API tools faq
paste
Advertisement
Guest User

Untitled

a guest
Nov 7th, 2014
248
0
Never
Add comment
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 20.40 KB | None | 0 0
raw download clone embed print report
  1. <?php
  2.  
  3. /**************************************************************************
  4. * Add more spiders to your database *
  5. ***************************************************************************/
  6. /*MyBB Plugin: Video Link v 0.1
  7. ---------------------------------------------------------------------------------------------------------
  8. http://www.edidesign.ca
  9. ---------------------------------------------------------------------------------------------------------
  10. Author : Thor2705
  11. http://www.edidesign.ca
  12. Please do not remove the links from this file. Please report any sites that are not working on mybb comunity forum
  13. */
  14.  
  15. if(!defined("IN_MYBB"))
  16. {
  17. die("Direct initialization of this file is not allowed.<br /><br />Please make sure IN_MYBB is defined.");
  18. }
  19.  
  20. $plugins->add_hook("admin_config_spiders_add", "addbot");
  21.  
  22. function addbot_info() {
  23. return array(
  24. "name" => "Add Spiders",
  25. "description" => "Add more spiders to your data base",
  26.  
  27. "website" => "http://www.edidesign.ca",
  28.  
  29. "author" => "Thor2705",
  30.  
  31. "authorsite" => "http://www.edidesign.ca",
  32.  
  33. "version" => "0.1",
  34.  
  35. "guid" => "66cc039962c0412fadbb48be13e96e37",
  36.  
  37. "compatibility" => "14*",
  38. );
  39. }
  40.  
  41.  
  42.  
  43.  
  44.  
  45. function addbot_activate()
  46. {
  47.  
  48. // Globals
  49. global $db;
  50.  
  51. $addSpider = array(
  52. // The big spiders detected by MyBB by default Website Description
  53. /* array(
  54. 'name' => 'Google',
  55. 'useragent' => 'Googlebot'), //http://www.google.com The main spider used by Google
  56. array(
  57. 'name' => 'Msn',
  58. 'useragent' => 'MsnBot'
  59. ), */ //http://www.msn.com The main spider used by MSN
  60. /* array(
  61. 'name' => 'Yahoo!',
  62. 'useragent' => 'Slurp'), */ //http://www.yahoo.com Worlds most aggressive spider
  63.  
  64. // MAJOR spiders Website Description
  65. array(
  66. 'name' => 'Ask',
  67. 'useragent' => 'Teoma'), // http://www.ask.com Spider for Ask Search Engine
  68. array(
  69. 'name' => 'Baidu',
  70. 'useragent' => 'Baiduspider'), // http://www.baidu.com Spider for Chinese search engine
  71. array(
  72. 'name' => 'GigaBot',
  73. 'useragent' => 'Gigabot'), // http://www.gigablast.com Another heavily travelled spider
  74. array(
  75. 'name' => 'Google-AdSense',
  76. 'useragent' => 'Mediapartners-Google'), // http://www.google.com Spider related to Adsense/Adwords
  77. array(
  78. 'name' => 'Google-Adwords',
  79. 'useragent' => 'AdsBot-Google'), // http://www.google.com Spider related to Adwords
  80. array(
  81. 'name' => 'Google-SA',
  82. 'useragent' => 'gsa-crawler'), // http://www.google.com Google Search Appliance Spider
  83. array(
  84. 'name' => 'Google-Image',
  85. 'useragent' => 'Googlebot-Image'), // http://www.google.com Spider for google image search
  86. array(
  87. 'name' => 'InternetArchive',
  88. 'useragent' => 'ia_archiver-web.archive.org'),// http://www.archive.org Way back When machine Spider
  89. array(
  90. 'name' => 'Alexa',
  91. 'useragent' => 'ia_archiver'), // http://www.alexa.com *Must be detected after Internet Archive
  92. array(
  93. 'name' => 'Omgili',
  94. 'useragent' => 'omgilibot'), // http://www.omgili.com Extremely aggressive Messageboard/forum Spider
  95. array(
  96. 'name' => 'Speedy Spider',
  97. 'useragent' => 'Speedy Spider'), // http://www.entireweb.com Entire web spider
  98. array(
  99. 'name' => 'Yahoo',
  100. 'useragent' => 'yahoo'), // http://www.yahoo.com For Yahoo Publisher Network (a variety in use)
  101. array(
  102. 'name' => 'Yahoo JP',
  103. 'useragent' => 'Y!J'), // http://www.yahoo.co.jp Spider for Yahoo Japan
  104.  
  105. // Checkers/Testers/Robots Website Description
  106. array(
  107. 'name' => 'DeadLinksChecker',
  108. 'useragent' => 'link validator'), // http://www.dead-links.com/ Checks your site for dead/bad links
  109. array(
  110. 'name' => 'W3C Validator',
  111. 'useragent' => 'W3C_Validator'), // http://validator.w3.org Checks standards validity of any html/xhtml page
  112. array(
  113. 'name' => 'W3C CSSValidator',
  114. 'useragent' => 'W3C_CSS_Validator'), // http://jigsaw.w3.org/css-validator/ Checks standards validity of css stylesheets
  115. array(
  116. 'name' => 'W3C FeedValidator',
  117. 'useragent' => 'FeedValidator'), // http://validator.w3.org/feed/ Checks standards validity of atom/rss feeds
  118. array(
  119. 'name' => 'W3C LinkChecker',
  120. 'useragent' => 'W3C-checklink'), // http://validator.w3.org/checklink Checks links on any html/xhtml page are valid
  121. array(
  122. 'name' => 'W3C mobileOK',
  123. 'useragent' => 'W3C-mobileOK'), // http://www.w3.org/2006/07/mobileok-ddc Checks page for how good it is for mobiles
  124. array(
  125. 'name' => 'W3C P3PValidator',
  126. 'useragent' => 'P3P Validator'), // http://www.w3.org/P3P/validator.html Checks something??
  127.  
  128. // Feed readers Website Description
  129. array(
  130. 'name' => 'Bloglines',
  131. 'useragent' => 'Bloglines'), // http://www.bloglines.com Spider for blog/rich web content (owned by Ask)
  132. array(
  133. 'name' => 'Feedburner',
  134. 'useragent' => 'Feedburner'), // http://www.feedburner.com Another RSS feed reader
  135.  
  136. // Website Thumbnail/Snapshot/Thumbshot takers Website Description
  137. array(
  138. 'name' => 'SnapBot',
  139. 'useragent' => 'Snapbot'), // http://www.snap.com Shapshots provider
  140. array(
  141. 'name' => 'Picsearch',
  142. 'useragent' => 'psbot'), // http://www.picsearch.com Picture/Image Search Engine
  143. array(
  144. 'name' => 'Websnapr',
  145. 'useragent' => 'Websnapr'), // http://www.websnapr.com Snapshot/site screenshot taker
  146.  
  147. // More MINOR Spiders/Robots Website Description
  148. array(
  149. 'name' => 'AllTheWeb',
  150. 'useragent' => 'FAST-WebCrawler'), // http://www.alltheweb.com Spider for alltheweb (now owned by Yahoo)
  151. array(
  152. 'name' => 'Altavista',
  153. 'useragent' => 'Scooter'), // http://www.altavista.com Another Major Search Engine spider
  154. array(
  155. 'name' => 'Asterias',
  156. 'useragent' => 'asterias'), // http://www.aol.com Media Spider
  157. array(
  158. 'name' => '192bot',
  159. 'useragent' => '192.comAgent'), // http://www.192.com Spider to index for 192.com
  160. array(
  161. 'name' => 'AbachoBot',
  162. 'useragent' => 'ABACHOBot'), // http://www.abacho.com Spider for multi language search engine/translator
  163. array(
  164. 'name' => 'Abdcatos',
  165. 'useragent' => 'abcdatos'), // http://www.abcdatos.com/botlink/ Spider for Italian Search Engine
  166. array(
  167. 'name' => 'Acoon',
  168. 'useragent' => 'Acoon'), // http://www.acoon.de Spider for small search engine
  169. array(
  170. 'name' => 'Accoona',
  171. 'useragent' => 'Accoona'), // http://www.accoona.com Spider for Accoona
  172. array(
  173. 'name' => 'BecomeBot',
  174. 'useragent' => 'BecomeBot'), // http://www.become.com Shopping/Products type search engine
  175. array(
  176. 'name' => 'BlogRefsBot',
  177. 'useragent' => 'BlogRefsBot'), // http://www.blogrefs.com/about/bloggers Blogs related spider
  178. array(
  179. 'name' => 'Daumoa',
  180. 'useragent' => 'Daumoa'), // http://ws.daum.net/aboutkr.html South Korean Search Engine Spider
  181. array(
  182. 'name' => 'DuckDuckBot',
  183. 'useragent' => 'DuckDuckBot'), // http://duckduckgo.com/duckduckbot.html Spider for small search engine
  184. array(
  185. 'name' => 'Exabot',
  186. 'useragent' => 'Exabot'), // http://www.exalead.com Spider for small search engine
  187. array(
  188. 'name' => 'Furl',
  189. 'useragent' => 'Furlbot'), // http://www.furl.net Spider for Furl social bookmarking site
  190. array(
  191. 'name' => 'FyperSpider',
  192. 'useragent' => 'FyberSpider'), // http://www.fybersearch.com Spider for Small Search Engine
  193. array(
  194. 'name' => 'Geona',
  195. 'useragent' => 'GeonaBot'), // http://www.geona.com Spider for another small search engine
  196. array(
  197. 'name' => 'GirafaBot',
  198. 'useragent' => 'Girafabot'), // http://www.girafa.com/ Thumbshot provider
  199. array(
  200. 'name' => 'GoSeeBot',
  201. 'useragent' => 'GoSeeBot'), // http://www.gosee.com/bot.html Spider for small search engine
  202. array(
  203. 'name' => 'Ichiro',
  204. 'useragent' => 'ichiro'), // http://help.goo.ne.jp/door/crawler.html Spider for Japanese search engine
  205. array(
  206. 'name' => 'LapozzBot',
  207. 'useragent' => 'LapozzBot'), // http://www.lapozz.hu Spider for Hungarian search engine
  208. array(
  209. 'name' => 'Looksmart',
  210. 'useragent' => 'WISENutbot'), // http://www.looksmart.com Spider related to advertising
  211. array(
  212. 'name' => 'Lycos',
  213. 'useragent' => 'Lycos_Spider'), // http://www.lycos.com Spider for search engine
  214. array(
  215. 'name' => 'Majestic12',
  216. 'useragent' => 'MJ12bot/v2'), // http://www.majestic12.co.uk/ Distributed Search Engine Project
  217. array(
  218. 'name' => 'MLBot',
  219. 'useragent' => 'MLBot'), // http://www.metadatalabs.com/ Media indexing spider
  220. array(
  221. 'name' => 'MSRBOT',
  222. 'useragent' => 'msrbot'), // http://research.microsoft.com/research/sv/msrbot/ Microsoft Research bot
  223. array(
  224. 'name' => 'MSR-ISRCCrawler',
  225. 'useragent' => 'MSR-ISRCCrawler'), // http://www.microsoft.com/research/ Another Microsoft Research bot
  226. array(
  227. 'name' => 'Naver',
  228. 'useragent' => 'NaverBot'), // http://www.naver.com South Korean Search Engine Spider
  229. array(
  230. 'name' => 'Naver',
  231. 'useragent' => 'Yeti'), // http://www.naver.com Another NaverBot for the South Korean Search Engine
  232. array(
  233. 'name' => 'NoxTrumBot',
  234. 'useragent' => 'noxtrumbot'), // http://www.noxtrum.com Spider for Spanish search engine
  235. array(
  236. 'name' => 'OmniExplorer',
  237. 'useragent' => 'OmniExplorer_Bot'), // http://www.omni-explorer.com/ Spider
  238. array(
  239. 'name' => 'OnetSzukaj',
  240. 'useragent' => 'OnetSzukaj'), // http://szukaj.onet.pl Polish Search Engine Spider
  241. array(
  242. 'name' => 'ScrubTheWeb',
  243. 'useragent' => 'Scrubby'), // http://www.scrubtheweb.com Spider for Scrub the web
  244. array(
  245. 'name' => 'SearchSight',
  246. 'useragent' => 'SearchSight'), // http://www.searchsite.com Another search engine
  247. array(
  248. 'name' => 'Seeqpod',
  249. 'useragent' => 'Seeqpod'), // http://www.seeqpod.com Spider for search engine (the google for mp3 files)
  250. array(
  251. 'name' => 'Shablast',
  252. 'useragent' => 'ShablastBot'), // http://www.shablast.com Spider for a small search engine
  253. array(
  254. 'name' => 'SitiDiBot',
  255. 'useragent' => 'SitiDiBot'), // http://www.sitidi.net Spider for italian Sitidi search engine
  256. array(
  257. 'name' => 'Slider',
  258. 'useragent' => 'silk/1.0'), // http://www.slider.com Spider for Slider, but it only spiders DMOZ entries
  259. array(
  260. 'name' => 'Sogou',
  261. 'useragent' => 'Sogou'), // http://www.sogou.com Spider for Chinese search engine
  262. array(
  263. 'name' => 'Sosospider',
  264. 'useragent' => 'Sosospider'), // http://help.soso.com/webspider.htm Non-english search engine
  265. array(
  266. 'name' => 'StackRambler',
  267. 'useragent' => 'StackRambler'), // http://www.rambler.ru/doc/robots.shtml Spider for Russian portal/search engine
  268. array(
  269. 'name' => 'SurveyBot',
  270. 'useragent' => 'SurveyBot'), // http://www.domaintools.com Probe for website statistics (WhoIs Source)
  271. array(
  272. 'name' => 'Touche',
  273. 'useragent' => 'Touche'), // http://www.touche.com.ve Another small search engine
  274. array(
  275. 'name' => 'Walhello',
  276. 'useragent' => 'appie'), // http://www.wahello.com/ Spider for wahello
  277. array(
  278. 'name' => 'WebAlta',
  279. 'useragent' => 'WebAlta'), // http://www.webalta.net Russian Search Engine
  280. array(
  281. 'name' => 'Wisponbot',
  282. 'useragent' => 'wisponbot'), // http://www.wispon.com Korean Search Engine
  283. array(
  284. 'name' => 'YacyBot',
  285. 'useragent' => 'yacybot'), // http://www.yacy.com Crawler for distributed search engine
  286. array(
  287. 'name' => 'YodaoBot',
  288. 'useragent' => 'YodaoBot'), // http://www.yodao.com Spider for Chinese Search Engine
  289.  
  290. // Google-Wanna-Be's - Spiders/Robots for Startups Website Description
  291. array(
  292. 'name' => 'Charlotte',
  293. 'useragent' => 'Charlotte'), // http://www.searchme.com/support/ Spider for new search engine (in beta)
  294. array(
  295. 'name' => 'DiscoBot',
  296. 'useragent' => 'DiscoBot'), // http://discoveryengine.com/discobot.html Spider for new search engine startup
  297. array(
  298. 'name' => 'EnaBot',
  299. 'useragent' => 'EnaBot'), // http://www.enaball.com/crawler.html Experimental new spider
  300. array(
  301. 'name' => 'Gaisbot',
  302. 'useragent' => 'Gaisbot'), // http://gais.cs.ccu.edu.tw/robot.php Spider for search engine startup
  303. array(
  304. 'name' => 'Kalooga',
  305. 'useragent' => 'kalooga'), // http://www.kalooga.com Spider for new media search engine (in beta)
  306. array(
  307. 'name' => 'ScoutJet',
  308. 'useragent' => 'ScoutJet'), // http://www.scoutjet.com/ Spider for new search engine (by the DMOZ founders)
  309. array(
  310. 'name' => 'TinEye',
  311. 'useragent' => 'TinEye'), // http://tineye.com/crawler.html Spider for search engine startup
  312. array(
  313. 'name' => 'Twiceler',
  314. 'useragent' => 'twiceler'), // http://www.cuill.com/twiceler/robot.html Experimental Spider, (aggressive)
  315.  
  316. // Software Website Description
  317. array(
  318. 'name' => 'GSiteCrawler',
  319. 'useragent' => 'GSiteCrawler'), // http://www.gsitecrawler.com/ Windows Based Sitemap Generator Software
  320. array(
  321. 'name' => 'HTTrack',
  322. 'useragent' => 'HTTrack'), // http://www.httrack.com HTTrack Website Copier - Offline Browser
  323. array(
  324. 'name' => 'Wget',
  325. 'useragent' => 'Wget'), // http://www.gnu.org/software/wget/ GNU software to retrieve files
  326. // Reason for detecting these: They can be very intensive. So seeing them in use, enables you to block if necessary.
  327.  
  328. );
  329.  
  330.  
  331. // Grab all the existing spiders to match against
  332. $query = $db->query("SELECT useragent
  333. FROM ".TABLE_PREFIX."spiders");
  334. /* $db->query('', '
  335. SELECT useragent
  336. FROM mybb_spiders',
  337. array()
  338. );
  339. */
  340. $knownspiders = $db->fetch_array($query);
  341. // $a = mysql_num_rows($request);
  342. /* if (db_num_rows($request) != "false")
  343. {
  344. // Store all found spiders in an array
  345. while ($row = $db->fetch_array($request))
  346. $knownspiders[] = $row['useragent'];
  347. }
  348. */
  349. // Now go through spider in the db
  350. foreach($addSpider as $spider)
  351. {
  352. // If doesn't already exist in the table, then add it
  353. if(!in_array($spider['useragent'], $knownspiders))
  354. {
  355. // Now add each spider
  356. $spiders_group = array(
  357. 'name' => addslashes($spider['name']),
  358. 'useragent' => addslashes($spider['useragent']),
  359. );
  360. $db->insert_query("spiders", $spiders_group);
  361. }
  362. }
  363.  
  364. //Unset everything
  365. unset($addSpider, $spider, $knownspiders);
  366.  
  367. }
  368. function addbot()
  369. {
  370. }
  371. function addbot_deactivate()
  372. {
  373. global $db;
  374. $addSpider = array(
  375. // The big spiders detected by MyBB by default
  376. // array('Google', 'Googlebot'),
  377. // array('Msn', 'MsnBot'),
  378. // array('Yahoo!', 'Slurp'),
  379. // MAJOR spiders
  380. array(
  381. 'name' => 'Ask',
  382. 'useragent' => 'Teoma'),
  383. array(
  384. 'name' => 'Baidu',
  385. 'useragent' => 'Baiduspider'),
  386. array(
  387. 'name' => 'GigaBot',
  388. 'useragent' => 'Gigabot'),
  389. array(
  390. 'name' => 'Google-AdSense',
  391. 'useragent' => 'Mediapartners-Google'),
  392. array(
  393. 'name' => 'Google-Adwords',
  394. 'useragent' => 'AdsBot-Google'),
  395. array(
  396. 'name' => 'Google-SA',
  397. 'useragent' => 'gsa-crawler'),
  398. array(
  399. 'name' => 'Google-Image',
  400. 'useragent' => 'Googlebot-Image'),
  401. array(
  402. 'name' => 'InternetArchive',
  403. 'useragent' => 'ia_archiver-web.archive.org'),
  404. array(
  405. 'name' => 'Alexa',
  406. 'useragent' => 'ia_archiver'),
  407. array(
  408. 'name' => 'Omgili',
  409. 'useragent' => 'omgilibot'),
  410. array(
  411. 'name' => 'Speedy Spider',
  412. 'useragent' => 'Speedy Spider'),
  413. array(
  414. 'name' => 'Yahoo',
  415. 'useragent' => 'yahoo'),
  416. array(
  417. 'name' => 'Yahoo JP',
  418. 'useragent' => 'Y!J'),
  419. // Checkers/Testers/Robots
  420. array(
  421. 'name' => 'DeadLinksChecker',
  422. 'useragent' => 'link validator'),
  423. array(
  424. 'name' => 'W3C Validator',
  425. 'useragent' => 'W3C_Validator'),
  426. array(
  427. 'name' => 'W3C CSSValidator',
  428. 'useragent' => 'W3C_CSS_Validator'),
  429. array(
  430. 'name' => 'W3C FeedValidator',
  431. 'useragent' => 'FeedValidator'),
  432. array(
  433. 'name' => 'W3C LinkChecker',
  434. 'useragent' => 'W3C-checklink'),
  435. array(
  436. 'name' => 'W3C mobileOK',
  437. 'useragent' => 'W3C-mobileOK'),
  438. array(
  439. 'name' => 'W3C P3PValidator',
  440. 'useragent' => 'P3P Validator'),
  441.  
  442. // Feed readers
  443. array(
  444. 'name' => 'Bloglines',
  445. 'useragent' => 'Bloglines'),
  446. array(
  447. 'name' => 'Feedburner',
  448. 'useragent' => 'Feedburner'),
  449. // Website Thumbnail/Snapshot/Thumbshot takers
  450. array(
  451. 'name' => 'SnapBot',
  452. 'useragent' => 'Snapbot'),
  453. array(
  454. 'name' => 'Picsearch',
  455. 'useragent' => 'psbot'),
  456. array(
  457. 'name' => 'Websnapr',
  458. 'useragent' => 'Websnapr'),
  459. // More MINOR Spiders/Robots
  460. array(
  461. 'name' => 'AllTheWeb',
  462. 'useragent' => 'FAST-WebCrawler'),
  463. array(
  464. 'name' => 'Altavista',
  465. 'useragent' => 'Scooter'),
  466. array(
  467. 'name' => 'Asterias',
  468. 'useragent' => 'asterias'),
  469. array(
  470. 'name' => '192bot',
  471. 'useragent' => '192.comAgent'),
  472. array(
  473. 'name' => 'AbachoBot',
  474. 'useragent' => 'ABACHOBot'),
  475. array(
  476. 'name' => 'Abdcatos',
  477. 'useragent' => 'abcdatos'),
  478. array(
  479. 'name' => 'Acoon',
  480. 'useragent' => 'Acoon'),
  481. array(
  482. 'name' => 'Accoona',
  483. 'useragent' => 'Accoona'),
  484. array(
  485. 'name' => 'BecomeBot',
  486. 'useragent' => 'BecomeBot'),
  487. array(
  488. 'name' => 'BlogRefsBot',
  489. 'useragent' => 'BlogRefsBot'),
  490. array(
  491. 'name' => 'Daumoa',
  492. 'useragent' => 'Daumoa'),
  493. array(
  494. 'name' => 'DuckDuckBot',
  495. 'useragent' => 'DuckDuckBot'),
  496. array(
  497. 'name' => 'Exabot',
  498. 'useragent' => 'Exabot'),
  499. array(
  500. 'name' => 'Furl',
  501. 'useragent' => 'Furlbot'),
  502. array(
  503. 'name' => 'FyperSpider',
  504. 'useragent' => 'FyberSpider'),
  505. array(
  506. 'name' => 'Geona',
  507. 'useragent' => 'GeonaBot'),
  508. array(
  509. 'name' => 'GirafaBot',
  510. 'useragent' => 'Girafabot'),
  511. array(
  512. 'name' => 'GoSeeBot',
  513. 'useragent' => 'GoSeeBot'),
  514. array(
  515. 'name' => 'Ichiro',
  516. 'useragent' => 'ichiro'),
  517. array(
  518. 'name' => 'LapozzBot',
  519. 'useragent' => 'LapozzBot'),
  520. array(
  521. 'name' => 'Looksmart',
  522. 'useragent' => 'WISENutbot'),
  523. array(
  524. 'name' => 'Lycos',
  525. 'useragent' => 'Lycos_Spider'),
  526. array(
  527. 'name' => 'Majestic12',
  528. 'useragent' => 'MJ12bot/v2'),
  529. array(
  530. 'name' => 'MLBot',
  531. 'useragent' => 'MLBot'),
  532. array(
  533. 'name' => 'MSRBOT',
  534. 'useragent' => 'msrbot'),
  535. array(
  536. 'name' => 'MSR-ISRCCrawler',
  537. 'useragent' => 'MSR-ISRCCrawler'),
  538. array(
  539. 'name' => 'Naver',
  540. 'useragent' => 'NaverBot'),
  541. array(
  542. 'name' => 'Naver',
  543. 'useragent' => 'Yeti'),
  544. array(
  545. 'name' => 'NoxTrumBot',
  546. 'useragent' => 'noxtrumbot'),
  547. array(
  548. 'name' => 'OmniExplorer',
  549. 'useragent' => 'OmniExplorer_Bot'),
  550. array(
  551. 'name' => 'OnetSzukaj',
  552. 'useragent' => 'OnetSzukaj'),
  553. array(
  554. 'name' => 'ScrubTheWeb',
  555. 'useragent' => 'Scrubby'),
  556. array(
  557. 'name' => 'SearchSight',
  558. 'useragent' => 'SearchSight'),
  559. array(
  560. 'name' => 'Seeqpod',
  561. 'useragent' => 'Seeqpod'),
  562. array(
  563. 'name' => 'Shablast',
  564. 'useragent' => 'ShablastBot'),
  565. array(
  566. 'name' => 'SitiDiBot',
  567. 'useragent' => 'SitiDiBot'),
  568. array(
  569. 'name' => 'Slider',
  570. 'useragent' => 'silk/1.0'),
  571. array(
  572. 'name' => 'Sogou',
  573. 'useragent' => 'Sogou'),
  574. array(
  575. 'name' => 'Sosospider',
  576. 'useragent' => 'Sosospider'),
  577. array(
  578. 'name' => 'StackRambler',
  579. 'useragent' => 'StackRambler'),
  580. array(
  581. 'name' => 'SurveyBot',
  582. 'useragent' => 'SurveyBot'),
  583. array(
  584. 'name' => 'Touche',
  585. 'useragent' => 'Touche'),
  586. array(
  587. 'name' => 'Walhello',
  588. 'useragent' => 'appie'),
  589. array(
  590. 'name' => 'WebAlta',
  591. 'useragent' => 'WebAlta'),
  592. array(
  593. 'name' => 'Wisponbot',
  594. 'useragent' => 'wisponbot'),
  595. array(
  596. 'name' => 'YacyBot',
  597. 'useragent' => 'yacybot'),
  598. array(
  599. 'name' => 'YodaoBot',
  600. 'useragent' => 'YodaoBot'),
  601. // Google-Wanna-Be's - Spiders/Robots for Startups
  602. array(
  603. 'name' => 'Charlotte',
  604. 'useragent' => 'Charlotte'),
  605. array(
  606. 'name' => 'DiscoBot',
  607. 'useragent' => 'DiscoBot'),
  608. array(
  609. 'name' => 'EnaBot',
  610. 'useragent' => 'EnaBot'),
  611. array(
  612. 'name' => 'Gaisbot',
  613. 'useragent' => 'Gaisbot'),
  614. array(
  615. 'name' => 'Kalooga',
  616. 'useragent' => 'kalooga'),
  617. array(
  618. 'name' => 'ScoutJet',
  619. 'useragent' => 'ScoutJet'),
  620. array(
  621. 'name' => 'TinEye',
  622. 'useragent' => 'TinEye'),
  623. array(
  624. 'name' => 'Twiceler',
  625. 'useragent' => 'twiceler'),
  626. // Software
  627. array(
  628. 'name' => 'GSiteCrawler',
  629. 'useragent' => 'GSiteCrawler'),
  630. array(
  631. 'name' => 'HTTrack',
  632. 'useragent' => 'HTTrack'),
  633. array(
  634. 'name' => 'Wget',
  635. 'useragent' => 'Wget'),
  636. // Reason for detecting these: They can be very intensive. So seeing them in use, enables you to block if necessary.
  637.  
  638. );
  639.  
  640.  
  641. // Now go through spider in the mo
  642. foreach($addSpider as $spider)
  643. {
  644.  
  645.  
  646. // Now delete each spider
  647. $spiders_group = array(
  648.  
  649. 'name' => addslashes($spider['name']),
  650.  
  651. );
  652. $db->write_query("DELETE FROM ".TABLE_PREFIX."spiders WHERE name='".$spiders_group['name']."'");
  653.  
  654.  
  655. }
  656. }
  657.  
  658. ?>
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement
Public Pastes
Advertisement
We use cookies for various purposes including analytics. By continuing to use Pastebin, you agree to our use of cookies as described in the Cookies Policy.  OK, I Understand
Not a member of Pastebin yet?
Sign Up, it unlocks many cool features!