Evenfire is an intelligence layer platform for running LLM agents that take real actions, on your infrastructure
TypeScript
// @vitest-environment jsdom
import { afterEach, describe, expect, it, vi } from 'vitest'
import { fireEvent, render, screen, waitFor } from '@testing-library/react'
import { CommandPalette } from '.'
const scrollIntoView = vi.fn()
afterEach(() => {
vi.restoreAllMocks()
scrollIntoView.mockReset()
})
describe('CommandPalette arrow navigation', () => {
it('scrolls each newly selected command into the visible list area', async () => {
Object.defineProperty(HTMLElement.prototype, 'scrollIntoView', {
configurable: true,
value: scrollIntoView,
})
render(
<CommandPalette
platform="darwin"
isEligible={() => true}
onClose={vi.fn()}
onExecute={vi.fn()}
/>
)
const input = screen.getByRole('textbox', { name: 'Search commands' })
const options = screen.getAllByRole('option')
await waitFor(() => expect(options[0]?.getAttribute('aria-selected')).toBe('true'))
scrollIntoView.mockClear()
fireEvent.keyDown(input, { key: 'ArrowDown' })
await waitFor(() => {
expect(options[1]?.getAttribute('aria-selected')).toBe('true')
expect(scrollIntoView).toHaveBeenLastCalledWith({ block: 'nearest' })
})
scrollIntoView.mockClear()
fireEvent.keyDown(input, { key: 'ArrowUp' })
await waitFor(() => {
expect(options[0]?.getAttribute('aria-selected')).toBe('true')
expect(scrollIntoView).toHaveBeenLastCalledWith({ block: 'nearest' })
})
})
})