API
Due to being inspired by the preact-testing-library you can check its page for more information.
There are several key differences, to be aware of.
render
The render
function takes in a function that returns a Solid Component, rather
than simply the component itself.
const results = render(() => <YourComponent />, options)
Solid.js does not re-render, it merely executes side effects triggered by
reactive state that change the DOM, therefore there is no rerender
method. You
can use global signals to manipulate your test component in a way that causes it
to update.
In addition to the original API, the render function of this testing library
supports a convenient location
option that will set up an in-memory router
pointing at the specified location. Since this setup is not synchronous, you
need to first use asynchronous queries (findBy
) after employing it:
it('uses params', async () => {
const App = () => (
<>
<Route path="/ids/:id" component={() => <p>Id: {useParams()?.id}</p>} />
<Route path="/" component={() => <p>Start</p>} />
</>
)
const {findByText} = render(() => <App />, {location: 'ids/1234'})
expect(await findByText('Id: 1234')).not.toBeFalsy()
})
It uses @solidjs/router
, so if you want to use a different router, you should
consider the wrapper
option instead. If you attempt to use this without having
the package installed, you will receive an error message.
renderHook
Solid.js external reactive state does not require any DOM elements to run in, so
our renderHook
call to test hooks in the context of a component (if your hook
does not require the context of a component, createRoot
should suffice to test
the reactive behavior; for convenience, we also have createEffect
, which is
described in the Async methods
section) has no container
,
baseElement
or queries in its options or return value. Instead, it has an
owner
to be used with
runWithOwner
if
required. It also exposes a cleanup
function, though this is already
automatically called after the test is finished.
function renderHook<Args extends any[], Result>(
hook: (...args: Args) => Result,
options: {
initialProps?: Args,
wrapper?: Component<{ children: JSX.Element }>
}
) => {
result: Result;
owner: Owner | null;
cleanup: () => void;
}
This can be used to easily test a hook / primitive:
const {result} = renderHook(createResult)
expect(result).toBe(true)
If you are using a wrapper
with renderHook
, make sure it will always
return props.children
- especially if you are using a context with
asynchronous code together with <Show>
, because this is required to get the
value from the hook and it is only obtained synchronously once and you will
otherwise only get undefined
and wonder why this is the case.
renderDirective
Solid.js supports
custom directives, which is a
convenient pattern to tie custom behavior to elements, so we also have a
renderDirective
call, which augments renderHook
to take a directive as first
argument, accept an initialValue
for the argument and a targetElement
(string, HTMLElement or function returning an HTMLElement) in the options
and
also returns arg
and setArg
to read and manipulate the argument of the
directive.
function renderDirective<
Arg extends any,
Elem extends HTMLElement
>(
directive: (ref: Elem, arg: Accessor<Arg>) => void,
options?: {
...renderOptions,
initialValue: Arg,
targetElement:
| Lowercase<Elem['nodeName']>
| Elem
| (() => Elem)
}
): Result & { arg: Accessor<Arg>, setArg: Setter<Arg> };
This allows for very effective and concise testing of directives:
const {asFragment, setArg} = renderDirective(myDirective)
expect(asFragment()).toBe('<div data-directive="works"></div>')
setArg('perfect')
expect(asFragment()).toBe('<div data-directive="perfect"></div>')
Async methods
Solid.js reactive changes are pretty instantaneous, so there is rarely need to
use waitFor(…)
, await findByRole(…)
and other asynchronous queries to test
the rendered result, except for transitions, suspense, resources and router
navigation.
Solid.js manages side effects with different variants of createEffect
. While
you can use waitFor
to test asynchronous effects, it uses polling instead of
allowing Solid's reactivity to trigger the next step. In order to simplify
testing those asynchronous effects, we have a testEffect
helper that
complements the hooks for directives and hooks:
testEffect(fn: (done: (result: T) => void) => void, owner?: Owner): Promise<T>
// use it like this:
test("testEffect allows testing an effect asynchronously", () => {
const [value, setValue] = createSignal(0);
return testEffect(done => createEffect((run: number = 0) => {
if (run === 0) {
expect(value()).toBe(0);
setValue(1);
} else if (run === 1) {
expect(value()).toBe(1);
done();
}
return run + 1;
}));
});
It allows running the effect inside a defined owner that is received as an
optional second argument. This can be useful in combination with renderHook
,
which gives you an owner field in its result. The return value is a Promise with
the value given to the done()
callback. You can either await the result for
further assertions or return it to your test runner.
Known issues
If you are using vitest
, then tests might fail, because
the packages solid-js
, and @solidjs/router
(if used) need to be loaded only
once, and they could be loaded both through the internal vite
server and
through node. Typical bugs that happen because of this is that dispose is
supposedly undefined, or the router could not be loaded.
Since version 2.8.2, our vite plugin has gained the capability to configure everything for testing, so you should only need extra configuration for globals, coverage, etc.