boraarslan / mapper-influence Goto Github PK

https://doc.rust-lang.org/cargo/reference/workspaces.html#the-dependencies-table

Since we use multiple crates for the project, workspace dependencies would make it easier for us to manage all of the dependencies.

less than 10 fps when the videos are visible 🤔

Steps to reproduce:

1. Login

Redirect does not provide anything to use on frontend to represent auth state.

Currently, there are no tests to cover new database queries and API endpoints. It's good idea to use mockall package.

missing fields:
pending_count: number
follower_count: number
flag: { code: "TR", name: "Türkiye" } #65

get_user function should return:

• user id
• user name
• avatar url
• background url
• user title
• number of users influenced
• user group array:
  • group color

    hexcode

  • group short name

    ex: NAT

  • group long name

    ex: Nomination Assesment Team

We want to store the timestamp of creation and the last modification of the influence.

API should return both properties

A /login and /logout endpoint is required.

Login endpoint

// Request 
POST API_URL/login
body: {
  code: "xxx",
}

// Response 201
{
  token: "xxx",
  user: { // This is the bare minimum info to populate the header. {id} will be used as a link
    id: number,
    username: string,
    avatarUrl: string,
    groups?: Group[] // This might not be needed, maybe can be used to color the avatar border
  }
}

Logout endpoint

// Request
DELETE API_URL/logout
body: {
  token: "xxx"
}

// Response 204
no content

Two endpoints are needed:

1 - Base user info

Returns basic user information such as user id, name and avatar url.
This is just a proxy for the osu api, there is probably a function that implements the api request already.

2 - Extended user info

Returns the information needed to populate the profile page.
This includes their bio, influences and mentions on top of the base user info.

Up to you if you want to include the base info in the extended info endpoint.
If you want a more specific return type, tag me.

Currently the user endpoint is missing the total number of times that other users added the user as influencer and the list of influences that the user has added themselves.
extends #136

Currently the api request fails because auth function does not check for expired sessions.
Add refresh token logic and force to re-authenticate if both are expired.

The leaderboard is shown in the homepage, so it shouldnt have authorization requirement
Number of users returned could be a dynamic number from the query, with max and min values of course.

Returned array of values:

• user data

  same as get_user

• influence count

  number of users who added this person

array should be ordered by influence count

Backend should control and handle expired cookies

• Delete cookie when user session is expired

Our init_user test on mi-db is broken. It should be fixed.

Success or Fail are way too vague, we want to know why auth request is failed

when opening the site while logged in, the landing page is shown until the user is fetched.
we can redirect the user to /dashboard before getting the user info from the backend

we can redirect the user to /dashboard after logging in as well, easy to change in env vars

It is possible to manipulate the scopes when redirected to Osu! OAuth page. For example, if a user removes public from the query parameter scope and we try to make an Osu! Request that depends on that scope with that user's token, API fails because of insufficient permissions.

We want to know if the user granted us the correct permissions before letting them use the site.

Currently, the latest version fails to compile. I did not spend too much time debugging so it might be a very simple fix.

Current logs are under HTTP namespace, this needs to be changed

Currently all of the colors used in the site is in one file. For example: Colors used in the user card component could be defined in that folder.

Current endpoints arent enough to construct the profile page.
Connected to this issue: #65

Needed fields:

• boolean for if the user has already added this person

  can use the cookie to get the requesting user,
  can be a separate endpoint

• influence data
  • influence array of:

      - influenced user  ->  same with get_user data
      - influence strength
      - influence description
      - example maps  ->  should have the function to pull map info
    

  • mentions array of:

      - connected user  ->  same with get_user data
      - boolean if the user entered a description
    

• map count info

  graved, pending, ranked etc. maps

• user bio

  moved from get_user to this endpoint

• featured maps

  area under bio in which the user can showcase their maps, this can either be just the map id or map information such as song metadata and difficulty name

This is the information needed to fill the profile page. Reference design:

this issue is observed only once and needs to be reproduced first

Currently, we use the user id to generate their cards. Those ids are fetching their user data individually.
The current return is this:

type LeaderboardResponse = {
  id: number;
  user_name: string;
  profile_picture: string;
  ranked_map_count: number;
  influence_count?: number;
};

We only need the id and probably influence count. We could access the influence count from the user data so it is not mandatory. Would be nice to have though.
An alternative is to have all of the base user data to immediately render the users.

Proposed return type:

type LeaderboardResponse = {
  id: number;
  influence_count: number;
};

cargo-deny CI check has warnings, therefore it fails even on main branch.

Need to eliminate the need for it in the card username displays. There is a new card design which has enough space:
connected issue: #65

either need pagination or dynamic size query, can be hardcoded for now

would be nice to have map leaderboards alongside user leaderboards

returns array of:

{
  user: // base user data for card
  count: // influence count for scoring 
}

or for maps

{
  map: // base map data for card
  count: // influence count for scoring 
}

SQLx creates a separate database for each test annotated with #[sqlx::test], and sometimes our test suite fails because SQLx can't connect to the database. Ideally, we want as few tests as possible so we won't get database connection errors when running the tests.

One way to solve this would be merging similar tests into one test. Such as:

#[sqlx::test]
async fn test_insert_user(db: PgPool) { /* ... */ }

#[sqlx::test]
async fn test_insert_no_bio(db: PgPool) { /* ... */ }

#[sqlx::test]
async fn test_insert_same_id(db: PgPool) { /* ... */ }

// These three tests can be merged into one test

#[sqlx::test]
async fn test_insert_user(db: PgPool) { /* We test all of the insert scenarios here */ }

Steps:

1. Click login or go to /login
2. Click cancel

Currently, our CI workflow is very redundant and it makes it harder for us to edit/update/add jobs. Ideally, we want to use a matrix to set up the redundant CI commands.

In addition to that our tests are dependent on various services so it might be better to use our build commands instead of cargo actions. (not really sure about this)

there should be no body in a get request according to https://www.rfc-editor.org/rfc/rfc2616#section-4.3
no need to have an optional id

the endpoint should return the base user data (see #65)
with that, a boolean for whether the user has already registered to the site or not would be nice.

could use path parameters for request because there is a 13 char limit to usernames:

GET .../search/:string

[
  {
    "user": { ... },
    "registered": true
  }
]

2 problems:

1. Empty user_id request not working. Should be returning user connected to the cookie sent
2. Both valid and invalid user_id throws Unsupported type `core::option::Option<i64>

Currently, we use Make to simplify the development workflow. Make is great and enough, but I personally would like to have a justfile in the project.

just : (https://github.com/casey/just)

Currently we have a function to lookup beatmaps with their difficulty id, we cannot find maps with just a setid
v2/beatmapsets/[id]

will be used to switch between localhost and production domain
mapperinfluence.me should always be allowed