Hexadecimal and decimal represenations of characters are converted to the utf-8 character by the oneform editor. A   is transformed to a blank. If there are serveral   they are converted to blanks and they are shown as one blank. So it's impossible to have serveral leading blanks as in lenya 1.2.x
The change happens when the document is written. If the document already contains a   it is displayed correctly in the editor.
The content parameter contains the   entity when the usecase handler is called.
I guess the problem is the DOM conversion. Unfortunately it is necessary for link rewriting.
Using SAX instead would be virtually impossible because we use XPath to specify the link attributes, which doesn't sound like a good idea now :( Maybe we should allow a different, more SAX-compatible way to specifiy the link attributes, similar to the configuration of the LinkRewritingTransformer classes: <link-attribute namespace="http://www.w3.org/1999/xhtml" element="a" attribute="href"/>
(In reply to comment #4) > Maybe we should allow a different, more SAX-compatible way to specifiy the link > attributes, similar to the configuration of the LinkRewritingTransformer > classes: BTW, in Lenya 2.2 I introduced two classes for this purpose: * LinkRewriteAttributes * LinkRewriteAttribute A LinkRewriteAttributes object (singleton) is basically the declaration of the link attributes of a document type. I think thas can be generalized.
what part of the rewriting behaviour is the problem? imho, an entity should be equivalent to the corresponding utf-8 character, and any transformation between those two is a valid identity operation in the semantic space (if not the lexical one). i wonder what lambert is trying to accomplish here? would it help to optionally convert everything that is not 7bit-ascii into numerical entities?