C#如何用正则表达式过滤网页的JavaScript代码？

C#如何用正则表达式过滤网页的JavaScript代码？例如：用户用记事本打开一个.htm网页。复制里面的代码到TextBox1里。点击Button1后，过滤掉里面的JavaScript代码。在TextBox2里显示过滤后的结果。这个用C#如何实现？代码在下面：

解决方案 »

免费领取超大流量手机卡，每月29元包185G流量+100分钟通话, 中国电信官方发货

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3c.org/TR/1999/REC-html401-19991224/loose.dtd">

<HTML xmlns="http://www.w3.org/1999/xhtml" xmlns:fb =
"http://www.facebook.com/2008/fbml"><HEAD><TITLE>Google Chrome - Reviews and free Google Chrome downloads at Download.com</TITLE>
<META http-equiv=Content-Type content="text/html; charset=gb2312">
<SCRIPT src="1.files/oreo.moo.rb.combined.js" type=text/javascript></SCRIPT><SCRIPT type=text/javascript>            UserVars = new LocalVars.UserVars({
                loggedIn: ' ',
                userName: ' ',
                ursRegId: ' ',
                rememberMe: ' ',
                confirmed: ' '
            });

            PageVars = new LocalVars.PageVars({
                pageType: '3640',
                nodeId: '2356',
                siteId: '4',
                assetId: '11015161',
                pageNumber: '',
                channelId: '20',
                editionId: '3',
                brandId: '6',
                breadcrumb: '20:2137:2356',
                userIP: '114.240.162.243',
                guid: 'slH0fwoPjAUAAGFgNtsAAABo'
               });
        </SCRIPT>
<LINK rev=stylesheet media=screen href="1.files/matrix.css"
type=text/css rel=stylesheet><LINK rev=stylesheet media=print
href="1.files/print.css" type=text/css rel=stylesheet>
<META
content="Come to CNET Download.com for free and safe Google Chrome downloads. Browse the Web, Google-style."
name=description>
<META content="Google Chrome, Web Browsers, downloads, software, trial, free"
name=keywords><LINK rev=stylesheet href="1.files/default.css" type=text/css
rel=stylesheet>
<SCRIPT src="1.files/download.global.framework.compressed.js"
type=text/javascript></SCRIPT><SCRIPT src="1.files/download.tron.title.detail2.compressed.js"
type=text/javascript></SCRIPT><META content="MSHTML 6.00.2900.2912" name=GENERATOR></HEAD>
<BODY class="siteId4 pageType3640"><IMG
style="LEFT: 0px; POSITION: absolute; TOP: 0px" src=""> <IMG
style="LEFT: 0px; POSITION: absolute; TOP: 0px" height=0 alt=""
src="1.files/dotclear.gif" width=0>
<SCRIPT src="1.files/dw.js" type=text/javascript></SCRIPT><SCRIPT type=text/javascript>
            DW.pageParams = {
                siteid: '4',
                edid:   '3',
                prtnr:  'CNET Networks, Inc.',
                ptid:   '3640',
                onid:   '2356',
                asid:   '11015161',
                pgnbr:  '',
                astId:  '1'
            };
            DW.regSilo = '1';

        </SCRIPT><SCRIPT type=text/javascript>
    DW.trackClicks();
</SCRIPT>

<DIV id=rb_bodyWrap>
<DIV id=rb_shell>
<DIV id=rb_header name="rb_header" section="hdr">
<DIV id=headerContents>
<UL class=section id=brandNavBar section="brandnav">
  <LI class=brandNav>
  <UL>
    <LI class="siteId1 "><A
    href="http://www.cnet.com/2001-1_1-0.html"><SPAN>home</SPAN></A> </LI>
    <LI class="siteId7 "><A
    href="http://reviews.cnet.com/2001-1_7-0.html"><SPAN>reviews</SPAN></A>
</LI>
    <LI class="siteId3 "><A
    href="http://news.cnet.com/2001-1_3-0.html"><SPAN>news</SPAN></A> </LI>
    <LI class="siteId4 active"><A
    href="http://download.cnet.com/windows/"><SPAN>downloads</SPAN></A> </LI>
    <LI class="siteId53 "><A
    href="http://cnettv.cnet.com/2000-1_53-0.html"><SPAN>cnet tv</SPAN></A>
  </LI></UL></LI>
  <LI class=rubics>
  <DIV id=rubicsTextAd>
  <DIV class=rubics_netxp1_main id=rbx_netxp1_parent>
  <SCRIPT type=text/javascript>
function do_rbx_ctrk_netxp1 (aElem, iTrkId, cTrkPrefix) {
var imgElem = document.getElementById(iTrkId);
if (!imgElem) {
return true;
} if (!aElem.rbxCPrefixed) {
aElem.rbxCPrefixed = true;
var destUri = encodeURIComponent(aElem.href);
var newUri = cTrkPrefix + destUri;
aElem.href = newUri;
}
return true;
}
function do_rbx_itrk_netxp1 (parentId, iTrkId, iTrkUri) {
if (!document.images) {
return false;
} var imgId = iTrkId;
var imgElem = document.getElementById(imgId); if (!imgElem) { var parentElem = document.getElementById(parentId);
if (!parentElem) {
return false;
} var imgElem = document.createElement('img');
if (!imgElem) {
return false;
} imgElem.id = imgId;
imgElem.src = iTrkUri;
imgElem.alt = '';
imgElem.width = '1';
imgElem.height = '1'; parentElem.appendChild(imgElem);
}
return true;
}
</SCRIPT>
  On The Insider: <A class=xpromolink
  onclick="return do_rbx_ctrk_netxp1(this, 'rbx_netxp1_1', 'http://dw.com.com/rubicsclk?ver=2&ts=2009.04.17.22.37.44&edId=3&onId=2356&ptId=3640&sId=4&appId=19&offId=5869&unitId=45&poolId=1&f1=1&f2=%2d0&f3=%2d0&alg=8&&opt=1&linkPos=1&destUrl=');"
  href="http://www.theinsider.com/news/2019525_Michael_Jackson_Auction_Off_the_Block"
  target=_top>Michael Jackson Auction Cancelled</A>
  <SCRIPT type=text/javascript>
//<![CDATA[
do_rbx_itrk_netxp1('rbx_netxp1_parent', 'rbx_netxp1_1', 'http://dw.com.com/rubicsimp/c.gif?ver=2&ts=2009.04.17.22.37.44&edId=3&onId=2356&ptId=3640&sId=4&appId=19&unitId=45&poolId=1&f1=1&f2=%2d0&f3=%2d0&alg=8&opt=1&off=5869,-1');
//]]>
</SCRIPT>
   </DIV></DIV></LI>
public static string NoHTML(string Htmlstring)
        {
            //删除脚本
            Htmlstring = Regex.Replace(Htmlstring, @"<script[^>]*?>.*?</script>", "", RegexOptions.IgnoreCase);
            //删除HTML
            Htmlstring = Regex.Replace(Htmlstring, @"<(.[^>]*)>", "", RegexOptions.IgnoreCase);
            Htmlstring = Regex.Replace(Htmlstring, @"([\r\n])[\s]+", "", RegexOptions.IgnoreCase);
            Htmlstring = Regex.Replace(Htmlstring, @"-->", "", RegexOptions.IgnoreCase);
            Htmlstring = Regex.Replace(Htmlstring, @"<!--.*", "", RegexOptions.IgnoreCase);            Htmlstring = Regex.Replace(Htmlstring, @"&(quot|#34);", "\"", RegexOptions.IgnoreCase);
            Htmlstring = Regex.Replace(Htmlstring, @"&(amp|#38);", "&", RegexOptions.IgnoreCase);
            Htmlstring = Regex.Replace(Htmlstring, @"&(lt|#60);", "<", RegexOptions.IgnoreCase);
            Htmlstring = Regex.Replace(Htmlstring, @"&(gt|#62);", ">", RegexOptions.IgnoreCase);
            Htmlstring = Regex.Replace(Htmlstring, @"&(nbsp|#160);", " ", RegexOptions.IgnoreCase);
            Htmlstring = Regex.Replace(Htmlstring, @"&(iexcl|#161);", "\xa1", RegexOptions.IgnoreCase);
            Htmlstring = Regex.Replace(Htmlstring, @"&(cent|#162);", "\xa2", RegexOptions.IgnoreCase);
            Htmlstring = Regex.Replace(Htmlstring, @"&(pound|#163);", "\xa3", RegexOptions.IgnoreCase);
            Htmlstring = Regex.Replace(Htmlstring, @"&(copy|#169);", "\xa9", RegexOptions.IgnoreCase);
            Htmlstring = Regex.Replace(Htmlstring, @"&#(\d+);", "", RegexOptions.IgnoreCase);            Htmlstring.Replace("<", "");
            Htmlstring.Replace(">", "");
            Htmlstring.Replace("\r\n", "");
            Htmlstring = HttpContext.Current.Server.HtmlEncode(Htmlstring).Trim();            return Htmlstring;
        }
Regex.Replace(Htmlstring, @" <script[^>]*?>.*? </script>", "", RegexOptions.IgnoreCase);
光这句还不行，还得替换所有类似 onclick="alert();"的东西,加上
Htmlstring = Regex.Replace(Htmlstring, "on.*?\"", "", RegexOptions.IgnoreCase);
Regex.Replace(Htmlstring, @" <script[^>]*?>.*? </script>", "", RegexOptions.IgnoreCase);
有错误，应该为
Regex.Replace(Htmlstring, @" <script.*?</script>", "", RegexOptions.IgnoreCase);
private void button1_Click(object sender, EventArgs e)
        {
            textBox2.Text = Regex.Replace(textBox1.Text, @"<script[^>]*?>[\s\S]*<\/script>", "", RegexOptions.IgnoreCase);
        }
我不贴了。你打开Google的首页就行了。不要用IE保存到硬盘里。直接“查看”--“源文件”。
试下行不？肯定是不行的。
回复太快！如果您是恶意刷楼，将会受到严厉惩罚！