Author: epbernard Date: 2006-11-01 16:32:11 -0500 (Wed, 01 Nov 2006) New Revision: 10678 Modified: branches/Lucene_Integration/HibernateExt/metadata/src/java/org/hibernate/lucene/DocumentBuilder.java branches/Lucene_Integration/HibernateExt/metadata/src/java/org/hibernate/lucene/query/IteratorImpl.java branches/Lucene_Integration/HibernateExt/metadata/src/java/org/hibernate/lucene/query/LuceneQueryImpl.java branches/Lucene_Integration/HibernateExt/metadata/src/test/org/hibernate/lucene/test/query/LuceneQueryTest.java Log: ANN-460 filter by classes on queries Modified: branches/Lucene_Integration/HibernateExt/metadata/src/java/org/hibernate/lucene/DocumentBuilder.java =================================================================== --- branches/Lucene_Integration/HibernateExt/metadata/src/java/org/hibernate/lucene/DocumentBuilder.java 2006-11-01 19:07:24 UTC (rev 10677) +++ branches/Lucene_Integration/HibernateExt/metadata/src/java/org/hibernate/lucene/DocumentBuilder.java 2006-11-01 21:32:11 UTC (rev 10678) @@ -219,7 +219,7 @@ } { Field classField = - new Field( CLASS_FIELDNAME, instanceClass.getName(), Field.Store.YES, Field.Index.NO ); + new Field( CLASS_FIELDNAME, instanceClass.getName(), Field.Store.YES, Field.Index.UN_TOKENIZED ); doc.add( classField ); idBridge.set( idKeywordName, id, doc, Field.Store.YES, Field.Index.UN_TOKENIZED, idBoost ); } Modified: branches/Lucene_Integration/HibernateExt/metadata/src/java/org/hibernate/lucene/query/IteratorImpl.java =================================================================== --- branches/Lucene_Integration/HibernateExt/metadata/src/java/org/hibernate/lucene/query/IteratorImpl.java 2006-11-01 19:07:24 UTC (rev 10677) +++ branches/Lucene_Integration/HibernateExt/metadata/src/java/org/hibernate/lucene/query/IteratorImpl.java 2006-11-01 21:32:11 UTC (rev 10678) @@ -2,6 +2,7 @@ package org.hibernate.lucene.query; import java.util.Iterator; +import java.util.List; import org.hibernate.Session; @@ -10,15 +11,15 @@ */ public class IteratorImpl implements Iterator { - private final EntityInfo[] entityInfos; + private final List<EntityInfo> entityInfos; private final Session session; private int index = 0; private final int size; - public IteratorImpl(EntityInfo[] entityInfos, Session session) { + public IteratorImpl(List<EntityInfo> entityInfos, Session session) { this.entityInfos = entityInfos; this.session = session; - this.size = entityInfos.length; + this.size = entityInfos.size(); } public boolean hasNext() { @@ -26,13 +27,13 @@ } public Object next() { - Object object = session.get( entityInfos[index].clazz, entityInfos[index].id ); + Object object = session.get( entityInfos.get(index).clazz, entityInfos.get(index).id ); index++; return object; } public void remove() { //TODO this is theorically doable - throw new UnsupportedOperationException( "Cannot remove from a lucene query interator" ); + throw new UnsupportedOperationException( "Cannot remove from a lucene query iterator" ); } } Modified: branches/Lucene_Integration/HibernateExt/metadata/src/java/org/hibernate/lucene/query/LuceneQueryImpl.java =================================================================== --- branches/Lucene_Integration/HibernateExt/metadata/src/java/org/hibernate/lucene/query/LuceneQueryImpl.java 2006-11-01 19:07:24 UTC (rev 10677) +++ branches/Lucene_Integration/HibernateExt/metadata/src/java/org/hibernate/lucene/query/LuceneQueryImpl.java 2006-11-01 21:32:11 UTC (rev 10678) @@ -18,7 +18,11 @@ import org.apache.lucene.search.IndexSearcher; import org.apache.lucene.search.MultiSearcher; import org.apache.lucene.search.Searcher; +import org.apache.lucene.search.BooleanQuery; +import org.apache.lucene.search.TermQuery; +import org.apache.lucene.search.BooleanClause; import org.apache.lucene.store.Directory; +import org.apache.lucene.index.Term; import org.hibernate.Hibernate; import org.hibernate.HibernateException; import org.hibernate.LockMode; @@ -41,6 +45,7 @@ private static final Log log = LogFactory.getLog( LuceneQueryImpl.class ); private org.apache.lucene.search.Query luceneQuery; private Class[] classes; + private Set<Class> classesAndSubclasses; private Integer firstResult; private Integer maxResults; private int resultSize; @@ -71,18 +76,18 @@ Searcher searcher = buildSearcher( listener ); Hits hits; try { - hits = searcher.search( luceneQuery ); + org.apache.lucene.search.Query query = filterQueryByClasses(luceneQuery); + hits = searcher.search( query ); setResultSize( hits ); int first = first(); int max = max( first, hits ); - EntityInfo[] entityInfos = new EntityInfo[max - first + 1]; + List<EntityInfo> entityInfos = new ArrayList<EntityInfo>(max - first + 1); for ( int index = first; index <= max; index++ ) { Document document = hits.doc( index ); EntityInfo entityInfo = new EntityInfo(); entityInfo.clazz = DocumentBuilder.getDocumentClass( document ); - //FIXME should check that clazz match classes but this complexify a lot the firstResult/maxResult entityInfo.id = DocumentBuilder.getDocumentId( listener, entityInfo.clazz, document ); - entityInfos[index - first] = entityInfo; + entityInfos.add(entityInfo); } return new IteratorImpl( entityInfos, (Session) this.session ); } @@ -106,7 +111,8 @@ Searcher searcher = buildSearcher( listener ); Hits hits; try { - hits = searcher.search( luceneQuery ); + org.apache.lucene.search.Query query = filterQueryByClasses( luceneQuery ); + hits = searcher.search( query ); setResultSize( hits ); int first = first(); int max = max( first, hits ); @@ -134,7 +140,9 @@ Searcher searcher = buildSearcher( listener ); Hits hits; try { - hits = searcher.search( luceneQuery ); + org.apache.lucene.search.Query query = filterQueryByClasses(luceneQuery); + log.warn(query.toString()); + hits = searcher.search( query ); setResultSize( hits ); int first = first(); int max = max( first, hits ); @@ -143,10 +151,10 @@ for ( int index = first; index <= max; index++ ) { Document document = hits.doc( index ); Class clazz = DocumentBuilder.getDocumentClass( document ); - //FIXME should check that clazz match classes but this complexify a lot the firstResult/maxResult Serializable id = DocumentBuilder.getDocumentId( listener, clazz, document ); result.add( sess.load( clazz, id ) ); - //use load to benefit from the batch-size (but facing some proxy casting issues... + //use load to benefit from the batch-size + //we don't face proxy casting issues since the exact class is extracted from the index } //then initialize the objects for ( Object element : result ) { @@ -167,11 +175,33 @@ } } + private org.apache.lucene.search.Query filterQueryByClasses(org.apache.lucene.search.Query luceneQuery) { + //A query filter is more practical than a manual class filtering post query (esp on scrollable resultsets) + //it also probably minimise the memory footprint + if (classesAndSubclasses == null) { + return luceneQuery; + } + else { + BooleanQuery classFilter = new BooleanQuery(); + //annihilate the scoring impact of DocumentBuilder.CLASS_FIELDNAME + classFilter.setBoost( 0 ); + for (Class clazz : classesAndSubclasses) { + Term t = new Term( DocumentBuilder.CLASS_FIELDNAME, clazz.getName() ); + TermQuery termQuery = new TermQuery( t ); + classFilter.add( termQuery, BooleanClause.Occur.SHOULD ); + } + BooleanQuery filteredQuery = new BooleanQuery(); + filteredQuery.add( luceneQuery, BooleanClause.Occur.MUST ); + filteredQuery.add( classFilter, BooleanClause.Occur.MUST ); + return filteredQuery; + } + } + private int max(int first, Hits hits) { return maxResults == null ? - first + hits.length() - 1 : + hits.length() - 1 : maxResults + first < hits.length() ? - first + maxResults : + first + maxResults - 1 : hits.length() - 1; } @@ -181,6 +211,7 @@ 0; } + //TODO change classesAndSubclasses by side effect, which is a mismatch with the Searcher return, fix that. private Searcher buildSearcher(LuceneEventListener listener) { Map<Class, DocumentBuilder<Object>> builders = listener.getDocumentBuilders(); Set<Directory> directories = new HashSet<Directory>(); @@ -189,6 +220,7 @@ for ( DocumentBuilder builder : builders.values() ) { directories.add( builder.getDirectoryProvider().getDirectory() ); } + classesAndSubclasses = null; } else { Set<Class> involvedClasses = new HashSet<Class>( classes.length ); @@ -203,6 +235,7 @@ if ( builder == null ) throw new HibernateException( "Not a mapped entity: " + clazz ); directories.add( builder.getDirectoryProvider().getDirectory() ); } + classesAndSubclasses = involvedClasses; } //set up the searcher Modified: branches/Lucene_Integration/HibernateExt/metadata/src/test/org/hibernate/lucene/test/query/LuceneQueryTest.java =================================================================== --- branches/Lucene_Integration/HibernateExt/metadata/src/test/org/hibernate/lucene/test/query/LuceneQueryTest.java 2006-11-01 19:07:24 UTC (rev 10677) +++ branches/Lucene_Integration/HibernateExt/metadata/src/test/org/hibernate/lucene/test/query/LuceneQueryTest.java 2006-11-01 21:32:11 UTC (rev 10678) @@ -56,11 +56,57 @@ assertTrue( Hibernate.isInitialized( element ) ); s.delete( element ); } - for (Object element : s.createQuery( "from Book" ).list() ) s.delete( element ); + for (Object element : s.createQuery( "from java.lang.Object" ).list() ) s.delete( element ); tx.commit(); s.close(); } + public void testFirstMax() throws Exception { + LuceneSession s = new LuceneSession( openSession() ); + Transaction tx = s.beginTransaction(); + Clock clock = new Clock(1, "Seiko"); + s.save( clock ); + clock = new Clock( 2, "Festina"); + s.save( clock ); + Book book = new Book(1, "La chute de la petite reine a travers les yeux de Festina", "La chute de la petite reine a travers les yeux de Festina, blahblah"); + s.save(book); + book = new Book(2, "La gloire de mon p�re", "Les deboires de mon p�re en v�lo"); + s.save(book); + tx.commit(); + s.clear(); + tx = s.beginTransaction(); + QueryParser parser = new QueryParser("title", new StopAnalyzer() ); + + Query query = parser.parse( "summary:Festina Or brand:Seiko" ); + org.hibernate.Query hibQuery = s.createLuceneQuery( query, Clock.class, Book.class ); + hibQuery.setFirstResult( 1 ); + List result = hibQuery.list(); + assertNotNull( result ); + assertEquals( "first result no max result", 1, result.size() ); + + hibQuery.setFirstResult( 0 ); + hibQuery.setMaxResults( 1 ); + result = hibQuery.list(); + assertNotNull( result ); + assertEquals( "max result set", 1, result.size() ); + + hibQuery.setFirstResult( 0 ); + hibQuery.setMaxResults( 3 ); + result = hibQuery.list(); + assertNotNull( result ); + assertEquals( "max result out of limit", 2, result.size() ); + + hibQuery.setFirstResult( 2 ); + hibQuery.setMaxResults( 3 ); + result = hibQuery.list(); + assertNotNull( result ); + assertEquals( "first result out of limit", 0, result.size() ); + + for (Object element : s.createQuery( "from java.lang.Object" ).list() ) s.delete( element ); + tx.commit(); + s.close(); + } + public void testIterator() throws Exception { LuceneSession s = new LuceneSession( openSession() ); Transaction tx = s.beginTransaction(); @@ -93,7 +139,7 @@ s.delete( result.next() ); } assertEquals( 2, index ); - for (Object element : s.createQuery( "from Book" ).list() ) s.delete( element ); + for (Object element : s.createQuery( "from java.lang.Object" ).list() ) s.delete( element ); tx.commit(); s.close(); } @@ -136,7 +182,7 @@ while ( result.next() ) { s.delete( result.get()[0] ); } - for (Object element : s.createQuery( "from Book" ).list() ) s.delete( element ); + for (Object element : s.createQuery( "from java.lang.Object" ).list() ) s.delete( element ); tx.commit(); s.close(); } @@ -144,6 +190,8 @@ public void testMultipleEntityPerIndex() throws Exception { LuceneSession s = new LuceneSession( openSession() ); Transaction tx = s.beginTransaction(); + Clock clock = new Clock(1, "Seiko"); + s.save( clock ); Book book = new Book(1, "La chute de la petite reine a travers les yeux de Festina", "La chute de la petite reine a travers les yeux de Festina, blahblah"); s.save(book); AlternateBook alternateBook = new AlternateBook(1, "La chute de la petite reine a travers les yeux de Festina"); @@ -158,8 +206,30 @@ List result = hibQuery.list(); assertNotNull( result ); assertEquals( "Query with explicit class filter", 1, result.size() ); + + query = parser.parse( "summary:Festina" ); + hibQuery = s.createLuceneQuery( query, Clock.class, Book.class ); + Iterator it = hibQuery.iterate(); + assertTrue( it.hasNext() ); + assertNotNull( it.next() ); + assertFalse( it.hasNext() ); - query = parser.parse( "summary:Festina" ); + query = parser.parse( "summary:Festina" ); + hibQuery = s.createLuceneQuery( query, Clock.class, Book.class ); + ScrollableResults sr = hibQuery.scroll(); + assertTrue( sr.first() ); + assertNotNull( sr.get() ); + assertFalse( sr.next() ); + sr.close(); + + query = parser.parse( "summary:Festina OR brand:seiko" ); + hibQuery = s.createLuceneQuery( query, Clock.class, Book.class ); + hibQuery.setMaxResults( 2 ); + result = hibQuery.list(); + assertNotNull( result ); + assertEquals( "Query with explicit class filter and limit", 2, result.size() ); + + query = parser.parse( "summary:Festina" ); hibQuery = s.createLuceneQuery( query ); result = hibQuery.list(); assertNotNull( result );