Assignments_ids with `()` in them fail in formgrader

[perllaghu]

Discovered when using notebook: 6.4.11 and nbgrader: 0.7.1 under linux

Tested & confirmed using notebook: 6.4.12 and nbgrader: 0.8.1 under linux

Expected behavior

When collecting notebooks known to be submitted, the # Submissions column should give a value, and a link to be able to grade submissions

Actual behavior

When collecting notebooks known to be submitted, the # Submissions column remains 0

Cause of the behavior

There are a number of places that use regexp to disassemble a path to find details of course-code, assignment-code, student-code, and/or notebooks.

There is a problem if any of those elements have characters in them that can be interpreted as a "regex special character" [. $ * + ? {} \ [] | () ... and various combinations involving the above.]

For example, if an assignment is called 123 (abc) - it will happily flow through the system, however formgrader will always count 0 students as NbGrader.api.get_submitted_students() has a regexp <snip>/submitted/(?P<student_id>.*)/123 (abc) ... and this matches 123 abc, with the abc in a MatchObject, *not 123 (abc)

Tested solution

Increase the number of characters tested for validation in assignment_id

• Edit server_extensions/formgrader/static/js/manage_assignments.js in the var createAssignment section (used in the formgrader UI)
• Edit coursedir.py in def _validate_assignment_id(.. (needed for the CLI commands)

I suggest the following characters will confuse the regexp system, and should be excluded: +()/\\[]{}$^

[perllaghu]

After coming across people who use braces and slashes in course codes.... I've changed a tack on this: convert relevant low ASCII characters into high-order equivalent Unicode characters.

This actually drops the clause that assignments can't have +.... we transform it to ➕ (Heavy Plus Sign U+2795)

so, in theory, abc (12/34*) + {not|really?}[^e$] becomes abc （12∕34★） ➕ ｛not❘really❔｝［＾e＄］ :)