unitgrade-devel 0.1.1

A set of tools to develop unitgrade reports and evaluate them

Unitgrade-devel

Note: This is the development version of unitgrade. If you are a student, please see http://gitlab.compute.dtu.dk/tuhe/unitgrade.

Unitgrade is an automatic report and exam evaluation framework that enables instructors to offer automatically evaluated programming assignments. Unitgrade is build on pythons unittest framework so that the tests can be specified in a familiar syntax and will integrate with any modern IDE. What it offers beyond unittest is the ability to collect tests in reports (for automatic evaluation) and an easy and 100% safe mechanism for verifying the students results and creating additional, hidden tests. A powerful cache system allows instructors to automatically create test-answers based on a working solution.

• 100% Python unittest compatible
• No external configuration files: Just write a unittest
• No unnatural limitations: Use any package or framework. If you can unittest it, it works.
• Tests are quick to run and will integrate with your IDE
• Cache and hint-system makes tests easy to develop
• Granular security model:
  • Students get public unittests for easy development of solutions
  • Students use a tamper-resistant file to create submissions which are uploaded
  • Instructors can automatically verify the students solution using a Docker VM and run hidden tests
• Automatic Moss anti-plagiarism detection
• CMU Autolab integration (Experimental)

Using unitgrade

The examples can be found in the /examples/ directory: https://gitlab.compute.dtu.dk/tuhe/unitgrade_private/examples

A simple example

Unitgrade makes the following assumptions:

• Your code is in python
• Whatever you want to do can be specified as a unittest

Although not required, it is recommended you maintain two version of the code:

• A fully-working version (i.e. all tests pass)
• A public version distributed to students (some code removed))

I use codesnipper (see http://gitlab.compute.dtu.dk/tuhe/snipper) to synchronize the two versions automatically.
Let's look at an example. Suppose our course is called cs101, in which case we make three files in our private folder instructor:

instructor/cs101/homework.py # This contains the students homework
instructor/cs101/report1.py  # This contains the tests
instructor/cs101/deploy.py   # A private file to deploy the tests

The homework

The homework is just any old python code you would give to the students. For instance:

def reverse_list(mylist): #!f
    """
    Given a list 'mylist' returns a list consisting of the same elements in reverse order. E.g.
    reverse_list([1,2,3]) should return [3,2,1] (as a list).
    """
    return list(reversed(mylist))

def add(a,b): #!f
    """ Given two numbers `a` and `b` this function should simply return their sum:
    > add(a,b) = a+b """
    return a+b

if __name__ == "__main__":
    # Problem 1: Write a function which add two numbers
    print(f"Your result of 2 + 2 = {add(2,2)}")
    print(f"Reversing a small list", reverse_list([2,3,5,7]))

The test:

The test consists of individual problems and a report-class. The tests themselves are just regular Unittest (we will see a slightly smarter idea in a moment). For instance:

from looping import reverse_list, add
import unittest


class Week1(unittest.TestCase):
    def test_add(self):
        self.assertEqual(add(2, 2), 4)
        self.assertEqual(add(-100, 5), -95)

    def test_reverse(self):
        self.assertEqual(reverse_list([1, 2, 3]), [3, 2, 1])

A number of tests can be collected into a Report, which will allow us to assign points to the tests and use the more advanced features of the framework later. A complete, minimal example:

from src.unitgrade2.unitgrade2 import Report
from src.unitgrade2 import evaluate_report_student
from looping import reverse_list, add
import unittest


class Week1(unittest.TestCase):
    def test_add(self):
        self.assertEqual(add(2, 2), 4)
        self.assertEqual(add(-100, 5), -95)

    def test_reverse(self):
        self.assertEqual(reverse_list([1, 2, 3]), [3, 2, 1])


import cs101


class Report1(Report):
    title = "CS 101 Report 1"
    questions = [(Week1, 10)]  # Include a single question for 10 credits.
    pack_imports = [cs101]


if __name__ == "__main__":
    # Uncomment to simply run everything as a unittest:
    # unittest.main(verbosity=2)
    evaluate_report_student(Report1())

Deployment

The above is all you need if you simply want to use the framework as a self-check: Students can run the code and see how well they did. In order to begin using the framework for evaluation we need to create a bit more structure. We do that by deploying the report class as follows:

from report1 import Report1
from unitgrade_private2.hidden_create_files import setup_grade_file_report
from snipper import snip_dir
import shutil

if __name__ == "__main__":
    setup_grade_file_report(Report1, minify=False, obfuscate=False, execute=False)

    # Deploy the files using snipper: https://gitlab.compute.dtu.dk/tuhe/snipper
    snip_dir.snip_dir(source_dir="../programs", dest_dir="../../students/programs", clean_destination_dir=True, exclude=['__pycache__', '*.token', 'deploy.py'])

• The first line creates the report1_grade.py script and any additional data files needed by the tests (none in this case)
• The second line set up the students directory (remember, we have included the solutions!) and remove the students solutions. You can check the results in the students folder.

Using the framework as a student

You can now upload the student directory to the students. The students can run their tests either by running cs101.report1 in their IDE or by typing:

python -m cs101.report1

in the command line. This produces a detailed output of the test and the program is 100% compatible with a debugger. When the students are happy with their output they can run (using command line or IDE):

python -m cs101.report1_grade

This runs an identical set of tests, but produces a .token file the students can upload to get credit.

• The report1_grade.py includes all tests and the main parts of the framework and is obfuscated by default. You can apply a much strong level of protection by using e.g. pyarmor.
• The report1_token.token file includes the outcome of the tests, the time taken, and all python source code in the package. In other words, the file can be used for manual grading, for plagirism detection and for detecting tampering.
• You can easily use the framework to include output of functions.
• See below for how to validate the students results

How safe is Unitgrade?

Cheating within the framework is probably best done by manually editing the .token-file or by creating a broken set of tests. This involves risk of being trivially detected, for instance because tests have the wrong runtime, but more importantly the framework automatically pack all the used source code and so if a student is cheating, there is no way to hide it for an instructor who looks at the results. If the program is used in conjunction with automatic plagiarism software, cheating therefore involves both breaking the framework, and creating 'false' solutions which statistically match other students solutions, and then hope nobody bothers to check the output. The bottom line is that I think plain old plagiarism is a much more significant risk, and one the framework reduces relative to other project work by demanding the source code is included.

If this is not enough you have two options: You can either use pyarmor to create a very difficult challenge for a prospective hacker, or you can simply validate the students results as shown below.

Example 2: The framework

One of the main advantages of unitgrade over web-based autograders it that tests are really easy to develop and maintain. To take advantage of this, we simply change the class the questions inherit from to UTestCase (this is still a unittest.TestCase) and we can make use of the chache system. As an example:

class Week1(UTestCase):
    """ The first question for week 1. """
    def test_add(self):
        from cs102.homework1 import add
        self.assertEqualC(add(2,2))
        self.assertEqualC(add(-100, 5))

    def test_reverse(self):
        from cs102.homework1 import reverse_list
        """ Reverse a list """ # Add a title to the test.
        self.assertEqualC(reverse_list([1,2,3]))

Note we have changed the test-function to self.assertEqualC (the C is for cache) and dropped the expected result. What unitgrade will do is to evaluate the test on the working version of the code, compute the results of the test, and allow them to be available to the user. All this happens in the deploy.py script from before.

There are other ways to send the output to the user. For instance:

class Question2(UTestCase):
    """ Second problem """
    @cache
    def my_reversal(self, ls):
        # The '@cache' decorator ensures the function is not run on the *students* computer
        # Instead the code is run on the teachers computer and the result is passed on with the
        # other pre-computed results -- i.e. this function will run regardless of how the student happens to have
        # implemented reverse_list.
        from cs102.homework1 import reverse_list
        return reverse_list(ls)

    def test_reverse_tricky(self):
        ls = ("butterfly", 4, 1)
        ls2 = self.my_reversal( tuple(ls) ) # This will always produce the right result.
        ls3 = self.my_reversal( tuple([1,2,3]) )  # Also works; the cache respects input arguments.
        self.assertEqualC(self.my_reversal( tuple(ls2) )) # This will actually test the students code.
        return ls

This code showcase the @cache decorator. What it does is it computes the output of the function on your computer and allows that result to be availble to students (the input arguments must be immutable). This may seem odd, but it is very helpful

• if you have exercises that depend on each other, and you want students to have access to the expected result of older methods which they may not have implemented correctly.
• If you want to use functions the students write to set up appropriate tests without giving away the solution

Furthermore, one of the test now has a return value, which will be automatically included in the .token file.

Example 3: Hidden and secure tests

To use unitgrade as a true autograder you both want security nobody tampered with your tests (or the .token files), and also that the students implementations didn't just detect what input was being used and return the correct answer. To do that you need hidden tests and external validation.

Our new testclass looks like this:

from src.unitgrade2.unitgrade2 import UTestCase, Report, hide
from src.unitgrade2 import evaluate_report_student


class Week1(UTestCase):
    """ The first question for week 1. """

    def test_add(self):
        from cs103.homework1 import add
        self.assertEqualC(add(2, 2))
        self.assertEqualC(add(-100, 5))

    @hide
    def test_add_hidden(self):
        # This is a hidden test. The @hide-decorator will allow unitgrade to remove the test.
        # See the output in the student directory for more information.
        from cs103.homework1 import add
        self.assertEqualC(add(2, 2))


import cs103


class Report3(Report):
    title = "CS 101 Report 3"
    questions = [(Week1, 20)]  # Include a single question for 10 credits.
    pack_imports = [cs103]


if __name__ == "__main__":
    evaluate_report_student(Report3())

This test is stored as report3_complete.py. Note the @hide decorator which will tell the framework that test (and all code) should be hidden from the user.

In order to use the hidden tests, we first need a version for the students without them. This can be done by changing the deploy.py script as follows:

def deploy_student_files():
    setup_grade_file_report(Report3, minify=False, obfuscate=False, execute=False)
    Report3.reset()

    fout, ReportWithoutHidden = remove_hidden_methods(Report3, outfile="report3.py")
    setup_grade_file_report(ReportWithoutHidden, minify=False, obfuscate=False, execute=False)
    sdir = "../../students/cs103"
    snip_dir(source_dir="../cs103", dest_dir=sdir, clean_destination_dir=True, exclude=['__pycache__', '*.token', 'deploy.py', 'report3_complete*.py'])
    return sdir


if __name__ == "__main__":
    # Step 1: Deploy the students files and return the directory they were written to
    student_directory = deploy_student_files()

This script first compiles the report3_complete_grade.py-script (which we will use) and then remove the hidden methods and compiles the students script report3_grade.py-script. Finally, we synchronize with the s student folder, which now contains no traces of our hidden method -- not in any of the sources files or the data files.

The next step is optional, but we quickly simulate that the student runs his script and we get a link to the .token file:

os.system("cd ../../students && python -m cs103.report3_grade")
student_token_file = glob.glob(student_directory + "/*.token")[0]

This is the file we assume the student uploads. The external validation can be carried out as follows:

def run_student_code_on_docker(Dockerfile, student_token_file):
    token = docker_run_token_file(Dockerfile_location=Dockerfile,
                          host_tmp_dir=os.path.dirname(Dockerfile) + "/tmp",
                          student_token_file=student_token_file,
                          instructor_grade_script="report3_complete_grade.py")
    with open(token, 'rb') as f:
        results = pickle.load(f)
    return results

if __name__ == "__main__":
    # Step 3: Compile the Docker image (obviously you will only do this once; add your packages to requirements.txt).
    Dockerfile = os.path.dirname(__file__) + "/../unitgrade-docker/Dockerfile"
    os.system("cd ../unitgrade-docker && docker build --tag unitgrade-docker .")

    # Step 4: Test the students .token file and get the results-token-file. Compare the contents with the students_token_file:
    checked_token = run_student_code_on_docker(Dockerfile, student_token_file)

    # Let's quickly compare the students score to what we got (the dictionary contains all relevant information including code).
    with open(student_token_file, 'rb') as f:
        results = pickle.load(f)
    print("Student's score was:", results['total'])
    print("My independent evaluation of the students score was", checked_token['total'])

These steps compile a Docker image (you can easily add whatever packages you need) and runs our project3_complete_grade.py script on the students source code (as taken from the token file).

The last lines load the result and compare the score -- in this case both will return 0 points, and any dissimilarity in the results should be immediate cause for concern.

• Docker prevents students from doing mailicious things to your computer and allows the results to be reproducible by TAs.

Moss plagiarism detection

You can easily apply Moss to the students token files. First get moss from https://theory.stanford.edu/~aiken/moss/ and create two directories:

whitelist/   # Whitelisted files. Code from these files are part of the handouts to students
submissions/ # Where you dump student submissions.

The whitelist directory is optional, and the submissions directory contains student submissions (one folder per student):

/submissions/<student-id-1>/..
/submissions/<student-id-2>/..

The files in the whitelist/student directory can be either .token files (which are unpacked) or python files, and they may contain subdirectories: Everything will be unpacked and flattened. The simplest way to set it up is simply to download all files from DTU learn as a zip-file and unzip it somewhere. When done just call moss as follows:

from unitgrade_private2.plagiarism.mossit import moss_it, get_id

if __name__ == "__main__":
    # Extract the moss id from the perl script:
    id = get_id("../../../02465private/admin/moss.pl")

    # moss_id should be a string containing an integer, i.e. "2434222134".
    moss_it(whitelist_dir="whitelist", submissions_dir="student_submissions", moss_id=id)

This will generate a report. You can view the example including the report here: https://lab.compute.dtu.dk/tuhe/unitgrade_private/-/tree/master/examples/example_moss

Smart hinting

To help students get started, unitgrade will collect hints to solve failed tests from across the codebase and display them. Consider the following homework where two problems depends on each other and the instructor has given a couple of hints: (example taken from example_hints):

def find_primes(n): #!f
    """
    Return a list of all primes up to (and including) n
    Hints:
        * Remember to return a *list* (and not a tuple or numpy ndarray)
        * Remember to include n if n is a prime
        * The first few primes are 2, 3, 5, ...
    """
    primes = [p for p in range(2, n+1) if is_prime(n) ]
    return primes

def is_prime(n): #!f
    """
    Return true iff n is a prime
    Hints:
        * A number if a prime if it has no divisors
        * You can check if k divides n using the modulo-operator. I.e. n % k == True if k divides n.
    """
    for k in range(2, n):
        if k % n == 0:
            return False
    return True

The report_file is simply as follows:

from unitgrade2 import Report, UTestCase, evaluate_report_student
from homework1 import find_primes
import homework1

class Week1(UTestCase):
    def test_find_all_primes(self):
        """
        Hints:
            * Insert a breakpoint and check what your function find_primes(4) actually outputs
        """
        self.assertEqual(find_primes(4), [2,3])

class Report1Hints(Report):
    title = "CS 106 Report 1"
    questions = [(Week1, 10)]  # Include a single question for 10 credits.
    pack_imports = [homework1] # Unitgrade will recursively include all .py files from "cs101flat"

if __name__ == "__main__":
    evaluate_report_student(Report1Hints())

When students run this homework it will fail and display the hints from the two methods:

What happens behind the scenes is that a code-coverage tool is run on the instructors computer to determine which methods are actually used in solving a problem, and then the hint-texts of those methods (and none other) are collected and displayed. This feature requires no external configuration; simply write Hints: in the source code.

Citing

@online{unitgrade_devel,
	title={Unitgrade-devel (0.1.0): \texttt{pip install unitgrade-devel}},
	url={https://lab.compute.dtu.dk/tuhe/unitgrade_private},
	urldate = {2021-09-08}, 
	month={9},
	publisher={Technical University of Denmark (DTU)},
	author={Tue Herlau},
	year={2021},
}