Commit ·
f63c37f
1
Parent(s): a374db7
fix: 无头模式下代理不生效
Browse files
src/services/browser_captcha.py
CHANGED
|
@@ -4,12 +4,37 @@
|
|
| 4 |
"""
|
| 5 |
import asyncio
|
| 6 |
import time
|
| 7 |
-
|
|
|
|
| 8 |
from playwright.async_api import async_playwright, Browser, BrowserContext
|
| 9 |
|
| 10 |
from ..core.logger import debug_logger
|
| 11 |
|
| 12 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
class BrowserCaptchaService:
|
| 14 |
"""浏览器自动化获取 reCAPTCHA token(单例模式)"""
|
| 15 |
|
|
@@ -60,9 +85,15 @@ class BrowserCaptchaService:
|
|
| 60 |
]
|
| 61 |
}
|
| 62 |
|
| 63 |
-
# 如果有代理,添加代理配置
|
| 64 |
if proxy_url:
|
| 65 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 66 |
|
| 67 |
self.browser = await self.playwright.chromium.launch(**launch_options)
|
| 68 |
self._initialized = True
|
|
|
|
| 4 |
"""
|
| 5 |
import asyncio
|
| 6 |
import time
|
| 7 |
+
import re
|
| 8 |
+
from typing import Optional, Dict
|
| 9 |
from playwright.async_api import async_playwright, Browser, BrowserContext
|
| 10 |
|
| 11 |
from ..core.logger import debug_logger
|
| 12 |
|
| 13 |
|
| 14 |
+
def parse_proxy_url(proxy_url: str) -> Optional[Dict[str, str]]:
|
| 15 |
+
"""解析代理URL,分离协议、主机、端口、认证信息
|
| 16 |
+
|
| 17 |
+
Args:
|
| 18 |
+
proxy_url: 代理URL,格式:protocol://[username:password@]host:port
|
| 19 |
+
|
| 20 |
+
Returns:
|
| 21 |
+
代理配置字典,包含server、username、password(如果有认证)
|
| 22 |
+
"""
|
| 23 |
+
proxy_pattern = r'^(socks5|http|https)://(?:([^:]+):([^@]+)@)?([^:]+):(\d+)$'
|
| 24 |
+
match = re.match(proxy_pattern, proxy_url)
|
| 25 |
+
|
| 26 |
+
if match:
|
| 27 |
+
protocol, username, password, host, port = match.groups()
|
| 28 |
+
proxy_config = {'server': f'{protocol}://{host}:{port}'}
|
| 29 |
+
|
| 30 |
+
if username and password:
|
| 31 |
+
proxy_config['username'] = username
|
| 32 |
+
proxy_config['password'] = password
|
| 33 |
+
|
| 34 |
+
return proxy_config
|
| 35 |
+
return None
|
| 36 |
+
|
| 37 |
+
|
| 38 |
class BrowserCaptchaService:
|
| 39 |
"""浏览器自动化获取 reCAPTCHA token(单例模式)"""
|
| 40 |
|
|
|
|
| 85 |
]
|
| 86 |
}
|
| 87 |
|
| 88 |
+
# 如果有代理,解析并添加代理配置
|
| 89 |
if proxy_url:
|
| 90 |
+
proxy_config = parse_proxy_url(proxy_url)
|
| 91 |
+
if proxy_config:
|
| 92 |
+
launch_options['proxy'] = proxy_config
|
| 93 |
+
auth_info = "auth=yes" if 'username' in proxy_config else "auth=no"
|
| 94 |
+
debug_logger.log_info(f"[BrowserCaptcha] 代理配置: {proxy_config['server']} ({auth_info})")
|
| 95 |
+
else:
|
| 96 |
+
debug_logger.log_warning(f"[BrowserCaptcha] 代理URL格式错误: {proxy_url}")
|
| 97 |
|
| 98 |
self.browser = await self.playwright.chromium.launch(**launch_options)
|
| 99 |
self._initialized = True
|