Apache Nifi Jolt Transform : Almost there but not quite..

[#1]

I'm trying to transform

{"namespace": "glmm.data",
 "type": "record",
 "name": "glmmobs",
 "fields": [
{"name": "uid", "type": "int"},
{"name": "user", "type": "string"},
{"name": "job", "type": "string"},
{"name": "fb1_a", "type": "float"},
{"name": "fb1_b", "type": "float"},
{"name": "fb2_a", "type": "int"},
{"name": "fb2_b", "type": "int"},
{"name": "fb2_c", "type": "string"}
]
}

into

{"namespace": "glmm.data",
"type": "record",
"name": "glmmobs",
"fields": [
{"name": "uid", "type": "int"},
{"name": "user", "type": "string"},
{"name": "job", "type": "string"},
{"name": "fb1", "type": "record", 
  "fields": [ {"name": "fb1_a", "type": "float"},
              {"name": "fb1_b", "type": "float"}]
},
{"name": "fb2", "type": "record",
  "fields": [ {"name": "fb2_a", "type": "int"},
              {"name": "fb2_b", "type": "int"},
              {"name": "fb2_c", "type": "string"}]
}
]
}

By using the following spec, I'm able to get close to what I want but not quite.. 

[
{
"operation": "shift",
"spec": {
"namespace": "namespace",
"type": "type",
"name": "name",

"fields": {
"0": "&1[0]",
"1": "&1[1]",
"2": "&1[2]",
"3": "&1[3].fb1[0]",
"4": "&1[3].fb1[1]", 
"5": "&1[4].fb2[0]",
"6": "&1[4].fb2[1]",
"7": "&1[4].fb2[2]"
}
}
}
]

>>>

{
"namespace": "glmm.data",
"type": "record",
"name": "glmmobs",
"fields": [
{"name": "uid","type": "int"}, 
{"name": "user","type": "string"}, 
{"name": "job","type": "string"}, 
{"fb1": [{"name": "fb1_a","type": "float"}, 
        {"name": "fb1_b","type": "float"}]
}, {
"fb2": [{"name": "fb2_a","type": "int"}, 
        {"name": "fb2_b","type": "int"}, 
        {"name": "fb2_c","type": "string"}]
}]
}

Can somebody please help?  (An additional nice to have is not having to hardcode the "fb1", "fb2" etc.)

Thanks in advance..

Rao

[#2]

Got it.. I think.  But, is this convoluted?

[
  {
    "operation": "shift",
    "spec": {

      "namespace":"namespace",
        "type":"type",
        "name":"name",
      "fields": {
      "0": "&1[0]",
      "1": "&1[1]",
      "2": "&1[2]",
      "3": "&1[3].fb1[0]",
      "4": "&1[3].fb1[1]", 
      "5": "&1[4].fb2[0]",
      "6": "&1[4].fb2[1]",
      "7": "&1[4].fb2[2]"
    }
    }
  }
  ,
  {
    "operation": "shift",
    "spec": {

      "namespace":"namespace",
        "type":"type",
        "name":"name",
          "fields": {
	      "0": "&1[0]",
	      "1": "&1[1]",
	      "2": "&1[2]",
	      "3": {
            "*": {
                  "$":"fields.[#3].name",
                  "@":"fields.[#3].fields"
                 }
          },
	      "4": {
            "*": {
                  "$":"fields.[#3].name",
                  "@":"fields.[#3].fields"
                 }
          }
    	}
    }}
]

>>

{
	"namespace": "glmm.data",
	"type": "record",
	"name": "glmmobs",
	"fields": [{
		"name": "uid",
		"type": "int"
	}, {
		"name": "user",
		"type": "string"
	}, {
		"name": "job",
		"type": "string"
	}, {
		"name": "fb1",
		"fields": [{
			"name": "fb1_a",
			"type": "float"
		}, {
			"name": "fb1_b",
			"type": "float"
		}]
	}, {
		"name": "fb2",
		"fields": [{
			"name": "fb2_a",
			"type": "int"
		}, {
			"name": "fb2_b",
			"type": "int"
		}, {
			"name": "fb2_c",
			"type": "string"
		}]
	}]
}

thanks,

Rao