概述

提供云开发 AI 接入能力，快速接入大模型和 Agent。

---

基础使用示例

初始化配置

Publishable Key 可前往 云开发平台/API Key 配置 中生成

类型声明

function ai(): AI;

返回值

返回新创建的 AI 实例。

import cloudbase from "@cloudbase/js-sdk";

// 初始化
const app = cloudbase.init({
  env: "your-env-id", // 替换为您的环境ID
  region: "ap-shanghai", // 地域，默认为上海
  accessKey: "", // 填入生成的 Publishable Key
});

// 如果填入了 accessKey，则不需要此步骤
await app.auth.signInAnonymously();

const ai = app.ai();

基础文本生成

// 基础文本生成示例
async function generateText() {
  const model = ai.createModel("cloudbase");
  const result = await model.generateText({
    model: "hy3-preview",
    messages: [{ role: "user", content: "你好，请你介绍一下李白" }],
  });

  console.log("生成的文本:", result.text);
  console.log("消耗的token:", result.usage);
}

流式文本生成

// 流式文本生成示例
async function streamText() {
  const model = ai.createModel("cloudbase");
  const result = await model.streamText({
    model: "hy3-preview",
    messages: [{ role: "user", content: "1+1等于多少" }],
  });

  for await (let chunk of result.textStream) {
    console.log("收到文本块:", chunk);
  }
}

Agent 对话

// Agent 对话示例
async function chatWithAgent() {
  const res = await ai.bot.sendMessage({
    botId: "botId-xxx",
    msg: "你好，请介绍一下你自己",
    history: [],
  });

  for await (let text of res.textStream) {
    console.log("Agent 回复:", text);
  }
}

---

AI

用于创建 AI 模型的类。

createModel

function createModel(model: string): ChatModel;

创建指定的 AI 模型。

• 创建一个新的 AI 模型实例
• 返回一个实现了 ChatModel 抽象类的模型实例
• 该实例提供 AI 生成文本相关能力

参数

model

string

模型标识符，如 'cloudbase'

返回

ChatModel

ChatModel

实现了 ChatModel 抽象类的模型实例，提供 AI 生成文本相关能力

示例

基础使用

const model = ai.createModel("cloudbase");

---

bot

挂载了 Bot 类的实例，上面集合了一系列与 Agent 交互的方法。具体可参考 Bot 类 的详细文档。

使用示例

const agentList = await ai.bot.list({ pageNumber: 1, pageSize: 10 });

---

registerFunctionTool

function registerFunctionTool(functionTool: FunctionTool): void;

注册函数工具。在进行大模型调用时，可以告知大模型可用的函数工具，当大模型的响应被解析为工具调用时，会自动调用对应的函数工具。

参数

functionTool

FunctionTool

要注册的函数工具定义

namestring

工具名称，用于大模型识别和调用

descriptionstring

工具功能描述，清晰的描述有助于大模型理解工具用途

fnCallableFunction

工具执行函数，当大模型调用工具时执行此函数

parametersobject

工具参数定义，使用 JSON Schema 格式

type

string

参数类型，通常为 'object'

properties

object

参数属性定义

required

Array<string>

必填参数列表

返回

void

undefined

无返回值

示例

基础工具注册

// 定义获取天气的工具
const getWeatherTool = {
  name: "get_weather",
  description: "返回某个城市的天气信息。调用示例：get_weather({city: '北京'})",
  fn: ({ city }) => `${city}的天气是：秋高气爽！！！`,
  parameters: {
    type: "object",
    properties: {
      city: {
        type: "string",
        description: "要查询的城市",
      },
    },
    required: ["city"],
  },
};

// 注册工具
ai.registerFunctionTool(getWeatherTool);

// 使用工具进行对话
const model = ai.createModel("cloudbase");
const result = await model.generateText({
  model: "hy3-preview",
  tools: [getWeatherTool],
  messages: [
    {
      role: "user",
      content: "请告诉我北京的天气状况",
    },
  ],
});

console.log(result.text);

多工具集成系统

// 多工具集成系统
class ToolIntegrationSystem {
  private registeredTools: Map<string, FunctionTool> = new Map();

  // 注册多个工具
  registerTools(tools: FunctionTool[]): void {
    tools.forEach((tool) => {
      ai.registerFunctionTool(tool);
      this.registeredTools.set(tool.name, tool);
      console.log(`✓ 工具注册成功: ${tool.name}`);
    });
  }

  // 获取所有可用工具
  getAvailableTools(): FunctionTool[] {
    return Array.from(this.registeredTools.values());
  }

  // 智能对话助手
  async smartAssistant(question: string): Promise<string> {
    const model = ai.createModel("cloudbase");
    const result = await model.generateText({
      model: "hy3-preview",
      tools: this.getAvailableTools(),
      messages: [
        {
          role: "user",
          content: question,
        },
      ],
    });

    return result.text;
  }

  // 工具使用统计
  getToolUsageStats(): Map<string, number> {
    const stats = new Map<string, number>();
    // 在实际应用中，这里可以记录工具调用次数
    return stats;
  }
}

// 定义多个工具
const tools = [
  {
    name: "get_weather",
    description: "获取城市天气信息",
    fn: ({ city }) => `${city}的天气：晴，25℃`,
    parameters: {
      type: "object",
      properties: { city: { type: "string", description: "城市名称" } },
      required: ["city"],
    },
  },
  {
    name: "calculate",
    description: "执行数学计算",
    fn: ({ expression }) => `计算结果：${eval(expression)}`,
    parameters: {
      type: "object",
      properties: { expression: { type: "string", description: "数学表达式" } },
      required: ["expression"],
    },
  },
  {
    name: "search_info",
    description: "搜索信息",
    fn: ({ keyword }) => `搜索结果：关于${keyword}的信息`,
    parameters: {
      type: "object",
      properties: { keyword: { type: "string", description: "搜索关键词" } },
      required: ["keyword"],
    },
  },
];

// 使用示例
const toolSystem = new ToolIntegrationSystem();
toolSystem.registerTools(tools);

// 智能对话
const answer = await toolSystem.smartAssistant(
  "计算一下 15*8+20 的结果，并告诉我北京的天气"
);
console.log("助手回答:", answer);

动态工具管理

// 动态工具管理系统
class DynamicToolManager {
  private toolRegistry: Map<string, FunctionTool> = new Map();

  // 动态注册工具
  registerTool(tool: FunctionTool): void {
    ai.registerFunctionTool(tool);
    this.toolRegistry.set(tool.name, tool);
    console.log(`工具注册: ${tool.name}`);
  }

  // 动态卸载工具
  unregisterTool(toolName: string): boolean {
    if (this.toolRegistry.has(toolName)) {
      this.toolRegistry.delete(toolName);
      console.log(`工具卸载: ${toolName}`);
      return true;
    }
    return false;
  }

  // 根据上下文动态选择工具
  async contextualAssistant(
    context: string,
    question: string
  ): Promise<string> {
    const relevantTools = this.selectRelevantTools(context, question);

    const model = ai.createModel("cloudbase");
    const result = await model.generateText({
      model: "hy3-preview",
      tools: relevantTools,
      messages: [
        {
          role: "system",
          content: `当前上下文：${context}`,
        },
        {
          role: "user",
          content: question,
        },
      ],
    });

    return result.text;
  }

  // 根据上下文选择相关工具
  private selectRelevantTools(
    context: string,
    question: string
  ): FunctionTool[] {
    const tools = Array.from(this.toolRegistry.values());

    // 简单的关键词匹配算法
    const keywords = [...context.split(" "), ...question.split(" ")];

    return tools.filter((tool) => {
      return keywords.some((keyword) =>
        tool.description.toLowerCase().includes(keyword.toLowerCase())
      );
    });
  }

  // 热更新工具
  updateTool(toolName: string, updatedTool: FunctionTool): boolean {
    if (this.toolRegistry.has(toolName)) {
      this.unregisterTool(toolName);
      this.registerTool(updatedTool);
      console.log(`工具更新: ${toolName}`);
      return true;
    }
    return false;
  }

  // 获取工具列表
  listTools(): string[] {
    return Array.from(this.toolRegistry.keys());
  }
}

// 使用示例
const toolManager = new DynamicToolManager();

// 注册初始工具
toolManager.registerTool({
  name: "weather",
  description: "获取天气信息",
  fn: ({ city }) => `${city}天气：晴`,
  parameters: {
    type: "object",
    properties: { city: { type: "string" } },
    required: ["city"],
  },
});

// 动态添加新工具
toolManager.registerTool({
  name: "translate",
  description: "翻译文本",
  fn: ({ text, to }) => `${text} 翻译为 ${to}：示例翻译结果`,
  parameters: {
    type: "object",
    properties: {
      text: { type: "string" },
      to: { type: "string" },
    },
    required: ["text", "to"],
  },
});

// 上下文感知对话
const answer = await toolManager.contextualAssistant(
  "天气和翻译相关的问题",
  "把'你好'翻译成英语，并告诉我上海的天气"
);
console.log("智能回答:", answer);

// 查看可用工具
console.log("可用工具:", toolManager.listTools());

---

ChatModel

这个抽象类描述了 AI 生文模型类提供的接口。

generateText

function generateText(data: BaseChatModelInput): Promise<{
  rawResponses: Array<unknown>;
  text: string;
  messages: Array<ChatModelMessage>;
  usage: Usage;
  error?: unknown;
}>;

调用大模型生成文本。

• 向大模型发送消息并获取生成的文本响应
• 支持完整的对话上下文管理
• 返回详细的调用信息和 token 消耗统计

参数

data

BaseChatModelInput

大模型输入参数，包含模型配置和消息内容

modelstring

模型名称，如 'cloudbase'

messagesArray<ChatModelMessage>

消息列表，包含对话上下文

temperature可选number

采样温度，控制输出的随机性，范围 0-1

topP可选number

温度采样，考虑概率质量为 top_p 的标记的结果

tools可选Array<FunctionTool>

大模型可用的工具列表

toolChoice可选string

指定大模型选择工具的方式

maxSteps可选number

请求大模型的最大次数

onStepFinish可选(prop: IOnStepFinish) => unknown

当对大模型的一次请求完成时，出发的回调函数。

返回

Promise

object

大模型生成文本的响应结果

textstring

大模型生成的文本内容

rawResponsesArray<unknown>

大模型的完整返回值，包含详细数据

messagesArray<ChatModelMessage>

本次调用的完整消息列表

usageUsage

本次调用消耗的 token 统计

error可选unknown

调用过程中产生的错误信息

示例

基础文本生成

const model = ai.createModel("cloudbase");
const result = await model.generateText({
  model: "hy3-preview",
  messages: [{ role: "user", content: "你好，请你介绍一下李白" }],
});

console.log("生成的文本:", result.text);
console.log("消耗的token:", result.usage);

多轮对话

const model = ai.createModel("cloudbase");

// 多轮对话示例
const messages = [
  { role: "user", content: "你好，我想了解中国古代文学" },
  {
    role: "assistant",
    content:
      "中国古代文学源远流长，从诗经楚辞到唐诗宋词，都有着丰富的文化内涵。",
  },
  { role: "user", content: "那你能介绍一下唐诗的特点吗？" },
];

const result = await model.generateText({
  model: "hy3-preview",
  messages: messages,
  temperature: 0.7,
  topP: 0.9,
});

console.log("唐诗介绍:", result.text);

带工具调用

// 注册天气查询工具
ai.registerFunctionTool({
  name: "get_weather",
  description: "查询城市天气信息",
  fn: ({ city }) => `${city}的天气是：晴朗，25℃`,
  parameters: {
    type: "object",
    properties: {
      city: {
        type: "string",
        description: "要查询的城市名称",
      },
    },
    required: ["city"],
  },
});

const model = ai.createModel("cloudbase");
const result = await model.generateText({
  model: "hy3-preview",
  messages: [{ role: "user", content: "请告诉我北京的天气状况" }],
  tools: [getWeatherTool],
  toolChoice: "auto",
});

console.log("大模型响应:", result.text);

---

streamText

function streamText(data: BaseChatModelInput): Promise<StreamTextResult>;

以流式调用大模型生成文本。

• 流式调用时，生成的文本及其他响应数据会通过 SSE 返回，该接口的返回值对 SSE 做了不同程度的封装，开发者能根据实际需求获取到文本流和完整数据流。
• 以流式方式调用大模型生成文本，支持实时获取增量内容

参数

data

BaseChatModelInput

大模型输入参数，包含模型配置和消息内容

modelstring

模型名称，如 'cloudbase'

messagesArray<ChatModelMessage>

消息列表，包含对话上下文

temperature可选number

采样温度，控制输出的随机性，范围 0-1

topP可选number

温度采样，考虑概率质量为 top_p 的标记的结果

tools可选Array<FunctionTool>

大模型可用的工具列表

toolChoice可选string

指定大模型选择工具的方式

maxSteps可选number

请求大模型的最大次数

onStepFinish可选(prop: IOnStepFinish) => unknown

当对大模型的一次请求完成时，出发的回调函数。

返回

Promise

StreamTextResult

流式文本生成的结果，包含文本流和数据流

textStreamReadableStream<string>

以流式返回的大模型生成文本，可获取到生成的增量文本

dataStreamReadableStream<DataChunk>

以流式返回的大模型响应数据，可获取到生成的增量数据

DataChunk

object

流式调用返回的数据块

choicesArray<object>

模型选择结果

finish_reason

string

模型终止推断的原因

delta

ChatModelMessage

本次请求的消息增量

usageUsage

本次请求消耗的 token

rawResponseunknown

大模型返回的原始回复

messagesPromise<ChatModelMessage[]>

本次调用的完整消息列表

usagePromise<Usage>

本次调用消耗的 token

error可选unknown

调用过程中产生的错误

示例

基础流式调用

const model = ai.createModel("cloudbase");
const result = await model.streamText({
  model: "hy3-preview",
  messages: [{ role: "user", content: "你好，请你介绍一下李白" }],
});

// 获取文本流
for await (let chunk of result.textStream) {
  console.log("收到文本块:", chunk);
}

// 1
// 加
// 1
// 的结果
// 是
// 2
// 。

// 获取数据流
for await (let data of result.dataStream) {
  console.log("收到数据块:", data);
}

// {created: 1723013866, id: "a95a54b5c5d2144eb700e60d0dfa5c98", model: "hy3-preview", version: "202404011000", choices: Array(1), …}
// {created: 1723013866, id: "a95a54b5c5d2144eb700e60d0dfa5c98", model: "hy3-preview", version: "202404011000", choices: Array(1), …}
// {created: 1723013866, id: "a95a54b5c5d2144eb700e60d0dfa5c98", model: "hy3-preview", version: "202404011000", choices: Array(1), …}
// {created: 1723013866, id: "a95a54b5c5d2144eb700e60d0dfa5c98", model: "hy3-preview", version: "202404011000", choices: Array(1), …}
// {created: 1723013866, id: "a95a54b5c5d2144eb700e60d0dfa5c98", model: "hy3-preview", version: "202404011000", choices: Array(1), …}
// {created: 1723013866, id: "a95a54b5c5d2144eb700e60d0dfa5c98", model: "hy3-preview", version: "202404011000", choices: Array(1), …}
// {created: 1723013866, id: "a95a54b5c5d2144eb700e60d0dfa5c98", model: "hy3-preview", version: "202404011000", choices: Array(1), …}
// {created: 1723013866, id: "a95a54b5c5d2144eb700e60d0dfa5c98", model: "hy3-preview", version: "202404011000", choices: Array(1), …}

实时显示应用

// 在网页中实时显示流式内容
async function displayStreamingText(question: string) {
  const model = ai.createModel("cloudbase");
  const result = await model.streamText({
    model: "hy3-preview",
    messages: [{ role: "user", content: question }],
  });

  const outputElement = document.getElementById("output");
  outputElement.innerHTML = "";

  for await (let chunk of result.textStream) {
    outputElement.innerHTML += chunk;
    // 滚动到最新内容
    outputElement.scrollTop = outputElement.scrollHeight;
  }
}

// 使用示例
displayStreamingText("请详细介绍一下唐诗的发展历程");

进度监控

// 监控流式生成进度
async function monitorStreamProgress(question: string) {
  const model = ai.createModel("cloudbase");
  const result = await model.streamText({
    model: "hy3-preview",
    messages: [{ role: "user", content: question }],
  });

  let totalChunks = 0;
  let startTime = Date.now();

  for await (let chunk of result.textStream) {
    totalChunks++;
    console.log(`第${totalChunks}个文本块: ${chunk}`);
  }

  const endTime = Date.now();
  console.log(
    `生成完成，共${totalChunks}个文本块，耗时${endTime - startTime}ms`
  );

  // 获取完整消息和token统计
  const messages = await result.messages;
  const usage = await result.usage;

  console.log("完整对话:", messages);
  console.log("Token消耗:", usage);
}

---

Bot

用于与 Agent 交互的类。

get

function get(props: { botId: string }): Promise<BotInfo>;

获取某个 Agent 的信息。

• 根据 Agent ID 获取详细的 Agent 信息
• 返回 Agent 的基本配置、欢迎语、头像等完整信息

参数

props

object

获取 Agent 信息的参数

botIdstring

要获取信息的 Agent 的 ID

返回

Promise

BotInfo

Agent 的详细信息

botIdstring

Agent ID

namestring

Agent 名称

introductionstring

Agent 简介

welcomeMessagestring

Agent 欢迎语

avatarstring

Agent 头像链接

backgroundstring

Agent 聊天背景图链接

isNeedRecommendboolean

Agent 回答后是否推荐问题

typestring

Agent 类型

示例

基础使用

const res = await ai.bot.get({ botId: "botId-xxx" });
console.log("Agent 信息:", res);

应用启动配置

// 应用启动时加载 Agent 配置
async function initializeAgent(botId: string) {
  try {
    const agentInfo = await ai.bot.get({ botId });

    // 设置应用界面
    document.title = `${agentInfo.name} - AI助手`;

    // 显示欢迎信息
    const welcomeElement = document.getElementById("welcome");
    welcomeElement.innerHTML = `
      <div class="agent-header">
        <img src="${agentInfo.avatar}" alt="${agentInfo.name}" class="avatar" />
        <h2>${agentInfo.name}</h2>
        <p>${agentInfo.introduction}</p>
      </div>
    `;

    // 设置聊天背景
    if (agentInfo.background) {
      document.body.style.backgroundImage = `url(${agentInfo.background})`;
    }

    return agentInfo;
  } catch (error) {
    console.error("加载 Agent 配置失败:", error);
    throw error;
  }
}

// 使用示例
initializeAgent("bot-27973647");

多 Agent 管理

// 管理多个 Agent 的配置
class AgentManager {
  private agents: Map<string, BotInfo> = new Map();

  async loadAgent(botId: string) {
    if (this.agents.has(botId)) {
      return this.agents.get(botId);
    }

    const agentInfo = await ai.bot.get({ botId });
    this.agents.set(botId, agentInfo);
    return agentInfo;
  }

  getAgentList() {
    return Array.from(this.agents.values());
  }

  async switchAgent(botId: string) {
    const agentInfo = await this.loadAgent(botId);

    // 更新界面显示
    this.updateUI(agentInfo);

    return agentInfo;
  }

  private updateUI(agentInfo: BotInfo) {
    // 更新界面元素
    console.log("切换到 Agent:", agentInfo.name);
  }
}

// 使用示例
const agentManager = new AgentManager();
await agentManager.loadAgent("bot-27973647");
await agentManager.loadAgent("bot-27973648");

---

list

function list(props: {
  name: string;
  introduction: string;
  information: string;
  enable: boolean;
  pageSize: number;
  pageNumber: number;
}): Promise<AgentListResult>;

批量获取多个 Agent 的信息。

• 查询和筛选可用的 Agent 列表
• 支持分页查询和条件筛选
• 返回 Agent 的基本信息和配置详情
• 适用于构建 Agent 选择器、Agent 管理界面等应用

参数

props

object

查询 Agent 列表的参数

pageNumbernumber

分页下标，从1开始

pageSizenumber

分页大小，每页显示的 Agent 数量

enableboolean

Agent 是否启用，true（启用）或 false（禁用）

namestring

Agent 名字，用于模糊查询

informationstring

Agent 信息，用于模糊查询

introductionstring

Agent 简介，用于模糊查询

返回

Promise

AgentListResult

Agent 列表查询结果

totalnumber

Agent 总数

botListArray<object>

Agent 列表

botId

string

Agent ID

name

string

Agent 名称

introduction

string

Agent 简介

welcomeMessage

string

Agent 欢迎语

avatar

string

Agent 头像链接

background

string

Agent 聊天背景图链接

isNeedRecommend

boolean

Agent 回答后是否推荐问题

type

string

Agent 类型

示例

基础查询

const agentList = await ai.bot.list({
  pageNumber: 1,
  pageSize: 10,
  name: "",
  enable: true,
  information: "",
  introduction: "",
});

console.log("Agent 总数:", agentList.total);
console.log("Agent 列表:", agentList.botList);

条件筛选

// 根据条件筛选 Agent
async function searchAgents(keyword: string, page: number = 1) {
  const result = await ai.bot.list({
    pageNumber: page,
    pageSize: 20,
    name: keyword,
    enable: true,
    information: keyword,
    introduction: keyword,
  });

  console.log(`找到 ${result.total} 个匹配的 Agent`);

  result.botList.forEach((agent) => {
    console.log(`- ${agent.name}: ${agent.introduction}`);
  });

  return result;
}

// 使用示例
await searchAgents("翻译"); // 搜索翻译相关的 Agent
await searchAgents("客服", 2); // 搜索客服相关的 Agent，第二页

Agent 选择器

// 构建 Agent 选择器界面
class AgentSelector {
  private currentPage: number = 1;
  private pageSize: number = 12;

  async loadAgents(keyword: string = "") {
    const result = await ai.bot.list({
      pageNumber: this.currentPage,
      pageSize: this.pageSize,
      name: keyword,
      enable: true,
      information: keyword,
      introduction: keyword,
    });

    this.displayAgents(result.botList);
    this.updatePagination(result.total);

    return result;
  }

  private displayAgents(agents: any[]) {
    const container = document.getElementById("agent-list");
    container.innerHTML = "";

    agents.forEach((agent) => {
      const agentCard = document.createElement("div");
      agentCard.className = "agent-card";
      agentCard.innerHTML = `
        <img src="${agent.avatar}" alt="${agent.name}" class="agent-avatar" />
        <div class="agent-info">
          <h3>${agent.name}</h3>
          <p>${agent.introduction}</p>
          <button onclick="selectAgent('${agent.botId}')">选择</button>
        </div>
      `;
      container.appendChild(agentCard);
    });
  }

  private updatePagination(total: number) {
    const pagination = document.getElementById("pagination");
    const totalPages = Math.ceil(total / this.pageSize);

    pagination.innerHTML = `
      <button onclick="prevPage()" ${
        this.currentPage <= 1 ? "disabled" : ""
      }>上一页</button>
      <span>第 ${this.currentPage} 页，共 ${totalPages} 页</span>
      <button onclick="nextPage()" ${
        this.currentPage >= totalPages ? "disabled" : ""
      }>下一页</button>
    `;
  }

  async nextPage() {
    this.currentPage++;
    await this.loadAgents();
  }

  async prevPage() {
    if (this.currentPage > 1) {
      this.currentPage--;
      await this.loadAgents();
    }
  }

  selectAgent(botId: string) {
    console.log("选择的 Agent ID:", botId);
    // 开始与选中的 Agent 对话
    this.startChat(botId);
  }

  private startChat(botId: string) {
    // 实现与 Agent 开始对话的逻辑
    console.log("开始与 Agent 对话:", botId);
  }
}

// 使用示例
const selector = new AgentSelector();
await selector.loadAgents(); // 加载第一页 Agent

// 搜索特定类型的 Agent
await selector.loadAgents("写作助手");

---

sendMessage

function sendMessage(props: {
  botId: string;
  msg: string;
  history: Array<{
    role: string;
    content: string;
  }>;
}): Promise<StreamResult>;

与 Agent 进行对话。

• 响应会通过 SSE 返回，该接口的返回值对 SSE 做了不同程度的封装，开发者能根据实际需求获取到文本流和完整数据流。
• 支持多轮对话上下文管理
• 返回流式响应，支持实时获取 Agent 回复
• 适用于构建聊天机器人、智能助手等应用

参数

props

object

发送消息的参数

botIdstring

Agent ID

msgstring

此次对话要发送的消息内容

historyArray<object>

在此次对话前发生的聊天记录

object

Object

聊天记录

rolestring

本聊天信息的发送角色（user/assistant）

contentstring

本聊天信息的内容

返回

Promise

StreamResult

流式对话结果，包含文本流和数据流

textStreamAsyncIterable<string>

以流式返回的 Agent 生成文本，可获取到生成的增量文本

dataStreamAsyncIterable<AgentStreamChunk>

以流式返回的 Agent 生成数据，可获取到生成的增量数据

AgentStreamChunk

object

Agent 流式对话返回的数据块

creatednumber

对话时间戳

record_idstring

对话记录ID

modelstring

大模型类型

versionstring

大模型版本

typestring

回复类型：text（主要回答内容）、thinking（思考过程）、search（查询结果）、knowledge（知识库）

rolestring

对话角色，响应中固定为 assistant

contentstring

对话内容

finish_reasonstring

对话结束标志，continue（对话未结束）、stop（对话结束）

reasoning_contentstring

深度思考内容（仅deepseek-r1不为空字符串）

usageobject

token使用量

prompt_tokens

number

表示prompt的tokens数，多次返回中保持不变

completion_tokens

number

回答的token总数，在流式返回中表示到目前为止所有completion的tokens总数，多次返回中持续累加

total_tokens

number

表示prompt_tokens和completion_tokens之和

knowledge_basestring[]

对话中使用到的知识库

search_infoobject

搜索结果信息，需要开启联网查询

search_results

object[]

搜索引文信息

indexstring

搜索引文序号

titlestring

搜索引文标题

urlstring

搜索引文链接

示例

基础对话

const res = await ai.bot.sendMessage({
  botId: "botId-xxx",
  history: [{ content: "你是李白。", role: "user" }],
  msg: "你好",
});

// 获取文本流
for await (let text of res.textStream) {
  console.log("Agent 回复:", text);
}

// 获取数据流
for await (let data of res.dataStream) {
  console.log("详细数据:", data);
}

多轮对话

// 多轮对话示例
class ChatSession {
  private history: Array<{ role: string; content: string }> = [];

  async sendMessage(botId: string, message: string) {
    const res = await ai.bot.sendMessage({
      botId: botId,
      history: this.history,
      msg: message,
    });

    let fullResponse = "";

    // 实时显示回复
    for await (let chunk of res.textStream) {
      fullResponse += chunk;
      this.updateDisplay(fullResponse);
    }

    // 记录对话历史
    this.history.push({ role: "user", content: message });
    this.history.push({ role: "assistant", content: fullResponse });

    return fullResponse;
  }

  private updateDisplay(text: string) {
    const chatArea = document.getElementById("chat-area");
    chatArea.innerHTML = text;
    chatArea.scrollTop = chatArea.scrollHeight;
  }

  clearHistory() {
    this.history = [];
  }

  getHistory() {
    return this.history;
  }
}

// 使用示例
const chatSession = new ChatSession();
await chatSession.sendMessage("botId-xxx", "你好，请介绍一下你自己");
await chatSession.sendMessage("botId-xxx", "你能帮我写一首诗吗？");

实时聊天应用

// 构建实时聊天应用
class ChatApp {
  private botId: string;
  private conversationId: string;

  constructor(botId: string) {
    this.botId = botId;
    this.setupEventListeners();
  }

  private setupEventListeners() {
    const sendBtn = document.getElementById("send-btn");
    const input = document.getElementById("message-input") as HTMLInputElement;

    sendBtn.addEventListener("click", () => {
      this.sendMessage(input.value);
      input.value = "";
    });

    input.addEventListener("keypress", (e) => {
      if (e.key === "Enter") {
        this.sendMessage(input.value);
        input.value = "";
      }
    });
  }

  async sendMessage(message: string) {
    if (!message.trim()) return;

    // 显示用户消息
    this.displayMessage("user", message);

    try {
      const res = await ai.bot.sendMessage({
        botId: this.botId,
        history: this.getChatHistory(),
        msg: message,
      });

      // 显示流式回复
      const assistantDiv = this.displayMessage("assistant", "");

      for await (let chunk of res.textStream) {
        assistantDiv.innerHTML += chunk;
        assistantDiv.scrollIntoView({ behavior: "smooth" });
      }
    } catch (error) {
      this.displayMessage("error", "发送消息失败: " + error.message);
    }
  }

  private displayMessage(role: string, content: string): HTMLDivElement {
    const chatContainer = document.getElementById("chat-container");
    const messageDiv = document.createElement("div");
    messageDiv.className = `message ${role}`;
    messageDiv.innerHTML = content;

    chatContainer.appendChild(messageDiv);
    chatContainer.scrollTop = chatContainer.scrollHeight;

    return messageDiv;
  }

  private getChatHistory() {
    const messages = document.querySelectorAll(".message");
    const history = [];

    for (let message of messages) {
      const role = message.classList.contains("user") ? "user" : "assistant";
      const content = message.innerHTML;
      history.push({ role, content });
    }

    return history;
  }
}

// 使用示例
const chatApp = new ChatApp("botId-xxx");

---

getChatRecords

function getChatRecords(props: {
  botId: string;
  sort: string;
  pageSize: number;
  pageNumber: number;
}): Promise<ChatRecordsResult>;

获取聊天记录。

• 获取指定 Agent 的历史聊天记录
• 支持分页查询和排序
• 返回完整的对话历史信息
• 适用于构建聊天历史查看功能

参数

props

object

获取聊天记录的参数

botIdstring

Agent ID

sortstring

排序方式，asc（升序）或 desc（降序）

pageSizenumber

分页大小，每页显示的记录数

pageNumbernumber

分页下标，从1开始

返回

Promise

ChatRecordsResult

聊天记录查询结果

totalnumber

对话总数

recordListArray<object>

对话记录列表

record

Record

对话记录

botIdstring

Agent ID

recordIdstring

对话ID，由系统生成

rolestring

对话中的角色（user/assistant）

contentstring

对话内容

conversationstring

用户标识

typestring

对话数据类型

imagestring

对话生成的图片链接

triggerSrcstring

对话发起来源

replyTostring

对话回复的记录ID

createTimestring

对话时间

示例

基础使用

const records = await ai.bot.getChatRecords({
  botId: "botId-xxx",
  pageNumber: 1,
  pageSize: 10,
  sort: "asc",
});

console.log("总记录数:", records.total);
console.log("记录列表:", records.recordList);

分页查询

// 分页查询聊天记录
class ChatHistoryManager {
  private botId: string;
  private currentPage: number = 1;
  private pageSize: number = 20;

  constructor(botId: string) {
    this.botId = botId;
  }

  async loadPage(pageNumber: number) {
    const records = await ai.bot.getChatRecords({
      botId: this.botId,
      pageNumber: pageNumber,
      pageSize: this.pageSize,
      sort: "desc", // 按时间倒序排列
    });

    this.currentPage = pageNumber;
    return records;
  }

  async nextPage() {
    return this.loadPage(this.currentPage + 1);
  }

  async previousPage() {
    if (this.currentPage > 1) {
      return this.loadPage(this.currentPage - 1);
    }
    return null;
  }

  getTotalPages(totalRecords: number) {
    return Math.ceil(totalRecords / this.pageSize);
  }
}

// 使用示例
const historyManager = new ChatHistoryManager("botId-xxx");
const firstPage = await historyManager.loadPage(1);
console.log("第1页记录:", firstPage.recordList);

const secondPage = await historyManager.nextPage();
console.log("第2页记录:", secondPage.recordList);

聊天历史界面

// 构建聊天历史查看界面
class ChatHistoryViewer {
  private botId: string;

  constructor(botId: string) {
    this.botId = botId;
    this.setupUI();
  }

  private setupUI() {
    const historyContainer = document.getElementById("history-container");
    const prevBtn = document.getElementById("prev-btn");
    const nextBtn = document.getElementById("next-btn");
    const pageInfo = document.getElementById("page-info");

    prevBtn.addEventListener("click", () => this.previousPage());
    nextBtn.addEventListener("click", () => this.nextPage());
  }

  async loadHistory(pageNumber: number = 1) {
    const records = await ai.bot.getChatRecords({
      botId: this.botId,
      pageNumber: pageNumber,
      pageSize: 10,
      sort: "desc",
    });

    this.displayRecords(records.recordList);
    this.updatePageInfo(pageNumber, records.total);
  }

  private displayRecords(records: any[]) {
    const container = document.getElementById("records-container");
    container.innerHTML = "";

    records.forEach((record) => {
      const recordDiv = document.createElement("div");
      recordDiv.className = "chat-record";
      recordDiv.innerHTML = `
        <div class="record-header">
          <span class="role">${record.role}</span>
          <span class="time">${record.createTime}</span>
        </div>
        <div class="content">${record.content}</div>
      `;
      container.appendChild(recordDiv);
    });
  }

  private updatePageInfo(currentPage: number, total: number) {
    const pageInfo = document.getElementById("page-info");
    const totalPages = Math.ceil(total / 10);
    pageInfo.textContent = `第${currentPage}页，共${totalPages}页，总计${total}条记录`;
  }

  async nextPage() {
    const currentPage = parseInt(
      document.getElementById("page-info").textContent.match(/第(\d+)页/)[1]
    );
    await this.loadHistory(currentPage + 1);
  }

  async previousPage() {
    const currentPage = parseInt(
      document.getElementById("page-info").textContent.match(/第(\d+)页/)[1]
    );
    if (currentPage > 1) {
      await this.loadHistory(currentPage - 1);
    }
  }
}

// 使用示例
const historyViewer = new ChatHistoryViewer("botId-xxx");
await historyViewer.loadHistory(1);

---

sendFeedback

function sendFeedback(props: {
  userFeedback: IUserFeedback;
  botId: string;
}): Promise<void>;

发送对某条聊天记录的反馈信息。

• 对指定的聊天记录进行评价和反馈
• 支持评分、评论、标签等多种反馈方式
• 帮助改进 Agent 的回答质量
• 适用于构建用户反馈系统

参数

props

object

发送反馈的参数

userFeedbackIUserFeedback

用户反馈信息，详见 IUserFeedback 类型定义

botId

string

Agent ID

recordId

string

聊天记录ID

comment

string

用户评论

rating

number

用户评分（1-5分）

tags

string[]

用户反馈的标签数组

aiAnswer

string

Agent 的回答内容

input

string

用户输入的问题

type

string

反馈类型，upvote（点赞）或 downvote（点踩）

botIdstring

将要反馈的 Agent ID

返回

Promise

void

无返回值

示例

基础反馈

const res = await ai.bot.sendFeedback({
  userFeedback: {
    botId: "botId-xxx",
    recordId: "recordId-xxx",
    comment: "非常棒",
    rating: 5,
    tags: ["优美"],
    aiAnswer: "落英缤纷",
    input: "来个成语",
    type: "upvote",
  },
  botId: "botId-xxx",
});

console.log("反馈发送成功");

用户反馈系统

// 构建用户反馈系统
class FeedbackSystem {
  private botId: string;

  constructor(botId: string) {
    this.botId = botId;
    this.setupFeedbackUI();
  }

  private setupFeedbackUI() {
    const feedbackButtons = document.querySelectorAll(".feedback-btn");

    feedbackButtons.forEach((button) => {
      button.addEventListener("click", (e) => {
        const recordId = e.target.dataset.recordId;
        const type = e.target.dataset.type; // upvote 或 downvote
        this.showFeedbackDialog(recordId, type);
      });
    });
  }

  private showFeedbackDialog(recordId: string, type: string) {
    const dialog = document.getElementById("feedback-dialog");
    const submitBtn = document.getElementById("submit-feedback");
    const ratingInput = document.getElementById("rating") as HTMLInputElement;
    const commentInput = document.getElementById(
      "comment"
    ) as HTMLTextAreaElement;
    const tagsInput = document.getElementById("tags") as HTMLInputElement;

    dialog.style.display = "block";

    submitBtn.onclick = async () => {
      await this.submitFeedback(recordId, type, {
        rating: parseInt(ratingInput.value),
        comment: commentInput.value,
        tags: tagsInput.value.split(",").map((tag) => tag.trim()),
      });
      dialog.style.display = "none";
    };
  }

  async submitFeedback(
    recordId: string,
    type: string,
    feedback: {
      rating: number;
      comment: string;
      tags: string[];
    }
  ) {
    try {
      await ai.bot.sendFeedback({
        userFeedback: {
          botId: this.botId,
          recordId: recordId,
          comment: feedback.comment,
          rating: feedback.rating,
          tags: feedback.tags,
          aiAnswer: this.getCurrentAnswer(),
          input: this.getCurrentQuestion(),
          type: type,
        },
        botId: this.botId,
      });

      this.showSuccessMessage("感谢您的反馈！");
    } catch (error) {
      this.showErrorMessage("反馈发送失败: " + error.message);
    }
  }

  private getCurrentAnswer(): string {
    // 获取当前对话的回答内容
    const answerElement = document.querySelector(
      ".assistant-message:last-child"
    );
    return answerElement?.textContent || "";
  }

  private getCurrentQuestion(): string {
    // 获取当前对话的问题内容
    const questionElement = document.querySelector(".user-message:last-child");
    return questionElement?.textContent || "";
  }

  private showSuccessMessage(message: string) {
    const notification = document.createElement("div");
    notification.className = "notification success";
    notification.textContent = message;
    document.body.appendChild(notification);
    setTimeout(() => notification.remove(), 3000);
  }

  private showErrorMessage(message: string) {
    const notification = document.createElement("div");
    notification.className = "notification error";
    notification.textContent = message;
    document.body.appendChild(notification);
    setTimeout(() => notification.remove(), 5000);
  }
}

// 使用示例
const feedbackSystem = new FeedbackSystem("botId-xxx");

批量反馈处理

// 批量处理用户反馈
class BatchFeedbackProcessor {
  private botId: string;
  private feedbackQueue: Array<{
    recordId: string;
    type: string;
    rating: number;
    comment: string;
    tags: string[];
  }> = [];

  constructor(botId: string) {
    this.botId = botId;
    this.startProcessing();
  }

  addFeedback(
    recordId: string,
    type: string,
    rating: number,
    comment: string,
    tags: string[]
  ) {
    this.feedbackQueue.push({ recordId, type, rating, comment, tags });
  }

  private async startProcessing() {
    setInterval(async () => {
      if (this.feedbackQueue.length > 0) {
        await this.processBatch();
      }
    }, 5000); // 每5秒处理一次
  }

  private async processBatch() {
    const batch = this.feedbackQueue.splice(0, 10); // 每次处理最多10个反馈

    const promises = batch.map((feedback) => {
      return ai.bot.sendFeedback({
        userFeedback: {
          botId: this.botId,
          recordId: feedback.recordId,
          comment: feedback.comment,
          rating: feedback.rating,
          tags: feedback.tags,
          aiAnswer: "", // 需要根据实际情况获取
          input: "", // 需要根据实际情况获取
          type: feedback.type,
        },
        botId: this.botId,
      });
    });

    try {
      await Promise.all(promises);
      console.log(`成功处理 ${batch.length} 个反馈`);
    } catch (error) {
      console.error("批量处理反馈失败:", error);
      // 将失败的反馈重新加入队列
      this.feedbackQueue.unshift(...batch);
    }
  }

  getQueueSize() {
    return this.feedbackQueue.length;
  }
}

// 使用示例
const feedbackProcessor = new BatchFeedbackProcessor("botId-xxx");

// 添加多个反馈
feedbackProcessor.addFeedback("record-1", "upvote", 5, "回答很棒", [
  "准确",
  "详细",
]);
feedbackProcessor.addFeedback("record-2", "downvote", 2, "回答不够准确", [
  "不准确",
]);

console.log("当前队列大小:", feedbackProcessor.getQueueSize());

---

getFeedback

function getFeedback(props: {
  botId: string;
  type: string;
  sender: string;
  senderFilter: string;
  minRating: number;
  maxRating: number;
  from: number;
  to: number;
  pageSize: number;
  pageNumber: number;
}): Promise<FeedbackResult>;

获取已存在的反馈信息。

• 查询指定 Agent 的用户反馈记录
• 支持多种过滤条件：时间范围、评分范围、用户过滤等
• 返回分页的反馈结果和统计信息
• 适用于构建反馈分析和管理系统

参数

props

object

查询反馈信息的参数

botIdstring

Agent ID

typestring

用户反馈类型，upvote（点赞）或 downvote（点踩）

senderstring

评论创建用户

senderFilterstring

评论创建用户过滤关系，include（包含）、exclude（不包含）、equal（等于）、unequal（不等于）、prefix（前缀）

minRatingnumber

最低评分

maxRatingnumber

最高评分

fromnumber

开始时间戳

tonumber

结束时间戳

pageSizenumber

分页大小

pageNumbernumber

分页下标

返回

Promise

FeedbackResult

反馈查询结果

feedbackListArray<object>

反馈查询结果列表

feedback

可选

Feedback

反馈查询结果

recordIdstring

对话记录ID

typestring

用户反馈类型，upvote（点赞）或 downvote（点踩）

botIdstring

Agent ID

commentstring

用户评论

ratingnumber

用户评分

tagsstring[]

用户反馈的标签数组

inputstring

用户输入的问题

aiAnswerstring

Agent 的回答

totalnumber

反馈总数

示例

基础查询

const res = await ai.bot.getFeedback({
  botId: "botId-xxx",
  from: 0,
  to: 0,
  maxRating: 4,
  minRating: 3,
  pageNumber: 1,
  pageSize: 10,
  sender: "user-a",
  senderFilter: "include",
  type: "upvote",
});

console.log("反馈总数:", res.total);
console.log("反馈列表:", res.feedbackList);

反馈分析系统

// 构建反馈分析系统
class FeedbackAnalyzer {
  private botId: string;

  constructor(botId: string) {
    this.botId = botId;
  }

  async analyzeFeedback(timeRange: { from: number; to: number }) {
    const feedback = await ai.bot.getFeedback({
      botId: this.botId,
      from: timeRange.from,
      to: timeRange.to,
      minRating: 1,
      maxRating: 5,
      pageNumber: 1,
      pageSize: 1000, // 获取足够多的数据进行分析
      sender: "",
      senderFilter: "include",
      type: "",
    });

    return this.generateReport(feedback.feedbackList);
  }

  private generateReport(feedbackList: any[]) {
    const report = {
      total: feedbackList.length,
      upvotes: feedbackList.filter((f) => f.type === "upvote").length,
      downvotes: feedbackList.filter((f) => f.type === "downvote").length,
      averageRating: this.calculateAverageRating(feedbackList),
      commonTags: this.getCommonTags(feedbackList),
      ratingDistribution: this.getRatingDistribution(feedbackList),
      topComments: this.getTopComments(feedbackList),
    };

    return report;
  }

  private calculateAverageRating(feedbackList: any[]): number {
    const ratings = feedbackList.map((f) => f.rating).filter((r) => r > 0);
    return ratings.length > 0
      ? ratings.reduce((a, b) => a + b, 0) / ratings.length
      : 0;
  }

  private getCommonTags(feedbackList: any[]): string[] {
    const tagCounts = new Map<string, number>();

    feedbackList.forEach((feedback) => {
      feedback.tags?.forEach((tag) => {
        tagCounts.set(tag, (tagCounts.get(tag) || 0) + 1);
      });
    });

    return Array.from(tagCounts.entries())
      .sort((a, b) => b[1] - a[1])
      .slice(0, 10)
      .map((entry) => entry[0]);
  }

  private getRatingDistribution(feedbackList: any[]): Record<number, number> {
    const distribution = { 1: 0, 2: 0, 3: 0, 4: 0, 5: 0 };

    feedbackList.forEach((feedback) => {
      if (feedback.rating >= 1 && feedback.rating <= 5) {
        distribution[feedback.rating]++;
      }
    });

    return distribution;
  }

  private getTopComments(feedbackList: any[]): any[] {
    return feedbackList
      .filter((f) => f.comment && f.comment.length > 0)
      .sort((a, b) => b.rating - a.rating)
      .slice(0, 5);
  }
}

// 使用示例
const analyzer = new FeedbackAnalyzer("botId-xxx");
const report = await analyzer.analyzeFeedback({
  from: Date.now() - 7 * 24 * 60 * 60 * 1000, // 最近7天
  to: Date.now(),
});

console.log("反馈分析报告:", report);

反馈管理界面

// 构建反馈管理界面
class FeedbackManager {
  private botId: string;

  constructor(botId: string) {
    this.botId = botId;
    this.setupUI();
  }

  private setupUI() {
    const filterForm = document.getElementById("feedback-filter-form");
    const resultsContainer = document.getElementById("feedback-results");
    const exportBtn = document.getElementById("export-feedback");

    filterForm.addEventListener("submit", (e) => {
      e.preventDefault();
      this.loadFeedback(this.getFilterValues());
    });

    exportBtn.addEventListener("click", () => {
      this.exportFeedback();
    });
  }

  private getFilterValues() {
    const formData = new FormData(
      document.getElementById("feedback-filter-form")
    );
    return {
      type: formData.get("type") as string,
      minRating: parseInt(formData.get("minRating") as string),
      maxRating: parseInt(formData.get("maxRating") as string),
      from: this.parseDate(formData.get("fromDate") as string),
      to: this.parseDate(formData.get("toDate") as string),
      pageNumber: 1,
      pageSize: 50,
    };
  }

  private parseDate(dateString: string): number {
    return dateString ? new Date(dateString).getTime() : 0;
  }

  async loadFeedback(filters: any) {
    try {
      const res = await ai.bot.getFeedback({
        botId: this.botId,
        ...filters,
        sender: "",
        senderFilter: "include",
      });

      this.displayResults(res.feedbackList, res.total);
    } catch (error) {
      this.showError("加载反馈失败: " + error.message);
    }
  }

  private displayResults(feedbackList: any[], total: number) {
    const container = document.getElementById("feedback-results");
    container.innerHTML = "";

    const summary = document.createElement("div");
    summary.className = "feedback-summary";
    summary.innerHTML = `共找到 ${total} 条反馈记录`;
    container.appendChild(summary);

    feedbackList.forEach((feedback) => {
      const feedbackCard = this.createFeedbackCard(feedback);
      container.appendChild(feedbackCard);
    });
  }

  private createFeedbackCard(feedback: any): HTMLDivElement {
    const card = document.createElement("div");
    card.className = `feedback-card ${feedback.type}`;

    card.innerHTML = `
      <div class="feedback-header">
        <span class="type ${feedback.type}">${
      feedback.type === "upvote" ? "👍" : "👎"
    }</span>
        <span class="rating">评分: ${feedback.rating}/5</span>
        <span class="time">${new Date(
          feedback.createTime
        ).toLocaleString()}</span>
      </div>
      <div class="feedback-content">
        <div class="question"><strong>问题:</strong> ${feedback.input}</div>
        <div class="answer"><strong>回答:</strong> ${feedback.aiAnswer}</div>
        <div class="comment"><strong>评论:</strong> ${feedback.comment}</div>
        <div class="tags"><strong>标签:</strong> ${
          feedback.tags?.join(", ") || "无"
        }</div>
      </div>
    `;

    return card;
  }

  private async exportFeedback() {
    const res = await ai.bot.getFeedback({
      botId: this.botId,
      from: 0,
      to: Date.now(),
      minRating: 1,
      maxRating: 5,
      pageNumber: 1,
      pageSize: 10000,
      sender: "",
      senderFilter: "include",
      type: "",
    });

    const csv = this.convertToCSV(res.feedbackList);
    this.downloadCSV(csv, `feedback_${this.botId}_${Date.now()}.csv`);
  }

  private convertToCSV(feedbackList: any[]): string {
    const headers = ["类型", "评分", "问题", "回答", "评论", "标签", "时间"];
    const rows = feedbackList.map((f) => [
      f.type,
      f.rating,
      f.input,
      f.aiAnswer,
      f.comment,
      f.tags?.join(";"),
      new Date(f.createTime).toLocaleString(),
    ]);

    return [headers, ...rows]
      .map((row) => row.map((cell) => `"${cell}"`).join(","))
      .join("\n");
  }

  private downloadCSV(csv: string, filename: string) {
    const blob = new Blob([csv], { type: "text/csv" });
    const url = URL.createObjectURL(blob);
    const a = document.createElement("a");
    a.href = url;
    a.download = filename;
    a.click();
    URL.revokeObjectURL(url);
  }

  private showError(message: string) {
    const errorDiv = document.createElement("div");
    errorDiv.className = "error-message";
    errorDiv.textContent = message;
    document.getElementById("feedback-results").appendChild(errorDiv);
  }
}

// 使用示例
const feedbackManager = new FeedbackManager("botId-xxx");

---

uploadFiles

function uploadFiles(props: {
  botId: string;
  fileList: Array<{
    fileId: string;
    fileName: string;
    type: "file";
  }>;
}): Promise<void>;

将云存储中的文件上传至 Agent，用于进行文档聊天。

• 支持将云存储中的文件上传到指定的 Agent
• 上传后的文件可用于文档聊天功能
• 支持批量上传多个文件
• 适用于构建文档问答、文件分析等应用场景

参数

props

object

文件上传参数

botIdstring

Agent ID，指定要上传文件的 Agent

fileListArray<object>

要上传的文件列表

fileId

string

云存储文件 ID，格式如 cloud://xxx.docx

fileName

string

文件名，用于显示和识别

type

string

文件类型，目前只支持 'file'

返回

Promise

void

无返回值

示例

基础上传

// 上传单个文件到指定 Agent
await ai.bot.uploadFiles({
  botId: "botId-xxx",
  fileList: [
    {
      fileId: "cloud://document.docx",
      fileName: "document.docx",
      type: "file",
    },
  ],
});

console.log("文件上传成功");

批量上传与文档聊天

// 批量上传多个文件
async function uploadAndChat(
  botId: string,
  files: Array<{ fileId: string; fileName: string }>
) {
  // 上传文件
  await ai.bot.uploadFiles({
    botId: botId,
    fileList: files.map((file) => ({
      ...file,
      type: "file" as const,
    })),
  });

  console.log("文件上传完成，开始文档聊天");

  // 基于上传的文件进行聊天
  const res = await ai.bot.sendMessage({
    botId: botId,
    msg: "请分析这些文档的主要内容",
    files: files.map((f) => f.fileId),
  });

  // 流式输出回答
  for await (let text of res.textStream) {
    console.log(text);
  }
}

// 使用示例
await uploadAndChat("your-bot-id", [
  { fileId: "cloud://report.pdf", fileName: "季度报告.pdf" },
  { fileId: "cloud://data.xlsx", fileName: "数据表格.xlsx" },
]);

文档问答系统

// 构建文档问答系统
class DocumentQASystem {
  private botId: string;
  private uploadedFiles: string[] = [];

  constructor(botId: string) {
    this.botId = botId;
  }

  async uploadDocuments(files: Array<{ fileId: string; fileName: string }>) {
    await ai.bot.uploadFiles({
      botId: this.botId,
      fileList: files.map((file) => ({
        ...file,
        type: "file" as const,
      })),
    });

    this.uploadedFiles.push(...files.map((f) => f.fileId));
    console.log(`已上传 ${files.length} 个文件`);
  }

  async askQuestion(question: string): Promise<string> {
    const res = await ai.bot.sendMessage({
      botId: this.botId,
      msg: question,
      files: this.uploadedFiles,
    });

    let fullAnswer = "";
    for await (let text of res.textStream) {
      fullAnswer += text;
    }

    return fullAnswer;
  }

  async interactiveChat() {
    console.log("文档问答系统已启动，输入 'exit' 退出");

    while (true) {
      const question = await this.getUserInput("请输入问题: ");

      if (question.toLowerCase() === "exit") {
        console.log("再见！");
        break;
      }

      console.log("思考中...");
      const answer = await this.askQuestion(question);
      console.log("回答:", answer);
      console.log("---");
    }
  }

  private getUserInput(prompt: string): Promise<string> {
    return new Promise((resolve) => {
      // 在实际应用中，这里可以是浏览器输入框或命令行输入
      const input = prompt;
      resolve(input);
    });
  }
}

// 使用示例
const qaSystem = new DocumentQASystem("doc-bot-id");

// 上传文档
await qaSystem.uploadDocuments([
  { fileId: "cloud://manual.pdf", fileName: "产品手册.pdf" },
  { fileId: "cloud://faq.docx", fileName: "常见问题.docx" },
]);

// 开始交互式问答
await qaSystem.interactiveChat();

---

getRecommendQuestions

function getRecommendQuestions(props: {
  botId: string;
  name: string;
  introduction: string;
  agentSetting: string;
  msg: string;
  history: Array<{
    role: string;
    content: string;
  }>;
}): Promise<StreamResult>;

获取推荐的问题，基于对话上下文智能生成相关问题建议。

• 根据当前对话内容和历史记录智能推荐相关问题
• 支持流式返回，实时获取推荐问题
• 适用于构建智能对话助手、聊天机器人等应用
• 帮助用户发现更多相关话题，提升对话体验

参数

props

object

获取推荐问题的参数

botIdstring

Agent ID，指定要获取推荐问题的 Agent

namestring

Agent 名称

introductionstring

Agent 简介

agentSettingstring

Agent 设定信息

msgstring

用户发送的最新消息

historyArray<object>

历史对话记录

role

string

对话角色，user 或 assistant

content

string

对话内容

返回

Promise

StreamResult

流式返回的推荐问题结果

textStreamAsyncIterable<string>

以流式返回的推荐问题文本

dataStreamAsyncIterable<AgentStreamChunk>

以流式返回的详细对话数据

AgentStreamChunk

object

Agent 流式对话返回的数据块

creatednumber

对话时间戳

record_idstring

对话记录ID

modelstring

大模型类型

versionstring

大模型版本

typestring

回复类型：text（主要回答内容）、thinking（思考过程）、search（查询结果）、knowledge（知识库）

rolestring

对话角色，响应中固定为 assistant

contentstring

对话内容

finish_reasonstring

对话结束标志，continue（对话未结束）、stop（对话结束）

reasoning_contentstring

深度思考内容（仅deepseek-r1不为空字符串）

usageobject

token使用量

prompt_tokens

number

表示prompt的tokens数，多次返回中保持不变

completion_tokens

number

回答的token总数，在流式返回中表示到目前为止所有completion的tokens总数，多次返回中持续累加

total_tokens

number

表示prompt_tokens和completion_tokens之和

knowledge_basestring[]

对话中使用到的知识库

search_infoobject

搜索结果信息，需要开启联网查询

search_results

object[]

搜索引文信息

indexstring

搜索引文序号

titlestring

搜索引文标题

urlstring

搜索引文链接

示例

基础使用

// 获取推荐问题
const res = await ai.bot.getRecommendQuestions({
  botId: "botId-xxx",
  name: "智能助手",
  introduction: "我是一个智能对话助手",
  agentSetting: "专业、友好的AI助手",
  msg: "你好，我想了解人工智能",
  history: [
    { content: "你是谁啊", role: "user" },
    { content: "我是一个智能助手，可以帮助你解答各种问题", role: "assistant" },
  ],
});

// 流式获取推荐问题
for await (let question of res.textStream) {
  console.log("推荐问题:", question);
}

智能对话助手

// 构建智能对话助手，自动推荐相关问题
class SmartChatAssistant {
  private botId: string;
  private conversationHistory: Array<{ role: string; content: string }> = [];

  constructor(botId: string) {
    this.botId = botId;
  }

  async sendMessage(message: string): Promise<string> {
    // 添加用户消息到历史记录
    this.conversationHistory.push({ role: "user", content: message });

    // 发送消息并获取回答
    const response = await ai.bot.sendMessage({
      botId: this.botId,
      msg: message,
      history: this.conversationHistory,
    });

    let assistantReply = "";
    for await (let text of response.textStream) {
      assistantReply += text;
    }

    // 添加助手回答到历史记录
    this.conversationHistory.push({
      role: "assistant",
      content: assistantReply,
    });

    return assistantReply;
  }

  async getRecommendedQuestions(): Promise<string[]> {
    if (this.conversationHistory.length === 0) {
      return [];
    }

    const latestMessage =
      this.conversationHistory[this.conversationHistory.length - 1].content;

    const res = await ai.bot.getRecommendQuestions({
      botId: this.botId,
      name: "智能助手",
      introduction: "智能对话助手",
      agentSetting: "",
      msg: latestMessage,
      history: this.conversationHistory,
    });

    const questions: string[] = [];
    for await (let question of res.textStream) {
      questions.push(question);
    }

    return questions;
  }

  async interactiveChat() {
    console.log(
      "智能助手已启动，输入 'exit' 退出，输入 'suggest' 获取推荐问题"
    );

    while (true) {
      const userInput = await this.getUserInput("请输入: ");

      if (userInput.toLowerCase() === "exit") {
        console.log("再见！");
        break;
      }

      if (userInput.toLowerCase() === "suggest") {
        console.log("正在为您推荐相关问题...");
        const questions = await this.getRecommendedQuestions();
        console.log("推荐问题:");
        questions.forEach((q, i) => console.log(`${i + 1}. ${q}`));
        continue;
      }

      console.log("思考中...");
      const reply = await this.sendMessage(userInput);
      console.log("助手回答:", reply);

      // 自动推荐相关问题
      console.log("\n您可能还想了解:");
      const questions = await this.getRecommendedQuestions();
      questions.slice(0, 3).forEach((q, i) => console.log(`${i + 1}. ${q}`));
      console.log("---");
    }
  }

  private getUserInput(prompt: string): Promise<string> {
    return new Promise((resolve) => {
      // 实际应用中这里可以是浏览器输入框
      const input = prompt;
      resolve(input);
    });
  }
}

// 使用示例
const assistant = new SmartChatAssistant("smart-bot-id");
await assistant.interactiveChat();

电商客服场景

// 电商客服场景下的推荐问题应用
class EcommerceCustomerService {
  private botId: string;

  constructor(botId: string) {
    this.botId = botId;
  }

  async handleCustomerQuery(
    query: string
  ): Promise<{ answer: string; suggestions: string[] }> {
    const response = await ai.bot.sendMessage({
      botId: this.botId,
      msg: query,
      history: [],
    });

    let answer = "";
    for await (let text of response.textStream) {
      answer += text;
    }

    // 获取相关推荐问题
    const recommendRes = await ai.bot.getRecommendQuestions({
      botId: this.botId,
      name: "电商客服",
      introduction: "专业的电商客服助手",
      agentSetting: "专注于解答电商相关问题",
      msg: query,
      history: [
        { role: "user", content: query },
        { role: "assistant", content: answer },
      ],
    });

    const suggestions: string[] = [];
    for await (let suggestion of recommendRes.textStream) {
      suggestions.push(suggestion);
    }

    return { answer, suggestions };
  }

  // 处理常见电商场景
  async handleProductInquiry(productName: string) {
    const query = `我想了解 ${productName} 这个产品`;
    const result = await this.handleCustomerQuery(query);

    console.log("客服回答:", result.answer);
    console.log("\n您可能还想了解:");
    result.suggestions.forEach((suggestion, index) => {
      console.log(`${index + 1}. ${suggestion}`);
    });

    return result;
  }

  async handleOrderInquiry(orderId: string) {
    const query = `查询订单 ${orderId} 的状态`;
    const result = await this.handleCustomerQuery(query);

    console.log("订单状态:", result.answer);
    console.log("\n相关服务:");
    result.suggestions.forEach((suggestion, index) => {
      console.log(`${index + 1}. ${suggestion}`);
    });

    return result;
  }
}

// 使用示例
const customerService = new EcommerceCustomerService("ecommerce-bot-id");

// 处理产品咨询
await customerService.handleProductInquiry("iPhone 15");

// 处理订单查询
await customerService.handleOrderInquiry("ORD20231226001");

---

createConversation

function createConversation(props: {
  botId: string;
  title?: string;
}): Promise<IConversation>;

创建与 Agent 的新对话，建立独立的对话会话。

• 为指定的 Agent 创建新的对话会话
• 支持自定义对话标题，便于管理和识别
• 每个对话会话独立存储对话历史
• 适用于多用户、多场景的对话管理需求

参数

props

object

创建对话的参数

botIdstring

Agent ID，指定要创建对话的 Agent

title可选string

对话标题，用于标识对话内容，可选

返回

Promise

IConversation

创建的对话信息

conversationIdstring

对话 ID，用于后续对话操作

titlestring

对话标题

createTimestring

对话创建时间

updateTimestring

对话最后更新时间

示例

基础创建

// 创建新的对话会话
const conversation = await ai.bot.createConversation({
  botId: "botId-xxx",
  title: "技术咨询对话",
});

console.log("对话创建成功:", conversation.conversationId);
console.log("对话标题:", conversation.title);

多用户对话管理

// 多用户对话管理系统
class ConversationManager {
  private botId: string;
  private userConversations: Map<string, string> = new Map(); // userId -> conversationId

  constructor(botId: string) {
    this.botId = botId;
  }

  async createUserConversation(
    userId: string,
    title?: string
  ): Promise<string> {
    const conversation = await ai.bot.createConversation({
      botId: this.botId,
      title: title || `${userId}的对话`,
    });

    this.userConversations.set(userId, conversation.conversationId);
    console.log(`为用户 ${userId} 创建对话: ${conversation.conversationId}`);

    return conversation.conversationId;
  }

  async getUserConversation(userId: string): Promise<string | null> {
    return this.userConversations.get(userId) || null;
  }

  async sendMessageToUser(userId: string, message: string): Promise<string> {
    let conversationId = await this.getUserConversation(userId);

    if (!conversationId) {
      conversationId = await this.createUserConversation(userId);
    }

    const response = await ai.bot.sendMessage({
      botId: this.botId,
      msg: message,
      conversationId: conversationId,
    });

    let reply = "";
    for await (let text of response.textStream) {
      reply += text;
    }

    return reply;
  }

  async getConversationHistory(userId: string) {
    const conversationId = await this.getUserConversation(userId);
    if (!conversationId) {
      return [];
    }

    const conversation = await ai.bot.getConversation({
      botId: this.botId,
      conversationId: conversationId,
    });

    return conversation.history || [];
  }
}

// 使用示例
const manager = new ConversationManager("support-bot-id");

// 为用户创建对话
await manager.createUserConversation("user123", "技术支持对话");

// 发送消息
const reply = await manager.sendMessageToUser("user123", "我的账户有问题");
console.log("助手回复:", reply);

// 获取对话历史
const history = await manager.getConversationHistory("user123");
console.log("对话历史:", history);

多主题对话管理

// 多主题对话管理系统
class MultiTopicConversationManager {
  private botId: string;
  private topicConversations: Map<string, string> = new Map(); // topic -> conversationId

  constructor(botId: string) {
    this.botId = botId;
  }

  async createTopicConversation(topic: string): Promise<string> {
    const conversation = await ai.bot.createConversation({
      botId: this.botId,
      title: `关于${topic}的讨论`,
    });

    this.topicConversations.set(topic, conversation.conversationId);
    console.log(`创建主题对话: ${topic} -> ${conversation.conversationId}`);

    return conversation.conversationId;
  }

  async switchTopic(topic: string, message: string): Promise<string> {
    let conversationId = this.topicConversations.get(topic);

    if (!conversationId) {
      conversationId = await this.createTopicConversation(topic);
    }

    const response = await ai.bot.sendMessage({
      botId: this.botId,
      msg: message,
      conversationId: conversationId,
    });

    let reply = "";
    for await (let text of response.textStream) {
      reply += text;
    }

    return reply;
  }

  async getTopicSummary(topic: string): Promise<string> {
    const conversationId = this.topicConversations.get(topic);
    if (!conversationId) {
      return "该主题暂无对话记录";
    }

    const conversation = await ai.bot.getConversation({
      botId: this.botId,
      conversationId: conversationId,
    });

    const history = conversation.history || [];
    const messageCount = history.length;
    const lastMessage =
      history.length > 0 ? history[history.length - 1].content : "无";

    return `主题: ${topic}\n对话数量: ${messageCount}\n最后消息: ${lastMessage}`;
  }

  listTopics(): string[] {
    return Array.from(this.topicConversations.keys());
  }
}

// 使用示例
const topicManager = new MultiTopicConversationManager("multi-topic-bot-id");

// 在不同主题间切换对话
const techReply = await topicManager.switchTopic(
  "技术问题",
  "如何解决这个bug？"
);
console.log("技术问题回复:", techReply);

const businessReply = await topicManager.switchTopic(
  "商务合作",
  "我们想谈合作"
);
console.log("商务合作回复:", businessReply);

// 回到技术问题继续对话
const techReply2 = await topicManager.switchTopic(
  "技术问题",
  "还有其他解决方案吗？"
);
console.log("技术问题继续:", techReply2);

// 查看主题摘要
console.log("所有主题:", topicManager.listTopics());
const summary = await topicManager.getTopicSummary("技术问题");
console.log("技术问题摘要:\n", summary);

---

getConversation

function getConversation(props: {
  botId: string;
  pageSize?: number;
  pageNumber?: number;
  isDefault?: boolean;
}): Promise<ConversationListResult>;

获取 Agent 的对话列表，支持分页查询和条件筛选。

• 查询指定 Agent 的所有对话记录
• 支持分页查询，便于管理大量对话
• 可筛选默认对话或所有对话
• 适用于对话管理、历史记录查看等场景

参数

props

object

获取对话列表的参数

botIdstring

Agent ID，指定要获取对话的 Agent

pageSize可选number

分页大小，每页显示的对话数量，默认 10

pageNumber可选number

分页下标，从 1 开始，默认 1

isDefault可选boolean

是否只获取默认对话，false 为获取所有对话

返回

Promise

ConversationListResult

对话列表查询结果

totalnumber

对话总数

conversationsArray<IConversation>

对话列表

conversationId

string

对话 ID

title

string

对话标题

createTime

string

对话创建时间

updateTime

string

对话最后更新时间

messageCount

number

对话中的消息数量

示例

基础查询

// 获取对话列表
const result = await ai.bot.getConversation({
  botId: "botId-xxx",
  pageSize: 10,
  pageNumber: 1,
  isDefault: false,
});

console.log(`共有 ${result.total} 个对话`);
result.conversations.forEach((conversation, index) => {
  console.log(
    `${index + 1}. ${conversation.title} (${conversation.messageCount} 条消息)`
  );
});

对话管理系统

// 对话管理系统
class ConversationManagementSystem {
  private botId: string;

  constructor(botId: string) {
    this.botId = botId;
  }

  async getAllConversations(): Promise<IConversation[]> {
    let allConversations: IConversation[] = [];
    let pageNumber = 1;
    const pageSize = 50;

    while (true) {
      const result = await ai.bot.getConversation({
        botId: this.botId,
        pageSize: pageSize,
        pageNumber: pageNumber,
        isDefault: false,
      });

      allConversations = allConversations.concat(result.conversations);

      if (result.conversations.length < pageSize) {
        break;
      }

      pageNumber++;
    }

    return allConversations;
  }

  async searchConversations(keyword: string): Promise<IConversation[]> {
    const allConversations = await this.getAllConversations();

    return allConversations.filter((conversation) =>
      conversation.title.toLowerCase().includes(keyword.toLowerCase())
    );
  }

  async getRecentConversations(days: number = 7): Promise<IConversation[]> {
    const allConversations = await this.getAllConversations();
    const cutoffDate = new Date();
    cutoffDate.setDate(cutoffDate.getDate() - days);

    return allConversations.filter((conversation) => {
      const updateTime = new Date(conversation.updateTime);
      return updateTime >= cutoffDate;
    });
  }

  async getConversationStats(): Promise<{
    total: number;
    recent: number;
    averageMessages: number;
  }> {
    const allConversations = await this.getAllConversations();
    const recentConversations = await this.getRecentConversations(7);

    const totalMessages = allConversations.reduce(
      (sum, conv) => sum + conv.messageCount,
      0
    );
    const averageMessages =
      allConversations.length > 0 ? totalMessages / allConversations.length : 0;

    return {
      total: allConversations.length,
      recent: recentConversations.length,
      averageMessages: Math.round(averageMessages),
    };
  }

  async exportConversationList(): Promise<string> {
    const allConversations = await this.getAllConversations();

    const csv = [
      "对话ID,标题,创建时间,最后更新时间,消息数量",
      ...allConversations.map(
        (conv) =>
          `${conv.conversationId},"${conv.title}",${conv.createTime},${conv.updateTime},${conv.messageCount}`
      ),
    ].join("\n");

    return csv;
  }
}

// 使用示例
const manager = new ConversationManagementSystem("management-bot-id");

// 获取对话统计
const stats = await manager.getConversationStats();
console.log(`总对话数: ${stats.total}`);
console.log(`最近7天对话数: ${stats.recent}`);
console.log(`平均消息数: ${stats.averageMessages}`);

// 搜索对话
const searchResults = await manager.searchConversations("技术");
console.log(`找到 ${searchResults.length} 个技术相关对话`);

// 导出对话列表
const csvData = await manager.exportConversationList();
console.log("对话列表导出完成");

批量对话分析

// 批量对话分析工具
class ConversationAnalyzer {
  private botId: string;

  constructor(botId: string) {
    this.botId = botId;
  }

  async analyzeConversationPatterns(): Promise<{
    timeDistribution: { hour: number; count: number }[];
    messageLengthStats: { avgLength: number; maxLength: number };
    topicDistribution: Map<string, number>;
  }> {
    const allConversations = await this.getAllConversations();

    // 时间分布分析
    const timeDistribution = Array.from({ length: 24 }, (_, hour) => ({
      hour,
      count: allConversations.filter((conv) => {
        const hour = new Date(conv.createTime).getHours();
        return hour === hour;
      }).length,
    }));

    // 消息长度统计
    const messageLengths = allConversations.map((conv) => conv.messageCount);
    const avgLength =
      messageLengths.reduce((sum, len) => sum + len, 0) / messageLengths.length;
    const maxLength = Math.max(...messageLengths);

    // 主题分布分析（基于标题关键词）
    const topicDistribution = new Map<string, number>();
    const topics = ["技术", "商务", "客服", "产品", "其他"];

    topics.forEach((topic) => {
      const count = allConversations.filter((conv) =>
        conv.title.includes(topic)
      ).length;
      topicDistribution.set(topic, count);
    });

    return {
      timeDistribution,
      messageLengthStats: { avgLength, maxLength },
      topicDistribution,
    };
  }

  async getConversationTrends(): Promise<{
    dailyTrends: { date: string; count: number }[];
    weeklyTrends: { week: string; count: number }[];
  }> {
    const allConversations = await this.getAllConversations();

    // 按日期分组
    const dailyCounts = new Map<string, number>();
    allConversations.forEach((conv) => {
      const date = new Date(conv.createTime).toISOString().split("T")[0];
      dailyCounts.set(date, (dailyCounts.get(date) || 0) + 1);
    });

    const dailyTrends = Array.from(dailyCounts.entries())
      .map(([date, count]) => ({
        date,
        count,
      }))
      .sort((a, b) => a.date.localeCompare(b.date));

    // 按周分组
    const weeklyCounts = new Map<string, number>();
    allConversations.forEach((conv) => {
      const date = new Date(conv.createTime);
      const weekStart = new Date(date);
      weekStart.setDate(date.getDate() - date.getDay());
      const weekKey = weekStart.toISOString().split("T")[0];
      weeklyCounts.set(weekKey, (weeklyCounts.get(weekKey) || 0) + 1);
    });

    const weeklyTrends = Array.from(weeklyCounts.entries())
      .map(([week, count]) => ({
        week,
        count,
      }))
      .sort((a, b) => a.week.localeCompare(b.week));

    return { dailyTrends, weeklyTrends };
  }

  private async getAllConversations(): Promise<IConversation[]> {
    let allConversations: IConversation[] = [];
    let pageNumber = 1;
    const pageSize = 100;

    while (true) {
      const result = await ai.bot.getConversation({
        botId: this.botId,
        pageSize: pageSize,
        pageNumber: pageNumber,
        isDefault: false,
      });

      allConversations = allConversations.concat(result.conversations);

      if (result.conversations.length < pageSize) {
        break;
      }

      pageNumber++;
    }

    return allConversations;
  }
}

// 使用示例
const analyzer = new ConversationAnalyzer("analytics-bot-id");

// 分析对话模式
const patterns = await analyzer.analyzeConversationPatterns();
console.log("对话时间分布:", patterns.timeDistribution);
console.log("消息长度统计:", patterns.messageLengthStats);
console.log("主题分布:", patterns.topicDistribution);

// 获取对话趋势
const trends = await analyzer.getConversationTrends();
console.log("每日趋势:", trends.dailyTrends);
console.log("每周趋势:", trends.weeklyTrends);

---

deleteConversation

function deleteConversation(props: {
  botId: string;
  conversationId: string;
}): Promise<void>;

删除指定的对话会话，清理对话历史记录。

• 永久删除指定的对话会话
• 清理对话中的所有消息记录
• 释放存储空间，优化系统性能
• 适用于对话管理、数据清理等场景

参数

props

object

删除对话的参数

botIdstring

Agent ID，指定要删除对话的 Agent

conversationIdstring

要删除的对话 ID

返回

Promise

void

删除操作完成，无返回值

示例

基础删除

// 删除指定的对话
await ai.bot.deleteConversation({
  botId: "botId-xxx",
  conversationId: "conv-123",
});

console.log("对话删除成功");

批量对话清理

// 批量对话清理工具
class ConversationCleaner {
  private botId: string;

  constructor(botId: string) {
    this.botId = botId;
  }

  async deleteOldConversations(daysThreshold: number = 30): Promise<number> {
    const allConversations = await this.getAllConversations();
    const cutoffDate = new Date();
    cutoffDate.setDate(cutoffDate.getDate() - daysThreshold);

    const oldConversations = allConversations.filter((conversation) => {
      const updateTime = new Date(conversation.updateTime);
      return updateTime < cutoffDate;
    });

    console.log(
      `找到 ${oldConversations.length} 个超过 ${daysThreshold} 天的旧对话`
    );

    // 批量删除旧对话
    let deletedCount = 0;
    for (const conversation of oldConversations) {
      try {
        await ai.bot.deleteConversation({
          botId: this.botId,
          conversationId: conversation.conversationId,
        });
        deletedCount++;
        console.log(`已删除对话: ${conversation.title}`);
      } catch (error) {
        console.error(`删除对话失败: ${conversation.title}`, error);
      }
    }

    console.log(`成功删除 ${deletedCount} 个旧对话`);
    return deletedCount;
  }

  async deleteEmptyConversations(): Promise<number> {
    const allConversations = await this.getAllConversations();
    const emptyConversations = allConversations.filter(
      (conversation) => conversation.messageCount === 0
    );

    console.log(`找到 ${emptyConversations.length} 个空对话`);

    let deletedCount = 0;
    for (const conversation of emptyConversations) {
      try {
        await ai.bot.deleteConversation({
          botId: this.botId,
          conversationId: conversation.conversationId,
        });
        deletedCount++;
      } catch (error) {
        console.error(`删除空对话失败: ${conversation.title}`, error);
      }
    }

    console.log(`成功删除 ${deletedCount} 个空对话`);
    return deletedCount;
  }

  async deleteConversationsByTitle(keyword: string): Promise<number> {
    const allConversations = await this.getAllConversations();
    const matchingConversations = allConversations.filter((conversation) =>
      conversation.title.toLowerCase().includes(keyword.toLowerCase())
    );

    console.log(
      `找到 ${matchingConversations.length} 个包含"${keyword}"的对话`
    );

    let deletedCount = 0;
    for (const conversation of matchingConversations) {
      try {
        await ai.bot.deleteConversation({
          botId: this.botId,
          conversationId: conversation.conversationId,
        });
        deletedCount++;
        console.log(`已删除: ${conversation.title}`);
      } catch (error) {
        console.error(`删除对话失败: ${conversation.title}`, error);
      }
    }

    return deletedCount;
  }

  private async getAllConversations(): Promise<IConversation[]> {
    let allConversations: IConversation[] = [];
    let pageNumber = 1;
    const pageSize = 100;

    while (true) {
      const result = await ai.bot.getConversation({
        botId: this.botId,
        pageSize: pageSize,
        pageNumber: pageNumber,
        isDefault: false,
      });

      allConversations = allConversations.concat(result.conversations);

      if (result.conversations.length < pageSize) {
        break;
      }

      pageNumber++;
    }

    return allConversations;
  }
}

// 使用示例
const cleaner = new ConversationCleaner("cleanup-bot-id");

// 删除30天前的旧对话
const oldDeleted = await cleaner.deleteOldConversations(30);
console.log(`清理了 ${oldDeleted} 个旧对话`);

// 删除空对话
const emptyDeleted = await cleaner.deleteEmptyConversations();
console.log(`清理了 ${emptyDeleted} 个空对话`);

// 删除包含特定关键词的对话
const keywordDeleted = await cleaner.deleteConversationsByTitle("测试");
console.log(`清理了 ${keywordDeleted} 个测试对话`);

用户数据管理

// 用户数据管理系统
class UserDataManager {
  private botId: string;
  private userConversations: Map<string, string[]> = new Map(); // userId -> conversationIds

  constructor(botId: string) {
    this.botId = botId;
  }

  async deleteUserConversations(userId: string): Promise<number> {
    const userConversationIds = this.userConversations.get(userId) || [];

    console.log(`为用户 ${userId} 删除 ${userConversationIds.length} 个对话`);

    let deletedCount = 0;
    for (const conversationId of userConversationIds) {
      try {
        await ai.bot.deleteConversation({
          botId: this.botId,
          conversationId: conversationId,
        });
        deletedCount++;
        console.log(`已删除对话: ${conversationId}`);
      } catch (error) {
        console.error(`删除对话失败: ${conversationId}`, error);
      }
    }

    // 从内存中移除用户对话记录
    this.userConversations.delete(userId);

    console.log(`用户 ${userId} 的数据清理完成，删除了 ${deletedCount} 个对话`);
    return deletedCount;
  }

  async deleteAllUserData(): Promise<number> {
    const allUserIds = Array.from(this.userConversations.keys());
    let totalDeleted = 0;

    console.log(`开始清理 ${allUserIds.length} 个用户的数据`);

    for (const userId of allUserIds) {
      const deletedCount = await this.deleteUserConversations(userId);
      totalDeleted += deletedCount;
    }

    console.log(`总共清理了 ${totalDeleted} 个对话`);
    return totalDeleted;
  }

  async deleteUserDataByCondition(
    condition: (userId: string, conversationIds: string[]) => boolean
  ): Promise<number> {
    const usersToDelete: string[] = [];

    for (const [userId, conversationIds] of this.userConversations.entries()) {
      if (condition(userId, conversationIds)) {
        usersToDelete.push(userId);
      }
    }

    console.log(`找到 ${usersToDelete.length} 个符合条件的用户`);

    let totalDeleted = 0;
    for (const userId of usersToDelete) {
      const deletedCount = await this.deleteUserConversations(userId);
      totalDeleted += deletedCount;
    }

    return totalDeleted;
  }

  // 添加用户对话记录
  addUserConversation(userId: string, conversationId: string) {
    if (!this.userConversations.has(userId)) {
      this.userConversations.set(userId, []);
    }
    this.userConversations.get(userId)!.push(conversationId);
  }

  // 获取用户对话统计
  getUserStats(): { totalUsers: number; totalConversations: number } {
    let totalConversations = 0;
    for (const conversationIds of this.userConversations.values()) {
      totalConversations += conversationIds.length;
    }

    return {
      totalUsers: this.userConversations.size,
      totalConversations: totalConversations,
    };
  }
}

// 使用示例
const dataManager = new UserDataManager("data-management-bot-id");

// 添加用户对话记录
dataManager.addUserConversation("user1", "conv-123");
dataManager.addUserConversation("user1", "conv-456");
dataManager.addUserConversation("user2", "conv-789");

// 查看统计
const stats = dataManager.getUserStats();
console.log(`用户数: ${stats.totalUsers}, 对话数: ${stats.totalConversations}`);

// 删除特定用户数据
const deletedCount = await dataManager.deleteUserConversations("user1");
console.log(`删除了用户1的 ${deletedCount} 个对话`);

// 条件删除：删除对话数超过阈值的用户
const conditionDeleted = await dataManager.deleteUserDataByCondition(
  (userId, conversationIds) => conversationIds.length > 5
);
console.log(`删除了 ${conditionDeleted} 个对话数超过5的用户数据`);

---

speechToText

function speechToText(props: {
  botId: string;
  engSerViceType: string;
  voiceFormat: string;
  url: string;
  isPreview?: boolean;
}): Promise<SpeechToTextResult>;

将语音文件转换为文字，支持多种音频格式和语言识别。

• 支持多种音频格式：MP3、WAV、AAC 等
• 支持多种语言识别引擎
• 可处理本地或远程音频文件
• 适用于语音助手、会议记录、语音笔记等场景

参数

props

object

语音转文字的参数

botIdstring

Agent ID，指定进行语音识别的 Agent

engSerViceTypestring

语音识别引擎类型，如 '16k_zh'（中文16kHz）、'8k_en'（英文8kHz）

voiceFormatstring

音频文件格式，支持 'mp3'、'wav'、'aac'、'amr' 等

urlstring

音频文件 URL，支持本地文件路径或远程 URL

isPreview可选boolean

是否为预览模式，预览模式可能有限制或水印

返回

Promise

SpeechToTextResult

语音识别结果

textstring

识别出的文本内容

confidencenumber

识别置信度，0-1之间的数值

durationnumber

音频时长（秒）

languagestring

识别出的语言类型

示例

基础语音识别

// 将语音文件转换为文字
const result = await ai.bot.speechToText({
  botId: "speech-bot-id",
  engSerViceType: "16k_zh",
  voiceFormat: "mp3",
  url: "https://example.com/audio.mp3",
});

console.log("识别结果:", result.text);
console.log("置信度:", result.confidence);
console.log("音频时长:", result.duration, "秒");
console.log("识别语言:", result.language);

语音助手应用

// 语音助手应用
class VoiceAssistant {
  private botId: string;

  constructor(botId: string) {
    this.botId = botId;
  }

  async processVoiceCommand(audioUrl: string): Promise<{
    text: string;
    response: string;
    action?: string;
  }> {
    // 语音转文字
    const speechResult = await ai.bot.speechToText({
      botId: this.botId,
      engSerViceType: "16k_zh",
      voiceFormat: "mp3",
      url: audioUrl,
    });

    console.log("语音识别结果:", speechResult.text);

    // 根据识别结果生成回复
    const response = await ai.bot.sendMessage({
      botId: this.botId,
      msg: speechResult.text,
    });

    let assistantReply = "";
    for await (let text of response.textStream) {
      assistantReply += text;
    }

    // 分析是否需要执行特定操作
    const action = this.analyzeCommand(speechResult.text);

    return {
      text: speechResult.text,
      response: assistantReply,
      action: action,
    };
  }

  async createVoiceNote(audioUrl: string, title?: string): Promise<string> {
    const result = await ai.bot.speechToText({
      botId: this.botId,
      engSerViceType: "16k_zh",
      voiceFormat: "mp3",
      url: audioUrl,
    });

    const noteContent = `
语音笔记 - ${title || new Date().toLocaleString()}
音频时长: ${result.duration}秒
识别置信度: ${Math.round(result.confidence * 100)}%

内容:
${result.text}
    `;

    console.log("语音笔记创建成功:");
    console.log(noteContent);

    return noteContent;
  }

  async batchTranscribe(
    audioFiles: Array<{ url: string; name: string }>
  ): Promise<Map<string, string>> {
    const results = new Map<string, string>();

    console.log(`开始批量转换 ${audioFiles.length} 个音频文件`);

    for (const file of audioFiles) {
      try {
        const result = await ai.bot.speechToText({
          botId: this.botId,
          engSerViceType: "16k_zh",
          voiceFormat: this.getFileFormat(file.url),
          url: file.url,
        });

        results.set(file.name, result.text);
        console.log(`✓ ${file.name}: 转换成功`);
      } catch (error) {
        console.error(`✗ ${file.name}: 转换失败`, error);
        results.set(file.name, "转换失败");
      }
    }

    console.log("批量转换完成");
    return results;
  }

  private analyzeCommand(text: string): string | undefined {
    const commands = {
      天气: "weather",
      时间: "time",
      播放: "play",
      停止: "stop",
      搜索: "search",
    };

    for (const [keyword, action] of Object.entries(commands)) {
      if (text.includes(keyword)) {
        return action;
      }
    }

    return undefined;
  }

  private getFileFormat(url: string): string {
    const extension = url.split(".").pop()?.toLowerCase();
    return extension === "wav" ? "wav" : "mp3";
  }
}

// 使用示例
const voiceAssistant = new VoiceAssistant("voice-bot-id");

// 处理语音命令
const commandResult = await voiceAssistant.processVoiceCommand(
  "voice-command.mp3"
);
console.log("语音命令:", commandResult.text);
console.log("助手回复:", commandResult.response);
if (commandResult.action) {
  console.log("执行操作:", commandResult.action);
}

// 创建语音笔记
const note = await voiceAssistant.createVoiceNote(
  "meeting-recording.mp3",
  "会议记录"
);

// 批量转换音频文件
const audioFiles = [
  { url: "audio1.mp3", name: "采访录音1" },
  { url: "audio2.wav", name: "会议录音2" },
  { url: "audio3.mp3", name: "讲座录音3" },
];
const transcriptions = await voiceAssistant.batchTranscribe(audioFiles);

会议记录系统

// 智能会议记录系统
class MeetingTranscriptionSystem {
  private botId: string;

  constructor(botId: string) {
    this.botId = botId;
  }

  async transcribeMeeting(
    audioUrl: string,
    participants: string[] = []
  ): Promise<{
    transcript: string;
    summary: string;
    actionItems: string[];
    speakerAnalysis: Map<string, number>;
  }> {
    // 语音转文字
    const speechResult = await ai.bot.speechToText({
      botId: this.botId,
      engSerViceType: "16k_zh",
      voiceFormat: "mp3",
      url: audioUrl,
    });

    console.log("会议录音转换完成，时长:", speechResult.duration, "秒");

    // 生成会议摘要
    const summaryPrompt = `请为以下会议内容生成摘要，提取关键决策和行动项：\n\n${speechResult.text}`;
    const summaryResponse = await ai.bot.sendMessage({
      botId: this.botId,
      msg: summaryPrompt,
    });

    let summary = "";
    for await (let text of summaryResponse.textStream) {
      summary += text;
    }

    // 提取行动项
    const actionPrompt = `从以下会议内容中提取具体的行动项和负责人：\n\n${speechResult.text}`;
    const actionResponse = await ai.bot.sendMessage({
      botId: this.botId,
      msg: actionPrompt,
    });

    let actionText = "";
    for await (let text of actionResponse.textStream) {
      actionText += text;
    }

    const actionItems = actionText
      .split("\n")
      .filter((item) => item.trim().length > 0);

    // 分析发言分布（简单版本）
    const speakerAnalysis = this.analyzeSpeakerDistribution(
      speechResult.text,
      participants
    );

    return {
      transcript: speechResult.text,
      summary: summary,
      actionItems: actionItems,
      speakerAnalysis: speakerAnalysis,
    };
  }

  async generateMeetingMinutes(
    audioUrl: string,
    meetingTitle: string,
    participants: string[]
  ): Promise<string> {
    const meetingData = await this.transcribeMeeting(audioUrl, participants);

    const minutes = `
会议纪要
========

会议标题: ${meetingTitle}
会议时长: ${
      meetingData.transcript.length > 0
        ? "约" + Math.ceil(meetingData.transcript.length / 200) + "分钟"
        : "未知"
    }
参会人员: ${participants.join(", ")}

会议摘要
--------
${meetingData.summary}

会议记录
--------
${meetingData.transcript}

行动项
------
${meetingData.actionItems
  .map((item, index) => `${index + 1}. ${item}`)
  .join("\n")}

发言统计
--------
${Array.from(meetingData.speakerAnalysis.entries())
  .map(([speaker, count]) => `${speaker}: ${count} 次发言`)
  .join("\n")}
    `;

    console.log("会议纪要生成完成");
    return minutes;
  }

  private analyzeSpeakerDistribution(
    transcript: string,
    participants: string[]
  ): Map<string, number> {
    const distribution = new Map<string, number>();

    // 为每个参与者初始化计数
    participants.forEach((participant) => {
      distribution.set(participant, 0);
    });

    // 简单分析：统计参与者名字出现的次数
    participants.forEach((participant) => {
      const regex = new RegExp(participant, "gi");
      const count = (transcript.match(regex) || []).length;
      distribution.set(participant, count);
    });

    return distribution;
  }
}

// 使用示例
const meetingSystem = new MeetingTranscriptionSystem("meeting-bot-id");

// 生成会议纪要
const minutes = await meetingSystem.generateMeetingMinutes(
  "meeting-recording.mp3",
  "季度规划会议",
  ["张三", "李四", "王五"]
);

console.log(minutes);

// 直接转录会议
const meetingData = await meetingSystem.transcribeMeeting("team-meeting.mp3");
console.log("会议摘要:", meetingData.summary);
console.log("行动项:", meetingData.actionItems);

---

textToSpeech

function textToSpeech(props: {
  botId: string;
  voiceType: number;
  text: string;
  isPreview?: boolean;
}): Promise<TextToSpeechResult>;

将文本转换为语音，支持多种语音类型和音色选择。

• 支持多种语音类型和音色选择
• 可调节语速、音调等参数
• 支持长文本分段转换
• 适用于语音播报、有声读物、语音助手等场景

参数

props

object

文字转语音的参数

botIdstring

Agent ID，指定进行语音合成的 Agent

voiceTypenumber

语音类型，1-标准女声，2-标准男声，3-情感女声，4-情感男声等

textstring

要转换为语音的文本内容

isPreview可选boolean

是否为预览模式，预览模式可能有时间限制或水印

返回

Promise

TextToSpeechResult

文字转语音的结果

audioUrlstring

生成的音频文件 URL

durationnumber

音频时长（秒）

formatstring

音频格式，如 'mp3'、'wav' 等

sizenumber

音频文件大小（字节）

示例

基础语音合成

// 将文本转换为语音
const result = await ai.bot.textToSpeech({
  botId: "tts-bot-id",
  voiceType: 1, // 标准女声
  text: "你好，我是AI语音助手，很高兴为您服务。",
});

console.log("语音合成完成:");
console.log("音频地址:", result.audioUrl);
console.log("音频时长:", result.duration, "秒");
console.log("文件格式:", result.format);
console.log("文件大小:", result.size, "字节");

智能语音助手

// 智能语音助手
class VoiceAssistant {
  private botId: string;
  private voiceType: number;

  constructor(botId: string, voiceType: number = 1) {
    this.botId = botId;
    this.voiceType = voiceType;
  }

  async speakText(text: string): Promise<string> {
    const result = await ai.bot.textToSpeech({
      botId: this.botId,
      voiceType: this.voiceType,
      text: text,
    });

    console.log(`语音合成完成: ${result.duration}秒`);

    // 在实际应用中，这里可以播放音频
    this.playAudio(result.audioUrl);

    return result.audioUrl;
  }

  async createAudioBook(
    chapters: Array<{ title: string; content: string }>
  ): Promise<Map<string, string>> {
    const audioFiles = new Map<string, string>();

    console.log(`开始制作有声读物，共 ${chapters.length} 章`);

    for (const [index, chapter] of chapters.entries()) {
      console.log(`正在合成第 ${index + 1} 章: ${chapter.title}`);

      try {
        const result = await ai.bot.textToSpeech({
          botId: this.botId,
          voiceType: this.voiceType,
          text: chapter.content,
        });

        audioFiles.set(chapter.title, result.audioUrl);
        console.log(`✓ 第 ${index + 1} 章合成完成，时长: ${result.duration}秒`);
      } catch (error) {
        console.error(`✗ 第 ${index + 1} 章合成失败`, error);
      }
    }

    console.log("有声读物制作完成");
    return audioFiles;
  }

  async generateInteractiveStory(
    storyTitle: string,
    storyContent: string[]
  ): Promise<{
    audioFiles: string[];
    totalDuration: number;
  }> {
    const audioFiles: string[] = [];
    let totalDuration = 0;

    console.log(`开始生成交互式故事: ${storyTitle}`);

    for (const [index, paragraph] of storyContent.entries()) {
      console.log(`合成第 ${index + 1} 段内容`);

      const result = await ai.bot.textToSpeech({
        botId: this.botId,
        voiceType: this.voiceType,
        text: paragraph,
      });

      audioFiles.push(result.audioUrl);
      totalDuration += result.duration;

      console.log(`✓ 第 ${index + 1} 段合成完成，时长: ${result.duration}秒`);
    }

    console.log(`故事生成完成，总时长: ${totalDuration}秒`);
    return { audioFiles, totalDuration };
  }

  async createVoiceAnnouncement(
    message: string,
    repeat: number = 1
  ): Promise<string[]> {
    const announcements: string[] = [];

    console.log(`生成语音播报，重复 ${repeat} 次`);

    for (let i = 0; i < repeat; i++) {
      const result = await ai.bot.textToSpeech({
        botId: this.botId,
        voiceType: this.voiceType,
        text: message,
      });

      announcements.push(result.audioUrl);
      console.log(`✓ 第 ${i + 1} 次播报生成完成`);
    }

    return announcements;
  }

  private playAudio(audioUrl: string): void {
    // 在实际应用中，这里可以实现音频播放逻辑
    console.log("播放音频:", audioUrl);

    // 示例：使用HTML5 Audio API播放
    // const audio = new Audio(audioUrl);
    // audio.play();
  }

  // 设置语音类型
  setVoiceType(voiceType: number): void {
    this.voiceType = voiceType;
    console.log(`语音类型已设置为: ${voiceType}`);
  }

  // 获取支持的语音类型
  getSupportedVoices(): Array<{ id: number; name: string }> {
    return [
      { id: 1, name: "标准女声" },
      { id: 2, name: "标准男声" },
      { id: 3, name: "情感女声" },
      { id: 4, name: "情感男声" },
      { id: 5, name: "儿童声音" },
      { id: 6, name: "老年声音" },
    ];
  }
}

// 使用示例
const voiceAssistant = new VoiceAssistant("voice-bot-id", 1);

// 简单语音播报
const audioUrl = await voiceAssistant.speakText("欢迎使用语音助手系统！");

// 制作有声读物
const chapters = [
  { title: "第一章", content: "这是一个故事的开头..." },
  { title: "第二章", content: "故事继续发展..." },
  { title: "第三章", content: "故事的高潮部分..." },
];
const audioBook = await voiceAssistant.createAudioBook(chapters);

// 生成交互式故事
const storyContent = [
  "从前有一个小村庄...",
  "村庄里住着一位聪明的老人...",
  "老人教会了村民很多知识...",
];
const story = await voiceAssistant.generateInteractiveStory(
  "智慧老人",
  storyContent
);

// 创建语音播报
const announcements = await voiceAssistant.createVoiceAnnouncement(
  "请注意，系统即将维护",
  3
);

多语言语音系统

// 多语言语音合成系统
class MultiLanguageVoiceSystem {
  private botId: string;
  private languageVoices: Map<string, number> = new Map();

  constructor(botId: string) {
    this.botId = botId;
    this.initializeLanguageVoices();
  }

  private initializeLanguageVoices(): void {
    this.languageVoices.set("zh-CN", 1); // 中文-标准女声
    this.languageVoices.set("en-US", 2); // 英文-标准男声
    this.languageVoices.set("ja-JP", 3); // 日语-情感女声
    this.languageVoices.set("ko-KR", 4); // 韩语-情感男声
    this.languageVoices.set("fr-FR", 1); // 法语-标准女声
    this.languageVoices.set("de-DE", 2); // 德语-标准男声
  }

  async synthesizeMultilingualText(
    texts: Map<string, string>
  ): Promise<Map<string, string>> {
    const results = new Map<string, string>();

    console.log(`开始合成多语言文本，共 ${texts.size} 种语言`);

    for (const [language, text] of texts.entries()) {
      const voiceType = this.languageVoices.get(language) || 1;

      console.log(`合成 ${language} 语言文本`);

      try {
        const result = await ai.bot.textToSpeech({
          botId: this.botId,
          voiceType: voiceType,
          text: text,
        });

        results.set(language, result.audioUrl);
        console.log(`✓ ${language} 语言合成完成，时长: ${result.duration}秒`);
      } catch (error) {
        console.error(`✗ ${language} 语言合成失败`, error);
        results.set(language, "合成失败");
      }
    }

    return results;
  }

  async createInternationalAnnouncement(
    message: string
  ): Promise<Map<string, string>> {
    const translations = await this.translateMessage(message);
    return this.synthesizeMultilingualText(translations);
  }

  async generateLanguageLearningAudio(
    language: string,
    phrases: string[]
  ): Promise<Array<{ phrase: string; audioUrl: string }>> {
    const learningAudios: Array<{ phrase: string; audioUrl: string }> = [];
    const voiceType = this.languageVoices.get(language) || 1;

    console.log(`生成 ${language} 语言学习音频，共 ${phrases.length} 个短语`);

    for (const phrase of phrases) {
      try {
        const result = await ai.bot.textToSpeech({
          botId: this.botId,
          voiceType: voiceType,
          text: phrase,
        });

        learningAudios.push({
          phrase: phrase,
          audioUrl: result.audioUrl,
        });

        console.log(`✓ 短语 "${phrase}" 音频生成完成`);
      } catch (error) {
        console.error(`✗ 短语 "${phrase}" 音频生成失败`, error);
      }
    }

    return learningAudios;
  }

  async createTourGuideAudio(
    language: string,
    attractions: Array<{ name: string; description: string }>
  ): Promise<Map<string, string>> {
    const guideAudios = new Map<string, string>();
    const voiceType = this.languageVoices.get(language) || 1;

    console.log(
      `生成 ${language} 语言导游音频，共 ${attractions.length} 个景点`
    );

    for (const attraction of attractions) {
      const description = `${attraction.name}：${attraction.description}`;

      try {
        const result = await ai.bot.textToSpeech({
          botId: this.botId,
          voiceType: voiceType,
          text: description,
        });

        guideAudios.set(attraction.name, result.audioUrl);
        console.log(`✓ 景点 "${attraction.name}" 导游音频生成完成`);
      } catch (error) {
        console.error(`✗ 景点 "${attraction.name}" 导游音频生成失败`, error);
      }
    }

    return guideAudios;
  }

  private async translateMessage(
    message: string
  ): Promise<Map<string, string>> {
    // 在实际应用中，这里可以调用翻译API
    const translations = new Map<string, string>();

    translations.set("zh-CN", message); // 中文
    translations.set("en-US", "Hello, this is an announcement."); // 英文
    translations.set("ja-JP", "こんにちは、これはアナウンスです。"); // 日语
    translations.set("ko-KR", "안녕하세요, 공지사항입니다."); // 韩语

    return translations;
  }

  // 添加新的语言支持
  addLanguageSupport(languageCode: string, voiceType: number): void {
    this.languageVoices.set(languageCode, voiceType);
    console.log(`添加语言支持: ${languageCode} -> 语音类型 ${voiceType}`);
  }

  // 获取支持的语言列表
  getSupportedLanguages(): string[] {
    return Array.from(this.languageVoices.keys());
  }
}

// 使用示例
const multiLangSystem = new MultiLanguageVoiceSystem("multilang-bot-id");

// 多语言文本合成
const multilingualTexts = new Map([
  ["zh-CN", "欢迎使用多语言语音系统"],
  ["en-US", "Welcome to the multilingual voice system"],
  ["ja-JP", "多言語音声システムへようこそ"],
]);
const audioResults = await multiLangSystem.synthesizeMultilingualText(
  multilingualTexts
);

// 国际公告播报
const announcements = await multiLangSystem.createInternationalAnnouncement(
  "系统维护通知"
);

// 语言学习音频
const englishPhrases = ["Hello", "How are you?", "Thank you", "Goodbye"];
const learningAudios = await multiLangSystem.generateLanguageLearningAudio(
  "en-US",
  englishPhrases
);

// 导游音频
const attractions = [
  { name: "故宫", description: "中国古代宫殿建筑群，世界文化遗产" },
  { name: "长城", description: "中国古代军事防御工程，世界奇迹" },
];
const guideAudios = await multiLangSystem.createTourGuideAudio(
  "zh-CN",
  attractions
);

---

getTextToSpeechResult

function getTextToSpeechResult(props: {
  botId: string;
  taskId: string;
  isPreview?: boolean;
}): Promise<TextToSpeechResult>;

获取文字转语音任务的结果，查询语音合成任务的完成状态和生成的音频文件。

• 查询异步语音合成任务的状态和结果
• 获取生成的音频文件信息
• 支持预览模式和正式模式
• 适用于批量语音合成、长文本处理等异步场景

参数

props

object

获取语音合成结果的参数

botIdstring

Agent ID，指定进行语音合成的 Agent

taskIdstring

语音合成任务 ID，由 textToSpeech 方法返回

isPreview可选boolean

是否为预览模式，预览模式可能有时间限制或水印

返回

Promise

TextToSpeechResult

语音合成任务结果

audioUrlstring

生成的音频文件 URL

durationnumber

音频时长（秒）

formatstring

音频格式，如 'mp3'、'wav' 等

sizenumber

音频文件大小（字节）

statusstring

任务状态：'processing'（处理中）、'completed'（已完成）、'failed'（失败）

progressnumber

任务进度百分比，0-100之间的数值

示例

基础查询

// 查询语音合成任务结果
const result = await ai.bot.getTextToSpeechResult({
  botId: "tts-bot-id",
  taskId: "task-123456",
});

console.log("任务状态:", result.status);
console.log("任务进度:", result.progress, "%");

if (result.status === "completed") {
  console.log("音频地址:", result.audioUrl);
  console.log("音频时长:", result.duration, "秒");
  console.log("文件格式:", result.format);
  console.log("文件大小:", result.size, "字节");
} else if (result.status === "processing") {
  console.log("任务仍在处理中，请稍后重试");
} else if (result.status === "failed") {
  console.log("任务处理失败");
}

批量任务监控

// 批量语音合成任务监控系统
class BatchTTSMonitor {
  private botId: string;
  private taskQueue: Map<string, { text: string; createTime: Date }> =
    new Map();

  constructor(botId: string) {
    this.botId = botId;
  }

  // 添加任务到队列
  addTask(taskId: string, text: string): void {
    this.taskQueue.set(taskId, {
      text: text,
      createTime: new Date(),
    });
    console.log(`任务 ${taskId} 已添加到队列`);
  }

  // 监控任务进度
  async monitorTaskProgress(
    taskId: string,
    maxWaitTime: number = 300
  ): Promise<TextToSpeechResult> {
    const startTime = Date.now();
    const checkInterval = 5000; // 5秒检查一次

    console.log(`开始监控任务 ${taskId} 进度`);

    while (true) {
      const result = await ai.bot.getTextToSpeechResult({
        botId: this.botId,
        taskId: taskId,
      });

      console.log(
        `任务 ${taskId} 状态: ${result.status}, 进度: ${result.progress}%`
      );

      if (result.status === "completed") {
        console.log(`任务 ${taskId} 已完成，音频时长: ${result.duration}秒`);
        this.taskQueue.delete(taskId);
        return result;
      }

      if (result.status === "failed") {
        console.error(`任务 ${taskId} 处理失败`);
        this.taskQueue.delete(taskId);
        throw new Error(`语音合成任务失败: ${taskId}`);
      }

      // 检查是否超时
      if (Date.now() - startTime > maxWaitTime * 1000) {
        console.error(`任务 ${taskId} 监控超时`);
        throw new Error(`语音合成任务超时: ${taskId}`);
      }

      // 等待一段时间后重试
      await this.sleep(checkInterval);
    }
  }

  // 批量监控所有任务
  async monitorAllTasks(): Promise<Map<string, TextToSpeechResult>> {
    const results = new Map<string, TextToSpeechResult>();
    const taskIds = Array.from(this.taskQueue.keys());

    console.log(`开始监控 ${taskIds.length} 个任务`);

    for (const taskId of taskIds) {
      try {
        const result = await this.monitorTaskProgress(taskId);
        results.set(taskId, result);
      } catch (error) {
        console.error(`任务 ${taskId} 监控失败:`, error);
      }
    }

    console.log(`任务监控完成，成功: ${results.size}/${taskIds.length}`);
    return results;
  }

  // 获取任务统计信息
  getTaskStats(): {
    total: number;
    completed: number;
    processing: number;
    failed: number;
  } {
    const stats = {
      total: this.taskQueue.size,
      completed: 0,
      processing: 0,
      failed: 0,
    };

    // 在实际应用中，这里可以根据任务状态进行统计
    return stats;
  }

  // 清理已完成的任务
  cleanupCompletedTasks(): void {
    const completedTaskIds: string[] = [];

    // 在实际应用中，这里可以清理已完成的任务记录
    console.log(`清理了 ${completedTaskIds.length} 个已完成任务`);
  }

  private sleep(ms: number): Promise<void> {
    return new Promise((resolve) => setTimeout(resolve, ms));
  }
}

// 使用示例
const monitor = new BatchTTSMonitor("batch-tts-bot-id");

// 添加任务到监控队列
monitor.addTask("task-001", "这是第一个语音合成任务");
monitor.addTask("task-002", "这是第二个语音合成任务");
monitor.addTask("task-003", "这是第三个语音合成任务");

// 监控单个任务
const result = await monitor.monitorTaskProgress("task-001");
console.log("任务结果:", result.audioUrl);

// 批量监控所有任务
const allResults = await monitor.monitorAllTasks();

// 查看任务统计
const stats = monitor.getTaskStats();
console.log("任务统计:", stats);

长文本语音合成

// 长文本语音合成系统
class LongTextTTSProcessor {
  private botId: string;
  private voiceType: number;

  constructor(botId: string, voiceType: number = 1) {
    this.botId = botId;
    this.voiceType = voiceType;
  }

  // 分段处理长文本
  async processLongText(
    longText: string,
    maxSegmentLength: number = 500
  ): Promise<string[]> {
    const segments = this.splitText(longText, maxSegmentLength);
    const audioUrls: string[] = [];

    console.log(`开始处理长文本，共 ${segments.length} 段`);

    for (const [index, segment] of segments.entries()) {
      console.log(`处理第 ${index + 1}/${segments.length} 段文本`);

      try {
        // 提交语音合成任务
        const result = await ai.bot.textToSpeech({
          botId: this.botId,
          voiceType: this.voiceType,
          text: segment,
        });

        // 等待任务完成
        const finalResult = await this.waitForTaskCompletion(result.taskId);
        audioUrls.push(finalResult.audioUrl);

        console.log(
          `✓ 第 ${index + 1} 段处理完成，时长: ${finalResult.duration}秒`
        );
      } catch (error) {
        console.error(`✗ 第 ${index + 1} 段处理失败`, error);
      }
    }

    console.log(`长文本处理完成，生成 ${audioUrls.length} 个音频文件`);
    return audioUrls;
  }

  // 等待任务完成
  private async waitForTaskCompletion(
    taskId: string,
    maxWaitTime: number = 600
  ): Promise<TextToSpeechResult> {
    const startTime = Date.now();
    const checkInterval = 3000; // 3秒检查一次

    while (true) {
      const result = await ai.bot.getTextToSpeechResult({
        botId: this.botId,
        taskId: taskId,
      });

      if (result.status === "completed") {
        return result;
      }

      if (result.status === "failed") {
        throw new Error(`语音合成任务失败: ${taskId}`);
      }

      // 检查超时
      if (Date.now() - startTime > maxWaitTime * 1000) {
        throw new Error(`语音合成任务超时: ${taskId}`);
      }

      console.log(`任务 ${taskId} 进度: ${result.progress}%`);
      await this.sleep(checkInterval);
    }
  }

  // 文本分段
  private splitText(text: string, maxLength: number): string[] {
    const segments: string[] = [];
    let currentSegment = "";

    const sentences = text.split(/[。！？.!?]/);

    for (const sentence of sentences) {
      const trimmedSentence = sentence.trim();
      if (!trimmedSentence) continue;

      if (currentSegment.length + trimmedSentence.length + 1 <= maxLength) {
        currentSegment += (currentSegment ? "。" : "") + trimmedSentence;
      } else {
        if (currentSegment) {
          segments.push(currentSegment + "。");
        }
        currentSegment = trimmedSentence;
      }
    }

    if (currentSegment) {
      segments.push(currentSegment + "。");
    }

    return segments;
  }

  // 合并音频文件（在实际应用中需要专门的音频处理工具）
  async mergeAudioFiles(audioUrls: string[]): Promise<string> {
    console.log(`开始合并 ${audioUrls.length} 个音频文件`);

    // 在实际应用中，这里可以调用音频合并API
    // 这里返回第一个音频URL作为示例
    return audioUrls[0];
  }

  // 生成有声读物
  async createAudioBook(
    bookTitle: string,
    chapters: string[]
  ): Promise<Map<string, string[]>> {
    const bookAudios = new Map<string, string[]>();

    console.log(`开始制作有声读物: ${bookTitle}，共 ${chapters.length} 章`);

    for (const [index, chapter] of chapters.entries()) {
      console.log(`处理第 ${index + 1} 章`);

      const chapterAudios = await this.processLongText(chapter);
      bookAudios.set(`第${index + 1}章`, chapterAudios);

      console.log(
        `✓ 第 ${index + 1} 章处理完成，生成 ${chapterAudios.length} 个音频片段`
      );
    }

    console.log(`有声读物制作完成: ${bookTitle}`);
    return bookAudios;
  }

  private sleep(ms: number): Promise<void> {
    return new Promise((resolve) => setTimeout(resolve, ms));
  }
}

// 使用示例
const longTextProcessor = new LongTextTTSProcessor("longtext-bot-id", 1);

// 处理长文本
const longText = "这是一段很长的文本内容，需要被分成多个段落进行语音合成...";
const audioSegments = await longTextProcessor.processLongText(longText);

// 制作有声读物
const bookChapters = [
  "第一章：故事的开端...",
  "第二章：情节发展...",
  "第三章：高潮部分...",
  "第四章：结局...",
];
const audioBook = await longTextProcessor.createAudioBook(
  "我的故事",
  bookChapters
);

// 合并音频文件
const mergedAudio = await longTextProcessor.mergeAudioFiles(audioSegments);
console.log("合并后的音频:", mergedAudio);

---

完整类型定义

IBotCreateConversation

interface IBotCreateConversation {
  botId: string;
  title?: string;
}

---

IBotGetConversation

interface IBotGetConversation {
  botId: string;
  pageSize?: number;
  pageNumber?: number;
  isDefault?: boolean;
}

---

IBotDeleteConversation

interface IBotDeleteConversation {
  botId: string;
  conversationId: string;
}

---

IBotSpeechToText

interface IBotSpeechToText {
  botId: string;
  engSerViceType: string;
  voiceFormat: string;
  url: string;
  isPreview?: boolean;
}

---

IBotTextToSpeech

interface IBotTextToSpeech {
  botId: string;
  voiceType: number;
  text: string;
  isPreview?: boolean;
}

---

IBotGetTextToSpeechResult

interface IBotGetTextToSpeechResult {
  botId: string;
  taskId: string;
  isPreview?: boolean;
}

---

BaseChatModelInput

interface BaseChatModelInput {
  model: string;
  messages: Array<ChatModelMessage>;
  temperature?: number;
  topP?: number;
  tools?: Array<FunctionTool>;
  toolChoice?: "none" | "auto" | "custom";
  maxSteps?: number;
  onStepFinish?: (prop: IOnStepFinish) => unknown;
}

BaseChatModelInput 属性名	类型	说明
model	string	模型名称。
messages	Array<ChatModelMessage>	消息列表。
temperature	number	采样温度，控制输出的随机性。
topP	number	温度采样，即模型考虑概率质量为 top_p 的标记的结果。
tools	Array<FunctionTool>	大模型可用的工具列表。
toolChoice	string	指定大模型选择工具的方式。
maxSteps	number	请求大模型的最大次数。
onStepFinish	(prop: IOnStepFinish) => unknown	当对大模型的一次请求完成时，出发的回调函数。

---

BotInfo

interface BotInfo {
  botId: string;
  name: string;
  introduction: string;
  agentSetting: string;
  welcomeMessage: string;
  avatar: string;
  background: string;
  tags: Array<string>;
  isNeedRecommend: boolean;
  knowledgeBase: Array<string>;
  type: string;
  initQuestions: Array<string>;
  enable: true;
}

---

IUserFeedback

interface IUserFeedback {
  recordId: string;
  type: string;
  botId: string;
  comment: string;
  rating: number;
  tags: Array<string>;
  input: string;
  aiAnswer: string;
}

---

ChatModelMessage

type ChatModelMessage =
  | UserMessage
  | SystemMessage
  | AssistantMessage
  | ToolMessage;

UserMessage

type UserMessage = {
  role: "user";
  content: string;
};

SystemMessage

type SystemMessage = {
  role: "system";
  content: string;
};

AssistantMessage

type AssistantMessage = {
  role: "assistant";
  content?: string;
  tool_calls?: Array<ToolCall>;
};

ToolMessage

type ToolMessage = {
  role: "tool";
  tool_call_id: string;
  content: string;
};

---

ToolCall

export type ToolCall = {
  id: string;
  type: string;
  function: { name: string; arguments: string };
};

---

FunctionTool

工具定义类型。

type FunctionTool = {
  name: string;
  description: string;
  fn: CallableFunction;
  parameters: object;
};

FunctionTool 属性名	类型	说明
name	string	工具名称。
description	string	工具的描述。清楚的工具描述有助于大模型认识工具的用途。
fn	CallableFunction	工具的执行函数。当 AI SDK 解析出大模型的响应需要该工具调用时，会调用此函数，并将结果返回给大模型。
parameters	object	工具执行函数的入参。需要使用 JSON Schema 的格式定义入参。

---

IOnStepFinish

大模型响应后出发的回调函数的入参类型。

interface IOnStepFinish {
  messages: Array<ChatModelMessage>;
  text?: string;
  toolCall?: ToolCall;
  toolResult?: unknown;
  finishReason?: string;
  stepUsage?: Usage;
  totalUsage?: Usage;
}

IOnStepFinish 属性名	类型	说明
messages	Array<ChatModelMessage>	到当前步骤为止所有的消息列表。
text	string	当前响应的文本。
toolCall	ToolCall	当前响应调用的工具。
toolResult	unknown	对应的工具调用结果。
finishReason	string	大模型推理结束的原因。
stepUsage	Usage	当前步骤所花费的 token。
totalUsage	Usage	到当前步骤为止所花费的总 token。

---

Usage

type Usage = {
  completion_tokens: number;
  prompt_tokens: number;
  total_tokens: number;
};

---

IConversation

Agent 会话。

interface IConversation {
  id: string;
  envId: string;
  ownerUin: string;
  userId: string;
  conversationId: string;
  title: string;
  startTime: string; // date-time format
  createTime: string;
  updateTime: string;
}