Requirements

@traversable/zod-test has 2 peer dependencies:

1. zod (v4)
2. fast-check

Usage

$ pnpm add -D @traversable/zod-test zod fast-check

Here's an example of importing the library:

import { z } from 'zod'
import { zxTest } from '@traversable/zod-test'

// see below for specifc examples

Track record

@traversabe/zod-test has found several upstream bugs in zod:

1. Security exploit: z.object pollutes the global Object prototype

• Issue
• Sandbox

2. Bug: z.literal escaping bug

• Issue
• Sandbox

3. Bug: "Diagonal" objects passed to z.enum produce false negatives

• Issue
• Sandbox

4. Bug: z.file output type incompatible with globalThis.File

• Issue
• Sandbox

Table of contents

• zxTest.fuzz
• zxTest.seedToSchema
• zxTest.seedToValidData
• zxTest.seedToInvalidData
• zxTest.seedToValidDataGenerator
• zxTest.seedToInvalidDataGenerator
• zxTest.SeedGenerator
• zxTest.SeedValidDataGenerator
• zxTest.SeedInvalidDataGenerator

zxTest.fuzz

Convert a Zod schema into a fast-check arbitrary.

Configure how fuzzed values will be generated via the 2nd argument (options).

Override individual arbitraries via the 3rd argument (overrides).

[!NOTE]

zxTest.fuzz is the only schema-to-generator function that has itself been fuzz tested to ensure that no matter what schema you give it, the data-generator it returns will always produce valid data.

This excludes schemas that make it impossible to generate valid data, for example:

• z.never
• z.nonoptional(z.undefined())
• z.enum([])
• z.union([])
• z.intersection(z.number(), z.string())

Example

import * as vi from 'vitest'
import * as fc from 'fast-check'
import { fuzz } from '@traversable/zod-test'

const Schema = z.record(
  z.string(), 
  z.union(
    z.number(),
    z.string(),
  )
)

const generator = fuzz(
  Schema, 
  { record: { minKeys: 1 }, number: { noDefaultInfinity: true } },
  { string: () => fc.stringMatching(/[\S\s]+[\S]+/) },
)

vi.test('fuzz test example', () => {
  fc.assert(
    fc.property(generator, (data) => {
      vi.assert.doesNotThrow(() => Schema.parse(data))
    }),
    { numRuns: 1_000 }
  )
})

See also

• the fast-check docs

zxTest.seedToSchema

Use zxTest.seedToSchema to convert a seed generated by zxTest.SeedGenerator into a zod schema that satisfies the configuration options you specified.

Example

import { zxTest } from '@traversable/zod-test'
import * as fc from 'fast-check'

const builder = zxTest.SeedGenerator()['*']
const [mySeed] = fc.sample(builder.object, 1)

const mySchema = zxTest.seedToSchema(mySeed)
//    ^? const mySchema: z.ZodType

zxTest.seedToValidData

Use zxTest.seedToValidData to convert a seed generated by zxTest.SeedGenerator into data that satisfies the schema that the seed represents.

Example

import { zxTest } from '@traversable/zod-test'
import * as fc from 'fast-check'

const builder = zxTest.SeedGenerator()['*']
const [mySeed] = fc.sample(builder.object, 1)

const mySchema = zxTest.seedToSchema(mySeed)
//    ^? const mySchema: z.ZodType

const validData = zxTest.seedToValidData(mySeed)

mySchema.parse(validData) // will never throw

zxTest.seedToInvalidData

Use zxTest.seedToInvalidData to convert a seed generated by zxTest.SeedGenerator into data that does not satisfy the schema that the seed represents.

Example

import { zxTest } from '@traversable/zod-test'
import * as fc from 'fast-check'

const builder = zxTest.SeedGenerator()['*']
const [mySeed] = fc.sample(builder.object, 1)

const mySchema = zxTest.seedToSchema(mySeed)
//    ^? const mySchema: z.ZodType

const invalidData = zxTest.seedToValidData(mySeed)

mySchema.parse(invalidData) // should always throw

zxTest.seedToValidDataGenerator

Like zxTest.seedToValidData, except zxTest.seedToValidDataGenerator accepts a seed and returns a valid data arbitrary (which can then be used to produce valid data).

Example

import { zxTest } from '@traversable/zod-test'
import * as fc from 'fast-check'

const builder = zxTest.SeedGenerator()['*']
const [mySeed] = fc.sample(builder.object, 1)

const mySchema = zxTest.seedToSchema(mySeed)
//    ^? const mySchema: z.ZodType

const validDataGenerator = zxTest.seedToValidDataGenerator(mySeed)
const [validData] = fc.sample(validDataGenerator, 1)

mySchema.parse(validData) // will never throw

zxTest.seedToInvalidDataGenerator

Like zxTest.seedToInvalidData, except zxTest.seedToValidDataGenerator accepts a seed and returns an invalid data arbitrary (which can then be used to produce invalid data).

Example

import type * as z from 'zod'
import * as fc from 'fast-check'
import { zxTest } from '@traversable/zod-test'

const builder = zxTest.SeedGenerator()['*']
const [mySeed] = fc.sample(builder.object, 1)

const mySchema = zxTest.seedToSchema(mySeed)
//    ^? const mySchema: z.ZodType

const invalidDataGenerator = zxTest.seedToInvalidDataGenerator(mySeed)
const [invalidData] = fc.sample(invalidDataGenerator, 1)

mySchema.parse(invalidData) // will always throw

zxTest.SeedGenerator

[!NOTE]

zxTest.SeedGenerator is fairly low-level. All of the other exports of this library have been implemented in terms of zxTest.SeedGenerator.

Generates a configurable, pseudo-random "seed builder".

• Use zxTest.seedToSchema to convert a seed into a zod schema
• Use zxTest.seedToValidData to convert a seed into valid data
• Use zxTest.seedToInvalidData to convert a seed into invalid data

Example

import { zxTest } from '@traversable/zod-test'
import * as fc from 'fast-check'

const builder = zxTest.SeedGenerator({
  include: ["boolean", "string", "object"],
  // 𐙘 use `include` to only include certain schema types
  exclude: ["boolean", "any"],
  // 𐙘 use `exclude` to exclude certain schema types altogether (overrides `include`)
  object: { maxKeys: 5 },
  // 𐙘 specific arbitraries are configurable by name
})

// included schemas are present as properties on your generator...
builder.string
builder.object

// ...excluded schemas are not present...
builder.boolean // 🚫 TypeError

// ...a special wildcard `"*"` property (pronounced "surprise me") is always present:
builder["*"]

/**
 * `fast-check` will generate a seed, which is a data structure containing
 * integers that represent a kind of AST.
 *
 * To use a seed, you need to pass it to an interpreter like `zxTest.seedToSchema`,
 * `zxTest.seedToValidData` or `zxTest.seedToInvalidData`:
 */

const [mySeed] = fc.sample(builder.object, 1)

const mySchema = zxTest.seedToSchema(mySeed)
//    ^? const mySchema: z.ZodType

const validData = zxTest.seedToValidData(mySeed)
//    ^? since the `mySeed` was also used to generate `mySchema`,
//       parsing `validData` should always succeed

const invalidData = zxTest.seedToInvalidData(mySeed)
//    ^? since the `mySeed` was also used to generate `mySchema`,
//       parsing `invalidData` should always fail

zxTest.SeedValidDataGenerator

Like zxTest.SeedGenerator, except zxTest.SeedValidDataGenerator comes pre-configured to exclude schemas that make it impossible to reliably generate valid data.

[!NOTE]

zxTest.SeedValidDataGenerator does not accept any options. If you need more fine-grained control of the schemas being generated, use zxTest.SeedGenerator.

zxTest.SeedInvalidDataGenerator

Like zxTest.SeedGenerator, except zxTest.SeedValidDataGenerator comes pre-configured to exclude schemas that make it impossible to reliably generate invalid data.

[!NOTE]

zxTest.SeedInvalidDataGenerator does not accept any options. If you need more fine-grained control of the schemas being generated, use zxTest.SeedGenerator.