n n Official Scala website n Scala API n

Transcription

1 n Quiz 8 Announcements n Rainbow grades: HW1-8, Quiz1-6, Exam1-2 n Still grading: HW9, Quiz 7 Scala n HW10 due today n HW11 out today, due Friday Fall 18 CSCI 4430, A Milanova 1 Today s Lecture Outline n Scala: a functional and object-oriented programming language n Key ideas n : cons, map/reduce, immutable types n Scala is functional: anonymous functions, higherorder functions, call-by-name parameters n Scala is object-oriented: classes, singleton objects, case classes and pattern matching, traits and mixins n And much more 3 Scala Resources n n Official Scala website n index.html n Scala API n n Tutorial n Getting started: tutorial + slides Fall 18 CSCI 4430, A Milanova 4 Scala n General-purpose language n Multi-paradigm: focus is on functional and object-oriented programming n Designed by Martin Odersky at EPFL, first version around 2001 n Built on the Java Virtual Machine (JVM) n Used at Twitter, Linkedin, Spark big data Scala n Model for variables? n Reference model n Scoping discipline? n Static scoping n Typing discipline? n Static typing. Type annotations, some type inference n Subtype, and (explicit) parametric polymorphism n Parameter passing? n Mixed modes. Default by-value, allows by-name Fall 18 CSCI 4430/6430, A Milanova 5 Fall 18 CSCI 4430/6430, A Milanova 6 1

2 Getting Started n Download Scala n On Mac: brew install scala n Scala REPL: n $ scala n scala> val li = 1::Nil li: List[Int] = List(1) // interpreter response Compiling and running Scala n $ scalac Expressions.scala n $ scala Expressions 7 Defining Variables n vars can be reassigned scala> var x = 1 x: Int = 1 scala> x = 42 x: Int = 42 n vals cannot be reassigned scala> val y = 1 y: Int = 1 scala> y = 42 <console>:12: error: reassignment to val y = 42 8 Basic Control Structures n If-then-else scala> val n = 10 scala> if (n%2==0) print( even ) else print( odd ) scala> if (n%2==0) { print( even ) else { print( odd ) n While scala> var x = 0 scala> while (x < 10) { println(x); x = x+1; n In Scala, everything is an expression! scala> var a = while (x < 10) { println(x); x = x+1; a: Unit = () // essentially Java s void 9 n are important in Scala too! scala> var li = List(1,2,3) li: List[Int] = List(1, 2, 3) scala> var li2 = 1::2::3::Nil li2: List[Int] = List(1, 2, 3) n We can have elements of different types scala> var li3 = List(1,List(2),3) li3: List[Any] = List(1, List(2), 3) :: is Cons! n Any is the supertype of all Scala types Fall 18 CSCI 4430/6430, A Milanova 10 n Cons and append scala> val li1 = List(1,2) scala> val li2 = List(3,4) scala> li2::li3 // yields what? res0: List[Any] = List(List(1, 2), 3, 4) ::: is append! scala> val li2:::li3 res1: List[Int] = List(1, 2, 3, 4) Fall 18 CSCI 4430/6430, A Milanova 11 n map, foldleft, foldright, filter are built in n In Scala, everything is an object n map, foldleft, etc. are applied on list objects scala> li1 = List(1,2,3,4) def foldleft[b](z: B)(op: (B, A) B): B foldleft is a function applied on receiver List[A]. It takes z of type B, and a binary scala> li1.map(_+1) operation op; it returns result of type B. res5: List[Int] = List(2, 3, 4, 5) scala> li1.foldleft(0)(_+_) // same semantics as our foldl res6: Int = 10 scala> li1.foldright(0)(_-_) 12 res7: Int = -2 2

3 n are immutable! n Scala supports immutable types (, Strings) as a defense against call-by-sharing n Just as in Scheme, :: and ::: create new lists n Arrays are mutable scala> li1(3) = 5 error (..) is subscript scala> val a = Array(1,2,3,4) a: Array[Int] = Array(1, 2, 3, 4) scala> a(3) = 5 13 n List (sequence) comprehensions n In Scala, known as for comprehensions n for (seq) yield expression n seq includes generators, may include filters scala> def all_pairs(lis1:list[int], lis2:list[int]) = for (x<-lis1; y<-lis2) yield ((x,y)) all_pairs: (lis1: List[Int], lis2: List[Int])List[(Int, Int)] scala> all_pairs(list(1,2,3),list(6,5,4)) // yields what? res9: List[(Int, Int)] = List((1,6), (1,5), (1,4), (2,6), (2,5), (2,4), (3,6), (3,5), (3,4)) 14 and Streams Stream cons Lecture Outline n Scala streams are like lists, however, elements are computed lazily n Can create infinite lists such as [1..] in Haskell scala> def fun (n:int) : Stream[Int] = n #:: fun(n+1) fun: (n: Int)Stream[Int] // must specify return type Stream[Int] scala> fun(1).take(10) // yields what? scala> fun(1)(100) // yields what? res3: Int = 101 Compare to def fun2 (n:int) : List[Int] = n :: fun2(n+1) fun2(1) Fall 18 CSCI 4430/6430, A Milanova 15 n Scala: a functional and object-oriented programming language n Key ideas n : cons, map/reduce, immutable types n Scala is functional: anonymous functions, higherorder functions, call-by-name parameters n Scala is object-oriented: classes, singleton objects, case classes and pattern matching, traits and mixins Fall 18 CSCI 4430, A Milanova 16 Functions n Functions that are part of an object are called methods (more on classes, case classes and singleton objects later) object HelloWorld { def main(args: Array[String]) { println("hello, world!") n In Scala, functions, just like any other value, are objects Fall 18 CSCI 4430/6430, A Milanova 17 Nested Functions and Scoping var i = 1; var j = 3; middle and outer return 3-Int tuple def outer() : (Int,Int,Int) = { var i = 2; def middle(k:int) : (Int,Int,Int) = { def inner() { var i = 4; println(i); inner(); (i,j,k) // middle returns tuple middle(j) // outer returns tuple 18 println (outer()); println (i+","+j); 3

4 Defining Functions scala> def square(x:int) = x*x square: (x: Int)Int // Scala infers return type n Or scala> def square(x:int) = { x*x square: (x: Int)Int // Also, def square(x:int) : Int = x*x scala> square(5) res0: Int = 25 n But not scala> def square(x:int) { x*x square: (x: Int)Unit 19 Higher-order Functions n Of course, higher-order functions everywhere n Scala is a functional language scala> def plusone(x:int) = x+1 plusone: (x: Int)Int // Note that Scala infers return type scala> def apply_n[a](f:a=>a,n:int,x:a) : A = if (n==0) x else apply_n(f,n-1,f(x)) apply_n: [A](f: A => A, n: Int, x: A)A // Can t infer types! scala> apply_n(plusone,10,0) res10: Int = 10 A is a type parameter Fall 18 CSCI 4430/6430, A Milanova 20 Anonymous Functions n Functions are first-class values (objects) n Can be passed as arguments, returned, assigned scala> var lis = List(1,2,3,4) scala> lis.map((x:int) => x+1) // yields what? res3: List[Int] = List(2, 3, 4, 5) scala> val plusone = (x:int) => x+1 plusone: Int => Int = $$Lambda n Different from def: scala> def plusone(x:int) = x+1 plusone: (x: Int)Int 21 Type Inference n Scala does only simple type inference n No rule to say when it can infer, and when it can t scala> val plusone = x => x+1 <console>:11: error: missing parameter type n Have to type parameter scala> val plusone = (x:int) => x+1 n Scala can infer type if anonymous function immediately applied Fall 18 CSCI 4430/6430, A Milanova 22 Call-by-name Parameters HW10 Example (Attempt 1) n By default Scala uses call-by-value scala> def square(x:int) = x*x scala> square(1+2) 1+2 = 3 3*3 = 9 n Can specify call-by-name by using => scala> def square(x: =>Int) = x*x scala> square(1+2) (1+2)*(1+2) = 3*(1+2) 3* def compute(first : Int, last : Int, incr : Int, i : =>Int, term : =>Double) = { var result : Double = 0.0; i = first; while (i <= last) { result = result + term; i = i + incr; result Doesn t work! Parameters are val, i can t be reassigned Homework assumes value model (by-reference parameter passing) Scala uses reference model var i = 0; var a = Array(1,2,3); val s = compute(0,2,1,i,a(i)); 4

5 HW10 Example (Attempt 2) def compute(first:int, last:int, incr:int, j: Int=>Unit, term: =>Double) = { var result : Double = 0.0; var i : Int = first; while (i <= last) { result = result + term; // term evaluated in ref. env. of caller j(incr); // incements main s i i = i+incr; inc is a closure! result var i : Int = 0; var a = Array(1,2,3); def inc(incr:int) { i = i+incr; a(i) evaluated in caller ref. env. print(compute(0,2,1,inc _,a(i))); 25 Lecture Outline n Scala: a functional and object-oriented programming language n Key ideas n : cons, map/reduce, immutable types n Scala is functional: anonymous functions, higherorder functions, call-by-name parameters n Scala is object-oriented: classes, singleton objects, case classes and pattern matching, traits and mixins Fall 18 CSCI 4430, A Milanova 26 Objects n In Scala, everything is an object n Including functions and numbers! n Unlike Java, no distinction between primitive types (value model), and reference types n Numbers are objects scala> 1+2*3 n Or, we can use instance call syntax scala> (1).+((2).*(3)) // yields what? scala> (2::Nil).::(1) // yeilds what? Fall 18 CSCI 4430/6430, A Milanova 27 Objects n Functions are objects too n Means they are first-class values, can be passed, returned and assigned n Anonymous functions, curried functions, higherorder functions, etc. n Essential characteristic of Scala, renders Scala a functional language Fall 18 CSCI 4430/6430, A Milanova 28 Objects n Singleton objects have a single instance. In Java, we needed the Singleton pattern object HelloWorld { def main(args: Array[String]) { println("hello, world!") n Object declaration entails a singleton object n Above declares both a class and an object n No static methods/fields in Scala! Fall 18 CSCI 4430/6430, A Milanova 29 Classes n Scala is object-oriented (concept of class) class Complex(real: Double, imaginary: Double) { def re() = real // or we can write def re = real def im() = imaginary // def im = imaginary n Syntax is similar to Java n Can specify fields in header, real, imaginary n val c = new Complex(1.5, 2.3) n Can add override String tostring() = "" + re + (if (im < 0) "-" else "+") + im + "i" Fall /6430, A Milanova (based on scala-lang.org tutorial) 30 5

6 Case Classes n Class Tree represents ASTs of arithmetic expressions (e.g., x+x+(1+y)) abstract class Tree case class Sum(l: Tree, r: Tree) extends Tree case class Var(n: String) extends Tree case class Const(v: Int) extends Tree n Can be viewed as a classical OO hierarchy (subtype polymorphism, dynamic dispatch)! n Can be viewed as an Algebraic Data Type (as in Haskell, a tagged union of products)! Fall /6430, A Milanova (based on scala-lang.org tutorial) 31 Case Classes n Can write Const(5) instead of new Const(5) n Getter functions are automatically defined for the constructor parameters (e.g., can write c.v) n Default equals and hashcode provided; work on the structure of the instances, not on their identity n Default tostring provided; prints the value in a source form (e.g., the tree for expression x+1 prints as Sum(Var(x),Const(1))) n Instances of these classes can be decomposed through pattern matching Fall 18 CSCI 4430/6430, A Milanova (based on scala-lang.org tutorial) 32 Pattern Matching n We already know what this is! def eval(t: Tree, env: Environment): Int = t match { case Sum(l, r) => eval(l, env) + eval(r, env) case Var(n) => env(n) case Const(v) => v n Just as in Haskell n Test: does value t match pattern (e.g., Sum) n Binding: if yes, then bind components to corresponding variables (e.g., l an r) Fall /6430, A Milanova (based on scala-lang.org tutorial) 33 Pattern Matching n What design patterns jump to mind? n Patterns for traversal of composite structures n Classical GoF Interpreter pattern n GoF Interpreter defined eval methods per node n Easy to add new node, difficult to add new operation n Pattern matching n Easy to add new operation, difficult to add new class/node n In that sense, similar to Procedural/Visitor 34 HW11 Traits In Scala, == goes to equals, i.e., value equality! n Traversal of composite expressions n A-a-again n This time in Scala n One problem: traversal using pattern matching over a hierarchy of case classes n Another problem: Visitor-like traversals using a higher-order method Fall 18 CSCI 4430/6430, A Milanova 35 n A class can mix in code from one or more traits trait Ord { def < (that: Any): Boolean def <=(that: Any): Boolean = (this < that) (this == that) def > (that: Any): Boolean =!(this <= that) def >=(that: Any): Boolean =!(this < that) n Similar to the Java interface, but can have code n Above example creates a new type Ord n In example <=, >, >= are defined in terms of < Fall 18 CSCI 4430/6430, A Milanova (example from scala-lang.org tutorial) 36 6

7 Date extends Ord n A class can mix in code from one or more traits class Date(y: Int, m: Int, d: Int) extends Ord { def year = y def month = m def day = d override def tostring(): String = year + "-" + month + "-" + day n In Scala, override annotation is mandatory when overriding a concrete method Fall 18 CSCI 4430/6430, A Milanova (example from scala-lang.org tutorial) 37 Date extends Ord override def equals(that: Any): Boolean = that.isinstanceof[date] && { val o = that.asinstanceof[date] o.day == day && o.month == month && o.year == year def <(that: Any): Boolean = { if (!that.isinstanceof[date]) sys.error("cannot compare " + that + " and a Date ) val o = that.asinstanceof[date] (year < o.year) (year == o.year && (month < o.month (month == o.month && day < o.day))) Fall 18 CSCI 4430/6430, A Milanova (example from scala-lang.org tutorial) 38 Traits and Mixins abstract class AbsIterator { type T def hasnext: Boolean def next(): T class StringIterator(s: String) extends AbsIterator { type T = Char private var i = 0 def hasnext = i < s.length // must implement member def next() = { // must implement member val ch = s charat i i += 1 ch 39 Traits and Mixins trait RichIterator extends AbsIterator { def foreach(f: T => Unit): Unit = while (hasnext) f(next()) class RichStringIterator extends StringIterator with RichIterator {... n Can mix-in functionality from different traits n class D extends A with B with C { n As in Java, one extends multiple with (implements) n More flexible (and more complex) than Java n Trait can extend an abstract class n There may be conflicting trait members Fall 18 CSCI 4430/6430, A Milanova 40 7