includes/media/DjVu.php

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