includes/media/DjVuImage.php

   1 <?php
   2 /**
   3  * DjVu image handler.
   4  *
   5  * Copyright © 2006 Brion Vibber <brion@pobox.com>
   6  * https://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  * @ingroup Media
  25  */
  26
  27 /**
  28  * Support for detecting/validating DjVu image files and getting
  29  * some basic file metadata (resolution etc)
  30  *
  31  * File format docs are available in source package for DjVuLibre:
  32  * http://djvulibre.djvuzone.org/
  33  *
  34  * @ingroup Media
  35  */
  36 class DjVuImage {
  37         /**
  38          * @const DJVUTXT_MEMORY_LIMIT Memory limit for the DjVu description software
  39          */
  40         const DJVUTXT_MEMORY_LIMIT = 300000;
  41
  42         /**
  43          * Constructor
  44          *
  45          * @param string $filename The DjVu file name.
  46          */
  47         function __construct( $filename ) {
  48                 $this->mFilename = $filename;
  49         }
  50
  51         /**
  52          * Check if the given file is indeed a valid DjVu image file
  53          * @return bool
  54          */
  55         public function isValid() {
  56                 $info = $this->getInfo();
  57
  58                 return $info !== false;
  59         }
  60
  61         /**
  62          * Return data in the style of getimagesize()
  63          * @return array|bool Array or false on failure
  64          */
  65         public function getImageSize() {
  66                 $data = $this->getInfo();
  67
  68                 if ( $data !== false ) {
  69                         $width = $data['width'];
  70                         $height = $data['height'];
  71
  72                         return [ $width, $height, 'DjVu',
  73                                 "width=\"$width\" height=\"$height\"" ];
  74                 }
  75
  76                 return false;
  77         }
  78
  79         // ---------
  80
  81         /**
  82          * For debugging; dump the IFF chunk structure
  83          */
  84         function dump() {
  85                 $file = fopen( $this->mFilename, 'rb' );
  86                 $header = fread( $file, 12 );
  87                 // @todo FIXME: Would be good to replace this extract() call with
  88                 // something that explicitly initializes local variables.
  89                 extract( unpack( 'a4magic/a4chunk/NchunkLength', $header ) );
  90                 /** @var string $chunk
  91                  * @var string $chunkLength */
  92                 echo "$chunk $chunkLength\n";
  93                 $this->dumpForm( $file, $chunkLength, 1 );
  94                 fclose( $file );
  95         }
  96
  97         private function dumpForm( $file, $length, $indent ) {
  98                 $start = ftell( $file );
  99                 $secondary = fread( $file, 4 );
 100                 echo str_repeat( ' ', $indent * 4 ) . "($secondary)\n";
 101                 while ( ftell( $file ) - $start < $length ) {
 102                         $chunkHeader = fread( $file, 8 );
 103                         if ( $chunkHeader == '' ) {
 104                                 break;
 105                         }
 106                         // @todo FIXME: Would be good to replace this extract() call with
 107                         // something that explicitly initializes local variables.
 108                         extract( unpack( 'a4chunk/NchunkLength', $chunkHeader ) );
 109                         /** @var string $chunk
 110                          * @var string $chunkLength */
 111                         echo str_repeat( ' ', $indent * 4 ) . "$chunk $chunkLength\n";
 112
 113                         if ( $chunk == 'FORM' ) {
 114                                 $this->dumpForm( $file, $chunkLength, $indent + 1 );
 115                         } else {
 116                                 fseek( $file, $chunkLength, SEEK_CUR );
 117                                 if ( $chunkLength & 1 == 1 ) {
 118                                         // Padding byte between chunks
 119                                         fseek( $file, 1, SEEK_CUR );
 120                                 }
 121                         }
 122                 }
 123         }
 124
 125         function getInfo() {
 126                 MediaWiki\suppressWarnings();
 127                 $file = fopen( $this->mFilename, 'rb' );
 128                 MediaWiki\restoreWarnings();
 129                 if ( $file === false ) {
 130                         wfDebug( __METHOD__ . ": missing or failed file read\n" );
 131
 132                         return false;
 133                 }
 134
 135                 $header = fread( $file, 16 );
 136                 $info = false;
 137
 138                 if ( strlen( $header ) < 16 ) {
 139                         wfDebug( __METHOD__ . ": too short file header\n" );
 140                 } else {
 141                         // @todo FIXME: Would be good to replace this extract() call with
 142                         // something that explicitly initializes local variables.
 143                         extract( unpack( 'a4magic/a4form/NformLength/a4subtype', $header ) );
 144
 145                         /** @var string $magic
 146                          * @var string $subtype
 147                          * @var string $formLength
 148                          * @var string $formType */
 149                         if ( $magic != 'AT&T' ) {
 150                                 wfDebug( __METHOD__ . ": not a DjVu file\n" );
 151                         } elseif ( $subtype == 'DJVU' ) {
 152                                 // Single-page document
 153                                 $info = $this->getPageInfo( $file );
 154                         } elseif ( $subtype == 'DJVM' ) {
 155                                 // Multi-page document
 156                                 $info = $this->getMultiPageInfo( $file, $formLength );
 157                         } else {
 158                                 wfDebug( __METHOD__ . ": unrecognized DJVU file type '$formType'\n" );
 159                         }
 160                 }
 161                 fclose( $file );
 162
 163                 return $info;
 164         }
 165
 166         private function readChunk( $file ) {
 167                 $header = fread( $file, 8 );
 168                 if ( strlen( $header ) < 8 ) {
 169                         return [ false, 0 ];
 170                 } else {
 171                         // @todo FIXME: Would be good to replace this extract() call with
 172                         // something that explicitly initializes local variables.
 173                         extract( unpack( 'a4chunk/Nlength', $header ) );
 174
 175                         /** @var string $chunk
 176                          * @var string $length */
 177                         return [ $chunk, $length ];
 178                 }
 179         }
 180
 181         private function skipChunk( $file, $chunkLength ) {
 182                 fseek( $file, $chunkLength, SEEK_CUR );
 183
 184                 if ( $chunkLength & 0x01 == 1 && !feof( $file ) ) {
 185                         // padding byte
 186                         fseek( $file, 1, SEEK_CUR );
 187                 }
 188         }
 189
 190         private function getMultiPageInfo( $file, $formLength ) {
 191                 // For now, we'll just look for the first page in the file
 192                 // and report its information, hoping others are the same size.
 193                 $start = ftell( $file );
 194                 do {
 195                         list( $chunk, $length ) = $this->readChunk( $file );
 196                         if ( !$chunk ) {
 197                                 break;
 198                         }
 199
 200                         if ( $chunk == 'FORM' ) {
 201                                 $subtype = fread( $file, 4 );
 202                                 if ( $subtype == 'DJVU' ) {
 203                                         wfDebug( __METHOD__ . ": found first subpage\n" );
 204
 205                                         return $this->getPageInfo( $file );
 206                                 }
 207                                 $this->skipChunk( $file, $length - 4 );
 208                         } else {
 209                                 wfDebug( __METHOD__ . ": skipping '$chunk' chunk\n" );
 210                                 $this->skipChunk( $file, $length );
 211                         }
 212                 } while ( $length != 0 && !feof( $file ) && ftell( $file ) - $start < $formLength );
 213
 214                 wfDebug( __METHOD__ . ": multi-page DJVU file contained no pages\n" );
 215
 216                 return false;
 217         }
 218
 219         private function getPageInfo( $file ) {
 220                 list( $chunk, $length ) = $this->readChunk( $file );
 221                 if ( $chunk != 'INFO' ) {
 222                         wfDebug( __METHOD__ . ": expected INFO chunk, got '$chunk'\n" );
 223
 224                         return false;
 225                 }
 226
 227                 if ( $length < 9 ) {
 228                         wfDebug( __METHOD__ . ": INFO should be 9 or 10 bytes, found $length\n" );
 229
 230                         return false;
 231                 }
 232                 $data = fread( $file, $length );
 233                 if ( strlen( $data ) < $length ) {
 234                         wfDebug( __METHOD__ . ": INFO chunk cut off\n" );
 235
 236                         return false;
 237                 }
 238
 239                 // @todo FIXME: Would be good to replace this extract() call with
 240                 // something that explicitly initializes local variables.
 241                 extract( unpack(
 242                         'nwidth/' .
 243                         'nheight/' .
 244                         'Cminor/' .
 245                         'Cmajor/' .
 246                         'vresolution/' .
 247                         'Cgamma', $data ) );
 248
 249                 # Newer files have rotation info in byte 10, but we don't use it yet.
 250
 251                 /** @var string $width
 252                  * @var string $height
 253                  * @var string $major
 254                  * @var string $minor
 255                  * @var string $resolution
 256                  * @var string $length
 257                  * @var string $gamma */
 258                 return [
 259                         'width' => $width,
 260                         'height' => $height,
 261                         'version' => "$major.$minor",
 262                         'resolution' => $resolution,
 263                         'gamma' => $gamma / 10.0 ];
 264         }
 265
 266         /**
 267          * Return an XML string describing the DjVu image
 268          * @return string|bool
 269          */
 270         function retrieveMetaData() {
 271                 global $wgDjvuToXML, $wgDjvuDump, $wgDjvuTxt;
 272
 273                 if ( !$this->isValid() ) {
 274                         return false;
 275                 }
 276
 277                 if ( isset( $wgDjvuDump ) ) {
 278                         # djvudump is faster as of version 3.5
 279                         # http://sourceforge.net/tracker/index.php?func=detail&aid=1704049&group_id=32953&atid=406583
 280                         $cmd = wfEscapeShellArg( $wgDjvuDump ) . ' ' . wfEscapeShellArg( $this->mFilename );
 281                         $dump = wfShellExec( $cmd );
 282                         $xml = $this->convertDumpToXML( $dump );
 283                 } elseif ( isset( $wgDjvuToXML ) ) {
 284                         $cmd = wfEscapeShellArg( $wgDjvuToXML ) . ' --without-anno --without-text ' .
 285                                 wfEscapeShellArg( $this->mFilename );
 286                         $xml = wfShellExec( $cmd );
 287                 } else {
 288                         $xml = null;
 289                 }
 290                 # Text layer
 291                 if ( isset( $wgDjvuTxt ) ) {
 292                         $cmd = wfEscapeShellArg( $wgDjvuTxt ) . ' --detail=page ' . wfEscapeShellArg( $this->mFilename );
 293                         wfDebug( __METHOD__ . ": $cmd\n" );
 294                         $retval = '';
 295                         $txt = wfShellExec( $cmd, $retval, [], [ 'memory' => self::DJVUTXT_MEMORY_LIMIT ] );
 296                         if ( $retval == 0 ) {
 297                                 # Strip some control characters
 298                                 $txt = preg_replace( "/[\013\035\037]/", "", $txt );
 299                                 $reg = <<<EOR
 300                                         /\(page\s[\d-]*\s[\d-]*\s[\d-]*\s[\d-]*\s*"
 301                                         ((?>    # Text to match is composed of atoms of either:
 302                                           \\\\. # - any escaped character
 303                                           |     # - any character different from " and \
 304                                           [^"\\\\]+
 305                                         )*?)
 306                                         "\s*\)
 307                                         | # Or page can be empty ; in this case, djvutxt dumps ()
 308                                         \(\s*()\)/sx
 309 EOR;
 310                                 $txt = preg_replace_callback( $reg, [ $this, 'pageTextCallback' ], $txt );
 311                                 $txt = "<DjVuTxt>\n<HEAD></HEAD>\n<BODY>\n" . $txt . "</BODY>\n</DjVuTxt>\n";
 312                                 $xml = preg_replace( "/<DjVuXML>/", "<mw-djvu><DjVuXML>", $xml, 1 );
 313                                 $xml = $xml . $txt . '</mw-djvu>';
 314                         }
 315                 }
 316
 317                 return $xml;
 318         }
 319
 320         function pageTextCallback( $matches ) {
 321                 # Get rid of invalid UTF-8, strip control characters
 322                 $val = htmlspecialchars( UtfNormal\Validator::cleanUp( stripcslashes( $matches[1] ) ) );
 323                 $val = str_replace( [ "\n", '�' ], [ '&#10;', '' ], $val );
 324                 return '<PAGE value="' . $val . '" />';
 325         }
 326
 327         /**
 328          * Hack to temporarily work around djvutoxml bug
 329          * @param string $dump
 330          * @return string
 331          */
 332         function convertDumpToXML( $dump ) {
 333                 if ( strval( $dump ) == '' ) {
 334                         return false;
 335                 }
 336
 337                 $xml = <<<EOT
 338 <?xml version="1.0" ?>
 339 <!DOCTYPE DjVuXML PUBLIC "-//W3C//DTD DjVuXML 1.1//EN" "pubtext/DjVuXML-s.dtd">
 340 <DjVuXML>
 341 <HEAD></HEAD>
 342 <BODY>
 343 EOT;
 344
 345                 $dump = str_replace( "\r", '', $dump );
 346                 $line = strtok( $dump, "\n" );
 347                 $m = false;
 348                 $good = false;
 349                 if ( preg_match( '/^( *)FORM:DJVU/', $line, $m ) ) {
 350                         # Single-page
 351                         if ( $this->parseFormDjvu( $line, $xml ) ) {
 352                                 $good = true;
 353                         } else {
 354                                 return false;
 355                         }
 356                 } elseif ( preg_match( '/^( *)FORM:DJVM/', $line, $m ) ) {
 357                         # Multi-page
 358                         $parentLevel = strlen( $m[1] );
 359                         # Find DIRM
 360                         $line = strtok( "\n" );
 361                         while ( $line !== false ) {
 362                                 $childLevel = strspn( $line, ' ' );
 363                                 if ( $childLevel <= $parentLevel ) {
 364                                         # End of chunk
 365                                         break;
 366                                 }
 367
 368                                 if ( preg_match( '/^ *DIRM.*indirect/', $line ) ) {
 369                                         wfDebug( "Indirect multi-page DjVu document, bad for server!\n" );
 370
 371                                         return false;
 372                                 }
 373                                 if ( preg_match( '/^ *FORM:DJVU/', $line ) ) {
 374                                         # Found page
 375                                         if ( $this->parseFormDjvu( $line, $xml ) ) {
 376                                                 $good = true;
 377                                         } else {
 378                                                 return false;
 379                                         }
 380                                 }
 381                                 $line = strtok( "\n" );
 382                         }
 383                 }
 384                 if ( !$good ) {
 385                         return false;
 386                 }
 387
 388                 $xml .= "</BODY>\n</DjVuXML>\n";
 389
 390                 return $xml;
 391         }
 392
 393         function parseFormDjvu( $line, &$xml ) {
 394                 $parentLevel = strspn( $line, ' ' );
 395                 $line = strtok( "\n" );
 396
 397                 # Find INFO
 398                 while ( $line !== false ) {
 399                         $childLevel = strspn( $line, ' ' );
 400                         if ( $childLevel <= $parentLevel ) {
 401                                 # End of chunk
 402                                 break;
 403                         }
 404
 405                         if ( preg_match(
 406                                 '/^ *INFO *\[\d*\] *DjVu *(\d+)x(\d+), *\w*, *(\d+) *dpi, *gamma=([0-9.-]+)/',
 407                                 $line,
 408                                 $m
 409                         ) ) {
 410                                 $xml .= Xml::tags(
 411                                         'OBJECT',
 412                                         [
 413                                                 # 'data' => '',
 414                                                 # 'type' => 'image/x.djvu',
 415                                                 'height' => $m[2],
 416                                                 'width' => $m[1],
 417                                                 # 'usemap' => '',
 418                                         ],
 419                                         "\n" .
 420                                                 Xml::element( 'PARAM', [ 'name' => 'DPI', 'value' => $m[3] ] ) . "\n" .
 421                                                 Xml::element( 'PARAM', [ 'name' => 'GAMMA', 'value' => $m[4] ] ) . "\n"
 422                                 ) . "\n";
 423
 424                                 return true;
 425                         }
 426                         $line = strtok( "\n" );
 427                 }
 428
 429                 # Not found
 430                 return false;
 431         }
 432 }