Parse/Flatten json document stored as text in parquet file (mapping data flow)

Hello,

I have a file stored in ADL in parquet format that has a column that contains json data (stored as text) stored like below. I need to flatten/parse it using mapping data flow and I have a really hard time doing it.

The expected result is 

[
  {
    "question": {
      "id": "e7e68630-53f2-49ea-8348-0c25aaf1e572"
    },
    "value": ["abc"]
  },
  {
    "question": {
      "id": "2a9ce734-e4b9-436a-b00e-067968a35301"
    },
    "value": ["bcd"]
  },
  {
    "question": {
      "id": "f6567968-ee60-4fe0-b90c-72114403e3f8"
    },
    "value": ["xxx"]
  },
  {
    "question": {
      "id": "8906e60b-4f07-4930-bdf6-5cef7d584d18"
    },
    "value": ["yyy"]
  },
  {
    "question": {
      "id": "d0916cbb-d858-4401-b036-931a06be3a67"
    },
    "value": ["ddd"]
  },
  {
    "question": {
      "id": "d3e838ed-d26d-4830-8b22-a5fa7a95322d"
    },
    "value": ["bbb"]
  },
  {
    "question": {
      "id": "5a077e11-304a-4053-ad1b-fa5bfc5fe156"
    },
    "value": ["ggg"]
  }
]