Scala: Pattern matching, Concepts and Implementations

SCALA
PATTERN MATCHING, CONCEPTS AND
IMPLEMENTATIONS
Mustapha Michrafy PhD
M. MICHRAFY & C. VALLIEZ
Contact the authors :
datascience.km@gmail.com
datascience.km@gmail.com1
Cyril Valliez PhD
- We thank Oussama Lachiri for helpful discussions.

Context
This study was presented at the seminar
« Data Science Principales, Tools and
Applications » at Cermsem laboratory
M. MICHRAFY & C. VALLIEZ datascience.km@gmail.com2

Goal
This study aims to present the pattern
matching and its implementation with
Scala

Prerequisite
Knowledge of
• Scala programming language
• Functional programming notion

Agenda
1. Basic pattern matching
2. Pattern matching and for loop
3. Pattern alternative: operator « | »
4. Pattern guards
5. Pattern guards and logical operator OR
6. Pattern matching and recursive function
7. Pattern matching on type
8. Pattern matching on tuple
9. Pattern matching on Option
10. Pattern matching on Immutable Collection
11. Pattern matching on Immutable Map
12. Pattern matching on Immutable Set
13. Pattern matching on List
14. Pattern matching on case class
15. Nested Pattern Matching in Case Classes
16. Matching on regular expression

Basic pattern matching : switch-case
def getChoicegetChoicegetChoicegetChoice(choicechoicechoicechoice: Int): String = choicechoicechoicechoice matchmatchmatchmatch {
case 0 => "no"
case 1 => "yes"
case ____ => "error"
}
if choice = 0 then "no"
else if choice = 1 then "yes"
else then "error"
Pattern matching
getChoice(1) //> res0: String = yes
getChoice(0) //> res1: String = no
getChoice(3) //> res2: String = error
Testing
The wildcard pattern “_”
matches any object whatsoever.

Pattern matching and initialization
val choicechoicechoicechoice = 0
var mesg ==== choicechoicechoicechoice match {
case 0 => "no"
case 1 => "yes"
case _ => "error"
}
println(mesg) //> no
We can use pattern matching for initializing a variable.

Pattern matching and for loop
val choiceschoiceschoiceschoices = List(0,1,10)
for( choicechoicechoicechoice <- choices) {
choice match {
case 0 => println("no")
case 1 => println("yes")
case _ => println("error")
}
}
//> no
//> yes
//>error
We can use pattern matching with for loop

Pattern alternative: operator « | »
casecasecasecase 0000 =>=>=>=> """"no"no"no"no"
casecasecasecase 1111 => "no"=> "no"=> "no"=> "no"
case 2 => "yes"
}
if choice in {0,1} then "no"
else if choice = 2 then "oui"
else then "error"
Pattern matching
« | » : OR
With “|” , we can have multiple tests on a single line
casecasecasecase 0000 | 1 =>| 1 =>| 1 =>| 1 => """"no"no"no"no"
case 2 => "yes"
}
getChoice(0) //>res1: String = no
getChoice(1) //>res2: String = no
Testing

Pattern guards
def sign(x:Double) : Byte = x match {
case x if x < 0if x < 0if x < 0if x < 0 => -1
case x if x > 0if x > 0if x > 0if x > 0 => 1
case _ => 0
}
-1 if x < 0
sing(x) = 1 if x > 0
0 sinon
Pattern matching
sign(0) //> res1: Byte = 0
sign(2) //> res2: Byte = 1
sign(-2) //> res3: Byte = -1
Testing
A pattern guard comes after a
pattern and starts with an if.
If a pattern guard is present, the
match succeeds only if the
guard evaluates to true

Pattern guards and logical operator OR
def OR(p:Boolean, q: Boolean) : Boolean = (p, q)(p, q)(p, q)(p, q) match {
case (x, y) if x==true | y==trueif x==true | y==trueif x==true | y==trueif x==true | y==true => true
case _ => false
}
OR(false, false) //> res1: Boolean = false
OR(false, true) //> res2: Boolean = true
p q p OR q
T T T
T F T
F T T
F F F
If (p=T | q=T) then OR(p,q)=true
else OR(p,q) =false
Testing
Pattern matching
Transform multiple parameters into one tuple (p,q)
1
1

Pattern matching and recursive function
1 if n = 0
n*(n-1)! otherwise
n! ====
def factorialfactorialfactorialfactorial(n: Int): Int = n match {
case 0 => 1
case _ => n*factorialfactorialfactorialfactorial(n-1)
}
def factorialfactorialfactorialfactorial(n:Int) : Int = {
def fcfcfcfc(n:Int, p:Int) : Int = ((((n,pn,pn,pn,p)))) match {
case (0,c) => c
case (m,c) => fc(m-1,p*m)
}
fcfcfcfc(n,1)
}
factorial(0) //> res 1: Int = 1
factorial(4) //> res 2: Int = 24
1 Using a tuple (n,p)
2 Initialize an accumulator p to 1
Tail-recursive function

Pattern matching on type
def matchByTypematchByTypematchByTypematchByType(z: Any): String = z match {
case b:Boolean => s"A boolean : $b"
case d:Double => s"A double : $d"
case _ => " Unknown type "
}
1
2
3
1 If z is Boolean type, b takes its value.
2 If z is the type Double , d takes its value.
3 Use wildcard when z is neither a Boolean type nor a Double type
A typed pattern x:T consists of a pattern variable x and a type pattern T.
The type of x is the type pattern T, where each type variable and wildcard
is replaced by a fresh, unknown type.

Pattern matching on tuple
def detectZeroInTupledetectZeroInTupledetectZeroInTupledetectZeroInTuple(x: Tuple2[Any,Any]):String = x match {
case (0,0|(0,0)) => "Both values zero."
case (0,b) => s"First value is 0 in (0,$b)"
case (a, 0) => s"Second value is 0 in ($a, 0)"
case _ => "Both nonzero : " + x
}
1
The tuple pattern matches input in tuple form and enables the tuple to be
decomposed into its constituent elements by using pattern matching
variables for each position in the tuple.
detectZeroInTuple((0,(0,0))) //> res1: String = Both values zero.
detectZeroInTuple((0,0)) //> res2: String = Both values zero.
detectZeroInTuple((0,11)) //> res3: String = First value is 0 in (0,11)
1 Match a first pattern if x=(0,0) or x=(0,(0,0)).

Pattern matching on Option
def getValueOptiongetValueOptiongetValueOptiongetValueOption[Any](x:Option[Any]) : String = x match{
case Some(x) => s"Some $x"
case None => "None "
}
1
Option -sealed abstract class- represents optional values.
Instances of Option are either an instance of SomeSomeSomeSome or the
object None.None.None.None.
getValueOption(Option("hello")) //> Some hello
getValueOption(Option(null)) //> None
getValueOption(Option(())) //> Some ()
getValueOption(Option(List(10,11))) //Some List(10, 11))
1 Match a Some(x) if x represents an existing value.
Option
Some None
c
c o
2 Match a None if x represents a non-existing value.
2

Pattern matching on immutable collection
import scala.collection.immutable._
def getTypegetTypegetTypegetType(x:Traversable[Any]) : String = x match{
case s:Set[Any] => s"A Immutable Set : $s"
case q:Seq[Any] => s"A Immutable Seq : $q"
case _ => "A Immutable Map : $x"
}
getType(Set(0,2,4)) //> A Immutable Set : Set(0, 2, 4)
getType(1 to 3) //> Immutable Seq : Range(1, 2, 3)
getType(HashMap[Int,Int](1->13)) //> A Immutable Map : Map(1 -> 13)
2 Match a pattern then the type of x is returned
Iterable
Set
Map
Seq
t
t
t
t
traversable
t
1
1
2
3
3 Match wildcard then type map is returned
This hierarchy is valid for mutable and immutable collections

Pattern matching on Immutable Map
def getTypeOfMapgetTypeOfMapgetTypeOfMapgetTypeOfMap(x:Map[Int,Int]) : String = x match{
case _:HashMap[Int,Int] => "A HashMap "
case _:TreeMap[Int,Int] => "A TreeMap "
case _ => "A ListMap "
}
getTypeOfMap(HashMap[Int,Int](1->12,2->10)) //>A HashMap
getTypeOfMap(TreeMap[Int,Int](1->3,6->6)) //> A TreeMap
getTypeOfMap(ListMap[Int,Int](1->13,6->16)) //> A ListMap
1 if x is neither a HashMap nor a TreeMap then x is a ListMap
Map
HashMap
ListMap
TreeMap
t
c
c
c
This hierarchy is valid only for the immutable Map
Map is a Trait and all (children) are implementation
classes
1

Pattern matching on immutable Set
import scala.collection.immutable.BitSet._
def getTypeOfSetgetTypeOfSetgetTypeOfSetgetTypeOfSet(x:Set[Int]) : String = x match{
case h:HashSet[Int] => "A HashSet " + (h.size,h.sum)
case t:TreeSet[Int] => "A TreeSet " + (t.size,t.sum)
case l:ListSet[Int] => "A ListSet " + (l.size,l.sum)
case z:AnyRef => "type : " + z.getClass.getName
}
getTypeOfSet(new BitSet1(10)) //> type : scala.collection.immutable.BitSet$BitSet1
getTypeOfSet(HashSet(1,10,12)) //> A HashSet (3,23)
getTypeOfSet(TreeSet(1,10)) //> A TreeSet (2,11)
getTypeOfSet(ListSet(1,11,12)) //> A ListSet (3,24)
Set
HashSet
ListSet
BitSet
t
c
c
c
This hierarchy is valid only for the immutable Set.
The Set is a Trait. BitSet is an abstract class and all children are
the class.
TreeSet
c
SortedSet
t
The subclasses of the abstract class BitSet are BitSet1, BitSet2 and BitSetN

Pattern matching on List 1/2
def sizeListsizeListsizeListsizeList[A](ls : List[A]) : Int = ls match{
case Nil => 0
case _::q => sizeList(q) + 1
}
The following function computes the length of a list.
It is based on two constructors NilNilNilNil and “::”.
sizeList(List()) //> res40: Int = 0
sizeList(List(1)) //> res41: Int = 1
sizeList(List(1,(1,2),3,4)) //> res42: Int = 4
1 If the list is empty, Nil, then we return 0.
2 If ls represents a no-empty list then the length of list is
(1+length(q)), where q represents the tail of the list.
1
2

Pattern matching on List 2/2
def identifyListeidentifyListeidentifyListeidentifyListe(ls: List[Any]): String = ls match {
case Nil => "empty list"
case _::Nil => "list with single element"
case _::_::Nil => "list has two elements"
case _ => "list has at least tree elements"
}
All lists are built from two constructors : NilNilNilNil and “::”....
NilNilNilNil represents a empty list
:::::::: builds a non empty list characterized by a head and a tail.
identifyListe(Nil) //> empty list
identifyListe(List((1,2))) //> list with single element
identifyListe(List("Scala", 3)) //> list has two elements
identifyListe(List(1,2,3)) //> list has at least tree elements

Pattern matching on case class
case class PersonPersonPersonPerson(name: String, age: Int)
def validateAgevalidateAgevalidateAgevalidateAge(p: Person): Option[String] = p match {
case Person(name, age) if age > 18 => Some(s"Hi $name! age $age")
case _ => None
}
Case classes are classes that get to String, hashCode, and equals methods
automatically. It provide a recursive decomposition mechanism via pattern
matching
validateAge(Person("Alice", 19)) //> Some(Hi Alice! age 19)
validateAge(Person("Bill", 15)) //> None
2 If the age is greater than18 , then we return the name and age of the person.
3 If the age of the person is less than 18 , then we return None.
2
3
1
1 We define a case class of a person with name and age.

Nested Pattern Matching in Case Classes
case class Address(street: String, city: String, country: String)
case class Person(name: String, age: Int, address:Address)
def checkPrscheckPrscheckPrscheckPrs(p: Person): Option[String] = p match {
case Person(name, age, null) if age > 18 => Some(s"Hi $name! age $age")
case Person(name, age, Address(_,_,ctr)) => Some(s"Hi $name! age $age $ctr")
case _ => None
}
checkPrs(Person("Alice", 19,null)) //> Some(Hi Alice! age 19)
checkPrs(Person("Bill", 15, Address("Scala Lane", "Chicago", "USA"))) //> Some(Hi Bill! age 15 USA)
2 If age > 18 & address = null, then we return the name and age of the person.
3 If the address != null then we return the name, age and the city of the person.
3
2
1 We define two case classes Address and Person. Person contains a subclass Address.
1
Case classes can contain other case classes, and the pattern matching can be nested.

Matching on regular expression
val email = """(.*)@(.*).(.*)""".r
val date = """(dddd)-(dd)-(dd)""".r
def checkPattern(x:String) : Option[String] = x match{
case email(name,domain,extension) => Some(s"$name $domain $extension")
case date(year, month, day) => Some(s"$year $month $day")
case _ => None
}
A regular expression is used to determine whether a string matches a pattern and,
if it does, to extract or transform the parts that match. Scala supports regular
expressions through Regex class available in the scala.util.matching package.
2
3
1
4
1 We define a regular expressions for email and date by using the method rmethod rmethod rmethod r2
3 If regex of email is matched then extract name, domain and extension
4 If regex of date is matched then extract year, month and day
checkPattern("info@scala.fr") //> Some(info scala fr)
checkPattern("2016-03-26") //> Some(2016 03 26)

References
• Scala Language Specification, http://www.scala-
lang.org/files/archive/spec/2.11/
• Documentation for the Scala standard library,
http://www.scala-lang.org/api/current/
• Beginning Scala, by Vishal Layka, David Pollak, 2015
• Programming in Scala, by Martin Odersky and al., 2011
• Programming Scala, by Dean Wampler, Alex Payne, 2014

Mustapha Michrafy PhD
Data Scientist and Expert Optim
at ATOS
Contact : datascience.km@gmail.com
Cyril Valliez PhD
Software Architect at Bouygues

Scala: Pattern matching, Concepts and Implementations

More Related Content

Scala: Pattern matching, Concepts and Implementations