Reputation: 9549
My stream function sometime returns null
, when I do collect them How to delete that null
return?
versions.stream().map(vs->{
if(vs.matches("^matched string$")) {
...
return new VersionNumber(tmp[0], tmp[1], tmp[2]));
}
return null;
}).flatMap(Optional::stream).collect(Collectors.toList());
For this stream functions, if all matched is false
, I mean if all the function inside the map()
method, it will rise NullPointException
.
How to make this stream not rise an exception and when all elements are null
make it to return an empty list or null
?
Upvotes: 1
Views: 1020
Reputation: 6573
As of java 16, there is also Stream.mapMulti
:
Returns a stream consisting of the results of replacing each element of this stream with multiple elements, specifically zero or more elements.
(emphasis mine)
Using this, your example becomes:
versions.stream().mapMulti((vs, consumer)->{
if(vs.matches("^matched string$")) {
...
consumer.accept(new VersionNumber(tmp[0], tmp[1], tmp[2]));
}
}).collect(Collectors.toList());
Note that the null values are not emitted downstream at all, so there is no need for filtering or intermediate streams for flat mapping.
Another example:
record A(String name, long count) {}
public static void main(String[] args) {
Stream.of(new A("apple", 1), new A("orange", 3), new A("banana", 0))
.mapMulti((bucket, consumer) -> {
for (int i = 0; i < bucket.count(); i++) {
consumer.accept(bucket.name());
}
})
.forEachOrdered(System.out::println);
}
Prints:
apple
orange
orange
orange
3 oranges, but no banana
Upvotes: 0
Reputation: 28988
if all matched is false, I mean if all the function in .map method, it will rise a
NullPointException
Since you're getting a NullPointerException
that means that your code compiles and running. Therefore, I assume that the "code" inside the map()
operation isn't your actual code because we can't treat elements in the of Stream<VersionNumber>
as optionals.
flatMap(Optional::stream)
will not cause a compilation error only if preceding map()
produces a result of type Optional<VersionNumber>
.
Hence, my suggestion is simple: never return null
in place of optional. It's completely wrong when an optional object might be null
, it should either contain a value or be empty, but it should never be null
.
And as @Andy Thomas has pointed out in the comment, there's no reason to utilize Optional
if the code inside the map()
really creates the object on the spot like new VersionNumber(tmp[0], tmp[1], tmp[2]));
. In this case, wrapping the result of the map()
with an Optional
will be a misuse of optional. Return either VersionNumber
instance or null
. And then apply a filter:
.map(vs -> {
if (vs.matches("^matched string$")) {
...
return new VersionNumber(tmp[0], tmp[1], tmp[2]);
}
return null;
})
.filter(Objects::nonNull)
.collect(Collectors.toList());
You have to resort to the usage of optional only if inside the map you're actually not creating VersionNumber
as you've shown, but, for instance, making an API call which return an Optional<VersionNumber>
if some condition is meat.
.map(vs -> {
if (vs.matches("^matched string$")) {
...
return getVersionNumber(tmp[0], tmp[1], tmp[2]); // method returing Optional<VersionNumber>
}
return Optional.empty();
})
.flatMap(Optional::stream)
.collect(Collectors.toList());
In such a case, there's no need to apply filtering. Optional.stream()
when invoked on an empty optional produces an empty stream.
Upvotes: 2
Reputation: 54148
You can filter afterward
versions.stream().map(vs->{
if(vs.matches("^matched string$")) {
...
return new VersionNumber(tmp[0], tmp[1], tmp[2]));
}
return null;
}).filter(Objects::nonNull).flatMap(Optional::stream).collect(Collectors.toList());
But that is wiser to filter the sooner
versions.stream().filter(vs -> vs.matches("^matched string$"))
.map(vs->{
...
return new VersionNumber(tmp[0], tmp[1], tmp[2]));
}).flatMap(Optional::stream).collect(Collectors.toList());
Upvotes: 3