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