表达式 控件 html,正则表达式能过滤html元素你知道吗

/**

* filter all html element.

* For example:< a href="www.sohu.com/test">hello!< /a>

* The filter result is :hello!

* Notice:This method filter the text between "< " and ">"

* @param element

* @return

*/

public static String getTxtWithoutHTMLElement (String element)

{

// String reg="< [^< |^>]+>";

// return element.replaceAll(reg,"");

if(null==element||"".equals(element.trim()))

{

return element;

}

Pattern pattern=Pattern.compile("< [^< |^>]*>");

Matcher matcher=pattern.matcher(element);

StringBuffer txt=new StringBuffer();

while(matcher.find())

{

String group=matcher.group();

if(group.matches("< [\\s]*>"))

{

matcher.appendReplacement(txt,group);

}

else

{

matcher.appendReplacement(txt,"");

}

}

matcher.appendTail(txt);

repaceEntities(txt,"&","&");

repaceEntities(txt,"< ","< ");

repaceEntities(txt,">",">");

repaceEntities(txt,""","\"");

repaceEntities(txt," ","");

return txt.toString();

}

下面是测试用例：

public void testGetTxtWithoutHTMLElement ()

{

assertEquals("test",ExcelHssfView.getTxtWithoutHTMLElement("< a href='a/test'>test< /a>"));

assertEquals("test",ExcelHssfView.getTxtWithoutHTMLElement("< a href='a/test'>test"));

assertEquals("test",ExcelHssfView.getTxtWithoutHTMLElement("< input type='text'>test< /input>"));

assertEquals("test",ExcelHssfView.getTxtWithoutHTMLElement("< p>test"));

assertEquals("test",ExcelHssfView.getTxtWithoutHTMLElement("< table>< tr>< td>test< /td>< /tr>< /table>"));

assertEquals("te< st",ExcelHssfView.getTxtWithoutHTMLElement("< p>te< st"));

assertEquals("te>st",ExcelHssfView.getTxtWithoutHTMLElement("< p>te>st"));

assertEquals("tst",ExcelHssfView.getTxtWithoutHTMLElement("< p>t< e>st"));

assertEquals("t< st",ExcelHssfView.getTxtWithoutHTMLElement("< p>t< < e>st"));

assertEquals("< >test",ExcelHssfView.getTxtWithoutHTMLElement("< p>< >test"));

assertEquals("< >test",ExcelHssfView.getTxtWithoutHTMLElement("< p>< >test"));

assertEquals("< < >test",ExcelHssfView.getTxtWithoutHTMLElement("< p>< < >test"));

assertEquals("test",ExcelHssfView.getTxtWithoutHTMLElement("< table>< tr>< td> test< /td>< /tr>< /table>"));

}

标签：

本站文章除注明转载外，均为本站原创或翻译。欢迎任何形式的转载，但请务必注明出处、不得修改原文相关链接，如果存在内容上的异议请邮件反馈至hey@evget.com