piotrm/albacross

backend-recruitment-task

Albacross - backend-recruitment-task

Testing the solution

The solution has been deployed to Heroku, therefore there is no need to set it up locally in order to see the results.

Swagger

There is a description for the API prepared in swagger that can be found here: https://app.swaggerhub.com/apis/pmatuszkiewicz/AlbacrossAPI/1.0.0

Swagger has a feature that allows to test the API with provided parameters straight from the documentation. In order to do it:

1. Select the endpoint
2. Press "Try it out"
3. Provide parameters
4. Execute

Browser

The response from the endpoints is also available via browser under: http://albacrossrecru.herokuapp.com/api/v1/

cURL

Third option is to use cURL.

With headers:

curl -i -H "Accept: application/json" -H "Content-Type: application/json" -X GET https://albacrossrecru.herokuapp.com/api/v1/employees\?page\=1&per_page\=15&salary_from\=50000&salary_to=54000

curl -i -H "Accept: application/json" -H "Content-Type: application/json" -X GET https://albacrossrecru.herokuapp.com/api/v1/departments/d009/active_employees\?page\=1

Without headers (pretty print works on OS X):

curl -H "Accept: application/json" -H "Content-Type: application/json" -X GET  https://albacrossrecru.herokuapp.com/api/v1/departments/d005/active_employees | python -mjson.tool

Progress Log

-> https://github.com/piotrm/albacross/commit/455f9496ae59d1080507bfa1a2cad381b57ed5fc

• Generates schema.rb from external DB's schema by rake db:schema:dump to make sure the schema does not get out of sync during cumbersome migrations' creation process
• Adds native_enum gem to enable MySQL's enum type that is not handled natively by Rails

-> https://github.com/piotrm/albacross/commit/0c177d7cae5ea10dd7f75c9c06d3d3fa0fe5a0a8

• Adds composite_primary_keys gem to enable composite primary keys for DeptEmp and DeptManager

-> https://github.com/piotrm/albacross/commit/f04b642802b6dff6c2a21c26412190a7c3d1a7d0

• Sets has_many :through association only for Department - DeptEmp - Employee trio, not for Department - DeptManager - Employee because Rails does not allow multiple has_many :through associations between two models (e.g. Department/Employee) - it raises ActiveRecord::HasManyThroughOrderError
• Creates DepartmentEmployee model for DeptEmp table and DepartmentManager for DeptManager table, breaking Convention over Configuration paradigm in the context of model naming for the sake of clarity. It is worth mentioning that the conventional naming (DeptEmp/DeptManager) would not work out of the box, because the tables in the external DB are named dept_emp and dept_manager and Rails would look for plurals - dept_emps and dept_managers.

-> https://github.com/piotrm/albacross/commit/9ce7e98a59c90017ce716e3dc0340d73ba229cda

• Changes the source of native_enum gem to the fork that consist of a fix for Rails 5.0. Using gem from rubygems causes incorrect limit interpretation during dev / test DB creation, making specs unusable.

-> https://github.com/piotrm/albacross/commit/4dfa91c15a95c963d6ca9a37a71d47f9d04f32af

• Exposes endpoint that allows to fetch Employees along with Departments, Titles and Salaries. It allows to fetch both all of the Employees (with pagination - either with specified page / per_page prams or with default ones page: 1 and per_page: 10) and Employees filtered by the salary range by providing salary_from or/and salary_to parameters.
• Initial concept circled around making the enpoint compliant with the json:api spec, but it then turned out that with given DB schema it is impossible because one of the major requirements of json:api is to have :id which is obviously missing. Ultimately, the API's response is close to what json:api offers, with similar structure and amount of meta-information.
• The logic responsible for fetching the data has been moved to FetchEmployees service object in order to keep the controller lightweight.
• input_sanitizer gem helps to ensure that the parameters' format is proper.

-> https://github.com/piotrm/albacross/commit/9ee75bd19452cc1f57a113741fd9a91fd2571bc2

• Exposes endpoint that allows to fetch active Employees along with Departments, Titles and Salaries. Active, because there are recorded activities (clicks, views) for them. Also in this case the pagination has been implemented. The response is formated in the same manner as in case of /api/v1/employees.
• The logic responsible for fetching the employees is located in FetchActiveEmployees service object.
• The logic responsible for retrieving stats from Elasticsearch is located in SeekEmployeeStat service object, not to make mess in the Department model. I am not the fan of "fat model, skinny controller" approach - I prefer moving responsibilites to other class (libs, service objects, domain classes etc.)

-> https://github.com/piotrm/albacross/commit/28d13e0848c10888264827f4643f31b39db434d1

• Moves ResponseFormatter from /lib to app/lib in order to leverage file autoloading on production environment

-> https://github.com/piotrm/albacross/commit/b8d0fef61cf9e9097344397adb07b8e1cd4afc56

• Replaces hardcoded Elasticsearch host for ENV variable - in the same manner as in case of DB credentials.

-> https://github.com/piotrm/albacross/commit/8851e216656950e59ae369d31a2ac697d56f01a1

• Sets additional constraints on page and per_page parameters - they has to be positive integers. In case there will be 0 or less than 0 the endpoints will return 400 Bad Request. In order to set the constraints input_sanitizer gem is used.