includes/DjVuImage.php

   1 <?php
   2 /**
   3  * DjVu image handler
   4  *
   5  * Copyright © 2006 Brion Vibber <brion@pobox.com>
   6  * http://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  */
  25
  26 /**
  27  * Support for detecting/validating DjVu image files and getting
  28  * some basic file metadata (resolution etc)
  29  *
  30  * File format docs are available in source package for DjVuLibre:
  31  * http://djvulibre.djvuzone.org/
  32  *
  33  * @ingroup Media
  34  */
  35 class DjVuImage {
  36         function __construct( $filename ) {
  37                 $this->mFilename = $filename;
  38         }
  39
  40         /**
  41          * Check if the given file is indeed a valid DjVu image file
  42          * @return bool
  43          */
  44         public function isValid() {
  45                 $info = $this->getInfo();
  46                 return $info !== false;
  47         }
  48
  49
  50         /**
  51          * Return data in the style of getimagesize()
  52          * @return array or false on failure
  53          */
  54         public function getImageSize() {
  55                 $data = $this->getInfo();
  56
  57                 if( $data !== false ) {
  58                         $width  = $data['width'];
  59                         $height = $data['height'];
  60
  61                         return array( $width, $height, 'DjVu',
  62                                 "width=\"$width\" height=\"$height\"" );
  63                 }
  64                 return false;
  65         }
  66
  67         // ---------
  68
  69         /**
  70          * For debugging; dump the IFF chunk structure
  71          */
  72         function dump() {
  73                 $file = fopen( $this->mFilename, 'rb' );
  74                 $header = fread( $file, 12 );
  75                 // FIXME: Would be good to replace this extract() call with something that explicitly initializes local variables.
  76                 extract( unpack( 'a4magic/a4chunk/NchunkLength', $header ) );
  77                 echo "$chunk $chunkLength\n";
  78                 $this->dumpForm( $file, $chunkLength, 1 );
  79                 fclose( $file );
  80         }
  81
  82         private function dumpForm( $file, $length, $indent ) {
  83                 $start = ftell( $file );
  84                 $secondary = fread( $file, 4 );
  85                 echo str_repeat( ' ', $indent * 4 ) . "($secondary)\n";
  86                 while( ftell( $file ) - $start < $length ) {
  87                         $chunkHeader = fread( $file, 8 );
  88                         if( $chunkHeader == '' ) {
  89                                 break;
  90                         }
  91                         // FIXME: Would be good to replace this extract() call with something that explicitly initializes local variables.
  92                         extract( unpack( 'a4chunk/NchunkLength', $chunkHeader ) );
  93                         echo str_repeat( ' ', $indent * 4 ) . "$chunk $chunkLength\n";
  94
  95                         if( $chunk == 'FORM' ) {
  96                                 $this->dumpForm( $file, $chunkLength, $indent + 1 );
  97                         } else {
  98                                 fseek( $file, $chunkLength, SEEK_CUR );
  99                                 if( $chunkLength & 1 == 1 ) {
 100                                         // Padding byte between chunks
 101                                         fseek( $file, 1, SEEK_CUR );
 102                                 }
 103                         }
 104                 }
 105         }
 106
 107         function getInfo() {
 108                 wfSuppressWarnings();
 109                 $file = fopen( $this->mFilename, 'rb' );
 110                 wfRestoreWarnings();
 111                 if( $file === false ) {
 112                         wfDebug( __METHOD__ . ": missing or failed file read\n" );
 113                         return false;
 114                 }
 115
 116                 $header = fread( $file, 16 );
 117                 $info = false;
 118
 119                 if( strlen( $header ) < 16 ) {
 120                         wfDebug( __METHOD__ . ": too short file header\n" );
 121                 } else {
 122                         // FIXME: Would be good to replace this extract() call with something that explicitly initializes local variables.
 123                         extract( unpack( 'a4magic/a4form/NformLength/a4subtype', $header ) );
 124
 125                         if( $magic != 'AT&T' ) {
 126                                 wfDebug( __METHOD__ . ": not a DjVu file\n" );
 127                         } elseif( $subtype == 'DJVU' ) {
 128                                 // Single-page document
 129                                 $info = $this->getPageInfo( $file, $formLength );
 130                         } elseif( $subtype == 'DJVM' ) {
 131                                 // Multi-page document
 132                                 $info = $this->getMultiPageInfo( $file, $formLength );
 133                         } else  {
 134                                 wfDebug( __METHOD__ . ": unrecognized DJVU file type '$formType'\n" );
 135                         }
 136                 }
 137                 fclose( $file );
 138                 return $info;
 139         }
 140
 141         private function readChunk( $file ) {
 142                 $header = fread( $file, 8 );
 143                 if( strlen( $header ) < 8 ) {
 144                         return array( false, 0 );
 145                 } else {
 146                         // FIXME: Would be good to replace this extract() call with something that explicitly initializes local variables.
 147                         extract( unpack( 'a4chunk/Nlength', $header ) );
 148                         return array( $chunk, $length );
 149                 }
 150         }
 151
 152         private function skipChunk( $file, $chunkLength ) {
 153                 fseek( $file, $chunkLength, SEEK_CUR );
 154
 155                 if( $chunkLength & 0x01 == 1 && !feof( $file ) ) {
 156                         // padding byte
 157                         fseek( $file, 1, SEEK_CUR );
 158                 }
 159         }
 160
 161         private function getMultiPageInfo( $file, $formLength ) {
 162                 // For now, we'll just look for the first page in the file
 163                 // and report its information, hoping others are the same size.
 164                 $start = ftell( $file );
 165                 do {
 166                         list( $chunk, $length ) = $this->readChunk( $file );
 167                         if( !$chunk ) {
 168                                 break;
 169                         }
 170
 171                         if( $chunk == 'FORM' ) {
 172                                 $subtype = fread( $file, 4 );
 173                                 if( $subtype == 'DJVU' ) {
 174                                         wfDebug( __METHOD__ . ": found first subpage\n" );
 175                                         return $this->getPageInfo( $file, $length );
 176                                 }
 177                                 $this->skipChunk( $file, $length - 4 );
 178                         } else {
 179                                 wfDebug( __METHOD__ . ": skipping '$chunk' chunk\n" );
 180                                 $this->skipChunk( $file, $length );
 181                         }
 182                 } while( $length != 0 && !feof( $file ) && ftell( $file ) - $start < $formLength );
 183
 184                 wfDebug( __METHOD__ . ": multi-page DJVU file contained no pages\n" );
 185                 return false;
 186         }
 187
 188         private function getPageInfo( $file, $formLength ) {
 189                 list( $chunk, $length ) = $this->readChunk( $file );
 190                 if( $chunk != 'INFO' ) {
 191                         wfDebug( __METHOD__ . ": expected INFO chunk, got '$chunk'\n" );
 192                         return false;
 193                 }
 194
 195                 if( $length < 9 ) {
 196                         wfDebug( __METHOD__ . ": INFO should be 9 or 10 bytes, found $length\n" );
 197                         return false;
 198                 }
 199                 $data = fread( $file, $length );
 200                 if( strlen( $data ) < $length ) {
 201                         wfDebug( __METHOD__ . ": INFO chunk cut off\n" );
 202                         return false;
 203                 }
 204
 205                 // FIXME: Would be good to replace this extract() call with something that explicitly initializes local variables.
 206                 extract( unpack(
 207                         'nwidth/' .
 208                         'nheight/' .
 209                         'Cminor/' .
 210                         'Cmajor/' .
 211                         'vresolution/' .
 212                         'Cgamma', $data ) );
 213                 # Newer files have rotation info in byte 10, but we don't use it yet.
 214
 215                 return array(
 216                         'width' => $width,
 217                         'height' => $height,
 218                         'version' => "$major.$minor",
 219                         'resolution' => $resolution,
 220                         'gamma' => $gamma / 10.0 );
 221         }
 222
 223         /**
 224          * Return an XML string describing the DjVu image
 225          * @return string
 226          */
 227         function retrieveMetaData() {
 228                 global $wgDjvuToXML, $wgDjvuDump, $wgDjvuTxt;
 229                 wfProfileIn( __METHOD__ );
 230
 231                 if ( isset( $wgDjvuDump ) ) {
 232                         # djvudump is faster as of version 3.5
 233                         # http://sourceforge.net/tracker/index.php?func=detail&aid=1704049&group_id=32953&atid=406583
 234                         wfProfileIn( 'djvudump' );
 235                         $cmd = wfEscapeShellArg( $wgDjvuDump ) . ' ' . wfEscapeShellArg( $this->mFilename );
 236                         $dump = wfShellExec( $cmd );
 237                         $xml = $this->convertDumpToXML( $dump );
 238                         wfProfileOut( 'djvudump' );
 239                 } elseif ( isset( $wgDjvuToXML ) ) {
 240                         wfProfileIn( 'djvutoxml' );
 241                         $cmd = wfEscapeShellArg( $wgDjvuToXML ) . ' --without-anno --without-text ' .
 242                                 wfEscapeShellArg( $this->mFilename );
 243                         $xml = wfShellExec( $cmd );
 244                         wfProfileOut( 'djvutoxml' );
 245                 } else {
 246                         $xml = null;
 247                 }
 248                 # Text layer
 249                 if ( isset( $wgDjvuTxt ) ) {
 250                         wfProfileIn( 'djvutxt' );
 251                         $cmd = wfEscapeShellArg( $wgDjvuTxt ) . ' --detail=page ' . wfEscapeShellArg( $this->mFilename ) ;
 252                         wfDebug( __METHOD__.": $cmd\n" );
 253                         $retval = '';
 254                         $txt = wfShellExec( $cmd, $retval );
 255                         wfProfileOut( 'djvutxt' );
 256                         if( $retval == 0) {
 257                                 # Strip some control characters
 258                                 $txt = preg_replace( "/[\013\035\037]/", "", $txt );
 259                                 $reg = <<<EOR
 260                                         /\(page\s[\d-]*\s[\d-]*\s[\d-]*\s[\d-]*\s*"
 261                                         ((?>    # Text to match is composed of atoms of either:
 262                                           \\\\. # - any escaped character
 263                                           |     # - any character different from " and \
 264                                           [^"\\\\]+
 265                                         )*?)
 266                                         "\s*\)
 267                                         | # Or page can be empty ; in this case, djvutxt dumps ()
 268                                         \(\s*()\)/sx
 269 EOR;
 270                                 $txt = preg_replace_callback( $reg, array( $this, 'pageTextCallback' ), $txt );
 271                                 $txt = "<DjVuTxt>\n<HEAD></HEAD>\n<BODY>\n" . $txt . "</BODY>\n</DjVuTxt>\n";
 272                                 $xml = preg_replace( "/<DjVuXML>/", "<mw-djvu><DjVuXML>", $xml );
 273                                 $xml = $xml . $txt. '</mw-djvu>' ;
 274                         }
 275                 }
 276                 wfProfileOut( __METHOD__ );
 277                 return $xml;
 278         }
 279
 280         function pageTextCallback( $matches ) {
 281                 # Get rid of invalid UTF-8, strip control characters
 282                 return '<PAGE value="' . htmlspecialchars( UtfNormal::cleanUp( $matches[1] ) ) . '" />';
 283         }
 284
 285         /**
 286          * Hack to temporarily work around djvutoxml bug
 287          */
 288         function convertDumpToXML( $dump ) {
 289                 if ( strval( $dump ) == '' ) {
 290                         return false;
 291                 }
 292
 293                 $xml = <<<EOT
 294 <?xml version="1.0" ?>
 295 <!DOCTYPE DjVuXML PUBLIC "-//W3C//DTD DjVuXML 1.1//EN" "pubtext/DjVuXML-s.dtd">
 296 <DjVuXML>
 297 <HEAD></HEAD>
 298 <BODY>
 299 EOT;
 300
 301                 $dump = str_replace( "\r", '', $dump );
 302                 $line = strtok( $dump, "\n" );
 303                 $m = false;
 304                 $good = false;
 305                 if ( preg_match( '/^( *)FORM:DJVU/', $line, $m ) ) {
 306                         # Single-page
 307                         if ( $this->parseFormDjvu( $line, $xml ) ) {
 308                                 $good = true;
 309                         } else {
 310                                 return false;
 311                         }
 312                 } elseif ( preg_match( '/^( *)FORM:DJVM/', $line, $m ) ) {
 313                         # Multi-page
 314                         $parentLevel = strlen( $m[1] );
 315                         # Find DIRM
 316                         $line = strtok( "\n" );
 317                         while ( $line !== false ) {
 318                                 $childLevel = strspn( $line, ' ' );
 319                                 if ( $childLevel <= $parentLevel ) {
 320                                         # End of chunk
 321                                         break;
 322                                 }
 323
 324                                 if ( preg_match( '/^ *DIRM.*indirect/', $line ) ) {
 325                                         wfDebug( "Indirect multi-page DjVu document, bad for server!\n" );
 326                                         return false;
 327                                 }
 328                                 if ( preg_match( '/^ *FORM:DJVU/', $line ) ) {
 329                                         # Found page
 330                                         if ( $this->parseFormDjvu( $line, $xml ) ) {
 331                                                 $good = true;
 332                                         } else {
 333                                                 return false;
 334                                         }
 335                                 }
 336                                 $line = strtok( "\n" );
 337                         }
 338                 }
 339                 if ( !$good ) {
 340                         return false;
 341                 }
 342
 343                 $xml .= "</BODY>\n</DjVuXML>\n";
 344                 return $xml;
 345         }
 346
 347         function parseFormDjvu( $line, &$xml ) {
 348                 $parentLevel = strspn( $line, ' ' );
 349                 $line = strtok( "\n" );
 350
 351                 # Find INFO
 352                 while ( $line !== false ) {
 353                         $childLevel = strspn( $line, ' ' );
 354                         if ( $childLevel <= $parentLevel ) {
 355                                 # End of chunk
 356                                 break;
 357                         }
 358
 359                         if ( preg_match( '/^ *INFO *\[\d*\] *DjVu *(\d+)x(\d+), *\w*, *(\d+) *dpi, *gamma=([0-9.-]+)/', $line, $m ) ) {
 360                                 $xml .= Xml::tags( 'OBJECT',
 361                                         array(
 362                                                 #'data' => '',
 363                                                 #'type' => 'image/x.djvu',
 364                                                 'height' => $m[2],
 365                                                 'width' => $m[1],
 366                                                 #'usemap' => '',
 367                                         ),
 368                                         "\n" .
 369                                         Xml::element( 'PARAM', array( 'name' => 'DPI', 'value' => $m[3] ) ) . "\n" .
 370                                         Xml::element( 'PARAM', array( 'name' => 'GAMMA', 'value' => $m[4] ) ) . "\n"
 371                                 ) . "\n";
 372                                 return true;
 373                         }
 374                         $line = strtok( "\n" );
 375                 }
 376                 # Not found
 377                 return false;
 378         }
 379 }