去除HTML标签

最新推荐文章于 2024-11-26 13:54:39 发布

原创最新推荐文章于 2024-11-26 13:54:39 发布 · 107 阅读

0 ·

CC 4.0 BY-SA版权

文章标签：

#去除HTML标签

java 专栏收录该内容

3 篇文章

订阅专栏

 /**
	 * 去除str中的HTML标签
	 * 
	 * @param str
	 * @return
	 */
	public static String formatHtml(String str) {
		if (str == null) {
			return "";
		}
		str = str.toLowerCase();
		str = str.replaceAll("</?span[^>]*>", "");
		str = str.replaceAll("&#[^>]*;", "");
		str = str.replaceAll("</?marquee[^>]*>", "");
		str = str.replaceAll("</?object[^>]*>", "");
		str = str.replaceAll("</?param[^>]*>", "");
		str = str.replaceAll("</?embed[^>]*>", "");
		str = str.replaceAll("</?table[^>]*>", "");
		str = str.replaceAll("&nbsp;", "");
		str = str.replaceAll("</?tr[^>]*>", "");
		str = str.replaceAll("</?th[^>]*>", "");
		str = str.replaceAll("</?p[^>]*>", "");
		str = str.replaceAll("</?a[^>]*>", "");
		str = str.replaceAll("</?img[^>]*>", "");
		str = str.replaceAll("</?tbody[^>]*>", "");
		str = str.replaceAll("</?li[^>]*>", "");
		str = str.replaceAll("</?div[^>]*>", "");
		str = str.replaceAll("</?td[^>]*>", "");
		str = str.replaceAll("</?script[^>]*>", "");
		str = str.replaceAll("(javascript|jscript|vbscript|vbs):", "");
		str = str.replaceAll("on(mouse|exit|error|click|key)", "");
		str = str.replaceAll("<\\?xml[^>]*>", "");
		str = str.replaceAll("<\\?[a-z]+:[^>]*>", "");
		str = str.replaceAll("</?font[^>]*>", "");
		str = str.replaceAll("</?b[^>]*>", "");
		str = str.replaceAll("</?u[^>]*>", "");
		str = str.replaceAll("</?i[^>]*>", "");
		str = str.replaceAll("</?strong[^>]*>", "");
		str = str.replaceAll("</?(a|A)( .*?>|>)", "");
		return str;
	}