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 = "2.0";
  29
  30         var $fileFactory = array( 'ForeignAPIFile', 'newFromTitle' );
  31         /* Check back with Commons after a day */
  32         var $apiThumbCacheExpiry = 86400;
  33         /* Redownload thumbnail files after a month */
  34         var $fileCacheExpiry = 2629743;
  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         function newFile( $title, $time = false ) {
  80                 if ( $time ) {
  81                         return false;
  82                 }
  83                 return parent::newFile( $title, $time );
  84         }
  85
  86 /**
  87  * No-ops
  88  */
  89         function storeBatch( $triplets, $flags = 0 ) {
  90                 return false;
  91         }
  92         function storeTemp( $originalName, $srcPath ) {
  93                 return false;
  94         }
  95         function append( $srcPath, $toAppendPath, $flags = 0 ){
  96                 return false;
  97         }
  98         function publishBatch( $triplets, $flags = 0 ) {
  99                 return false;
 100         }
 101         function deleteBatch( $sourceDestPairs ) {
 102                 return false;
 103         }
 104
 105
 106         function fileExistsBatch( $files, $flags = 0 ) {
 107                 $results = array();
 108                 foreach ( $files as $k => $f ) {
 109                         if ( isset( $this->mFileExists[$k] ) ) {
 110                                 $results[$k] = true;
 111                                 unset( $files[$k] );
 112                         } elseif( self::isVirtualUrl( $f ) ) {
 113                                 # TODO! FIXME! We need to be able to handle virtual
 114                                 # URLs better, at least when we know they refer to the
 115                                 # same repo.
 116                                 $results[$k] = false;
 117                                 unset( $files[$k] );
 118                         }
 119                 }
 120
 121                 $results = $this->fetchImageQuery( array( 'titles' => implode( $files, '|' ),
 122                                                                                         'prop' => 'imageinfo' ) );
 123                 if( isset( $data['query']['pages'] ) ) {
 124                         $i = 0;
 125                         foreach( $files as $key => $file ) {
 126                                 $results[$key] = $this->mFileExists[$key] = !isset( $data['query']['pages'][$i]['missing'] );
 127                                 $i++;
 128                         }
 129                 }
 130         }
 131         function getFileProps( $virtualUrl ) {
 132                 return false;
 133         }
 134
 135         function fetchImageQuery( $query ) {
 136                 global $wgMemc;
 137
 138                 $query = array_merge( $query,
 139                         array(
 140                                 'format' => 'json',
 141                                 'action' => 'query',
 142                                 'redirects' => 'true'
 143                         ) );
 144                 if ( $this->mApiBase ) {
 145                         $url = wfAppendQuery( $this->mApiBase, $query );
 146                 } else {
 147                         $url = $this->makeUrl( $query, 'api' );
 148                 }
 149
 150                 if( !isset( $this->mQueryCache[$url] ) ) {
 151                         $key = $this->getLocalCacheKey( 'ForeignAPIRepo', 'Metadata', md5( $url ) );
 152                         $data = $wgMemc->get( $key );
 153                         if( !$data ) {
 154                                 $data = self::httpGet( $url );
 155                                 if ( !$data ) {
 156                                         return null;
 157                                 }
 158                                 $wgMemc->set( $key, $data, 3600 );
 159                         }
 160
 161                         if( count( $this->mQueryCache ) > 100 ) {
 162                                 // Keep the cache from growing infinitely
 163                                 $this->mQueryCache = array();
 164                         }
 165                         $this->mQueryCache[$url] = $data;
 166                 }
 167                 return FormatJson::decode( $this->mQueryCache[$url], true );
 168         }
 169
 170         function getImageInfo( $data ) {
 171                 if( $data && isset( $data['query']['pages'] ) ) {
 172                         foreach( $data['query']['pages'] as $info ) {
 173                                 if( isset( $info['imageinfo'][0] ) ) {
 174                                         return $info['imageinfo'][0];
 175                                 }
 176                         }
 177                 }
 178                 return false;
 179         }
 180
 181         function findBySha1( $hash ) {
 182                 $results = $this->fetchImageQuery( array(
 183                                                                                 'aisha1base36' => $hash,
 184                                                                                 'aiprop'       => ForeignAPIFile::getProps(),
 185                                                                                 'list'         => 'allimages', ) );
 186                 $ret = array();
 187                 if ( isset( $results['query']['allimages'] ) ) {
 188                         foreach ( $results['query']['allimages'] as $img ) {
 189                                 // 1.14 was broken, doesn't return name attribute
 190                                 if( !isset( $img['name'] ) ) {
 191                                         continue;
 192                                 }
 193                                 $ret[] = new ForeignAPIFile( Title::makeTitle( NS_FILE, $img['name'] ), $this, $img );
 194                         }
 195                 }
 196                 return $ret;
 197         }
 198
 199         function getThumbUrl( $name, $width=-1, $height=-1, &$result ) {
 200                 $data = $this->fetchImageQuery( array(
 201                         'titles' => 'File:' . $name,
 202                         'iiprop' => 'url|timestamp',
 203                         'iiurlwidth' => $width,
 204                         'iiurlheight' => $height,
 205                         'prop' => 'imageinfo' ) );
 206                 $info = $this->getImageInfo( $data );
 207
 208                 if( $data && $info && isset( $info['thumburl'] ) ) {
 209                         wfDebug( __METHOD__ . " got remote thumb " . $info['thumburl'] . "\n" );
 210                         $result = $info;
 211                         return $info['thumburl'];
 212                 } else {
 213                         return false;
 214                 }
 215         }
 216
 217         /*
 218          * Return the imageurl from cache if possible
 219          *
 220          * If the url has been requested today, get it from cache
 221          * Otherwise retrieve remote thumb url, check for local file.
 222          *
 223          * @param $name String is a dbkey form of a title
 224          * @param $width
 225          * @param $height
 226          */
 227         function getThumbUrlFromCache( $name, $width, $height ) {
 228                 global $wgMemc;
 229
 230                 if ( !$this->canCacheThumbs() ) {
 231                         return $this->getThumbUrl( $name, $width, $height );
 232                 }
 233                 $key = $this->getLocalCacheKey( 'ForeignAPIRepo', 'ThumbUrl', $name );
 234                 $sizekey = "$width:$height";
 235
 236                 /* Get the array of urls that we already know */
 237                 $knownThumbUrls = $wgMemc->get($key);
 238                 if( !$knownThumbUrls ) {
 239                         /* No knownThumbUrls for this file */
 240                         $knownThumbUrls = array();
 241                 } else {
 242                         if( isset( $knownThumbUrls[$sizekey] ) ) {
 243                                 wfDebug("Got thumburl from local cache. {$knownThumbUrls[$sizekey]} \n");
 244                                 return $knownThumbUrls[$sizekey];
 245                         }
 246                         /* This size is not yet known */
 247                 }
 248
 249                 $metadata = null;
 250                 $foreignUrl = $this->getThumbUrl( $name, $width, $height, $metadata );
 251
 252                 if( !$foreignUrl ) {
 253                         wfDebug( __METHOD__ . " Could not find thumburl\n" );
 254                         return false;
 255                 }
 256
 257                 // We need the same filename as the remote one :)
 258                 $fileName = rawurldecode( pathinfo( $foreignUrl, PATHINFO_BASENAME ) );
 259                 if( !$this->validateFilename( $fileName ) ) {
 260                         wfDebug( __METHOD__ . " The deduced filename $fileName is not safe\n" );
 261                         return false;
 262                 }
 263                 $localPath =  $this->getZonePath( 'thumb' ) . "/" . $this->getHashPath( $name ) . $name;
 264                 $localFilename = $localPath . "/" . $fileName;
 265                 $localUrl =  $this->getZoneUrl( 'thumb' ) . "/" . $this->getHashPath( $name ) . rawurlencode( $name ) . "/" . rawurlencode( $fileName );
 266
 267                 if( file_exists( $localFilename ) && isset( $metadata['timestamp'] ) ) {
 268                         wfDebug( __METHOD__ . " Thumbnail was already downloaded before\n" );
 269                         $modified = filemtime( $localFilename );
 270                         $remoteModified = strtotime( $metadata['timestamp'] );
 271                         $current = time();
 272                         $diff = abs( $modified - $current );
 273                         if( $remoteModified < $modified && $diff < $this->fileCacheExpiry ) {
 274                                 /* Use our current and already downloaded thumbnail */
 275                                 $knownThumbUrls["$width:$height"] = $localUrl;
 276                                 $wgMemc->set( $key, $knownThumbUrls, $this->apiThumbCacheExpiry );
 277                                 return $localUrl;
 278                         }
 279                         /* There is a new Commons file, or existing thumbnail older than a month */
 280                 }
 281                 $thumb = self::httpGet( $foreignUrl );
 282                 if( !$thumb ) {
 283                         wfDebug( __METHOD__ . " Could not download thumb\n" );
 284                         return false;
 285                 }
 286                 if ( !is_dir($localPath) ) {
 287                         if( !wfMkdirParents($localPath) ) {
 288                                 wfDebug(  __METHOD__ . " could not create directory $localPath for thumb\n" );
 289                                 return $foreignUrl;
 290                         }
 291                 }
 292
 293                 # FIXME: Delete old thumbs that aren't being used. Maintenance script?
 294                 wfSuppressWarnings();
 295                 if( !file_put_contents( $localFilename, $thumb ) ) {
 296                         wfRestoreWarnings();
 297                         wfDebug( __METHOD__ . " could not write to thumb path\n" );
 298                         return $foreignUrl;
 299                 }
 300                 wfRestoreWarnings();
 301                 $knownThumbUrls[$sizekey] = $localUrl;
 302                 $wgMemc->set( $key, $knownThumbUrls, $this->apiThumbCacheExpiry );
 303                 wfDebug( __METHOD__ . " got local thumb $localUrl, saving to cache \n" );
 304                 return $localUrl;
 305         }
 306
 307         /**
 308          * @see FileRepo::getZoneUrl()
 309          */
 310         function getZoneUrl( $zone ) {
 311                 switch ( $zone ) {
 312                         case 'public':
 313                                 return $this->url;
 314                         case 'thumb':
 315                                 return $this->thumbUrl;
 316                         default:
 317                                 return parent::getZoneUrl( $zone );
 318                 }
 319         }
 320
 321         /**
 322          * Get the local directory corresponding to one of the three basic zones
 323          */
 324         function getZonePath( $zone ) {
 325                 switch ( $zone ) {
 326                         case 'public':
 327                                 return $this->directory;
 328                         case 'thumb':
 329                                 return $this->thumbDir;
 330                         default:
 331                                 return false;
 332                 }
 333         }
 334
 335         /**
 336          * Are we locally caching the thumbnails?
 337          * @return bool
 338          */
 339         public function canCacheThumbs() {
 340                 return ( $this->apiThumbCacheExpiry > 0 );
 341         }
 342
 343         /**
 344          * The user agent the ForeignAPIRepo will use.
 345          */
 346         public static function getUserAgent() {
 347                 return Http::userAgent() . " ForeignAPIRepo/" . self::VERSION;
 348         }
 349
 350         /**
 351          * Like a Http:get request, but with custom User-Agent.
 352          * @see Http:get
 353          */
 354         public static function httpGet( $url, $timeout = 'default', $options = array() ) {
 355                 $options['timeout'] = $timeout;
 356                 /* Http::get */
 357                 $url = wfExpandUrl( $url );
 358                 wfDebug( "ForeignAPIRepo: HTTP GET: $url\n" );
 359                 $options['method'] = "GET";
 360
 361                 if ( !isset( $options['timeout'] ) ) {
 362                         $options['timeout'] = 'default';
 363                 }
 364
 365                 $req = MWHttpRequest::factory( $url, $options );
 366                 $req->setUserAgent( ForeignAPIRepo::getUserAgent() );
 367                 $status = $req->execute();
 368
 369                 if ( $status->isOK() ) {
 370                         return $req->getContent();
 371                 } else {
 372                         return false;
 373                 }
 374         }
 375 }