includes/media/DjVu.php

   1 <?php
   2 /**
   3  * Handler for DjVu images
   4  *
   5  * @file
   6  * @ingroup Media
   7  */
   8
   9 /**
  10  * Handler for DjVu images
  11  *
  12  * @ingroup Media
  13  */
  14 class DjVuHandler extends ImageHandler {
  15
  16         /**
  17          * @return bool
  18          */
  19         function isEnabled() {
  20                 global $wgDjvuRenderer, $wgDjvuDump, $wgDjvuToXML;
  21                 if ( !$wgDjvuRenderer || ( !$wgDjvuDump && !$wgDjvuToXML ) ) {
  22                         wfDebug( "DjVu is disabled, please set \$wgDjvuRenderer and \$wgDjvuDump\n" );
  23                         return false;
  24                 } else {
  25                         return true;
  26                 }
  27         }
  28
  29         /**
  30          * @param $file
  31          * @return bool
  32          */
  33         function mustRender( $file ) {
  34                 return true;
  35         }
  36
  37         /**
  38          * @param $file
  39          * @return bool
  40          */
  41         function isMultiPage( $file ) {
  42                 return true;
  43         }
  44
  45         /**
  46          * @return array
  47          */
  48         function getParamMap() {
  49                 return array(
  50                         'img_width' => 'width',
  51                         'img_page' => 'page',
  52                 );
  53         }
  54
  55         /**
  56          * @param $name
  57          * @param $value
  58          * @return bool
  59          */
  60         function validateParam( $name, $value ) {
  61                 if ( in_array( $name, array( 'width', 'height', 'page' ) ) ) {
  62                         if ( $value <= 0 ) {
  63                                 return false;
  64                         } else {
  65                                 return true;
  66                         }
  67                 } else {
  68                         return false;
  69                 }
  70         }
  71
  72         /**
  73          * @param $params
  74          * @return bool|string
  75          */
  76         function makeParamString( $params ) {
  77                 $page = isset( $params['page'] ) ? $params['page'] : 1;
  78                 if ( !isset( $params['width'] ) ) {
  79                         return false;
  80                 }
  81                 return "page{$page}-{$params['width']}px";
  82         }
  83
  84         /**
  85          * @param $str
  86          * @return array|bool
  87          */
  88         function parseParamString( $str ) {
  89                 $m = false;
  90                 if ( preg_match( '/^page(\d+)-(\d+)px$/', $str, $m ) ) {
  91                         return array( 'width' => $m[2], 'page' => $m[1] );
  92                 } else {
  93                         return false;
  94                 }
  95         }
  96
  97         /**
  98          * @param $params
  99          * @return array
 100          */
 101         function getScriptParams( $params ) {
 102                 return array(
 103                         'width' => $params['width'],
 104                         'page' => $params['page'],
 105                 );
 106         }
 107
 108         /**
 109          * @param $image File
 110          * @param  $dstPath
 111          * @param  $dstUrl
 112          * @param  $params
 113          * @param int $flags
 114          * @return MediaTransformError|ThumbnailImage|TransformParameterError
 115          */
 116         function doTransform( $image, $dstPath, $dstUrl, $params, $flags = 0 ) {
 117                 global $wgDjvuRenderer, $wgDjvuPostProcessor;
 118
 119                 // Fetch XML and check it, to give a more informative error message than the one which
 120                 // normaliseParams will inevitably give.
 121                 $xml = $image->getMetadata();
 122                 if ( !$xml ) {
 123                         $width = isset( $params['width'] ) ? $params['width'] : 0;
 124                         $height = isset( $params['height'] ) ? $params['height'] : 0;
 125                         return new MediaTransformError( 'thumbnail_error', $width, $height,
 126                                 wfMsg( 'djvu_no_xml' ) );
 127                 }
 128
 129                 if ( !$this->normaliseParams( $image, $params ) ) {
 130                         return new TransformParameterError( $params );
 131                 }
 132                 $width = $params['width'];
 133                 $height = $params['height'];
 134                 $srcPath = $image->getPath();
 135                 $page = $params['page'];
 136                 if ( $page > $this->pageCount( $image ) ) {
 137                         return new MediaTransformError( 'thumbnail_error', $width, $height, wfMsg( 'djvu_page_error' ) );
 138                 }
 139
 140                 if ( $flags & self::TRANSFORM_LATER ) {
 141                         return new ThumbnailImage( $image, $dstUrl, $width, $height, $dstPath, $page );
 142                 }
 143
 144                 if ( !wfMkdirParents( dirname( $dstPath ), null, __METHOD__ ) ) {
 145                         return new MediaTransformError( 'thumbnail_error', $width, $height, wfMsg( 'thumbnail_dest_directory' ) );
 146                 }
 147
 148                 # Use a subshell (brackets) to aggregate stderr from both pipeline commands
 149                 # before redirecting it to the overall stdout. This works in both Linux and Windows XP.
 150                 $cmd = '(' . wfEscapeShellArg( $wgDjvuRenderer ) . " -format=ppm -page={$page}" .
 151                         " -size={$params['physicalWidth']}x{$params['physicalHeight']} " .
 152                         wfEscapeShellArg( $srcPath );
 153                 if ( $wgDjvuPostProcessor ) {
 154                         $cmd .= " | {$wgDjvuPostProcessor}";
 155                 }
 156                 $cmd .= ' > ' . wfEscapeShellArg($dstPath) . ') 2>&1';
 157                 wfProfileIn( 'ddjvu' );
 158                 wfDebug( __METHOD__.": $cmd\n" );
 159                 $retval = '';
 160                 $err = wfShellExec( $cmd, $retval );
 161                 wfProfileOut( 'ddjvu' );
 162
 163                 $removed = $this->removeBadFile( $dstPath, $retval );
 164                 if ( $retval != 0 || $removed ) {
 165                         wfDebugLog( 'thumbnail',
 166                                 sprintf( 'thumbnail failed on %s: error %d "%s" from "%s"',
 167                                         wfHostname(), $retval, trim($err), $cmd ) );
 168                         return new MediaTransformError( 'thumbnail_error', $width, $height, $err );
 169                 } else {
 170                         return new ThumbnailImage( $image, $dstUrl, $width, $height, $dstPath, $page );
 171                 }
 172         }
 173
 174         /**
 175          * Cache an instance of DjVuImage in an Image object, return that instance
 176          *
 177          * @return DjVuImage
 178          */
 179         function getDjVuImage( $image, $path ) {
 180                 if ( !$image ) {
 181                         $deja = new DjVuImage( $path );
 182                 } elseif ( !isset( $image->dejaImage ) ) {
 183                         $deja = $image->dejaImage = new DjVuImage( $path );
 184                 } else {
 185                         $deja = $image->dejaImage;
 186                 }
 187                 return $deja;
 188         }
 189
 190         /**
 191          * Cache a document tree for the DjVu XML metadata
 192          * @param $image File
 193          */
 194         function getMetaTree( $image , $gettext = false ) {
 195                 if ( isset( $image->dejaMetaTree ) ) {
 196                         return $image->dejaMetaTree;
 197                 }
 198
 199                 $metadata = $image->getMetadata();
 200                 if ( !$this->isMetadataValid( $image, $metadata ) ) {
 201                         wfDebug( "DjVu XML metadata is invalid or missing, should have been fixed in upgradeRow\n" );
 202                         return false;
 203                 }
 204                 wfProfileIn( __METHOD__ );
 205
 206                 wfSuppressWarnings();
 207                 try {
 208                         // Set to false rather than null to avoid further attempts
 209                         $image->dejaMetaTree = false;
 210                         $image->djvuTextTree = false;
 211                         $tree = new SimpleXMLElement( $metadata );
 212                         if( $tree->getName() == 'mw-djvu' ) {
 213                                 foreach($tree->children() as $b){
 214                                         if( $b->getName() == 'DjVuTxt' ) {
 215                                                 $image->djvuTextTree = $b;
 216                                         }
 217                                         elseif ( $b->getName() == 'DjVuXML' ) {
 218                                                 $image->dejaMetaTree = $b;
 219                                         }
 220                                 }
 221                         } else {
 222                                 $image->dejaMetaTree = $tree;
 223                         }
 224                 } catch( Exception $e ) {
 225                         wfDebug( "Bogus multipage XML metadata on '{$image->getName()}'\n" );
 226                 }
 227                 wfRestoreWarnings();
 228                 wfProfileOut( __METHOD__ );
 229                 if( $gettext ) {
 230                         return $image->djvuTextTree;
 231                 } else {
 232                         return $image->dejaMetaTree;
 233                 }
 234         }
 235
 236         function getImageSize( $image, $path ) {
 237                 return $this->getDjVuImage( $image, $path )->getImageSize();
 238         }
 239
 240         function getThumbType( $ext, $mime, $params = null ) {
 241                 global $wgDjvuOutputExtension;
 242                 static $mime;
 243                 if ( !isset( $mime ) ) {
 244                         $magic = MimeMagic::singleton();
 245                         $mime = $magic->guessTypesForExtension( $wgDjvuOutputExtension );
 246                 }
 247                 return array( $wgDjvuOutputExtension, $mime );
 248         }
 249
 250         function getMetadata( $image, $path ) {
 251                 wfDebug( "Getting DjVu metadata for $path\n" );
 252                 return $this->getDjVuImage( $image, $path )->retrieveMetaData();
 253         }
 254
 255         function getMetadataType( $image ) {
 256                 return 'djvuxml';
 257         }
 258
 259         function isMetadataValid( $image, $metadata ) {
 260                 return !empty( $metadata ) && $metadata != serialize(array());
 261         }
 262
 263         function pageCount( $image ) {
 264                 $tree = $this->getMetaTree( $image );
 265                 if ( !$tree ) {
 266                         return false;
 267                 }
 268                 return count( $tree->xpath( '//OBJECT' ) );
 269         }
 270
 271         function getPageDimensions( $image, $page ) {
 272                 $tree = $this->getMetaTree( $image );
 273                 if ( !$tree ) {
 274                         return false;
 275                 }
 276
 277                 $o = $tree->BODY[0]->OBJECT[$page-1];
 278                 if ( $o ) {
 279                         return array(
 280                                 'width' => intval( $o['width'] ),
 281                                 'height' => intval( $o['height'] )
 282                         );
 283                 } else {
 284                         return false;
 285                 }
 286         }
 287
 288         function getPageText( $image, $page ){
 289                 $tree = $this->getMetaTree( $image, true );
 290                 if ( !$tree ) {
 291                         return false;
 292                 }
 293
 294                 $o = $tree->BODY[0]->PAGE[$page-1];
 295                 if ( $o ) {
 296                         $txt = $o['value'];
 297                         return $txt;
 298                 } else {
 299                         return false;
 300                 }
 301
 302         }
 303
 304 }