yiyan-browser-agent 1.0.22 → 1.0.24

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "yiyan-browser-agent",
-  "version": "1.0.22",
+  "version": "1.0.24",
   "description": "AI coding agent powered by Yiyan (文心一言) via browser automation — no API key needed",
   "main": "src/index.js",
   "bin": {

package/src/browser-manager.js

@@ -10,35 +10,47 @@ const config       = require('./config');
 
 const CDP_PORT_FILE = path.join(os.homedir(), '.yiyan-agent', 'cdp-port.json');
 const CDP_PORT = 9222;
+const SESSION_DIR = path.join(os.homedir(), '.yiyan-agent', 'session');
 
 class BrowserManager {
   /**
    * Get browser - connect existing or launch new
    */
   static async getInstance() {
+    // Ensure directories exist
+    fs.mkdirSync(SESSION_DIR, { recursive: true });
+
     // Try connect existing browser
     try {
       if (fs.existsSync(CDP_PORT_FILE)) {
-        const portInfo = JSON.parse(fs.readFileSync(CDP_PORT_FILE, 'utf8'));
-        const browserURL = `http://localhost:${portInfo.port || CDP_PORT}`;
+        const endpointInfo = JSON.parse(fs.readFileSync(CDP_PORT_FILE, 'utf8'));
+        const wsEndpoint = endpointInfo.wsEndpoint;
 
-        logger.info('Connecting to existing browser...');
-        const browser = await chromium.connectOverCDP(browserURL, { timeout: 5000 });
+        if (!wsEndpoint) {
+          throw new Error('No wsEndpoint in file');
+        }
 
-        // Create NEW page (tab) for this task
-        const page = await browser.newPage();
+        logger.info('Connecting to existing browser...');
+        const browser = await chromium.connect({ wsEndpoint, timeout: 5000 });
+
+        // Create NEW context and page (tab) for this task
+        const context = await browser.newContext({
+          viewport: { width: 1280, height: 900 },
+          userAgent: 'Mozilla/5.0 AppleWebKit/537.36 Chrome/124.0.0.0 Safari/537.36',
+        });
+        const page = await context.newPage();
         await page.goto(config.YIYAN_URL, { waitUntil: 'domcontentloaded', timeout: 15000 });
         await page.waitForTimeout(500);
 
-        logger.success('Connected! Using new tab.');
-        return { browser, page, isNew: false };
+        logger.success('Connected! New tab opened.');
+        return { browser, context, page, isNew: false };
       }
     } catch (err) {
       logger.warn('Connection failed, launching new browser...');
       try { fs.unlinkSync(CDP_PORT_FILE); } catch {}
     }
 
-    // Launch new browser with CDP
+    // Launch new browser with CDP port
     return await this._launchNew();
   }
 
@@ -46,53 +58,47 @@ class BrowserManager {
    * Launch new browser with CDP port
    */
   static async _launchNew() {
-    logger.info('Launching new browser...');
-
-    // Ensure session directory
-    const sessionDir = path.join(os.homedir(), '.yiyan-agent', 'session');
-    fs.mkdirSync(sessionDir, { recursive: true });
+    logger.info('Launching new browser server...');
 
-    // Launch with persistent context + CDP port
-    const context = await chromium.launchPersistentContext(sessionDir, {
+    // Use launchServer to create independent browser process
+    // Browser will NOT close when Node.js process exits
+    const browserServer = await chromium.launchServer({
       headless: false,
-      viewport: { width: 1280, height: 900 },
       args: [
-        `--remote-debugging-port=${CDP_PORT}`,
         '--disable-blink-features=AutomationControlled',
         '--no-first-run',
         '--no-sandbox',
       ],
-      userAgent: 'Mozilla/5.0 AppleWebKit/537.36 Chrome/124.0.0.0 Safari/537.36',
     });
 
-    // Get the browser from context
-    const browser = context.browser();
+    // Get the WebSocket endpoint
+    const wsEndpoint = browserServer.wsEndpoint();
+    const port = wsEndpoint.match(/:(\d+)/)?.[1] || CDP_PORT;
 
-    // Save CDP port
+    // Save endpoint info
     fs.writeFileSync(CDP_PORT_FILE, JSON.stringify({
-      port: CDP_PORT,
-      pid: process.pid,
+      wsEndpoint: wsEndpoint,
+      port: port,
       launchedAt: Date.now()
     }));
 
-    // Use existing page or create new
-    const pages = context.pages();
-    const page = pages.length > 0 ? pages[0] : await context.newPage();
+    logger.success('Browser server started!');
+
+    // Connect to our own server
+    const browser = await chromium.connect({ wsEndpoint });
+
+    // Create context and page
+    const context = await browser.newContext({
+      viewport: { width: 1280, height: 900 },
+      userAgent: 'Mozilla/5.0 AppleWebKit/537.36 Chrome/124.0.0.0 Safari/537.36',
+    });
+    const page = await context.newPage();
 
     await page.goto(config.YIYAN_URL, { waitUntil: 'domcontentloaded', timeout: 15000 });
     await page.waitForTimeout(800);
 
-    logger.success('Browser launched! Tab ready.');
-    return { browser, context, page, isNew: true };
-  }
-
-  /**
-   * Close current page (tab) but keep browser running
-   */
-  static async closePage(page) {
-    try {
-      await page.close();
-    } catch {}
+    logger.success('Tab ready.');
+    return { browser, context, page, isNew: true, browserServer };
   }
 
   /**
@@ -100,7 +106,7 @@ class BrowserManager {
    */
   static async forceClose() {
     try { fs.unlinkSync(CDP_PORT_FILE); } catch {}
-    // Browser will close when user does Ctrl+C or manually
+    // Browser continues running independently
   }
 }