Agent/test_browser_config.py

#!/usr/bin/env python3
"""
使用 Browser + BrowserConfig 方式测试 browser-use
参考 GitHub issue #1520 的解决方案
"""

import asyncio
import json
from pathlib import Path
from datetime import datetime


async def test_browser_config():
    """使用 BrowserConfig 测试"""
    print("="*60)
    print("🧪 测试 browser-use (BrowserConfig 方式)")
    print("="*60)
    print()

    from browser_use import Browser, BrowserConfig
    from browser_use.browser.context import BrowserContextConfig

    try:
        # 创建上下文配置
        print("📌 步骤 1: 创建配置...")
        context_cfg = BrowserContextConfig(
            disable_security=True,
            minimum_wait_page_load_time=0.5,
            wait_for_network_idle_page_load_time=0.5,
        )

        # 创建浏览器配置
        browser = Browser(
            config=BrowserConfig(
                chrome_instance_path='/Applications/Google Chrome.app/Contents/MacOS/Google Chrome',
                new_context_config=context_cfg,
                headless=False  # 非无头模式，便于调试
            )
        )
        print("✅ 配置已创建\n")

        # 获取浏览器会话
        print("📌 步骤 2: 获取浏览器会话...")
        session = await browser.get_session()
        print("✅ 浏览器会话已获取\n")

        # 导入工具
        from browser_use import Tools
        tools = Tools()

        # 导航到百度
        print("📌 步骤 3: 导航到百度...")
        result = await tools.navigate(
            url="https://www.baidu.com",
            browser_session=session
        )
        print(f"✅ {result.long_term_memory}\n")

        await asyncio.sleep(2)

        # 搜索
        search_keyword = "Python 教程"
        search_url = f"https://www.baidu.com/s?wd={search_keyword}"

        print(f"📌 步骤 4: 搜索 '{search_keyword}'...")
        result = await tools.navigate(
            url=search_url,
            browser_session=session
        )
        print(f"✅ {result.long_term_memory}\n")

        await asyncio.sleep(3)

        # 滚动页面
        print("📌 步骤 5: 滚动页面...")
        await tools.scroll(
            down=True,
            pages=1.0,
            browser_session=session
        )
        await asyncio.sleep(2)
        print("✅ 页面滚动完成\n")

        # 提取数据
        print("📌 步骤 6: 提取搜索结果...")
        extract_js = """
        (function(){
            try {
                const results = [];
                const resultItems = document.querySelectorAll('#content_left > div[class*="result"]');

                resultItems.forEach((item, index) => {
                    if (index >= 10) return;

                    const titleEl = item.querySelector('h3 a, .t a');
                    const title = titleEl ? titleEl.textContent.trim() : '';
                    const link = titleEl ? titleEl.href : '';
                    const summaryEl = item.querySelector('.c-abstract, .content-right_8Zs40');
                    const summary = summaryEl ? summaryEl.textContent.trim() : '';
                    const sourceEl = item.querySelector('.c-color-gray, .source_1Vdff');
                    const source = sourceEl ? sourceEl.textContent.trim() : '';

                    if (title || link) {
                        results.push({
                            index: index + 1,
                            title: title,
                            link: link,
                            summary: summary.substring(0, 200),
                            source: source
                        });
                    }
                });

                return {
                    success: true,
                    count: results.length,
                    keyword: 'Python 教程',
                    timestamp: new Date().toISOString(),
                    results: results
                };
            } catch (e) {
                return {
                    success: false,
                    error: e.message
                };
            }
        })()
        """

        result = await tools.evaluate(
            code=extract_js,
            browser_session=session
        )

        # 解析结果
        output = result.extracted_content or str(result.metadata)
        if isinstance(output, str) and output.startswith("Result: "):
            output = output[8:]

        data = json.loads(output) if isinstance(output, str) else output

        if data.get('success'):
            print(f"✅ 成功提取 {data.get('count', 0)} 条结果\n")

            # 保存数据
            json_file = Path("baidu.json")
            with open(json_file, 'w', encoding='utf-8') as f:
                json.dump(data, f, ensure_ascii=False, indent=2)
            print(f"✅ 数据已保存: {json_file}\n")

            # 显示前3条结果
            if data.get('results'):
                print("📋 前3条结果:")
                for item in data['results'][:3]:
                    print(f"  {item.get('index')}. {item.get('title', '无标题')[:50]}...")
                print()
        else:
            print(f"⚠️  提取失败: {data.get('error')}\n")

        # 保存 HTML
        print("📌 步骤 7: 保存页面 HTML...")
        cdp = await session.get_or_create_cdp_session()
        html_result = await cdp.cdp_client.send.Runtime.evaluate(
            params={'expression': 'document.documentElement.outerHTML'},
            session_id=cdp.session_id
        )
        html_content = html_result.get('result', {}).get('value', '')

        html_file = Path("baidu_page.html")
        with open(html_file, 'w', encoding='utf-8') as f:
            f.write(f"<!-- 保存时间: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')} -->\n")
            f.write(html_content)

        print(f"✅ HTML 已保存: {html_file}")
        print(f"   大小: {len(html_content):,} 字符\n")

        print("="*60)
        print("🎉 测试成功！")
        print("="*60)
        print("✅ browser-use 使用 BrowserConfig 方式正常工作")
        print("✅ 生成文件:")
        print("   • baidu.json")
        print("   • baidu_page.html")
        print("="*60)

        # 关闭浏览器
        await browser.close()

    except Exception as e:
        print(f"\n❌ 错误: {e}")
        import traceback
        traceback.print_exc()


if __name__ == "__main__":
    asyncio.run(test_browser_config())