I'm working on it.. :)
I had a converter that created MML from HTML, but I never got to do the nested
list conversion, since MML (and FrameMaker) didn't allow me to do nested lists
without nasty hacks. Anyway, I was only using to print html documents.
What I'm doing is to parse html into paragraph structures which consists of
string components. Each individual component can be anything, such as a
string, image, math string etc. Each string component inherits the string and
image context of the paragraph, but can overriden if necessary, eg. italicized
text.
In addition I would like to include paragraph groups, so that eg. items of a
list can belong to a specific group, whereas plain paragraphs belongs to the
default group. Text inside blockquote could belong to another group, allowing
it to be completely independent of the rest of the text. Groups can then be
hierarcial, and lists can be recursively indented.
Rendering hints can be embedded as components, eg. a hint that if the
paragraph were to be printed on a postscript printer, a linebreak and a
pagebreak would occur. It was the idea that hints could be associated with
different devices, eg. dps, ps and X. I was thinking about creating a
paragraph textsource and an associated paragraph oriented text widget,
allowing viewing, selection and editing. The first step would be to just print
paragraphs as postscript text.
Actually, html is just a external representation.
A good flexible system would allow overriding of the default paragraph styles
as defined in the html specs. Eg. that the address paragraph usually renders
as italicized and right justified, but for this particular page, it renders in
som other way.
So much for my own approach. I'm about to select and start on my master
thesis, so this project will probably be put on ice. If there is anyone
interested in participating in creating such a textsource and probably a
texteditor widget, drop me a line. The best thing would be if I could do my
master thesis associated with such work, but this isn't a very big
university..
> Many people perhaps, but not those who are fighting a rearguard action to
> keep page description features from creeping into SGML-based systems.
But event if you want complete sgml compliance in browsers, you will still
need formatting information. How would you do that if not specifying a default
sgml dtd with implicit rendering specification, such as html?
-- Torgeir @ http://www.ii.uib.no/~torgeir/