lib/ExternalReferrer.php

   1 <?php
   2
   3
   4 /**
   5  * Detect external referrers
   6  * Currently only search engines, and highlight the searched item.
   7  *
   8  * Todo:
   9  *   store all external referrers in (rotatable) log/db for a RecentReferrers plugin.
  10  */
  11 if (!function_exists('isExternalReferrer')) { // also defined in stdlib.php
  12   function isExternalReferrer(&$request) {
  13     if ($referrer = $request->get('HTTP_REFERER')) {
  14         $home = SCRIPT_NAME; // was SERVER_URL, check sister wiki's: same host but other other script url
  15         if (substr(strtolower($referrer),0,strlen($home)) == strtolower($home)) return false;
  16         require_once 'lib/ExternalReferrer.php';
  17         $se = new SearchEngines();
  18         return $se->parseSearchQuery($referrer);
  19     }
  20     return false;
  21   }
  22 }
  23
  24 class SearchEngines {
  25
  26     var $searchEngines =
  27     array(
  28           "search.sli.sympatico.ca/" => array("engine" => "Sympatico", "query1" => "query=", "query2" => "", "url" => "http://www1.sympatico.ca/"),
  29           "www.search123.com/cgi-bin/" => array("engine" => "Search123", "query1" => "query=", "query2" => "", "url" => "http://www.search123.com/"),
  30           "search.dogpile.com" => array("engine" => "Dogpile", "query1" => "q=", "query2" => "", "url" => "http://www.dogpile.com"),
  31           "vivisimo." => array("engine" => "Vivisimo", "query1" => "query=", "query2" => "", "url" => "http://www.vivisimo.com"),
  32           "webindex.sanook.com" => array("engine" => "Sanook", "query1" => "d1=", "query2" => "", "url" => "http://www.sanook.com/"),
  33           "tiscali.cz/search" => array("engine" => "JANAS", "query1" => "query=", "query2" => "", "url" => "http://www.tiscali.cz/"),
  34           "teoma.com/gs?" => array("engine" => "Teoma", "query1" => "terms=", "query2" => "", "url" => "http://www.teoma.com/"),
  35           "redbox." => array("engine" => "RedBox", "query1" => "srch=", "query2" => "", "url" => "http://www.redbox.cz/"),
  36           "globetrotter.net" => array("engine" => "Telus Network - Globetrotter.net", "query1" => "string=", "query2" => "", "url" => "http://www.globetrotter.net/"),
  37           "myto.com" => array("engine" => "Telus Network - myTO.com", "query1" => "string=", "query2" => "", "url" => "http://www.myto.com/"),
  38           "alberta.com" => array("engine" => "Telus Network - Alberta.com", "query1" => "string=", "query2" => "", "url" => "http://www.alberta.com/"),
  39           "mybc.com" => array("engine" => "Telus Network - myBC.com", "query1" => "string=", "query2" => "", "url" => "http://www.mybc.com/"),
  40           "monstercrawler." => array("engine" => "MonsterCrawler", "query1" => "qry=", "query2" => "", "url" => "http://www.monstercrawler.com/"),
  41           "allthesites." => array("engine" => "All the Sites", "query1" => "query=", "query2" => "", "url" => "http://www.althesites.com/"),
  42           "suche.web" => array("engine" => "Web.de", "query1" => "su=", "query2" => "", "url" => "http://www.web.de/"),
  43           "rediff." => array("engine" => "reDiff", "query1" => "MT=", "query2" => "", "url" => "http://www.rediff.com/"),
  44           "evreka." => array("engine" => "Evreka", "query1" => "q=", "query2" => "", "url" => "http://evreka.suomi24.fi/"),
  45           "findia." => array("engine" => "Findia", "query1" => "query=", "query2" => "", "url" => "http://www.findia.net/"),
  46           "av.yahoo" => array("engine" => "Yahoo", "query1" => "p=", "query2" => "", "url" => "http://www.yahoo.com/"),
  47           "google.yahoo" => array("engine" => "Yahoo", "query1" => "p=", "query2" => "", "url" => "http://www.yahoo.com/"),
  48           "yahoo." => array("engine" => "Yahoo", "query1" => "q=", "query2" => "", "url" => "http://www.yahoo.com/"),
  49           "aol." => array("engine" => "AOL Search", "query1" => "query=", "query2" => "", "url" => "http://search.aol.com/"),
  50           "about." => array("engine" => "About", "query1" => "terms=", "query2" => "", "url" => "http://www.about.com/"),
  51           "altavista." => array("engine" => "Altavista", "query1" => "q=", "query2" => "", "url" => "http://www.altavista.com/"),
  52           "directhit." => array("engine" => "DirectHit", "query1" => "qry=", "query2" => "", "url" => "http://www.directhit.com/"),
  53           "lk=webcrawler" => array("engine" => "Webcrawler", "query1" => "s=", "query2" => "", "url" => "http://www.webcrawler.com/"),
  54           "excite." => array("engine" => "Excite", "query1" => "search=", "query2" => "", "url" => "http://www.excite.com/"),
  55           "alltheweb." => array("engine" => "All the Web", "query1" => "query=", "query2" => "q=", "url" => "http://www.alltheweb.com/"),
  56           "netscape." => array("engine" => "Netscape", "query1" => "search=", "query2" => "", "url" => "http://search.netscape.com/"),
  57           "google." => array("engine" => "Google", "query1" => "q=", "query2" => "query=", "url" => "http://www.google.com/"),
  58           "?partner=go_home" => array("engine" => "Infoseek/Go", "query1" => "Keywords=", "query2" => "", "url" => "http://www.go.com/"),
  59           "nbci." => array("engine" => "NBCi", "query1" => "Keywords=", "query2" => "", "url" => "http://www.nbci.com/"),
  60           "goto." => array("engine" => "GoTo", "query1" => "Keywords=", "query2" => "", "url" => "http://www.goto.com/"),
  61           "hotbot." => array("engine" => "HotBot", "query1" => "MT=", "query2" => "", "url" => "http://hotbot.lycos.com/"),
  62           "iwon." => array("engine" => "IWon", "query1" => "searchfor=", "query2" => "", "url" => "http://home.iwon.com/index_gen.html"),
  63           "looksmart." => array("engine" => "Looksmart", "query1" => "key=", "query2" => "", "url" => "http://www.looksmart.com/"),
  64           "lycos." => array("engine" => "Lycos", "query1" => "query=", "query2" => "", "url" => "http://www.lycos.com/"),
  65           "msn." => array("engine" => "MSN", "query1" => "q=", "query2" => "", "url" => "http://search.msn.com/"),
  66           "dmoz." => array("engine" => "Dmoz", "query1" => "search=", "query2" => "", "url" => "http://www.dmoz.org/"),
  67
  68           );
  69
  70     /**
  71      * parseSearchQuery(url)
  72      * Parses the passed refering url looking for search engine data.  If search info is found,
  73      * the method determines the name of the search engine, it's URL, and the search keywords
  74      * used in the search. This information is returned in an associative array with the following
  75      * keys:
  76      * @returns array engine, engine_url, query
  77      * @public
  78      */
  79     function parseSearchQuery($url) {
  80         // test local referrers
  81         if (DEBUG & _DEBUG_REMOTE) {
  82             $this->searchEngines[strtolower(SERVER_URL)] = array("engine" => "DEBUG", "query1" => "s=", "query2" => "", "url" => SCRIPT_NAME);
  83             $this->searchEngines['http://localhost'] = array("engine" => "DEBUG", "query1" => "s=", "query2" => "", "url" => SCRIPT_NAME);
  84         }
  85         $url = strtolower($url);
  86         $ref = $url;
  87         while (list($key,$var) = @each($this->searchEngines)) {
  88             if (strstr($ref, $key)) {
  89                 unset($ref);
  90                 $ref["engine"] = $var["engine"];
  91                 $query1 =  $var["query1"];
  92                 $query2 =  $var["query2"];
  93                 $ref["engine_url"] = $var["url"];
  94                 break;
  95             }
  96         }
  97         reset($this->searchEngines);
  98         if ($ref == $url) {
  99             return false;
 100         }
 101         $url = @parse_url(strtolower($url));
 102         if (!empty($url["query"]))
 103             $url = $url["query"];
 104         if ($query1 and @stristr($url, $query1)) {
 105              $query = @explode($query1, $url);
 106         }
 107         else if ($query2 and @stristr($url, $query2)) {
 108             $query = explode($query2, $url);
 109         }
 110         if (!empty($query)) {
 111             $query = @explode("&", $query[1]);
 112             $ref["query"] = @urldecode($query[0]);
 113         }
 114         return $ref;
 115     }
 116 }
 117
 118 // Local Variables:
 119 // mode: php
 120 // tab-width: 8
 121 // c-basic-offset: 4
 122 // c-hanging-comment-ender-p: nil
 123 // indent-tabs-mode: nil
 124 // End: