Within a given JSON I have to adapt Dates to an ISO format, and I do not get the right approach.

This is my input json:

{
  "itemNo": "1010333",
  "productNo": "myProduct",
  "description": "myItem-01",
  "attributes": [
    {
      "identifier": "FIRST_DATE",
      "type": "attribute",
      "DataType": "DATE",
      "values": [
        {
          "value": "05.12.2022"
        }
      ]
    },
    {
      "identifier": "mytimestamp",
      "type": "attribute",
      "DataType": "DATEANDTIME",
      "values": [
        {
          "value": "17.07.2023 13:13:10"
        }
      ]
    }
  ]
}

and this is the desired output
when DataType is DATE => parse String of value to yyyy_MM-ss
when DataType is DATEANDTIME => parse String of value to yyyy_MM-ss’T’hh:mm:ss

{
  "itemNo": "1010333",
  "productNo": "myProduct",
  "description": "myItem-01",
  "attributes": [
    {
      "identifier": "FIRST_DATE",
      "type": "attribute",
      "DataType": "DATE",
      "values": [
        {
          "value": "2022-12-05"
        }
      ]
    },
    {
      "identifier": "mytimestamp",
      "type": "attribute",
      "DataType": "DATEANDTIME",
      "values": [
        {
          "value": "2023-07-17'T'13:13:10"
        }
      ]
    }
  ]
}

Unfortuantely I cannot provide the slightest approach here, because I do not know wether to start with “shift” or with “modify-overwrite-beta”, due to the fact, that all depends on the DataType.
If it is not a good idea to transform Dates within JOLT I can try it with another nifi processor.Currently my aim is, to proceed as much as possible with JOLT. Thanks in advance for your opinion.