Title: | Utility Functions for 'PrairieLearn' and R |
---|---|
Description: | 'PrairieLearn' is an online problem-driven learning system for creating homeworks and tests. This package adds some helper functions for using it along with R as we are currently doing for <https://stat430.com>. |
Authors: | Dirk Eddelbuettel and Alton Barbehenn |
Maintainer: | Dirk Eddelbuettel <[email protected]> |
License: | GPL (>= 2) |
Version: | 0.0.2.3 |
Built: | 2024-08-11 04:50:04 UTC |
Source: | https://github.com/stat447/plr |
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.
get_question_details(dir, pattern = "^test.*\\.[rR]$")
get_question_details(dir, pattern = "^test.*\\.[rR]$")
dir |
Directory containing the test files for a question |
pattern |
A regular expression identifying test files in the directory |
A data.frame object with colums name, file, and max_points
Helper function to format result object returned to PL
message_to_test_result(msg, max_pts = 100)
message_to_test_result(msg, max_pts = 100)
msg |
Character variable with the error or warning received |
max_pts |
Optional numeric variable with maximal attainable points, usuall 100 |
A data.frame object with four elements as expected by PL
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.
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)
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)
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 |
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 |
path |
Optional path to a file that should be hidden before evaluation happens. It is then unhidden on exit. |
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.
A value of the expr
sourced from file
and evaluated by
uid
, or NULL in case of error.
## Not run: n <- sample(3:20, 1) # random payload res <- source_and_eval_safe("code/fib.R", fib(n), "ag") ## End(Not run)
## Not run: n <- sample(3:20, 1) # random payload res <- source_and_eval_safe("code/fib.R", fib(n), "ag") ## End(Not run)
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.
test_question(tests_dir = "/grade/tests/tests", results_file = "results.json")
test_question(tests_dir = "/grade/tests/tests", results_file = "results.json")
tests_dir |
Directory containing the test files for a question |
results_file |
JSON file into which results are written |
The results data.frame is returned, but the functions is invoked for its side-effect of creating the JSON file