(int a) -> a * 2; // Calculate the double of a a -> a * 2; // or simply without type(a, b) -> a + b; // Sum of 2 parametersIf the lambda is more than one expression we can use { } and return
(x, y) -> { int sum = x + y; int avg = sum / 2; return avg; }A lambda expression cannot stand alone in Java, it need to be associated to a functional interface.
interface MyMath { int getDoubleOf(int a); } MyMath d = a -> a * 2; // associated to the interface d.getDoubleOf(4); // is 8All examples with "list" use :
List<String> list = [Bohr, Darwin, Galilei, Tesla, Einstein, Newton]sort sort(list, comparator)
list.sort((a, b) -> a.length() - b.length()) list.sort(Comparator.comparing(n -> n.length())); // same list.sort(Comparator.comparing(String::length)); // same //> [Bohr, Tesla, Darwin, Newton, Galilei, Einstein]removeIf
list.removeIf(w -> w.length() < 6); //> [Darwin, Galilei, Einstein, Newton]merge merge(key, value, remappingFunction)
Map<String, String> names = new HashMap<>(); names.put("Albert", "Ein?"); names.put("Marie", "Curie"); names.put("Max", "Plank"); // Value "Albert" exists // {Marie=Curie, Max=Plank, Albert=Einstein} names.merge("Albert", "stein", (old, val) -> old.substring(0, 3) + val); // Value "Newname" don't exists // {Marie=Curie, Newname=stein, Max=Plank, Albert=Einstein} names.merge("Newname", "stein", (old, val) -> old.substring(0, 3) + val);Allows to reference methods (and constructors) without executing them
// Lambda Form: getPrimes(numbers, a -> StaticMethod.isPrime(a)); // Method Reference: getPrimes(numbers, StaticMethod::isPrime);| Method Reference | Lambda Form |
|---|---|
StaticMethod::isPrime | n -> StaticMethod.isPrime(n) |
String::toUpperCase | (String w) -> w.toUpperCase() |
String::compareTo | (String s, String t) -> s.compareTo(t) |
System.out::println | x -> System.out.println(x) |
Double::new | n -> new Double(n) |
String[]::new | (int n) -> new String[n] |
Similar to collections, but
- They don't store their own data
- The data comes from elsewhere (collection, file, db, web, ...)
- immutable (produce new streams)
- lazy (only computes what is necessary !)
// Will compute just 3 "filter" Stream<String> longNames = list .filter(n -> n.length() > 8) .limit(3);Create a new stream
Stream<Integer> stream = Stream.of(1, 2, 3, 5, 7, 11); Stream<String> stream = Stream.of("Jazz", "Blues", "Rock"); Stream<String> stream = Stream.of(myArray); // or from an array list.stream(); // or from a list // Infinit stream [0; inf[ Stream<Integer> integers = Stream.iterate(0, n -> n + 1);Collecting results
// Collect into an array (::new is the constructor reference) String[] myArray = stream.toArray(String[]::new); // Collect into a List or Set List<String> myList = stream.collect(Collectors.toList()); Set<String> mySet = stream.collect(Collectors.toSet()); // Collect into a String String str = list.collect(Collectors.joining(", "));map map(mapper)
Applying a function to each element
// Apply "toLowerCase" for each element res = stream.map(w -> w.toLowerCase()); res = stream.map(String::toLowerCase); //> bohr darwin galilei tesla einstein newton res = Stream.of(1,2,3,4,5).map(x -> x + 1); //> 2 3 4 5 6filter filter(predicate)
Retains elements that match the predicate
// Filter elements that begin with "E" res = stream.filter(n -> n.substring(0, 1).equals("E")); //> Einstein res = Stream.of(1,2,3,4,5).filter(x -> x < 3); //> 1 2reduce
Reduce the elements to a single value
String reduced = stream .reduce("", (acc, el) -> acc + "|" + el); //> |Bohr|Darwin|Galilei|Tesla|Einstein|Newtonlimit limit(maxSize) The n first elements
res = stream.limit(3); //> Bohr Darwin Galileiskip Discarding the first n elements
res = strem.skip(2); // skip Bohr and Darwin //> Galilei Tesla Einstein Newtondistinct Remove duplicated elemetns
res = Stream.of(1,0,0,1,0,1).distinct(); //> 1 0sorted Sort elements (must be Comparable)
res = stream.sorted(); //> Bohr Darwin Einstein Galilei Newton Tesla allMatch
// Check if there is a "e" in each elements boolean res = words.allMatch(n -> n.contains("e"));anyMatch: Check if there is a "e" in an element
noneMatch: Check if there is no "e" in elements
parallel Returns an equivalent stream that is parallel
findAny faster than findFirst on parallel streams
Wrappers (like Stream) are inefficients. It requires a lot of unboxing and boxing for each element. Better to use IntStream, DoubleStream, etc.
Creation
IntStream stream = IntStream.of(1, 2, 3, 5, 7); stream = IntStream.of(myArray); // from an array stream = IntStream.range(5, 80); // range from 5 to 80 Random gen = new Random(); IntStream rand = gen(1, 9); // stream of randomsUse mapToX (mapToObj, mapToDouble, etc.) if the function yields Object, double, etc. values.
Collectors.groupingBy
// Groupe by length Map<Integer, List<String>> groups = stream .collect(Collectors.groupingBy(w -> w.length())); //> 4=[Bohr], 5=[Tesla], 6=[Darwin, Newton], ...Collectors.toSet
// Same as before but with Set ... Collectors.groupingBy( w -> w.substring(0, 1), Collectors.toSet()) ...Collectors.counting Count the number of values in a group
Collectors.summing__ summingInt, summingLong, summingDouble to sum group values
Collectors.averaging__ averagingInt, averagingLong, ...
// Average length of each element of a group Collectors.averagingInt(String::length)PS: Don't forget Optional (like Map<T, Optional<T>>) with some Collection methods (like Collectors.maxBy).
Creation
Stream<String> parStream = list.parallelStream(); Stream<String> parStream = Stream.of(myArray).parallel();unordered Can speed up the limit or distinct
stream.parallelStream().unordered().distinct();PS: Work with the streams library. Eg. use filter(x -> x.length() < 9) instead of a forEach with an if.
In Java, it is common to use null to denote absence of result. Problems when no checks: NullPointerException.
// Optional<String> contains a string or nothing Optional<String> res = stream .filter(w -> w.length() > 10) .findFirst(); // length of the value or "" if nothing int length = res.orElse("").length(); // run the lambda if there is a value res.ifPresent(v -> results.add(v));Return an Optional
Optional<Double> squareRoot(double x) { if (x >= 0) { return Optional.of(Math.sqrt(x)); } else { return Optional.empty(); } }Note on inferance limitations
interface Pair<A, B> { A first(); B second(); }A steam of type Stream<Pair<String, Long>> :
stream.sorted(Comparator.comparing(Pair::first)) // okstream.sorted(Comparator.comparing(Pair::first).thenComparing(Pair::second)) // dont work
Java cannot infer type for the .comparing(Pair::first) part and fallback to Object, on which Pair::first cannot be applied.
The required type for the whole expression cannot be propagated through the method call (.thenComparing) and used to infer type of the first part.
Type must be given explicitly.
stream.sorted( Comparator.<Pair<String, Long>, String>comparing(Pair::first) .thenComparing(Pair::second) ) // okThis cheat sheet was based on the lecture of Cay Horstmann http://horstmann.com/heig-vd/spring2015/poo/