API

Each function calls the Vitest function of the same name. vitest-chain is ESM only, with vitest as a peer dependency.

Function	Alias	Chains into	Vitest's arguments
describe(name, fn)	suite	nothing	—
beforeAll(fn)		beforeAll, beforeEach, test/it, afterEach, afterAll	(context, suite)
beforeEach(fn)		beforeEach, test/it, afterEach	(testContext)
test(name, fn)	it	nothing	(testContext)
afterEach(fn)		afterEach	(testContext)
afterAll(fn)		afterAll	(context, suite)

describe and test also have .only and .skip. Hooks and tests take only a callback (and a name), not vitest's timeout or options arguments, and there is no .each, .concurrent, .todo or .skipIf: use vitest's own functions for those.

Callbacks that aren't chained receive Vitest's arguments. Chained ones receive the chain's context first, then Vitest's arguments. A hook may return a plain object, or a Promise of one, to add its keys to the context. Returning nothing (undefined or null) keeps the context unchanged. Only the object's own enumerable properties are copied, so return { db } rather than a class instance, whose methods would be missing from the context. A number or a string, and an array, function, Map, Set, Date or RegExp, is a type error. Types can't tell a class instance from a plain object, so a hook that returns one (or any object whose prototype isn't Object.prototype or null) fails with HookContextError. An arrow function without braces (beforeEach(() => db.clear())) returns, and merges, whatever the call returns.

Requirements and errors

• Vitest must run hooks in declaration order: set sequence.hooks: 'list' in its config. A hook that runs before one it's chained to fails with HookOrderError.
• Declaring a suite, hook or test while a hook or test is running throws.
• If a hook fails, the afterEach/afterAll hooks chained to it still run, with the context produced before the failure.