svn commit: r1352959 - in /jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine: index/HashIndexTable.java index/IndexFactory.java index/IndexTable.java index/LinearIndex.java iterator/QueryIter2.java

[rv...@apache.org]

Author: rvesse
Date: Fri Jun 22 16:28:45 2012
New Revision: 1352959

URL: http://svn.apache.org/viewvc?rev=1352959&view=rev
Log:
Applied Paul Gearon's further patch for JENA-266 to add test cases for new data structures introduced to speed up MINUS evaluation

Modified:
    jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/HashIndexTable.java
    jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/IndexFactory.java
    jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/IndexTable.java
    jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/LinearIndex.java
    jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/iterator/QueryIter2.java

Modified: jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/HashIndexTable.java
URL: http://svn.apache.org/viewvc/jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/HashIndexTable.java?rev=1352959&r1=1352958&r2=1352959&view=diff
==============================================================================
--- jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/HashIndexTable.java (original)
+++ jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/HashIndexTable.java Fri Jun 22 16:28:45 2012
@@ -29,6 +29,12 @@ import com.hp.hpl.jena.sparql.core.Var;
 import com.hp.hpl.jena.sparql.engine.QueryIterator;
 import com.hp.hpl.jena.sparql.engine.binding.Binding;
 
+/**
+ * Indexes bindings so that they can be search for quickly when a binding to all the
+ * variables is provided. If a binding to only some of the known variables is provided
+ * then the index still works, but will search linearly.
+ * @author Paul Gearon
+ */
 public class HashIndexTable implements IndexTable {
 
 	final private Set<Key> table ;

Modified: jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/IndexFactory.java
URL: http://svn.apache.org/viewvc/jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/IndexFactory.java?rev=1352959&r1=1352958&r2=1352959&view=diff
==============================================================================
--- jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/IndexFactory.java (original)
+++ jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/IndexFactory.java Fri Jun 22 16:28:45 2012
@@ -24,6 +24,11 @@ import com.hp.hpl.jena.sparql.core.Var;
 import com.hp.hpl.jena.sparql.engine.QueryIterator;
 import com.hp.hpl.jena.sparql.engine.index.HashIndexTable.MissingBindingException ;
 
+/**
+ * Creates {@link com.hp.hpl.jena.sparql.engine.index.IndexTable}s for use by
+ * {@link com.hp.hpl.jena.sparql.engine.iterator.QueryIterMinus}.
+ * @author Paul Gearon
+ */
 public class IndexFactory {
 
 	public static IndexTable createIndex(Set<Var> commonVars, QueryIterator data)

Modified: jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/IndexTable.java
URL: http://svn.apache.org/viewvc/jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/IndexTable.java?rev=1352959&r1=1352958&r2=1352959&view=diff
==============================================================================
--- jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/IndexTable.java (original)
+++ jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/IndexTable.java Fri Jun 22 16:28:45 2012
@@ -20,6 +20,13 @@ package com.hp.hpl.jena.sparql.engine.in
 
 import com.hp.hpl.jena.sparql.engine.binding.Binding;
 
+/**
+ * Interface for indexes that are used for identifying matching
+ * {@link com.hp.hpl.jena.sparql.engine.binding.Binding}s when
+ * {@link com.hp.hpl.jena.sparql.engine.iterator.QueryIterMinus} is trying to determine
+ * which Bindings need to be removed.
+ * @author Paul Gearon
+ */
 public interface IndexTable {
 
 	public abstract boolean containsCompatibleWithSharedDomain(Binding binding);

Modified: jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/LinearIndex.java
URL: http://svn.apache.org/viewvc/jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/LinearIndex.java?rev=1352959&r1=1352958&r2=1352959&view=diff
==============================================================================
--- jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/LinearIndex.java (original)
+++ jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/index/LinearIndex.java Fri Jun 22 16:28:45 2012
@@ -30,6 +30,24 @@ import com.hp.hpl.jena.sparql.engine.Que
 import com.hp.hpl.jena.sparql.engine.binding.Binding;
 import com.hp.hpl.jena.sparql.engine.binding.BindingHashMap;
 
+/**
+ * A slow "index" that looks for data by searching linearly through a set.
+ * Only used when the indexed data contains fewer bound variables than expected.
+ * Note that this class is only used for a MINUS operation that is removing data
+ * with potentially unbound values, and is therefore rarely used.
+ * 
+ * TODO: If this index starts to be used more often then consider various options for
+ *       indexing on the known bound variables.
+ *       One possibility is for each variable (found in commonVars) to take
+ *       the value of a var/value pair and TreeMap this to a set of Bindings that it occurs in.
+ *       This would offer a reduced set to search, and set intersections may also work
+ *       (intersections like this could be done on Binding reference equality rather than value).
+ *       TreeMap is suggested here, since there would be commonVars.size() maps, which would take
+ *       a lot of heap, particularly since performance of this class is only an issue when the
+ *       data to search is significant.
+ * @author Paul Gearon
+ */
+
 public class LinearIndex implements IndexTable {
 
 	final Set<Var> commonVars ;
@@ -79,7 +97,7 @@ public class LinearIndex implements Inde
 		return false;
 	}
 
-	private static Binding toBinding(HashIndexTable.Key key, Map<Var,Integer> mappings)
+	static Binding toBinding(HashIndexTable.Key key, Map<Var,Integer> mappings)
 	{
 		Node[] values = key.getNodes() ;
 		BindingHashMap b = new BindingHashMap() ;

Modified: jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/iterator/QueryIter2.java
URL: http://svn.apache.org/viewvc/jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/iterator/QueryIter2.java?rev=1352959&r1=1352958&r2=1352959&view=diff
==============================================================================
--- jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/iterator/QueryIter2.java (original)
+++ jena/trunk/jena-arq/src/main/java/com/hp/hpl/jena/sparql/engine/iterator/QueryIter2.java Fri Jun 22 16:28:45 2012
@@ -27,7 +27,7 @@ import com.hp.hpl.jena.sparql.serializer
 import com.hp.hpl.jena.sparql.util.Utils ;
 
 /**
- * This class makrs a QueryIter that takes one QueryIterator as input. */
+ * This class makrs a QueryIter that takes two QueryIterators as input. */
 public abstract class QueryIter2 extends QueryIter
 {
     private QueryIterator leftInput ;