tcomb-validation

What is tcomb-validation?

tcomb-validation is a library for validating JavaScript values with static types. It provides a way to define types and validate data against those types, ensuring that your data conforms to the expected structure.

What are tcomb-validation's main functionalities?

Type Definitions

You can define types using tcomb-validation. In this example, a 'Person' type is defined with 'name' as a string and 'age' as a number.

const t = require('tcomb-validation');

const Person = t.struct({
  name: t.String,
  age: t.Number
});

const person = Person({ name: 'John Doe', age: 30 });

Validation

tcomb-validation allows you to validate data against defined types. In this example, the validation fails because 'age' is a string instead of a number.

const t = require('tcomb-validation');

const Person = t.struct({
  name: t.String,
  age: t.Number
});

const result = t.validate({ name: 'John Doe', age: '30' }, Person);
console.log(result.isValid()); // false

Custom Types

You can create custom types using refinements. In this example, a 'Positive' type is defined to ensure the number is greater than zero.

const t = require('tcomb-validation');

const Positive = t.refinement(t.Number, n => n > 0, 'Positive');

const result = t.validate(10, Positive);
console.log(result.isValid()); // true

Other packages similar to tcomb-validation

joi

Joi is a powerful schema description language and data validator for JavaScript. It allows you to create blueprints or schemas for JavaScript objects to ensure validation. Compared to tcomb-validation, Joi offers a more extensive set of validation rules and is widely used in the Node.js community.

yup

Yup is a JavaScript schema builder for value parsing and validation. It is similar to Joi but is often preferred for its simplicity and ease of use, especially in React applications. Yup provides a more modern API and better integration with form libraries like Formik.

ajv

AJV (Another JSON Schema Validator) is a JSON schema validator that is very fast and supports JSON Schema draft-07. It is more focused on JSON schema validation compared to tcomb-validation, which is more about type validation in JavaScript.

README

A general purpose JavaScript validation library based on type combinators

Features

• concise yet expressive syntax
• validates native types, refinements, objects, lists and tuples, enums, unions, dicts, intersections
• validates structures with arbitrary level of nesting
• detailed informations on failed validations
• lightweight alternative to JSON Schema
• reuse your domain model written with tcomb

Documentation

• Basic usage
  • Primitives
  • Refinements
  • Objects
  • Lists and tuples
  • Enums
  • Unions
  • Dicts
  • Intersections
  • Nested structures
• Customise error messages
• Use cases
  • Form validation
  • JSON schema
• API reference

Basic usage

If you don't know how to define types with tcomb you may want to take a look at its README file.

The main function is validate:

validate(value, type, [options]) -> ValidationResult

• value the value to validate
• type a type defined with the tcomb library
• options (optional) is an object with the following keys
  • path: Array<string | number> path prefix for validation
  • context: any passed to getValidationErrorMessage (useful for i18n)
  • strict: boolean (default false) if true no additional properties are allowed while validating structs

returns a ValidationResult object containing the result of the validation

Note.

• options can be an array (as path prefix) for backward compatibility (deprecated)

Example

var t = require('tcomb-validation');
var validate = t.validate;

validate(1, t.String).isValid();   // => false
validate('a', t.String).isValid(); // => true

You can inspect the result to quickly identify what's wrong:

var result = validate(1, t.String);
result.isValid();             // => false
result.firstError().message;  // => 'Invalid value 1 supplied to String'

// see `result.errors` to inspect all errors

Primitives

// null and undefined
validate('a', t.Nil).isValid();       // => false
validate(null, t.Nil).isValid();      // => true
validate(undefined, t.Nil).isValid(); // => true

// strings
validate(1, t.String).isValid();   // => false
validate('a', t.String).isValid(); // => true

// numbers
validate('a', t.Number).isValid(); // => false
validate(1, t.Number).isValid();   // => true

// booleans
validate(1, t.Boolean).isValid();    // => false
validate(true, t.Boolean).isValid(); // => true

// optional values
validate(null, maybe(t.String)).isValid(); // => true
validate('a', maybe(t.String)).isValid();  // => true
validate(1, maybe(t.String)).isValid();    // => false

// functions
validate(1, t.Function).isValid();              // => false
validate(function () {}, t.Function).isValid(); // => true

// dates
validate(1, t.Date).isValid();           // => false
validate(new Date(), t.Date).isValid();  // => true

// regexps
validate(1, t.RegExp).isValid();    // => false
validate(/^a/, t.RegExp).isValid(); // => true

Refinements

You can express more fine-grained contraints with the refinement syntax:

// a predicate is a function with signature: (x) -> boolean
var predicate = function (x) { return x >= 0; };

// a positive number
var Positive = t.refinement(t.Number, predicate);

validate(-1, Positive).isValid(); // => false
validate(1, Positive).isValid();  // => true

Objects

Structs

// an object with two numerical properties
var Point = t.struct({
  x: t.Number,
  y: t.Number
});

validate(null, Point).isValid();            // => false
validate({x: 0}, Point).isValid();          // => false, y is missing
validate({x: 0, y: 'a'}, Point).isValid();  // => false, y is not a number
validate({x: 0, y: 0}, Point).isValid();    // => true
validate({x: 0, y: 0, z: 0}, Point, { strict: true }).isValid(); // => false, no additional properties are allowed

Interfaces

Differences from structs

• also checks prototype keys

var Serializable = t.interface({
  serialize: t.Function
});

validate(new Point(...), Serializable).isValid(); // => false

Point.prototype.serialize = function () { ... }

validate(new Point(...), Serializable).isValid(); // => true

Lists and tuples

Lists

// a list of strings
var Words = t.list(t.String);

validate(null, Words).isValid();                  // => false
validate(['hello', 1], Words).isValid();          // => false, [1] is not a string
validate(['hello', 'world'], Words).isValid();    // => true

Tuples

// a tuple (width x height)
var Size = t.tuple([Positive, Positive]);

validate([1], Size).isValid();      // => false, height missing
validate([1, -1], Size).isValid();  // => false, bad height
validate([1, 2], Size).isValid();   // => true

Enums

var CssTextAlign = t.enums.of('left right center justify');

validate('bottom', CssTextAlign).isValid(); // => false
validate('left', CssTextAlign).isValid();   // => true

Unions

var CssLineHeight = t.union([t.Number, t.String]);

validate(null, CssLineHeight).isValid();    // => false
validate(1.4, CssLineHeight).isValid();     // => true
validate('1.2em', CssLineHeight).isValid(); // => true

Dicts

// a dictionary of numbers
var Country = t.enums.of(['IT', 'US'], 'Country');
var Warranty = t.dict(Country, t.Number, 'Warranty');

validate(null, Warranty).isValid();             // => false
validate({a: 2}, Warranty).isValid();           // => false, ['a'] is not a Country
validate({US: 2, IT: 'a'}, Warranty).isValid(); // => false, ['IT'] is not a number
validate({US: 2, IT: 1}, Warranty).isValid();   // => true

Intersections

var Min = t.refinement(t.String, function (s) { return s.length > 2; }, 'Min');
var Max = t.refinement(t.String, function (s) { return s.length < 5; }, 'Max');
var MinMax = t.intersection([Min, Max], 'MinMax');

MinMax.is('abc'); // => true
MinMax.is('a'); // => false
MinMax.is('abcde'); // => false

Nested structures

You can validate structures with an arbitrary level of nesting:

var Post = t.struct({
  title: t.String,
  content: t.String,
  tags: Words
});

var mypost = {
  title: 'Awesome!',
  content: 'You can validate structures with arbitrary level of nesting',
  tags: ['validation', 1] // <-- ouch!
};

validate(mypost, Post).isValid();             // => false
validate(mypost, Post).firstError().message;  // => 'tags[1] is `1`, should be a `Str`'

Customise error messages

You can customise the validation error message defining a function getValidationErrorMessage(value, path, context) on the type constructor:

var ShortString = t.refinement(t.String, function (s) {
  return s.length < 3;
});

ShortString.getValidationErrorMessage = function (value) {
  if (!value) {
    return 'Required';
  }
  if (value.length >= 3) {
    return 'Too long my friend';
  }
};

validate('abc', ShortString).firstError().message; // => 'Too long my friend'

How to keep DRY?

In order to keep the validation logic in one place, one may define a custom combinator:

function mysubtype(type, getValidationErrorMessage, name) {
  var Subtype = t.refinement(type, function (x) {
    return !t.String.is(getValidationErrorMessage(x));
  }, name);
  Subtype.getValidationErrorMessage = getValidationErrorMessage;
  return Subtype;
}

var ShortString = mysubtype(t.String, function (s) {
  if (!s) {
    return 'Required';
  }
  if (s.length >= 3) {
    return 'Too long my friend';
  }
});

Use cases

Form validation

Let's design the process for a simple sign in form:

var SignInInfo = t.struct({
  username: t.String,
  password: t.String
});

// retrieves values from the UI
var formValues = {
  username: $('#username').val().trim() || null,
  password: $('#password').val().trim() || null
};

// if formValues = {username: null, password: 'password'}
var result = validate(formValues, SignInInfo);
result.isValid();             // => false
result.firstError().message;  // => 'Invalid value null supplied to /username: String'

JSON schema

If you don't want to use a JSON Schema validator or it's not applicable, you can just use this lightweight library in a snap. This is the JSON Schema example of http://jsonschemalint.com/

{
  "type": "object",
  "properties": {
    "foo": {
      "type": "number"
    },
    "bar": {
      "type": "string",
      "enum": [
        "a",
        "b",
        "c"
      ]
    }
  }
}

and the equivalent tcomb-validation counterpart:

var Schema = t.struct({
  foo: t.Number,
  bar: t.enums.of('a b c')
});

let's validate the example JSON:

var json = {
  "foo": "this is a string, not a number",
  "bar": "this is a string that isn't allowed"
};

validate(json, Schema).isValid(); // => false

// the returned errors are:
- Invalid value "this is a string, not a number" supplied to /foo: Number
- Invalid value "this is a string that isn't allowed" supplied to /bar: "a" | "b" | "c"

Note: A feature missing in standard JSON Schema is the powerful refinement syntax.

Api reference

ValidationResult

ValidationResult represents the result of a validation. It containes the following fields:

• errors: a list of ValidationError if validation fails
• value: an instance of type if validation succeded

// the definition of `ValidationError`
var ValidationError = t.struct({
  message: t.String,                        // a default message for developers
  actual: t.Any,                            // the actual value being validated
  expected: t.Function,                     // the type expected
  path: list(t.union([t.String, t.Number])) // the path of the value
}, 'ValidationError');

// the definition of `ValidationResult`
var ValidationResult = t.struct({
  errors: list(ValidationError),
  value: t.Any
}, 'ValidationResult');

#isValid()

Returns true if there are no errors.

validate('a', t.String).isValid(); // => true

#firstError()

Returns an object that contains an error message or null if validation succeeded.

validate(1, t.String).firstError().message; // => 'value is `1`, should be a `Str`'

validate(value, type, [options]) -> ValidationResult

• value the value to validate
• type a type defined with the tcomb library
• options (optional) is an object with the following keys
  • path: Array<string | number> path prefix for validation
  • context: any passed to getValidationErrorMessage (useful for i18n)
  • strict: boolean (default false) if true no additional properties are allowed while validating structs

Tests

Run npm test

License

The MIT License (MIT)

Changelog

v3.4.1

• Bug Fix
  • TypeScript: fix recursive type definition error (@RedRoserade)