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