fix: 避免复用浏览器时跨域 Cookie 过长导致请求失败

连接已有 Chrome 会把整个浏览器上下文的 cookie 带入平台 client。
除 xhs 外，多数平台仍直接读取全量 cookies，导致请求头过长并放大跨域污染。
本次将各平台的 cookie 读取统一收口到平台域名，并补上基础回归测试。

Constraint: 必须继续复用用户真实浏览器里的平台登录态
Rejected: 仅修复 xhs | 其他平台在连接已有浏览器时仍会携带超长 Cookie
Confidence: high
Scope-risk: moderate
Reversibility: clean
Directive: 后续新增平台或调整 update_cookies 和 create client 流程时，只按平台域名读取 cookies
Tested: uv run pytest test/test_utils.py; python3 -m compileall tools/crawler_util.py media_platform/douyin/core.py media_platform/douyin/client.py media_platform/kuaishou/core.py media_platform/kuaishou/client.py media_platform/bilibili/core.py media_platform/bilibili/client.py media_platform/zhihu/core.py media_platform/zhihu/client.py media_platform/tieba/core.py media_platform/tieba/client.py media_platform/xhs/core.py media_platform/xhs/client.py media_platform/weibo/core.py media_platform/weibo/client.py test/test_utils.py
Not-tested: 各平台在真实 CDP 浏览器连接下的端到端抓取流程

tools/crawler_util.py

@@ -34,7 +34,7 @@ from typing import Dict, List, Optional, Tuple, cast
 
 import httpx
 from PIL import Image, ImageDraw, ImageShow
-from playwright.async_api import Cookie, Page
+from playwright.async_api import BrowserContext, Cookie, Page
 
 from . import utils
 from .httpx_util import make_async_client
@@ -145,6 +145,17 @@ def convert_cookies(cookies: Optional[List[Cookie]]) -> Tuple[str, Dict]:
     return cookies_str, cookie_dict
 
 
+async def convert_browser_context_cookies(
+    browser_context: BrowserContext, urls: Optional[List[str]] = None
+) -> Tuple[str, Dict]:
+    cookies = (
+        await browser_context.cookies(urls=urls)
+        if urls
+        else await browser_context.cookies()
+    )
+    return convert_cookies(cookies)
+
+
 def convert_str_cookie_to_dict(cookie_str: str) -> Dict:
     cookie_dict: Dict[str, str] = dict()
     if not cookie_str: