$ip = $_SERVER['REMOTE_ADDR'];
$ua = $_SERVER['HTTP_USER_AGENT'];
if ($ip =="10.10.10.101"){
#ini_set('display_errors', 1);
#ini_set('display_startup_errors', 1);
#error_reporting(E_ALL);
}
$actual_link = $_SERVER['PHP_SELF'];
$list = file("/var/wwwroot/westnet.ca/bannedipan.txt", FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES);
#if (preg_match('/bot|semrush|riddler|qwant|curl|dataprovider|riddler|daum|get|spider|find|java|majesticsEO|teoma|contaxe|yandex|libwww-perl|abacho|accona|AddThis|AdsBot|ahoy|AhrefsBot|AISearchBot|alexa|altavista|anthill|appie|applebot|arale|araneo|AraybOt|ariadne|arks|aspseek|ATN_Worldwide|Atomz|baiduspider|baidu|bbot|bingbot|bing|Bjaaland|BlackWidow|BotLink|bot|boxseabot|bspider|calif|CCBot|ChinaClaw|christcrawler|CMC\/0\.01|combine|confuzzledbot|contaxe|CoolBot|cosmos|crawler|crawlpaper|crawl|curl|cusco|cyberspyder|cydralspider|dataprovider|digger|DIIbot|DotBot|downloadexpress|DragonBot|DuckDuckBot|dwcp|EasouSpider|ebiness|ecollector|elfinbot|esculapio|ESI|esther|eStyle|Ezooms|facebookexternalhit|facebook|facebot|fastcrawler|FatBot|FDSE|FELIX IDE|fetch|fido|find|Firefly|fouineur|Freecrawl|froogle|gammaSpider|gazz|gcreep|geona|Getterrobo-Plus|get|girafabot|golem|grabber|GrabNet|griffon|Gromit|gulliver|gulper|hambot|havIndex|hotwired|htdig|HTTrack|ia_archiver|iajabot|IDBot|Informant|InfoSeek|InfoSpiders|INGRID\/0\.1|inktomi|inspectorwww|Internet Cruiser Robot|irobot|Iron33|JBot|jcrawler|Jeeves|jobo|KDD\-Explorer|KIT\-Fireball|ko_yappo_robot|label\-grabber|larbin|legs|libwww-perl|linkedin|Linkidator|linkwalker|Lockon|logo_gif_crawler|Lycos|m2e|majesticsEO|marvin|mattie|mediafox|mediapartners|MerzScope|MindCrawler|MJ12bot|mod_pagespeed|moget|Motor|msnbot|muncher|muninn|MuscatFerret|MwdSearch|NationalDirectory|naverbot|NEC\-MeshExplorer|NetcraftSurveyAgent|NetScoop|NetSeer|newscan\-online|nil|none|Nutch|ObjectsSearch|Occam|openstat.ru\/Bot|packrat|pageboy|ParaSite|patric|pegasus|perlcrawler|phpdig|piltdownman|Pimptrain|pingdom|pinterest|pjspider|PlumtreeWebAccessor|PortalBSpider|psbot|rambler|Raven|RHCS|RixBot|roadrunner|Robbie|robi|RoboCrawl|robofox|Scooter|Scrubby|Search\-AU|searchprocess|search|SemrushBot|Senrigan|seznambot|Shagseeker|sharp\-info\-agent|sift|SimBot|Site Valet|SiteSucker|skymob|SLCrawler\/2\.0|snooper|solbot|speedy|spider_monkey|SpiderBot\/1\.0|spiderline|spider|suke|tach_bw|TechBOT|TechnoratiSnoop|templeton|teoma|titin|topiclink|twitterbot|twitter|UdmSearch|Ukonline|UnwindFetchor|URL_Spider_SQL|urlck|urlresolver|Valkyrie libwww\-perl|verticrawl|Victoria|void\-bot|Voyager|VWbot_K|wapspider|WebBandit\/1\.0|webcatcher|WebCopier|WebFindBot|WebLeacher|WebMechanic|WebMoose|webquest|webreaper|webspider|webs|WebWalker|WebZip|wget|whowhere|winona|wlm|WOLP|woriobot|WWWC|XGET|xing|yandex|yeti|Zeus/i', $_SERVER['HTTP_USER_AGENT'])){
#if (preg_match('/(?!\googlebot)|bot|semrush|riddler|curl|dataprovider|riddler|daum|get|spider|find|java|majesticsEO|teoma|contaxe|yandex|libwww-perl/i', $_SERVER['HTTP_USER_AGENT'])) {
#if (preg_match('/bot|semrush|riddler|curl|dataprovider|riddler|daum|get|spider|find|java|majesticsEO|teoma|contaxe|yandex|libwww-perl/i', $_SERVER['HTTP_USER_AGENT'])) {
#if (preg_match('/bot|semrush|riddler|qwant|curl|dataprovider|riddler|daum|get|spider|find|java|majesticsEO|teoma|contaxe|yandex|libwww-perl|abacho|accona|AddThis|AdsBot|ahoy|AhrefsBot|AISearchBot|alexa|altavista|anthill|appie|applebot|arale|araneo|AraybOt|ariadne|arks|aspseek|ATN_Worldwide|Atomz|baiduspider|baidu|bbot|bingbot|bing|Bjaaland|BlackWidow|BotLink|bot|boxseabot|bspider|calif|CCBot|ChinaClaw|christcrawler|CMC\/0\.01|combine|confuzzledbot|contaxe|CoolBot|cosmos|crawler|crawlpaper|crawl|curl|cusco|cyberspyder|cydralspider|dataprovider|digger|DIIbot|DotBot|downloadexpress|DragonBot|DuckDuckBot|dwcp|EasouSpider|ebiness|ecollector|elfinbot|esculapio|ESI|esther|eStyle|Ezooms|facebookexternalhit|facebook|facebot|fastcrawler|FatBot|FDSE|FELIX IDE|fetch|fido|find|Firefly|fouineur|Freecrawl|froogle|gammaSpider|gazz|gcreep|geona|Getterrobo-Plus|get|girafabot|golem|grabber|GrabNet|griffon|Gromit|gulliver|gulper|hambot|havIndex|hotwired|htdig|HTTrack|ia_archiver|iajabot|IDBot|Informant|InfoSeek|InfoSpiders|INGRID\/0\.1|inktomi|inspectorwww|Internet Cruiser Robot|irobot|Iron33|JBot|jcrawler|Jeeves|jobo|KDD\-Explorer|KIT\-Fireball|ko_yappo_robot|label\-grabber|larbin|legs|libwww-perl|linkedin|Linkidator|linkwalker|Lockon|logo_gif_crawler|Lycos|m2e|majesticsEO|marvin|mattie|mediafox|mediapartners|MerzScope|MindCrawler|MJ12bot|mod_pagespeed|moget|Motor|msnbot|muncher|muninn|MuscatFerret|MwdSearch|NationalDirectory|naverbot|NEC\-MeshExplorer|NetcraftSurveyAgent|NetScoop|NetSeer|newscan\-online|nil|none|Nutch|ObjectsSearch|Occam|openstat.ru\/Bot|packrat|pageboy|ParaSite|patric|pegasus|perlcrawler|phpdig|piltdownman|Pimptrain|pingdom|pinterest|pjspider|PlumtreeWebAccessor|PortalBSpider|psbot|rambler|Raven|RHCS|RixBot|roadrunner|Robbie|robi|RoboCrawl|robofox|Scooter|Scrubby|Search\-AU|searchprocess|search|SemrushBot|Senrigan|seznambot|Shagseeker|sharp\-info\-agent|sift|SimBot|Site Valet|SiteSucker|skymob|SLCrawler\/2\.0|snooper|solbot|speedy|spider_monkey|SpiderBot\/1\.0|spiderline|spider|suke|tach_bw|TechBOT|TechnoratiSnoop|templeton|teoma|titin|topiclink|twitterbot|twitter|UdmSearch|Ukonline|UnwindFetchor|URL_Spider_SQL|urlck|urlresolver|Valkyrie libwww\-perl|verticrawl|Victoria|void\-bot|Voyager|VWbot_K|wapspider|WebBandit\/1\.0|webcatcher|WebCopier|WebFindBot|WebLeacher|WebMechanic|WebMoose|webquest|webreaper|webspider|webs|WebWalker|WebZip|wget|whowhere|winona|wlm|WOLP|woriobot|WWWC|XGET|xing|yahoo|YandexBot|YandexMobileBot|yandex|yeti|Zeus/i', $_SERVER['HTTP_USER_AGENT'])){
#if (preg_match('/bot|crawl|curl|qwant|dataprovider|search|get|spider|find|java|majesticsEO|teoma|contaxe|yandex|libwww-perl/i', $_SERVER['HTTP_USER_AGENT'], $matches)) {
if (preg_match('/^(?!.*\bgooglebot\b)(?!.*\bslurp\b)(?!.*\byandex\b)(?!.*\bbing\b).*?(?:bot|crawl|qwant|curl|dataprovider|search|get|spider|find|java|majesticsEO|yahoo|teoma|contaxe|yandex|libwww-perl|abacho|accona|AddThis|AdsBot|ahoy|AhrefsBot|AISearchBot|alexa|altavista|anthill|appie|arale|araneo|AraybOt|ariadne|arks|aspseek|ATN_Worldwide|Atomz|baiduspider|baidu|bbot|bingbot|bing|Bjaaland|BlackWidow|BotLink|bot|boxseabot|bspider|calif|CCBot|ChinaClaw|christcrawler|CMC\/0\.01|combine|confuzzledbot|contaxe|CoolBot|cosmos|crawler|crawlpaper|crawl|curl|cusco|cyberspyder|cydralspider|dataprovider|digger|DIIbot|DotBot|downloadexpress|DragonBot|DuckDuckBot|dwcp|EasouSpider|ebiness|ecollector|elfinbot|esculapio|ESI|esther|eStyle|Ezooms|facebot|fastcrawler|FatBot|FDSE|FELIX IDE|fetch|fido|find|Firefly|fouineur|Freecrawl|froogle|gammaSpider|gazz|gcreep|geona|Getterrobo-Plus|get|girafabot|golem|grabber|GrabNet|griffon|Gromit|gulliver|gulper|hambot|havIndex|hotwired|htdig|HTTrack|ia_archiver|iajabot|IDBot|Informant|InfoSeek|InfoSpiders|INGRID\/0\.1|inktomi|inspectorwww|Internet Cruiser Robot|irobot|Iron33|JBot|jcrawler|Jeeves|jobo|KDD\-Explorer|KIT\-Fireball|ko_yappo_robot|label\-grabber|larbin|legs|libwww-perl|linkedin|Linkidator|linkwalker|Lockon|logo_gif_crawler|Lycos|m2e|majesticsEO|marvin|mattie|mediafox|mediapartners|MerzScope|MindCrawler|MJ12bot|mod_pagespeed|moget|Motor|msnbot|muncher|muninn|MuscatFerret|MwdSearch|NationalDirectory|naverbot|NEC\-MeshExplorer|NetcraftSurveyAgent|NetScoop|NetSeer|newscan\-online|nil|none|Nutch|ObjectsSearch|Occam|openstat.ru\/Bot|packrat|pageboy|ParaSite|patric|pegasus|perlcrawler|phpdig|piltdownman|Pimptrain|pingdom|pinterest|pjspider|PlumtreeWebAccessor|PortalBSpider|psbot|rambler|Raven|RHCS|RixBot|roadrunner|Robbie|robi|RoboCrawl|robofox|Scooter|Scrubby|Search\-AU|searchprocess|search|SemrushBot|Senrigan|seznambot|Shagseeker|sharp\-info\-agent|sift|SimBot|Site Valet|SiteSucker|skymob|SLCrawler\/2\.0|snooper|solbot|speedy|spider_monkey|SpiderBot\/1\.0|spiderline|spider|suke|tach_bw|TechBOT|TechnoratiSnoop|templeton|teoma|titin|topiclink|UdmSearch|Ukonline|UnwindFetchor|URL_Spider_SQL|urlck|urlresolver|Valkyrie libwww\-perl|verticrawl|Victoria|void\-bot|Voyager|VWbot_K|wapspider|WebBandit\/1\.0|webcatcher|WebCopier|WebFindBot|WebLeacher|WebMechanic|WebMoose|webquest|webreaper|webspider|webs|WebWalker|WebZip|wget|whowhere|winona|wlm|WOLP|woriobot|WWWC|XGET|xing|yeti|Zeus)/i', $_SERVER['HTTP_USER_AGENT'], $matches)) {
$arr = $matches;
$uafound = $arr['0'];
#echo "Useragent Match: $uafound\n";
#Goodbots "Googlebot" OR "bingbot" OR "Yahoo" OR "Slurp" OR "DuckDuckBot"
if (in_array($ip, $list) ){
#echo "$ip FOUND IN FILE $ua $uafound";
}else{
#echo "$ip NOT FOUND IN FILE $ua $uafound";
if ($ip !=="10.10.10.101" ){
#$ht = "Deny from $ip";
file_put_contents("/var/wwwroot/westnet.ca/bannedipan.txt", $ip."\n", FILE_APPEND);
#file_put_contents("/var/wwwroot/westnet.ca/.htaccess", $ht."\n", FILE_APPEND);
file_put_contents("/var/wwwroot/westnet.ca/bannedbots.txt", "$ip $ua Useragent Match: $uafound\n", FILE_APPEND);
}
}
/*
header("HTTP/1.1 503 Service Temporarily Unavailable");
header("Status: 503 Service Temporarily Unavailable");
header("Retry-After: 3600");
echo '
Site is temporarily unavailable due to maintenance
Site is temporarily unavailable due to maintenance
We expect to have the site back up within 4 hours.
';
exit;
*/
#}
}#goodbotcheck
/*
if (in_array($ip, $list) ){
#echo "$ip FOUND IN FILE, last array";
header("HTTP/1.1 503 Service Temporarily Unavailable");
header("Status: 503 Service Temporarily Unavailable");
header("Retry-After: 3600");
echo '
Site is temporarily unavailable due to maintenance
Site is temporarily unavailable due to maintenance
We expect to have the site back up within 4 hours.