Adverity coding challenge

Hello there. I am very pleased to have you here, interested in my work. I imagine the next section is by no means an epiphany for you, but still - who doesn't appreciate some ready to copy-paste set of commands, huh? Just to speed things up a bit. So let's start with...

Application preparation

First, we need to build it. With JDK 11 of any flavour. In order to build the app simply run

$ mvn clean package

Once completed application can be started by issuing lame

java -jar .\target\adverity-1.0.jar

For those who prefer docker way of life an alternative approach has been prepared.

Type in:

$ docker build -t adverity:1.0 .

to build the image and then to start the container run:

$ docker run -d -p 8080:8080 adverity:1.0

It doesn't do much else, I just needed this in order to send it to Amazon's ECR, but since I already have it...

Now application should be up and running. If not then maybe you're looking for someone other than me ;) Or not, your call.

Assuming it works though - you can move to the next section and see what amazing stuff you can do with this app!

API

What can we do with it? Not that much, I admit. It's a tech demonstrator after all, not a production-grade service. But it can do the following:

Query metrics as daily trends

GET /impressions

GET /clicks

Both of these accept request parameters from=yyyy-MM-dd and to=yyy-MM-dd. Those parameters are not required, but be aware that if you don't specify them yourself an app will provide you with default ones (beginning of time and today)

CTR for dimension

GET /ctr

It will give you the CTRs computed in context of given parameter dimension={DATASOURCE|CAMPAIGN}. You don't actually need to put in whole dimension name. You can also try dimension=data or dimension=d, as long as it is a beginning part of dimension name - you're good.

Summarized metrics by dimension

GET /metric/{metric}/byDatasource/{datasourceName}

GET /metric/{metric}/byCampaign/{campaignName}

Should you be interested in summarized metric (CLICKS or IMPRESSIONS) for given dimension (DATASOURCE or CAMPAIGN) in a specified time window (from=yyyy-MM-dd and to=yyyy-MM-dd) then this is the endpoint for you.

And yes, you can be lazy effort-efficient here as well (e.g. i instead of IMPRESSIONS and so on)

Adding entries

POST /add

This endpoint can be called when you want to add some more data to application storage. It needs to be provided as request parameter file and can not exceed 10MB of size. File must be of type csv with column headers Datasource,Campaign,Daily,Clicks,Impressions Please note that during import all queries are blocked with service returning HTTP 503, even if initializing request receives a response.

AWS

For showoff purposes I created an AWS deployment of the service. It is available under: http://ec2-54-154-185-220.eu-west-1.compute.amazonaws.com:8080

Dev Decisions Background

• Dockerfile is prepared assuming that if you have Docker you also have Maven and JDK 11. Hopefully it is a fair assumption ;)
• Spring because extreme performance has not been considered a priority but optimal development speed has.
• Lombok because it's OOtB in Intellij and noone loves writing boilerplate
• opencsv because it does the job and I used it couple of times before
• Solution is built as DB-agnostic. This approach allows for (relatively) easy portability to any DB, while InMemory repository has proven extremely useful for TDD.
• In order to increase performance the repository works on shallow copies. It results in increased risk to data integrity. Since application is for demo purposes only and performs non-modifying queries I considered it acceptable.