How to flatten all items from a nested Java Collection into a single List?

Given a complex nested collection of objects such as:

Set<List<Map<String, List<Object>>>> complexNestedCollection;

Does a generic method exist to flatten this out and get a single List of all Objects contained within?

A few details:

  1. The list shouldn't include collection objects themselves or map keys - only the values at the lowest level.
  2. It should follow the same ordering where possible - so in the example, items in the list would be in order, whereas ordering of maps/sets would depend on the implementation.
  3. It could optionally exclude duplicates
  4. UPDATE: It should ideally detect/handle circular references at any level, e.g. a List<List<Object>> where the outer List contains itself as a member. (Credit to Adrian Jałoszewski for mentioning this in the comments below).

Note: The actual use case is to get all Strings from a List<List<String>>, which can be done easily enough with two loops but it made me wonder about the general case.


Assuming that you use Java 8, you could do that with the Stream API thanks to the method flatMap(Function<? super T,? extends Stream<? extends R>> mapper) as next:

// 1. Convert the Set as a Stream of List<Map<String, List<Object>>>
// 2. Extract the elements of the lists to get a Stream of Map<String, List<Object>>
// 3. Extract values of the maps to get a Stream of List<Object>
// 4. Extract the elements of the lists to get a Stream of Object
// 5. Get rid of duplicates
// 6. Collect the result as a List of Object
List<Object> result = complexNestedCollection.stream()
    .flatMap(List::stream)
    .flatMap(m -> m.values().stream())
    .flatMap(List::stream)
    .distinct()
    .collect(Collectors.toList());

<R> Stream<R> flatMap(Function<? super T,? extends Stream<? extends R>> mapper)

Returns a stream consisting of the results of replacing each element of this stream with the contents of a mapped stream produced by applying the provided mapping function to each element. Each mapped stream is closed after its contents have been placed into this stream. (If a mapped stream is null an empty stream is used, instead.)


For previous versions of Java, you can still use FluentIterable from Google Guava to replace the Stream and use transformAndConcat(Function<? super E,? extends Iterable<? extends T>> function) instead of flatMap to flatten your collection.

The previous code snippet would then be rewritten as next:

List<Object> result =
    new ArrayList<>(
        new LinkedHashSet<>(
            FluentIterable.from(complexNestedCollection)
                .transformAndConcat(
                    new Function<List<Map<String, List<Object>>>, Iterable<Map<String, List<Object>>>> () {
                        public Iterable<Map<String, List<Object>>> apply(final List<Map<String, List<Object>>> input) {
                            return input;
                        }
                    }
                ).transformAndConcat(
                    new Function<Map<String, List<Object>>, Iterable<List<Object>>> () {
                        public Iterable<List<Object>> apply(final Map<String, List<Object>> input) {
                            return input.values();
                        }
                    }
                ).transformAndConcat(
                    new Function<List<Object>, Iterable<Object>> () {
                        public Iterable<Object> apply(final List<Object> input) {
                            return input;
                        }
                    }
                ).toList()
        )
    );

I'm not sure if this exact implementation would work, as it's full of unchecked warnings and other dangerous stuff, but you should get the general idea.

public static Set<Object> recursiveExtract(Object stuff) {

    Set<Object> set = new HashSet<Object>();

    if(stuff instanceof Iterable) {
        for(Object o : (Iterable<?>)stuff) {
            set.addAll(recursiveExtract(o));
        }
    } else if(stuff instanceof Map) {
        for(Object o : ((Map<?, ? extends Object>) stuff).values()) {
            set.addAll(recursiveExtract(o));
        }
    } else {
        set.add(stuff);
    }

    return set;
}

You can also use List<Object> if you insist on List, but then you could get duplicate results, or LinkedHashSet<Object> if you care about the order.


Please instead of downvotes, give me suggestions for improvement. It's nicer.