root v2 at this point

[ivinjabraham]

Fixes #24. And I'll be rewriting some of the current system in hopes to clean up some of the unintuitive stuff that crept in. I could make these into two separate PRs, but I'd rather not let the latter be delayed after decoupling.

Please take code reviews more seriously so that we don't have to keep rewriting stuff.

Changelog

• remove soon-to-be outdated information in readme: With Shuttle gone, we'll have a much simpler setup.
• remove shuttle dependency
• reduce codebase to basic working web server template: Since a large part will be changed, I'll need to incrementally test out each feature which requires me to get rid of all the features temporarily.
• remove migrations: Old schema, changing it as it was inconsistent and weak in terms of integrity. Also, what are these migrations?! Three migrations just to rename a column? Create a table only to instantly delete and create it again? You do know there's a way to rename an existing table right? Sqlx::migrate! embeds the migrations into the binary, so unnecessary migrations will bloat the binary.
• add database backup to gitignore: Dumped all the data (as of Jan 12) from shuttle.
• remove unused state from router: Don't see much of a need for this right now.
• ignore .env files
• add migration to create initial tables for member, streaks and attendance: Reflects the new schema.
• remove shuttle workflows
• rename db module to models in order to better reflect its contents: This directory has nothing to do with the database, it's just models used by SQLx to serialize/deserialize between Rust and the DB.
• add graphql interface back to the server: Will need to disable/lock this in the future. Adding just for development purposes.
• minor refinements to loading env variables, db pool configuration and CORSLayer: Restricted requests to only come from home, DB pool now has at least 2 connections up at start and dotenv is made to expect to find .env since we're not using any other deployment configuration for env vars.
• uncomment scheduled task
• rename scheduled_task to daily_task
• rename reference to db module to models, was left out earlier
• use kolkata timezone instead of local time
• remove leaderboard and active projects feature: rip.
• documentation changes and additions
• ignore .log files: We have logs now.
• rewrite graphql interface: Don't tell anyone but this basically rewrites the whole thing.
• remove tests: Will add this back in the future.
• add re-exports for less verbose imports
• add BIND_ADDRESS env variable to avoid hardcoding address
• improve documentation in main.rs
• refactor main.rs to be more high-level by abstracing away its processes
• rename endpoints for more clarity
• complete todo to handle error when failing to fetch members
• add documentation to models
• fix graphiql missing graphql schema
• update documentation for root
  ...and more
  These are some pretty chonky and important commits, worth reviewing twice:
• rewrite midnight attendance records insertion and required models
• rewrite attendance record insertion and updation logic

TODO

• Remove shuttle workflows.
• Restore old data.
• Rewrite the logic based on the previous schema to work on the new schema.
• Rewrite the GraphQL interface to be more end-user friendly
• Rewrite some of the comments and/or document most of the codebase. A lot of functions make assumptions that are not specified anywhere, I only saw Swayam commenting stuff well.
• Restrict CORSLayer
• Solve, if any, lint issues
• Update docs to be more comprehensive of the entire system and it's applications. Ideally, we have a "why" for every feature we have in Root. Need information on the DB schema. Additionally, development setup and process should be made more clear as well.
• Reset streaks mutation.

Future Work

I'll be removing some of the stuff that isn't actively used now to simplify (lol) this pull request. This will be a list of stuff to do later as well as the list of stuff that I removed.

• Anything to do with ActiveProjects.

• The leaderboard features in main are a huge mess starting from the relations and it seems whoever reviewed it did not pay attention. Hence, I think it's just better if @swayam-agrahari rewrote it at a later date. Some of the problems include:

  • DB Schema containing derived fields like problems_solved, confusing columns like contests_participated and total_contests, fields like best_rank being non-nullable.
  • Overfetching data in fetch_leetcode_stats.
  • update_leaderboard doing way too many things for one function.
  • The algorithm for ranking is not correct?
  • Data inconsistency issues when using the add_or_update mutation for username updation.
    This is ignoring pedantic, but still important, problems like fetch_{x}_stats being functions that insert data into the DB unlike what the name implies. Additionally, since tests are mainly for the leaderboard feature, that was removed too.

• HMAC or any other form of "authentication" in the request itself. The solution to this is obviously to prevent users from accessing the API unless authorized, not to send auth information in requests. This means we'll need to integrate basic CRUD operations on Home for members (and other models in the future). Note: markAttendance still has HMAC verification.