如何在Elasticsearch/OpenSearch中利用脚本字段计算值实现过滤？

核心问题原因 #

script_fields是在查询结果返回阶段计算的，无法在过滤/查询阶段被引用，所以直接用它做range过滤行不通。

解决方案1：使用Script Filter直接计算过滤 #

不需要提前索引计算值，直接在bool的filter里用脚本判断计算结果是否符合范围。

Elasticsearch/OpenSearch 查询示例 #

{
  "_source": ["*"],
  "query": {
    "function_score": {
      "query": {
        "bool": {
          "filter": {
            "script": {
              "source": "params.total_stay * doc['default_rate'].value >= params.min_fee && params.total_stay * doc['default_rate'].value <= params.max_fee",
              "params": {
                "total_stay": 12,
                "min_fee": 12,
                "max_fee": 100
              }
            }
          }
        }
      },
      "boost_mode": "replace",
      "score_mode": "first"
    }
  },
  "script_fields": {
    "total_stay_fee": {
      "script": {
        "source": "params.total_stay * doc['default_rate'].value",
        "params": {
          "total_stay": 12
        }
      }
    }
  }
}

Searchkick 中的实现（Ruby on Rails） #

Model.search(
  "*",
  body: {
    query: {
      function_score: {
        query: {
          bool: {
            filter: {
              script: {
                source: "params.total_stay * doc['default_rate'].value >= params.min_fee && params.total_stay * doc['default_rate'].value <= params.max_fee",
                params: { total_stay: 12, min_fee: 12, max_fee: 100 }
              }
            }
          }
        },
        boost_mode: "replace",
        score_mode: "first"
      }
    },
    script_fields: {
      total_stay_fee: {
        script: {
          source: "params.total_stay * doc['default_rate'].value",
          params: { total_stay: 12 }
        }
      }
    }
  }
)

解决方案2：使用Runtime Fields（推荐） #

Runtime Fields 是在查询阶段动态计算的虚拟字段，可像普通字段一样用于过滤、排序和聚合，且无需修改现有索引结构。

Elasticsearch/OpenSearch 查询示例 #

{
  "_source": ["*"],
  "runtime_mappings": {
    "total_stay_fee": {
      "type": "double",
      "script": {
        "source": "emit(params.total_stay * doc['default_rate'].value)",
        "params": {
          "total_stay": 12
        }
      }
    }
  },
  "query": {
    "function_score": {
      "query": {
        "bool": {
          "filter": {
            "range": {
              "total_stay_fee": {
                "gte": 12,
                "lte": 100
              }
            }
          }
        }
      },
      "boost_mode": "replace",
      "score_mode": "first"
    }
  }
}

Searchkick 中的实现 #

Model.search(
  "*",
  body: {
    runtime_mappings: {
      total_stay_fee: {
        type: "double",
        script: {
          source: "emit(params.total_stay * doc['default_rate'].value)",
          params: { total_stay: 12 }
        }
      }
    },
    query: {
      function_score: {
        query: {
          bool: {
            filter: {
              range: {
                total_stay_fee: { gte: 12, lte: 100 }
              }
            }
          }
        },
        boost_mode: "replace",
        score_mode: "first"
      }
    }
  }
)

兼容性说明 #

以上两种方案均支持Elasticsearch 7.x+ 和 OpenSearch 1.x+，可通过对应API直接实现。

内容的提问来源于stack exchange，提问作者Code father