playwright-automation

$npx mdskill add aAAaqwq/AGI-Super-Team/playwright-automation

Automate browser tasks without human intervention.

  • Executes crawling, data collection, and UI testing in headless mode.
  • Integrates with Bash and Python Playwright libraries.
  • Decides execution based on cron scheduling and resource constraints.
  • Delivers results via logs and generated screenshots or PDFs.

SKILL.md

.github/skills/playwright-automationView on GitHub ↗
---
name: playwright-automation
description: Playwright 浏览器自动化。用于自动化爬虫、数据采集、表单填写、UI 测试等需要浏览器自动化的场景。无需人工干预,适合 cron 定时任务。
allowed-tools: Bash, Exec, Read, Write
---

# Playwright 浏览器自动化

## 概述

Playwright 是一个强大的浏览器自动化工具,可以模拟真实用户操作,支持:
- 无头浏览器模式(后台运行)
- 数据采集和爬虫
- 表单自动填写
- UI 自动化测试
- 截图和 PDF 生成

## 为什么需要 Playwright

### ⚠️ 资源清理原则(强制)

**所有涉及浏览器的 cron 任务完成后,必须自动关闭 Chrome/Chromium 进程!**

```python
# 任务结束时必须执行
import subprocess

# 方式1: 显式关闭浏览器
await browser.close()
await context.close()

# 方式2: 强制清理残留进程(推荐在脚本结束时调用)
subprocess.run(['pkill', '-f', 'chrome'], capture_output=True)
subprocess.run(['pkill', '-f', 'chromium'], capture_output=True)
```

**原因**: 避免内存泄漏和资源占用,防止 Gateway CPU 100% 过载

### 与 browser tool 的区别

| 特性 | browser tool | Playwright |
|------|-------------|-----------|
| 需要用户参与 | ✅ 需要手动打开浏览器 | ❌ 完全自动 |
| 适合定时任务 | ❌ | ✅ |
| 后台运行 | ❌ | ✅ |
| 调试友好 | ✅ 可视化操作 | ⚠️ 需要日志 |
| 无需安装 | ✅ 已集成 | ❌ 需要安装 |

### 使用场景

**使用 Playwright**:
- ✅ 定时监控(cron 任务)
- ✅ 大规模数据采集
- ✅ 无人值守运行
- ✅ 生产环境部署

**使用 browser tool**:
- ✅ 交互式调试
- ✅ 需要人工决策的操作
- ✅ 一次性任务
- ✅ 绕过复杂验证码

## 快速开始

### 1. 安装 Playwright

```bash
# 安装 Python 包
pip install playwright

# 安装浏览器(Chromium)
playwright install chromium

# 验证安装
python3 -c "from playwright.sync_api import sync_playwright; print('✅ 安装成功')"
```

### 2. 基本使用

#### 同步 API(简单任务)

```python
from playwright.sync_api import sync_playwright

with sync_playwright() as p:
    browser = p.chromium.launch(headless=True)  # headless=False 显示浏览器
    page = browser.new_page()
    page.goto('https://example.com')
    print(page.title())
    browser.close()
```

#### 异步 API(推荐,性能更好)

```python
import asyncio
from playwright.async_api import async_playwright

async def main():
    async with async_playwright() as p:
        browser = await p.chromium.launch(headless=True)
        page = await browser.new_page()
        await page.goto('https://example.com')
        title = await page.title()
        print(title)
        await browser.close()

asyncio.run(main())
```

## 常用功能

### 1. 页面导航

```python
# 等待加载完成
await page.goto('https://example.com', wait_until='domcontentloaded')

# 等待选项:
# - 'load' - 页面完全加载
# - 'domcontentloaded' - DOM 加载完成(更快)
# - 'networkidle' - 网络空闲(最慢但最稳)
```

### 2. 元素定位

```python
# CSS 选择器
await page.click('button.submit')
await page.fill('input[name="username"]', 'myuser')

# XPath
await page.click('xpath=//button[@type="submit"]')

# 文本选择器
await page.click('text=登录')

# 组合选择器
await page.click('div.login-form >> text=登录')
```

### 3. 提取数据

```python
# 获取文本
text = await page.text_content('h1.title')

# 获取属性
href = await page.get_attribute('a.link', 'href')

# 获取多个元素
items = await page.query_selector_all('div.item')
for item in items:
    text = await item.text_content()
    print(text)

# 执行 JavaScript
result = await page.evaluate('() => document.title')

# 获取整个 HTML
html = await page.content()
```

### 4. 表单操作

```python
# 填写表单
await page.fill('input[name="username"]', 'myuser')
await page.fill('input[name="password"]', 'mypass')
await page.click('button[type="submit"]')

# 下拉选择
await page.select_option('select#country', 'China')

# 复选框
await page.check('input#agree')

# 上传文件
await page.set_input_files('input[type="file"]', 'path/to/file.pdf')
```

### 5. 等待策略

```python
# 等待元素出现
await page.wait_for_selector('div.result', timeout=5000)

# 等待导航
async with page.expect_navigation():
    await page.click('a.link')

# 等待特定条件
await page.wait_for_function('() => document.title.includes("加载完成")')

# 固定延迟
import asyncio
await asyncio.sleep(2)  # 等待 2 秒
```

### 6. 滚动和交互

```python
# 滚动到页面底部
await page.evaluate('window.scrollTo(0, document.body.scrollHeight)')

# 滚动到元素
await page.locator('div.footer').scroll_into_view_if_needed()

# 鼠标悬停
await page.hover('div.menu')

# 拖拽
await page.drag_and_drop('div.draggable', 'div.dropzone')
```

### 7. 截图和 PDF

```python
# 截图
await page.screenshot(path='screenshot.png')

# 全页截图
await page.screenshot(path='full.png', full_page=True)

# 元素截图
await page.locator('div.content').screenshot(path='element.png')

# 生成 PDF
await page.pdf(path='page.pdf', format='A4')
```

### 8. 处理弹窗

```python
# 接受 alert
async with page.expect_event('dialog') as dialog_info:
    await page.click('button')
dialog = await dialog_info.value
await dialog.accept()

# 输入 prompt
async with page.expect_event('dialog') as dialog_info:
    await page.click('button')
dialog = await dialog_info.value
await dialog.accept('my input')
```

## 反爬策略

### 1. User-Agent 轮换

```python
user_agents = [
    'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36',
    'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36',
    'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36'
]

browser = await p.chromium.launch(
    user_agent=random.choice(user_agents)
)
```

### 2. 随机延迟

```python
import random
import asyncio

# 在操作之间添加随机延迟
await asyncio.sleep(random.uniform(2, 5))
```

### 3. Cookie 保存和加载

```python
# 第一次登录后保存 Cookie
context = await browser.new_context()
await page.goto('https://example.com/login')
# ... 登录操作 ...
await context.storage_state(path='cookies.json')

# 后续使用保存的 Cookie
context = await browser.new_context(storage_state='cookies.json')
```

### 4. 代理设置

```python
browser = await p.chromium.launch(
    proxy={
        'server': 'http://proxy.example.com:8080',
        'username': 'user',
        'password': 'pass'
    }
)
```

### 5. 浏览器上下文隔离

```python
# 创建独立的上下文(相当于无痕模式)
context = await browser.new_context(
    viewport={'width': 1920, 'height': 1080},
    user_agent='Custom UA',
    locale='zh-CN'
)
page = await context.new_page()
```

## 调试技巧

### 1. 显示浏览器

```python
# 开启有头模式,可以看到操作过程
browser = await p.chromium.launch(headless=False, slow_mo=1000)
# slow_mo=1000 会在每个操作间延迟 1 秒
```

### 2. 截图调试

```python
# 在关键步骤截图
await page.goto('https://example.com')
await page.screenshot(path='step1.png')

await page.click('button')
await page.screenshot(path='step2.png')
```

### 3. 查看日志

```python
# 监听控制台消息
page.on('console', lambda msg: print(f'Console: {msg.text}'))

# 监听网络请求
page.on('request', lambda request: print(f'Request: {request.url}'))
page.on('response', lambda response: print(f'Response: {response.status}'))
```

### 4. Playwright Inspector

```bash
# 启动 Inspector 模式
PWDEBUG=1 python3 your_script.py
```

## 常见问题

### Q: 如何处理验证码?

**A**: 几种方案
1. 使用打码平台(超级鹰、若快打码)
2. 手动处理:暂停等待用户输入
3. 降级:用 browser tool 让用户手动操作

```python
# 方案 2: 手动处理
input("遇到验证码,请在浏览器中完成,然后按回车继续...")
await asyncio.sleep(2)  # 等待验证通过
```

### Q: 元素找不到怎么办?

**A**: 检查以下几点
1. 是否在 iframe 中(需要切换)
2. 是否动态加载(需要等待)
3. 选择器是否正确

```python
# 切换到 iframe
frame = page.frame('iframe-id')
await frame.click('button')

# 等待动态加载
await page.wait_for_selector('div.dynamic-content')
```

### Q: 如何提高性能?

**A**:
1. 使用异步 API
2. 并发多个页面
3. 减少不必要的等待

```python
# 并发多个页面
async def fetch(url):
    page = await browser.new_page()
    await page.goto(url)
    # ...

tasks = [fetch(url) for url in urls]
await asyncio.gather(*tasks)
```

## 与 OpenClaw 集成

### 在 healthcare-monitor 中的使用

```python
# scraper_free.py
from playwright.async_api import async_playwright

async with async_playwright() as p:
    browser = await p.chromium.launch(headless=True)
    context = await browser.new_context(
        user_agent="Mozilla/5.0 ..."
    )
    page = await context.new_page()
    await page.goto(url)
    # ... 采集数据 ...
    await browser.close()
```

### 与 browser tool 配合

1. **Playwright** → 日常自动监控
2. **browser tool** → 调试和异常处理

## 资源

- 官方文档: https://playwright.dev/python/
- API 参考: https://playwright.dev/python/docs/api/class-playwright
- 示例代码: `~/clawd/skills/playwright-automation/examples/`

## 快速命令

```bash
# 安装
pip install playwright && playwright install chromium

# 运行脚本
python3 script.py

# 调试模式
PWDEBUG=1 python3 script.py

# 查看版本
playwright --version
```

---

**记住**: Playwright 让你的自动化任务完全无人值守!🚀

More from aAAaqwq/AGI-Super-Team

SkillDescription
a-fund-monitor监控 A 股基金实时估值与盘后净值,自动判断交易日并生成提醒或分析。
add-leadAdd company/person/relationship to CRM
ads-agentAI-агент для управления Facebook рекламой. Вызывай для анализа, оптимизации, создания кампаний и отчётов.
afrexai-compliance-auditRun internal compliance audits against major governance and security
afrexai-personal-financeComplete personal finance system — budgeting, debt payoff, investing, tax optimization, net worth tracking, and financial independence planning. Use when managing money, building wealth, paying off debt, planning retirement, or optimizing taxes. Zero dependencies.
after-salesUse when managing post-purchase experience, building customer loyalty, or increasing repeat purchases
agent-contactsAI agent contacts — add, list, remove MCP contacts. Use when someone gives an agent URL, or when you need to view/remove contacts.
agent-model-switcher批量查看和切换子 agent 的模型配置,用于统一调整多 agent 的 provider/model 设置。
agent-patternsFormat SPAWN REQUEST messages to launch parallel agents, generate structured agent status reports, and define communication protocols within the sprint system. Use when the user needs to coordinate multiple agents, format spawn requests, produce agent reports, or establish inter-agent communication patterns.
agent-task-confirmUse when confirming whether a dispatched agent task was actually received, activated, and progressing after sessions_send or other task handoff actions.