Fri Sep 5 15:57:31 CEST 2008


Trying to typeset Marycela's book.  Making some routines to process
.fodt XML files.  Load and save work.

This could save as a nice frontend for Melissa's writer website.
Instead of asking people to create HTML files, maybe they should just
stick with wordprocessors?

This looks like it's doable, if the style used is a bit restricted.
Marycela uses a lot of `space and tab' formatting, which is difficult
to recover properly.

Let's try to artificially create some documents and see how they look
in XML.

Hmm.. Coming from the structured documents side, it seems odd that
text is not structured.  Text is a sequence of paragraphs, where some
paragraphs might be tagged with a "Heading" style.  This is then
collected to make table-of-contents etc..

.odt is a JAR (ZIP) file.  Apparently the flat XML is not the standard
one.  And paragraphs and headers are separated: 'text:p and 'text:h
tags.  ZIP is supported in PLT Scheme through dherman/zip


Works like a charm:

(define (load-odt filename)
   (read-zip-directory filename)
   (lambda (name dir port)
       (read-xml port))))))