[developers] MRS XML support

Ben Waldron benjamin.waldron at cl.cam.ac.uk
Sat May 6 22:16:52 CEST 2006


Ann Copestake wrote:

>As usual, testing and bug reporting soon would be appreciated.
>`Funny' characters in rel names etc are a possible issue in terms of
>XML validity - we should perhaps test for these in the sem-i so that
>users can be notified when the lexicon is validated.  I would leave
>this up to Ben, however.
>  
>
The clean solution is to escape special XML characters when serializing 
to XML, so that the MRS XML contains eg.

    <spred>_nasty&lt;&gt;&apos;&quot;&amp;_1_rel</spred>

rather than the following invalid XML:

    <spred>_nasty<>'"&_1_rel</spred>

The function lkb::xml-escape can take care of this. Eg.

    (format stream "<spred>~a</spred>" (lkb::xml-escape pred))

in place of

    (format stream "<spred>~a</spred>" pred)

Shall I commit the necessary changes to the output methods of the 
'mrs-xml' class in mrs/basemrs.lisp?

- Ben



More information about the developers mailing list