Perx Parser Technical Challenge

This is my submission for the Perx Health postfix notation parsing technical challenge, written in Elixir.

Installation

The first step is to clone the repo from GitHub

$ git clone git@github.com:lukerollans/perx_parser.git
$ cd perx_parser

The next step is to install Elixir and Erlang/OTP on your machine. You can use asdf if you have it installed

$ cd perx_parser
$ asdf install

If you don't, you'll need to follow the Elixir installation instructions found here

Note: Make sure you also install Erlang/OTP, the steps for which are also in the above link.

Once you have Elixir and Erlang/OTP successfully installed, you can install the app's dependencies.

Note: There are none except for the standard library, so this step is essentially just a formality.

$ mix deps.get

Finally, to ensure everything is running nicely, you should run the app's test suite and ensure it is green

$ mix test
.........

Finished in 0.04 seconds (0.04s async, 0.00s sync)
9 tests, 0 failures

Randomized with seed 68826

Deliverables

Invocation instructions

I've set up the deliverable functionality to be invoked via a "mix task". This is similar to a rake task in a Rails app or a custom script in a Node app.

To invoke it, run the following (while also supplying some CSV data via stdin)

$ cat sample.csv | mix parse_csv

The accompanying sample.csv found in this repo can be used as input data as demonstrated above, however any correctly formatted CSV should work.

Report

The challenge took me approximately two hours to complete. I chose Elixir not just because it's one of my favourite languages and one I'm comfortable with, but primarily because I feel the problem was much easier to solve with a functional and immutable approach rather than something like Node. This is just my opinion, I'm sure for other people the opposite may be true!

I approached the problem from a functional point of view. Pulling in each line, splitting each line up in to cells and printing back out the result of the postfix notation evaluation in to a new list.

There is one design decision I think can be improved, but it's more of a performance concern rather than an architectural concern. At the moment, the entirety of CSV data is loaded in to memory before any processing begins. In Elixir land, this could easily be updated such that as each line is read in and concurrently evaluated while subsequent lines are being read. I felt this would have been a good example of premature optimisation, however.