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::class,
  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::class, '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 = $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 string[] $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 false
 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 ForeignAPIFile[]
 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 $height
 260          * @param array|null &$result Output-only parameter, guaranteed to become an array
 261          * @param string $otherParams
 262          *
 263          * @return string|false
 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|null $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                 // We can't check the local cache using FileRepo functions because
 336                 // we override fileExistsBatch(). We have to use the FileBackend directly.
 337                 $backend = $this->getBackend(); // convenience
 338
 339                 if ( !$this->canCacheThumbs() ) {
 340                         $result = null; // can't pass "null" by reference, but it's ok as default value
 341                         return $this->getThumbUrl( $name, $width, $height, $result, $params );
 342                 }
 343                 $key = $this->getLocalCacheKey( 'ForeignAPIRepo', 'ThumbUrl', $name );
 344                 $sizekey = "$width:$height:$params";
 345
 346                 /* Get the array of urls that we already know */
 347                 $knownThumbUrls = $this->wanCache->get( $key );
 348                 if ( !$knownThumbUrls ) {
 349                         /* No knownThumbUrls for this file */
 350                         $knownThumbUrls = [];
 351                 } elseif ( isset( $knownThumbUrls[$sizekey] ) ) {
 352                         wfDebug( __METHOD__ . ': Got thumburl from local cache: ' .
 353                                 "{$knownThumbUrls[$sizekey]} \n" );
 354
 355                         return $knownThumbUrls[$sizekey];
 356                 }
 357
 358                 $metadata = null;
 359                 $foreignUrl = $this->getThumbUrl( $name, $width, $height, $metadata, $params );
 360
 361                 if ( !$foreignUrl ) {
 362                         wfDebug( __METHOD__ . " Could not find thumburl\n" );
 363
 364                         return false;
 365                 }
 366
 367                 // We need the same filename as the remote one :)
 368                 $fileName = rawurldecode( pathinfo( $foreignUrl, PATHINFO_BASENAME ) );
 369                 if ( !$this->validateFilename( $fileName ) ) {
 370                         wfDebug( __METHOD__ . " The deduced filename $fileName is not safe\n" );
 371
 372                         return false;
 373                 }
 374                 $localPath = $this->getZonePath( 'thumb' ) . "/" . $this->getHashPath( $name ) . $name;
 375                 $localFilename = $localPath . "/" . $fileName;
 376                 $localUrl = $this->getZoneUrl( 'thumb' ) . "/" . $this->getHashPath( $name ) .
 377                         rawurlencode( $name ) . "/" . rawurlencode( $fileName );
 378
 379                 if ( $backend->fileExists( [ 'src' => $localFilename ] )
 380                         && isset( $metadata['timestamp'] )
 381                 ) {
 382                         wfDebug( __METHOD__ . " Thumbnail was already downloaded before\n" );
 383                         $modified = $backend->getFileTimestamp( [ 'src' => $localFilename ] );
 384                         $remoteModified = strtotime( $metadata['timestamp'] );
 385                         $current = time();
 386                         $diff = abs( $modified - $current );
 387                         if ( $remoteModified < $modified && $diff < $this->fileCacheExpiry ) {
 388                                 /* Use our current and already downloaded thumbnail */
 389                                 $knownThumbUrls[$sizekey] = $localUrl;
 390                                 $this->wanCache->set( $key, $knownThumbUrls, $this->apiThumbCacheExpiry );
 391
 392                                 return $localUrl;
 393                         }
 394                         /* There is a new Commons file, or existing thumbnail older than a month */
 395                 }
 396
 397                 $thumb = self::httpGet( $foreignUrl, 'default', [], $mtime );
 398                 if ( !$thumb ) {
 399                         wfDebug( __METHOD__ . " Could not download thumb\n" );
 400
 401                         return false;
 402                 }
 403
 404                 # @todo FIXME: Delete old thumbs that aren't being used. Maintenance script?
 405                 $backend->prepare( [ 'dir' => dirname( $localFilename ) ] );
 406                 $params = [ 'dst' => $localFilename, 'content' => $thumb ];
 407                 if ( !$backend->quickCreate( $params )->isOK() ) {
 408                         wfDebug( __METHOD__ . " could not write to thumb path '$localFilename'\n" );
 409
 410                         return $foreignUrl;
 411                 }
 412                 $knownThumbUrls[$sizekey] = $localUrl;
 413
 414                 $ttl = $mtime
 415                         ? $this->wanCache->adaptiveTTL( $mtime, $this->apiThumbCacheExpiry )
 416                         : $this->apiThumbCacheExpiry;
 417                 $this->wanCache->set( $key, $knownThumbUrls, $ttl );
 418                 wfDebug( __METHOD__ . " got local thumb $localUrl, saving to cache \n" );
 419
 420                 return $localUrl;
 421         }
 422
 423         /**
 424          * @see FileRepo::getZoneUrl()
 425          * @param string $zone
 426          * @param string|null $ext Optional file extension
 427          * @return string
 428          */
 429         function getZoneUrl( $zone, $ext = null ) {
 430                 switch ( $zone ) {
 431                         case 'public':
 432                                 return $this->url;
 433                         case 'thumb':
 434                                 return $this->thumbUrl;
 435                         default:
 436                                 return parent::getZoneUrl( $zone, $ext );
 437                 }
 438         }
 439
 440         /**
 441          * Get the local directory corresponding to one of the basic zones
 442          * @param string $zone
 443          * @return bool|null|string
 444          */
 445         function getZonePath( $zone ) {
 446                 $supported = [ 'public', 'thumb' ];
 447                 if ( in_array( $zone, $supported ) ) {
 448                         return parent::getZonePath( $zone );
 449                 }
 450
 451                 return false;
 452         }
 453
 454         /**
 455          * Are we locally caching the thumbnails?
 456          * @return bool
 457          */
 458         public function canCacheThumbs() {
 459                 return ( $this->apiThumbCacheExpiry > 0 );
 460         }
 461
 462         /**
 463          * The user agent the ForeignAPIRepo will use.
 464          * @return string
 465          */
 466         public static function getUserAgent() {
 467                 return Http::userAgent() . " ForeignAPIRepo/" . self::VERSION;
 468         }
 469
 470         /**
 471          * Get information about the repo - overrides/extends the parent
 472          * class's information.
 473          * @return array
 474          * @since 1.22
 475          */
 476         function getInfo() {
 477                 $info = parent::getInfo();
 478                 $info['apiurl'] = $this->getApiUrl();
 479
 480                 $query = [
 481                         'format' => 'json',
 482                         'action' => 'query',
 483                         'meta' => 'siteinfo',
 484                         'siprop' => 'general',
 485                 ];
 486
 487                 $data = $this->httpGetCached( 'SiteInfo', $query, 7200 );
 488
 489                 if ( $data ) {
 490                         $siteInfo = FormatJson::decode( $data, true );
 491                         $general = $siteInfo['query']['general'];
 492
 493                         $info['articlepath'] = $general['articlepath'];
 494                         $info['server'] = $general['server'];
 495
 496                         if ( isset( $general['favicon'] ) ) {
 497                                 $info['favicon'] = $general['favicon'];
 498                         }
 499                 }
 500
 501                 return $info;
 502         }
 503
 504         /**
 505          * Like a HttpRequestFactory::get request, but with custom User-Agent.
 506          * @see HttpRequestFactory::get
 507          * @todo Can this use HttpRequestFactory::get() but just pass the 'userAgent' option?
 508          * @param string $url
 509          * @param string $timeout
 510          * @param array $options
 511          * @param int|bool &$mtime Resulting Last-Modified UNIX timestamp if received
 512          * @return bool|string
 513          */
 514         public static function httpGet(
 515                 $url, $timeout = 'default', $options = [], &$mtime = false
 516         ) {
 517                 $options['timeout'] = $timeout;
 518                 /* Http::get */
 519                 $url = wfExpandUrl( $url, PROTO_HTTP );
 520                 wfDebug( "ForeignAPIRepo: HTTP GET: $url\n" );
 521                 $options['method'] = "GET";
 522
 523                 if ( !isset( $options['timeout'] ) ) {
 524                         $options['timeout'] = 'default';
 525                 }
 526
 527                 $req = MWHttpRequest::factory( $url, $options, __METHOD__ );
 528                 $req->setUserAgent( self::getUserAgent() );
 529                 $status = $req->execute();
 530
 531                 if ( $status->isOK() ) {
 532                         $lmod = $req->getResponseHeader( 'Last-Modified' );
 533                         $mtime = $lmod ? wfTimestamp( TS_UNIX, $lmod ) : false;
 534
 535                         return $req->getContent();
 536                 } else {
 537                         $logger = LoggerFactory::getInstance( 'http' );
 538                         $logger->warning(
 539                                 $status->getWikiText( false, false, 'en' ),
 540                                 [ 'caller' => 'ForeignAPIRepo::httpGet' ]
 541                         );
 542
 543                         return false;
 544                 }
 545         }
 546
 547         /**
 548          * @return string
 549          * @since 1.23
 550          */
 551         protected static function getIIProps() {
 552                 return implode( '|', self::$imageInfoProps );
 553         }
 554
 555         /**
 556          * HTTP GET request to a mediawiki API (with caching)
 557          * @param string $target Used in cache key creation, mostly
 558          * @param array $query The query parameters for the API request
 559          * @param int $cacheTTL Time to live for the memcached caching
 560          * @return string|null
 561          */
 562         public function httpGetCached( $target, $query, $cacheTTL = 3600 ) {
 563                 if ( $this->mApiBase ) {
 564                         $url = wfAppendQuery( $this->mApiBase, $query );
 565                 } else {
 566                         $url = $this->makeUrl( $query, 'api' );
 567                 }
 568
 569                 return $this->wanCache->getWithSetCallback(
 570                         $this->getLocalCacheKey( static::class, $target, md5( $url ) ),
 571                         $cacheTTL,
 572                         function ( $curValue, &$ttl ) use ( $url ) {
 573                                 $html = self::httpGet( $url, 'default', [], $mtime );
 574                                 if ( $html !== false ) {
 575                                         $ttl = $mtime ? $this->wanCache->adaptiveTTL( $mtime, $ttl ) : $ttl;
 576                                 } else {
 577                                         $ttl = $this->wanCache->adaptiveTTL( $mtime, $ttl );
 578                                         $html = null; // caches negatives
 579                                 }
 580
 581                                 return $html;
 582                         },
 583                         [ 'pcTTL' => WANObjectCache::TTL_PROC_LONG ]
 584                 );
 585         }
 586
 587         /**
 588          * @param callable $callback
 589          * @throws MWException
 590          */
 591         function enumFiles( $callback ) {
 592                 throw new MWException( 'enumFiles is not supported by ' . static::class );
 593         }
 594
 595         /**
 596          * @throws MWException
 597          */
 598         protected function assertWritableRepo() {
 599                 throw new MWException( static::class . ': write operations are not supported.' );
 600         }
 601 }