d3ro-voice/packages/api-client/__tests__/transcribe.test.ts

// packages/api-client/__tests__/transcribe.test.ts
// transcribeAudio 함수 단위 및 통합 인터페이스 테스트

import { describe, it, expect, vi, beforeEach } from 'vitest'
import { transcribeAudio } from '../src/transcribe'

describe('transcribeAudio', () => {
  beforeEach(() => {
    vi.restoreAllMocks()
  })

  it('Blob 입력 시 multipart/form-data로 /api/stt/transcribe를 호출하여 전사 결과를 반환한다', async () => {
    const mockAudioBlob = new Blob(['mock-audio-content'], { type: 'audio/webm' })
    const mockResponse = {
      text: '안녕하세요, 클라우드 음성 전사 테스트입니다.',
      confidence: 0.99,
      language: 'ko',
      durationSeconds: 2.5,
      provider: 'groq',
      modelId: 'whisper-large-v3-turbo',
      latencyMs: 145,
      cost: 0.000021,
    }

    const fetchMock = vi.fn().mockResolvedValue({
      ok: true,
      status: 200,
      json: async () => mockResponse,
    })
    vi.stubGlobal('fetch', fetchMock)

    const result = await transcribeAudio({
      audio: mockAudioBlob,
      language: 'ko',
      apiBaseUrl: 'http://localhost:5000',
      token: 'test-token',
    })

    expect(fetchMock).toHaveBeenCalledTimes(1)
    const [url, options] = fetchMock.mock.calls[0]
    expect(url).toBe('http://localhost:5000/api/stt/transcribe')
    expect(options.method).toBe('POST')
    expect(options.headers['Authorization']).toBe('Bearer test-token')
    expect(options.body).toBeInstanceOf(FormData)

    expect(result.text).toBe('안녕하세요, 클라우드 음성 전사 테스트입니다.')
    expect(result.provider).toBe('groq')
    expect(result.durationSeconds).toBe(2.5)
    expect(result.latencyMs).toBe(145)
  })

  it('Base64 문자열 입력 시 JSON 페이로드로 전송한다', async () => {
    const mockBase64 = 'UklGRiQAAABXQVZFZm10IBAAAAABAAEAQB8AAEAfAAABAAgAZGF0YQAAAAA='
    const mockResponse = {
      text: 'Base64 전사 성공',
      confidence: 0.98,
      language: 'ko',
      durationSeconds: 1.0,
      provider: 'openai',
      modelId: 'whisper-1',
      latencyMs: 320,
      cost: 0.0001,
    }

    const fetchMock = vi.fn().mockResolvedValue({
      ok: true,
      status: 200,
      json: async () => mockResponse,
    })
    vi.stubGlobal('fetch', fetchMock)

    const result = await transcribeAudio({
      audio: mockBase64,
      language: 'ko',
      prompt: '의료 용어 가이드',
      apiBaseUrl: 'http://localhost:5000',
    })

    expect(fetchMock).toHaveBeenCalledTimes(1)
    const [url, options] = fetchMock.mock.calls[0]
    expect(url).toBe('http://localhost:5000/api/stt/transcribe')
    expect(options.headers['Content-Type']).toBe('application/json')
    const body = JSON.parse(options.body)
    expect(body.audioBase64).toBe(mockBase64)
    expect(body.language).toBe('ko')
    expect(body.initialPrompt).toBe('의료 용어 가이드')

    expect(result.text).toBe('Base64 전사 성공')
  })

  it('서버 응답이 4xx/5xx 실패 시 에러를 throw한다', async () => {
    vi.stubGlobal(
      'fetch',
      vi.fn().mockResolvedValue({
        ok: false,
        status: 500,
        text: async () => 'Internal STT Proxy Failure',
      })
    )

    await expect(
      transcribeAudio({
        audio: new Blob(['data']),
        apiBaseUrl: 'http://localhost:5000',
      })
    ).rejects.toThrow('STT transcription failed (500): Internal STT Proxy Failure')
  })
})