facet refactoring, default collector base post implementation

automatically implement post based on collector

Author: kimchy

src/main/java/org/elasticsearch/common/lucene/docset/ContextDocIdSet.java

@@ -23,6 +23,7 @@
 import org.apache.lucene.search.DocIdSet;
 
 /**
+ * A holder for a {@link DocIdSet} and the {@link AtomicReaderContext} it is associated with.
  */
 public class ContextDocIdSet {
 

src/main/java/org/elasticsearch/search/facet/FacetExecutor.java

@@ -20,6 +20,7 @@
 package org.elasticsearch.search.facet;
 
 import org.apache.lucene.search.DocIdSet;
+import org.apache.lucene.search.DocIdSetIterator;
 import org.apache.lucene.search.Filter;
 import org.apache.lucene.search.Scorer;
 import org.elasticsearch.common.lucene.docset.AndDocIdSet;
@@ -34,8 +35,8 @@
  * A facet processor ends up actually executing the relevant facet for a specific
  * search request.
  * <p/>
- * The facet executor requires at least one of the {@link #collector()} or {@link #post()} methods to be
- * implemented.
+ * The facet executor requires at least the {@link #collector()} method to be implemented,
+ * with an optional {@link #post()} implementation if specific optimizations can be done.
  */
 public abstract class FacetExecutor {
 
@@ -46,6 +47,9 @@ public static abstract class Post {
 
  public abstract void executePost(List<ContextDocIdSet> docSets) throws IOException;
 
+ /**
+ * A filtered post execution.
+ */
  public static class Filtered extends Post {
 
  private final Post post;
@@ -71,6 +75,32 @@ public void executePost(List<ContextDocIdSet> docSets) throws IOException {
  post.executePost(filteredEntries);
  }
  }
+
+ /**
+ * A {@link FacetExecutor.Collector} based post.
+ */
+ public static class Collector extends Post {
+
+ private final FacetExecutor.Collector collector;
+
+ public Collector(FacetExecutor.Collector collector) {
+ this.collector = collector;
+ }
+
+ @Override
+ public void executePost(List<ContextDocIdSet> docSets) throws IOException {
+ for (int i = 0; i < docSets.size(); i++) {
+ ContextDocIdSet docSet = docSets.get(i);
+ collector.setNextReader(docSet.context);
+ DocIdSetIterator it = docSet.docSet.iterator();
+ int doc;
+ while ((doc = it.nextDoc()) != DocIdSetIterator.NO_MORE_DOCS) {
+ collector.collect(doc);
+ }
+ }
+ collector.postCollection();
+ }
+ }
  }
 
  /**
@@ -117,7 +147,12 @@ public static enum Mode {
  public abstract Collector collector();
 
  /**
- * A post based facet that executes the facet using the aggregated docs relevant.
+ * A post based facet that executes the facet using the aggregated docs. By default
+ * uses the {@link Post.Collector} based implementation.
+ * <p/>
+ * Can be overridden if a more optimized non collector based implementation can be implemented.
  */
- public abstract Post post();
+ public Post post() {
+ return new Post.Collector(collector());
+ }
 }

src/main/java/org/elasticsearch/search/facet/datehistogram/CountDateHistogramFacetExecutor.java

@@ -21,17 +21,14 @@
 
 import gnu.trove.map.hash.TLongLongHashMap;
 import org.apache.lucene.index.AtomicReaderContext;
-import org.apache.lucene.search.DocIdSetIterator;
 import org.elasticsearch.common.CacheRecycler;
 import org.elasticsearch.common.joda.TimeZoneRounding;
-import org.elasticsearch.common.lucene.docset.ContextDocIdSet;
 import org.elasticsearch.index.fielddata.IndexNumericFieldData;
 import org.elasticsearch.index.fielddata.LongValues;
 import org.elasticsearch.search.facet.FacetExecutor;
 import org.elasticsearch.search.facet.InternalFacet;
 
 import java.io.IOException;
-import java.util.List;
 
 /**
  * A date histogram facet collector that uses the same field as the key as well as the
@@ -58,32 +55,11 @@ public Collector collector() {
  return new Collector();
  }
 
- @Override
- public Post post() {
- return new Post();
- }
-
  @Override
  public InternalFacet buildFacet(String facetName) {
  return new InternalCountDateHistogramFacet(facetName, comparatorType, counts, true);
  }
 
- class Post extends FacetExecutor.Post {
-
- @Override
- public void executePost(List<ContextDocIdSet> docSets) throws IOException {
- DateHistogramProc histoProc = new DateHistogramProc(counts, tzRounding);
- for (ContextDocIdSet docSet : docSets) {
- LongValues values = indexFieldData.load(docSet.context).getLongValues();
- DocIdSetIterator it = docSet.docSet.iterator();
- int doc;
- while ((doc = it.nextDoc()) != DocIdSetIterator.NO_MORE_DOCS) {
- values.forEachValueInDoc(doc, histoProc);
- }
- }
- }
- }
-
  class Collector extends FacetExecutor.Collector {
 
  private LongValues values;

src/main/java/org/elasticsearch/search/facet/datehistogram/ValueDateHistogramFacetExecutor.java

@@ -20,10 +20,8 @@
 package org.elasticsearch.search.facet.datehistogram;
 
 import org.apache.lucene.index.AtomicReaderContext;
-import org.apache.lucene.search.DocIdSetIterator;
 import org.elasticsearch.common.CacheRecycler;
 import org.elasticsearch.common.joda.TimeZoneRounding;
-import org.elasticsearch.common.lucene.docset.ContextDocIdSet;
 import org.elasticsearch.common.trove.ExtTLongObjectHashMap;
 import org.elasticsearch.index.fielddata.DoubleValues;
 import org.elasticsearch.index.fielddata.IndexNumericFieldData;
@@ -32,7 +30,6 @@
 import org.elasticsearch.search.facet.InternalFacet;
 
 import java.io.IOException;
-import java.util.List;
 
 /**
  * A histogram facet collector that uses different fields for the key and the value.
@@ -60,34 +57,11 @@ public Collector collector() {
  return new Collector();
  }
 
- @Override
- public Post post() {
- return new Post();
- }
-
  @Override
  public InternalFacet buildFacet(String facetName) {
  return new InternalFullDateHistogramFacet(facetName, comparatorType, entries, true);
  }
 
- class Post extends FacetExecutor.Post {
-
- @Override
- public void executePost(List<ContextDocIdSet> docSets) throws IOException {
- DateHistogramProc histoProc = new DateHistogramProc(tzRounding, ValueDateHistogramFacetExecutor.this.entries);
- for (ContextDocIdSet docSet : docSets) {
- LongValues keyValues = keyIndexFieldData.load(docSet.context).getLongValues();
- histoProc.valueValues = valueIndexFieldData.load(docSet.context).getDoubleValues();
-
- DocIdSetIterator it = docSet.docSet.iterator();
- int doc;
- while ((doc = it.nextDoc()) != DocIdSetIterator.NO_MORE_DOCS) {
- keyValues.forEachValueInDoc(doc, histoProc);
- }
- }
- }
- }
-
  class Collector extends FacetExecutor.Collector {
 
  private final DateHistogramProc histoProc;

src/main/java/org/elasticsearch/search/facet/datehistogram/ValueScriptDateHistogramFacetExecutor.java

@@ -20,11 +20,9 @@
 package org.elasticsearch.search.facet.datehistogram;
 
 import org.apache.lucene.index.AtomicReaderContext;
-import org.apache.lucene.search.DocIdSetIterator;
 import org.apache.lucene.search.Scorer;
 import org.elasticsearch.common.CacheRecycler;
 import org.elasticsearch.common.joda.TimeZoneRounding;
-import org.elasticsearch.common.lucene.docset.ContextDocIdSet;
 import org.elasticsearch.common.trove.ExtTLongObjectHashMap;
 import org.elasticsearch.index.fielddata.IndexNumericFieldData;
 import org.elasticsearch.index.fielddata.LongValues;
@@ -33,7 +31,6 @@
 import org.elasticsearch.search.facet.InternalFacet;
 
 import java.io.IOException;
-import java.util.List;
 
 /**
  * A histogram facet collector that uses the same field as the key as well as the
@@ -62,33 +59,11 @@ public Collector collector() {
  return new Collector();
  }
 
- @Override
- public Post post() {
- return new Post();
- }
-
  @Override
  public InternalFacet buildFacet(String facetName) {
  return new InternalFullDateHistogramFacet(facetName, comparatorType, entries, true);
  }
 
- class Post extends FacetExecutor.Post {
-
- @Override
- public void executePost(List<ContextDocIdSet> docSets) throws IOException {
- DateHistogramProc histoProc = new DateHistogramProc(tzRounding, valueScript, ValueScriptDateHistogramFacetExecutor.this.entries);
- for (ContextDocIdSet entry : docSets) {
- LongValues keyValues = keyIndexFieldData.load(entry.context).getLongValues();
- valueScript.setNextReader(entry.context);
- DocIdSetIterator it = entry.docSet.iterator();
- int doc;
- while ((doc = it.nextDoc()) != DocIdSetIterator.NO_MORE_DOCS) {
- keyValues.forEachValueInDoc(doc, histoProc);
- }
- }
- }
- }
-
  class Collector extends FacetExecutor.Collector {
 
  private final DateHistogramProc histoProc;

src/main/java/org/elasticsearch/search/facet/geodistance/GeoDistanceFacetExecutor.java

@@ -60,11 +60,6 @@ public Collector collector() {
  return new Collector();
  }
 
- @Override
- public Post post() {
- return null;
- }
-
  @Override
  public InternalFacet buildFacet(String facetName) {
  return new InternalGeoDistanceFacet(facetName, entries);

src/main/java/org/elasticsearch/search/facet/geodistance/ScriptGeoDistanceFacetExecutor.java

@@ -50,11 +50,6 @@ public Collector collector() {
  return new Collector();
  }
 
- @Override
- public Post post() {
- return null;
- }
-
  class Collector extends GeoDistanceFacetExecutor.Collector {
 
  private Aggregator scriptAggregator;

src/main/java/org/elasticsearch/search/facet/geodistance/ValueGeoDistanceFacetExecutor.java

@@ -48,11 +48,6 @@ public Collector collector() {
  return new Collector();
  }
 
- @Override
- public Post post() {
- return null;
- }
-
  class Collector extends GeoDistanceFacetExecutor.Collector {
 
  Collector() {

src/main/java/org/elasticsearch/search/facet/histogram/CountHistogramFacetExecutor.java

@@ -55,11 +55,6 @@ public Collector collector() {
  return new Collector();
  }
 
- @Override
- public Post post() {
- return null;
- }
-
  @Override
  public InternalFacet buildFacet(String facetName) {
  return new InternalCountHistogramFacet(facetName, comparatorType, counts, true);

src/main/java/org/elasticsearch/search/facet/histogram/FullHistogramFacetExecutor.java

@@ -55,11 +55,6 @@ public Collector collector() {
  return new Collector();
  }
 
- @Override
- public Post post() {
- return null;
- }
-
  @Override
  public InternalFacet buildFacet(String facetName) {
  return new InternalFullHistogramFacet(facetName, comparatorType, entries, true);