Fixed an issue where and IndexOutOfBoundsException was thrown when a date_/histogram aggregation was defined on unmapped field and also had a sub aggregation. The root cause there was that in such case, the estimated bucket count was 0, and the code was not designed to handle that well.

Closes #5179

src/main/java/org/elasticsearch/search/aggregations/AggregatorFactories.java

@@ -73,9 +73,13 @@ public class AggregatorFactories {
                 ObjectArray<Aggregator> aggregators;
 
                 {
-                    aggregators = BigArrays.newObjectArray(estimatedBucketsCount, context.pageCacheRecycler());
+                    // if estimated count is zero, we at least create a single aggregator.
+                    // The estimated count is just an estimation and we can't rely on how it's estimated (that is, an
+                    // estimation of 0 should not imply that we'll end up without any buckets)
+                    long arraySize = estimatedBucketsCount > 0 ?  estimatedBucketsCount : 1;
+                    aggregators = BigArrays.newObjectArray(arraySize , context.pageCacheRecycler());
                     aggregators.set(0, first);
-                    for (long i = 1; i < estimatedBucketsCount; ++i) {
+                    for (long i = 1; i < arraySize; ++i) {
                         aggregators.set(i, createAndRegisterContextAware(parent.context(), factory, parent, estimatedBucketsCount));
                     }
                 }

src/test/java/org/elasticsearch/search/aggregations/CombiTests.java

@@ -25,18 +25,21 @@ import org.elasticsearch.action.index.IndexRequestBuilder;
 import org.elasticsearch.action.search.SearchResponse;
 import org.elasticsearch.common.settings.ImmutableSettings;
 import org.elasticsearch.common.settings.Settings;
+import org.elasticsearch.search.aggregations.bucket.histogram.Histogram;
 import org.elasticsearch.search.aggregations.bucket.missing.Missing;
 import org.elasticsearch.search.aggregations.bucket.terms.Terms;
 import org.elasticsearch.test.ElasticsearchIntegrationTest;
+import org.hamcrest.Matchers;
 import org.junit.Test;
 
 import java.util.Collection;
 
 import static org.elasticsearch.common.xcontent.XContentFactory.jsonBuilder;
-import static org.elasticsearch.search.aggregations.AggregationBuilders.missing;
+import static org.elasticsearch.search.aggregations.AggregationBuilders.*;
-import static org.elasticsearch.search.aggregations.AggregationBuilders.terms;
 import static org.elasticsearch.test.hamcrest.ElasticsearchAssertions.assertSearchResponse;
 import static org.hamcrest.CoreMatchers.equalTo;
+import static org.hamcrest.CoreMatchers.is;
+import static org.hamcrest.core.IsNull.notNullValue;
 
 /**
  *
@@ -110,4 +113,34 @@ public class CombiTests extends ElasticsearchIntegrationTest {
         }
         assertTrue(values.isEmpty());
     }
+
+
+    /**
+     * Some top aggs (eg. date_/histogram) that are executed on unmapped fields, will generate an estimate count of buckets - zero.
+     * when the sub aggregator is then created, it will take this estimation into account. This used to cause
+     * and an ArrayIndexOutOfBoundsException...
+     */
+    @Test
+    public void subAggregationForTopAggregationOnUnmappedField() throws Exception {
+
+        prepareCreate("idx").addMapping("type", jsonBuilder()
+                .startObject()
+                .startObject("type").startObject("properties")
+                    .startObject("name").field("type", "string").endObject()
+                    .startObject("value").field("type", "integer").endObject()
+                .endObject().endObject()
+                .endObject()).execute().actionGet();
+
+        ensureSearchable("idx");
+
+        SearchResponse searchResponse = client().prepareSearch("idx")
+                .addAggregation(histogram("values").field("value1").interval(1)
+                        .subAggregation(terms("names").field("name")))
+                .execute().actionGet();
+
+        assertThat(searchResponse.getHits().getTotalHits(), Matchers.equalTo(0l));
+        Histogram values = searchResponse.getAggregations().get("values");
+        assertThat(values, notNullValue());
+        assertThat(values.getBuckets().isEmpty(), is(true));
+    }
 }