Create Grading Service Part 2 (POST /cs3213/feedback_error and a GET endpoint to obtain submission with feedbacks)

[tryyang2001]

Pull Request

Description

Part 2 of grading service implementation. Created the API wrapper for the POST /cs3213/feedback_error endpoint. Also created a GET /questions/:questionId/submissions?studentId=xxx endpoint to allow retrieving generated feedbacks for the same submission without calling the ITS API again.

To explain further, the wrapper for POST /cs3213/feedback_error endpoint has the API path of /grading/api/feedback/generate, which also uses POST method. Instead of taking non-human-readable code parser as input request body, it only takes the JSON-friendly actual source code, the language (only support "python" or "c"), the question id and the student id. In short, the request body to POST /grading/api/feedback/generate should have this format, see example:

{
  "language": "python",
  "student_code": "def is_odd(x):\r\n    if x % 2 == 1:\r\n        return False\r\n    else:\r\n        return True",
  "question_id": "some-question-id",
  "student_id": "some-student-id-in-integer-value"
}

The response of this generateFeedback endpoint is shown in the screenshot section.

Related Issue(s)

Closes #20

Screenshots (if applicable)

Above is the example request and response of GET /grading/api/questions/:questionId/submissions endpoint.

And this is the example request and response of POST /grading/api/feedback/generate endpoint.

Checklist

• I have checked that the changes included in the PR are intended to merge to master or any destination branch.
• I have verified that the new changes do not break any existing functionalities, unless the new changes are intended and have approved by the team.
• I will take care of the merging and delete the side-branch after the PR is merged.

Additional Notes/References

As the next step of action, I will add more proper error checking, create unit tests, and the API docs for grading service. One potential issue is that calling POST /grading/api/feedback/generate is extremely slow, as ITS API takes quite long to generate the output, and the code parser can be very long.

[vivienherq]

LGTM!

[hhchinh2002]

LGTM! beside few very minor code quality that I left reviews, all sample requests from Postman json file work for me, however:

• due to complexity of the techstack you are using, maybe can consider create a README.md file with few simple instruction on how to build/run the service locally
• add yarn prisma generate to the package.json file scripts under build

[hhchinh2002]

This look like DRY to me cause this piece of code appear in other place, I think you can consider exporting an method that set response status and json body from ITSPostParserError.ts and invoke it

[tryyang2001]

Oh okay, tbh the actual reason was that the ITS response on error is not very clear and consistent, so I was just using a simple way to distinguish what happens... Will polish this in the future.