BadFileLookup.php

Go to the documentation of this file.

<?php
 
namespace MediaWiki;
 
use BagOStuff;
use MalformedTitleException;
use MediaWiki\HookContainer\HookContainer;
use MediaWiki\HookContainer\HookRunner;
use MediaWiki\Linker\LinkTarget;
use RepoGroup;
use TitleParser;
 
class BadFileLookup {
    private $listCallback;
 
    private $cache;
 
    private $repoGroup;
 
    private $titleParser;
 
    private $badFiles;
 
    private $hookRunner;
 
    public function __construct(
        callable $listCallback,
        BagOStuff $cache,
        RepoGroup $repoGroup,
        TitleParser $titleParser,
        HookContainer $hookContainer
    ) {
        $this->listCallback = $listCallback;
        $this->cache = $cache;
        $this->repoGroup = $repoGroup;
        $this->titleParser = $titleParser;
        $this->hookRunner = new HookRunner( $hookContainer );
    }
 
    public function isBadFile( $name, LinkTarget $contextTitle = null ) {
        // Handle redirects; callers almost always hit RepoGroup::findFile() anyway,
        // so just use that method because it has a fast process cache.
        $file = $this->repoGroup->findFile( $name );
        // XXX If we don't find the file we also don't replace spaces by underscores or otherwise
        // validate or normalize the title, is this right?
        if ( $file ) {
            $name = $file->getTitle()->getDBkey();
        }
 
        // Run the extension hook
        $bad = false;
        if ( !$this->hookRunner->onBadImage( $name, $bad ) ) {
            return (bool)$bad;
        }
 
        if ( $this->badFiles === null ) {
            $list = ( $this->listCallback )();
            $key = $this->cache->makeKey( 'bad-image-list', sha1( $list ) );
            $this->badFiles = $this->cache->getWithSetCallback(
                $key,
                BagOStuff::TTL_DAY,
                function () use ( $list ) {
                    return $this->buildBadFilesList( $list );
                }
            );
        }
 
        return isset( $this->badFiles[$name] ) && ( !$contextTitle ||
            !isset( $this->badFiles[$name][$contextTitle->getNamespace()][$contextTitle->getDBkey()] ) );
    }
 
    private function buildBadFilesList( string $list ): array {
        $ret = [];
        $lines = explode( "\n", $list );
        foreach ( $lines as $line ) {
            // List items only
            if ( substr( $line, 0, 1 ) !== '*' ) {
                continue;
            }
 
            // Find all links
            $m = [];
            // XXX What is the ':?' doing in the regex? Why not let the TitleParser strip it?
            if ( !preg_match_all( '/\[\[:?(.*?)\]\]/', $line, $m ) ) {
                continue;
            }
 
            $fileDBkey = null;
            $exceptions = [];
            foreach ( $m[1] as $i => $titleText ) {
                try {
                    $title = $this->titleParser->parseTitle( $titleText );
                } catch ( MalformedTitleException $e ) {
                    continue;
                }
                if ( $i == 0 ) {
                    $fileDBkey = $title->getDBkey();
                } else {
                    $exceptions[$title->getNamespace()][$title->getDBkey()] = true;
                }
            }
 
            if ( $fileDBkey !== null ) {
                $ret[$fileDBkey] = $exceptions;
            }
        }
        return $ret;
    }
}