对于平时的group by 我们都知道使用 AggregationBuilder 可以解决

AggregationBuilder aggregationBuilder = 
AggregationBuilders.terms("nameAgg").field("name.keyword").size(Integer.MAX_VALUE) //1
     .subAggregation(AggregationBuilders.terms("jobAgg").field("job.keyword").size(Integer.MAX_VALUE) //2
          .subAggregation(AggregationBuilders.avg("ageAgg").field("age")) //3
               .subAggregation(AggregationBuilders.count("totalNum").field("name.keyword"))); //4
 
searchSourceBuilder.aggregation(aggregationBuilder);
以上代码的解释如下

    (1) 首先按照name分组,terms括号里面是聚合名字,随便起,field为聚合的字段名;之所以加了.keyword是因为不加聚合的时候会报fielddata属性没有设置为true;

{
    "error": {
        "root_cause": [{
            "type": "illegal_argument_exception",
            "reason": "Fielddata is disabled on text fields by     default.Set fielddata = true on[name] in order to load fielddata in memory by uninverting the    inverted index.Note that this can however use significant memory.Alternatively use a keyword field    instead."
        }],
        "type ": "search_phase_execution_exception ",
        "reason ": "    all shards failed ",
        "phase ": "    query ",
        "grouped": true,
        "failed_shards": [{
                    "shard": 0,
                    "index": "school",
                    "node": "H7VIRoOwS8mws78T-0Ce-Q",
                    "reason": {
                        "type": "illegal_argument_exception",
                        "reason": "Fielddata is disabled on text fields by default. Set 
                        fielddata = true on[name] in order to load fielddata in memory by uninverting the inverted index.Note that
                        this can however use significant memory.Alternatively use a keyword field instead.
                        "}}]},"
                        status ":400}

因为映射模板将string类型的字段存进elasticsearch时,一个字符串字段有两个类型,一个text类型,分词类型;一个keyword类型,不分词类型;所以加上.keyword就可以正常聚合了,对于es2.x版本有可能不分词的类型为.raw;注意甄别。后面size参数默认为10,貌似是最多聚合10个,我肯定想要聚合全部数据,就填最大值

    (2) 紧接着在name分组的基础上按job分组,属于nameAgg聚合的子聚合,后面的都属于前面的子聚合

    (3) 分组完,紧接着统计各组平均年龄,由于年龄属于long类型,不用加.keyword,从这里以后都要注意括号的位置,.subAggregation跟在谁的后面一定要搞清楚,搞混淆结果会不一样

    (4) 实际上这一步不需要,因为elasticsearch在分组聚合完自动会计算当前分组下有多少doc_count

但是有时候会涉及到一些子查询(嵌套查询)

SumBucketPipelineAggregationBuilder pipelineAggregationBuilder = 
     PipelineAggregatorBuilders.sumBucket("countTotalNum","jobAgg>totalNum");//第二个参数为聚合路径
 
//如果有需要管道聚合,可以在上面的分组上继续.subAggregation(pipelineAggregationBuilder);

接下来开始将查询和聚合条件放入search中:

String query = searchSourceBuilder.toString();
 
Search search = new Search.Builder(query).addIndex("school").addType("student").build(); 
 
SearchResult result = client.execute(search); 
  接下来就是取结果,最主要是聚合怎么取

//首先取最外层的聚合,拿到桶
List<TermsAggregation.Entry> nameAgg =
        result.getAggregations().getTermsAggregation("nameAgg").getBuckets();
//循环每一个桶,拿到里面的聚合,再拿桶
for (TermsAggregation.Entry entry : nameAgg) {
    List<TermsAggregation.Entry> jobAgg = entry.getTermsAggregation("jobAgg").getBuckets();
    //循环每一个桶,拿到里面的聚合,再拿桶
    for (TermsAggregation.Entry jobEntry : jobAgg) {
        //取到每个分组里的平均年龄
        long avgAge = jobEntry.getAvgAggregation("ageAgg").getAvg(); 
 
        //其实这里已经能获取doc_count了,所以聚合计算总数那一步可以省略
        long count = jobEntry.getCount();
        ........
        //其他操作
        ........
    }
    ..........
    //其他操作
    ..........

Logo

为开发者提供学习成长、分享交流、生态实践、资源工具等服务,帮助开发者快速成长。

更多推荐