validate-any

Validate Any

Validate Any is a type validator mainly for Typescript (also works with Javascript) available here. Ever faced those issues where you're trying to make sure a type any is an instance of an interface? This is the issue this package was designed to solve. With this package, you can safely assert the type for an object and return customised errors if the types are incorrect.

Motivation

I hate it when I am unable to validate the request types of JSON objects in Express requests. I wanted to be able to reject a request before my request handlers used the request body. I also didn't like casting the type of the body in the request.

Features

• validate function to validate the data type of a variable
• Validate Any can read the function types of
  • Strings
  • Numbers
  • Booleans
  • Undefined
  • Null
  • Lists
  • Objects
  • Classes
• Validate Any is built to automatically return you the typescript annotations generated by the schema you pass, in the data property
• Detailed error messages to tell you where any validation errors occured
• API Middleware to simplify its usage with Express or NextJS

Installation

With yarn

$ npm install validate-any

With npm

$ yarn add validate-any

Usage

Import the module like this:

// Typescript
import { validate } from "validate-any"

// Javascript
const { validate } = require("validate-any")

How the validate function works

The validate function is the main validator in this package. It is a function which can take 2 (3: optional) parameters.

validate parameters:

Number	Type	Description
1	any	The object we are checking
2	Validator	A specific rule to compare the object to (more about this later)
3	string?	The name of the root object when logs display errors. Defaults to * as root

The type Validator is something you don't need to worry about. Just know that it is the Typescript type for a rule. Rules can look like STRING() or NUMBER().

validate returns an object containing:

Property	Type	Description
success	boolean	Whether the validation of the object was a success or failure. true if success, false if failure
errors	iValidationError[]	The list of corrections to make if any
data	T	The data you passed in with type annotations from the schema you passed in

iValidationError is the data type for an error

Property	Type	Description
location	string	Where in the object did an error occur. E.g. "* > settings > wallpaper"
message	string	Description of what the error is
expected	string	The expected type of value
value	T	The value given

In the examples below, you will see in the error object I only show the error message and display a "..." after it.
This is because I don't want to display unnecessary information.
However, every single error will have all the properties defined here

Also do note that the location started with a *.
This is the name of the root defined here

Making a rule

There are infinite combinations of rules we can make. The complexity of the rule only depends on how much code you are willing to write.

Validation basics with STRING()

Here is how to validate a string

console.log(validate("string", STRING()))
// > { success: true, errors: [], data: "string" }

console.log(validate(0, STRING()))
// > {
// >      success: false,
// >      errors: [ { message: 'Value is not of the correct type', ... } ],
// >      data: undefined
// > }

The function STRING with nothing in the parameters represents a type string. This method can also take in items in the parameters:

Type	Description
(empty)	validates if the input is a string
RegExp	validates if the input is a string and matches the RegExp
...string[]	validates if the input is a string and matches any of the given strings

import { validate, STRING } from "validate-any"

console.log(validate("string", STRING()))
// > { success: true, errors: [], data: "string" }
// This returned true because "string" is a string

console.log(validate(0, STRING()))
// > {
// >      success: false,
// >      errors: [ { message: 'Value is not of the correct type', ... } ],
// >      data: undefined
// > }
// This returned false because 0 is not a string

console.log(validate("string", STRING(/^string$/)))
// > { success: true, errors: [], data: "string" }
// This returned true because "string" matches the RegExp /^string$/

console.log(validate("string", STRING(/^something-else$/)))
// > {
// >      success: false,
// >      errors: [ { message: 'Value does not match the defined RegExp pattern', ... } ],
// >      data: undefined
// > }
// This returned false because "string" didn't match the RegExp /^something-else$/

console.log(validate("string", STRING("does", "the", "string", "match", "?")))
// > { success: true, errors: [], data: "string" }
// This returned true because "string" was passed into STRING() as a parameter

console.log(validate("string", STRING("doesn't", "match"), "my-string"))
// > {
// >      success: false,
// >      errors: [ { message: `Value doesn't match anything in the defined set of strings`, ... } ],
// >      data: undefined
// > }
// This returns false because "string" wasn't passed into STRING() as a parameter

Validating a number with NUMBER()

Because STRING() validates strings, it's obvious that NUMBER() validates numbers. NUMBER() works the same was as STRING() except allows a different set of parameters:

Type	Description
(empty)	validates if the input is a number
...number[]	validates if the input is a number and matches any of the given numbers

import { validate, NUMBER } from "validate-any"

console.log(validate(3, NUMBER()))
// > { success: true, errors: [], data: 3 }

console.log(validate("string", NUMBER()))
// > {
// >      success: false,
// >      errors: [ { message: 'Value is not of the correct type', ... } ],
// >      data: undefined
// > }

console.log(validate(3, NUMBER(1, 2, 3, 4, 5)))
// > { success: true, errors: [], data: 3 }

console.log(validate(3, NUMBER(6, 7, 8, 9, 10)))
// > {
// >      success: false,
// >      errors: [ { message: 'Value doesn't match anything in the defined set of numbers', ... } ],
// >      data: undefined
// > }

Validating a boolean with BOOLEAN()

BOOLEAN() allows comparison of booleans only

Type	Description
(empty)	validates if the input is a boolean
boolean	validates if the input is a boolean and if the booleans are equal

import { validate, BOOLEAN } from "validate-any"

console.log(validate(true, BOOLEAN()))
// > { success: true, errors: [], data: true }

console.log(validate("string", BOOLEAN()))
// > {
// >      success: false,
// >      errors: [ { message: 'Value is not of the correct type', ... } ],
// >      data: undefined
// > }

console.log(validate(true, BOOLEAN(true)))
// > { success: true, errors: [], data: true }

console.log(validate(false, BOOLEAN(true)))
// > {
// >      success: false,
// >      errors: [ { message: 'Value is not allowed', ... } ],
// >      data: undefined
// > }

Validating null with NULL()

NULL() doesn't allow variations of the parameters

Type	Description
(empty)	validates if the input is a null

import { validate, NULL } from "validate-any"

console.log(validate(null, NULL()))
// > { success: true, errors: [], data: null }

console.log(validate(undefined, NULL()))
// > {
// >      success: false,
// >      errors: [ { message: 'Value is not allowed', ... } ],
// >      data: undefined
// > }

Validating undefined with UNDEFINED()

Just like NULL(), UNDEFINED() doesn't allow variations of the parameters

Type	Description
(empty)	validates if the input is a undefined

import { validate, UNDEFINED } from "validate-any"

console.log(validate(undefined, UNDEFINED()))
// > { success: true, errors: [], data: undefined }

console.log(validate(null, UNDEFINED()))
// > {
// >      success: false,
// >      errors: [ { message: 'Value is not allowed', ... } ],
// >      data: undefined
// > }

Validating a list with LIST()

This one's a bit more complicated. LIST() allows a few sets of parameters:

Type	Description
(empty)	validates if the input is a list
...Validator[]	validates if the input is a list and checks if all items in the list match at least 1 of the Rules stated

import { validate, LIST, STRING, NUMBER } from "validate-any"

console.log(validate([1, 2, 3, 4, 5], LIST()))
// > { success: true, errors: [], data: [ 1, 2, 3, 4, 5 ] }

console.log(validate({ property: "value" }, LIST()))
// > {
// >      success: false,
// >      errors: [ { message: 'Value is not of the correct type', ... } ],
// >      data: undefined
// > }

console.log(validate(["one", "two", "three"], LIST(STRING())))
// > { success: true, errors: [], data: [ "one", "two", "three" ] }

console.log(validate([1, "two", 3], LIST(NUMBER())))
// > {
// >      success: false,
// >      errors: [ { message: 'Value is not of the correct type', ... } ],
// >      data: undefined
// > }

console.log(validate([1, "two", []], LIST(STRING(), NUMBER(), LIST())))
// > { success: true, errors: [], data: [ 1, "two", [] ] }
// And yes we also can verify LIST() within LIST()

console.log(validate([1, "two", null], LIST(STRING(), NUMBER())))
// > {
// >      success: false,
// >      errors: [ { message: 'Value is not of the correct type', ... } ],
// >      data: undefined
// > }

const usernames = ["jack", "_jack", "-jack"]
console.log(validate(usernames, LIST(STRING(/^[a-zA-Z]/))))
// > {
// >     success: false,
// >     errors: [
// >         { message: 'Value is not of the correct type', ... },
// >         { message: 'Value is not of the correct type', ... }
// >     ],
// >     data: undefined
// > }
// Not every username matched /^[a-zA-Z]/

const codes = [34, 76, 92]
console.log(validate(codes, LIST(NUMBER(34, 76, 92))))
// > { success: true, errors: [], data: [ 34, 76, 92 ] }
// Every code matched items in [34, 76, 92]

This way, we can make checking of list types much more detailed

Validating an object with OBJECT()

We can use OBJECT() to validate objects. OBJECT() only allows 1 optional parameter which maps out what the properties will look like

import { validate, OBJECT } from "validate-any"

console.log(validate({ property: "value" }, OBJECT()))
// > { success: true, errors: [], data: { property: "value" } }
// Since { property: "value" } is an object, validate() returned true

console.log(validate({ property: "value" }, OBJECT({})))
// > {
// >     success: false,
// >     errors: [ { message: 'Object has unknown property which is defined', ... } ],
// >     data: undefined
// > }
// The rule of {} means the object must have no properties

console.log(validate({ property: "value" }, OBJECT({ property: STRING() })))
// > { success: true, errors: [], data: { property: "value" } }
// We set the OBJECT's params to an object with a property "property" and a value "value"
// Since "value" matches STRING(), validate() returned true

console.log(
	validate(
		{
			property: "value"
		},
		OBJECT({
			prop: STRING()
		})
	)
)
// > {
// >     success: false,
// >     errors: [
// >         { message: 'Object requires this property but is missing', ... },
// >         { message: 'Object has unknown property which is defined', ... }
// >     ],
// >     data: undefined
// > }
// Since there is no property for the type validation "prop", we got an error
// Since there is no type validation for the property "property", we got an error

console.log(
	validate(
		{
			property: "value",
			layer: {
				deepProperty: ["", 0, null, undefined, false]
			}
		},
		OBJECT({
			property: STRING(),
			layer: OBJECT({
				deepProperty: LIST(
					STRING(),
					NUMBER(0),
					NULL(),
					UNDEFINED(),
					BOOLEAN(false)
				)
			})
		})
	)
)
// > {
// >      success: true,
// >      errors: [],
// >      data: { property: 'value', layer: { deepProperty: [Array] } }
// > }
// We can even nest OBJECT() in OBJECT()

Validating the or operation with OR()

If you want either of a few rules to match, use the OR() operator. This function takes multiple parameters:

Type	Description
...Validator[]	A list of rules to test on the input

import { validate, OR, STRING, NUMBER, BOOLEAN } from "validate-any"

console.log(validate("string", OR()))
// > Error: Expected developer to provide at least 1 rule for the OR operation
// An OR operation only works with at least one input

console.log(validate("string", OR(STRING(), NUMBER())))
// > { success: true, errors: [], data: "string" }

console.log(validate("string", OR(BOOLEAN(), NUMBER())))
// > {
// >     success: false,
// >     errors: [ { message: 'Value does not match any of the validators defined', ... } ],
// >     data: undefined
// > }

Using withValidBody with Express or Next

You can also import the module as a middleware to be used with Express or Next. This way, you can verify the types of the req.body before invalid types mess your code up

import { OBJECT, STRING, withValidBody } from "validate-any"

// Express
app.post(
	"/body",
	withValidBody(OBJECT({ usnm: STRING(), pswd: STRING() }), (req, res) => {
		const { usnm, pswd } = req.body
		console.log(`Username: ${usnm}`, `Password: ${pswd}`)
		res.end()
	})
)

// Next
export default withValidBody(
	OBJECT({ usnm: STRING(), pswd: STRING() }),
	(req, res) => {
		const { usnm, pswd } = req.body
		console.log(`Username: ${usnm}`, `Password: ${pswd}`)
		res.end()
	}
)

The withValidBody takes in 2 parameters:

Number	Type	Description
1	Validator	Rule to compare the req.body with
2	handler	Handler to handle the request if it works

Because of the middleware, in Typescript you can now safely use type assertions. Also, now for both Typescript and Javascript, you can safely use the variables like they are the defined types and not have to worry about invalid types crashing your server!

Using withValidQuery with Express or Next

The withValidQuery works the same way as withValidBody but it validates the req.query instead.

Testing

Validate Any's individual validators are tested with Jest. To run the tests, run the command

$ npm run test

Built with

• TypeScript
  • 
  • 
• Jest
  • 
  • 
  • 
  • 
• Miscellaneous
  •