1 40 package org.dspace.search; 41 42 import java.io.Reader ; 43 import java.util.Set ; 44 45 import org.apache.lucene.analysis.Analyzer; 46 import org.apache.lucene.analysis.LowerCaseFilter; 47 import org.apache.lucene.analysis.PorterStemFilter; 48 import org.apache.lucene.analysis.StopFilter; 49 import org.apache.lucene.analysis.TokenStream; 50 import org.apache.lucene.analysis.standard.StandardFilter; 51 52 56 public class DSAnalyzer extends Analyzer 57 { 58 62 private static final String [] STOP_WORDS = 63 { 64 65 "a", "am", "and", "are", "as", "at", "be", "but", "by", "for", 67 "if", "in", "into", "is", "it", "no", "not", "of", "on", "or", 68 "the", "to", "was" 69 76 }; 77 78 81 final static private Set stopSet = StopFilter.makeStopSet(STOP_WORDS); 82 83 86 public final TokenStream tokenStream(String fieldName, final Reader reader) 87 { 88 TokenStream result = new DSTokenizer(reader); 89 90 result = new StandardFilter(result); 91 result = new LowerCaseFilter(result); 92 result = new StopFilter(result, stopSet); 93 result = new PorterStemFilter(result); 94 95 return result; 96 } 97 } 98 | Popular Tags |