Skip to main content

Transform custom types into JSON-compatible data.

Works with
This package works with Bun
This package works with Cloudflare Workers
This package works with Node.js
This package works with Deno
This package works with Browsers
JSR Score
3 weeks ago (3.0.1)

🏯 ZenJSON Travis (.com) npm bundle size Codecov

Transform custom types into JSON-compatible data.

Have you ever tried to do something like this:

JSON.stringify({ createdAt: new Date() });

This works but the date is converted into a string so when you JSON.parse the result you don't get a Date back.

ZenJSON allow you to transform you data into a new one that can safely be stringifyed. Of course you can also do the reverse operation to get back your original data.

Here is a gist:

import { restore, sanitize } from "";

const data = {
  createdAt: new Date(),
  nested: {
    array: [Infinity, undefined, NaN],

// sanitize your data before calling JSON.stringify
const str = JSON.stringify(sanitize(data));

// restore afer calling JSON.parse to get back the original data
const parsed = restore(JSON.parse(str));

assert(parsed.createdAt instanceof Date);
assert(parsed.nested.array[0] === Infinity);


# npm
npm install zenjson
# yarb
yarn add zenjson

You can also use this package in Deno / Recent Browser using ESM import like this:

import { restore, sanitize } from "";

// don't forget to fix the version, for example
// open to find the latest one

Suported data type

By default ZenJSON supports the following types:

  • Any data normally suported in JSON (string, number, boolean, null, object, array)
  • Date
  • Special numbers: Infinity, -Infinity and NaN
  • undefined

How does it works

When you call sanitize, ZenJSON will traverse your data and match it against a list of custom types. When a value matches a type, it will replace the value into a tuple of two elements (the first element is the name of the type, the second is the sanitized value).

For example if you call sanitize(NaN) it will produce the following result: ['number', 'NaN'].

If you then call the restore function with this tuple: restore(['number', 'NaN']) it will return NaN.

Note: Both sanitize and restore create a shallow copy, even if no change is made.

Custom types

You can provide your own custom types and even replace the default ones using the createSanitize and createRestore functions.

These functions take one argument: the list of supported types. The defaults supported types are exposed as defaultTypes.

  • The sanitize function correspond to createSanitize(defaultTypes)
  • The restore function correspond to createRestore(defaultTypes)

If you use one of the create function you probably want to add your own custom type:

import {
} from "";

const sanitize = createSanitize([
  // copy the default types
  // add your own type (keep reading to find out how to create this)

Defining a custom type

Note: If you use TypeScript you can use the ICustomType export to defined you types.

A custom type is an object with the following properties:

  • name: This identify your type. It must be unique. The defaults types uses the following names: undefined, number, array, date.
  • check: A function that receive a value and return true if it has the correct type.
  • sanitize: A function that receive the value and must return a JSON-compatible version of it (for example (date) => date.toISOString()).
  • restore: A function that receive the sanitized value and must return the restored value (for example (dateStr) => parseISO(dateStr)).

Here is an example of a custom type that handles BigInt:

import {
} from "";

const bigintType = {
  name: "bigint",
  check: (val) => typeof val === "bigint",
  sanitize: (val) => val.toString(),
  restore: (val) => BigInt(val),

const sanitize = createSanitize([...defaultTypes, bigintType]);
const restore = createRestore([...defaultTypes, bigintType]);

const data = { num: 123456789123456789123456n };
const sanitized = sanitize(data); // { num: ['bigint', '123456789123456789123456'] }
const restored = restore(sanitized); // { num: 123456789123456789123456n }

Note: The check, sanitize and restore also receive a second parameter: ctx. This is an object that contains some usefull data and functions.

Take a look a the types.ts file to see how the defaults types are implemented.

By the way, each one of the default types are exported and can be used indivdually. They are four types:

  • dateType (name: date) handles Date objects
  • undefinedType (name: undefined) handles undefined values
  • specialNumberType (name: number) handles NaN, Infinity and -Infinity
  • arrayType (name: array) handles special cases (see below)

The arrayType custom type

You might be wonderring what happens if your data looks like one of the sanitized tuple, for example you might have something like this:

const data = {
  keys: ["date", "time"],

This object might cause an error because the restore function will interpret the ['date', 'time'] as a sanitized value and will try to transform it into a date.

To solve this ZenJSON include the arrayType in the list of default types. This type will tranform any array that looks like a sanitized tuple into a tuple ['array', __THE_ARRAY__].

In the example above the result of sanitize(data) is:

const result = {
  keys: ["array", ["date", "time"]],

Which is correctly restored into the original object.

Important: When you define your own types you should always include the arrayType to avoid the problem desribed above.

Handling nested data

ZenJSON handles nested array and object but it will stop as soon as a custom type matches. If you want to sanitize data inside f your custom type, you can use the ctx.sanitize and ctx.restore.

Here is an example of a custom type for Set:

const setType = {
  name: "set",
  check: (val) => val && val instanceof Set,
  sanitize: (val, ctx) => {
    // extract list of values
    const setValues = Array.from(set.values());
    // use ctx.sanitize to sanitize items inside the set
    return => ctx.sanitize(item));
  restore: (val, ctx) => {
    const restoredValues = => ctx.restore(item));
    return new Set(restoredValues);

Using ctx.state

For more advanced use case you can use the ctx.state to store data outside of the sanitized / restored data. This object is a ITypedMap (a WeakMap that use a special key to enforce type safety).

When you call sanitize or restore you can pass a second argument that will be used as the initial state of the ctx.state.

const sanitize = createSanitize([
const state = createTypedMap();
const sanitized = sanitize(data, state);
// do something with the state

Take a look at this test file for an example.

Built and signed on
GitHub Actions
View transparency log

Add Package

deno add @dldc/zenjson

Import symbol

import * as mod from "@dldc/zenjson";

Add Package

npx jsr add @dldc/zenjson

Import symbol

import * as mod from "@dldc/zenjson";

Add Package

yarn dlx jsr add @dldc/zenjson

Import symbol

import * as mod from "@dldc/zenjson";

Add Package

pnpm dlx jsr add @dldc/zenjson

Import symbol

import * as mod from "@dldc/zenjson";

Add Package

bunx jsr add @dldc/zenjson

Import symbol

import * as mod from "@dldc/zenjson";