--- language: Scala filename: learnscala.scala contributors: - ["George Petrov", "http://github.com/petrovg"] - ["Dominic Bou-Samra", "http://dbousamra.github.com"] - ["Geoff Liu", "http://geoffliu.me"] - ["Ha-Duong Nguyen", "http://reference-error.org"] --- Scala - the scalable language ```scala ///////////////////////////////////////////////// // 0. Basics ///////////////////////////////////////////////// /* Setup Scala: 1) Download Scala - http://www.scala-lang.org/downloads 2) Unzip/untar to your favorite location and put the bin subdir in your `PATH` environment variable */ /* Try the REPL Scala has a tool called the REPL (Read-Eval-Print Loop) that is analogous to commandline interpreters in many other languages. You may type any Scala expression, and the result will be evaluated and printed. The REPL is a very handy tool to test and verify code. Use it as you read this tutorial to quickly explore concepts on your own. */ // Start a Scala REPL by running `scala`. You should see the prompt: $ scala scala> // By default each expression you type is saved as a new numbered value scala> 2 + 2 res0: Int = 4 // Default values can be reused. Note the value type displayed in the result.. scala> res0 + 2 res1: Int = 6 // Scala is a strongly typed language. You can use the REPL to check the type // without evaluating an expression. scala> :type (true, 2.0) (Boolean, Double) // REPL sessions can be saved scala> :save /sites/repl-test.scala // Files can be loaded into the REPL scala> :load /sites/repl-test.scala Loading /sites/repl-test.scala... res2: Int = 4 res3: Int = 6 // You can search your recent history scala> :h? 1 2 + 2 2 res0 + 2 3 :save /sites/repl-test.scala 4 :load /sites/repl-test.scala 5 :h? // Now that you know how to play, let's learn a little scala... ///////////////////////////////////////////////// // 1. Basics ///////////////////////////////////////////////// // Single-line comments start with two forward slashes /* Multi-line comments, as you can already see from above, look like this. */ // Printing, and forcing a new line on the next print println("Hello world!") println(10) // Hello world! // 10 // Printing, without forcing a new line on next print print("Hello world") print(10) // Hello world10 // Declaring values is done using either var or val. // val declarations are immutable, whereas vars are mutable. Immutability is // a good thing. val x = 10 // x is now 10 x = 20 // error: reassignment to val var y = 10 y = 20 // y is now 20 /* Scala is a statically typed language, yet note that in the above declarations, we did not specify a type. This is due to a language feature called type inference. In most cases, Scala compiler can guess what the type of a variable is, so you don't have to type it every time. We can explicitly declare the type of a variable like so: */ val z: Int = 10 val a: Double = 1.0 // Notice automatic conversion from Int to Double, result is 10.0, not 10 val b: Double = 10 // Boolean values true false // Boolean operations !true // false !false // true true == false // false 10 > 5 // true // Math is as per usual 1 + 1 // 2 2 - 1 // 1 5 * 3 // 15 6 / 2 // 3 6 / 4 // 1 6.0 / 4 // 1.5 6 / 4.0 // 1.5 // Evaluating an expression in the REPL gives you the type and value of the result 1 + 7 /* The above line results in: scala> 1 + 7 res29: Int = 8 This means the result of evaluating 1 + 7 is an object of type Int with a value of 8 Note that "res29" is a sequentially generated variable name to store the results of the expressions you typed, your output may differ. */ "Scala strings are surrounded by double quotes" 'a' // A Scala Char // 'Single quote strings don't exist' <= This causes an error // Strings have the usual Java methods defined on them "hello world".length "hello world".substring(2, 6) "hello world".replace("C", "3") // They also have some extra Scala methods. See also: scala.collection.immutable.StringOps "hello world".take(5) "hello world".drop(5) // String interpolation: notice the prefix "s" val n = 45 s"We have $n apples" // => "We have 45 apples" // Expressions inside interpolated strings are also possible val a = Array(11, 9, 6) s"My second daughter is ${a(0) - a(2)} years old." // => "My second daughter is 5 years old." s"We have double the amount of ${n / 2.0} in apples." // => "We have double the amount of 22.5 in apples." s"Power of 2: ${math.pow(2, 2)}" // => "Power of 2: 4" // Formatting with interpolated strings with the prefix "f" f"Power of 5: ${math.pow(5, 2)}%1.0f" // "Power of 5: 25" f"Square root of 122: ${math.sqrt(122)}%1.4f" // "Square root of 122: 11.0454" // Raw strings, ignoring special characters. raw"New line feed: \n. Carriage return: \r." // => "New line feed: \n. Carriage return: \r." // Some characters need to be "escaped", e.g. a double quote inside a string: "They stood outside the \"Rose and Crown\"" // => "They stood outside the "Rose and Crown"" // Triple double-quotes let strings span multiple rows and contain quotes val html = """

Press belo', Joe

""" ///////////////////////////////////////////////// // 2. Functions ///////////////////////////////////////////////// // Functions are defined like so: // // def functionName(args...): ReturnType = { body... } // // If you come from more traditional languages, notice the omission of the // return keyword. In Scala, the last expression in the function block is the // return value. def sumOfSquares(x: Int, y: Int): Int = { val x2 = x * x val y2 = y * y x2 + y2 } // The { } can be omitted if the function body is a single expression: def sumOfSquaresShort(x: Int, y: Int): Int = x * x + y * y // Syntax for calling functions is familiar: sumOfSquares(3, 4) // => 25 // You can use parameters names to specify them in different order def subtract(x: Int, y: Int): Int = x - y subtract(10, 3) // => 7 subtract(y=10, x=3) // => -7 // In most cases (with recursive functions the most notable exception), function // return type can be omitted, and the same type inference we saw with variables // will work with function return values: def sq(x: Int) = x * x // Compiler can guess return type is Int // Functions can have default parameters: def addWithDefault(x: Int, y: Int = 5) = x + y addWithDefault(1, 2) // => 3 addWithDefault(1) // => 6 // Anonymous functions look like this: (x: Int) => x * x // Unlike defs, even the input type of anonymous functions can be omitted if the // context makes it clear. Notice the type "Int => Int" which means a function // that takes Int and returns Int. val sq: Int => Int = x => x * x // Anonymous functions can be called as usual: sq(10) // => 100 // If each argument in your anonymous function is // used only once, Scala gives you an even shorter way to define them. These // anonymous functions turn out to be extremely common, as will be obvious in // the data structure section. val addOne: Int => Int = _ + 1 val weirdSum: (Int, Int) => Int = (_ * 2 + _ * 3) addOne(5) // => 6 weirdSum(2, 4) // => 16 // The return keyword exists in Scala, but it only returns from the inner-most // def that surrounds it. // WARNING: Using return in Scala is error-prone and should be avoided. // It has no effect on anonymous functions. For example: def foo(x: Int): Int = { val anonFunc: Int => Int = { z => if (z > 5) return z // This line makes z the return value of foo! else z + 2 // This line is the return value of anonFunc } anonFunc(x) // This line is the return value of foo } ///////////////////////////////////////////////// // 3. Flow Control ///////////////////////////////////////////////// 1 to 5 val r = 1 to 5 r.foreach(println) r foreach println // NB: Scala is quite lenient when it comes to dots and brackets - study the // rules separately. This helps write DSLs and APIs that read like English // Why doesn't `println` need any parameters here? Stay tuned for Functional Programming below! (5 to 1 by -1) foreach (println) // A while loop var i = 0 while (i < 10) { println("i " + i); i += 1 } while (i < 10) { println("i " + i); i += 1 } // Yes, again. What happened? Why? i // Show the value of i. Note that while is a loop in the classical sense - // it executes sequentially while changing the loop variable. while is very // fast, but using the combinators and comprehensions above is easier // to understand and parallelize // A do-while loop i = 0 do { println("i is still less than 10") i += 1 } while (i < 10) // Recursion is the idiomatic way of repeating an action in Scala (as in most // other functional languages). // Recursive functions need an explicit return type, the compiler can't infer it. // Here it's Unit, which is analagous to a `void` return type in Java def showNumbersInRange(a: Int, b: Int): Unit = { print(a) if (a < b) showNumbersInRange(a + 1, b) } showNumbersInRange(1, 14) // Conditionals val x = 10 if (x == 1) println("yeah") if (x == 10) println("yeah") if (x == 11) println("yeah") if (x == 11) println("yeah") else println("nay") println(if (x == 10) "yeah" else "nope") val text = if (x == 10) "yeah" else "nope" ///////////////////////////////////////////////// // 4. Data Structures ///////////////////////////////////////////////// val a = Array(1, 2, 3, 5, 8, 13) a(0) // Int = 1 a(3) // Int = 5 a(21) // Throws an exception val m = Map("fork" -> "tenedor", "spoon" -> "cuchara", "knife" -> "cuchillo") m("fork") // java.lang.String = tenedor m("spoon") // java.lang.String = cuchara m("bottle") // Throws an exception val safeM = m.withDefaultValue("no lo se") safeM("bottle") // java.lang.String = no lo se val s = Set(1, 3, 7) s(0) // Boolean = false s(1) // Boolean = true /* Look up the documentation of map here - * http://www.scala-lang.org/api/current/index.html#scala.collection.immutable.Map * and make sure you can read it */ // Tuples (1, 2) (4, 3, 2) (1, 2, "three") (a, 2, "three") // Why have this? val divideInts = (x: Int, y: Int) => (x / y, x % y) // The function divideInts gives you the result and the remainder divideInts(10, 3) // (Int, Int) = (3,1) // To access the elements of a tuple, use _._n where n is the 1-based index of // the element val d = divideInts(10, 3) // (Int, Int) = (3,1) d._1 // Int = 3 d._2 // Int = 1 // Alternatively you can do multiple-variable assignment to tuple, which is more // convenient and readable in many cases val (div, mod) = divideInts(10, 3) div // Int = 3 mod // Int = 1 ///////////////////////////////////////////////// // 5. Object Oriented Programming ///////////////////////////////////////////////// /* Aside: Everything we've done so far in this tutorial has been simple expressions (values, functions, etc). These expressions are fine to type into the command-line interpreter for quick tests, but they cannot exist by themselves in a Scala file. For example, you cannot have just "val x = 5" in a Scala file. Instead, the only top-level constructs allowed in Scala are: - objects - classes - case classes - traits And now we will explain what these are. */ // classes are similar to classes in other languages. Constructor arguments are // declared after the class name, and initialization is done in the class body. class Dog(br: String) { // Constructor code here var breed: String = br // Define a method called bark, returning a String def bark = "Woof, woof!" // Values and methods are assumed public. "protected" and "private" keywords // are also available. private def sleep(hours: Int) = println(s"I'm sleeping for $hours hours") // Abstract methods are simply methods with no body. If we uncomment the // def line below, class Dog would need to be declared abstract like so: // abstract class Dog(...) { ... } // def chaseAfter(what: String): String } val mydog = new Dog("greyhound") println(mydog.breed) // => "greyhound" println(mydog.bark) // => "Woof, woof!" // The "object" keyword creates a type AND a singleton instance of it. It is // common for Scala classes to have a "companion object", where the per-instance // behavior is captured in the classes themselves, but behavior related to all // instance of that class go in objects. The difference is similar to class // methods vs static methods in other languages. Note that objects and classes // can have the same name. object Dog { def allKnownBreeds = List("pitbull", "shepherd", "retriever") def createDog(breed: String) = new Dog(breed) } // Case classes are classes that have extra functionality built in. A common // question for Scala beginners is when to use classes and when to use case // classes. The line is quite fuzzy, but in general, classes tend to focus on // encapsulation, polymorphism, and behavior. The values in these classes tend // to be private, and only methods are exposed. The primary purpose of case // classes is to hold immutable data. They often have few methods, and the // methods rarely have side-effects. case class Person(name: String, phoneNumber: String) // Create a new instance. Note cases classes don't need "new" val george = Person("George", "1234") val kate = Person("Kate", "4567") // With case classes, you get a few perks for free, like getters: george.phoneNumber // => "1234" // Per field equality (no need to override .equals) Person("George", "1234") == Person("Kate", "1236") // => false // Easy way to copy // otherGeorge == Person("George", "9876") val otherGeorge = george.copy(phoneNumber = "9876") // And many others. Case classes also get pattern matching for free, see below. // Traits // Similar to Java interfaces, traits define an object type and method // signatures. Scala allows partial implementation of those methods. // Constructor parameters are not allowed. Traits can inherit from other // traits or classes without parameters. trait Dog { def breed: String def color: String def bark: Boolean = true def bite: Boolean } class SaintBernard extends Dog { val breed = "Saint Bernard" val color = "brown" def bite = false } scala> b res0: SaintBernard = SaintBernard@3e57cd70 scala> b.breed res1: String = Saint Bernard scala> b.bark res2: Boolean = true scala> b.bite res3: Boolean = false // A trait can also be used as Mixin. The class "extends" the first trait, // but the keyword "with" can add additional traits. trait Bark { def bark: String = "Woof" } trait Dog { def breed: String def color: String } class SaintBernard extends Dog with Bark { val breed = "Saint Bernard" val color = "brown" } scala> val b = new SaintBernard b: SaintBernard = SaintBernard@7b69c6ba scala> b.bark res0: String = Woof ///////////////////////////////////////////////// // 6. Pattern Matching ///////////////////////////////////////////////// // Pattern matching is a powerful and commonly used feature in Scala. Here's how // you pattern match a case class. NB: Unlike other languages, Scala cases do // not need breaks, fall-through does not happen. def matchPerson(person: Person): String = person match { // Then you specify the patterns: case Person("George", number) => "We found George! His number is " + number case Person("Kate", number) => "We found Kate! Her number is " + number case Person(name, number) => "We matched someone : " + name + ", phone : " + number } // Regular expressions are also built in. // Create a regex with the `r` method on a string: val email = "(.*)@(.*)".r // Pattern matching might look familiar to the switch statements in the C family // of languages, but this is much more powerful. In Scala, you can match much // more: def matchEverything(obj: Any): String = obj match { // You can match values: case "Hello world" => "Got the string Hello world" // You can match by type: case x: Double => "Got a Double: " + x // You can specify conditions: case x: Int if x > 10000 => "Got a pretty big number!" // You can match case classes as before: case Person(name, number) => s"Got contact info for $name!" // You can match regular expressions: case email(name, domain) => s"Got email address $name@$domain" // You can match tuples: case (a: Int, b: Double, c: String) => s"Got a tuple: $a, $b, $c" // You can match data structures: case List(1, b, c) => s"Got a list with three elements and starts with 1: 1, $b, $c" // You can nest patterns: case List(List((1, 2, "YAY"))) => "Got a list of list of tuple" // Match any case (default) if all previous haven't matched case _ => "Got unknown object" } // In fact, you can pattern match any object with an "unapply" method. This // feature is so powerful that Scala lets you define whole functions as // patterns: val patternFunc: Person => String = { case Person("George", number) => s"George's number: $number" case Person(name, number) => s"Random person's number: $number" } ///////////////////////////////////////////////// // 7. Functional Programming ///////////////////////////////////////////////// // Scala allows methods and functions to return, or take as parameters, other // functions or methods. val add10: Int => Int = _ + 10 // A function taking an Int and returning an Int List(1, 2, 3) map add10 // List(11, 12, 13) - add10 is applied to each element // Anonymous functions can be used instead of named functions: List(1, 2, 3) map (x => x + 10) // And the underscore symbol, can be used if there is just one argument to the // anonymous function. It gets bound as the variable List(1, 2, 3) map (_ + 10) // If the anonymous block AND the function you are applying both take one // argument, you can even omit the underscore List("Dom", "Bob", "Natalia") foreach println // Combinators // Using `s` from above: // val s = Set(1, 3, 7) s.map(sq) val sSquared = s. map(sq) sSquared.filter(_ < 10) sSquared.reduce (_+_) // The filter function takes a predicate (a function from A -> Boolean) and // selects all elements which satisfy the predicate List(1, 2, 3) filter (_ > 2) // List(3) case class Person(name: String, age: Int) List( Person(name = "Dom", age = 23), Person(name = "Bob", age = 30) ).filter(_.age > 25) // List(Person("Bob", 30)) // Certain collections (such as List) in Scala have a `foreach` method, // which takes as an argument a type returning Unit - that is, a void method val aListOfNumbers = List(1, 2, 3, 4, 10, 20, 100) aListOfNumbers foreach (x => println(x)) aListOfNumbers foreach println // For comprehensions for { n <- s } yield sq(n) val nSquared2 = for { n <- s } yield sq(n) for { n <- nSquared2 if n < 10 } yield n for { n <- s; nSquared = n * n if nSquared < 10} yield nSquared /* NB Those were not for loops. The semantics of a for loop is 'repeat', whereas a for-comprehension defines a relationship between two sets of data. */ ///////////////////////////////////////////////// // 8. Implicits ///////////////////////////////////////////////// /* WARNING WARNING: Implicits are a set of powerful features of Scala, and * therefore it is easy to abuse them. Beginners to Scala should resist the * temptation to use them until they understand not only how they work, but also * best practices around them. We only include this section in the tutorial * because they are so commonplace in Scala libraries that it is impossible to * do anything meaningful without using a library that has implicits. This is * meant for you to understand and work with implicits, not declare your own. */ // Any value (vals, functions, objects, etc) can be declared to be implicit by // using the, you guessed it, "implicit" keyword. Note we are using the Dog // class from section 5 in these examples. implicit val myImplicitInt = 100 implicit def myImplicitFunction(breed: String) = new Dog("Golden " + breed) // By itself, implicit keyword doesn't change the behavior of the value, so // above values can be used as usual. myImplicitInt + 2 // => 102 myImplicitFunction("Pitbull").breed // => "Golden Pitbull" // The difference is that these values are now eligible to be used when another // piece of code "needs" an implicit value. One such situation is implicit // function arguments: def sendGreetings(toWhom: String)(implicit howMany: Int) = s"Hello $toWhom, $howMany blessings to you and yours!" // If we supply a value for "howMany", the function behaves as usual sendGreetings("John")(1000) // => "Hello John, 1000 blessings to you and yours!" // But if we omit the implicit parameter, an implicit value of the same type is // used, in this case, "myImplicitInt": sendGreetings("Jane") // => "Hello Jane, 100 blessings to you and yours!" // Implicit function parameters enable us to simulate type classes in other // functional languages. It is so often used that it gets its own shorthand. The // following two lines mean the same thing: // def foo[T](implicit c: C[T]) = ... // def foo[T : C] = ... // Another situation in which the compiler looks for an implicit is if you have // obj.method(...) // but "obj" doesn't have "method" as a method. In this case, if there is an // implicit conversion of type A => B, where A is the type of obj, and B has a // method called "method", that conversion is applied. So having // myImplicitFunction above in scope, we can say: "Retriever".breed // => "Golden Retriever" "Sheperd".bark // => "Woof, woof!" // Here the String is first converted to Dog using our function above, and then // the appropriate method is called. This is an extremely powerful feature, but // again, it is not to be used lightly. In fact, when you defined the implicit // function above, your compiler should have given you a warning, that you // shouldn't do this unless you really know what you're doing. ///////////////////////////////////////////////// // 9. Misc ///////////////////////////////////////////////// // Importing things import scala.collection.immutable.List // Import all "sub packages" import scala.collection.immutable._ // Import multiple classes in one statement import scala.collection.immutable.{List, Map} // Rename an import using '=>' import scala.collection.immutable.{List => ImmutableList} // Import all classes, except some. The following excludes Map and Set: import scala.collection.immutable.{Map => _, Set => _, _} // Java classes can also be imported. Scala syntax can be used import java.swing.{JFrame, JWindow} // Your programs entry point is defined in an scala file using an object, with a // single method, main: object Application { def main(args: Array[String]): Unit = { // stuff goes here. } } // Files can contain multiple classes and objects. Compile with scalac // Input and output // To read a file line by line import scala.io.Source for(line <- Source.fromFile("myfile.txt").getLines()) println(line) // To write a file use Java's PrintWriter val writer = new PrintWriter("myfile.txt") writer.write("Writing line for line" + util.Properties.lineSeparator) writer.write("Another line here" + util.Properties.lineSeparator) writer.close() ``` ## Further resources * [Scala for the impatient](http://horstmann.com/scala/) * [Twitter Scala school](http://twitter.github.io/scala_school/) * [The scala documentation](http://docs.scala-lang.org/) * [Try Scala in your browser](http://scalatutorials.com/tour/) * Join the [Scala user group](https://groups.google.com/forum/#!forum/scala-user)