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         var $fileFactory = array( 'ForeignAPIFile', 'newFromTitle' );
  26         /* Check back with Commons after a day */
  27         var $apiThumbCacheExpiry = 86400;
  28         /* Redownload thumbnail files after a month */
  29         var $fileCacheExpiry = 2629743;
  30         /* Local image directory */
  31         var $directory;
  32         var $thumbDir;
  33
  34         protected $mQueryCache = array();
  35         protected $mFileExists = array();
  36
  37         function __construct( $info ) {
  38                 parent::__construct( $info );
  39                 global $wgUploadDirectory;
  40
  41                 // http://commons.wikimedia.org/w/api.php
  42                 $this->mApiBase = isset( $info['apibase'] ) ? $info['apibase'] : null;
  43                 $this->directory = isset( $info['directory'] ) ? $info['directory'] : $wgUploadDirectory;
  44
  45                 if( isset( $info['apiThumbCacheExpiry'] ) ) {
  46                         $this->apiThumbCacheExpiry = $info['apiThumbCacheExpiry'];
  47                 }
  48                 if( isset( $info['fileCacheExpiry'] ) ) {
  49                         $this->fileCacheExpiry = $info['fileCacheExpiry'];
  50                 }
  51                 if( !$this->scriptDirUrl ) {
  52                         // hack for description fetches
  53                         $this->scriptDirUrl = dirname( $this->mApiBase );
  54                 }
  55                 // If we can cache thumbs we can guess sane defaults for these
  56                 if( $this->canCacheThumbs() && !$this->url ) {
  57                         global $wgLocalFileRepo;
  58                         $this->url = $wgLocalFileRepo['url'];
  59                 }
  60                 if( $this->canCacheThumbs() && !$this->thumbUrl ) {
  61                         $this->thumbUrl = $this->url . '/thumb';
  62                 }
  63                 if ( isset( $info['thumbDir'] ) ) {
  64                         $this->thumbDir =  $info['thumbDir'];
  65                 } else {
  66                         $this->thumbDir = "{$this->directory}/thumb";
  67                 }
  68         }
  69
  70         /**
  71          * Per docs in FileRepo, this needs to return false if we don't support versioned
  72          * files. Well, we don't.
  73          */
  74         function newFile( $title, $time = false ) {
  75                 if ( $time ) {
  76                         return false;
  77                 }
  78                 return parent::newFile( $title, $time );
  79         }
  80
  81 /**
  82  * No-ops
  83  */
  84         function storeBatch( $triplets, $flags = 0 ) {
  85                 return false;
  86         }
  87         function storeTemp( $originalName, $srcPath ) {
  88                 return false;
  89         }
  90         function append( $srcPath, $toAppendPath, $flags = 0 ){
  91                 return false;
  92         }
  93         function publishBatch( $triplets, $flags = 0 ) {
  94                 return false;
  95         }
  96         function deleteBatch( $sourceDestPairs ) {
  97                 return false;
  98         }
  99
 100
 101         function fileExistsBatch( $files, $flags = 0 ) {
 102                 $results = array();
 103                 foreach ( $files as $k => $f ) {
 104                         if ( isset( $this->mFileExists[$k] ) ) {
 105                                 $results[$k] = true;
 106                                 unset( $files[$k] );
 107                         } elseif( self::isVirtualUrl( $f ) ) {
 108                                 # TODO! FIXME! We need to be able to handle virtual
 109                                 # URLs better, at least when we know they refer to the
 110                                 # same repo.
 111                                 $results[$k] = false;
 112                                 unset( $files[$k] );
 113                         }
 114                 }
 115
 116                 $results = $this->fetchImageQuery( array( 'titles' => implode( $files, '|' ),
 117                                                                                         'prop' => 'imageinfo' ) );
 118                 if( isset( $data['query']['pages'] ) ) {
 119                         $i = 0;
 120                         foreach( $files as $key => $file ) {
 121                                 $results[$key] = $this->mFileExists[$key] = !isset( $data['query']['pages'][$i]['missing'] );
 122                                 $i++;
 123                         }
 124                 }
 125         }
 126         function getFileProps( $virtualUrl ) {
 127                 return false;
 128         }
 129
 130         function fetchImageQuery( $query ) {
 131                 global $wgMemc;
 132
 133                 $query = array_merge( $query,
 134                         array(
 135                                 'format' => 'json',
 136                                 'action' => 'query',
 137                                 'redirects' => 'true'
 138                         ) );
 139                 if ( $this->mApiBase ) {
 140                         $url = wfAppendQuery( $this->mApiBase, $query );
 141                 } else {
 142                         $url = $this->makeUrl( $query, 'api' );
 143                 }
 144
 145                 if( !isset( $this->mQueryCache[$url] ) ) {
 146                         $key = $this->getLocalCacheKey( 'ForeignAPIRepo', 'Metadata', md5( $url ) );
 147                         $data = $wgMemc->get( $key );
 148                         if( !$data ) {
 149                                 $data = Http::get( $url );
 150                                 if ( !$data ) {
 151                                         return null;
 152                                 }
 153                                 $wgMemc->set( $key, $data, 3600 );
 154                         }
 155
 156                         if( count( $this->mQueryCache ) > 100 ) {
 157                                 // Keep the cache from growing infinitely
 158                                 $this->mQueryCache = array();
 159                         }
 160                         $this->mQueryCache[$url] = $data;
 161                 }
 162                 return FormatJson::decode( $this->mQueryCache[$url], true );
 163         }
 164
 165         function getImageInfo( $data ) {
 166                 if( $data && isset( $data['query']['pages'] ) ) {
 167                         foreach( $data['query']['pages'] as $info ) {
 168                                 if( isset( $info['imageinfo'][0] ) ) {
 169                                         return $info['imageinfo'][0];
 170                                 }
 171                         }
 172                 }
 173                 return false;
 174         }
 175
 176         function findBySha1( $hash ) {
 177                 $results = $this->fetchImageQuery( array(
 178                                                                                 'aisha1base36' => $hash,
 179                                                                                 'aiprop'       => ForeignAPIFile::getProps(),
 180                                                                                 'list'         => 'allimages', ) );
 181                 $ret = array();
 182                 if ( isset( $results['query']['allimages'] ) ) {
 183                         foreach ( $results['query']['allimages'] as $img ) {
 184                                 // 1.14 was broken, doesn't return name attribute
 185                                 if( !isset( $img['name'] ) ) {
 186                                         continue;
 187                                 }
 188                                 $ret[] = new ForeignAPIFile( Title::makeTitle( NS_FILE, $img['name'] ), $this, $img );
 189                         }
 190                 }
 191                 return $ret;
 192         }
 193
 194         function getThumbUrl( $name, $width=-1, $height=-1, &$result ) {
 195                 $data = $this->fetchImageQuery( array(
 196                         'titles' => 'File:' . $name,
 197                         'iiprop' => 'url|timestamp',
 198                         'iiurlwidth' => $width,
 199                         'iiurlheight' => $height,
 200                         'prop' => 'imageinfo' ) );
 201                 $info = $this->getImageInfo( $data );
 202
 203                 if( $data && $info && isset( $info['thumburl'] ) ) {
 204                         wfDebug( __METHOD__ . " got remote thumb " . $info['thumburl'] . "\n" );
 205                         $result = $info;
 206                         return $info['thumburl'];
 207                 } else {
 208                         return false;
 209                 }
 210         }
 211
 212         /*
 213          * Return the imageurl from cache if possible
 214          *
 215          * If the url has been requested today, get it from cache
 216          * Otherwise retrieve remote thumb url, check for local file.
 217          *
 218          * @param $name String is a dbkey form of a title
 219          * @param $width
 220          * @param $height
 221          */
 222         function getThumbUrlFromCache( $name, $width, $height ) {
 223                 global $wgMemc;
 224
 225                 if ( !$this->canCacheThumbs() ) {
 226                         return $this->getThumbUrl( $name, $width, $height );
 227                 }
 228                 $key = $this->getLocalCacheKey( 'ForeignAPIRepo', 'ThumbUrl', $name, $width );
 229
 230                 $thumbUrl = $wgMemc->get($key);
 231                 if ( $thumbUrl ) {
 232                         wfDebug("Got thumb from local cache. $thumbUrl \n");
 233                         return $thumbUrl;
 234                 }
 235                 else {
 236                         $metadata = null;
 237                         $foreignUrl = $this->getThumbUrl( $name, $width, $height, $metadata );
 238
 239                         if( !$foreignUrl ) {
 240                                 wfDebug( __METHOD__ . " Could not find thumburl\n" );
 241                                 return false;
 242                         }
 243
 244                         // We need the same filename as the remote one :)
 245                         $fileName = rawurldecode( pathinfo( $foreignUrl, PATHINFO_BASENAME ) );
 246                         if( !$this->validateFilename( $fileName ) ) {
 247                                 wfDebug( __METHOD__ . " The deduced filename $fileName is not safe\n" );
 248                                 return false;
 249                         }
 250                         $localPath =  $this->getZonePath( 'thumb' ) . "/" . $this->getHashPath( $name ) . $name;
 251                         $localFilename = $localPath . "/" . $fileName;
 252                         $localUrl =  $this->getZoneUrl( 'thumb' ) . "/" . $this->getHashPath( $name ) . urlencode( $name ) . "/" . urlencode( $fileName );
 253
 254                         if( file_exists( $localFilename ) && isset( $metadata['timestamp'] ) ) {
 255                                 wfDebug( __METHOD__ . " Thumbnail was already downloaded before\n" );
 256                                 $modified = filemtime( $localFilename );
 257                                 $remoteModified = strtotime( $metadata['timestamp'] );
 258                                 $current = time();
 259                                 $diff = abs( $modified - $current );
 260                                 if( $remoteModified < $modified && $diff < $this->fileCacheExpiry ) {
 261                                         /* Use our current and already downloaded thumbnail */
 262                                         $wgMemc->set( $key, $localUrl, $this->apiThumbCacheExpiry );
 263                                         return $localUrl;
 264                                 }
 265                                 /* There is a new Commons file, or existing thumbnail older than a month */
 266                         }
 267                         $thumb = Http::get( $foreignUrl );
 268                         if( !$thumb ) {
 269                                 wfDebug( __METHOD__ . " Could not download thumb\n" );
 270                                 return false;
 271                         }
 272                         if ( !is_dir($localPath) ) {
 273                                 if( !wfMkdirParents($localPath) ) {
 274                                         wfDebug(  __METHOD__ . " could not create directory $localPath for thumb\n" );
 275                                         return $foreignUrl;
 276                                 }
 277                         }
 278
 279                         # FIXME: Delete old thumbs that aren't being used. Maintenance script?
 280                         wfSuppressWarnings();
 281                         if( !file_put_contents( $localFilename, $thumb ) ) {
 282                                 wfRestoreWarnings();
 283                                 wfDebug( __METHOD__ . " could not write to thumb path\n" );
 284                                 return $foreignUrl;
 285                         }
 286                         wfRestoreWarnings();
 287                         $wgMemc->set( $key, $localUrl, $this->apiThumbCacheExpiry );
 288                         wfDebug( __METHOD__ . " got local thumb $localUrl, saving to cache \n" );
 289                         return $localUrl;
 290                 }
 291         }
 292
 293         /**
 294          * @see FileRepo::getZoneUrl()
 295          */
 296         function getZoneUrl( $zone ) {
 297                 switch ( $zone ) {
 298                         case 'public':
 299                                 return $this->url;
 300                         case 'thumb':
 301                                 return $this->thumbUrl;
 302                         default:
 303                                 return parent::getZoneUrl( $zone );
 304                 }
 305         }
 306
 307         /**
 308          * Get the local directory corresponding to one of the three basic zones
 309          */
 310         function getZonePath( $zone ) {
 311                 switch ( $zone ) {
 312                         case 'public':
 313                                 return $this->directory;
 314                         case 'thumb':
 315                                 return $this->thumbDir;
 316                         default:
 317                                 return false;
 318                 }
 319         }
 320
 321         /**
 322          * Are we locally caching the thumbnails?
 323          * @return bool
 324          */
 325         public function canCacheThumbs() {
 326                 return ( $this->apiThumbCacheExpiry > 0 );
 327         }
 328 }