length );
#region [.detect particle-that.]
var len = _Length - PARTICLE_THAT_LENGTH;
if (0 < len)
{
//ends-with "-ТО"
if ((*(_CTM + *(_WordToUpperBufferPtrBase + len)) & CharType.IsHyphen) == CharType.IsHyphen &&
(*(_WordToUpperBufferPtrBase + ++len) == T_RU) &&
(*(_WordToUpperBufferPtrBase + ++len) == O_RU)
)
{
//need split particle-that
if (!_ParticleThatExclusion.Contains(valueUpper))
{
#region [.create word with split particle-that.]
len = _Length - PARTICLE_THAT_LENGTH;
valueUpper = new string( _WordToUpperBufferPtrBase, 0, len );
#region [.create word without particle-that.]
var _valueOriginal = new string( _BASE, _StartIndex, len );
var _word = new word_t()
{
startIndex = _StartIndex,
length = len,
valueOriginal = _valueOriginal,
valueUpper = valueUpper,
};
#endregion
#region [.nerInputType.]
_word.nerInputType = _NerInputTypeProcessor.GetNerInputType(startPtr, len);
#endregion
#region [.posTaggerInputType.]
//if ( _Make_PosTagger )
{
var _result = _PosTaggerInputTypeProcessor.GetResult(startPtr, len, _word);
_word.posTaggerInputType = _result.posTaggerInputType;
_word.posTaggerExtraWordType = _result.posTaggerExtraWordType;
_word.posTaggerFirstCharIsUpper = ((*(_CTM + *startPtr) & CharType.IsUpper) == CharType.IsUpper);
_word.posTaggerLastValueUpperInNumeralChain = (_result.posTaggerLastValueUpperInNumeralChainIsValueOriginal)
? _word.valueUpper : _result.posTaggerLastValueUpperInNumeralChain;
}
#endregion
#region [.put-2-list.]
_Words.Add(_word);
#endregion
#region [.create word particle-that.]
var startIndex = _StartIndex + len;
//_valueOriginal = new string( _BASE, startIndex, 1 );
_word = new word_t()
{
startIndex = startIndex,
length = DUSH_LENGTH,
valueOriginal = DUSH,
valueUpper = DUSH,
posTaggerInputType = PosTaggerInputType.Dush,
posTaggerExtraWordType = PosTaggerExtraWordType.Punctuation,
//nerInputType = NerInputType.O,
};
_Words.Add(_word);
startIndex++;
//valueUpper = new string( chars_ptr, len, 2 );
//_valueOriginal = new string( _BASE, startIndex, 2 );
_word = new word_t()
{
startIndex = startIndex,
length = THAT_LENGTH,
valueOriginal = THAT_LOWER,
valueUpper = THAT_UPPER,
posTaggerInputType = PosTaggerInputType.O,
//nerInputType = NerInputType.O,
};
_Words.Add(_word);
#endregion
#region [.inctement start-index.]
_StartIndex += _Length;
_Length = 0;
#endregion
#region [.return.]
return;
#endregion
#endregion
}
}
}
#endregion
#endregion
#region [.create word.]
var valueOriginal = new string( _BASE, _StartIndex, _Length );
var word = new word_t()
{
startIndex = _StartIndex,
length = _Length,
valueOriginal = valueOriginal,
valueUpper = valueUpper,
};
#endregion
#region [.nerInputType.]
word.nerInputType = _NerInputTypeProcessor.GetNerInputType(_BASE + _StartIndex, _Length);
#endregion
#region [.posTaggerInputType.]
//if ( _Make_PosTagger )
{
var result = _PosTaggerInputTypeProcessor.GetResult(startPtr, _Length, word);
word.posTaggerInputType = result.posTaggerInputType;
word.posTaggerExtraWordType = result.posTaggerExtraWordType;
word.posTaggerFirstCharIsUpper = ((*(_CTM + *startPtr) & CharType.IsUpper) == CharType.IsUpper);
word.posTaggerLastValueUpperInNumeralChain = (result.posTaggerLastValueUpperInNumeralChainIsValueOriginal)
? word.valueUpper : result.posTaggerLastValueUpperInNumeralChain;
}
#endregion
#region [.put-2-list.]
_Words.Add(word);
#endregion
#region [.inctement start-index.]
_StartIndex += _Length;
_Length = 0;
#endregion
}
}
