avantifellows / quiz-backend Goto Github PK

We need a variable called 'testPurpose' to accommodate values such as Full Syllabus Test, Major Test, Part Test, Chapter Test, Hiring Test, and Evaluation Test.

Add checks to ensure that multiple devices do not use the same quiz at the same time.

• This creates issues in session logic.

Session route logic assumes that there is only one session running at a time wrt (user_id + quiz_id) combination

This can help us track temporal information of how users are interacting with the quiz.
Can also help to sort users based on time of submission.

Wrong scorecard shown when users do not go through all questions. This happens because scorecard is computed based on marking_scheme field of question. And not all questions have marking_scheme.

When preparing subset without details, we do not project marking scheme.

Fix will be to project it -- so that all questions (whether subset is fetched or not) -- will contain marking scheme

Currently the cloudformation stack (API Gateway + Lambda function running FastAPI + S3 function to store changeset files) are all based in us-east-1.

To reduce latency as much as possible, we should move all of this infra to ap-south-1.
Luckily, it's just a one liner change as AWS SAM handles all of that for us

A new view that allows users to key-in their answers quickly on a single page. This is useful when users solve quizzes offline.

Make changes in backend for a new question type. The question looks like the image below

The task is to match A,B,C,D in colA with P,Q,R,S,T in colB

Options would look like the image below:

• Backend code should also incorporate marking scheme changes for this question format.

Note that this is effectively like 4 multi-answer-questions put under one question.

Add feature to have a marking scheme for each question set.

This allows for different sets with different scores / weightages.

Currently, we update session answer whenever a patch request is made.
It would be useful to provide a way to update all session answers at once. For example, when a test ends, we could sync all answers at once to be doubly sure of saving answers that users picked.

Related to #16

For example, if question type is numerical-integer, the correct answer should be an integer.
If the question type is single-choice, the answer should be a list.

Currently we allow answer to be int | string | list | null which is ok for basic validation, but not question-type specific

This will allow us to maintain a link to the source of the problem (only CMS for now, but more sources can come later).

Currently, we are storing a dummy-event object everytime frontend sends a dummy event.
We just need the timestamp of the new dummy events that frontend sends.

This will also reduce events array size.

Set up pre-commit in CI, so that pre-commit check runs during PR creation

Make the tests run when a PR is created and also calculate the coverage for the same so that we can get a report of how much of the code is covered by tests,

This helps in increasing the stakes of quiz completion. Making it much closer to real-world testing scenarios.

Caveat: keep the countdown on server / backend instead of frontend -- since user browsers may have different clocks and definitions of datetime

You can find more details in this ADR https://www.notion.so/avantifellows/ADR-4-Sync-quiz-prod-DB-to-dev-staging-DB-f4141cbe5ff543cd90f1c7b6216718a9

• Add instructions in README.md to run a mongodb host process in the background.

create_quiz POST request returns entire quiz. Instead return only the quiz_id. This reduces load on network.

We can compress the json responses so that loading time is reduced.

• API call to create an organization.
• Auto create an API key for each org.

This can help in analyzing sessions that are derived from the same CMS id