宏注释可覆盖Scala函数的toString [英] Macro annotation to override toString of Scala function

问题描述

如何编写看起来像@named("+2") _ + 2的用法的宏注释并产生:

How to write macro annotation which looks in usage like @named("+2") _ + 2 and produces:

new (Int => Int) {
  override def toString(): String = "+2"
  def apply(x: Int): Int = x + 2
}

推荐答案

正确的语法是((_: Int) + 2): @named("+2").不幸的是，注释表达式的宏注释无法扩展.

Correct syntax is ((_: Int) + 2): @named("+2"). Unfortunately macro annotations annotating expressions don't expand.

最简单的方法就是使用

object Named { 
  def build[T, R](name: String)(applyFunc: T => R): T => R = new (T => R) { 
    override def toString() = name
    def apply(x: T): R = applyFunc(x) 
  }
}

没有任何宏.

否则， Scalameta 可以扩展表达式的注释:

Otherwise Scalameta can expand annotations on expressions:

build.sbt (有关源生成的sbt文档为

build.sbt (sbt documentation about generation of sources is here)

ThisBuild / name := "scalametademo"

lazy val commonSettings = Seq(
  scalaVersion := "2.13.1",
)

lazy val annotations = project
  .settings(
    commonSettings,
  )

lazy val helpers = project
  .settings(
    commonSettings,
  )

lazy val in = project
  .dependsOn(annotations)
  .settings(
    commonSettings,
  )

lazy val out = project
  .dependsOn(helpers)
  .settings(
    sourceGenerators in Compile += Def.task {
      Generator.gen(
        inputDir  = sourceDirectory.in(in, Compile).value,
        outputDir = sourceManaged.in(Compile).value
      )
    }.taskValue,

    commonSettings,
  )

project/build.sbt

libraryDependencies += "org.scalameta" %% "scalameta" % "4.3.0"

project/Generator.scala

import sbt._

object Generator {
  def gen(inputDir: File, outputDir: File): Seq[File] = {
    val finder: PathFinder = inputDir ** "*.scala"

    for(inputFile <- finder.get) yield {
      val inputStr = IO.read(inputFile)
      val outputFile = outputDir / inputFile.toURI.toString.stripPrefix(inputDir.toURI.toString)
      val outputStr = Transformer.transform(inputStr)
      IO.write(outputFile, outputStr)
      outputFile
    }
  }
}

project/Transformer.scala

import scala.meta._

object Transformer {
  val getNamedAnnotationParam: PartialFunction[Mod, Lit] = {
    case mod"@named(...${List(List(s: Lit))})" => s
  }

  val isNamedAnnotated: Mod => Boolean = getNamedAnnotationParam.lift(_).isDefined

  def transform(input: String): String = transform(input.parse[Source].get).toString

  def transform(input: Tree): Tree = input.transform {
    case q"package $eref { ..$stats }" =>
      val stats1 = stats.filter {
        case q"import ..${List(importer"annotations.{..$importeesnel}")}" => false
        case _ => true
      }

      q"package $eref { ..$stats1 }"

    case q"$expr: ..@$annotsnel" if annotsnel.exists(isNamedAnnotated) =>
      val annotsnel1 = annotsnel.filterNot(isNamedAnnotated)
      val name = annotsnel.collect(getNamedAnnotationParam).head

      val expr1 = expr match {
        case q"(..$params) => $expr2" =>
          val params1 = params.map {
            case param"..$mods $name: ${Some(tpe)} = $expropt" => 
              param"..$mods $name: ${Some(tpe)} = $expropt"
            case param"..$mods $name: ${None} = $expropt" => 
              param"..$mods $name: scala.Any = $expropt"
          }

          val domain = params1.map {
            case param"..$mods $name: $tpeopt = $expropt" => tpeopt.get
          }

          q"""
               val typed = com.example.helpers.${Term.Name("TypedFunction" + params.length)}($expr)

               new ((..$domain) => typed.CoDomain) {
                 override def toString(): String = $name
                 def apply(..$params1): typed.CoDomain = $expr2
               }
             """

        case e => e
      }

      if (annotsnel1.nonEmpty)
        q"$expr1: ..@$annotsnel1"
      else q"$expr1"
  }
}

annotations/src/main/scala/com/example/annotations/named.scala

package com.example.annotations

import scala.annotation.StaticAnnotation

class named(name: String) extends StaticAnnotation

helpers/src/main/scala/com/example/helpers/TypedFunctions.scala

package com.example.helpers

sealed trait TypedFunctions[_CoDomain] {
  type CoDomain = _CoDomain
}

case class TypedFunction0[_CoDomain](f: () => _CoDomain) extends TypedFunctions[_CoDomain]
case class TypedFunction1[_Domain,  _CoDomain](f: _Domain => _CoDomain) extends TypedFunctions[_CoDomain]
case class TypedFunction2[_Domain1, _Domain2,  _CoDomain](f: (_Domain1, _Domain2) => _CoDomain) extends TypedFunctions[_CoDomain]
case class TypedFunction3[_Domain1, _Domain2, _Domain3, _CoDomain](f: (_Domain1, _Domain2, _Domain3) => _CoDomain) extends TypedFunctions[_CoDomain]

in/src/main/scala/com/example/App.scala

package com.example

import annotations.named

object App {
  (((x: Int) => x + 2): @named("+2"))

  (((x: Int, y: Int) => x + y): @named("+"))
}

out/target/scala-2.13/src_managed/main/scala/com/example/App.scala (在sbt "; project out; clean; compile"之后)

package com.example
object App {
  {
    val typed = com.example.helpers.TypedFunction1 { (x: Int) => x + 2 }
    new (Int => typed.CoDomain) {
      override def toString(): String = "+2"
      def apply(x: Int): typed.CoDomain = x + 2
    }
  }
  {
    val typed = com.example.helpers.TypedFunction2 { (x: Int, y: Int) => x + y }
    new ((Int, Int) => typed.CoDomain) {
      override def toString(): String = "+"
      def apply(x: Int, y: Int): typed.CoDomain = x + y
    }
  }
}

另一个示例是如何在Scala中合并多个导入?

这篇关于宏注释可覆盖Scala函数的toString的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！