[xml] parser/tokenizer [was Re: support for xml schema in libxml2]

[Frans Englich <frans englich telia com>]

On Tuesday 04 July 2006 16:32, Buchcik, Kasimier wrote:
> Hi,
>
> > -----Original Message-----
> > From: xml-bounces gnome org [mailto:xml-bounces gnome org] On
> > Behalf Of boss gregerhaga net
>
> [...]
>
> > I'd like to help. Quite honestly:I have very little knowledge of the
> > big picture about XML, XSLT and related technologies. I have used
> > libxml2 recently, see my site
> > http://www.gregerhaga.net/qxrss-1.2.6-dox for my rss reader, which
> > uses it ( perhaps in a stupid way but still*grin*).
> >
> > So:what can i do to contribute? What is the learning curve
> > like for an almost newbie in XML but with experience from C?
> >
> > Greger
>
> I'll list the things I would do, when I would be about to
> put the rope around my neck - as you do (hmm, we _are_ scary):
>
> 1) A good starter for XML stuff is http://www.w3schools.com/
>   Besides other things, it teaches (rudimentally) about XSLT 2.0,
>   XQuery 1.0 and XPath 2.0.
>
> 2) Google around for some deeper tutorials, articles, etc.; but don't
>   get lost, since it might become boring.
>
> 3) Read the specs: http://www.w3.org/Style/XSL/ ; but again, don't get
>   lost, since it might become boring and we'll never see you again.
>
> 4) Get in touch with Libxml2's XPath code. Write an XPath test app and
>   debugg the parser and evaluation code, while switching back and
>   fro to the specs (XPath 1.0 in this case) and trying to understand
>   why the code does what it does. Write down the obervations, new ideas,
>   things you would have done differently, etc. Maybe, just for learning,
>   also add a piece of self-invented syntax to the XPath machinery, in
>   order to get a feeling how to change/extend things without breaking
> them.
>
> 5) I think one will need to create a scenario for oneself, where
>   one actually _needs_ to implement XPath 2.0/XQuery 1.0/XSLT 2.0.
>   This can either be the fact that you already told all your
> girl-friends
>   that you're gonna do it (this might be not enough a reason), some
>   idealistic reasons, some business-related issues, etc.
>
> 6) Wear heavy chains for about 2 weeks, then buy some t-shirts with
>   XPath 2.0/XQuery 1.0/XSLT 2.0 written on it in fancy letters.
>
> 7) Try to annoy people with questions, bugzilla-entries, etc.
>    See http://www.xmlsoft.org/bugs.html
>
> But I don't know how to actually start writing such new code. Daniel
> will have to tell us (or at least me) what he expects; I would need to
> know to what extent new code for XPath 2.0 can be written and what
> existing code should be rather extended; so questions like:
> - Should the existing expression-parsing code be extended (is this
>   possible?), or should we write on new one for XPath 2.0?

One could consider Bison combined with a hand written tokenizer(that's how 
Patternist do it). Here's an XQuery grammar; path expressions and (some) XML 
literals are missing, but other than that it's quite complete:

http://websvn.kde.org/trunk/kdenonbeta/kdom/patternist/parser/QueryTransformParser.ypp?rev=557332&view=log

The tokenizer is in the same dir, but I doubt it would be useful.

[...]
> Finally, I think that there's no need not know much about
> XML-related technology to be able to code much stuff for e.g. the
> XPath 2.0 module.

Grasping the specs can be daunting. It was that for me. And is..


Cheers,

                Frans