Add unit tests

[taylortom]

Writing tests

Instructions for writing tests in this module.

Stack

• Test runner: node:test (built into Node.js)
• Assertions: node:assert/strict

No external test dependencies are needed. Do not introduce Mocha, Jest, or other test frameworks.

Note: Some older modules in the project use Mocha + Should.js. New tests in this module use the built-in Node.js test library instead.

File placement and naming

• Tests live in a tests/ directory at the module root
• One test file per source module, named <moduleName>.spec.js
• Test data/fixtures go in tests/data/

my-module/
├── lib/
│   ├── myModule.js
│   └── myUtils.js
└── tests/
    ├── data/
    │   └── fixtures.json
    ├── myModule.spec.js
    └── myUtils.spec.js

File structure

Every test file follows this structure:

import { describe, it, before } from 'node:test'
import assert from 'node:assert/strict'
import MyModule from '../lib/myModule.js'

describe('My Module', () => {
  let instance

  before(() => {
    instance = new MyModule()
  })

  describe('#methodName()', () => {
    it('should describe expected behaviour', () => {
      assert.equal(instance.methodName(), 'expected')
    })
  })
})

Key rules:

• Import describe, it, before, after etc. from node:test
• Import assert from node:assert/strict (strict mode uses deepStrictEqual by default)
• Use ES module import syntax — this project is "type": "module"
• Group tests by method using nested describe blocks, prefixed with # for instance methods
• Store shared state in let variables scoped to the describe block

Assertions

Use node:assert/strict. Common patterns:

// equality (uses Object.is)
assert.equal(actual, expected)

// deep equality (objects/arrays)
assert.deepEqual(actual, { key: 'value' })

// booleans
assert.equal(result, true)
assert.equal(result, false)

// truthy / falsy
assert.ok(value)
assert.ok(!value)

// type checks
assert.equal(typeof value, 'object')
assert.ok(Array.isArray(value))
assert.ok(value instanceof MyClass)

// expected errors (sync)
assert.throws(() => dangerousCall(), { name: 'TypeError' })

// expected errors (async)
await assert.rejects(asyncCall(), { name: 'Error' })

// not equal
assert.notEqual(a, b)
assert.notDeepEqual(obj1, obj2)

Dynamic test generation

When testing a function against multiple inputs, generate tests in a loop:

const validInputs = ['a@b.com', 'test@example.org']
const invalidInputs = ['not-an-email', '@missing.user']

validInputs.forEach((input) => {
  it(`should accept ${input}`, () => {
    assert.equal(validate(input), true)
  })
})

invalidInputs.forEach((input) => {
  it(`should reject ${input}`, () => {
    assert.equal(validate(input), false)
  })
})

Async tests

node:test supports async/await directly:

it('should connect successfully', async () => {
  const result = await instance.connect()
  assert.equal(typeof result, 'object')
})

Test data

Place fixture files in tests/data/. Use import or fs to load them:

import { readFileSync } from 'node:fs'
import { join, dirname } from 'node:path'
import { fileURLToPath } from 'node:url'

const __dirname = dirname(fileURLToPath(import.meta.url))
const fixtures = JSON.parse(readFileSync(join(__dirname, 'data', 'fixtures.json')))

What to test

• All public methods on exported classes and utilities
• Both success and error paths
• Edge cases (empty input, missing arguments, invalid types)
• That errors are thrown or returned where expected (use assert.throws / assert.rejects)

What NOT to do

• Don't test private/internal methods (prefixed with _)
• Don't add external test dependencies — node:test and node:assert are sufficient
• Don't write tests that depend on execution order between describe blocks
• Don't mock more than necessary — prefer testing real behaviour

Add script to package.json

The tests should be accessible via an npm script in package.json:

"scripts": {
  "test": "node --test tests/"
}

Add GitHub workflow

The following workflow should be added to /.github:

name: Tests
on: push
jobs:
  default:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@master
      - uses: actions/setup-node@master
        with:
          node-version: 'lts/*'
          cache: 'npm'
      - run: npm ci
      - run: npm test

Running tests

# run the test suite
npm test

# run the linter
npx standard

Both must pass before submitting a pull request.