Streaming.md

Streaming the IntervalTree

When iterating over a Java Collection, it is pretty common to do so by using an Iterator. There are many discussion on how to iterate over an amount of items, e.g., 1, 2.

Besides the nicely readable code, Streams may have positive or negative impact on the performance of someones application. Until version 1.6.3 the IntervalTree provided "only" a stream() and iterator(), as needed by the Collection interface, whereby the stream() implementation was based on the iterator(). Both of these methods enabled the user to iterate over the full content of the IntervalTree.

Since 1.6.3 the IntervalTree also provides a overlapStream() method, which "pushes" the overlapping intervals through the stream.

Examples:

    public void example() {
        final IntervalTree tree = IntervalTreeBuilder.newBuilder()
                .usePredefinedType(IntervalType.NUMBER, false)
                .collectIntervals(interval -> new ListIntervalCollection())
                .enableWriteCollections()
                .build();

        // let's add 1_000_000 intervals from 0, 1_000_000 -> 99, 1_000_000 and a random id
        IntStream.range(0, 1_000_000).forEach(i ->
                tree.insert(new IdInterval<>(UUID.randomUUID(), i, 1_000_000)));

        // Now let's find 3 intervals, which:
        //   1. overlap with [1, 10_000] (which are all), and
        //   2. have an `id` (or whatever attribute) that starts with "1"
        // The streaming will ensure stop any further calculation, when 3 intervals are found
        final List<IdInterval<UUID, Integer>> res = tree.overlapStream(new IntegerInterval(1, 10_000))
                .map(interval -> (IdInterval<UUID, Integer>) interval)
                .filter(interval -> interval.getId().toString().startsWith("1"))
                .limit(3)
                .collect(Collectors.toList());

        // output
        System.out.println(res);
    }

Further Information

• Processing Data with Java SE 8 Streams