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