Re: [Exist-open] handling same path but multiple namespaces depending on data provider

[Joe W. <jo...@gm...>]

Hi Guillaume,

Short of normalizing the data, namespace wildcards are the most general
purpose solution for path expressions.

You can also use the local-name() function, e.g.,

  /*[local-name() eq "table"]/*[local-name() eq "tr"]/*[local-name() eq
"td"]

To really be sure you're not inadvertently including other namespaces, you
could declare all of the namespaces and explicitly select these in your
node tests with the union operator:


/(v12:table|v13:table|v14:table)/(v12:tr|v13:tr|v14:tr)/(v12:td/v13:td/v14:td)

It's hard to comment on performance concerns without knowing what kinds of
queries you need to run, but you might consider eXist 5's fields and facets
as part of your arsenal: you could construct a field with an expression
that selects the content you need, and then you could query the field
without worrying about which namespace the source came from. (I think.)

Joe

On Thu, Feb 4, 2021 at 11:26 AM Guillaume Mella <
gui...@un...> wrote:

> Hello,
>
>   I have a module which consumes VOTable (table/tr/td with meta data for
> Astronomy) using a standardised TableAccesProtocol. But depending on the
> data provider the returned document may get multiple namespaces depending
> on the version they implement, eg.:
>
> declare namespace votable="http://www.ivoa.net/xml/VOTable/v1.2"
> <http://www.ivoa.net/xml/VOTable/v1.2>; (: could be v1.3 v1.4 :)
>
> Most data part are the same in the various versions and I would like to
> support it dynamically without branching/duplication/copy.
>
> I tried with the wildcard namespace using *:tag but didn't investigate to
> know more about performance issues.
>
> Is there any best practices, hints or documentation pointers for such use
> case ?
>
>
> Cheers,
>
> --
> Guillaume Mella - Ingénieur CNRS
> OSUG UMS832 :  http://www.osug.fr  -  JMMC : http://www.jmmc.fr/~mellag
> OSUG-D, 122 rue de la piscine, 38 400 Saint Martin d’Hères, France
> ☎ +33 (0)4 76 63 58 22
>
> _______________________________________________
> Exist-open mailing list
> Exi...@li...
> https://lists.sourceforge.net/lists/listinfo/exist-open
>