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