Re: encode_for_xml and CDATA

[Tibor Simko]

On Thu, 22 Apr 2010, Victor Engmark wrote:
>>>> from invenio.textutils import encode_for_xml
>>>> encode_for_xml('<subfield code="a">Learn about &lt;![CDATA[ lalala ]]&gt; 
>>>> blocks. These are great!</subfield>')
>>>> '&lt;subfield code="a"&gt;Learn about &amp;lt;![CDATA[ lalala ]]&amp;gt; 
>>>> blocks. These are great!&lt;/subfield&gt;'
>
> That should do it, no?

Let's test the full upload-index-download cycle for a real-life test ADS
record.  Dunno what CDATA contains in Benoit's use case, and whether it
is wanted to store it `as is' in Invenio DB tables.  For example, I'd
say they should rather be stored dereferenced, whenever possible,
similarly as we dereference \u03A3 into Σ during upload; this makes
indexing etc simpler.  But maybe the real-life use case is different.

Benoit, can you please send a concrete MARCXML snippet to be loaded, and
specify how you would like to see it stored/indexed/exported?
Alternatively, please just take the XML encoding bits from Victor's
public branch (vengmark/webtag), and test on your own end.

Best regards
--
Tibor Simko