immutable object validator

Usage no npm install needed!

<script type="module">
  import valleydate from '';



  • you can find all the functions under hoplon.types in the new module.

  • valleydate will still be available, but will not receive any updates.

  • in case newer features are needed in your project, please refactor any code to use hoplon.types instead of valleydate.


npm install valleydate
////| | much install
npm install sourcevault/valleydate#dist

Build Status

valleydate is a functional approach to schema validation that puts composability and extensibility as it's core feature.

  1. Introduction
  2. Initializing Validator
  3. Chainable Functions
  4. Creating Custom Basetypes
  5. Context Variable
  6. Helper Validators
  7. .flatato
  8. common pitfall

.. quick examples ..

🟡 Object with required properties foo and bar.

var IS = require("valleydate")

var V = IS.required("foo","bar")


  continue: false,
  error: true,
  value: {},
  message: [ 'foo', 'bar' ],
  path: [ 'foo' ]

🟡 Object with required properties name age and address, with address having required fields of city and country.

var IS = require("valleydate")

var address = IS.required("city","country")

var V = IS.required("address","name","age")

var sample =


  continue: false,
  error: true,
  value: { name: 'Fred', age: 30, address: { city: 'foocity' } },
  message: [ 'city', 'country' ],
  path: [ 'address', 'country' ]

🟢 Table 1 - method names and their mapping to which underlying type check.

obj            Object
arr            Array
undef          Undefined
bool           Boolean
null           Null
num            Number
str            String
fun            Function
arg            Argument
cont           continue
err            error
alt            alternative


.. why another schema validator ?

  • Monadic chainable functions.

  • custom validators that are easy to build and extend.

valleydate exposes few key operators for creating data validators, for handling arbitrary complex data types.

We start by defining our basetypes:

  • num,arr,str,null,bool,undef,arg,obj and fun.

.. then chainable units :

  • and,or,alt,map,on.

.. and finally consumption units :

  • cont/edit,jam, err and fix.

Initializing Validator

Each validator chain starts with a basetype.

var V = IS.num
V(1) // {continue: true, error: false, value:1}
var V = IS.obj
V({}) // {continue: true, error: false, value:{}}
var V = IS.arr
V([]) // {continue: true, error: false, value:[]}
var V = IS.obj
V([]) // {continue: false, error: true, message:"not an array",path:[]}

The return object will always return .continue, .error and .value. First two are boolean, and will always be opposite in value. The final output is kept in the .value attribute.

⚠️ .value may be modified if consumption units are used in the chain , so be careful. ⚠️

If {cotinue:false,error:true,...} the return object would also have attributes .message and .path, both are Array , with message values :

  • message- that passes along error messages from the validator.
  • path - in case the input is of type array or object, the path within the object where the validator function failed.

Chainable Functions

After initilizating a validator with its basetype, you are returned a unit object that can be chained ( infinitely ) using a few operators.

These operators all accept custom validators but also other valleydate objects.

- and

  • when validators need to be combined, and data has to satisfy conditions set by both validator.

  • a common situation is validating string enums.

var G7 = new Set([

var valG7 = function(s){
  if (G7.has(s)){
   return true
  else {
   return [false,"not in G7"]
var isG7 = IS.str.and(valG7)


//{ continue: true, error: false, value: 'UK' }


/*{ continue: false,
  error: true,
  message: [ 'not in G7' ],
  value: 'Spain'

⛔️ valG7 is a custom validator in the above example, they can be any function that returns boolean or [boolean,string].

- or

  • when validators need to be combined, here data can satisfy either validator.

  • a useful example would be accepting a single string or multiple strings in an array to define ipaddress to use in an application.

var canbeIP = IS.str.or(

- alt

  • functionally similar to or using either condition but the result ( or error ) is merged with upstream validator chain.
var canbeIP = IS.str.or(

- map

⛔️ .map only works for basetype Array, Object and Argument. ⛔️
  • map allows to run validators on each value in an array or object.

  • an example of this would be an object of names with age.

var example = {

A validator for it would look something like this :

var ratifydata =;

- on

⛔️ .on only works for basetype Array, Object and Argument. ⛔️
  • apply validator to specific value in an object or array.

  • if there are multiple on, instead of chaining them, you could just pass an object with the validator for each key.

var V = IS.obj


// Also ...

var V1 = IS.obj.on({foo:IS.num,bar:IS.num})


// Also ...

var V2 = IS.obj.on(["foo","bar"],IS.num)


- cont

Alias: edit

  • accepts functions that run based on output of validation.

  • After validating some data, it needs to be consumed ( if valid ) or throw an error.

  • .cont/edit,jam,fix and err are consumption unit function that can be used to do just that.

  • return value of consumption units are important, they replace some parts of return object.

using the IP example from above :

var sendData = function(data){...}

var data = ["",""]

var V = canbeIP
.cont(sendDate) // <-- only this is called as data is valid

🟡 .cont can be used to making values consistent, using the IP address validator from above :

IS = require("valleydate")

var canbeIP =
.or(IS.str.cont (x) => [x]) // <-- we want string to go inside an array
// so we do not have to do extra prcessing downstream.

var ret = canbeIP.auth("")

//{error: false, continue: true, value: ['']}
//                                           ↑  ↑  ↑
//                                       value is an array

- fix

  • When errors can be dealt with locally without being passed upstream.

  • Used commonly in creating default, using the IP address from above :

IS = require("valleydate")

var canbeIP =
.or(IS.string.cont((x) => [x]))

var ret = canbeIP.auth(null)

console.log(ret) // [""]

- err

  • When validation fails, callback provided to .err is invoked.

  • The return value of .err replaces the .error message to be sent upstream.

- jam

  • jam allows to "jam" (raise an error) within a validation chain.

  • The return value of .jam replaces the .error message to be sent upstream.

Creating Custom Basetypes

In case defaults are not sufficient, clean validators can be easily created.

  1. create a validator function with return types :
  • boolean
  • [boolean,any]
  1. provide it as first argument into valleydate as shown below :
var IS = require("valleydate")

var simpleEmail = function(value){

  var isemail = value.match (/[\w-]+@([\w-]+\.)+[\w-]+/)

  if (isemail) {return true}
  else {return [false,"not a valid email address"] }


var isEmail = IS(simpleEmail)

// isEmail is now a valleydate validator which means it gets

// .and, .or, .cont, .err , .jam and .fix methods.


Context Variable

  • .auth actually accepts any number of arguments.

  • but expects the first argument to be what needs to be validated.

🟡 so, what does valleydate do with the extra arguments ?

  • It simply passes it downstream ( as subsequent ) arguments in case they need them.

  • We refer to these extra arguments as context variables.

  • In cases where .map of .on are used, the context variables are appended with the key value.

🟡 These context variables are useful in two important ways :

  • data needs to be provided to .err to create better error message, it could be things like filename.

  • .map, on modification is index / key dependant.

Helper Validators

Some validators are common enough to be added in core.

  • required - accepts a list of strings and checks if they are not undefined in an object.

  • restricted - checks if object has properties that are restricted to provided keys. examples

  • int - checks if input is a integer

🟡 using int :

var IS = require("valleydate")
//{continue:true,error:false,value:1} //{continue:false,error:true,message:['not an integer']}
//{continue:false,error:true,message:['not an integer']}


  • maybe namespace can be used to validate optional value that conform to a type.

  • The function exposed through maybe.* using :

var IS = require("valleydate")

var V =

V.auth(undefined) // { continue: true, error: false, value: undefined }

V.auth(2) // { continue: true, error: false, value: 2}

V.auth("foo bar")

  continue: false,
  error: true,
  message: [ 'not an integer ( or number )', 'not undefined' ],
  value: 'foo bar'

🟢 All possible primitive and helper function provided in core.

// how to see both helper and primitive validators
> console.log((require("valleydate")))
int.neg              int.pos
list.ofint           list.ofnum
list.ofstr           maybe.arr
maybe.bool           maybe.boolnum          maybe.list.ofint
maybe.list.ofnum     maybe.list.ofstr
maybe.null           maybe.num
maybe.obj            maybe.str
maybe.undef          not.arr
not.null             not.num
not.obj              not.str
not.undef            arg
arr                  bool
boolnum              fun
flatato              null
num                  obj
reqres               required
restricted           str
undef                undefnull


.err function by default gives the raw chain of errors.

flatting it gets quite messy 🤷🏼‍♂️.

valleydate provides a helper function .flatato to smoothly flatten raw error values.

but it requires your messages to follow a specific message passing protocol :

  • error value should always be an array.

  • first value of said array should always be a string that starts with a colon ":".

  • to help with sorting, a number can be provided after a second colon to tell flatato the hierarchy of your messages.

// Examples of message that flatato matches against
  [' value is not tuple type.']

  ['length',' value is not tuple type.']

  ['innertype',' value is not tuple type.']

.. common pitfall ..

  1. why does mutating variable in function does not change it downstream ?

each value is rewritten at every return, so for example using context variable to try and change a value will lead to confusing output.

# .. in livescript instead of javascript ..

V = be.obj.on \foo,
  (foo,__,data) -> = "i got changed !"

data = {foo:void}

torn = (V data,data).value

console.log torn #{foo:undefined} 🡐 ( wont change, can't change )

It's one of the trade off of having hidden mutability, it's easy to avoid such "bugs" by restricting the use of the chainable functions for their stated purpose ( e.g don't use .and to edit variables, use .edit instead ).


  • Code released under BSD-3-Clause Licence.
  • Documentation and Images released under CC BY-NC-ND 4.0.
  • details can be found here.