Re: [basex-talk] DITA project
Christian, I love the contains token! Didn't know about that.
The reason I'm looking for a switch instead of the if then else is more for
readability, there are a lot of elements and classes in DITA, when you're
indenting to character 2xx it starts to get hard to manage! Maybe I'll bug
people on the xquery side with this case. DITA is popular enough that it
should matter.
Mark, good point, values for @ class in DITA start and end with spaces, and
you always match on space-value-space. I omitted the spaces to focus on my
current issue, but it looks like you have a bionic eye for finding
potential issues :-P.
Thank you for your feedback on this!
On Fri, Mar 13, 2015 at 4:52 AM, Christian Grün christian.gr...@gmail.com
wrote:
A little addition: XQuery 3.1 provides the function contains-token to
request single values of attributes. It basically does what Marc
proposed [1]:
let $input := a class='ancestor field'/
let $class := $input/@class
return if(contains-token($class, 'ancestor')) then (
1
) else if(contains-token($class, 'descendant')) then (
2
) else (
0
)
This is one more way to do it:
let $input := a class='ancestor field'/
let $tokens := map:merge(
tokenize($input/@class, '\s+') ! map { .: true() }
)
return if($tokens?ancestor) then (
1
) else if($tokens?descendant) then (
2
) else (
0
)
Best,
Christian
[1] http://docs.basex.org/wiki/XQuery_3.1#fn:contains-token
On Thu, Mar 12, 2015 at 10:45 PM, Marc van Grootel
marc.van.groo...@gmail.com wrote:
Hi France,
Typeswitch is not the right tool for this.
From the spec:
[74] TypeswitchExpr ::= typeswitch ( Expr ) CaseClause+
default ($ VarName)? return ExprSingle
[75] CaseClause ::= case ($ VarName as)? SequenceTypeUnion
return ExprSingle
[76] SequenceTypeUnion ::= SequenceType (| SequenceType)*
As the name implies it is meant for checking (sequence) types and not
for arbitrary XPath expressions.
Also, using contains isn't watertight as contains('foobar','foo') would
match as well. I would probably use a function like this
declare function in-class($node as element(), $class as xs:string) as
xs:boolean {
$class = tokenize($node/@class,'\s+')
}
Do check performance in your situation. In case you need to do many
checks on the same class attribute you may want to bind the tokenized value
list with a let instead of using this function.
I also remember that Michael Kay is looking into improving on exactly
this use case. But that doesn't help you now.
Cheers,
--Marc
On 12 mrt. 2015, at 20:33, France Baril france.ba...@architextus.com
wrote:
Hi,
I'm working on a new DITA project. The DITA standard is used for
technical documentation. It creates XML models with inheritance by using
@class=ancestor parent child. If the child should behave like it's
ancestor the XSL would say:
template match=contains(@class, 'ancestor') ... /template
The advantage is that the model can evolve and when new elements are
added, you only need to code transformations for the differences.
I am trying to figure out if I can use type switching with contains in
attribute. Search gets me no syntax for something like this:
typeswitch ($node)
case attribute(contains(class, 'ancestor'))
Maybe I should register to the xquery group to get an answer, but since
I'm already here, I though I should ask, and maybe there is a BaseX
specific option. I mean other that an unmanageable number of 'if then else'
statements.
Thanks!
--
France Baril
Architecte documentaire / Documentation architect
france.ba...@architextus.com
--
France Baril
Architecte documentaire / Documentation architect
france.ba...@architextus.com
Re: [basex-talk] DITA project
Hi France, If you really need multi-branching if then else the language doesnt require increasing indentation. if (cond1) then expr1 else if (cond2) then expr2 else expr3 In my opinion this is just as good as using switch, i've seen examples like this even in the spec. Of course readability suffers if there's a lot going on in the expressions such as complete flwor expressions but that's just a matter of breaking up the code in smaller (re-usable) pieces/functions. Breaking up / refactoring your logic to avoid massive if then else or switch constructs is advisable anyway. But I don't know how your code looks so I cannot say. hth --Marc
Re: [basex-talk] DITA project
Hmmm! Sometimes, we get stuck in the simplest things. I see you can actually follow the pattern pretty well without indent. I tend to do everything in small functions the only time I get more then a function call is when I need different calls based on ancestors, and that's also pretty short. Thank you Mark. On Fri, Mar 13, 2015 at 12:01 PM, Marc van Grootel marc.van.groo...@gmail.com wrote: Hi France, If you really need multi-branching if then else the language doesnt require increasing indentation. if (cond1) then expr1 else if (cond2) then expr2 else expr3 In my opinion this is just as good as using switch, i've seen examples like this even in the spec. Of course readability suffers if there's a lot going on in the expressions such as complete flwor expressions but that's just a matter of breaking up the code in smaller (re-usable) pieces/functions. Breaking up / refactoring your logic to avoid massive if then else or switch constructs is advisable anyway. But I don't know how your code looks so I cannot say. hth --Marc -- France Baril Architecte documentaire / Documentation architect france.ba...@architextus.com
Re: [basex-talk] DITA project
Hi France,
Typeswitch is not the right tool for this.
From the spec:
[74]TypeswitchExpr ::= typeswitch ( Expr ) CaseClause+
default ($ VarName)? return ExprSingle
[75]CaseClause ::= case ($ VarName as)?
SequenceTypeUnion return ExprSingle
[76]SequenceTypeUnion ::= SequenceType (| SequenceType)*
As the name implies it is meant for checking (sequence) types and not for
arbitrary XPath expressions.
Also, using contains isn't watertight as contains('foobar','foo') would match
as well. I would probably use a function like this
declare function in-class($node as element(), $class as xs:string) as
xs:boolean {
$class = tokenize($node/@class,'\s+')
}
Do check performance in your situation. In case you need to do many checks on
the same class attribute you may want to bind the tokenized value list with a
let instead of using this function.
I also remember that Michael Kay is looking into improving on exactly this use
case. But that doesn't help you now.
Cheers,
--Marc
On 12 mrt. 2015, at 20:33, France Baril france.ba...@architextus.com wrote:
Hi,
I'm working on a new DITA project. The DITA standard is used for technical
documentation. It creates XML models with inheritance by using
@class=ancestor parent child. If the child should behave like it's ancestor
the XSL would say:
template match=contains(@class, 'ancestor') ... /template
The advantage is that the model can evolve and when new elements are added,
you only need to code transformations for the differences.
I am trying to figure out if I can use type switching with contains in
attribute. Search gets me no syntax for something like this:
typeswitch ($node)
case attribute(contains(class, 'ancestor'))
Maybe I should register to the xquery group to get an answer, but since I'm
already here, I though I should ask, and maybe there is a BaseX specific
option. I mean other that an unmanageable number of 'if then else' statements.
Thanks!
--
France Baril
Architecte documentaire / Documentation architect
france.ba...@architextus.com
