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