如何使用Java 8 Stream扩展并重新组合List列表？

Question

I have a list of the Class A, that includes a List itself.

public class A {
    public double val;
    public String id;
    public List<String> names = new ArrayList<String>();
    public A(double v, String ID, String name)
    {
        val = v;
        id = ID;
        names.add(name);
    }

static public List<A> createAnExample()
{
    List<A> items  = new ArrayList<A>();

    items.add(new A(8.0,"x1","y11"));
    items.add(new A(12.0, "x2", "y21"));
    items.add(new A(24.0,"x3","y31"));
    items.get(0).names.add("y12");
    items.get(1).names.add("y11");
    items.get(1).names.add("y31");
    items.get(2).names.add("y11");
    items.get(2).names.add("y32");
    items.get(2).names.add("y33");
    return  items;
}

The aim is to sum over average val per id over the List. I added the code in Main function by using some Java 8 stream. My question is how can I rewrite it in a more elegant way without using the second Array and the for loop.

static public void main(String[] args) {
    List<A> items = createAnExample();

    List<A> items2 = new ArrayList<A>();
    for (int i = 0; i < items.size(); i++) {
        List<String> names = items.get(i).names;
        double v = items.get(i).val / names.size();
        String itemid = items.get(i).id;
        for (String n : names) {
            A item = new A(v, itemid, n);
            items2.add(item);
        }
    }
    Map<String, Double> x = items2.stream().collect(Collectors.groupingBy(item ->
            item.names.isEmpty() ? "NULL" : item.names.get(0), Collectors.summingDouble(item -> item.val)));
    for (Map.Entry entry : x.entrySet())
        System.out.println(entry.getKey() + " --> " + entry.getValue());
}

Answer 1

You can do it with flatMap :

x = items.stream()
    .flatMap(a -> a.names.stream()
        .map(n -> new AbstractMap.SimpleEntry<>(n, a.val / a.names.size()))
    ).collect(groupingBy(
        Map.Entry::getKey, summingDouble(Map.Entry::getValue)
    ));

If you find yourself dealing with problems like these often, consider a static method to create a Map.Entry :

static<K,V> Map.Entry<K,V> entry(K k, V v) {
    return new AbstractMap.SimpleImmutableEntry<>(k,v);
}

Then you would have a less verbose .map(n -> entry(n, a.val/a.names.size()))

Answer 2

In my free StreamEx library which extends standard Stream API there are special operations which help building such complex maps. Using the StreamEx your problem can be solved like this:

Map<String, Double> x = StreamEx.of(createAnExample())
    .mapToEntry(item -> item.names, item -> item.val / item.names.size())
    .flatMapKeys(List::stream)
    .grouping(Collectors.summingDouble(v -> v));

Here mapToEntry creates stream of map entries (so-called EntryStream ) where keys are lists of names and values are averaged vals. Next we use flatMapKeys to flatten the keys leaving values as is (so we have stream of Entry<String, Double> ). Finally we group them together summing the values for repeating keys.