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                 # Eliminate Blanks at start
  51                 $term = ereg_replace('[[:blank:]]*', '', $term);
  52
  53                 # Exact match? No need to look further.
  54                 $title = Title::newFromText( $term );
  55                 if ( $title->getNamespace() == NS_SPECIAL || 0 != $title->getArticleID() ) {
  56                         return $title;
  57                 }
  58
  59                 # Now try all lower case (i.e. first letter capitalized)
  60                 #
  61                 $title = Title::newFromText( strtolower( $term ) );
  62                 if ( 0 != $title->getArticleID() ) {
  63                         return $title;
  64                 }
  65
  66                 # Now try capitalized string
  67                 #
  68                 $title = Title::newFromText( ucwords( strtolower( $term ) ) );
  69                 if ( 0 != $title->getArticleID() ) {
  70                         return $title;
  71                 }
  72
  73                 # Now try all upper case
  74                 #
  75                 $title = Title::newFromText( strtoupper( $term ) );
  76                 if ( 0 != $title->getArticleID() ) {
  77                         return $title;
  78                 }
  79
  80                 # Entering an IP address goes to the contributions page
  81                 if ( preg_match( '/^(user:)?\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}$/', strtolower($term) ) ) {
  82                         $title = Title::makeTitle( NS_SPECIAL, "Contributions/" . $term );
  83                         return $title;
  84                 }
  85
  86                 # Entering a user goes to the user page whether it's there or not
  87                 if ( preg_match( '/^user:/', strtolower($term) ) ) {
  88                         if (User::idFromName($term)) {
  89                           $title = Title::newFromURL( $term );
  90                           return $title;
  91                         }
  92                 }
  93
  94                 return NULL;
  95         }
  96
  97         function legalSearchChars() {
  98                 return "A-Za-z_'0-9\\x80-\\xFF\\-";
  99         }
 100
 101         /**
 102          * Set the maximum number of results to return
 103          * and how many to skip before returning the first.
 104          *
 105          * @param int $limit
 106          * @param int $offset
 107          * @access public
 108          */
 109         function setLimitOffset( $limit, $offset = 0 ) {
 110                 $this->limit = IntVal( $limit );
 111                 $this->offset = IntVal( $offset );
 112         }
 113
 114         /**
 115          * Set which namespaces the search should include.
 116          * Give an array of namespace index numbers.
 117          *
 118          * @param array $namespaces
 119          * @access public
 120          */
 121         function setNamespaces( $namespaces ) {
 122                 $this->namespaces = $namespaces;
 123         }
 124
 125         /**
 126          * Make a list of searchable namespaces and their canonical names.
 127          * @return array
 128          * @access public
 129          */
 130         function searchableNamespaces() {
 131                 global $wgContLang;
 132                 $arr = array();
 133                 foreach( $wgContLang->getNamespaces() as $ns => $name ) {
 134                         if( $ns >= NS_MAIN ) {
 135                                 $arr[$ns] = $name;
 136                         }
 137                 }
 138                 return $arr;
 139         }
 140
 141         /**
 142          * Fetch an array of regular expression fragments for matching
 143          * the search terms as parsed by this engine in a text extract.
 144          *
 145          * @return array
 146          * @access public
 147          */
 148         function termMatches() {
 149                 return $this->searchTerms;
 150         }
 151
 152         /**
 153          * Return a 'cleaned up' search string
 154          *
 155          * @return string
 156          * @access public
 157          */
 158         function filter( $text ) {
 159                 $lc = $this->legalSearchChars();
 160                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 161         }
 162
 163         /**
 164          * Return a partial WHERE clause to exclude redirects, if so set
 165          * @return string
 166          * @access private
 167          */
 168         function queryRedirect() {
 169                 if( $this->showRedirects ) {
 170                         return 'AND cur_is_redirect=0';
 171                 } else {
 172                         return '';
 173                 }
 174         }
 175
 176         /**
 177          * Return a partial WHERE clause to limit the search to the given namespaces
 178          * @return string
 179          * @access private
 180          */
 181         function queryNamespaces() {
 182                 $namespaces = implode( ',', $this->namespaces );
 183                 if ($namespaces == '') {
 184                         $namespaces = '0';
 185                 }
 186                 return 'AND page_namespace IN (' . $namespaces . ')';
 187         }
 188
 189         /**
 190          * Return a LIMIT clause to limit results on the query.
 191          * @return string
 192          * @access private
 193          */
 194         function queryLimit() {
 195                 return $this->db->limitResult( $this->limit, $this->offset );
 196         }
 197
 198         /**
 199          * Does not do anything for generic search engine
 200          * subclasses may define this though
 201          * @return string
 202          * @access private
 203          */
 204         function queryRanking($filteredTerm,$fulltext) {
 205                 return "";
 206         }
 207
 208         /**
 209          * Construct the full SQL query to do the search.
 210          * The guts shoulds be constructed in queryMain()
 211          * @param string $filteredTerm
 212          * @param bool $fulltext
 213          * @access private
 214          */
 215         function getQuery( $filteredTerm, $fulltext ) {
 216                 return $this->queryMain( $filteredTerm, $fulltext ) . ' ' .
 217                         $this->queryRedirect() . ' ' .
 218                         $this->queryNamespaces() . ' ' .
 219                         $this->queryRanking($filteredTerm, $fulltext) . ' ' .
 220                         $this->queryLimit();
 221         }
 222
 223         /**
 224          * Load up the appropriate search engine class for the currently
 225            * active database backend, and return a configured instance.
 226            *
 227            * @return SearchEngine
 228            * @access private
 229         */
 230         function create() {
 231                 global $wgDBtype, $wgDBmysql4, $wgSearchType;
 232                 if( $wgDBtype == 'mysql' ) {
 233                         if( $wgDBmysql4 ) {
 234                                 $class = 'SearchMySQL4';
 235                                 require_once( 'SearchMySQL4.php' );
 236                         } else {
 237                                 $class = 'SearchMysql3';
 238                                 require_once( 'SearchMySQL3.php' );
 239                         }
 240                 } else if ( $wgDBtype == 'PostgreSQL' ) {
 241                         $class = 'SearchTsearch2';
 242                         require_once( 'SearchTsearch2.php' );
 243                 } else {
 244                         $class = 'SearchEngineDummy';
 245                 }
 246                 $search = new $class( wfGetDB( DB_SLAVE ) );
 247                 $search->setLimitOffset(0,0);
 248                 return $search;
 249         }
 250
 251
 252 }
 253
 254 /**
 255  * @package MediaWiki
 256  */
 257 class SearchEngineDummy {
 258         function search( $term ) {
 259                 return null;
 260         }
 261 }
 262