| 1 |
<?php |
| 2 |
|
| 3 |
class RobotDetector { |
| 4 |
private $robots = array( |
| 5 |
'abcdatos' => true, |
| 6 |
'acme-spider' => true, |
| 7 |
'ahoythehomepagefinder' => true, |
| 8 |
'Alkaline' => true, |
| 9 |
'anthill' => true, |
| 10 |
'appie' => true, |
| 11 |
'arachnophilia' => true, |
| 12 |
'arale' => true, |
| 13 |
'araneo' => true, |
| 14 |
'araybot' => true, |
| 15 |
'architext' => true, |
| 16 |
'aretha' => true, |
| 17 |
'ariadne' => true, |
| 18 |
'arks' => true, |
| 19 |
'aspider' => true, |
| 20 |
'atn' => true, |
| 21 |
'atomz' => true, |
| 22 |
'auresys' => true, |
| 23 |
'backrub' => true, |
| 24 |
'bayspider' => true, |
| 25 |
'bbot' => true, |
| 26 |
'bigbrother' => true, |
| 27 |
'bingbot' => true, |
| 28 |
'bjaaland' => true, |
| 29 |
'blackwidow' => true, |
| 30 |
'blindekuh' => true, |
| 31 |
'Bloodhound' => true, |
| 32 |
'borg-bot' => true, |
| 33 |
'boxseabot' => true, |
| 34 |
'brightnet' => true, |
| 35 |
'bspider' => true, |
| 36 |
'cactvschemistryspider' => true, |
| 37 |
'calif' => true, |
| 38 |
'curl' => true, |
| 39 |
'cassandra' => true, |
| 40 |
'cgireader' => true, |
| 41 |
'checkbot' => true, |
| 42 |
'christcrawler' => true, |
| 43 |
'churl' => true, |
| 44 |
'cienciaficcion' => true, |
| 45 |
'cmc' => true, |
| 46 |
'Collective' => true, |
| 47 |
'combine' => true, |
| 48 |
'confuzzledbot' => true, |
| 49 |
'coolbot' => true, |
| 50 |
'core' => true, |
| 51 |
'cosmos' => true, |
| 52 |
'cruiser' => true, |
| 53 |
'cusco' => true, |
| 54 |
'cyberspyder' => true, |
| 55 |
'cydralspider' => true, |
| 56 |
'desertrealm' => true, |
| 57 |
'deweb' => true, |
| 58 |
'dienstspider' => true, |
| 59 |
'digger' => true, |
| 60 |
'diibot' => true, |
| 61 |
'directhit' => true, |
| 62 |
'dnabot' => true, |
| 63 |
'download_express' => true, |
| 64 |
'dragonbot' => true, |
| 65 |
'dwcp' => true, |
| 66 |
'e-collector' => true, |
| 67 |
'ebiness' => true, |
| 68 |
'eit' => true, |
| 69 |
'elfinbot' => true, |
| 70 |
'emacs' => true, |
| 71 |
'emcspider' => true, |
| 72 |
'esculapio' => true, |
| 73 |
'esther' => true, |
| 74 |
'evliyacelebi' => true, |
| 75 |
'nzexplorer' => true, |
| 76 |
'fastcrawler' => true, |
| 77 |
'fdse' => true, |
| 78 |
'felix' => true, |
| 79 |
'ferret' => true, |
| 80 |
'fetchrover' => true, |
| 81 |
'fido' => true, |
| 82 |
'finnish' => true, |
| 83 |
'fireball' => true, |
| 84 |
'fish' => true, |
| 85 |
'fouineur' => true, |
| 86 |
'francoroute' => true, |
| 87 |
'freecrawl' => true, |
| 88 |
'funnelweb' => true, |
| 89 |
'gama' => true, |
| 90 |
'gazz' => true, |
| 91 |
'gcreep' => true, |
| 92 |
'getbot' => true, |
| 93 |
'geturl' => true, |
| 94 |
'golem' => true, |
| 95 |
'googlebot' => true, |
| 96 |
'grapnel' => true, |
| 97 |
'griffon' => true, |
| 98 |
'gromit' => true, |
| 99 |
'gulliver' => true, |
| 100 |
'gulperbot' => true, |
| 101 |
'hambot' => true, |
| 102 |
'harvest' => true, |
| 103 |
'havindex' => true, |
| 104 |
'hi' => true, |
| 105 |
'hometown' => true, |
| 106 |
'wired-digital' => true, |
| 107 |
'htdig' => true, |
| 108 |
'htmlgobble' => true, |
| 109 |
'hyperdecontextualizer' => true, |
| 110 |
'iajabot' => true, |
| 111 |
'ibm' => true, |
| 112 |
'iconoclast' => true, |
| 113 |
'Ilse' => true, |
| 114 |
'imagelock' => true, |
| 115 |
'incywincy' => true, |
| 116 |
'informant' => true, |
| 117 |
'infoseek' => true, |
| 118 |
'infoseeksidewinder' => true, |
| 119 |
'infospider' => true, |
| 120 |
'inspectorwww' => true, |
| 121 |
'intelliagent' => true, |
| 122 |
'irobot' => true, |
| 123 |
'iron33' => true, |
| 124 |
'israelisearch' => true, |
| 125 |
'javabee' => true, |
| 126 |
'JBot' => true, |
| 127 |
'jcrawler' => true, |
| 128 |
'askjeeves' => true, |
| 129 |
'jobo' => true, |
| 130 |
'jobot' => true, |
| 131 |
'joebot' => true, |
| 132 |
'jubii' => true, |
| 133 |
'jumpstation' => true, |
| 134 |
'kapsi' => true, |
| 135 |
'katipo' => true, |
| 136 |
'kdd' => true, |
| 137 |
'kilroy' => true, |
| 138 |
'ko_yappo_robot' => true, |
| 139 |
'larbin' => true, |
| 140 |
'legs' => true, |
| 141 |
'linkidator' => true, |
| 142 |
'linkscan' => true, |
| 143 |
'linkwalker' => true, |
| 144 |
'lockon' => true, |
| 145 |
'lycos' => true, |
| 146 |
'macworm' => true, |
| 147 |
'magpie' => true, |
| 148 |
'marvin' => true, |
| 149 |
'mattie' => true, |
| 150 |
'mediafox' => true, |
| 151 |
'merzscope' => true, |
| 152 |
'meshexplorer' => true, |
| 153 |
'MindCrawler' => true, |
| 154 |
'mnogosearch' => true, |
| 155 |
'moget' => true, |
| 156 |
'momspider' => true, |
| 157 |
'monster' => true, |
| 158 |
'motor' => true, |
| 159 |
'msnbot' => true, |
| 160 |
'muncher' => true, |
| 161 |
'muninn' => true, |
| 162 |
'muscatferret' => true, |
| 163 |
'mwdsearch' => true, |
| 164 |
'myweb' => true, |
| 165 |
'NDSpider' => true, |
| 166 |
'netcarta' => true, |
| 167 |
'netmechanic' => true, |
| 168 |
'netscoop' => true, |
| 169 |
'newscan-online' => true, |
| 170 |
'nhse' => true, |
| 171 |
'nomad' => true, |
| 172 |
'northstar' => true, |
| 173 |
'objectssearch' => true, |
| 174 |
'occam' => true, |
| 175 |
'octopus' => true, |
| 176 |
'OntoSpider' => true, |
| 177 |
'openfind' => true, |
| 178 |
'orb_search' => true, |
| 179 |
'packrat' => true, |
| 180 |
'pageboy' => true, |
| 181 |
'parasite' => true, |
| 182 |
'patric' => true, |
| 183 |
'pegasus' => true, |
| 184 |
'perignator' => true, |
| 185 |
'perlcrawler' => true, |
| 186 |
'phantom' => true, |
| 187 |
'phpdig' => true, |
| 188 |
'piltdownman' => true, |
| 189 |
'pimptrain' => true, |
| 190 |
'pioneer' => true, |
| 191 |
'pitkow' => true, |
| 192 |
'pjspider' => true, |
| 193 |
'pka' => true, |
| 194 |
'plumtreewebaccessor' => true, |
| 195 |
'poppi' => true, |
| 196 |
'portalb' => true, |
| 197 |
'psbot' => true, |
| 198 |
'Puu' => true, |
| 199 |
'python' => true, |
| 200 |
'raven' => true, |
| 201 |
'rbse' => true, |
| 202 |
'resumerobot' => true, |
| 203 |
'rhcs' => true, |
| 204 |
'rixbot' => true, |
| 205 |
'roadrunner' => true, |
| 206 |
'robbie' => true, |
| 207 |
'robi' => true, |
| 208 |
'robocrawl' => true, |
| 209 |
'robofox' => true, |
| 210 |
'robozilla' => true, |
| 211 |
'roverbot' => true, |
| 212 |
'rules' => true, |
| 213 |
'safetynetrobot' => true, |
| 214 |
'scooter' => true, |
| 215 |
'search_au' => true, |
| 216 |
'search-info' => true, |
| 217 |
'searchprocess' => true, |
| 218 |
'senrigan' => true, |
| 219 |
'sgscout' => true, |
| 220 |
'shaggy' => true, |
| 221 |
'shaihulud' => true, |
| 222 |
'sift' => true, |
| 223 |
'simbot' => true, |
| 224 |
'site-valet' => true, |
| 225 |
'sitetech' => true, |
| 226 |
'skymob' => true, |
| 227 |
'slcrawler' => true, |
| 228 |
'slurp' => true, |
| 229 |
'smartspider' => true, |
| 230 |
'snooper' => true, |
| 231 |
'solbot' => true, |
| 232 |
'speedy' => true, |
| 233 |
'spider_monkey' => true, |
| 234 |
'spiderbot' => true, |
| 235 |
'spiderline' => true, |
| 236 |
'spiderman' => true, |
| 237 |
'spiderview' => true, |
| 238 |
'spry' => true, |
| 239 |
'ssearcher' => true, |
| 240 |
'suke' => true, |
| 241 |
'suntek' => true, |
| 242 |
'sven' => true, |
| 243 |
'sygol' => true, |
| 244 |
'tach_bw' => true, |
| 245 |
'tarantula' => true, |
| 246 |
'tarspider' => true, |
| 247 |
'tcl' => true, |
| 248 |
'techbot' => true, |
| 249 |
'templeton' => true, |
| 250 |
'titin' => true, |
| 251 |
'titan' => true, |
| 252 |
'tkwww' => true, |
| 253 |
'tlspider' => true, |
| 254 |
'ucsd' => true, |
| 255 |
'udmsearch' => true, |
| 256 |
'uptimebot' => true, |
| 257 |
'urlck' => true, |
| 258 |
'us' => true, |
| 259 |
'valkyrie' => true, |
| 260 |
'verticrawl' => true, |
| 261 |
'victoria' => true, |
| 262 |
'visionsearch' => true, |
| 263 |
'voidbot' => true, |
| 264 |
'voyager' => true, |
| 265 |
'vwbot' => true, |
| 266 |
'w3index' => true, |
| 267 |
'w3m2' => true, |
| 268 |
'wallpaper' => true, |
| 269 |
'wanderer' => true, |
| 270 |
'wapspider' => true, |
| 271 |
'webbandit' => true, |
| 272 |
'webcatcher' => true, |
| 273 |
'webcopy' => true, |
| 274 |
'webfetcher' => true, |
| 275 |
'webfoot' => true, |
| 276 |
'webinator' => true, |
| 277 |
'weblayers' => true, |
| 278 |
'weblinker' => true, |
| 279 |
'webmirror' => true, |
| 280 |
'webmoose' => true, |
| 281 |
'webquest' => true, |
| 282 |
'webreader' => true, |
| 283 |
'webreaper' => true, |
| 284 |
'webs' => true, |
| 285 |
'websnarf' => true, |
| 286 |
'webspider' => true, |
| 287 |
'webvac' => true, |
| 288 |
'webwalk' => true, |
| 289 |
'webwalker' => true, |
| 290 |
'webwatch' => true, |
| 291 |
'wget' => true, |
| 292 |
'whatuseek' => true, |
| 293 |
'whowhere' => true, |
| 294 |
'wlm' => true, |
| 295 |
'wmir' => true, |
| 296 |
'wolp' => true, |
| 297 |
'wombat' => true, |
| 298 |
'worm' => true, |
| 299 |
'wwwc' => true, |
| 300 |
'wz101' => true, |
| 301 |
'xget' => true, |
| 302 |
'Nederland.zoek' => true, |
| 303 |
'nutch' => true |
| 304 |
); |
| 305 |
|
| 306 |
public function is_robot() { |
| 307 |
$user_agent = empty( $_SERVER['HTTP_USER_AGENT'] ) ? '' : sanitize_text_field( wp_unslash( $_SERVER['HTTP_USER_AGENT'] ) ); |
| 308 |
|
| 309 |
$common_browsers = 'Chrome|Firefox|MSIE|Safari|Opera'; |
| 310 |
if ( preg_match( '/' . $common_browsers . '/i', $user_agent ) ) { |
| 311 |
return false; |
| 312 |
} |
| 313 |
|
| 314 |
$agent_name = preg_replace( '#^(.+?)/.*$#', '$1', $user_agent ); |
| 315 |
|
| 316 |
return ! empty( $this->robots[ $agent_name ] ); |
| 317 |
} |
| 318 |
} |
| 319 |
|
