SearchUpdate.php

Go to the documentation of this file.

<?php
use MediaWiki\MediaWikiServices;

class SearchUpdate implements DeferrableUpdate {
    private $id = 0;

    private $title;

    private $content;

    private $page;

    public function __construct( $id, $title, $c = null ) {
        if ( is_string( $title ) ) {
            wfDeprecated( __METHOD__ . " with a string for the title", '1.34' );
            $this->title = Title::newFromText( $title );
            if ( $this->title === null ) {
                throw new InvalidArgumentException( "Cannot construct the title: $title" );
            }
        } else {
            $this->title = $title;
        }

        $this->id = $id;
        // is_string() check is back-compat for ApprovedRevs
        if ( is_string( $c ) ) {
            wfDeprecated( __METHOD__ . " with a string for the content", '1.34' );
            $c = new TextContent( $c );
        } elseif ( is_bool( $c ) ) {
            wfDeprecated( __METHOD__ . " with a boolean for the content", '1.34' );
            $c = null;
        }
        $this->content = $c;
    }

    public function doUpdate() {
        $services = MediaWikiServices::getInstance();
        $config = $services->getSearchEngineConfig();

        if ( $config->getConfig()->get( 'DisableSearchUpdate' ) || !$this->id ) {
            return;
        }

        $seFactory = $services->getSearchEngineFactory();
        foreach ( $config->getSearchTypes() as $type ) {
            $search = $seFactory->create( $type );
            if ( !$search->supports( 'search-update' ) ) {
                continue;
            }

            $normalTitle = $this->getNormalizedTitle( $search );

            if ( $this->getLatestPage() === null ) {
                $search->delete( $this->id, $normalTitle );
                continue;
            } elseif ( $this->content === null ) {
                $search->updateTitle( $this->id, $normalTitle );
                continue;
            }

            $text = $this->content !== null ? $this->content->getTextForSearchIndex() : '';
            $text = $this->updateText( $text, $search );

            # Perform the actual update
            $search->update( $this->id, $normalTitle, $search->normalizeText( $text ) );
        }
    }

    public function updateText( $text, SearchEngine $se = null ) {
        $services = MediaWikiServices::getInstance();
        $contLang = $services->getContentLanguage();
        # Language-specific strip/conversion
        $text = $contLang->normalizeForSearch( $text );
        $se = $se ?: $services->newSearchEngine();
        $lc = $se->legalSearchChars() . '&#;';

        # Strip HTML markup
        $text = preg_replace( "/<\\/?\\s*[A-Za-z][^>]*?>/",
            ' ', $contLang->lc( " " . $text . " " ) );
        $text = preg_replace( "/(^|\\n)==\\s*([^\\n]+)\\s*==(\\s)/sD",
            "\\1\\2 \\2 \\2\\3", $text ); # Emphasize headings

        # Strip external URLs
        $uc = "A-Za-z0-9_\\/:.,~%\\-+&;#?!=()@\\x80-\\xFF";
        $protos = "http|https|ftp|mailto|news|gopher";
        $pat = "/(^|[^\\[])({$protos}):[{$uc}]+([^{$uc}]|$)/";
        $text = preg_replace( $pat, "\\1 \\3", $text );

        $p1 = "/([^\\[])\\[({$protos}):[{$uc}]+]/";
        $p2 = "/([^\\[])\\[({$protos}):[{$uc}]+\\s+([^\\]]+)]/";
        $text = preg_replace( $p1, "\\1 ", $text );
        $text = preg_replace( $p2, "\\1 \\3 ", $text );

        # Internal image links
        $pat2 = "/\\[\\[image:([{$uc}]+)\\.(gif|png|jpg|jpeg)([^{$uc}])/i";
        $text = preg_replace( $pat2, " \\1 \\3", $text );

        $text = preg_replace( "/([^{$lc}])([{$lc}]+)]]([a-z]+)/",
            "\\1\\2 \\2\\3", $text ); # Handle [[game]]s

        # Strip all remaining non-search characters
        $text = preg_replace( "/[^{$lc}]+/", " ", $text );

        $text = strrev( preg_replace( "/ s'([{$lc}]+)/", " s'\\1 \\1", strrev( $text ) ) );
        $text = strrev( preg_replace( "/ 's([{$lc}]+)/", " s\\1", strrev( $text ) ) );

        # Strip wiki '' and '''
        $text = preg_replace( "/''[']*/", " ", $text );

        return $text;
    }

    private function getLatestPage() {
        if ( !isset( $this->page ) ) {
            $this->page = WikiPage::newFromID( $this->id, WikiPage::READ_LATEST );
        }

        return $this->page;
    }

    private function getNormalizedTitle( SearchEngine $search ) {
        $contLang = MediaWikiServices::getInstance()->getContentLanguage();
        $ns = $this->title->getNamespace();
        $title = $this->title->getText();

        $lc = $search->legalSearchChars() . '&#;';
        $t = $contLang->normalizeForSearch( $title );
        $t = preg_replace( "/[^{$lc}]+/", ' ', $t );
        $t = $contLang->lc( $t );

        # Handle 's, s'
        $t = preg_replace( "/([{$lc}]+)'s( |$)/", "\\1 \\1's ", $t );
        $t = preg_replace( "/([{$lc}]+)s'( |$)/", "\\1s ", $t );

        $t = preg_replace( "/\\s+/", ' ', $t );

        if ( $ns == NS_FILE ) {
            $t = preg_replace( "/ (png|gif|jpg|jpeg|ogg)$/", "", $t );
        }

        return $search->normalizeText( trim( $t ) );
    }
}