自定义排序 RankedQuery

RankedQuery 在原始的查询结果集上增加了二次自定义排位功能，可用于 BadCase 的快速处理或垂搜场景的结果集处理。

示例

准备一些测试数据，例如从1到10的序列

POST _bulk
{ "index" : { "_index" : "greeting","_type":"_doc", "_id" : "1" } }
{ "number" : 1 }
{ "index" : { "_index" : "greeting","_type":"_doc", "_id" : "2" } }
{ "number" : 2 }
{ "index" : { "_index" : "greeting","_type":"_doc", "_id" : "3" } }
{ "number" : 3 }
{ "index" : { "_index" : "greeting","_type":"_doc", "_id" : "4" } }
{ "number" : 4 }
{ "index" : { "_index" : "greeting","_type":"_doc", "_id" : "5" } }
{ "number" : 5 }
{ "index" : { "_index" : "greeting","_type":"_doc", "_id" : "6" } }
{ "number" : 6 }
{ "index" : { "_index" : "greeting","_type":"_doc", "_id" : "7" } }
{ "number" : 7 }
{ "index" : { "_index" : "greeting","_type":"_doc", "_id" : "8" } }
{ "number" : 8 }
{ "index" : { "_index" : "greeting","_type":"_doc", "_id" : "9" } }
{ "number" : 9 }
{ "index" : { "_index" : "greeting","_type":"_doc", "_id" : "10" } }
{ "number" : 10 }

执行简单查询

GET greeting/_search
{
  "query": {
    "match_all": {}
  }
}

结果集默认按_id排序

 "hits" : [
      {
        "_source" : {"number" : 1}
      },
      {
        "_source" : {"number" : 2}
      },
      {
        "_source" : {"number" : 3}
      },
      ...

1. 文档置顶

将 _id 为 7 和 8 的文档置顶

GET greeting/_search
{
  "query": {
    "ranked": {
      "query": {  
        "match_all": {}
      },
      "rank": {
        "top": ["7","8"]
      }
    }
  }
}

结果集如下

 "hits" : [
      {
        "_source" : {"number" : 7}
      },
      {
        "_source" : {"number" : 8}
      },
      {
        "_source" : {"number" : 1}
      },
      {
        "_source" : {"number" : 2}
      },      
      ...

2. 移除文档

将_id 为 1 和 2 的文档移除

GET greeting/_search
{
  "query": {
    "ranked": {
      "query": {  
        "match_all": {}
      },
      "rank": {
        "block" : [ "1", "2" ]
      }
    }
  }
}

返回

 "hits" : [
      {
        "_source" : {"number" : 3}
      },
      {
        "_source" : {"number" : 4}
      },
      {
        "_source" : {"number" : 5}
      },
      ...

3. 指定文档位置

将 _id 为 1 的文档排在第 3 位 ，将 _id 为 9 的文档排在第 2 位

GET greeting/_search
{
  "query": {
    "ranked": {
      "query": {  
        "match_all": {}
      },
      "rank": {
        "pos": [ {"1" : 3}, {"9" : 2} ] 
      }
    }
  }
}

返回

 "hits" : [
      {
        "_source" : {"number" : 2}
      },
      {
        "_source" : {"number" : 9}
      },
      {
        "_source" : {"number" : 1}
      },
      {
        "_source" : {"number" : 3}
      },      
      ...

API

参数	说明	类型	默认值
top	指定_id的文档尽量置顶	string []	-
block	指定_id的文档在结果集中排除	string []	-
pos	精确指定文档排列位置，优先级高于 top	{ string : number } []	-

限制

• block 最多处理 1024 个文档。
• top+pos 合计最多处理 1024 个文档。

影响

• 被处理的文档将丧失原始分 _score。