Excel Vba 正则URL协议、域名、端口号、页面路径

打开Excel,Shift + F11 打开VBA编辑器，注意高度时必须先保存，否则会提示无法高度外部程序。

Sub RegexURL()

Dim url As String
Dim regex As Object
url = "http://www.baidu.com/a/b/index.html"
Set regex = CreateObject("vbscript.regexp")
regex.Global = True
regex.Pattern = "(\w+)://([^/:]+)(:\d*)?([^# ]*)"
MsgBox regex.Replace(url, "使用协议/主域名/端口号/页面：[$1],[$2],[$3],[$4]")

End Sub

            Regex reg = new Regex(@"(\w+):\/\/([^/:]+)(:\d*)?([^# ]*)");

            List<string> items = new List<string>() { 
                "http://www.baidu.com",
                "http://www.baidu.com/index.html",
                "http://www.fengcao.baidu.cn/index/index.html",
                "ftp://www.baidu.cn/index/index.html",
                "ftp://www.baidu.com.cn",
                "http://www.net.cn",
            };

            StringBuilder sb = new StringBuilder ();
            foreach (string item in items)
            { 
                foreach(Match match in reg.Matches(item))
                {
                    //for(int i=0;i<match.Groups.Count ;i++)
                    //{
                        //sb.Append(match.Groups[2]);
                    MessageBox.Show(GetServerDomain(match.Groups[2].ToString ()));
                    //}
                }
                sb.AppendLine("--");
            }

            string result = sb.ToString();

       public bool IsNumeric(string str)
        {
            try { int i = Convert.ToInt32(str); return true; }
            catch { return false; }
        }

        public string GetServerDomain(string url)
        {
            string str = url.ToLower();//此处获取值转换为小写
            if (str.IndexOf('.') > 0)
            {
                string[] strArr = str.Split('.');
                string lastStr = strArr.GetValue(strArr.Length - 1).ToString();
                if (IsNumeric(lastStr)) //如果最后一位是数字，那么说明是IP地址
                {
                    return str.Replace(".", ""); //替换.为纯数字输出 
                }
                else //否则为域名
                {
                    string[] domainRules = ".com.cn|.net.cn|.org.cn|.gov.cn|.com|.net|.cn|.org|.cc|.me|.tel|.mobi|.asia|.biz|.info|.name|.tv|.hk|.公司|.中国|.网络".Split('|');
                    string findStr = string.Empty;
                    string replaceStr = string.Empty;
                    string returnStr = string.Empty;
                    for (int i = 0; i < domainRules.Length; i++)
                    {
                        if (str.EndsWith(domainRules[i].ToLower())) //如果最后有找到匹配项
                        {
                            findStr = domainRules[i].ToString(); //www.eieboom.com
                            replaceStr = str.Replace(findStr, ""); //将匹配项替换为空，便于再次判断
                            if (replaceStr.IndexOf('.') > 0) //存在二级域名或者三级，比如：www.eieboom
                            {
                                string[] replaceArr = replaceStr.Split('.'); // www eieboom
                                returnStr = replaceArr.GetValue(replaceArr.Length - 1).ToString() + findStr;
                                return returnStr;
                            }
                            else //eieboom
                            {
                                returnStr = replaceStr + findStr; //连接起来输出为：eieboom.com
                                return returnStr;
                            };
                        }
                        else
                        { returnStr = str; }
                    }
                    return returnStr;
                }
            }
            else
            {
                return str;
            }
        }