Karya Angkatan: Behind The Scene And User Input Messiness

Posted on 2021-11-21 11:00


In my faculty, for freshmens we are introduced with a bunch of events and tasks to further bond ourselves. One of it is Karya Angkatan, in which we are tasked to make something where everyone is able to contribute. For our year, we decided to follow last year's path and make a yearbook site. The deadline was long, around 3 months, so it is a pretty relaxed task. Here, I will talk about the technology being used in the site's back-end and how everything will be done, really.

Evaluation From Previous Work

Before I start on planning things, I decided to evaluate myself from previous work, specifically Tutor Angkatan. Upon Tutor Angkatan's release, there are one major problem: It was slow. This is caused due to little to no caching being used, and the number of request that is sent from the front-end. Learning from this, I decided to make my caching system more aggresive.

Tech Stack

For the back-end, we used the following technologies:

  • Python >=3.8 (3.7 works, but not tested)
  • FastAPI Framework
  • Ormar ORM
  • Redis

The stack was picked as such because of the following reason:

  • I am enforcing static typing
    Yes, static typing in a dynamic language, this is done to reduce the number of silly errors that might happen in testing or even in production.
  • Asyncronous While may not be very superior, this gives me the opportunity to use uvloop, which is the same tech Node.js uses, and from the benchmark it looks like it is performing much better than other frameworks.
  • Enforcing Documentation Literacy
    Due to them being such a new stack, not a lot of tutorials are available online, so your only option is to understand the official documentation and even the code behind it itself. I know I might be pushing it a bit too far, but it seems to be working just fine. It would benefit them as well.

Development Process

The flow for development is to create a git branch, then open up a Pull Request for it. Just like how you would do in professional environment. I will be the one reviewing it. Testing is mandatory as well, we use Pytest for that.

The process actually went really well, I am genuinely surprised with it, props to my team for that! We finished everything around 3-4 weeks before deadline, so there is a lot of room for hotfixes without being rushed.

Gathering Data

And... here comes the messiest part of all of this. Asking every single freshman for their data. For context, here are the things I am asking from them:

  • Student Identifier ID
  • Name
  • Username
  • Selected Program
  • House
  • Place and Date of Birth
  • Hobby
  • IT Interests (Hardcoded choices)
  • Instagram without the @ (optional)
  • Line ID (optional)
  • "About Me"
  • Future Message
  • Picture of themselves, in 3:4 aspect ratio, with file naming format being "ID_FotoDiri.ext"
  • Introduction video, portrait, possibly 16:9 aspect ratio, uploaded to YouTube

Yes, a lot, I know. However these are required for everything that we need to do for this task. It is submitted through Google Form, and I very well know that it will be messy. I sent them the form around 3 weeks before site's deadline, with the form's deadline to submit being a week before site's deadline.

A week passed, less than 10% has responded. 3 days later, 32.81% has responded. I then decide to split Introduction Video out of it and make a new form for that, so they will still have 2 days to respond. 2 days later, 94.34% has responded. This is good, but theres quite a few that needs to respond, so I poke them personally and there's only <3% that hasn't responded up until the very second I write this.

Now, I have one week to do everything. I need to import them to the server. However, I am very well know that there is going to be challenges, and here are those:

  • Impersonation or Trolls
    Fake entries, impersonation, and everything is definitely a big threat of mine. Obviously I don't want the site to have rick roll or even straight up porn in it.
  • People will add @ in Instagram ID regardless.
    I think this is self-explanatory, people can't read.
  • About Me and Future Message being inappropriate
  • Picture not being 3:4 aspect ratio, or not even portrait in the first place.
  • Picture not using the correct format
  • Picture being stupidly heavy

With these in mind, I have decided to ensure that I have all of those checks in place.

  1. Impersonation or Trolls
    The plan here is to check whether the name and ID is valid. To do so, I fetched from official documents and gathered everyone's ID and name, and then check whether the names match up.
  2. Instagram with @
    Well, simply check whether it has @ and just cut it, really.
  3. Inappropriate messages
    For this one, I have asked about 13 people to check every single messages, and ask them to change or remove the inappropriate bits.
  4. Picture Problem
    Yeah, this one is a hard one. There are about >100 people that does not follow the specifications (that's about 25%), and 30 people sent me landscape pictures. Hell, that's not even portrait!
    The plan then is to simply ignore those. I don't give a damn.

  5. Incorrect File Naming Here, I decided to iterate on every single entry in the spreadsheet, and rename the file to follow the correct naming scheme with a script.
  6. Heavy Pictures I'll just rescale those and compress it.

With all these in mind, the scripts are available here: GitHub Gist

And so, with all of these done, I can finally deploy the back-end into my server.

This task has been pretty tiring for me, especially on the last few days before deadline. However I had fun learning new things and explore new things to solve problems. I look forward into working in more projects in this university.

That's all, have a nice day!