feat: initial release v0.3.0

tests/unit/providers/bailian-tts.test.ts

@@ -0,0 +1,145 @@
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { synthesizeWithBailianTTS } from '@/lib/providers/bailian/tts'
+
+function buildWavBuffer(durationMs: number): Buffer {
+  const sampleRate = 8000
+  const channels = 1
+  const bitsPerSample = 16
+  const byteRate = sampleRate * channels * (bitsPerSample / 8)
+  const dataLength = Math.round((durationMs / 1000) * byteRate)
+  const pcmData = Buffer.alloc(dataLength, 0)
+  const output = Buffer.alloc(44 + dataLength)
+
+  output.write('RIFF', 0, 'ascii')
+  output.writeUInt32LE(36 + dataLength, 4)
+  output.write('WAVE', 8, 'ascii')
+  output.write('fmt ', 12, 'ascii')
+  output.writeUInt32LE(16, 16)
+  output.writeUInt16LE(1, 20)
+  output.writeUInt16LE(channels, 22)
+  output.writeUInt32LE(sampleRate, 24)
+  output.writeUInt32LE(byteRate, 28)
+  output.writeUInt16LE(channels * (bitsPerSample / 8), 32)
+  output.writeUInt16LE(bitsPerSample, 34)
+  output.write('data', 36, 'ascii')
+  output.writeUInt32LE(dataLength, 40)
+  pcmData.copy(output, 44)
+
+  return output
+}
+
+describe('bailian tts synthesis', () => {
+  beforeEach(() => {
+    vi.restoreAllMocks()
+  })
+
+  it('synthesizes one segment and returns wav buffer', async () => {
+    const wav = buildWavBuffer(120)
+    const fetchMock = vi.fn(async (input: string) => {
+      if (input.includes('/multimodal-generation/generation')) {
+        return {
+          ok: true,
+          text: async () => JSON.stringify({
+            output: {
+              audio: {
+                url: 'https://audio.example/segment-1.wav',
+              },
+            },
+            usage: { characters: 10 },
+            request_id: 'req-1',
+          }),
+        }
+      }
+      if (input === 'https://audio.example/segment-1.wav') {
+        return {
+          ok: true,
+          status: 200,
+          arrayBuffer: async () => wav.buffer.slice(wav.byteOffset, wav.byteOffset + wav.byteLength),
+        }
+      }
+      throw new Error(`unexpected fetch url: ${input}`)
+    })
+    vi.stubGlobal('fetch', fetchMock as unknown as typeof fetch)
+
+    const result = await synthesizeWithBailianTTS({
+      text: '你好，世界',
+      voiceId: 'voice_1',
+    }, 'bl-key')
+
+    expect(result.success).toBe(true)
+    expect(result.audioData).toBeDefined()
+    expect(result.audioDuration).toBe(120)
+    expect(result.audioUrl).toBe('https://audio.example/segment-1.wav')
+    expect(result.characters).toBe(10)
+    expect(fetchMock).toHaveBeenCalledTimes(2)
+  })
+
+  it('splits text over 600 chars and merges audio segments', async () => {
+    const wavA = buildWavBuffer(100)
+    const wavB = buildWavBuffer(200)
+    let generationCallCount = 0
+    const fetchMock = vi.fn(async (input: string) => {
+      if (input.includes('/multimodal-generation/generation')) {
+        generationCallCount += 1
+        const audioUrl = generationCallCount === 1
+          ? 'https://audio.example/segment-a.wav'
+          : 'https://audio.example/segment-b.wav'
+        return {
+          ok: true,
+          text: async () => JSON.stringify({
+            output: {
+              audio: { url: audioUrl },
+            },
+            usage: { characters: 600 },
+            request_id: `req-${generationCallCount}`,
+          }),
+        }
+      }
+      if (input === 'https://audio.example/segment-a.wav') {
+        return {
+          ok: true,
+          status: 200,
+          arrayBuffer: async () => wavA.buffer.slice(wavA.byteOffset, wavA.byteOffset + wavA.byteLength),
+        }
+      }
+      if (input === 'https://audio.example/segment-b.wav') {
+        return {
+          ok: true,
+          status: 200,
+          arrayBuffer: async () => wavB.buffer.slice(wavB.byteOffset, wavB.byteOffset + wavB.byteLength),
+        }
+      }
+      throw new Error(`unexpected fetch url: ${input}`)
+    })
+    vi.stubGlobal('fetch', fetchMock as unknown as typeof fetch)
+
+    const result = await synthesizeWithBailianTTS({
+      text: 'a'.repeat(601),
+      voiceId: 'voice_2',
+    }, 'bl-key')
+
+    expect(result.success).toBe(true)
+    expect(result.audioData).toBeDefined()
+    expect(result.audioDuration).toBe(300)
+    expect(result.audioUrl).toBeUndefined()
+    expect(result.characters).toBe(1200)
+    expect(generationCallCount).toBe(2)
+  })
+
+  it('fails explicitly when voiceId is missing', async () => {
+    const fetchMock = vi.fn()
+    vi.stubGlobal('fetch', fetchMock as unknown as typeof fetch)
+
+    const result = await synthesizeWithBailianTTS({
+      text: 'hello',
+      voiceId: '',
+    }, 'bl-key')
+
+    expect(result).toEqual({
+      success: false,
+      error: 'BAILIAN_TTS_VOICE_ID_REQUIRED',
+    })
+    expect(fetchMock).not.toHaveBeenCalled()
+  })
+})
+