The goal of this task was to build a service that, given latitude and longitude of an object, its type and marketing type, returns a list of similar objects with their prices.
It was implemented with Ruby on Rails.
For querying distances between properties, the activerecord-postgres-earthdistance gem was used which leverages the PostgreSQL extension earthdistance
This project was built with Ruby 2.5.1 and Rails 5.2.1 (--api mode).
In order to run this app, PostgreSQL 9.6+ should be installed, as well as Redis (optional) for testing of caching
In order to setup the database, first the properties.sql needs to be copied into db/dump folder (which is in .gitignore). Then run:
rake db:create
rake db:import
rake db:migrate
which should populate the database with the data, as well as add the needed PostgreSQL extensions and indexes to the datatbase
In order to enable caching locally with Redis, run rails dev:cache. Right now the serialized objects are being cached with the fast_jsonapi gem.
Note: maybe a better implementation would have been to add updated_at field to the table and leverage caching of the queries (although the sizes would have been bigger)
It has only 1 endpoint:
GET /api/v1/find
which accepts 4 params; lat, lng, property_type, marketing_type. In the current implementation all of them are mandatory. Example
{
lng: '13.4132',
lat: '52.5219',
property_type: 'apartment',
marketing_type: 'sell'
}
And the response is :
{"data": [
{ "id": "4", "type": "property",
"attributes": {
"house_number": "1",
"street": "Alexanderplatz Straße",
"city": "Berlin", "zip_code": "10178",
"lat": "52.5219", "lng": "13.4132",
"price": "990000.0"},
"relationships": {}
}
]}