Elasticsearch | es_分组-分页-TransportClient实现

说在前面：

Elasticsearch Java API 有四种实现方式：分别是 TransportClient、RestClient、Jest、Spring Data Elasticsearch。
本文使用第一种方式，也就是 TransportClient 的方式进行实现。想要了解其他三种的方式可以看一下这篇文章：https://blog.youkuaiyun.com/qq_33314107/article/details/80725913

总结：

1、group 之后不能自动分页，需要手动设置；

2、size 需要指定，否则会出错。

---

group 之后分页、排序实现

需要手动截取分页对应范围内的数据。

比如：这里

倒序，获取到数据集的第 (currentPage-1) * limit 到 currentPage * limit 条记录；

升序，获取到数据集的 第 buckets.size() - (currentPage - 1) * limit 到 buckets.size() - currentPage * limit 条记录。

// 获取到 response 之后
Aggregations aggregations = response.getAggregations();
Terms carids =  aggregations.get("group_car_bayId");
List<? extends Terms.Bucket> buckets =  carids.getBuckets();
List<carBean> listCarTgs = new ArrayList<>();
carBean carBean ;
// buckets 全部数据，分页就是取固定位置的 limit 条数据
// 默认按照统计之后的数量倒序， 如果要正序，则第一页从最后一条开始取
if(buckets.size()>0) {
    int i=0;
    if("desc".equalsIgnoreCase(order)) {// 倒序
        for(Terms.Bucket bucket : buckets){
            if(i++<(currentPage-1) * limit){
                continue;
            }
            if (i > currentPage * limit) {
                break;
            }
            carBean = new carBean();
            carBean.setPassTimes((int)bucket.getDocCount());
            carBean.setBayId(bucket.getKeyAsString().split("_")[2]);
            carBean.setPlateNumber(bucket.getKeyAsString().split("_")[0]);
            carBean.setPlateType(bucket.getKeyAsString().split("_")[1]);
            listCarTgs.add(carBean);
        }
    }else if("asc".equalsIgnoreCase(order)) {// 升序
        for(i = buckets.size() - 1; i >= 0; i--){
            if(i < buckets.size() - currentPage * limit){
                break;
            }
            if(i > buckets.size() - (currentPage - 1) * limit) continue;;
            carBean = new carBean();
            carBean.setPassTimes((int)buckets.get(i).getDocCount());
            carBean.setBayId(buckets.get(i).getKeyAsString().split("_")[2]);
            carBean.setPlateNumber(buckets.get(i).getKeyAsString().split("_")[0]);
            carBean.setPlateType(buckets.get(i).getKeyAsString().split("_")[1]);
            listCarTgs.add(carBean);
        }
    }
}

单个 group

注意：需要设置 size

TermsAggregationBuilder tb=  AggregationBuilders.terms("group_bayId").field("bay_id").size(Integer.MAX_VALUE);

多个 group

以脚本的形式

TermsAggregationBuilder tb=  AggregationBuilders.terms("group_carId").script(new Script("doc['plateNumber'].value+'_'+doc['plateType'].value"));
            tb.subAggregation(AggregationBuilders.topHits("max_time").sort("reportTime", SortOrder.DESC).size(1));

再比如：三个 group

BoolQueryBuilder filter = QueryBuilders.boolQuery();
if (carList != null && carList.size() >0) {
    filter.must(QueryBuilders.termsQuery("car_plate_number", carList.stream().map(SimpleCar:: getPlateNumber).collect(Collectors.toList())));
}
if (startTime != null && endTime != null) {
    filter.filter(QueryBuilders.rangeQuery("timestamp").gt(startTime.getTime()).lt(endTime.getTime()));
} else if (startTime != null) {
    filter.filter(QueryBuilders.rangeQuery("timestamp").gt(startTime.getTime()));
} else if (endTime != null) {
    filter.filter(QueryBuilders.rangeQuery("timestamp").lt(endTime.getTime()));
}
FieldSortBuilder sort = SortBuilders.fieldSort("transit_times").order("asc".equalsIgnoreCase(order)?SortOrder.ASC:SortOrder.DESC);
TermsAggregationBuilder termsAggregationBuilder =  AggregationBuilders.terms("group_car_bayId")
        .script(new Script("doc['car_plate_number'].value+'_'+doc['car_plate_type'].value + '_' +doc['bay_id'].value")).size(Integer.MAX_VALUE);
SearchResponse response = search(filter, sort, termsAggregationBuilder, elasticsearchProperties.getTgsIndex(), elasticsearchProperties.getTgsType(),
        (currentPage-1) * limit, 0);
Aggregations aggregations = response.getAggregations();

例子2：

TermsAggregationBuilder tb=  AggregationBuilders.terms("group_bayId").field("bay_id").size(Integer.MAX_VALUE);
    tb.order(BucketOrder.count(false));