public override bool IncrementToken()
{
if (input.IncrementToken())
{
char[] termBuffer = termAtt.TermBuffer();
var termText = new String(termBuffer, 0, termAtt.TermLength());
byte[] collationKey = cultureInfo.CompareInfo.GetSortKey(termText).KeyData;
int encodedLength = IndexableBinaryStringTools_UsingArrays.GetEncodedLength(collationKey);
if (encodedLength > termBuffer.Length)
{
termAtt.ResizeTermBuffer(encodedLength);
}
termAtt.SetTermLength(encodedLength);
IndexableBinaryStringTools_UsingArrays.Encode(collationKey, termBuffer);
return(true);
}
return(false);
}
public static IEnumerable <string> TokensFromAnalysis(Analyzer analyzer, String text)
{
TokenStream stream = analyzer.TokenStream("contents", new StringReader(text));
List <string> result = new List <string>();
TermAttribute tokenAttr = (TermAttribute)stream.GetAttribute(typeof(TermAttribute));
while (stream.IncrementToken())
{
Console.WriteLine("Buffer:={0}, Length:={1}, Term:={2}".FormatWith(tokenAttr.TermBuffer(), tokenAttr.TermLength(), tokenAttr.Term()));
result.Add(tokenAttr.Term());
}
//tokenAttr.
stream.End();
stream.Close();
return(result);
}
public override bool IncrementToken()
{
if (input.IncrementToken() == false)
{
return(false);
}
var termBuffer = _termAtt.TermBuffer();
var termText = new string(termBuffer, 0, _termAtt.TermLength());
var collationKey = GetCollationKey(termText);
var encodedLength = IndexableBinaryStringTools_UsingArrays.GetEncodedLength(collationKey);
if (encodedLength > termBuffer.Length)
{
termBuffer = _termAtt.ResizeTermBuffer(encodedLength);
}
_termAtt.SetTermLength(encodedLength);
IndexableBinaryStringTools_UsingArrays.Encode(collationKey, termBuffer);
return(true);
}
/** Returns the next token in the stream, or null at EOS. */
public override bool IncrementToken()
{
while (true)
{
if (_curTermBuffer == null)
{
if (!input.IncrementToken())
{
return(false);
}
else
{
_curTermBuffer = (char[])_termAtt.TermBuffer().Clone();
_curTermLength = _termAtt.TermLength();
_curGramSize = _minGram;
_curPos = 0;
_tokStart = _offsetAtt.StartOffset;
}
}
while (_curGramSize <= _maxGram)
{
if (_curPos + _curGramSize <= _curTermLength)
{ // while there is input
ClearAttributes();
_termAtt.SetTermBuffer(_curTermBuffer, _curPos, _curGramSize);
_offsetAtt.SetOffset(_tokStart + _curPos, _tokStart + _curPos + _curGramSize);
_curPos++;
return(true);
}
_curGramSize++; // increase n-gram size
_curPos = 0;
}
_curTermBuffer = null;
}
}
