X-Git-Url: http://git.heureux-cyclage.org/?a=blobdiff_plain;f=includes%2FDjVuImage.php;h=59f49af68ca8aaf1a79f40cd00a101de17abebf9;hb=274d317a0ca2777a5c5fae2007c53811397540f5;hp=7637b32feaac82b87e53598e7833b9a7762923c2;hpb=e6dc4bc95e14cdf316f33006f3afa772b4a339fe;p=lhc%2Fweb%2Fwiklou.git diff --git a/includes/DjVuImage.php b/includes/DjVuImage.php index 7637b32fea..59f49af68c 100644 --- a/includes/DjVuImage.php +++ b/includes/DjVuImage.php @@ -1,13 +1,8 @@ + * Copyright © 2006 Brion Vibber * http://www.mediawiki.org/ * * This program is free software; you can redistribute it and/or modify @@ -25,14 +20,23 @@ * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. * http://www.gnu.org/copyleft/gpl.html * - * @package MediaWiki + * @file */ +/** + * Support for detecting/validating DjVu image files and getting + * some basic file metadata (resolution etc) + * + * File format docs are available in source package for DjVuLibre: + * http://djvulibre.djvuzone.org/ + * + * @ingroup Media + */ class DjVuImage { function __construct( $filename ) { $this->mFilename = $filename; } - + /** * Check if the given file is indeed a valid DjVu image file * @return bool @@ -41,27 +45,27 @@ class DjVuImage { $info = $this->getInfo(); return $info !== false; } - - + + /** * Return data in the style of getimagesize() * @return array or false on failure */ public function getImageSize() { $data = $this->getInfo(); - + if( $data !== false ) { $width = $data['width']; $height = $data['height']; - + return array( $width, $height, 'DjVu', "width=\"$width\" height=\"$height\"" ); } return false; } - + // --------- - + /** * For debugging; dump the IFF chunk structure */ @@ -74,7 +78,7 @@ class DjVuImage { $this->dumpForm( $file, $chunkLength, 1 ); fclose( $file ); } - + private function dumpForm( $file, $length, $indent ) { $start = ftell( $file ); $secondary = fread( $file, 4 ); @@ -87,7 +91,7 @@ class DjVuImage { // FIXME: Would be good to replace this extract() call with something that explicitly initializes local variables. extract( unpack( 'a4chunk/NchunkLength', $chunkHeader ) ); echo str_repeat( ' ', $indent * 4 ) . "$chunk $chunkLength\n"; - + if( $chunk == 'FORM' ) { $this->dumpForm( $file, $chunkLength, $indent + 1 ); } else { @@ -99,23 +103,25 @@ class DjVuImage { } } } - + function getInfo() { + wfSuppressWarnings(); $file = fopen( $this->mFilename, 'rb' ); + wfRestoreWarnings(); if( $file === false ) { wfDebug( __METHOD__ . ": missing or failed file read\n" ); return false; } - + $header = fread( $file, 16 ); $info = false; - + if( strlen( $header ) < 16 ) { wfDebug( __METHOD__ . ": too short file header\n" ); } else { // FIXME: Would be good to replace this extract() call with something that explicitly initializes local variables. extract( unpack( 'a4magic/a4form/NformLength/a4subtype', $header ) ); - + if( $magic != 'AT&T' ) { wfDebug( __METHOD__ . ": not a DjVu file\n" ); } elseif( $subtype == 'DJVU' ) { @@ -131,7 +137,7 @@ class DjVuImage { fclose( $file ); return $info; } - + private function readChunk( $file ) { $header = fread( $file, 8 ); if( strlen( $header ) < 8 ) { @@ -142,16 +148,16 @@ class DjVuImage { return array( $chunk, $length ); } } - + private function skipChunk( $file, $chunkLength ) { fseek( $file, $chunkLength, SEEK_CUR ); - + if( $chunkLength & 0x01 == 1 && !feof( $file ) ) { // padding byte fseek( $file, 1, SEEK_CUR ); } } - + private function getMultiPageInfo( $file, $formLength ) { // For now, we'll just look for the first page in the file // and report its information, hoping others are the same size. @@ -161,7 +167,7 @@ class DjVuImage { if( !$chunk ) { break; } - + if( $chunk == 'FORM' ) { $subtype = fread( $file, 4 ); if( $subtype == 'DJVU' ) { @@ -174,18 +180,18 @@ class DjVuImage { $this->skipChunk( $file, $length ); } } while( $length != 0 && !feof( $file ) && ftell( $file ) - $start < $formLength ); - + wfDebug( __METHOD__ . ": multi-page DJVU file contained no pages\n" ); return false; } - + private function getPageInfo( $file, $formLength ) { list( $chunk, $length ) = $this->readChunk( $file ); if( $chunk != 'INFO' ) { wfDebug( __METHOD__ . ": expected INFO chunk, got '$chunk'\n" ); return false; } - + if( $length < 9 ) { wfDebug( __METHOD__ . ": INFO should be 9 or 10 bytes, found $length\n" ); return false; @@ -195,7 +201,7 @@ class DjVuImage { wfDebug( __METHOD__ . ": INFO chunk cut off\n" ); return false; } - + // FIXME: Would be good to replace this extract() call with something that explicitly initializes local variables. extract( unpack( 'nwidth/' . @@ -205,7 +211,7 @@ class DjVuImage { 'vresolution/' . 'Cgamma', $data ) ); # Newer files have rotation info in byte 10, but we don't use it yet. - + return array( 'width' => $width, 'height' => $height, @@ -219,18 +225,155 @@ class DjVuImage { * @return string */ function retrieveMetaData() { - global $wgDjvuToXML; - if ( isset( $wgDjvuToXML ) ) { - $cmd = $wgDjvuToXML . ' --without-anno --without-text ' . + global $wgDjvuToXML, $wgDjvuDump, $wgDjvuTxt; + wfProfileIn( __METHOD__ ); + + if ( isset( $wgDjvuDump ) ) { + # djvudump is faster as of version 3.5 + # http://sourceforge.net/tracker/index.php?func=detail&aid=1704049&group_id=32953&atid=406583 + wfProfileIn( 'djvudump' ); + $cmd = wfEscapeShellArg( $wgDjvuDump ) . ' ' . wfEscapeShellArg( $this->mFilename ); + $dump = wfShellExec( $cmd ); + $xml = $this->convertDumpToXML( $dump ); + wfProfileOut( 'djvudump' ); + } elseif ( isset( $wgDjvuToXML ) ) { + wfProfileIn( 'djvutoxml' ); + $cmd = wfEscapeShellArg( $wgDjvuToXML ) . ' --without-anno --without-text ' . wfEscapeShellArg( $this->mFilename ); $xml = wfShellExec( $cmd ); + wfProfileOut( 'djvutoxml' ); } else { $xml = null; } + # Text layer + if ( isset( $wgDjvuTxt ) ) { + wfProfileIn( 'djvutxt' ); + $cmd = wfEscapeShellArg( $wgDjvuTxt ) . ' --detail=page ' . wfEscapeShellArg( $this->mFilename ) ; + wfDebug( __METHOD__.": $cmd\n" ); + $retval = ''; + $txt = wfShellExec( $cmd, $retval ); + wfProfileOut( 'djvutxt' ); + if( $retval == 0) { + # Get rid of invalid UTF-8, strip control characters + $txt = UtfNormal::cleanUp( $txt ); + $txt = preg_replace( "/[\013\035\037]/", "", $txt ); + $reg = << # Text to match is composed of atoms of either: + \\\\. # - any escaped character + | # - any character different from " and \ + [^"\\\\]+ + )*?) + "\s*\) + | # Or page can be empty ; in this case, djvutxt dumps () + \(\s*()\)/sx +EOR; + $txt = preg_replace_callback( $reg, array( $this, 'pageTextCallback' ), $txt ); + $txt = "\n\n\n" . $txt . "\n\n"; + $xml = preg_replace( "//", "", $xml ); + $xml = $xml . $txt. '' ; + } + } + wfProfileOut( __METHOD__ ); return $xml; } - -} + function pageTextCallback( $matches ) { + return ''; + } + + /** + * Hack to temporarily work around djvutoxml bug + */ + function convertDumpToXML( $dump ) { + if ( strval( $dump ) == '' ) { + return false; + } + + $xml = << + + + + +EOT; + + $dump = str_replace( "\r", '', $dump ); + $line = strtok( $dump, "\n" ); + $m = false; + $good = false; + if ( preg_match( '/^( *)FORM:DJVU/', $line, $m ) ) { + # Single-page + if ( $this->parseFormDjvu( $line, $xml ) ) { + $good = true; + } else { + return false; + } + } elseif ( preg_match( '/^( *)FORM:DJVM/', $line, $m ) ) { + # Multi-page + $parentLevel = strlen( $m[1] ); + # Find DIRM + $line = strtok( "\n" ); + while ( $line !== false ) { + $childLevel = strspn( $line, ' ' ); + if ( $childLevel <= $parentLevel ) { + # End of chunk + break; + } + + if ( preg_match( '/^ *DIRM.*indirect/', $line ) ) { + wfDebug( "Indirect multi-page DjVu document, bad for server!\n" ); + return false; + } + if ( preg_match( '/^ *FORM:DJVU/', $line ) ) { + # Found page + if ( $this->parseFormDjvu( $line, $xml ) ) { + $good = true; + } else { + return false; + } + } + $line = strtok( "\n" ); + } + } + if ( !$good ) { + return false; + } + + $xml .= "\n\n"; + return $xml; + } + + function parseFormDjvu( $line, &$xml ) { + $parentLevel = strspn( $line, ' ' ); + $line = strtok( "\n" ); + + # Find INFO + while ( $line !== false ) { + $childLevel = strspn( $line, ' ' ); + if ( $childLevel <= $parentLevel ) { + # End of chunk + break; + } -?> + if ( preg_match( '/^ *INFO *\[\d*\] *DjVu *(\d+)x(\d+), *\w*, *(\d+) *dpi, *gamma=([0-9.-]+)/', $line, $m ) ) { + $xml .= Xml::tags( 'OBJECT', + array( + #'data' => '', + #'type' => 'image/x.djvu', + 'height' => $m[2], + 'width' => $m[1], + #'usemap' => '', + ), + "\n" . + Xml::element( 'PARAM', array( 'name' => 'DPI', 'value' => $m[3] ) ) . "\n" . + Xml::element( 'PARAM', array( 'name' => 'GAMMA', 'value' => $m[4] ) ) . "\n" + ) . "\n"; + return true; + } + $line = strtok( "\n" ); + } + # Not found + return false; + } +}