Bad Html Entities encoding in Text Styling

The problem is visible when using XDocReport text styling feature.
For example, render a Note inside a OpenOffice document.

XDocReport is used to generate ODT XML structure from the HTML of the Note.
Because of a missing encoding inside XDocreport TextStyling (may be my fault), the generate XML document may be not valid.

Example:

source Html :

<p> A &amp; B </p>

3 Text nodes are read :

 A 

 & 

 B 

In the second node the entity is decoded.

Then in the generated XML the & is not encoded.

xml output :

 
<text:p>  
  <text:span> A </text:span>
  <text:span> & </text:span>
  <text:span> B </text:span>
</text:p>

This XML is invalid, so OOo/LibreOffice will crash when loading the doc.

See XdocReport issue : http://code.google.com/p/xdocreport/issues/detail?id=173

5.6 :
do a quick fix like "remove the Html Entities that would break XML if used in decoded form"

5.7 :
align on 1.0.0 version of XDocReport