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