#!/bin/bash

# 百度搜索任务 - 使用 browser-use CDP 方式（修复版）
# 此脚本会自动启动 Chrome 并执行搜索任务

echo "=========================================="
echo "🚀 百度搜索任务 (browser-use CDP 方式)"
echo "=========================================="
echo

# 设置变量
CHROME_PATH="/Applications/Google Chrome.app/Contents/MacOS/Google Chrome"
DEBUG_PORT=9222
USER_DATA_DIR="/tmp/chrome-debug-profile-baidu"

# 检查 Chrome 是否存在
if [ ! -f "$CHROME_PATH" ]; then
    echo "❌ 错误：Chrome 未找到"
    exit 1
fi

# 关闭现有的 Chrome 进程
echo "📌 步骤 1: 检查并关闭现有 Chrome 调试进程..."
pkill -f "remote-debugging-port=$DEBUG_PORT" 2>/dev/null
sleep 2
echo "✅ 已关闭现有进程"
echo

# 创建用户数据目录
echo "📌 步骤 2: 准备用户数据目录..."
mkdir -p "$USER_DATA_DIR"
echo "✅ 目录已创建: $USER_DATA_DIR"
echo

# 启动 Chrome
echo "📌 步骤 3: 启动 Chrome（远程调试模式）..."
"$CHROME_PATH" \
    --remote-debugging-port=$DEBUG_PORT \
    --user-data-dir="$USER_DATA_DIR" \
    --no-first-run \
    --no-default-browser-check \
    > /dev/null 2>&1 &

CHROME_PID=$!
echo "✅ Chrome 已启动 (PID: $CHROME_PID)"
echo

# 等待 Chrome 启动
echo "⏳ 等待 Chrome 启动完成..."
MAX_WAIT=30
WAITED=0
while [ $WAITED -lt $MAX_WAIT ]; do
    if curl -s http://localhost:$DEBUG_PORT/json/version > /dev/null 2>&1; then
        echo "✅ Chrome 调试端口已就绪"
        break
    fi
    sleep 1
    WAITED=$((WAITED + 1))
done
echo

if [ $WAITED -ge $MAX_WAIT ]; then
    echo "❌ 错误：Chrome 启动超时"
    kill $CHROME_PID 2>/dev/null
    exit 1
fi

# 验证 CDP
echo "📌 步骤 4: 验证 CDP 连接..."
CDP_INFO=$(curl -s http://localhost:$DEBUG_PORT/json/version)
if [ -n "$CDP_INFO" ]; then
    echo "✅ CDP 连接成功"
else
    echo "❌ 错误：CDP 连接失败"
    kill $CHROME_PID 2>/dev/null
    exit 1
fi
echo

# 运行 Python 脚本
echo "=========================================="
echo "🐍 运行 Python 脚本"
echo "=========================================="
echo

python3 << 'PYTHON_END'
import asyncio
import json
import sys
import os
from pathlib import Path
from datetime import datetime

# 添加项目路径
sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))

async def run_search():
    """执行搜索任务"""
    print("🚀 开始执行搜索任务...\n")

    # 项目根目录
    project_root = Path.cwd()

    try:
        # 导入 browser-use
        from browser_use import Tools
        from browser_use.browser import BrowserProfile, BrowserSession

        # 连接到 Chrome
        print("📌 连接到 Chrome (CDP: http://localhost:9222)...")
        browser_session = BrowserSession(
            browser_profile=BrowserProfile(
                cdp_url='http://localhost:9222',
                is_local=True
            )
        )

        # 重要：启动 browser session
        await browser_session.start()

        tools = Tools()
        print("✅ 已连接并启动 BrowserSession\n")

        # 导航到百度
        print("📌 导航到百度首页...")
        result = await tools.navigate(
            url="https://www.baidu.com",
            browser_session=browser_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"📌 搜索: {search_keyword}")
        result = await tools.navigate(
            url=search_url,
            browser_session=browser_session
        )
        print(f"✅ {result.long_term_memory}\n")
        await asyncio.sleep(3)

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

        # 提取数据
        print("📌 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=browser_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")

            # 保存数据
            json_file = project_root / "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")

            # 显示预览
            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')}")

        # 保存 HTML
        print("📌 保存页面 HTML...")
        cdp = await browser_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 = project_root / "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("🎉 任务完成！")
        print("="*60)
        print("生成文件:")
        print("  • baidu.json")
        print("  • baidu_page.html")
        print("="*60)

        # 停止 browser session
        await browser_session.stop()

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

asyncio.run(run_search())
PYTHON_END

PYTHON_EXIT_CODE=$?

echo
echo

# 询问是否关闭 Chrome
echo "=========================================="
echo "清理"
echo "=========================================="
echo
read -p "是否关闭 Chrome？(y/N): " -n 1 -r
echo
if [[ $REPLY =~ ^[Yy]$ ]]; then
    echo "🔄 正在关闭 Chrome..."
    kill $CHROME_PID 2>/dev/null
    sleep 2
    pkill -f "remote-debugging-port=$DEBUG_PORT" 2>/dev/null
    echo "✅ Chrome 已关闭"
else
    echo "💡 Chrome 保持运行状态 (PID: $CHROME_PID)"
    echo "   手动关闭命令: kill $CHROME_PID"
fi

echo
echo "=========================================="
echo "✨ 脚本执行完成"
echo "=========================================="

exit $PYTHON_EXIT_CODE