Update 测试.py

2024-09-03 17:46:49 +08:00
parent 78628a068f
commit 4079c8c702
1 changed files with 3 additions and 1 deletions
@@ -84,6 +84,8 @@ for url in urls:
    urls_all = re.findall(pattern, page_content)
    # 去重得到唯一的URL列表
    unique_urls = set(urls_all)
    # 排除包含特定子字符串的域名
    unique_urls = {u for u in unique_urls if "fofa.info" not in u}
    valid_urls = []
    # 多线程获取可用url