ForeignAPIRepo.php

Go to the documentation of this file.

<?php
use MediaWiki\Logger\LoggerFactory;

class ForeignAPIRepo extends FileRepo {
    /* 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. */
    const VERSION = "2.1";

    protected static $imageInfoProps = [
        'url',
        'timestamp',
    ];

    protected $fileFactory = [ ForeignAPIFile::class, 'newFromTitle' ];
    protected $apiThumbCacheExpiry = 86400; // 1 day (24*3600)

    protected $fileCacheExpiry = 2592000; // 1 month (30*24*3600)

    protected $mFileExists = [];

    private $mApiBase;

    function __construct( $info ) {
        global $wgLocalFileRepo;
        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 = $info['fileCacheExpiry'];
        }
        if ( !$this->scriptDirUrl ) {
            // hack for description fetches
            $this->scriptDirUrl = dirname( $this->mApiBase );
        }
        // If we can cache thumbs we can guess sane defaults for these
        if ( $this->canCacheThumbs() && !$this->url ) {
            $this->url = $wgLocalFileRepo['url'];
        }
        if ( $this->canCacheThumbs() && !$this->thumbUrl ) {
            $this->thumbUrl = $this->url . '/thumb';
        }
    }

    function getApiUrl() {
        return $this->mApiBase;
    }

    function newFile( $title, $time = false ) {
        if ( $time ) {
            return false;
        }

        return parent::newFile( $title, $time );
    }

    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;
    }

    function getFileProps( $virtualUrl ) {
        return [];
    }

    function fetchImageQuery( $query ) {
        global $wgLanguageCode;

        $query = array_merge( $query,
            [
                'format' => 'json',
                'action' => 'query',
                'redirects' => 'true'
            ] );

        if ( !isset( $query['uselang'] ) ) { // uselang is unset or null
            $query['uselang'] = $wgLanguageCode;
        }

        $data = $this->httpGetCached( 'Metadata', $query );

        if ( $data ) {
            return FormatJson::decode( $data, true );
        } else {
            return null;
        }
    }

    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;
    }

    function findBySha1( $hash ) {
        $results = $this->fetchImageQuery( [
            'aisha1base36' => $hash,
            'aiprop' => ForeignAPIFile::getProps(),
            'list' => 'allimages',
        ] );
        $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;
    }

    function getThumbUrl( $name, $width = -1, $height = -1, &$result = null, $otherParams = '' ) {
        $data = $this->fetchImageQuery( [
            'titles' => 'File:' . $name,
            'iiprop' => self::getIIProps(),
            'iiurlwidth' => $width,
            'iiurlheight' => $height,
            'iiurlparam' => $otherParams,
            'prop' => 'imageinfo' ] );
        $info = $this->getImageInfo( $data );

        if ( $data && $info && isset( $info['thumburl'] ) ) {
            wfDebug( __METHOD__ . " got remote thumb " . $info['thumburl'] . "\n" );
            $result = $info;

            return $info['thumburl'];
        } else {
            return false;
        }
    }

    function getThumbError( $name, $width = -1, $height = -1, $otherParams = '', $lang = null ) {
        $data = $this->fetchImageQuery( [
            'titles' => 'File:' . $name,
            'iiprop' => self::getIIProps(),
            'iiurlwidth' => $width,
            'iiurlheight' => $height,
            'iiurlparam' => $otherParams,
            'prop' => 'imageinfo',
            'uselang' => $lang,
        ] );
        $info = $this->getImageInfo( $data );

        if ( $data && $info && isset( $info['thumberror'] ) ) {
            wfDebug( __METHOD__ . " got remote thumb error " . $info['thumberror'] . "\n" );

            return new MediaTransformError(
                'thumbnail_error_remote',
                $width,
                $height,
                $this->getDisplayName(),
                $info['thumberror'] // already parsed message from foreign repo
            );
        } else {
            return false;
        }
    }

    function getThumbUrlFromCache( $name, $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 $this->getThumbUrl( $name, $width, $height, $result, $params );
        }
        $key = $this->getLocalCacheKey( 'ForeignAPIRepo', 'ThumbUrl', $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]} \n" );

            return $knownThumbUrls[$sizekey];
        }

        $metadata = null;
        $foreignUrl = $this->getThumbUrl( $name, $width, $height, $metadata, $params );

        if ( !$foreignUrl ) {
            wfDebug( __METHOD__ . " Could not find thumburl\n" );

            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\n" );

            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\n" );
            $modified = $backend->getFileTimestamp( [ 'src' => $localFilename ] );
            $remoteModified = strtotime( $metadata['timestamp'] );
            $current = time();
            $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 = self::httpGet( $foreignUrl, 'default', [], $mtime );
        if ( !$thumb ) {
            wfDebug( __METHOD__ . " Could not download thumb\n" );

            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'\n" );

            return $foreignUrl;
        }
        $knownThumbUrls[$sizekey] = $localUrl;

        $ttl = $mtime
            ? $this->wanCache->adaptiveTTL( $mtime, $this->apiThumbCacheExpiry )
            : $this->apiThumbCacheExpiry;
        $this->wanCache->set( $key, $knownThumbUrls, $ttl );
        wfDebug( __METHOD__ . " got local thumb $localUrl, saving to cache \n" );

        return $localUrl;
    }

    function getZoneUrl( $zone, $ext = null ) {
        switch ( $zone ) {
            case 'public':
                return $this->url;
            case 'thumb':
                return $this->thumbUrl;
            default:
                return parent::getZoneUrl( $zone, $ext );
        }
    }

    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 static function getUserAgent() {
        return Http::userAgent() . " ForeignAPIRepo/" . self::VERSION;
    }

    function getInfo() {
        $info = parent::getInfo();
        $info['apiurl'] = $this->getApiUrl();

        $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( $general['favicon'] ) ) {
                $info['favicon'] = $general['favicon'];
            }
        }

        return $info;
    }

    public static function httpGet(
        $url, $timeout = 'default', $options = [], &$mtime = false
    ) {
        $options['timeout'] = $timeout;
        /* Http::get */
        $url = wfExpandUrl( $url, PROTO_HTTP );
        wfDebug( "ForeignAPIRepo: HTTP GET: $url\n" );
        $options['method'] = "GET";

        if ( !isset( $options['timeout'] ) ) {
            $options['timeout'] = 'default';
        }

        $req = MWHttpRequest::factory( $url, $options, __METHOD__ );
        $req->setUserAgent( self::getUserAgent() );
        $status = $req->execute();

        if ( $status->isOK() ) {
            $lmod = $req->getResponseHeader( 'Last-Modified' );
            $mtime = $lmod ? wfTimestamp( TS_UNIX, $lmod ) : false;

            return $req->getContent();
        } else {
            $logger = LoggerFactory::getInstance( 'http' );
            $logger->warning(
                $status->getWikiText( false, false, 'en' ),
                [ 'caller' => 'ForeignAPIRepo::httpGet' ]
            );

            return false;
        }
    }

    protected static function getIIProps() {
        return implode( '|', self::$imageInfoProps );
    }

    public function httpGetCached( $target, $query, $cacheTTL = 3600 ) {
        if ( $this->mApiBase ) {
            $url = wfAppendQuery( $this->mApiBase, $query );
        } else {
            $url = $this->makeUrl( $query, 'api' );
        }

        return $this->wanCache->getWithSetCallback(
            $this->getLocalCacheKey( static::class, $target, md5( $url ) ),
            $cacheTTL,
            function ( $curValue, &$ttl ) use ( $url ) {
                $html = self::httpGet( $url, 'default', [], $mtime );
                if ( $html !== false ) {
                    $ttl = $mtime ? $this->wanCache->adaptiveTTL( $mtime, $ttl ) : $ttl;
                } else {
                    $ttl = $this->wanCache->adaptiveTTL( $mtime, $ttl );
                    $html = null; // caches negatives
                }

                return $html;
            },
            [ 'pcGroup' => 'http-get:3', 'pcTTL' => WANObjectCache::TTL_PROC_LONG ]
        );
    }

    function enumFiles( $callback ) {
        throw new MWException( 'enumFiles is not supported by ' . static::class );
    }

    protected function assertWritableRepo() {
        throw new MWException( static::class . ': write operations are not supported.' );
    }
}