20070712 搜索引擎 蜘蛛 捕捉
http://www.yippeesoft.com

网上下载的一个PHP捕捉，修改了保存路径和搜索引擎蜘蛛类别判断
<?php
/**
 * @author      Tmpbin (tmpbin@163.com)
 * @copyright   Www.Ruanzai.Net
 * @package     ccount.php
 * @version     1.0
 * @lastupdate  2006-12-10
 * @link        http://www.ruanzai.net
 */

$datadir=$_SERVER[SCRIPT_FILENAME];
 //日志所在目录，请用绝对路径如 d:/host/web/log/
$len=strlen($path)-9;
 $datadir=substr($path,0,$len);

 session_start();
 $datafile=$datadir."system_bot.php";
 $lang = $_SERVER[\'HTTP_ACCEPT_LANGUAGE\'];
 $urlhead .= "Accept-Encoding: ".$_SERVER[\'HTTP_ACCEPT_ENCODING\']."&line;";
 $urlhead .= "User-Agent: ".$_SERVER[\'HTTP_USER_AGENT\']."&line;";
 $urlhead .= "Host: ".$_SERVER[\'HTTP_HOST\']."&line;";
 $cookie = $_SERVER[\'HTTP_COOKIE\'];
 $time=time();

 $userip = $_SERVER["HTTP_X_FORWARDED_FOR"];
 If ($userip == "") $userip = $_SERVER["REMOTE_ADDR"];

 $urlPar = $_SERVER["SCRIPT_NAME"];
 if (trim($_SERVER["QUERY_STRING"]) != "")    
 $urlPar = $urlPar."?".$_SERVER["QUERY_STRING"];
 $info_cot=$time."&line;".$userip."&line;".$urlPar."&line;".$urlhead."\\n";
if (empty($cookie)) &leftsign;

if (empty($lang))&leftsign;
 $fnum=@fopen("$datafile","a");
 flock($fnum,LOCK_EX);
 $file_data=fwrite($fnum,$info_cot);
 fclose($fnum);
 &rightsign;
&rightsign;

?>

function get_naps_bot($iee)
&leftsign;
        $useragent = strtolower($iee);
                      
        if (strpos($useragent, \’googlebot\’) !== false)&leftsign;
                return \’Googlebot\’;
        &rightsign;
      
        if (strpos($useragent, \’msnbot\’) !== false)&leftsign;
                return \’MSNbot\’;
        &rightsign;
      
        if (strpos($useragent, \’slurp\’) !== false)&leftsign;
                return \’Yahoobot\’;
        &rightsign;
      
        if (strpos($useragent, \’baiduspider\’) !== false)&leftsign;
                return \’Baiduspider\’;
        &rightsign;
      
        if (strpos($useragent, \’sohu-search\’) !== false)&leftsign;
                return \’Sohubot\’;
        &rightsign;
      
        if (strpos($useragent, \’lycos\’) !== false)&leftsign;
                return \’Lycos\’;
        &rightsign;
      
        if (strpos($useragent, \’robozilla\’) !== false)&leftsign;
                return \’Robozilla\’;
        &rightsign;      
        return $iee;
&rightsign;

$thisprog="log.php";
$_SESSION[\'login\']="ok";
if ($job=="") &leftsign;
$a=2;
$str=file("$datafile");
$count = count($str);
for ($i=0;$i<$count;$i++)&leftsign;
     $file.= $str[$i];
  &rightsign;
$new_info=explode("\\n",$file);
$count=count($new_info);
  for ($i=0; $i<$count-1; $i++) &leftsign;
list($time,$ip,$url,$ae,$ie,$host)=explode("&line;","$new_info[$i]");
$time=date("Y-m-d(H:i:s)",$time);;
$ie=get_naps_bot($ie);
#$iet=get_naps_bot($ie);
$show=$show."<tr class=\’row$a\’><td width=10%>$time</td><td width=5%>$ip
</td><td width=20%>$url</td><td width=30% title=\\"$iet\\">$ie</td><td width=5%>$ae</td><td width=20%>$host</td></tr>\\n";
$a=$a-1;
if ($a<1) $a=2;
&rightsign;

 print <<<EOT
<table width=98% align=center cellSpacing=1 cellPadding=1 style="border-collapse: collapse;background: #E5ECFA; border: 1px solid #5296FF ">
  <tr align=\’center\’>共$count 条日志
     <td style="color: #000000"><b>机器人到访日志查看</b></td>
    </tr>
<tr>
 <td>
 <table width=99% align=center cellSpacing=1 cellPadding=1 style="border-collapse: collapse;background: #E5ECFA; border: 1px solid #5296FF ">
<tr class=\’rowcat\’>
<td width=10%>&nbsp;到访时间
</td><td width=5%>蜘蛛Ip
</td><td width=20%>访问地址
</td><td width=30%>蜘蛛信息
</td><td width=5%>Accept-Encoding
</td><td width=20%>主机
</td></tr>

$show

历史博文

标签：, , , ,
四月 6, 2008 at 11:55 下午 by yippee 1,021 次
Category: Soft
Tags: , , , ,