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; /* 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         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         function fileExistsBatch( $files, $flags = 0 ) {
 106                 $results = array();
 107                 foreach ( $files as $k => $f ) {
 108                         if ( isset( $this->mFileExists[$k] ) ) {
 109                                 $results[$k] = true;
 110                                 unset( $files[$k] );
 111                         } elseif( self::isVirtualUrl( $f ) ) {
 112                                 # TODO! FIXME! We need to be able to handle virtual
 113                                 # URLs better, at least when we know they refer to the
 114                                 # same repo.
 115                                 $results[$k] = false;
 116                                 unset( $files[$k] );
 117                         }
 118                 }
 119
 120                 $data = $this->fetchImageQuery( array( 'titles' => implode( $files, '|' ),
 121                                                                                         'prop' => 'imageinfo' ) );
 122                 if( isset( $data['query']['pages'] ) ) {
 123                         $i = 0;
 124                         foreach( $files as $key => $file ) {
 125                                 $results[$key] = $this->mFileExists[$key] = !isset( $data['query']['pages'][$i]['missing'] );
 126                                 $i++;
 127                         }
 128                 }
 129                 return $results;
 130         }
 131
 132         function getFileProps( $virtualUrl ) {
 133                 return false;
 134         }
 135
 136         function fetchImageQuery( $query ) {
 137                 global $wgMemc;
 138
 139                 $query = array_merge( $query,
 140                         array(
 141                                 'format' => 'json',
 142                                 'action' => 'query',
 143                                 'redirects' => 'true'
 144                         ) );
 145                 if ( $this->mApiBase ) {
 146                         $url = wfAppendQuery( $this->mApiBase, $query );
 147                 } else {
 148                         $url = $this->makeUrl( $query, 'api' );
 149                 }
 150
 151                 if( !isset( $this->mQueryCache[$url] ) ) {
 152                         $key = $this->getLocalCacheKey( 'ForeignAPIRepo', 'Metadata', md5( $url ) );
 153                         $data = $wgMemc->get( $key );
 154                         if( !$data ) {
 155                                 $data = self::httpGet( $url );
 156                                 if ( !$data ) {
 157                                         return null;
 158                                 }
 159                                 $wgMemc->set( $key, $data, 3600 );
 160                         }
 161
 162                         if( count( $this->mQueryCache ) > 100 ) {
 163                                 // Keep the cache from growing infinitely
 164                                 $this->mQueryCache = array();
 165                         }
 166                         $this->mQueryCache[$url] = $data;
 167                 }
 168                 return FormatJson::decode( $this->mQueryCache[$url], true );
 169         }
 170
 171         function getImageInfo( $data ) {
 172                 if( $data && isset( $data['query']['pages'] ) ) {
 173                         foreach( $data['query']['pages'] as $info ) {
 174                                 if( isset( $info['imageinfo'][0] ) ) {
 175                                         return $info['imageinfo'][0];
 176                                 }
 177                         }
 178                 }
 179                 return false;
 180         }
 181
 182         function findBySha1( $hash ) {
 183                 $results = $this->fetchImageQuery( array(
 184                                                                                 'aisha1base36' => $hash,
 185                                                                                 'aiprop'       => ForeignAPIFile::getProps(),
 186                                                                                 'list'         => 'allimages', ) );
 187                 $ret = array();
 188                 if ( isset( $results['query']['allimages'] ) ) {
 189                         foreach ( $results['query']['allimages'] as $img ) {
 190                                 // 1.14 was broken, doesn't return name attribute
 191                                 if( !isset( $img['name'] ) ) {
 192                                         continue;
 193                                 }
 194                                 $ret[] = new ForeignAPIFile( Title::makeTitle( NS_FILE, $img['name'] ), $this, $img );
 195                         }
 196                 }
 197                 return $ret;
 198         }
 199
 200         function getThumbUrl( $name, $width = -1, $height = -1, &$result = null ) {
 201                 $data = $this->fetchImageQuery( array(
 202                         'titles' => 'File:' . $name,
 203                         'iiprop' => 'url|timestamp',
 204                         'iiurlwidth' => $width,
 205                         'iiurlheight' => $height,
 206                         'prop' => 'imageinfo' ) );
 207                 $info = $this->getImageInfo( $data );
 208
 209                 if( $data && $info && isset( $info['thumburl'] ) ) {
 210                         wfDebug( __METHOD__ . " got remote thumb " . $info['thumburl'] . "\n" );
 211                         $result = $info;
 212                         return $info['thumburl'];
 213                 } else {
 214                         return false;
 215                 }
 216         }
 217
 218         /*
 219          * Return the imageurl from cache if possible
 220          *
 221          * If the url has been requested today, get it from cache
 222          * Otherwise retrieve remote thumb url, check for local file.
 223          *
 224          * @param $name String is a dbkey form of a title
 225          * @param $width
 226          * @param $height
 227          */
 228         function getThumbUrlFromCache( $name, $width, $height ) {
 229                 global $wgMemc;
 230
 231                 if ( !$this->canCacheThumbs() ) {
 232                         return $this->getThumbUrl( $name, $width, $height );
 233                 }
 234                 $key = $this->getLocalCacheKey( 'ForeignAPIRepo', 'ThumbUrl', $name );
 235                 $sizekey = "$width:$height";
 236
 237                 /* Get the array of urls that we already know */
 238                 $knownThumbUrls = $wgMemc->get($key);
 239                 if( !$knownThumbUrls ) {
 240                         /* No knownThumbUrls for this file */
 241                         $knownThumbUrls = array();
 242                 } else {
 243                         if( isset( $knownThumbUrls[$sizekey] ) ) {
 244                                 wfDebug("Got thumburl from local cache. {$knownThumbUrls[$sizekey]} \n");
 245                                 return $knownThumbUrls[$sizekey];
 246                         }
 247                         /* This size is not yet known */
 248                 }
 249
 250                 $metadata = null;
 251                 $foreignUrl = $this->getThumbUrl( $name, $width, $height, $metadata );
 252
 253                 if( !$foreignUrl ) {
 254                         wfDebug( __METHOD__ . " Could not find thumburl\n" );
 255                         return false;
 256                 }
 257
 258                 // We need the same filename as the remote one :)
 259                 $fileName = rawurldecode( pathinfo( $foreignUrl, PATHINFO_BASENAME ) );
 260                 if( !$this->validateFilename( $fileName ) ) {
 261                         wfDebug( __METHOD__ . " The deduced filename $fileName is not safe\n" );
 262                         return false;
 263                 }
 264                 $localPath =  $this->getZonePath( 'thumb' ) . "/" . $this->getHashPath( $name ) . $name;
 265                 $localFilename = $localPath . "/" . $fileName;
 266                 $localUrl =  $this->getZoneUrl( 'thumb' ) . "/" . $this->getHashPath( $name ) . rawurlencode( $name ) . "/" . rawurlencode( $fileName );
 267
 268                 if( file_exists( $localFilename ) && isset( $metadata['timestamp'] ) ) {
 269                         wfDebug( __METHOD__ . " Thumbnail was already downloaded before\n" );
 270                         $modified = filemtime( $localFilename );
 271                         $remoteModified = strtotime( $metadata['timestamp'] );
 272                         $current = time();
 273                         $diff = abs( $modified - $current );
 274                         if( $remoteModified < $modified && $diff < $this->fileCacheExpiry ) {
 275                                 /* Use our current and already downloaded thumbnail */
 276                                 $knownThumbUrls["$width:$height"] = $localUrl;
 277                                 $wgMemc->set( $key, $knownThumbUrls, $this->apiThumbCacheExpiry );
 278                                 return $localUrl;
 279                         }
 280                         /* There is a new Commons file, or existing thumbnail older than a month */
 281                 }
 282                 $thumb = self::httpGet( $foreignUrl );
 283                 if( !$thumb ) {
 284                         wfDebug( __METHOD__ . " Could not download thumb\n" );
 285                         return false;
 286                 }
 287                 if ( !is_dir($localPath) ) {
 288                         if( !wfMkdirParents($localPath) ) {
 289                                 wfDebug(  __METHOD__ . " could not create directory $localPath for thumb\n" );
 290                                 return $foreignUrl;
 291                         }
 292                 }
 293
 294                 # FIXME: Delete old thumbs that aren't being used. Maintenance script?
 295                 wfSuppressWarnings();
 296                 if( !file_put_contents( $localFilename, $thumb ) ) {
 297                         wfRestoreWarnings();
 298                         wfDebug( __METHOD__ . " could not write to thumb path\n" );
 299                         return $foreignUrl;
 300                 }
 301                 wfRestoreWarnings();
 302                 $knownThumbUrls[$sizekey] = $localUrl;
 303                 $wgMemc->set( $key, $knownThumbUrls, $this->apiThumbCacheExpiry );
 304                 wfDebug( __METHOD__ . " got local thumb $localUrl, saving to cache \n" );
 305                 return $localUrl;
 306         }
 307
 308         /**
 309          * @see FileRepo::getZoneUrl()
 310          */
 311         function getZoneUrl( $zone ) {
 312                 switch ( $zone ) {
 313                         case 'public':
 314                                 return $this->url;
 315                         case 'thumb':
 316                                 return $this->thumbUrl;
 317                         default:
 318                                 return parent::getZoneUrl( $zone );
 319                 }
 320         }
 321
 322         /**
 323          * Get the local directory corresponding to one of the three basic zones
 324          */
 325         function getZonePath( $zone ) {
 326                 switch ( $zone ) {
 327                         case 'public':
 328                                 return $this->directory;
 329                         case 'thumb':
 330                                 return $this->thumbDir;
 331                         default:
 332                                 return false;
 333                 }
 334         }
 335
 336         /**
 337          * Are we locally caching the thumbnails?
 338          * @return bool
 339          */
 340         public function canCacheThumbs() {
 341                 return ( $this->apiThumbCacheExpiry > 0 );
 342         }
 343
 344         /**
 345          * The user agent the ForeignAPIRepo will use.
 346          */
 347         public static function getUserAgent() {
 348                 return Http::userAgent() . " ForeignAPIRepo/" . self::VERSION;
 349         }
 350
 351         /**
 352          * Like a Http:get request, but with custom User-Agent.
 353          * @see Http:get
 354          */
 355         public static function httpGet( $url, $timeout = 'default', $options = array() ) {
 356                 $options['timeout'] = $timeout;
 357                 /* Http::get */
 358                 $url = wfExpandUrl( $url );
 359                 wfDebug( "ForeignAPIRepo: HTTP GET: $url\n" );
 360                 $options['method'] = "GET";
 361
 362                 if ( !isset( $options['timeout'] ) ) {
 363                         $options['timeout'] = 'default';
 364                 }
 365
 366                 $req = MWHttpRequest::factory( $url, $options );
 367                 $req->setUserAgent( ForeignAPIRepo::getUserAgent() );
 368                 $status = $req->execute();
 369
 370                 if ( $status->isOK() ) {
 371                         return $req->getContent();
 372                 } else {
 373                         return false;
 374                 }
 375         }
 376 }