includes/filerepo/ForeignAPIRepo.php

   1 <?php
   2 /**
   3  * Foreign repository accessible through api.php requests.
   4  *
   5  * @file
   6  * @ingroup FileRepo
   7  */
   8
   9 /**
  10  * A foreign repository with a remote MediaWiki with an API thingy
  11  *
  12  * Example config:
  13  *
  14  * $wgForeignFileRepos[] = array(
  15  *   'class'                  => 'ForeignAPIRepo',
  16  *   'name'                   => 'shared',
  17  *   'apibase'                => 'http://en.wikipedia.org/w/api.php',
  18  *   'fetchDescription'       => true, // Optional
  19  *   'descriptionCacheExpiry' => 3600,
  20  * );
  21  *
  22  * @ingroup FileRepo
  23  */
  24 class ForeignAPIRepo extends FileRepo {
  25         /* This version string is used in the user agent for requests and will help
  26          * server maintainers in identify ForeignAPI usage.
  27          * Update the version every time you make breaking or significant changes. */
  28         const VERSION = "2.1";
  29
  30         var $fileFactory = array( 'ForeignAPIFile', 'newFromTitle' );
  31         /* Check back with Commons after a day */
  32         var $apiThumbCacheExpiry = 86400; /* 24*60*60 */
  33         /* Redownload thumbnail files after a month */
  34         var $fileCacheExpiry = 2592000; /* 86400*30 */
  35
  36         protected $mQueryCache = array();
  37         protected $mFileExists = array();
  38
  39         function __construct( $info ) {
  40                 global $wgLocalFileRepo;
  41                 parent::__construct( $info );
  42
  43                 // http://commons.wikimedia.org/w/api.php
  44                 $this->mApiBase = isset( $info['apibase'] ) ? $info['apibase'] : null;
  45
  46                 if( isset( $info['apiThumbCacheExpiry'] ) ) {
  47                         $this->apiThumbCacheExpiry = $info['apiThumbCacheExpiry'];
  48                 }
  49                 if( isset( $info['fileCacheExpiry'] ) ) {
  50                         $this->fileCacheExpiry = $info['fileCacheExpiry'];
  51                 }
  52                 if( !$this->scriptDirUrl ) {
  53                         // hack for description fetches
  54                         $this->scriptDirUrl = dirname( $this->mApiBase );
  55                 }
  56                 // If we can cache thumbs we can guess sane defaults for these
  57                 if( $this->canCacheThumbs() && !$this->url ) {
  58                         $this->url = $wgLocalFileRepo['url'];
  59                 }
  60                 if( $this->canCacheThumbs() && !$this->thumbUrl ) {
  61                         $this->thumbUrl = $this->url . '/thumb';
  62                 }
  63         }
  64
  65         /**
  66          * Per docs in FileRepo, this needs to return false if we don't support versioned
  67          * files. Well, we don't.
  68          *
  69          * @return File
  70          */
  71         function newFile( $title, $time = false ) {
  72                 if ( $time ) {
  73                         return false;
  74                 }
  75                 return parent::newFile( $title, $time );
  76         }
  77
  78         /**
  79          * No-ops
  80          * @return bool
  81          */
  82
  83         function storeBatch( $triplets, $flags = 0 ) {
  84                 return false;
  85         }
  86
  87         function storeTemp( $originalName, $srcPath ) {
  88                 return false;
  89         }
  90
  91         function concatenate( $fileList, $targetPath, $flags = 0 ){
  92                 return false;
  93         }
  94
  95         function append( $srcPath, $toAppendPath, $flags = 0 ){
  96                 return false;
  97         }
  98
  99         function appendFinish( $toAppendPath ){
 100                 return false;
 101         }
 102
 103         function publishBatch( $triplets, $flags = 0 ) {
 104                 return false;
 105         }
 106
 107         function deleteBatch( $sourceDestPairs ) {
 108                 return false;
 109         }
 110
 111         function fileExistsBatch( $files, $flags = 0 ) {
 112                 $results = array();
 113                 foreach ( $files as $k => $f ) {
 114                         if ( isset( $this->mFileExists[$k] ) ) {
 115                                 $results[$k] = true;
 116                                 unset( $files[$k] );
 117                         } elseif( self::isVirtualUrl( $f ) ) {
 118                                 # @todo FIXME: We need to be able to handle virtual
 119                                 # URLs better, at least when we know they refer to the
 120                                 # same repo.
 121                                 $results[$k] = false;
 122                                 unset( $files[$k] );
 123                         }
 124                 }
 125
 126                 $data = $this->fetchImageQuery( array( 'titles' => implode( $files, '|' ),
 127                                                                                         'prop' => 'imageinfo' ) );
 128                 if( isset( $data['query']['pages'] ) ) {
 129                         $i = 0;
 130                         foreach( $files as $key => $file ) {
 131                                 $results[$key] = $this->mFileExists[$key] = !isset( $data['query']['pages'][$i]['missing'] );
 132                                 $i++;
 133                         }
 134                 }
 135                 return $results;
 136         }
 137
 138         function getFileProps( $virtualUrl ) {
 139                 return false;
 140         }
 141
 142         function fetchImageQuery( $query ) {
 143                 global $wgMemc;
 144
 145                 $query = array_merge( $query,
 146                         array(
 147                                 'format' => 'json',
 148                                 'action' => 'query',
 149                                 'redirects' => 'true'
 150                         ) );
 151                 if ( $this->mApiBase ) {
 152                         $url = wfAppendQuery( $this->mApiBase, $query );
 153                 } else {
 154                         $url = $this->makeUrl( $query, 'api' );
 155                 }
 156
 157                 if( !isset( $this->mQueryCache[$url] ) ) {
 158                         $key = $this->getLocalCacheKey( 'ForeignAPIRepo', 'Metadata', md5( $url ) );
 159                         $data = $wgMemc->get( $key );
 160                         if( !$data ) {
 161                                 $data = self::httpGet( $url );
 162                                 if ( !$data ) {
 163                                         return null;
 164                                 }
 165                                 $wgMemc->set( $key, $data, 3600 );
 166                         }
 167
 168                         if( count( $this->mQueryCache ) > 100 ) {
 169                                 // Keep the cache from growing infinitely
 170                                 $this->mQueryCache = array();
 171                         }
 172                         $this->mQueryCache[$url] = $data;
 173                 }
 174                 return FormatJson::decode( $this->mQueryCache[$url], true );
 175         }
 176
 177         function getImageInfo( $data ) {
 178                 if( $data && isset( $data['query']['pages'] ) ) {
 179                         foreach( $data['query']['pages'] as $info ) {
 180                                 if( isset( $info['imageinfo'][0] ) ) {
 181                                         return $info['imageinfo'][0];
 182                                 }
 183                         }
 184                 }
 185                 return false;
 186         }
 187
 188         function findBySha1( $hash ) {
 189                 $results = $this->fetchImageQuery( array(
 190                                                                                 'aisha1base36' => $hash,
 191                                                                                 'aiprop'       => ForeignAPIFile::getProps(),
 192                                                                                 'list'         => 'allimages', ) );
 193                 $ret = array();
 194                 if ( isset( $results['query']['allimages'] ) ) {
 195                         foreach ( $results['query']['allimages'] as $img ) {
 196                                 // 1.14 was broken, doesn't return name attribute
 197                                 if( !isset( $img['name'] ) ) {
 198                                         continue;
 199                                 }
 200                                 $ret[] = new ForeignAPIFile( Title::makeTitle( NS_FILE, $img['name'] ), $this, $img );
 201                         }
 202                 }
 203                 return $ret;
 204         }
 205
 206         function getThumbUrl( $name, $width = -1, $height = -1, &$result = null, $otherParams = '' ) {
 207                 $data = $this->fetchImageQuery( array(
 208                         'titles' => 'File:' . $name,
 209                         'iiprop' => 'url|timestamp',
 210                         'iiurlwidth' => $width,
 211                         'iiurlheight' => $height,
 212                         'iiurlparam'  => $otherParams,
 213                         'prop' => 'imageinfo' ) );
 214                 $info = $this->getImageInfo( $data );
 215
 216                 if( $data && $info && isset( $info['thumburl'] ) ) {
 217                         wfDebug( __METHOD__ . " got remote thumb " . $info['thumburl'] . "\n" );
 218                         $result = $info;
 219                         return $info['thumburl'];
 220                 } else {
 221                         return false;
 222                 }
 223         }
 224
 225         /**
 226          * Return the imageurl from cache if possible
 227          *
 228          * If the url has been requested today, get it from cache
 229          * Otherwise retrieve remote thumb url, check for local file.
 230          *
 231          * @param $name String is a dbkey form of a title
 232          * @param $width
 233          * @param $height
 234          * @param String $param Other rendering parameters (page number, etc) from handler's makeParamString.
 235          * @return bool|string
 236          */
 237         function getThumbUrlFromCache( $name, $width, $height, $params="" ) {
 238                 global $wgMemc;
 239
 240                 if ( !$this->canCacheThumbs() ) {
 241                         $result = null; // can't pass "null" by reference, but it's ok as default value
 242                         return $this->getThumbUrl( $name, $width, $height, $result, $params );
 243                 }
 244                 $key = $this->getLocalCacheKey( 'ForeignAPIRepo', 'ThumbUrl', $name );
 245                 $sizekey = "$width:$height:$params";
 246
 247                 /* Get the array of urls that we already know */
 248                 $knownThumbUrls = $wgMemc->get($key);
 249                 if( !$knownThumbUrls ) {
 250                         /* No knownThumbUrls for this file */
 251                         $knownThumbUrls = array();
 252                 } else {
 253                         if( isset( $knownThumbUrls[$sizekey] ) ) {
 254                                 wfDebug( __METHOD__ . ': Got thumburl from local cache: ' .
 255                                         "{$knownThumbUrls[$sizekey]} \n");
 256                                 return $knownThumbUrls[$sizekey];
 257                         }
 258                         /* This size is not yet known */
 259                 }
 260
 261                 $metadata = null;
 262                 $foreignUrl = $this->getThumbUrl( $name, $width, $height, $metadata, $params );
 263
 264                 if( !$foreignUrl ) {
 265                         wfDebug( __METHOD__ . " Could not find thumburl\n" );
 266                         return false;
 267                 }
 268
 269                 // We need the same filename as the remote one :)
 270                 $fileName = rawurldecode( pathinfo( $foreignUrl, PATHINFO_BASENAME ) );
 271                 if( !$this->validateFilename( $fileName ) ) {
 272                         wfDebug( __METHOD__ . " The deduced filename $fileName is not safe\n" );
 273                         return false;
 274                 }
 275                 $localPath =  $this->getZonePath( 'thumb' ) . "/" . $this->getHashPath( $name ) . $name;
 276                 $localFilename = $localPath . "/" . $fileName;
 277                 $localUrl =  $this->getZoneUrl( 'thumb' ) . "/" . $this->getHashPath( $name ) . rawurlencode( $name ) . "/" . rawurlencode( $fileName );
 278
 279                 if( $this->fileExists( $localFilename ) && isset( $metadata['timestamp'] ) ) {
 280                         wfDebug( __METHOD__ . " Thumbnail was already downloaded before\n" );
 281                         $modified = $this->getFileTimestamp( $localFilename );
 282                         $remoteModified = strtotime( $metadata['timestamp'] );
 283                         $current = time();
 284                         $diff = abs( $modified - $current );
 285                         if( $remoteModified < $modified && $diff < $this->fileCacheExpiry ) {
 286                                 /* Use our current and already downloaded thumbnail */
 287                                 $knownThumbUrls[$sizekey] = $localUrl;
 288                                 $wgMemc->set( $key, $knownThumbUrls, $this->apiThumbCacheExpiry );
 289                                 return $localUrl;
 290                         }
 291                         /* There is a new Commons file, or existing thumbnail older than a month */
 292                 }
 293                 $thumb = self::httpGet( $foreignUrl );
 294                 if( !$thumb ) {
 295                         wfDebug( __METHOD__ . " Could not download thumb\n" );
 296                         return false;
 297                 }
 298
 299                 # @todo FIXME: Delete old thumbs that aren't being used. Maintenance script?
 300                 wfSuppressWarnings();
 301                 $backend = $this->getBackend();
 302                 $op = array( 'op' => 'create', 'dst' => $localFilename, 'content' => $thumb );
 303                 if( !$backend->doOperation( $op )->isOK() ) {
 304                         wfRestoreWarnings();
 305                         wfDebug( __METHOD__ . " could not write to thumb path\n" );
 306                         return $foreignUrl;
 307                 }
 308                 wfRestoreWarnings();
 309                 $knownThumbUrls[$sizekey] = $localUrl;
 310                 $wgMemc->set( $key, $knownThumbUrls, $this->apiThumbCacheExpiry );
 311                 wfDebug( __METHOD__ . " got local thumb $localUrl, saving to cache \n" );
 312                 return $localUrl;
 313         }
 314
 315         /**
 316          * @see FileRepo::getZoneUrl()
 317          * @return String
 318          */
 319         function getZoneUrl( $zone ) {
 320                 switch ( $zone ) {
 321                         case 'public':
 322                                 return $this->url;
 323                         case 'thumb':
 324                                 return $this->thumbUrl;
 325                         default:
 326                                 return parent::getZoneUrl( $zone );
 327                 }
 328         }
 329
 330         /**
 331          * Get the local directory corresponding to one of the basic zones
 332          * @return bool|null|string
 333          */
 334         function getZonePath( $zone ) {
 335                 $supported = array( 'public', 'thumb' );
 336                 if ( in_array( $zone, $supported ) ) {
 337                         return parent::getZonePath( $zone );
 338                 }
 339                 return false;
 340         }
 341
 342         /**
 343          * Are we locally caching the thumbnails?
 344          * @return bool
 345          */
 346         public function canCacheThumbs() {
 347                 return ( $this->apiThumbCacheExpiry > 0 );
 348         }
 349
 350         /**
 351          * The user agent the ForeignAPIRepo will use.
 352          * @return string
 353          */
 354         public static function getUserAgent() {
 355                 return Http::userAgent() . " ForeignAPIRepo/" . self::VERSION;
 356         }
 357
 358         /**
 359          * Like a Http:get request, but with custom User-Agent.
 360          * @see Http:get
 361          * @return bool|String
 362          */
 363         public static function httpGet( $url, $timeout = 'default', $options = array() ) {
 364                 $options['timeout'] = $timeout;
 365                 /* Http::get */
 366                 $url = wfExpandUrl( $url, PROTO_HTTP );
 367                 wfDebug( "ForeignAPIRepo: HTTP GET: $url\n" );
 368                 $options['method'] = "GET";
 369
 370                 if ( !isset( $options['timeout'] ) ) {
 371                         $options['timeout'] = 'default';
 372                 }
 373
 374                 $req = MWHttpRequest::factory( $url, $options );
 375                 $req->setUserAgent( ForeignAPIRepo::getUserAgent() );
 376                 $status = $req->execute();
 377
 378                 if ( $status->isOK() ) {
 379                         return $req->getContent();
 380                 } else {
 381                         return false;
 382                 }
 383         }
 384
 385         function enumFiles( $callback ) {
 386                 throw new MWException( 'enumFiles is not supported by ' . get_class( $this ) );
 387         }
 388 }