Flatten a json datastructure in javascript

Question

I have a json response coming from a service that's a bit of a pain to work with. How can I go about flattening this structure a bit so that "not_restricted":{"doc_count": 976465} is changed to "not_restricted":976465? I'd perfer a function that can detect this kind of a structure in all types of json docs and modify the result.

This is a the general structure of the json:

{
    "took": 159,
    "timed_out": false,
    "_shards": {
        "total": 6,
        "successful": 6,
        "failed": 0
    },
    "hits": {
        "total": 4909332,
        "max_score": 1,
        "hits": [
            {
                ...
            },
            {
                ...
            },
            {
                ...
            }
        ]
    },
    "aggregations": {
        "index_types": {
            "doc_count_error_upper_bound": 0,
            "sum_other_doc_count": 0,
            "buckets": [
                {
                    "key": "client",
                    "doc_count": 1958205,
                    "not_restricted": {
                        "doc_count": 976465
                    },
                    "restricted": {
                        "doc_count": 981740
                    }
                },
                {
                    "key": "ultimateparent",
                    "doc_count": 1616164,
                    "not_restricted": {
                        "doc_count": 741059
                    },
                    "restricted": {
                        "doc_count": 875105
                    }
                },
                {
                    "key": "facility",
                    "doc_count": 1334963,
                    "not_restricted": {
                        "doc_count": 914090
                    },
                    "restricted": {
                        "doc_count": 420872
                    }
                }
            ]
        }
    }
}

Answer 1

You can very easily do this by using the reviver parameter of JSON.parse, which allows you to filter every object in the JSON:

var res = JSON.parse(json, function(k, o) {
    if (Object(o) !== o) return o; // primitive values
    var keys = Object.keys(o);
    if (keys.length == 1 && keys[0] == "doc_count")
        return o.doc_count;
    else
        return o;
});

You can also test k to include "restricted" or so if you want to make sure that doc_count singletons are only replaced in such properties.

Answer 2

You can use LoDash to make your life easier. Include it, and then you can use _.each

_.each(data.aggregations.index_types.buckets, function(n){
  n.not_restricted = n.not_restricted.doc_count;
  n.restricted = n.restricted.doc_count;
})

Answer 3

If you want to go thru the object and flatten all 1 property objects, you can do:

for (var i = 0; i < data.aggregations["index_types"].buckets.length; i++) {
    var obj = data.aggregations["index_types"].buckets[i];
    for (var key in obj) {
        if (typeof obj[key] === "object" && Object.keys(obj[key]).length === 1) {
            //1 property object!
            obj[key] = obj[key][Object.keys(obj[key])[0]];
        } 
    }
}

Demo: http://jsfiddle.net/jkm1vmbm/

Answer 4

Since you said that the JSON are actually strings, if all you need is that specific replacement:

Replacing `{"doc_count": NUMBER}` with that `NUMBER`

What you can do is just use a regex to replace what you want:

s.replace(/({"doc_count":([0-9]+)})/g, '$2')

Which works like:

(                   // start capturing group 1
{\"doc_count\":     // match the string {\"doc_count\":
    (               // start capturing group 2
    [0-9]+          // match one or more digits
    )               // close capturing group 2
}                   // add the rest of the string (a closing bracket)  

Then you can use $2 in your replace() to extract only the number from the capturing group 2.

var json = {
  "aggregations": {
    "index_types": {
      "doc_count_error_upper_bound": 0,
      "sum_other_doc_count": 0,
      "buckets": [{
        "key": "client",
        "doc_count": 1958205,
        "not_restricted": {
          "doc_count": 976465
        },
        "restricted": {
          "doc_count": 981740
        }
      }, {
        "key": "ultimateparent",
        "doc_count": 1616164,
        "not_restricted": {
          "doc_count": 741059
        },
        "restricted": {
          "doc_count": 875105
        }
      }, {
        "key": "facility",
        "doc_count": 1334963,
        "not_restricted": {
          "doc_count": 914090
        },
        "restricted": {
          "doc_count": 420872
        }
      }]
    }
  }
};

var s = JSON.stringify(json);
// replace all instances of {doc_count: NUMBER} with that NUMBER
console.log(s.replace(/({"doc_count":([0-9]+)})/g, '$2'));

I also noticed that you have properties doc_count outside of restricted and non_restricted. If you only want to replace the ones within a property restricted or non_restricted, You can use this regex:

("((?:not_)?restricted"):{"doc_count":([0-9]+)})

That added part at the beginning basically means "starts with 'restricted' with an optional 'not_' before it"

It's used like:

s.replace(/("((?:not_)?restricted"):{"doc_count":([0-9]+)})/g, '"$2:$3')

var json = {
  "aggregations": {
    "index_types": {
      "doc_count_error_upper_bound": 0,
      "sum_other_doc_count": 0,
      "buckets": [{
        "key": "client",
        "doc_count": 1958205,
        "not_restricted": {
          "doc_count": 976465
        },
        "restricted": {
          "doc_count": 981740
        }
      }, {
        "key": "ultimateparent",
        "doc_count": 1616164,
        "not_restricted": {
          "doc_count": 741059
        },
        "restricted": {
          "doc_count": 875105
        }
      }, {
        "key": "facility",
        "doc_count": 1334963,
        "not_restricted": {
          "doc_count": 914090
        },
        "restricted": {
          "doc_count": 420872
        }
      }]
    }
  }
};

var s = JSON.stringify(json);
console.log(s.replace(/("((?:not_)?restricted"):{"doc_count":([0-9]+)})/g, '"$2:$3'));

NOTE: using ?: for the (?:not_) makes the part in parenthesis a non-capturing group so it doesn't get counted in the capturing groups. The term "restricted" with the optional "not_" before it is now contained in group 2 and the number value is in group 3. We also need to prepend a " and combine the two groups with a :.