Reputation: 1988
we have
private def insertUpdateDeleteFormDsList(dsList : List[FormDefinitionDataSourceRequestModel])(implicit formDefId:Int,subject:Subject,session: Session) : (List[(Int,Int)],Seq[FormDefinitionDataSourceRequestModel],Seq[FormDefinitionDataSourceRequestModel]) = {
val incomingIds = dsList.map( ds => (ds.dataSourceId,ds.dsTypeId) )
val existingIds = formDefinitionDatasources.filter(_.tenantId === subject.tenantId).filter(_.formDefId === formDefId).map( ds => (ds.dataSourceId,ds.dataSourceTypeId) ).list
val idsForDeletion = existingIds diff incomingIds
val idsForInsertion = incomingIds diff existingIds
val idsForUpdate = existingIds diff idsForDeletion
val insertList = dsList.flatMap{ t => idsForInsertion collectFirst{ case (dsId,dsType) if(dsId == t.dataSourceId && dsType == t.dsTypeId)=> t} }
val updateList = dsList.flatMap{t=>idsForUpdate collectFirst {case (dsId,dsType) if(dsId == t.dataSourceId && dsType == t.dsTypeId)=> t}}
(idsForDeletion,updateList,insertList)
}
And other similar methods as
private def insertUpdateDelDataInstances(instances: List[Instance])(implicit subject: Subject, session: Session): (Seq[Instance], Seq[Instance], Seq[Instance]) = {
val incomingIds = instances.map(_.id)
val existingIds = dataSourceInstanceNew.filter(_.tenantId === subject.tenantId).map(_.id).list
val idsForDeletion = existingIds diff incomingIds
val idsForInsertion = incomingIds diff existingIds
val idsForUpdate = existingIds diff idsForDeletion
val deleteList = instances.flatMap{ t => idsForDeletion collectFirst{ case id if(id == t.id)=> t} }
val insertList = instances.flatMap{ t => idsForInsertion collectFirst{ case id if(id == t.id)=> t} }
val updateList = instances.flatMap{t=>idsForUpdate collectFirst {case id if(id === t.id)=> t}}
(deleteList,updateList,insertList)
}
There are similar method occurrences elsewhere. Every time List[T]
will be passed as a method argument where T
is always a case class
. now how val incomingIds
gets constructed is dependent on specific case class
attributes.
We want to create a generalized function which can accept a List[T]
and possibly incomingIds
and return a desired tuple to avoid writing similar looking boilerplate every time.
If say logic was to "always" use T
case class
's id
attribute then I can easily create a parent trait
with id
and have all case class
es mixin the trait - but that's not the case here. preparing val incomingIds
depend on different case class
attributes depending on where in the code they are called from.
Illustrating below
def generalizedInsertUpdateDeleteList[T](data:List[T],incomingIds:List[Int], existingIds:List[Int] )(implicit subject: Subject, session:Session) = {
val idsForDeletion = existingIds diff incomingIds
val idsForInsertion = incomingIds diff existingIds
val idsForUpdate = existingIds diff idsForDeletion
/*
//what's the best way to generalize comparison inside collectFirst?
//to use case class attribute names from `T`. Was thinking if Structural type can help but not sure if that
//can quite work unless there is a way to pass in arguments in a structural type?
val deleteList = data.flatMap{ t => idsForDeletion collectFirst{ case id if(id == t.id)=> t} }
val insertList = data.flatMap{ t => idsForInsertion collectFirst{ case id if(id == t.id)=> t} }
val updateList = data.flatMap{ t => idsForUpdate collectFirst {case id if(id === t.id)=> t}}
*/
Can shapeless help here if no other cleaner way to achieve this using standard scala/scalaz APIs?
Upvotes: 2
Views: 121
Reputation: 9820
You could create a type class with a PartialFunction
which can be used inside collectFirst
.
trait IUD[T, IdType] {
// returns a partial function which will be used in collectFirst
def collectId(t: T): PartialFunction[IdType, T]
}
We can create IUD
instances for your two methods :
// I chose (Long, Long) as type of (ds.dataSourceId,ds.dsTypeId)
type FormModel = FormDefinitionDataSourceRequestModel
implicit object FormModelIUD extends IUD[FormModel, (Long, Long)] {
def collectId(t: FormModel): PartialFunction[(Long, Long), FormModel] = {
case (dsId,dsType) if(dsId == t.dataSourceId && dsType == t.dsTypeId) => t
}
}
implicit object InstanceIUD extends IUD[Instance, Int] {
def collectId(t: Instance): PartialFunction[Int, Instance] = {
case id if id == t.id => t
}
}
We can use the IUD
type class in your generalizedInsertUpdateDeleteList
method :
def generalizedIUDList[T, IdType](
data: List[T], incomingIds: List[IdType], existingIds: List[IdType]
)(implicit
subject: Subject, session: Session, iud: IUD[T, IdType]
) = {
val idsForDeletion = existingIds diff incomingIds
val idsForInsertion = incomingIds diff existingIds
val idsForUpdate = existingIds diff idsForDeletion
def filterIds(ids: List[IdType]) =
data.flatMap(instance => ids collectFirst(iud.collectId(instance)) )
val deleteList = filterIds(idsForDeletion)
val insertList = filterIds(idsForInsertion)
val updateList = filterIds(idsForUpdate)
(deleteList,updateList,insertList)
}
Upvotes: 1
Reputation: 10428
collectFirst
accepts a PartialFunction
, in your case PartialFunction[Int, T]
I think?
You could pass the partial functions in as parameters to your generalizedInsertUpdateDeleteList
method, then only these would need to be defined each time.
Upvotes: 0
Reputation: 139048
Shapeless's records provide a type-safe way to abstract over case classes with a particular member name. For example:
import shapeless._, ops.record.Selector
def getId[A, R <: HList](a: A)(implicit
gen: LabelledGeneric.Aux[A, R],
sel: Selector[R, Witness.`'id`.T]
): sel.Out = sel(gen.to(a))
And then:
scala> case class Foo(id: String)
defined class Foo
scala> case class Bar(id: Int, name: String)
defined class Bar
scala> getId(Foo("12345"))
res0: String = 12345
scala> getId(Bar(123, "bar"))
res1: Int = 123
If you need to constrain the type of the id
member, you can use Selector.Aux
:
def getIntId[A, R <: HList](a: A)(implicit
gen: LabelledGeneric.Aux[A, R],
sel: Selector.Aux[R, Witness.`'id`.T, Int]
): Int = sel(gen.to(a))
Now getIntId(Bar(123, "bar"))
will compile, but getIntId(Foo("12345"))
won't.
Upvotes: 3