[Python] 纯文本查看 复制代码
import requests import re from datetime import datetime, timedelta # 目标网页 URL url = 'https://www.rarlab.com/' # 发送 HTTP GET 请求获取网页内容 response = requests.get(url) response.raise_for_status() # 确保请求成功 # 使用正则表达式查找连接文字为“Chinese Simplified”的下载链接 pattern = r'href="(/rar/winrar-x64-[^"]+sc\.exe)"[^>]*>Chinese Simplified<' match = re.search(pattern, response.text, re.IGNORECASE) if match: # 补全下载链接 download_link = f"https://www.rarlab.com{match.group(1)}" print(f"找到下载链接: {download_link}") # 发送 HTTP HEAD 请求获取文件的服务器时间 head_response = requests.head(download_link) server_date = head_response.headers.get('Last-Modified') if server_date: # 将服务器时间转换为 datetime 对象 server_date = datetime.strptime(server_date, '%a, %d %b %Y %H:%M:%S %Z') print(f"文件在服务器的时间: {server_date}") # 日期往前推一个月 target_date = server_date - timedelta(days=30) file_name = match.group(1).split('/')[-1] for i in range(60): # 最多加60天 # 生成新的日期格式 new_date_str = target_date.strftime('%Y%m%d') new_url_x64 = f"https://www.win-rar.com/fileadmin/winrar-versions/sc/sc{new_date_str}/wrr/{file_name}" new_url_x32 = new_url_x64.replace('x64', 'x32') print(f"尝试访问: {new_url_x64}") # 发送 HTTP HEAD 请求测试 x64 文件是否存在 head_response_x64 = requests.head(new_url_x64) if head_response_x64.status_code == 200: print(f"x64 文件存在: {new_url_x64}") # 下载 x64 文件 download_response_x64 = requests.get(new_url_x64) download_response_x64.raise_for_status() # 确保下载成功 # 保存 x64 文件到当前目录 with open(file_name, 'wb') as file: file.write(download_response_x64.content) print(f"x64 文件已下载到当前目录: {file_name}") # 发送 HTTP HEAD 请求测试 x32 文件是否存在 head_response_x32 = requests.head(new_url_x32) if head_response_x32.status_code == 200: print(f"x32 文件存在: {new_url_x32}") # 下载 x32 文件 download_response_x32 = requests.get(new_url_x32) download_response_x32.raise_for_status() # 确保下载成功 # 保存 x32 文件到当前目录 file_name_x32 = file_name.replace('x64', 'x32') with open(file_name_x32, 'wb') as file: file.write(download_response_x32.content) print(f"x32 文件已下载到当前目录: {file_name_x32}") else: print(f"x32 文件不存在: {new_url_x32}") break else: print(f"文件不存在: {new_url_x64}") target_date += timedelta(days=1) else: print("无法获取文件的服务器时间") else: print("未找到连接文字为'Chinese Simplified'的下载链接")
