«

[Warning, XSLT noob.]

In the XML source I'm working with, some node values that should be
identical, vary slightly from value to value. The differences are in
whitespace (some have a carriage return mid-string).

Ideally, I'd like to normalize-space() all of the values of the
underlying data, so that I later use and reuse the cleaned data
throughout the rest of the transformation (so I could properly use
distinct-values() on that list, etc.).

Is there a way to modify the underlying data, as opposed to having to
clean it each time you come to it?

Thanks,
Jamie

Jamie Jackson wrote:
> Is there a way to modify the underlying data, as opposed to having to
> clean it each time you come to it?

You can't alter the source document -- expect by taking a two-pass
approach, one pass to create the altered source and then a second pass
to style from that. The two passes can be done in a single stylesheet,
with some work, if you're either using XSLT2 or are willing to use the
EXSLT node-set() function -- build the altered tree in a variable, then
style from that -- but of course this doubles your memory use and means
touching everything twice, so it may not be a net win.

Joe Kesselman wrote:
> Jamie Jackson wrote:
> > Is there a way to modify the underlying data, as opposed to having to
> > clean it each time you come to it?
>
> You can't alter the source document -- expect by taking a two-pass
> approach, one pass to create the altered source and then a second pass
> to style from that. The two passes can be done in a single stylesheet,
> with some work, if you're either using XSLT2 or are willing to use the
> EXSLT node-set() function -- build the altered tree in a variable, then
> style from that -- but of course this doubles your memory use and means
> touching everything twice, so it may not be a net win.

Okay, it's probably not worth the trouble, but good to know, thanks.

I've come up with a decent, simple workaround for my current issue, so
I'm set for today.

Thanks,
Jamie