SearchPostgres.php

Go to the documentation of this file.

<?php
use MediaWiki\MediaWikiServices;
use MediaWiki\Revision\SlotRecord;
use Wikimedia\Rdbms\IDatabase;
 
class SearchPostgres extends SearchDatabase {
    protected function doSearchTitleInDB( $term ) {
        $q = $this->searchQuery( $term, 'titlevector' );
        $olderror = error_reporting( E_ERROR );
        $dbr = $this->lb->getConnectionRef( DB_REPLICA );
        $resultSet = $dbr->query( $q, 'SearchPostgres', IDatabase::QUERY_SILENCE_ERRORS );
        error_reporting( $olderror );
        return new SqlSearchResultSet( $resultSet, $this->searchTerms );
    }
 
    protected function doSearchTextInDB( $term ) {
        $q = $this->searchQuery( $term, 'textvector' );
        $olderror = error_reporting( E_ERROR );
        $dbr = $this->lb->getConnectionRef( DB_REPLICA );
        $resultSet = $dbr->query( $q, 'SearchPostgres', IDatabase::QUERY_SILENCE_ERRORS );
        error_reporting( $olderror );
        return new SqlSearchResultSet( $resultSet, $this->searchTerms );
    }
 
    private function parseQuery( $term ) {
        wfDebug( "parseQuery received: $term" );
 
        // No backslashes allowed
        $term = preg_replace( '/\\\/', '', $term );
 
        // Collapse parens into nearby words:
        $term = preg_replace( '/\s*\‍(\s*/', ' (', $term );
        $term = preg_replace( '/\s*\‍)\s*/', ') ', $term );
 
        // Treat colons as word separators:
        $term = preg_replace( '/:/', ' ', $term );
 
        $searchstring = '';
        $m = [];
        if ( preg_match_all( '/([-!]?)(\S+)\s*/', $term, $m, PREG_SET_ORDER ) ) {
            foreach ( $m as $terms ) {
                if ( strlen( $terms[1] ) ) {
                    $searchstring .= ' & !';
                }
                if ( strtolower( $terms[2] ) === 'and' ) {
                    $searchstring .= ' & ';
                } elseif ( strtolower( $terms[2] ) === 'or' || $terms[2] === '|' ) {
                    $searchstring .= ' | ';
                } elseif ( strtolower( $terms[2] ) === 'not' ) {
                    $searchstring .= ' & !';
                } else {
                    $searchstring .= " & $terms[2]";
                }
            }
        }
 
        // Strip out leading junk
        $searchstring = preg_replace( '/^[\s\&\|]+/', '', $searchstring );
 
        // Remove any doubled-up operators
        $searchstring = preg_replace( '/([\!\&\|]) +(?:[\&\|] +)+/', "$1 ", $searchstring );
 
        // Remove any non-spaced operators (e.g. "Zounds!")
        $searchstring = preg_replace( '/([^ ])[\!\&\|]/', "$1", $searchstring );
 
        // Remove any trailing whitespace or operators
        $searchstring = preg_replace( '/[\s\!\&\|]+$/', '', $searchstring );
 
        // Remove unnecessary quotes around everything
        $searchstring = preg_replace( '/^[\'"](.*)[\'"]$/', "$1", $searchstring );
 
        // Quote the whole thing
        $dbr = $this->lb->getConnectionRef( DB_REPLICA );
        $searchstring = $dbr->addQuotes( $searchstring );
 
        wfDebug( "parseQuery returned: $searchstring" );
 
        return $searchstring;
    }
 
    private function searchQuery( $term, $fulltext ) {
        # Get the SQL fragment for the given term
        $searchstring = $this->parseQuery( $term );
 
        // We need a separate query here so gin does not complain about empty searches
        $sql = "SELECT to_tsquery($searchstring)";
        $dbr = $this->lb->getConnectionRef( DB_REPLICA );
        $res = $dbr->query( $sql, __METHOD__ );
        if ( !$res ) {
            // TODO: Better output (example to catch: one 'two)
            die( "Sorry, that was not a valid search string. Please go back and try again" );
        }
        $top = $res->fetchRow()[0];
 
        $this->searchTerms = [];
        $slotRoleStore = MediaWikiServices::getInstance()->getSlotRoleStore();
        if ( $top === "" ) { // e.g. if only stopwords are used XXX return something better
            $query = "SELECT page_id, page_namespace, page_title, 0 AS score " .
                "FROM page p, revision r, slots s, content c, \"text\" pc " .
                "WHERE p.page_latest = r.rev_id " .
                "AND s.slot_revision_id = r.rev_id " .
                "AND s.slot_role_id = " . $slotRoleStore->getId( SlotRecord::MAIN ) . " " .
                "AND c.content_id = s.slot_content_id " .
                "AND pc.old_id = substring( c.content_address from '^tt:([0-9]+)$' )::int " .
                "AND 1=0";
        } else {
            $m = [];
            if ( preg_match_all( "/'([^']+)'/", $top, $m, PREG_SET_ORDER ) ) {
                foreach ( $m as $terms ) {
                    $this->searchTerms[$terms[1]] = $terms[1];
                }
            }
 
            $query = "SELECT page_id, page_namespace, page_title, " .
                "ts_rank($fulltext, to_tsquery($searchstring), 5) AS score " .
                "FROM page p, revision r, slots s, content c, \"text\" pc " .
                "WHERE p.page_latest = r.rev_id " .
                "AND s.slot_revision_id = r.rev_id " .
                "AND s.slot_role_id = " . $slotRoleStore->getId( SlotRecord::MAIN ) . " " .
                "AND c.content_id = s.slot_content_id " .
                "AND pc.old_id = substring( c.content_address from '^tt:([0-9]+)$' )::int " .
                "AND $fulltext @@ to_tsquery($searchstring)";
        }
        // Namespaces - defaults to main
        if ( $this->namespaces !== null ) { // null -> search all
            if ( count( $this->namespaces ) < 1 ) {
                $query .= ' AND page_namespace = ' . NS_MAIN;
            } else {
                $namespaces = $dbr->makeList( $this->namespaces );
                $query .= " AND page_namespace IN ($namespaces)";
            }
        }
 
        $query .= " ORDER BY score DESC, page_id DESC";
 
        $query .= $dbr->limitResult( '', $this->limit, $this->offset );
 
        wfDebug( "searchQuery returned: $query" );
 
        return $query;
    }
 
    // Most of the work of these two functions are done automatically via triggers
 
    public function update( $pageid, $title, $text ) {
        // We don't want to index older revisions
        $slotRoleStore = MediaWikiServices::getInstance()->getSlotRoleStore();
        $sql = "UPDATE \"text\" SET textvector = NULL " .
            "WHERE textvector IS NOT NULL " .
            "AND old_id IN " .
            "(SELECT DISTINCT substring( c.content_address from '^tt:([0-9]+)$' )::int AS old_rev_text_id " .
            " FROM content c, slots s, revision r " .
            " WHERE r.rev_page = $pageid " .
            " AND s.slot_revision_id = r.rev_id " .
            " AND s.slot_role_id = " . $slotRoleStore->getId( SlotRecord::MAIN ) . " " .
            " AND c.content_id = s.slot_content_id " .
            " ORDER BY old_rev_text_id DESC OFFSET 1)";
 
        $dbw = $this->lb->getConnectionRef( DB_PRIMARY );
        $dbw->query( $sql, __METHOD__ );
 
        return true;
    }
 
    public function updateTitle( $id, $title ) {
        return true;
    }
}