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