In order to run the app you will need go1.16+ and optionally make

In Makefile you will find command to run, test and lint the application.

• make run: to execute the server
• make test: to execute tests
• make lint: to execute linter

The applications writes a log in the current directory, with name: skus- {timestamp}.log

I also provide a simple script that sends random skus to the server (make run-client). It accepts two parameters for customization but should be run directly with go run:

• -count n: indicate the number of skus to send
• -terminate: will send a terminate sequence when all skus are sent

For example (server must be up to work):

go run cmd/client/main.go -count 123 -terminate

feeder
├── cmd/ -> application entrypoints
├── internal/feeder/ -> application specific code
├── out/ -> file for app output files
└── pkg/ -> application shareable code
    ├── server/ -> server that listens for incoming connections
    ├── store/ -> store that deduplicates strings 
    └── utils/ -> some utility code

I tried to follow Go standard layout:

Application entrypoints are placed in cmd/ directory, and one folder for each entrypoint.

In pkg/ folder, i placed code that could be shared among different internal service or maybe promoted to a library. It doesn't contain any business logic and is reusable.

in internal/ folder, there is the "business logic" of the application, this code would not be shared between services, for that it keeps in internal "private" folder.

As I understand, a client will only send ONE sku or terminate sequence, at first moment I supposed that a client could send many skus, but reading many times I understand it's not.

Test doesn't specify if sku is case-sensitive, or if all characters must be uppercase (like in examples). I decided to allow uppercase and lowercase, but match duplicated in case-insensitive way. For this, I transform all skus to uppercase when they are stored

It's obious we need some sort of storage in order to check for duplicates and finally count valid, duplicateds and bad skus. I decided to implement a simple in memory store. It keeps skus sorted to optimize searching for duplicates.

It allows to subscribe a function that will be called each time it gets a new not duplicated value, so we can log them into a file