113 lines
3.4 KiB
Markdown
113 lines
3.4 KiB
Markdown
---
|
||
name: browser-chromium
|
||
version: 1.1.0
|
||
description: 基于 patchright + CloakBrowser 双引擎的反检测 Chromium 浏览器自动化
|
||
---
|
||
|
||
# browser-chromium (v1.1.0)
|
||
|
||
基于 patchright + CloakBrowser 双引擎的反检测浏览器,适合需要绕过检测的网站。
|
||
|
||
**新增 v1.1.0**: CloakBrowser C++级源码补丁引擎,可通过 `--engine cloak` 切换
|
||
|
||
## 双引擎架构
|
||
|
||
| 引擎 | 命令 | 原理 | 适用场景 |
|
||
|------|------|------|---------|
|
||
| **patchright** (默认) | `--engine patchright` | JS注入反检测,兼容Playwright API | 普通反爬站(Twitter/X等) |
|
||
| **CloakBrowser** | `--engine cloak` | C++源码级Chromium补丁,最强反检测 | Cloudflare/Turnstile/高防护站 |
|
||
| **auto** | `--engine auto` | 自动选择(检测到防护站自动切CloakBrowser) | 通用 |
|
||
|
||
## 与 browser-desktop 的区别
|
||
|
||
| 维度 | browser-desktop | browser-chromium |
|
||
|------|----------------|-----------------|
|
||
| 浏览器 | Firefox (Playwright) | Chromium (patchright + CloakBrowser) |
|
||
| 反检测 | 无 | 多层指纹伪装 |
|
||
| 适用场景 | 普通网站 | 反爬/反机器人检测/高防护站 |
|
||
| 速度 | 正常 | 正常 |
|
||
|
||
## 安装
|
||
|
||
```bash
|
||
pip install patchright --break-system-packages
|
||
patchright install chromium
|
||
pip install cloakbrowser --break-system-packages
|
||
pip install crawlee --break-system-packages # 指纹套件
|
||
```
|
||
|
||
## 使用
|
||
|
||
```bash
|
||
# 标准浏览 (patchright 引擎)
|
||
python3 /home/yh6788/.openclaw/skills/browser-chromium/scripts/browse.py "https://x.com"
|
||
|
||
# 截图
|
||
python3 /home/yh6788/.openclaw/skills/browser-chromium/scripts/browse.py "https://example.com" --screenshot
|
||
|
||
# 高防护站 (CloakBrowser 引擎)
|
||
python3 /home/yh6788/.openclaw/skills/browser-chromium/scripts/browse.py "https://cloudflare.com" --engine cloak
|
||
|
||
# 自动选择引擎
|
||
python3 /home/yh6788/.openclaw/skills/browser-chromium/scripts/browse.py "https://x.com" --engine auto
|
||
|
||
# 带代理
|
||
python3 /home/yh6788/.openclaw/skills/browser-chromium/scripts/browse.py "https://example.com" --proxy "http://192.168.0.66:7890"
|
||
|
||
# 检测指纹强度
|
||
python3 /home/yh6788/.openclaw/skills/browser-chromium/scripts/browse.py --check-fingerprint --engine patchright
|
||
```
|
||
|
||
## API(Python 脚本中引用)
|
||
|
||
```python
|
||
from patchright.sync_api import sync_playwright
|
||
|
||
with sync_playwright() as p:
|
||
browser = p.chromium.launch(headless=True)
|
||
page = browser.new_page()
|
||
page.goto("https://x.com")
|
||
page.screenshot()
|
||
browser.close()
|
||
```
|
||
|
||
## CloakBrowser 用法
|
||
|
||
```python
|
||
from cloakbrowser import CloakBrowser
|
||
|
||
cb = CloakBrowser(headless=True, proxy="http://192.168.0.66:7890")
|
||
page = cb.page
|
||
page.goto("https://cloudflare.com")
|
||
print(page.content())
|
||
```
|
||
|
||
## 引擎选择建议
|
||
|
||
```
|
||
简单站/登录站 → patchright (默认,Cookie持久化)
|
||
Cloudflare/Turnstile → cloak
|
||
未知站 → auto
|
||
```
|
||
|
||
## 持久化会话
|
||
|
||
Cookie 自动保存到 `~/.openclaw/browser-chromium-cookies.json`,下次打开自动复用登录状态。(CloakBrowser模式除外)
|
||
|
||
## 依赖生态
|
||
|
||
| 库 | 用途 | 状态 |
|
||
|----|------|------|
|
||
| patchright | 主引擎 | ✅ |
|
||
| playwright | patchright 依赖 | ✅ |
|
||
| cloakbrowser | C++级反检测 | ✅ |
|
||
| crawlee | fingerprint_suite | ✅ |
|
||
| parsel | crawlee 依赖 | ✅ |
|
||
| jaro-winkler | crawlee 依赖 | ✅ |
|
||
|
||
## 注意事项
|
||
|
||
- patchright 只支持 Chromium,Firefox/WebKit 不支持
|
||
- 如遇检测,优先尝试 `--engine cloak` 或加代理
|
||
- CloakBrowser 会话不持久化(每次新建)
|