Java 8: vertical slicing an array of Lists

Question

I'm learning Java 8 lambdas and streams. So, I got an array of lists having varying lengths. Lists contain integers.

What would be the best way to collect vertical slices in another list of lists i.e collect all integers with index 0 from all original lists in slice 0, index 1 in slice 1 and so an until the length of the longest list (filling zeros for shorter lists)

I know it's trivial to hand code a couple of traditional loops for this but what about doing this using Java 8 features?

Holger · Accepted Answer · 2015-01-08 11:38:16Z

4

That's a pretty interesting question - thanks for posting. I'm sure you'll see some interesting answers. Here's my attempt:

List<Integer> source[];
List<List<Integer>> slices = IntStream.range(0, Arrays.stream(source).mapToInt(List::size).max().getAsInt())
     .mapToObj(index -> Arrays.stream(source).map(list -> list.size() > index ? list.get(index) : 0)
         .collect(Collectors.toList()))
     .collect(Collectors.toList())

edited Jan 8, 2015 at 11:38

Holger

301k43 gold badges481 silver badges827 bronze badges

answered Jan 8, 2015 at 10:04

sprinter

28.1k8 gold badges51 silver badges81 bronze badges

Sign up to request clarification or add additional context in comments.

5 Comments

SubOptimal Over a year ago

It doesn't compile. The compiler error for max() is required: Comparator<? super Integer>.

Holger Over a year ago

@SubOptimal: I took the freedom to fix the compiler error as the solution should work in general.

sprinter Over a year ago

Oops forgot to map to int. Thanks for fixing Holger.

tshymbra Over a year ago

@sprinter got the code, now what about "inverted" solution i.e. which in the first line goes through stream of lists in the source array and via a running index collects values into slices from each list. I'm not sure firing up internal stream all over again on each index in IntStream.range is very effective.

Holger Over a year ago

@tshymbra: if you want the “inverted solution”, look at my answer. But it doesn’t matter how many Streams are created as it’s not the stream itself that matters as a Stream is only a lightweight object. What matters is that it is iterating over the source again for every index, hence performs max list size × number of lists operations but that’s unavoidable. In theory, the order of iterations might have an implication on cache usage but the effects are not really predictable and might differ from application to application. So recall, “optimize only if you have real performance problems”

Holger · Accepted Answer · 2015-01-08 12:27:52Z

Here is a solution which does not insert padding values for shorter Lists:

List<List<Integer>> slices = Stream.of(source).flatMap(l->
    IntStream.range(0, l.size()).mapToObj(i->new int[]{i, l.get(i)})
).collect(collectingAndThen(
    groupingBy(a->a[0], TreeMap::new, mapping(a->a[1], toList())),
    m->new ArrayList<>(m.values()))
);

this can be expanded to a zero-padding version like this:

int maxSize=IntStream.range(0,source.length).map(i->source[i].size()).max().orElse(0);
List<List<Integer>> slices = Stream.of(source).flatMap(l->
    Stream.concat(
      IntStream.range(0, l.size()).mapToObj(i->new int[]{i, l.get(i)}),
      IntStream.range(l.size(), maxSize).mapToObj(i->new int[]{i, 0})
    )
).collect(collectingAndThen(
    groupingBy(a->a[0], TreeMap::new, mapping(a->a[1], toList())),
    m->new ArrayList<>(m.values()))
);

Both solutions assume that you do import static java.util.stream.Collectors.*; as otherwise the code becomes really unreadable.

Collectives™ on Stack Overflow

Java 8: vertical slicing an array of Lists

2 Answers 2

5 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

5 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related