d3ro-voice/apps/desktop/tests/main/services/VoiceModeService.test.ts

// tests/main/services/VoiceModeService.test.ts
// 상태 머신 전이 + 이중 조건 플러시 + accidentalPress 테스트

import { describe, it, expect, beforeEach, vi } from 'vitest'
import { EventEmitter } from 'events'
import { RecognitionState, AudioState } from '@d3ro/core/types'
import { TIMING } from '@d3ro/core/constants'

// 모든 하위 서비스 모킹
vi.mock('../../../src/main/services/LoggerService', () => ({
  getLogger: () => ({
    info: vi.fn(),
    warn: vi.fn(),
    error: vi.fn(),
    debug: vi.fn()
  })
}))

const mockSTT = {
  initialize: vi.fn(() => Promise.resolve()),
  transcribe: vi.fn(() =>
    Promise.resolve({ text: '테스트 전사', segments: [], language: 'ko', duration: 2, processingTime: 500 })
  ),
  getStatus: vi.fn(() => ({ state: 'ready', modelId: 'base', uptime: 0 })),
  // 프리플라이트 검증(모델 설치 여부)용 — 기본은 설치된 상태로 목킹
  getModels: vi.fn(() => [
    { id: 'base', name: 'Base', sizeBytes: 0, downloaded: true, languages: [], accuracy: 2, speed: 4 },
  ]),
  on: vi.fn(),
  off: vi.fn()
}

vi.mock('../../../src/main/services/LocalSTTService', () => ({
  getLocalSTTService: () => mockSTT,
  resetLocalSTTServiceForTests: () => undefined,
}))

const audioBus = new EventEmitter()
const mockAudio = {
  start: vi.fn(() => Promise.resolve()),
  stop: vi.fn(() => Promise.resolve()),
  on: vi.fn((ev: string, fn: (...args: unknown[]) => void) => {
    audioBus.on(ev, fn)
  }),
  off: vi.fn((ev: string, fn: (...args: unknown[]) => void) => {
    audioBus.off(ev, fn)
  }),
}

vi.mock('../../../src/main/services/AudioCaptureService', () => ({
  getAudioCaptureService: () => mockAudio
}))

const mockKeyBinding = {
  on: vi.fn(),
  off: vi.fn()
}

vi.mock('../../../src/main/services/KeyBindingService', () => ({
  getKeyBindingService: () => mockKeyBinding
}))

vi.mock('../../../src/main/services/ConfigService', () => ({
  configGet: vi.fn((key: string) => {
    const defaults: Record<string, unknown> = {
      sttModelId: 'base',
      defaultLLMAction: 'refine',
      ollamaServerUrl: 'http://localhost:11434',
      llmModelId: 'gemma4:e4b'
    }
    return defaults[key]
  })
}))

const mockTextInsert = {
  insertText: vi.fn(() => Promise.resolve({ success: true, method: 'clipboard', textLength: 10, durationMs: 50 }))
}

vi.mock('../../../src/main/services/TextInsertService', () => ({
  getTextInsertService: () => mockTextInsert
}))

const mockLLM = {
  isAvailable: vi.fn(() => false),
  processText: vi.fn(() => Promise.resolve('다듬어진 텍스트')),
  on: vi.fn(),
  off: vi.fn()
}

vi.mock('../../../src/main/services/LocalLLMService', () => ({
  getLocalLLMService: () => mockLLM
}))

let getVoiceModeService: () => ReturnType<typeof import('../../../src/main/services/VoiceModeService')['getVoiceModeService']>

beforeEach(async () => {
  vi.resetModules()
  vi.clearAllMocks()
  audioBus.removeAllListeners()
  mockSTT.initialize.mockResolvedValue(undefined as never)
  mockSTT.transcribe.mockResolvedValue({
    text: '테스트 전사',
    segments: [],
    language: 'ko',
    duration: 2,
    processingTime: 500,
  } as never)
  mockLLM.processText.mockResolvedValue('다듬어진 텍스트' as never)
  const mod = await import('../../../src/main/services/VoiceModeService')
  mod.resetVoiceModeServiceForTests()
  getVoiceModeService = mod.getVoiceModeService
})

describe('VoiceModeService', () => {
  describe('상태 머신', () => {
    it('초기 상태는 IDLE이다', () => {
      const svc = getVoiceModeService()
      const state = svc.getState()

      expect(state.recognitionState).toBe(RecognitionState.IDLE)
      expect(state.audioState).toBe(AudioState.IDLE)
      expect(state.sessionId).toBeNull()
    })

    it('startSession 호출 시 PREPARING으로 전이한다', async () => {
      const svc = getVoiceModeService()
      const stateChanges: RecognitionState[] = []

      svc.on('recognition-state-changed', (payload: { current: RecognitionState }) => {
        stateChanges.push(payload.current)
      })

      await svc.startSession('dictation')

      // PREPARING → CONNECTING → READY 순서
      expect(stateChanges[0]).toBe(RecognitionState.PREPARING)
      expect(stateChanges).toContain(RecognitionState.CONNECTING)
    })

    it('isActive는 세션이 활성일 때 true이다', async () => {
      const svc = getVoiceModeService()
      expect(svc.isActive).toBe(false)

      // startSession은 완전 비동기이므로 await 후 세션 활성 확인
      await svc.startSession('dictation')
      expect(svc.isActive).toBe(true)
    })
  })

  describe('accidentalPress', () => {
    it('700ms 미만 세션은 자동 취소된다', async () => {
      const svc = getVoiceModeService()
      let cancelReason: string | null = null

      svc.on('session-cancelled', (payload: { reason: string }) => {
        cancelReason = payload.reason
      })

      // 세션 시작 즉시 종료 (700ms 미만)
      await svc.startSession('dictation')
      await svc.stopSession()

      expect(cancelReason).toBe('too-short')
    })
  })

  describe('STT 준비 대기', () => {
    it('전사가 시작된 뒤에는 readiness timeout 으로 세션을 취소하지 않는다', async () => {
      let resolveInit: (() => void) | undefined
      mockSTT.initialize.mockImplementation(
        () =>
          new Promise<void>((resolve) => {
            resolveInit = resolve
          }),
      )
      mockSTT.transcribe.mockResolvedValue({
        text: '안녕하세요',
        segments: [],
        language: 'ko',
        duration: 2,
        processingTime: 50,
      } as never)

      const svc = getVoiceModeService()
      let cancelReason: string | null = null
      let completed = false
      svc.on('session-cancelled', (payload: { reason: string }) => {
        cancelReason = payload.reason
      })
      const completedPromise = new Promise<void>((resolve) => {
        svc.once('session-completed', () => {
          completed = true
          resolve()
        })
      })

      await svc.startSession('dictation')
      audioBus.emit('audio-data', { buffer: Buffer.alloc(16000 * 2) })
      await new Promise((r) => setTimeout(r, 850))
      await svc.stopSession()

      resolveInit?.()
      await Promise.race([completedPromise, new Promise((r) => setTimeout(r, 1000))])

      expect(cancelReason).not.toBe('timeout')
      expect(completed).toBe(true)
    })
  })

  describe('cancelSession', () => {
    it('user 취소로 세션을 종료한다', async () => {
      const svc = getVoiceModeService()
      let cancelReason: string | null = null

      svc.on('session-cancelled', (payload: { reason: string }) => {
        cancelReason = payload.reason
      })

      await svc.startSession('dictation')
      svc.cancelSession()

      expect(cancelReason).toBe('user')
    })
  })

  describe('getState', () => {
    it('현재 상태를 VoiceState 형태로 반환한다', () => {
      const svc = getVoiceModeService()
      const state = svc.getState()

      expect(state).toHaveProperty('recognitionState')
      expect(state).toHaveProperty('audioState')
      expect(state).toHaveProperty('mode')
      expect(state).toHaveProperty('sessionId')
      expect(state).toHaveProperty('recordingStartedAt')
    })
  })

  describe('터미널 상태', () => {
    it('cancelSession 후 _resetToIdle의 200ms 딜레이 후 IDLE로 전이한다', async () => {
      vi.useFakeTimers()
      const svc = getVoiceModeService()

      await svc.startSession('dictation')
      svc.cancelSession()

      // 200ms 딜레이로 IDLE 전이 예약됨
      vi.advanceTimersByTime(250)

      const state = svc.getState()
      expect(state.recognitionState).toBe(RecognitionState.IDLE)

      vi.useRealTimers()
    })
  })
})

describe('TIMING 상수', () => {
  it('핵심 타이밍 값이 Speakly 패턴과 일치한다', () => {
    expect(TIMING.MIN_AUDIO_DURATION).toBe(700)
    expect(TIMING.DOUBLE_PRESS_DURATION).toBe(300)
    expect(TIMING.POST_RECORDING_WAIT).toBe(4000)
    expect(TIMING.POST_RECORDING_WAIT_BUFFERED).toBe(6000)
    expect(TIMING.ABSOLUTE_MAX_WAIT).toBe(120000)
    expect(TIMING.AUDIO_LEVEL_INTERVAL).toBe(100)
  })
})