SearchPostgres.php

Go to the documentation of this file.

<?php
class SearchPostgres extends SearchEngine {

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

    function searchTitle( $term ) {
        $q = $this->searchQuery( $term, 'titlevector', 'page_title' );
        $olderror = error_reporting( E_ERROR );
        $resultSet = $this->db->resultObject( $this->db->query( $q, 'SearchPostgres', true ) );
        error_reporting( $olderror );
        if ( !$resultSet ) {
            // Needed for "Query requires full scan, GIN doesn't support it"
            return new SearchResultTooMany();
        }
        return new PostgresSearchResultSet( $resultSet, $this->searchTerms );
    }

    function searchText( $term ) {
        $q = $this->searchQuery( $term, 'textvector', 'old_text' );
        $olderror = error_reporting( E_ERROR );
        $resultSet = $this->db->resultObject( $this->db->query( $q, 'SearchPostgres', true ) );
        error_reporting( $olderror );
        if ( !$resultSet ) {
            return new SearchResultTooMany();
        }
        return new PostgresSearchResultSet( $resultSet, $this->searchTerms );
    }

    function parseQuery( $term ) {

        wfDebug( "parseQuery received: $term \n" );

        ## 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 = array();
        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' 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
        $searchstring = $this->db->addQuotes( $searchstring );

        wfDebug( "parseQuery returned: $searchstring \n" );

        return $searchstring;

    }

    function searchQuery( $term, $fulltext, $colname ) {
        # 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)";
        $res = $this->db->query( $SQL );
        if ( !$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();
        $top = $top[0];

        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, pagecontent c WHERE p.page_latest = r.rev_id " .
                "AND r.rev_text_id = c.old_id AND 1=0";
        }
        else {
            $m = array();
            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, pagecontent c WHERE p.page_latest = r.rev_id " .
            "AND r.rev_text_id = c.old_id AND $fulltext @@ to_tsquery($searchstring)";
        }

        ## Redirects
        if ( !$this->showRedirects ) {
            $query .= ' AND page_is_redirect = 0';
        }

        ## Namespaces - defaults to 0
        if ( !is_null( $this->namespaces ) ) { // null -> search all
            if ( count( $this->namespaces ) < 1 ) {
                $query .= ' AND page_namespace = 0';
            } else {
                $namespaces = $this->db->makeList( $this->namespaces );
                $query .= " AND page_namespace IN ($namespaces)";
            }
        }

        $query .= " ORDER BY score DESC, page_id DESC";

        $query .= $this->db->limitResult( '', $this->limit, $this->offset );

        wfDebug( "searchQuery returned: $query \n" );

        return $query;
    }

    ## Most of the work of these two functions are done automatically via triggers

    function update( $pageid, $title, $text ) {
        ## We don't want to index older revisions
        $SQL = "UPDATE pagecontent SET textvector = NULL WHERE old_id IN " .
                "(SELECT rev_text_id FROM revision WHERE rev_page = " . intval( $pageid ) .
                " ORDER BY rev_text_id DESC OFFSET 1)";
        $this->db->query( $SQL );
        return true;
    }

    function updateTitle( $id, $title ) {
        return true;
    }

} ## end of the SearchPostgres class

class PostgresSearchResult extends SearchResult {
    function __construct( $row ) {
        parent::__construct( $row );
        $this->score = $row->score;
    }
    function getScore() {
        return $this->score;
    }
}

class PostgresSearchResultSet extends SqlSearchResultSet {
    function __construct( $resultSet, $terms ) {
        parent::__construct( $resultSet, $terms );
    }

    function next() {
        $row = $this->mResultSet->fetchObject();
        if ( $row === false ) {
            return false;
        } else {
            return new PostgresSearchResult( $row );
        }
    }
}