cdmlib-services/src/main/java/eu/etaxonomy/cdm/api/service/search/LuceneMultiSearch.java

   1 // $Id$
   2 /**
   3 * Copyright (C) 2011 EDIT
   4 * European Distributed Institute of Taxonomy
   5 * http://www.e-taxonomy.eu
   6 *
   7 * The contents of this file are subject to the Mozilla Public License Version 1.1
   8 * See LICENSE.TXT at the top of this package for the full license terms.
   9 */
  10 package eu.etaxonomy.cdm.api.service.search;
  11
  12 import java.util.ArrayList;
  13 import java.util.Arrays;
  14 import java.util.HashSet;
  15 import java.util.List;
  16 import java.util.Set;
  17
  18 import org.apache.log4j.Logger;
  19 import org.apache.lucene.analysis.Analyzer;
  20 import org.apache.lucene.index.IndexReader;
  21 import org.apache.lucene.index.MultiReader;
  22 import org.apache.lucene.search.BooleanClause.Occur;
  23 import org.apache.lucene.search.BooleanQuery;
  24 import org.apache.lucene.search.IndexSearcher;
  25 import org.hibernate.search.Search;
  26 import org.hibernate.search.SearchFactory;
  27 import org.hibernate.search.indexes.IndexReaderAccessor;
  28
  29 import eu.etaxonomy.cdm.model.common.CdmBase;
  30
  31 /**
  32  * A LuceneSearch which allow to run a union like search on multiple indexes at once.
  33  * Internally a {@link MultiReader} is being used.
  34  *
  35  * @author Andreas Kohlbecker
  36  * @date Dec 21, 2011
  37  *
  38  */
  39 public class LuceneMultiSearch extends LuceneSearch {
  40
  41     public static final Logger logger = Logger.getLogger(LuceneMultiSearch.class);
  42
  43     private final Set<Class<? extends CdmBase>> directorySelectClasses = new HashSet<Class<? extends CdmBase>>();
  44
  45
  46     /**
  47      * @param luceneSearch the searches to execute together as a union like search
  48      * @throws Exception
  49      */
  50     public LuceneMultiSearch(LuceneSearch... luceneSearch) throws LuceneMultiSearchException {
  51         session = luceneSearch[0].session;
  52
  53         BooleanQuery query = new BooleanQuery();
  54
  55         Set<String> highlightFields = new HashSet<String>();
  56
  57         for(LuceneSearch search : luceneSearch){
  58             this.directorySelectClasses.add(search.getDirectorySelectClass());
  59             query.add(search.getQuery(), Occur.SHOULD);
  60             highlightFields.addAll(Arrays.asList(search.getHighlightFields()));
  61             if(search.getClazz() != null){
  62                 if(getClazz() != null){
  63                     throw new LuceneMultiSearchException("LuceneMultiSearch can only handle once class restriction, but multiple given: " + getClazz() + ", " + search.getClazz());
  64                 }
  65                 setClazz(search.getClazz());
  66             }
  67         }
  68
  69         this.highlightFields = highlightFields.toArray(new String[highlightFields.size()]);
  70         this.query = query;
  71     }
  72
  73     /**
  74      * @return
  75      */
  76     @Override
  77     public IndexSearcher getSearcher() {
  78
  79         if(searcher == null){
  80
  81             SearchFactory searchFactory = Search.getFullTextSession(session).getSearchFactory();
  82             List<IndexReader> readers = new ArrayList<IndexReader>();
  83             for(Class<? extends CdmBase> type : directorySelectClasses){
  84                    //OLD
  85 //                DirectoryProvider[] directoryProviders = searchFactory.getDirectoryProviders(type);
  86 //                logger.info(directoryProviders[0].getDirectory().toString());
  87
  88 //                ReaderProvider readerProvider = searchFactory.getReaderProvider();
  89                 IndexReaderAccessor ira = searchFactory.getIndexReaderAccessor();
  90                 IndexReader reader = ira.open(type);
  91 //              readers.add(readerProvider.openReader(directoryProviders[0]));
  92                 readers.add(reader);
  93             }
  94             if(readers.size() > 1){
  95                 MultiReader multireader = new MultiReader(readers.toArray(new IndexReader[readers.size()]), true);
  96                 searcher = new IndexSearcher(multireader);
  97             } else {
  98                 searcher = new IndexSearcher(readers.get(0));
  99             }
 100         }
 101
 102         return searcher;
 103     }
 104
 105     /**
 106      * does exactly the same as {@link LuceneSearch#getAnalyzer()} but perform
 107      * an additional check to assure that all indexes are using the same
 108      * analyzer
 109      *
 110      * @return
 111      */
 112     @Override
 113     public Analyzer getAnalyzer() {
 114         SearchFactory searchFactory = Search.getFullTextSession(session).getSearchFactory();
 115         Analyzer analyzer = null;
 116         for(Class<? extends CdmBase> type : directorySelectClasses){
 117             Analyzer a = searchFactory.getAnalyzer(type);
 118             if(isEqual(analyzer, a)){
 119                 throw new RuntimeException("The LuceneMultiSearch must only be used on indexes which are using the same Analyzer.");
 120             }
 121             analyzer = a;
 122         }
 123         return analyzer;
 124     }
 125
 126     /**
 127      * @param analyzer
 128      * @param a
 129      * @return
 130      */
 131     private boolean isEqual(Analyzer analyzer, Analyzer a) {
 132         // FIXME PatternAnalyzers must be compared by Pattern also other analyzers must be compared by their properties
 133         return analyzer != null && !analyzer.getClass().equals(a.getClass());
 134     }
 135
 136 }