@mingxy/ocosay 1.0.0

package/src/core/stream-reader.ts

@@ -0,0 +1,190 @@
+/**
+ * StreamReader - 流式文本缓冲与句子边界检测
+ * 
+ * 功能：
+ * - 订阅 TuiEventBus 的 message.part.delta 事件
+ * - 缓冲区满或遇到句子结束符时触发 textReady 事件
+ * - 支持超时强制发送
+ */
+
+import { EventEmitter } from 'events'
+import { StreamState, TTSError, TTSErrorCode } from './types'
+
+export class StreamReader extends EventEmitter {
+  private state: StreamState = StreamState.IDLE
+  private buffer: string = ''
+  private sessionID?: string
+  private messageID?: string
+  private partID?: string
+  private timeoutHandle?: NodeJS.Timeout
+  
+  constructor(
+    private bufferSize: number = 30,
+    private bufferTimeout: number = 2000
+  ) {
+    super()
+  }
+
+  /**
+   * 启动流式监听
+   * 将状态从 IDLE 切换到 BUFFERING，开始监听事件
+   */
+  start(): void {
+    if (this.state === StreamState.IDLE) {
+      this.state = StreamState.BUFFERING
+      this.emit('streamStart')
+    }
+  }
+  
+  /**
+   * 处理 message.part.delta 事件
+   */
+  handleDelta(sessionID: string, messageID: string, partID: string, delta: string): void {
+    if (this.state === StreamState.IDLE) {
+      this.state = StreamState.BUFFERING
+      this.sessionID = sessionID
+      this.messageID = messageID
+      this.partID = partID
+      this.emit('streamStart')
+    }
+    
+    this.buffer += delta
+    this.resetTimeout()
+    
+    if (this.shouldFlush()) {
+      this.flushBuffer()
+    }
+  }
+  
+  /**
+   * 处理流结束
+   */
+  handleEnd(): void {
+    if (this.state === StreamState.ENDED) {
+      return
+    }
+    if (this.buffer.length > 0) {
+      this.flushBuffer()
+    }
+    this.state = StreamState.ENDED
+    this.clearTimeout()
+    this.emit('streamEnd')
+  }
+  
+  /**
+   * 处理错误
+   */
+  handleError(error: TTSError): void {
+    this.clearTimeout()
+    this.state = StreamState.IDLE
+    this.buffer = ''
+    this.emit('streamError', error)
+  }
+  
+  /**
+   * 重置缓冲器
+   */
+  reset(): void {
+    this.state = StreamState.IDLE
+    this.buffer = ''
+    this.sessionID = undefined
+    this.messageID = undefined
+    this.partID = undefined
+    this.clearTimeout()
+  }
+  
+  /**
+   * 判断是否应该刷新缓冲区
+   * 条件：
+   * 1. 包含句子结束符（任何长度）
+   * 2. 缓冲区长度 >= bufferSize
+   */
+  private shouldFlush(): boolean {
+    // 句子结束标记：。！？.!?……（中文句号、感叹号、问号、省略号）
+    const sentenceEnd = /[。！？.!?]|……/
+    if (sentenceEnd.test(this.buffer)) {
+      return true
+    }
+    // 缓冲区达到阈值
+    if (this.buffer.length >= this.bufferSize) {
+      return true
+    }
+    return false
+  }
+  
+  /**
+   * 刷新缓冲区，发送textReady事件
+   */
+  private flushBuffer(): void {
+    const text = this.buffer.trim()
+    if (text.length > 0) {
+      this.emit('textReady', text)
+    }
+    this.buffer = ''
+    this.resetTimeout()
+  }
+  
+  /**
+   * 重置超时计时器
+   */
+  private resetTimeout(): void {
+    this.clearTimeout()
+    this.timeoutHandle = setTimeout(() => {
+      if (this.buffer.length > 0) {
+        this.flushBuffer()
+      }
+    }, this.bufferTimeout)
+  }
+  
+  /**
+   * 清除超时计时器
+   */
+  private clearTimeout(): void {
+    if (this.timeoutHandle) {
+      clearTimeout(this.timeoutHandle)
+      this.timeoutHandle = undefined
+    }
+  }
+  
+  /**
+   * 获取当前状态
+   */
+  getState(): StreamState {
+    return this.state
+  }
+  
+  /**
+   * 检查流是否处于活跃状态
+   */
+  isActive(): boolean {
+    return this.state === StreamState.BUFFERING
+  }
+  
+  /**
+   * 获取当前缓冲区内容
+   */
+  getBuffer(): string {
+    return this.buffer
+  }
+  
+  /**
+   * 获取当前会话ID
+   */
+  getSessionID(): string | undefined {
+    return this.sessionID
+  }
+  
+  /**
+   * 获取当前消息ID
+   */
+  getMessageID(): string | undefined {
+    return this.messageID
+  }
+  
+  /**
+   * 获取当前分块ID
+   */
+  getPartID(): string | undefined {
+    return this.partID
+  }
+}

package/src/core/streaming-synthesizer.ts

@@ -0,0 +1,123 @@
+/**
+ * StreamingSynthesizer - 流式合成器
+ * 
+ * 功能：
+ * - 接收 StreamReader 发来的文本（通过 synthesize 方法）
+ * - 调用 TTSProvider 的流式合成接口
+ * - 将返回的音频 chunk 传递给下游（StreamPlayer）
+ * 
+ * 数据流：
+ * StreamReader.textReady → StreamingSynthesizer.synthesize() → StreamPlayer (边收边播)
+ */
+
+import { EventEmitter } from 'events'
+import { TTSProvider, TTSError, TTSErrorCode, StreamingSynthesizerOptions, AudioResult } from './types'
+
+export interface StreamingSynthesizerEvents {
+  on(event: 'chunk', handler: (chunk: Buffer) => void): void
+  on(event: 'error', handler: (error: TTSError) => void): void
+  on(event: 'done', handler: () => void): void
+}
+
+export class StreamingSynthesizer extends EventEmitter {
+  private audioChunks: Buffer[] = []
+  
+  constructor(private options: StreamingSynthesizerOptions) {
+    super()
+  }
+  
+  /**
+   * 发送文本片段进行合成
+   * 调用 provider.speak() 并处理返回的音频流
+   */
+  async synthesize(text: string): Promise<void> {
+    if (!text || text.trim().length === 0) {
+      return
+    }
+    
+    try {
+      const result = await this.options.provider.speak(text, {
+        model: 'stream',
+        voice: this.options.voice,
+        speed: this.options.speed,
+        volume: this.options.volume,
+        pitch: this.options.pitch
+      })
+      
+      await this.processAudioResult(result)
+      
+      this.emit('done')
+    } catch (error) {
+      const ttsError = error instanceof TTSError 
+        ? error 
+        : new TTSError(
+            error instanceof Error ? error.message : 'Synthesis failed',
+            'UNKNOWN' as TTSErrorCode,
+            this.options.provider.name,
+            error
+          )
+      this.emit('error', ttsError)
+    }
+  }
+  
+  /**
+   * 处理 AudioResult，根据 audioData 类型进行相应处理
+   */
+  private async processAudioResult(result: AudioResult): Promise<void> {
+    if (result.isStream && result.audioData instanceof ReadableStream) {
+      // 流式数据：ReadableStream
+      await this.processReadableStream(result.audioData)
+    } else if (Buffer.isBuffer(result.audioData)) {
+      // 非流式数据：Buffer
+      this.emitChunk(result.audioData)
+    }
+  }
+  
+  /**
+   * 处理 ReadableStream，逐chunk emit
+   */
+  private async processReadableStream(stream: ReadableStream): Promise<void> {
+    const reader = stream.getReader()
+    
+    try {
+      while (true) {
+        const { done, value } = await reader.read()
+        
+        if (done) {
+          break
+        }
+        
+        if (value) {
+          const chunk = Buffer.isBuffer(value) ? value : Buffer.from(value)
+          this.emitChunk(chunk)
+        }
+      }
+    } finally {
+      reader.releaseLock()
+    }
+  }
+  
+  /**
+   * emit chunk 并累积
+   */
+  private emitChunk(chunk: Buffer): void {
+    this.audioChunks.push(chunk)
+    this.emit('chunk', chunk)
+  }
+  
+  /**
+   * 重置状态
+   * 清空累积的音频数据
+   */
+  reset(): void {
+    this.audioChunks = []
+  }
+  
+  /**
+   * 获取累积的音频数据
+   * 返回所有已接收的 chunk
+   */
+  getAudioChunks(): Buffer[] {
+    return [...this.audioChunks]
+  }
+}

package/src/core/types.ts

@@ -0,0 +1,185 @@
+/**
+ * TTS Core Types
+ * 核心类型定义
+ */
+
+export enum TTSErrorCode {
+  NETWORK = 'NETWORK',
+  AUTH = 'AUTH',
+  QUOTA = 'QUOTA',
+  INVALID_VOICE = 'INVALID_VOICE',
+  INVALID_PARAMS = 'INVALID_PARAMS',
+  PLAYER_ERROR = 'PLAYER_ERROR',
+  UNKNOWN = 'UNKNOWN'
+}
+
+export class TTSError extends Error {
+  constructor(
+    message: string,
+    code: TTSErrorCode,
+    provider: string,
+    details?: unknown
+  ) {
+    super(message)
+    this.name = 'TTSError'
+    this.code = code
+    this.provider = provider
+    this.details = details
+  }
+  
+  code: TTSErrorCode
+  provider: string
+  details?: unknown
+}
+
+export interface Voice {
+  id: string
+  name: string
+  language?: string
+  gender?: 'male' | 'female' | 'neutral'
+  previewUrl?: string
+}
+
+export interface TTSCapabilities {
+  speak: true
+  voiceClone?: boolean
+  stream?: boolean
+  voiceList?: boolean
+  sync?: boolean
+  async?: boolean
+}
+
+export type SynthesisModel = 'sync' | 'async' | 'stream'
+
+export interface SpeakOptions {
+  voice?: string
+  model?: SynthesisModel
+  speed?: number
+  volume?: number
+  pitch?: number
+  sourceVoice?: string
+}
+
+export interface AudioResult {
+  audioData: Buffer | ReadableStream
+  sampleRate?: number
+  channels?: number
+  duration?: number
+  format: string
+  isStream: boolean
+}
+
+export type TTSEvent = 
+  | 'start'
+  | 'end'
+  | 'error'
+  | 'progress'
+  | 'pause'
+  | 'resume'
+  | 'stop'
+
+export interface SpeakerEvents {
+  on(event: 'start', handler: (text: string) => void): void
+  on(event: 'end', handler: (text: string) => void): void
+  on(event: 'error', handler: (error: TTSError) => void): void
+  on(event: 'progress', handler: (progress: { current: number; total: number }) => void): void
+  on(event: 'pause', handler: () => void): void
+  on(event: 'resume', handler: () => void): void
+  on(event: 'stop', handler: () => void): void
+  off(event: TTSEvent, handler: Function): void
+}
+
+export interface TTSProvider {
+  name: string
+  capabilities: TTSCapabilities
+  
+  initialize(): Promise<void>
+  destroy(): Promise<void>
+  
+  speak(text: string, options?: SpeakOptions): Promise<AudioResult>
+  pause(): Promise<void>
+  resume(): Promise<void>
+  stop(): Promise<void>
+  listVoices(): Promise<Voice[]>
+  
+  getCapabilities(): TTSCapabilities
+}
+
+// ============================================================================
+// 配置相关类型
+// ============================================================================
+
+export interface GlobalConfig {
+  defaultProvider: string
+  defaultModel?: SynthesisModel
+  defaultVoice?: string
+}
+
+export interface ProviderConfig {
+  enabled?: boolean
+  apiKey?: string
+  [key: string]: unknown
+}
+
+export interface OcosayConfig {
+  enabled?: boolean
+  autoPlay?: boolean
+  autoRead?: boolean
+  streamMode?: boolean
+  streamBufferSize?: number
+  streamBufferTimeout?: number
+  provider?: string
+  ttsModel?: string
+  baseURL?: string
+  speed?: number
+  volume?: number
+  pitch?: number
+}
+
+// ============================================================================
+// 流式朗读相关类型
+// ============================================================================
+
+export enum StreamState {
+  IDLE = 'idle',
+  BUFFERING = 'buffering',
+  STREAMING = 'streaming',
+  ENDED = 'ended'
+}
+
+export interface OcosayStreamConfig {
+  enabled: boolean
+  autoPlay: boolean
+  autoRead: boolean
+  streamMode: boolean
+  streamBufferSize: number
+  streamBufferTimeout: number
+  provider: string
+  voiceId?: string
+  ttsModel?: string
+  baseURL?: string
+  speed?: number
+  volume?: number
+  pitch?: number
+  apiKey?: string
+}
+
+export interface StreamReaderEvents {
+  onTextReady: (text: string) => void
+  onStreamStart: () => void
+  onStreamEnd: () => void
+  onStreamError: (error: TTSError) => void
+}
+
+export interface StreamingSynthesizerOptions {
+  provider: TTSProvider
+  voice?: string
+  speed?: number
+  volume?: number
+  pitch?: number
+}
+
+export interface StreamPlayerOptions {
+  format?: 'mp3' | 'wav' | 'flac'
+  onProgress?: (bytesReceived: number) => void
+}