/gradescope-scripts

Scripts to clean and bulk upload programming assignments to gradescope

Primary LanguagePythonMIT LicenseMIT

gradescope-scripts

These scripts are written to prepare student submissions as downloaded from CourseWeb to be uploaded to GradeScope for the purposes of

  • automatic or manual grading [1,3,4]
  • plagiarism detection [2]

Getting Started

Prerequisites

Setting Up

  • Clone this repository to a suitable location e.g.~/gs

  • Unzip the student submissions into a folder e.g. ~/gs/submissions

    The student submissions should be in the format of: ~/gs/submissions/<batch>/<student-details>/<student-id>.zip

Running

  • Login to GradeScope, open your course, then assignment, and note the course-id and assignment-id by the browser URL displayed

    gradescope.com/courses/<course-id>/assignment/<assignment-id>

  • Run the script clean-all.py using the following command:

       $ python clean-all.py ~/gs/submissions

  • The script will creates a CSV file students.csv. This could be used to bulk-upload students to GradeScope using the interface provided at:

    gradescope.com/courses/<course-id>/memberships

  • The submissions should now be re-organised into the following format

    ~/gs/submissions/<student-id>/<student-id>.java The <student-id>.java contains a merged version of all the source files belonging to the student. This yields more accurate similarity comparisons between code for plagiarism detection with less false positives.

  • Upload the submissions to GradeScope using the script upload-all.py from within the submissions folder

      $ cd submissions
  $ python ../upload-all.py

What Happens

This is an abstract explanation of what the program does. In-depth details could be understood by readingv the code

  • clean-all.py

    1. Traverse the submission folder hierarchy and generate the student list as CSV
    2. Moves all the student submission folders from inside batch folders to the submission root e.g. from ~/gs/submissions/<batch>/<student-submission> to ~/gs/submissions/<student-submission>
    3. The student submission folder names are a combination of names and student-ids, rename these to just student-ids
    4. Recursively extract all the zip files found inside the folder hierarchy. This includes zips inside zips as well
    5. Recursively remove all non-source files - at the moment we assume source is contained inside .java or .txt files only. Unfortunately if students have put source files in other formats such as WORD or PDF, they'll be removed too. This also renames all .txt extensions to .java extensions.
    6. Move all source files to the student-submission root e.g. directly under ~/gs/submissions/<student-id>
    7. Removes all empty folders
    8. Lists all student submissions with no files in them - this means all the files inside them were removed due to reasons mentioned above
    9. if merge==True merges all the source files inside each student submission folder into one source file. This helps when comparing submissions during plagiarism checking, and yields better results.

  • upload-all.py

    1. Logs into GradeScope using given credentials
    2. Captures the course-id and assignment-id from the user
    3. Loops through all the student folders
    4. For each student folder, uploads the student submission to the given course/assignment using HTTP post requests

Notes

The merge feature could be turned off by:

  • setting merge = False in upload-all.py, and
  • commenting out the function call to merge_files() in clean-all.py

This will result in multiple source files inside the student folder within submissions

~/gs/submissions/<student-id>/[files]

Limitations

  • Supports extraction of only zip, jar, and rar files in the submission hierarchy
  • Students' source code should be included within .java or .txt files, any files without extensions will be removed, along with any code written inside word or PDF documents
  • Limited to the folder hierarchy structure of CourseWeb only. Generalising this would require rewriting a significant part of the code
  • Limited to the CSV format specified by GradeScope
  • Limited to JAVA coding assignments - this can be generalised easily to other languages

Future Work

The following items are in the development pipeline

  • Migrate all code to python3
  • Instead of modifying the existing submissions folder, work the changes on a deep copy
  • Replace calls to subprocess with suitable python libraries
  • Use inner functions where possible - e.g. for recursive operations
  • Generalising to a wider set of courses, assignments, and submissions
    • configurable folder structures
    • configurable filtering parameters for source code (e.g. .c, .py, etc.)
  • Support more zip file formats
    • e.g. .7z, .tar.gz
  • Support more filetypes where source code might reside in
    • e.g. .docx, .pdf, screenshots (using OCR), etc.
  • Configurations via command line arguments to run parts of the code
    • E.g. allow the user to run cleanup steps 1,2,5, skipping 3,4

Credits | References