Deep "flatReduce" for WebFlux?

Question

I have defined an operator flatReduce() which is to reduce() what flatMap() is to map():

public class FlatReduce {
    public static <V, W> Mono<V> flatReduce(Mono<V> initial, Iterable<W> items, BiFunction<V, W, Mono<V>> accumulator) {
        for (W item : items) {
            initial = initial.flatMap(v -> accumulator.apply(v, item));
        }
        return initial;
    }

    public static void main(String[] args) {
        flatReduce(Mono.just(1), IntStream.range(0, 4000).mapToObj(it -> it).collect(toList()), (a, b) -> Mono.just(a + b)).block();
    }
}

This yields deeply nested flatMaps and I have observed stack overflows. Is there any way to work around this (maybe turning this into continuation style)?

Thanks.

This sounds super interesting, but could you please explain in more detail and with an example, what flatReduce does? *"to `reduce()` what `flatMap()` is to `map()`"* evokes things in my mind, but none of those things sound very practical, nor sound like they could be implemented efficiently. — Stef, Sep 05 '21 at 13:09

score 1 · Accepted Answer · answered Jan 04 '21 at 13:23

Found a workaround by converting the Mono to CompletableFuture.

    @Test
    public void runTest() {
        System.out.println(flatReduce(Mono.just(1), IntStream.range(1, 500000).mapToObj(it -> it).collect(toList()), (a, b) -> Mono.just(a + 1)).block());
    }

    private static <V, W> Mono<V> flatReduce(Mono<V> initial, List<W> items, BiFunction<V, W, Mono<V>> accumulator) {
        return Mono.fromCompletionStage(flatReduceWithFuture(initial.toFuture(), items, (v, w) -> accumulator.apply(v, w).toFuture()));
    }

    private static <V, W> CompletableFuture<V> flatReduceWithFuture(CompletableFuture<V> initial, List<W> items, BiFunction<V, W, CompletableFuture<V>> accumulator) {
        for (W item : items) {
            initial = initial.thenCompose(x -> accumulator.apply(x, item));
        }
        return initial;
    }

Deep "flatReduce" for WebFlux?

1 Answers1