标签:
1、过滤文本中的HTML标签
1 /// <summary> 2 /// 清除文本中Html的标签 3 /// </summary> 4 /// <param name="Content"></param> 5 /// <returns></returns> 6 public static string ClearHtml(string Content) 7 { 8 Content = ReplaceHtml("&#[^>]*;", "", Content); 9 Content = ReplaceHtml("</?marquee[^>]*>", "", Content); 10 Content = ReplaceHtml("</?object[^>]*>", "", Content); 11 Content = ReplaceHtml("</?param[^>]*>", "", Content); 12 Content = ReplaceHtml("</?embed[^>]*>", "", Content); 13 Content = ReplaceHtml("</?table[^>]*>", "", Content); 14 Content = ReplaceHtml(" ", "", Content); 15 Content = ReplaceHtml("</?tr[^>]*>", "", Content); 16 Content = ReplaceHtml("</?th[^>]*>", "", Content); 17 Content = ReplaceHtml("</?p[^>]*>", "", Content); 18 Content = ReplaceHtml("</?a[^>]*>", "", Content); 19 Content = ReplaceHtml("</?img[^>]*>", "", Content); 20 Content = ReplaceHtml("</?tbody[^>]*>", "", Content); 21 Content = ReplaceHtml("</?li[^>]*>", "", Content); 22 Content = ReplaceHtml("</?span[^>]*>", "", Content); 23 Content = ReplaceHtml("</?div[^>]*>", "", Content); 24 Content = ReplaceHtml("</?th[^>]*>", "", Content); 25 Content = ReplaceHtml("</?td[^>]*>", "", Content); 26 Content = ReplaceHtml("</?script[^>]*>", "", Content); 27 Content = ReplaceHtml("(javascript|jscript|vbscript|vbs):", "", Content); 28 Content = ReplaceHtml("on(mouse|exit|error|click|key)", "", Content); 29 Content = ReplaceHtml("<\\?xml[^>]*>", "", Content); 30 Content = ReplaceHtml("<\\/?[a-z]+:[^>]*>", "", Content); 31 Content = ReplaceHtml("</?font[^>]*>", "", Content); 32 Content = ReplaceHtml("</?b[^>]*>", "", Content); 33 Content = ReplaceHtml("</?u[^>]*>", "", Content); 34 Content = ReplaceHtml("</?i[^>]*>", "", Content); 35 Content = ReplaceHtml("</?strong[^>]*>", "", Content); 36 37 Content = Regex.Replace(Content.Trim(), "\\s+", " "); 38 string clearHtml = Content; 39 return clearHtml; 40 } 41 42 /// <summary> 43 /// 清除文本中的Html标签 44 /// </summary> 45 /// <param name="patrn">要替换的标签正则表达式</param> 46 /// <param name="strRep">替换为的内容</param> 47 /// <param name="content">要替换的内容</param> 48 /// <returns></returns> 49 private static string ReplaceHtml(string patrn, string strRep, string content) 50 { 51 if (string.IsNullOrEmpty(content)) 52 { 53 content = ""; 54 } 55 Regex rgEx = new Regex(patrn, RegexOptions.IgnoreCase); 56 string strTxt = rgEx.Replace(content, strRep); 57 return strTxt; 58 }
标签:
原文地址:http://www.cnblogs.com/loyung/p/4625974.html