On 7/13/06, Volker Haas
<volker.haas(a)brainbot.com> wrote:
Hi Robert
Robert Scott Horning wrote:
I am curious how you plan to compile these books
in a format that looks
good on paper and not strictly as a web format. While this can be
automated to an extent, I do think there are some issues that come from
trying to move content from a web format to a printed page, and not all
of these can be completely automated.
As you have pointed out, an automatic conversion of html (or mediawiki
markup) into latex can not be automated in a way that the book is
absolutely perfect.
But right now we are pretty satisfied with the results - even though we
are still working on improvments of the conversion.
The conversion is done with a parser we developed from scratch - as
mentioned in a previous post. The mediawiki markup is translated
into an internal representation which then gets transformed to latex (to
be more precise "context") - this is the hard part. It is slightly
easier to transform the internal representation back to html since some
css-style information can be maintained.
As I've mentioned before, I'd love to see a Wikimedia project which
does exactly this. Then the latex could be edited collaboratively to
make things more "absolutely perfect". It's nice to see it's at least
somewhat possible, though I'd say the quality of the previews right
now is fairly low.
My wiki2xml converter is in the subversion repository. It is a set of
scripts to convert MediaWiki markup into XML, and from there into other
formats, including plain text, HTML, DocBook, and ODT
(OpenDocument/OpenOffice format). OpenOffice is also open source and can
generate PDFs natively.