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  * Very hacky and inefficient
  12  * do not use except for testing :D
  13  *
  14  * Example config:
  15  *
  16  * $wgForeignFileRepos[] = array(
  17  *   'class'                  => 'ForeignAPIRepo',
  18  *   'name'                   => 'shared',
  19  *   'apibase'                => 'http://en.wikipedia.org/w/api.php',
  20  *   'fetchDescription'       => true, // Optional
  21  *   'descriptionCacheExpiry' => 3600,
  22  * );
  23  *
  24  * @ingroup FileRepo
  25  */
  26 class ForeignAPIRepo extends FileRepo {
  27         var $fileFactory = array( 'ForeignAPIFile', 'newFromTitle' );
  28         /* Check back with Commons after a day */
  29         var $apiThumbCacheExpiry = 86400;
  30         /* Redownload thumbnail files after a month */
  31         var $fileCacheExpiry = 2629743;
  32
  33         protected $mQueryCache = array();
  34         protected $mFileExists = array();
  35
  36         function __construct( $info ) {
  37                 parent::__construct( $info );
  38
  39                 // http://commons.wikimedia.org/w/api.php
  40                 $this->mApiBase = isset( $info['apibase'] ) ? $info['apibase'] : null;
  41
  42                 if( isset( $info['apiThumbCacheExpiry'] ) ) {
  43                         $this->apiThumbCacheExpiry = $info['apiThumbCacheExpiry'];
  44                 }
  45                 if( !$this->scriptDirUrl ) {
  46                         // hack for description fetches
  47                         $this->scriptDirUrl = dirname( $this->mApiBase );
  48                 }
  49                 // If we can cache thumbs we can guess sane defaults for these
  50                 if( $this->canCacheThumbs() && !$this->url ) {
  51                         global $wgLocalFileRepo;
  52                         $this->url = $wgLocalFileRepo['url'];
  53                 }
  54                 if( $this->canCacheThumbs() && !$this->thumbUrl ) {
  55                         $this->thumbUrl = $this->url . '/thumb';
  56                 }
  57         }
  58
  59         /**
  60          * Per docs in FileRepo, this needs to return false if we don't support versioned
  61          * files. Well, we don't.
  62          */
  63         function newFile( $title, $time = false ) {
  64                 if ( $time ) {
  65                         return false;
  66                 }
  67                 return parent::newFile( $title, $time );
  68         }
  69
  70 /**
  71  * No-ops
  72  */
  73         function storeBatch( $triplets, $flags = 0 ) {
  74                 return false;
  75         }
  76         function storeTemp( $originalName, $srcPath ) {
  77                 return false;
  78         }
  79         function append( $srcPath, $toAppendPath, $flags = 0 ){
  80                 return false;
  81         }
  82         function publishBatch( $triplets, $flags = 0 ) {
  83                 return false;
  84         }
  85         function deleteBatch( $sourceDestPairs ) {
  86                 return false;
  87         }
  88
  89
  90         function fileExistsBatch( $files, $flags = 0 ) {
  91                 $results = array();
  92                 foreach ( $files as $k => $f ) {
  93                         if ( isset( $this->mFileExists[$k] ) ) {
  94                                 $results[$k] = true;
  95                                 unset( $files[$k] );
  96                         } elseif( self::isVirtualUrl( $f ) ) {
  97                                 # TODO! FIXME! We need to be able to handle virtual
  98                                 # URLs better, at least when we know they refer to the
  99                                 # same repo.
 100                                 $results[$k] = false;
 101                                 unset( $files[$k] );
 102                         }
 103                 }
 104
 105                 $results = $this->fetchImageQuery( array( 'titles' => implode( $files, '|' ),
 106                                                                                         'prop' => 'imageinfo' ) );
 107                 if( isset( $data['query']['pages'] ) ) {
 108                         $i = 0;
 109                         foreach( $files as $key => $file ) {
 110                                 $results[$key] = $this->mFileExists[$key] = !isset( $data['query']['pages'][$i]['missing'] );
 111                                 $i++;
 112                         }
 113                 }
 114         }
 115         function getFileProps( $virtualUrl ) {
 116                 return false;
 117         }
 118
 119         function fetchImageQuery( $query ) {
 120                 global $wgMemc;
 121
 122                 $query = array_merge( $query,
 123                         array(
 124                                 'format' => 'json',
 125                                 'action' => 'query',
 126                                 'redirects' => 'true'
 127                         ) );
 128                 if ( $this->mApiBase ) {
 129                         $url = wfAppendQuery( $this->mApiBase, $query );
 130                 } else {
 131                         $url = $this->makeUrl( $query, 'api' );
 132                 }
 133
 134                 if( !isset( $this->mQueryCache[$url] ) ) {
 135                         $key = $this->getLocalCacheKey( 'ForeignAPIRepo', 'Metadata', md5( $url ) );
 136                         $data = $wgMemc->get( $key );
 137                         if( !$data ) {
 138                                 $data = Http::get( $url );
 139                                 if ( !$data ) {
 140                                         return null;
 141                                 }
 142                                 $wgMemc->set( $key, $data, 3600 );
 143                         }
 144
 145                         if( count( $this->mQueryCache ) > 100 ) {
 146                                 // Keep the cache from growing infinitely
 147                                 $this->mQueryCache = array();
 148                         }
 149                         $this->mQueryCache[$url] = $data;
 150                 }
 151                 return FormatJson::decode( $this->mQueryCache[$url], true );
 152         }
 153
 154         function getImageInfo( $data ) {
 155                 if( $data && isset( $data['query']['pages'] ) ) {
 156                         foreach( $data['query']['pages'] as $info ) {
 157                                 if( isset( $info['imageinfo'][0] ) ) {
 158                                         return $info['imageinfo'][0];
 159                                 }
 160                         }
 161                 }
 162                 return false;
 163         }
 164
 165         function findBySha1( $hash ) {
 166                 $results = $this->fetchImageQuery( array(
 167                                                                                 'aisha1base36' => $hash,
 168                                                                                 'aiprop'       => ForeignAPIFile::getProps(),
 169                                                                                 'list'         => 'allimages', ) );
 170                 $ret = array();
 171                 if ( isset( $results['query']['allimages'] ) ) {
 172                         foreach ( $results['query']['allimages'] as $img ) {
 173                                 // 1.14 was broken, doesn't return name attribute
 174                                 if( !isset( $img['name'] ) ) {
 175                                         continue;
 176                                 }
 177                                 $ret[] = new ForeignAPIFile( Title::makeTitle( NS_FILE, $img['name'] ), $this, $img );
 178                         }
 179                 }
 180                 return $ret;
 181         }
 182
 183         function getThumbUrl( $name, $width=-1, $height=-1, &$result ) {
 184                 $data = $this->fetchImageQuery( array(
 185                         'titles' => 'File:' . $name,
 186                         'iiprop' => 'url|timestamp',
 187                         'iiurlwidth' => $width,
 188                         'iiurlheight' => $height,
 189                         'prop' => 'imageinfo' ) );
 190                 $info = $this->getImageInfo( $data );
 191
 192                 if( $data && $info && isset( $info['thumburl'] ) ) {
 193                         wfDebug( __METHOD__ . " got remote thumb " . $info['thumburl'] . "\n" );
 194                         $result = $info;
 195                         return $info['thumburl'];
 196                 } else {
 197                         return false;
 198                 }
 199         }
 200
 201         /*
 202          * Return the imageurl from cache if possible
 203          *
 204          * If the url has been requested today, get it from cache
 205          * Otherwise retrieve remote thumb url, check for local file.
 206          *
 207          * @param $name String is a dbkey form of a title
 208          * @param $width
 209          * @param $height
 210          */
 211         function getThumbUrlFromCache( $name, $width, $height ) {
 212                 global $wgMemc, $wgUploadPath, $wgServer, $wgUploadDirectory;
 213
 214                 if ( !$this->canCacheThumbs() ) {
 215                         return $this->getThumbUrl( $name, $width, $height );
 216                 }
 217
 218                 $key = $this->getLocalCacheKey( 'ForeignAPIRepo', 'ThumbUrl', $name, $width );
 219
 220                 $thumbUrl = $wgMemc->get($key);
 221                 if ( $thumbUrl ) {
 222                         wfDebug("Got thumb from local cache. $thumbUrl \n");
 223                         return $thumbUrl;
 224                 }
 225                 else {
 226                         $metadata = null;
 227                         $foreignUrl = $this->getThumbUrl( $name, $width, $height, $metadata );
 228                         // We need the same filename as the remote one :)
 229                         $fileName = rawurldecode( pathinfo( $foreignUrl, PATHINFO_BASENAME ) );
 230                         $path = 'thumb/' . $this->getHashPath( $name ) . $name . "/";
 231                         $localFilename = $wgUploadDirectory . '/' . $path . $fileName;
 232                         $localUrl =  $wgServer . $wgUploadPath . '/' . $path . $fileName;
 233
 234                         if( !$foreignUrl ) {
 235                                 wfDebug( __METHOD__ . " Could not find thumburl\n" );
 236                                 return false;
 237                         }
 238                         if( file_exists( $localFilename ) && isset( $metadata['timestamp'] ) ) {
 239                                 wfDebug( __METHOD__ . " Thumbnail was already downloaded before\n" );
 240                                 $modified = filemtime( $localFilename );
 241                                 $remoteModified = strtotime( $metadata['timestamp'] );
 242                                 $diff = abs( $modified - $remoteModified );
 243                                 if( $remoteModified < $modified && $diff < $this->fileCacheExpiry ) {
 244                                         /* Use our current already downloaded thumbnail */
 245                                         $wgMemc->set( $key, $localUrl, $this->apiThumbCacheExpiry );
 246                                         return $localUrl;
 247                                 }
 248                                 /* There is a new Commons file, or existing thumbnail older than a month */
 249
 250                         }
 251                         $thumb = Http::get( $foreignUrl );
 252                         if( !$thumb ) {
 253                                 wfDebug( __METHOD__ . " Could not download thumb\n" );
 254                                 return false;
 255                         }
 256                         if ( !is_dir($wgUploadDirectory . '/' . $path) ) {
 257                                 if( !wfMkdirParents($wgUploadDirectory . '/' . $path) ) {
 258                                         wfDebug(  __METHOD__ . " could not create directory for thumb\n" );
 259                                         return $foreignUrl;
 260                                 }
 261                         }
 262
 263                         # FIXME: Delete old thumbs that aren't being used. Maintenance script?
 264                         wfSuppressWarnings();
 265                         if( !file_put_contents($localFilename, $thumb ) ) {
 266                                 wfRestoreWarnings();
 267                                 wfDebug( __METHOD__ . " could not write to thumb path\n" );
 268                                 return $foreignUrl;
 269                         }
 270                         wfRestoreWarnings();
 271                         $wgMemc->set( $key, $localUrl, $this->apiThumbCacheExpiry );
 272                         wfDebug( __METHOD__ . " got local thumb $localUrl, saving to cache \n" );
 273                         return $localUrl;
 274                 }
 275         }
 276
 277         /**
 278          * @see FileRepo::getZoneUrl()
 279          */
 280         function getZoneUrl( $zone ) {
 281                 switch ( $zone ) {
 282                         case 'public':
 283                                 return $this->url;
 284                         case 'thumb':
 285                                 return $this->thumbUrl;
 286                         default:
 287                                 return parent::getZoneUrl( $zone );
 288                 }
 289         }
 290
 291         /**
 292          * Are we locally caching the thumbnails?
 293          * @return bool
 294          */
 295         public function canCacheThumbs() {
 296                 return ( $this->apiThumbCacheExpiry > 0 );
 297         }
 298 }