deepspider 0.2.11 → 0.2.12

package/src/agent/core/StreamHandler.js

@@ -0,0 +1,263 @@
+/**
+ * DeepSpider - 流式输出处理器
+ * 处理 Agent 的流式事件
+ */
+
+import { isApiServiceError, isToolSchemaError } from '../errors/ErrorClassifier.js';
+import { RetryManager, sleep } from './RetryManager.js';
+
+// DeepSeek 特殊标记清理
+const DSML_PATTERN = /｜DSML｜/g;
+function cleanDSML(text) {
+  return text ? text.replace(DSML_PATTERN, '') : text;
+}
+
+// 人工介入配置
+const INTERVENTION_CONFIG = {
+  idleTimeoutMs: 120000,  // 2分钟无响应触发提示
+  checkIntervalMs: 30000, // 30秒检测一次
+};
+
+export class StreamHandler {
+  constructor({ agent, config, panelBridge, riskTools = [], debug = () => {} }) {
+    this.agent = agent;
+    this.config = config;
+    this.panelBridge = panelBridge;
+    this.riskTools = riskTools;
+    this.debug = debug;
+    this.retryManager = new RetryManager();
+  }
+
+  /**
+   * 流式对话 - 显示思考过程（带重试）
+   */
+  async chatStream(input, retryCount = 0) {
+    let finalResponse = '';
+    let lastEventTime = Date.now();
+    let eventCount = 0;
+    let lastToolCall = null;
+
+    // 重置面板状态
+    this.panelBridge.reset();
+    await this.panelBridge.setBusy(true);
+
+    this.debug(`chatStream: 开始处理, 输入长度=${input.length}`);
+
+    // 心跳检测
+    let interventionNotified = false;
+    const heartbeat = this._createHeartbeat(
+      () => lastEventTime,
+      () => eventCount,
+      () => lastToolCall,
+      () => interventionNotified,
+      (v) => { interventionNotified = v; }
+    );
+
+    try {
+      this.debug('chatStream: 创建事件流');
+      const eventStream = await this.agent.streamEvents(
+        { messages: [{ role: 'user', content: input }] },
+        { ...this.config, version: 'v2' }
+      );
+
+      this.debug('chatStream: 开始遍历事件');
+      for await (const event of eventStream) {
+        lastEventTime = Date.now();
+        eventCount++;
+
+        if (event.event === 'on_tool_start') {
+          lastToolCall = event.name;
+        }
+
+        await this._handleStreamEvent(event);
+
+        if (event.event === 'on_chat_model_end' && event.name === 'ChatOpenAI') {
+          const output = event.data?.output;
+          if (output?.content) {
+            finalResponse = output.content;
+            this.debug(`chatStream: 收到最终响应, 长度=${finalResponse.length}`);
+          }
+        }
+      }
+
+      clearInterval(heartbeat);
+      console.log(`\n[完成] 共处理 ${eventCount} 个事件`);
+
+      await this.panelBridge.flushPanelText();
+      await this.panelBridge.finalizeMessage('assistant');
+      await this.panelBridge.setBusy(false);
+
+      this.debug(`chatStream: 完成, 响应长度=${finalResponse.length}`);
+      return finalResponse || '[无响应]';
+    } catch (error) {
+      clearInterval(heartbeat);
+      return this._handleError(error, input, eventCount, lastToolCall, retryCount);
+    }
+  }
+
+  /**
+   * 从检查点恢复流式对话
+   */
+  async chatStreamResume(retryCount = 0) {
+    let finalResponse = '';
+    let lastEventTime = Date.now();
+    let eventCount = 0;
+
+    await this.panelBridge.setBusy(true);
+    this.debug(`chatStreamResume: 从检查点恢复, retryCount=${retryCount}`);
+
+    const heartbeat = setInterval(() => {
+      const elapsed = Math.round((Date.now() - lastEventTime) / 1000);
+      if (elapsed > 30) {
+        console.log(`\n[心跳] 恢复中，已等待 ${elapsed}s`);
+      }
+    }, 30000);
+
+    try {
+      const eventStream = await this.agent.streamEvents(
+        { messages: [] },
+        { ...this.config, version: 'v2' }
+      );
+
+      for await (const event of eventStream) {
+        lastEventTime = Date.now();
+        eventCount++;
+        await this._handleStreamEvent(event);
+
+        if (event.event === 'on_chat_model_end' && event.name === 'ChatOpenAI') {
+          const output = event.data?.output;
+          if (output?.content) {
+            finalResponse = output.content;
+          }
+        }
+      }
+
+      clearInterval(heartbeat);
+      await this.panelBridge.flushPanelText();
+      await this.panelBridge.setBusy(false);
+      console.log(`\n[恢复完成] 共处理 ${eventCount} 个事件`);
+      return finalResponse || '[无响应]';
+    } catch (error) {
+      clearInterval(heartbeat);
+      await this.panelBridge.setBusy(false);
+      const errMsg = error.message || String(error);
+      console.error(`\n[恢复失败] ${errMsg}`);
+
+      if (isApiServiceError(errMsg) && this.retryManager.canRetry(retryCount)) {
+        const delay = this.retryManager.getDelay(retryCount);
+        console.log(`\n[重试 ${retryCount + 1}/${this.retryManager.maxRetries}] ${delay}ms 后再次恢复...`);
+        await sleep(delay);
+        return this.chatStreamResume(retryCount + 1);
+      }
+
+      return `恢复失败: ${errMsg}`;
+    }
+  }
+
+  /**
+   * 创建心跳检测定时器
+   */
+  _createHeartbeat(getLastEventTime, getEventCount, getLastToolCall, getNotified, setNotified) {
+    return setInterval(() => {
+      const elapsed = Math.round((Date.now() - getLastEventTime()) / 1000);
+      if (elapsed > 30) {
+        console.log(`\n[心跳] 已等待 ${elapsed}s, 事件数=${getEventCount()}, 最后工具=${getLastToolCall() || '无'}`);
+      }
+
+      const isRiskTool = getLastToolCall() && this.riskTools.includes(getLastToolCall());
+      if (elapsed * 1000 > INTERVENTION_CONFIG.idleTimeoutMs && !getNotified() && isRiskTool) {
+        setNotified(true);
+        const msg = '⚠️ 页面操作后长时间无响应，可能遇到验证码或风控，请检查浏览器';
+        console.log('\n[提示] ' + msg);
+        this.panelBridge.sendToPanel('system', msg).catch(() => {});
+      }
+    }, INTERVENTION_CONFIG.checkIntervalMs);
+  }
+
+  /**
+   * 处理流式事件
+   */
+  async _handleStreamEvent(event) {
+    const { event: eventType, name, data } = event;
+
+    // 过滤内部事件
+    if (name?.startsWith('ChannelWrite') ||
+        name?.startsWith('Branch') ||
+        name?.includes('Middleware') ||
+        name === 'RunnableSequence' ||
+        name === 'model_request' ||
+        name === 'tools') {
+      return;
+    }
+
+    this.debug(`handleStreamEvent: ${eventType}, name=${name}`);
+
+    switch (eventType) {
+      case 'on_chat_model_stream':
+        let chunk = data?.chunk?.content;
+        if (chunk && typeof chunk === 'string') {
+          chunk = cleanDSML(chunk);
+          process.stdout.write(chunk);
+          await this.panelBridge.appendToPanel(chunk);
+        }
+        break;
+
+      case 'on_tool_start':
+        this.debug('handleStreamEvent: 工具开始，先刷新缓冲区');
+        await this.panelBridge.flushPanelText();
+        this.panelBridge.hasStartedAssistantMsg = false;
+        const input = data?.input || {};
+        const inputStr = typeof input === 'string' ? input : JSON.stringify(input);
+        const preview = inputStr.length > 100 ? inputStr.slice(0, 100) + '...' : inputStr;
+        console.log(`\n[调用] ${name}(${preview})`);
+        await this.panelBridge.sendToPanel('system', `[调用] ${name}`);
+        break;
+
+      case 'on_tool_end':
+        const output = data?.output;
+        let result = '';
+        this.debug(`on_tool_end: name=${name}, output type=${typeof output}`);
+
+        if (typeof output === 'string') {
+          result = output.slice(0, 80);
+        } else if (output?.content) {
+          result = String(output.content).slice(0, 80);
+        }
+        if (result) {
+          console.log(`[结果] ${result}${result.length >= 80 ? '...' : ''}`);
+          await this.panelBridge.sendToPanel('system', `[结果] ${result.slice(0, 50)}${result.length > 50 ? '...' : ''}`);
+        }
+        break;
+    }
+  }
+
+  /**
+   * 处理错误和重试
+   */
+  async _handleError(error, input, eventCount, lastToolCall, retryCount) {
+    const errMsg = error.message || String(error);
+    await this.panelBridge.setBusy(false);
+    console.error(`\n[异常] 事件数=${eventCount}, 最后工具=${lastToolCall || '无'}, 错误: ${errMsg}`);
+
+    if (this.retryManager.canRetry(retryCount)) {
+      if (isApiServiceError(errMsg)) {
+        const delay = this.retryManager.getDelay(retryCount);
+        console.log(`\n[重试 ${retryCount + 1}/${this.retryManager.maxRetries}] API错误，${delay}ms 后从检查点恢复...`);
+        await this.panelBridge.sendToPanel('system',
+          `服务暂时不可用，${Math.round(delay/1000)}s 后重试 (${retryCount + 1}/${this.retryManager.maxRetries})`);
+        await sleep(delay);
+        return this.chatStreamResume(retryCount + 1);
+      }
+
+      if (isToolSchemaError(errMsg)) {
+        console.log(`\n[重试 ${retryCount + 1}/${this.retryManager.maxRetries}] 工具参数错误，发送修正请求...`);
+        await this.panelBridge.sendToPanel('system',
+          `工具调用失败，正在修正 (${retryCount + 1}/${this.retryManager.maxRetries})`);
+        const resumeInput = `工具调用失败: ${errMsg}\n请检查参数格式并重试。`;
+        return this.chatStream(resumeInput, retryCount + 1);
+      }
+    }
+
+    return `错误: ${errMsg}`;
+  }
+}

package/src/agent/core/index.js

@@ -0,0 +1,7 @@
+/**
+ * DeepSpider - Core 模块索引
+ */
+
+export { StreamHandler } from './StreamHandler.js';
+export { RetryManager, sleep } from './RetryManager.js';
+export { PanelBridge } from './PanelBridge.js';

package/src/agent/errors/ErrorClassifier.js

@@ -0,0 +1,43 @@
+/**
+ * DeepSpider - 错误分类器
+ * 判断错误类型，决定重试策略
+ */
+
+/**
+ * 判断是否为工具参数错误（需要 LLM 修正）
+ */
+export function isToolSchemaError(errMsg) {
+  return /did not match expected schema|Invalid input|tool input/i.test(errMsg);
+}
+
+/**
+ * 判断是否为 API 服务错误（可直接重试）
+ */
+export function isApiServiceError(errMsg) {
+  return /503|502|429|rate limit|无可用渠道|timeout|ECONNRESET|ETIMEDOUT/i.test(errMsg);
+}
+
+/**
+ * 判断是否为浏览器错误
+ */
+export function isBrowserError(errMsg) {
+  return /Target closed|page closed|context closed|browser disconnected/i.test(errMsg);
+}
+
+/**
+ * 判断是否为网络错误
+ */
+export function isNetworkError(errMsg) {
+  return /ENOTFOUND|ECONNREFUSED|network|fetch failed/i.test(errMsg);
+}
+
+/**
+ * 获取错误类型
+ */
+export function classifyError(errMsg) {
+  if (isToolSchemaError(errMsg)) return 'tool_schema';
+  if (isApiServiceError(errMsg)) return 'api_service';
+  if (isBrowserError(errMsg)) return 'browser';
+  if (isNetworkError(errMsg)) return 'network';
+  return 'unknown';
+}

package/src/agent/errors/SpiderError.js

@@ -0,0 +1,68 @@
+/**
+ * DeepSpider - 结构化错误类型
+ */
+
+/**
+ * 基础错误类
+ */
+export class SpiderError extends Error {
+  constructor(message, code, details = {}) {
+    super(message);
+    this.name = 'SpiderError';
+    this.code = code;
+    this.details = details;
+  }
+
+  toJSON() {
+    return {
+      name: this.name,
+      code: this.code,
+      message: this.message,
+      details: this.details,
+    };
+  }
+}
+
+/**
+ * API 服务错误 - 503/502/429 等
+ */
+export class ApiServiceError extends SpiderError {
+  constructor(message, details = {}) {
+    super(message, 'API_SERVICE_ERROR', details);
+    this.name = 'ApiServiceError';
+    this.retryable = true;
+  }
+}
+
+/**
+ * 工具参数错误 - 工具调用参数不匹配
+ */
+export class ToolSchemaError extends SpiderError {
+  constructor(message, details = {}) {
+    super(message, 'TOOL_SCHEMA_ERROR', details);
+    this.name = 'ToolSchemaError';
+    this.retryable = true;
+  }
+}
+
+/**
+ * 浏览器错误 - 页面关闭、上下文丢失等
+ */
+export class BrowserError extends SpiderError {
+  constructor(message, details = {}) {
+    super(message, 'BROWSER_ERROR', details);
+    this.name = 'BrowserError';
+    this.retryable = false;
+  }
+}
+
+/**
+ * 网络错误 - 连接失败、DNS 解析等
+ */
+export class NetworkError extends SpiderError {
+  constructor(message, details = {}) {
+    super(message, 'NETWORK_ERROR', details);
+    this.name = 'NetworkError';
+    this.retryable = true;
+  }
+}

package/src/agent/errors/index.js

@@ -0,0 +1,19 @@
+/**
+ * DeepSpider - Errors 模块索引
+ */
+
+export {
+  isToolSchemaError,
+  isApiServiceError,
+  isBrowserError,
+  isNetworkError,
+  classifyError,
+} from './ErrorClassifier.js';
+
+export {
+  SpiderError,
+  ApiServiceError,
+  ToolSchemaError,
+  BrowserError,
+  NetworkError,
+} from './SpiderError.js';