jq: How to convert flat json into nested one

Question

I've been trying to re-arrange a pretty flat json into structure with more depth, so far without any success. Here's my source data:

[ { "id": "27", "time": "2017-12-21 07:24:00", "service_name": "prices", "version": "61f4u8e", "event": "Success" }, { "id": "28", "time": "2017-12-21 07:23:00", "service_name": "prices", "version": "21c2f7d", "event": "Fail" }, { "id": "29", "time": "2017-12-21 07:21:00", "service_name": "cart", "version": "9ff24c4", "event": "Success" } ]

The result should look like this:

[ { "name": "cart", "data": [ { "date": "2017-12-21 07:21:00", "details": { "event": "Success", "version": "9ff24c4" } } ] }, { "name": "prices", "data": [ { "date": "2017-12-21 07:24:00", "details": { "event": "Success", "version": "61f4u8e" } }, { "date": "2017-12-21 07:23:00", "details": { "event": "Fail", "version": "21c2f7d" } } ] } ]

So, basically grouping by service_name and moving the other fields around a little.

So far I've managed to group, but only one level:

group_by(.service_name) | map({name: .[0].service_name, data: .}) How can I shuffle around the other values, assuming the input will always have all fields as in the example above?

Answer 1

jq solution:

jq '[group_by(.service_name)[] 
    | .[0].service_name as $name 
    | { name: $name, 
        data: map({ date: .time, 
                    details: {event, version}}) 
      }]' jsonfile

The output:

[
  {
    "name": "cart",
    "data": [
      {
        "date": "2017-12-21 07:21:00",
        "details": {
          "event": "Success",
          "version": "9ff24c4"
        }
      }
    ]
  },
  {
    "name": "prices",
    "data": [
      {
        "date": "2017-12-21 07:24:00",
        "details": {
          "event": "Success",
          "version": "61f4u8e"
        }
      },
      {
        "date": "2017-12-21 07:23:00",
        "details": {
          "event": "Fail",
          "version": "21c2f7d"
        }
      }
    ]
  }
]