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