使用 JOLT 破坏嵌套动态 JSON 数组

Question

I'm looking for flattening nested JSON file into SQL ready format.

JSON file's content:

{
  "ProductLine": [
    "Product 1",
    "Product 2"
  ],
  "Purchase": 364,
  "Cancel": [
    140,
    2
  ]
}

My current transformation:

[
  {
    "operation": "shift",
    "spec": {
      "*": {
        "*": {
          "@": "[#2].&2"
        }
      }
    }
  }
]

Desired output:

[
  {
    "ProductLine": "Product 1",
    "Purchase": 364,
    "Cancel": 140
  },
  {
    "ProductLine": "Product 2",
    "Cancel": 2
  }
]

The difficulty is that arrays can change, sometimes "Cancel" can be an array or sometimes "Purchase" block can be nested.

Answer 1

You can use this spec:

If Purchase or cancel be an array or not, this works

[
  {
    "operation": "cardinality",
    "spec": {
      "*": "MANY"
    }
  },
  {
    "operation": "shift",
    "spec": {
      "ProductLine": {
        "*": {
          "*": {
            "@1": "[&2].&3",
            "@(3,Purchase[&1])": "[&2].Purchase",
            "@(3,Cancel[&1])": "[&2].Cancel"
          }
        }
      }
    }
  }
]

First, change all values to the array. Now you can loop on the ProductLine and get other fields from Purchase and Cancel .

Update: The following answer has been obtained in collaboration with Barbaros Özhan . Special thanks.

[
  {
    "operation": "cardinality",
    "spec": {
      "*": "MANY"
    }
  },
  {
    "operation": "shift",
    "spec": {
      "*": {
        "*": {
          "@": "[#2].&2"
        }
      }
    }
  }
]

Answer 2

We can pick Purchase at a different( outer ) level such as

[
  {
    "operation": "shift",
    "spec": {
      "*": {
        "*": {
          "@": "[#2].&2" 
        }
      },
      "Purchase": "[#].&"// at two level less than the inner object
    }
  }
]

the demo one the site http://jolt-demo.appspot.com/ is

Edit : Considering array indeterminance for the attributes, you can use the following spec alternatively

[
  { //reform two separate objects
    "operation": "shift",
    "spec": {
      "@": "orj",
      "*": "non_array.&.@0[]"
    }
  },
  { // in order to keep the non-array values as the first component of the newly formed array(s) 
    "operation": "sort"
  },
  {
    "operation": "shift",
    "spec": {
      "*": { //the topmost level 
        "*": { //level for the keys
          "*": "&1[]" //match keys and values to convert non-arrays to arrays
        }
      }
    }
  },
  {// pick the first component for the non-array(s)
    "operation": "modify-overwrite-beta",
    "spec": {
      "*": {
        "*": "=firstElement"
      }
    }
  },
  { // apply the original spec after having got individual array values
    "operation": "shift",
    "spec": {
      "*": {
        "*": {
          "@": "[#2].&2"
        }
      }
    }
  },
  { //get rid of the attributes with null values
    "operation": "modify-overwrite-beta",
    "spec": {
      "*": "=recursivelySquashNulls"
    }
  }
]

or another straightforward alternative would be using your original spec after applying cardinality spec such as

[
  {
    "operation": "cardinality",
    "spec": {
      "*": "MANY"
    }
  },
  {
    "operation": "shift",
    "spec": {
      "*": {
        "*": {
          "@": "[#2].&2"
        }
      }
    }
  }
]