Runtime validation with a plain function

Use case

Validate API responses using a simple function without any schema library.

Smallest working example

import { createClient } from '@parcely/core'

const http = createClient({ baseURL: 'https://api.example.com' })

interface User {
  id: string
  name: string
}

function validateUser(input: unknown): User {
  if (
    typeof input === 'object' && input !== null &&
    'id' in input && typeof (input as User).id === 'string' &&
    'name' in input && typeof (input as User).name === 'string'
  ) {
    return input as User
  }
  throw new Error('Invalid user data')
}

const { data } = await http.get('/users/me', {
  validate: validateUser,
})
// data is typed as User

How it works

parcely's Validator<T> type accepts (input: unknown) => T functions. The function receives the parsed response body. If it returns a value, that value becomes data. If it throws, parcely wraps the error in HttpError with code: 'ERR_VALIDATION'.

Error handling

import { isHttpError } from '@parcely/core'

try {
  await http.get('/users/me', { validate: validateUser })
} catch (err) {
  if (isHttpError(err) && err.code === 'ERR_VALIDATION') {
    console.log('Validation failed:', err.cause)
  }
}

Axios equivalent

Axios does not support validation. You would validate after the response:

// axios:
const { data } = await http.get('/users/me')
const user = validateUser(data) // manual

// parcely (integrated):
const { data: user } = await http.get('/users/me', { validate: validateUser })

Notes and gotchas

• The function validator is the simplest option -- no library needed.
• Resolution order for the Validator<T> type: Standard Schema v1 (detected via ~standard property), then .parse() method, then function call.
• The validator runs after response parsing and before the response envelope is returned.