使用场景
使用es聚合时,有时还需要获取query(或filter) 的相关文档结果(数据)。
比如统计各个地区编码的营业额,得到了聚合的统计结果,还想知道query结果中对应的地区名称,并根据营业额进行排序,
这时可以使用 top_hits。
top_hits属性
top_hits有以下的属性:
from - 从第几个结果开始获取。size - 每个桶返回的query结果的数量。默认情况下,返回前三个匹配的结果。sort - 根据字段进行排序。默认情况下,按主查询的分数排序。
top_hits的DSL
格式如下:
{"size" : 0,"query" : { },"aggregations" : {"自己命名的聚合名称" : {"terms" : {"field" : "聚合字段","size" : 10000,"order" : {"_term" : "asc"}},"aggregations" : {"hits" : {"top_hits" : {"sort": [{"排序字段": {"order": "desc"}}],"from" : 0,"size" : 5}},"自己命名的聚合统计的名称" : {"sum" : {"field" : "聚合统计字段"}}}}}
}
示例如下:
{"size" : 0,"query" : { },"aggregations" : {"agg_area" : {"terms" : {"field" : "area","size" : 10000,"order" : {"_term" : "asc"}},"aggregations" : {"hits" : {"top_hits" : {"sort": [{"amount": {"order": "desc"}}],"from" : 0,"size" : 5}},"area_sum" : {"sum" : {"field" : "amount"}}}}}
}
top_hits的java代码
java代码格式:
public static String getTopHitsDSL() {SearchSourceBuilder searchSourceBuilder = SearchSourceBuilder.searchSource();AggregationBuilder areaCodeAgg = AggregationBuilders.terms(自己命名的聚合名称).field(聚合字段).order(Terms.Order.aggregation("_term", true)).size(10000).subAggregation(AggregationBuilders.topHits("hits").sort(排序字段).size(5)).subAggregation(AggregationBuilders.sum(自己命名的聚合统计的名称).field(聚合字段));return searchSourceBuilder.query().aggregation(areaCodeAgg).size(0).toString();}
如下所示:
public static String getTopHitsDSL() {SearchSourceBuilder searchSourceBuilder = SearchSourceBuilder.searchSource();AggregationBuilder areaCodeAgg = AggregationBuilders.terms("agg_area").field("area").order(Terms.Order.aggregation("_term", true)).size(10000).subAggregation(AggregationBuilders.topHits("hits").sort("amount").size(5)).subAggregation(AggregationBuilders.sum("area_sum").field("amount"));return searchSourceBuilder.query().aggregation(areaCodeAgg).size(0).toString();}
参考资料:
http://itindex.net/detail/60468-elasticsearch-top-hits