Vedanth
Vedanth

Reputation: 285

Java stream filter items of specific index

I'm looking for a concise way to filter out items in a List at a particular index. My example input looks like this:

List<Double> originalList = Arrays.asList(0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0);
List<Integer> filterIndexes = Arrays.asList(2, 4, 6, 8);

I want to filter out items at index 2, 4, 6, 8. I have a for loop that skips items that match the index but I was hoping there would be an easy way of doing it using streams. The final result would look like that:

List<Double> filteredList = Arrays.asList(0.0, 1.0, 3.0, 5.0, 7.0, 9.0, 10.0);

Upvotes: 17

Views: 30843

Answers (4)

Oleg Cherednik
Oleg Cherednik

Reputation: 18253

Probably you can use simple Iterator?

Stream<Integer> rows = Stream.of(1, 2, 3);
Iterator<Integer> it = rows.iterator();
int pos = 0;

while (it.hasNext()) {
    Integer num = it.next();

    if (pos++ == 0) {
        // this is a head
    } else {
        // this is a record
    }
}

Upvotes: 0

slartidan
slartidan

Reputation: 21608

If you sort your indexes descending, then you can use java.util.List.remove(int) to remove the items.

List<Double> originalList = new ArrayList<>(Arrays.asList(0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0));
List<Integer> filterIndexes = Arrays.asList(2, 4, 6, 8);

filterIndexes.stream()

    // remove higher indixes first, because each remove affects all indexes to the right
    .sorted(Comparator.reverseOrder())

    // make sure to use remove(int) not remove(Object) in java.util.List to use indexes
    .mapToInt(Integer::intValue)

    // remove each index
    .forEach(originalList::remove);

// print results
originalList.forEach(System.out::println);

Upvotes: 3

Tagir Valeev
Tagir Valeev

Reputation: 100349

If your filteredIndexes list is presorted, you can avoid checking every element in this way:

List<Double> filteredList = IntStream.rangeClosed(0, filterIndexes.size())
    .mapToObj(idxPos -> idxPos == 0 
           ? originalList.subList(0, filterIndexes.get(idxPos)) 
           : idxPos == filterIndexes.size() 
           ? originalList.subList(filterIndexes.get(idxPos-1)+1, originalList.size()) 
           : originalList.subList(filterIndexes.get(idxPos-1)+1, filterIndexes.get(idxPos)))
    .flatMap(List::stream)
    .collect(Collectors.toList());

Here we create a number of sublists which contain all the elements between the filtered indices, then just flatten them into the single final list. For big input (e.g. a million of numbers) this solution could be magnitudes faster than one proposed by @AlexisC.

Upvotes: 6

Alexis C.
Alexis C.

Reputation: 93902

You can generate an IntStream to mimic the indices of the original list, then remove the ones that are in the filteredIndexes list and then map those indices to their corresponding element in the list (a better way would be to have a HashSet<Integer> for indices since they are unique by definition so that contains is a constant time operation).

List<Double> filteredList = 
    IntStream.range(0, originalList.size())
             .filter(i -> !filterIndexes.contains(i))
             .mapToObj(originalList::get)
             .collect(Collectors.toList());

Upvotes: 30

Related Questions