Type check allows use of types to automatically add checking code when types are annotated.
type_define()
is used to define a new type. The check
argument specifies a function used to verify the objects type. type_check
adds the checks to a specific function.
type.numeric <- type_define(check = is.numeric)
f <- type_check(function(x = ? numeric) x)
f(1)
#> [1] 1
f("txt")
#> Error: `x` is a `character` not a `numeric`.
Types are defined as methods of the type
generic. This means they follow the same properties as normal S3 methods and can be exported and imported to and from packages like all other functions.
The error
argument is used to specify a custom error message for a type.
type.numeric <- type_define(
check = is.numeric,
error = function(obj_name, obj_value, type) {
sprintf("%s: '%s' is not a number!", obj_name, obj_value)
})
f <- type_check(function(x = ? numeric) x)
f("txt")
#> Error: x: 'txt' is not a number!
When writing a package adding a call to typeCheck::type_check_package()
anywhere outside a function will add type checks to all functions in the package. Functions without type annotations are unaltered.
This means it is easy to add annotations in a stepwise process to existing packages.
If you are using roxygen2 You can use the following importFrom
statement (or use the equivalent importFrom()
call directly in the NAMESPACE
file.)
f <- function(x = ? numeric) x
#' @importFrom typeCheck type type_define
typeCheck::type_check_package()