Re: [xsl] two regexp related questions

[Julian Reschke <julian.reschke@xxxxxx>]

On 2011-05-19 23:36, Michael Kay wrote:
> On 19/05/2011 18:45, Julian Reschke wrote:
> > Hi there,
> >
> > I've got two regexp-related questions.
> >
> > 1) Is it correct that XSLT/XPath2's regular expressions do not support
> > non-capturing groups (as shown in
> > <http://www.exampledepot.com/egs/java.util.regex/NoGroup.html>)?
>
> Yes. This is being added in 3.0.
> > 2) With respect to analyze-string, and the captured regex-groups:
> >
> > I'm using a regex like
> >
> > ([A-Z]+) = ([A-Z]+) ( ; ([A-Z]+) = ([A-Z]+) )*
> >
> > for matching things like
> >
> > a=b;c=d;e=f
>
> Why not use the regex
>
> ([A-Z]+) = ([A-Z]+)
>
> and rely on xsl:matching-substring being called multiple times?

My bad. I simplified the example too much; the first segment has a 
(slightly) different syntax.

> Alternatively, first use tokenize() to split on the semicolons, and then
> put each token through xsl:analyze-string.

See above; the separator character can occur in quoted values; I 
simplified too much.

> Personally, I find long regexes very difficult to get right, and try to
> split the task up into small pieces that I can understand and debug.

Indeed.

I'm now assembling them piece by piece from smaller sub-expressions, 
each of which matching a part of the grammar I'm trying to parse.

Thanks for the feedback, Julian