[hibernate-commits] Hibernate SVN: r15535 - in search/trunk/src: test/org/hibernate/search/test/filter and 1 other directory.

Author: epbernard Date: 2008-11-09 01:34:45 -0500 (Sun, 09 Nov 2008) New Revision: 15535 Added: search/trunk/src/java/org/hibernate/search/filter/AndDocIdSet.java search/trunk/src/java/org/hibernate/search/filter/EmptyDocIdBitSet.java Modified: search/trunk/src/java/org/hibernate/search/filter/CachingWrapperFilter.java search/trunk/src/java/org/hibernate/search/filter/ChainedFilter.java search/trunk/src/test/org/hibernate/search/test/filter/ExcludeAllFilter.java Log: HSEARCH-289 move to DocIdSet use. Added: search/trunk/src/java/org/hibernate/search/filter/AndDocIdSet.java =================================================================== --- search/trunk/src/java/org/hibernate/search/filter/AndDocIdSet.java (rev 0) +++ search/trunk/src/java/org/hibernate/search/filter/AndDocIdSet.java 2008-11-09 06:34:45 UTC (rev 15535) @@ -0,0 +1,141 @@ +package org.hibernate.search.filter; + +import java.io.IOException; +import java.util.BitSet; +import java.util.List; +import static java.lang.Math.max; + +import org.apache.lucene.search.DocIdSet; +import org.apache.lucene.search.DocIdSetIterator; +import org.apache.lucene.util.DocIdBitSet; + +/** + * A DocIdSet built as applying "AND" operation to a list of other DocIdSet. + * The DocIdSetIterator returned will return only document ids contained + * in all DocIdSet handed to the constructor. + * + * @author Sanne Grinovero + */ +public class AndDocIdSet extends DocIdSet { + + private DocIdBitSet docIdBitSet; + private final List<DocIdSet> andedDocIdSets; + + public AndDocIdSet(List<DocIdSet> andedDocIdSets) { + if ( andedDocIdSets == null || andedDocIdSets.size() < 2 ) + throw new IllegalArgumentException( "To \"and\" some DocIdSet they should be at least 2" ); + this.andedDocIdSets = andedDocIdSets; + } + + private synchronized void buildBitset() throws IOException { + if ( docIdBitSet != null ) return; // double check for concurrent initialization + //TODO if all andedDocIdSets are actually DocIdBitSet, use their internal BitSet instead of next algo. + //TODO if some andedDocIdSets are DocIdBitSet, merge them first. + int size = andedDocIdSets.size(); + DocIdSetIterator[] iterators = new DocIdSetIterator[size]; + int[] positions = new int[size]; + boolean valuesExist = true; + int maxIndex = 0; + for (int i=0; i<size; i++) { + // build all iterators + DocIdSetIterator iterator = andedDocIdSets.get(i).iterator(); + iterators[i] = iterator; + // and move to first position + boolean nextExists = iterator.next(); + if ( ! nextExists ) { + valuesExist = false; + break; + } + int currentFilterValue = iterator.doc(); + positions[i] = currentFilterValue; + // find the initial maximum position + maxIndex = max( maxIndex, currentFilterValue ); + } + BitSet bitSet = new BitSet(); + if ( valuesExist ) { // skip further processing if some idSet is empty + do { + if ( allSame( positions ) ) { + // enable a bit if all idSets agree on it: + bitSet.set( maxIndex ); + maxIndex++; + } + maxIndex = advance( iterators, positions, maxIndex ); + } while ( maxIndex != -1 ); // -1 means the end of some bitSet has been reached (end condition) + } + docIdBitSet = new DocIdBitSet( bitSet ); + } + + /** + * Have all DocIdSetIterator having current doc id minor than currentMaxPosition + * skip to at least this position. + * @param iterators + * @param positions + * @return maximum position of all DocIdSetIterator after the operation, or -1 when at least one reached the end. + * @throws IOException + */ + private final int advance(final DocIdSetIterator[] iterators, final int[] positions, int currentMaxPosition) throws IOException { + for (int i=0; i<positions.length; i++) { + if ( positions[i] != currentMaxPosition ) { + boolean validPosition = iterators[i].skipTo( currentMaxPosition ); + if ( ! validPosition ) + return -1; + positions[i] = iterators[i].doc(); + currentMaxPosition = max( currentMaxPosition, positions[i] ); + } + } + return currentMaxPosition; + } + + /** + * see if all DocIdSetIterator stopped at the same position. + * @param positions the array of current positions. + * @return true if all DocIdSetIterator agree on the current docId. + */ + private final boolean allSame(final int[] positions) { + int base = positions[0]; + for (int i=1; i<positions.length; i++) { + if ( base != positions[i] ) + return false; + } + return true; + } + + @Override + public DocIdSetIterator iterator() { + return new AndingDocIdSetIterator(); + } + + private class AndingDocIdSetIterator extends DocIdSetIterator { + + private DocIdSetIterator iterator; + + @Override + public int doc() { + // should never happen when respecting interface contract; otherwise I + // prefer a NPE than a hard to debug return 0. + assert iterator != null : "Illegal state, can't be called before next() or skipTo(int)"; + return iterator.doc(); + } + + @Override + public boolean next() throws IOException { + ensureInitialized(); //can't initialize before as it would not be allowed to throw IOException + return iterator.next(); + } + + @Override + public boolean skipTo(int target) throws IOException { + ensureInitialized(); //can't initialize before as it would not be allowed to throw IOException + return iterator.skipTo( target ); + } + + private void ensureInitialized() throws IOException { + if ( docIdBitSet == null ) buildBitset(); + if ( iterator == null ) { + iterator = docIdBitSet.iterator(); + } + } + + } + +} Modified: search/trunk/src/java/org/hibernate/search/filter/CachingWrapperFilter.java =================================================================== --- search/trunk/src/java/org/hibernate/search/filter/CachingWrapperFilter.java 2008-11-08 22:25:28 UTC (rev 15534) +++ search/trunk/src/java/org/hibernate/search/filter/CachingWrapperFilter.java 2008-11-09 06:34:45 UTC (rev 15535) @@ -5,6 +5,7 @@ import java.util.BitSet; import org.apache.lucene.index.IndexReader; +import org.apache.lucene.search.DocIdSet; import org.apache.lucene.search.Filter; import org.slf4j.Logger; @@ -27,53 +28,56 @@ public static final int DEFAULT_SIZE = 5; - private final int size; - /** * The cache using soft references in order to store the filter bit sets. */ - private transient SoftLimitMRUCache cache; + private final SoftLimitMRUCache cache; private final Filter filter; /** - * @param filter - * Filter to cache results of + * @param filter Filter to cache results of */ public CachingWrapperFilter(Filter filter) { this(filter, DEFAULT_SIZE); } /** - * @param filter - * Filter to cache results of + * @param filter Filter to cache results of */ public CachingWrapperFilter(Filter filter, int size) { this.filter = filter; - this.size = size; + log.debug( "Initialising SoftLimitMRUCache with hard ref size of {}", size ); + this.cache = new SoftLimitMRUCache( size ); } + @Override public BitSet bits(IndexReader reader) throws IOException { - if (cache == null) { - log.debug("Initialising SoftLimitMRUCache with hard ref size of {}", size); - cache = new SoftLimitMRUCache(size); + throw new UnsupportedOperationException(); + /* BitSet cached = (BitSet) cache.get(reader); + if (cached != null) { + return cached; } - - //memory barrier ensure cache == null will not always stay true on concurrent threads - synchronized (cache) { // check cache - BitSet cached = (BitSet) cache.get(reader); - if (cached != null) { + final BitSet bits = filter.bits(reader); + cache.put(reader, bits); + return bits; */ + } + + @Override + public DocIdSet getDocIdSet(IndexReader reader) throws IOException { + DocIdSet cached = (DocIdSet) cache.get( reader ); + if ( cached != null ) { + return cached; + } + synchronized (cache) { + cached = (DocIdSet) cache.get( reader ); + if ( cached != null ) { return cached; } + final DocIdSet docIdSet = filter.getDocIdSet( reader ); + cache.put( reader, docIdSet ); + return docIdSet; } - - final BitSet bits = filter.bits(reader); - - synchronized (cache) { // update cache - cache.put(reader, bits); - } - - return bits; } public String toString() { Modified: search/trunk/src/java/org/hibernate/search/filter/ChainedFilter.java =================================================================== --- search/trunk/src/java/org/hibernate/search/filter/ChainedFilter.java 2008-11-08 22:25:28 UTC (rev 15534) +++ search/trunk/src/java/org/hibernate/search/filter/ChainedFilter.java 2008-11-09 06:34:45 UTC (rev 15535) @@ -6,6 +6,7 @@ import java.util.ArrayList; import java.io.IOException; +import org.apache.lucene.search.DocIdSet; import org.apache.lucene.search.Filter; import org.apache.lucene.index.IndexReader; import org.hibernate.annotations.common.AssertionFailure; @@ -14,6 +15,7 @@ * @author Emmanuel Bernard */ public class ChainedFilter extends Filter { + private static final long serialVersionUID = -6153052295766531920L; private final List<Filter> chainedFilters = new ArrayList<Filter>(); @@ -23,6 +25,8 @@ } public BitSet bits(IndexReader reader) throws IOException { + throw new UnsupportedOperationException(); + /* if (chainedFilters.size() == 0) throw new AssertionFailure("Chainedfilter has no filters to chain for"); //we need to copy the first BitSet because BitSet is modified by .logicalOp Filter filter = chainedFilters.get( 0 ); @@ -31,7 +35,26 @@ result.and( chainedFilters.get( index ).bits( reader ) ); } return result; + */ } + + @Override + public DocIdSet getDocIdSet(IndexReader reader) throws IOException { + int size = chainedFilters.size(); + if ( size == 0 ) { + throw new AssertionFailure( "Chainedfilter has no filters to chain for" ); + } + else if ( size == 1 ) { + return chainedFilters.get(0).getDocIdSet(reader); + } + else { + List<DocIdSet> subSets = new ArrayList<DocIdSet>( size ); + for ( Filter f : chainedFilters ) { + subSets.add( f.getDocIdSet( reader ) ); + } + return new AndDocIdSet( subSets ); + } + } public String toString() { StringBuilder sb = new StringBuilder("ChainedFilter ["); Added: search/trunk/src/java/org/hibernate/search/filter/EmptyDocIdBitSet.java =================================================================== --- search/trunk/src/java/org/hibernate/search/filter/EmptyDocIdBitSet.java (rev 0) +++ search/trunk/src/java/org/hibernate/search/filter/EmptyDocIdBitSet.java 2008-11-09 06:34:45 UTC (rev 15535) @@ -0,0 +1,46 @@ +package org.hibernate.search.filter; + +import java.io.IOException; + +import org.apache.lucene.search.DocIdSet; +import org.apache.lucene.search.DocIdSetIterator; + +public class EmptyDocIdBitSet extends DocIdSet { + + public static final DocIdSet instance = new EmptyDocIdBitSet(); + + private final DocIdSetIterator iterator = new EmptyDocIdSetIterator(); + + private EmptyDocIdBitSet(){ + // is singleton + } + + @Override + public DocIdSetIterator iterator() { + return iterator; + } + + /** + * implements a DocIdSetIterator for an empty DocIdSet + * As it is empty it also is stateless and so it can be reused. + */ + private static class EmptyDocIdSetIterator extends DocIdSetIterator { + + @Override + public int doc() { + throw new IllegalStateException("Should never be called"); + } + + @Override + public boolean next() throws IOException { + return false; + } + + @Override + public boolean skipTo(int target) throws IOException { + return false; + } + + } + +} Modified: search/trunk/src/test/org/hibernate/search/test/filter/ExcludeAllFilter.java =================================================================== --- search/trunk/src/test/org/hibernate/search/test/filter/ExcludeAllFilter.java 2008-11-08 22:25:28 UTC (rev 15534) +++ search/trunk/src/test/org/hibernate/search/test/filter/ExcludeAllFilter.java 2008-11-09 06:34:45 UTC (rev 15535) @@ -4,20 +4,33 @@ import java.util.BitSet; import java.io.IOException; +import org.apache.lucene.search.DocIdSet; import org.apache.lucene.search.Filter; import org.apache.lucene.index.IndexReader; +import org.hibernate.search.filter.EmptyDocIdBitSet; /** * @author Emmanuel Bernard */ @SuppressWarnings("serial") public class ExcludeAllFilter extends Filter { + + //ugly but useful for test purposes private static volatile boolean done = false; + @Override public BitSet bits(IndexReader reader) throws IOException { - if (done) throw new IllegalStateException("Called twice"); + if ( done ) throw new IllegalStateException( "Called twice" ); BitSet bitSet = new BitSet( reader.maxDoc() ); done = true; return bitSet; } + + @Override + public DocIdSet getDocIdSet(IndexReader reader) throws IOException { + if ( done ) throw new IllegalStateException( "Called twice" ); + done = true; + return EmptyDocIdBitSet.instance; + } + }