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          * @static
  46          * @param string $term
  47          * @return Title
  48          * @access private
  49          */
  50         function getNearMatch( $term ) {
  51                 # Exact match? No need to look further.
  52                 $title = Title::newFromText( $term );
  53                 if ( $title->getNamespace() == NS_SPECIAL || 0 != $title->getArticleID() ) {
  54                         return $title;
  55                 }
  56
  57                 # Now try all lower case (i.e. first letter capitalized)
  58                 #
  59                 $title = Title::newFromText( strtolower( $term ) );
  60                 if ( 0 != $title->getArticleID() ) {
  61                         return $title;
  62                 }
  63
  64                 # Now try capitalized string
  65                 #
  66                 $title = Title::newFromText( ucwords( strtolower( $term ) ) );
  67                 if ( 0 != $title->getArticleID() ) {
  68                         return $title;
  69                 }
  70
  71                 # Now try all upper case
  72                 #
  73                 $title = Title::newFromText( strtoupper( $term ) );
  74                 if ( 0 != $title->getArticleID() ) {
  75                         return $title;
  76                 }
  77
  78                 $title = Title::newFromText( $term );
  79
  80                 # Entering an IP address goes to the contributions page
  81                 if ( ( $title->getNamespace() == NS_USER && User::isIP($title->getText() ) )
  82                         || User::isIP( trim( $term ) ) ) {
  83                         return Title::makeTitle( NS_SPECIAL, "Contributions/" . $title->getDbkey() );
  84                 }
  85
  86
  87                 # Entering a user goes to the user page whether it's there or not
  88                 if ( $title->getNamespace() == NS_USER ) {
  89                         return $title;
  90                 }
  91
  92                 # Quoted term? Try without the quotes...
  93                 if( preg_match( '/^"([^"]+)"$/', $term, $matches ) ) {
  94                         return SearchEngine::getNearMatch( $matches[1] );
  95                 }
  96
  97                 return NULL;
  98         }
  99
 100         function legalSearchChars() {
 101                 return "A-Za-z_'0-9\\x80-\\xFF\\-";
 102         }
 103
 104         /**
 105          * Set the maximum number of results to return
 106          * and how many to skip before returning the first.
 107          *
 108          * @param int $limit
 109          * @param int $offset
 110          * @access public
 111          */
 112         function setLimitOffset( $limit, $offset = 0 ) {
 113                 $this->limit = IntVal( $limit );
 114                 $this->offset = IntVal( $offset );
 115         }
 116
 117         /**
 118          * Set which namespaces the search should include.
 119          * Give an array of namespace index numbers.
 120          *
 121          * @param array $namespaces
 122          * @access public
 123          */
 124         function setNamespaces( $namespaces ) {
 125                 $this->namespaces = $namespaces;
 126         }
 127
 128         /**
 129          * Make a list of searchable namespaces and their canonical names.
 130          * @return array
 131          * @access public
 132          */
 133         function searchableNamespaces() {
 134                 global $wgContLang;
 135                 $arr = array();
 136                 foreach( $wgContLang->getNamespaces() as $ns => $name ) {
 137                         if( $ns >= NS_MAIN ) {
 138                                 $arr[$ns] = $name;
 139                         }
 140                 }
 141                 return $arr;
 142         }
 143
 144         /**
 145          * Fetch an array of regular expression fragments for matching
 146          * the search terms as parsed by this engine in a text extract.
 147          *
 148          * @return array
 149          * @access public
 150          */
 151         function termMatches() {
 152                 return $this->searchTerms;
 153         }
 154
 155         /**
 156          * Return a 'cleaned up' search string
 157          *
 158          * @return string
 159          * @access public
 160          */
 161         function filter( $text ) {
 162                 $lc = $this->legalSearchChars();
 163                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 164         }
 165
 166         /**
 167          * Return a partial WHERE clause to exclude redirects, if so set
 168          * @return string
 169          * @access private
 170          */
 171         function queryRedirect() {
 172                 if( $this->showRedirects ) {
 173                         return 'AND cur_is_redirect=0';
 174                 } else {
 175                         return '';
 176                 }
 177         }
 178
 179         /**
 180          * Return a partial WHERE clause to limit the search to the given namespaces
 181          * @return string
 182          * @access private
 183          */
 184         function queryNamespaces() {
 185                 $namespaces = implode( ',', $this->namespaces );
 186                 if ($namespaces == '') {
 187                         $namespaces = '0';
 188                 }
 189                 return 'AND page_namespace IN (' . $namespaces . ')';
 190         }
 191
 192         /**
 193          * Return a LIMIT clause to limit results on the query.
 194          * @return string
 195          * @access private
 196          */
 197         function queryLimit() {
 198                 return $this->db->limitResult( $this->limit, $this->offset );
 199         }
 200
 201         /**
 202          * Does not do anything for generic search engine
 203          * subclasses may define this though
 204          * @return string
 205          * @access private
 206          */
 207         function queryRanking($filteredTerm,$fulltext) {
 208                 return "";
 209         }
 210
 211         /**
 212          * Construct the full SQL query to do the search.
 213          * The guts shoulds be constructed in queryMain()
 214          * @param string $filteredTerm
 215          * @param bool $fulltext
 216          * @access private
 217          */
 218         function getQuery( $filteredTerm, $fulltext ) {
 219                 return $this->queryMain( $filteredTerm, $fulltext ) . ' ' .
 220                         $this->queryRedirect() . ' ' .
 221                         $this->queryNamespaces() . ' ' .
 222                         $this->queryRanking($filteredTerm, $fulltext) . ' ' .
 223                         $this->queryLimit();
 224         }
 225
 226         /**
 227          * Load up the appropriate search engine class for the currently
 228            * active database backend, and return a configured instance.
 229            *
 230            * @return SearchEngine
 231            * @access private
 232         */
 233         function create() {
 234                 global $wgDBtype, $wgDBmysql4, $wgSearchType;
 235                 if( $wgDBtype == 'mysql' ) {
 236                         if( $wgDBmysql4 ) {
 237                                 $class = 'SearchMySQL4';
 238                                 require_once( 'SearchMySQL4.php' );
 239                         } else {
 240                                 $class = 'SearchMysql3';
 241                                 require_once( 'SearchMySQL3.php' );
 242                         }
 243                 } else if ( $wgDBtype == 'PostgreSQL' ) {
 244                         $class = 'SearchTsearch2';
 245                         require_once( 'SearchTsearch2.php' );
 246                 } else {
 247                         $class = 'SearchEngineDummy';
 248                 }
 249                 $search = new $class( wfGetDB( DB_SLAVE ) );
 250                 $search->setLimitOffset(0,0);
 251                 return $search;
 252         }
 253
 254
 255 }
 256
 257 /**
 258  * @package MediaWiki
 259  */
 260 class SearchEngineDummy {
 261         function search( $term ) {
 262                 return null;
 263         }
 264 }
 265