正则表达式彻底去除HTML

 public static String RMHTML(String Htmlstring)

 {

/**

转载请注明:

PowerBy:Lulu

Www.HotCity.Cn

*/

// 删除脚本

 Htmlstring = RegexPattern("<//s*?script[^>]*>[//s//S]*?<//s*?///s*?script//s*?>","",Htmlstring);

// 删除HTML

 Htmlstring = RegexPattern("<([^>]*)>", "",Htmlstring);

 Htmlstring = RegexPattern( "([/r/n])[//s]+", "",Htmlstring);

 Htmlstring = RegexPattern( "-->", "",Htmlstring);

 Htmlstring = RegexPattern( "<!--.*", "",Htmlstring);

 Htmlstring = RegexPattern( "&(quot|#34);", "/"",Htmlstring);

 Htmlstring = RegexPattern( "&(amp|#38);", "&",Htmlstring);

 Htmlstring = RegexPattern( "&(lt|#60);", "<",Htmlstring);

 Htmlstring = RegexPattern( "&(gt|#62);", ">",Htmlstring);

 Htmlstring = RegexPattern( "&(nbsp|#160);", " ",Htmlstring);

 Htmlstring = RegexPattern( "&(iexcl|#161);", "//xa1",Htmlstring);

 Htmlstring = RegexPattern( "&(cent|#162);", "//xa2",Htmlstring);

 Htmlstring = RegexPattern( "&(pound|#163);", "//xa3",Htmlstring);

 Htmlstring = RegexPattern( "&(copy|#169);", "//xa9",Htmlstring);

 Htmlstring = RegexPattern( "&#(//d+);", "",Htmlstring);

 

 Htmlstring = RegexPattern("<", "",Htmlstring);

 Htmlstring = RegexPattern(">", "",Htmlstring);

 //Htmlstring.replace("/r/n", "",Htmlstring);

 

 return Htmlstring;

 }
   public static String RegexPattern(String pattern,String str,String content){

       if(pattern!=null && !pattern.equals("")){

           Pattern p = Pattern.compile(pattern,2); //参数2表示大小写不区分

           Matcher m = p.matcher(content);
           content=m.replaceAll(str);
 

       }
       return content;      
   }