includes/SearchEngine.php

   1 <?php
   2 /**
   3  * Contain a class for special pages
   4  * @package MediaWiki
   5  * @subpackage Search
   6  */
   7
   8 /**
   9  * @package MediaWiki
  10  */
  11 class SearchEngine {
  12         var $limit = 10;
  13         var $offset = 0;
  14         var $searchTerms = array();
  15         var $namespaces = array( 0 );
  16         var $showRedirects = false;
  17
  18         /**
  19          * Perform a full text search query and return a result set.
  20          *
  21          * @param string $term - Raw search term
  22          * @param array $namespaces - List of namespaces to search
  23          * @return ResultWrapper
  24          * @access public
  25          */
  26         function searchText( $term ) {
  27                 return $this->db->resultObject( $this->db->query( $this->getQuery( $this->filter( $term ), true ) ) );
  28         }
  29
  30         /**
  31          * Perform a title-only search query and return a result set.
  32          *
  33          * @param string $term - Raw search term
  34          * @param array $namespaces - List of namespaces to search
  35          * @return ResultWrapper
  36          * @access public
  37          */
  38         function searchTitle( $term ) {
  39                 return $this->db->resultObject( $this->db->query( $this->getQuery( $this->filter( $term ), false ) ) );
  40         }
  41
  42         /**
  43          * If an exact title match can be find, or a very slightly close match,
  44          * return the title. If no match, returns NULL.
  45          *
  46          * @static
  47          * @param string $term
  48          * @return Title
  49          * @access private
  50          */
  51         function getNearMatch( $term ) {
  52                 # Exact match? No need to look further.
  53                 $title = Title::newFromText( $term );
  54                 if ( $title->getNamespace() == NS_SPECIAL || 0 != $title->getArticleID() ) {
  55                         return $title;
  56                 }
  57
  58                 # Now try all lower case (i.e. first letter capitalized)
  59                 #
  60                 $title = Title::newFromText( strtolower( $term ) );
  61                 if ( 0 != $title->getArticleID() ) {
  62                         return $title;
  63                 }
  64
  65                 # Now try capitalized string
  66                 #
  67                 $title = Title::newFromText( ucwords( strtolower( $term ) ) );
  68                 if ( 0 != $title->getArticleID() ) {
  69                         return $title;
  70                 }
  71
  72                 # Now try all upper case
  73                 #
  74                 $title = Title::newFromText( strtoupper( $term ) );
  75                 if ( 0 != $title->getArticleID() ) {
  76                         return $title;
  77                 }
  78
  79                 $title = Title::newFromText( $term );
  80
  81                 # Entering an IP address goes to the contributions page
  82                 if ( ( $title->getNamespace() == NS_USER && User::isIP($title->getText() ) )
  83                         || User::isIP( trim( $term ) ) ) {
  84                         return Title::makeTitle( NS_SPECIAL, "Contributions/" . $title->getDbkey() );
  85                 }
  86
  87
  88                 # Entering a user goes to the user page whether it's there or not
  89                 if ( $title->getNamespace() == NS_USER ) {
  90                         return $title;
  91                 }
  92
  93                 # Quoted term? Try without the quotes...
  94                 if( preg_match( '/^"([^"]+)"$/', $term, $matches ) ) {
  95                         return SearchEngine::getNearMatch( $matches[1] );
  96                 }
  97
  98                 return NULL;
  99         }
 100
 101         function legalSearchChars() {
 102                 return "A-Za-z_'0-9\\x80-\\xFF\\-";
 103         }
 104
 105         /**
 106          * Set the maximum number of results to return
 107          * and how many to skip before returning the first.
 108          *
 109          * @param int $limit
 110          * @param int $offset
 111          * @access public
 112          */
 113         function setLimitOffset( $limit, $offset = 0 ) {
 114                 $this->limit = IntVal( $limit );
 115                 $this->offset = IntVal( $offset );
 116         }
 117
 118         /**
 119          * Set which namespaces the search should include.
 120          * Give an array of namespace index numbers.
 121          *
 122          * @param array $namespaces
 123          * @access public
 124          */
 125         function setNamespaces( $namespaces ) {
 126                 $this->namespaces = $namespaces;
 127         }
 128
 129         /**
 130          * Make a list of searchable namespaces and their canonical names.
 131          * @return array
 132          * @access public
 133          */
 134         function searchableNamespaces() {
 135                 global $wgContLang;
 136                 $arr = array();
 137                 foreach( $wgContLang->getNamespaces() as $ns => $name ) {
 138                         if( $ns >= NS_MAIN ) {
 139                                 $arr[$ns] = $name;
 140                         }
 141                 }
 142                 return $arr;
 143         }
 144
 145         /**
 146          * Fetch an array of regular expression fragments for matching
 147          * the search terms as parsed by this engine in a text extract.
 148          *
 149          * @return array
 150          * @access public
 151          */
 152         function termMatches() {
 153                 return $this->searchTerms;
 154         }
 155
 156         /**
 157          * Return a 'cleaned up' search string
 158          *
 159          * @return string
 160          * @access public
 161          */
 162         function filter( $text ) {
 163                 $lc = $this->legalSearchChars();
 164                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 165         }
 166
 167         /**
 168          * Return a partial WHERE clause to exclude redirects, if so set
 169          * @return string
 170          * @access private
 171          */
 172         function queryRedirect() {
 173                 if( $this->showRedirects ) {
 174                         return 'AND cur_is_redirect=0';
 175                 } else {
 176                         return '';
 177                 }
 178         }
 179
 180         /**
 181          * Return a partial WHERE clause to limit the search to the given namespaces
 182          * @return string
 183          * @access private
 184          */
 185         function queryNamespaces() {
 186                 $namespaces = implode( ',', $this->namespaces );
 187                 if ($namespaces == '') {
 188                         $namespaces = '0';
 189                 }
 190                 return 'AND page_namespace IN (' . $namespaces . ')';
 191         }
 192
 193         /**
 194          * Return a LIMIT clause to limit results on the query.
 195          * @return string
 196          * @access private
 197          */
 198         function queryLimit() {
 199                 return $this->db->limitResult( $this->limit, $this->offset );
 200         }
 201
 202         /**
 203          * Does not do anything for generic search engine
 204          * subclasses may define this though
 205          * @return string
 206          * @access private
 207          */
 208         function queryRanking($filteredTerm,$fulltext) {
 209                 return "";
 210         }
 211
 212         /**
 213          * Construct the full SQL query to do the search.
 214          * The guts shoulds be constructed in queryMain()
 215          * @param string $filteredTerm
 216          * @param bool $fulltext
 217          * @access private
 218          */
 219         function getQuery( $filteredTerm, $fulltext ) {
 220                 return $this->queryMain( $filteredTerm, $fulltext ) . ' ' .
 221                         $this->queryRedirect() . ' ' .
 222                         $this->queryNamespaces() . ' ' .
 223                         $this->queryRanking($filteredTerm, $fulltext) . ' ' .
 224                         $this->queryLimit();
 225         }
 226
 227         /**
 228          * Load up the appropriate search engine class for the currently
 229            * active database backend, and return a configured instance.
 230            *
 231            * @return SearchEngine
 232            * @access private
 233         */
 234         function create() {
 235                 global $wgDBtype, $wgDBmysql4, $wgSearchType;
 236                 if( $wgDBtype == 'mysql' ) {
 237                         if( $wgDBmysql4 ) {
 238                                 $class = 'SearchMySQL4';
 239                                 require_once( 'SearchMySQL4.php' );
 240                         } else {
 241                                 $class = 'SearchMysql3';
 242                                 require_once( 'SearchMySQL3.php' );
 243                         }
 244                 } else if ( $wgDBtype == 'PostgreSQL' ) {
 245                         $class = 'SearchTsearch2';
 246                         require_once( 'SearchTsearch2.php' );
 247                 } else {
 248                         $class = 'SearchEngineDummy';
 249                 }
 250                 $search = new $class( wfGetDB( DB_SLAVE ) );
 251                 $search->setLimitOffset(0,0);
 252                 return $search;
 253         }
 254
 255
 256 }
 257
 258 /**
 259  * @package MediaWiki
 260  */
 261 class SearchEngineDummy {
 262         function search( $term ) {
 263                 return null;
 264         }
 265 }
 266