R help - Oct 2009 - inverse currying

Dear list,

I have the following function,

sugar = function(fun, id = "id"){
  ff <- formals(fun)
  if( id %in% names(ff))
    stop("id is part of args(fun)")
  formals(fun) <- c(unlist(ff), alist(id=))
  fun
}

which one may use on a function foo,

foo = function(x){
  x
}

sugar(foo) # results in the extended closure,

function (x, id)
{
    x
}

Its limitation (other than not working with .Primitives) is the 'id'
tag that I add in the formals of fun(). I don't know how to create a
alist(id=) pairlist where id can be changed. I tried the usual bquote
and substitute approach but they don't seem to work here. I suppose I
could do something like,

parse(text = paste("alist(",id, "=)", sep=""))

but this is usually not recommended.

Any ideas?

Best regards,

baptiste

After some more digging (grep "alist" R-devel/ ), I've come up
with this,

tools:::as.alist.symbol("x")

sugar = function(fun, id = "id"){
 ff <- formals(fun)
 if( id %in% names(ff))
   stop(paste(id, "is part of args(fun)"))

 new.arg <- tools:::as.alist.symbol(id)

 formals(fun) <- c(unlist(ff), new.arg)
 fun

}

foo = function(x, a=1){
 x
}

sugar(foo)
sugar(foo, 'a')
sugar(sugar(foo))

sugar(foo, 'my.new.arg')


Best,

baptiste



2009/10/1 baptiste auguie <baptiste.auguie at
googlemail.com>:
> Dear list,
>
> I have the following function,
>
> sugar = function(fun, id = "id"){
> ?ff <- formals(fun)
> ?if( id %in% names(ff))
> ? ?stop("id is part of args(fun)")
> ?formals(fun) <- c(unlist(ff), alist(id=))
> ?fun
> }
>
> which one may use on a function foo,
>
> foo = function(x){
> ?x
> }
>
> sugar(foo) # results in the extended closure,
>
> function (x, id)
> {
> ? ?x
> }
>
> Its limitation (other than not working with .Primitives) is the
'id'
> tag that I add in the formals of fun(). I don't know how to create a
> alist(id=) pairlist where id can be changed. I tried the usual bquote
> and substitute approach but they don't seem to work here. I suppose I
> could do something like,
>
> parse(text = paste("alist(",id, "=)",
sep=""))
>
> but this is usually not recommended.
>
> Any ideas?
>
> Best regards,
>
> baptiste
>