@Test
public void script_MultiValued() throws Exception {
SearchResponse response =
client()
.prepareSearch("idx")
.setTypes("type")
.addAggregation(
terms("terms")
.executionHint(randomExecutionHint())
.script("doc['" + MULTI_VALUED_FIELD_NAME + "'].values"))
.execute()
.actionGet();
assertSearchResponse(response);
Terms terms = response.getAggregations().get("terms");
assertThat(terms, notNullValue());
assertThat(terms.getName(), equalTo("terms"));
assertThat(terms.getBuckets().size(), equalTo(6));
for (int i = 0; i < 6; i++) {
Terms.Bucket bucket = terms.getBucketByKey("val" + i);
assertThat(bucket, notNullValue());
assertThat(key(bucket), equalTo("val" + i));
if (i == 0 || i == 5) {
assertThat(bucket.getDocCount(), equalTo(1l));
} else {
assertThat(bucket.getDocCount(), equalTo(2l));
}
}
}
@Test
public void testChildrenAggs() throws Exception {
SearchResponse searchResponse =
client()
.prepareSearch("test")
.setQuery(matchQuery("randomized", true))
.addAggregation(
terms("category")
.field("category")
.size(0)
.subAggregation(
children("to_comment")
.childType("comment")
.subAggregation(
terms("commenters")
.field("commenter")
.size(0)
.subAggregation(topHits("top_comments")))))
.get();
assertSearchResponse(searchResponse);
Terms categoryTerms = searchResponse.getAggregations().get("category");
assertThat(categoryTerms.getBuckets().size(), equalTo(categoryToControl.size()));
for (Map.Entry<String, Control> entry1 : categoryToControl.entrySet()) {
Terms.Bucket categoryBucket = categoryTerms.getBucketByKey(entry1.getKey());
assertThat(categoryBucket.getKey(), equalTo(entry1.getKey()));
assertThat(categoryBucket.getDocCount(), equalTo((long) entry1.getValue().articleIds.size()));
Children childrenBucket = categoryBucket.getAggregations().get("to_comment");
assertThat(childrenBucket.getName(), equalTo("to_comment"));
assertThat(childrenBucket.getDocCount(), equalTo((long) entry1.getValue().commentIds.size()));
assertThat(
(long) childrenBucket.getProperty("_count"),
equalTo((long) entry1.getValue().commentIds.size()));
Terms commentersTerms = childrenBucket.getAggregations().get("commenters");
assertThat((Terms) childrenBucket.getProperty("commenters"), sameInstance(commentersTerms));
assertThat(
commentersTerms.getBuckets().size(),
equalTo(entry1.getValue().commenterToCommentId.size()));
for (Map.Entry<String, Set<String>> entry2 :
entry1.getValue().commenterToCommentId.entrySet()) {
Terms.Bucket commentBucket = commentersTerms.getBucketByKey(entry2.getKey());
assertThat(commentBucket.getKey(), equalTo(entry2.getKey()));
assertThat(commentBucket.getDocCount(), equalTo((long) entry2.getValue().size()));
TopHits topHits = commentBucket.getAggregations().get("top_comments");
for (SearchHit searchHit : topHits.getHits().getHits()) {
assertThat(entry2.getValue().contains(searchHit.getId()), is(true));
}
}
}
}
@Test
public void singleValueField_WithMaxSize() throws Exception {
SearchResponse response =
client()
.prepareSearch("idx")
.setTypes("high_card_type")
.addAggregation(
terms("terms")
.executionHint(randomExecutionHint())
.field(SINGLE_VALUED_FIELD_NAME)
.size(20)
.order(
Terms.Order.term(
true))) // we need to sort by terms cause we're checking the first 20
// values
.execute()
.actionGet();
assertSearchResponse(response);
Terms terms = response.getAggregations().get("terms");
assertThat(terms, notNullValue());
assertThat(terms.getName(), equalTo("terms"));
assertThat(terms.getBuckets().size(), equalTo(20));
for (int i = 0; i < 20; i++) {
Terms.Bucket bucket = terms.getBucketByKey("val" + Strings.padStart(i + "", 3, '0'));
assertThat(bucket, notNullValue());
assertThat(key(bucket), equalTo("val" + Strings.padStart(i + "", 3, '0')));
assertThat(bucket.getDocCount(), equalTo(1l));
}
}
@Test
public void singleValuedField_OrderedBySingleValueSubAggregationAsc() throws Exception {
boolean asc = true;
SearchResponse response =
client()
.prepareSearch("idx")
.setTypes("type")
.addAggregation(
terms("terms")
.executionHint(randomExecutionHint())
.field(SINGLE_VALUED_FIELD_NAME)
.order(Terms.Order.aggregation("avg_i", asc))
.subAggregation(avg("avg_i").field("i")))
.execute()
.actionGet();
assertSearchResponse(response);
Terms terms = response.getAggregations().get("terms");
assertThat(terms, notNullValue());
assertThat(terms.getName(), equalTo("terms"));
assertThat(terms.getBuckets().size(), equalTo(5));
int i = 0;
for (Terms.Bucket bucket : terms.getBuckets()) {
assertThat(bucket, notNullValue());
assertThat(key(bucket), equalTo("val" + i));
assertThat(bucket.getDocCount(), equalTo(1l));
Avg avg = bucket.getAggregations().get("avg_i");
assertThat(avg, notNullValue());
assertThat(avg.getValue(), equalTo((double) i));
i++;
}
}
@Test
public void singleValueField_OrderedByTermDesc() throws Exception {
SearchResponse response =
client()
.prepareSearch("idx")
.setTypes("type")
.addAggregation(
terms("terms")
.executionHint(randomExecutionHint())
.field(SINGLE_VALUED_FIELD_NAME)
.order(Terms.Order.term(false)))
.execute()
.actionGet();
assertSearchResponse(response);
Terms terms = response.getAggregations().get("terms");
assertThat(terms, notNullValue());
assertThat(terms.getName(), equalTo("terms"));
assertThat(terms.getBuckets().size(), equalTo(5));
int i = 4;
for (Terms.Bucket bucket : terms.getBuckets()) {
assertThat(bucket, notNullValue());
assertThat(key(bucket), equalTo("val" + i));
assertThat(bucket.getDocCount(), equalTo(1l));
i--;
}
}
@Test
public void multiValuedField_WithValueScript_NotUnique() throws Exception {
SearchResponse response =
client()
.prepareSearch("idx")
.setTypes("type")
.addAggregation(
terms("terms")
.executionHint(randomExecutionHint())
.field(MULTI_VALUED_FIELD_NAME)
.script("_value.substring(0,3)"))
.execute()
.actionGet();
assertSearchResponse(response);
Terms terms = response.getAggregations().get("terms");
assertThat(terms, notNullValue());
assertThat(terms.getName(), equalTo("terms"));
assertThat(terms.getBuckets().size(), equalTo(1));
Terms.Bucket bucket = terms.getBucketByKey("val");
assertThat(bucket, notNullValue());
assertThat(key(bucket), equalTo("val"));
assertThat(bucket.getDocCount(), equalTo(5l));
}
public Map<String, Long> countByField(IndexField indexField, FilterBuilder filter) {
Map<String, Long> counts = new HashMap<>();
SearchRequestBuilder request =
client
.prepareSearch(this.getIndexName())
.setTypes(this.getIndexType())
.setQuery(QueryBuilders.filteredQuery(QueryBuilders.matchAllQuery(), filter))
.setSize(0)
.addAggregation(
AggregationBuilders.terms(indexField.field())
.field(indexField.field())
.order(Terms.Order.count(false))
.size(Integer.MAX_VALUE)
.minDocCount(0));
SearchResponse response = request.get();
Terms values = response.getAggregations().get(indexField.field());
for (Terms.Bucket value : values.getBuckets()) {
counts.put(value.getKey(), value.getDocCount());
}
return counts;
}
@Test
public void script_SingleValue_WithSubAggregator_Inherited() throws Exception {
SearchResponse response =
client()
.prepareSearch("idx")
.setTypes("type")
.addAggregation(
terms("terms")
.executionHint(randomExecutionHint())
.script("doc['" + SINGLE_VALUED_FIELD_NAME + "'].value")
.subAggregation(count("count")))
.execute()
.actionGet();
assertSearchResponse(response);
Terms terms = response.getAggregations().get("terms");
assertThat(terms, notNullValue());
assertThat(terms.getName(), equalTo("terms"));
assertThat(terms.getBuckets().size(), equalTo(5));
for (int i = 0; i < 5; i++) {
Terms.Bucket bucket = terms.getBucketByKey("val" + i);
assertThat(bucket, notNullValue());
assertThat(key(bucket), equalTo("val" + i));
assertThat(bucket.getDocCount(), equalTo(1l));
ValueCount valueCount = bucket.getAggregations().get("count");
assertThat(valueCount, notNullValue());
assertThat(valueCount.getValue(), equalTo(1l));
}
}
void assertBasicAggregationWorks(String indexName) {
// histogram on a long
SearchResponse searchRsp =
client()
.prepareSearch(indexName)
.addAggregation(AggregationBuilders.histogram("histo").field("long_sort").interval(10))
.get();
ElasticsearchAssertions.assertSearchResponse(searchRsp);
Histogram histo = searchRsp.getAggregations().get("histo");
assertNotNull(histo);
long totalCount = 0;
for (Histogram.Bucket bucket : histo.getBuckets()) {
totalCount += bucket.getDocCount();
}
assertEquals(totalCount, searchRsp.getHits().getTotalHits());
// terms on a boolean
searchRsp =
client()
.prepareSearch(indexName)
.addAggregation(AggregationBuilders.terms("bool_terms").field("bool"))
.get();
Terms terms = searchRsp.getAggregations().get("bool_terms");
totalCount = 0;
for (Terms.Bucket bucket : terms.getBuckets()) {
totalCount += bucket.getDocCount();
}
assertEquals(totalCount, searchRsp.getHits().getTotalHits());
}
@Test
public void partiallyUnmapped() throws Exception {
SearchResponse response =
client()
.prepareSearch("idx", "idx_unmapped")
.setTypes("type")
.addAggregation(
terms("terms").executionHint(randomExecutionHint()).field(SINGLE_VALUED_FIELD_NAME))
.execute()
.actionGet();
assertSearchResponse(response);
Terms terms = response.getAggregations().get("terms");
assertThat(terms, notNullValue());
assertThat(terms.getName(), equalTo("terms"));
assertThat(terms.getBuckets().size(), equalTo(5));
for (int i = 0; i < 5; i++) {
Terms.Bucket bucket = terms.getBucketByKey("val" + i);
assertThat(bucket, notNullValue());
assertThat(key(bucket), equalTo("val" + i));
assertThat(bucket.getDocCount(), equalTo(1l));
}
}
@Test
public void testSimpleSubAggregations() throws Exception {
final String query =
String.format(
"SELECT /*! DOCS_WITH_AGGREGATION(10) */ * FROM %s/account GROUP BY (gender), (state) ",
TEST_INDEX);
SqlElasticSearchRequestBuilder select = getSearchRequestBuilder(query);
SearchResponse response = (SearchResponse) select.get();
Aggregations result = response.getAggregations();
Terms gender = result.get("gender");
for (Terms.Bucket genderBucket : gender.getBuckets()) {
String genderKey = genderBucket.getKey();
Assert.assertTrue("Gender should be m or f", genderKey.equals("m") || genderKey.equals("f"));
}
Assert.assertEquals(2, gender.getBuckets().size());
Terms state = result.get("state");
for (Terms.Bucket stateBucket : state.getBuckets()) {
if (stateBucket.getKey().equalsIgnoreCase("ak")) {
Assert.assertTrue("There are 22 entries for state ak", stateBucket.getDocCount() == 22);
}
}
Assert.assertEquals(response.getHits().totalHits(), 1000);
Assert.assertEquals(response.getHits().hits().length, 10);
}
@Test
public void stringTermsNestedIntoPerBucketAggregator() throws Exception {
// no execution hint so that the logic that decides whether or not to use ordinals is executed
SearchResponse response =
client()
.prepareSearch("idx")
.setTypes("type")
.addAggregation(
filter("filter")
.filter(termFilter(MULTI_VALUED_FIELD_NAME, "val3"))
.subAggregation(terms("terms").field(MULTI_VALUED_FIELD_NAME)))
.execute()
.actionGet();
assertThat(response.getFailedShards(), equalTo(0));
Filter filter = response.getAggregations().get("filter");
Terms terms = filter.getAggregations().get("terms");
assertThat(terms, notNullValue());
assertThat(terms.getName(), equalTo("terms"));
assertThat(terms.getBuckets().size(), equalTo(3));
for (int i = 2; i <= 4; i++) {
Terms.Bucket bucket = terms.getBucketByKey("val" + i);
assertThat(bucket, notNullValue());
assertThat(key(bucket), equalTo("val" + i));
assertThat(bucket.getDocCount(), equalTo(i == 3 ? 2L : 1L));
}
}
public static LinkedHashMap<String, Long> termsToMap(Terms terms) {
LinkedHashMap<String, Long> map = new LinkedHashMap<>();
List<Terms.Bucket> buckets = terms.getBuckets();
for (Terms.Bucket bucket : buckets) {
map.put(bucket.getKeyAsString(), bucket.getDocCount());
}
return map;
}
// Duel between histograms and scripted terms
public void testDuelTermsHistogram() throws Exception {
createIndex("idx");
final int numDocs = scaledRandomIntBetween(500, 5000);
final int maxNumTerms = randomIntBetween(10, 2000);
final int interval = randomIntBetween(1, 100);
final Integer[] values = new Integer[maxNumTerms];
for (int i = 0; i < values.length; ++i) {
values[i] = randomInt(maxNumTerms * 3) - maxNumTerms;
}
for (int i = 0; i < numDocs; ++i) {
XContentBuilder source =
jsonBuilder().startObject().field("num", randomDouble()).startArray("values");
final int numValues = randomInt(4);
for (int j = 0; j < numValues; ++j) {
source = source.value(randomFrom(values));
}
source = source.endArray().endObject();
client().prepareIndex("idx", "type").setSource(source).execute().actionGet();
}
assertNoFailures(
client()
.admin()
.indices()
.prepareRefresh("idx")
.setIndicesOptions(IndicesOptions.lenientExpandOpen())
.execute()
.get());
SearchResponse resp =
client()
.prepareSearch("idx")
.addAggregation(
terms("terms")
.field("values")
.collectMode(randomFrom(SubAggCollectionMode.values()))
.script("floor(_value / interval)")
.param("interval", interval)
.size(maxNumTerms))
.addAggregation(histogram("histo").field("values").interval(interval))
.execute()
.actionGet();
assertSearchResponse(resp);
Terms terms = resp.getAggregations().get("terms");
assertThat(terms, notNullValue());
Histogram histo = resp.getAggregations().get("histo");
assertThat(histo, notNullValue());
assertThat(terms.getBuckets().size(), equalTo(histo.getBuckets().size()));
for (Terms.Bucket bucket : terms.getBuckets()) {
final long key = bucket.getKeyAsNumber().longValue() * interval;
final Histogram.Bucket histoBucket = histo.getBucketByKey(key);
assertEquals(bucket.getDocCount(), histoBucket.getDocCount());
}
}
@Test
public void singleValuedField_OrderedBySingleBucketSubAggregationAsc() throws Exception {
boolean asc = randomBoolean();
SearchResponse response =
client()
.prepareSearch("idx")
.setTypes("type")
.addAggregation(
terms("tags")
.executionHint(randomExecutionHint())
.field("tag")
.order(Terms.Order.aggregation("filter", asc))
.subAggregation(filter("filter").filter(FilterBuilders.matchAllFilter())))
.execute()
.actionGet();
assertSearchResponse(response);
Terms tags = response.getAggregations().get("tags");
assertThat(tags, notNullValue());
assertThat(tags.getName(), equalTo("tags"));
assertThat(tags.getBuckets().size(), equalTo(2));
Iterator<Terms.Bucket> iters = tags.getBuckets().iterator();
Terms.Bucket tag = iters.next();
assertThat(tag, notNullValue());
assertThat(key(tag), equalTo(asc ? "less" : "more"));
assertThat(tag.getDocCount(), equalTo(asc ? 2l : 3l));
Filter filter = tag.getAggregations().get("filter");
assertThat(filter, notNullValue());
assertThat(filter.getDocCount(), equalTo(asc ? 2l : 3l));
tag = iters.next();
assertThat(tag, notNullValue());
assertThat(key(tag), equalTo(asc ? "more" : "less"));
assertThat(tag.getDocCount(), equalTo(asc ? 3l : 2l));
filter = tag.getAggregations().get("filter");
assertThat(filter, notNullValue());
assertThat(filter.getDocCount(), equalTo(asc ? 3l : 2l));
}
@Test
public void multiValuedField_WithValueScript_WithInheritedSubAggregator() throws Exception {
SearchResponse response =
client()
.prepareSearch("idx")
.setTypes("type")
.addAggregation(
terms("terms")
.executionHint(randomExecutionHint())
.field(MULTI_VALUED_FIELD_NAME)
.script("'foo_' + _value")
.subAggregation(count("count")))
.execute()
.actionGet();
assertSearchResponse(response);
Terms terms = response.getAggregations().get("terms");
assertThat(terms, notNullValue());
assertThat(terms.getName(), equalTo("terms"));
assertThat(terms.getBuckets().size(), equalTo(6));
for (int i = 0; i < 6; i++) {
Terms.Bucket bucket = terms.getBucketByKey("foo_val" + i);
assertThat(bucket, notNullValue());
assertThat(key(bucket), equalTo("foo_val" + i));
if (i == 0 | i == 5) {
assertThat(bucket.getDocCount(), equalTo(1l));
ValueCount valueCount = bucket.getAggregations().get("count");
assertThat(valueCount, notNullValue());
assertThat(valueCount.getValue(), equalTo(2l));
} else {
assertThat(bucket.getDocCount(), equalTo(2l));
ValueCount valueCount = bucket.getAggregations().get("count");
assertThat(valueCount, notNullValue());
assertThat("term[" + key(bucket) + "]", valueCount.getValue(), equalTo(4l));
}
}
}
public void testNestNestedAggs() throws Exception {
SearchResponse response =
client()
.prepareSearch("idx_nested_nested_aggs")
.addAggregation(
nested("level1", "nested1")
.subAggregation(
terms("a")
.field("nested1.a")
.collectMode(aggCollectionMode)
.subAggregation(
nested("level2", "nested1.nested2")
.subAggregation(sum("sum").field("nested1.nested2.b")))))
.get();
assertSearchResponse(response);
Nested level1 = response.getAggregations().get("level1");
assertThat(level1, notNullValue());
assertThat(level1.getName(), equalTo("level1"));
assertThat(level1.getDocCount(), equalTo(2L));
StringTerms a = level1.getAggregations().get("a");
Terms.Bucket bBucket = a.getBucketByKey("a");
assertThat(bBucket.getDocCount(), equalTo(1L));
Nested level2 = bBucket.getAggregations().get("level2");
assertThat(level2.getDocCount(), equalTo(1L));
Sum sum = level2.getAggregations().get("sum");
assertThat(sum.getValue(), equalTo(2d));
a = level1.getAggregations().get("a");
bBucket = a.getBucketByKey("b");
assertThat(bBucket.getDocCount(), equalTo(1L));
level2 = bBucket.getAggregations().get("level2");
assertThat(level2.getDocCount(), equalTo(1L));
sum = level2.getAggregations().get("sum");
assertThat(sum.getValue(), equalTo(2d));
}
private void assertDocCountErrorWithinBounds(
int size, SearchResponse accurateResponse, SearchResponse testResponse) {
Terms accurateTerms = accurateResponse.getAggregations().get("terms");
assertThat(accurateTerms, notNullValue());
assertThat(accurateTerms.getName(), equalTo("terms"));
assertThat(accurateTerms.getDocCountError(), equalTo(0L));
Terms testTerms = testResponse.getAggregations().get("terms");
assertThat(testTerms, notNullValue());
assertThat(testTerms.getName(), equalTo("terms"));
assertThat(testTerms.getDocCountError(), greaterThanOrEqualTo(0L));
Collection<Bucket> testBuckets = testTerms.getBuckets();
assertThat(testBuckets.size(), lessThanOrEqualTo(size));
assertThat(accurateTerms.getBuckets().size(), greaterThanOrEqualTo(testBuckets.size()));
for (Terms.Bucket testBucket : testBuckets) {
assertThat(testBucket, notNullValue());
Terms.Bucket accurateBucket = accurateTerms.getBucketByKey(testBucket.getKeyAsString());
assertThat(accurateBucket, notNullValue());
assertThat(accurateBucket.getDocCountError(), equalTo(0L));
assertThat(testBucket.getDocCountError(), lessThanOrEqualTo(testTerms.getDocCountError()));
assertThat(
testBucket.getDocCount() + testBucket.getDocCountError(),
greaterThanOrEqualTo(accurateBucket.getDocCount()));
assertThat(
testBucket.getDocCount() - testBucket.getDocCountError(),
lessThanOrEqualTo(accurateBucket.getDocCount()));
}
for (Terms.Bucket accurateBucket : accurateTerms.getBuckets()) {
assertThat(accurateBucket, notNullValue());
Terms.Bucket testBucket = accurateTerms.getBucketByKey(accurateBucket.getKeyAsString());
if (testBucket == null) {
assertThat(accurateBucket.getDocCount(), lessThanOrEqualTo(testTerms.getDocCountError()));
}
}
}
/*
* 所有姓氏为"Smith"的员工的最大的共同点(兴趣爱好)是什么
* 聚合变成只包含和查询语句想匹配的文档
*/
@Test
public void queryAggregationTest() {
Client client = ClientTemplate.getClient();
SearchResponse response =
client
.prepareSearch("fccs")
.setTypes("employee")
.setQuery(QueryBuilders.matchQuery("lastName", "Smith"))
.addAggregation(AggregationBuilders.terms("by_interests").field("interests"))
.get();
Terms terms = response.getAggregations().get("by_interests");
for (Terms.Bucket bucket : terms.getBuckets()) {
System.out.println("key:" + bucket.getKey() + " doc_count:" + bucket.getDocCount());
}
client.close();
}
@Test
public void testMetric_topLevel() throws Exception {
SearchResponse response =
client()
.prepareSearch("idx")
.addAggregation(
terms("terms")
.field("tag")
.subAggregation(sum("sum").field(SINGLE_VALUED_FIELD_NAME)))
.addAggregation(maxBucket("max_bucket").setBucketsPaths("terms>sum"))
.execute()
.actionGet();
assertSearchResponse(response);
Terms terms = response.getAggregations().get("terms");
assertThat(terms, notNullValue());
assertThat(terms.getName(), equalTo("terms"));
List<Terms.Bucket> buckets = terms.getBuckets();
assertThat(buckets.size(), equalTo(interval));
List<String> maxKeys = new ArrayList<>();
double maxValue = Double.NEGATIVE_INFINITY;
for (int i = 0; i < interval; ++i) {
Terms.Bucket bucket = buckets.get(i);
assertThat(bucket, notNullValue());
assertThat((String) bucket.getKey(), equalTo("tag" + (i % interval)));
assertThat(bucket.getDocCount(), greaterThan(0l));
Sum sum = bucket.getAggregations().get("sum");
assertThat(sum, notNullValue());
if (sum.value() > maxValue) {
maxValue = sum.value();
maxKeys = new ArrayList<>();
maxKeys.add(bucket.getKeyAsString());
} else if (sum.value() == maxValue) {
maxKeys.add(bucket.getKeyAsString());
}
}
InternalBucketMetricValue maxBucketValue = response.getAggregations().get("max_bucket");
assertThat(maxBucketValue, notNullValue());
assertThat(maxBucketValue.getName(), equalTo("max_bucket"));
assertThat(maxBucketValue.value(), equalTo(maxValue));
assertThat(maxBucketValue.keys(), equalTo(maxKeys.toArray(new String[maxKeys.size()])));
}
// Response helpers
protected Multimap<String, FacetValue> processAggregations(Aggregations aggregations) {
Multimap<String, FacetValue> stats = ArrayListMultimap.create();
if (aggregations != null) {
for (Aggregation aggregation : aggregations.asList()) {
if (aggregation instanceof StringTerms) {
for (Terms.Bucket value : ((Terms) aggregation).getBuckets()) {
FacetValue facetValue = new FacetValue(value.getKey(), value.getDocCount());
stats.put(aggregation.getName(), facetValue);
}
} else if (aggregation instanceof InternalValueCount) {
InternalValueCount count = (InternalValueCount) aggregation;
FacetValue facetValue = new FacetValue(count.getName(), count.getValue());
stats.put(count.getName(), facetValue);
}
}
}
return stats;
}
@Override
public void testOrderByEmptyAggregation() throws Exception {
SearchResponse searchResponse =
client()
.prepareSearch("idx")
.setQuery(matchAllQuery())
.addAggregation(
terms("terms")
.field("value")
.order(
Terms.Order.compound(
Terms.Order.aggregation("filter>percentiles.99", true)))
.subAggregation(
filter("filter", termQuery("value", 100))
.subAggregation(
percentiles("percentiles")
.method(PercentilesMethod.TDIGEST)
.field("value"))))
.get();
assertHitCount(searchResponse, 10);
Terms terms = searchResponse.getAggregations().get("terms");
assertThat(terms, notNullValue());
List<Terms.Bucket> buckets = terms.getBuckets();
assertThat(buckets, notNullValue());
assertThat(buckets.size(), equalTo(10));
for (int i = 0; i < 10; i++) {
Terms.Bucket bucket = buckets.get(i);
assertThat(bucket, notNullValue());
assertThat(bucket.getKeyAsNumber(), equalTo((long) i + 1));
assertThat(bucket.getDocCount(), equalTo(1L));
Filter filter = bucket.getAggregations().get("filter");
assertThat(filter, notNullValue());
assertThat(filter.getDocCount(), equalTo(0L));
Percentiles percentiles = filter.getAggregations().get("percentiles");
assertThat(percentiles, notNullValue());
assertThat(percentiles.percentile(99), equalTo(Double.NaN));
}
}
@Test
public void testSubAggregations() throws Exception {
Set expectedAges =
new HashSet<>(ContiguousSet.create(Range.closed(20, 40), DiscreteDomain.integers()));
final String query =
String.format(
"SELECT /*! DOCS_WITH_AGGREGATION(10) */"
+ " * FROM %s/account GROUP BY (gender, age), (state) LIMIT 0,10",
TEST_INDEX);
Map<String, Set<Integer>> buckets = new HashMap<>();
SqlElasticSearchRequestBuilder select = getSearchRequestBuilder(query);
SearchResponse response = (SearchResponse) select.get();
Aggregations result = response.getAggregations();
Terms gender = result.get("gender");
for (Terms.Bucket genderBucket : gender.getBuckets()) {
String genderKey = genderBucket.getKey();
buckets.put(genderKey, new HashSet<Integer>());
Terms ageBuckets = (Terms) genderBucket.getAggregations().get("age");
for (Terms.Bucket ageBucket : ageBuckets.getBuckets()) {
buckets.get(genderKey).add(Integer.parseInt(ageBucket.getKey()));
}
}
Assert.assertEquals(2, buckets.keySet().size());
Assert.assertEquals(expectedAges, buckets.get("m"));
Assert.assertEquals(expectedAges, buckets.get("f"));
Terms state = result.get("state");
for (Terms.Bucket stateBucket : state.getBuckets()) {
if (stateBucket.getKey().equalsIgnoreCase("ak")) {
Assert.assertTrue("There are 22 entries for state ak", stateBucket.getDocCount() == 22);
}
}
Assert.assertEquals(response.getHits().totalHits(), 1000);
Assert.assertEquals(response.getHits().hits().length, 10);
}
@Override
public void testOrderByEmptyAggregation() throws Exception {
SearchResponse searchResponse =
client()
.prepareSearch("idx")
.setQuery(matchAllQuery())
.addAggregation(
terms("terms")
.field("value")
.order(Order.compound(Order.aggregation("filter>stats.avg", true)))
.subAggregation(
filter("filter", termQuery("value", 100))
.subAggregation(stats("stats").field("value"))))
.get();
assertHitCount(searchResponse, 10);
Terms terms = searchResponse.getAggregations().get("terms");
assertThat(terms, notNullValue());
List<Terms.Bucket> buckets = terms.getBuckets();
assertThat(buckets, notNullValue());
assertThat(buckets.size(), equalTo(10));
for (int i = 0; i < 10; i++) {
Terms.Bucket bucket = buckets.get(i);
assertThat(bucket, notNullValue());
assertThat(bucket.getKeyAsNumber(), equalTo((long) i + 1));
assertThat(bucket.getDocCount(), equalTo(1L));
Filter filter = bucket.getAggregations().get("filter");
assertThat(filter, notNullValue());
assertThat(filter.getDocCount(), equalTo(0L));
Stats stats = filter.getAggregations().get("stats");
assertThat(stats, notNullValue());
assertThat(stats.getMin(), equalTo(Double.POSITIVE_INFINITY));
assertThat(stats.getMax(), equalTo(Double.NEGATIVE_INFINITY));
assertThat(stats.getAvg(), equalTo(Double.NaN));
assertThat(stats.getSum(), equalTo(0.0));
assertThat(stats.getCount(), equalTo(0L));
}
}
@Override
public Map<String, Integer> countTermsByField(SearchCriteria searchCriteria, String field) {
ElasticSearchCriteria criteria = (ElasticSearchCriteria) searchCriteria;
if (criteria == null) return null;
SearchRequestBuilder srb = criteria2builder(criteria);
srb.setFrom(0);
srb.setSize(0);
TermsBuilder tb = AggregationBuilders.terms(field);
tb.field(field);
srb.addAggregation(tb);
try {
SearchResponse response = srb.execute().get();
StringTerms aggr = response.getAggregations().get(field);
Map<String, Integer> result = new LinkedHashMap<>();
for (Terms.Bucket bucket : aggr.getBuckets()) {
result.put(bucket.getKey(), (int) bucket.getDocCount());
}
return result;
} catch (Exception e) {
logger.error(e.getMessage(), e);
}
return Collections.emptyMap();
}
@Test
public void singleValueField_WithRegexFiltering_WithFlags() throws Exception {
// include without exclude
// we should be left with: val000, val001, val002, val003, val004, val005, val006, val007,
// val008, val009
// with case insensitive flag on the include regex
SearchResponse response =
client()
.prepareSearch("idx")
.setTypes("high_card_type")
.addAggregation(
terms("terms")
.field(SINGLE_VALUED_FIELD_NAME)
.include("VAL00.+", Pattern.CASE_INSENSITIVE))
.execute()
.actionGet();
assertSearchResponse(response);
Terms terms = response.getAggregations().get("terms");
assertThat(terms, notNullValue());
assertThat(terms.getName(), equalTo("terms"));
assertThat(terms.getBuckets().size(), equalTo(10));
for (int i = 0; i < 10; i++) {
Terms.Bucket bucket = terms.getBucketByKey("val00" + i);
assertThat(bucket, notNullValue());
assertThat(key(bucket), equalTo("val00" + i));
assertThat(bucket.getDocCount(), equalTo(1l));
}
// include and exclude
// we should be left with: val002, val003, val004, val005, val006, val007, val008, val009
// with multi-flag masking on the exclude regex
response =
client()
.prepareSearch("idx")
.setTypes("high_card_type")
.addAggregation(
terms("terms")
.field(SINGLE_VALUED_FIELD_NAME)
.include("val00.+")
.exclude(
"( val000 | VAL001 )#this is a comment",
Pattern.CASE_INSENSITIVE | Pattern.COMMENTS))
.execute()
.actionGet();
assertSearchResponse(response);
terms = response.getAggregations().get("terms");
assertThat(terms, notNullValue());
assertThat(terms.getName(), equalTo("terms"));
assertThat(terms.getBuckets().size(), equalTo(8));
for (int i = 2; i < 10; i++) {
Terms.Bucket bucket = terms.getBucketByKey("val00" + i);
assertThat(bucket, notNullValue());
assertThat(key(bucket), equalTo("val00" + i));
assertThat(bucket.getDocCount(), equalTo(1l));
}
// exclude without include
// we should be left with: val000, val001, val002, val003, val004, val005, val006, val007,
// val008, val009
// with a "no flag" flag
response =
client()
.prepareSearch("idx")
.setTypes("high_card_type")
.addAggregation(
terms("terms").field(SINGLE_VALUED_FIELD_NAME).exclude("val0[1-9]+.+", 0))
.execute()
.actionGet();
assertSearchResponse(response);
terms = response.getAggregations().get("terms");
assertThat(terms, notNullValue());
assertThat(terms.getName(), equalTo("terms"));
assertThat(terms.getBuckets().size(), equalTo(10));
for (int i = 0; i < 10; i++) {
Terms.Bucket bucket = terms.getBucketByKey("val00" + i);
assertThat(bucket, notNullValue());
assertThat(key(bucket), equalTo("val00" + i));
assertThat(bucket.getDocCount(), equalTo(1l));
}
}
public void testNestedSameDocIdProcessedMultipleTime() throws Exception {
assertAcked(
prepareCreate("idx4")
.setSettings(
Settings.builder()
.put(SETTING_NUMBER_OF_SHARDS, 1)
.put(SETTING_NUMBER_OF_REPLICAS, 0))
.addMapping(
"product",
"categories",
"type=text",
"name",
"type=text",
"property",
"type=nested"));
ensureGreen("idx4");
client()
.prepareIndex("idx4", "product", "1")
.setSource(
jsonBuilder()
.startObject()
.field("name", "product1")
.field("categories", "1", "2", "3", "4")
.startArray("property")
.startObject()
.field("id", 1)
.endObject()
.startObject()
.field("id", 2)
.endObject()
.startObject()
.field("id", 3)
.endObject()
.endArray()
.endObject())
.get();
client()
.prepareIndex("idx4", "product", "2")
.setSource(
jsonBuilder()
.startObject()
.field("name", "product2")
.field("categories", "1", "2")
.startArray("property")
.startObject()
.field("id", 1)
.endObject()
.startObject()
.field("id", 5)
.endObject()
.startObject()
.field("id", 4)
.endObject()
.endArray()
.endObject())
.get();
refresh();
SearchResponse response =
client()
.prepareSearch("idx4")
.setTypes("product")
.addAggregation(
terms("category")
.field("categories")
.subAggregation(
nested("property", "property")
.subAggregation(terms("property_id").field("property.id"))))
.get();
assertNoFailures(response);
assertHitCount(response, 2);
Terms category = response.getAggregations().get("category");
assertThat(category.getBuckets().size(), equalTo(4));
Terms.Bucket bucket = category.getBucketByKey("1");
assertThat(bucket.getDocCount(), equalTo(2L));
Nested property = bucket.getAggregations().get("property");
assertThat(property.getDocCount(), equalTo(6L));
Terms propertyId = property.getAggregations().get("property_id");
assertThat(propertyId.getBuckets().size(), equalTo(5));
assertThat(propertyId.getBucketByKey("1").getDocCount(), equalTo(2L));
assertThat(propertyId.getBucketByKey("2").getDocCount(), equalTo(1L));
assertThat(propertyId.getBucketByKey("3").getDocCount(), equalTo(1L));
assertThat(propertyId.getBucketByKey("4").getDocCount(), equalTo(1L));
assertThat(propertyId.getBucketByKey("5").getDocCount(), equalTo(1L));
bucket = category.getBucketByKey("2");
assertThat(bucket.getDocCount(), equalTo(2L));
property = bucket.getAggregations().get("property");
assertThat(property.getDocCount(), equalTo(6L));
propertyId = property.getAggregations().get("property_id");
assertThat(propertyId.getBuckets().size(), equalTo(5));
assertThat(propertyId.getBucketByKey("1").getDocCount(), equalTo(2L));
assertThat(propertyId.getBucketByKey("2").getDocCount(), equalTo(1L));
assertThat(propertyId.getBucketByKey("3").getDocCount(), equalTo(1L));
assertThat(propertyId.getBucketByKey("4").getDocCount(), equalTo(1L));
assertThat(propertyId.getBucketByKey("5").getDocCount(), equalTo(1L));
bucket = category.getBucketByKey("3");
assertThat(bucket.getDocCount(), equalTo(1L));
property = bucket.getAggregations().get("property");
assertThat(property.getDocCount(), equalTo(3L));
propertyId = property.getAggregations().get("property_id");
assertThat(propertyId.getBuckets().size(), equalTo(3));
assertThat(propertyId.getBucketByKey("1").getDocCount(), equalTo(1L));
assertThat(propertyId.getBucketByKey("2").getDocCount(), equalTo(1L));
assertThat(propertyId.getBucketByKey("3").getDocCount(), equalTo(1L));
bucket = category.getBucketByKey("4");
assertThat(bucket.getDocCount(), equalTo(1L));
property = bucket.getAggregations().get("property");
assertThat(property.getDocCount(), equalTo(3L));
propertyId = property.getAggregations().get("property_id");
assertThat(propertyId.getBuckets().size(), equalTo(3));
assertThat(propertyId.getBucketByKey("1").getDocCount(), equalTo(1L));
assertThat(propertyId.getBucketByKey("2").getDocCount(), equalTo(1L));
assertThat(propertyId.getBucketByKey("3").getDocCount(), equalTo(1L));
}
@Test
public void singleValuedField_OrderedBySubAggregationAsc_MultiHierarchyLevels() throws Exception {
boolean asc = randomBoolean();
SearchResponse response =
client()
.prepareSearch("idx")
.setTypes("type")
.addAggregation(
terms("tags")
.executionHint(randomExecutionHint())
.field("tag")
.order(Terms.Order.aggregation("filter1>filter2>stats.max", asc))
.subAggregation(
filter("filter1")
.filter(FilterBuilders.matchAllFilter())
.subAggregation(
filter("filter2")
.filter(FilterBuilders.matchAllFilter())
.subAggregation(stats("stats").field("i")))))
.execute()
.actionGet();
assertSearchResponse(response);
Terms tags = response.getAggregations().get("tags");
assertThat(tags, notNullValue());
assertThat(tags.getName(), equalTo("tags"));
assertThat(tags.getBuckets().size(), equalTo(2));
Iterator<Terms.Bucket> iters = tags.getBuckets().iterator();
// the max for "more" is 2
// the max for "less" is 4
Terms.Bucket tag = iters.next();
assertThat(tag, notNullValue());
assertThat(key(tag), equalTo(asc ? "more" : "less"));
assertThat(tag.getDocCount(), equalTo(asc ? 3l : 2l));
Filter filter1 = tag.getAggregations().get("filter1");
assertThat(filter1, notNullValue());
assertThat(filter1.getDocCount(), equalTo(asc ? 3l : 2l));
Filter filter2 = filter1.getAggregations().get("filter2");
assertThat(filter2, notNullValue());
assertThat(filter2.getDocCount(), equalTo(asc ? 3l : 2l));
Stats stats = filter2.getAggregations().get("stats");
assertThat(stats, notNullValue());
assertThat(stats.getMax(), equalTo(asc ? 2.0 : 4.0));
tag = iters.next();
assertThat(tag, notNullValue());
assertThat(key(tag), equalTo(asc ? "less" : "more"));
assertThat(tag.getDocCount(), equalTo(asc ? 2l : 3l));
filter1 = tag.getAggregations().get("filter1");
assertThat(filter1, notNullValue());
assertThat(filter1.getDocCount(), equalTo(asc ? 2l : 3l));
filter2 = filter1.getAggregations().get("filter2");
assertThat(filter2, notNullValue());
assertThat(filter2.getDocCount(), equalTo(asc ? 2l : 3l));
stats = filter2.getAggregations().get("stats");
assertThat(stats, notNullValue());
assertThat(stats.getMax(), equalTo(asc ? 4.0 : 2.0));
}
// Test based on: https://github.com/elastic/elasticsearch/issues/9280
public void testParentFilterResolvedCorrectly() throws Exception {
XContentBuilder mapping =
jsonBuilder()
.startObject()
.startObject("provider")
.startObject("properties")
.startObject("comments")
.field("type", "nested")
.startObject("properties")
.startObject("cid")
.field("type", "long")
.endObject()
.startObject("identifier")
.field("type", "keyword")
.endObject()
.startObject("tags")
.field("type", "nested")
.startObject("properties")
.startObject("tid")
.field("type", "long")
.endObject()
.startObject("name")
.field("type", "keyword")
.endObject()
.endObject()
.endObject()
.endObject()
.endObject()
.startObject("dates")
.field("type", "object")
.startObject("properties")
.startObject("day")
.field("type", "date")
.field("format", "dateOptionalTime")
.endObject()
.startObject("month")
.field("type", "object")
.startObject("properties")
.startObject("end")
.field("type", "date")
.field("format", "dateOptionalTime")
.endObject()
.startObject("start")
.field("type", "date")
.field("format", "dateOptionalTime")
.endObject()
.startObject("label")
.field("type", "keyword")
.endObject()
.endObject()
.endObject()
.endObject()
.endObject()
.endObject()
.endObject()
.endObject();
assertAcked(
prepareCreate("idx2")
.setSettings(
Settings.builder()
.put(SETTING_NUMBER_OF_SHARDS, 1)
.put(SETTING_NUMBER_OF_REPLICAS, 0))
.addMapping("provider", mapping));
ensureGreen("idx2");
List<IndexRequestBuilder> indexRequests = new ArrayList<>(2);
indexRequests.add(
client()
.prepareIndex("idx2", "provider", "1")
.setSource(
"{\"dates\": {\"month\": {\"label\": \"2014-11\", \"end\": \"2014-11-30\", \"start\": \"2014-11-01\"}, \"day\": \"2014-11-30\"}, \"comments\": [{\"cid\": 3,\"identifier\": \"29111\"}, {\"cid\": 4,\"tags\": [{\"tid\" :44,\"name\": \"Roles\"}], \"identifier\": \"29101\"}]}"));
indexRequests.add(
client()
.prepareIndex("idx2", "provider", "2")
.setSource(
"{\"dates\": {\"month\": {\"label\": \"2014-12\", \"end\": \"2014-12-31\", \"start\": \"2014-12-01\"}, \"day\": \"2014-12-03\"}, \"comments\": [{\"cid\": 1, \"identifier\": \"29111\"}, {\"cid\": 2,\"tags\": [{\"tid\" : 22, \"name\": \"DataChannels\"}], \"identifier\": \"29101\"}]}"));
indexRandom(true, indexRequests);
SearchResponse response =
client()
.prepareSearch("idx2")
.setTypes("provider")
.addAggregation(
terms("startDate")
.field("dates.month.start")
.subAggregation(
terms("endDate")
.field("dates.month.end")
.subAggregation(
terms("period")
.field("dates.month.label")
.subAggregation(
nested("ctxt_idfier_nested", "comments")
.subAggregation(
filter(
"comment_filter",
termQuery("comments.identifier", "29111"))
.subAggregation(
nested("nested_tags", "comments.tags")
.subAggregation(
terms("tag")
.field(
"comments.tags.name"))))))))
.get();
assertNoFailures(response);
assertHitCount(response, 2);
Terms startDate = response.getAggregations().get("startDate");
assertThat(startDate.getBuckets().size(), equalTo(2));
Terms.Bucket bucket = startDate.getBucketByKey("2014-11-01T00:00:00.000Z");
assertThat(bucket.getDocCount(), equalTo(1L));
Terms endDate = bucket.getAggregations().get("endDate");
bucket = endDate.getBucketByKey("2014-11-30T00:00:00.000Z");
assertThat(bucket.getDocCount(), equalTo(1L));
Terms period = bucket.getAggregations().get("period");
bucket = period.getBucketByKey("2014-11");
assertThat(bucket.getDocCount(), equalTo(1L));
Nested comments = bucket.getAggregations().get("ctxt_idfier_nested");
assertThat(comments.getDocCount(), equalTo(2L));
Filter filter = comments.getAggregations().get("comment_filter");
assertThat(filter.getDocCount(), equalTo(1L));
Nested nestedTags = filter.getAggregations().get("nested_tags");
assertThat(nestedTags.getDocCount(), equalTo(0L)); // This must be 0
Terms tags = nestedTags.getAggregations().get("tag");
assertThat(tags.getBuckets().size(), equalTo(0)); // and this must be empty
bucket = startDate.getBucketByKey("2014-12-01T00:00:00.000Z");
assertThat(bucket.getDocCount(), equalTo(1L));
endDate = bucket.getAggregations().get("endDate");
bucket = endDate.getBucketByKey("2014-12-31T00:00:00.000Z");
assertThat(bucket.getDocCount(), equalTo(1L));
period = bucket.getAggregations().get("period");
bucket = period.getBucketByKey("2014-12");
assertThat(bucket.getDocCount(), equalTo(1L));
comments = bucket.getAggregations().get("ctxt_idfier_nested");
assertThat(comments.getDocCount(), equalTo(2L));
filter = comments.getAggregations().get("comment_filter");
assertThat(filter.getDocCount(), equalTo(1L));
nestedTags = filter.getAggregations().get("nested_tags");
assertThat(nestedTags.getDocCount(), equalTo(0L)); // This must be 0
tags = nestedTags.getAggregations().get("tag");
assertThat(tags.getBuckets().size(), equalTo(0)); // and this must be empty
}
@Test
public void singleValueField_WithRegexFiltering() throws Exception {
// include without exclude
// we should be left with: val000, val001, val002, val003, val004, val005, val006, val007,
// val008, val009
SearchResponse response =
client()
.prepareSearch("idx")
.setTypes("high_card_type")
.addAggregation(terms("terms").field(SINGLE_VALUED_FIELD_NAME).include("val00.+"))
.execute()
.actionGet();
assertSearchResponse(response);
Terms terms = response.getAggregations().get("terms");
assertThat(terms, notNullValue());
assertThat(terms.getName(), equalTo("terms"));
assertThat(terms.getBuckets().size(), equalTo(10));
for (int i = 0; i < 10; i++) {
Terms.Bucket bucket = terms.getBucketByKey("val00" + i);
assertThat(bucket, notNullValue());
assertThat(key(bucket), equalTo("val00" + i));
assertThat(bucket.getDocCount(), equalTo(1l));
}
// include and exclude
// we should be left with: val002, val003, val004, val005, val006, val007, val008, val009
response =
client()
.prepareSearch("idx")
.setTypes("high_card_type")
.addAggregation(
terms("terms")
.field(SINGLE_VALUED_FIELD_NAME)
.include("val00.+")
.exclude("(val000|val001)"))
.execute()
.actionGet();
assertSearchResponse(response);
terms = response.getAggregations().get("terms");
assertThat(terms, notNullValue());
assertThat(terms.getName(), equalTo("terms"));
assertThat(terms.getBuckets().size(), equalTo(8));
for (int i = 2; i < 10; i++) {
Terms.Bucket bucket = terms.getBucketByKey("val00" + i);
assertThat(bucket, notNullValue());
assertThat(key(bucket), equalTo("val00" + i));
assertThat(bucket.getDocCount(), equalTo(1l));
}
// exclude without include
// we should be left with: val000, val001, val002, val003, val004, val005, val006, val007,
// val008, val009
response =
client()
.prepareSearch("idx")
.setTypes("high_card_type")
.addAggregation(terms("terms").field(SINGLE_VALUED_FIELD_NAME).exclude("val0[1-9]+.+"))
.execute()
.actionGet();
assertSearchResponse(response);
terms = response.getAggregations().get("terms");
assertThat(terms, notNullValue());
assertThat(terms.getName(), equalTo("terms"));
assertThat(terms.getBuckets().size(), equalTo(10));
for (int i = 0; i < 10; i++) {
Terms.Bucket bucket = terms.getBucketByKey("val00" + i);
assertThat(bucket, notNullValue());
assertThat(key(bucket), equalTo("val00" + i));
assertThat(bucket.getDocCount(), equalTo(1l));
}
}