Serializes and deserializes otherwise valid JSON objects containing circular references into and from a specialized JSON format.

Smaller, faster, and able to produce on average a reduced output too, flatted is the new, bloatless, ESM and CJS compatible, circular JSON parser.

It has now reached V1 and it implements the exact same JSON API.

Please note CircularJSON is in maintenance only and flatted is its successor.

A usage example:

var object = {};
object.arr = [
  object, object
];
object.arr.push(object.arr);
object.obj = object;

var serialized = CircularJSON.stringify(object);
// '{"arr":["~","~","~arr"],"obj":"~"}'
// NOTE: CircularJSON DOES NOT parse JS
// it handles receiver and reviver callbacks

var unserialized = CircularJSON.parse(serialized);
// { arr: [ [Circular], [Circular] ],
// obj: [Circular] }

unserialized.obj === unserialized;
unserialized.arr[0] === unserialized;
unserialized.arr.pop() === unserialized.arr;

A quick summary:

• new in version 0.5, you can specify a JSON parser different from JSON itself. CircularJSON.parser = ABetterJSON; is all you need.
• uses ~ as a special prefix symbol to denote which parent the reference belongs to (i.e. ~root~child1~child2)
• reasonably fast in both serialization and deserialization
• compact serialization for easier and slimmer transportation across environments
• tested and covered over nasty structures too
• compatible with all JavaScript engines

npm install --save circular-json

'use strict';

var
  CircularJSON = require('circular-json'),
  obj = { foo: 'bar' },
  str
;
  
obj.self = obj;
str = CircularJSON.stringify(obj);

There are no dependencies.

• Global: <build/circular-json.js>
• AMD: <build/circular-json.amd.js>
• CommonJS: <build/circular-json.node.js>

(generated via gitstrap)

<script src="build/circular-json.js"></script>

'use strict';

var CircularJSON = window.CircularJSON
  , obj = { foo: 'bar' }
  , str
  ;
  
obj.self = obj;
str = CircularJSON.stringify(obj);

NOTE: Platforms without native JSON (i.e. MSIE <= 8) requires json3.js or similar.

It is also a bad idea to CircularJSON.parse(JSON.stringify(object)) because of those manipulation used in CircularJSON.stringify() able to make parsing safe and secure.

As summary: CircularJSON.parse(CircularJSON.stringify(object)) is the way to go, same is for JSON.parse(JSON.stringify(object)).

It's the same as native JSON, except the fourth parameter placeholder, which circular references to be replaced with "[Circular]" (i.e. for logging).

• CircularJSON.stringify(object, replacer, spacer, placeholder)
• CircularJSON.parse(string, reviver)

Bear in mind JSON.parse(CircularJSON.stringify(object)) will work but not produce the expected output.

The module json-stringify-safe seems to be for console.log() but it's completely pointless for JSON.parse(), being latter one unable to retrieve back the initial structure. Here an example:

// a logged object with circular references
{
  "circularRef": "[Circular]",
  "list": [
    "[Circular]",
    "[Circular]"
  ]
}
// what do we do with above output ?

Just type this in your node console: var o = {}; o.a = o; console.log(o);. The output will be { a: [Circular] } ... good, but that ain't really solving the problem.

However, if that's all you need, the function used to create that kind of output is probably faster than CircularJSON and surely fits in less lines of code.

So here the thing: circular references can be wrong but, if there is a need for them, any attempt to ignore them or remove them can be considered just a failure.

Not because the method is bad or it's not working, simply because the circular info, the one we needed and used in the first place, is lost!

In this case, CircularJSON does even more than just solve circular and recursions: it maps all same objects so that less memory is used as well on deserialization as less bandwidth too! It's able to redefine those references back later on so the way we store is the way we retrieve and in a reasonably performant way, also trusting the snappy and native JSON methods to iterate.