OD format and localizable contents

[JC Helary <fusion@mx6.tiki.ne.jp>]

(3 mails combined in one. Already sent to xml-dev on OOo, but I  
thought maybe this list was more relevant)

I would like to know if there is a "simple" way to identify:

-block level textual information
-inline level textual information
-localizable subflows present in tags attributes values.

This, to parse an OD file and be able to properly segment the text  
for use in a CAT tool.

By "subflows" I mean information that is not between tags but inside  
tags as attributes values ( text:name="something like alternative  
text for graphic items for ex").

It looks like there is a very wide range of possible  
<text:"attributes" and I found that <text:sequence seems to be inline  
as well as <text:user-defined

I am looking for an extensive list of such attributes and their  
characteristics (block or inline).

The point being that OD needs a proper parsing of its localisable  
data so that computer assisted translation tool (the apps translators  
use everyday) give proper access to the proper data within their  
translation framework.

It would be nice if OD provided such "meta" information as too what  
is localizable, what is not and how that fits in the block/inline/ 
subflow that most translation related standards (TMX/XLIFF etc) are  
based on.

Of course, developers can always check the specification and take  
guesses as to what is what, just like what I trying to do right now,  
but I'd say that's could be part of the OD specification to provide  
easy access to such meta information to make sure there is proper  
implementation on the CAT tool development side.

Regards,
Jean-Christophe Helary