求助：编写Jolt Spec实现嵌套JSON转换输出扁平化数组

问题表现 #

进行JSON结构转换时，嵌套的库存层级字段未按预期打平，同存储位置下的多条库存数据被聚合为单对象下的数组字段，未实现「单条库存对应一个独立输出对象」的扁平化效果。

输入样例结构 #

输入JSON包含三层嵌套数组：根节点下的存储位置数组data.storageLocations，每个存储位置下包含库存数组stocks，每个库存条目下又包含数量数组quantity，同时存在全局元数据、商品公共字段需要下沉到每一条扁平化后的库存记录中。

{
  "metadata": {
    "timestamp": "XXXXX"
  },
  "data": {
    "product": {
      "id": "XXXXX"
    },
    "storageLocations": [
      {
        "id": "XXXXX",
        "stocks": [
          {
            "stockStatus": "XXXXX",
            "quantity": [{"uom": "XXXXX","isoUom": "XXXXX","amount": "XXXXX"}]
          },
          {
            "stockStatus": "XXXXX",
            "quantity": [{"uom": "XXXXX","isoUom": "XXXXX","amount": "XXXXX"}]
          },
          {
            "stockStatus": "XXXXX",
            "quantity": [{"uom": "XXXXX","isoUom": "XXXXX","amount": "XXXXX"}]
          }
        ],
        "dateModified": "XXXXX"
      },
      {
        "id": "XXXXX",
        "stocks": [
          {
            "stockStatus": "XXXXX",
            "quantity": [{"uom": "XXXXX","isoUom": "XXXXX","amount": "XXXXX"}]
          },
          {
            "stockStatus": "XXXXX",
            "quantity": [{"uom": "XXXXX","isoUom": "XXXXX","amount": "XXXXX"}]
          },
          {
            "stockStatus": "XXXXX",
            "quantity": [{"uom": "XXXXX","isoUom": "XXXXX","amount": "XXXXX"}]
          }
        ],
        "dateModified": "XXXXX"
      },
      {
        "id": "XXXXX",
        "stocks": [
          {
            "stockStatus": "XXXXX",
            "quantity": [{"uom": "XXXXX","isoUom": "XXXXX","amount": "XXXXX"}]
          },
          {
            "stockStatus": "XXXXX",
            "quantity": [{"uom": "XXXXX","isoUom": "XXXXX","amount": "XXXXX"}]
          },
          {
            "stockStatus": "XXXXX",
            "quantity": [{"uom": "XXXXX","isoUom": "XXXXX","amount": "XXXXX"}]
          }
        ],
        "dateModified": "XXXXX"
      }
    ],
    "dateModified": "XXXXX",
    "temporaryNegativeStock": "XXXXX"
  }
}

原有Spec问题 #

原有Spec直接在最内层quantity层级使用[&5]作为输出索引，该索引仅对应存储位置的数组下标，同个存储位置下的3条库存数据都会被写入同一个输出对象的同路径下，Jolt默认会将同路径多值合并为数组，最终导致库存字段全部变为数组形式。
原有错误Spec：

[
  {
    "operation": "shift",
    "spec": {
      "data": {
        "storageLocations": {
          "*": {
            "@(3,metadata.timestamp)": "[&1].LastUpdated",
            "@(2,product.id)": "[&1].MaterialCode",
            "@(2,product.baseUoM)": "[&1].BaseUoM",
            "@(2,product.isoBaseUom)": "[&1].ISOBaseUoM",
            "@(2,location.id)": "[&1].LocationId",
            "@(id)": "[&1].StorageLocationId",
            "@(dateModified)": "[&1].StorageLocationLastUpdated",
            "stocks": {
              "*": {
                "quantity": {
                  "*": {
                    "@(amount)": "[&5].StockAmount",
                    "@(isoUom)": "[&5].StockISOUoM",
                    "@(uom)": "[&5].StockUoM",
                    "@(2,stockStatus)": "[&5].StockStatus"
                  }
                }
              }
            }
          }
        }
      }
    }
  }
]

错误输出表现 #

输出数组长度与存储位置数量一致（共3个对象），每个对象内的库存字段均为长度3的数组，未打平到单条库存粒度。

[
  {
    "LastUpdated": "XXXXX",
    "MaterialCode": "na-XXXXX",
    "StorageLocationId": "XXXXX",
    "StorageLocationLastUpdated": "XXXXX",
    "StockAmount": ["0.000","0.000","0.000"],
    "StockISOUoM": ["XXXXX","XXXXX","XXXXX"],
    "StockUoM": ["XXXXX","XXXXX","XXXXX"],
    "StockStatus": ["XXXXX","XXXXX","XXXXX"]
  },
  {
    "LastUpdated": "XXXXX",
    "MaterialCode": "na-XXXXX",
    "StorageLocationId": "XXXXX",
    "StorageLocationLastUpdated": "XXXXX",
    "StockAmount": ["0.000","0.000","0.000"],
    "StockISOUoM": ["XXXXX","XXXXX","XXXXX"],
    "StockUoM": ["XXXXX","XXXXX","XXXXX"],
    "StockStatus": ["XXXXX","XXXXX","XXXXX"]
  },
  {
    "LastUpdated": "XXXXX",
    "MaterialCode": "na-XXXXX",
    "StorageLocationId": "XXXXX",
    "StorageLocationLastUpdated": "XXXXX",
    "StockAmount": ["0.000","0.000","0.000"],
    "StockISOUoM": ["XXXXX","XXXXX","XXXXX"],
    "StockUoM": ["XXXXX","XXXXX","XXXXX"],
    "StockStatus": ["XXXXX","XXXXX","XXXXX"]
  }
]

正确实现方案 #

采用两步shift转换，第一步先归集所有层级数据到临时路径，避免索引计算混乱；第二步遍历到最内层库存条目，使用自增索引生成独立输出对象，确保每条库存对应一条记录。
可用的Jolt Spec如下：

[
  {
    "operation": "shift",
    "spec": {
      "metadata": {
        "timestamp": "tmp.timestamp"
      },
      "data": {
        "product": {
          "id": "tmp.productId",
          "baseUoM": "tmp.baseUoM",
          "isoBaseUom": "tmp.isoBaseUom"
        },
        "storageLocations": {
          "*": {
            "id": "tmp.locations[&1].id",
            "dateModified": "tmp.locations[&1].dateModified",
            "stocks": {
              "*": {
                "stockStatus": "tmp.locations[&3].stocks[&1].stockStatus",
                "quantity": {
                  "*": {
                    "uom": "tmp.locations[&5].stocks[&3].uom",
                    "isoUom": "tmp.locations[&5].stocks[&3].isoUom",
                    "amount": "tmp.locations[&5].stocks[&3].amount"
                  }
                }
              }
            }
          }
        }
      }
    }
  },
  {
    "operation": "shift",
    "spec": {
      "tmp": {
        "locations": {
          "*": {
            "stocks": {
              "*": {
                "@(4,timestamp)": "[#4].LastUpdated",
                "@(4,productId)": "[#4].MaterialCode",
                "@(4,baseUoM)": "[#4].BaseUoM",
                "@(4,isoBaseUom)": "[#4].ISOBaseUoM",
                "@(2,id)": "[#4].StorageLocationId",
                "@(2,dateModified)": "[#4].StorageLocationLastUpdated",
                "amount": "[#4].StockAmount",
                "isoUom": "[#4].StockISOUoM",
                "uom": "[#4].StockUoM",
                "stockStatus": "[#4].StockStatus"
              }
            }
          }
        }
      }
    }
  }
]

输出效果 #

最终输出数组长度为「存储位置数 * 每个存储位置下的库存条目数」，每个对象为单条库存的全量扁平化数据，所有字段均为单值，不存在数组聚合问题。

内容的提问来源于stack exchange，提问作者Thiyagaraj Narayanan