jointz

Zero dependency universal TypeScript validation library. Similar interface to Joi but without all the bloat, and built for browsers and node with zero dependencies.

Unlike Joi, the goal is to write validation code only once, in the code where it's used, and from that can export all the necessary information (TypeScript types, JSON schema).

Features

• Written in TypeScript
• Zero dependencies, tiny, and well tested
• Pretty fast
• Infer TypeScript types from validators
• Produce JSON schema from validators
• Supports any, string, number, array, tuple, constant, boolean, or and object validation
• Implement your own validator and use it with any of the other validators
• Fluid immutable interface
• Targets both browsers and node

Installation

You can install it via npm

npm i --save jointz

or yarn

yarn add jointz

Usage

Import the default export from jointz

import jointz from 'jointz';

Then use it to construct validators

const ThingValidator = jointz.object({
  id: jointz.string().uuid(),
  name: jointz.string().minLength(3).maxLength(100)
}).requiredKeys(['id', 'name']);

The validator can now be used to check for errors. Validator#validate returns an array of validation errors, which is empty if the value passes validation.

const myObject = { id: 'abc', name: 'hello world!' };

const errors = ThingValidator.validate(myObject); // expect an error because id is not a uuid

if (errors.length) {
  // Fail
} else {
  // Continue
}

You can also generate TypeScript types from your validators, using Infer, or get the resulting type from validation via checkValid.

type Thing = Infer<typeof ThingValidator>;

const myObject: unknown = { id: 'abc', name: 'hello world!' };

try {
  const thing: Thing = ThingValidator.checkValid(myObject);
} catch (validationError) {
  console.log(validationError.errors);
}

jointz validators also expose the type guard function isValid on every validator.

const myObject: unknown = { id: 'abc', name: 'hello world!' };

if (ThingValidator.isValid(myObject)) {
    // This works because myObject is a valid Thing
    const id: string = myObject.id;
}

JSON Schema

All validators expose a method #toJsonSchema(): JSONSchema7 which return the JSON schema corresponding to the given type.

Errors

Errors match the following interface:

interface ValidationError {
  // Array of keys indicating where the validation failed. This is empty if top level validation failed.
  path: Array<string | number>;
  // The error message describing the failed validation.
  message: string;
  // The value that failed validation. This value may not be defined, e.g. in the case of missing required keys.
  value?: unknown;
}

A single validator can produce many errors. However, validators will only produce relevant errors, e.g. a number validator that checks a number is a multiple of 2 will not produce an additional error about the multiple when validating a string.

Validator	Example	Error.path	Error.message	Error.value
jointz.number()	'abc'	[]	'must be a number'	'abc'
jointz.string()	3	[]	'must be a string'	3
jointz.object({ abc: jointz.string() })	{abc:3}	['abc']	'must be a string'	3
jointz.object({ arr: jointz.array(jointz.number().multipleOf(2)) })	{arr:[2,'5',8]}	['arr',1]	'must be a number'	'5'
jointz.object({ arr: jointz.array(jointz.number().multipleOf(2)) })	{arr:[2,5,8]}	['arr',1]	'number was not a multiple of 2'	5