- Default to the `*ByRole` queries when testing components because it encourages testing with accessibility concerns in mind.
- Alternatively, you could use `*ByLabelText` queries for testing components. However, we recommend the `*ByRole` queries because they are [more robust](https://testing-library.com/docs/queries/bylabeltext/#name).
We use the [user-event](https://testing-library.com/docs/user-event/intro) library for simulating user interactions during testing. This library is preferred over the built-in `fireEvent` method, as it more accurately mirrors real user interactions with elements.
There are two important considerations when working with `userEvent`:
1. All methods in `userEvent` are asynchronous, and thus require the use of `await` when called.
1. Directly calling methods from `userEvent` may not be supported in future versions. As such, it's necessary to first call `userEvent.setup()` prior to the tests. This method returns a `userEvent` instance, complete with all its methods. This setup process can be simplified using a utility function:
- [screen.debug()](https://testing-library.com/docs/queries/about/#screendebug) - This function prints a human-readable representation of the document's DOM tree when called without arguments, or the DOM tree of specific node or nodes when provided with arguments. It is internally using `console.log` to log the output to terminal.
- [Testing Playground](https://testing-playground.com/) - An interactive sandbox that allows testing of which queries work with specific HTML elements.
- [prettyDOM logRoles](https://testing-library.com/docs/dom-testing-library/api-debugging/#prettydom) - A utility function that prints out all the implicit ARIA roles for a given DOM tree.
Here, the [OrgRolePicker](https://github.com/grafana/grafana/blob/38863844e7ac72c7756038a1097f89632f9985ff/public/app/features/admin/OrgRolePicker.tsx) component is used as an example. This component essentially serves as a wrapper for the `Select` component, complete with its own set of options.
It is a recommended practice to query `Select` components by using a label. Add a `label` element and provide the `htmlFor` prop with a matching `inputId`. Alternatively, you can specify `aria-label` on the `Select` statement.
At times, it might be necessary to verify that the `Select` component is displaying the correct options. In such instances, the best solution is to click the `Select` component and match the desired option using the `*ByText` query.
```tsx
it('should have an "Editor" option', async () => {
To simplify the process of selecting an option from a `Select` component, there is a `selectOptionInTest` utility function. This function is a wrapper over the [react-select-event](https://testing-library.com/docs/ecosystem-react-select-event/) package.
The recommended approach for mocking the `window` object is to use [Jest spies](https://jestjs.io/docs/jest-object). Jest's spy functions provide a built-in mechanism for restoring mocks. This feature eliminates the need to manually save a reference to the `window` object.
Here's what the test looks like for this [OrgPicker](https://github.com/grafana/grafana/blob/38863844e7ac72c7756038a1097f89632f9985ff/public/app/core/components/Select/OrgPicker.tsx) component, which uses `AsyncSelect` under the hood: