public virtual void TestTwoTrailingHolesTriShingleWithTokenFiller()
{
// Analyzing "purple wizard of the", where of and the are removed as a
// stopwords, leaving two trailing holes:
Token[] inputTokens = new Token[] { CreateToken("purple", 0, 6), CreateToken("wizard", 7, 13) };
ShingleFilter filter = new ShingleFilter(new CannedTokenStream(2, 20, inputTokens), 2, 3);
filter.SetFillerToken("--");
AssertTokenStreamContents(filter, new string[] { "purple", "purple wizard", "purple wizard --", "wizard", "wizard --", "wizard -- --" }, new int[] { 0, 0, 0, 7, 7, 7 }, new int[] { 6, 13, 20, 13, 20, 20 }, new int[] { 1, 0, 0, 1, 0, 0 }, 20);
filter = new ShingleFilter(new CannedTokenStream(2, 20, inputTokens), 2, 3);
filter.SetFillerToken("");
AssertTokenStreamContents(filter, new string[] { "purple", "purple wizard", "purple wizard ", "wizard", "wizard ", "wizard " }, new int[] { 0, 0, 0, 7, 7, 7 }, new int[] { 6, 13, 20, 13, 20, 20 }, new int[] { 1, 0, 0, 1, 0, 0 }, 20);
filter = new ShingleFilter(new CannedTokenStream(2, 20, inputTokens), 2, 3);
filter.SetFillerToken(null);
AssertTokenStreamContents(filter, new string[] { "purple", "purple wizard", "purple wizard ", "wizard", "wizard ", "wizard " }, new int[] { 0, 0, 0, 7, 7, 7 }, new int[] { 6, 13, 20, 13, 20, 20 }, new int[] { 1, 0, 0, 1, 0, 0 }, 20);
filter = new ShingleFilter(new CannedTokenStream(2, 20, inputTokens), 2, 3);
filter.SetFillerToken(null);
filter.SetTokenSeparator(null);
AssertTokenStreamContents(filter, new string[] { "purple", "purplewizard", "purplewizard", "wizard", "wizard", "wizard" }, new int[] { 0, 0, 0, 7, 7, 7 }, new int[] { 6, 13, 20, 13, 20, 20 }, new int[] { 1, 0, 0, 1, 0, 0 }, 20);
}
public override TokenStream Create(TokenStream input)
{
ShingleFilter r = new ShingleFilter(input, minShingleSize, maxShingleSize);
r.SetOutputUnigrams(outputUnigrams);
r.SetOutputUnigramsIfNoShingles(outputUnigramsIfNoShingles);
r.SetTokenSeparator(tokenSeparator);
r.SetFillerToken(fillerToken);
return(r);
}
protected override TokenStreamComponents WrapComponents(string fieldName, TokenStreamComponents components)
{
ShingleFilter filter = new ShingleFilter(components.TokenStream, minShingleSize, maxShingleSize);
filter.SetMinShingleSize(minShingleSize);
filter.SetMaxShingleSize(maxShingleSize);
filter.SetTokenSeparator(tokenSeparator);
filter.SetOutputUnigrams(outputUnigrams);
filter.SetOutputUnigramsIfNoShingles(outputUnigramsIfNoShingles);
filter.SetFillerToken(fillerToken);
return(new TokenStreamComponents(components.Tokenizer, filter));
}