How to Set up your Own Dependency Injection with Reader

Photo by Frank Wang

There are many ways to create dependency injections. Dependency injection helps decoupled one object from another, bypassing a dependency to a framework or an object during run time. Therefore, the client doesn’t need to find what dependency needs to supply to the object or framework - instead, the system tells the client what dependency they need to supply to the object.

In Scala, there are many ways to create dependency injection. Today, I will talk about using Reader as a Monad in Cats for creating dependency inject, and how Reader can help you quickly create configuration validation for your system.

Imagine if you need to create a validation system that can easily pick and choose various kinds of operations for validating configuration in a system. You can create a method to validate each component in the configuration.

def isCorrectEmail(id:Long, email:String, repo: Repository) : Boolean = {
  if(repo.findId(id)) repo.getId(id) ....
}

In this case, if one day a new configuration attribute is needed to validate, or a specialized check from the configuration to make it pass, you will need to go to the code and change the methods.

With more changes in the configuration file, you need to go to the file and change the code each time. Moreover, every single check is coupled with one another, making it hard to update in the future.

Reader to the rescue

Cats Reader Data type can help you chain multiple operations together, and produce a significant computation that accepts a configuration as one parameter. It would be best if you inputted the configuration, and it runs as we specified.

Let’s take a look at the definition of Reader[A, B]:

object Reader {
  def apply[A,B](f:A => B) : Reader[A,B] =  ReaderT[Id,A,B]
}

It meant when you create a Reader[A, B], it gets callback function of A type and returns B type.

Example of creating a Reader:

case class Cat(sound: String)

// extracting the sound from the Cat class
val retrieveSound : Reader[Cat, String] = Reader {cat => 
  cat.sound
}

When you instantiate the Reader type, you pass in a callback function of the configuration that you are receiving, in this case Cat, and retrieve one of the attributes inside, in this case, cat.sound.

After you define the Reader type, you can call it by running the method run and supply the configuration as a parameter of run:

// calling the cat
println(retrieveSound.run(Cat("meow"))) // meow

If you read until this point, you probably pause for a moment and think, “What does Reader type Monad so especially with just regular validation raw function?”

The answer, my friend, is that the Monad type is what makes it unique.

With being a Monad, the Reader type can call map and flatMap to chain operation, and modify the operation from an existing Reader.

Usually, you create a set of Reader that will accept the same type of configuration. Then, you can combine and chain them with flatMap and map, and then use run and supply that configuration at the end.

map

The map method modifies the computation inside the Reader, bypassing the function to the result of the callback function.

// from previous Cat example
val checkSound: Reader[Cat, Boolean] = retrieveSound.map(_ == "meow")

checkSound.run(Cat("bark")) // false

flatMap

The flatMap operation is what makes Reader so powerful. It helps combine operations that depend on the same input type.

val greet:Reader[Cat, String] = Reader{cat =>
  s"hello $cat"
}

val sound:Reader[Cat, String] = Reader{cat =>
  s"${cat.sound} ${cat.sound}"
}

val greetAndSound = for{
  g <- greet
  check <- checkSound
  s <- sound
} yield {
  if(check) g + s else "sound is not right"
}

val result = greetAndSound.run(Cat("meow"))
println(result)
val notRight = greetAndSound.run(Cat("bark"))
println(notRight)

Reader in Action

Reader can set up all your operations and inject your operation with a configuration as a parameter.

Let’s create a validation check function, given an id and an email, check whether there is an email corresponding to the id in the database. Let’s create our Repository case class:

case class Repository(userDB:Map[Long, String], emailDB: Map[String, List[String]])

userDB maps id, ‘Long, to username,String. emailDB` maps username to a list of email.

First, we need to create a Reader type getUser that gets an id and retrieve the username.

def getUser(id:Long):Reader[Repository, Option[String]] = Reader{repo =>
    repo.userDB.get(id)
  }

Then, we create getEmail which will get the username and return a list of email.

def getEmail(username:Option[String]): Reader[Respository, List[String]] = Reader {repo =>
    repo.emailDB.getOrElse(username.fold("none")(st => st), List.empty[String])
  }

Lastly, we can have Reader that checkIfEmailMatch that takes in a parameter of id and email and check if the id contains the email address. With this method, we combine getUser method and getEmail. Then, we check if the List of emails matches the email that is input in the parameter.

def checkIfEmailMatch(id:Long, email:String):UserReader[Boolean] = for {
    usernameOption <- getUser(id)
    emails <- getEmail(usernameOption)
  } yield {

    emails.contains(email)
  }

How do you execute checkIfEmailMatch?

Let’s set up the main method:

val userDB = Map(
    1L -> "john",
    2L -> "jane",
    3L -> "kate"
)
val emailDB = Map(
  "john" -> List("[email protected]"),
  "jane" -> List("[email protected]", "[email protected]"),
  "kate" -> List("[email protected]", "[email protected]")
)

val repo = Repository(userDB,emailDB)
val res = checkIfEmailMatch(2L, "[email protected]").run(repo)
println(res)

We set up all the operations, started from getUser, then getEmail. We, then, combine the two operations in checkIfEmail exist. These all take in 1 configuration file, which is the repo. When you call the run method, it executes the operation.

Takeaway

  • Reader provides a tool for dependency injection - by setting up all the operation that takes in the same configuration.
  • Readers are most useful when we want to construct a batch program that can easily represent as function; defer injection of a known parameter, and isolate the parts of the program that we want to test.
  • By setting your system as readers, you can represent your readers as a pure function, and do map or flatMap for combining them.

Like this Article?

Sign up for my newsletter to get notified for new articles!


Related Posts

5 Anti Pattern for Writing Code in a Functional Programming Language

No 1. Nested Asynchronous Function

Use This Mantra to Decide whether You Want to go to Big Tech or a Startup

If you want to Go Deep, Go for Big Tech. If you want to Go Wide, Go for Startup.

Why Do Functional Programmers Prefer For-Comprehension Over Imperative Code Block

Short Answer - Readability

Does Seniority Title Really Matters in Tech

The answer is always - depends

An Interview Question that Truly Tests your Experience as a Software Engineer

questions that is not based on Leetcode