elasticsearch - 获取带有排序嵌套对象的文档

Question

我有以下映射:http://pastebin.com/FgtC5aaw(包括样本数据)

http://localhost:9200/databases/database/1

{
 "columns": [
   {
     "name": "First column",
     "position": 1
   },
   {
     "name": "Second column",
     "position": 2
   },
   {
     "name": "Third column",
     "position": 3
   }
 ],
 "name": "Database name"
}

我想得到的是“数据库”文档，但是每个带有列的列均按columns.position字段降序排列，因此:

{
    "name": "Database name",
    "columns": [
        {
            "name": "Third column",
            "position": 3
        },
        {
            "name": "Second column",
            "position": 2
        },
        {
            "name": "First column",
            "position": 1
        }
    ]
}

我尝试在“columns.position”上使用“sort”，但是我认为它只会对嵌套列上的所有“数据库”文档进行排序。

在Elasticsearch中这样的结果可能吗？

Answer 1

如果事先知道，您将始终需要按nested的降序检索columns.position对象列，您只需使用已经正确排序的columns列表对数据库文档进行索引即可。

如果不是这种情况，有一种方法可以使用 columns 对嵌套的inner_hits列表进行排序，如下所示:

curl -XPOST localhost:9200/databases/database/_search -d '{
  "_source": false,            <---- set this to true to get database fields too
  "query": {
    "nested": {
      "path": "columns",
      "query": {
        "match_all": {}
      },
      "inner_hits": {
        "sort": {
          "columns.position": "desc"
        }
      }
    }
  }
}'

这将产生:

{
  ...
  "hits" : {
    "total" : 1,
    "max_score" : 1.0,
    "hits" : [ {
      "_index" : "databases",
      "_type" : "database",
      "_id" : "1",
      "_score" : 1.0,
      "inner_hits" : {
        "columns" : {
          "hits" : {
            "total" : 3,
            "max_score" : null,
            "hits" : [ {
              "_index" : "databases",
              "_type" : "database",
              "_id" : "1",
              "_nested" : {
                "field" : "columns",
                "offset" : 2
              },
              "_score" : null,
              "_source":{"name":"Third column","position":3},
              "sort" : [ 3 ]
            }, {
              "_index" : "databases",
              "_type" : "database",
              "_id" : "1",
              "_nested" : {
                "field" : "columns",
                "offset" : 1
              },
              "_score" : null,
              "_source":{"name":"Second column","position":2},
              "sort" : [ 2 ]
            }, {
              "_index" : "databases",
              "_type" : "database",
              "_id" : "1",
              "_nested" : {
                "field" : "columns",
                "offset" : 0
              },
              "_score" : null,
              "_source":{"name":"First column","position":1},
              "sort" : [ 1 ]
            } ]
          }
        }
      }
    } ]
  }
}