160969 [Help] Exception occurred while adding document xxx.xhtml to index.

2 files changed, 34 insertions, 1 deletions

diff --git a/org.eclipse.help.base/src/org/eclipse/help/internal/xhtml/DynamicXHTMLProcessor.java b/org.eclipse.help.base/src/org/eclipse/help/internal/xhtml/DynamicXHTMLProcessor.java
index 2fb68d733..bf1db1d57 100644
--- a/org.eclipse.help.base/src/org/eclipse/help/internal/xhtml/DynamicXHTMLProcessor.java
+++ b/org.eclipse.help.base/src/org/eclipse/help/internal/xhtml/DynamicXHTMLProcessor.java
@@ -47,6 +47,7 @@ public class DynamicXHTMLProcessor {
 						xmlProcessor = new XMLProcessor(new DOMProcessorHandler[] {
 								new IncludeHandler(locale),
 								new ExtensionHandler(locale),
+								new XHTMLCharsetHandler(),
 								new FilterHandler()
 						});
 					}
@@ -55,7 +56,8 @@ public class DynamicXHTMLProcessor {
 				if (xmlProcessorNoFilter == null) {
 					xmlProcessorNoFilter = new XMLProcessor(new DOMProcessorHandler[] {
 							new IncludeHandler(locale),
-							new ExtensionHandler(locale)
+							new ExtensionHandler(locale),
+							new XHTMLCharsetHandler()
 					});
 				}
 				return xmlProcessorNoFilter.process(buf, href);
diff --git a/org.eclipse.help.base/src/org/eclipse/help/internal/xhtml/XHTMLCharsetHandler.java b/org.eclipse.help.base/src/org/eclipse/help/internal/xhtml/XHTMLCharsetHandler.java
new file mode 100644
index 000000000..fb636b1b0
--- /dev/null
+++ b/org.eclipse.help.base/src/org/eclipse/help/internal/xhtml/XHTMLCharsetHandler.java
@@ -0,0 +1,31 @@
+package org.eclipse.help.internal.xhtml;
+
+import org.eclipse.help.internal.dynamic.DOMProcessorHandler;
+import org.w3c.dom.Element;
+
+/*
+ * Converts the charset in XHTML meta tag to UTF-8. This is the encoding
+ * output by the XMLProcessor, and we need the charset in the meta tags
+ * to match, otherwise browsers will be confused.
+ */
+public class XHTMLCharsetHandler extends DOMProcessorHandler {
+
+	private static final String ELEMENT_META = "meta"; //$NON-NLS-1$
+	private static final String ATTRIBUTE_CONTENT = "content"; //$NON-NLS-1$
+	private static final String PREFIX_CHARSET = "text/html; charset="; //$NON-NLS-1$
+	private static final String ENCODING_UTF8 = "UTF-8"; //$NON-NLS-1$
+
+	/* (non-Javadoc)
+	 * @see org.eclipse.help.internal.dynamic.DOMProcessorHandler#handle(org.w3c.dom.Element, java.lang.String)
+	 */
+	public short handle(Element elem, String id) {
+		if (ELEMENT_META.equals(elem.getNodeName())) {
+			String content = elem.getAttribute(ATTRIBUTE_CONTENT);
+			if (content.startsWith(PREFIX_CHARSET)) {
+				elem.setAttribute(ATTRIBUTE_CONTENT, PREFIX_CHARSET + ENCODING_UTF8);
+				return HANDLED_CONTINUE;
+			}
+		}
+		return UNHANDLED;
+	}
+}