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