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         /* Local image directory */
  36         var $directory;
  37         var $thumbDir;
  38
  39         protected $mQueryCache = array();
  40         protected $mFileExists = array();
  41
  42         function __construct( $info ) {
  43                 parent::__construct( $info );
  44                 global $wgUploadDirectory;
  45
  46                 // http://commons.wikimedia.org/w/api.php
  47                 $this->mApiBase = isset( $info['apibase'] ) ? $info['apibase'] : null;
  48                 $this->directory = isset( $info['directory'] ) ? $info['directory'] : $wgUploadDirectory;
  49
  50                 if( isset( $info['apiThumbCacheExpiry'] ) ) {
  51                         $this->apiThumbCacheExpiry = $info['apiThumbCacheExpiry'];
  52                 }
  53                 if( isset( $info['fileCacheExpiry'] ) ) {
  54                         $this->fileCacheExpiry = $info['fileCacheExpiry'];
  55                 }
  56                 if( !$this->scriptDirUrl ) {
  57                         // hack for description fetches
  58                         $this->scriptDirUrl = dirname( $this->mApiBase );
  59                 }
  60                 // If we can cache thumbs we can guess sane defaults for these
  61                 if( $this->canCacheThumbs() && !$this->url ) {
  62                         global $wgLocalFileRepo;
  63                         $this->url = $wgLocalFileRepo['url'];
  64                 }
  65                 if( $this->canCacheThumbs() && !$this->thumbUrl ) {
  66                         $this->thumbUrl = $this->url . '/thumb';
  67                 }
  68                 if ( isset( $info['thumbDir'] ) ) {
  69                         $this->thumbDir =  $info['thumbDir'];
  70                 } else {
  71                         $this->thumbDir = "{$this->directory}/thumb";
  72                 }
  73         }
  74
  75         /**
  76          * Per docs in FileRepo, this needs to return false if we don't support versioned
  77          * files. Well, we don't.
  78          *
  79          * @return File
  80          */
  81         function newFile( $title, $time = false ) {
  82                 if ( $time ) {
  83                         return false;
  84                 }
  85                 return parent::newFile( $title, $time );
  86         }
  87
  88         /**
  89          * No-ops
  90          */
  91
  92         function storeBatch( $triplets, $flags = 0 ) {
  93                 return false;
  94         }
  95
  96         function storeTemp( $originalName, $srcPath ) {
  97                 return false;
  98         }
  99
 100         function concatenate( $fileList, $targetPath, $flags = 0 ){
 101                 return false;
 102         }
 103
 104         function append( $srcPath, $toAppendPath, $flags = 0 ){
 105                 return false;
 106         }
 107
 108         function appendFinish( $toAppendPath ){
 109                 return false;
 110         }
 111
 112         function publishBatch( $triplets, $flags = 0 ) {
 113                 return false;
 114         }
 115
 116         function deleteBatch( $sourceDestPairs ) {
 117                 return false;
 118         }
 119
 120         function fileExistsBatch( $files, $flags = 0 ) {
 121                 $results = array();
 122                 foreach ( $files as $k => $f ) {
 123                         if ( isset( $this->mFileExists[$k] ) ) {
 124                                 $results[$k] = true;
 125                                 unset( $files[$k] );
 126                         } elseif( self::isVirtualUrl( $f ) ) {
 127                                 # @todo FIXME: We need to be able to handle virtual
 128                                 # URLs better, at least when we know they refer to the
 129                                 # same repo.
 130                                 $results[$k] = false;
 131                                 unset( $files[$k] );
 132                         }
 133                 }
 134
 135                 $data = $this->fetchImageQuery( array( 'titles' => implode( $files, '|' ),
 136                                                                                         'prop' => 'imageinfo' ) );
 137                 if( isset( $data['query']['pages'] ) ) {
 138                         $i = 0;
 139                         foreach( $files as $key => $file ) {
 140                                 $results[$key] = $this->mFileExists[$key] = !isset( $data['query']['pages'][$i]['missing'] );
 141                                 $i++;
 142                         }
 143                 }
 144                 return $results;
 145         }
 146
 147         function getFileProps( $virtualUrl ) {
 148                 return false;
 149         }
 150
 151         function fetchImageQuery( $query ) {
 152                 global $wgMemc;
 153
 154                 $query = array_merge( $query,
 155                         array(
 156                                 'format' => 'json',
 157                                 'action' => 'query',
 158                                 'redirects' => 'true'
 159                         ) );
 160                 if ( $this->mApiBase ) {
 161                         $url = wfAppendQuery( $this->mApiBase, $query );
 162                 } else {
 163                         $url = $this->makeUrl( $query, 'api' );
 164                 }
 165
 166                 if( !isset( $this->mQueryCache[$url] ) ) {
 167                         $key = $this->getLocalCacheKey( 'ForeignAPIRepo', 'Metadata', md5( $url ) );
 168                         $data = $wgMemc->get( $key );
 169                         if( !$data ) {
 170                                 $data = self::httpGet( $url );
 171                                 if ( !$data ) {
 172                                         return null;
 173                                 }
 174                                 $wgMemc->set( $key, $data, 3600 );
 175                         }
 176
 177                         if( count( $this->mQueryCache ) > 100 ) {
 178                                 // Keep the cache from growing infinitely
 179                                 $this->mQueryCache = array();
 180                         }
 181                         $this->mQueryCache[$url] = $data;
 182                 }
 183                 return FormatJson::decode( $this->mQueryCache[$url], true );
 184         }
 185
 186         function getImageInfo( $data ) {
 187                 if( $data && isset( $data['query']['pages'] ) ) {
 188                         foreach( $data['query']['pages'] as $info ) {
 189                                 if( isset( $info['imageinfo'][0] ) ) {
 190                                         return $info['imageinfo'][0];
 191                                 }
 192                         }
 193                 }
 194                 return false;
 195         }
 196
 197         function findBySha1( $hash ) {
 198                 $results = $this->fetchImageQuery( array(
 199                                                                                 'aisha1base36' => $hash,
 200                                                                                 'aiprop'       => ForeignAPIFile::getProps(),
 201                                                                                 'list'         => 'allimages', ) );
 202                 $ret = array();
 203                 if ( isset( $results['query']['allimages'] ) ) {
 204                         foreach ( $results['query']['allimages'] as $img ) {
 205                                 // 1.14 was broken, doesn't return name attribute
 206                                 if( !isset( $img['name'] ) ) {
 207                                         continue;
 208                                 }
 209                                 $ret[] = new ForeignAPIFile( Title::makeTitle( NS_FILE, $img['name'] ), $this, $img );
 210                         }
 211                 }
 212                 return $ret;
 213         }
 214
 215         function getThumbUrl( $name, $width = -1, $height = -1, &$result = null, $otherParams = '' ) {
 216                 $data = $this->fetchImageQuery( array(
 217                         'titles' => 'File:' . $name,
 218                         'iiprop' => 'url|timestamp',
 219                         'iiurlwidth' => $width,
 220                         'iiurlheight' => $height,
 221                         'iiurlparam'  => $otherParams,
 222                         'prop' => 'imageinfo' ) );
 223                 $info = $this->getImageInfo( $data );
 224
 225                 if( $data && $info && isset( $info['thumburl'] ) ) {
 226                         wfDebug( __METHOD__ . " got remote thumb " . $info['thumburl'] . "\n" );
 227                         $result = $info;
 228                         return $info['thumburl'];
 229                 } else {
 230                         return false;
 231                 }
 232         }
 233
 234         /**
 235          * Return the imageurl from cache if possible
 236          *
 237          * If the url has been requested today, get it from cache
 238          * Otherwise retrieve remote thumb url, check for local file.
 239          *
 240          * @param $name String is a dbkey form of a title
 241          * @param $width
 242          * @param $height
 243          * @param String $param Other rendering parameters (page number, etc) from handler's makeParamString.
 244          */
 245         function getThumbUrlFromCache( $name, $width, $height, $params="" ) {
 246                 global $wgMemc;
 247
 248                 if ( !$this->canCacheThumbs() ) {
 249                         $result = null; // can't pass "null" by reference, but it's ok as default value
 250                         return $this->getThumbUrl( $name, $width, $height, $result, $params );
 251                 }
 252                 $key = $this->getLocalCacheKey( 'ForeignAPIRepo', 'ThumbUrl', $name );
 253                 $sizekey = "$width:$height:$params";
 254
 255                 /* Get the array of urls that we already know */
 256                 $knownThumbUrls = $wgMemc->get($key);
 257                 if( !$knownThumbUrls ) {
 258                         /* No knownThumbUrls for this file */
 259                         $knownThumbUrls = array();
 260                 } else {
 261                         if( isset( $knownThumbUrls[$sizekey] ) ) {
 262                                 wfDebug( __METHOD__ . ': Got thumburl from local cache: ' .
 263                                         "{$knownThumbUrls[$sizekey]} \n");
 264                                 return $knownThumbUrls[$sizekey];
 265                         }
 266                         /* This size is not yet known */
 267                 }
 268
 269                 $metadata = null;
 270                 $foreignUrl = $this->getThumbUrl( $name, $width, $height, $metadata, $params );
 271
 272                 if( !$foreignUrl ) {
 273                         wfDebug( __METHOD__ . " Could not find thumburl\n" );
 274                         return false;
 275                 }
 276
 277                 // We need the same filename as the remote one :)
 278                 $fileName = rawurldecode( pathinfo( $foreignUrl, PATHINFO_BASENAME ) );
 279                 if( !$this->validateFilename( $fileName ) ) {
 280                         wfDebug( __METHOD__ . " The deduced filename $fileName is not safe\n" );
 281                         return false;
 282                 }
 283                 $localPath =  $this->getZonePath( 'thumb' ) . "/" . $this->getHashPath( $name ) . $name;
 284                 $localFilename = $localPath . "/" . $fileName;
 285                 $localUrl =  $this->getZoneUrl( 'thumb' ) . "/" . $this->getHashPath( $name ) . rawurlencode( $name ) . "/" . rawurlencode( $fileName );
 286
 287                 if( file_exists( $localFilename ) && isset( $metadata['timestamp'] ) ) {
 288                         wfDebug( __METHOD__ . " Thumbnail was already downloaded before\n" );
 289                         $modified = filemtime( $localFilename );
 290                         $remoteModified = strtotime( $metadata['timestamp'] );
 291                         $current = time();
 292                         $diff = abs( $modified - $current );
 293                         if( $remoteModified < $modified && $diff < $this->fileCacheExpiry ) {
 294                                 /* Use our current and already downloaded thumbnail */
 295                                 $knownThumbUrls[$sizekey] = $localUrl;
 296                                 $wgMemc->set( $key, $knownThumbUrls, $this->apiThumbCacheExpiry );
 297                                 return $localUrl;
 298                         }
 299                         /* There is a new Commons file, or existing thumbnail older than a month */
 300                 }
 301                 $thumb = self::httpGet( $foreignUrl );
 302                 if( !$thumb ) {
 303                         wfDebug( __METHOD__ . " Could not download thumb\n" );
 304                         return false;
 305                 }
 306                 if ( !is_dir($localPath) ) {
 307                         if( !wfMkdirParents( $localPath, null, __METHOD__ ) ) {
 308                                 wfDebug(  __METHOD__ . " could not create directory $localPath for thumb\n" );
 309                                 return $foreignUrl;
 310                         }
 311                 }
 312
 313                 # @todo FIXME: Delete old thumbs that aren't being used. Maintenance script?
 314                 wfSuppressWarnings();
 315                 if( !file_put_contents( $localFilename, $thumb ) ) {
 316                         wfRestoreWarnings();
 317                         wfDebug( __METHOD__ . " could not write to thumb path\n" );
 318                         return $foreignUrl;
 319                 }
 320                 wfRestoreWarnings();
 321                 $knownThumbUrls[$sizekey] = $localUrl;
 322                 $wgMemc->set( $key, $knownThumbUrls, $this->apiThumbCacheExpiry );
 323                 wfDebug( __METHOD__ . " got local thumb $localUrl, saving to cache \n" );
 324                 return $localUrl;
 325         }
 326
 327         /**
 328          * @see FileRepo::getZoneUrl()
 329          */
 330         function getZoneUrl( $zone ) {
 331                 switch ( $zone ) {
 332                         case 'public':
 333                                 return $this->url;
 334                         case 'thumb':
 335                                 return $this->thumbUrl;
 336                         default:
 337                                 return parent::getZoneUrl( $zone );
 338                 }
 339         }
 340
 341         /**
 342          * Get the local directory corresponding to one of the three basic zones
 343          */
 344         function getZonePath( $zone ) {
 345                 switch ( $zone ) {
 346                         case 'public':
 347                                 return $this->directory;
 348                         case 'thumb':
 349                                 return $this->thumbDir;
 350                         default:
 351                                 return false;
 352                 }
 353         }
 354
 355         /**
 356          * Are we locally caching the thumbnails?
 357          * @return bool
 358          */
 359         public function canCacheThumbs() {
 360                 return ( $this->apiThumbCacheExpiry > 0 );
 361         }
 362
 363         /**
 364          * The user agent the ForeignAPIRepo will use.
 365          */
 366         public static function getUserAgent() {
 367                 return Http::userAgent() . " ForeignAPIRepo/" . self::VERSION;
 368         }
 369
 370         /**
 371          * Like a Http:get request, but with custom User-Agent.
 372          * @see Http:get
 373          */
 374         public static function httpGet( $url, $timeout = 'default', $options = array() ) {
 375                 $options['timeout'] = $timeout;
 376                 /* Http::get */
 377                 $url = wfExpandUrl( $url, PROTO_HTTP );
 378                 wfDebug( "ForeignAPIRepo: HTTP GET: $url\n" );
 379                 $options['method'] = "GET";
 380
 381                 if ( !isset( $options['timeout'] ) ) {
 382                         $options['timeout'] = 'default';
 383                 }
 384
 385                 $req = MWHttpRequest::factory( $url, $options );
 386                 $req->setUserAgent( ForeignAPIRepo::getUserAgent() );
 387                 $status = $req->execute();
 388
 389                 if ( $status->isOK() ) {
 390                         return $req->getContent();
 391                 } else {
 392                         return false;
 393                 }
 394         }
 395 }