SearchSqlite.php

Go to the documentation of this file.

<?php
class SearchSqlite extends SearchEngine {

        protected $db;

        function __construct( $db ) {
                parent::__construct( $db );
        }

        function fulltextSearchSupported() {
                return $this->db->checkForEnabledSearch();
        }

        function parseQuery( $filteredText, $fulltext ) {
                global $wgContLang;
                $lc = SearchEngine::legalSearchChars(); // Minus format chars
                $searchon = '';
                $this->searchTerms = array();

                $m = array();
                if( preg_match_all( '/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/',
                          $filteredText, $m, PREG_SET_ORDER ) ) {
                        foreach( $m as $bits ) {
                                @list( /* all */, $modifier, $term, $nonQuoted, $wildcard ) = $bits;

                                if( $nonQuoted != '' ) {
                                        $term = $nonQuoted;
                                        $quote = '';
                                } else {
                                        $term = str_replace( '"', '', $term );
                                        $quote = '"';
                                }

                                if( $searchon !== '' ) {
                                        $searchon .= ' ';
                                }

                                // Some languages such as Serbian store the input form in the search index,
                                // so we may need to search for matches in multiple writing system variants.
                                $convertedVariants = $wgContLang->autoConvertToAllVariants( $term );
                                if( is_array( $convertedVariants ) ) {
                                        $variants = array_unique( array_values( $convertedVariants ) );
                                } else {
                                        $variants = array( $term );
                                }

                                // The low-level search index does some processing on input to work
                                // around problems with minimum lengths and encoding in MySQL's
                                // fulltext engine.
                                // For Chinese this also inserts spaces between adjacent Han characters.
                                $strippedVariants = array_map(
                                        array( $wgContLang, 'normalizeForSearch' ),
                                        $variants );

                                // Some languages such as Chinese force all variants to a canonical
                                // form when stripping to the low-level search index, so to be sure
                                // let's check our variants list for unique items after stripping.
                                $strippedVariants = array_unique( $strippedVariants );

                                $searchon .= $modifier;
                                if( count( $strippedVariants) > 1 )
                                        $searchon .= '(';
                                foreach( $strippedVariants as $stripped ) {
                                        if( $nonQuoted && strpos( $stripped, ' ' ) !== false ) {
                                                // Hack for Chinese: we need to toss in quotes for
                                                // multiple-character phrases since normalizeForSearch()
                                                // added spaces between them to make word breaks.
                                                $stripped = '"' . trim( $stripped ) . '"';
                                        }
                                        $searchon .= "$quote$stripped$quote$wildcard ";
                                }
                                if( count( $strippedVariants) > 1 )
                                        $searchon .= ')';

                                // Match individual terms or quoted phrase in result highlighting...
                                // Note that variants will be introduced in a later stage for highlighting!
                                $regexp = $this->regexTerm( $term, $wildcard );
                                $this->searchTerms[] = $regexp;
                        }

                } else {
                        wfDebug( __METHOD__ . ": Can't understand search query '{$filteredText}'\n" );
                }

                $searchon = $this->db->strencode( $searchon );
                $field = $this->getIndexField( $fulltext );
                return " $field MATCH '$searchon' ";
        }

        function regexTerm( $string, $wildcard ) {
                global $wgContLang;

                $regex = preg_quote( $string, '/' );
                if( $wgContLang->hasWordBreaks() ) {
                        if( $wildcard ) {
                                // Don't cut off the final bit!
                                $regex = "\b$regex";
                        } else {
                                $regex = "\b$regex\b";
                        }
                } else {
                        // For Chinese, words may legitimately abut other words in the text literal.
                        // Don't add \b boundary checks... note this could cause false positives
                        // for latin chars.
                }
                return $regex;
        }

        public static function legalSearchChars() {
                return "\"*" . parent::legalSearchChars();
        }

        function searchText( $term ) {
                return $this->searchInternal( $term, true );
        }

        function searchTitle( $term ) {
                return $this->searchInternal( $term, false );
        }

        protected function searchInternal( $term, $fulltext ) {
                global $wgCountTotalSearchHits, $wgContLang;

                if ( !$this->fulltextSearchSupported() ) {
                        return null;
                }

                $filteredTerm = $this->filter( $wgContLang->lc( $term ) );
                $resultSet = $this->db->query( $this->getQuery( $filteredTerm, $fulltext ) );

                $total = null;
                if( $wgCountTotalSearchHits ) {
                        $totalResult = $this->db->query( $this->getCountQuery( $filteredTerm, $fulltext ) );
                        $row = $totalResult->fetchObject();
                        if( $row ) {
                                $total = intval( $row->c );
                        }
                        $totalResult->free();
                }

                return new SqliteSearchResultSet( $resultSet, $this->searchTerms, $total );
        }


        function queryRedirect() {
                if( $this->showRedirects ) {
                        return '';
                } else {
                        return 'AND page_is_redirect=0';
                }
        }

        function queryNamespaces() {
                if( is_null($this->namespaces) )
                        return '';  # search all
                if ( !count( $this->namespaces ) ) {
                        $namespaces = '0';
                } else {
                        $namespaces = $this->db->makeList( $this->namespaces );
                }
                return 'AND page_namespace IN (' . $namespaces . ')';
        }

        function limitResult( $sql ) {
                return $this->db->limitResult( $sql, $this->limit, $this->offset );
        }

        function getQuery( $filteredTerm, $fulltext ) {
                return $this->limitResult(
                        $this->queryMain( $filteredTerm, $fulltext ) . ' ' .
                        $this->queryRedirect() . ' ' .
                        $this->queryNamespaces()
                );
        }

        function getIndexField( $fulltext ) {
                return $fulltext ? 'si_text' : 'si_title';
        }

        function queryMain( $filteredTerm, $fulltext ) {
                $match = $this->parseQuery( $filteredTerm, $fulltext );
                $page        = $this->db->tableName( 'page' );
                $searchindex = $this->db->tableName( 'searchindex' );
                return "SELECT $searchindex.rowid, page_namespace, page_title " .
                        "FROM $page,$searchindex " .
                        "WHERE page_id=$searchindex.rowid AND $match";
        }

        function getCountQuery( $filteredTerm, $fulltext ) {
                $match = $this->parseQuery( $filteredTerm, $fulltext );
                $page        = $this->db->tableName( 'page' );
                $searchindex = $this->db->tableName( 'searchindex' );
                return "SELECT COUNT(*) AS c " .
                        "FROM $page,$searchindex " .
                        "WHERE page_id=$searchindex.rowid AND $match" .
                        $this->queryRedirect() . ' ' .
                        $this->queryNamespaces();
        }

        function update( $id, $title, $text ) {
                if ( !$this->fulltextSearchSupported() ) {
                        return;
                }
                // @todo: find a method to do it in a single request,
                // couldn't do it so far due to typelessness of FTS3 tables.
                $dbw = wfGetDB( DB_MASTER );

                $dbw->delete( 'searchindex', array( 'rowid' => $id ), __METHOD__ );

                $dbw->insert( 'searchindex',
                        array(
                                'rowid' => $id,
                                'si_title' => $title,
                                'si_text' => $text
                        ), __METHOD__ );
        }

    function updateTitle( $id, $title ) {
                if ( !$this->fulltextSearchSupported() ) {
                        return;
                }
                $dbw = wfGetDB( DB_MASTER );

                $dbw->update( 'searchindex',
                        array( 'si_title' => $title ),
                        array( 'rowid'  => $id ),
                        __METHOD__ );
        }
}

class SqliteSearchResultSet extends SqlSearchResultSet {
        function __construct( $resultSet, $terms, $totalHits=null ) {
                parent::__construct( $resultSet, $terms );
                $this->mTotalHits = $totalHits;
        }

        function getTotalHits() {
                return $this->mTotalHits;
        }
}