feat: 浏览器控制工具全面增强 — browser_open(wait_selector)/screenshot(full_page+selector)/extract(selector+max_chars)/click(wait滚动到视图)/type(clear+submit)/scroll(selector)/新增browser_wait

src/main/browser.ts

@@ -1,5 +1,5 @@
 /**
- * Browser - Agent 浏览器控制 (v5.0)
+ * Browser - Agent 浏览器控制 (v5.1 增强版)
  * 通过隐藏 BrowserWindow 实现网页加载、截图、JS 执行
  */
 
@@ -52,10 +52,25 @@ function getAgentBrowser(): BrowserWindow {
 /** 页面加载超时（毫秒） */
 const PAGE_LOAD_TIMEOUT = 30_000;
 
+/** 等待 CSS 选择器出现在页面中（最多 waitMs 毫秒） */
+async function waitForSelector(win: BrowserWindow, selector: string, waitMs = 10_000): Promise<boolean> {
+  const start = Date.now();
+  while (Date.now() - start < waitMs) {
+    try {
+      const found = await win.webContents.executeJavaScript(`
+        (() => { const el = document.querySelector(${JSON.stringify(selector)}); return el !== null; })()
+      `, true);
+      if (found) return true;
+    } catch { /* retry */ }
+    await new Promise(r => setTimeout(r, 300));
+  }
+  return false;
+}
+
 /** 打开 URL */
-export async function browserOpen(url: string): Promise<{ success: boolean; title?: string; url?: string; error?: string }> {
+export async function browserOpen(url: string, waitSelector?: string): Promise<{ success: boolean; title?: string; url?: string; waited?: boolean; error?: string }> {
   try {
-    // 如果已有 BrowserWindow 且加载了不同 URL，先关闭重建，避免状态残留
+    // 如果已有 BrowserWindow 且加载了不同 URL，先关闭重建
     if (agentBrowser && !agentBrowser.isDestroyed()) {
       const currentUrl = agentBrowser.webContents.getURL();
       if (currentUrl !== 'about:blank' && currentUrl !== url) {
@@ -68,7 +83,6 @@ export async function browserOpen(url: string): Promise<{ success: boolean; titl
     const win = getAgentBrowser();
     sendLog('info', `🌐 browser 打开`, url.slice(0, 100));
 
-    // 等待页面实际加载完成，而非固定延时
     const loadPromise = new Promise<void>((resolve, reject) => {
       const timeout = setTimeout(() => {
         win.webContents.removeListener('did-finish-load', onFinish);
@@ -88,9 +102,18 @@ export async function browserOpen(url: string): Promise<{ success: boolean; titl
     });
 
     await loadPromise;
+
+    let waited = false;
+    // 等待特定元素出现（用于 SPA 页面）
+    if (waitSelector) {
+      sendLog('info', `🌐 browser 等待元素`, waitSelector);
+      waited = await waitForSelector(win, waitSelector);
+      if (!waited) sendLog('warn', `🌐 browser 等待超时`, `未找到: ${waitSelector}`);
+    }
+
     const title = win.webContents.getTitle();
-    sendLog('success', `🌐 browser 已加载`, title);
-    return { success: true, title, url: win.webContents.getURL() };
+    sendLog('success', `🌐 browser 已加载`, title + (waited ? ' [等待元素完成]' : ''));
+    return { success: true, title, url: win.webContents.getURL(), waited };
   } catch (err) {
     sendLog('error', `🌐 browser 打开失败`, (err as Error).message);
     return { success: false, error: (err as Error).message };
@@ -98,15 +121,73 @@ export async function browserOpen(url: string): Promise<{ success: boolean; titl
 }
 
 /** 截图 */
-export async function browserScreenshot(): Promise<{ success: boolean; png?: string; error?: string }> {
+export async function browserScreenshot(params: { full_page?: boolean; selector?: string } = {}): Promise<{ success: boolean; png?: string; width?: number; height?: number; error?: string }> {
   try {
     const ready = ensureBrowserReady();
     if (!ready.ok) return { success: false, error: ready.error };
     const win = ready.win!;
+
+    if (params.selector) {
+      // 截取特定元素的截图
+      const rect = await win.webContents.executeJavaScript(`
+        (() => {
+          const el = document.querySelector(${JSON.stringify(params.selector)});
+          if (!el) return null;
+          const r = el.getBoundingClientRect();
+          return { x: r.x, y: r.y, width: r.width, height: r.height };
+        })()
+      `, true) as { x: number; y: number; width: number; height: number } | null;
+
+      if (!rect || rect.width === 0 || rect.height === 0) {
+        return { success: false, error: `未找到元素或元素不可见: ${params.selector}` };
+      }
+
+      const image = await win.webContents.capturePage({ x: Math.round(rect.x), y: Math.round(rect.y), width: Math.round(rect.width), height: Math.round(rect.height) });
+      const buffer = image.toPNG();
+      sendLog('info', `🌐 browser 截图(元素)`, `${params.selector} → ${buffer.length} bytes, ${rect.width}x${rect.height}`);
+      return { success: true, png: buffer.toString('base64'), width: Math.round(rect.width), height: Math.round(rect.height) };
+    }
+
+    if (params.full_page) {
+      // 全页面截图：先滚动获取完整高度，截完恢复
+      const pageSize = await win.webContents.executeJavaScript(`
+        (() => {
+          const w = Math.max(document.documentElement.scrollWidth, document.body.scrollWidth, window.innerWidth);
+          const h = Math.max(document.documentElement.scrollHeight, document.body.scrollHeight, window.innerHeight);
+          const origScroll = { x: window.scrollX, y: window.scrollY };
+          window.scrollTo(0, 0);
+          return { width: w, height: h, origScroll };
+        })()
+      `, true) as { width: number; height: number; origScroll: { x: number; y: number } };
+
+      // 分块截图拼接（超过窗口高度的页面）
+      const viewHeight = 800;
+      const chunks: Buffer[] = [];
+      for (let y = 0; y < pageSize.height; y += viewHeight) {
+        await win.webContents.executeJavaScript(`window.scrollTo(0, ${y})`, true);
+        await new Promise(r => setTimeout(r, 200));
+        const image = await win.webContents.capturePage({
+          x: 0, y: 0,
+          width: pageSize.width,
+          height: Math.min(viewHeight, pageSize.height - y)
+        });
+        chunks.push(image.toPNG());
+      }
+      // 恢复滚动位置
+      await win.webContents.executeJavaScript(`window.scrollTo(${pageSize.origScroll.x}, ${pageSize.origScroll.y})`, true);
+
+      // 拼接所有截图（简单合并 base64）
+      const fullPng = Buffer.concat(chunks).toString('base64');
+      sendLog('success', `🌐 browser 截图(全页)`, `${pageSize.width}x${pageSize.height}, ${fullPng.length} chars`);
+      return { success: true, png: fullPng, width: pageSize.width, height: pageSize.height };
+    }
+
+    // 视口截图（默认）
     const image = await win.webContents.capturePage();
     const buffer = image.toPNG();
-    sendLog('info', `🌐 browser 截图`, `${buffer.length} bytes`);
-    return { success: true, png: buffer.toString('base64') };
+    const size = image.getSize();
+    sendLog('info', `🌐 browser 截图(视口)`, `${buffer.length} bytes, ${size.width}x${size.height}`);
+    return { success: true, png: buffer.toString('base64'), width: size.width, height: size.height };
   } catch (err) {
     sendLog('error', `🌐 browser 截图失败`, (err as Error).message);
     return { success: false, error: (err as Error).message };
@@ -130,12 +211,13 @@ export async function browserEvaluate(js: string): Promise<{ success: boolean; r
   }
 }
 
-/** 提取页面文本和链接 */
-export async function browserExtract(): Promise<{ success: boolean; title?: string; text?: string; links?: Array<{text: string; href: string}>; error?: string }> {
+/** 提取页面内容 */
+export async function browserExtract(params: { selector?: string; max_chars?: number } = {}): Promise<{ success: boolean; title?: string; text?: string; links?: Array<{text: string; href: string}>; error?: string }> {
   try {
     const ready = ensureBrowserReady();
     if (!ready.ok) return { success: false, error: ready.error };
     const win = ready.win!;
+    const maxChars = params.max_chars || 15000;
 
     const result = await win.webContents.executeJavaScript(`
       (() => {
@@ -143,24 +225,35 @@ export async function browserExtract(): Promise<{ success: boolean; title?: stri
         const getText = (el) => {
           if (!el) return '';
           const clone = el.cloneNode(true);
-          clone.querySelectorAll('script,style,noscript,iframe').forEach(e => e.remove());
+          clone.querySelectorAll('script,style,noscript,iframe,svg').forEach(e => e.remove());
           return clone.innerText?.trim() || '';
         };
-        const body = getText(document.body);
-        const text = body.length > 15000 ? body.slice(0, 15000) + '\\\\n... (已截断)' : body;
+
+        let root = document.body;
+        ${params.selector ? `
+        const selEl = document.querySelector(${JSON.stringify(params.selector)});
+        if (selEl) root = selEl;
+        ` : ''}
+
+        const body = getText(root);
+        const maxC = ${maxChars};
+        const text = body.length > maxC ? body.slice(0, maxC) + '\\\\n... (已截断)' : body;
+
         const links = [];
-        document.querySelectorAll('a[href]').forEach(a => {
+        root.querySelectorAll('a[href]').forEach(a => {
           const href = a.href;
-          const text = a.textContent?.trim();
-          if (href && text && href.startsWith('http')) {
-            links.push({ text: text.slice(0, 100), href });
+          const t = a.textContent?.trim();
+          if (href && t && href.startsWith('http')) {
+            links.push({ text: t.slice(0, 100), href });
           }
         });
+
         return { title, text, links: links.slice(0, 50) };
       })()
     `, true);
 
-    sendLog('success', `🌐 browser 提取完成`, `标题: ${result.title}, 文本: ${result.text?.length || 0}字, 链接: ${result.links?.length || 0}个`);
+    const extractType = params.selector ? `selector:${params.selector}` : '全文';
+    sendLog('success', `🌐 browser 提取完成(${extractType})`, `标题: ${result.title}, 文本: ${result.text?.length || 0}字, 链接: ${result.links?.length || 0}个`);
     return { success: true, ...result };
   } catch (err) {
     sendLog('error', `🌐 browser 提取失败`, (err as Error).message);
@@ -169,17 +262,25 @@ export async function browserExtract(): Promise<{ success: boolean; title?: stri
 }
 
 /** 点击元素 */
-export async function browserClick(selector: string): Promise<{ success: boolean; error?: string }> {
+export async function browserClick(selector: string, wait = false): Promise<{ success: boolean; error?: string }> {
   try {
     const ready = ensureBrowserReady();
     if (!ready.ok) return { success: false, error: ready.error };
     const win = ready.win!;
+
+    if (wait) {
+      const found = await waitForSelector(win, selector);
+      if (!found) return { success: false, error: `等待超时: 未找到元素 ${selector}` };
+      sendLog('info', `🌐 browser 等待完成`, selector);
+    }
+
     sendLog('info', `🌐 browser 点击`, selector);
     await win.webContents.executeJavaScript(`
       (() => {
         const sel = ${JSON.stringify(selector)};
         const el = document.querySelector(sel);
         if (!el) throw new Error('元素未找到: ' + sel);
+        el.scrollIntoView({ behavior: 'smooth', block: 'center' });
         el.click();
       })()
     `, true);
@@ -193,7 +294,7 @@ export async function browserClick(selector: string): Promise<{ success: boolean
 }
 
 /** 输入文本 */
-export async function browserType(selector: string, text: string): Promise<{ success: boolean; error?: string }> {
+export async function browserType(selector: string, text: string, clear = true, submit = false): Promise<{ success: boolean; error?: string }> {
   try {
     const ready = ensureBrowserReady();
     if (!ready.ok) return { success: false, error: ready.error };
@@ -206,12 +307,21 @@ export async function browserType(selector: string, text: string): Promise<{ suc
         const el = document.querySelector(sel);
         if (!el) throw new Error('元素未找到: ' + sel);
         el.focus();
-        el.value = val;
+        ${clear ? `el.value = val;` : `el.value += val;`}
         el.dispatchEvent(new Event('input', { bubbles: true }));
         el.dispatchEvent(new Event('change', { bubbles: true }));
+        ${submit ? `
+        const form = el.closest('form');
+        if (form) {
+          form.dispatchEvent(new Event('submit', { bubbles: true, cancelable: true }));
+        } else {
+          el.dispatchEvent(new KeyboardEvent('keydown', { key: 'Enter', code: 'Enter', keyCode: 13, bubbles: true }));
+        }
+        ` : ''}
       })()
     `, true);
-    sendLog('success', `🌐 browser 输入完成`, selector);
+    await new Promise(r => setTimeout(r, submit ? 1000 : 300));
+    sendLog('success', `🌐 browser 输入完成`, selector + (submit ? ' [已提交]' : ''));
     return { success: true };
   } catch (err) {
     sendLog('error', `🌐 browser 输入失败`, `${selector}: ${(err as Error).message}`);
@@ -220,11 +330,29 @@ export async function browserType(selector: string, text: string): Promise<{ suc
 }
 
 /** 滚动页面 */
-export async function browserScroll(direction: 'down' | 'up' | 'top' | 'bottom'): Promise<{ success: boolean; error?: string }> {
+export async function browserScroll(params: { direction?: string; selector?: string } = {}): Promise<{ success: boolean; error?: string }> {
   try {
     const ready = ensureBrowserReady();
     if (!ready.ok) return { success: false, error: ready.error };
     const win = ready.win!;
+    const direction = params.direction || 'down';
+
+    if (params.selector) {
+      // 滚动到特定元素
+      const found = await win.webContents.executeJavaScript(`
+        (() => {
+          const el = document.querySelector(${JSON.stringify(params.selector)});
+          if (!el) return false;
+          el.scrollIntoView({ behavior: 'smooth', block: 'center' });
+          return true;
+        })()
+      `, true);
+      if (!found) return { success: false, error: `未找到元素: ${params.selector}` };
+      await new Promise(r => setTimeout(r, 500));
+      sendLog('info', `🌐 browser 滚动到元素`, params.selector);
+      return { success: true };
+    }
+
     const js = direction === 'down' ? 'window.scrollBy(0, 500)'
       : direction === 'up' ? 'window.scrollBy(0, -500)'
       : direction === 'top' ? 'window.scrollTo(0, 0)'
@@ -239,6 +367,38 @@ export async function browserScroll(direction: 'down' | 'up' | 'top' | 'bottom')
   }
 }
 
+/** 等待条件 */
+export async function browserWait(params: { selector?: string; time_ms?: number }): Promise<{ success: boolean; found?: boolean; error?: string }> {
+  try {
+    const ready = ensureBrowserReady();
+    if (!ready.ok) return { success: false, error: ready.error };
+    const win = ready.win!;
+
+    if (params.selector) {
+      sendLog('info', `🌐 browser 等待元素`, params.selector);
+      const found = await waitForSelector(win, params.selector, params.time_ms || 10_000);
+      if (found) {
+        sendLog('success', `🌐 browser 元素已出现`, params.selector);
+      } else {
+        sendLog('warn', `🌐 browser 等待超时`, params.selector);
+      }
+      return { success: true, found };
+    }
+
+    if (params.time_ms) {
+      sendLog('info', `🌐 browser 等待`, `${params.time_ms}ms`);
+      await new Promise(r => setTimeout(r, params.time_ms));
+      return { success: true };
+    }
+
+    // 默认等待 1 秒
+    await new Promise(r => setTimeout(r, 1000));
+    return { success: true };
+  } catch (err) {
+    return { success: false, error: (err as Error).message };
+  }
+}
+
 /** 关闭浏览器 */
 export function browserClose(): void {
   if (agentBrowser && !agentBrowser.isDestroyed()) {

src/main/ipc.ts

@@ -48,7 +48,7 @@ import {
   handleGit,
   handleCompress
 } from './tool-handlers.js';
-import { browserOpen, browserScreenshot, browserEvaluate, browserExtract, browserClick, browserType, browserScroll, browserClose } from './browser.js';
+import { browserOpen, browserScreenshot, browserEvaluate, browserExtract, browserClick, browserType, browserScroll, browserClose, browserWait } from './browser.js';
 import { startServer, stopServer, stopAllServers, callTool, getAllTools, getServerStatuses, refreshTools, setMCPTimeout } from './mcp-manager.js';
 import { getAllowedDirs, getBlockedDirs, setAllowedDirs } from './tool-security.js';
 import { startProcess, killProcess, getWorkspaceDir, setWorkspaceDir, listWorkspaceDir } from './workspace.js';
@@ -190,14 +190,15 @@ export async function setupIPC(): Promise<void> {
         case 'read_multiple_files':result = await handleReadMultipleFiles(args as { paths: string[]; max_chars_per_file?: number }); break;
         case 'git':                result = await handleGit(args as { action: string; path?: string; files?: string[]; message?: string; branch?: string; remote?: string; remote_url?: string; count?: number; all?: boolean; staged?: boolean; new_branch?: boolean; delete_branch?: boolean; force?: boolean; url?: string }); break;
         case 'compress':           result = await handleCompress(args as { action: string; path: string; destination?: string; format?: string }); break;
-        // v5.0 Browser 控制
-        case 'browser_open':       result = await browserOpen(args.url as string); break;
-        case 'browser_screenshot': result = await browserScreenshot(); break;
+        // v5.1 Browser 控制（增强版）
+        case 'browser_open':       result = await browserOpen(args.url as string, args.wait_selector as string | undefined); break;
+        case 'browser_screenshot': result = await browserScreenshot({ full_page: args.full_page as boolean, selector: args.selector as string }); break;
         case 'browser_evaluate':   result = await browserEvaluate(args.js as string); break;
-        case 'browser_extract':    result = await browserExtract(); break;
-        case 'browser_click':      result = await browserClick(args.selector as string); break;
-        case 'browser_type':       result = await browserType(args.selector as string, args.text as string); break;
-        case 'browser_scroll':     result = await browserScroll((args.direction as 'down' | 'up' | 'top' | 'bottom') || 'down'); break;
+        case 'browser_extract':    result = await browserExtract({ selector: args.selector as string, max_chars: args.max_chars as number }); break;
+        case 'browser_click':      result = await browserClick(args.selector as string, (args.wait as boolean) || false); break;
+        case 'browser_type':       result = await browserType(args.selector as string, args.text as string, args.clear !== false, args.submit as boolean || false); break;
+        case 'browser_scroll':     result = await browserScroll({ direction: args.direction as string, selector: args.selector as string }); break;
+        case 'browser_wait':       result = await browserWait({ selector: args.selector as string, time_ms: args.time_ms as number }); break;
         case 'browser_close':      browserClose(); result = { success: true }; break;
         default:                   return { success: false, error: `未知工具: ${toolName}` };
       }

src/renderer/services/tool-registry.ts

@@ -508,16 +508,19 @@ export const TOOL_DEFINITIONS: ToolDefinition[] = [
   // ══════════════════════════════════════════════
   // v5.0 新增工具：浏览器控制
   // ══════════════════════════════════════════════
+  // v5.1 Browser 控制工具（增强版）
+  // ══════════════════════════════════════════════
   {
     type: 'function',
     function: {
       name: 'browser_open',
-      description: 'Open a URL in the agent browser. Loads the page and waits for rendering. Returns the page title and final URL.',
+      description: 'Open a URL in the agent browser. Waits for page load. Use wait_selector to wait for a specific element (useful for SPA pages). Returns page title and URL.',
       parameters: {
         type: 'object',
         required: ['url'],
         properties: {
-          url: { type: 'string', description: 'The URL to open (http/https).' }
+          url: { type: 'string', description: 'The URL to open (http/https).' },
+          wait_selector: { type: 'string', description: 'CSS selector to wait for after page load. Useful for SPA/dynamic pages. Default: none.' }
         }
       }
     }
@@ -526,15 +529,21 @@ export const TOOL_DEFINITIONS: ToolDefinition[] = [
     type: 'function',
     function: {
       name: 'browser_screenshot',
-      description: 'Take a screenshot of the current browser page. Returns the image as base64 PNG. Use this to see what the page looks like.',
-      parameters: { type: 'object', properties: {} }
+      description: 'Take a screenshot of the current page. Supports viewport (default), full_page (entire scrollable page), or selector (specific element only). Returns base64 PNG.',
+      parameters: {
+        type: 'object',
+        properties: {
+          full_page: { type: 'boolean', description: 'Capture the entire scrollable page. Default: false.' },
+          selector: { type: 'string', description: 'CSS selector of element to capture. Overrides full_page.' }
+        }
+      }
     }
   },
   {
     type: 'function',
     function: {
       name: 'browser_evaluate',
-      description: 'Execute JavaScript code in the browser page context. Use to interact with the page, read DOM elements, fill forms, click buttons, etc. Returns the result of the last expression.',
+      description: 'Execute JavaScript in the browser page. Returns the result as JSON string. Use to read DOM, extract data, or interact programmatically.',
       parameters: {
         type: 'object',
         required: ['js'],
@@ -548,20 +557,27 @@ export const TOOL_DEFINITIONS: ToolDefinition[] = [
     type: 'function',
     function: {
       name: 'browser_extract',
-      description: 'Extract all text content and links from the current browser page. Returns title, full text, and up to 50 links.',
-      parameters: { type: 'object', properties: {} }
+      description: 'Extract text and links from the current page. Use selector to extract only from a specific element (e.g. "main", "#content", "article"). Without selector, extracts full page. Returns title, text, and up to 50 links.',
+      parameters: {
+        type: 'object',
+        properties: {
+          selector: { type: 'string', description: 'CSS selector to extract from. Default: entire body.' },
+          max_chars: { type: 'integer', description: 'Max characters to return. Default: 15000.' }
+        }
+      }
     }
   },
   {
     type: 'function',
     function: {
       name: 'browser_click',
-      description: 'Click an element on the page using a CSS selector.',
+      description: 'Click an element by CSS selector. Use wait=true to wait up to 10s for the element to appear first (useful after page navigation). Scrolls element into view before clicking.',
       parameters: {
         type: 'object',
         required: ['selector'],
         properties: {
-          selector: { type: 'string', description: 'CSS selector for the element to click.' }
+          selector: { type: 'string', description: 'CSS selector for the element to click.' },
+          wait: { type: 'boolean', description: 'Wait for element to appear before clicking (max 10s). Default: false.' }
         }
       }
     }
@@ -570,13 +586,15 @@ export const TOOL_DEFINITIONS: ToolDefinition[] = [
     type: 'function',
     function: {
       name: 'browser_type',
-      description: 'Type text into an input field on the page using a CSS selector.',
+      description: 'Type text into an input field. Use clear=false to append instead of replacing. Use submit=true to press Enter or submit the parent form after typing.',
       parameters: {
         type: 'object',
         required: ['selector', 'text'],
         properties: {
           selector: { type: 'string', description: 'CSS selector for the input element.' },
-          text: { type: 'string', description: 'Text to type into the input.' }
+          text: { type: 'string', description: 'Text to type into the input.' },
+          clear: { type: 'boolean', description: 'Clear existing text first. Default: true.' },
+          submit: { type: 'boolean', description: 'Submit the form or press Enter after typing. Default: false.' }
         }
       }
     }
@@ -585,11 +603,26 @@ export const TOOL_DEFINITIONS: ToolDefinition[] = [
     type: 'function',
     function: {
       name: 'browser_scroll',
-      description: 'Scroll the browser page.',
+      description: 'Scroll the page. Use direction for up/down/top/bottom, or selector to scroll a specific element into view.',
       parameters: {
         type: 'object',
         properties: {
-          direction: { type: 'string', enum: ['down', 'up', 'top', 'bottom'], description: 'Scroll direction. Default: down.' }
+          direction: { type: 'string', enum: ['down', 'up', 'top', 'bottom'], description: 'Scroll direction. Default: down.' },
+          selector: { type: 'string', description: 'CSS selector to scroll into view (overrides direction).' }
+        }
+      }
+    }
+  },
+  {
+    type: 'function',
+    function: {
+      name: 'browser_wait',
+      description: 'Wait for a condition. Use selector to wait for an element to appear, or time_ms for a fixed delay. Default: 1 second.',
+      parameters: {
+        type: 'object',
+        properties: {
+          selector: { type: 'string', description: 'CSS selector to wait for. Returns found: true/false.' },
+          time_ms: { type: 'integer', description: 'Fixed wait time in milliseconds. Default: 1000.' }
         }
       }
     }
@@ -598,7 +631,7 @@ export const TOOL_DEFINITIONS: ToolDefinition[] = [
     type: 'function',
     function: {
       name: 'browser_close',
-      description: 'Close the agent browser and release resources.',
+      description: 'Close the agent browser and free resources.',
       parameters: { type: 'object', properties: {} }
     }
   }