去掉HTML标记

　Pattern pattern = Pattern.compile("<.+?>", Pattern.DOTALL);

  Matcher matcher = pattern.matcher("<a href=\"index.html\">主页</a>");
  String string = matcher.replaceAll("");

  System.out.println(string);

 

 

 

　Pattern pattern = Pattern.compile("href=\"(.+?)\"");
  Matcher matcher = pattern.matcher("＜a href=\"index.html\"＞主页＜/a＞");
  if(matcher.find())
  System.out.println(matcher.group(1));