http://www.turnitin.com/robot/crawlerinfo.html) // FAST-WebCrawler/3.x Multimedia (mm dash crawler at fast dot no) // Firefly/1.0 (compatible; Mozilla 4.0; MSIE 5.5) // Googlebot/2.1 (+ ">http://www.googlebot.com/bot.html) // Scrubby/2.2 ( ">http://www.scrubtheweb.com/) // psbot/0.1 (+ ">http://www.picsearch.com/bot.html) // NG/1.0 // URL_Spider_Pro/3.0 ( ">http://www.innerprise.net/usp-spider.asp)" // Pompos/1.3 ">http://dir.com/pompos.html // Szukacz/1.5 (robot; www.szukacz.pl/jakdzialarobot.html; info@szukacz.pl) // ASPseek/1.2.10 // NPBot-1/2.0 // NetResearchServer/2.7(loopimprovements.com/robot.html) // dloader(NaverRobot)/1.0 // Mozilla/4.0 compatible ZyBorg/1.0 Daily Refresh Beta-d03 (wn.zyborg@looksmart.net; // Mozilla/2.0 (compatible; Ask Jeeves/Teoma) // Mozilla/5.0 (Slurp/si; slurp@inktomi.com; ">http://www.inktomi.com/slurp.html) // Mozilla/5.0 [en] (compatible; Gulper Web Bot 0.2.4 www.ecsl.cs.sunysb.edu/~maxim/cgi-bin/Link/GulperBot) // Mozilla/4.0 (compatible; MSIE 5.0; Windows NT; Girafabot; girafabot at girafa dot com; // Mozilla/4.0 (efp@gmx.net) // Mozilla/4.5 (compatible; HTTrack 3.0x; Windows 98) // PingALink Monitoring Services 1.0 (http://www.pingalink.com) // IlTrovatore-Setaccio (+ ">http://www.iltrovatore.it) // Mercator-2.0 // appie 1.1 (www.walhello.com) // larbin_2.6.2 (larbin2.6.2@unspecified.mail) // OWR_Crawler 0.1 // search.ch V1.4.2 (spiderman@search.ch; // WebFilter Robot 1.0 // Openfind data gatherer, Openbot/3.0+(robot-response@openfind.com.tw;+ // LinkWalker // Internet Explorer 5.5 // BaiDuSpider // Mozilla/4.0 (compatible; B-l-i-t-z-B-O-T) // B l i t z B O T @ t r i c u s . n e t (Mozilla compatible) // sitecheck.internetseer.com (For more info see: ">http://sitecheck.internetseer.com) // http://www.almaden.ibm.com/cs/crawler   [c01] // ia_archiver // Nutch // Mozilla // HeinrichderMiragoRobot // dumbBot var $uastring; var $brand; var $version; var $bot = false; function userAgent($ua_string = "") { // *** constructor *** if (strlen($ua_string)) { $this->uastring = $ua_string; } else { // read raw UA string $this->uastring = $_SERVER["HTTP_USER_AGENT"]; } // get UA brand and version $this->brand = "Unknown"; $this->version = 0; if (ereg("([0-9a-zA-Z\.()-]+)/([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = $regs[1]; // this is a reasonable default :) $this->version = $regs[2]; // this is a reasonable default :) } if (ereg("Netscape6/([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "Netscape"; $this->version = $regs[1]; } elseif (ereg("Netscape/([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "Netscape"; $this->version = $regs[1]; } elseif (ereg("Chimera/([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "Chimera"; $this->version = $regs[1]; } elseif (ereg("Phoenix/([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "Phoenix"; $this->version = $regs[1]; } elseif (ereg("Mozilla Firebird/([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "Mozilla Firebird"; $this->version = $regs[1]; } elseif (ereg("Galeon/([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "Galeon"; $this->version = $regs[1]; } elseif (ereg("rv:([0-9a-zA-Z\.+]+)", $this->uastring, $regs) && strstr($this->uastring, "Mozilla/") && strstr($this->uastring, "Gecko/")) { $this->brand = "Mozilla"; $this->version = $regs[1]; } elseif (ereg("Opera[ /]([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "Opera"; $this->version = $regs[1]; } elseif (ereg("OmniWeb/([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "OmniWeb"; $this->version = $regs[1]; } elseif (ereg("Konqueror/([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "Konqueror"; $this->version = $regs[1]; } elseif (ereg("Safari/([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "Safari"; $this->version = $regs[1]; } elseif (ereg("AppleWebKit/([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "AppleWebKit"; $this->version = $regs[1]; } elseif (ereg("ZyBorg/([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "ZyBorg"; $this->version = $regs[1]; $this->bot = true; } elseif (ereg("Ask Jeeves/([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "Ask Jeeves"; $this->version = $regs[1]; $this->bot = true; } elseif (ereg("Slurp/([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "Slurp"; $this->version = $regs[1]; $this->bot = true; } elseif (ereg("Openbot/([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "Openbot"; $this->version = $regs[1]; $this->bot = true; } elseif (ereg("Mercator-([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "Mercator"; $this->version = $regs[1]; $this->bot = true; } elseif (ereg("appie ([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "appie"; $this->version = $regs[1]; $this->bot = true; } elseif (ereg("larbin_([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "larbin"; $this->version = $regs[1]; $this->bot = true; } elseif (ereg("Gulper Web Bot ([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "Gulper Web Bot"; $this->version = $regs[1]; $this->bot = true; } elseif (ereg("HTTrack ([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "HTTrack"; $this->version = $regs[1]; $this->bot = true; } elseif (ereg("OWR_Crawler ([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "OWR_Crawler"; $this->version = $regs[1]; $this->bot = true; } elseif (ereg("search.ch ([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "search.ch"; $this->version = $regs[1]; $this->bot = true; } elseif (ereg("WebFilter Robot ([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "WebFilter Robot"; $this->version = $regs[1]; $this->bot = true; } elseif (ereg("^LinkWalker", $this->uastring)) { $this->brand = "LinkWalker"; $this->version = ""; $this->bot = true; } elseif (ereg("^Internet Explorer 5.5", $this->uastring)) { $this->brand = "Unknown bot (IE5.5)"; $this->version = ""; $this->bot = true; } elseif (ereg("^Mozilla[\s ]*$", $this->uastring)) { $this->brand = "Unknown bot (Mozilla)"; $this->version = ""; $this->bot = true; } elseif (ereg("^BaiDuSpider", $this->uastring)) { $this->brand = "BaiDuSpider"; $this->version = ""; $this->bot = true; } elseif (ereg("^ia_archiver", $this->uastring)) { $this->brand = "ia_archiver"; $this->version = ""; $this->bot = true; } elseif (ereg("^PingALink", $this->uastring)) { $this->brand = "PingALink"; $this->version = ""; $this->bot = true; } elseif (ereg("^IlTrovatore-Setaccio", $this->uastring)) { $this->brand = "IlTrovatore-Setaccio"; $this->version = ""; $this->bot = true; } elseif (ereg("^Nutch", $this->uastring)) { $this->brand = "Nutch"; $this->version = ""; $this->bot = true; } elseif (ereg("^dumbBot", $this->uastring)) { $this->brand = "dumbBot"; $this->version = ""; $this->bot = true; } elseif (ereg("http://www.almaden.ibm.com/cs/crawler", $this->uastring)) { $this->brand = "almaden crawler"; $this->version = ""; $this->bot = true; } elseif (ereg("B-l-i-t-z-B-O-T", $this->uastring) || ereg("B l i t z B O T @ t r i c u s . n e t", $this->uastring)) { $this->brand = "BlitzBOT"; $this->version = ""; $this->bot = true; } elseif (ereg("sitecheck.internetseer.com", $this->uastring)) { $this->brand = "internetseer"; $this->version = ""; $this->bot = true; } elseif (ereg("Girafabot", $this->uastring)) { $this->brand = "Girafabot"; $this->version = ""; $this->bot = true; } elseif (ereg("efp@gmx.net", $this->uastring)) { $this->brand = "efp"; $this->version = ""; $this->bot = true; } elseif (ereg("HeinrichderMiragoRobot", $this->uastring)) { $this->brand = "HeinrichderMiragoRobot"; $this->version = ""; $this->bot = true; } elseif (ereg("MSIE ([0-9a-zA-Z\.+]+)", $this->uastring, $regs)) { $this->brand = "Microsoft Internet Explorer"; $this->version = $regs[1]; } elseif (ereg("Mozilla/([0-9a-zA-Z\.+]+)", $this->uastring, $regs) && !strstr($this->uastring, "compatible;") && !strstr($this->uastring, "Gecko/")) { $this->brand = "Netscape"; $this->version = $regs[1]; if (intval($this->version) == 4) { $this->brand .= " Communicator"; } } $botArray = array("Scooter","Spinne","Vagabondo","TurnitinBot","FAST-WebCrawler","Firefly","Googlebot", "Scrubby","psbot","NG","URL_Spider_Pro","Pompos","Szukacz","ASPseek","NPBot-1", "dloader(NaverRobot)","NetResearchServer","","","","","","",""); if (in_array($this->brand, $botArray)) { $this->bot = true; } } function getBrand() { return $this->brand; } function getVersion() { return $this->version; } function isbot() { return $this->bot; } function isns() { // set it static so that we don't have to call it that often static $is_ns; if (!isset($is_ns)) { $is_ns = false; if (strstr($this->brand, "Netscape")) { $is_ns = true; } } return $is_ns; } function isns4() { // set it static so that we don't have to call it that often static $is_ns4; if (!isset($is_ns4)) { $is_ns4 = false; if (strstr($this->brand, "Netscape") && (intval($this->version) == 4)) { $is_ns4 = true; } } return $is_ns4; } function isie() { // set it static so that we don't have to call it that often static $is_ie; if (!isset($is_ie)) { $is_ie = false; if (strstr($this->brand, "Internet Explorer")) { $is_ie = true; } } return $is_ie; } function geckobased() { // set it static so that we don't have to call it that often static $is_gecko; if (!isset($is_gecko)) { $is_gecko = false; if (strstr($this->uastring, "Gecko/")) { $is_gecko = true; } } return $is_gecko; } function geckodate() { // set it static so that we don't have to call it that often static $gdate; if (!isset($gdate)) { $gdate = 0; if (ereg("Gecko/([0-9]+)", $this->uastring, $regs)) { $gdate = $regs[1]; } } return $gdate; } function khtmlbased() { // set it static so that we don't have to call it that often static $is_khtml; if (!isset($is_khtml)) { $is_khtml = false; if (strstr($this->brand, "Konqueror") || strstr($this->brand, "Safari") || strstr($this->brand, "AppleWebKit")) { $is_khtml = true; } } return $is_khtml; } } ?>