java8 Stream的filter（）和map（）方法是否使用迭代？

Question

I have a POJO in Person.java file:

public class Person {
    private String name;
    private int age;

    public Person(String n, int a) {
        name = n;
        age = a;
    }

    public String getName() {
        return name;
    }

    public int getAge() {
        return age;
    }

    public boolean isAdult() {
        return getAge() >= 18;    
    }
}

And then I have a Demo.java file which creates a list of persons and uses streams to filter and print content from the list:

import java.util.*;

public class Demo {
    public static void main(String[] args) {
        List<Person> people = createPeople();
        List<String> names = people.stream()
                                   .filter(person -> person.isAdult())
                                   .map(person -> person.getName())
                                   .collect(toList());
        System.out.println(names);
    }

    private static List<Person> createPeople() {
        List<Person> people = new ArrayList<>();
        people.add("John", 19);
        people.add("Joe", 21);
        people.add("Jill", 16);
        people.add("Sylvester", 18);
        people.add("Hillary", 17);
        people.add("Donald", 4);

        return people;
    }
}  

I wanted to know:

1> Does filter() and map() internally use a loop to iterate over all the Person objects in the List people ?

2> If yes, do they loop over all the objects in the list two different times (1st iteration by filter() and other by map() )?

3> If yes again, if I add another map() or filter() method, will it loop over all the objects again for the third time?

4> If yes again, then how is it different performance wise from our traditional imperative style coding (in-fact, in traditional imperative style, we could do all the filtering and mapping in 1 single loop most of the times. So performance wise, imperative style coding would perform better than streams in such a case.)?

PS: If there is a No to any of the above questions, please add an explanation regarding how things work then.

One more: Is there a difference in iteration done by the stream internally and the iteration we do in imperative style? Please shed some more light to my knowledge with some detailed explaination.

Answer 1

First, your code doesn't compile, because map() returns a Stream , not a List . You must end the stream chain with a terminal operation, and both filter() and map() are intermediate operations . Says so right there in the javadoc. In your case, you need to add .collect(Collectors.toList()) to make it compile and run ok.

1> Does filter() and map() internally use a loop to iterate over all the Person objects in the List people?

No. The terminal operation is doing the looping.

Since questions 2 to 4 assumes a Yes answer, they have no answer.

If there is a No to any of the above questions, please add an explanation regarding how things work then.

Read the documentation , or search the web. It's pretty well explained.

Is there a difference in iteration done by the stream internally and the iteration we do in imperative style?

Yes, eg streams can utilize parallel thread execution. Even single-threaded there is a difference in how the entire operation works, though logically, at a high level, they are essentially the same.

---

Example

In your code, with the collect() call added, the equivalent imperative code would be:

List<String> names = new ArrayList<>();
for (Person person : people)
    if (person.isAdult())
        names.add(person.getName());

To compare to what the stream logic does, first you define the lambdas passed to filter() and map() :

Predicate<Person>        filter = person -> person.isAdult();
Function<Person, String> map    = person -> person.getName();

Then you get the Collector by calling Collectors.toList() , and retrieve the objects it provide:

Collector<String, List<String>, List<String>> collector = (Collector) Collectors.toList();
Supplier<List<String>>               supplier    = collector.supplier();
BiConsumer<List<String>, String>     accumulator = collector.accumulator();
Function<List<String>, List<String>> finisher    = collector.finisher();

Now, the stream() call basically provides an Iterator (it's actually a Spliterator ) and the collect call will iterate, so combined they are equivalent to the for loop. I won't cover the full logic of how Stream , Spliterator and collect() works. Search the web if you need more detail on that.

So, the imperative for loop above becomes:

List<String> list = supplier.get();        // list = new ArrayList<>()
for (Person person : people)               // collect() loop using Spliterator from stream()
    if (filter.test(person)) {             // if (person.isAdult())
        String value = map.apply(person);  // value = person.getName()
        accumulator.accept(list, value);   // list.add(value)
    }
List<String> names = finisher.apply(list); // names = list