Schematox

Schematox is a lightweight typesafe schema defined parser/validator. All schemas are JSON compatible.

Instead of supporting all possible JS/TS data structures, the library is focusing on fixed set of schema types: string, number, boolean, literal, object, array, union. Each schema can have parameters: optional, nullable, description. Each primitive schema has "brand" parameter as mean of making its subject type nominal. The rest parameters is schema specific range limiters.

Library supports static schema definition which means your schemas could be completely independent from schematox. One could use such schemas as source for generation other structures like DB models.

Features:

• Statically defined JSON compatible schema
• Check defined schema correctness using non generic type "Schema"
• Programmatically defined schema (struct)
• Schema subject verification methods:
  • parse: constructs new object based on the given schema and subject
  • validate: checks and returns reference to the original schema subject
  • guard: validates and narrows schema subject type in the current scope
• Ether-style error handling (no unexpected throws)
• First-class support for branded primitives (primitive nominal types alias)

Check out github issues to know what we are planning to support soon.

Currently we on version 0. The public API is mostly defined however few thing left before the first major release:

• Record and tuple schema support
• Allow parser to replace value before it's validated (similar to coercing concept)
• Clearly defined supported versions of typescript/node
• Support "deno" runtime and publish package on "deno.land"
• Have a benchmark that compares library performance with other parsers

The library is small so exploring README.md is enough for understanding its API, checkout limitations/examples and you good to go:

• Install
• Limitations
• Static schema example
• Programmatic schema example
• Example for all supported schema types
  • String
  • Number
  • Boolean
  • Literal
  • Object
  • Array
  • Union
• Schema parameters
• Error shape

Install

npm install schematox

Limitations

Currently we support max 7 layers of depth for compound schema type: object, array, union.

Because of this we can check structural correctness of the statically defined schema using non generic type Schema. Important detail is that union schema type is also compound type so each nested definition counts as +1 layer of depth.

Static schema example

Statically defined schema:

import { parse, validate, guard } from 'schematox'
import type { Schema } from 'schematox'

export const userSchema = {
  type: 'object',
  of: {
    id: {
      type: 'string',
      brand: ['idFor', 'User'],
    },
    name: { type: 'string' },
  },
} as const satisfies Schema

const subject = {
  id: '1' as SubjectType<typeof userSchema.id>,
  name: 'John',
} as unknown

const parsed = parse(userSchema, subject)

if (parsed.left) {
  throw Error('Not expected')
}

console.log(parsed.right) // { id: '1', name: 'John' }

const validated = validate(userSchema, subject)

if (validated.left) {
  throw Error('Not expected')
}

console.log(validated.right) // { id: '1', name: 'John' }

if (guard(userSchema, subject)) {
  // { id: string & { __idFor: 'User' }; name: string }
  subject
}

Programmatic schema example

Same schema but defined programmatically:

import { object, string } from 'schematox'
import type { SubjectType } from 'schematox'

const struct = object({
  id: string().brand('idFor', 'User'),
  name: string(),
})

const subject = { id: '1', name: 'John' } as unknown

const parsed = struct.parse(subject)

if (parsed.left) {
  throw Error('Not expected')
}

console.log(parsed.right) // { id: '1', name: 'John' }

const validated = struct.validate(subject)

if (validated.left) {
  throw Error('Not expected')
}

console.log(validated.right) // { id: '1', name: 'John' }

if (struct.guard(subject)) {
  // { id: string & { __idFor: 'User' }; name: string }
  subject
}

All programmatically defined schemas are the same as static, one just needs to access it through __schema key. We can mix static/programmatic schemas either accessing it through __schema or wrap it by { __schema: T } if consumer is programmatic schema.

Example for all supported schema types

We distinguish two main categories of schema units:

• primitive: string, number, boolean, literal
• compound: object, array, union

Any schema share optional/nullable/description parameters. Any compound schema could have any other schema type as its member including itself. Any primitive schema can have "brand" parameter.

String

const schema = {
  type: 'string',
  optional: true,
  nullable: true,
  brand: ['x', 'y'],
  minLength: 1,
  maxLength: 2,
  description: 'x',
} as const satisfies Schema

const struct = string()
  .optional()
  .nullable()
  .brand('x', 'y')
  .minLength(1)
  .maxLength(2)
  .description('x')

// (string & { __x: 'y' }) | undefined | null
type FromSchema = SubjectType<typeof schema>
type FromStruct = SubjectType<typeof struct>

Number

const schema = {
  type: 'number',
  optional: true,
  nullable: true,
  brand: ['x', 'y'],
  min: 1,
  max: 2,
  description: 'x',
} as const satisfies Schema

const struct = number()
  .optional()
  .nullable()
  .brand('x', 'y')
  .min(1)
  .max(2)
  .description('x')

// (number & { __x: 'y' }) | undefined | null
type FromSchema = SubjectType<typeof schema>
type FromStruct = SubjectType<typeof struct>
//

Boolean

const schema = {
  type: 'boolean',
  optional: true,
  nullable: true,
  brand: ['x', 'y'],
  description: 'x',
} as const satisfies Schema

const struct = boolean() //
  .optional()
  .nullable()
  .brand('x', 'y')
  .description('x')

// (boolean & { __x: 'y' }) | undefined | null
type FromSchema = SubjectType<typeof schema>
type FromStruct = SubjectType<typeof struct>

Literal

Could be string/number/boolean literal

const schema = {
  type: 'literal',
  of: 'x',
  optional: true,
  nullable: true,
  brand: ['x', 'y'],
  description: 'x',
} as const satisfies Schema

const struct = literal('x') //
  .optional()
  .nullable()
  .brand('x', 'y')
  .description('x')

// ('x' & { __x: 'y' }) | undefined | null
type FromSchema = SubjectType<typeof schema>
type FromStruct = SubjectType<typeof struct>

Object

const schema = {
  type: 'object',
  of: {
    x: { type: 'string' },
    y: { type: 'number' },
  },
  optional: true,
  nullable: true,
  description: 'x',
} as const satisfies Schema

const struct = object({
  x: string(),
  y: number(),
})
  .optional()
  .nullable()
  .description('x')

// { x: string; y: number } | undefined | null
type FromSchema = SubjectType<typeof schema>
type FromStruct = SubjectType<typeof struct>

Array

const schema = {
  type: 'array',
  of: { type: 'string' },
  optional: true,
  minLength: 1,
  maxLength: 1000,
  description: 'x',
} as const satisfies Schema

const struct = array(string())
  .optional()
  .nullable()
  .minLength(1)
  .maxLength(1000)
  .description('x')

// string[] | undefined | null
type FromSchema = SubjectType<typeof schema>
type FromStruct = SubjectType<typeof struct>

Union

const schema = {
  type: 'union',
  of: [{ type: 'string' }, { type: 'number' }],
  optional: true,
  nullable: true,
  description: 'x',
} as const satisfies Schema

const struct = union([string(), number()])
  .optional()
  .nullable()
  .description('x')

// string | number | undefined | null
type FromSchema = SubjectType<typeof schema>
type FromStruct = SubjectType<typeof struct>

Schema parameters

• optional?: boolean – does undefined is valid value
• nullable?: boolean – does null is valid value
• brand?: [string, string] – make primitive type nominal "['idFor', 'User'] -> T & { __idFor: 'User' }"
• minLength/maxLength/min/max – schema type dependent limiting characteristics
• description?: string – description of the particular schema property which can be used to provide more detailed information for the user/developer on validation/parse error

Error shape

Nested schema example. Subject 0 is invalid, should be a string:

import { object, array, string } from 'schematox'

const struct = object({
  x: object({
    y: array(
      object({
        z: string(),
      })
    ),
  }),
})

const result = struct.parse({ x: { y: [{ z: 0 }] } })

The result.left shape is:

[
  {
    "code": "INVALID_TYPE",
    "schema": { "type": "string" },
    "subject": 0,
    "path": ["x", "y", 0, "z"]
  }
]

It's always an array with at least one entry. Each entry includes:

• code: Specifies either INVALID_TYPE (when schema subject or default value don't meet schema type specifications), or INVALID_RANGE (when min/max or minLength/maxLength schema requirements aren't met).
• schema: The specific section of schema where the invalid value is found.
• subject: The specific part of the validated subject where the invalid value exists.
• path: Traces the route from the root to the error subject, with strings as keys and numbers as array indexes.