In Data-Camp/testwhat: Easily write submission correctness tests for R exercises

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  eval = FALSE
)

Some exercises require a student to hand-craft a function to make their code DRY. With check_fun_def() and a range of related helper check_ functions you can check the correctness of these user-defined functions.

As an example, have a look at the following solution:

# Define my_fun
my_fun <- function(a, b) {
  abs(a) + abs(b)
}

The following SCT checks the function definition and whether the function was properly called:

ex() %>% check_fun_def("my_fun") %>% {
  check_arguments(.)
  check_call(., 1, 2) %>% check_result() %>% check_equal()
  check_call(., -1, 2) %>% check_result() %>% check_equal()
  check_call(., 1, -2) %>% check_result() %>% check_equal()
  check_call(., -1, -2) %>% check_result() %>% check_equal()
  check_body(.) %>% {
    check_function(., "abs", index = 1)
    check_function(., "abs", index = 2)
  }
}

• check_fun_def checks whether my_fun was defined by the student. If this is the case, it will produce a substate that contains a reference to the function arguments and its body.
• check_arguments() checks whether the student specified the same number of arguments as the function definition in the solution did.
• check_call() will create an R expression that calls the function my_fun() with the arguments specified in it. As an example, check_call(., 1, 2) will generate the expression my_fun(1, 2). check_result() will then execute this expression (in both student and solution workspace), and focus on the result of the expression. Finally, check_equal() verifies whether these call results correspond.
• check_body() zooms in on the body of the function definition (as if the body of the function was the only code in the student submission). If, for example, the student used abs() outside the function definition, this would be ignored. Notice here that no arguments are specified. If we did this, check_function() would compare the actual values of the arguments. This is information that is not available, as a and b are not variables that are available in the workspace, only in the scope of the function.

Note: in addition to check_result(), you can also pipe the result of check_call() into check_output() and check_error(), and follow up with check_equal(). These functions will respectively check the output and error the function call generated when executed in the student/solution environment.

Data-Camp/testwhat documentation built on June 24, 2022, 9:59 p.m.