frist

2026-01-26 15:04:02 +08:00
commit 4813449f9c
31 changed files with 8439 additions and 0 deletions
--- a/core/challenge.py
+++ b/core/challenge.py
@@ -0,0 +1,274 @@
+"""
+Cloudflare Turnstile 验证码解决器
+
+Cloudflare Turnstile 是一种新型验证码系统，用于替代传统的 reCAPTCHA
+当触发时会返回 403 状态码并显示 "Just a moment" 页面
+
+⚠️ 本模块提供预留接口，用户根据需要配置解决方案
+"""
+
+from typing import Optional, Dict, Any
+from utils.logger import logger
+
+
+class CloudflareSolver:
+    """
+    Cloudflare Turnstile 验证码解决器
+
+    ⚠️ 预留接口 - 用户根据实际情况选择是否实现
+
+    可能的解决方案:
+    1. 使用高质量住宅代理（推荐，成本较低）
+    2. 集成打码平台（如 2captcha, capsolver）
+    3. 使用浏览器自动化 + undetected-chromedriver
+    4. 等待一段时间后重试（部分情况有效）
+    """
+
+    # Turnstile 相关常量
+    TURNSTILE_SITE_KEY = "0x4AAAAAAADnPIDROrmt1Wwj"  # OpenAI 的 Turnstile site key（需要从实际页面提取）
+
+    @staticmethod
+    def detect_challenge(response) -> bool:
+        """
+        检测响应是否为 Cloudflare Turnstile 挑战
+
+        参数:
+            response: HTTP 响应对象（来自 requests 或 curl_cffi）
+
+        返回:
+            True 如果检测到 Cloudflare 挑战，否则 False
+
+        检测特征:
+        - 状态码 403
+        - 响应体包含 "Just a moment", "Checking your browser" 等文本
+        - 包含 Cloudflare 相关 JavaScript
+        """
+        if response.status_code != 403:
+            return False
+
+        body = response.text.lower()
+        cloudflare_keywords = [
+            "just a moment",
+            "checking your browser",
+            "cloudflare",
+            "cf-challenge",
+            "turnstile",
+            "ray id"
+        ]
+
+        detected = any(keyword in body for keyword in cloudflare_keywords)
+
+        if detected:
+            logger.warning("Cloudflare Turnstile challenge detected")
+            # 尝试提取 Ray ID（用于调试）
+            ray_id = CloudflareSolver._extract_ray_id(response.text)
+            if ray_id:
+                logger.info(f"Cloudflare Ray ID: {ray_id}")
+
+        return detected
+
+    @staticmethod
+    async def solve(session, target_url: str, **kwargs) -> Optional[str]:
+        """
+        解决 Cloudflare Turnstile 挑战
+
+        ⚠️ 预留接口 - 用户需要根据实际需求实现
+
+        参数:
+            session: OAISession 实例
+            target_url: 触发挑战的目标 URL
+            **kwargs: 其他可能需要的参数（如 site_key, action 等）
+
+        返回:
+            cf_clearance Cookie 值 或 Turnstile response token
+
+        抛出:
+            NotImplementedError: 用户需要实现此方法
+
+        集成示例:
+        ```python
+        # 方案 1: 使用 2captcha 打码平台
+        from twocaptcha import TwoCaptcha
+        solver = TwoCaptcha('YOUR_API_KEY')
+        result = solver.turnstile(
+            sitekey=CloudflareSolver.TURNSTILE_SITE_KEY,
+            url=target_url
+        )
+        return result['code']
+
+        # 方案 2: 使用 capsolver
+        import capsolver
+        capsolver.api_key = "YOUR_API_KEY"
+        solution = capsolver.solve({
+            "type": "AntiTurnstileTaskProxyLess",
+            "websiteURL": target_url,
+            "websiteKey": CloudflareSolver.TURNSTILE_SITE_KEY,
+        })
+        return solution['token']
+
+        # 方案 3: 使用浏览器自动化
+        from selenium import webdriver
+        from undetected_chromedriver import Chrome
+        driver = Chrome()
+        driver.get(target_url)
+        # 等待 Cloudflare 自动通过...
+        cf_clearance = driver.get_cookie('cf_clearance')['value']
+        return cf_clearance
+        ```
+        """
+        logger.warning(
+            f"Cloudflare challenge detected at {target_url}, but solver not configured"
+        )
+
+        raise NotImplementedError(
+            "❌ Cloudflare solver not implemented.\n\n"
+            "This is OPTIONAL. Only implement if you encounter frequent 403 errors.\n\n"
+            "Recommended solutions:\n"
+            "1. Use high-quality residential proxies (easiest)\n"
+            "2. Integrate captcha solving service (2captcha, capsolver)\n"
+            "3. Use browser automation (undetected-chromedriver)\n"
+            "4. Retry with different proxy/IP\n\n"
+            f"Target URL: {target_url}\n"
+            f"Site Key: {CloudflareSolver.TURNSTILE_SITE_KEY}\n\n"
+            "Example implementation location: core/challenge.py -> solve()"
+        )
+
+    @staticmethod
+    def _extract_ray_id(html: str) -> Optional[str]:
+        """
+        从 Cloudflare 错误页面提取 Ray ID（用于调试）
+
+        Ray ID 格式示例: 84a1b2c3d4e5f678-LAX
+
+        参数:
+            html: Cloudflare 错误页面的 HTML 内容
+
+        返回:
+            Ray ID 字符串，未找到则返回 None
+        """
+        import re
+        match = re.search(r'Ray ID: ([a-f0-9-]+)', html, re.IGNORECASE)
+        if match:
+            return match.group(1)
+
+        # 尝试其他格式
+        match = re.search(r'ray id[:\s]+([a-f0-9-]+)', html, re.IGNORECASE)
+        if match:
+            return match.group(1)
+
+        return None
+
+    @staticmethod
+    def should_retry(response) -> bool:
+        """
+        判断是否应该重试请求（针对 Cloudflare 挑战）
+
+        某些情况下，简单地等待几秒后重试即可通过
+
+        参数:
+            response: HTTP 响应对象
+
+        返回:
+            True 如果建议重试，否则 False
+        """
+        if not CloudflareSolver.detect_challenge(response):
+            return False
+
+        # 如果是轻量级挑战（JavaScript challenge），重试可能有效
+        # 如果是 Turnstile 验证码，重试无效，需要解决验证码
+        body = response.text.lower()
+
+        # JavaScript challenge 特征（可以重试）
+        js_challenge_keywords = ["checking your browser", "please wait"]
+        has_js_challenge = any(kw in body for kw in js_challenge_keywords)
+
+        # Turnstile 验证码特征（需要解决，重试无效）
+        turnstile_keywords = ["turnstile", "cf-turnstile"]
+        has_turnstile = any(kw in body for kw in turnstile_keywords)
+
+        if has_js_challenge and not has_turnstile:
+            logger.info("Detected JavaScript challenge, retry may work")
+            return True
+        else:
+            logger.warning("Detected Turnstile captcha, retry unlikely to work")
+            return False
+
+    @staticmethod
+    def get_bypass_headers() -> Dict[str, str]:
+        """
+        获取可能帮助绕过 Cloudflare 的额外 HTTP 头
+
+        这些 Header 可以提高通过率，但不保证 100% 有效
+
+        返回:
+            额外的 HTTP 头字典
+        """
+        return {
+            "Cache-Control": "max-age=0",
+            "Upgrade-Insecure-Requests": "1",
+            "Sec-Fetch-User": "?1",
+            "Sec-Fetch-Dest": "document",
+            "Sec-Fetch-Mode": "navigate",
+            "Sec-Fetch-Site": "none",
+            "Priority": "u=0, i",
+        }
+
+
+class CaptchaSolver:
+    """
+    通用验证码解决器（预留接口）
+
+    支持多种验证码类型:
+    - Cloudflare Turnstile
+    - reCAPTCHA v2/v3
+    - hCaptcha
+    - 图片验证码
+    """
+
+    def __init__(self, api_key: Optional[str] = None, provider: str = "2captcha"):
+        """
+        初始化验证码解决器
+
+        参数:
+            api_key: 打码平台 API Key
+            provider: 打码平台名称 ("2captcha", "capsolver", "anticaptcha")
+        """
+        self.api_key = api_key
+        self.provider = provider
+
+        if not api_key:
+            logger.warning("CaptchaSolver initialized without API key")
+
+    async def solve_turnstile(
+        self,
+        site_key: str,
+        page_url: str,
+        **kwargs
+    ) -> Optional[str]:
+        """
+        解决 Turnstile 验证码（预留接口）
+
+        参数:
+            site_key: Turnstile site key
+            page_url: 页面 URL
+            **kwargs: 其他参数
+
+        返回:
+            Turnstile response token
+        """
+        if not self.api_key:
+            raise ValueError("API key not configured")
+
+        logger.info(f"Solving Turnstile captcha with {self.provider}...")
+
+        # TODO: 用户集成实际的打码平台 API
+        raise NotImplementedError(
+            f"Turnstile solver not implemented for provider: {self.provider}"
+        )
+
+
+# 导出主要接口
+__all__ = [
+    "CloudflareSolver",
+    "CaptchaSolver",
+]