Autograder

An automated grader for C++ projects!

Features

• Unzip all submissions in one go
• Ensure certain files are not tampered with in each submission
• Compile, run, and check all programs for memory leaks all at once
• Apply points/penalties automatically for:
  • Compilation errors
  • Warnings
  • Crashing
  • Memory leaks
  • Unit test results
• Compare work for plagiarism using Moss
• Output all grades with feedback in a CSV
• Email grades to every student with their individual feedback

Example:

Prerequisites

• GCC & Make
• Python 3
• Valgrind

Installation

1. Clone the repo and cd into it:

   git clone https://github.com/KRMisha/Autograder.git
   cd Autograder

2. Install the required packages

   • With a venv (recommended):

     python3 -m venv .venv
     source .venv/bin/activate # or .venv\Scripts\activate.bat on Windows
     pip install -r requirements.txt

   • Globally:

     pip install mosspy

Setup

1. Download all the student submissions from your university's web portal (e.g. Moodle). Save them to the zipped directory.

   The zipped directory should contain a list of folders each named with the submission's team name. Inside each of these team folders should be the respective team's turned in work as a zip archive. The name of the zip archive does not need to follow a specific convention.

   For example:

   zipped
   ├── S1_G4
   │   └── finalversion.zip
   ├── S2_G8
   │   └── 1234567_1234568_2.zip
   ├── S2_G9
   │   └── lab.zip
   └── ...
   

2. Fill in the tab-delimited teams_info.csv file with each student's information. For example:

   LastName	FirstName	ID	Email	Team
   Last1	First1	1234567	first1.last1@email.com	S2_G8
   Last2	First2	1234568	first2.last2@email.com	S2_G8
   Last3	First3	1234569	first3.last3@email.com	S2_G9
   Last4	First4	1234570	first4.last4@email.com	S2_G9
   ...	...	...	...	...

   Make sure to keep the header row with the named fields shown above.

   The Team column's values should match the folder names inside the zipped directory. This is how the Autograder will match submissions with their author.

3. In the master_files directory, add any files you wish to have automatically copied to every team's submission before compiling (overwriting them if they already exist).

   This is useful for making sure each project has a Makefile, input text files, or that certain source files have not been tampered with by students (such as unit tests). Any files ending in a .txt file extension will automatically be copied into the built executable's directory.

   Tip: use the provided Makefile (repo) to avoid needing to tweak autograder.py.

4. Edit the config.py file according to your needs:

   • Edit the "Grading settings" section to change the penalties associated with certain errors.
   • Change the MOSS_USER_ID variable to your Moss ID (see the Moss instructions to learn how to obtain a Moss ID)
   • Change SMTP_SERVER and SMTP_PORT to your email server's settings.
   • Change SENDER_EMAIL to the email address from which you will send the grades.
   • Change EMAIL_SUBJECT and EMAIL_CONTENT with the desired information.

Usage

Autograding

Unzip and autograde all submissions in a single step:

./autograder.py

This will first unzip all the zipped submissions into a new unzipped directory. Then, the following will be done for each submission:

• Copy the unzipped files and master files to a team folder inside the grading directory
• Compile the program
• Output any compilation errors and warnings to make_stderr.txt (depending on the warning options set in the Makefile)
• Run the program and save its output to program_output.txt (with both stdout and stderr)
• Parse the test results from the program output
• Output any memory leaks detected with Valgrind to valgrind_stderr.txt
• Create a points_to_rm.txt file. This file will contain the breakdown of points which have automatically been subtracted for compilation errors, warnings, crashes, leaks, or failed unit tests.

The team folders will be placed in either of the compilation_failed, crashed or working subdirectories of the grading folder depending on their state.

If an error occurs and you try to rerun autograder.py, the existing output folders will be detected and the script will halt to prevent accidentally overwriting everything. To force it to run again, use the following command:

./autograder.py force

Manually grading style

Once each submission has been autograded and the points_to_rm.txt file has been created in each team's grading folder, you can now manually check each submission to grade aspects such as code style or good practices.

To do so, simply open each submission in the grading directory. To remove points with feedback, add a line to the team's points_to_rm.txt file with the following format, where the # represents any decimal number:

-#: Explanation for penalty

To grant bonus points, simply use a + sign rather than a - sign.

Calculating grades

To calculate students' grades based on the feedback in each points_to_rm.txt file, use the following command:

./calculate_grades.py

This will create two CSV files in the grades directory. They both contain the same information but present it differently:

• individual_grades.csv: all the grades with feedback, one line per student
• team_grades.csv: all the grades with feedback, one line per team

Checking for plagiarism

To check for plagiarism between teams by detecting excessive similarity, use the following command:

./check_plagiarism.py

This will compare all the files in the unzipped directory and output the report of similarity to moss_report/report. Simply open moss_report/report/index.html in your web browser to view the results.

To force the report to be regenerated if it has already been run, add the force option when invoking check_plagiarism.py.

Sending grades by email

Finally, to send each student their grade and feedback as saved in grading/individual_grades.csv, use the following command:

./email_grades.py

After you provide your email login credentials at the prompt, an email will be sent to each student with their grade.

Make sure to have set the email's desired content in config.py beforehand!

Folder hierarchy overview

.
├── grades
│   ├── individual_grades.csv               # Final grades for each student
│   └── team_grades.csv                     # Final grades grouped by team
├── grading
│   ├── compilation_failed                  # Programs which did not compile
│   │   ├── S1_G6
│   │   │   ├── <team files>
│   │   │   ├── compilation_stderr.txt      # Make stderr output (compilation errors and warnings)
│   │   │   ├── compilation_stdout.txt      # Make stdout output (what compiled successfully)
│   │   │   └── points_to_rm.txt            # List of all removed points and feedback
│   │   └── ...
│   ├── crashed                             # Programs which compiled but crashed
│   │   ├── S1_G5
│   │   │   ├── <team files>
│   │   │   ├── compilation_stderr.txt
│   │   │   ├── compilation_stdout.txt
│   │   │   ├── points_to_rm.txt
│   │   │   └── program_output.txt          # The program output (both stdout and stderr)
│   │   └── ...
│   └── working                             # Programs which ran successfully
│       ├── S1_G1
│       │   ├── <team files>
│       │   ├── compilation_stderr.txt
│       │   ├── compilation_stdout.txt
│       │   ├── points_to_rm.txt
│       │   ├── program_output.txt
│       │   └── valgrind_stderr.txt         # The stderr output for valgrind
│       └── ...
├── master_files                            # The master files copied into each submission
│   └── <files>
├── moss_report
│   ├── concatenated_files
│   └── report                              # Plagiarism report generated by Moss
├── unzipped                                # Unzipped submissions with added master files
│   ├── S1_G1
│   └── ...
├── zipped                                  # Zipped submissions as downloaded
│   ├── S1_G1
│   └── ...
├── .gitignore
├── autograder.py
├── calculate_grades.py
├── check_plagiarism
├── config.py
├── email_grades.py
├── README.md
├── requirements.txt
└── teams_info.csv                          # CSV with each student's information

License

MIT