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 use MediaWiki\Shell\Shell;
  25
  26 /**
  27  * Handler for DjVu images
  28  *
  29  * @ingroup Media
  30  */
  31 class DjVuHandler extends ImageHandler {
  32         const EXPENSIVE_SIZE_LIMIT = 10485760; // 10MiB
  33
  34         /**
  35          * @return bool
  36          */
  37         public function isEnabled() {
  38                 global $wgDjvuRenderer, $wgDjvuDump, $wgDjvuToXML;
  39                 if ( !$wgDjvuRenderer || ( !$wgDjvuDump && !$wgDjvuToXML ) ) {
  40                         wfDebug( "DjVu is disabled, please set \$wgDjvuRenderer and \$wgDjvuDump\n" );
  41
  42                         return false;
  43                 } else {
  44                         return true;
  45                 }
  46         }
  47
  48         /**
  49          * @param File $file
  50          * @return bool
  51          */
  52         public function mustRender( $file ) {
  53                 return true;
  54         }
  55
  56         /**
  57          * True if creating thumbnails from the file is large or otherwise resource-intensive.
  58          * @param File $file
  59          * @return bool
  60          */
  61         public function isExpensiveToThumbnail( $file ) {
  62                 return $file->getSize() > static::EXPENSIVE_SIZE_LIMIT;
  63         }
  64
  65         /**
  66          * @param File $file
  67          * @return bool
  68          */
  69         public function isMultiPage( $file ) {
  70                 return true;
  71         }
  72
  73         /**
  74          * @return array
  75          */
  76         public function getParamMap() {
  77                 return [
  78                         'img_width' => 'width',
  79                         'img_page' => 'page',
  80                 ];
  81         }
  82
  83         /**
  84          * @param string $name
  85          * @param mixed $value
  86          * @return bool
  87          */
  88         public function validateParam( $name, $value ) {
  89                 if ( $name === 'page' && trim( $value ) !== (string)intval( $value ) ) {
  90                         // Extra junk on the end of page, probably actually a caption
  91                         // e.g. [[File:Foo.djvu|thumb|Page 3 of the document shows foo]]
  92                         return false;
  93                 }
  94                 if ( in_array( $name, [ 'width', 'height', 'page' ] ) ) {
  95                         if ( $value <= 0 ) {
  96                                 return false;
  97                         } else {
  98                                 return true;
  99                         }
 100                 } else {
 101                         return false;
 102                 }
 103         }
 104
 105         /**
 106          * @param array $params
 107          * @return bool|string
 108          */
 109         public function makeParamString( $params ) {
 110                 $page = $params['page'] ?? 1;
 111                 if ( !isset( $params['width'] ) ) {
 112                         return false;
 113                 }
 114
 115                 return "page{$page}-{$params['width']}px";
 116         }
 117
 118         /**
 119          * @param string $str
 120          * @return array|bool
 121          */
 122         public function parseParamString( $str ) {
 123                 $m = false;
 124                 if ( preg_match( '/^page(\d+)-(\d+)px$/', $str, $m ) ) {
 125                         return [ 'width' => $m[2], 'page' => $m[1] ];
 126                 } else {
 127                         return false;
 128                 }
 129         }
 130
 131         /**
 132          * @param array $params
 133          * @return array
 134          */
 135         protected function getScriptParams( $params ) {
 136                 return [
 137                         'width' => $params['width'],
 138                         'page' => $params['page'],
 139                 ];
 140         }
 141
 142         /**
 143          * @param File $image
 144          * @param string $dstPath
 145          * @param string $dstUrl
 146          * @param array $params
 147          * @param int $flags
 148          * @return MediaTransformError|ThumbnailImage|TransformParameterError
 149          */
 150         function doTransform( $image, $dstPath, $dstUrl, $params, $flags = 0 ) {
 151                 global $wgDjvuRenderer, $wgDjvuPostProcessor;
 152
 153                 if ( !$this->normaliseParams( $image, $params ) ) {
 154                         return new TransformParameterError( $params );
 155                 }
 156                 $width = $params['width'];
 157                 $height = $params['height'];
 158                 $page = $params['page'];
 159
 160                 if ( $flags & self::TRANSFORM_LATER ) {
 161                         $params = [
 162                                 'width' => $width,
 163                                 'height' => $height,
 164                                 'page' => $page
 165                         ];
 166
 167                         return new ThumbnailImage( $image, $dstUrl, $dstPath, $params );
 168                 }
 169
 170                 if ( !wfMkdirParents( dirname( $dstPath ), null, __METHOD__ ) ) {
 171                         return new MediaTransformError(
 172                                 'thumbnail_error',
 173                                 $width,
 174                                 $height,
 175                                 wfMessage( 'thumbnail_dest_directory' )
 176                         );
 177                 }
 178
 179                 // Get local copy source for shell scripts
 180                 // Thumbnail extraction is very inefficient for large files.
 181                 // Provide a way to pool count limit the number of downloaders.
 182                 if ( $image->getSize() >= 1e7 ) { // 10MB
 183                         $work = new PoolCounterWorkViaCallback( 'GetLocalFileCopy', sha1( $image->getName() ),
 184                                 [
 185                                         'doWork' => function () use ( $image ) {
 186                                                 return $image->getLocalRefPath();
 187                                         }
 188                                 ]
 189                         );
 190                         $srcPath = $work->execute();
 191                 } else {
 192                         $srcPath = $image->getLocalRefPath();
 193                 }
 194
 195                 if ( $srcPath === false ) { // Failed to get local copy
 196                         wfDebugLog( 'thumbnail',
 197                                 sprintf( 'Thumbnail failed on %s: could not get local copy of "%s"',
 198                                         wfHostname(), $image->getName() ) );
 199
 200                         return new MediaTransformError( 'thumbnail_error',
 201                                 $params['width'], $params['height'],
 202                                 wfMessage( 'filemissing' )
 203                         );
 204                 }
 205
 206                 # Use a subshell (brackets) to aggregate stderr from both pipeline commands
 207                 # before redirecting it to the overall stdout. This works in both Linux and Windows XP.
 208                 $cmd = '(' . Shell::escape(
 209                         $wgDjvuRenderer,
 210                         "-format=ppm",
 211                         "-page={$page}",
 212                         "-size={$params['physicalWidth']}x{$params['physicalHeight']}",
 213                         $srcPath );
 214                 if ( $wgDjvuPostProcessor ) {
 215                         $cmd .= " | {$wgDjvuPostProcessor}";
 216                 }
 217                 $cmd .= ' > ' . Shell::escape( $dstPath ) . ') 2>&1';
 218                 wfDebug( __METHOD__ . ": $cmd\n" );
 219                 $retval = '';
 220                 $err = wfShellExec( $cmd, $retval );
 221
 222                 $removed = $this->removeBadFile( $dstPath, $retval );
 223                 if ( $retval != 0 || $removed ) {
 224                         $this->logErrorForExternalProcess( $retval, $err, $cmd );
 225                         return new MediaTransformError( 'thumbnail_error', $width, $height, $err );
 226                 } else {
 227                         $params = [
 228                                 'width' => $width,
 229                                 'height' => $height,
 230                                 'page' => $page
 231                         ];
 232
 233                         return new ThumbnailImage( $image, $dstUrl, $dstPath, $params );
 234                 }
 235         }
 236
 237         /**
 238          * Cache an instance of DjVuImage in an Image object, return that instance
 239          *
 240          * @param File|FSFile $image
 241          * @param string $path
 242          * @return DjVuImage
 243          */
 244         function getDjVuImage( $image, $path ) {
 245                 if ( !$image ) {
 246                         $deja = new DjVuImage( $path );
 247                 } elseif ( !isset( $image->dejaImage ) ) {
 248                         $deja = $image->dejaImage = new DjVuImage( $path );
 249                 } else {
 250                         $deja = $image->dejaImage;
 251                 }
 252
 253                 return $deja;
 254         }
 255
 256         /**
 257          * Get metadata, unserializing it if necessary.
 258          *
 259          * @param File $file The DjVu file in question
 260          * @return string XML metadata as a string.
 261          * @throws MWException
 262          */
 263         private function getUnserializedMetadata( File $file ) {
 264                 $metadata = $file->getMetadata();
 265                 if ( substr( $metadata, 0, 3 ) === '<?xml' ) {
 266                         // Old style. Not serialized but instead just a raw string of XML.
 267                         return $metadata;
 268                 }
 269
 270                 Wikimedia\suppressWarnings();
 271                 $unser = unserialize( $metadata );
 272                 Wikimedia\restoreWarnings();
 273                 if ( is_array( $unser ) ) {
 274                         if ( isset( $unser['error'] ) ) {
 275                                 return false;
 276                         } elseif ( isset( $unser['xml'] ) ) {
 277                                 return $unser['xml'];
 278                         } else {
 279                                 // Should never ever reach here.
 280                                 throw new MWException( "Error unserializing DjVu metadata." );
 281                         }
 282                 }
 283
 284                 // unserialize failed. Guess it wasn't really serialized after all,
 285                 return $metadata;
 286         }
 287
 288         /**
 289          * Cache a document tree for the DjVu XML metadata
 290          * @param File $image
 291          * @param bool $gettext DOCUMENT (Default: false)
 292          * @return bool|SimpleXMLElement
 293          */
 294         public function getMetaTree( $image, $gettext = false ) {
 295                 if ( $gettext && isset( $image->djvuTextTree ) ) {
 296                         return $image->djvuTextTree;
 297                 }
 298                 if ( !$gettext && isset( $image->dejaMetaTree ) ) {
 299                         return $image->dejaMetaTree;
 300                 }
 301
 302                 $metadata = $this->getUnserializedMetadata( $image );
 303                 if ( !$this->isMetadataValid( $image, $metadata ) ) {
 304                         wfDebug( "DjVu XML metadata is invalid or missing, should have been fixed in upgradeRow\n" );
 305
 306                         return false;
 307                 }
 308
 309                 $trees = $this->extractTreesFromMetadata( $metadata );
 310                 $image->djvuTextTree = $trees['TextTree'];
 311                 $image->dejaMetaTree = $trees['MetaTree'];
 312
 313                 if ( $gettext ) {
 314                         return $image->djvuTextTree;
 315                 } else {
 316                         return $image->dejaMetaTree;
 317                 }
 318         }
 319
 320         /**
 321          * Extracts metadata and text trees from metadata XML in string form
 322          * @param string $metadata XML metadata as a string
 323          * @return array
 324          */
 325         protected function extractTreesFromMetadata( $metadata ) {
 326                 Wikimedia\suppressWarnings();
 327                 try {
 328                         // Set to false rather than null to avoid further attempts
 329                         $metaTree = false;
 330                         $textTree = false;
 331                         $tree = new SimpleXMLElement( $metadata, LIBXML_PARSEHUGE );
 332                         if ( $tree->getName() == 'mw-djvu' ) {
 333                                 /** @var SimpleXMLElement $b */
 334                                 foreach ( $tree->children() as $b ) {
 335                                         if ( $b->getName() == 'DjVuTxt' ) {
 336                                                 // @todo File::djvuTextTree and File::dejaMetaTree are declared
 337                                                 // dynamically. Add a public File::$data to facilitate this?
 338                                                 $textTree = $b;
 339                                         } elseif ( $b->getName() == 'DjVuXML' ) {
 340                                                 $metaTree = $b;
 341                                         }
 342                                 }
 343                         } else {
 344                                 $metaTree = $tree;
 345                         }
 346                 } catch ( Exception $e ) {
 347                         wfDebug( "Bogus multipage XML metadata\n" );
 348                 }
 349                 Wikimedia\restoreWarnings();
 350
 351                 return [ 'MetaTree' => $metaTree, 'TextTree' => $textTree ];
 352         }
 353
 354         function getImageSize( $image, $path ) {
 355                 return $this->getDjVuImage( $image, $path )->getImageSize();
 356         }
 357
 358         public function getThumbType( $ext, $mime, $params = null ) {
 359                 global $wgDjvuOutputExtension;
 360                 static $mime;
 361                 if ( !isset( $mime ) ) {
 362                         $magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
 363                         $mime = $magic->guessTypesForExtension( $wgDjvuOutputExtension );
 364                 }
 365
 366                 return [ $wgDjvuOutputExtension, $mime ];
 367         }
 368
 369         public function getMetadata( $image, $path ) {
 370                 wfDebug( "Getting DjVu metadata for $path\n" );
 371
 372                 $xml = $this->getDjVuImage( $image, $path )->retrieveMetaData();
 373                 if ( $xml === false ) {
 374                         // Special value so that we don't repetitively try and decode a broken file.
 375                         return serialize( [ 'error' => 'Error extracting metadata' ] );
 376                 } else {
 377                         return serialize( [ 'xml' => $xml ] );
 378                 }
 379         }
 380
 381         function getMetadataType( $image ) {
 382                 return 'djvuxml';
 383         }
 384
 385         public function isMetadataValid( $image, $metadata ) {
 386                 return !empty( $metadata ) && $metadata != serialize( [] );
 387         }
 388
 389         public function pageCount( File $image ) {
 390                 $info = $this->getDimensionInfo( $image );
 391
 392                 return $info ? $info['pageCount'] : false;
 393         }
 394
 395         public function getPageDimensions( File $image, $page ) {
 396                 $index = $page - 1; // MW starts pages at 1
 397
 398                 $info = $this->getDimensionInfo( $image );
 399                 if ( $info && isset( $info['dimensionsByPage'][$index] ) ) {
 400                         return $info['dimensionsByPage'][$index];
 401                 }
 402
 403                 return false;
 404         }
 405
 406         protected function getDimensionInfo( File $file ) {
 407                 $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
 408                 return $cache->getWithSetCallback(
 409                         $cache->makeKey( 'file-djvu', 'dimensions', $file->getSha1() ),
 410                         $cache::TTL_INDEFINITE,
 411                         function () use ( $file ) {
 412                                 $tree = $this->getMetaTree( $file );
 413                                 return $this->getDimensionInfoFromMetaTree( $tree );
 414                         },
 415                         [ 'pcTTL' => $cache::TTL_INDEFINITE ]
 416                 );
 417         }
 418
 419         /**
 420          * Given an XML metadata tree, returns dimension information about the document
 421          * @param bool|SimpleXMLElement $metatree The file's XML metadata tree
 422          * @return bool|array
 423          */
 424         protected function getDimensionInfoFromMetaTree( $metatree ) {
 425                 if ( !$metatree ) {
 426                         return false;
 427                 }
 428
 429                 $dimsByPage = [];
 430                 $count = count( $metatree->xpath( '//OBJECT' ) );
 431                 for ( $i = 0; $i < $count; $i++ ) {
 432                         $o = $metatree->BODY[0]->OBJECT[$i];
 433                         if ( $o ) {
 434                                 $dimsByPage[$i] = [
 435                                         'width' => (int)$o['width'],
 436                                         'height' => (int)$o['height'],
 437                                 ];
 438                         } else {
 439                                 $dimsByPage[$i] = false;
 440                         }
 441                 }
 442
 443                 return [ 'pageCount' => $count, 'dimensionsByPage' => $dimsByPage ];
 444         }
 445
 446         /**
 447          * @param File $image
 448          * @param int $page Page number to get information for
 449          * @return bool|string Page text or false when no text found.
 450          */
 451         function getPageText( File $image, $page ) {
 452                 $tree = $this->getMetaTree( $image, true );
 453                 if ( !$tree ) {
 454                         return false;
 455                 }
 456
 457                 $o = $tree->BODY[0]->PAGE[$page - 1];
 458                 if ( $o ) {
 459                         $txt = $o['value'];
 460
 461                         return $txt;
 462                 } else {
 463                         return false;
 464                 }
 465         }
 466 }