MediaWiki  REL1_22
DjVuImage.php
Go to the documentation of this file.
00001 <?php
00036 class DjVuImage {
00042     function __construct( $filename ) {
00043         $this->mFilename = $filename;
00044     }
00045 
00049     const DJVUTXT_MEMORY_LIMIT = 300000;
00050 
00055     public function isValid() {
00056         $info = $this->getInfo();
00057         return $info !== false;
00058     }
00059 
00064     public function getImageSize() {
00065         $data = $this->getInfo();
00066 
00067         if ( $data !== false ) {
00068             $width = $data['width'];
00069             $height = $data['height'];
00070 
00071             return array( $width, $height, 'DjVu',
00072                 "width=\"$width\" height=\"$height\"" );
00073         }
00074         return false;
00075     }
00076 
00077     // ---------
00078 
00082     function dump() {
00083         $file = fopen( $this->mFilename, 'rb' );
00084         $header = fread( $file, 12 );
00085         // @todo FIXME: Would be good to replace this extract() call with something that explicitly initializes local variables.
00086         extract( unpack( 'a4magic/a4chunk/NchunkLength', $header ) );
00087         echo "$chunk $chunkLength\n";
00088         $this->dumpForm( $file, $chunkLength, 1 );
00089         fclose( $file );
00090     }
00091 
00092     private function dumpForm( $file, $length, $indent ) {
00093         $start = ftell( $file );
00094         $secondary = fread( $file, 4 );
00095         echo str_repeat( ' ', $indent * 4 ) . "($secondary)\n";
00096         while ( ftell( $file ) - $start < $length ) {
00097             $chunkHeader = fread( $file, 8 );
00098             if ( $chunkHeader == '' ) {
00099                 break;
00100             }
00101             // @todo FIXME: Would be good to replace this extract() call with something that explicitly initializes local variables.
00102             extract( unpack( 'a4chunk/NchunkLength', $chunkHeader ) );
00103             echo str_repeat( ' ', $indent * 4 ) . "$chunk $chunkLength\n";
00104 
00105             if ( $chunk == 'FORM' ) {
00106                 $this->dumpForm( $file, $chunkLength, $indent + 1 );
00107             } else {
00108                 fseek( $file, $chunkLength, SEEK_CUR );
00109                 if ( $chunkLength & 1 == 1 ) {
00110                     // Padding byte between chunks
00111                     fseek( $file, 1, SEEK_CUR );
00112                 }
00113             }
00114         }
00115     }
00116 
00117     function getInfo() {
00118         wfSuppressWarnings();
00119         $file = fopen( $this->mFilename, 'rb' );
00120         wfRestoreWarnings();
00121         if ( $file === false ) {
00122             wfDebug( __METHOD__ . ": missing or failed file read\n" );
00123             return false;
00124         }
00125 
00126         $header = fread( $file, 16 );
00127         $info = false;
00128 
00129         if ( strlen( $header ) < 16 ) {
00130             wfDebug( __METHOD__ . ": too short file header\n" );
00131         } else {
00132             // @todo FIXME: Would be good to replace this extract() call with something that explicitly initializes local variables.
00133             extract( unpack( 'a4magic/a4form/NformLength/a4subtype', $header ) );
00134 
00135             if ( $magic != 'AT&T' ) {
00136                 wfDebug( __METHOD__ . ": not a DjVu file\n" );
00137             } elseif ( $subtype == 'DJVU' ) {
00138                 // Single-page document
00139                 $info = $this->getPageInfo( $file, $formLength );
00140             } elseif ( $subtype == 'DJVM' ) {
00141                 // Multi-page document
00142                 $info = $this->getMultiPageInfo( $file, $formLength );
00143             } else {
00144                 wfDebug( __METHOD__ . ": unrecognized DJVU file type '$formType'\n" );
00145             }
00146         }
00147         fclose( $file );
00148         return $info;
00149     }
00150 
00151     private function readChunk( $file ) {
00152         $header = fread( $file, 8 );
00153         if ( strlen( $header ) < 8 ) {
00154             return array( false, 0 );
00155         } else {
00156             // @todo FIXME: Would be good to replace this extract() call with something that explicitly initializes local variables.
00157             extract( unpack( 'a4chunk/Nlength', $header ) );
00158             return array( $chunk, $length );
00159         }
00160     }
00161 
00162     private function skipChunk( $file, $chunkLength ) {
00163         fseek( $file, $chunkLength, SEEK_CUR );
00164 
00165         if ( $chunkLength & 0x01 == 1 && !feof( $file ) ) {
00166             // padding byte
00167             fseek( $file, 1, SEEK_CUR );
00168         }
00169     }
00170 
00171     private function getMultiPageInfo( $file, $formLength ) {
00172         // For now, we'll just look for the first page in the file
00173         // and report its information, hoping others are the same size.
00174         $start = ftell( $file );
00175         do {
00176             list( $chunk, $length ) = $this->readChunk( $file );
00177             if ( !$chunk ) {
00178                 break;
00179             }
00180 
00181             if ( $chunk == 'FORM' ) {
00182                 $subtype = fread( $file, 4 );
00183                 if ( $subtype == 'DJVU' ) {
00184                     wfDebug( __METHOD__ . ": found first subpage\n" );
00185                     return $this->getPageInfo( $file, $length );
00186                 }
00187                 $this->skipChunk( $file, $length - 4 );
00188             } else {
00189                 wfDebug( __METHOD__ . ": skipping '$chunk' chunk\n" );
00190                 $this->skipChunk( $file, $length );
00191             }
00192         } while ( $length != 0 && !feof( $file ) && ftell( $file ) - $start < $formLength );
00193 
00194         wfDebug( __METHOD__ . ": multi-page DJVU file contained no pages\n" );
00195         return false;
00196     }
00197 
00198     private function getPageInfo( $file, $formLength ) {
00199         list( $chunk, $length ) = $this->readChunk( $file );
00200         if ( $chunk != 'INFO' ) {
00201             wfDebug( __METHOD__ . ": expected INFO chunk, got '$chunk'\n" );
00202             return false;
00203         }
00204 
00205         if ( $length < 9 ) {
00206             wfDebug( __METHOD__ . ": INFO should be 9 or 10 bytes, found $length\n" );
00207             return false;
00208         }
00209         $data = fread( $file, $length );
00210         if ( strlen( $data ) < $length ) {
00211             wfDebug( __METHOD__ . ": INFO chunk cut off\n" );
00212             return false;
00213         }
00214 
00215         // @todo FIXME: Would be good to replace this extract() call with something that explicitly initializes local variables.
00216         extract( unpack(
00217             'nwidth/' .
00218             'nheight/' .
00219             'Cminor/' .
00220             'Cmajor/' .
00221             'vresolution/' .
00222             'Cgamma', $data ) );
00223         # Newer files have rotation info in byte 10, but we don't use it yet.
00224 
00225         return array(
00226             'width' => $width,
00227             'height' => $height,
00228             'version' => "$major.$minor",
00229             'resolution' => $resolution,
00230             'gamma' => $gamma / 10.0 );
00231     }
00232 
00237     function retrieveMetaData() {
00238         global $wgDjvuToXML, $wgDjvuDump, $wgDjvuTxt;
00239         wfProfileIn( __METHOD__ );
00240 
00241         if ( isset( $wgDjvuDump ) ) {
00242             # djvudump is faster as of version 3.5
00243             # http://sourceforge.net/tracker/index.php?func=detail&aid=1704049&group_id=32953&atid=406583
00244             wfProfileIn( 'djvudump' );
00245             $cmd = wfEscapeShellArg( $wgDjvuDump ) . ' ' . wfEscapeShellArg( $this->mFilename );
00246             $dump = wfShellExec( $cmd );
00247             $xml = $this->convertDumpToXML( $dump );
00248             wfProfileOut( 'djvudump' );
00249         } elseif ( isset( $wgDjvuToXML ) ) {
00250             wfProfileIn( 'djvutoxml' );
00251             $cmd = wfEscapeShellArg( $wgDjvuToXML ) . ' --without-anno --without-text ' .
00252                 wfEscapeShellArg( $this->mFilename );
00253             $xml = wfShellExec( $cmd );
00254             wfProfileOut( 'djvutoxml' );
00255         } else {
00256             $xml = null;
00257         }
00258         # Text layer
00259         if ( isset( $wgDjvuTxt ) ) {
00260             wfProfileIn( 'djvutxt' );
00261             $cmd = wfEscapeShellArg( $wgDjvuTxt ) . ' --detail=page ' . wfEscapeShellArg( $this->mFilename );
00262             wfDebug( __METHOD__ . ": $cmd\n" );
00263             $retval = '';
00264             $txt = wfShellExec( $cmd, $retval, array(), array( 'memory' => self::DJVUTXT_MEMORY_LIMIT ) );
00265             wfProfileOut( 'djvutxt' );
00266             if ( $retval == 0 ) {
00267                 # Strip some control characters
00268                 $txt = preg_replace( "/[\013\035\037]/", "", $txt );
00269                 $reg = <<<EOR
00270                     /\(page\s[\d-]*\s[\d-]*\s[\d-]*\s[\d-]*\s*"
00271                     ((?>    # Text to match is composed of atoms of either:
00272                       \\\\. # - any escaped character
00273                       |     # - any character different from " and \
00274                       [^"\\\\]+
00275                     )*?)
00276                     "\s*\)
00277                     | # Or page can be empty ; in this case, djvutxt dumps ()
00278                     \(\s*()\)/sx
00279 EOR;
00280                 $txt = preg_replace_callback( $reg, array( $this, 'pageTextCallback' ), $txt );
00281                 $txt = "<DjVuTxt>\n<HEAD></HEAD>\n<BODY>\n" . $txt . "</BODY>\n</DjVuTxt>\n";
00282                 $xml = preg_replace( "/<DjVuXML>/", "<mw-djvu><DjVuXML>", $xml, 1 );
00283                 $xml = $xml . $txt . '</mw-djvu>';
00284             }
00285         }
00286         wfProfileOut( __METHOD__ );
00287         return $xml;
00288     }
00289 
00290     function pageTextCallback( $matches ) {
00291         # Get rid of invalid UTF-8, strip control characters
00292         return '<PAGE value="' . htmlspecialchars( UtfNormal::cleanUp( $matches[1] ) ) . '" />';
00293     }
00294 
00299     function convertDumpToXML( $dump ) {
00300         if ( strval( $dump ) == '' ) {
00301             return false;
00302         }
00303 
00304         $xml = <<<EOT
00305 <?xml version="1.0" ?>
00306 <!DOCTYPE DjVuXML PUBLIC "-//W3C//DTD DjVuXML 1.1//EN" "pubtext/DjVuXML-s.dtd">
00307 <DjVuXML>
00308 <HEAD></HEAD>
00309 <BODY>
00310 EOT;
00311 
00312         $dump = str_replace( "\r", '', $dump );
00313         $line = strtok( $dump, "\n" );
00314         $m = false;
00315         $good = false;
00316         if ( preg_match( '/^( *)FORM:DJVU/', $line, $m ) ) {
00317             # Single-page
00318             if ( $this->parseFormDjvu( $line, $xml ) ) {
00319                 $good = true;
00320             } else {
00321                 return false;
00322             }
00323         } elseif ( preg_match( '/^( *)FORM:DJVM/', $line, $m ) ) {
00324             # Multi-page
00325             $parentLevel = strlen( $m[1] );
00326             # Find DIRM
00327             $line = strtok( "\n" );
00328             while ( $line !== false ) {
00329                 $childLevel = strspn( $line, ' ' );
00330                 if ( $childLevel <= $parentLevel ) {
00331                     # End of chunk
00332                     break;
00333                 }
00334 
00335                 if ( preg_match( '/^ *DIRM.*indirect/', $line ) ) {
00336                     wfDebug( "Indirect multi-page DjVu document, bad for server!\n" );
00337                     return false;
00338                 }
00339                 if ( preg_match( '/^ *FORM:DJVU/', $line ) ) {
00340                     # Found page
00341                     if ( $this->parseFormDjvu( $line, $xml ) ) {
00342                         $good = true;
00343                     } else {
00344                         return false;
00345                     }
00346                 }
00347                 $line = strtok( "\n" );
00348             }
00349         }
00350         if ( !$good ) {
00351             return false;
00352         }
00353 
00354         $xml .= "</BODY>\n</DjVuXML>\n";
00355         return $xml;
00356     }
00357 
00358     function parseFormDjvu( $line, &$xml ) {
00359         $parentLevel = strspn( $line, ' ' );
00360         $line = strtok( "\n" );
00361 
00362         # Find INFO
00363         while ( $line !== false ) {
00364             $childLevel = strspn( $line, ' ' );
00365             if ( $childLevel <= $parentLevel ) {
00366                 # End of chunk
00367                 break;
00368             }
00369 
00370             if ( preg_match( '/^ *INFO *\[\d*\] *DjVu *(\d+)x(\d+), *\w*, *(\d+) *dpi, *gamma=([0-9.-]+)/', $line, $m ) ) {
00371                 $xml .= Xml::tags( 'OBJECT',
00372                     array(
00373                         #'data' => '',
00374                         #'type' => 'image/x.djvu',
00375                         'height' => $m[2],
00376                         'width' => $m[1],
00377                         #'usemap' => '',
00378                     ),
00379                     "\n" .
00380                     Xml::element( 'PARAM', array( 'name' => 'DPI', 'value' => $m[3] ) ) . "\n" .
00381                     Xml::element( 'PARAM', array( 'name' => 'GAMMA', 'value' => $m[4] ) ) . "\n"
00382                 ) . "\n";
00383                 return true;
00384             }
00385             $line = strtok( "\n" );
00386         }
00387         # Not found
00388         return false;
00389     }
00390 }