diff options
Diffstat (limited to 'includes/DjVuImage.php')
| -rw-r--r-- | includes/DjVuImage.php | 134 |
1 files changed, 123 insertions, 11 deletions
diff --git a/includes/DjVuImage.php b/includes/DjVuImage.php index 3b8a68ba..1e423565 100644 --- a/includes/DjVuImage.php +++ b/includes/DjVuImage.php @@ -1,11 +1,6 @@ <?php + /** - * 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/ - * * * Copyright (C) 2006 Brion Vibber <brion@pobox.com> * http://www.mediawiki.org/ @@ -25,9 +20,17 @@ * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. * http://www.gnu.org/copyleft/gpl.html * - * @package MediaWiki */ +/** + * 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/ + * + * @addtogroup Media + */ class DjVuImage { function __construct( $filename ) { $this->mFilename = $filename; @@ -68,6 +71,7 @@ class DjVuImage { function dump() { $file = fopen( $this->mFilename, 'rb' ); $header = fread( $file, 12 ); + // FIXME: Would be good to replace this extract() call with something that explicitly initializes local variables. extract( unpack( 'a4magic/a4chunk/NchunkLength', $header ) ); echo "$chunk $chunkLength\n"; $this->dumpForm( $file, $chunkLength, 1 ); @@ -83,6 +87,7 @@ class DjVuImage { if( $chunkHeader == '' ) { break; } + // 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"; @@ -111,6 +116,7 @@ class DjVuImage { 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' ) { @@ -134,6 +140,7 @@ class DjVuImage { if( strlen( $header ) < 8 ) { return array( false, 0 ); } else { + // FIXME: Would be good to replace this extract() call with something that explicitly initializes local variables. extract( unpack( 'a4chunk/Nlength', $header ) ); return array( $chunk, $length ); } @@ -192,6 +199,7 @@ class DjVuImage { return false; } + // FIXME: Would be good to replace this extract() call with something that explicitly initializes local variables. extract( unpack( 'nwidth/' . 'nheight/' . @@ -214,17 +222,121 @@ class DjVuImage { * @return string */ function retrieveMetaData() { - global $wgDjvuToXML; - if ( isset( $wgDjvuToXML ) ) { - $cmd = $wgDjvuToXML . ' --without-anno --without-text ' . + global $wgDjvuToXML, $wgDjvuDump; + 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; } return $xml; } - + + /** + * Hack to temporarily work around djvutoxml bug + */ + function convertDumpToXML( $dump ) { + if ( strval( $dump ) == '' ) { + return false; + } + + $xml = <<<EOT +<?xml version="1.0" ?> +<!DOCTYPE DjVuXML PUBLIC "-//W3C//DTD DjVuXML 1.1//EN" "pubtext/DjVuXML-s.dtd"> +<DjVuXML> +<HEAD></HEAD> +<BODY> +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 .= "</BODY>\n</DjVuXML>\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; + } } |