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                         return new MediaTransformError( 'thumbnail_error', @$params['width'], @$params['height'],
 124                                 wfMsg( 'djvu_no_xml' ) );
 125                 }
 126
 127                 if ( !$this->normaliseParams( $image, $params ) ) {
 128                         return new TransformParameterError( $params );
 129                 }
 130                 $width = $params['width'];
 131                 $height = $params['height'];
 132                 $srcPath = $image->getPath();
 133                 $page = $params['page'];
 134                 if ( $page > $this->pageCount( $image ) ) {
 135                         return new MediaTransformError( 'thumbnail_error', $width, $height, wfMsg( 'djvu_page_error' ) );
 136                 }
 137
 138                 if ( $flags & self::TRANSFORM_LATER ) {
 139                         return new ThumbnailImage( $image, $dstUrl, $width, $height, $dstPath, $page );
 140                 }
 141
 142                 if ( !wfMkdirParents( dirname( $dstPath ) ) ) {
 143                         return new MediaTransformError( 'thumbnail_error', $width, $height, wfMsg( 'thumbnail_dest_directory' ) );
 144                 }
 145
 146                 # Use a subshell (brackets) to aggregate stderr from both pipeline commands
 147                 # before redirecting it to the overall stdout. This works in both Linux and Windows XP.
 148                 $cmd = '(' . wfEscapeShellArg( $wgDjvuRenderer ) . " -format=ppm -page={$page} -size={$width}x{$height} " .
 149                         wfEscapeShellArg( $srcPath );
 150                 if ( $wgDjvuPostProcessor ) {
 151                         $cmd .= " | {$wgDjvuPostProcessor}";
 152                 }
 153                 $cmd .= ' > ' . wfEscapeShellArg($dstPath) . ') 2>&1';
 154                 wfProfileIn( 'ddjvu' );
 155                 wfDebug( __METHOD__.": $cmd\n" );
 156                 $retval = '';
 157                 $err = wfShellExec( $cmd, $retval );
 158                 wfProfileOut( 'ddjvu' );
 159
 160                 $removed = $this->removeBadFile( $dstPath, $retval );
 161                 if ( $retval != 0 || $removed ) {
 162                         wfDebugLog( 'thumbnail',
 163                                 sprintf( 'thumbnail failed on %s: error %d "%s" from "%s"',
 164                                         wfHostname(), $retval, trim($err), $cmd ) );
 165                         return new MediaTransformError( 'thumbnail_error', $width, $height, $err );
 166                 } else {
 167                         return new ThumbnailImage( $image, $dstUrl, $width, $height, $dstPath, $page );
 168                 }
 169         }
 170
 171         /**
 172          * Cache an instance of DjVuImage in an Image object, return that instance
 173          *
 174          * @return DjVuImage
 175          */
 176         function getDjVuImage( $image, $path ) {
 177                 if ( !$image ) {
 178                         $deja = new DjVuImage( $path );
 179                 } elseif ( !isset( $image->dejaImage ) ) {
 180                         $deja = $image->dejaImage = new DjVuImage( $path );
 181                 } else {
 182                         $deja = $image->dejaImage;
 183                 }
 184                 return $deja;
 185         }
 186
 187         /**
 188          * Cache a document tree for the DjVu XML metadata
 189          * @param $image File
 190          */
 191         function getMetaTree( $image , $gettext = false ) {
 192                 if ( isset( $image->dejaMetaTree ) ) {
 193                         return $image->dejaMetaTree;
 194                 }
 195
 196                 $metadata = $image->getMetadata();
 197                 if ( !$this->isMetadataValid( $image, $metadata ) ) {
 198                         wfDebug( "DjVu XML metadata is invalid or missing, should have been fixed in upgradeRow\n" );
 199                         return false;
 200                 }
 201                 wfProfileIn( __METHOD__ );
 202
 203                 wfSuppressWarnings();
 204                 try {
 205                         // Set to false rather than null to avoid further attempts
 206                         $image->dejaMetaTree = false;
 207                         $image->djvuTextTree = false;
 208                         $tree = new SimpleXMLElement( $metadata );
 209                         if( $tree->getName() == 'mw-djvu' ) {
 210                                 foreach($tree->children() as $b){
 211                                         if( $b->getName() == 'DjVuTxt' ) {
 212                                                 $image->djvuTextTree = $b;
 213                                         }
 214                                         elseif ( $b->getName() == 'DjVuXML' ) {
 215                                                 $image->dejaMetaTree = $b;
 216                                         }
 217                                 }
 218                         } else {
 219                                 $image->dejaMetaTree = $tree;
 220                         }
 221                 } catch( Exception $e ) {
 222                         wfDebug( "Bogus multipage XML metadata on '$image->name'\n" );
 223                 }
 224                 wfRestoreWarnings();
 225                 wfProfileOut( __METHOD__ );
 226                 if( $gettext ) {
 227                         return $image->djvuTextTree;
 228                 } else {
 229                         return $image->dejaMetaTree;
 230                 }
 231         }
 232
 233         function getImageSize( $image, $path ) {
 234                 return $this->getDjVuImage( $image, $path )->getImageSize();
 235         }
 236
 237         function getThumbType( $ext, $mime, $params = null ) {
 238                 global $wgDjvuOutputExtension;
 239                 static $mime;
 240                 if ( !isset( $mime ) ) {
 241                         $magic = MimeMagic::singleton();
 242                         $mime = $magic->guessTypesForExtension( $wgDjvuOutputExtension );
 243                 }
 244                 return array( $wgDjvuOutputExtension, $mime );
 245         }
 246
 247         function getMetadata( $image, $path ) {
 248                 wfDebug( "Getting DjVu metadata for $path\n" );
 249                 return $this->getDjVuImage( $image, $path )->retrieveMetaData();
 250         }
 251
 252         function getMetadataType( $image ) {
 253                 return 'djvuxml';
 254         }
 255
 256         function isMetadataValid( $image, $metadata ) {
 257                 return !empty( $metadata ) && $metadata != serialize(array());
 258         }
 259
 260         function pageCount( $image ) {
 261                 $tree = $this->getMetaTree( $image );
 262                 if ( !$tree ) {
 263                         return false;
 264                 }
 265                 return count( $tree->xpath( '//OBJECT' ) );
 266         }
 267
 268         function getPageDimensions( $image, $page ) {
 269                 $tree = $this->getMetaTree( $image );
 270                 if ( !$tree ) {
 271                         return false;
 272                 }
 273
 274                 $o = $tree->BODY[0]->OBJECT[$page-1];
 275                 if ( $o ) {
 276                         return array(
 277                                 'width' => intval( $o['width'] ),
 278                                 'height' => intval( $o['height'] )
 279                         );
 280                 } else {
 281                         return false;
 282                 }
 283         }
 284
 285         function getPageText( $image, $page ){
 286                 $tree = $this->getMetaTree( $image, true );
 287                 if ( !$tree ) {
 288                         return false;
 289                 }
 290
 291                 $o = $tree->BODY[0]->PAGE[$page-1];
 292                 if ( $o ) {
 293                         $txt = $o['value'];
 294                         return $txt;
 295                 } else {
 296                         return false;
 297                 }
 298
 299         }
 300
 301 }