Using Moshi with multiple input fields - kotlin

I have some JSON that looks like this:
{
"name" : "Credit Card",
"code" : "AUD",
"value" : 1000
}
and am using Moshi to unmarshall this into a data structure like:
data class Account(
#Json(name = "name")
val name: String,
#Json(name = "currency")
val currency: String,
#Json(name = "value")
val value: Int
)
Everything works well. However, I really would like to extract the currency and value parameters into a separate Money object. So my model looks more like:
data class Money(
#Json(name = "currency")
val currency: String,
#Json(name = "value")
val value: Int
)
data class Account(
#Json(name = "name")
val name: String,
#Json(name = "???")
val money: Money
)
The challenge I'm struggling with is how to annotate things so that the Money object can be given two different fields (currency and value) that come from the same level as the parent account.
Do I need to create an intermediate "unmarshalling" object called, say, MoshiAccount and then use a custom adapter to convert that to my real Account object?
I saw How to deseralize an int array into a custom class with Moshi? which looks close (except that in that case, the adapted object (VideoSize) only needs a single field as input... in my case, I need both currency and value)
Any thoughts or suggestions would be much appreciated. Thanks

Moshi's adapters can morph your JSON structure for you.
object ADAPTER {
private class FlatAccount(
val name: String,
val currency: String,
val value: Int
)
#FromJson private fun fromJson(json: FlatAccount): Account {
return Account(json.name, Money(json.currency, json.value))
}
#ToJson private fun toJson(account: Account): FlatAccount {
return FlatAccount(account.name, account.money.currency, account.money.value)
}
}
Don't forget to add the adapter to your Moshi instance.
val moshi = Moshi.Builder().add(Account.ADAPTER).add(KotlinJsonAdapterFactory()).build()
val adapter = moshi.adapter(Account::class.java)

Related

kotlin data class constructors not getting picked up

I am creating a data class in kotlin as such
data class User(val name: String, val age: Int)
{
constructor(name: String, age: Int, size: String): this(name, age) {
}
}
In my main function, I can access the objects as such:
fun main(){
val x = User("foo", 5, "M")
println(x.name)
println(x.age)
println(x.size) // does not work
}
My problem is that I can't get access to size.
What I am trying to do is, create a data class where top level params are the common items that will be accessed, and in the constructors, have additional params that fit certain situations. The purpose is so that I can do something like
// something along the lines of
if (!haveSize()){
val person = User("foo", 5, "M")
} else {
val person = User("foo", 5)
}
}
Any ideas?
In Kotlin you do not need separate constructors for defining optional constructor params. You can define them all in a single constructor with default values or make them nullable, like this:
data class User(val name: String, val age: Int, val size: String = "M")
fun main(){
val x = User("foo", 5, "L")
val y = User("foo", 5)
println(x.size) // "L" from call site
println(y.size) // "M" from default param
}
You can not access size variable, because this is from secondary construct, but we have alternative variant.
data class User(var name: String, var age: Int) {
var size: String
init {
size = "size"
}
constructor(name: String, age: Int, size: String) : this(name, age) {
this.size = size
}
}
In short, you want to have one property that can be one of a limited number of options. This could be solved using generics, or sealed inheritance.
Generics
Here I've added an interface, MountDetails, with a generic parameter, T. There's a single property, val c, which is of type T.
data class User(
val mountOptions: MountOptions,
val mountDetails: MountDetails<*>,
)
data class MountOptions(
val a: String,
val b: String
)
interface MountDetails<T : Any> {
val c: T
}
data class MountOneDetails(override val c: Int) : MountDetails<Int>
data class MountTwoDetails(override val c: String) : MountDetails<String>
Because the implementations MountDetails (MountOneDetails and MountTwoDetails) specify the type of T to be Int or String, val c can always be accessed.
fun anotherCaller(user: User) {
println(user.mountOptions.a)
println(user.mountOptions.b)
println(user.mountDetails)
}
fun main() {
val mt = MountOptions("foo", "bar")
val mountOneDetails = MountOneDetails(111)
anotherCaller(User(mt, mountOneDetails))
val mountTwoDetails = MountTwoDetails("mount two")
anotherCaller(User(mt, mountTwoDetails))
}
Output:
foo
bar
MountOneDetails(c=111)
foo
bar
MountTwoDetails(c=mount two)
Generics have downsides though. If there are lots of generic parameters it's messy, and it can be difficult at runtime to determine the type of classes thanks to type-erasure.
Sealed inheritance
Since you only have a limited number of mount details, a much neater solution is sealed classes and interfaces.
data class User(val mountOptions: MountOptions)
sealed interface MountOptions {
val a: String
val b: String
}
data class MountOneOptions(
override val a: String,
override val b: String,
val integerData: Int,
) : MountOptions
data class MountTwoOptions(
override val a: String,
override val b: String,
val stringData: String,
) : MountOptions
The benefit here is that there's fewer classes, and the typings are more specific. It's also easy to add or remove an additional mount details, and any exhaustive when statements will cause a compiler error.
fun anotherCaller(user: User) {
println(user.mountOptions.a)
println(user.mountOptions.b)
// use an exhaustive when to determine the actual type
when (user.mountOptions) {
is MountOneOptions -> println(user.mountOptions.integerData)
is MountTwoOptions -> println(user.mountOptions.stringData)
// no need for an 'else' branch
}
}
fun main() {
val mountOne = MountOneOptions("foo", "bar", 111)
anotherCaller(User(mountOne))
val mountTwo = MountTwoOptions("foo", "bar", "mount two")
anotherCaller(User(mountTwo))
}
Output:
foo
bar
111
foo
bar
mount two
This is really the "default values" answer provided by Hubert Grzeskowiak adjusted to your example:
data class OneDetails(val c: Int)
data class TwoDetails(val c: String)
data class MountOptions(val a: String, val b: String)
data class User(
val mountOptions: MountOptions,
val detailsOne: OneDetails? = null,
val detailsTwo: TwoDetails? = null
)
fun main() {
fun anotherCaller(user: User) = println(user)
val mt = MountOptions("foo", "bar")
val one = OneDetails(1)
val two = TwoDetails("2")
val switch = "0"
when (switch) {
"0" -> anotherCaller(User(mt))
"1" -> anotherCaller(User(mt, detailsOne = one))
"2" -> anotherCaller(User(mt, detailsTwo = two))
"12" -> anotherCaller(User(mt, detailsOne = one, detailsTwo = two))
else -> throw IllegalArgumentException(switch)
}
}

cant understand this kotlin companion object code

I went through this code example but i can't get it to run neither do understand what it does exactly.
data class Order(
val id: String,
val name: String,
val data:String
)
data class OrderResponse(
val id: String,
val name: String,
val data: String
) {
companion object {
fun Order.toOrderResponse() = OrderResponse(
id = id,
name = name,
data = data ?: "",
)
}
}
The function in the companion object extends Order with a help function to turn Order instances into OrderResponse instances. So for example like
val order = Order("a", "b", "c")
val orderResponse = order.toOrderResponse()

Kotlin Data class setter using class fields

I have a redis entity using a kotlin data class and its Id should be a combination of few other fields from the same class.
Can this be achieved by defining setter for Id field instead of computing it outside of data class?
#RedisHash("Game")
data class Game(
#Id
val generatedId: String = "Default_ID",
val name: String,
val location: String,
val homeTeam: String,
val awayTeam: String
)
// want something like this
var generatedId : String = "DEFAULT_ID"
get() = "${name}${location}"
// or even better
var generated_Id : String = "${name}${location}"
Did you try to do something like this?
#RedisHash("Game")
data class Game(
val name: String,
val location: String,
val homeTeam: String,
val awayTeam: String,
#Id
val generatedId: String = "${name}${location}"
)

What is the easiest way to print a Kotlin data class as compilable code?

I'd like to be able to turn an instance of a fairly simple Kotlin data class into a String that could be copy and pasted into a Kotlin file and would compile.
For example, given these data classes:
data class Parent(val name: String, val age: Int, val children: Set<Child>)
data class Child(val name: String, val age: Int)
I would like a function from any data class to String such that:
toCompilableString(
Parent("Joe", 34, setOf(Child("Amy", 4), Child("Bob", 7)))
)
would return
"""Parent("Joe", 34, setOf(Child("Amy", 4), Child("Bob", 7)))"""
Does such a thing exist?
We can override the behaviour of toString to output in the desired format:
fun main() {
var amy = Child(name="Amy",age=4)
var bob = Child(name="Bob",age=7)
var joe = Parent(name="Joe", age=34, children=setOf(amy, bob))
print(joe) // outputs "Parent("Joe", 34, setOf(Child("Amy", 4), Child("Bob", 7))"
}
data class Parent(val name: String, val age: Int, val children: Set<Child>) {
override fun toString() = "Parent(\"$name\", $age, setOf(${children.joinToString()})"
}
data class Child(val name: String, val age: Int) {
override fun toString() = "Child(\"$name\", $age)"
}
With the help of joinToString(), this will output in the format "Parent("Joe", 34, setOf(Child("Amy", 4), Child("Bob", 7))".
If you really love pain, there are reflection tools made specially for such things. I made a small function that will generate what you need:
fun dataClassToString(instance: Any) {
val sb = StringBuilder()
sb.append("data class ${instance::class.qualifiedName} (")
var prefix = ""
instance::class.memberProperties.forEach{
sb.append(prefix)
prefix = ","
sb.append("${it.name} = ${it.getter.call(instance)}")
}
sb.append(")")
println(sb.toString())
}
The only problem with this function is that for your parent class it generates the following:
data class Parent (age = 34,children = [Child(name=Amy, age=4), Child(name=Bob, age=7)],name = Joe)
Internally set is represented as array, however if you know for sure that you will have only sets or arrays, you can easily check what type it is and append that type when creating the set. You can also check if this is a data class and append it instead of hardcoded string.

Kotlin data class + Gson: optional field

I have the following data class in Kotlin:
import com.google.gson.annotations.SerializedName
data class RouteGroup(
#SerializedName("name") var name: String,
#SerializedName("id") var id: Int
)
Sometimes I need to create an object with both fields, sometimes with only one of them.
How can I do this?
EDIT
This is not the duplicate of this question: Can Kotlin data class have more than one constructor?
That question shows how to set a default value for a field. But in my case, I don't need to serialize the field with the default value. I want a field to be serialized only when I explicitly assign a value to it.
it is easy you have to use the nullable operator
import com.google.gson.annotations.SerializedName
data class RouteGroup #JvmOverloads constructor(
#SerializedName("name") var name: String? = null,
#SerializedName("id") var id: Int? = null
)
You may need something like this:
sealed class RouteGroup
data class RouteGroupWithName(
#SerializedName("name") var name: String
) : RouteGroup()
data class RouteGroupWithId(
#SerializedName("id") var id: Int
) : RouteGroup()
data class RouteGroupWithNameAndId(
#SerializedName("name") var name: String,
#SerializedName("id") var id: Int
) : RouteGroup()
EDIT 1:
Or you can use nullable fields and named parameters like this:
data class RouteGroup(
#SerializedName("name") var name: String? = null,
#SerializedName("id") var id: Int? = null
)
val routeGroupWithName = RouteGroup(name = "example")
val routeGroupWithId = RouteGroup(id = 2)
val routeGroupWithNameAndId = RouteGroup(id = 2, name = "example")