Package 'plr'

Extract question name and score from header

Description

This function is inspired by the roxygen2 decoration of source files with content used to create the manual and help files. Here we expect two tags @title with the displayed title of the question, and @score with the number of available points.

Usage

get_question_details(dir, pattern = "^test.*\\.[rR]$")

Arguments

dir	Directory containing the test files for a question
pattern	A regular expression identifying test files in the directory

Value

A data.frame object with colums name, file, and max_points

---

Helper function to format result object returned to PL

Description

Helper function to format result object returned to PL

Usage

message_to_test_result(msg, max_pts = 100)

Arguments

msg	Character variable with the error or warning received
max_pts	Optional numeric variable with maximal attainable points, usuall 100

Value

A data.frame object with four elements as expected by PL

---

Wrapper to source a file and safely evaluate an expression

Description

We assume all files surrounding the to be evaluated files have different user ids and file modes not allowing the supplied user id to read them. One way to do that is to just set all files within the evaluation directories to root:root removing group and others the rights to read (or write or execute). We therefore also chmod the supplied file back to mode “0644” ensuring that the file can be read so that the expression can be evaluated—but nothing else should be in reach.

Usage

source_and_eval_safe(file, expr, uid = NULL)

eval_safe_as(expr, uid = NULL)

source_and_eval_safe_with_hiding(file, expr, uid = NULL, path = NULL)

Arguments

file	A filename with an R file to be source, typically containing the student code to be evaluated safely.
expr	An expression to be evaluate by eval_safe, typically the name of the sane of the function containing the student code plus the argument supplied from the test runner.
uid	Optional numeric or character user id identifying the user id with (presumably lower) privileges as which the code is running; the numeric uid is obtained via user_info is a character is supplied. Note that using this argument requires being the ‘root’ user.
path	Optional path to a file that should be hidden before evaluation happens. It is then unhidden on exit.

Details

The source_and_eval_safe_with_hiding variant can hide a given file, for example containing a reference answer, but assigning it to a unique temporary name so that it cannot be sourced.

The eval_safe_as convenience function fetches the (numeric) user id before calling unix::eval_safe; it is equivalent to source_and_eval_safe but does not involve a file.

Note that you must run these functions as the ‘root’ user in order to set the uid.

Value

A value of the expr sourced from file and evaluated by uid, or NULL in case of error.

Examples

## Not run: 
n <- sample(3:20, 1)         # random payload
res <- source_and_eval_safe("code/fib.R", fib(n), "ag")

## End(Not run)

---

Run a whole question and report aggregate results

Description

This function is the equivalent of the pltest.R script which, given a directory runs the tests file therein and reports the results in a JSON file for PrairieLearn to consume.

Usage

test_question(tests_dir = "/grade/tests/tests",
  results_file = "results.json")

Arguments

tests_dir	Directory containing the test files for a question
results_file	JSON file into which results are written

Value

The results data.frame is returned, but the functions is invoked for its side-effect of creating the JSON file

---