Caution

This project is now deprecated as Valibot now has an official JSON schema converter in the @valibot/to-json-schema. The command line interface is moved to the project @gcornut/cli-valibot-to-json-schema.

@gcornut/valibot-json-schema

CLI and JS utility to convert Valibot schemas to JSON schema (draft 07).

Some of the features of Valibot can't be converted to JSON schema. JS-specific types like blob or nan obviously can't have an equivalent.
See supported features for more info.

Command Line Tool

This lib exports a CLI that can be used to quickly convert JS modules defining valibot schemas into a JSON schema.

# Convert valibot schemas to JSON schema
npx @gcornut/valibot-json-schema to-json-schema ./schemas.js

This outputs a conversion of the Valibot schemas into JSON schema. If the default export is a Valibot schemas, it is used as the root definition. Other exported Valibot schemas are converted in the JSON schema definitions section.

See detailed input and output:

Input file ./schemas.js:

import * as v from 'valibot';

export const AString = v.string();
const AnObject = v.object({ aString: AString });
export default AnObject;

Output conversion:

{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "definitions": {
    "AString": {
      "type": "string"
    }
  },
  "properties": {
    "aString": {
      "$ref": "#/definitions/AString"
    }
  },
  "required": [
    "aString"
  ],
  "type": "object"
}

AnObject is the default export in the source module, so it is converted as the root definition. AString is exported separately, so it is exported to the definitions section.

ESM and TypeScript input module

The valibot-json-schema CLI loads the input JS module using standard NodeJS CommonJS require. This means you will have issues with ESM or TypeScript modules.

To remedy that, you will either have to preinstall ebuild-runner and esbuild (so that the program can use them) or use a Node-compatible runtime that support these modules (ex: bun, replacing npx with bunx).

Example:

# Convert from TS/ESM module using bunx
bunx @gcornut/valibot-json-schema to-json-schema ./schemas.ts

# Convert from TS/ESM module with esbuild-runner preinstalled
npm install esbuild esbuild-runner
npx @gcornut/valibot-json-schema to-json-schema ./schemas.ts

# Convert from TS/ESM module using `yarn dlx` multi-package feature 
yarn dlx -p esbuild -p esbuild-runner -p @gcornut/valibot-json-schema valibot-json-schema to-json-schema ./schemas.ts

CLI parameters

Use -o <file> option to output the JSON schema to a file instead of stdout.

Use -t <type> option to select the root definitions from the module exports (defaults to the "default" export). Example: -t foo.bar will get the property bar on the foo export of the input JS module.

Use -d <type> option to select the definitions from the module exports (instead of using all non-default export). Example: -d foo.bar will get the property bar on the foo export of the input JS module.

See Advanced options for more options.

Programmatic usage

Use the toJSONSchema function to convert a Valibot schema into JSON schema (v7):

import { toJSONSchema } from '@gcornut/valibot-json-schema';
import { string } from 'valibot';

toJSONSchema({ schema: string() })
// {
//    $schema: 'http://json-schema.org/draft-07/schema#',
//    type: 'string',
// }

If you have many named schemas to convert, you can use the definitions option schema to list as a plain object so that both the schema names (object keys) and schema definitions (object values) are exported to JSON schema.

import { toJSONSchema } from '@gcornut/valibot-json-schema';
import { number } from 'valibot';

const Number = number();
toJSONSchema({ definitions: { Number } });
// {
//    $schema: 'http://json-schema.org/draft-07/schema#',
//    definitions: { Number: { type: 'number' } },
// }

See Advanced options for more options.

Advanced options

Strict object types

Using the strictObjectTypes option (--strictObjectTypes in the CLI), you can force all object type block unknown keys (additionalProperties: false).

Ignore unknown validation

You might want to have advanced validation in your valibot schema that JSON schema can't support ( like custom()). By default, toJSONSchema() will throw an error to let you know these can't be converted to JSON schema.

Using the ignoreUnknownValidation option (--ignoreUnknownValidation in the CLI), you can ignore unsupported valibot validations.

If you preferred to only ignore a specific type of validation, you can use the customValidationConversion option.

Common type strategies

Many JS types don't have a direct equivalent in JSON and toJSONSchema() will throw an error when encountering any of those. However, some of these types have common strategies to convert them.

Using the dateStrategy option (--dateStrategy in the CLI), you can choose to convert JS Date objects into a string ( with the date-time format) or into an integer (with the unix-time format).

Using the undefinedStrategy option (--undefinedStrategy in the CLI), you can choose to convert undefined to the " any" schema (no validation) or to the null type.

Using the bigintStrategy option (--bigintStrategy in the CLI), you can choose to convert JS BigInt values to an integer (with the int64 format) or into a string.

Custom schema conversion

In some cases, you might want to provide custom schema conversion to override standard schema conversion or to add conversion of unsupported schemas.

Using the customSchemaConversion option, you can provide a custom schema converter for any valibot schema type.

Example: converting JS Set schema into an array schema

import { set, string } from 'valibot';

toJSONSchema({
    schema: set(string()),
    customSchemaConversion: {
        // Treat set type like an array
        set: (schema, converter) => converter(array(schema.value))
    },
})
// => { type: 'array', items: { type: 'string' } }

Custom validation conversion

Using the customValidationConversion option, you can provide a custom validation converter for any valibot validation type.

Example: ignoring the valibot custom() validation on all string schema

import { string, custom, pipe } from 'valibot';

toJSONSchema({
    schema: pipe(string(), custom(myCustomStringValidation)),
    customValidationConversion: {
        string: { custom: () => ({}) }
    },
})
// => { type: 'string' }

Supported features

Some converted schema might have slightly different behavior in a JSON schema validator, especially on string formats (email, ipv4, date, etc.) since their implementation is different from valibot implementation.

Here is the list of supported Valibot features (some have partial support):

feature	status
any schema	✅ supported
null schema	✅ supported
number schema	✅ supported
bigint schema	⚠️ with bigintStrategy option provided
string schema	✅ supported
boolean schema	✅ supported
literal schema	⚠️ partial: only JSON-compatible literal are supported
nullable schema	✅ supported
nullish schema	✅ supported
optional schema	✅ supported
never schema	⚠️ partial: only inside object rest or tuple rest params
picklist schema	⚠️ partial: only JSON-compatible literal are supported
enum_ schema	✅ supported
union schema	✅ supported
variant schema	⚠️ partial: converted just like an union
intersect schema	✅ supported
array schema	✅ supported
tuple schema	✅ supported
object schema	✅ supported
record schema	⚠️ partial: only string key are allowed, applicable to plain object only, not arrays
lazy schema	⚠️ partial: only if the schema inside is referenced in definitions
date schema	⚠️ with dateStrategy option provided
length validation	✅ supported
maxLength validation	✅ supported
minLength validation	✅ supported
regex validation	⚠️ partial: only on RegExp features supported by JSON schema
value validation	⚠️ partial: only on string, number and boolean
minValue validation	⚠️ partial: only on number
maxValue validation	⚠️ partial: only on number
multipleOf validation	✅ supported
integer validation	✅ supported
email validation	✅ supported (JSON schema format)
isoDate validation	✅ supported (JSON schema format)
isoTimestamp validation	✅ supported (JSON schema format)
ipv4 validation	✅ supported (JSON schema format)
ipv6 validation	✅ supported (JSON schema format)
uuid validation	✅ supported (JSON schema format)
description metadata	✅ supported