TypeScript icon, indicating that this package has built-in type declarations

5.0.0 • Public • Published


Data structure and rule validation engine. Robust model schema for JS objects.

npm version Code Climate Travis

Universal, ultra fast and lightweight (4Kb!), Skematic enables you to design, format and validate data according to rules and conditions specified as simple config models, for browser and Node.js.

  • Design: structure your data models as config objects
  • Format: transform, generate and modify data structures
  • Validate: check that arbitrary data conforms to rules

A basic example:

// -- Define a simple data structure
const Hero = {
  name:    {rules: {minLength: 4}, errors: 'Bad name!'},
  shouts:  {transform: val => val.trim().toUpperCase()},
  skill:   {default: 3, required: true, rules: {isNumber: true}},
  updated: {generate:}
// -- Format some data
Skematic.format(Hero, {shouts: '  woo   '})
// {shouts: 'WOO', skill: 3, updated: 1426937159385}
// -- Validate an object
Skematic.validate(Hero, {name: 'Zim'})
// {valid: false, errors: {name: ['Bad name!'], skill: ['Failed: required']}}

Also fully supports Typescript:

interface ISimpleHero {
  name: string
const SimpleHero: Skematic.Model<ISuperHero> {
  name{ requiredtrue },
// Typescript Error:
// Object literal may only specify known properties, 
// and 'sOmeJUNKK' does not exist in type ISimpleHero


npm install --save skematic

Import to your project:

// CommonJS modules
const Skematic = require('skematic')
// OR using ES6 Module imports
import Skematic from 'skematic'

To use in a browser:

<script src="node_modules/skematic/build/skematic.min.js"></script>

Compatibility Note: Skematic is written in ES6 but compiled down to ES5 and works across all modern browsers (IE9+, Chrome, Firefox, Safari evergreens). Please note that the ES5 Object.keys() method is not supported by IE7 & 8, so to use Skematic in these fossil browsers, you'll need to install es5-shim (and worship Satan 🤘).


The API surface is small by design, with two primary methods:

  • .format( model, data [, opts]) - see Format
  • .validate( model, data [, opts] ) - see Validate


Model configuration

Skematic provides keys to define rules and conditions for your data model. Config keys are all optional.


  • default {any} value to apply if no value is set/passed
  • lock {Boolean} disallows/strips value on format (unlock format opts to override)
  • transform {Function} a function to transform a value (not called if value is undefined or null)
  • generate {Object|Function} enables computing a value from functions
  • show {String|Array} string scopes required to show field on format (hides if not met)
  • write {String|Array} scopes required to validate this field being set (fails validation if scopes aren't matching)
  • model {Object} declare sub-model defining this value (see "Sub-model")


  • rules {Object} validation rules: {rules: {min: 3, max: 11}}
  • errors {Object|String} error messages for rules
  • required {Boolean} flag if property MUST be set and/or provided


  • allowNull {Boolean} Accept null values (no other validation applied) or set to false to force a NOT NULL condition (no undefined or null values permitted). Designed to:
    • a) false: enables setting required (which ordinarily passes for null) while disallowing null as a value.
    • b) true: enables accepting null without triggering any other rule validation (ie. 'null' becomes a valid value)
  • primaryKey {Boolean} flag to indicate whether this field is the primary key (id field), used in conjunction with mapIdFrom format option to allow transposing your datastore id to some other field on your data model (eg. Mongo's _id can be mapped to the field you set primaryKey: true on)

Note: See format()'s order of execution for which formatting changes get applied in what order.

Simple examples

A basic data model:

const Hero = {
  name: {
    default: 'Genericman',
    required: true,
    rules: {maxLength: 140, minLength: 4},
    errors: {maxLength: 'Too long', minLength: 'Shorty!'}
// Generate a record by passing null/undefined to `format(Model, null)`
// -> {name: 'Genericman'}
Skematic.validate(Hero, {name: 'Spiderman'})
// -> {valid: true, errors: null}
Skematic.validate(Hero, {name: 'Moo'})
// -> {valid: false, errors: {name: ['Shorty!']]}}

Typically you'll create a more complete data model to represent your application objects, with several fields to format and validate:

const Hero = {
  name: HeroNameField,
  skill: {default: 0}
Skematic.validate(Hero, {name: 'Spiderman', skill: 15})
// -> {valid: true, errors: null}
Skematic.validate(Hero, {name: 'Moo'})
// -> {valid: false, errors: {name: ['Shorty!']}


Several validation rules are built in. Custom rules are defined as functions that receive the field value and return pass/fail (true/false). Notably, 'required' is passed as a property option, rather than a rule.

Important: rules ONLY run when the value of the field is defined (i.e. NOT undefined). If a value is undefined on your data, no rules are applied. You can force a value to be provided by add the required: true flag to your model.

The other available validators are:

  • .min - The lowest permitted number
  • .max - The highest permitted number
  • .minLength - The shortest string permitted
  • .maxLength - The longest string permitted
  • .eq - Value must be strictly equal
  • .neq - Value must not equal
  • .oneOf - Value must be one of the values in the list of elements
  • .notOneOf - Value must NOT be in the list of elements
  • .has - List of elements contains the value
  • .hasNot - List of elements does NOT contain the value
  • .isEmail - no parameters: Is the string an email
  • .isUrl - no parameters: Is the string a URL
  • .isAlpha - no parameters: Checks value is an ALPHA string (abcd...)
  • .isAlphaNum - no parameters: Checks value is AlphaNumeric (abc..012..9)
  • .isNumber - no parameters: Checks value is a Number (NaN fails this test)
  • .isString - no parameters: Checks value is of type String
  • .match - String must match regexp
  • .notMatch - String must NOT match regexp
  • .isEmpty - set to true to check the value is empty
  • .notEmpty - set to true to check the value is not empty

Custom rules can be applied by providing your own validation functions that accept a value to test and return a Boolean (pass/fail).

Note: The required rule has a special shorthand to declare it directly on the model:

const modelProp = {default: 'Boom!', required: true}

Declare rules key as follows:

const User = {
  name: {
    rules: {minLength: 5}
Skematic.validate(User, {name: 'Zim'})
// -> {valid: false, errors: {name: ['Failed: minLength']}}
Skematic.validate(User, {name: 'Bunnylord'})
// -> {valid: true, errors: null}

Custom Rules

You can mix in Custom rules that have access to the rest of the data model via this. For example:

const User = {
  name: {
    rules: {
      // A built in validation
      minLength: 5,
      // Your own custom validator (accepts `value` to test, returns Boolean)
      // Note: MUST use `function () {}` notation to access correct `this`
      onlyFastBunnylord: function myCustomCheck (value) {
        // See us access the `speed` prop in our check:
        return value === 'Bunnylord' && this.speed > 5
  speed: {default: 5}
// Wrong name
Skematic.validate(User, {name: 'Zim', speed: 10})
// -> {valid: false, errors: {name: ['Failed: minLength', 'Failed: onlyFastBunnylord']}}
// Too slow!
Skematic.validate(User, {name: 'Bunnylord', speed: 3})
// -> {valid: false, errors: {name: ['Failed: onlyFastBunnylord']}}
Skematic.validate(User, {name: 'Bunnylord', speed: 10})
// -> {vaid: true, errors: null}

Custom error messages

Custom error messages can be declared per rule name: {errors: {'$ruleName': 'Custom message'}}

Provide a default message if no specific error message exists for that rule:

  errors: {
    max: 'Too large',
    default: 'Validation failed'

Usage example:

const User = {
  name: {
    rules: {minLength: 5},
    errors: {minLength: 'Name too short!'}
// Using a value test:
Skematic.validate(, 'Zim')
// -> {valid:false, errors:['Name too short!']}
// Using a keyed object value test:
Skematic.validate(User, {name:'Zim'})
// -> {valid:false, errors:{name:['Name too short!']}}

Note: You can create error messages for custom rules too. Just use the same key you used to define the custom rule. {rules: {myCustom: val => false}, errors: {myCustom: 'Always fails!'}}

Rules can be combined, and you can declare a string message on errors to apply to any and all errors:

const User = {
  name: {
    rules: {minLength: 5, maxLength: 10},
    errors: 'Name must be between 5 and 10 characters'


Computed values - Skematic keys can generate values using functions referenced in the generate directive.

The simplest usage is to specify generate as a function:

{generate: () =>}

You may also pass generate a config object with properties:

Legend: field - {Type} default: Description

  • ops {Array} of fn objects {fn [, args]) or functions. The first function in the list is passed the value of the object being formatted. The output of each function is passed as the first parameter of the next.
  • preserve {Boolean} false: OPTIONAL Preserves a provided value and does not overwrite if set to true. (If left as false, generate will always replace the provided value). Note: undefined values treated as being NOT SET - use null to pass 'no value'
  • require {Boolean false: OPTIONAL Ensures that value is only generated if the field exists on the provided data.
  • once {Boolean} false: OPTIONAL Flag this field to only generate if .format() is called with the option once:true. Useful for fields like "created".

Unless instructed otherwise (via flags) generate will compute a value every time and overwrite any provided value. To preserve any provided value set preserve: true (note that undefined is treated as not set, use null to provide a no-value). To only generate a value when the key for that field is provided, set require: true. To manually run generators based on a flag provided to format, set {once: true} on the model field, (and run format(Model, data, {once: true}).


const Hero = {
  updated: {
    generate: {
      // The ops array lists fn objects or functions
      ops: [
        // A fn object specifies `fn` and `args`
        {fn: myFunc, args: []},
        // , {fn...}, etc etc
        // And here is a raw function with no args, it will be passed
        // the output of the last `fn` as its first parameter
      // Optional flag: preserves a provided value
      // (default: false)
      preserve: false,
      // Optional flag: ONLY generate if provided a field on data
      // (default: false)
      require: false,
      // Optional flag: Require passing {once:true} to format to compute value
      // (default: false)
      once: true

That looks like a mouthful - but if we pass the raw functions and assume default settings for the other flags, the above collapses to:

const Hero = {
  updated: {generate: {ops: [myFunc, anotherFn], once: true}}


A property can be formatted to another model (essentially, a complex object), or array of models.

// A "post" would have comments made up of `owner_id, body`
const Post = {
  comments: { 
    model: {
      owner_id: {lock: true},
      body: {rules: {minLength: 25, }}
// Or, a simple scalar array of "tags" (an array of strings):
const Picture = {
  url: {rules: {isURL: true}},
  tags: {model: {rules: {minLength: 3}}}

All the model validations and checks assigned to the sub-model (comments) will be correctly cast and enforced when the parent (post) has any of its validation routines called.


A model can declare any one of its fields as the primary key (the id field) to be used for its data objects. This can be used in conjunction with Skematic.format() in order to modify an incoming data collection and map a pre-existing id field (say for example "_id") to the primaryKey.

This is useful for data stores that use their own id fields (eg. MongoDB uses '_id').

const propSchema = {
  prop_id: {primaryKey: true},
  name: {type: Skematic.STRING}
// Example default results from data store:
let data = [{_id: '512314', name: 'power'}, {_id: '519910', name: 'speed'}]
Skematic.format(propSchema, {mapIdFrom: '_id'}, data)
// -> [{prop_id: '512314', name: 'power'}, {prop_id: '519910', name: 'speed'}]

Note: Your data store might automatically use a particular field name for its identifying purposes (usually "id"). If you know you're using a datastore that defaults its id field to a given key, you can simply reuse this field name in your model. Specifying primaryKey is simply a way to force data models into using a given key.


Format creates and returns a conformed data structure based on the model and input data provided.

Side-effect free, format never mutates data

Skematic.format(model [, data] [, opts])
// -> {formattedData}

Special case: Passing format no data will cause format to create blank record based on your model format(model), including defaults and generated fields. You can pass options too, as follows: format(model, null, {defaults: false})


  • model: The model to format against
  • data: The data object to format. If null or undefined, format will attempt to create data to return
  • opts: [Optional] options hash (see below)
Skematic.format(Hero) // create a data block
// -> {name: 'Genericman'}
Skematic.format(Hero, {name: 'Zim'})
// -> {name: 'Zim'}
// Or with options
Skematic.format(Hero, {name: 'Zim', junk: '!'}, {strict: true})
// -> {name: 'Zim'}

Format options

Format options include:

Legend: field - {Type} - default: Description

  • scopes - {String|Array} - undefined: List of scopes that toggle .show model fields on format() (See validate() for .write scopes)
  • unscope - {Boolean} - false: Ignores 'show' of scopes (ie. shows all fields)
  • strict - {Boolean} - false: Strips any fields not declared on model
  • sparse - {Boolean} - false: Only process fields on the provided data, rather than all fields on the entire model
  • defaults - {Boolean} - true: Set default values on 'empty' fields. Toggle to false to disable.
  • generate - {Boolean} - true: Enable/disable generating new values - see Design:generate
  • once - {Boolean} - false: Run generator functions set to {once: true} - see Design:generate
  • transform {Boolean} - true: Toggle to false to cancel modifying values
  • unlock - {Boolean} - false: Unlocks 'lock'ed model fields (ie. no longer stripped, allows for overwriting).
  • strip - {Array} - []: Remove fields with matching values from data
  • mapIdFrom - {String} - undefined: Maps a primary key field from the field name provided (requires a primaryKey field set on the model)

Format order of updates

Format applies these options in significant order:

  1. scopes: Checks scope match - hides field if the check fails
  2. lock: Strip locked fields (unless {unlock: true} provided)
  3. sparse: Only processes keys on the provided data (not the whole model)
  4. defaults: Apply default values
  5. generate: Compute and apply generated values
  6. transform: Run transform functions on values
  7. strip: Removes field with matching values after all other formatting
  8. mapIdFrom: Sets the id field on data to be on the 'primaryKey'

Meaning if you have an uppercase transform, it will run AFTER your generate methods, thus uppercasing whatever they produce.

Format examples:

const myModel = {
  mod_id: {primaryKey: true},
  rando: {generate: {ops: Math.random, once: true}},
  power: {default: 5},
  name: {default: 'zim', transform: val => val.toUpperCase()},
  secret: {show: 'admin'}
Skematic.format(myModel, {}, {once: true})
// -> {rando: 0.24123545, power: 5, name: 'ZIM'}
Skematic.format(myModel, {}) // (model, data)
// -> {power: 5, name: 'ZIM}
Skematic.format(myModel, {}, {defaults: false})
// -> {}
Skematic.format(myModel, {rando: undefined, power: 'x'}, {strip: [undefined, 'x']})
// -> {name: 'ZIM'}
Skematic.format(myModel, {name: 'Zim', secret: 'hi!'}, {scopes: ['admin']})
// -> {name: 'ZIM', secret: 'hi!'}
Skematic.format(myModel, {name: 'Zim', secret: 'hi!'}, {scopes: ['not:admin']})
// -> {name: 'ZIM'}
Skematic.format(myModel, {name: 'Gir'}, {sparse: true})
// -> {name: 'GIR'}
Skematic.format(myModel, {_id: '12345'}, {mapIdFrom: '_id'})
// -> {mod_id: '12345', power: 5, name: 'ZIM'}


Validation applies any rules specified in the model fields to the provided data and returns an object {valid, errors}:

Skematic.validate(model, data [, opts])
// -> {valid: <Boolean>, errors: {$key: [errors<String>]} | null}


  • model: The model to validate against
  • data: The data object to validate
  • opts: [Optional] options hash (see below)
Skematic.validate(Hero, {name: 'Zim'})
// Or with options
Skematic.validate(Hero, {name: 'Zim'}, {sparse: true})

Returns an object {valid: $boolean, errors: $object|$array|null} where the errors key may be:

  • null - no errors
  • array - of errors if validating a scalar (string, number, etc)
  • object - hash of errors when validating a data object

Validate options include:

Legend: field - {Type} - default: Description

  • scopes - {String|Array} - undefined: List of scopes that will be tested against .write model fields for matches. Errors if scopes don't meet.
  • unscope - {Boolean} - false: Ignores any scope requirements on the model
  • strict - {Boolean} - false: Validates that all keys provided by data are defined on the model as well as valid (prevents validating/accepting extraneous fields)
  • sparse - {Boolean} - false: Only process fields on the provided data, rather than all fields on the entire model. This will skip required fields on your model if those fields are not present on your data. Can be useful for only validating subsets of models.
  • keyCheckOnly - {Boolean} - false: Overrides normal validation and ONLY checks user data keys are all defined on model. Useful to ensure user is not sending bogus keys. @see Format options: strict to simply strip unknown keys.


Skematic is written in ES6+.

Developing Skemetic requires installing all dependencies:

npm install

Run the tests:

npm test

Note: Generated API docs can be found in the npm installed package under docs/index.html. Otherwise generate them using npm run docs

Benchmarks: The perf/benchmark.js is simply a check to ensure you haven't destroyed performance: npm run benchmark. Skematic runs at several tens of thousands of complex validations per second on basic hardware.

Code conventions based on Standard.



Contributions to Skematic are welcome.

  • Maintain the existing code style conventions
  • Ensure your code passes Standard lint npm run lint
  • Include tests that fail without your code, and pass with it
  • Add documentation (JSDoc for functions, README updates, etc)
  • Open a pull request


Copyright 2017 @cayuu v2+ Released under the ISC License (ISC)

Package Sidebar


npm i skematic

Weekly Downloads






Unpacked Size

3.03 MB

Total Files


Last publish


  • cayuu