We Keep Coding

Transforming a while loop to a stream in Java 8

As an exercise I'm converting some old code to functional streams. I don't know much about streams. It seems like it should be simple to convert this code, but I haven't had much luck. The method starts at a given integer, passes it to isPrime, which returns true if it's prime. and then hands off the new(next) prime number to be printed. If isPrime is false i is incremented and we check the next integer.
private static int nextPrime(final int number) {
int i = number + 1;
while (!isPrime(i)) {
i++;
}
return i;
}

I see no reason to use a Stream for this other than to take advantage of parallelism (if the primes happen to be very far apart, but this won't be true within int bounds, so there's essentially no benefit).
You can iterate over an IntStream of ascending integers (starting from number + 1) and filter only the prime numbers. When one is inevitably found, you can return the first.
private static int nextPrime(final int number) {
return IntStream.iterate(number + 1, i -> i + 1)
.filter(Test::isPrime)
.findFirst()
.getAsInt();
}
Note: The class that I used to test this is called Test, as seen by the method reference. You should change that to your class name.

Based on this answer, for an object like an Enumeration<T> for example, where you can only call .hasMoreElements() & .nextElement(), you can use this kind of code :
public static <T> Stream<T> enumerationAsStream(Enumeration<T> e) {
return StreamSupport.stream(
new Spliterators.AbstractSpliterator<T>(Long.MAX_VALUE, Spliterator.ORDERED) {
public boolean tryAdvance(Consumer<? super T> action) {
if(e.hasMoreElements()) {
action.accept(e.nextElement());
return true;
}
return false;
}
public void forEachRemaining(Consumer<? super T> action) {
while(e.hasMoreElements()) action.accept(e.nextElement());
}
}, false);
}

Is it possible to define an optional flow or exception-like behaviour in Java 8 streams API?

Let us have a stream of objects, resulting from a sequence of operations (e.g. mapping, filtering, flatmapping, etc.). Now I want to do a certain operation on them, but only if a given predicate is true. Otherwise I want to immediately return something else.
A simple example. I have a stream of different food objects. If all of them are edible I want to perform a cook operation on them and return the list of cooked food. But if any of them turns out to not be edible I want to immediately return an empty list.
Few solutions come to my mind, but I am not satisfied with any of them.
I could first perform an allMatch operation with isEdible predicate on the stream, but it will result in terminating it and I would need to repeat preliminary operations once more.
I could persist the collection that is the result of preliminary operations before checking the edibility, but therefore I need to perform them for all elements. Which is suboptimal, because it may turn out, that the first of them is not edible and allMatch would return much, much earlier.
Or I could design a hacky reduce routine, but it would also be unable to stop processing elements when predicate fails.
What I hope for is something like the code below. Is it possible with current API?
source.stream()
// some operations
.ifAny(food -> !food.isEdible(), new LinkedList<Food>())
// other operations if previous step not failed
.peek(food -> food.prepare())
.collect(Collectors.toList());

Not exactly what you wanted, but using the ternary operator will make your two step solution look cleaner, it also should have optimal performance:
return source.stream()
.allMatch(this::isEdible)
? source.stream()
.map(this::prepare()) // do stuff
.collect(Collectors.toList())
: Collections.emptyList();

Here's a method I just cooked up. It's a class that wraps Collector and checks if each element passes a predicate along the way. At the end, if any element failed, it returns the default return instead of the collected return.
On the downside you still have to iterate through every element even if the first one fails the predicate, but on the upside:
Only have to iterate once through each element
Simple in-line syntax
Easily integrates with existing stream and collector syntax and use cases.
So based on those I would imagine it would be worth it.
public class SatisfyableCollector<T,A,R> implements Collector<T,A,R> {
private Predicate<T> predicate; //The predicate used to test each element
private boolean elmHasFailed; //True once an element has failed the predicate
private Collector<T,A,R> collector; //The collector this wraps
private R defaultResult; //The result to return at the end if an element failed
public static <T,A,R> SatisfyableCollector<T,A,R> of(Collector<T,A,R> collector, Predicate<T> predicate, R defaultResult) {
return new SatisfyableCollector<>(collector, predicate, defaultResult);
}
private SatisfyableCollector(Collector<T,A,R> collector, Predicate<T> predicate, R defaultResult) {
this.predicate = predicate;
this.collector = collector;
this.defaultResult = defaultResult;
elmHasFailed = false;
}
#Override
public Supplier<A> supplier() {
return collector.supplier();
}
#Override
/** Before accumulating the new element t, check it against the predicate */
public BiConsumer<A, T> accumulator() {
return (a, t) -> {
if (! predicate.test(t)) {
elmHasFailed = true;
}
collector.accumulator().accept(a,t);
};
}
#Override
public BinaryOperator<A> combiner() {
return collector.combiner();
}
#Override
/** At the end, check if something failed. If so, return the default.
* Otherwise the wrapped collector's finisher.
*/
public Function<A, R> finisher() {
return (a) -> elmHasFailed ? defaultResult : collector.finisher().apply(a);
}
#Override
/** Make sure IDENTITY_FINISH isn't present, or finisher() won't be called */
public Set<Characteristics> characteristics() {
Set<Characteristics> originalSet = collector.characteristics();
if (! originalSet.contains(Characteristics.IDENTITY_FINISH)) return originalSet;
else {
HashSet<Characteristics> set = new HashSet<>(collector.characteristics()); //Make new set so we can modify it.
set.remove(Characteristics.IDENTITY_FINISH); //Make sure finisher() is called
return Collections.unmodifiableSet(set);
}
}
}
And can be used like this:
public static void main(String[] args) {
Stream<Integer> stream = Arrays.asList(1,2,3,4,5,6,7,8).stream();
//Create a list out of the stream *if* every element is even. Return empty otherwise
List<Integer> listOrEmpty = stream.collect(SatisfyableCollector.of(Collectors.toList(), (x) -> x%2 == 0, new ArrayList<>())));
System.out.println(listOrEmpty);
}
This currently doesn't handle exception throwing gracefully, but that shouldn't be too hard to add.

Does Java 8 have cached support for suppliers?

The guava library has it's own Supplier which does not extend Java 8 Supplier. Also guava provides a cache for suppliers - Suppliers#memoize.
Is there something similar, but for Java 8 Suppliers?

There's no built-in Java function for memoization, though it's not very hard to implement it, for example, like this:
public static <T> Supplier<T> memoize(Supplier<T> delegate) {
AtomicReference<T> value = new AtomicReference<>();
return () -> {
T val = value.get();
if (val == null) {
val = value.updateAndGet(cur -> cur == null ?
Objects.requireNonNull(delegate.get()) : cur);
}
return val;
};
}
Note that different implementation approaches exist. The above implementation may call the delegate several times if the memoized supplier requested simultaneously several times from the different threads. Sometimes such implementation is preferred over the explicit synchronization with lock. If lock is preferred, then DCL could be used:
public static <T> Supplier<T> memoizeLock(Supplier<T> delegate) {
AtomicReference<T> value = new AtomicReference<>();
return () -> {
T val = value.get();
if (val == null) {
synchronized(value) {
val = value.get();
if (val == null) {
val = Objects.requireNonNull(delegate.get());
value.set(val);
}
}
}
return val;
};
}
Also note, as #LouisWasserman correctly mentioned in comments, you can easily transform JDK supplier into Guava supplier and vice versa using method reference:
java.util.function.Supplier<String> jdkSupplier = () -> "test";
com.google.common.base.Supplier<String> guavaSupplier = jdkSupplier::get;
java.util.function.Supplier<String> jdkSupplierBack = guavaSupplier::get;
So it's not a big problem to switch between Guava and JDK functions.

The simplest solution would be
public static <T> Supplier<T> memoize(Supplier<T> original) {
ConcurrentHashMap<Object, T> store=new ConcurrentHashMap<>();
return ()->store.computeIfAbsent("dummy", key->original.get());
}
However, the simplest is not always the most efficient.
If you want a clean and efficient solution, resorting to an anonymous inner class to hold the mutable state will pay off:
public static <T> Supplier<T> memoize1(Supplier<T> original) {
return new Supplier<T>() {
Supplier<T> delegate = this::firstTime;
boolean initialized;
public T get() {
return delegate.get();
}
private synchronized T firstTime() {
if(!initialized) {
T value=original.get();
delegate=() -> value;
initialized=true;
}
return delegate.get();
}
};
}
This uses a delegate supplier which will do the first time operation and afterwards, replace itself with a supplier that unconditionally returns the captured result of the first evaluation. Since it has final fields semantics, it can be unconditionally returned without any additional synchronization.
Inside the synchronized method firstTime(), there is still an initialized flag needed because in case of concurrent access during initialization, multiple threads may wait at the method’s entry before the delegate has been replaced. Hence, these threads need to detect that the initialization has been done already. All subsequent accesses will read the new delegate supplier and get the value quickly.

A simple wrapper for Guava 20 on Java 8:
static <T> java.util.function.Supplier<T> memoize(java.util.function.Supplier<? extends T> supplier) {
return com.google.common.base.Suppliers.memoize(supplier::get)::get;
}

How to short-circuit reduce on Stream?

Suppose I have a stream of boolean values and the reduce operation that I am writing is || (OR). Can I write it in a way such that the evaluation of at least some of the elements is abandoned if a true value is encountered?
I am looking for some amount of optimization (perhaps if it is a parallel stream), not necessarily full optimization although the latter would be awesome.

I suspect you want this type of construct.
// stop when any element evaluates to true
boolean any = stream.anyMatch(t -> t);
You can check this with peek
Stream.of(1, 2, 3, 4).peek(System.out::println).anyMatch(i -> i == 2);
prints
1
2
For a parallel example
AtomicInteger count = new AtomicInteger();
IntStream.range(0, 1000).parallel().peek(t -> count.incrementAndGet()).anyMatch(i -> i == 2);
System.out.println("count: " + count);
prints a number like
count: 223
The exact number varies.
For a referencePipeline, the anyMatch calls
#Override
public final boolean anyMatch(Predicate<? super P_OUT> predicate) {
return evaluate(MatchOps.makeRef(predicate, MatchOps.MatchKind.ANY));
}
which calls this
public static <T> TerminalOp<T, Boolean> makeRef(Predicate<? super T> predicate,
MatchKind matchKind) {
Objects.requireNonNull(predicate);
Objects.requireNonNull(matchKind);
class MatchSink extends BooleanTerminalSink<T> {
MatchSink() {
super(matchKind);
}
#Override
public void accept(T t) {
if (!stop && predicate.test(t) == matchKind.stopOnPredicateMatches) {
stop = true;
value = matchKind.shortCircuitResult;
}
}
}
return new MatchOp<>(StreamShape.REFERENCE, matchKind, MatchSink::new);
}
where you can start to see the short circuiting code.

Functional style of Java 8's Optional.ifPresent and if-not-Present?

In Java 8, I want to do something to an Optional object if it is present, and do another thing if it is not present.
if (opt.isPresent()) {
System.out.println("found");
} else {
System.out.println("Not found");
}
This is not a 'functional style', though.
Optional has an ifPresent() method, but I am unable to chain an orElse() method.
Thus, I cannot write:
opt.ifPresent( x -> System.out.println("found " + x))
.orElse( System.out.println("NOT FOUND"));
In reply to #assylias, I don't think Optional.map() works for the following case:
opt.map( o -> {
System.out.println("while opt is present...");
o.setProperty(xxx);
dao.update(o);
return null;
}).orElseGet( () -> {
System.out.println("create new obj");
dao.save(new obj);
return null;
});
In this case, when opt is present, I update its property and save to the database. When it is not available, I create a new obj and save to the database.
Note in the two lambdas I have to return null.
But when opt is present, both lambdas will be executed. obj will be updated, and a new object will be saved to the database . This is because of the return null in the first lambda. And orElseGet() will continue to execute.

If you are using Java 9+, you can use ifPresentOrElse() method:
opt.ifPresentOrElse(
value -> System.out.println("Found: " + value),
() -> System.out.println("Not found")
);

For me the answer of #Dane White is OK, first I did not like using Runnable but I could not find any alternatives.
Here another implementation I preferred more:
public class OptionalConsumer<T> {
private Optional<T> optional;
private OptionalConsumer(Optional<T> optional) {
this.optional = optional;
}
public static <T> OptionalConsumer<T> of(Optional<T> optional) {
return new OptionalConsumer<>(optional);
}
public OptionalConsumer<T> ifPresent(Consumer<T> c) {
optional.ifPresent(c);
return this;
}
public OptionalConsumer<T> ifNotPresent(Runnable r) {
if (!optional.isPresent()) {
r.run();
}
return this;
}
}
Then:
Optional<Any> o = Optional.of(...);
OptionalConsumer.of(o).ifPresent(s -> System.out.println("isPresent " + s))
.ifNotPresent(() -> System.out.println("! isPresent"));
Update 1:
the above solution for the traditional way of development when you have the value and want to process it but what if I want to define the functionality and the execution will be then, check below enhancement;
public class OptionalConsumer<T> implements Consumer<Optional<T>> {
private final Consumer<T> c;
private final Runnable r;
public OptionalConsumer(Consumer<T> c, Runnable r) {
super();
this.c = c;
this.r = r;
}
public static <T> OptionalConsumer<T> of(Consumer<T> c, Runnable r) {
return new OptionalConsumer(c, r);
}
#Override
public void accept(Optional<T> t) {
if (t.isPresent()) {
c.accept(t.get());
}
else {
r.run();
}
}
Then could be used as:
Consumer<Optional<Integer>> c = OptionalConsumer.of(
System.out::println,
() -> System.out.println("Not fit")
);
IntStream.range(0, 100)
.boxed()
.map(i -> Optional.of(i)
.filter(j -> j % 2 == 0))
.forEach(c);
In this new code you have 3 things:
can define the functionality before the existing of an object easy.
not creating object reference for each Optional, only one, you have so less memory than less GC.
it is implementing consumer for better usage with other components.
By the way, now its name is more descriptive it is actually Consumer<Optional<?>>

Java 9 introduces
ifPresentOrElse if a value is present, performs the given action with the value, otherwise performs the given empty-based action.
See excellent Optional in Java 8 cheat sheet.
It provides all answers for most use cases.
Short summary below
ifPresent() - do something when Optional is set
opt.ifPresent(x -> print(x));
opt.ifPresent(this::print);
filter() - reject (filter out) certain Optional values.
opt.filter(x -> x.contains("ab")).ifPresent(this::print);
map() - transform value if present
opt.map(String::trim).filter(t -> t.length() > 1).ifPresent(this::print);
orElse()/orElseGet() - turning empty Optional to default T
int len = opt.map(String::length).orElse(-1);
int len = opt.
map(String::length).
orElseGet(() -> slowDefault()); //orElseGet(this::slowDefault)
orElseThrow() - lazily throw exceptions on empty Optional
opt.
filter(s -> !s.isEmpty()).
map(s -> s.charAt(0)).
orElseThrow(IllegalArgumentException::new);

An alternative is:
System.out.println(opt.map(o -> "Found")
.orElse("Not found"));
I don't think it improves readability though.
Or as Marko suggested, use a ternary operator:
System.out.println(opt.isPresent() ? "Found" : "Not found");

Another solution would be to use higher-order functions as follows
opt.<Runnable>map(value -> () -> System.out.println("Found " + value))
.orElse(() -> System.out.println("Not Found"))
.run();

There isn't a great way to do it out of the box. If you want to be using your cleaner syntax on a regular basis, then you can create a utility class to help out:
public class OptionalEx {
private boolean isPresent;
private OptionalEx(boolean isPresent) {
this.isPresent = isPresent;
}
public void orElse(Runnable runner) {
if (!isPresent) {
runner.run();
}
}
public static <T> OptionalEx ifPresent(Optional<T> opt, Consumer<? super T> consumer) {
if (opt.isPresent()) {
consumer.accept(opt.get());
return new OptionalEx(true);
}
return new OptionalEx(false);
}
}
Then you can use a static import elsewhere to get syntax that is close to what you're after:
import static com.example.OptionalEx.ifPresent;
ifPresent(opt, x -> System.out.println("found " + x))
.orElse(() -> System.out.println("NOT FOUND"));

If you can use only Java 8 or lower:
1) if you don't have spring-data the best way so far is:
opt.<Runnable>map(param -> () -> System.out.println(param))
.orElse(() -> System.out.println("no-param-specified"))
.run();
Now I know it's not so readable and even hard to understand for someone, but looks fine for me personally and I don't see another nice fluent way for this case.
2) if you're lucky enough and you can use spring-data the best way is
Optionals#ifPresentOrElse:
Optionals.ifPresentOrElse(opt, System.out::println,
() -> System.out.println("no-param-specified"));
If you can use Java 9, you should definitely go with:
opt.ifPresentOrElse(System.out::println,
() -> System.out.println("no-param-specified"));

You cannot call orElse after ifPresent, the reason is, orElse is called on an optiional but ifPresent returns void. So the best approach to achieve is ifPresentOrElse.
It could be like this:
op.ifPresentOrElse(
(value)
-> { System.out.println(
"Value is present, its: "
+ value); },
()
-> { System.out.println(
"Value is empty"); });

The described behavior can be achieved by using Vavr (formerly known as Javaslang), an object-functional library for Java 8+, that implements most of Scala constructs (being Scala a more expressive language with a way richer type system built on JVM). It is a very good library to add to your Java projects to write pure functional code.
Vavr provides the Option monad that provides functions to work with the Option type such as:
fold: to map the value of the option on both cases (defined/empty)
onEmpty: allows to execute a Runnable when option is empty
peek: allows to consume the value of the option (when defined).
and it is also Serializable on the contrary of Optional which means you can safely use it as method argument and instance member.
Option follows the monad laws at difference to the Java's Optional "pseudo-monad" and provides a richer API. And of course you can make it from a Java's Optional (and the other way around): Option.ofOptional(javaOptional) –Vavr is focused on interoperability.
Going to the example:
// AWESOME Vavr functional collections (immutable for the gread good :)
// fully convertible to Java's counterparts.
final Map<String, String> map = Map("key1", "value1", "key2", "value2");
final Option<String> opt = map.get("nonExistentKey"); // you're safe of null refs!
final String result = opt.fold(
() -> "Not found!!!", // Option is None
val -> "Found the value: " + val // Option is Some(val)
);
Moreover, all Vavr types are convertible to its Java counterparts, for the sake of the example: Optional javaOptional = opt.toJava(), very easy :) Of course the conversion also exists in the other way: Option option = Option.ofOptional(javaOptional).
N.B. Vavr offers a io.vavr.API class with a lot of convenient static methods =)
Further reading
Null reference, the billion dollar mistake
N.B. This is only a very little example of what Vavr offers (pattern matching, streams a.k.a. lazy evaluated lists, monadic types, immutable collections,...).

The problem here:
optional
.map(object -> {
System.out.println("If present.");
return null;
})
.orElseGet( () -> {
System.out.println("If empty.");
return null;
});
Is that map() converts the null returned by the first function to empty(); it then returns empty(). As it returns empty(), it prompts the invocation of the second function. Note that orElseGet() does not convert the null returned by the second function to empty(), so it will return null.
See the implementation of map():
public<U> Optional<U> map(Function<? super T, ? extends U> mapper) {
Objects.requireNonNull(mapper);
if (!isPresent())
return empty();
else {
return Optional.ofNullable(mapper.apply(value));
}
}
And the implementation of orElseGet():
public T orElseGet(Supplier<? extends T> other) {
return value != null ? value : other.get();
}
Thus when executed:
if optional.isPresent(), the system will print If present., then If empty., and the expression will evaluate to null.
if !optional.isPresent(), the system will print If empty., and the expression will evaluate to null.
If the function provided to map() returned any other value - any other value - the code would work as you expect, with the function provided to map() being executed if isPresent() and the function provided to orElseGet() if !isPresent():
For example, this:
optional
.map(data -> {
System.out.println("If present.");
return 0;
})
.orElseGet( () -> {
System.out.println("If empty.");
return 0;
});
When executed:
if optional.isPresent(), the system will print If present., and the expression will evaluate to 0.
if !optional.isPresent(), the system will print If empty., and the expression will evaluate to 0.
If your specific case, I suggest that your insert and update methods return, say, the persisted object, or the id of the persisted object, or something similarly useful; then you can use code similar to this:
final Object persist = optional
.map(object -> {
System.out.println("If present.");
return update(object);
})
.orElseGet( () -> {
System.out.println("If empty.");
return insert(new Object());
});

Another solution could be following:
This is how you use it:
final Opt<String> opt = Opt.of("I'm a cool text");
opt.ifPresent()
.apply(s -> System.out.printf("Text is: %s\n", s))
.elseApply(() -> System.out.println("no text available"));
Or in case you in case of the opposite use case is true:
final Opt<String> opt = Opt.of("This is the text");
opt.ifNotPresent()
.apply(() -> System.out.println("Not present"))
.elseApply(t -> /*do something here*/);
This are the ingredients:
Little modified Function interface, just for the "elseApply" method
Optional enhancement
A little bit of curring :-)
The "cosmetically" enhanced Function interface.
#FunctionalInterface
public interface Fkt<T, R> extends Function<T, R> {
default R elseApply(final T t) {
return this.apply(t);
}
}
And the Optional wrapper class for enhancement:
public class Opt<T> {
private final Optional<T> optional;
private Opt(final Optional<T> theOptional) {
this.optional = theOptional;
}
public static <T> Opt<T> of(final T value) {
return new Opt<>(Optional.of(value));
}
public static <T> Opt<T> of(final Optional<T> optional) {
return new Opt<>(optional);
}
public static <T> Opt<T> ofNullable(final T value) {
return new Opt<>(Optional.ofNullable(value));
}
public static <T> Opt<T> empty() {
return new Opt<>(Optional.empty());
}
private final BiFunction<Consumer<T>, Runnable, Void> ifPresent = (present, notPresent) -> {
if (this.optional.isPresent()) {
present.accept(this.optional.get());
} else {
notPresent.run();
}
return null;
};
private final BiFunction<Runnable, Consumer<T>, Void> ifNotPresent = (notPresent, present) -> {
if (!this.optional.isPresent()) {
notPresent.run();
} else {
present.accept(this.optional.get());
}
return null;
};
public Fkt<Consumer<T>, Fkt<Runnable, Void>> ifPresent() {
return Opt.curry(this.ifPresent);
}
public Fkt<Runnable, Fkt<Consumer<T>, Void>> ifNotPresent() {
return Opt.curry(this.ifNotPresent);
}
private static <X, Y, Z> Fkt<X, Fkt<Y, Z>> curry(final BiFunction<X, Y, Z> function) {
return (final X x) -> (final Y y) -> function.apply(x, y);
}
}
This should do the trick and could serve as a basic template how to deal with such requirements.
The basic idea here is following. In a non functional style programming world you would probably implement a method taking two parameter where the first is a kind of runnable code which should be executed in case the value is available and the other parameter is the runnable code which should be run in case the value is not available. For the sake of better readability, you can use curring to split the function of two parameter in two functions of one parameter each. This is what I basically did here.
Hint: Opt also provides the other use case where you want to execute a piece of code just in case the value is not available. This could be done also via Optional.filter.stuff but I found this much more readable.
Hope that helps!
Additional Info:
There is another way to have say "if then else" using currying:
public static <X, Y> Function<Predicate<X>, Function<Function<X, Y>, Function<Function<X, Y>, Y>>> ifThenElse(X input) {
return (final Predicate<X> pred) -> (final Function<X, Y> ifPresent) -> (final Function<X, Y> ifNotPresent) -> pred.test(input) ? ifPresent.apply(input) : ifNotPresent.apply(input);
}
This way it is possible to say:
final String result = ifThenElse("fancy")
.apply(input -> input.contains("fancy")) /* test */
.apply(input -> input.toUpperCase()) /* if-case */
.apply(input -> input.toLowerCase()); /* else-case */

In case you want store the value:
Pair.of<List<>, List<>> output = opt.map(details -> Pair.of(details.a, details.b))).orElseGet(() -> Pair.of(Collections.emptyList(), Collections.emptyList()));

Supposing that you have a list and avoiding the isPresent() issue (related with optionals) you could use .iterator().hasNext() to check if not present.