ES中数据类型转换处理

原创

周银辉

发布于 2025-02-13 16:10:39

1250

文章被收录于专栏：elasticsearch常见问题elasticsearch常见问题

测试数据类型存储

PUT my_type_index
{
  "mappings": {
    "properties": {
      "price":{
        "type":"float"
      }
    }
  }
}

正常存储数字

PUT my_type_index/_doc/1
{
  "price":7.8
}

这里能把字符串数据存进去

PUT my_type_index/_doc/2
{
  "price":"9.4"
}

字符串无法存储

PUT my_type_index/_doc/3
{
  "price":"change the data type"
}

查询到数据即有数字，又有字符串

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

如何才能只存储数字,字段中使用coerce

DELETE my_type_index
PUT my_type_index
{
  "mappings": {
    "properties": {
      "price": {
        "type": "float",
        "coerce": false
      }
    }
  }
}

这个时候就无法写入了

PUT my_type_index/_doc/1
{
  "price":"9.4"
}

针对已有的数据如何处理呢？可以通过reindex + pipeline

DELETE my_type_index2
PUT my_type_index2
{
  "mappings": {
    "properties": {
      "price":{
        "type":"float",
        "coerce": false
      }
    }
  }
}

这是是不会直接成功的，必须先转换

POST _reindex
{
  "source": {
    "index": "my_type_index"
  },
  "dest": {
    "index": "my_type_index2"
  }
}

创建pipeline

PUT _ingest/pipeline/my_convert_test_pipe
{
  "description": "convert string to float", 
  "processors": [
    {
      "convert": {
        "field": "price",
        "type": "float"
      }
    }
  ]
}

再次重建

POST _reindex
{
  "source": {
    "index": "my_type_index"
  },
  "dest": {
    "index": "my_type_index2",
    "pipeline": "my_convert_test_pipe"
  }
}