I have 4 methods with one logic and 4 possible type mapping:
def convertStringToString(in: String): String = ???
def convertIntToString(in: Int): String = ???
def convertIntToInt(in: Int): Int = ???
def convertStringToInt(in: String): Int = ???
I want to generalize input and output type and write logic in one methods. Tried to generelize input parameter:
def convertToInt[IN](in: IN): Int = in match {
case x: String if x.forall(_.isDigit) => x.toInt
case y: Int => y
case _ => 0
}
def convertToString[IN](in: IN): String = convertToInt[IN](in).toString
Could you help me to generalize second:
def convertToInt[IN, OUT](in: IN): OUT = ???
If you really wanted to, you could have something typeclass-based:
def convert[I, O](in: I)(implicit c: ConversionRule[I, O]): O = {
if (c.isConvertible(in)) c.convert(in)
else c.zero
}
trait ConversionRule[I, O] {
def isConvertible(in: I): Boolean
def convert(in: I): O
def zero: O // Could possibly derive the zero from, e.g., a cats Monoid instance where such exists
}
The eagle-eyed may notice that the isConvertible/convert methods match the contract of PartialFunction[I, O]'s isDefinedAt/apply, so may as well just use PartialFunction (and rewrite convert with isDefinedAt/apply)
trait ConversionRule[I, O] extends PartialFunction[I, O] {
def zero: O
}
zero can be implemented in terms of PartialFunction.applyOrElse, but for the case where zero is constant (which is the case where referential transparency is preserved), this is much faster.
Smart constructors can be defined:
object ConversionRule {
def apply[I, O](zeroValue: O)(pf: PartialFunction[I, O]): ConversionRule[I, O] =
new ConversionRule[I, O] {
override def apply(i: I): O = pf(i)
override def isDefinedAt(i: I): Boolean = pf.isDefinedAt(i)
val zero: O = zeroValue
}
def totalConversion[I, O](f: I => O): ConversionRule[I, O] =
new ConversionRule[I, O] {
override def apply(i: I) = f(i)
override def isDefinedAt(i: I) = true
override def zero: O = throw new AssertionError("Should not call since conversion is defined")
}
// Might want to put this in a `LowPriorityImplicits` trait which this object extends
implicit def identityConversion[I]: ConversionRule[I, I] =
totalConversion(identity)
}
identityConversion means that a convertIntToInt gets automatically generated.
convertStringToInt can then be defined as
implicit val stringToIntConversion = ConversionRule[String, Int](0) {
case x if x.forAll(_.isDigit) => x.toInt
}
One can define a toString based conversion (basically the non-lawful Show proposed for alleycats):
implicit def genericToString[I]: ConversionRule[I, String] =
ConversionRule.totalConversionRule(_.toString)
And it should then be possible to define a stringViaInt ConversionRule derivation like:
implicit def stringViaInt[I, O](implicit toInt: ConversionRule[I, Int]): ConversionRule[I, String] =
convert(convert(in)(toInt))
The only really useful thing this provides is an opt-in to usage of implicit conversions. Whether that's enough of a gain to justify? shrug
(Disclaimer: only the scala compiler in my head has attempted to compile this)
Edit:
Last revision was deemed unhelpful as it did not contain necessary information that help narrow down my issue. hence the need to also include the AST.
Below is a library in its entirety that allows parsing and writing of play-json's json based on user defined schema; Similar to what Scala's slick offers for database columns to some extent:
import scala.language.higherKinds
import play.api.libs.functional.syntax._
import play.api.libs.json._
import scala.language.{higherKinds, implicitConversions}
type PathNodes = List[PathNode]
sealed trait Field[A] {
def pathNodes: PathNodes
def jsPath: JsPath = JsPath(pathNodes)
def relativePath: JsPath = JsPath(List(pathNodes.last))
def format: Format[A]
def nestedFormatter(path: JsPath): OFormat[A]
def nestedFormat: OFormat[A] = nestedFormatter(relativePath)
}
case class PlainField[A: Format](prefix: PathNodes) extends Field[A] {
override def pathNodes: PathNodes = prefix
def format: Format[A] = implicitly[Format[A]]
override def nestedFormatter(path: JsPath): OFormat[A] = path.format(format)
}
abstract class JsonSchema[T](val _prefix: PathNodes) extends Field[T] with SchemaExtensionMethods {
override def pathNodes: PathNodes = _prefix
def format: OFormat[T]
protected def plain[A: Format](name: String): PlainField[A] = PlainField[A](_prefix :+ KeyPathNode(name))
protected def nested[N](name: String, factory: PathNodes => N): N = factory(_prefix :+ KeyPathNode(name))
protected def nested[B, G <: JsonSchema[B]](name: String)(implicit sm: HasJsonSchema[B, G]): G = sm.apply(_prefix :+ KeyPathNode(name))
override def nestedFormatter(path: JsPath): OFormat[T] = path.format(format)
}
case class Optional[F, A](field: F)(implicit ev: F <:< Field[A]) extends Field[Option[A]] {
override def pathNodes: PathNodes = field.pathNodes
override def format: Format[Option[A]] = {
implicit val writes: Writes[Option[A]] = JsPath.writeNullable(field.format)
implicit val reads: Reads[Option[A]] = JsPath.readNullable(field.format)
implicitly[Format[Option[A]]]
}
def map[G, B](f: F => G)(implicit ev: G <:< Field[B]): Optional[G, B] = new Optional[G, B](f(field))
def flatMap[G <: Field[B], B](f: F => Optional[G, B]): Optional[G, B] = f(field)
override def nestedFormatter(path: JsPath): OFormat[Option[A]] = path.formatNullable(field.format)
}
case class Collection[F, A](field: F)(implicit ev: F <:< Field[A], repath: Repath[F]) extends Field[Seq[A]] {
override def pathNodes: PathNodes = field.pathNodes
override def format: Format[Seq[A]] = {
implicit val writes: Writes[Seq[A]] = Writes.seq(field.format)
implicit val reads: Reads[Seq[A]] = Reads.seq(field.format)
implicitly[Format[Seq[A]]]
}
def apply(idx: Int): F = implicitly[Repath[F]].apply(field, IdxPathNode(idx))
override def nestedFormatter(path: JsPath): OFormat[Seq[A]] = path.format(format)
}
class FormatExtensionMethods[T](val arg: T) {
def <>[A, B, Fun](apply: Fun, unapply: B => Option[A])(implicit jss: JsonShape[A, B, T, Fun]): OFormat[B] = jss.format(arg, apply, unapply andThen (_.get))
}
class FieldExtensionMethods[F](val field: F) {
def optional[A](implicit ev: F <:< Field[A]): Optional[F, A] = new Optional[F, A](field)
def sequence[A](implicit ev: F <:< Field[A], repath: Repath[F]): Collection[F, A] = new Collection[F, A](field)
}
trait SchemaExtensionMethods {
implicit def formatExtensionMethods[M](t: M): FormatExtensionMethods[M] = new FormatExtensionMethods[M](t)
implicit def fieldExtensionMethods[M, A](t: M): FieldExtensionMethods[M] = new FieldExtensionMethods[M](t)
}
trait Repath[F] {
def apply(f: F, node: PathNode): F
}
object Repath {
implicit def plain[T]: Repath[PlainField[T]] = new Repath[PlainField[T]] {
override def apply(t: PlainField[T], node: PathNode): PlainField[T] =
PlainField[T](t.pathNodes :+ node)(t.format)
}
implicit def schema[S <: JsonSchema[_]](implicit sm: HasJsonSchema[_, S]): Repath[S] = new Repath[S] {
override def apply(t: S, node: PathNode): S =
sm.apply(t.pathNodes :+ node)
}
implicit def option[F <: Field[T] : Repath, T]: Repath[Optional[F, T]] = new Repath[Optional[F, T]] {
override def apply(t: Optional[F, T], node: PathNode): Optional[F, T] =
new Optional[F, T](implicitly[Repath[F]].apply(t.field, node))
}
implicit def sequence[F <: Field[T] : Repath, T]: Repath[Collection[F, T]] = new Repath[Collection[F, T]] {
override def apply(t: Collection[F, T], node: PathNode): Collection[F, T] =
new Collection[F, T](implicitly[Repath[F]].apply(t.field, node))
}
}
trait JsonShape[A, B, -T, Func] {
def format(t: T, apply: Func, unapply: B => A): OFormat[B]
}
object JsonShape {
type F[T] = Field[T]
implicit def cc1[A, B]: JsonShape[A, B, F[A], (A) => B] = (t: F[A], apply: (A) => B, unapply: B => A) => {
val name = t.pathNodes.last.asInstanceOf[KeyPathNode].key
OFormat[B](
Reads[B](jsv => (jsv \ name).validate[A](t.format).map(apply)),
OWrites[B](b => JsObject(Map(name -> Json.toJson(unapply(b))(t.format))))
)
}
implicit def cc2[T1, T2, B]: JsonShape[(T1, T2), B, (F[T1], F[T2]), (T1, T2) => B] = (t: (F[T1], F[T2]), apply: (T1, T2) => B, unapply: B => (T1, T2)) => {
(
t._1.nestedFormat and
t._2.nestedFormat
) (apply, unapply)
}
implicit def cc3[T1, T2, T3, B]: JsonShape[(T1, T2, T3), B, (F[T1], F[T2], F[T3]), (T1, T2, T3) => B] = (t: (F[T1], F[T2], F[T3]), apply: (T1, T2, T3) => B, unapply: B => (T1, T2, T3)) => {
(
t._1.nestedFormat and
t._2.nestedFormat and
t._3.nestedFormat
) (apply, unapply)
}
//this goes up to 22
}
abstract class HasJsonSchema[T, +S <: JsonSchema[T]](val apply: PathNodes => S) extends OFormat[T] {
val root: S = apply(Nil)
def format: OFormat[T] = root.format
def writes(o: T): JsObject = root.format.writes(o)
def reads(json: JsValue): JsResult[T] = root.format.reads(json)
}
Now let's write a small piece of client code that reproduce the issue:
case class MessageSchema(prefix: PathNodes) extends JsonSchema[Message](prefix) {
def underlying = plain[String]("underlying")
//def underlying = plain[String]("underlying").optional if I wanted the field to be Option[String]
//def underlying = plain[String]("underlying").sequence if I wanted the field to be Seq[String]
override def format = underlying <> (Message.apply _, Message.unapply)
}
case class Message(underlying: String)
object Message {
implicit object sm extends HasJsonSchema[Message, MessageSchema](MessageSchema.apply)
}
case class LanguageTaggedSchema[T, S <: JsonSchema[T]](prefix: PathNodes)(implicit evT: HasJsonSchema[T, S]) extends JsonSchema[LanguageTagged[T]](prefix) {
def lang = plain[String]("lang")
def data: S = nested("data")(evT)
def format = (lang, data) <> (LanguageTagged.apply[T] _, LanguageTagged.unapply[T])
}
case class LanguageTagged[T](lang: String, data: T)
object LanguageTagged {
implicit def schemaMapper[T, S <: JsonSchema[T]](implicit ev: HasJsonSchema[T, S]): HasJsonSchema[LanguageTagged[T], LanguageTaggedSchema[T, S]] =
new HasJsonSchema[LanguageTagged[T], LanguageTaggedSchema[T, S]](LanguageTaggedSchema.apply[T, S]) {}
}
def toJson[T, S <: JsonSchema[T]](a: T)(implicit ev: HasJsonSchema[T, S]): JsValue = Json.toJson(a)(ev.format)
toJson(Message("hi")) //Ok!
toJson(LanguageTagged("en", Message("hi"))) //Ok!
//or simply write
Json.toJson(LanguageTagged("en", Message("hi")))
//and if i wanted to traverse a json path i would do:
val schema = implicitly[HasJsonSchema[LanguageTagged[Message],LanguageTaggedSchema[Message,MessageSchema]]].root
schema.data.underlying.jsPath
//prints: res2: play.api.libs.json.JsPath = /data/underlying
//Now to where the problem starts:
def getSchema[T, S <: JsonSchema[T]](a: T)(implicit ev: HasJsonSchema[T, S]): S = ev.root
getSchema(Message("hi")) //Ok!
getSchema(LanguageTagged("en", Message("hi"))) //Not Ok but why?
//Error:(211, 11) could not find implicit value for
//parameter ev: A$A6.this.HasJsonSchema[A$A6.this.LanguageTagged[A$A6.this.Message],S]
//getSchema(LanguageTagged("en", Message("hi")));//
//^
I have a huge suspicion that the compiler runs into issues because of the bounded type of S inHasJsonSchema[T, S <: JsonSchema[T]] when infering the implicit type S. and so far only in that specific situation as shown on the last line of all the code. as a dubugging attempt I created a similar situation and realized that if the type S was not bounded I wouldn't have this issue. Any sort of solution that refactors the code such that it doesn't depend on bounded types or one that simply solves the implicit resolution is appreciated
What You're trying to achieve cannot be done with subtyping. You should use type-classes instead, a more in-depth explanation:
http://danielwestheide.com/blog/2013/02/06/the-neophytes-guide-to-scala-part-12-type-classes.html
Having the following code:
import shapeless.{::, Generic, HList, HNil, Lazy}
object Problem {
trait In {
def bar: Double
}
trait A {
def foo: Int
type I <: In
}
/////////////////////////////////////////////////////
final case class A1In(d: Double) extends In {
override def bar: Double = 1.1 + d
}
final case class A1() extends A {
override def foo: Int = 1
override type I = A1In
}
final case class A2In(d: Double) extends In {
override def bar: Double = 1.1 + d
}
final case class A2() extends A {
override def foo: Int = 1
override type I = A2In
}
final case class AListIn[T <: HList](items: T)(implicit ev: isIn[T]) extends In {
override def bar = 1.1
}
final case class AList[T <: HList](items: T)(implicit ev: isA[T]) extends A {
override def foo: Int = 555
override type I = AListIn[???]
}
trait isA[T] {
def aux_foo(value: T): Int
}
trait isIn[T] {
def aux_bar(value: T): Double
}
/////////////////////////////////////////////////////
def alloc(a: A): In = ????
def usage() = {
val a1: A1 = A1()
val a2: A2 = A2()
val l: AList[::[A1, ::[A2, ::[A1, HNil]]]] = AList(a1 :: a2 :: a1 :: HNil)
val a1In: A1In = A1In(1.2)
val a2In: A2In = A2In(9.3)
val lIn: AListIn[::[A2In, ::[A1In, HNil]]] = AListIn(a2In :: a1In :: HNil)
}
}
How can I fix it so it works as expected?
E.g how do I get correct type in place of ??? which is a proper type HList being result of applying isA -> isIn type mapping. The mapping must follow the natural association of A -> In mapping defined as type I <: In in trait A
And how to implement alloc function which for any concrete instance of In will produce corresponding instance of A?
Should concrete implementations of In be path dependent types of corresponding As?
Below is the code for isA the code for isIn is analogous
trait isA[T] {
def aux_foo(value: T): Int
}
object isA {
// "Summoner" method
def apply[T](implicit enc: isA[T]): isA[T] = enc
// "Constructor" method
def instance[T](func: T => Int): isA[T] = new isA[T] {
override def aux_foo(value: T): Int = func(value)
}
implicit def a1Encoder: isA[A1] = instance(i => 4)
implicit def a2Encoder: isA[A2] = instance(i => 9)
implicit def hnilEncoder: isA[HNil] = instance(hnil => 0)
implicit def hlistEncoder[H, T <: HList](implicit
hInstance: Lazy[isA[H]],
tInstance: isA[T]
): isA[H :: T] = instance {
case h :: t => hInstance.value.aux_foo(h) + tInstance.aux_foo(t)
}
implicit def genericInstance[A, R](implicit
generic: Generic.Aux[A, R],
rInstance: Lazy[isA[R]]
): isA[A] = instance { value => rInstance.value.aux_foo(generic.to(value)) }
}
I have following question: having typeclass for ADT derived with LabelledTypeClassCompanion (where both product and coproduct are correctly defined) inside Scala object, why compiler can't find instance for datatype itself (coproduct), but is able to do so for parameters of specific data constructors (product) when called inside the same object and assigned to val?
Below is a snippet for Show typeclass which is mostly borrowed from corresponding Shapeless example:
import shapeless._
object ShowGeneric {
trait Show[T] {
def show(t: T): String
}
object Show extends LabelledTypeClassCompanion[Show] {
implicit def intShow: Show[Int] = new Show[Int] {
override def show(i: Int): String = i.toString
}
implicit def booleanShow: Show[Boolean] = new Show[Boolean] {
override def show(b: Boolean): String = b.toString
}
implicit def listShow[A](implicit showA: Show[A]): Show[List[A]] = new Show[List[A]] {
override def show(l: List[A]): String = l.map(showA.show).mkString("List(", ", ", ")")
}
object typeClass extends LabelledTypeClass[Show] {
override def emptyProduct: Show[HNil] = new Show[HNil] {
override def show(t: HNil): String = ""
}
override def product[H, T <: HList](name: String, sh: Show[H], st: Show[T]): Show[H :: T] = new Show[H :: T] {
override def show(t: H :: T): String = {
val head = s"$name = ${sh.show(t.head)}"
val tail = st.show(t.tail)
if(tail.isEmpty) head else s"$head, $tail"
}
}
override def coproduct[L, R <: Coproduct](name: String, cl: => Show[L], cr: => Show[R]): Show[L :+: R] = new Show[L :+: R] {
override def show(t: L :+: R): String = t match {
case Inl(l) => s"$name(${cl.show(l)})"
case Inr(r) => cr.show(r)
}
}
override def emptyCoproduct: Show[CNil] = new Show[CNil] {
override def show(t: CNil): String = ""
}
override def project[F, G](instance: => Show[G], to: F => G, from: G => F): Show[F] = new Show[F] {
override def show(t: F): String = instance.show(to(t))
}
}
}
implicit class ShowOps[T](t: T)(implicit showT: Show[T]) {
def show: String = showT.show(t)
}
sealed trait Whatever
case class IntBool(i: Int, b: Boolean) extends Whatever
case class BoolInt(b: Boolean, i: Int) extends Whatever
case class BoolListInt(b: Boolean, l: List[Int]) extends Whatever
def showWhatever(whatever: Whatever)(implicit showWhatever: Show[Whatever]): String = {
showWhatever.show(whatever)
}
val stringBoolInt = BoolInt(false, 100).show // Compiles
val stringShowWhatever = showWhatever(BoolInt(false, 100)) // Doesn't compile: could not find implicit value for parameter showWhatever:ShowGeneric.Show[ShowGeneric.Whatever]
}
Thanks in advance for your answers!
The trait TraversableLike[+A, +Repr] allows one to make a collection where some functions will return a Repr, while others continue to return the type parameter That on the function. Is there a way to define a CustomCollection[A] where functions like map, ++, and others will default That as Repr if not inferred otherwise?
Here is a code snippet that hopefully describes what I would like:
case class CustomCollection[A](list: List[A]) extends TraversableLike[A, CustomCollection[A]] {
protected[this] def newBuilder = new CustomCollectionBuilder[A]
def foreach[U](f: (A) => U) {list foreach f}
def seq = list
}
class CustomCollectionBuilder[A] extends mutable.Builder[A, CustomCollection[A]] {
private val list = new mutable.ListBuffer[A]()
def += (elem: A): this.type = {
list += elem
this
}
def clear() {list.clear()}
def result(): CustomCollection[A] = CustomCollection(list.result())
}
object CustomCollection extends App {
val customCollection = CustomCollection(List(1, 2, 3))
println(customCollection filter {x => x == 1}) // CustomCollection(1)
println(customCollection map {x => x + 1}) // non-empty iterator
}
I would like the last line to be CustomCollection(2, 3, 4).
You need to set up a companion object which provides the refined CanBuildFrom instance:
import collection.TraversableLike
import collection.generic.{CanBuildFrom, GenericCompanion, GenericTraversableTemplate,
TraversableFactory}
import collection.mutable.{Builder, ListBuffer}
object CustomCollection extends TraversableFactory[CustomCollection] {
def newBuilder[A] = new CustomCollectionBuilder[A]
implicit def canBuildFrom[A]: CanBuildFrom[Coll, A, CustomCollection[A]] =
new CanBuildFrom[Coll, A, CustomCollection[A]] {
def apply(): Builder[A, CustomCollection[A]] = new CustomCollectionBuilder()
def apply(from: Coll): Builder[A, CustomCollection[A]] = apply()
}
}
case class CustomCollection[A](list: List[A]) extends Traversable[A]
with TraversableLike[A, CustomCollection[A]]
with GenericTraversableTemplate[A, CustomCollection] {
override def companion: GenericCompanion[CustomCollection] = CustomCollection
def foreach[U](f: A => U) { list foreach f }
override def seq = list
}
class CustomCollectionBuilder[A] extends Builder[A, CustomCollection[A]] {
private val list = new ListBuffer[A]()
def += (elem: A): this.type = {
list += elem
this
}
def clear() {list.clear()}
def result(): CustomCollection[A] = CustomCollection(list.result())
}
val customCollection = CustomCollection(List(1, 2, 3))
val f = customCollection filter {x => x == 1} // CustomCollection[Int]
val m = customCollection map {x => x + 1} // CustomCollection[Int]