今天给大家转载一篇关于ES聚合相关的文章,是利用Java API实现的。因为公司最近要上搜索引擎相关的功能,所以最近一直在学习es相关的内容。基本内容有:组合查询、聚合、分页、权重设置、数据同步方案、索引创建规则方案、分词、分片规则、批量操作等,后期会陆陆续续和大家介绍ES这些相关的内容的。
转载自:https://blog.csdn.net/carlislelee/article/details/52598022
以球员信息为例,player索引的player type包含5个字段,姓名,年龄,薪水,球队,场上位置。
index的mapping为:
  1. "mappings": {
  2. "quote": {
  3. "properties": {
  4. "adj_close": {
  5. "type": "long"
  6. },
  7. "open": {
  8. "type": "long"
  9. },
  10. "symbol": {
  11. "index": "not_analyzed",
  12. "type": "string"
  13. },
  14. "volume": {
  15. "type": "long"
  16. },
  17. "high": {
  18. "type": "long"
  19. },
  20. "low": {
  21. "type": "long"
  22. },
  23. "date": {
  24. "format": "strict_date_optional_time||epoch_millis",
  25. "type": "date"
  26. },
  27. "close": {
  28. "type": "long"
  29. }
  30. },
  31. "_all": {
  32. "enabled": false
  33. }
  34. }
  35. }

索引中的全部数据:
nameagesalaryteamposition
james333000cavsf
irving252000cavpg
curry291000warpg
thompson262000warsg
green262000warpf
garnett401000timpf
towns21500timc
lavin21300timsg
wigins20500timsf
首先,初始化Builder:
SearchRequestBuilder sbuilder = client.prepareSearch("player").setTypes("player");
接下来举例说明各种聚合操作的实现方法,因为在es的api中,多字段上的聚合操作需要用到子聚合(subAggregation),初学者可能找不到方法(网上资料比较少,笔者在这个问题上折腾了两天,最后度了源码才彻底搞清楚T_T),后边会特意说明多字段聚合的实现方法。另外,聚合后的排序也会单独说明。

1. group by/count

例如要计算每个球队的球员数,如果使用SQL语句,应表达如下:
select team, count(*) as player_count from player group by team;
ES的java api:
  1. TermsBuilder teamAgg= AggregationBuilders.terms("player_count ").field("team");
  2. sbuilder.addAggregation(teamAgg);
  3. SearchResponse response = sbuilder.execute().actionGet();

2.group by多个field

例如要计算每个球队每个位置的球员数,如果使用SQL语句,应表达如下:
select team, position, count(*) as pos_count from player group by team, position;
ES的java api:
  1. TermsBuilder teamAgg= AggregationBuilders.terms("player_count ").field("team");
  2. TermsBuilder posAgg= AggregationBuilders.terms("pos_count").field("position");
  3. sbuilder.addAggregation(teamAgg.subAggregation(posAgg));
  4. SearchResponse response = sbuilder.execute().actionGet();

3.max/min/sum/avg

例如要计算每个球队年龄最大/最小/总/平均的球员年龄,如果使用SQL语句,应表达如下:
select team, max(age) as max_age from player group by team;
ES的java api:
  1. TermsBuilder teamAgg= AggregationBuilders.terms("player_count ").field("team");
  2. MaxBuilder ageAgg= AggregationBuilders.max("max_age").field("age");
  3. sbuilder.addAggregation(teamAgg.subAggregation(ageAgg));
  4. SearchResponse response = sbuilder.execute().actionGet();

4.对多个field求max/min/sum/avg

例如要计算每个球队球员的平均年龄,同时又要计算总年薪,如果使用SQL语句,应表达如下:
select team, avg(age)as avg_age, sum(salary) as total_salary from player group by team;
ES的java api:
  1. TermsBuilder teamAgg= AggregationBuilders.terms("team");
  2. AvgBuilder ageAgg= AggregationBuilders.avg("avg_age").field("age");
  3. SumBuilder salaryAgg= AggregationBuilders.avg("total_salary ").field("salary");
  4. sbuilder.addAggregation(teamAgg.subAggregation(ageAgg).subAggregation(salaryAgg));
  5. SearchResponse response = sbuilder.execute().actionGet();

5.聚合后对Aggregation结果排序

例如要计算每个球队总年薪,并按照总年薪倒序排列,如果使用SQL语句,应表达如下:
select team, sum(salary) as total_salary from player group by team order by total_salary desc;
ES的java api:
  1. TermsBuilder teamAgg= AggregationBuilders.terms("team").order(Order.aggregation("total_salary ", false);
  2. SumBuilder salaryAgg= AggregationBuilders.avg("total_salary ").field("salary");
  3. sbuilder.addAggregation(teamAgg.subAggregation(salaryAgg));
  4. SearchResponse response = sbuilder.execute().actionGet();

需要特别注意的是,排序是在TermAggregation处执行的,Order.aggregation函数的第一个参数是aggregation的名字,第二个参数是boolean型,true表示正序,false表示倒序。

6.Aggregation结果条数的问题

默认情况下,search执行后,仅返回10条聚合结果,如果想反悔更多的结果,需要在构建TermsBuilder 时指定size:
TermsBuilder teamAgg= AggregationBuilders.terms("team").size(15);

7.Aggregation结果的解析/输出

得到response后:
  1. <span style="white-space:pre"> </span>Map<String, Aggregation> aggMap = response.getAggregations().asMap();
  2. StringTerms teamAgg= (StringTerms) aggMap.get("keywordAgg");
  3. Iterator<Bucket> teamBucketIt = teamAgg.getBuckets().iterator();
  4. while (teamBucketIt .hasNext()) {
  5. Bucket buck = teamBucketIt .next();
  6. //球队名
  7. String team = buck.getKey();
  8. //记录数
  9. long count = buck.getDocCount();
  10. //得到所有子聚合
  11. Map subaggmap = buck.getAggregations().asMap();
  12. //avg值获取方法
  13. double avg_age= ((InternalAvg) subaggmap.get("avg_age")).getValue();
  14. //sum值获取方法
  15. double total_salary = ((InternalSum) subaggmap.get("total_salary")).getValue();
  16. //...
  17. //max/min以此类推
  18. }

8. 总结

综上,聚合操作主要是调用了SearchRequestBuilder的addAggregation方法,通常是传入一个TermsBuilder,子聚合调用TermsBuilder的subAggregation方法,可以添加的子聚合有TermsBuilder、SumBuilder、AvgBuilder、MaxBuilder、MinBuilder等常见的聚合操作。

从实现上来讲,SearchRequestBuilder在内部保持了一个私有的 SearchSourceBuilder实例, SearchSourceBuilder内部包含一个List<AbstractAggregationBuilder>,每次调用addAggregation时会调用 SearchSourceBuilder实例,添加一个AggregationBuilder。
同样的,TermsBuilder也在内部保持了一个List<AbstractAggregationBuilder>,调用addAggregation方法(来自父类addAggregation)时会添加一个AggregationBuilder。有兴趣的读者也可以阅读源码的实现。

如果有任何问题,欢迎一起讨论,如果文中有什么错误,欢迎批评指正。

注:文中使用的Elastic Search API版本为2.3.2

想要更多干货、技术猛料的孩子,快点拿起手机扫码关注我,我在这里等你哦~



本文转载:CSDN博客