Best Matched Restaurants - DK

Running

• Make sure you have Docker installed
• Run docker compose build
• Run docker compose run
• The server should now be running at http://localhost:4000
• You can either access it (and the implemented endpoints) in the browser, via CURL, or something else similar

Implemented endpoints & parameters

• /api/restaurants returns a list of the 5 restaurants that best matched the following parameters (all optional):
  • max_distance: integer number, sets maximum (inclusive) distance
  • min_rating: integer number, sets minimum (inclusive) customer_rating
  • max_price: integer number, sets maximum (inclusive) price
  • cuisine_id: integer number, determines what the cuisine_id must be
  • name: string, determines a total or partial query for the restaurant name

An example of a full URL for the endpoint above would be http://localhost:4000/api/restaurants?max_distance=3&cuisine_id=15&min_rating=3&max_price=30

Running the tests

Setup

• Make sure you have ASDF, ASDF-Erlang, and ASDF-Elixir installed
• Run asdf install from within the project folder
• Run mix deps.get

Running

• Run mix test

Assumptions

• Even though the UI for a cuisine filter would probably be some sort of dropdown that lists the name of each cuisine, generally the cuisine's ID would still be accessible and would be preferred to communicate with the api over the name. Hence, I made the parameter be cuisine_id and not cuisine_name.

• Since I'm not using the cuisines.csv to map from name to ID, I've decided to go the opposite way to find use for it: search for the cuisine with the given ID, and append a cuisine object to the restaurant object.

Considerations

UI

I did not build a UI, but will describe roughly how I would've built one:

The base of the UI would be a sort of table listing the best matched restaurants (initial render would include a sorted but unfiltered list). Either above or below it, a group of inputs: a text input for the name, a number input for the maximum distance, a number input for the minimum rating, a number input for the maximum price, and a dropdown for the cuisine. Upon updating a value (but debouncing the text/number inputs), the table content would be updated to match the new filters. Resetting an input to its default value sets the parameter to null, which tells the server to not filter by that field.

Docker

Setting up a new language to run locally is always prone to take more time than one would like to, so I opted to provide a way to execute the server without having to setup Elixir/Erlang locally.

Database

I considered building the solution by building a database with the CSVs instead of reading them directly. I opted not to do so mostly becaue of time, but also because Elixir Streams seemed OK for this amount of data, and I'd imagine describing how that could work to be showcase enough:

• Create tables for both Restaurants and Cuisines;
• The relationship between them is already explicit by the cuisine_id field in a restaurant: Restaurants belong to one Cuisine and Cuisines have many Restaurants;
• Run a migration to import the data from the Cuisines CSV into the Cuisines table;
• Run a migration to import the data from the Restaurants CSV into the Restaurants table;
• Everywhere in the code where a CSV is read directly and filters are applied to the respective Streams, use Ecto instead to read from the database.