DictValidator

See source code
Table of contents

Extends Validator<Record<Key, Value>>.

Validator for dictionary/map objects where both keys and values are validated. Useful for validating objects used as key-value stores.

class DictValidator<Key extends string, Value> extends Validator<
  Record<Key, Value>
> {}

Example

const scoreDict = new DictValidator(T.string, T.number)
const scores = scoreDict.validate({
  alice: 100,
  bob: 85,
  charlie: 92,
})
// scores is typed as Record<string, number>

Constructor

Creates a new DictValidator.

keyValidator - Validator for object keys valueValidator - Validator for object values

Parameters

NameDescription

keyValidator

Validatable<Key>

valueValidator

Validatable<Value>

Properties

keyValidator

readonly
readonly keyValidator: Validatable<Key>

validateUsingKnownGoodVersionFn

readonlyoptional
readonly validateUsingKnownGoodVersionFn?:
  | undefined
  | ValidatorUsingKnownGoodVersionFn<T>

validationFn

readonly
readonly validationFn: ValidatorFn<T>

valueValidator

readonly
readonly valueValidator: Validatable<Value>

Methods

check( )

Adds an additional validation check without changing the resulting value type. Can be called with just a check function, or with a name for better error messages.

check(name: string, checkFn: (value: T) => void): Validator<T>

Example

import { T, ValidationError } from '@tldraw/validate'

// Basic check without name
const evenNumber = T.number.check((value) => {
  if (value % 2 !== 0) {
    throw new ValidationError('Expected even number')
  }
})

// Named checks for better error messages in complex validators
const shapePositionValidator = T.object({
  x: T.number.check('finite', (value) => {
    if (!Number.isFinite(value)) {
      throw new ValidationError('Position must be finite')
    }
  }),
  y: T.number.check('within-bounds', (value) => {
    if (value < -10000 || value > 10000) {
      throw new ValidationError(
        'Position must be within bounds (-10000 to 10000)'
      )
    }
  }),
})

// Error will be: "At x (check finite): Position must be finite"

Parameters

NameDescription

name

string

Name for the check (used in error messages)

checkFn

(value: T) => void

Function that validates the value (should throw on invalid input)

Returns

Validator<T>

A new validator with the additional check


isValid( )

Type guard that checks if a value is valid without throwing an error.

isValid(value: unknown): value is T

Example

import { T } from '@tldraw/validate'

function processUserInput(input: unknown) {
  if (T.string.isValid(input)) {
    // input is now typed as string within this block
    return input.toUpperCase()
  }
  if (T.number.isValid(input)) {
    // input is now typed as number within this block
    return input.toFixed(2)
  }
  throw new Error('Expected string or number')
}

Parameters

NameDescription

value

unknown

The value to check

Returns

value is T

True if the value is valid, false otherwise


nullable( )

Returns a new validator that also accepts null values.

nullable(): Validator<null | T>

Example

import { T } from '@tldraw/validate'

const assetValidator = T.object({
  id: T.string,
  name: T.string,
  src: T.srcUrl.nullable(), // Can be null if not loaded yet
  mimeType: T.string.nullable(),
})

const asset = assetValidator.validate({
  id: 'image-123',
  name: 'photo.jpg',
  src: null, // Valid - asset not loaded yet
  mimeType: 'image/jpeg',
})

optional( )

Returns a new validator that also accepts undefined values.

optional(): Validator<T | undefined>

Example

import { T } from '@tldraw/validate'

const shapeConfigValidator = T.object({
  type: T.literal('rectangle'),
  x: T.number,
  y: T.number,
  label: T.string.optional(), // Optional property
  metadata: T.object({ created: T.string }).optional(),
})

// Both of these are valid:
const shape1 = shapeConfigValidator.validate({ type: 'rectangle', x: 0, y: 0 })
const shape2 = shapeConfigValidator.validate({
  type: 'rectangle',
  x: 0,
  y: 0,
  label: 'My Shape',
})

refine( )

Creates a new validator by refining this validator with additional logic that can transform the validated value to a new type.

refine<U>(otherValidationFn: (value: T) => U): Validator<U>

Example

import { T, ValidationError } from '@tldraw/validate'

// Transform string to ensure it starts with a prefix
const prefixedIdValidator = T.string.refine((id) => {
  return id.startsWith('shape:') ? id : `shape:${id}`
})

const id1 = prefixedIdValidator.validate('rectangle-123') // Returns "shape:rectangle-123"
const id2 = prefixedIdValidator.validate('shape:circle-456') // Returns "shape:circle-456"

// Parse and validate JSON strings
const jsonValidator = T.string.refine((str) => {
  try {
    return JSON.parse(str)
  } catch {
    throw new ValidationError('Invalid JSON string')
  }
})

Parameters

NameDescription

otherValidationFn

(value: T) => U

Function that transforms/validates the value to type U

Returns

Validator<U>

A new validator that validates to type U


validate( )

Validates an unknown value and returns it with the correct type. The returned value is guaranteed to be referentially equal to the passed value.

validate(value: unknown): T

Example

import { T } from '@tldraw/validate'

const name = T.string.validate('Alice') // Returns "Alice" as string
const title = T.string.validate('') // Returns "" (empty strings are valid)

// These will throw ValidationError:
T.string.validate(123) // Expected string, got a number
T.string.validate(null) // Expected string, got null
T.string.validate(undefined) // Expected string, got undefined

Parameters

NameDescription

value

unknown

The unknown value to validate

Returns

T

The validated value with type T


validateUsingKnownGoodVersion( )

Performance-optimized validation using a previously validated value. If the new value is referentially equal to the known good value, returns the known good value immediately.

validateUsingKnownGoodVersion(knownGoodValue: T, newValue: unknown): T

Example

import { T } from '@tldraw/validate'

const userValidator = T.object({
  name: T.string,
  settings: T.object({ theme: T.literalEnum('light', 'dark') }),
})

const user = userValidator.validate({
  name: 'Alice',
  settings: { theme: 'light' },
})

// Later, with partially changed data:
const newData = { name: 'Alice', settings: { theme: 'dark' } }
const updated = userValidator.validateUsingKnownGoodVersion(user, newData)
// Only validates the changed 'theme' field for better performance

Parameters

NameDescription

knownGoodValue

T

A previously validated value

newValue

unknown

The new value to validate

Returns

T

The validated value, potentially reusing the known good value


Prev
ArrayOfValidator
Next
ObjectValidator