dépôts / lhc / web / wiklou.git / commitdiff
? search:
summary | shortlog | log | commit | commitdiff | tree
raw | patch | inline | side by side (parent: a1837b2)
fix text layer extraction (bug 21526); patch by Simon Lipp
authorThomasV <thomasv@users.mediawiki.org>
Wed, 7 Jul 2010 11:05:00 +0000 (11:05 +0000)
committerThomasV <thomasv@users.mediawiki.org>
Wed, 7 Jul 2010 11:05:00 +0000 (11:05 +0000)
includes/DjVuImage.php patch | blob | history

diff --git a/includes/DjVuImage.php b/includes/DjVuImage.php
index 75df0fd..f14d8d6 100644 (file)
--- a/includes/DjVuImage.php
+++ b/includes/DjVuImage.php
@@ -259,8 +259,21 @@ class DjVuImage {
                                        $txt = UtfNormal::cleanUp( $txt );
                                }
                                $txt = preg_replace( "/[\013\035\037]/", "", $txt );
-                               $txt = htmlspecialchars($txt);
-                               $txt = preg_replace( "/\((page\s[\d-]*\s[\d-]*\s[\d-]*\s[\d-]*\s*\&quot;([^<]*?)\&quot;\s*|)\)/s", "<PAGE value=\"$2\" />", $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( $reg,
+                                       create_function('$matches', 'return \'<PAGE value="\'.htmlspecialchars($matches[1]).\'" />\';'),
+                                       $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>' ;
Wiklou, le Wiki du Wiklou