portal/patient/fwk/libs/verysimple/Search/Google.php

   1 <?php
   2 /** @package    verysimple::Search */
   3
   4 /**
   5  * include required files
   6  */
   7 require_once("SearchEngine.php");
   8 require_once('SOAP/Client.php'); // PEAR::SOAP::Client
   9
  10 /**
  11  * This is an implmentation of a SearchRank/SearchEngine for Google.
  12  * It provides
  13  * a convenient way to execute search queries, get results, ranking and inbound
  14  * links from google
  15  *
  16  * example
  17  * $google = new Google($API_KEY); // Google API Key must be obtained from Google
  18  *
  19  * $rank = $google->GetRank($url,$searchterm,$MAX_RESULTS);
  20  * $links = $google->GetLinks($url);
  21  *
  22  * @package verysimple::Search
  23  * @author VerySimple Inc.
  24  * @copyright 1997-2007 VerySimple, Inc.
  25  * @license http://www.gnu.org/licenses/lgpl.html LGPL
  26  * @version 1.0
  27  */
  28 class Google extends SearchEngine
  29 {
  30
  31     /**
  32      * Returns the inbound links for the given url
  33      *
  34      * @param string $url
  35      *          the url for which you want to find the rank
  36      * @param string $max
  37      *          the maximum number to return
  38      * @return SearchResult
  39      */
  40     function GetLinks($url, $max = 10)
  41     {
  42         return $this->DoSearch("link:" . $url, 0, $max);
  43     }
  44
  45     /**
  46      * Returns the estimated number of inbound links for the given url
  47      *
  48      * @param string $url
  49      *          the url for which you want to find the rank
  50      * @return int
  51      */
  52     function GetLinkCount($url)
  53     {
  54         return $this->GetLinks($url, 1)->estimatedTotalResultsCount;
  55     }
  56
  57     /**
  58      * Given a url and result returned by DoSearch, returns the rank that url appears on the page
  59      *
  60      * @param string $url
  61      *          the url for which you want to find the rank
  62      * @param string $query
  63      *          the search query
  64      * @param int $maxresults
  65      *          (default 30) the max number of results to search for.
  66      * @return SearchRank
  67      */
  68     function GetRank($url, $query, $maxresults = 30)
  69     {
  70         $current_page = 0;
  71         $result_counter = 0;
  72         $page_size = 10;
  73
  74         $rank = new SearchRank();
  75
  76         $rank->Query = $query;
  77
  78         while ($rank->Position == 0 && $result_counter <= $maxresults) {
  79             $current_page ++;
  80
  81             $result = $this->DoSearch($query, $result_counter, $page_size);
  82
  83             // only loop through as many results as we have
  84             if ($result->estimatedTotalResultsCount < $maxresults) {
  85                 $maxresults = $result->estimatedTotalResultsCount;
  86             }
  87
  88             if ($current_page == 1 && $result->estimatedTotalResultsCount > 0) {
  89                 $rank->EstimatedResults = $result->estimatedTotalResultsCount;
  90                 $rank->TopRankedUrl = $result->resultElements [0]->URL;
  91                 $rank->TopRankedSnippet = $result->resultElements [0]->snippet;
  92                 $rank->TopRankedTitle = $result->resultElements [0]->title;
  93             }
  94
  95             $cp_rank = $this->GetPositionOnPage($url, $result);
  96
  97             if ($cp_rank->Position) {
  98                 // we found a match
  99                 $rank->Page = $current_page;
 100                 $rank->Position = $cp_rank->Position * $current_page;
 101                 $rank->Title = $cp_rank->Title;
 102                 $rank->Snippet = $cp_rank->Snippet;
 103                 $rank->Url = $cp_rank->Url;
 104             }
 105
 106             $result_counter += $page_size;
 107             $result = null;
 108         }
 109
 110         return $rank;
 111     }
 112
 113     /**
 114      * Given a url and result returned by DoSearch, returns the rank that url appears on the page
 115      *
 116      * @param string $url
 117      * @param Result $result
 118      * @return SearchRank
 119      */
 120     function GetPositionOnPage($url, &$result)
 121     {
 122         $rank = new SearchRank();
 123         $counter = 0;
 124
 125         foreach ($result->resultElements as $element) {
 126             $counter ++;
 127             // print "<div>$url :: $counter = ".$element->URL."</div>";
 128             $normalizedurl = str_replace("/", "\\/", $url);
 129
 130             if ($rank->Position == 0 && preg_match("/$normalizedurl/i", $element->URL)) {
 131                 $rank->Position = ( int ) $counter;
 132                 $rank->Title = $element->title;
 133                 $rank->Snippet = $element->snippet;
 134                 $rank->Url = $element->URL;
 135             }
 136         }
 137
 138         return $rank;
 139     }
 140
 141     /**
 142      * Queries google for the specified search query
 143      *
 144      * @param string $query
 145      * @param int $start
 146      * @param int $max
 147      * @param bool $filter
 148      * @param bool $restrict
 149      * @param bool $safe
 150      * @param string $lr
 151      * @param string $ie
 152      * @param string $oe
 153      * @throws Exception
 154      */
 155     function DoSearch($query, $start = 0, $max = 10, $filter = false, $restrict = "", $safe = false, $lr = "", $ie = "", $oe = "")
 156     {
 157         $result = null;
 158         $continue = true;
 159         $counter = 0;
 160
 161         while ($continue) {
 162             $counter ++;
 163             try {
 164                 $result = $this->_googleSoap($query, $start, $max, $filter, $restrict, $safe, $lr, $ie, $oe);
 165                 $continue = false;
 166             } catch (exception $ex) {
 167                 $this->FailedRequests ++;
 168
 169                 if (preg_match("/Daily limit/i", $ex->getMessage()) || preg_match("/Invalid/i", $ex->getMessage())) {
 170                     throw $ex;
 171                 }
 172
 173                 if ($counter > 2) {
 174                     throw new Exception("Tried to contact Google API $counter times without success:" . $ex->getMessage());
 175                 }
 176             }
 177         }
 178
 179         return $result;
 180     }
 181
 182     /**
 183      * Makes SOAP request to google
 184      */
 185     private function _googleSoap($query, $start = 0, $max = 10, $filter = false, $restrict = "", $safe = false, $lr = "", $ie = "", $oe = "")
 186     {
 187         $wsdl = new SOAP_WSDL('http://api.google.com/GoogleSearch.wsdl');
 188         $soapclient = $wsdl->getProxy();
 189
 190         if (get_class($soapclient) != "SOAP_Fault") {
 191             $result = $soapclient->doGoogleSearch($this->Key, $query, $start, $max, $filter, $restrict, $safe, $lr, $ie, $oe);
 192         } else {
 193             throw new Exception("SOAP_Fault: " . $soapclient->message);
 194         }
 195
 196         if (PEAR::isError($result)) {
 197             throw new Exception("PEAR Exception: " . $result->message);
 198         } else {
 199             return $result;
 200         }
 201     }
 202 }