MediaWiki  REL1_23
SearchMssql.php
Go to the documentation of this file.
00001 <?php
00028 class SearchMssql extends SearchDatabase {
00036     function searchText( $term ) {
00037         $resultSet = $this->db->resultObject( $this->db->query( $this->getQuery( $this->filter( $term ), true ) ) );
00038         return new MssqlSearchResultSet( $resultSet, $this->searchTerms );
00039     }
00040 
00048     function searchTitle( $term ) {
00049         $resultSet = $this->db->resultObject( $this->db->query( $this->getQuery( $this->filter( $term ), false ) ) );
00050         return new MssqlSearchResultSet( $resultSet, $this->searchTerms );
00051     }
00052 
00059     function queryNamespaces() {
00060         $namespaces = implode( ',', $this->namespaces );
00061         if ( $namespaces == '' ) {
00062             $namespaces = '0';
00063         }
00064         return 'AND page_namespace IN (' . $namespaces . ')';
00065     }
00066 
00074     function queryLimit( $sql ) {
00075         return $this->db->limitResult( $sql, $this->limit, $this->offset );
00076     }
00077 
00084     function queryRanking( $filteredTerm, $fulltext ) {
00085         return ' ORDER BY ftindex.[RANK] DESC'; // return ' ORDER BY score(1)';
00086     }
00087 
00096     function getQuery( $filteredTerm, $fulltext ) {
00097         return $this->queryLimit( $this->queryMain( $filteredTerm, $fulltext ) . ' ' .
00098             $this->queryNamespaces() . ' ' .
00099             $this->queryRanking( $filteredTerm, $fulltext ) . ' ' );
00100     }
00101 
00108     function getIndexField( $fulltext ) {
00109         return $fulltext ? 'si_text' : 'si_title';
00110     }
00111 
00120     function queryMain( $filteredTerm, $fulltext ) {
00121         $match = $this->parseQuery( $filteredTerm, $fulltext );
00122         $page = $this->db->tableName( 'page' );
00123         $searchindex = $this->db->tableName( 'searchindex' );
00124 
00125         return 'SELECT page_id, page_namespace, page_title, ftindex.[RANK]' .
00126             "FROM $page,FREETEXTTABLE($searchindex , $match, LANGUAGE 'English') as ftindex " .
00127             'WHERE page_id=ftindex.[KEY] ';
00128     }
00129 
00133     function parseQuery( $filteredText, $fulltext ) {
00134         global $wgContLang;
00135         $lc = SearchEngine::legalSearchChars();
00136         $this->searchTerms = array();
00137 
00138         # @todo FIXME: This doesn't handle parenthetical expressions.
00139         $m = array();
00140         $q = array();
00141 
00142         if ( preg_match_all( '/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/',
00143             $filteredText, $m, PREG_SET_ORDER ) ) {
00144             foreach ( $m as $terms ) {
00145                 $q[] = $terms[1] . $wgContLang->normalizeForSearch( $terms[2] );
00146 
00147                 if ( !empty( $terms[3] ) ) {
00148                     $regexp = preg_quote( $terms[3], '/' );
00149                     if ( $terms[4] ) {
00150                         $regexp .= "[0-9A-Za-z_]+";
00151                     }
00152                 } else {
00153                     $regexp = preg_quote( str_replace( '"', '', $terms[2] ), '/' );
00154                 }
00155                 $this->searchTerms[] = $regexp;
00156             }
00157         }
00158 
00159         $searchon = $this->db->strencode( join( ',', $q ) );
00160         $field = $this->getIndexField( $fulltext );
00161         return "$field, '$searchon'";
00162     }
00163 
00173     function update( $id, $title, $text ) {
00174         // We store the column data as UTF-8 byte order marked binary stream
00175         // because we are invoking the plain text IFilter on it so that, and we want it
00176         // to properly decode the stream as UTF-8.  SQL doesn't support UTF8 as a data type
00177         // but the indexer will correctly handle it by this method.  Since all we are doing
00178         // is passing this data to the indexer and never retrieving it via PHP, this will save space
00179         $table = $this->db->tableName( 'searchindex' );
00180         $utf8bom = '0xEFBBBF';
00181         $si_title = $utf8bom . bin2hex( $title );
00182         $si_text = $utf8bom . bin2hex( $text );
00183         $sql = "DELETE FROM $table WHERE si_page = $id;";
00184         $sql .= "INSERT INTO $table (si_page, si_title, si_text) VALUES ($id, $si_title, $si_text)";
00185         return $this->db->query( $sql, 'SearchMssql::update' );
00186     }
00187 
00196     function updateTitle( $id, $title ) {
00197         $table = $this->db->tableName( 'searchindex' );
00198 
00199         // see update for why we are using the utf8bom
00200         $utf8bom = '0xEFBBBF';
00201         $si_title = $utf8bom . bin2hex( $title );
00202         $sql = "DELETE FROM $table WHERE si_page = $id;";
00203         $sql .= "INSERT INTO $table (si_page, si_title, si_text) VALUES ($id, $si_title, 0x00)";
00204         return $this->db->query( $sql, 'SearchMssql::updateTitle' );
00205     }
00206 }
00207 
00211 class MssqlSearchResultSet extends SearchResultSet {
00212     function __construct( $resultSet, $terms ) {
00213         $this->mResultSet = $resultSet;
00214         $this->mTerms = $terms;
00215     }
00216 
00217     function termMatches() {
00218         return $this->mTerms;
00219     }
00220 
00221     function numRows() {
00222         return $this->mResultSet->numRows();
00223     }
00224 
00225     function next() {
00226         $row = $this->mResultSet->fetchObject();
00227         if ( $row === false ) {
00228             return false;
00229         }
00230         return new SearchResult( $row );
00231     }
00232 }