elasticsearch7.8父子结构设计以及mapping映射

elasticsearch7.8父子结构设计以及mapping映射

设计父子结构

建立index1索引，父：parent，子：child
父包含id, parentName字段（数据：父亲1，父亲2）
子包含id, childName, parentId字段（数据：父亲1有一儿一女，父亲2有一儿）
relation表示关系，名称自定义，type:join 表示父子结构

PUT index1
PUT index1/_mapping
{
  "properties": {
    "parent": {
      "properties": {
        "id": {
          "type": "integer"
        },
        "parentName": {
          "type": "text"
        }
      }
    },
    "child": {
      "properties": {
        "id": {
          "type": "integer"
        },
        "childName": {
          "type": "text"
        },
        "parentId": {
          "type": "integer"
        }
      }
    },
    "relation": {
      "type": "join",
      "relations": {
        "parent": [
          "child"
        ]
      }
    }
  }
}

建立父 parent 数据

POST index1/_doc/1
{
  "parent.id":1,
  "relation":"parent",
  "parent.parentName":"父亲1"
}
POST index1/_doc/2
{
  "parent.id":2,
  "relation":"parent",
  "parent.parentName":"父亲2"
}

建立父亲1的子数据

POST index1/_doc/101?routing=1
{ 
  "child.id":101,
  "relation":
  {
    "name":"child",
    "parent":1
  },
  "child.childName":"父亲1的儿子",
  "child.parentId":1
}
POST index1/_doc/102?routing=1
{
  "child.id":101,
  "relation":
  {
    "name":"child",
    "parent":1
  },
  "child.childName":"父亲1的女儿",
  "child.parentId":1
}

建立父亲2的子数据

POST index1/_doc/201?routing=2
{
  "child.id":201,
  "relation":
  {
    "name":"child",
    "parent":2
  },
  "child.childName":"父亲2的儿子",
  "child.parentId":2
}

测试

#以子查父
GET index1/_search
{
  "query": {
    "has_child": {
      "type": "child",
      "query": {
        "term": {
          "child.parentId": "1"
        }
      }
    }
  }
}

#带inner_hits 以子查父，查出父子级联信息
GET index1/_search
{
  "query": {
    "has_child": {
      "type": "child",
      "query": {
        "term": {
          "child.parentId": 1
        }
      },
      "inner_hits": {}
    }
  }
}

#以父查子
GET index1/_search
{
  "query": {
    "has_parent": {
      "parent_type": "parent",
      "query": {
        "term": {
          "parent.id": "1"
        }
        
        }
      }
    }  
}

查询结果：

使用RestHighLevelClient代码查询

 // 带子级查询
    @Test
    public void test1() throws IOException{
        SearchRequest searchRequest = new SearchRequest();
        searchRequest.indices("index1");
        HasChildQueryBuilder qb = JoinQueryBuilders.hasChildQuery(
                "child",                         //要查询的子类型
                QueryBuilders.termQuery("child.parentId","1"),
                ScoreMode.None
        ).innerHit(new InnerHitBuilder());
        SearchSourceBuilder searchSourceBuilder = new SearchSourceBuilder().query(qb);
        searchRequest.source(searchSourceBuilder);
        SearchResponse response = client.search(searchRequest, RequestOptions.DEFAULT);
        response.getHits().forEach(message->{
            System.out.println(message.getSourceAsString());
            message.getInnerHits().get("child").forEach(e ->{
                System.out.println(e.getSourceAsString());
            });
        });
    }

要注意的细节挺多，父子数据要在一个分配，索引建立子数据，必须设定routing为父id
注意父子结构容易出现的问题是，添加数据时，要注意带上 parent.id 这种关系，否则不会用我们自定义的映射mapping关系，es会自动匹配字段类型，添加完数据一定要 GET index1/_mapping查看结构是否变化了。