How to describe event structures and obtain information? - kotlin

Question

I want to write a DSL in kotlin that describes events in an event store to generate Java code, JSON examples, schema and documentation based on these descriptions. My current approach is:
A data class MyEvent that holds the structure of a event
data class MyEvent(val name: String, val version: Int, val content: Map<String, String>)
so that I can describe events like
val OrderCreated = MyEvent("OrderCreated", 1, mapOf("orderId" to "UUID", "nameOfProduct" to "String(1, 256)", "quantity" to "Integer"))
val OrderCancelled = MyEvent("OrderCancelled", 2, mapOf("orderId" to "UUID", "reason" to "String(100, 1000)"))
val OrderQuestioned = MyEvent("OrderQuestioned", 3, mapOf("orderId" to "UUID", "question" to "String(10, 1000)"))
Whereas "String(1, 256)" means it is of type String with a minimum of 1 and a maximum of 256 characters.
To iterate over all events to generate everything I want, I need to manually add each event to a list / set
fun scanForAllMyEventsInstances(): Set<MyEvent> {
return hashSetOf(OrderCreated, OrderCancelled, OrderQuestioned)
}
This way doesn't feel like the best way.
I don't like to be forced to add new events in scanForAllMyEventsInstances. I only want to describe a new event at one point.
I don't like to create an instance of MyEvent for each event. It's all "static" information.
So my question: How would you do that? I'd like to have some suggestions.
I use SpringBoot at the moment. So don't hesitate to suggest frameworks.

Answer 1

I think enum class is your friend here.
enum class MyEvent(val eventName: String, val version: Int, val content: Map<String, String>) {
OrderCreated("OrderCreated", 1, mapOf("orderId" to "UUID", "nameOfProduct" to "String(1, 256)", "quantity" to "Integer")),
OrderCancelled("OrderCancelled", 2, mapOf("orderId" to "UUID", "reason" to "String(100, 1000)")),
OrderQuestioned("OrderQuestioned", 3, mapOf("orderId" to "UUID", "question" to "String(10, 1000)"))
}
And you can iterate over instances
MyEvent.values().forEach { println(it.eventName) }

Answer 2

With the approach of enum classes, I added an interface so that I can organize plenty of (different) events
interface MyEvent2 {
val eventName: String
val version: Int
val content: Map<String, String>
}
enum class MyOrderEvents : MyEvent2 {
OrderCreated {
override val eventName = "OrderCreated"
override val version = 1
override val content = mapOf("orderId" to "UUID", "nameOfProduct" to "String(1, 256)", "quantity" to "Integer")
},
OrderCancelled {
override val eventName = "OrderCancelled"
override val version = 2
override val content = mapOf("orderId" to "UUID", "reason" to "String(100, 1000)")
},
OrderQuestioned {
override val eventName = "OrderQuestioned"
override val version = 3
override val content = mapOf("orderId" to "UUID", "question" to "String(10, 1000)")
}
}
enum class MyUserEvents : MyEvent2 {
UserCreated {
override val eventName = "UserCreated"
override val version = 1
override val content = mapOf("userId" to "UUID", "name" to "String(1, 256)")
},
UserDeleted {
override val eventName = "UserDeleted"
override val version = 2
override val content = mapOf("userId" to "UUID")
},
UserChanged {
override val eventName = "UserChanged"
override val version = 3
override val content = mapOf("userId" to "UUID", "newName" to "String(1, 256)")
}
}
I still need to add each enum class to a list/set.
fun scanForAllMyEventsInstances(): Set<MyEvent2> {
val allEvents = HashSet<MyEvent2>()
allEvents.addAll(MyOrderEvents.values())
allEvents.addAll(MyUserEvents.values())
return allEvents
}
fun main() {
scanForAllMyEventsInstances().forEach{e -> println(e.eventName)}
}
But only once per class. Each new event within that class is automatically recognized. That's fine for me.
remark: The val name should not be used when using an enum class because in enum classes the value name is already defined. That's why in MyEvent2 it's called eventName. Maybe I even can skip name / eventName because it's 1:1 to the enum name.
I will have a try with this.
PS: It's still not a nice DSL, I think. But optimizing this, I will do in a future task.