ES聚合之Pipeline聚合语法讲解

我的身前一尺是我的世界

已于 2022-09-12 16:08:57 修改

阅读量1.5k

点赞数 1

分类专栏： Elasticsearch 文章标签： es嵌套分组 es嵌套分桶

于 2022-09-12 16:08:28 首次发布

本文链接：https://blog.csdn.net/qq_39706570/article/details/126735610

版权

Elasticsearch 专栏收录该内容

16 篇文章 4 订阅

订阅专栏

目标

掌握Pipeline aggregations（管道聚合，相当于MySQL中的having ）的基本语法。它比另外两种类型的聚合用得较少，这里只举了少量案例供大家参考分析。Pipeline aggregations分为两类：

Parent（父级）：结果内嵌到现有的聚合结果中。
Sibling（兄弟级）：结果和现有的聚合结果同级。

ES版本信息

7.17.5

官方文档

Pipeline aggregationshttps://www.elastic.co/guide/en/elasticsearch/reference/7.17/search-aggregations-pipeline.html

实战

新增测试数据

#图书馆数据库
PUT /library_db
{
  "settings": {
    "index": {
      "analysis.analyzer.default.type": "ik_max_word"
    }
  }
}

PUT /library_db/_bulk
{"index":{"_id":"1"}}
{"id":1,"type":"玄幻","name":"诛仙","words_num":120,"chapter_num":600,"completion_time":"2000-09-01","author":"萧鼎","prices":32.12}
{"index":{"_id":"2"}}
{"id":2,"type":"玄幻","name":"诛仙前传：蛮荒行","words_num":30,"chapter_num":67,"completion_time":"2020-09-01","author":"萧鼎","prices":23.12}
{"index":{"_id":"3"}}
{"id":3,"type":"武侠","name":"天龙八部","words_num":80,"chapter_num":120,"completion_time":"1995-09-01","author":"金庸","prices":52.1}
{"index":{"_id":"4"}}
{"id":4,"type":"武侠","name":"射雕英雄传","words_num":67,"chapter_num":95,"completion_time":"1998-01-01","author":"金庸","prices":4.12}
{"index":{"_id":"5"}}
{"id":5,"type":"武侠","name":"神雕侠侣","words_num":75,"chapter_num":76,"completion_time":"2000-01-01","author":"金庸","prices":32.8}
{"index":{"_id":"6"}}
{"id":5,"type":"武侠","name":"倚天屠龙记","words_num":83,"chapter_num":130,"completion_time":"2003-01-01","author":"金庸","prices":100.12}
{"index":{"_id":"7"}}
{"id":7,"type":"玄幻","name":"凡人修仙传","words_num":600,"chapter_num":3000,"completion_time":"2018-01-01","author":"忘语","prices":120.12}
{"index":{"_id":"8"}}
{"id":8,"type":"玄幻","name":"魔天记","words_num":159,"chapter_num":400,"completion_time":"2019-01-01","author":"忘语","prices":11.12}
{"index":{"_id":"9"}}
{"id":9,"type":"都市异能","name":"黄金瞳","words_num":220,"chapter_num":400,"completion_time":"2019-01-01","author":"打眼","prices":74.5}
{"index":{"_id":"10"}}
{"id":10,"type":"玄幻","name":"将夜","words_num":210,"chapter_num":600,"completion_time":"2014-01-01","author":"血红","prices":32.0}
{"index":{"_id":"11"}}
{"id":11,"type":"军事","name":"亮剑","words_num":120,"chapter_num":100,"completion_time":"2012-01-01","author":"都梁","prices":15.0}

基本语法

需求一：求每个作者下面字数最多的小说的字数，按照作品数量降序排序。

分析：第一步求出每个作者的小说数量，按照降序排序；第二步求出每个作者下字数最多的小说的字数。

POST /library_db/_search
{
  "size": 0,
  "aggs": {
    "author_group": {
      "terms": {
        "field": "author.keyword",
        "size": 10,
        "order": {
          "_key": "desc"
        }
      },
      "aggs": {
        "max_words_num": {
          "max": {
            "field": "words_num"
          }
        }
      }
    }
  }
}

需求二：求小说的每个类型的平均字数，求出最大的平均字数。

POST /library_db/_search
{
  "size": 0,
  "aggs": {
    "author_group": {
      "terms": {
        "field": "type.keyword",
        "size": 10,
        "order": {
          "_key": "desc"
        }
      },
      "aggs": {
        "avg_words_num": {
          "avg": {
            "field": "words_num"
          }
        }
      }
    },
    "max_avg_words_num":{
      "max_bucket": {
        "buckets_path": "author_group>avg_words_num"
      }
    }
  }
}

百分数分桶聚合

需求：求小说的每个类型的平均字数，求出平均字数的百分位。

POST /library_db/_search
{
  "size": 0,
  "aggs": {
    "author_group": {
      "terms": {
        "field": "type.keyword",
        "size": 10,
        "order": {
          "_key": "desc"
        }
      },
      "aggs": {
        "avg_words_num": {
          "avg": {
            "field": "words_num"
          }
        }
      }
    },
    "percentiles_avg_words_num":{
      "percentiles_bucket": {
        "buckets_path": "author_group>avg_words_num",
        "percents": [ 25.0, 50.0, 75.0,100 ]  
      }
    }
  }
}