testing.md

Testing

The repo uses @oracle/suitecloud-unit-testing with Jest. Tests run against the compiled JavaScript, not the TypeScript source, which matches what NetSuite executes.

Running tests

npm test                       # All tests with coverage
npm run test:unit              # Unit projects only
npm test -- --watch            # Watch mode
npm test -- path/to/test.js    # Specific test file

suitecloud project:deploy runs the suite before deploying. Set SKIP_TESTS=true to bypass for an emergency deploy, but the default expectation is that tests pass.

Layout

__tests__/
  <domain>/
    fixtures/        Shared test data
    models/          Pure-logic model tests
    services/        (Service tests usually validated in sandbox, not unit tested)

The test tree mirrors src/TypeScripts/. Coverage is computed via collectCoverageFrom in jest.config.js, which excludes script entry points (anything that imports N/*) since those rely on the NetSuite runtime.

The compiled-JS pattern

Tests import the compiled output, not the .ts source:

const Foo = require('SuiteScripts/<domain>/models/Foo');

This means you must run npm run tsc (or tsc --watch) before tests pick up TS changes. npm test does not invoke the compiler.

Mocking N/* modules

Anything that imports a NetSuite module needs the module mocked before the unit under test is required:

jest.mock('N/log', () => ({
  debug: jest.fn(),
  error: jest.fn(),
  audit: jest.fn(),
}));

jest.mock('N/record', () => ({
  load: jest.fn(),
  save: jest.fn(),
}));

beforeEach(() => {
  jest.clearAllMocks();
});

const myScript = require('SuiteScripts/<domain>/myScript');

Reset mocks between tests with jest.clearAllMocks() so per-test expectations don't leak.

What to unit test, what to validate in sandbox

Layer	Coverage approach
Pure-logic models (no N/* imports)	Unit tested directly. No mocking needed.
Services (orchestrate N/* calls)	Light unit tests with mocks for branching logic. Real behavior verified in sandbox.
Scripts (entry points)	Excluded from coverage. Validated end-to-end in sandbox.

The bias is toward extracting logic into pure models so it can be tested without running NetSuite. Services and scripts stay thin.

Fixtures

__tests__/<domain>/fixtures/ holds shared test data. Treat fixtures as the source of truth for "what does a record look like in this domain", reuse them across tests rather than hand-rolling sample objects in each file.