Authorizer App

The Authorizer application authorizes a transaction for a specific account following a set of predefined rules.

The application aggregates all transactions within 2 minutes interval, once the aggregation interval completes, the transactions will be executed according to the time defined in transaction.

The application:

Appication is written in Scala and based on Akka actors .

running the Application

java -jar $PATH_TO_JAR/authorizer-assembly-1.0.jar

For graceful stop use the command stop as input or simply CTL+C

Building executable

Use sbt build tool https://www.scala-sbt.org/

run the command: sbt clean assembly or sbt assembly

Project structure

Domain Driven Design is the main design approach hence using a ubiquitous language trough the spec document given by Nubank and naming the application entities

Tests

All rules defined in the spec are covered in AthorizerSpec.scala

domain entities:

Authorizer

Aggregate the Account life states, and simulates state machine of the account life cycle and executes the operations according to the logic

Authorizer states :

waitForInit - the initial state, waits for Init account command , once the account is initilize with active-card true it fires timer accourding to transaction interval (2 minutes,can be configured) and change the state to run or to inactiveAccount (depends on the active-card argument)
run - accepts transactions and checks validation of the operation
inactiveAccount - in case the active-card:false, this state will respond with violation "card-not-active" in case of transaction

violations

contains the value objects that represents the different violations

Operations

contains the value objects that represents the different operations and responses

protocol

This package contains the json conventions protocol The application accepts the input and emits the output as json

utils

package that contains helpers for handling the configuration and datetime conventions gracefully

Publisher - simple actor that simply receives a message and emits it to the stdout as a json

Remarks

Akka actors are low resource and enables to achieve async and scalable operations easily.
Each actor is minimized to maintain single responsibility althogh it might be a little overhead or over engineering I do believe that it is correct in a real life apps.
In order to keep it simple I didn't use persisted actor or other persistence storage and maintained a single account with in memory state as requested.

123avi/authorizer