PHP判断来访是搜索引擎蜘蛛还是普通用户的代码小结

<?php
function checkrobot($useragent=''){
	static $kw_spiders = array('bot', 'crawl', 'spider' ,'slurp', 'sohu-search', 'lycos', 'robozilla');
	static $kw_browsers = array('msie', 'netscape', 'opera', 'konqueror', 'mozilla');

	$useragent = strtolower(empty($useragent) ? $_server['http_user_agent'] : $useragent);
	if(strpos($useragent, 'http://') === false && dstrpos($useragent, $kw_browsers)) return false;
	if(dstrpos($useragent, $kw_spiders)) return true;
	return false;
}
function dstrpos($string, $arr, $returnvalue = false) {
	if(empty($string)) return false;
	foreach((array)$arr as $v) {
		if(strpos($string, $v) !== false) {
			$return = $returnvalue ? $v : true;
			return $return;
		}
	}
	return false;
}
if(checkrobot()){
	echo '机器人爬虫';
}else{
	echo '人';
}
?>

<?php
if(!checkrobot()){
//do something
}
?>

$useragent = addslashes(strtolower($_server['http_user_agent']));

 if (strpos($useragent, 'googlebot')!== false){$bot = 'google';}
 elseif (strpos($useragent,'mediapartners-google') !== false){$bot = 'google adsense';}
 elseif (strpos($useragent,'baiduspider') !== false){$bot = 'baidu';}
 elseif (strpos($useragent,'sogou spider') !== false){$bot = 'sogou';}
 elseif (strpos($useragent,'sogou web') !== false){$bot = 'sogou web';}
 elseif (strpos($useragent,'sosospider') !== false){$bot = 'soso';}
 elseif (strpos($useragent,'360spider') !== false){$bot = '360spider';}
 elseif (strpos($useragent,'yahoo') !== false){$bot = 'yahoo';}
 elseif (strpos($useragent,'msn') !== false){$bot = 'msn';}
 elseif (strpos($useragent,'msnbot') !== false){$bot = 'msnbot';}
 elseif (strpos($useragent,'sohu') !== false){$bot = 'sohu';}
 elseif (strpos($useragent,'yodaobot') !== false){$bot = 'yodao';}
 elseif (strpos($useragent,'twiceler') !== false){$bot = 'twiceler';}
 elseif (strpos($useragent,'ia_archiver') !== false){$bot = 'alexa_';}
 elseif (strpos($useragent,'iaarchiver') !== false){$bot = 'alexa';}
 elseif (strpos($useragent,'slurp') !== false){$bot = '雅虎';}
 elseif (strpos($useragent,'bot') !== false){$bot = '其它蜘蛛';}
 if(isset($bot)){
   $fp = @fopen('bot.txt','a');
   fwrite($fp,date('y-m-d h:i:s')."t".$_server["remote_addr"]."t".$bot."t".'http://'.$_server['server_name'].$_server["request_uri"]."rn");
   fclose($fp);
 }

function is_crawler() { 
  $useragent = strtolower($_server['http_user_agent']); 
  $spiders = array( 
    'googlebot', // google 爬虫 
    'baiduspider', // 百度爬虫 
    'yahoo! slurp', // 雅虎爬虫 
    'yodaobot', // 有道爬虫 
    'msnbot' // bing爬虫 
    // 更多爬虫关键字 
  ); 
  foreach ($spiders as $spider) { 
    $spider = strtolower($spider); 
    if (strpos($useragent, $spider) !== false) { 
      return true; 
    } 
  } 
  return false; 
}

function iscrawler() { 
    echo $agent= strtolower($_server['http_user_agent']); 
    if (!empty($agent)) { 
        $spidersite= array( 
            "tencenttraveler", 
            "baiduspider+", 
            "baidugame", 
            "googlebot", 
            "msnbot", 
            "sosospider+", 
            "sogou web spider", 
            "ia_archiver", 
            "yahoo! slurp", 
            "youdaobot", 
            "yahoo slurp", 
            "msnbot", 
            "java (often spam bot)", 
            "baiduspider", 
            "voila", 
            "yandex bot", 
            "bspider", 
            "twiceler", 
            "sogou spider", 
            "speedy spider", 
            "google adsense", 
            "heritrix", 
            "python-urllib", 
            "alexa (ia archiver)", 
            "ask", 
            "exabot", 
            "custo", 
            "outfoxbot/yodaobot", 
            "yacy", 
            "surveybot", 
            "legs", 
            "lwp-trivial", 
            "nutch", 
            "stackrambler", 
            "the web archive (ia archiver)", 
            "perl tool", 
            "mj12bot", 
            "netcraft", 
            "msiecrawler", 
            "wget tools", 
            "larbin", 
            "fish search", 
        ); 
        foreach($spidersite as $val) { 
            $str = strtolower($val); 
            if (strpos($agent, $str) !== false) { 
                return true; 
            } 
        } 
    } else { 
        return false; 
    } 
} 
if (iscrawler()){ 
    echo "你好蜘蛛精！"; 
} 
else{ 
   echo "你不是蜘蛛精啊！"; 
} 

<?php
$flag = false;
$tmp = $_server['http_user_agent'];
if(strpos($tmp, 'googlebot') !== false){
  $flag = true;
} else if(strpos($tmp, 'baiduspider') >0){
  $flag = true;
} else if(strpos($tmp, 'yahoo! slurp') !== false){
  $flag = true;
} else if(strpos($tmp, 'msnbot') !== false){
  $flag = true;
} else if(strpos($tmp, 'sosospider') !== false){
  $flag = true;
} else if(strpos($tmp, 'yodaobot') !== false || strpos($tmp, 'outfoxbot') !== false){
  $flag = true;
} else if(strpos($tmp, 'sogou web spider') !== false || strpos($tmp, 'sogou orion spider') !== false){
  $flag = true;
} else if(strpos($tmp, 'fast-webcrawler') !== false){
  $flag = true;
} else if(strpos($tmp, 'gaisbot') !== false){
  $flag = true;
} else if(strpos($tmp, 'ia_archiver') !== false){
  $flag = true;
} else if(strpos($tmp, 'altavista') !== false){
  $flag = true;
} else if(strpos($tmp, 'lycos_spider') !== false){
  $flag = true;
} else if(strpos($tmp, 'inktomi slurp') !== false){
  $flag = true;
}
if($flag == false){
  header("location: http://www.jb51.net" . $_server['request_uri']);
  // 自动转到http://www.jb51.net 对应的网页
  // $_server['request_uri'] 为域名后面的路径
  // 或换成header("location: http://www.jb51.net/abc/d.php");
  exit();
}
?>