Fix variable name in doc comment added in r78192

[lhc/web/wiklou.git] / includes / DjVuImage.php
diff --git a/includes/DjVuImage.php b/includes/DjVuImage.php

index fbb2586..a55017e 100644 (file)
--- a/includes/DjVuImage.php
+++ b/includes/DjVuImage.php
@@ -1,8 +1,8 @@
  <?php
-
  /**
+ * DjVu image handler
   *
- * Copyright (C) 2006 Brion Vibber <brion@pobox.com>
+ * Copyright © 2006 Brion Vibber <brion@pobox.com>
   * http://www.mediawiki.org/
   *
   * This program is free software; you can redistribute it and/or modify
@@ -20,6 +20,7 @@
   * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
   * http://www.gnu.org/copyleft/gpl.html
   *
+ * @file
   */
  
  /**
@@ -247,12 +248,25 @@ class DjVuImage {
                         wfProfileIn( 'djvutxt' );
                         $cmd = wfEscapeShellArg( $wgDjvuTxt ) . ' --detail=page ' . wfEscapeShellArg( $this->mFilename ) ;
                         wfDebug( __METHOD__.": $cmd\n" );
+                       $retval = '';
                         $txt = wfShellExec( $cmd, $retval );
                         wfProfileOut( 'djvutxt' );
                         if( $retval == 0) {
-                               $txt = htmlspecialchars($txt);
-                               $txt = preg_replace( "/\(page\s\d*\s\d*\s\d*\s\d*\s*\&quot;(.*?)\&quot;\s*\)/s", "<PAGE value=\"$1\" />", $txt  );
-                               $txt = preg_replace( "/\(\)/", "<PAGE value=\"\" />", $txt );
+                               # Get rid of invalid UTF-8, strip control characters
+                               $txt = UtfNormal::cleanUp( $txt );
+                               $txt = preg_replace( "/[\013\035\037]/", "", $txt );
+                               $reg = <<<EOR
+                                       /\(page\s[\d-]*\s[\d-]*\s[\d-]*\s[\d-]*\s*"
+                                       ((?>    # Text to match is composed of atoms of either:
+                                         \\\\. # - any escaped character 
+                                         |     # - any character different from " and \
+                                         [^"\\\\]+
+                                       )*?)
+                                       "\s*\)
+                                       | # Or page can be empty ; in this case, djvutxt dumps ()
+                                       \(\s*()\)/sx
+EOR;
+                               $txt = preg_replace_callback( $reg, array( $this, 'pageTextCallback' ), $txt );
                                 $txt = "<DjVuTxt>\n<HEAD></HEAD>\n<BODY>\n" . $txt . "</BODY>\n</DjVuTxt>\n";
                                 $xml = preg_replace( "/<DjVuXML>/", "<mw-djvu><DjVuXML>", $xml );
                                 $xml = $xml . $txt. '</mw-djvu>' ;
@@ -261,6 +275,10 @@ class DjVuImage {
                 return $xml;
         }
  
+       function pageTextCallback( $matches ) {
+               return '<PAGE value="' . htmlspecialchars( $matches[1] ) . '" />';
+       }
+
         /**
          * Hack to temporarily work around djvutoxml bug
          */