例子1:
class User {
var name = "anonymous"
var age:Int = _
val country = "china"
def email = name + "@mail"
}
使用:
val u = new User
// var定义的属性可读可写
u.name = "qh"; u.age = 30
println(u.name + ", " + u.age) // "qh, 30"
// val 定义的属性只读不可写
u.country = "usa" // 报错
println(u.country) // "china"
// def 定义的是方法, 每次调用时重新计算
u.email // "qh@mail"
例子2:
// 定义
class Person(ln : String, fn : String, s : Person = null) {
def lastName = ln; // 用def定义后才是属性,ln,fn,s不可见
def firstName = fn;
def spouse = s;
def introduction() : String =
return (
("Hi, " + firstName + " " + lastName) +
(if (spouse != null) " and spouse, " + spouse.firstName + " " + spouse.lastName + "."
else ".")
);
}
// 调用
new Person("aa","bb", new Person("cc","dd")).introduction();
class c1(x:String) // 等同于:class c1(private var x:String)
val o1 = new c1("aaa")
o1.x // 报错,因为是private的,定义成 class c1(var x:String) 或者 class c1(val x:String) 才能这样用
例子1:
object construct1 {
class c1(name:String, age:Int) { // (1)直接在类定义处
def this() { this("anonymous", 20) } // (2)用this定义
def m1() = { printf("%s=%d\n", name, age) }
}
def main(args:Array[String]) = {
new c1().m1()
new c1("qh", 30).m1()
}
}
编译:fsc construct1.scala
运行:java construct1
例子2:继承中的构造方法:
class c2(name:String, age:Int, female:Boolean=false) extends c1(name,age) {
override def toString = { name + "," + age + "," + female }
}
不同于Java的使用 @Override,或者直接使用相同名字覆盖父类方法。
override def toString = { name + "," + age + "," + female }
如果是覆盖抽象方法,可以不用override关键字。
classes in Scala cannot have static members. Instead, Scala has singleton objects.
A singleton object definition looks like a class definition, except instead of the keyword class you use the keyword object.
Defining a singleton object doesn't define a type.
When a singleton object shares the same name with a class, it is called that class’s companion object. You must define both the class and its companion object in the same source file. The class is called the companion class of the singleton object. A class and its companion object can access each other's private members.
A singleton object is more than a holder of static methods, however. It is a first-class object. A singleton object is initialized the first time some code accesses it.
如:
Java |
Scala |
public class User { private String name; private User(String name) { this.name=name; } private static theUser = new User("Nick"); public static User instance() { return theUser; } } |
object User { var name:String = _ def apply(name:String){this.name=name; this} override def toString = "name: " + name } |
A singleton object that does not share the same name with a companion class is called a standalone object. You can use standalone objects for many purposes, including collecting related utility methods together, or defining an entry point to a Scala application.
Scala没有静态方法,类似静态方法的函数定义在object中:
object Stringx {
def left(s0:String, s:String) = ...
}
直接调用Stringx.left(s0, s),或者 Stringx left (s0, s)
定义在object中的implicit方法也能被直接调用:
例如:
--------- ImportSub.scala
object ImportSub {
def fac(n: Int) = 1 to n reduceLeft (_ * _)
implicit def foo(n: Int) = new { def ! = fac(n) }
}
--------- ImportMain.scala
import ImportSub._
object ImportMain {
def main(args : Array[String]) : Unit = {
println(5!) // 调用ImportSub中定义的implicit函数
}
}
例如:
case class Person(name:String, age:Int)
特殊之处:
l 新建类实例不用new Person(..),直接用Person("qh",20)
l 自动定义好getXX方法,Person("qh",20).name // "qh"
l 提供默认的toString(), Person("qh",20) // "Person(qh,20)"
l 结合类继承可以通过模式匹配进行分解
例子1:
abstract class Person
case class Man(power:Int) extends Person
case class Woman(beauty:Int, from:String) extends Person
val w1 = Woman(100,"china")
val w2 = w1.copy(from="usa") // Woman(100,"usa")
def f(t:Person) = t match {
case Man(x) => "man's power:" + x
case Woman(x,y) => y + " beauty:" + x
}
f(Man(100)) // man's power:100
f(Woman(90, "china")) // china beauty:90
注:基本类型直接可以用math case
例子2:可变的类状态
case class C1(var s: String, var ops: Int) {
def >>= (f: (String=>String)) = {
s = f(s) // s改变
ops += 1 // ops改变
this // 返回自身,可以连续调用
}
}
val C1(res, ops) = C1("ab", 0) >>= (_ * 3) >>= (_ drop 3)
// res="ab"->"ababab"->"bab", ops=0-> 0+1+1->2
细细体会构造对象,连续调用>>方法,然后match的这一系列过程。
例子3:用case class代替tuple
val p = ("qh",20) // p._1 = "qh", p._2 = 20;好处是简洁,但无意义
case class person(name:String, age:Int)
val p = person("qh",20) // p.name = "qh", p.age = 20; 好处是有名字,自说明,可读性强
例子4:用case class来描述元数据
xml的版本:
<todo name = "housework">
<item priority = "high">Clean the hose</item>
<item priority = "medium">Wash the dishes</item>
<item priority = "medium">Buy more soap</item>
</todo>
Scala的版本:
case class item(priority:String, s:String)
case class todo(name:String, items:List[item])
todo (name="housework",
items=item("high","Clean the house")::
item("medium","Wash the dishes")::
item("medium","Buy more soap")::Nil)
比如定义一个标识类(而不是字符串):
case object Start
case object Stop
Java中:
enum fruits { apple, banana, cherry }
在Scala中,则是:
sealed abstract class Fruits // sealed类似于java的final
case object Apple extends Fruits
case object Banana extends Fruits
case object Cherry extends Fruits
也可以是 case class
例子1(直接定义和使用属性):
class c {
var name = "anonymous" // var定义的是r/w的属性
val age = 20 // val定义的是只r属性
}
val o = new c
o.name = "qh"
o.name // "qh"
o.age = 10 // 错误
o.age // 20
o.
例子2(定义get/set方法):
class c2 {
@reflect.BeanProperty var name = "anonymous"
}
val o2 = new c2
o2.name = "qh" // 也可以直接存取
o2.name // "qh"
o2.setName("james") // 增加了set/get方法
o2.getName() // "james"
Scala没有太特别的反射机制,使用java的即可,不过Scala在match..case中可以匹配类型:
case o:FooClass1 => ...
相关还有isInstanceOf[T], asInstanceOf[T]
例1(利用java的reflect):
"hello".getClass.getMethods.map(_.getName).toList.sortWith(_<_).mkString(", ")
例子2:
classOf[String] // 相当于java中的String.class
"aaa".isInstanceOf[String] // true
"aaa".asInstanceOf[String]
疑问:为什么是中括号而不是小括号呢?泛型?
注:trait [treit] n.特征,特点,特性
和Java的Interface类似,但可以定义实体方法,而非仅仅方法定义
trait可以看作有方法实现和字段的interface;代表一类事物的特性;
比如
Tom,可能是Engine 和Son两个trait的混合;
Sunny可能Sales、Son和Father三个trait的混合;
当在运行时往Son里面增加方法或者字段的时候,Tom和Sunny都得到增加的特性。
trait Runnable {
def run(): Unit;
}
只是用一个接口,就用extends:
class c1 extends Runnable {...}
2个接口(或一个继承一个接口),用with而不是implements如下:
class c1 extends c0 with Runnable {
def run(): Unit = {...}
}
一个类可以组合多个trait:
class c1 extends t1 with t2 with t3 {...}
class Human
class Child
trait Dad {
private var children:List[Child] = Nil
def add(child:Child) = child :: children
}
class Man1(name:String) extends Human with Dad // 静态mixin
class Man2(name:String) extends Human // 先不具备Dad trait
val m1 = new Man1("qh")
m1.add(new Child)
val m2 = new Man2("小孩")
// m2.add(new Child) // 报错
val m2$ = new Man2("james") with Dad // 动态mixin
m2$.add(new Child)
使用“+”“-”差异标记
Function[A, B]和Function[-A, +B]的区别图示:
Function[A,B] |
Function[-A,+B] |
trait Queue[T] {} |
非变 |
trait Queue[+T] {} |
协变 如果S extends A (S为子类型,A为父类型), 则Queue[S]为子类型,Queue[A]为父类型 S <: A => Queue[S] <: Queue[A] |
trait Queue[-T] {} |
逆变 如果S extends A (S为子类型,A为父类型) 则Queue[S]为父类型,Queue[A]为子类型,和协变互逆 S <: A => Queue[S] >: Queue[A] |
-A是A的子集,叫逆变
+B是B的超集,叫协变
|
|
<% |
foo[T <% Ordered[T]](...) 关系较弱:T能够隐式转换为Ordered[T] |
<: |
foo[T <: Ordered[T]](...) 关系较强:T必须是Ordered[T]的子类型,即T的类型范围小于Ordered[T],Ordered[T]为上界 |
>: |
foo[T >: A](...) 关系较强:T必须是A的父类型,即Tde类型范围大于A,A为下界 |
例子1:
trait c1[+T] { def m[K >: T](x:K) = x } |
trait c1[-T] { def m[K <: T](x:K) = x } |
object c2 extends c1[Int] c2.m(3) // 3 c2.m(3.0) // 3.0 c2.m("abc") // "abc" |
object c2 extends c1[Int] c2.m(3) // 3 c2.m(3.0) // 报错 c2.m("abc") // 报错 |
|
|
例子2:
// 非变
case class T1[T](e:T)
val v1:T1[java.lang.Integer] = new T1(100)
val v2:T1[java.lang.Integer] = v1
v2.e // 100
val v3:T1[java.lang.Number] = v1 // 报错
// 协变
case class T1[+T](e:T)
val v1:T1[java.lang.Integer] = new T1(100)
val v2:T1[java.lang.Integer] = v1
v2.e // 100
val v3:T1[java.lang.Number] = v1 // 合法
v3.e // 100
val v4:T1[java.lang.Integer] = v3 //非法
// 逆变
class T1[-T](e:T)
val v1:T1[java.lang.Number] = new T1(100)
val v2:T1[java.lang.Number] = v1
val v3:T1[java.lang.Integer] = v1 // 合法
val v4:T1[java.lang.Number] = v3 // 非法