app.py

import gradio as gr
import requests
import re
import time
import logging
from functools import lru_cache
from urllib.parse import urlparse, urljoin
from typing import Dict, Any, Optional, List
from dataclasses import dataclass
from datetime import datetime
from fastapi import FastAPI, Request
from fastapi.responses import StreamingResponse, RedirectResponse, JSONResponse
from fastapi.middleware.cors import CORSMiddleware

# 配置日志
logging.basicConfig(
    level=logging.INFO,
    format='%(asctime)s - %(levelname)s - %(message)s'
)

@dataclass
class ProxyResponse:
    """代理响应数据类"""
    status: int
    content: Any
    headers: Dict[str, str]
    redirect_url: Optional[str] = None
    error: Optional[str] = None

class Config:
    """配置类"""
    ASSET_URL = "https://1pages.nbid.bid/"
    PREFIX = "/"
    JSDELIVR = 0  # 是否使用 jsDelivr 镜像
    CACHE_TTL = 3600
    MAX_RETRIES = 3
    TIMEOUT = 20
    CHUNK_SIZE = 1024 * 1024  # 10KB 分块大小
    SIZE_LIMIT = 1024 * 1024 * 1024 * 99  # 999GB 文件大小限制
    
    RATE_LIMIT = {
        "window_ms": 15 * 60 * 1000,  # 15分钟
        "max": 200  # 限制每个IP最多100个请求
    }
    WHITE_LIST: List[str] = []  # 白名单
    BLACK_LIST: List[str] = []  # 黑名单
    PASS_LIST: List[str] = []   # 直接通过名单（使用 jsDelivr）
    
    # 请求头
    DEFAULT_HEADERS = {
        "User-Agent": "git/2.41.0",  # 使用 Git 用户代理
        "Accept": "*/*",
        "Accept-Encoding": "gzip, deflate, br"
    }
    
    # CORS设置
    CORS = {
        "allow_origins": ["*"],
        "allow_methods": ["GET", "POST", "OPTIONS", "HEAD"],
        "allow_headers": ["*"],
        "max_age": 1728000
    }
    
    # URL模式（完善匹配模式）
    PATTERNS = {
        # 发布和归档
        "releases": r"^(?:https?:\/\/)?github\.com\/(?P<author>.+?)\/(?P<repo>.+?)\/(?:releases|archive)\/.*$",
        # blob和raw文件
        "blob": r"^(?:https?:\/\/)?github\.com\/(?P<author>.+?)\/(?P<repo>.+?)\/(?:blob|raw)\/.*$",
        # Git操作
        "git": r"^(?:https?:\/\/)?github\.com\/(?P<author>.+?)\/(?P<repo>.+?)\/(?:info\/refs\?service=)?git-.*$",
        # Raw内容
        "raw": r"^(?:https?:\/\/)?raw\.(?:githubusercontent|github)\.com\/(?P<author>.+?)\/(?P<repo>.+?)\/.+?\/.+$",
        # Gist
        "gist": r"^(?:https?:\/\/)?gist\.(?:githubusercontent|github)\.com\/(?P<author>.+?)\/.+?\/.+$",
        # 标签
        "tags": r"^(?:https?:\/\/)?github\.com\/(?P<author>.+?)\/(?P<repo>.+?)\/tags.*$"
    }

class RateLimiter:
    """请求频率限制器"""
    def __init__(self):
        self.request_records: Dict[str, List[float]] = {}
    
    def is_allowed(self, ip: str) -> bool:
        now = time.time() * 1000
        window_start = now - Config.RATE_LIMIT["window_ms"]
        
        if ip not in self.request_records:
            self.request_records[ip] = []
        
        # 清理过期记录
        self.request_records[ip] = [t for t in self.request_records[ip] if t > window_start]
        
        if len(self.request_records[ip]) >= Config.RATE_LIMIT["max"]:
            return False
        
        self.request_records[ip].append(now)
        return True
    
    def get_remaining(self, ip: str) -> int:
        """获取剩余请求次数"""
        if ip not in self.request_records:
            return Config.RATE_LIMIT["max"]
        return Config.RATE_LIMIT["max"] - len(self.request_records[ip])

class GitHubProxy:
    """GitHub代理核心类"""
    def __init__(self):
        self.rate_limiter = RateLimiter()
        self.session = requests.Session()
        self.session.headers.update(Config.DEFAULT_HEADERS)
    
    def check_url(self, url: str) -> bool:
        """检查URL是否匹配GitHub模式"""
        return any(re.search(pattern, url, re.I) for pattern in Config.PATTERNS.values())
    
    def check_white_list(self, url: str) -> bool:
        """检查白名单"""
        if not Config.WHITE_LIST:
            return True
        return any(white_item in url for white_item in Config.WHITE_LIST)
    
    @lru_cache(maxsize=1000)
    def fetch_github_content(self, url: str, method: str = "GET", stream: bool = False) -> ProxyResponse:
        """获取GitHub内容（带缓存）"""
        try:
            response = self.session.request(
                method=method,
                url=url,
                timeout=Config.TIMEOUT,
                allow_redirects=False,
                stream=stream
            )
            
            headers = dict(response.headers)
            
            # 处理重定向
            if response.is_redirect:
                redirect_url = response.headers["Location"]
                if self.check_url(redirect_url):
                    redirect_url = Config.PREFIX + redirect_url
                return ProxyResponse(
                    status=response.status_code,
                    content="",
                    headers=headers,
                    redirect_url=redirect_url
                )
            
            # 如果是流式响应，直接返回response对象
            if stream:
                return ProxyResponse(
                    status=response.status_code,
                    content=response,  # 返回response对象以支持流式传输
                    headers=headers
                )
            
            # 检查是否是二进制内容
            content_type = response.headers.get('content-type', '')
            is_binary = not any(text_type in content_type.lower() for text_type in ['text', 'json', 'xml', 'html'])
            
            content = response.content if is_binary else response.text
            
            return ProxyResponse(
                status=response.status_code,
                content=content,
                headers=headers
            )
            
        except requests.Timeout:
            return ProxyResponse(
                status=504,
                content="Request Timeout",
                headers={},
                error="请求超时"
            )
        except Exception as e:
            logging.error(f"Fetch error: {str(e)}")
            return ProxyResponse(
                status=500,
                content=str(e),
                headers={},
                error="服务器内部错误"
            )
    
    def proxy_request(self, url: str, request: Request) -> Dict[str, Any]:
        """处理代理请求"""
        # 记录请求
        logging.info(f"Proxy request from {request.client.host} to {url}")
        
        # 检查频率限制
        if not self.rate_limiter.is_allowed(request.client.host):
            return {
                "status": 429,
                "content": "Too Many Requests",
                "headers": {},
                "error": "请求过于频繁，请稍后再试",
                "rate_limit": {
                    "remaining": self.rate_limiter.get_remaining(request.client.host),
                    "reset": int((time.time() * 1000 + Config.RATE_LIMIT["window_ms"]) / 1000)
                }
            }
        
        # 检查白名单
        if not self.check_white_list(url):
            return {
                "status": 403,
                "content": "Access Denied",
                "headers": {},
                "error": "访问被拒绝"
            }
        
        # 处理URL
        if not url.startswith(("http://", "https://")):
            url = "https://" + url
        
        # 检查URL是否为GitHub链接
        if not self.check_url(url):
            return {
                "status": 400,
                "content": "Invalid GitHub URL",
                "headers": {},
                "error": "无效的GitHub URL"
            }
        
        # 处理jsDelivr重定向
        if Config.JSDELIVR and re.search(Config.PATTERNS["blob"], url):
            url = url.replace("/blob/", "@").replace("github.com", "cdn.jsdelivr.net/gh")
            return {
                "status": 302,
                "content": "",
                "headers": {"Location": url},
                "redirect_url": url
            }
        
        # 获取内容
        response = self.fetch_github_content(url)
        result = {
            "status": response.status,
            "content": response.content,
            "headers": response.headers,
            "timestamp": datetime.now().isoformat()
        }
        
        if response.redirect_url:
            result["redirect_url"] = response.redirect_url
        if response.error:
            result["error"] = response.error
            
        return result

# 创建 FastAPI 应用
api = FastAPI()

# 配置 CORS
api.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

def create_interface():
    """创建Gradio界面"""
    proxy = GitHubProxy()
    
    @api.get("/{path:path}")
    async def proxy_download(request: Request, path: str):
        """处理直接代理请求"""
        if not path:
            return JSONResponse({"error": "无效的请求路径"}, status_code=400)
            
        # 检查并处理URL
        if not path.startswith(('http://', 'https://')):
            path = 'https://' + path
            
        try:
            # 检查URL格式
            match = None
            for pattern in Config.PATTERNS.values():
                if re.match(pattern, path):
                    match = re.match(pattern, path)
                    break
                    
            if not match:
                return JSONResponse({"error": "不支持的URL格式"}, status_code=400)
                
            # 检查白名单和黑名单
            author, repo = match.group('author', 'repo')
            if Config.WHITE_LIST and not any(
                (a == '*' or a == author) and (r == '*' or r == repo)
                for a, r in [x.split('/') for x in Config.WHITE_LIST]
            ):
                return JSONResponse({"error": "不在白名单中"}, status_code=403)
                
            if any(
                (a == '*' or a == author) and (r == '*' or r == repo)
                for a, r in [x.split('/') for x in Config.BLACK_LIST]
            ):
                return JSONResponse({"error": "在黑名单中"}, status_code=403)
                
            # 处理 jsDelivr 重定向
            use_jsdelivr = Config.JSDELIVR or any(
                (a == '*' or a == author) and (r == '*' or r == repo)
                for a, r in [x.split('/') for x in Config.PASS_LIST]
            )
            
            if use_jsdelivr and ('blob' in path or 'raw.githubusercontent.com' in path):
                # 转换为 jsDelivr URL
                if 'blob' in path:
                    path = path.replace('/blob/', '@').replace('github.com', 'cdn.jsdelivr.net/gh', 1)
                else:
                    path = re.sub(r'(\.com/.*?/.+?)/(.+?/)', r'\1@\2', path, 1)
                    path = path.replace('raw.githubusercontent.com', 'cdn.jsdelivr.net/gh', 1)
                return RedirectResponse(path)
                
            # 处理 blob 到 raw 的转换
            if 'blob' in path:
                path = path.replace('/blob/', '/raw/', 1)
                
            # 处理 Git 请求
            if 'git-upload-pack' in path or 'git-receive-pack' in path or 'info/refs' in path:
                headers = dict(request.headers)
                headers.update({
                    'User-Agent': 'git/2.41.0',
                    'Accept': 'application/x-git-upload-pack-result, */*',
                })
                proxy.session.headers.update(headers)
            
            # 获取代理响应
            response = proxy.proxy_request(path, request)
            
            if 'error' in response:
                return JSONResponse({"error": response['error']}, status_code=response['status'])
                
            if 'redirect_url' in response:
                return RedirectResponse(response['redirect_url'])
                
            # 获取文件内容（使用流式传输）
            proxy_response = proxy.fetch_github_content(path, stream=True)
            if proxy_response.error:
                return JSONResponse({"error": proxy_response.error}, status_code=proxy_response.status)
                
            # 检查文件大小限制
            content_length = int(proxy_response.headers.get('content-length', 0))
            if content_length > Config.SIZE_LIMIT:
                return RedirectResponse(path)
                
            # 返回流式响应
            headers = dict(proxy_response.headers)
            # 设置正确的内容类型
            if 'git-upload-pack' in path:
                headers['Content-Type'] = 'application/x-git-upload-pack-result'
            elif 'git-receive-pack' in path:
                headers['Content-Type'] = 'application/x-git-receive-pack-result'
            elif 'info/refs' in path:
                headers['Content-Type'] = 'application/x-git-upload-pack-advertisement'
            
            return StreamingResponse(
                proxy_response.content.iter_content(chunk_size=Config.CHUNK_SIZE),
                headers=headers,
                status_code=proxy_response.status
            )
                
        except Exception as e:
            logging.error(f"Proxy error: {str(e)}")
            return JSONResponse({"error": f"代理请求失败: {str(e)}"}, status_code=500)
    
    with gr.Blocks(title="GitHub Proxy", theme=gr.themes.Soft()) as blocks:
        gr.Markdown("""
        # 🚀 GitHub Proxy
        
        ### 使用方法
        1. 直接访问: `https://your-domain.com/github-url`
        2. 或者在下方输入GitHub URL进行测试
        
        ### 功能特点
        - ✨ 支持多种GitHub URL格式
        - 🔄 自动处理重定向
        - 💾 响应缓存
        - ⚡ CDN加速支持
        - 🛡️ 请求频率限制
        
        ### 支持的URL类型
        - GitHub Release/Archive
        - GitHub Raw/Blob
        - GitHub Gist
        - Raw GitHub Content
        """)
        
        with gr.Row():
            url_input = gr.Textbox(
                label="GitHub URL",
                placeholder="输入GitHub URL，例如：github.com/user/repo/blob/master/file.txt",
                scale=4
            )
            submit_btn = gr.Button("获取内容", scale=1)
        
        with gr.Row():
            with gr.Column():
                status = gr.Textbox(label="状态")
                headers = gr.JSON(label="响应头")
            with gr.Column():
                content = gr.Textbox(label="内容", max_lines=20)
                error = gr.Textbox(label="错误信息", visible=False)
        
        def handle_request(url: str, request: gr.Request):
            result = proxy.proxy_request(url, request)
            
            # 更新UI
            error_visible = "error" in result
            error_msg = result.get("error", "")
            
            return {
                status: f"状态码: {result['status']}",
                headers: result["headers"],
                content: result["content"],
                error: error_msg,
                error: gr.update(visible=error_visible, value=error_msg)
            }
        
        submit_btn.click(
            fn=handle_request,
            inputs=[url_input],
            outputs=[status, headers, content, error]
        )
        
        # 添加示例
        gr.Examples(
            examples=[
                ["github.com/microsoft/vscode/blob/main/README.md"],
                ["raw.githubusercontent.com/microsoft/vscode/main/README.md"],
                ["github.com/ollama/ollama/releases/download/v0.5.1/ollama-windows-amd64.zip"]
            ],
            inputs=url_input
        )
    
    # 将 Gradio 界面挂载到 FastAPI
    blocks.queue()
    api.mount("/ui", blocks)
    return api

if __name__ == "__main__":
    import uvicorn
    app = create_interface()
    uvicorn.run(app, host="0.0.0.0", port=7860)