ForeignAPIRepo.php

Go to the documentation of this file.

<?php
namespace MediaWiki\FileRepo;
 
use LogicException;
use MediaWiki\FileRepo\File\File;
use MediaWiki\FileRepo\File\ForeignAPIFile;
use MediaWiki\Json\FormatJson;
use MediaWiki\Linker\LinkTarget;
use MediaWiki\Logger\LoggerFactory;
use MediaWiki\MainConfigNames;
use MediaWiki\Media\MediaTransformError;
use MediaWiki\MediaWikiServices;
use MediaWiki\Page\PageIdentity;
use MediaWiki\Title\Title;
use RuntimeException;
use Wikimedia\FileBackend\FileBackend;
use Wikimedia\ObjectCache\WANObjectCache;
use Wikimedia\Timestamp\ConvertibleTimestamp;
use Wikimedia\Timestamp\TimestampFormat as TS;
 
class ForeignAPIRepo extends FileRepo implements IForeignRepoWithMWApi {
    /* This version string is used in the user agent for requests and will help
     * server maintainers in identify ForeignAPI usage.
     * Update the version every time you make breaking or significant changes. */
    private const VERSION = "2.1";
 
    protected $fileFactory = [ ForeignAPIFile::class, 'newFromTitle' ];
    protected $apiThumbCacheExpiry = 24 * 3600; // 1 day
 
    protected $fileCacheExpiry = 30 * 24 * 3600; // 1 month
 
    protected $apiMetadataExpiry = 4 * 3600; // 4 hours
 
    protected $mFileExists = [];
 
    private $mApiBase;
 
    private $userAgent;
 
    public function __construct( $info ) {
        $localFileRepo = MediaWikiServices::getInstance()->getMainConfig()
            ->get( MainConfigNames::LocalFileRepo );
        parent::__construct( $info );
 
        // https://commons.wikimedia.org/w/api.php
        $this->mApiBase = $info['apibase'] ?? null;
 
        if ( isset( $info['apiThumbCacheExpiry'] ) ) {
            $this->apiThumbCacheExpiry = $info['apiThumbCacheExpiry'];
        }
        if ( isset( $info['fileCacheExpiry'] ) ) {
            $this->fileCacheExpiry = $info['fileCacheExpiry'];
        }
        if ( isset( $info['apiMetadataExpiry'] ) ) {
            $this->apiMetadataExpiry = $info['apiMetadataExpiry'];
        }
        if ( isset( $info['userAgent'] ) ) {
            $this->userAgent = $info['userAgent'];
        }
        if ( !$this->scriptDirUrl ) {
            // hack for description fetches
            $this->scriptDirUrl = dirname( $this->mApiBase );
        }
        // If we can cache thumbs we can guess sensible defaults for these
        if ( $this->canCacheThumbs() && !$this->url ) {
            $this->url = $localFileRepo['url'];
        }
        if ( $this->canCacheThumbs() && !$this->thumbUrl ) {
            $this->thumbUrl = $this->url . '/thumb';
        }
    }
 
    public function newFile( $title, $time = false ) {
        if ( $time ) {
            return false;
        }
 
        return parent::newFile( $title, $time );
    }
 
    public function fileExistsBatch( array $files ) {
        $results = [];
        foreach ( $files as $k => $f ) {
            if ( isset( $this->mFileExists[$f] ) ) {
                $results[$k] = $this->mFileExists[$f];
                unset( $files[$k] );
            } elseif ( self::isVirtualUrl( $f ) ) {
                # @todo FIXME: We need to be able to handle virtual
                # URLs better, at least when we know they refer to the
                # same repo.
                $results[$k] = false;
                unset( $files[$k] );
            } elseif ( FileBackend::isStoragePath( $f ) ) {
                $results[$k] = false;
                unset( $files[$k] );
                wfWarn( "Got mwstore:// path '$f'." );
            }
        }
 
        $data = $this->fetchImageQuery( [
            'titles' => implode( '|', $files ),
            'prop' => 'imageinfo' ]
        );
 
        if ( isset( $data['query']['pages'] ) ) {
            # First, get results from the query. Note we only care whether the image exists,
            # not whether it has a description page.
            foreach ( $data['query']['pages'] as $p ) {
                $this->mFileExists[$p['title']] = ( $p['imagerepository'] !== '' );
            }
            # Second, copy the results to any redirects that were queried
            if ( isset( $data['query']['redirects'] ) ) {
                foreach ( $data['query']['redirects'] as $r ) {
                    $this->mFileExists[$r['from']] = $this->mFileExists[$r['to']];
                }
            }
            # Third, copy the results to any non-normalized titles that were queried
            if ( isset( $data['query']['normalized'] ) ) {
                foreach ( $data['query']['normalized'] as $n ) {
                    $this->mFileExists[$n['from']] = $this->mFileExists[$n['to']];
                }
            }
            # Finally, copy the results to the output
            foreach ( $files as $key => $file ) {
                $results[$key] = $this->mFileExists[$file];
            }
        }
 
        return $results;
    }
 
    public function getFileProps( $virtualUrl ) {
        return [];
    }
 
    public function fetchImageQuery( $query ) {
        $query = array_merge( $query, [
            'format' => 'json',
            'action' => 'query',
            'redirects' => 'true',
            // Optimization: Set uselang to fix CDN caching (T97096)
            // Optimization: Set uselang=content to improve CDN caching (T56037)
            // The only localised errors are unreachable via API.
            'uselang' => 'content',
        ] );
 
        $data = $this->httpGetCached( 'Metadata', $query, $this->apiMetadataExpiry );
 
        if ( $data ) {
            return FormatJson::decode( $data, true );
        } else {
            return null;
        }
    }
 
    public function getImageInfo( $data ) {
        if ( $data && isset( $data['query']['pages'] ) ) {
            foreach ( $data['query']['pages'] as $info ) {
                if ( isset( $info['imageinfo'][0] ) ) {
                    $return = $info['imageinfo'][0];
                    if ( isset( $info['pageid'] ) ) {
                        $return['pageid'] = $info['pageid'];
                    }
                    return $return;
                }
            }
        }
 
        return false;
    }
 
    public function findBySha1( $hash ) {
        $results = $this->fetchImageQuery( [
            'list' => 'allimages',
            'aiprop' => ForeignAPIFile::getProps(),
            'aisha1base36' => $hash,
        ] );
        $ret = [];
        if ( isset( $results['query']['allimages'] ) ) {
            foreach ( $results['query']['allimages'] as $img ) {
                // 1.14 was broken, doesn't return name attribute
                if ( !isset( $img['name'] ) ) {
                    continue;
                }
                $ret[] = new ForeignAPIFile( Title::makeTitle( NS_FILE, $img['name'] ), $this, $img );
            }
        }
 
        return $ret;
    }
 
    public function getThumbUrlFromCache( ForeignAPIFile $file, $width, $height, $params = "" ) {
        // We can't check the local cache using FileRepo functions because
        // we override fileExistsBatch(). We have to use the FileBackend directly.
        $backend = $this->getBackend(); // convenience
 
        if ( !$this->canCacheThumbs() ) {
            $result = null; // can't pass "null" by reference, but it's ok as default value
 
            return $file->getThumbUrlFromInfo( $width, $height, $result, $params );
        }
 
        // The dbkey form of a file page title
        $name = $file->getName();
        $key = $this->getLocalCacheKey( 'file-thumb-url', sha1( $name ) );
        $sizekey = "$width:$height:$params";
 
        /* Get the array of urls that we already know */
        $knownThumbUrls = $this->wanCache->get( $key );
        if ( !$knownThumbUrls ) {
            /* No knownThumbUrls for this file */
            $knownThumbUrls = [];
        } elseif ( isset( $knownThumbUrls[$sizekey] ) ) {
            wfDebug( __METHOD__ . ': Got thumburl from local cache: ' .
                "{$knownThumbUrls[$sizekey]}" );
 
            return $knownThumbUrls[$sizekey];
        }
 
        $metadata = null;
        $foreignUrl = $file->getThumbUrlFromInfo( $width, $height, $metadata, $params );
 
        if ( !$foreignUrl || $foreignUrl instanceof MediaTransformError ) {
            wfDebug( __METHOD__ . " Could not find thumburl" );
 
            return false;
        }
 
        // We need the same filename as the remote one :)
        $fileName = rawurldecode( pathinfo( $foreignUrl, PATHINFO_BASENAME ) );
        if ( !$this->validateFilename( $fileName ) ) {
            wfDebug( __METHOD__ . " The deduced filename $fileName is not safe" );
 
            return false;
        }
        $localPath = $this->getZonePath( 'thumb' ) . "/" . $this->getHashPath( $name ) . $name;
        $localFilename = $localPath . "/" . $fileName;
        $localUrl = $this->getZoneUrl( 'thumb' ) . "/" . $this->getHashPath( $name ) .
            rawurlencode( $name ) . "/" . rawurlencode( $fileName );
 
        if ( $backend->fileExists( [ 'src' => $localFilename ] )
            && isset( $metadata['timestamp'] )
        ) {
            wfDebug( __METHOD__ . " Thumbnail was already downloaded before" );
            $modified = (int)wfTimestamp( TS::UNIX, $backend->getFileTimestamp( [ 'src' => $localFilename ] ) );
            $remoteModified = (int)wfTimestamp( TS::UNIX, $metadata['timestamp'] );
            $current = (int)ConvertibleTimestamp::now( TS::UNIX );
            $diff = abs( $modified - $current );
            if ( $remoteModified < $modified && $diff < $this->fileCacheExpiry ) {
                /* Use our current and already downloaded thumbnail */
                $knownThumbUrls[$sizekey] = $localUrl;
                $this->wanCache->set( $key, $knownThumbUrls, $this->apiThumbCacheExpiry );
 
                return $localUrl;
            }
            /* There is a new Commons file, or existing thumbnail older than a month */
        }
 
        $thumb = $this->httpGet( $foreignUrl, 'default', [], $mtime );
        if ( !$thumb ) {
            wfDebug( __METHOD__ . " Could not download thumb" );
 
            return false;
        }
 
        # @todo FIXME: Delete old thumbs that aren't being used. Maintenance script?
        $backend->prepare( [ 'dir' => dirname( $localFilename ) ] );
        $params = [ 'dst' => $localFilename, 'content' => $thumb ];
        if ( !$backend->quickCreate( $params )->isOK() ) {
            wfDebug( __METHOD__ . " could not write to thumb path '$localFilename'" );
 
            return $foreignUrl;
        }
        $knownThumbUrls[$sizekey] = $localUrl;
        $this->wanCache->set( $key, $knownThumbUrls, $this->apiThumbCacheExpiry );
        wfDebug( __METHOD__ . " got local thumb $localUrl, saving to cache" );
 
        return $localUrl;
    }
 
    public function getZoneUrl( $zone, $ext = null ) {
        switch ( $zone ) {
            case 'public':
                return $this->url;
            case 'thumb':
                return $this->thumbUrl;
            default:
                return parent::getZoneUrl( $zone, $ext );
        }
    }
 
    public function getZonePath( $zone ) {
        $supported = [ 'public', 'thumb' ];
        if ( in_array( $zone, $supported ) ) {
            return parent::getZonePath( $zone );
        }
 
        return false;
    }
 
    public function canCacheThumbs() {
        return ( $this->apiThumbCacheExpiry > 0 );
    }
 
    public function getUserAgent() {
        $mediaWikiUA = MediaWikiServices::getInstance()->getHttpRequestFactory()->getUserAgent();
        $classVersion = self::VERSION;
        $extra = $this->userAgent !== null ? ' ' . $this->userAgent : '';
        return "$mediaWikiUA ForeignAPIRepo/$classVersion" . $extra;
    }
 
    public function getInfo() {
        $info = parent::getInfo();
        $info['apiurl'] = $this->mApiBase;
 
        $query = [
            'format' => 'json',
            'action' => 'query',
            'meta' => 'siteinfo',
            'siprop' => 'general',
        ];
 
        $data = $this->httpGetCached( 'SiteInfo', $query, 7200 );
 
        if ( $data ) {
            $siteInfo = FormatJson::decode( $data, true );
            $general = $siteInfo['query']['general'];
 
            $info['articlepath'] = $general['articlepath'];
            $info['server'] = $general['server'];
            if ( !isset( $info['favicon'] ) && isset( $general['favicon'] ) ) {
                $info['favicon'] = $general['favicon'];
            }
        }
 
        return $info;
    }
 
    public function httpGet(
        $url, $timeout = 'default', $options = [], &$mtime = false
    ) {
        $urlUtils = MediaWikiServices::getInstance()->getUrlUtils();
        $requestFactory = MediaWikiServices::getInstance()->getHttpRequestFactory();
 
        $options['timeout'] = $timeout;
        $url = $urlUtils->expand( $url, PROTO_HTTP );
        wfDebug( "ForeignAPIRepo: HTTP GET: $url" );
        if ( !$url ) {
            return false;
        }
        $options['method'] = "GET";
 
        if ( !isset( $options['timeout'] ) ) {
            $options['timeout'] = 'default';
        }
 
        $options['userAgent'] = $this->getUserAgent();
 
        $req = $requestFactory->create( $url, $options, __METHOD__ );
        $req->setHeader( 'Referer', $urlUtils->getCanonicalServer() );
        $status = $req->execute();
 
        if ( $status->isOK() ) {
            $lmod = $req->getResponseHeader( 'Last-Modified' );
            $mtime = $lmod ? (int)wfTimestamp( TS::UNIX, $lmod ) : false;
 
            return $req->getContent();
        } else {
            $logger = LoggerFactory::getInstance( 'http' );
            $logger->warning(
                $status->getWikiText( false, false, 'en' ),
                [ 'caller' => __METHOD__ ]
            );
 
            return false;
        }
    }
 
    public function httpGetCached( $attribute, $query, $cacheTTL = 3600 ) {
        if ( $this->mApiBase ) {
            $url = wfAppendQuery( $this->mApiBase, $query );
        } else {
            $url = $this->makeUrl( $query, 'api' );
        }
 
        return $this->wanCache->getWithSetCallback(
            // Allow reusing the same cached data across wikis (T285271).
            // This does not use getSharedCacheKey() because caching here
            // is transparent to client wikis (which are not expected to issue purges).
            $this->wanCache->makeGlobalKey( "filerepo-$attribute", sha1( $url ) ),
            $cacheTTL,
            function () use ( $url ) {
                $html = $this->httpGet( $url, 'default', [], $mtime );
                if ( $html === false ) {
                    $html = null; // caches negatives
                }
                return $html;
            },
            [ 'pcGroup' => 'http-get:3', 'pcTTL' => WANObjectCache::TTL_PROC_LONG ]
        );
    }
 
    public function enumFiles( $callback ): never {
        throw new RuntimeException( 'enumFiles is not supported by ' . static::class );
    }
 
    protected function assertWritableRepo(): never {
        throw new LogicException( static::class . ': write operations are not supported.' );
    }
}
 
class_alias( ForeignAPIRepo::class, 'ForeignAPIRepo' );