Re: ConTeXt to XML?

[Brooks Moses <bmoses@stanford.edu>]

At 09:18 AM 7/19/2005, Elena Fraboschi <elena@mendoza.math.indiana.edu> wrote:
>So, I have been delving into ConTeXt, and I like its syntax:  far 
>"cleaner" than LaTeX.  I have also read that there is work in progress
>to convert XML to ConTeXt --- my question is, any thoughts, hints,
>recommendations about reverse-engineering, that is, from ConTeXt to
>XML?  If the IUMJ switched production to ConTeXt, we would still want
>to preserve on "archival" copy in XML.

I'm reminded of a presentation that Kaveh Bazargan (from River Valley 
Technologies) gave at the PracticalTeX 2004 conference on a LaTeX/XML 
process they use, but unfortunately he hasn't provided any paper to go with 
his presentation.

One of the things, though, that I specifically remember was that, since 
they also were archiving XML, their process specifically converted the 
authors' LaTeX to XML (specifically with the mathematics in MathML, rather 
than as embedded LaTeX as I know some publishers do), and then converted 
that from XML back to LaTeX and thence to PDF (or Postscript, possibly) for 
actual production -- thus guaranteeing that the archival XML would in fact 
reproduce the production versions exactly.

With a workflow like that, I'm not certain how much conversion from ConTeXt 
to XML will be necessary -- you'll probably, at least for the time being, 
still be getting author submissions in LaTeX.  You could then convert that 
to XML using a version of your present process, adjust the XML as needed to 
fit your standards, archive the XML, and then dump the XML into an 
automated sort of process that converts it to ConTeXt and thence to PDF, 
html, and whatever else you need.

The only requirement, then, is that the XML -> ConTeXt -> PDF workflow not 
involve any manual adjustments to the ConTeXt code -- all manual changes 
would need to be made in the XML.  (That may be a good idea anyway, as it 
guarantees that the XML is always the "true" version.)

>(As of now we translate LaTeX to XML using "hermes", but it would
>not work with ConTeXt.)  If nothing is contemplated in the
>area of *from ConTeXt to XML", might this strategy work
>
>ConTeXt -> pdf - pdf to XML ?

I suspect that, insofar as it worked, you'd probably lose most of the 
metadata (this is a section header, this is a subsection header, etc.) 
unless it was a PDF to XML converter that was very specific to your 
particular PDF files.

...

On a different note, it may be worth pointing out that ConTeXt's support 
for typesetting of complicated equations -- that is, the sort of stuff for 
which one really wants the AMSmath package in LaTeX -- really isn't 
especially great; it's pretty much limited to the capabilities of Plain 
TeX.  I'm currently working (very slowly) on trying to improve this by 
doing a port of AMSmath to ConTeXt, but I have no idea how compatible that 
will be with ConTeXt's MathML/XML support, nor when I'd have it to a point 
where it would be ready for production work.

The other alternative I know of for doing math in XML is embedding bits of 
LaTeX code within the XML.  Since most "conversion to non-PDF formats" 
involves converting the math to bitmap images anyway (or, at least, any 
conversion to HTML for the web does!), I don't think that's really 
complicating things much.  It's certainly possible in theory to embed 
LaTeX-coded equations in ConTeXt, and there's a small bit of comment on 
this on the Wiki.  In practice, it may take a bit of coding to make work, 
but the amount won't be excessive.

...

And, on yet a third note, you might find it useful to talk to Steve 
Grathwohl at the Duke University Press -- he was also at the PracticalTeX 
2004 conference, and had clearly "caught the ConTeXt bug", and I suspect he 
may well have some useful practical experience in this area.

- Brooks