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         /**
  72          * @param $image File
  73          * @param  $dstPath
  74          * @param  $dstUrl
  75          * @param  $params
  76          * @param int $flags
  77          * @return MediaTransformError|ThumbnailImage|TransformParameterError
  78          */
  79         function doTransform( $image, $dstPath, $dstUrl, $params, $flags = 0 ) {
  80                 global $wgDjvuRenderer, $wgDjvuPostProcessor;
  81
  82                 // Fetch XML and check it, to give a more informative error message than the one which
  83                 // normaliseParams will inevitably give.
  84                 $xml = $image->getMetadata();
  85                 if ( !$xml ) {
  86                         return new MediaTransformError( 'thumbnail_error', @$params['width'], @$params['height'],
  87                                 wfMsg( 'djvu_no_xml' ) );
  88                 }
  89
  90                 if ( !$this->normaliseParams( $image, $params ) ) {
  91                         return new TransformParameterError( $params );
  92                 }
  93                 $width = $params['width'];
  94                 $height = $params['height'];
  95                 $srcPath = $image->getPath();
  96                 $page = $params['page'];
  97                 if ( $page > $this->pageCount( $image ) ) {
  98                         return new MediaTransformError( 'thumbnail_error', $width, $height, wfMsg( 'djvu_page_error' ) );
  99                 }
 100
 101                 if ( $flags & self::TRANSFORM_LATER ) {
 102                         return new ThumbnailImage( $image, $dstUrl, $width, $height, $dstPath, $page );
 103                 }
 104
 105                 if ( !wfMkdirParents( dirname( $dstPath ) ) ) {
 106                         return new MediaTransformError( 'thumbnail_error', $width, $height, wfMsg( 'thumbnail_dest_directory' ) );
 107                 }
 108
 109                 # Use a subshell (brackets) to aggregate stderr from both pipeline commands
 110                 # before redirecting it to the overall stdout. This works in both Linux and Windows XP.
 111                 $cmd = '(' . wfEscapeShellArg( $wgDjvuRenderer ) . " -format=ppm -page={$page} -size={$width}x{$height} " .
 112                         wfEscapeShellArg( $srcPath );
 113                 if ( $wgDjvuPostProcessor ) {
 114                         $cmd .= " | {$wgDjvuPostProcessor}";
 115                 }
 116                 $cmd .= ' > ' . wfEscapeShellArg($dstPath) . ') 2>&1';
 117                 wfProfileIn( 'ddjvu' );
 118                 wfDebug( __METHOD__.": $cmd\n" );
 119                 $retval = '';
 120                 $err = wfShellExec( $cmd, $retval );
 121                 wfProfileOut( 'ddjvu' );
 122
 123                 $removed = $this->removeBadFile( $dstPath, $retval );
 124                 if ( $retval != 0 || $removed ) {
 125                         wfDebugLog( 'thumbnail',
 126                                 sprintf( 'thumbnail failed on %s: error %d "%s" from "%s"',
 127                                         wfHostname(), $retval, trim($err), $cmd ) );
 128                         return new MediaTransformError( 'thumbnail_error', $width, $height, $err );
 129                 } else {
 130                         return new ThumbnailImage( $image, $dstUrl, $width, $height, $dstPath, $page );
 131                 }
 132         }
 133
 134         /**
 135          * Cache an instance of DjVuImage in an Image object, return that instance
 136          */
 137         function getDjVuImage( $image, $path ) {
 138                 if ( !$image ) {
 139                         $deja = new DjVuImage( $path );
 140                 } elseif ( !isset( $image->dejaImage ) ) {
 141                         $deja = $image->dejaImage = new DjVuImage( $path );
 142                 } else {
 143                         $deja = $image->dejaImage;
 144                 }
 145                 return $deja;
 146         }
 147
 148         /**
 149          * Cache a document tree for the DjVu XML metadata
 150          * @param $image File
 151          */
 152         function getMetaTree( $image , $gettext = false ) {
 153                 if ( isset( $image->dejaMetaTree ) ) {
 154                         return $image->dejaMetaTree;
 155                 }
 156
 157                 $metadata = $image->getMetadata();
 158                 if ( !$this->isMetadataValid( $image, $metadata ) ) {
 159                         wfDebug( "DjVu XML metadata is invalid or missing, should have been fixed in upgradeRow\n" );
 160                         return false;
 161                 }
 162                 wfProfileIn( __METHOD__ );
 163
 164                 wfSuppressWarnings();
 165                 try {
 166                         // Set to false rather than null to avoid further attempts
 167                         $image->dejaMetaTree = false;
 168                         $image->djvuTextTree = false;
 169                         $tree = new SimpleXMLElement( $metadata );
 170                         if( $tree->getName() == 'mw-djvu' ) {
 171                                 foreach($tree->children() as $b){
 172                                         if( $b->getName() == 'DjVuTxt' ) {
 173                                                 $image->djvuTextTree = $b;
 174                                         }
 175                                         else if ( $b->getName() == 'DjVuXML' ) {
 176                                                 $image->dejaMetaTree = $b;
 177                                         }
 178                                 }
 179                         } else {
 180                                 $image->dejaMetaTree = $tree;
 181                         }
 182                 } catch( Exception $e ) {
 183                         wfDebug( "Bogus multipage XML metadata on '$image->name'\n" );
 184                 }
 185                 wfRestoreWarnings();
 186                 wfProfileOut( __METHOD__ );
 187                 if( $gettext ) {
 188                         return $image->djvuTextTree;
 189                 } else {
 190                         return $image->dejaMetaTree;
 191                 }
 192         }
 193
 194         function getImageSize( $image, $path ) {
 195                 return $this->getDjVuImage( $image, $path )->getImageSize();
 196         }
 197
 198         function getThumbType( $ext, $mime, $params = null ) {
 199                 global $wgDjvuOutputExtension;
 200                 static $mime;
 201                 if ( !isset( $mime ) ) {
 202                         $magic = MimeMagic::singleton();
 203                         $mime = $magic->guessTypesForExtension( $wgDjvuOutputExtension );
 204                 }
 205                 return array( $wgDjvuOutputExtension, $mime );
 206         }
 207
 208         function getMetadata( $image, $path ) {
 209                 wfDebug( "Getting DjVu metadata for $path\n" );
 210                 return $this->getDjVuImage( $image, $path )->retrieveMetaData();
 211         }
 212
 213         function getMetadataType( $image ) {
 214                 return 'djvuxml';
 215         }
 216
 217         function isMetadataValid( $image, $metadata ) {
 218                 return !empty( $metadata ) && $metadata != serialize(array());
 219         }
 220
 221         function pageCount( $image ) {
 222                 $tree = $this->getMetaTree( $image );
 223                 if ( !$tree ) {
 224                         return false;
 225                 }
 226                 return count( $tree->xpath( '//OBJECT' ) );
 227         }
 228
 229         function getPageDimensions( $image, $page ) {
 230                 $tree = $this->getMetaTree( $image );
 231                 if ( !$tree ) {
 232                         return false;
 233                 }
 234
 235                 $o = $tree->BODY[0]->OBJECT[$page-1];
 236                 if ( $o ) {
 237                         return array(
 238                                 'width' => intval( $o['width'] ),
 239                                 'height' => intval( $o['height'] )
 240                         );
 241                 } else {
 242                         return false;
 243                 }
 244         }
 245
 246         function getPageText( $image, $page ){
 247                 $tree = $this->getMetaTree( $image, true );
 248                 if ( !$tree ) {
 249                         return false;
 250                 }
 251
 252                 $o = $tree->BODY[0]->PAGE[$page-1];
 253                 if ( $o ) {
 254                         $txt = $o['value'];
 255                         return $txt;
 256                 } else {
 257                         return false;
 258                 }
 259
 260         }
 261
 262 }