|
以下为引用的内容:
立即学习“PHP免费学习笔记(深入)”;
//使用Snoopy的抓页功能
include "Snoopy.class.php";
//目标URL
$url = $_REQUEST['targeturl'];
//将所有传递过来的参数列出来
$par = array();
$GetPost = array_merge($_POST,$_GET);
foreach($GetPost As $Key=>$Value){
if($Key!='targeturl'){
$Value = str_replace("%25","%",$Value);
array_push($par,($Key . "=" . $Value));
}
}
//判断目标URL是否带有?(即是否带有参数)
$cc = strpos($url,"?") ? "&" : "?";
//重组URL
$geturl = !$par ? $url : $url.$cc.implode("&",$par);
//抓取重组后的URL页面内容
$snoopy = new Snoopy;
$snoopy->fetch($geturl);
//替换目标内容中脚本里有可能替换父窗口地址的代码
$org = str_replace("top.location","top.title",$snoopy->results);
//尝试转换目标内容编码到UTF-8
$opt = iconv("gbk","utf-8",$org);
//判断目标内容编码为GBK或UTF-8
$ec = strlen($opt)?"gbk":"utf-8";
?> <script type="text/javascript"><br>
//封闭运行,避免与后面内容中脚本混乱<br>
(function(){<br>
var easyUTF8 = function(gbk){if(!gbk){return '';}var utf8 = [];for(var i=0;i<gbk.length;i++){var s_str = gbk.charAt(i);if(!(/^%u/i.test(escape(s_str)))){utf8.push(s_str);continue;}var s_char = gbk.charCodeAt(i);var b_char = s_char.toString(2).split('');var c_char = (b_char.length==15)?[0].concat(b_char):b_char;var a_b =[];a_b[0] = '1110'+c_char.splice(0,4).join('');a_b[1] = '10'+c_char.splice(0,6).join('');a_b[2] = '10'+c_char.splice(0,6).join('');for(var n=0;n<a_b.length;n++){utf8.push('%'+parseInt(a_b[n],2).toString(16).toUpperCase());}}return utf8.join('');};<br>
var getArgs = function(surl){var sarg = surl.split('?'),rv={};rv.filename=sarg[0];if(!sarg[1]){return rv;}var aarg=sarg[1].split('&'),atmp=[];for(var i=0;i<aarg.length;i++){atmp=aarg[i].split('=');rv[atmp[0]]=atmp[1];}return rv;};<br>
var createIPH = function(name,value){if(!name){return;}if(/msie/i.test(navigator.appVersion)){return document.createElement('<input type="hidden" name="'+name+'" value="'+value+'"/>');}else{var dfi = document.createElement('input');dfi.type = 'hidden';dfi.name = name;dfi.value = value;return dfi;}};<br>
//回显目标URL到父窗口文本框<br>
var dtu = top.document.getElementById('targeturl');<br>
if(dtu){dtu.value = '<?php echo $geturl;?>';}<br>
//目标URL及domain<br>
var sref = '<?php echo $url;?>';<br>
var sdomain = sref.match(/^http:\/\/[^\/]*/i)[0];<br>
//页面加载后执行下面的过程<br>
var process = function(){<br>
//抓取页面中所有链接<br>
var dlink = document.getElementsByTagName('a'),la = dlink.length;<br>
//抓取页面中所有表单<br>
var dform = document.getElementsByTagName('form'),lf = dform.length;<br>
//遍历所有链接,替换它们的href地址<br>
for(var i=0;i<la;i++){<br>
var src = dlink[i].href.toString().replace(/^http:\/\/www\.w3cgroup\.com(?:\/geturl)?/i,sdomain);<br>
var oargs = getArgs(src),ahref = [];<br>
//UTF-8编码参数值<br>
for(var d in oargs){<br>
if(!dd=='filename'!oargs[d]){continue;}<br>
ahref.push(d+'='+encodeURIComponent(easyUTF8(oargs[d])));<br>
}<br>
var ghref = ahref.length?oargs.filename+'?'+ahref.join('&'):oargs.filename;<br>
//重设链接地址<br>
dlink[i].href = 'http://www.w3cgroup.com/geturl/action_get.php?targeturl='+ghref;<br>
}<br>
//遍历所有表单,替换它们的action地址<br>
for(i=0;i<lf;i++){<br>
//抓取表单action并处理<br>
var src = dform[i].action.toString().replace(/^http:\/\/www\.w3cgroup\.com(?:\/geturl)?/i,sdomain);<br>
if(!(/^http/.test(src))){src = (/^\/.*$/.test(src))?(sdomain+src):(sdomain+'/'+src);}<br>
//创建一个隐藏域targeturl,值为上面处理的src地址<br>
var dfi = createIPH('targeturl',src);<br>
dform[i].appendChild(dfi);<br>
//创建一个隐藏域ie,值为utf-8,纯属为搜索引擎使用<br>
var dfi2 = createIPH('ie','utf-8');<br>
dform[i].appendChild(dfi2);<br>
//重设表单提交目标窗口<br>
dform[i].target = 'actioncontent';<br>
//重设表单action地址<br>
dform[i].action = 'http://www.w3cgroup.com/geturl/action_get.php';<br>
//重设表单onsubmit事件,用来UTF8编码字段值<br>
dform[i].onsubmit = function(){<br>
var dlms = this.elements,l = dlms.length-1,pn = '',pt = '',pv = '';<br>
for(var i=0;i<l;i++){<br>
pn = dlms[i].name,pt = dlms[i].type,pv = dlms[i].value;<br>
if(!pnpn=='targeturl'pn=='ie'){continue;}<br>
if(pt=='submit'pt=='reset'pt=='button'){<br>
dlms[i].value = encodeURIComponent(pv);<br>
}else{<br>
dlms[i].value = encodeURIComponent(easyUTF8(pv));<br>
}<br>
}<br>
};<br>
}<br>
};<br>
//绑定该过程到window.onload<br>
if(document.attachEvent){window.attachEvent('onload',process);}else{window.addEventListener('load',process,false);}<br>
})();<br>
</script>
//脚本放在输出内容前,避免内容中可能出现脚本错误而导致我们要做的事情被废掉
//输出抓到的目标页面内容
|