competemcgill / wisp-problems-microservice Goto Github PK

When retreiving documents from mongo, we can specify a field to sort by. In our case, we should sort by the most recently created problems/problemSets by default.

Ignore if this is meant as a feature.

Steps to reproduce

1. Login with an Admin account
2. Create a problem set
3. Create a problem and add it to that problem set
4. Delete the problem set
5. Go to the /problems page. The orphan problem will appear in the table.

Issuing a PUT request on /problems/:problemid results in a 500 error - upon further investigation, I've determined this is because the controller reassigns the value of the problemId from the MongoID which is expected on the request to the computed Platform-Problem hash.

CI for building and pushing images shouldn't run on branches other than master OR, if it runs on a branch other than master, it should tag the image with branch_name-short_sha instead of just short_sha.

Currently, the GET "/problems/:id/exists" route returns a 200 if a problem exists and a 404 if it doesn't. This causes some confusion in services that are using the route. Instead, the route should return status 200 for both cases with a boolean denoting if the problem exists in wisp or not.

The IProblem interface used to set the fields "platformMetadata.difficulty" and "source" as type "Difficulty" and "PlatformName" respectively. This offered no validation for the values in the controllers and restricted values to 0, 1, and 2 when setting the fields manually. Currently, if the request body's value is different from the ones in the enums, the interface does nothing and it's up to mongo to throw the errors -> status 500. The types of these fields has since been changed in the interface to "string" in order to accomodate setting the values to strings instead of the ints 0, 1, and 2. From there, we now need to add custom validators to check if the request body values are indeed expected values.

For custim validators, see the validateProblemSetIds function in src/validators/problem

Remove the deleted problem set's id from the associated users' problemSets field.

Check if referenced problemSet exists before creating the problem then send back an appropriate error message.

Currently, users can submit problemSetIds that aren't mongo ObjectIds. This leads to a status 500 since mongo throws an error when parsing problemSetIds.

platformProblemId (i.e. "1A" for Codeforces) should either always be uppercase or lowercase. Currently, either can be sent through the POST or PUT routes, and these result in different problemId hashes ("1A" vs "1a" hashes to different results).

Also need to check with wisp-users-microservice to see how we're doing it there, and we should standardize it there as well.

Currently, problemCount is initialized when a problem is added to a problemSet. However, this can cause issues when a problemSet is created with existing problems. The issue can be solved by initializing the problemCount field as soon as the problemSet is created.

We are missing a couple of await statements on the problem controller!

Currently the int problemCount in the problemSet model can be changed in a PUT request. This should not be the case, as then it could be manipulated to not properly represent the number of problems in a problemSet.

@idodin could we use the github short SHA as a tag instead of the full SHA?

Also I don’t think we need to push twice: right now we’re pushing once for the GitHub SHA tag and once for the “latest” tag. I believe that the images can have multiple tags, and can be pushed/referred to by either tag. And these tags persist when we push the image. @MohamedBeydoun can confirm this though

CRUD routes for Contest model.

Contest model ideas:

• Inherit from problemSet?
• Add more fields inside problemSet to indicate that it's a contest + store contest info?

In the model:

• Start and end time
• problemSet-like info
• Scoreboard/participants

Remove the deleted problem set's id from the associated problems' problemSetIds field.