Java 8 Stream API - reduce operator

Right after blog post covering Stream.flatMap() operator, we'll also cover another important Stream operator called - reduce. We'll go straight away with an example.

Say we have a simple Person Java Bean:

package rs.dodalovic.streams.reduce;

class Person {
    private String name;
    private Gender gender;
    private Integer age;

    public Person(String name, Gender gender, Integer age) {
        this.name = name;
        this.gender = gender;
        this.age = age;
    }

    static enum Gender {
        MALE, FEMALE;
    }

    public String getName() {
        return name;
    }

    public void setName(String name) {
        this.name = name;
    }

    public Gender getGender() {
        return gender;
    }

    public Integer getAge() {
        return age;
    }

    public void setAge(Integer age) {
        this.age = age;
    }

    @Override
    public String toString() {
        return "Person{" +
                "name='" + name + '\'' +
                ", gender=" + gender +
                '}';
    }
}

Our task is to iterate through list of persons and find:

Female person with the longest name and print the details
Longest female name only and print the details
Youngest male and print his/her age

Here's our reduce operator showcase right below:

package rs.dodalovic.streams.reduce;

import java.util.Arrays;
import java.util.List;
import java.util.Optional;
import java.util.OptionalInt;
import java.util.function.Predicate;

public class ReduceDemo {
    public static void main(String[] args) {

        final List<Person> persons = Arrays.asList(
                new Person("Jeanny", Person.Gender.FEMALE, 22),
                new Person("Jim", Person.Gender.MALE, 25),
                new Person("Alicia", Person.Gender.FEMALE, 32),
                new Person("Jennifer", Person.Gender.FEMALE, 38),
                new Person("Anastasia", Person.Gender.FEMALE, 19),
                new Person("John", Person.Gender.MALE, 22),
                new Person("Jimmy", Person.Gender.MALE, 38),
                new Person("Michael", Person.Gender.MALE, 15)
        );
        printFemaleWithLongestName(persons);
        printLongestFemaleNameOnly(persons);
        printYoungestMaleAge(persons);
        printYoungestMaleAgeUsingMin(persons);
    }

    private static void printFemaleWithLongestName(List<Person> persons) {
        final Optional<Person> personWithLongestName = persons.stream()
                .filter(onlyGender(Person.Gender.FEMALE))
                .reduce((person, person2) -> person.getName().length() > person2.getName().length() ? person : person2);
        personWithLongestName.ifPresent(person -> System.out.printf("Person with longest female name %s%n",
                personWithLongestName));
    }

    private static void printLongestFemaleNameOnly(List<Person> persons) {
        persons.stream()
                .filter(onlyGender(Person.Gender.FEMALE))
                .map(Person::getName)
                .reduce((person1Name, person2Name) -> person1Name.length() > person2Name.length() ? person1Name :
                        person2Name)
                .ifPresent(longestName -> System.out.printf("Longest female name: %s%n", longestName));
    }

    private static void printYoungestMaleAge(List<Person> persons) {
        final Integer youngestMaleAge = persons.stream()
                .filter(onlyGender(Person.Gender.MALE))
                .map(Person::getAge)
                .reduce(Integer.MAX_VALUE, (age1, age2) -> age1 < age2 ? age1 : age2);
        System.out.printf("Youngest male age: %d%n", youngestMaleAge);
    }

    private static void printYoungestMaleAgeUsingMin(List<Person> persons) {
        final OptionalInt youngest = persons.stream()
                .filter(onlyGender(Person.Gender.MALE))
                .mapToInt(Person::getAge)
                .min();
        youngest.ifPresent(value -> System.out.printf("Youngest male age: %d", value));
    }

    private static Predicate<Person> onlyGender(Person.Gender gender) {
        return person -> person.getGender() == gender;
    }
}

First - we initialize our sample person list that we'll operate on. To find the person with the longest name, we call printFemaleWithLongestName() method. What we do there is first taking stream out of list (call to stream() method), followed by filtering stream to keep Gender.FEMALE persons only and then calling our reduce() method. The parameter it receives is BinaryOperator<T, T, T>, which is basically a Function receiving two (same type) parameters, and returning value of that same type. This Function will be called either:

n-1 times, where n is number of elements in underlying collection, in case we didn't supply initial / default value to reduce() call
n times otherwise

So, in case we have collection of elements A, B, C, D - function will be called 3 times (if we haven't supplied initial value to reduce()) with parameters:

the first call will be with A, B elements of collection
the second call will take result of previous function call as the first value, and C as the second
similarly, the third call will take result of second function call as the first value, and D as the second

If we passed the initial/default value, function would have been called 4 times, with the only difference that the first call will have that initial/default value as the first parameter, and A as the second one. Afterwards algorithm stays the same.

In case of printFemaleWithLongestName() function, for each pair of persons combinations we ask:

(person, person2) -> person.getName().length() > person2.getName().length() ? person : person2

In case person's name is longer than person2's - we return the first person, otherwise - person2. So, after traversing each two pairs of persons, we'll at the end have as a result of call to reduce()

Optional.

Returning Optional is fine since it might be the case that list we're streaming through might be empty one, so we can't actually get the person with longest name there. That's why, in this case, we call ifPresent() at the end - passing it Consumer to do with our Person object whatever we like.

The other example, printLongestFemaleNameOnly , only differs that we're calling map() to transform Person instances to Person instance name field. After that we similarly call reduce operator, comparing lengths in there, and the same as the first time, if underlying collection wasn't empty, we extract from resulting Optional value, which represents longest female name in given list.

In case of printYoungestMaleAge method, we're first filter to keep males only by passing Predicate onlyGender(Person.Gender.FEMALE) to filter() call. Afterwards, we're transform underlying list ato convert to list of person ages by calling map() passing Person::getAge conversion function to it. After that we're calling reduce() function with slightly overloaded signature. This one accepts initial / default value as a first parameter, and BinaryOperator as second one. If underlying collection is empty, reduce() will return initial/default value passed to it (Integer.MAX_VALUE), otherwise - similarly what we had in our previous cases, we iterate through all pairs of Persons and return finally the one that's the youngest. The only difference, in this case is that if underlying collection is empty(), we won't return empty Optional but rather default value we passed as the first parameter to reduce() function.

Another, slightly easier way to achieve minimum age calculation is to use min() terminal operator on IntStream. This is demonstrated in printYoungestMaleAgeUsingMin method. The only differences are that instead of calling map(), we call mapToInt() which converts existing stream to IntStream, which is convenient since it has methods such as sum(), min(), max().

Source code as Github gist

Key takeaways

Reduce operator is terminal operator that helps us extract single value from underlying stream. Accumulator function is called either n-1 times for collection of n elements if we haven't provided initial / default value as the first parameter, or n times - if we did (see algorithm description above).

That was all for today! Hope you liked it!

Key takeaways​

Key takeaways