public List <string> GetNGrams(string TextFieldKey, string document)
{
List <string> ngrams = new List <string>();
if (LuceneService.NGrammer != null)
{
TokenStream stream = LuceneService.NGrammer.GetTokenStream(TextFieldKey, new StringReader(document));
//AttributeSource source = new AttributeSource();
//OffsetAttribute offsetAttribute = stream.AddAttribute<OffsetAttribute>();
var charTermAttribute = stream.AddAttribute <ICharTermAttribute>();
stream.Reset();
while (stream.IncrementToken())
{
//Token token = new Token();
//int startOffset = offsetAttribute.StartOffset;
//int endOffset = offsetAttribute.EndOffset;
String term = charTermAttribute.ToString();
ngrams.Add(term);
}
stream.ClearAttributes();
stream.End();
stream.Dispose();
return(ngrams);
}
else
{
throw new Exception("No ngrammer");
}
}
public static Dictionary <int, List <int> > GetTokenDataForDoc(string document)
{
Dictionary <int, List <int> > res = new Dictionary <int, List <int> >();
List <int> list = new List <int>();
TokenStream stream = LuceneService.Analyzer.GetTokenStream(ProjectInfo.TextFieldKey, new StringReader(document));
var index = 0;
var charTermAttribute = stream.AddAttribute <ICharTermAttribute>();
stream.Reset();
while (stream.IncrementToken())
{
//Token token = new Token();
//int startOffset = offsetAttribute.StartOffset;
//int endOffset = offsetAttribute.EndOffset;
String term = charTermAttribute.ToString();
index++;
list.Add(term.Length);
}
res.Add(index, list);
stream.ClearAttributes();
stream.End();
stream.Dispose();
return(res);
}
public static void AssertTokenStreamContents(TokenStream ts, String[] output, Int32[] startOffsets, Int32[] endOffsets, String[] types, Int32[] posIncrements, Int32? finalOffset) {
Assert.IsNotNull(output);
var checkClearAtt = (CheckClearAttributesAttribute) ts.AddAttribute(typeof (CheckClearAttributesAttribute));
Assert.IsTrue(ts.HasAttribute(typeof (TermAttribute)), "has no TermAttribute");
var termAtt = (TermAttribute) ts.GetAttribute(typeof (TermAttribute));
OffsetAttribute offsetAtt = null;
if (startOffsets != null || endOffsets != null || finalOffset != null) {
Assert.IsTrue(ts.HasAttribute(typeof (OffsetAttribute)), "has no OffsetAttribute");
offsetAtt = (OffsetAttribute) ts.GetAttribute(typeof (OffsetAttribute));
}
TypeAttribute typeAtt = null;
if (types != null) {
Assert.IsTrue(ts.HasAttribute(typeof (TypeAttribute)), "has no TypeAttribute");
typeAtt = (TypeAttribute) ts.GetAttribute(typeof (TypeAttribute));
}
PositionIncrementAttribute posIncrAtt = null;
if (posIncrements != null) {
Assert.IsTrue(ts.HasAttribute(typeof (PositionIncrementAttribute)), "has no PositionIncrementAttribute");
posIncrAtt = (PositionIncrementAttribute) ts.GetAttribute(typeof (PositionIncrementAttribute));
}
ts.Reset();
for (Int32 i = 0; i < output.Length; i++) {
// extra safety to enforce, that the state is not preserved and also assign bogus values
ts.ClearAttributes();
termAtt.SetTermBuffer("bogusTerm");
if (offsetAtt != null) offsetAtt.SetOffset(14584724, 24683243);
if (typeAtt != null) typeAtt.SetType("bogusType");
if (posIncrAtt != null) posIncrAtt.SetPositionIncrement(45987657);
checkClearAtt.GetAndResetClearCalled(); // reset it, because we called clearAttribute() before
Assert.IsTrue(ts.IncrementToken(), "token " + i + " does not exist");
Assert.IsTrue(checkClearAtt.GetAndResetClearCalled(), "clearAttributes() was not called correctly in TokenStream chain");
Assert.AreEqual(output[i], termAtt.Term(), "term " + i);
if (startOffsets != null)
Assert.AreEqual(startOffsets[i], offsetAtt.StartOffset(), "startOffset " + i);
if (endOffsets != null)
Assert.AreEqual(endOffsets[i], offsetAtt.EndOffset(), "endOffset " + i);
if (types != null)
Assert.AreEqual(types[i], typeAtt.Type(), "type " + i);
if (posIncrements != null)
Assert.AreEqual(posIncrements[i], posIncrAtt.GetPositionIncrement(), "posIncrement " + i);
}
Assert.IsFalse(ts.IncrementToken(), "end of stream");
ts.End();
if (finalOffset.HasValue)
Assert.AreEqual(finalOffset.Value, offsetAtt.EndOffset(), "finalOffset ");
ts.Close();
}
/// <summary>
/// TODO: rewrite tests not to use string comparison.
/// </summary>
private static string tsToString(TokenStream @in)
{
StringBuilder @out = new StringBuilder();
ICharTermAttribute termAtt = @in.AddAttribute <ICharTermAttribute>();
// extra safety to enforce, that the state is not preserved and also
// assign bogus values
@in.ClearAttributes();
termAtt.SetEmpty().Append("bogusTerm");
@in.Reset();
while (@in.IncrementToken())
{
if (@out.Length > 0)
{
@out.Append(' ');
}
@out.Append(termAtt.ToString());
@in.ClearAttributes();
termAtt.SetEmpty().Append("bogusTerm");
}
@in.Dispose();
return(@out.ToString());
}
public static void AssertTokenStreamContents(TokenStream ts, System.String[] output, int[] startOffsets, int[] endOffsets, System.String[] types, int[] posIncrements, int?finalOffset)
{
Assert.IsNotNull(output);
ICheckClearAttributesAttribute checkClearAtt = ts.AddAttribute <ICheckClearAttributesAttribute>();
Assert.IsTrue(ts.HasAttribute <ITermAttribute>(), "has no TermAttribute");
ITermAttribute termAtt = ts.GetAttribute <ITermAttribute>();
IOffsetAttribute offsetAtt = null;
if (startOffsets != null || endOffsets != null || finalOffset != null)
{
Assert.IsTrue(ts.HasAttribute <IOffsetAttribute>(), "has no OffsetAttribute");
offsetAtt = ts.GetAttribute <IOffsetAttribute>();
}
ITypeAttribute typeAtt = null;
if (types != null)
{
Assert.IsTrue(ts.HasAttribute <ITypeAttribute>(), "has no TypeAttribute");
typeAtt = ts.GetAttribute <ITypeAttribute>();
}
IPositionIncrementAttribute posIncrAtt = null;
if (posIncrements != null)
{
Assert.IsTrue(ts.HasAttribute <IPositionIncrementAttribute>(), "has no PositionIncrementAttribute");
posIncrAtt = ts.GetAttribute <IPositionIncrementAttribute>();
}
ts.Reset();
for (int i = 0; i < output.Length; i++)
{
// extra safety to enforce, that the state is not preserved and also assign bogus values
ts.ClearAttributes();
termAtt.SetTermBuffer("bogusTerm");
if (offsetAtt != null)
{
offsetAtt.SetOffset(14584724, 24683243);
}
if (typeAtt != null)
{
typeAtt.Type = "bogusType";
}
if (posIncrAtt != null)
{
posIncrAtt.PositionIncrement = 45987657;
}
checkClearAtt.GetAndResetClearCalled(); // reset it, because we called clearAttribute() before
Assert.IsTrue(ts.IncrementToken(), "token " + i + " does not exist");
Assert.IsTrue(checkClearAtt.GetAndResetClearCalled(), "clearAttributes() was not called correctly in TokenStream chain");
Assert.AreEqual(output[i], termAtt.Term, "term " + i);
if (startOffsets != null)
{
Assert.AreEqual(startOffsets[i], offsetAtt.StartOffset, "startOffset " + i);
}
if (endOffsets != null)
{
Assert.AreEqual(endOffsets[i], offsetAtt.EndOffset, "endOffset " + i);
}
if (types != null)
{
Assert.AreEqual(types[i], typeAtt.Type, "type " + i);
}
if (posIncrements != null)
{
Assert.AreEqual(posIncrements[i], posIncrAtt.PositionIncrement, "posIncrement " + i);
}
}
Assert.IsFalse(ts.IncrementToken(), "end of stream");
ts.End();
if (finalOffset.HasValue)
{
Assert.AreEqual(finalOffset, offsetAtt.EndOffset, "finalOffset ");
}
ts.Close();
}
/// <summary>
/// TODO: rewrite tests not to use string comparison.
/// </summary>
private static string tsToString(TokenStream @in)
{
StringBuilder @out = new StringBuilder();
ICharTermAttribute termAtt = @in.AddAttribute<ICharTermAttribute>();
// extra safety to enforce, that the state is not preserved and also
// assign bogus values
@in.ClearAttributes();
termAtt.SetEmpty().Append("bogusTerm");
@in.Reset();
while (@in.IncrementToken())
{
if (@out.Length > 0)
{
@out.Append(' ');
}
@out.Append(termAtt.ToString());
@in.ClearAttributes();
termAtt.SetEmpty().Append("bogusTerm");
}
@in.Dispose();
return @out.ToString();
}