高阶函数主要有两种:一种是将一个函数当做另外一个函数的参数(即函数参数);另外一种是返回值是函数的函数。这两种在本教程的第五节 函数与闭包中已经有所涉及,这里简单地回顾一下: (1)函数参数
//函数参数,即传入另一个函数的参数是函数 //((Int)=>String)=>String scala> def convertIntToString(f:(Int)=>String)=f(4) convertIntToString: (f: Int => String)String scala> convertIntToString((x:Int)=>x+" s") res32: String = 4 s(2)返回值是函数的函数
//高阶函数可以产生新的函数,即我们讲的函数返回值是一个函数 //(Double)=>((Double)=>Double) scala> def multiplyBy(factor:Double)=(x:Double)=>factor*x multiplyBy: (factor: Double)Double => Double scala> val x=multiplyBy(10) x: Double => Double = <function1> scala> x(50) res33: Double = 500.0Scala中的高阶函数可以说是无处不在,这点可以在Scala中的API文档中得到验证,下图给出的是Array数组的需要函数作为参数的API: 例如flatMap方法,下面是其API的详细内容:
def flatMap[B](f: (A) ⇒ GenTraversableOnce[B]): Array[B] [use case] Builds a new collection by applying a function to all elements of this array and using the elements of the resulting collections. //下面的代码给出了该函数的用法 For example: def getWords(lines: Seq[String]): Seq[String] = lines flatMap (line => line split "\\W+") The type of the resulting collection is guided by the static type of array. This might cause unexpected results sometimes. For example: // lettersOf will return a Seq[Char] of likely repeated letters, instead of a Set def lettersOf(words: Seq[String]) = words flatMap (word => word.toSet) // lettersOf will return a Set[Char], not a Seq def lettersOf(words: Seq[String]) = words.toSet flatMap (word => word.toSeq) // xs will be a an Iterable[Int] val xs = Map("a" -> List(11,111), "b" -> List(22,222)).flatMap(_._2) // ys will be a Map[Int, Int] val ys = Map("a" -> List(1 -> 11,1 -> 111), "b" -> List(2 -> 22,2 -> 222)).flatMap(_._2) //下面几行对该函数的参数进行了说明 B the element type of the returned collection. //指明f是函数,该函数传入的参数类型是A,返回类型是GenTraversableOnce[B] f the function to apply to each element. returns a new array resulting from applying the given collection-valued function f to each element of this array and concatenating the results.1 map函数 所有集合类型都存在map函数,例如Array的map函数的API具有如下形式:
def map[B](f: (A) ⇒ B): Array[B] 用途:Builds a new collection by applying a function to all elements of this array. B的含义:the element type of the returned collection. f的含义:the function to apply to each element. 返回:a new array resulting from applying the given function f to each element of this array and collecting the results. //这里面采用的是匿名函数的形式,字符串*n得到的是重复的n个字符串,这是scala中String操作的一个特点 scala> Array("spark","hive","hadoop").map((x:String)=>x*2) res3: Array[String] = Array(sparkspark, hivehive, hadoophadoop) //在函数与闭包那一小节,我们提到,上面的代码还可以简化 //省略匿名函数参数类型 scala> Array("spark","hive","hadoop").map((x)=>x*2) res4: Array[String] = Array(sparkspark, hivehive, hadoophadoop) //单个参数,还可以省去括号 scala> Array("spark","hive","hadoop").map(x=>x*2) res5: Array[String] = Array(sparkspark, hivehive, hadoophadoop) //参数在右边只出现一次的话,还可以用占位符的表示方式 scala> Array("spark","hive","hadoop").map(_*2) res6: Array[String] = Array(sparkspark, hivehive, hadoophadoop)List类型:
scala> val list=List("Spark"->1,"hive"->2,"hadoop"->2) list: List[(String, Int)] = List((Spark,1), (hive,2), (hadoop,2)) //写法1 scala> list.map(x=>x._1) res20: List[String] = List(Spark, hive, hadoop) //写法2 scala> list.map(_._1) res21: List[String] = List(Spark, hive, hadoop) scala> list.map(_._2) res22: List[Int] = List(1, 2, 2)Map类型:
//写法1 scala> Map("spark"->1,"hive"->2,"hadoop"->3).map(_._1) res23: scala.collection.immutable.Iterable[String] = List(spark, hive, hadoop) scala> Map("spark"->1,"hive"->2,"hadoop"->3).map(_._2) res24: scala.collection.immutable.Iterable[Int] = List(1, 2, 3) //写法2 scala> Map("spark"->1,"hive"->2,"hadoop"->3).map(x=>x._2) res25: scala.collection.immutable.Iterable[Int] = List(1, 2, 3) scala> Map("spark"->1,"hive"->2,"hadoop"->3).map(x=>x._1) res26: scala.collection.immutable.Iterable[String] = List(spark, hive, hadoop)2 flatMap函数
//写法1 scala> List(List(1,2,3),List(2,3,4)).flatMap(x=>x) res40: List[Int] = List(1, 2, 3, 2, 3, 4) //写法2 scala> List(List(1,2,3),List(2,3,4)).flatMap(x=>x.map(y=>y)) res41: List[Int] = List(1, 2, 3, 2, 3, 4)3 filter函数
scala> Array(1,2,4,3,5).filter(_>3) res48: Array[Int] = Array(4, 5) scala> List("List","Set","Array").filter(_.length>3) res49: List[String] = List(List, Array) scala> Map("List"->3,"Set"->5,"Array"->7).filter(_._2>3) res50: scala.collection.immutable.Map[String,Int] = Map(Set -> 5, Array -> 7)4 reduce函数
//写法1 scala> Array(1,2,4,3,5).reduce(_+_) res51: Int = 15 scala> List("Spark","Hive","Hadoop").reduce(_+_) res52: String = SparkHiveHadoop //写法2 scala> Array(1,2,4,3,5).reduce((x:Int,y:Int)=>{println(x,y);x+y}) (1,2) (3,4) (7,3) (10,5) res60: Int = 15 scala> Array(1,2,4,3,5).reduceLeft((x:Int,y:Int)=>{println(x,y);x+y}) (1,2) (3,4) (7,3) (10,5) res61: Int = 15 scala> Array(1,2,4,3,5).reduceRight((x:Int,y:Int)=>{println(x,y);x+y}) (3,5) (4,8) (2,12) (1,14) res62: Int = 155 fold函数
scala> Array(1,2,4,3,5).foldLeft(0)((x:Int,y:Int)=>{println(x,y);x+y}) (0,1) (1,2) (3,4) (7,3) (10,5) res66: Int = 15 scala> Array(1,2,4,3,5).foldRight(0)((x:Int,y:Int)=>{println(x,y);x+y}) (5,0) (3,5) (4,8) (2,12) (1,14) res67: Int = 15 scala> Array(1,2,4,3,5).foldLeft(0)(_+_) res68: Int = 15 scala> Array(1,2,4,3,5).foldRight(10)(_+_) res69: Int = 25 // /:相当于foldLeft scala> (0 /: Array(1,2,4,3,5)) (_+_) res70: Int = 15 scala> (0 /: Array(1,2,4,3,5)) ((x:Int,y:Int)=>{println(x,y);x+y}) (0,1) (1,2) (3,4) (7,3) (10,5) res72: Int = 156 scan函数
//从左扫描,每步的结果都保存起来,执行完成后生成数组 scala> Array(1,2,4,3,5).scanLeft(0)((x:Int,y:Int)=>{println(x,y);x+y}) (0,1) (1,2) (3,4) (7,3) (10,5) res73: Array[Int] = Array(0, 1, 3, 7, 10, 15) //从右扫描,每步的结果都保存起来,执行完成后生成数组 scala> Array(1,2,4,3,5).scanRight(0)((x:Int,y:Int)=>{println(x,y);x+y}) (5,0) (3,5) (4,8) (2,12) (1,14) res74: Array[Int] = Array(15, 14, 12, 8, 5, 0)在java的GUI编程中,在设置某个按钮的监听器的时候,我们常常会使用下面的代码(利用scala进行代码开发):
var counter=0; val button=new JButton("click") button.addActionListener(new ActionListener{ override def actionPerformed(event:ActionEvent){ counter+=1 } })上面代码在addActionListener方法中定义了一个实现了ActionListener接口的匿名内部类,代码中
new ActionListener{ override def actionPerformed(event:ActionEvent){ }这部分称为样板代码,即在任何实现该接口的类中都需要这样用,重复性较高,由于ActionListener接口只有一个actionPerformed方法,它被称为simple abstract method(SAM)。SAM转换是指只给addActionListener方法传递一个参数
button.addActionListener((event:ActionEvent)=>counter+=1) //并提供一个隐式转换,我们后面会具体讲隐式转换 implict def makeAction(action:(event:ActionEvent)=>Unit){ new ActionListener{ override def actionPerformed(event:ActionEvent){action(event)} }这样的话,在进行GUI编程的时候,可以省略非常多的样板代码,使代码更简洁。
在函数与闭包那一节中,我们定义了下面这样的一个函数
//mutiplyBy这个函数的返回值是一个函数 //该函数的输入是Doulbe,返回值也是Double scala> def multiplyBy(factor:Double)=(x:Double)=>factor*x multiplyBy: (factor: Double)Double => Double //返回的函数作为值函数赋值给变量x scala> val x=multiplyBy(10) x: Double => Double = <function1> //变量x现在可以直接当函数使用 scala> x(50) res33: Double = 500.0上述代码可以像这样使用:
scala> def multiplyBy(factor:Double)=(x:Double)=>factor*x multiplyBy: (factor: Double)Double => Double //这是高阶函数调用的另外一种形式 scala> multiplyBy(10)(50) res77: Double = 500.0那函数柯里化(curry)是怎么样的呢?其实就是将multiplyBy函数定义成如下形式
scala> def multiplyBy(factor:Double)(x:Double)=x*factor multiplyBy: (factor: Double)(x: Double)Double即通过(factor:Double)(x:Double)定义函数参数,该函数的调用方式如下:
//柯里化的函数调用方式 scala> multiplyBy(10)(50) res81: Double = 500.0 //但此时它不能像def multiplyBy(factor:Double)=(x:Double)=>factor*x函数一样,可以输入单个参数进行调用 scala> multiplyBy(10) <console>:10: error: missing arguments for method multiplyBy; follow this method with `_' if you want to treat it as a partially applied funct ion multiplyBy(10) ^错误提示函数multiplyBy缺少参数,如果要这么做的话,需要将其定义为偏函数
scala> multiplyBy(10)_ res79: Double => Double = <function1>那现在我们接着对偏函数进行介绍
在数组那一节中,我们讲到,Scala中的数组可以通过foreach方法将其内容打印出来,代码如下:
scala>Array("Hadoop","Hive","Spark")foreach(x=>println(x)) Hadoop Hive Spark //上面的代码等价于下面的代码 scala> def print(x:String)=println(x) print: (x: String)Unit scala> Array("Hadoop","Hive","Spark")foreach(print) Hadoop Hive Spark那什么是部分应用函数呢,所谓部分应用函数就是指,当函数有多个参数,而在我们使用该函数时我们不想提供所有参数(假设函数有3个函数),只提供0~2个参数,此时得到的函数便是部分应用函数,定义上述print函数的部分应用函数代码如下:
//定义print的部分应用函数 scala> val p=print _ p: String => Unit = <function1> scala> Array("Hadoop","Hive","Spark")foreach(p) Hadoop Hive Spark scala> Array("Hadoop","Hive","Spark")foreach(print _) Hadoop Hive Spark在上面的简化输出代码中,下划线_并不是占位符的作用,而是作为部分应用函数的定义符。前面我演示了一个参数的函数部分应用函数的定义方式,现在我们定义一个多个输入参数的函数,代码如下:
//定义一个求和函数 scala> def sum(x:Int,y:Int,z:Int)=x+y+z sum: (x: Int, y: Int, z: Int)Int //不指定任何参数的部分应用函数 scala> val s1=sum _ s1: (Int, Int, Int) => Int = <function3> scala> s1(1,2,3) res91: Int = 6 //指定两个参数的部分应用函数 scala> val s2=sum(1,_:Int,3) s2: Int => Int = <function1> scala> s2(2) res92: Int = 6 //指定一个参数的部分应用函数 scala> val s3=sum(1,_:Int,_:Int) s3: (Int, Int) => Int = <function2> scala> s3(2,3) res93: Int = 6在函数柯里化那部分,我们提到柯里化的multiplyBy函数输入单个参数,它并不会像没有柯里化的函数那样返回一个函数,而是会报错,如果需要其返回函数的话,需要定义其部分应用函数,代码如下:
//定义multiplyBy函数的部分应用函数,它返回的是一个函数 scala> val m=multiplyBy(10)_ m: Double => Double = <function1> scala> m(50) res94: Double = 500.0添加公众微信号,可以了解更多最新Spark、Scala相关技术资讯
