java Elasticsearch 進行嵌套子聚合

 

聚合子查詢：

TermsAggregationBuilder aggregation = AggregationBuilders.terms("dt_id").field("dt_ids").size(30000);
        String agg_area_field = "city_code";
        TermsAggregationBuilder area_field_builder = AggregationBuilders.terms("area_field").field(agg_area_field).size(30000);
        area_field_builder.subAggregation(aggregation);
        SearchResponse response = client.prepareSearch(index_name).setTypes("lw_devices")
                .setQuery(boolQuery)
                .addAggregation(area_field_builder)
                .execute()
                .actionGet();

 

以上demo的大致的意思就是，先對city_code字段進行聚合，然后對聚合結果，再用dt_ids字段進行嵌套聚合(子聚合) ， 相當於sql的兩個group by , 一個聚合嵌套於另一個聚合之內，可以用subAggregation方法進行關聯。

 

然后就是對聚合結果的遍歷：

 

        Terms terms = response.getAggregations().get("area_field");
        List<Terms.Bucket> buckets = terms.getBuckets();for (Terms.Bucket bucket : buckets) {
            //地區
            String code = (String) bucket.getKey();
            System.out.println(code);
            Aggregations aggregations = bucket.getAggregations();
            Terms dt_id = aggregations.get("dt_id");
            List<Terms.Bucket> buckets1 = dt_id.getBuckets();
            for (Terms.Bucket bucket1 : buckets1) {
                System.out.println(bucket1.getKey() + ":" + bucket1.getDocCount());
            }
        }

 

思路就是，先根據response拿到aggregation,然后根據標識符拿到對應的，聚合類型，然后得到bucket的集合。然后再通過bucket拿到aggregation， 然后以此循環往下提取聚合結果。