includes/media/DjVuHandler.php

   1 <?php
   2 /**
   3  * Handler for DjVu images.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Media
  22  */
  23 use MediaWiki\MediaWikiServices;
  24
  25 /**
  26  * Handler for DjVu images
  27  *
  28  * @ingroup Media
  29  */
  30 class DjVuHandler extends ImageHandler {
  31         const EXPENSIVE_SIZE_LIMIT = 10485760; // 10MiB
  32
  33         /**
  34          * @return bool
  35          */
  36         public function isEnabled() {
  37                 global $wgDjvuRenderer, $wgDjvuDump, $wgDjvuToXML;
  38                 if ( !$wgDjvuRenderer || ( !$wgDjvuDump && !$wgDjvuToXML ) ) {
  39                         wfDebug( "DjVu is disabled, please set \$wgDjvuRenderer and \$wgDjvuDump\n" );
  40
  41                         return false;
  42                 } else {
  43                         return true;
  44                 }
  45         }
  46
  47         /**
  48          * @param File $file
  49          * @return bool
  50          */
  51         public function mustRender( $file ) {
  52                 return true;
  53         }
  54
  55         /**
  56          * True if creating thumbnails from the file is large or otherwise resource-intensive.
  57          * @param File $file
  58          * @return bool
  59          */
  60         public function isExpensiveToThumbnail( $file ) {
  61                 return $file->getSize() > static::EXPENSIVE_SIZE_LIMIT;
  62         }
  63
  64         /**
  65          * @param File $file
  66          * @return bool
  67          */
  68         public function isMultiPage( $file ) {
  69                 return true;
  70         }
  71
  72         /**
  73          * @return array
  74          */
  75         public function getParamMap() {
  76                 return [
  77                         'img_width' => 'width',
  78                         'img_page' => 'page',
  79                 ];
  80         }
  81
  82         /**
  83          * @param string $name
  84          * @param mixed $value
  85          * @return bool
  86          */
  87         public function validateParam( $name, $value ) {
  88                 if ( $name === 'page' && trim( $value ) !== (string)intval( $value ) ) {
  89                         // Extra junk on the end of page, probably actually a caption
  90                         // e.g. [[File:Foo.djvu|thumb|Page 3 of the document shows foo]]
  91                         return false;
  92                 }
  93                 if ( in_array( $name, [ 'width', 'height', 'page' ] ) ) {
  94                         if ( $value <= 0 ) {
  95                                 return false;
  96                         } else {
  97                                 return true;
  98                         }
  99                 } else {
 100                         return false;
 101                 }
 102         }
 103
 104         /**
 105          * @param array $params
 106          * @return bool|string
 107          */
 108         public function makeParamString( $params ) {
 109                 $page = $params['page'] ?? 1;
 110                 if ( !isset( $params['width'] ) ) {
 111                         return false;
 112                 }
 113
 114                 return "page{$page}-{$params['width']}px";
 115         }
 116
 117         /**
 118          * @param string $str
 119          * @return array|bool
 120          */
 121         public function parseParamString( $str ) {
 122                 $m = false;
 123                 if ( preg_match( '/^page(\d+)-(\d+)px$/', $str, $m ) ) {
 124                         return [ 'width' => $m[2], 'page' => $m[1] ];
 125                 } else {
 126                         return false;
 127                 }
 128         }
 129
 130         /**
 131          * @param array $params
 132          * @return array
 133          */
 134         protected function getScriptParams( $params ) {
 135                 return [
 136                         'width' => $params['width'],
 137                         'page' => $params['page'],
 138                 ];
 139         }
 140
 141         /**
 142          * @param File $image
 143          * @param string $dstPath
 144          * @param string $dstUrl
 145          * @param array $params
 146          * @param int $flags
 147          * @return MediaTransformError|ThumbnailImage|TransformParameterError
 148          */
 149         function doTransform( $image, $dstPath, $dstUrl, $params, $flags = 0 ) {
 150                 global $wgDjvuRenderer, $wgDjvuPostProcessor;
 151
 152                 if ( !$this->normaliseParams( $image, $params ) ) {
 153                         return new TransformParameterError( $params );
 154                 }
 155                 $width = $params['width'];
 156                 $height = $params['height'];
 157                 $page = $params['page'];
 158
 159                 if ( $flags & self::TRANSFORM_LATER ) {
 160                         $params = [
 161                                 'width' => $width,
 162                                 'height' => $height,
 163                                 'page' => $page
 164                         ];
 165
 166                         return new ThumbnailImage( $image, $dstUrl, $dstPath, $params );
 167                 }
 168
 169                 if ( !wfMkdirParents( dirname( $dstPath ), null, __METHOD__ ) ) {
 170                         return new MediaTransformError(
 171                                 'thumbnail_error',
 172                                 $width,
 173                                 $height,
 174                                 wfMessage( 'thumbnail_dest_directory' )
 175                         );
 176                 }
 177
 178                 // Get local copy source for shell scripts
 179                 // Thumbnail extraction is very inefficient for large files.
 180                 // Provide a way to pool count limit the number of downloaders.
 181                 if ( $image->getSize() >= 1e7 ) { // 10MB
 182                         $work = new PoolCounterWorkViaCallback( 'GetLocalFileCopy', sha1( $image->getName() ),
 183                                 [
 184                                         'doWork' => function () use ( $image ) {
 185                                                 return $image->getLocalRefPath();
 186                                         }
 187                                 ]
 188                         );
 189                         $srcPath = $work->execute();
 190                 } else {
 191                         $srcPath = $image->getLocalRefPath();
 192                 }
 193
 194                 if ( $srcPath === false ) { // Failed to get local copy
 195                         wfDebugLog( 'thumbnail',
 196                                 sprintf( 'Thumbnail failed on %s: could not get local copy of "%s"',
 197                                         wfHostname(), $image->getName() ) );
 198
 199                         return new MediaTransformError( 'thumbnail_error',
 200                                 $params['width'], $params['height'],
 201                                 wfMessage( 'filemissing' )
 202                         );
 203                 }
 204
 205                 # Use a subshell (brackets) to aggregate stderr from both pipeline commands
 206                 # before redirecting it to the overall stdout. This works in both Linux and Windows XP.
 207                 $cmd = '(' . wfEscapeShellArg(
 208                         $wgDjvuRenderer,
 209                         "-format=ppm",
 210                         "-page={$page}",
 211                         "-size={$params['physicalWidth']}x{$params['physicalHeight']}",
 212                         $srcPath );
 213                 if ( $wgDjvuPostProcessor ) {
 214                         $cmd .= " | {$wgDjvuPostProcessor}";
 215                 }
 216                 $cmd .= ' > ' . wfEscapeShellArg( $dstPath ) . ') 2>&1';
 217                 wfDebug( __METHOD__ . ": $cmd\n" );
 218                 $retval = '';
 219                 $err = wfShellExec( $cmd, $retval );
 220
 221                 $removed = $this->removeBadFile( $dstPath, $retval );
 222                 if ( $retval != 0 || $removed ) {
 223                         $this->logErrorForExternalProcess( $retval, $err, $cmd );
 224                         return new MediaTransformError( 'thumbnail_error', $width, $height, $err );
 225                 } else {
 226                         $params = [
 227                                 'width' => $width,
 228                                 'height' => $height,
 229                                 'page' => $page
 230                         ];
 231
 232                         return new ThumbnailImage( $image, $dstUrl, $dstPath, $params );
 233                 }
 234         }
 235
 236         /**
 237          * Cache an instance of DjVuImage in an Image object, return that instance
 238          *
 239          * @param File|FSFile $image
 240          * @param string $path
 241          * @return DjVuImage
 242          */
 243         function getDjVuImage( $image, $path ) {
 244                 if ( !$image ) {
 245                         $deja = new DjVuImage( $path );
 246                 } elseif ( !isset( $image->dejaImage ) ) {
 247                         $deja = $image->dejaImage = new DjVuImage( $path );
 248                 } else {
 249                         $deja = $image->dejaImage;
 250                 }
 251
 252                 return $deja;
 253         }
 254
 255         /**
 256          * Get metadata, unserializing it if necessary.
 257          *
 258          * @param File $file The DjVu file in question
 259          * @return string XML metadata as a string.
 260          * @throws MWException
 261          */
 262         private function getUnserializedMetadata( File $file ) {
 263                 $metadata = $file->getMetadata();
 264                 if ( substr( $metadata, 0, 3 ) === '<?xml' ) {
 265                         // Old style. Not serialized but instead just a raw string of XML.
 266                         return $metadata;
 267                 }
 268
 269                 Wikimedia\suppressWarnings();
 270                 $unser = unserialize( $metadata );
 271                 Wikimedia\restoreWarnings();
 272                 if ( is_array( $unser ) ) {
 273                         if ( isset( $unser['error'] ) ) {
 274                                 return false;
 275                         } elseif ( isset( $unser['xml'] ) ) {
 276                                 return $unser['xml'];
 277                         } else {
 278                                 // Should never ever reach here.
 279                                 throw new MWException( "Error unserializing DjVu metadata." );
 280                         }
 281                 }
 282
 283                 // unserialize failed. Guess it wasn't really serialized after all,
 284                 return $metadata;
 285         }
 286
 287         /**
 288          * Cache a document tree for the DjVu XML metadata
 289          * @param File $image
 290          * @param bool $gettext DOCUMENT (Default: false)
 291          * @return bool|SimpleXMLElement
 292          */
 293         public function getMetaTree( $image, $gettext = false ) {
 294                 if ( $gettext && isset( $image->djvuTextTree ) ) {
 295                         return $image->djvuTextTree;
 296                 }
 297                 if ( !$gettext && isset( $image->dejaMetaTree ) ) {
 298                         return $image->dejaMetaTree;
 299                 }
 300
 301                 $metadata = $this->getUnserializedMetadata( $image );
 302                 if ( !$this->isMetadataValid( $image, $metadata ) ) {
 303                         wfDebug( "DjVu XML metadata is invalid or missing, should have been fixed in upgradeRow\n" );
 304
 305                         return false;
 306                 }
 307
 308                 $trees = $this->extractTreesFromMetadata( $metadata );
 309                 $image->djvuTextTree = $trees['TextTree'];
 310                 $image->dejaMetaTree = $trees['MetaTree'];
 311
 312                 if ( $gettext ) {
 313                         return $image->djvuTextTree;
 314                 } else {
 315                         return $image->dejaMetaTree;
 316                 }
 317         }
 318
 319         /**
 320          * Extracts metadata and text trees from metadata XML in string form
 321          * @param string $metadata XML metadata as a string
 322          * @return array
 323          */
 324         protected function extractTreesFromMetadata( $metadata ) {
 325                 Wikimedia\suppressWarnings();
 326                 try {
 327                         // Set to false rather than null to avoid further attempts
 328                         $metaTree = false;
 329                         $textTree = false;
 330                         $tree = new SimpleXMLElement( $metadata, LIBXML_PARSEHUGE );
 331                         if ( $tree->getName() == 'mw-djvu' ) {
 332                                 /** @var SimpleXMLElement $b */
 333                                 foreach ( $tree->children() as $b ) {
 334                                         if ( $b->getName() == 'DjVuTxt' ) {
 335                                                 // @todo File::djvuTextTree and File::dejaMetaTree are declared
 336                                                 // dynamically. Add a public File::$data to facilitate this?
 337                                                 $textTree = $b;
 338                                         } elseif ( $b->getName() == 'DjVuXML' ) {
 339                                                 $metaTree = $b;
 340                                         }
 341                                 }
 342                         } else {
 343                                 $metaTree = $tree;
 344                         }
 345                 } catch ( Exception $e ) {
 346                         wfDebug( "Bogus multipage XML metadata\n" );
 347                 }
 348                 Wikimedia\restoreWarnings();
 349
 350                 return [ 'MetaTree' => $metaTree, 'TextTree' => $textTree ];
 351         }
 352
 353         function getImageSize( $image, $path ) {
 354                 return $this->getDjVuImage( $image, $path )->getImageSize();
 355         }
 356
 357         public function getThumbType( $ext, $mime, $params = null ) {
 358                 global $wgDjvuOutputExtension;
 359                 static $mime;
 360                 if ( !isset( $mime ) ) {
 361                         $magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
 362                         $mime = $magic->guessTypesForExtension( $wgDjvuOutputExtension );
 363                 }
 364
 365                 return [ $wgDjvuOutputExtension, $mime ];
 366         }
 367
 368         public function getMetadata( $image, $path ) {
 369                 wfDebug( "Getting DjVu metadata for $path\n" );
 370
 371                 $xml = $this->getDjVuImage( $image, $path )->retrieveMetaData();
 372                 if ( $xml === false ) {
 373                         // Special value so that we don't repetitively try and decode a broken file.
 374                         return serialize( [ 'error' => 'Error extracting metadata' ] );
 375                 } else {
 376                         return serialize( [ 'xml' => $xml ] );
 377                 }
 378         }
 379
 380         function getMetadataType( $image ) {
 381                 return 'djvuxml';
 382         }
 383
 384         public function isMetadataValid( $image, $metadata ) {
 385                 return !empty( $metadata ) && $metadata != serialize( [] );
 386         }
 387
 388         public function pageCount( File $image ) {
 389                 $info = $this->getDimensionInfo( $image );
 390
 391                 return $info ? $info['pageCount'] : false;
 392         }
 393
 394         public function getPageDimensions( File $image, $page ) {
 395                 $index = $page - 1; // MW starts pages at 1
 396
 397                 $info = $this->getDimensionInfo( $image );
 398                 if ( $info && isset( $info['dimensionsByPage'][$index] ) ) {
 399                         return $info['dimensionsByPage'][$index];
 400                 }
 401
 402                 return false;
 403         }
 404
 405         protected function getDimensionInfo( File $file ) {
 406                 $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
 407                 return $cache->getWithSetCallback(
 408                         $cache->makeKey( 'file-djvu', 'dimensions', $file->getSha1() ),
 409                         $cache::TTL_INDEFINITE,
 410                         function () use ( $file ) {
 411                                 $tree = $this->getMetaTree( $file );
 412                                 return $this->getDimensionInfoFromMetaTree( $tree );
 413                         },
 414                         [ 'pcTTL' => $cache::TTL_INDEFINITE ]
 415                 );
 416         }
 417
 418         /**
 419          * Given an XML metadata tree, returns dimension information about the document
 420          * @param bool|SimpleXMLElement $metatree The file's XML metadata tree
 421          * @return bool|array
 422          */
 423         protected function getDimensionInfoFromMetaTree( $metatree ) {
 424                 if ( !$metatree ) {
 425                         return false;
 426                 }
 427
 428                 $dimsByPage = [];
 429                 $count = count( $metatree->xpath( '//OBJECT' ) );
 430                 for ( $i = 0; $i < $count; $i++ ) {
 431                         $o = $metatree->BODY[0]->OBJECT[$i];
 432                         if ( $o ) {
 433                                 $dimsByPage[$i] = [
 434                                         'width' => (int)$o['width'],
 435                                         'height' => (int)$o['height'],
 436                                 ];
 437                         } else {
 438                                 $dimsByPage[$i] = false;
 439                         }
 440                 }
 441
 442                 return [ 'pageCount' => $count, 'dimensionsByPage' => $dimsByPage ];
 443         }
 444
 445         /**
 446          * @param File $image
 447          * @param int $page Page number to get information for
 448          * @return bool|string Page text or false when no text found.
 449          */
 450         function getPageText( File $image, $page ) {
 451                 $tree = $this->getMetaTree( $image, true );
 452                 if ( !$tree ) {
 453                         return false;
 454                 }
 455
 456                 $o = $tree->BODY[0]->PAGE[$page - 1];
 457                 if ( $o ) {
 458                         $txt = $o['value'];
 459
 460                         return $txt;
 461                 } else {
 462                         return false;
 463                 }
 464         }
 465 }