[R] Correct way to handle class

[Rui Barradas]

Hello,

Write a function to set the new class attribute. If other changes need 
to be done, you can do them all in one function. For instance, the 
as.matrix method for atomic vectors coerces a vector to class "matrix" 
*and* gives it a dim attribute. The two changes are done in one call.

In the example below, there's a setter function and a print method for 
the new class. It can handle both atomic vectors and with NextMethod 
complicated objects such as data.frames.


as_ECFOCF <- function(x, ...) {
   if(!inherits(x, "ECFOCF"))
     class(x) <- c("ECFOCF", class(x))
   x
}
print.ECFOCF <- function(x, ...) {
   if(is.atomic(x))
     print.default(unclass(x))
   else
     NextMethod()
}

result2 <- "Je suis donc je pense"
resultL2 <- data.frame(A=c(1,2), B=c(3, 4))

result2 <- as_ECFOCF(result2)
class(result2)
#> [1] "ECFOCF"    "character"
result2
#> [1] "Je suis donc je pense"

resultL2 <- as_ECFOCF(resultL2)
class(resultL2)
#> [1] "ECFOCF"     "data.frame"
resultL2
#>   A B
#> 1 1 3
#> 2 2 4


Hope this helps,

Rui Barradas

Às 11:34 de 14/04/2022, Marc Girondot via R-help escreveu:
> Dear member of the list,
> 
> I search for the correct way to handle class.
> 
> I learn recently that inherits() was the correct way to test class:
> 
> inherits(x, what, which = FALSE)
> 
> For this part it is ok.
> 
> But now I have questions about the correct procedure to set class.
> 
> Previously I used:
> 
> result <- "Je suis donc je pense"
> class(result) <- "ECFOCF"
> 
> But I loose the previous class "character". It is still present in 
> mode(result) but sometimes it can be a problem.
> 
> For example here:
> 
>  > resultL <- data.frame(A=c(1,2), B=c(3, 4))
>  > class(resultL)
> [1] "data.frame"
>  > class(resultL) <- "ECFOCF"
>  > resultL
> $A
> [1] 1 2
> 
> $B
> [1] 3 4
> 
> attr(,"class")
> [1] "ECFOCF"
> attr(,"row.names")
> [1] 1 2
> 
> I lost the data.frame structure when I set the new class. It is seen as 
> a list by:
> 
>  > mode(resultL)
> [1] "list"
> 
> So I use:
> 
> result <- "Je suis donc je pense"
> class(result) <- unique(append("ECFOCF", class(result)))
> 
> I use append because I don't want loose the original class. I use 
> unique() to prevent the same class being put several times in the object.
> 
> However, Using this formula, I am not 100% sure the "ECFOCF" is the 
> first class. If it is not, it could prevent plot.ECFOCF to take in 
> charge the plot of this object.
> 
> An alternative could be:
> 
>  > result <- "Je suis donc je pense"
>  > class(result) <- unique(append("ECFOCF", class(result)[class(result) 
> != "ECFOCF"]))
>  > class(result)
> [1] "ECFOCF"    "character"
> 
> Then I am sure that "ECFOCF" class is the first one.
> 
> Is it the correct way to define class or I am in wrong direction?
> 
> Thanks a lot
> 
> Marc
> 
> ______________________________________________
> R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide 
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.