Author: scottbw
Date: Fri May 27 20:27:11 2011
New Revision: 1128443
URL: http://svn.apache.org/viewvc?rev=1128443&view=rev
Log:
Fixed a potential issue in the parser with double-encoding XML content,
particularly nested text nodes, when recursively extracting text from localized
text fields.
Modified:
incubator/wookie/trunk/parser/java/src/org/apache/wookie/w3c/impl/AbstractLocalizedEntity.java
Modified:
incubator/wookie/trunk/parser/java/src/org/apache/wookie/w3c/impl/AbstractLocalizedEntity.java
URL:
http://svn.apache.org/viewvc/incubator/wookie/trunk/parser/java/src/org/apache/wookie/w3c/impl/AbstractLocalizedEntity.java?rev=1128443&r1=1128442&r2=1128443&view=diff
==============================================================================
---
incubator/wookie/trunk/parser/java/src/org/apache/wookie/w3c/impl/AbstractLocalizedEntity.java
(original)
+++
incubator/wookie/trunk/parser/java/src/org/apache/wookie/w3c/impl/AbstractLocalizedEntity.java
Fri May 27 20:27:11 2011
@@ -13,6 +13,7 @@
*/
package org.apache.wookie.w3c.impl;
+import org.apache.commons.lang.StringEscapeUtils;
import org.apache.wookie.w3c.ILocalizedEntity;
import org.apache.wookie.w3c.IW3CXMLConfiguration;
import org.apache.wookie.w3c.util.LocalizationUtils;
@@ -99,8 +100,13 @@ public abstract class AbstractLocalizedE
content.append(getLocalizedTextContent((Element)node));
}
}
+ // Append text to the string
+ // First we have to unescape any XML special characters
so we don't
+ // double-encode them (e.g. ´ = ´) when
exporting to
+ // HTML or XML later
if (node instanceof Text){
- content.append(((Text)node).getText());
+ String text = ((Text)node).getText();
+
content.append(StringEscapeUtils.unescapeXml(text));
}
}
return UnicodeUtils.normalizeWhitespace(content.toString());
