Re: replacing a node in in-memory XML

[Thomas Lord]

Another follow-up for Wolfgang's good tip:

Think carefully about how many copies of a given
element your input might have, and how many copies
of a given output you want it to have.

Consider these three variations (for example):

    element {QName ($n)}
    {
      $n/@*,
      for $c in $n/node()
      typeswitch ($c)
      case element foo return <my-foo-replacement/>
      default return $c
    }

and

    element {QName ($n)}
    {
      $n/@*,
      for $c in $n/node()
      typeswitch ($c)
      case element foo return ()
      default return $c,
      <my-foo-replacement/>
    }
and

    element {QName ($n)}
    {
      $n/@*,
      for $c in $n/node()
      typeswitch ($c)
      case element foo return ()
      default return $c,
      if ($n/foo) then <my-foo-replacement/> else ()
    }

You can score a lot of good "robustness karma points" if
you make the right choices among variations like those.
(And, you can make distinctions like that in XSLT just fine
but I'm not sure that code comes out clearer than
the above examples!)

Also, don't forget to copy attributes (as illustrated)
when appropriate!  (And, you can do the same trick to
edit attributes, btw.)

-t


Hans-Juergen Rennau wrote:
> Hello Wolfgang,
>
> let me add a remark about the situation when the nodes to be modified cannot be identified by type *ALONE*. For example, the document might contain many community groups, and the services element is to be replaced only in the groups (2, 31, 32, 40). In such situations, I suggest to combine the "typeswitch recursion pattern" with a "node marking pattern": a) before starting the recursion, "mark" the modification targets by collecting them in a variable, b) within the relevant typeswitch branch use a node identity-based if/else. This approach scales rather well when things get more complicated (several element types are to be partially and heterogeneously modified).
>
> a) 
> declare variable $targetNodes_services := 
>    ($doc//communitygroup)[position() = (2, 31, 32, 40)]/group/services
>
> b)
> typeswitch ($node)
>    case $elem as element(services) return
>       if ($elem intersect $targetNodes_services) then <services>...</services>
>       else
>
> 		         element {node-name($elem)} {$elem/@*, $elem/node()/local:replace(.)}
>     ...
>
> The approach helped me when I had to modify WordML documents, highly repetitive as they are in terms of element names!
>
> With kind regards,
> Hans-Juergen
>
>
>
>
>
>        __________________________________ Ihr erstes Fernweh? Wo gibt es den schönsten Strand? www.yahoo.de/clever
>
> _______________________________________________
> http://x-query.com/mailman/listinfo/talk
> http://x-query.com/mailman/listinfo/talk
>