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