Elasticsearch - 如何过滤嵌套聚合桶？

Question

我正在尝试过滤嵌套聚合的桶.

制图:

{
  "dev.directory.3" : {
    "mappings" : {
      "profile" : {
        "properties" : {
          "events" : {
            "type" : "nested",
            "properties" : {
              "id" : {
                "type" : "integer"
              },
              "name" : {
                "type" : "string",
                "index" : "not_analyzed"
              },
            }
          },
          "title" : {
            "type" : "string"
          }
        }
      }
    }
  }
}

指数数据:

"hits" : {
    "total" : 1,
    "max_score" : 1.0,
    "hits" : [ {
      "_index" : "dev.directory.3",
      "_type" : "profile",
      "_id" : "1",
      "_score" : 1.0,
      "_source" : {
        "title" : "Project manager",
        "events" : [ 
          {
            "id" : 1,
            "name" : "Event A",
          }, 
          {
            "id" : 2,
            "name" : "Event B",
          },
          {
            "id" : 3,
            "name" : "Event C",
          },
          {
            "id" : 4,
            "name" : "Event D",
          } 
        ],
      }
    }
  ]
}

我正在使用此查询和聚合定义

{
    "query": {
        "nested": {
            "path": "events",
            "query": {
                "bool": {
                    "filter": [{
                        "terms": {
                            "events.id": [1, 2]
                        }
                    }]
                }
            },
            "inner_hits": {}
        }
    },
    "aggs": {
        "events.name12": {
            "filter": {},
            "aggs": {
                "inner": {
                    "nested": {
                        "path": "events"
                    },
                    "aggs": {
                        "events.name": {
                            "terms": {
                                "field": "events.name"
                            }
                        },
                        "events.name_count": {
                            "cardinality": {
                                "field": "events.name"
                            }
                        }
                    }
                }
            }
        }
    },
    "size": 20,
    "_source": ["email", "company_name", "events"]
}

我得到的是聚合结果是:

"aggregations": {
    "events.name12": {
      "doc_count": 2,
      "filtered": {
        "doc_count": 4,
        "events.name": {
          "buckets": [
            {
              "key": "Event A",
              "doc_count": 1
            },
            {
              "key": "Event B",
              "doc_count": 1
            },
            {
              "key": "Event C",
              "doc_count": 1
            },
            {
              "key": "Event D",
              "doc_count": 1
            }
          ]
        },
        "events.name_count": {
          "value": 4
        }
      }
    }

我只是为了提供事件ID而过滤桶列表,所以结果应该是:

"aggregations": {
    "events.name12": {
      "doc_count": 2,
      "filtered": {
        "doc_count": 2,
        "events.name": {
          "buckets": [
            {
              "key": "Event A",
              "doc_count": 1
            },
            {
              "key": "Event B",
              "doc_count": 1
            }
          ]
        },
        "events.name_count": {
          "value": 2
        }
      }
    }

Answer 1

你几乎就在那里,你只需要events.id在聚合中添加过滤器,如下所示:

{
  "query": {
    "nested": {
      "path": "events",
      "query": {
        "bool": {
          "filter": [
            {
              "terms": {
                "events.id": [
                  1,
                  2
                ]
              }
            }
          ]
        }
      },
      "inner_hits": {}
    }
  },
  "aggs": {
    "events.name12": {
      "nested": {
        "path": "events"
      },
      "aggs": {
        "inner": {
          "filter": {
            "terms": {
              "events.id": [
                1,
                2
              ]
            }
          },
          "aggs": {
            "events.name": {
              "terms": {
                "field": "events.name"
              }
            },
            "events.name_count": {
              "cardinality": {
                "field": "events.name"
              }
            }
          }
        }
      }
    }
  },
  "size": 20,
  "_source": [
    "email",
    "company_name",
    "events"
  ]
}

原因是您的查询将正确选择具有指定事件ID的嵌套事件的所有文档,但是,您的聚合将处理来自所有选定文档的所有嵌套事件.因此,您需要从聚合中没有正确ID的那些文档中过滤掉所有嵌套事件.