JOLT转换：如何有条件合并两个字段并保留其他属性？

需求说明 #

需要合并AccountNumber和EntityID两个字段，仅保留非空值（输入保证二者必有一个有效，且绝不会同时有值），同时保留其他所有相关属性。

输入示例 #

{
  "AccountNumber": "123456",
  "EntityID": "",
  "TransactionYear": "2023",
  "TransactionMonth": "03",
  "TransactionDay": "05",
  "TransactionHour": "10",
  "TransactionMinutes": "15",
  "TransactionSeconds": "30",
  "ComponentNumber": "12345",
  "Price": "250.00",
  "CurrencyCode": "USD"
}

目标输出 #

合并后AccountNumber为单个非空值，保留其他所有属性：

{
  "AccountNumber": "123456",
  "TransactionData": {
    "TransactionYear": "2023",
    "TransactionMonth": "03",
    "TransactionDay": "05",
    "TransactionHour": "10",
    "TransactionMinutes": "15",
    "TransactionSeconds": "30"
  },
  "NewPrice": {
    "ComponentNumber": "12345",
    "SpecialPrice": {
      "Price": "250.00",
      "CurrencyCode": "USD"
    }
  }
}

当前问题 #

原JOLT规范使用firstNotNull时，输出的AccountNumber是包含空字符串和有效值的数组，而非单个值：

{
  "AccountNumber": ["","123456"],
  "TransactionData": {
    "TransactionYear": "2023",
    "TransactionMonth": "03",
    "TransactionDay": "05",
    "TransactionHour": "10",
    "TransactionMinutes": "15",
    "TransactionSeconds": "30"
  },
  "NewPrice": {
    "ComponentNumber": "12345",
    "SpecialPrice": {
      "Price": "250.00",
      "CurrencyCode": "USD"
    }
  }
}

解决方案 #

方案1：利用Shift操作的条件过滤（推荐） #

直接在Shift步骤中添加非空判断，仅将非空字段映射到目标AccountNumber字段，无需额外临时字段和修改步骤：

[
  {
    "operation": "shift",
    "spec": {
      // 仅当AccountNumber非空时映射到目标字段
      "AccountNumber": {
        "?": {
          "@(2,AccountNumber)": "AccountNumber"
        }
      },
      // 仅当EntityID非空时映射到目标字段（二者不同时有值，不会冲突）
      "EntityID": {
        "?": {
          "@(2,EntityID)": "AccountNumber"
        }
      },
      "TransactionYear": "TransactionData.TransactionYear",
      "TransactionMonth": "TransactionData.TransactionMonth",
      "TransactionDay": "TransactionData.TransactionDay",
      "TransactionHour": "TransactionData.TransactionHour",
      "TransactionMinutes": "TransactionData.TransactionMinutes",
      "TransactionSeconds": "TransactionData.TransactionSeconds",
      "ComponentNumber": "NewPrice.ComponentNumber",
      "Price": "NewPrice.SpecialPrice.Price",
      "CurrencyCode": "NewPrice.SpecialPrice.CurrencyCode"
    }
  }
]

方案2：先将空字符串转为Null再使用firstNotNull #

如果需要保留临时字段的方式，可先将空字符串转为null，让firstNotNull能正确识别并选择第一个非空值：

[
  {
    "operation": "shift",
    "spec": {
      "AccountNumber": "Temp.AccountNumber",
      "EntityID": "Temp.EntityID",
      "TransactionYear": "TransactionData.TransactionYear",
      "TransactionMonth": "TransactionData.TransactionMonth",
      "TransactionDay": "TransactionData.TransactionDay",
      "TransactionHour": "TransactionData.TransactionHour",
      "TransactionMinutes": "TransactionData.TransactionMinutes",
      "TransactionSeconds": "TransactionData.TransactionSeconds",
      "ComponentNumber": "NewPrice.ComponentNumber",
      "Price": "NewPrice.SpecialPrice.Price",
      "CurrencyCode": "NewPrice.SpecialPrice.CurrencyCode"
    }
  },
  {
    "operation": "modify-overwrite-beta",
    "spec": {
      "Temp": {
        // 将空字符串转为null，使firstNotNull能正确筛选
        "AccountNumber": ["=isNull(@(1))", "@(1)"],
        "EntityID": ["=isNull(@(1))", "@(1)"]
      },
      "AccountNumber": "=firstNotNull(@(1,Temp.AccountNumber), @(1,Temp.EntityID))"
    }
  },
  {
    "operation": "remove",
    "spec": {
      "Temp": ""
    }
  }
]

问题原因说明 #

原规范中firstNotNull生成数组的核心原因是：JOLT将空字符串视为有效值，而非null。当传入两个参数（一个空字符串、一个有效值）时，firstNotNull不会筛选掉空字符串，而是将两个参数打包成数组返回。

内容的提问来源于stack exchange，提问作者Ryan