- Learn Scala Programming
- Slava Schmidt
- 401字
- 2021-06-10 19:35:49
Higher-order functions
So far, we have discussed function literals and we've created a printHash function, which we used to demonstrate different forms of passing functions into methods:
scala> def printHash(hasher: String => Int)(s: String): Unit =
println(hasher(s))
printHash: (hasher: String => Int)(s: String)Unit
The printHash takes two parameters: the hasher function and a string to hash. Or, in functional form:
scala> :type printHash _
(String => Int) => (String => Unit)
Our function is curried as it takes one argument (a String => Int function) and returns another function, String => Unit. The fact that printHash takes a function as an argument is reflected by saying that printHash is a higher order function (HOF). There is nothing else special about HOFs except the fact that one or multiple parameters are functions. They work just as normal functions, can be assigned and passed over, partially applied and be polymorphic:
def printHash[A](hasher: A => Int)(s: A): Unit = println(hasher(s))
In fact, HOFs usually apply function(s) given as parameter(s) to another parameter in some creative ways and therefore are almost always polymorphic.
Let's take another look at our printHash example. There is nothing in particular that requires a hasher function to calculate a hash; the function carried out by the hasher is independent of the logic of the printHash. Interestingly, this is the case more often than one would expect and it leads to the definition of HOF, for example:
def printer[A, B, C <: A](f: A => B)(a: C): Unit = println(f(a))
Our printing logic does not require the given function to have any specific type of an argument or result. The only limitation we need to enforce is that it is possible to call the function with the given argument, which we formulate with the type constraint C <: A. The nature of the function and an argument can also be anything, and it is common to use short neutral names when defining a HOF because of this. This is how our new definition can be used in practice:
scala> printer((_: String).hashCode)("HaHa")
2240498
scala> printer((_: Int) / 2)(42)
21
The compiler needs to know the type of the function, and therefore we need to define it as a part of the placeholder syntax. We can help the compiler by changing the order of the parameters of the function:
def printer[A, B, C <: A](a: C)(f: A => B): Unit = println(f(a))
With this definition, it will be possible to infer the C first and then use the inferred type to enforce the type of f:
scala> printer("HoHo")(_.length)
4
scala> printer(42)(identity)
42
The identity function is defined as def identity[A](x: A): A = x in the standard library.
- Java程序設計(慕課版)
- Python科學計算(第2版)
- Design Principles for Process:driven Architectures Using Oracle BPM and SOA Suite 12c
- C語言程序設計案例教程(第2版)
- Visual FoxPro程序設計教程
- Scratch 3游戲與人工智能編程完全自學教程
- Web程序設計(第二版)
- C++面向對象程序設計習題解答與上機指導(第三版)
- Android移動開發案例教程:基于Android Studio開發環境
- 從Excel到Python數據分析:Pandas、xlwings、openpyxl、Matplotlib的交互與應用
- JQuery風暴:完美用戶體驗
- 遠方:兩位持續創業者的點滴思考
- Android嵌入式系統程序開發(基于Cortex-A8)
- Yii2 By Example
- 創新工場講AI課:從知識到實踐