Enterprise agent platform for building, deploying, and operating AI teammates.
import React from "react"
import { act, cleanup, fireEvent, render, renderHook, screen, waitFor } from "@testing-library/react"
import { afterEach, beforeEach, describe, expect, it, vi } from "vitest"
const apiRequestMock = vi.hoisted(() => vi.fn())
const toastErrorMock = vi.hoisted(() => vi.fn())
vi.mock("@/lib/api-wrapper", async () => {
const actual = await vi.importActual<typeof import("@/lib/api-wrapper")>(
"@/lib/api-wrapper"
)
return {
...actual,
apiRequest: apiRequestMock,
}
})
vi.mock("@/lib/utils", async () => {
const actual = await vi.importActual<typeof import("@/lib/utils")>("@/lib/utils")
return {
...actual,
cn: (...classes: Array<string | false | null | undefined>) =>
classes.filter(Boolean).join(" "),
getApiUrl: () => "http://api.local",
getUploadApiUrl: () => "http://upload.local",
}
})
vi.mock("@/contexts/i18n-context", () => ({
useI18n: () => ({
t: (key: string) => key,
}),
}))
vi.mock("@/components/ui/sonner", () => ({
toast: {
error: toastErrorMock,
},
}))
vi.mock("lucide-react", () => {
const Icon = (props: React.SVGProps<SVGSVGElement>) => <svg {...props} />
return {
Loader2: Icon,
Mic: Icon,
Square: Icon,
}
})
import { VoiceInputController, useVoiceInputControls } from "./voice-input-controller"
function jsonResponse(data: unknown): Response {
return new Response(JSON.stringify(data), {
status: 200,
headers: { "Content-Type": "application/json" },
})
}
function setVisibleRect(
element: HTMLElement,
rect: Partial<DOMRect> = {}
): void {
const visibleRect = {
bottom: 50,
height: 40,
left: 10,
right: 210,
top: 10,
width: 200,
x: 10,
y: 10,
...rect,
}
element.getBoundingClientRect = () =>
({
...visibleRect,
toJSON: () => ({}),
}) as DOMRect
}
class MediaRecorderMock {
static isTypeSupported = vi.fn(() => true)
mimeType: string
ondataavailable: ((event: BlobEvent) => void) | null = null
onerror: ((event: Event) => void) | null = null
onstop: ((event: Event) => void) | null = null
state: RecordingState = "inactive"
constructor(_stream: MediaStream, options?: MediaRecorderOptions) {
this.mimeType = options?.mimeType || "audio/webm"
}
start(): void {
this.state = "recording"
}
stop(): void {
this.state = "inactive"
const blob = new Blob(["audio"], { type: this.mimeType })
this.ondataavailable?.({ data: blob } as BlobEvent)
this.onstop?.(new Event("stop"))
}
}
describe("VoiceInputController", () => {
beforeEach(() => {
apiRequestMock.mockReset()
toastErrorMock.mockReset()
apiRequestMock.mockImplementation((url: string) => {
if (url === "http://api.local/api/models/?category=speech&limit=1000") {
return Promise.resolve(jsonResponse([{ abilities: ["asr"] }]))
}
if (url === "http://upload.local/api/models/speech/transcribe") {
return Promise.resolve(jsonResponse({ text: "voice text" }))
}
return Promise.reject(new Error(`unexpected request: ${url}`))
})
Object.defineProperty(globalThis, "MediaRecorder", {
configurable: true,
value: MediaRecorderMock,
})
Object.defineProperty(window, "MediaRecorder", {
configurable: true,
value: MediaRecorderMock,
})
Object.defineProperty(navigator, "mediaDevices", {
configurable: true,
value: {
getUserMedia: vi.fn().mockResolvedValue({
getTracks: () => [{ stop: vi.fn() }],
} as unknown as MediaStream),
},
})
})
afterEach(() => {
cleanup()
})
it("requires a file ASR model even when a custom realtime row advertises ASR", async () => {
const realtime = {
model_id: "custom-live-model",
model_provider: "elevenlabs",
model_name: "scribe_v2_realtime",
abilities: ["asr"],
}
const batch = { ...realtime, model_id: "file-asr", model_name: "scribe_v2" }
const { result } = renderHook(() => useVoiceInputControls({ autoRefresh: false }))
apiRequestMock.mockResolvedValueOnce(jsonResponse([realtime]))
await act(() => result.current.refreshAvailability())
expect(result.current.hasAsrModel).toBe(false)
apiRequestMock.mockResolvedValueOnce(jsonResponse([realtime, batch]))
await act(() => result.current.refreshAvailability())
expect(result.current.hasAsrModel).toBe(true)
apiRequestMock.mockResolvedValueOnce(jsonResponse([realtime]))
await act(() => result.current.refreshAvailability())
expect(result.current.hasAsrModel).toBe(false)
})
it("inserts transcription into the field active when recording started", async () => {
render(
<>
<input aria-label="first prompt" data-voice-input="true" />
<input aria-label="second prompt" data-voice-input="true" />
<VoiceInputController />
</>
)
const first = screen.getByLabelText("first prompt") as HTMLInputElement
const second = screen.getByLabelText("second prompt") as HTMLInputElement
setVisibleRect(first)
setVisibleRect(second)
first.focus()
fireEvent.focusIn(first)
fireEvent.click(await screen.findByLabelText("voiceInput.start"))
await screen.findByLabelText("voiceInput.stop")
second.focus()
fireEvent.focusIn(second)
fireEvent.click(screen.getByLabelText("voiceInput.stop"))
await waitFor(() => {
expect(first).toHaveValue("voice text")
})
expect(second).toHaveValue("")
})
it("shows an error when voice recording is unsupported", async () => {
render(
<>
<input aria-label="prompt" data-voice-input="true" />
<VoiceInputController />
</>
)
const input = screen.getByLabelText("prompt")
setVisibleRect(input)
input.focus()
fireEvent.focusIn(input)
Object.defineProperty(globalThis, "MediaRecorder", {
configurable: true,
value: undefined,
})
Object.defineProperty(window, "MediaRecorder", {
configurable: true,
value: undefined,
})
Object.defineProperty(navigator, "mediaDevices", {
configurable: true,
value: undefined,
})
fireEvent.click(await screen.findByLabelText("voiceInput.start"))
expect(toastErrorMock).toHaveBeenCalledWith("voiceInput.errors.unsupported")
})
it("does not enable arbitrary contentEditable regions without opt-in", () => {
render(
<>
<div contentEditable suppressContentEditableWarning>
private draft
</div>
<VoiceInputController />
</>
)
const editable = screen.getByText("private draft")
setVisibleRect(editable)
editable.focus()
fireEvent.focusIn(editable)
expect(apiRequestMock).not.toHaveBeenCalled()
expect(screen.queryByLabelText("voiceInput.start")).not.toBeInTheDocument()
})
it("does not enable regular form fields without opt-in", () => {
render(
<>
<input aria-label="agent name" />
<VoiceInputController />
</>
)
const input = screen.getByLabelText("agent name")
setVisibleRect(input)
input.focus()
fireEvent.focusIn(input)
expect(apiRequestMock).not.toHaveBeenCalled()
expect(screen.queryByLabelText("voiceInput.start")).not.toBeInTheDocument()
})
it("positions contentEditable controls against the enclosing input shell", async () => {
render(
<>
<div data-testid="input-shell">
<div
contentEditable
data-voice-input="true"
role="textbox"
suppressContentEditableWarning
/>
</div>
<VoiceInputController />
</>
)
const shell = screen.getByTestId("input-shell")
const editor = screen.getByRole("textbox")
Object.defineProperty(editor, "isContentEditable", {
configurable: true,
value: true,
})
setVisibleRect(shell, {
bottom: 160,
height: 140,
left: 40,
right: 440,
top: 20,
width: 400,
x: 40,
y: 20,
})
setVisibleRect(editor, {
bottom: 82,
height: 40,
left: 60,
right: 130,
top: 42,
width: 70,
x: 60,
y: 42,
})
editor.focus()
fireEvent.focusIn(editor)
const button = await screen.findByLabelText("voiceInput.start")
expect(button).toHaveStyle({ left: "336px", top: "104px" })
})
it("shows textarea controls on hover against the enclosing input shell", async () => {
render(
<>
<div data-testid="input-shell">
<textarea aria-label="task prompt" data-voice-input="true" />
</div>
<VoiceInputController />
</>
)
const shell = screen.getByTestId("input-shell")
const textarea = screen.getByLabelText("task prompt")
setVisibleRect(shell, {
bottom: 160,
height: 140,
left: 40,
right: 440,
top: 20,
width: 400,
x: 40,
y: 20,
})
setVisibleRect(textarea, {
bottom: 82,
height: 40,
left: 60,
right: 130,
top: 42,
width: 70,
x: 60,
y: 42,
})
fireEvent.pointerOver(textarea)
const button = await screen.findByLabelText("voiceInput.start")
expect(button).toHaveStyle({ left: "336px", top: "104px" })
})
it("does not float controls over large framed textareas", () => {
render(
<>
<div data-testid="input-shell">
<textarea aria-label="large prompt" />
</div>
<VoiceInputController />
</>
)
const shell = screen.getByTestId("input-shell")
const textarea = screen.getByLabelText("large prompt")
setVisibleRect(shell, {
bottom: 180,
height: 160,
left: 40,
right: 440,
top: 20,
width: 400,
x: 40,
y: 20,
})
setVisibleRect(textarea, {
bottom: 160,
height: 120,
left: 60,
right: 420,
top: 40,
width: 360,
x: 60,
y: 40,
})
fireEvent.pointerOver(textarea)
expect(apiRequestMock).not.toHaveBeenCalled()
expect(screen.queryByLabelText("voiceInput.start")).not.toBeInTheDocument()
})
it("does not float controls over textareas with nearby action buttons", () => {
render(
<>
<div data-testid="input-shell">
<textarea aria-label="action prompt" />
<button type="button">Send</button>
</div>
<VoiceInputController />
</>
)
const shell = screen.getByTestId("input-shell")
const textarea = screen.getByLabelText("action prompt")
const button = screen.getByText("Send")
setVisibleRect(shell, {
bottom: 90,
height: 70,
left: 40,
right: 440,
top: 20,
width: 400,
x: 40,
y: 20,
})
setVisibleRect(textarea, {
bottom: 70,
height: 40,
left: 60,
right: 360,
top: 30,
width: 300,
x: 60,
y: 30,
})
setVisibleRect(button, {
bottom: 72,
height: 32,
left: 372,
right: 428,
top: 40,
width: 56,
x: 372,
y: 40,
})
fireEvent.pointerOver(textarea)
expect(apiRequestMock).not.toHaveBeenCalled()
expect(screen.queryByLabelText("voiceInput.start")).not.toBeInTheDocument()
})
})