ElasticSearch 在无痛脚本中舍入浮点值

Question

我在使用 ElasticSearch 的脚本功能时注意到舍入问题。我有以下映射：

{
  "my-index": {
    "mappings": {
      "my-document": {
        "properties": {
          "id": {
            "type": "long"
          },
          "foo": {
            "type": "float"
          }
        }
      }
    }
  }
}

和一堆文档索引。一份文档的 foo 字段值为 159944154.8644。当我发送以下查询时：

{
  "query": {
    "bool": {
      "must": [
        { "match": { "id": 42 } }
      ]
    }
  },
  "_source": {
    "includes": ["id","foo"]
  },
  "script_fields": {
    "foo_scripted": {
      "script": {
        "lang": "painless",
        "inline": "doc['foo'].value"
      }
    }
  }
}

我得到以下舍入值作为响应：

{
  "_index": "my-index",
  "_type": "my-document",
  "_id": "42",
  "_score": 1,
  "_source": {
    "foo": 1.599441548644E8, //= 159944154.8644 -> correct!
    "id": 42
  },
  "fields": {
    "foo_scripted": [
      1.5994416E8 //= 159944160.0 -> rounded :(
    ]
  }
}

知道如何解决这个问题吗？提前致谢。

ES 版本：5.3.0

Answer 1

映射将字段类型设置为 "float", which is a single-precision 32-bit IEEE 754 floating point number. This floating point representation is capable of storing at most 9 significant decimal digits。尝试将字段类型更改为 "double"，这是一个双精度 64 位 IEEE 754 浮点数。

ElasticSearch 在无痛脚本中舍入浮点值

ElasticSearch rounds float value in painless script

elasticsearch

elasticsearch-5

elasticsearch-painless