You are viewing a plain text version of this content. The canonical link for it is here.
Posted to oak-commits@jackrabbit.apache.org by ch...@apache.org on 2014/10/15 15:33:14 UTC

svn commit: r1632017 - in /jackrabbit/oak/trunk/oak-lucene/src: main/java/org/apache/jackrabbit/oak/plugins/index/lucene/ test/java/org/apache/jackrabbit/oak/plugins/index/lucene/

Author: chetanm
Date: Wed Oct 15 13:33:14 2014
New Revision: 1632017

URL: http://svn.apache.org/r1632017
Log:
OAK-2196 - Implement sorting based on Lucene sorting

Make use of DocValues for sorting based on Amit's approach.  orderedProps would be be stored in a separate field with DocVal type set.

Modified:
    jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/FieldNames.java
    jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/IndexDefinition.java
    jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/IndexPlanner.java
    jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LuceneIndexConstants.java
    jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LuceneIndexEditor.java
    jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LuceneIndexEditorContext.java
    jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LucenePropertyIndex.java
    jackrabbit/oak/trunk/oak-lucene/src/test/java/org/apache/jackrabbit/oak/plugins/index/lucene/LucenePropertyIndexTest.java

Modified: jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/FieldNames.java
URL: http://svn.apache.org/viewvc/jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/FieldNames.java?rev=1632017&r1=1632016&r2=1632017&view=diff
==============================================================================
--- jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/FieldNames.java (original)
+++ jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/FieldNames.java Wed Oct 15 13:33:14 2014
@@ -48,4 +48,16 @@ public final class FieldNames {
     public static final Set<String> PATH_SELECTOR = new HashSet<String>(
             Arrays.asList(PATH));
 
+    /**
+     * Encodes the field name such that it can be used for storing DocValue
+     * This is done such a field if used for both sorting and querying uses
+     * a different name for docvalue field
+     *
+     * @param name name to encode
+     * @return encoded field name
+     */
+    public static String createDocValFieldName(String name){
+        return ":dv" + name;
+    }
+
 }

Modified: jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/IndexDefinition.java
URL: http://svn.apache.org/viewvc/jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/IndexDefinition.java?rev=1632017&r1=1632016&r2=1632017&view=diff
==============================================================================
--- jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/IndexDefinition.java (original)
+++ jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/IndexDefinition.java Wed Oct 15 13:33:14 2014
@@ -28,6 +28,7 @@ import javax.jcr.PropertyType;
 
 import com.google.common.collect.ImmutableMap;
 import com.google.common.collect.ImmutableSet;
+import com.google.common.collect.Iterables;
 import com.google.common.collect.Maps;
 import com.google.common.collect.Sets;
 import org.apache.jackrabbit.oak.api.PropertyState;
@@ -42,6 +43,7 @@ import static org.apache.jackrabbit.oak.
 import static org.apache.jackrabbit.oak.plugins.index.lucene.LuceneIndexConstants.FULL_TEXT_ENABLED;
 import static org.apache.jackrabbit.oak.plugins.index.lucene.LuceneIndexConstants.INCLUDE_PROPERTY_NAMES;
 import static org.apache.jackrabbit.oak.plugins.index.lucene.LuceneIndexConstants.INCLUDE_PROPERTY_TYPES;
+import static org.apache.jackrabbit.oak.plugins.index.lucene.LuceneIndexConstants.ORDERED_PROP_NAMES;
 
 public class IndexDefinition {
     private static final Logger log = LoggerFactory.getLogger(IndexDefinition.class);
@@ -51,6 +53,8 @@ public class IndexDefinition {
 
     private final Set<String> includes;
 
+    private final Set<String> orderedProps;
+
     private final boolean fullTextEnabled;
 
     private final boolean storageEnabled;
@@ -78,12 +82,14 @@ public class IndexDefinition {
 
         this.excludes = toLowerCase(getMultiProperty(defn, EXCLUDE_PROPERTY_NAMES));
         this.includes = getMultiProperty(defn, INCLUDE_PROPERTY_NAMES);
+        this.orderedProps = getMultiProperty(defn, ORDERED_PROP_NAMES);
+
         this.fullTextEnabled = getOptionalValue(defn, FULL_TEXT_ENABLED, true);
         //Storage is disabled for non full text indexes
         this.storageEnabled = this.fullTextEnabled && getOptionalValue(defn, EXPERIMENTAL_STORAGE, true);
 
         Map<String, PropertyDefinition> propDefns = Maps.newHashMap();
-        for(String propName : includes){
+        for(String propName : Iterables.concat(includes, orderedProps)){
             if(defn.hasChildNode(propName)){
                 propDefns.put(propName, new PropertyDefinition(this, propName, defn.child(propName)));
             }
@@ -105,6 +111,10 @@ public class IndexDefinition {
         return (propertyTypes & (1 << type)) != 0;
     }
 
+    boolean isOrdered(String name) {
+        return orderedProps.contains(name);
+    }
+
     public NodeBuilder getDefinition() {
         return definition;
     }

Modified: jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/IndexPlanner.java
URL: http://svn.apache.org/viewvc/jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/IndexPlanner.java?rev=1632017&r1=1632016&r2=1632017&view=diff
==============================================================================
--- jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/IndexPlanner.java (original)
+++ jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/IndexPlanner.java Wed Oct 15 13:33:14 2014
@@ -27,8 +27,6 @@ import javax.annotation.CheckForNull;
 import org.apache.jackrabbit.oak.query.fulltext.FullTextExpression;
 import org.apache.jackrabbit.oak.spi.query.Filter;
 import org.apache.lucene.index.IndexReader;
-import org.slf4j.Logger;
-import org.slf4j.LoggerFactory;
 
 import static com.google.common.collect.Lists.newArrayListWithCapacity;
 import static org.apache.jackrabbit.oak.spi.query.Filter.PropertyRestriction;
@@ -129,7 +127,7 @@ public class IndexPlanner {
             //sorting can only be done for known/configured properties
             // and whose types are known
             //TODO Can sorting be done for array properties
-            if (defn.includeProperty(o.getPropertyName())
+            if (defn.includeProperty(o.getPropertyName()) || defn.isOrdered(o.getPropertyName())
                     && o.getPropertyType() != null
                     && !o.getPropertyType().isArray()) {
                 orderEntries.add(o); //Lucene can manage any order desc/asc

Modified: jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LuceneIndexConstants.java
URL: http://svn.apache.org/viewvc/jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LuceneIndexConstants.java?rev=1632017&r1=1632016&r2=1632017&view=diff
==============================================================================
--- jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LuceneIndexConstants.java (original)
+++ jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LuceneIndexConstants.java Wed Oct 15 13:33:14 2014
@@ -68,8 +68,14 @@ public interface LuceneIndexConstants {
 
     /**
      * Type of the property being indexed defined as part of property definition
-     * under the given index definition. Refer to {@ling javax.jcr.PropertyType}
+     * under the given index definition. Refer to {@link javax.jcr.PropertyType}
      * contants for the possible values
      */
     String PROP_TYPE = "propertyType";
+
+    /**
+     * Defines properties which would be used for ordering. If range queries are to
+     * be performed with same property then it must be part of include list also
+     */
+    String ORDERED_PROP_NAMES = "orderedProps";
 }

Modified: jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LuceneIndexEditor.java
URL: http://svn.apache.org/viewvc/jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LuceneIndexEditor.java?rev=1632017&r1=1632016&r2=1632017&view=diff
==============================================================================
--- jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LuceneIndexEditor.java (original)
+++ jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LuceneIndexEditor.java Wed Oct 15 13:33:14 2014
@@ -41,12 +41,16 @@ import org.apache.jackrabbit.oak.spi.sta
 import org.apache.jackrabbit.oak.spi.state.NodeState;
 import org.apache.lucene.analysis.Analyzer;
 import org.apache.lucene.document.Document;
+import org.apache.lucene.document.DoubleDocValuesField;
 import org.apache.lucene.document.DoubleField;
 import org.apache.lucene.document.Field;
 import org.apache.lucene.document.LongField;
+import org.apache.lucene.document.NumericDocValuesField;
+import org.apache.lucene.document.SortedDocValuesField;
 import org.apache.lucene.document.StringField;
 import org.apache.lucene.index.IndexWriter;
 import org.apache.lucene.search.PrefixQuery;
+import org.apache.lucene.util.BytesRef;
 import org.apache.tika.metadata.Metadata;
 import org.apache.tika.parser.ParseContext;
 import org.apache.tika.sax.WriteOutContentHandler;
@@ -200,9 +204,16 @@ public class LuceneIndexEditor implement
         boolean dirty = false;
         for (PropertyState property : state.getProperties()) {
             String pname = property.getName();
-            if (isVisible(pname)
-                    && context.includeProperty(pname)) {
 
+            if (!isVisible(pname)) {
+                continue;
+            }
+
+            if (context.getDefinition().isOrdered(pname)) {
+                dirty = addTypedOrderedFields(fields, property);
+            }
+
+            if (context.includeProperty(pname)) {
                 //In case of fulltext we also check if given type is enabled for indexing
                 //TODO Use context.includePropertyType however that cause issue. Need
                 //to make filtering based on type consistent both on indexing side and
@@ -267,8 +278,6 @@ public class LuceneIndexEditor implement
         for (int i = 0; i < property.count(); i++) {
             Field f = null;
             if (tag == Type.LONG.tag()) {
-                //TODO Distinguish fields which need to be used for search and for sort
-                //If a field is only used for Sort then it can be stored with less precision
                 f = new LongField(name, property.getValue(Type.LONG, i), Field.Store.NO);
             } else if (tag == Type.DATE.tag()) {
                 String date = property.getValue(Type.DATE, i);
@@ -288,6 +297,38 @@ public class LuceneIndexEditor implement
         return fieldAdded;
     }
 
+    private boolean addTypedOrderedFields(List<Field> fields, PropertyState property) throws CommitFailedException {
+        int tag = property.getType().tag();
+        String name = FieldNames.createDocValFieldName(property.getName());
+        boolean fieldAdded = false;
+        for (int i = 0; i < property.count(); i++) {
+            Field f = null;
+            if (tag == Type.LONG.tag()) {
+                //TODO Distinguish fields which need to be used for search and for sort
+                //If a field is only used for Sort then it can be stored with less precision
+                f = new NumericDocValuesField(name, property.getValue(Type.LONG, i));
+            } else if (tag == Type.DATE.tag()) {
+                String date = property.getValue(Type.DATE, i);
+                f = new NumericDocValuesField(name, FieldFactory.dateToLong(date));
+            } else if (tag == Type.DOUBLE.tag()) {
+                f = new DoubleDocValuesField(name, property.getValue(Type.DOUBLE, i));
+            } else if (tag == Type.BOOLEAN.tag()) {
+                f = new SortedDocValuesField(name,
+                        new BytesRef(property.getValue(Type.BOOLEAN, i).toString()));
+            } else if (tag == Type.STRING.tag()) {
+                f = new SortedDocValuesField(name,
+                        new BytesRef(property.getValue(Type.STRING, i)));
+            }
+
+            if (f != null) {
+                this.context.indexUpdate();
+                fields.add(f);
+                fieldAdded = true;
+            }
+        }
+        return fieldAdded;
+    }
+
     private static boolean isVisible(String name) {
         return name.charAt(0) != ':';
     }

Modified: jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LuceneIndexEditorContext.java
URL: http://svn.apache.org/viewvc/jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LuceneIndexEditorContext.java?rev=1632017&r1=1632016&r2=1632017&view=diff
==============================================================================
--- jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LuceneIndexEditorContext.java (original)
+++ jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LuceneIndexEditorContext.java Wed Oct 15 13:33:14 2014
@@ -174,4 +174,8 @@ public class LuceneIndexEditorContext {
     public int getPropertyTypes() {
         return definition.getPropertyTypes();
     }
+
+    public IndexDefinition getDefinition() {
+        return definition;
+    }
 }

Modified: jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LucenePropertyIndex.java
URL: http://svn.apache.org/viewvc/jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LucenePropertyIndex.java?rev=1632017&r1=1632016&r2=1632017&view=diff
==============================================================================
--- jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LucenePropertyIndex.java (original)
+++ jackrabbit/oak/trunk/oak-lucene/src/main/java/org/apache/jackrabbit/oak/plugins/index/lucene/LucenePropertyIndex.java Wed Oct 15 13:33:14 2014
@@ -392,11 +392,16 @@ public class LucenePropertyIndex impleme
         }
         IndexNode indexNode = acquireIndexNode(plan);
         try {
+            IndexDefinition defn = indexNode.getDefinition();
             SortField[] fields = new SortField[sortOrder.size()];
             for (int i = 0; i < sortOrder.size(); i++) {
                 OrderEntry oe = sortOrder.get(i);
                 boolean reverse = oe.getOrder() != OrderEntry.Order.ASCENDING;
-                fields[i] = new SortField(oe.getPropertyName(), toLuceneSortType(oe, indexNode.getDefinition()), reverse);
+                String propName = oe.getPropertyName();
+                if (defn.isOrdered(propName)){
+                    propName = FieldNames.createDocValFieldName(propName);
+                }
+                fields[i] = new SortField(propName, toLuceneSortType(oe, defn), reverse);
             }
             return new Sort(fields);
         } finally {

Modified: jackrabbit/oak/trunk/oak-lucene/src/test/java/org/apache/jackrabbit/oak/plugins/index/lucene/LucenePropertyIndexTest.java
URL: http://svn.apache.org/viewvc/jackrabbit/oak/trunk/oak-lucene/src/test/java/org/apache/jackrabbit/oak/plugins/index/lucene/LucenePropertyIndexTest.java?rev=1632017&r1=1632016&r2=1632017&view=diff
==============================================================================
--- jackrabbit/oak/trunk/oak-lucene/src/test/java/org/apache/jackrabbit/oak/plugins/index/lucene/LucenePropertyIndexTest.java (original)
+++ jackrabbit/oak/trunk/oak-lucene/src/test/java/org/apache/jackrabbit/oak/plugins/index/lucene/LucenePropertyIndexTest.java Wed Oct 15 13:33:14 2014
@@ -49,12 +49,16 @@ import org.junit.Test;
 
 import static com.google.common.collect.ImmutableSet.of;
 import static java.util.Arrays.asList;
+import static org.apache.jackrabbit.oak.api.Type.STRINGS;
 import static org.apache.jackrabbit.oak.plugins.index.IndexConstants.INDEX_DEFINITIONS_NAME;
 import static org.apache.jackrabbit.oak.plugins.index.IndexConstants.INDEX_DEFINITIONS_NODE_TYPE;
 import static org.apache.jackrabbit.oak.plugins.index.IndexConstants.REINDEX_PROPERTY_NAME;
 import static org.apache.jackrabbit.oak.plugins.index.IndexConstants.TYPE_PROPERTY_NAME;
+import static org.apache.jackrabbit.oak.plugins.index.lucene.LuceneIndexConstants.INCLUDE_PROPERTY_NAMES;
+import static org.apache.jackrabbit.oak.plugins.index.lucene.LuceneIndexConstants.ORDERED_PROP_NAMES;
 import static org.apache.jackrabbit.oak.plugins.index.lucene.LuceneIndexEditorTest.createCal;
 import static org.apache.jackrabbit.oak.plugins.index.property.OrderedIndex.OrderDirection;
+import static org.apache.jackrabbit.oak.plugins.memory.PropertyStates.createProperty;
 import static org.junit.Assert.assertEquals;
 import static org.junit.Assert.assertThat;
 import static org.junit.matchers.JUnitMatchers.containsString;
@@ -225,6 +229,22 @@ public class LucenePropertyIndexTest ext
         propIdx.setProperty(LuceneIndexConstants.PROP_TYPE, PropertyType.TYPENAME_LONG);
         root.commit();
 
+        assertSortedtLong();
+    }
+
+    @Test
+    public void sortQueriesWithLong_OrderedProps() throws Exception {
+        Tree idx = createIndex("test1", of("foo", "bar"));
+        idx.setProperty(createProperty(INCLUDE_PROPERTY_NAMES, of("bar"), STRINGS));
+        idx.setProperty(createProperty(ORDERED_PROP_NAMES, of("foo"), STRINGS));
+        Tree propIdx = idx.addChild("foo");
+        propIdx.setProperty(LuceneIndexConstants.PROP_TYPE, PropertyType.TYPENAME_LONG);
+        root.commit();
+
+        assertSortedtLong();
+    }
+
+    void assertSortedtLong() throws CommitFailedException {
         Tree test = root.getTree("/").addChild("test");
         List<Long> values = createLongs(NUMBER_OF_NODES);
         List<Tuple> tuples = Lists.newArrayListWithCapacity(values.size());
@@ -247,6 +267,22 @@ public class LucenePropertyIndexTest ext
         propIdx.setProperty(LuceneIndexConstants.PROP_TYPE, PropertyType.TYPENAME_DOUBLE);
         root.commit();
 
+        assertSortedDouble();
+    }
+
+    @Test
+    public void sortQueriesWithDouble_OrderedProps() throws Exception {
+        Tree idx = createIndex("test1", of("foo", "bar"));
+        idx.setProperty(createProperty(INCLUDE_PROPERTY_NAMES, of("bar"), STRINGS));
+        idx.setProperty(createProperty(ORDERED_PROP_NAMES, of("foo"), STRINGS));
+        Tree propIdx = idx.addChild("foo");
+        propIdx.setProperty(LuceneIndexConstants.PROP_TYPE, PropertyType.TYPENAME_DOUBLE);
+        root.commit();
+
+        assertSortedDouble();
+    }
+
+    void assertSortedDouble() throws CommitFailedException {
         Tree test = root.getTree("/").addChild("test");
         List<Double> values = createDoubles(NUMBER_OF_NODES);
         List<Tuple> tuples = Lists.newArrayListWithCapacity(values.size());
@@ -268,6 +304,21 @@ public class LucenePropertyIndexTest ext
         idx.addChild("foo");
         root.commit();
 
+        assertSortedString();
+    }
+
+    @Test
+    public void sortQueriesWithString_OrderedProps() throws Exception {
+        Tree idx = createIndex("test1", of("foo", "bar"));
+        idx.setProperty(createProperty(INCLUDE_PROPERTY_NAMES, of("bar"), STRINGS));
+        idx.setProperty(createProperty(ORDERED_PROP_NAMES, of("foo"), STRINGS));
+        idx.addChild("foo");
+        root.commit();
+
+        assertSortedString();
+    }
+
+    void assertSortedString() throws CommitFailedException {
         Tree test = root.getTree("/").addChild("test");
         List<String> values = createStrings(NUMBER_OF_NODES);
         List<Tuple> tuples = Lists.newArrayListWithCapacity(values.size());
@@ -290,6 +341,22 @@ public class LucenePropertyIndexTest ext
         propIdx.setProperty(LuceneIndexConstants.PROP_TYPE, PropertyType.TYPENAME_DATE);
         root.commit();
 
+        assertSortedDate();
+    }
+
+    @Test
+    public void sortQueriesWithDate_OrderedProps() throws Exception {
+        Tree idx = createIndex("test1", of("foo", "bar"));
+        idx.setProperty(createProperty(INCLUDE_PROPERTY_NAMES, of("bar"), STRINGS));
+        idx.setProperty(createProperty(ORDERED_PROP_NAMES, of("foo"), STRINGS));
+        Tree propIdx = idx.addChild("foo");
+        propIdx.setProperty(LuceneIndexConstants.PROP_TYPE, PropertyType.TYPENAME_DATE);
+        root.commit();
+
+        assertSortedDate();
+    }
+
+    void assertSortedDate() throws ParseException, CommitFailedException {
         Tree test = root.getTree("/").addChild("test");
         List<Calendar> values = createDates(NUMBER_OF_NODES);
         List<Tuple> tuples = Lists.newArrayListWithCapacity(values.size());