Challenge for Software Engineer - Big Data

Assumptions and Notes

There is already some sort of Ruby version manager being used (i.e. RVM, rbenv, chruby) and the local version of ruby for this project is set to 2.1.2.
The Ruby version has been coded into the Gemfile to avoid possibly running a different version.
As this is an exercise:
- Not knowing what flavor of DB might be running on the test machine(s), I stuck with SQLite as it is (usually) the most straightforward.
- Not knowing the preference of web server on the test machine(s), I stuck with WEBrick.
- The file will be able to be uploaded and processed in the scope of a single request.
- No styling was created.
- Since the upload model is only one field, I did not implement handling for error messages on its form.

Getting started

Make sure your ruby version manager is using a local version of 2.1.2

$ bundle install
$ bundle exec rake db:migrate
$ bundle exec rails s

Open your browser and navigate to http://localhost:3000

Running tests

The test suite can be run by using rake: $ bundle exec rake test

Approach

Authentication

I did take a look at also integrating OpenID authentication with Devise, but it's been so long since I implemented I felt it would exceed the time expectations for this exercise.

File Upload

Rather than have had the form post to a controller action and have the data processed there, I set up carrierwave so my thought process would be clearer and more could be shown.

I set it up so the data would get serialized to the model and then that model generates the appropriate records in a callback. This was mostly to separate the actions and make it clear that in a real-world app, those processes would be handled in the background and could be separated if desired.

Data Model

I set up the data model in such a way that if desired in the future, revenue could easily be broken down by purchaser or merchant in addition to by file that has been uploaded. I made sure to avoid ambiguous column names where applicable (i.e. quantity on the purchases table instead of count; changed my initial filename on the uploads table to file when I remembered carrierwave has an internal method called filename that it uses)

grumpit/data-engineering