lib/ExternalReferrer.php

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