CODEWITHSUNDEEP
javajava-8java-stream
V. Pantis
V. Pantis

Reputation: 115

Java stream filter sum of values

I have a class called MonitoredData with which describes an activity, its starting time and ending time. The attributes are activityLabel, startTime, endTime. I have to group and filter these activities, using streams, the activities which have the total duration of more than 10 hours. I managed to make the sum of durations and group them according to the activity using this:

Map<String, Long> map4 = new HashMap<String, Long>();

map4 = data.stream()
    .collect(
        Collectors.groupingBy(
            MonitoredData::getActivity, 
            Collectors.summingLong(MonitoredData::getDuration)
        )
    ); //getDuration returns end Time - startTime in milliseconds

But I haven't managed to add a filter. I tried using: 

.filter(Collectors.summingLong(MonitoredData::getDuration) > whatever)

but obviously it doesn't work. How can I solve this in order to make it return a Map<String, Long>?

Upvotes: 9

Views: 7898

Answers (4)

fps
fps

Reputation: 34460

I would first do as you've already done: I'd collect the stream of MonitoredData instances to a map, grouping by activity and summing the duration of each activity in each value:

Map<String, Long> map4 = data.stream()
    .collect(Collectors.groupingBy(
        MonitoredData::getActivity, 
        HashMap::new,
        Collectors.summingLong(MonitoredData::getDuration)));

The nuance is that I'm using the overloaded version of Collectors.groupingBy that accepts a factory for the map, because in the next step I want to remove the entries whose duration is less than 10 hours, and the spec doesn't guarantee that the map returned by the Collectors.groupingBy methods that take either one or two arguments is mutable.

This is how I'd remove the entries that don't match:

public static final long TEN_HOURS_MS = 10 * 60 * 60 * 1000;

map4.values().removeIf(v -> v < TEN_HOURS_MS);

If you want to do everything in a single line, you might want to use Collectors.collectingAndThen:

Map<String, Long> map4 = data.stream()
    .collect(Collectors.collectingAndThen(
        Collectors.groupingBy(
            MonitoredData::getActivity, 
            HashMap::new,
            Collectors.summingLong(MonitoredData::getDuration)),
        m -> { m.values().removeIf(v -> v < TEN_HOURS_MS); return m; } ));

Here I've used Collectors.collectingAndThen to modify the map returned by Collectors.groupingBy. And, within the finisher function, I've used Collection.removeIf, which takes a predicate and removes all the entries that match that predicate.

Upvotes: 6

holi-java
holi-java

Reputation: 30676

How about this? I answer on phone, you need test by yourself.

map = map.entrySet()
         .stream()
         .filter(it->TimeUnit.MILLISECONDS.toHours(it.getValue())>10)
         .collect(Collectors.toMap(Map.Entry::getKey,Map.Entry::getValue));

OR using Collectors#collectingAndThen:

map4 = data.stream()
.collect(Collectors.collectingAndThen(    
    Collectors.groupingBy(
        MonitoredData::getActivity, 
        Collectors.summingLong(MonitoredData::getDuration)
    ),
    r1 -> r1.entrySet().stream()
            .filter(it->TimeUnit.MILLISECONDS.toHours(it.getValue())>10)
            .collect(Collectors.toMap(Map.Entry::getKey,Map.Entry::getValue))
));

Upvotes: 4

NiVeR
NiVeR

Reputation: 9786

Add the below code after you got the map:

map4.entrySet().stream()
                .filter(a -> a.getValue() > whatever)
                .collect(Collectors.joining());

Upvotes: 2

Shadov
Shadov

Reputation: 5592

I think this is what you wanted, using Google Guava Maps class:

Map<String, Long> map = Maps.filterValues(data.stream()
            .collect(Collectors.groupingBy
                    (MonitoredData::getActivity, Collectors.summingLong
                            (MonitoredData::getDuration)
                    )), value -> value > 3);

You can obviously write your own method to filter map like that, but since it's already there, on so popular library... How to do it with pure streams: I don't know, but maybe this will be satisfying.

Upvotes: 3

Related Questions