JOLT转换嵌套数组与重复键：如何将记录展开为独立对象

需求说明 #

需将包含readings数组的JSON转换为每条记录独立的对象，每个对象包含meter、Ah、epoch_timestamp、copper_id（对应原uniqueID）字段，忽略原始的timestamp字段。现有shift操作将同一仪表的字段合并为数组，不符合需求。

原始JSON #

{"readings":[{"meter":"02319771","records":[{"timestamp":"28/06/2024 14:35:00","values":{"Ah":4350.947,"epoch_timestamp":1719578227,"uniqueID":"12005"}},{"timestamp":"28/06/2024 15:00:00","values":{"Ah":4353.141,"epoch_timestamp":1719579724,"uniqueID":"12005"}}]}]}

现有JOLT规范 #

[{"operation":"shift","spec":{"readings":{"*":{"meter":"[&1].meter_serial","records":{"*":{"values":{"epoch_timestamp":"[&4].&","uniqueID":"[&4].copper_id","Ah":"[&4].ah"}}}}}}]

当前转换结果 #

[{"meter_serial":"02319771","epoch_timestamp":[1719578227,1719579724],"copper_id":["12005","12005"],"ah":[4350.947,4353.141]}]

期望输出 #

[{"meter":"02319771","Ah":4350.947,"epoch_timestamp":1719578227,"copper_id":"12005"},{"meter":"02319771","Ah":4353.141,"epoch_timestamp":1719579724,"copper_id":"12005"}]

修正后的JOLT规范 #

[
  {
    "operation": "shift",
    "spec": {
      "readings": {
        "*": {
          "records": {
            "*": {
              "values": {
                "Ah": "[&3].[&1].Ah",
                "epoch_timestamp": "[&3].[&1].epoch_timestamp",
                "uniqueID": "[&3].[&1].copper_id"
              },
              "@(2,meter)": "[&3].[&1].meter"
            }
          }
        }
      }
    }
  },
  {
    "operation": "shift",
    "spec": {
      "*": {
        "*": "[]"
      }
    }
  }
]

规则解释 #

1. 第一个shift操作：
   • 用[&3].[&1]定位：&3对应外层readings的索引，&1对应当前record的索引，确保每条record生成独立的嵌套对象
   • 通过@(2,meter)将当前仪表的meter字段传递到对应record的对象中
   • 映射values下的字段到目标字段，将uniqueID重命名为copper_id
2. 第二个shift操作：
   • 将嵌套的二维数组扁平化，得到一维对象数组

内容的提问来源于stack exchange，提问作者Tokolosk