includes/filerepo/ForeignAPIRepo.php

   1 <?php
   2 /**
   3  * Foreign repository accessible through api.php requests.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup FileRepo
  22  */
  23
  24 use MediaWiki\Logger\LoggerFactory;
  25 use MediaWiki\MediaWikiServices;
  26
  27 /**
  28  * A foreign repository with a remote MediaWiki with an API thingy
  29  *
  30  * Example config:
  31  *
  32  * $wgForeignFileRepos[] = [
  33  *   'class'                  => ForeignAPIRepo::class,
  34  *   'name'                   => 'shared',
  35  *   'apibase'                => 'https://en.wikipedia.org/w/api.php',
  36  *   'fetchDescription'       => true, // Optional
  37  *   'descriptionCacheExpiry' => 3600,
  38  * ];
  39  *
  40  * @ingroup FileRepo
  41  */
  42 class ForeignAPIRepo extends FileRepo {
  43         /* This version string is used in the user agent for requests and will help
  44          * server maintainers in identify ForeignAPI usage.
  45          * Update the version every time you make breaking or significant changes. */
  46         const VERSION = "2.1";
  47
  48         /**
  49          * List of iiprop values for the thumbnail fetch queries.
  50          * @since 1.23
  51          */
  52         protected static $imageInfoProps = [
  53                 'url',
  54                 'timestamp',
  55         ];
  56
  57         protected $fileFactory = [ ForeignAPIFile::class, 'newFromTitle' ];
  58         /** @var int Check back with Commons after this expiry */
  59         protected $apiThumbCacheExpiry = 86400; // 1 day (24*3600)
  60
  61         /** @var int Redownload thumbnail files after this expiry */
  62         protected $fileCacheExpiry = 2592000; // 1 month (30*24*3600)
  63
  64         /** @var array */
  65         protected $mFileExists = [];
  66
  67         /** @var string */
  68         private $mApiBase;
  69
  70         /**
  71          * @param array|null $info
  72          */
  73         function __construct( $info ) {
  74                 global $wgLocalFileRepo;
  75                 parent::__construct( $info );
  76
  77                 // https://commons.wikimedia.org/w/api.php
  78                 $this->mApiBase = $info['apibase'] ?? null;
  79
  80                 if ( isset( $info['apiThumbCacheExpiry'] ) ) {
  81                         $this->apiThumbCacheExpiry = $info['apiThumbCacheExpiry'];
  82                 }
  83                 if ( isset( $info['fileCacheExpiry'] ) ) {
  84                         $this->fileCacheExpiry = $info['fileCacheExpiry'];
  85                 }
  86                 if ( !$this->scriptDirUrl ) {
  87                         // hack for description fetches
  88                         $this->scriptDirUrl = dirname( $this->mApiBase );
  89                 }
  90                 // If we can cache thumbs we can guess sane defaults for these
  91                 if ( $this->canCacheThumbs() && !$this->url ) {
  92                         $this->url = $wgLocalFileRepo['url'];
  93                 }
  94                 if ( $this->canCacheThumbs() && !$this->thumbUrl ) {
  95                         $this->thumbUrl = $this->url . '/thumb';
  96                 }
  97         }
  98
  99         /**
 100          * @return string
 101          * @since 1.22
 102          */
 103         function getApiUrl() {
 104                 return $this->mApiBase;
 105         }
 106
 107         /**
 108          * Per docs in FileRepo, this needs to return false if we don't support versioned
 109          * files. Well, we don't.
 110          *
 111          * @param Title $title
 112          * @param string|bool $time
 113          * @return File|false
 114          */
 115         function newFile( $title, $time = false ) {
 116                 if ( $time ) {
 117                         return false;
 118                 }
 119
 120                 return parent::newFile( $title, $time );
 121         }
 122
 123         /**
 124          * @param string[] $files
 125          * @return array
 126          */
 127         function fileExistsBatch( array $files ) {
 128                 $results = [];
 129                 foreach ( $files as $k => $f ) {
 130                         if ( isset( $this->mFileExists[$f] ) ) {
 131                                 $results[$k] = $this->mFileExists[$f];
 132                                 unset( $files[$k] );
 133                         } elseif ( self::isVirtualUrl( $f ) ) {
 134                                 # @todo FIXME: We need to be able to handle virtual
 135                                 # URLs better, at least when we know they refer to the
 136                                 # same repo.
 137                                 $results[$k] = false;
 138                                 unset( $files[$k] );
 139                         } elseif ( FileBackend::isStoragePath( $f ) ) {
 140                                 $results[$k] = false;
 141                                 unset( $files[$k] );
 142                                 wfWarn( "Got mwstore:// path '$f'." );
 143                         }
 144                 }
 145
 146                 $data = $this->fetchImageQuery( [
 147                         'titles' => implode( '|', $files ),
 148                         'prop' => 'imageinfo' ]
 149                 );
 150
 151                 if ( isset( $data['query']['pages'] ) ) {
 152                         # First, get results from the query. Note we only care whether the image exists,
 153                         # not whether it has a description page.
 154                         foreach ( $data['query']['pages'] as $p ) {
 155                                 $this->mFileExists[$p['title']] = ( $p['imagerepository'] !== '' );
 156                         }
 157                         # Second, copy the results to any redirects that were queried
 158                         if ( isset( $data['query']['redirects'] ) ) {
 159                                 foreach ( $data['query']['redirects'] as $r ) {
 160                                         $this->mFileExists[$r['from']] = $this->mFileExists[$r['to']];
 161                                 }
 162                         }
 163                         # Third, copy the results to any non-normalized titles that were queried
 164                         if ( isset( $data['query']['normalized'] ) ) {
 165                                 foreach ( $data['query']['normalized'] as $n ) {
 166                                         $this->mFileExists[$n['from']] = $this->mFileExists[$n['to']];
 167                                 }
 168                         }
 169                         # Finally, copy the results to the output
 170                         foreach ( $files as $key => $file ) {
 171                                 $results[$key] = $this->mFileExists[$file];
 172                         }
 173                 }
 174
 175                 return $results;
 176         }
 177
 178         /**
 179          * @param string $virtualUrl
 180          * @return false
 181          */
 182         function getFileProps( $virtualUrl ) {
 183                 return false;
 184         }
 185
 186         /**
 187          * @param array $query
 188          * @return string
 189          */
 190         function fetchImageQuery( $query ) {
 191                 global $wgLanguageCode;
 192
 193                 $query = array_merge( $query,
 194                         [
 195                                 'format' => 'json',
 196                                 'action' => 'query',
 197                                 'redirects' => 'true'
 198                         ] );
 199
 200                 if ( !isset( $query['uselang'] ) ) { // uselang is unset or null
 201                         $query['uselang'] = $wgLanguageCode;
 202                 }
 203
 204                 $data = $this->httpGetCached( 'Metadata', $query );
 205
 206                 if ( $data ) {
 207                         return FormatJson::decode( $data, true );
 208                 } else {
 209                         return null;
 210                 }
 211         }
 212
 213         /**
 214          * @param array $data
 215          * @return bool|array
 216          */
 217         function getImageInfo( $data ) {
 218                 if ( $data && isset( $data['query']['pages'] ) ) {
 219                         foreach ( $data['query']['pages'] as $info ) {
 220                                 if ( isset( $info['imageinfo'][0] ) ) {
 221                                         $return = $info['imageinfo'][0];
 222                                         if ( isset( $info['pageid'] ) ) {
 223                                                 $return['pageid'] = $info['pageid'];
 224                                         }
 225                                         return $return;
 226                                 }
 227                         }
 228                 }
 229
 230                 return false;
 231         }
 232
 233         /**
 234          * @param string $hash
 235          * @return ForeignAPIFile[]
 236          */
 237         function findBySha1( $hash ) {
 238                 $results = $this->fetchImageQuery( [
 239                         'aisha1base36' => $hash,
 240                         'aiprop' => ForeignAPIFile::getProps(),
 241                         'list' => 'allimages',
 242                 ] );
 243                 $ret = [];
 244                 if ( isset( $results['query']['allimages'] ) ) {
 245                         foreach ( $results['query']['allimages'] as $img ) {
 246                                 // 1.14 was broken, doesn't return name attribute
 247                                 if ( !isset( $img['name'] ) ) {
 248                                         continue;
 249                                 }
 250                                 $ret[] = new ForeignAPIFile( Title::makeTitle( NS_FILE, $img['name'] ), $this, $img );
 251                         }
 252                 }
 253
 254                 return $ret;
 255         }
 256
 257         /**
 258          * @param string $name
 259          * @param int $width
 260          * @param int $height
 261          * @param array|null &$result Output-only parameter, guaranteed to become an array
 262          * @param string $otherParams
 263          *
 264          * @return string|false
 265          */
 266         function getThumbUrl( $name, $width = -1, $height = -1, &$result = null, $otherParams = '' ) {
 267                 $data = $this->fetchImageQuery( [
 268                         'titles' => 'File:' . $name,
 269                         'iiprop' => self::getIIProps(),
 270                         'iiurlwidth' => $width,
 271                         'iiurlheight' => $height,
 272                         'iiurlparam' => $otherParams,
 273                         'prop' => 'imageinfo' ] );
 274                 $info = $this->getImageInfo( $data );
 275
 276                 if ( $data && $info && isset( $info['thumburl'] ) ) {
 277                         wfDebug( __METHOD__ . " got remote thumb " . $info['thumburl'] . "\n" );
 278                         $result = $info;
 279
 280                         return $info['thumburl'];
 281                 } else {
 282                         return false;
 283                 }
 284         }
 285
 286         /**
 287          * @param string $name
 288          * @param int $width
 289          * @param int $height
 290          * @param string $otherParams
 291          * @param string|null $lang Language code for language of error
 292          * @return bool|MediaTransformError
 293          * @since 1.22
 294          */
 295         function getThumbError( $name, $width = -1, $height = -1, $otherParams = '', $lang = null ) {
 296                 $data = $this->fetchImageQuery( [
 297                         'titles' => 'File:' . $name,
 298                         'iiprop' => self::getIIProps(),
 299                         'iiurlwidth' => $width,
 300                         'iiurlheight' => $height,
 301                         'iiurlparam' => $otherParams,
 302                         'prop' => 'imageinfo',
 303                         'uselang' => $lang,
 304                 ] );
 305                 $info = $this->getImageInfo( $data );
 306
 307                 if ( $data && $info && isset( $info['thumberror'] ) ) {
 308                         wfDebug( __METHOD__ . " got remote thumb error " . $info['thumberror'] . "\n" );
 309
 310                         return new MediaTransformError(
 311                                 'thumbnail_error_remote',
 312                                 $width,
 313                                 $height,
 314                                 $this->getDisplayName(),
 315                                 $info['thumberror'] // already parsed message from foreign repo
 316                         );
 317                 } else {
 318                         return false;
 319                 }
 320         }
 321
 322         /**
 323          * Return the imageurl from cache if possible
 324          *
 325          * If the url has been requested today, get it from cache
 326          * Otherwise retrieve remote thumb url, check for local file.
 327          *
 328          * @param string $name Is a dbkey form of a title
 329          * @param int $width
 330          * @param int $height
 331          * @param string $params Other rendering parameters (page number, etc)
 332          *   from handler's makeParamString.
 333          * @return bool|string
 334          */
 335         function getThumbUrlFromCache( $name, $width, $height, $params = "" ) {
 336                 $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
 337                 // We can't check the local cache using FileRepo functions because
 338                 // we override fileExistsBatch(). We have to use the FileBackend directly.
 339                 $backend = $this->getBackend(); // convenience
 340
 341                 if ( !$this->canCacheThumbs() ) {
 342                         $result = null; // can't pass "null" by reference, but it's ok as default value
 343                         return $this->getThumbUrl( $name, $width, $height, $result, $params );
 344                 }
 345                 $key = $this->getLocalCacheKey( 'ForeignAPIRepo', 'ThumbUrl', $name );
 346                 $sizekey = "$width:$height:$params";
 347
 348                 /* Get the array of urls that we already know */
 349                 $knownThumbUrls = $cache->get( $key );
 350                 if ( !$knownThumbUrls ) {
 351                         /* No knownThumbUrls for this file */
 352                         $knownThumbUrls = [];
 353                 } else {
 354                         if ( isset( $knownThumbUrls[$sizekey] ) ) {
 355                                 wfDebug( __METHOD__ . ': Got thumburl from local cache: ' .
 356                                         "{$knownThumbUrls[$sizekey]} \n" );
 357
 358                                 return $knownThumbUrls[$sizekey];
 359                         }
 360                         /* This size is not yet known */
 361                 }
 362
 363                 $metadata = null;
 364                 $foreignUrl = $this->getThumbUrl( $name, $width, $height, $metadata, $params );
 365
 366                 if ( !$foreignUrl ) {
 367                         wfDebug( __METHOD__ . " Could not find thumburl\n" );
 368
 369                         return false;
 370                 }
 371
 372                 // We need the same filename as the remote one :)
 373                 $fileName = rawurldecode( pathinfo( $foreignUrl, PATHINFO_BASENAME ) );
 374                 if ( !$this->validateFilename( $fileName ) ) {
 375                         wfDebug( __METHOD__ . " The deduced filename $fileName is not safe\n" );
 376
 377                         return false;
 378                 }
 379                 $localPath = $this->getZonePath( 'thumb' ) . "/" . $this->getHashPath( $name ) . $name;
 380                 $localFilename = $localPath . "/" . $fileName;
 381                 $localUrl = $this->getZoneUrl( 'thumb' ) . "/" . $this->getHashPath( $name ) .
 382                         rawurlencode( $name ) . "/" . rawurlencode( $fileName );
 383
 384                 if ( $backend->fileExists( [ 'src' => $localFilename ] )
 385                         && isset( $metadata['timestamp'] )
 386                 ) {
 387                         wfDebug( __METHOD__ . " Thumbnail was already downloaded before\n" );
 388                         $modified = $backend->getFileTimestamp( [ 'src' => $localFilename ] );
 389                         $remoteModified = strtotime( $metadata['timestamp'] );
 390                         $current = time();
 391                         $diff = abs( $modified - $current );
 392                         if ( $remoteModified < $modified && $diff < $this->fileCacheExpiry ) {
 393                                 /* Use our current and already downloaded thumbnail */
 394                                 $knownThumbUrls[$sizekey] = $localUrl;
 395                                 $cache->set( $key, $knownThumbUrls, $this->apiThumbCacheExpiry );
 396
 397                                 return $localUrl;
 398                         }
 399                         /* There is a new Commons file, or existing thumbnail older than a month */
 400                 }
 401
 402                 $thumb = self::httpGet( $foreignUrl, 'default', [], $mtime );
 403                 if ( !$thumb ) {
 404                         wfDebug( __METHOD__ . " Could not download thumb\n" );
 405
 406                         return false;
 407                 }
 408
 409                 # @todo FIXME: Delete old thumbs that aren't being used. Maintenance script?
 410                 $backend->prepare( [ 'dir' => dirname( $localFilename ) ] );
 411                 $params = [ 'dst' => $localFilename, 'content' => $thumb ];
 412                 if ( !$backend->quickCreate( $params )->isOK() ) {
 413                         wfDebug( __METHOD__ . " could not write to thumb path '$localFilename'\n" );
 414
 415                         return $foreignUrl;
 416                 }
 417                 $knownThumbUrls[$sizekey] = $localUrl;
 418
 419                 $ttl = $mtime
 420                         ? $cache->adaptiveTTL( $mtime, $this->apiThumbCacheExpiry )
 421                         : $this->apiThumbCacheExpiry;
 422                 $cache->set( $key, $knownThumbUrls, $ttl );
 423                 wfDebug( __METHOD__ . " got local thumb $localUrl, saving to cache \n" );
 424
 425                 return $localUrl;
 426         }
 427
 428         /**
 429          * @see FileRepo::getZoneUrl()
 430          * @param string $zone
 431          * @param string|null $ext Optional file extension
 432          * @return string
 433          */
 434         function getZoneUrl( $zone, $ext = null ) {
 435                 switch ( $zone ) {
 436                         case 'public':
 437                                 return $this->url;
 438                         case 'thumb':
 439                                 return $this->thumbUrl;
 440                         default:
 441                                 return parent::getZoneUrl( $zone, $ext );
 442                 }
 443         }
 444
 445         /**
 446          * Get the local directory corresponding to one of the basic zones
 447          * @param string $zone
 448          * @return bool|null|string
 449          */
 450         function getZonePath( $zone ) {
 451                 $supported = [ 'public', 'thumb' ];
 452                 if ( in_array( $zone, $supported ) ) {
 453                         return parent::getZonePath( $zone );
 454                 }
 455
 456                 return false;
 457         }
 458
 459         /**
 460          * Are we locally caching the thumbnails?
 461          * @return bool
 462          */
 463         public function canCacheThumbs() {
 464                 return ( $this->apiThumbCacheExpiry > 0 );
 465         }
 466
 467         /**
 468          * The user agent the ForeignAPIRepo will use.
 469          * @return string
 470          */
 471         public static function getUserAgent() {
 472                 return Http::userAgent() . " ForeignAPIRepo/" . self::VERSION;
 473         }
 474
 475         /**
 476          * Get information about the repo - overrides/extends the parent
 477          * class's information.
 478          * @return array
 479          * @since 1.22
 480          */
 481         function getInfo() {
 482                 $info = parent::getInfo();
 483                 $info['apiurl'] = $this->getApiUrl();
 484
 485                 $query = [
 486                         'format' => 'json',
 487                         'action' => 'query',
 488                         'meta' => 'siteinfo',
 489                         'siprop' => 'general',
 490                 ];
 491
 492                 $data = $this->httpGetCached( 'SiteInfo', $query, 7200 );
 493
 494                 if ( $data ) {
 495                         $siteInfo = FormatJson::decode( $data, true );
 496                         $general = $siteInfo['query']['general'];
 497
 498                         $info['articlepath'] = $general['articlepath'];
 499                         $info['server'] = $general['server'];
 500
 501                         if ( isset( $general['favicon'] ) ) {
 502                                 $info['favicon'] = $general['favicon'];
 503                         }
 504                 }
 505
 506                 return $info;
 507         }
 508
 509         /**
 510          * Like a Http:get request, but with custom User-Agent.
 511          * @see Http::get
 512          * @param string $url
 513          * @param string $timeout
 514          * @param array $options
 515          * @param int|bool &$mtime Resulting Last-Modified UNIX timestamp if received
 516          * @return bool|string
 517          */
 518         public static function httpGet(
 519                 $url, $timeout = 'default', $options = [], &$mtime = false
 520         ) {
 521                 $options['timeout'] = $timeout;
 522                 /* Http::get */
 523                 $url = wfExpandUrl( $url, PROTO_HTTP );
 524                 wfDebug( "ForeignAPIRepo: HTTP GET: $url\n" );
 525                 $options['method'] = "GET";
 526
 527                 if ( !isset( $options['timeout'] ) ) {
 528                         $options['timeout'] = 'default';
 529                 }
 530
 531                 $req = MWHttpRequest::factory( $url, $options, __METHOD__ );
 532                 $req->setUserAgent( self::getUserAgent() );
 533                 $status = $req->execute();
 534
 535                 if ( $status->isOK() ) {
 536                         $lmod = $req->getResponseHeader( 'Last-Modified' );
 537                         $mtime = $lmod ? wfTimestamp( TS_UNIX, $lmod ) : false;
 538
 539                         return $req->getContent();
 540                 } else {
 541                         $logger = LoggerFactory::getInstance( 'http' );
 542                         $logger->warning(
 543                                 $status->getWikiText( false, false, 'en' ),
 544                                 [ 'caller' => 'ForeignAPIRepo::httpGet' ]
 545                         );
 546
 547                         return false;
 548                 }
 549         }
 550
 551         /**
 552          * @return string
 553          * @since 1.23
 554          */
 555         protected static function getIIProps() {
 556                 return implode( '|', self::$imageInfoProps );
 557         }
 558
 559         /**
 560          * HTTP GET request to a mediawiki API (with caching)
 561          * @param string $target Used in cache key creation, mostly
 562          * @param array $query The query parameters for the API request
 563          * @param int $cacheTTL Time to live for the memcached caching
 564          * @return string|null
 565          */
 566         public function httpGetCached( $target, $query, $cacheTTL = 3600 ) {
 567                 if ( $this->mApiBase ) {
 568                         $url = wfAppendQuery( $this->mApiBase, $query );
 569                 } else {
 570                         $url = $this->makeUrl( $query, 'api' );
 571                 }
 572
 573                 $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
 574                 return $cache->getWithSetCallback(
 575                         $this->getLocalCacheKey( static::class, $target, md5( $url ) ),
 576                         $cacheTTL,
 577                         function ( $curValue, &$ttl ) use ( $url, $cache ) {
 578                                 $html = self::httpGet( $url, 'default', [], $mtime );
 579                                 if ( $html !== false ) {
 580                                         $ttl = $mtime ? $cache->adaptiveTTL( $mtime, $ttl ) : $ttl;
 581                                 } else {
 582                                         $ttl = $cache->adaptiveTTL( $mtime, $ttl );
 583                                         $html = null; // caches negatives
 584                                 }
 585
 586                                 return $html;
 587                         },
 588                         [ 'pcTTL' => $cache::TTL_PROC_LONG ]
 589                 );
 590         }
 591
 592         /**
 593          * @param callable $callback
 594          * @throws MWException
 595          */
 596         function enumFiles( $callback ) {
 597                 throw new MWException( 'enumFiles is not supported by ' . static::class );
 598         }
 599
 600         /**
 601          * @throws MWException
 602          */
 603         protected function assertWritableRepo() {
 604                 throw new MWException( static::class . ': write operations are not supported.' );
 605         }
 606 }