How can I enforce that a class has a companion object? - kotlin

What I want is this:
interface base {
abstract static fun foo()
}
class impl : base {
override static fun foo()
}
Normally, Kotlin solves problems using companion objects rather than static functions. But an interface can't define a requirement for a companion object with function. So how can I accomplish this? The code that uses this would look like
fun <T : base> bar() {
T.foo()
}
Any other way to get this behavior? Namely, that I can execute a function of a derivative of T, without knowing the specific type, and not assuming the derivative has a default constructor?
Edit
I was able to get this to do what I want by using value parameters of types that can be set on the companion objects of the classes I want to work with. An illustrative example of what I want to use this technique for.
import kotlin.reflect.full.*
interface DynamicBuilder {
fun build(sides: Int): Shape?
}
interface Shape {
companion object : DynamicBuilder {
override fun build(sides: Int) = null
}
}
abstract class Shape2D : Shape {
companion object : DynamicBuilder {
override fun build(sides: Int) = if(sides > 0) Square() else Circle()
}
}
abstract class Shape3D : Shape {
companion object : DynamicBuilder {
override fun build(sides: Int) = if(sides > 0) Cube() else Sphere()
}
}
class Square : Shape2D()
class Circle : Shape2D()
class Sphere : Shape3D()
class Cube : Shape3D()
fun Build(sides: Int, builder: DynamicBuilder): Shape? {
return builder.build(sides)
}
inline fun <reified T : Shape> Build(sides: Int): Shape? {
return Build(sides, T::class.companionObjectInstance as DynamicBuilder)
}
fun main() {
println(Build(0, Shape2D))
println(Build(4, Shape2D))
println(Build<Shape3D>(0))
println(Build<Shape3D>(6))
}
The goal is that I can create a new entire class of Shape, and have all the logic related to how it builds the concrete object contained in that file, rather than having some monolithic shared switch statement.

An interface can define a requirement for some object with function, and you can suggest it to be the companion object even if you can't force it to be.
interface BaseCompanion {
fun foo(): Unit
}
interface Base {
companion object : BaseCompanion {
fun foo() { println("in Base") }
}
fun companion(): BaseCompanion = Base
}
interface Derived : Base {
companion object : BaseCompanion {
fun foo() { println("in Derived") }
}
override fun companion() = Derived
}
// value parameter, not type parameter
fun bar(companion: BaseCompanion) {
companion.foo()
}
bar(Base)
bar(Derived)
The companion() function isn't actually used in this case, it's for when you want to access the companion from a Base instance:
fun baz(x: Base) {
x.companion().foo()
}
Another (unsafe) option is to define companion() using reflection.
fun companion() = this::class.companionObjectInstance as BaseCompanion
Plus: no need to explicitly override it in Derived; minuses: 1. will crash at runtime if you forget to create the companion or to extend BaseCompanion; 2. slower than non-reflection definition.

TL;TR:
How can I enforce that a class has a companion object?
You cannot.
Kotlin has no static methods. Even if it had them, they wouldn't be overridable, as they are not in Java. The same holds for companion objects. Kotlin code is eventually compiled to Java byte code, so what is not possible in Java won't be possible in Kotlin either.
Edit:
It's interesting to see what the compiler has to say about it. Consider the following snippet:
open class Base {
companion object {
fun test() {}
}
}
inline fun <reified T : Base> staticCall() {
T.test() // <-- ERROR
}
The error message:
Type parameter 'T' cannot have or inherit a companion object, so it cannot be on the left hand side of dot

Based on your updated question, it seems like what you want is usually achieved using the factory pattern. Alternatively you could also use dependency injection. There are many options without the usage of reflection.
Why shouldn't you use reflection?
There are a few reasons here and here and you can find more if you google it. Generally reflection was created for a specific purpose, to discover the functionality of a class that was unknown at compile time. You do not use it for this purpose, since your implementation requires you to know the class, in order to pass it as a reified generic parameter. If you do require to discover classes that you don't know at compile time, you can use dependency injection.
The simpler solution for your version is a factory pattern:
interface Shape
class Square : Shape
class Circle : Shape
class Sphere : Shape
class Cube : Shape
object ShapeFactory {
fun build2DShape(sides: Int): Shape {
if(sides > 0) Square() else Circle()
}
fun build3DShape(sides: Int): Shape {
if(sides > 0) Cube() else Sphere()
}
}
fun main() {
println(ShapeFactory.build2DShape(0))
println(ShapeFactory.build3DShape(0))
}
In short, Build<Shape3D>(0) is replaced by ShapeFactory.build3DShape(0). The caller still has to know that there are 3DShapes and where they are. The only thing that changed is that you do not require Reflection.
This requires the person calling the function to know of the existence of 2D and 3D shapes. Same as in your implementation with reflection. This way you can have all the logic how to create the shapes in the same file as the shapes. You could even make the factory call some functions in the companion object of the shape if you wish to do so. Your factory knows of the existence of those subclasses. But since you can put the factory in the same file as the subclasses, that doesn't split the logic to somewhere else.
If you want to delegate the deciding whether it is a 2D or a 3D shape to a subclass you can do the following:
interface Shape
class Square : Shape
class Circle : Shape
class Sphere : Shape
class Cube : Shape
object ShapeFactory {
fun build2DShape(sides: Int): Shape {
return if(sides > 0) Square() else Circle()
}
fun build3DShape(sides: Int): Shape {
return if(sides > 0) Cube() else Sphere()
}
}
fun getBuilder(dimensions: Int): (sides: Int) -> Shape {
if (dimensions == 2)
return ShapeFactory::build2DShape
else
return ShapeFactory::build3DShape
}
fun main() {
print (getBuilder(2)(3))
}

Related

Generics in Objects

I have a question about sealed class, generics and object.
Let's say I would like to model something like 3 finite cases with a sealed class something like this:
sealed class ChangeState<S> {
fun reduceState(state: S): S
}
data class SetState<S>(val newState: S) : ChangeState<S>() {
override fun reduce(state: S): S = newState
}
object NoStateChange : ChangeState<Nothing>() { // What do I specify here for ChangeState? Nothing?
override fun reduce(state: Nothing): Nothing {
throw Exception("This should never be called")
}
}
The goal is to provide a convenient way to define NoStateChange in a generic way that it can be used as following:
fun foo(i : Int) : ChangeState<Int> {
return if (i==0)
NoStateChange // Won't compile because return type is ChangeState<Nothing> but expected ChangeState<Int>
else
SetState(i)
}
Is there a way to do that with object and Generics somehow?
As pointed out by #Tenfour04 the issue is that out is needed but reduceState() would require in as well. However, reduceState() can be refactored out of the class hierarchy and moved to an extension function like that:
sealed class ChangeState<out S>
data class SetState<S>(val newState: S) : ChangeState<S>()
object NoStateChange : ChangeState<Nothing>()
fun <S> ChangeState<S>.reduce(state: S): S {
return when (val change = this) {
is SetState -> change.newState
is NoStateChange -> state
}
}

Kotlin class generics without duplication

Consider an abstract class:
abstract class PubSubSubscriber<T : Any>(private val topic: KClass<T>) : BackgroundFunction<PubSubMessage> {
abstract fun consume(payload: T)
override fun accept(message: PubSubMessage, context: Context) {
val json = String(Base64.getDecoder().decode(message.data.toByteArray()))
val payload = objectMapper.readValue(json, topic.java)
consume(payload)
}
}
And implementation:
class MySubscriber : PubSubSubscriber<Payload>(Payload::class) {
Is there a way to define such abstract class so that I don't have to repeat twice the Payload and Payload::class in the class definition?
Yes, with some reflection.
At construction time, we can extract the type parameter and assign it to a property that no longer needs to be given to the constructor:
abstract class PubSubSubscriber<T : Any> {
val topic: KClass<T> = extractTypeParam<T>(0).kotlin
private fun <X> extractTypeParam(paramIdx: Int): Class<X> {
require(PubSubSubscriber::class.java == javaClass.superclass) {
"PubSubSubscriber subclass $javaClass should directly extend PubSubSubscriber"
}
#Suppress("UNCHECKED_CAST")
return (javaClass.genericSuperclass as ParameterizedType).actualTypeArguments[paramIdx] as Class<X>
}
abstract fun consume(payload: T)
override fun accept(message: PubSubMessage, context: Context) {
val json = String(Base64.getDecoder().decode(message.data.toByteArray()))
val payload = objectMapper.readValue(json, topic.java)
consume(payload)
}
Note the following limitations:
A) this solution works only if MySubscriber directly extends from PubSubSubscriber. However, the given code can detect if that's not the case and warn about it (at runtime). In such cases, there are the following solutions:
MySubscriber falls back to providing a duplicate argument (essentially what you already had)
the direct superclass of MySubscriber can provide a similar detection mechanism
B) You call reflection code every time a MySubscriber instance is created. This may be too slow in certain contexts, but for many this is unproblematic.

Is there a standard Kotlin way to make just this one special-case object of a data class have a different toString() method?

Here's my attempt:
data class LineStyle(val thickness: Float) {
override fun toString() =
if (thickness == 0f) {
"NO_LINE"
} else {
"LineStyle(${thickness}f)"
}
companion object {
#JvmField
val NO_LINE = LineStyle(0f)
}
}
I'd rather override toString separately for the NO_LINE singleton, but don't see how. I have a lot of classes that have a zero instance.
There are a few different ways you could do this. You could use a sealed or open class as your actual LineStyle and have the NoLine as a companion object, forcing you to refer to it as LineStyle.NoLine. In that object, you could override the actual toString() method with whatever you want. The implementation would look something like this:
// sealed so it's easier to deal with in when statements
sealed class LineStyle(val thickness: Float) {
// defines a no line style
companion object NoLine: LineStyle(0f) {
override fun toString() = "NO_LINE"
}
}
fun main(args: Array<String>) {
// only able to access it as
LineStyle.NoLine
}
However, with this approach, you lose the data aspect of the class. The only other option is to do basically what you have done, instead, maybe use a when statement though for future extensibility. It's not the prettiest but it works:
data class LineStyle(val thickness: Float) {
// if you want extensibility, you could use a when statement
override fun toString() = if(this == NO_LINE) "NO_LINE" else super.toString()
companion object {
val NO_LINE = LineStyle(0f)
}
}

How kotlin delegation is useful?

I'm really confused about the kotlin delegation. Let me describe the regular polymorphism approach here which looks same like the kotlin delgation.
interface Base {
fun print()
}
class BaseImpl(val x: Int) : Base {
override fun print() { print(x) }
}
fun main(args: Array<String>) {
val b : Base = BaseImpl(10)
b.print() // prints 10
}
I can pass any implemented class of Base interface to b variable to call the method of specified class's object. Then what is the benefit of kotlin's delegation? Which is described here.
interface Base {
fun print()
}
class BaseImpl(val x: Int) : Base {
override fun print() { print(x) }
}
class Derived(b: Base) : Base by b // why extra line of code?
// if the above example works fine without it.
fun main(args: Array<String>) {
val b = BaseImpl(10)
Derived(b).print() // prints 10
}
I know this is the simple scenario where the both codes are working fine. There should be a benefit of delegation that's why kotlin introduced it. What is the difference? and how kotlin delegation can be useful? Please give me a working example to compare with polymorphism approach.
Also remember that you're not restricted to just one delegate. Kotlin's way of implementing delegation is similar to traits implementation in languages like Groovy. You can compose different functionality via delegates. Kotlin's way can also be considered more powerful because you can "plug in" different implementations too.
interface Marks {
fun printMarks()
}
class StdMarks() : Marks {
override fun printMarks() { println("printed marks") }
}
class CsvMarks() : Marks {
override fun printMarks() { println("printed csv marks") }
}
interface Totals {
fun printTotals()
}
class StdTotals : Totals {
override fun printTotals() { println("calculated and printed totals") }
}
class CheatTotals : Totals {
override fun printTotals() { println("calculated and printed higher totals") }
}
class Student(val studentId: Int, marks: Marks, totals: Totals)
: Marks by marks, Totals by totals
fun main(args:Array<String>) {
val student = Student(1,StdMarks(), StdTotals())
student.printMarks()
student.printTotals()
val cheater = Student(1,CsvMarks(), CheatTotals())
cheater.printMarks()
cheater.printTotals()
}
Output:
printed marks
calculated and printed totals
printed csv marks
calculated and printed higher totals
You can't do this with inheritance.
It is extremely useful for creating decorators and for object composition.
Joshua Bloch in Effective Java, 2nd Edition, Item 16 'Favor Composition Over Inheritance' shows a good example: inheritance is easy-to-break, and decorators are not.
Inheritance:
class LoggingList<E> : ArrayList<E>() {
override fun add(e: E): Boolean {
println("added $e")
return super.add(e)
}
override fun addAll(e: Collection<E>): Boolean {
println("added all: $e")
return super.addAll(e) // oops! Calls [add] internally.
}
}
Delegation:
class LoggingList<E>(delegate: MutableList<E>) : MutableList<E> by delegate {
override fun add(e: E): Boolean {
println("added $e")
return delegate.add(e)
}
override fun addAll(e: Collection<E>): Boolean {
println("added all: $e")
return delegate.addAll(e) // all OK
// it calls [delegate]'s [add] internally, not ours
}
}
It is useful because of the Delegation Pattern where most of the behavior can be the same as the target of the delegation (b) but you just want to override a subset of methods to act differently.
An example would be an InputStream implementation which delegates all work to another InputStream but overrides the close() method to not close the underlying stream. This could be implemented as:
class CloseGuardInputStream(private val base: InputStream)
: InputStream by base {
override fun close() {}
}
Following is the example :-
interface Mode{
val color:String
fun display()
}
class DarkMode(override val color:String) : Mode{
override fun display(){
println("Dark Mode..."+color)
}
}
class LightMode(override val color:String) : Mode {
override fun display() {
println("Light Mode..."+color)
}
}
class MyCustomMode(val mode: Mode): Mode{
override val color:String = mode.color
override fun display() {
mode.display()
}
}
Now, the custom mode can reuse display() function of both modes DarkMode & LightMode
fun main() {
MyCustomMode(DarkMode("CUSTOM_DARK_GRAY")).display()
MyCustomMode(LightMode("CUSTOM_LIGHT_GRAY")).display()
}
/* output:
Dark Mode...CUSTOM_DARK_GRAY
Light Mode...CUSTOM_LIGHT_GRAY
*/
Kotlin natively support delegation pattern.
Kotlin provides by keyword to specify the delegate object which our custom mode will be delegating to.
We can achieve the same result of the code above using by keyword.
class MyCustomMode(val mode: Mode): Mode by mode
fun main() {
MyCustomMode(DarkMode("CUSTOM_DARK_GRAY")).display()
MyCustomMode(LightMode("CUSTOM_LIGHT_GRAY")).display()
}
/* output:
Dark Mode...CUSTOM_DARK_GRAY
Light Mode...CUSTOM_LIGHT_GRAY
*/

How to specify "own type" as return type in Kotlin

Is there a way to specify the return type of a function to be the type of the called object?
e.g.
trait Foo {
fun bar(): <??> /* what to put here? */ {
return this
}
}
class FooClassA : Foo {
fun a() {}
}
class FooClassB : Foo {
fun b() {}
}
// this is the desired effect:
val a = FooClassA().bar() // should be of type FooClassA
a.a() // so this would work
val b = FooClassB().bar() // should be of type FooClassB
b.b() // so this would work
In effect, this would be roughly equivalent to instancetype in Objective-C or Self in Swift.
There's no language feature supporting this, but you can always use recursive generics (which is the pattern many libraries use):
// Define a recursive generic parameter Me
trait Foo<Me: Foo<Me>> {
fun bar(): Me {
// Here we have to cast, because the compiler does not know that Me is the same as this class
return this as Me
}
}
// In subclasses, pass itself to the superclass as an argument:
class FooClassA : Foo<FooClassA> {
fun a() {}
}
class FooClassB : Foo<FooClassB> {
fun b() {}
}
You can return something's own type with extension functions.
interface ExampleInterface
// Everything that implements ExampleInterface will have this method.
fun <T : ExampleInterface> T.doSomething(): T {
return this
}
class ClassA : ExampleInterface {
fun classASpecificMethod() {}
}
class ClassB : ExampleInterface {
fun classBSpecificMethod() {}
}
fun example() {
// doSomething() returns ClassA!
ClassA().doSomething().classASpecificMethod()
// doSomething() returns ClassB!
ClassB().doSomething().classBSpecificMethod()
}
You can use an extension method to achieve the "returns same type" effect. Here's a quick example that shows a base type with multiple type parameters and an extension method that takes a function which operates on an instance of said type:
public abstract class BuilderBase<A, B> {}
public fun <B : BuilderBase<*, *>> B.doIt(): B {
// Do something
return this
}
public class MyBuilder : BuilderBase<Int,String>() {}
public fun demo() {
val b : MyBuilder = MyBuilder().doIt()
}
Since extension methods are resolved statically (at least as of M12), you may need to have the extension delegate the actual implementation to its this should you need type-specific behaviors.
Recursive Type Bound
The pattern you have shown in the question is known as recursive type bound in the JVM world. A recursive type is one that includes a function that uses that type itself as a type for its parameter or its return value. In your example, you are using the same type for the return value by saying return this.
Example
Let's understand this with a simple and real example. We'll replace trait from your example with interface because trait is now deprecated in Kotlin. In this example, the interface VitaminSource returns different implementations of the sources of different vitamins.
In the following interface, you can see that its type parameter has itself as an upper bound. This is why it's known as recursive type bound:
VitaminSource.kt
interface VitaminSource<T: VitaminSource<T>> {
fun getSource(): T {
#Suppress("UNCHECKED_CAST")
return this as T
}
}
We suppress the UNCHECKED_CAST warning because the compiler can't possibly know whether we passed the same class name as a type argument.
Then we extend the interface with concrete implementations:
Carrot.kt
class Carrot : VitaminSource<Carrot> {
fun getVitaminA() = println("Vitamin A")
}
Banana.kt
class Banana : VitaminSource<Banana> {
fun getVitaminB() = println("Vitamin B")
}
While extending the classes, you must make sure to pass the same class to the interface otherwise you'll get ClassCastException at runtime:
class Banana : VitaminSource<Banana> // OK
class Banana : VitaminSource<Carrot> // No compiler error but exception at runtime
Test.kt
fun main() {
val carrot = Carrot().getSource()
carrot.getVitaminA()
val banana = Banana().getSource()
banana.getVitaminB()
}
That's it! Hope that helps.
Depending on the exact use case, scope functions can be a good alternative. For the builder pattern apply seems to be most useful because the context object is this and the result of the scope function is this as well.
Consider this example for a builder of List with a specialized builder subclass:
open class ListBuilder<E> {
// Return type does not matter, could also use Unit and not return anything
// But might be good to avoid that to not force users to use scope functions
fun add(element: E): ListBuilder<E> {
...
return this
}
fun buildList(): List<E> {
...
}
}
class EnhancedListBuilder<E>: ListBuilder<E>() {
fun addTwice(element: E): EnhancedListBuilder<E> {
addNTimes(element, 2)
return this
}
fun addNTimes(element: E, times: Int): EnhancedListBuilder<E> {
repeat(times) {
add(element)
}
return this
}
}
// Usage of builder:
val list = EnhancedListBuilder<String>().apply {
add("a") // Note: This would return only ListBuilder
addTwice("b")
addNTimes("c", 3)
}.buildList()
However, this only works if all methods have this as result. If one of the methods actually creates a new instance, then that instance would be discarded.
This is based on this answer to a similar question.
You can do it also via extension functions.
class Foo
fun <T: Foo>T.someFun(): T {
return this
}
Foo().someFun().someFun()