includes/SearchEngine.php

   1 <?php
   2 /**
   3  * Contain a class for special pages
   4  * @package MediaWiki
   5  */
   6
   7 /**
   8  * @package MediaWiki
   9  */
  10 class SearchEngine {
  11         var $limit = 10;
  12         var $offset = 0;
  13         var $searchTerms = array();
  14         var $namespaces = array( 0 );
  15         var $showRedirects = false;
  16
  17         /**
  18          * Perform a full text search query and return a result set.
  19          *
  20          * @param string $term - Raw search term
  21          * @param array $namespaces - List of namespaces to search
  22          * @return ResultWrapper
  23          * @access public
  24          */
  25         function searchText( $term ) {
  26                 return $this->db->resultObject( $this->db->query( $this->getQuery( $this->filter( $term ), true ) ) );
  27         }
  28
  29         /**
  30          * Perform a title-only search query and return a result set.
  31          *
  32          * @param string $term - Raw search term
  33          * @param array $namespaces - List of namespaces to search
  34          * @return ResultWrapper
  35          * @access public
  36          */
  37         function searchTitle( $term ) {
  38                 return $this->db->resultObject( $this->db->query( $this->getQuery( $this->filter( $term ), false ) ) );
  39         }
  40
  41         /**
  42          * If an exact title match can be find, or a very slightly close match,
  43          * return the title. If no match, returns NULL.
  44          *
  45          * @param string $term
  46          * @return Title
  47          * @access private
  48          */
  49         function getNearMatch( $term ) {
  50                 # Exact match? No need to look further.
  51                 $title = Title::newFromText( $term );
  52                 if ( $title->getNamespace() == NS_SPECIAL || 0 != $title->getArticleID() ) {
  53                         return $title;
  54                 }
  55
  56                 # Now try all lower case (i.e. first letter capitalized)
  57                 #
  58                 $title = Title::newFromText( strtolower( $term ) );
  59                 if ( 0 != $title->getArticleID() ) {
  60                         return $title;
  61                 }
  62
  63                 # Now try capitalized string
  64                 #
  65                 $title = Title::newFromText( ucwords( strtolower( $term ) ) );
  66                 if ( 0 != $title->getArticleID() ) {
  67                         return $title;
  68                 }
  69
  70                 # Now try all upper case
  71                 #
  72                 $title = Title::newFromText( strtoupper( $term ) );
  73                 if ( 0 != $title->getArticleID() ) {
  74                         return $title;
  75                 }
  76
  77                 # Entering an IP address goes to the contributions page
  78                 if ( preg_match( '/^\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}$/', $term ) ) {
  79                         $title = Title::makeTitle( NS_SPECIAL, "Contributions/" . $term );
  80                         return $title;
  81                 }
  82
  83                 return NULL;
  84         }
  85
  86         function legalSearchChars() {
  87                 return "A-Za-z_'0-9\\x80-\\xFF\\-";
  88         }
  89
  90         /**
  91          * Set the maximum number of results to return
  92          * and how many to skip before returning the first.
  93          *
  94          * @param int $limit
  95          * @param int $offset
  96          * @access public
  97          */
  98         function setLimitOffset( $limit, $offset = 0 ) {
  99                 $this->limit = IntVal( $limit );
 100                 $this->offset = IntVal( $offset );
 101         }
 102
 103         /**
 104          * Set which namespaces the search should include.
 105          * Give an array of namespace index numbers.
 106          *
 107          * @param array $namespaces
 108          * @access public
 109          */
 110         function setNamespaces( $namespaces ) {
 111                 $this->namespaces = $namespaces;
 112         }
 113
 114         /**
 115          * Make a list of searchable namespaces and their canonical names.
 116          * @return array
 117          * @access public
 118          */
 119         function searchableNamespaces() {
 120                 global $wgContLang;
 121                 $arr = array();
 122                 foreach( $wgContLang->getNamespaces() as $ns => $name ) {
 123                         if( $ns >= NS_MAIN ) {
 124                                 $arr[$ns] = $name;
 125                         }
 126                 }
 127                 return $arr;
 128         }
 129
 130         /**
 131          * Fetch an array of regular expression fragments for matching
 132          * the search terms as parsed by this engine in a text extract.
 133          *
 134          * @return array
 135          * @access public
 136          */
 137         function termMatches() {
 138                 return $this->searchTerms;
 139         }
 140
 141         /**
 142          * Return a 'cleaned up' search string
 143          *
 144          * @return string
 145          * @access public
 146          */
 147         function filter( $text ) {
 148                 $lc = $this->legalSearchChars();
 149                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 150         }
 151
 152         /**
 153          * Return a partial WHERE clause to exclude redirects, if so set
 154          * @return string
 155          * @access private
 156          */
 157         function queryRedirect() {
 158                 if( $this->showRedirects ) {
 159                         return 'AND cur_is_redirect=0';
 160                 } else {
 161                         return '';
 162                 }
 163         }
 164
 165         /**
 166          * Return a partial WHERE clause to limit the search to the given namespaces
 167          * @return string
 168          * @access private
 169          */
 170         function queryNamespaces() {
 171                 $namespaces = implode( ',', $this->namespaces );
 172                 if ($namespaces == '') {
 173                         $namespaces = '0';
 174                 }
 175                 return 'AND page_namespace IN (' . $namespaces . ')';
 176         }
 177
 178         /**
 179          * Return a LIMIT clause to limit results on the query.
 180          * @return string
 181          * @access private
 182          */
 183         function queryLimit() {
 184                 return $this->db->limitResult( $this->limit, $this->offset );
 185         }
 186
 187         /**
 188          * Does not do anything for generic search engine
 189          * subclasses may define this though
 190          * @return string
 191          * @access private
 192          */
 193         function queryRanking($filteredTerm,$fulltext) {
 194                 return "";
 195         }
 196
 197         /**
 198          * Construct the full SQL query to do the search.
 199          * The guts shoulds be constructed in queryMain()
 200          * @param string $filteredTerm
 201          * @param bool $fulltext
 202          * @access private
 203          */
 204         function getQuery( $filteredTerm, $fulltext ) {
 205                 return $this->queryMain( $filteredTerm, $fulltext ) . ' ' .
 206                         $this->queryRedirect() . ' ' .
 207                         $this->queryNamespaces() . ' ' .
 208                         $this->queryRanking($filteredTerm, $fulltext) . ' ' .
 209                         $this->queryLimit();
 210         }
 211
 212         /**
 213          * Load up the appropriate search engine class for the currently
 214            * active database backend, and return a configured instance.
 215            *
 216            * @return SearchEngine
 217            * @access private
 218         */
 219         function create() {
 220                 global $wgDBtype, $wgDBmysql4, $wgSearchType;
 221                 if( $wgDBtype == 'mysql' ) {
 222                         if( $wgDBmysql4 ) {
 223                                 $class = 'SearchMySQL4';
 224                                 require_once( 'SearchMySQL4.php' );
 225                         } else {
 226                                 $class = 'SearchMysql3';
 227                                 require_once( 'SearchMySQL3.php' );
 228                         }
 229                 } else if ( $wgDBtype == 'PostgreSQL' ) {
 230                         $class = 'SearchTsearch2';
 231                         require_once( 'SearchTsearch2.php' );
 232                 } else {
 233                         $class = 'SearchEngineDummy';
 234                 }
 235                 $search = new $class( wfGetDB( DB_SLAVE ) );
 236                 $search->setLimitOffset(0,0);
 237                 return $search;
 238         }
 239
 240
 241 }
 242
 243 /**
 244  * @package MediaWiki
 245  */
 246 class SearchEngineDummy {
 247         function search( $term ) {
 248                 return null;
 249         }
 250 }
 251