Group by a Collection attribute using Java Streams

Question

I have an object that contains a Collection of strings, let's say the languages that a person speaks.

public class Person {    private String name;    private int age;    private List<String> languagesSpoken;     // ... } 

Now, creating some instances like this...

Person p1 = new Person("Bob", 21, Arrays.asList("English", "French", "German")); Person p2 = new Person("Alice", 33, Arrays.asList("English", "Chinese", "Spanish")); Person p3 = new Person("Joe", 43, Arrays.asList("English", "Dutch", "Spanish", "German"));  //put them in list List<Person> people = Arrays.asList(p1,p2,p3); 

... what I want to have is a Map<String, List<Person>>, for every language listing the persons that speak the language:

["English" -> [p1, p2, p3],  "German"  -> [p1, p3],  etc. ] 

Of course this can be programmed easily in an imperative way, but how to do it the functional way with Java Streams? I have tried something like people.stream.collect(groupingBy(Person::getLanguagesSpoken)) but that of course gives me a Map<List<String>, List<Person>>. All the examples I could find, are using groupingBy on Primitives or Strings.

Answer 1

You can break the Person instances into pairs of language and Person (using flatMap) and then group them as required:

Map<String, List<Person>> langPersons =     people.stream()           .flatMap(p -> p.getLanguagesSpoken()                          .stream()                          .map(l -> new SimpleEntry<>(l,p)))           .collect(Collectors.groupingBy(Map.Entry::getKey,                                          Collectors.mapping(Map.Entry::getValue,                                                             Collectors.toList()))); 

Answer 2

This is possible to do without streams too, still using java-8 new features.

people.forEach(x -> {         x.getLanguagesSpoken().forEach(lang -> {             langPersons.computeIfAbsent(lang, ignoreMe -> new ArrayList<>()).add(x);         }); });