1 /*
2 * Copyright (c) 2012, 2013, Oracle and/or its affiliates. All rights reserved.
3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4 *
5 * This code is free software; you can redistribute it and/or modify it
6 * under the terms of the GNU General Public License version 2 only, as
7 * published by the Free Software Foundation. Oracle designates this
8 * particular file as subject to the "Classpath" exception as provided
9 * by Oracle in the LICENSE file that accompanied this code.
10 *
11 * This code is distributed in the hope that it will be useful, but WITHOUT
12 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
13 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
14 * version 2 for more details (a copy is included in the LICENSE file that
15 * accompanied this code).
16 *
17 * You should have received a copy of the GNU General Public License version
18 * 2 along with this work; if not, write to the Free Software Foundation,
19 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
20 *
21 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
22 * or visit www.oracle.com if you need additional information or have any
23 * questions.
24 */
25 package java.util.stream;
26
27 import java.util.AbstractMap;
28 import java.util.AbstractSet;
29 import java.util.ArrayList;
30 import java.util.Arrays;
31 import java.util.Collection;
32 import java.util.Collections;
33 import java.util.Comparator;
34 import java.util.DoubleSummaryStatistics;
35 import java.util.EnumSet;
36 import java.util.HashMap;
37 import java.util.HashSet;
38 import java.util.IntSummaryStatistics;
39 import java.util.Iterator;
40 import java.util.List;
41 import java.util.LongSummaryStatistics;
42 import java.util.Map;
43 import java.util.Objects;
44 import java.util.Optional;
45 import java.util.Set;
46 import java.util.StringJoiner;
47 import java.util.concurrent.ConcurrentHashMap;
48 import java.util.concurrent.ConcurrentMap;
49 import java.util.function.BiConsumer;
50 import java.util.function.BiFunction;
51 import java.util.function.BinaryOperator;
52 import java.util.function.Consumer;
53 import java.util.function.Function;
54 import java.util.function.Predicate;
55 import java.util.function.Supplier;
56 import java.util.function.ToDoubleFunction;
57 import java.util.function.ToIntFunction;
58 import java.util.function.ToLongFunction;
59
60 /**
61 * Implementations of {@link Collector} that implement various useful reduction
62 * operations, such as accumulating elements into collections, summarizing
63 * elements according to various criteria, etc.
64 *
65 * <p>The following are examples of using the predefined collectors to perform
66 * common mutable reduction tasks:
67 *
68 * <pre>{@code
69 * // Accumulate names into a List
70 * List<String> list = people.stream().map(Person::getName).collect(Collectors.toList());
71 *
72 * // Accumulate names into a TreeSet
73 * Set<String> set = people.stream().map(Person::getName).collect(Collectors.toCollection(TreeSet::new));
74 *
75 * // Convert elements to strings and concatenate them, separated by commas
76 * String joined = things.stream()
77 * .map(Object::toString)
78 * .collect(Collectors.joining(", "));
79 *
80 * // Compute sum of salaries of employee
81 * int total = employees.stream()
82 * .collect(Collectors.summingInt(Employee::getSalary)));
83 *
84 * // Group employees by department
85 * Map<Department, List<Employee>> byDept
86 * = employees.stream()
87 * .collect(Collectors.groupingBy(Employee::getDepartment));
88 *
89 * // Compute sum of salaries by department
90 * Map<Department, Integer> totalByDept
91 * = employees.stream()
92 * .collect(Collectors.groupingBy(Employee::getDepartment,
93 * Collectors.summingInt(Employee::getSalary)));
94 *
95 * // Partition students into passing and failing
96 * Map<Boolean, List<Student>> passingFailing =
97 * students.stream()
98 * .collect(Collectors.partitioningBy(s -> s.getGrade() >= PASS_THRESHOLD));
99 *
100 * }</pre>
101 *
102 * @since 1.8
103 */
104 public final class Collectors {
105
106 static final Set<Collector.Characteristics> CH_CONCURRENT_ID
107 = Collections.unmodifiableSet(EnumSet.of(Collector.Characteristics.CONCURRENT,
108 Collector.Characteristics.UNORDERED,
109 Collector.Characteristics.IDENTITY_FINISH));
110 static final Set<Collector.Characteristics> CH_CONCURRENT_NOID
111 = Collections.unmodifiableSet(EnumSet.of(Collector.Characteristics.CONCURRENT,
112 Collector.Characteristics.UNORDERED));
113 static final Set<Collector.Characteristics> CH_ID
114 = Collections.unmodifiableSet(EnumSet.of(Collector.Characteristics.IDENTITY_FINISH));
115 static final Set<Collector.Characteristics> CH_UNORDERED_ID
116 = Collections.unmodifiableSet(EnumSet.of(Collector.Characteristics.UNORDERED,
117 Collector.Characteristics.IDENTITY_FINISH));
118 static final Set<Collector.Characteristics> CH_NOID = Collections.emptySet();
119
120 private Collectors() { }
121
122 /**
123 * Returns a merge function, suitable for use in
124 * {@link Map#merge(Object, Object, BiFunction) Map.merge()} or
125 * {@link #toMap(Function, Function, BinaryOperator) toMap()}, which always
126 * throws {@code IllegalStateException}. This can be used to enforce the
127 * assumption that the elements being collected are distinct.
128 *
129 * @param <T> the type of input arguments to the merge function
130 * @return a merge function which always throw {@code IllegalStateException}
131 */
132 private static <T> BinaryOperator<T> throwingMerger() {
133 return (u,v) -> { throw new IllegalStateException(String.format("Duplicate key %s", u)); };
134 }
135
136 @SuppressWarnings("unchecked")
137 private static <I, R> Function<I, R> castingIdentity() {
138 return i -> (R) i;
139 }
140
141 /**
142 * Simple implementation class for {@code Collector}.
143 *
144 * @param <T> the type of elements to be collected
145 * @param <R> the type of the result
146 */
147 static class CollectorImpl<T, A, R> implements Collector<T, A, R> {
148 private final Supplier<A> supplier;
149 private final BiConsumer<A, T> accumulator;
150 private final BinaryOperator<A> combiner;
151 private final Function<A, R> finisher;
152 private final Set<Characteristics> characteristics;
153
154 CollectorImpl(Supplier<A> supplier,
155 BiConsumer<A, T> accumulator,
156 BinaryOperator<A> combiner,
157 Function<A,R> finisher,
158 Set<Characteristics> characteristics) {
159 this.supplier = supplier;
160 this.accumulator = accumulator;
161 this.combiner = combiner;
162 this.finisher = finisher;
163 this.characteristics = characteristics;
164 }
165
166 CollectorImpl(Supplier<A> supplier,
167 BiConsumer<A, T> accumulator,
168 BinaryOperator<A> combiner,
169 Set<Characteristics> characteristics) {
170 this(supplier, accumulator, combiner, castingIdentity(), characteristics);
171 }
172
173 @Override
174 public BiConsumer<A, T> accumulator() {
175 return accumulator;
176 }
177
178 @Override
179 public Supplier<A> supplier() {
180 return supplier;
181 }
182
183 @Override
184 public BinaryOperator<A> combiner() {
185 return combiner;
186 }
187
188 @Override
189 public Function<A, R> finisher() {
190 return finisher;
191 }
192
193 @Override
194 public Set<Characteristics> characteristics() {
195 return characteristics;
196 }
197 }
198
199 /**
200 * Returns a {@code Collector} that accumulates the input elements into a
201 * new {@code Collection}, in encounter order. The {@code Collection} is
202 * created by the provided factory.
203 *
204 * @param <T> the type of the input elements
205 * @param <C> the type of the resulting {@code Collection}
206 * @param collectionFactory a {@code Supplier} which returns a new, empty
207 * {@code Collection} of the appropriate type
208 * @return a {@code Collector} which collects all the input elements into a
209 * {@code Collection}, in encounter order
210 */
211 public static <T, C extends Collection<T>>
212 Collector<T, ?, C> toCollection(Supplier<C> collectionFactory) {
213 return new CollectorImpl<>(collectionFactory, Collection<T>::add,
214 (r1, r2) -> { r1.addAll(r2); return r1; },
215 CH_ID);
216 }
217
218 /**
219 * Returns a {@code Collector} that accumulates the input elements into a
220 * new {@code List}. There are no guarantees on the type, mutability,
221 * serializability, or thread-safety of the {@code List} returned; if more
222 * control over the returned {@code List} is required, use {@link #toCollection(Supplier)}.
223 *
224 * @param <T> the type of the input elements
225 * @return a {@code Collector} which collects all the input elements into a
226 * {@code List}, in encounter order
227 */
228 public static <T>
229 Collector<T, ?, List<T>> toList() {
230 return new CollectorImpl<>((Supplier<List<T>>) ArrayList::new, List::add,
231 (left, right) -> { left.addAll(right); return left; },
232 CH_ID);
233 }
234
235 /**
236 * Returns a {@code Collector} that accumulates the input elements into a
237 * new {@code Set}. There are no guarantees on the type, mutability,
238 * serializability, or thread-safety of the {@code Set} returned; if more
239 * control over the returned {@code Set} is required, use
240 * {@link #toCollection(Supplier)}.
241 *
242 * <p>This is an {@link Collector.Characteristics#UNORDERED unordered}
243 * Collector.
244 *
245 * @param <T> the type of the input elements
246 * @return a {@code Collector} which collects all the input elements into a
247 * {@code Set}
248 */
249 public static <T>
250 Collector<T, ?, Set<T>> toSet() {
251 return new CollectorImpl<>((Supplier<Set<T>>) HashSet::new, Set::add,
252 (left, right) -> { left.addAll(right); return left; },
253 CH_UNORDERED_ID);
254 }
255
256 /**
257 * Returns a {@code Collector} that concatenates the input elements into a
258 * {@code String}, in encounter order.
259 *
260 * @return a {@code Collector} that concatenates the input elements into a
261 * {@code String}, in encounter order
262 */
263 public static Collector<CharSequence, ?, String> joining() {
264 return new CollectorImpl<CharSequence, StringBuilder, String>(
265 StringBuilder::new, StringBuilder::append,
266 (r1, r2) -> { r1.append(r2); return r1; },
267 StringBuilder::toString, CH_NOID);
268 }
269
270 /**
271 * Returns a {@code Collector} that concatenates the input elements,
272 * separated by the specified delimiter, in encounter order.
273 *
274 * @param delimiter the delimiter to be used between each element
275 * @return A {@code Collector} which concatenates CharSequence elements,
276 * separated by the specified delimiter, in encounter order
277 */
278 public static Collector<CharSequence, ?, String> joining(CharSequence delimiter) {
279 return joining(delimiter, "", "");
280 }
281
282 /**
283 * Returns a {@code Collector} that concatenates the input elements,
284 * separated by the specified delimiter, with the specified prefix and
285 * suffix, in encounter order.
286 *
287 * @param delimiter the delimiter to be used between each element
288 * @param prefix the sequence of characters to be used at the beginning
289 * of the joined result
290 * @param suffix the sequence of characters to be used at the end
291 * of the joined result
292 * @return A {@code Collector} which concatenates CharSequence elements,
293 * separated by the specified delimiter, in encounter order
294 */
295 public static Collector<CharSequence, ?, String> joining(CharSequence delimiter,
296 CharSequence prefix,
297 CharSequence suffix) {
298 return new CollectorImpl<>(
299 () -> new StringJoiner(delimiter, prefix, suffix),
300 StringJoiner::add, StringJoiner::merge,
301 StringJoiner::toString, CH_NOID);
302 }
303
304 /**
305 * {@code BinaryOperator<Map>} that merges the contents of its right
306 * argument into its left argument, using the provided merge function to
307 * handle duplicate keys.
308 *
309 * @param <K> type of the map keys
310 * @param <V> type of the map values
311 * @param <M> type of the map
312 * @param mergeFunction A merge function suitable for
313 * {@link Map#merge(Object, Object, BiFunction) Map.merge()}
314 * @return a merge function for two maps
315 */
316 private static <K, V, M extends Map<K,V>>
317 BinaryOperator<M> mapMerger(BinaryOperator<V> mergeFunction) {
318 return (m1, m2) -> {
319 for (Map.Entry<K,V> e : m2.entrySet())
320 m1.merge(e.getKey(), e.getValue(), mergeFunction);
321 return m1;
322 };
323 }
324
325 /**
326 * Adapts a {@code Collector} accepting elements of type {@code U} to one
327 * accepting elements of type {@code T} by applying a mapping function to
328 * each input element before accumulation.
329 *
330 * @apiNote
331 * The {@code mapping()} collectors are most useful when used in a
332 * multi-level reduction, such as downstream of a {@code groupingBy} or
333 * {@code partitioningBy}. For example, given a stream of
334 * {@code Person}, to accumulate the set of last names in each city:
335 * <pre>{@code
336 * Map<City, Set<String>> lastNamesByCity
337 * = people.stream().collect(groupingBy(Person::getCity,
338 * mapping(Person::getLastName, toSet())));
339 * }</pre>
340 *
341 * @param <T> the type of the input elements
342 * @param <U> type of elements accepted by downstream collector
343 * @param <A> intermediate accumulation type of the downstream collector
344 * @param <R> result type of collector
345 * @param mapper a function to be applied to the input elements
346 * @param downstream a collector which will accept mapped values
347 * @return a collector which applies the mapping function to the input
348 * elements and provides the mapped results to the downstream collector
349 */
350 public static <T, U, A, R>
351 Collector<T, ?, R> mapping(Function<? super T, ? extends U> mapper,
352 Collector<? super U, A, R> downstream) {
353 BiConsumer<A, ? super U> downstreamAccumulator = downstream.accumulator();
354 return new CollectorImpl<>(downstream.supplier(),
355 (r, t) -> downstreamAccumulator.accept(r, mapper.apply(t)),
356 downstream.combiner(), downstream.finisher(),
357 downstream.characteristics());
358 }
359
360 /**
361 * Adapts a {@code Collector} to perform an additional finishing
362 * transformation. For example, one could adapt the {@link #toList()}
363 * collector to always produce an immutable list with:
364 * <pre>{@code
365 * List<String> people
366 * = people.stream().collect(collectingAndThen(toList(), Collections::unmodifiableList));
367 * }</pre>
368 *
369 * @param <T> the type of the input elements
370 * @param <A> intermediate accumulation type of the downstream collector
371 * @param <R> result type of the downstream collector
372 * @param <RR> result type of the resulting collector
373 * @param downstream a collector
374 * @param finisher a function to be applied to the final result of the downstream collector
375 * @return a collector which performs the action of the downstream collector,
376 * followed by an additional finishing step
377 */
378 public static<T,A,R,RR> Collector<T,A,RR> collectingAndThen(Collector<T,A,R> downstream,
379 Function<R,RR> finisher) {
380 Set<Collector.Characteristics> characteristics = downstream.characteristics();
381 if (characteristics.contains(Collector.Characteristics.IDENTITY_FINISH)) {
382 if (characteristics.size() == 1)
383 characteristics = Collectors.CH_NOID;
384 else {
385 characteristics = EnumSet.copyOf(characteristics);
386 characteristics.remove(Collector.Characteristics.IDENTITY_FINISH);
387 characteristics = Collections.unmodifiableSet(characteristics);
388 }
389 }
390 return new CollectorImpl<>(downstream.supplier(),
391 downstream.accumulator(),
392 downstream.combiner(),
393 downstream.finisher().andThen(finisher),
394 characteristics);
395 }
396
397 /**
398 * Returns a {@code Collector} accepting elements of type {@code T} that
399 * counts the number of input elements. If no elements are present, the
400 * result is 0.
401 *
402 * @implSpec
403 * This produces a result equivalent to:
404 * <pre>{@code
405 * reducing(0L, e -> 1L, Long::sum)
406 * }</pre>
407 *
408 * @param <T> the type of the input elements
409 * @return a {@code Collector} that counts the input elements
410 */
411 public static <T> Collector<T, ?, Long>
412 counting() {
413 return reducing(0L, e -> 1L, Long::sum);
414 }
415
416 /**
417 * Returns a {@code Collector} that produces the minimal element according
418 * to a given {@code Comparator}, described as an {@code Optional<T>}.
419 *
420 * @implSpec
421 * This produces a result equivalent to:
422 * <pre>{@code
423 * reducing(BinaryOperator.minBy(comparator))
424 * }</pre>
425 *
426 * @param <T> the type of the input elements
427 * @param comparator a {@code Comparator} for comparing elements
428 * @return a {@code Collector} that produces the minimal value
429 */
430 public static <T> Collector<T, ?, Optional<T>>
431 minBy(Comparator<? super T> comparator) {
432 return reducing(BinaryOperator.minBy(comparator));
433 }
434
435 /**
436 * Returns a {@code Collector} that produces the maximal element according
437 * to a given {@code Comparator}, described as an {@code Optional<T>}.
438 *
439 * @implSpec
440 * This produces a result equivalent to:
441 * <pre>{@code
442 * reducing(BinaryOperator.maxBy(comparator))
443 * }</pre>
444 *
445 * @param <T> the type of the input elements
446 * @param comparator a {@code Comparator} for comparing elements
447 * @return a {@code Collector} that produces the maximal value
448 */
449 public static <T> Collector<T, ?, Optional<T>>
450 maxBy(Comparator<? super T> comparator) {
451 return reducing(BinaryOperator.maxBy(comparator));
452 }
453
454 /**
455 * Returns a {@code Collector} that produces the sum of a integer-valued
456 * function applied to the input elements. If no elements are present,
457 * the result is 0.
458 *
459 * @param <T> the type of the input elements
460 * @param mapper a function extracting the property to be summed
461 * @return a {@code Collector} that produces the sum of a derived property
462 */
463 public static <T> Collector<T, ?, Integer>
464 summingInt(ToIntFunction<? super T> mapper) {
465 return new CollectorImpl<>(
466 () -> new int[1],
467 (a, t) -> { a[0] += mapper.applyAsInt(t); },
468 (a, b) -> { a[0] += b[0]; return a; },
469 a -> a[0], CH_NOID);
470 }
471
472 /**
473 * Returns a {@code Collector} that produces the sum of a long-valued
474 * function applied to the input elements. If no elements are present,
475 * the result is 0.
476 *
477 * @param <T> the type of the input elements
478 * @param mapper a function extracting the property to be summed
479 * @return a {@code Collector} that produces the sum of a derived property
480 */
481 public static <T> Collector<T, ?, Long>
482 summingLong(ToLongFunction<? super T> mapper) {
483 return new CollectorImpl<>(
484 () -> new long[1],
485 (a, t) -> { a[0] += mapper.applyAsLong(t); },
486 (a, b) -> { a[0] += b[0]; return a; },
487 a -> a[0], CH_NOID);
488 }
489
490 /**
491 * Returns a {@code Collector} that produces the sum of a double-valued
492 * function applied to the input elements. If no elements are present,
493 * the result is 0.
494 *
495 * <p>The sum returned can vary depending upon the order in which
496 * values are recorded, due to accumulated rounding error in
497 * addition of values of differing magnitudes. Values sorted by increasing
498 * absolute magnitude tend to yield more accurate results. If any recorded
499 * value is a {@code NaN} or the sum is at any point a {@code NaN} then the
500 * sum will be {@code NaN}.
501 *
502 * @param <T> the type of the input elements
503 * @param mapper a function extracting the property to be summed
504 * @return a {@code Collector} that produces the sum of a derived property
505 */
506 public static <T> Collector<T, ?, Double>
507 summingDouble(ToDoubleFunction<? super T> mapper) {
508 return new CollectorImpl<>(
509 () -> new double[2],
510 (a, t) -> { sumWithCompensation(a, mapper.applyAsDouble(t)); },
511 (a, b) -> { sumWithCompensation(a, b[0]); return sumWithCompensation(a, b[1]); },
512 a -> a[0],
513 CH_NOID);
514 }
515
516 /**
517 * Incorporate a new double value using Kahan summation /
518 * compensation summation.
519 *
520 * High-order bits of the sum are in intermediateSum[0], low-order
521 * bits of the sum are in intermediateSum[1], any additional
522 * elements are application-specific.
523 *
524 * @param intermediateSum the high-order and low-order words of the intermediate sum
525 * @param value the name value to be included in the running sum
526 */
527 static double[] sumWithCompensation(double[] intermediateSum, double value) {
528 double tmp = value - intermediateSum[1];
529 double sum = intermediateSum[0];
530 double velvel = sum + tmp; // Little wolf of rounding error
531 intermediateSum[1] = (velvel - sum) - tmp;
532 intermediateSum[0] = velvel;
533 return intermediateSum;
534 }
535
536
537 /**
538 * Returns a {@code Collector} that produces the arithmetic mean of an integer-valued
539 * function applied to the input elements. If no elements are present,
540 * the result is 0.
541 *
542 * @param <T> the type of the input elements
543 * @param mapper a function extracting the property to be summed
544 * @return a {@code Collector} that produces the sum of a derived property
545 */
546 public static <T> Collector<T, ?, Double>
547 averagingInt(ToIntFunction<? super T> mapper) {
548 return new CollectorImpl<>(
549 () -> new long[2],
550 (a, t) -> { a[0] += mapper.applyAsInt(t); a[1]++; },
551 (a, b) -> { a[0] += b[0]; a[1] += b[1]; return a; },
552 a -> (a[1] == 0) ? 0.0d : (double) a[0] / a[1], CH_NOID);
553 }
554
555 /**
556 * Returns a {@code Collector} that produces the arithmetic mean of a long-valued
557 * function applied to the input elements. If no elements are present,
558 * the result is 0.
559 *
560 * @param <T> the type of the input elements
561 * @param mapper a function extracting the property to be summed
562 * @return a {@code Collector} that produces the sum of a derived property
563 */
564 public static <T> Collector<T, ?, Double>
565 averagingLong(ToLongFunction<? super T> mapper) {
566 return new CollectorImpl<>(
567 () -> new long[2],
568 (a, t) -> { a[0] += mapper.applyAsLong(t); a[1]++; },
569 (a, b) -> { a[0] += b[0]; a[1] += b[1]; return a; },
570 a -> (a[1] == 0) ? 0.0d : (double) a[0] / a[1], CH_NOID);
571 }
572
573 /**
574 * Returns a {@code Collector} that produces the arithmetic mean of a double-valued
575 * function applied to the input elements. If no elements are present,
576 * the result is 0.
577 *
578 * <p>The average returned can vary depending upon the order in which
579 * values are recorded, due to accumulated rounding error in
580 * addition of values of differing magnitudes. Values sorted by increasing
581 * absolute magnitude tend to yield more accurate results. If any recorded
582 * value is a {@code NaN} or the sum is at any point a {@code NaN} then the
583 * average will be {@code NaN}.
584 *
585 * @param <T> the type of the input elements
586 * @param mapper a function extracting the property to be summed
587 * @return a {@code Collector} that produces the sum of a derived property
588 */
589 public static <T> Collector<T, ?, Double>
590 averagingDouble(ToDoubleFunction<? super T> mapper) {
591 return new CollectorImpl<>(
592 () -> new double[3],
593 (a, t) -> { sumWithCompensation(a, mapper.applyAsDouble(t)); a[2]++; },
594 (a, b) -> { sumWithCompensation(a, b[0]); sumWithCompensation(a, b[1]); a[2] += b[2]; return a; },
595 a -> (a[2] == 0) ? 0.0d : (a[0] / a[2]),
596 CH_NOID);
597 }
598
599 /**
600 * Returns a {@code Collector} which performs a reduction of its
601 * input elements under a specified {@code BinaryOperator} using the
602 * provided identity.
603 *
604 * @apiNote
605 * The {@code reducing()} collectors are most useful when used in a
606 * multi-level reduction, downstream of {@code groupingBy} or
607 * {@code partitioningBy}. To perform a simple reduction on a stream,
608 * use {@link Stream#reduce(Object, BinaryOperator)}} instead.
609 *
610 * @param <T> element type for the input and output of the reduction
611 * @param identity the identity value for the reduction (also, the value
612 * that is returned when there are no input elements)
613 * @param op a {@code BinaryOperator<T>} used to reduce the input elements
614 * @return a {@code Collector} which implements the reduction operation
615 *
616 * @see #reducing(BinaryOperator)
617 * @see #reducing(Object, Function, BinaryOperator)
618 */
619 public static <T> Collector<T, ?, T>
620 reducing(T identity, BinaryOperator<T> op) {
621 return new CollectorImpl<>(
622 boxSupplier(identity),
623 (a, t) -> { a[0] = op.apply(a[0], t); },
624 (a, b) -> { a[0] = op.apply(a[0], b[0]); return a; },
625 a -> a[0],
626 CH_NOID);
627 }
628
629 @SuppressWarnings("unchecked")
630 private static <T> Supplier<T[]> boxSupplier(T identity) {
631 return () -> (T[]) new Object[] { identity };
632 }
633
634 /**
635 * Returns a {@code Collector} which performs a reduction of its
636 * input elements under a specified {@code BinaryOperator}. The result
637 * is described as an {@code Optional<T>}.
638 *
639 * @apiNote
640 * The {@code reducing()} collectors are most useful when used in a
641 * multi-level reduction, downstream of {@code groupingBy} or
642 * {@code partitioningBy}. To perform a simple reduction on a stream,
643 * use {@link Stream#reduce(BinaryOperator)} instead.
644 *
645 * <p>For example, given a stream of {@code Person}, to calculate tallest
646 * person in each city:
647 * <pre>{@code
648 * Comparator<Person> byHeight = Comparator.comparing(Person::getHeight);
649 * Map<City, Person> tallestByCity
650 * = people.stream().collect(groupingBy(Person::getCity, reducing(BinaryOperator.maxBy(byHeight))));
651 * }</pre>
652 *
653 * @param <T> element type for the input and output of the reduction
654 * @param op a {@code BinaryOperator<T>} used to reduce the input elements
655 * @return a {@code Collector} which implements the reduction operation
656 *
657 * @see #reducing(Object, BinaryOperator)
658 * @see #reducing(Object, Function, BinaryOperator)
659 */
660 public static <T> Collector<T, ?, Optional<T>>
661 reducing(BinaryOperator<T> op) {
662 class OptionalBox implements Consumer<T> {
663 T value = null;
664 boolean present = false;
665
666 @Override
667 public void accept(T t) {
668 if (present) {
669 value = op.apply(value, t);
670 }
671 else {
672 value = t;
673 present = true;
674 }
675 }
676 }
677
678 return new CollectorImpl<T, OptionalBox, Optional<T>>(
679 OptionalBox::new, OptionalBox::accept,
680 (a, b) -> { if (b.present) a.accept(b.value); return a; },
681 a -> Optional.ofNullable(a.value), CH_NOID);
682 }
683
684 /**
685 * Returns a {@code Collector} which performs a reduction of its
686 * input elements under a specified mapping function and
687 * {@code BinaryOperator}. This is a generalization of
688 * {@link #reducing(Object, BinaryOperator)} which allows a transformation
689 * of the elements before reduction.
690 *
691 * @apiNote
692 * The {@code reducing()} collectors are most useful when used in a
693 * multi-level reduction, downstream of {@code groupingBy} or
694 * {@code partitioningBy}. To perform a simple map-reduce on a stream,
695 * use {@link Stream#map(Function)} and {@link Stream#reduce(Object, BinaryOperator)}
696 * instead.
697 *
698 * <p>For example, given a stream of {@code Person}, to calculate the longest
699 * last name of residents in each city:
700 * <pre>{@code
701 * Comparator<String> byLength = Comparator.comparing(String::length);
702 * Map<City, String> longestLastNameByCity
703 * = people.stream().collect(groupingBy(Person::getCity,
704 * reducing(Person::getLastName, BinaryOperator.maxBy(byLength))));
705 * }</pre>
706 *
707 * @param <T> the type of the input elements
708 * @param <U> the type of the mapped values
709 * @param identity the identity value for the reduction (also, the value
710 * that is returned when there are no input elements)
711 * @param mapper a mapping function to apply to each input value
712 * @param op a {@code BinaryOperator<U>} used to reduce the mapped values
713 * @return a {@code Collector} implementing the map-reduce operation
714 *
715 * @see #reducing(Object, BinaryOperator)
716 * @see #reducing(BinaryOperator)
717 */
718 public static <T, U>
719 Collector<T, ?, U> reducing(U identity,
720 Function<? super T, ? extends U> mapper,
721 BinaryOperator<U> op) {
722 return new CollectorImpl<>(
723 boxSupplier(identity),
724 (a, t) -> { a[0] = op.apply(a[0], mapper.apply(t)); },
725 (a, b) -> { a[0] = op.apply(a[0], b[0]); return a; },
726 a -> a[0], CH_NOID);
727 }
728
729 /**
730 * Returns a {@code Collector} implementing a "group by" operation on
731 * input elements of type {@code T}, grouping elements according to a
732 * classification function, and returning the results in a {@code Map}.
733 *
734 * <p>The classification function maps elements to some key type {@code K}.
735 * The collector produces a {@code Map<K, List<T>>} whose keys are the
736 * values resulting from applying the classification function to the input
737 * elements, and whose corresponding values are {@code List}s containing the
738 * input elements which map to the associated key under the classification
739 * function.
740 *
741 * <p>There are no guarantees on the type, mutability, serializability, or
742 * thread-safety of the {@code Map} or {@code List} objects returned.
743 * @implSpec
744 * This produces a result similar to:
745 * <pre>{@code
746 * groupingBy(classifier, toList());
747 * }</pre>
748 *
749 * @implNote
750 * The returned {@code Collector} is not concurrent. For parallel stream
751 * pipelines, the {@code combiner} function operates by merging the keys
752 * from one map into another, which can be an expensive operation. If
753 * preservation of the order in which elements appear in the resulting {@code Map}
754 * collector is not required, using {@link #groupingByConcurrent(Function)}
755 * may offer better parallel performance.
756 *
757 * @param <T> the type of the input elements
758 * @param <K> the type of the keys
759 * @param classifier the classifier function mapping input elements to keys
760 * @return a {@code Collector} implementing the group-by operation
761 *
762 * @see #groupingBy(Function, Collector)
763 * @see #groupingBy(Function, Supplier, Collector)
764 * @see #groupingByConcurrent(Function)
765 */
766 public static <T, K> Collector<T, ?, Map<K, List<T>>>
767 groupingBy(Function<? super T, ? extends K> classifier) {
768 return groupingBy(classifier, toList());
769 }
770
771 /**
772 * Returns a {@code Collector} implementing a cascaded "group by" operation
773 * on input elements of type {@code T}, grouping elements according to a
774 * classification function, and then performing a reduction operation on
775 * the values associated with a given key using the specified downstream
776 * {@code Collector}.
777 *
778 * <p>The classification function maps elements to some key type {@code K}.
779 * The downstream collector operates on elements of type {@code T} and
780 * produces a result of type {@code D}. The resulting collector produces a
781 * {@code Map<K, D>}.
782 *
783 * <p>There are no guarantees on the type, mutability,
784 * serializability, or thread-safety of the {@code Map} returned.
785 *
786 * <p>For example, to compute the set of last names of people in each city:
787 * <pre>{@code
788 * Map<City, Set<String>> namesByCity
789 * = people.stream().collect(groupingBy(Person::getCity,
790 * mapping(Person::getLastName, toSet())));
791 * }</pre>
792 *
793 * @implNote
794 * The returned {@code Collector} is not concurrent. For parallel stream
795 * pipelines, the {@code combiner} function operates by merging the keys
796 * from one map into another, which can be an expensive operation. If
797 * preservation of the order in which elements are presented to the downstream
798 * collector is not required, using {@link #groupingByConcurrent(Function, Collector)}
799 * may offer better parallel performance.
800 *
801 * @param <T> the type of the input elements
802 * @param <K> the type of the keys
803 * @param <A> the intermediate accumulation type of the downstream collector
804 * @param <D> the result type of the downstream reduction
805 * @param classifier a classifier function mapping input elements to keys
806 * @param downstream a {@code Collector} implementing the downstream reduction
807 * @return a {@code Collector} implementing the cascaded group-by operation
808 * @see #groupingBy(Function)
809 *
810 * @see #groupingBy(Function, Supplier, Collector)
811 * @see #groupingByConcurrent(Function, Collector)
812 */
813 public static <T, K, A, D>
814 Collector<T, ?, Map<K, D>> groupingBy(Function<? super T, ? extends K> classifier,
815 Collector<? super T, A, D> downstream) {
816 return groupingBy(classifier, HashMap::new, downstream);
817 }
818
819 /**
820 * Returns a {@code Collector} implementing a cascaded "group by" operation
821 * on input elements of type {@code T}, grouping elements according to a
822 * classification function, and then performing a reduction operation on
823 * the values associated with a given key using the specified downstream
824 * {@code Collector}. The {@code Map} produced by the Collector is created
825 * with the supplied factory function.
826 *
827 * <p>The classification function maps elements to some key type {@code K}.
828 * The downstream collector operates on elements of type {@code T} and
829 * produces a result of type {@code D}. The resulting collector produces a
830 * {@code Map<K, D>}.
831 *
832 * <p>For example, to compute the set of last names of people in each city,
833 * where the city names are sorted:
834 * <pre>{@code
835 * Map<City, Set<String>> namesByCity
836 * = people.stream().collect(groupingBy(Person::getCity, TreeMap::new,
837 * mapping(Person::getLastName, toSet())));
838 * }</pre>
839 *
840 * @implNote
841 * The returned {@code Collector} is not concurrent. For parallel stream
842 * pipelines, the {@code combiner} function operates by merging the keys
843 * from one map into another, which can be an expensive operation. If
844 * preservation of the order in which elements are presented to the downstream
845 * collector is not required, using {@link #groupingByConcurrent(Function, Supplier, Collector)}
846 * may offer better parallel performance.
847 *
848 * @param <T> the type of the input elements
849 * @param <K> the type of the keys
850 * @param <A> the intermediate accumulation type of the downstream collector
851 * @param <D> the result type of the downstream reduction
852 * @param <M> the type of the resulting {@code Map}
853 * @param classifier a classifier function mapping input elements to keys
854 * @param downstream a {@code Collector} implementing the downstream reduction
855 * @param mapFactory a function which, when called, produces a new empty
856 * {@code Map} of the desired type
857 * @return a {@code Collector} implementing the cascaded group-by operation
858 *
859 * @see #groupingBy(Function, Collector)
860 * @see #groupingBy(Function)
861 * @see #groupingByConcurrent(Function, Supplier, Collector)
862 */
863 public static <T, K, D, A, M extends Map<K, D>>
864 Collector<T, ?, M> groupingBy(Function<? super T, ? extends K> classifier,
865 Supplier<M> mapFactory,
866 Collector<? super T, A, D> downstream) {
867 Supplier<A> downstreamSupplier = downstream.supplier();
868 BiConsumer<A, ? super T> downstreamAccumulator = downstream.accumulator();
869 BiConsumer<Map<K, A>, T> accumulator = (m, t) -> {
870 K key = Objects.requireNonNull(classifier.apply(t), "element cannot be mapped to a null key");
871 A container = m.computeIfAbsent(key, k -> downstreamSupplier.get());
872 downstreamAccumulator.accept(container, t);
873 };
874 BinaryOperator<Map<K, A>> merger = Collectors.<K, A, Map<K, A>>mapMerger(downstream.combiner());
875 @SuppressWarnings("unchecked")
876 Supplier<Map<K, A>> mangledFactory = (Supplier<Map<K, A>>) mapFactory;
877
878 if (downstream.characteristics().contains(Collector.Characteristics.IDENTITY_FINISH)) {
879 return new CollectorImpl<>(mangledFactory, accumulator, merger, CH_ID);
880 }
881 else {
882 @SuppressWarnings("unchecked")
883 Function<A, A> downstreamFinisher = (Function<A, A>) downstream.finisher();
884 Function<Map<K, A>, M> finisher = intermediate -> {
885 intermediate.replaceAll((k, v) -> downstreamFinisher.apply(v));
886 @SuppressWarnings("unchecked")
887 M castResult = (M) intermediate;
888 return castResult;
889 };
890 return new CollectorImpl<>(mangledFactory, accumulator, merger, finisher, CH_NOID);
891 }
892 }
893
894 /**
895 * Returns a concurrent {@code Collector} implementing a "group by"
896 * operation on input elements of type {@code T}, grouping elements
897 * according to a classification function.
898 *
899 * <p>This is a {@link Collector.Characteristics#CONCURRENT concurrent} and
900 * {@link Collector.Characteristics#UNORDERED unordered} Collector.
901 *
902 * <p>The classification function maps elements to some key type {@code K}.
903 * The collector produces a {@code ConcurrentMap<K, List<T>>} whose keys are the
904 * values resulting from applying the classification function to the input
905 * elements, and whose corresponding values are {@code List}s containing the
906 * input elements which map to the associated key under the classification
907 * function.
908 *
909 * <p>There are no guarantees on the type, mutability, or serializability
910 * of the {@code Map} or {@code List} objects returned, or of the
911 * thread-safety of the {@code List} objects returned.
912 * @implSpec
913 * This produces a result similar to:
914 * <pre>{@code
915 * groupingByConcurrent(classifier, toList());
916 * }</pre>
917 *
918 * @param <T> the type of the input elements
919 * @param <K> the type of the keys
920 * @param classifier a classifier function mapping input elements to keys
921 * @return a concurrent, unordered {@code Collector} implementing the group-by operation
922 *
923 * @see #groupingBy(Function)
924 * @see #groupingByConcurrent(Function, Collector)
925 * @see #groupingByConcurrent(Function, Supplier, Collector)
926 */
927 public static <T, K>
928 Collector<T, ?, ConcurrentMap<K, List<T>>>
929 groupingByConcurrent(Function<? super T, ? extends K> classifier) {
930 return groupingByConcurrent(classifier, ConcurrentHashMap::new, toList());
931 }
932
933 /**
934 * Returns a concurrent {@code Collector} implementing a cascaded "group by"
935 * operation on input elements of type {@code T}, grouping elements
936 * according to a classification function, and then performing a reduction
937 * operation on the values associated with a given key using the specified
938 * downstream {@code Collector}.
939 *
940 * <p>This is a {@link Collector.Characteristics#CONCURRENT concurrent} and
941 * {@link Collector.Characteristics#UNORDERED unordered} Collector.
942 *
943 * <p>The classification function maps elements to some key type {@code K}.
944 * The downstream collector operates on elements of type {@code T} and
945 * produces a result of type {@code D}. The resulting collector produces a
946 * {@code Map<K, D>}.
947 *
948 * <p>For example, to compute the set of last names of people in each city,
949 * where the city names are sorted:
950 * <pre>{@code
951 * ConcurrentMap<City, Set<String>> namesByCity
952 * = people.stream().collect(groupingByConcurrent(Person::getCity,
953 * mapping(Person::getLastName, toSet())));
954 * }</pre>
955 *
956 * @param <T> the type of the input elements
957 * @param <K> the type of the keys
958 * @param <A> the intermediate accumulation type of the downstream collector
959 * @param <D> the result type of the downstream reduction
960 * @param classifier a classifier function mapping input elements to keys
961 * @param downstream a {@code Collector} implementing the downstream reduction
962 * @return a concurrent, unordered {@code Collector} implementing the cascaded group-by operation
963 *
964 * @see #groupingBy(Function, Collector)
965 * @see #groupingByConcurrent(Function)
966 * @see #groupingByConcurrent(Function, Supplier, Collector)
967 */
968 public static <T, K, A, D>
969 Collector<T, ?, ConcurrentMap<K, D>> groupingByConcurrent(Function<? super T, ? extends K> classifier,
970 Collector<? super T, A, D> downstream) {
971 return groupingByConcurrent(classifier, ConcurrentHashMap::new, downstream);
972 }
973
974 /**
975 * Returns a concurrent {@code Collector} implementing a cascaded "group by"
976 * operation on input elements of type {@code T}, grouping elements
977 * according to a classification function, and then performing a reduction
978 * operation on the values associated with a given key using the specified
979 * downstream {@code Collector}. The {@code ConcurrentMap} produced by the
980 * Collector is created with the supplied factory function.
981 *
982 * <p>This is a {@link Collector.Characteristics#CONCURRENT concurrent} and
983 * {@link Collector.Characteristics#UNORDERED unordered} Collector.
984 *
985 * <p>The classification function maps elements to some key type {@code K}.
986 * The downstream collector operates on elements of type {@code T} and
987 * produces a result of type {@code D}. The resulting collector produces a
988 * {@code Map<K, D>}.
989 *
990 * <p>For example, to compute the set of last names of people in each city,
991 * where the city names are sorted:
992 * <pre>{@code
993 * ConcurrentMap<City, Set<String>> namesByCity
994 * = people.stream().collect(groupingBy(Person::getCity, ConcurrentSkipListMap::new,
995 * mapping(Person::getLastName, toSet())));
996 * }</pre>
997 *
998 *
999 * @param <T> the type of the input elements
1000 * @param <K> the type of the keys
1001 * @param <A> the intermediate accumulation type of the downstream collector
1002 * @param <D> the result type of the downstream reduction
1003 * @param <M> the type of the resulting {@code ConcurrentMap}
1004 * @param classifier a classifier function mapping input elements to keys
1005 * @param downstream a {@code Collector} implementing the downstream reduction
1006 * @param mapFactory a function which, when called, produces a new empty
1007 * {@code ConcurrentMap} of the desired type
1008 * @return a concurrent, unordered {@code Collector} implementing the cascaded group-by operation
1009 *
1010 * @see #groupingByConcurrent(Function)
1011 * @see #groupingByConcurrent(Function, Collector)
1012 * @see #groupingBy(Function, Supplier, Collector)
1013 */
1014 public static <T, K, A, D, M extends ConcurrentMap<K, D>>
1015 Collector<T, ?, M> groupingByConcurrent(Function<? super T, ? extends K> classifier,
1016 Supplier<M> mapFactory,
1017 Collector<? super T, A, D> downstream) {
1018 Supplier<A> downstreamSupplier = downstream.supplier();
1019 BiConsumer<A, ? super T> downstreamAccumulator = downstream.accumulator();
1020 BinaryOperator<ConcurrentMap<K, A>> merger = Collectors.<K, A, ConcurrentMap<K, A>>mapMerger(downstream.combiner());
1021 @SuppressWarnings("unchecked")
1022 Supplier<ConcurrentMap<K, A>> mangledFactory = (Supplier<ConcurrentMap<K, A>>) mapFactory;
1023 BiConsumer<ConcurrentMap<K, A>, T> accumulator;
1024 if (downstream.characteristics().contains(Collector.Characteristics.CONCURRENT)) {
1025 accumulator = (m, t) -> {
1026 K key = Objects.requireNonNull(classifier.apply(t), "element cannot be mapped to a null key");
1027 A resultContainer = m.computeIfAbsent(key, k -> downstreamSupplier.get());
1028 downstreamAccumulator.accept(resultContainer, t);
1029 };
1030 }
1031 else {
1032 accumulator = (m, t) -> {
1033 K key = Objects.requireNonNull(classifier.apply(t), "element cannot be mapped to a null key");
1034 A resultContainer = m.computeIfAbsent(key, k -> downstreamSupplier.get());
1035 synchronized (resultContainer) {
1036 downstreamAccumulator.accept(resultContainer, t);
1037 }
1038 };
1039 }
1040
1041 if (downstream.characteristics().contains(Collector.Characteristics.IDENTITY_FINISH)) {
1042 return new CollectorImpl<>(mangledFactory, accumulator, merger, CH_CONCURRENT_ID);
1043 }
1044 else {
1045 @SuppressWarnings("unchecked")
1046 Function<A, A> downstreamFinisher = (Function<A, A>) downstream.finisher();
1047 Function<ConcurrentMap<K, A>, M> finisher = intermediate -> {
1048 intermediate.replaceAll((k, v) -> downstreamFinisher.apply(v));
1049 @SuppressWarnings("unchecked")
1050 M castResult = (M) intermediate;
1051 return castResult;
1052 };
1053 return new CollectorImpl<>(mangledFactory, accumulator, merger, finisher, CH_CONCURRENT_NOID);
1054 }
1055 }
1056
1057 /**
1058 * Returns a {@code Collector} which partitions the input elements according
1059 * to a {@code Predicate}, and organizes them into a
1060 * {@code Map<Boolean, List<T>>}.
1061 *
1062 * There are no guarantees on the type, mutability,
1063 * serializability, or thread-safety of the {@code Map} returned.
1064 *
1065 * @param <T> the type of the input elements
1066 * @param predicate a predicate used for classifying input elements
1067 * @return a {@code Collector} implementing the partitioning operation
1068 *
1069 * @see #partitioningBy(Predicate, Collector)
1070 */
1071 public static <T>
1072 Collector<T, ?, Map<Boolean, List<T>>> partitioningBy(Predicate<? super T> predicate) {
1073 return partitioningBy(predicate, toList());
1074 }
1075
1076 /**
1077 * Returns a {@code Collector} which partitions the input elements according
1078 * to a {@code Predicate}, reduces the values in each partition according to
1079 * another {@code Collector}, and organizes them into a
1080 * {@code Map<Boolean, D>} whose values are the result of the downstream
1081 * reduction.
1082 *
1083 * <p>There are no guarantees on the type, mutability,
1084 * serializability, or thread-safety of the {@code Map} returned.
1085 *
1086 * @param <T> the type of the input elements
1087 * @param <A> the intermediate accumulation type of the downstream collector
1088 * @param <D> the result type of the downstream reduction
1089 * @param predicate a predicate used for classifying input elements
1090 * @param downstream a {@code Collector} implementing the downstream
1091 * reduction
1092 * @return a {@code Collector} implementing the cascaded partitioning
1093 * operation
1094 *
1095 * @see #partitioningBy(Predicate)
1096 */
1097 public static <T, D, A>
1098 Collector<T, ?, Map<Boolean, D>> partitioningBy(Predicate<? super T> predicate,
1099 Collector<? super T, A, D> downstream) {
1100 BiConsumer<A, ? super T> downstreamAccumulator = downstream.accumulator();
1101 BiConsumer<Partition<A>, T> accumulator = (result, t) ->
1102 downstreamAccumulator.accept(predicate.test(t) ? result.forTrue : result.forFalse, t);
1103 BinaryOperator<A> op = downstream.combiner();
1104 BinaryOperator<Partition<A>> merger = (left, right) ->
1105 new Partition<>(op.apply(left.forTrue, right.forTrue),
1106 op.apply(left.forFalse, right.forFalse));
1107 Supplier<Partition<A>> supplier = () ->
1108 new Partition<>(downstream.supplier().get(),
1109 downstream.supplier().get());
1110 if (downstream.characteristics().contains(Collector.Characteristics.IDENTITY_FINISH)) {
1111 return new CollectorImpl<>(supplier, accumulator, merger, CH_ID);
1112 }
1113 else {
1114 Function<Partition<A>, Map<Boolean, D>> finisher = par ->
1115 new Partition<>(downstream.finisher().apply(par.forTrue),
1116 downstream.finisher().apply(par.forFalse));
1117 return new CollectorImpl<>(supplier, accumulator, merger, finisher, CH_NOID);
1118 }
1119 }
1120
1121 /**
1122 * Returns a {@code Collector} that accumulates elements into a
1123 * {@code Map} whose keys and values are the result of applying the provided
1124 * mapping functions to the input elements.
1125 *
1126 * <p>If the mapped keys contains duplicates (according to
1127 * {@link Object#equals(Object)}), an {@code IllegalStateException} is
1128 * thrown when the collection operation is performed. If the mapped keys
1129 * may have duplicates, use {@link #toMap(Function, Function, BinaryOperator)}
1130 * instead.
1131 *
1132 * @apiNote
1133 * It is common for either the key or the value to be the input elements.
1134 * In this case, the utility method
1135 * {@link java.util.function.Function#identity()} may be helpful.
1136 * For example, the following produces a {@code Map} mapping
1137 * students to their grade point average:
1138 * <pre>{@code
1139 * Map<Student, Double> studentToGPA
1140 * students.stream().collect(toMap(Functions.identity(),
1141 * student -> computeGPA(student)));
1142 * }</pre>
1143 * And the following produces a {@code Map} mapping a unique identifier to
1144 * students:
1145 * <pre>{@code
1146 * Map<String, Student> studentIdToStudent
1147 * students.stream().collect(toMap(Student::getId,
1148 * Functions.identity());
1149 * }</pre>
1150 *
1151 * @implNote
1152 * The returned {@code Collector} is not concurrent. For parallel stream
1153 * pipelines, the {@code combiner} function operates by merging the keys
1154 * from one map into another, which can be an expensive operation. If it is
1155 * not required that results are inserted into the {@code Map} in encounter
1156 * order, using {@link #toConcurrentMap(Function, Function)}
1157 * may offer better parallel performance.
1158 *
1159 * @param <T> the type of the input elements
1160 * @param <K> the output type of the key mapping function
1161 * @param <U> the output type of the value mapping function
1162 * @param keyMapper a mapping function to produce keys
1163 * @param valueMapper a mapping function to produce values
1164 * @return a {@code Collector} which collects elements into a {@code Map}
1165 * whose keys and values are the result of applying mapping functions to
1166 * the input elements
1167 *
1168 * @see #toMap(Function, Function, BinaryOperator)
1169 * @see #toMap(Function, Function, BinaryOperator, Supplier)
1170 * @see #toConcurrentMap(Function, Function)
1171 */
1172 public static <T, K, U>
1173 Collector<T, ?, Map<K,U>> toMap(Function<? super T, ? extends K> keyMapper,
1174 Function<? super T, ? extends U> valueMapper) {
1175 return toMap(keyMapper, valueMapper, throwingMerger(), HashMap::new);
1176 }
1177
1178 /**
1179 * Returns a {@code Collector} that accumulates elements into a
1180 * {@code Map} whose keys and values are the result of applying the provided
1181 * mapping functions to the input elements.
1182 *
1183 * <p>If the mapped
1184 * keys contains duplicates (according to {@link Object#equals(Object)}),
1185 * the value mapping function is applied to each equal element, and the
1186 * results are merged using the provided merging function.
1187 *
1188 * @apiNote
1189 * There are multiple ways to deal with collisions between multiple elements
1190 * mapping to the same key. The other forms of {@code toMap} simply use
1191 * a merge function that throws unconditionally, but you can easily write
1192 * more flexible merge policies. For example, if you have a stream
1193 * of {@code Person}, and you want to produce a "phone book" mapping name to
1194 * address, but it is possible that two persons have the same name, you can
1195 * do as follows to gracefully deals with these collisions, and produce a
1196 * {@code Map} mapping names to a concatenated list of addresses:
1197 * <pre>{@code
1198 * Map<String, String> phoneBook
1199 * people.stream().collect(toMap(Person::getName,
1200 * Person::getAddress,
1201 * (s, a) -> s + ", " + a));
1202 * }</pre>
1203 *
1204 * @implNote
1205 * The returned {@code Collector} is not concurrent. For parallel stream
1206 * pipelines, the {@code combiner} function operates by merging the keys
1207 * from one map into another, which can be an expensive operation. If it is
1208 * not required that results are merged into the {@code Map} in encounter
1209 * order, using {@link #toConcurrentMap(Function, Function, BinaryOperator)}
1210 * may offer better parallel performance.
1211 *
1212 * @param <T> the type of the input elements
1213 * @param <K> the output type of the key mapping function
1214 * @param <U> the output type of the value mapping function
1215 * @param keyMapper a mapping function to produce keys
1216 * @param valueMapper a mapping function to produce values
1217 * @param mergeFunction a merge function, used to resolve collisions between
1218 * values associated with the same key, as supplied
1219 * to {@link Map#merge(Object, Object, BiFunction)}
1220 * @return a {@code Collector} which collects elements into a {@code Map}
1221 * whose keys are the result of applying a key mapping function to the input
1222 * elements, and whose values are the result of applying a value mapping
1223 * function to all input elements equal to the key and combining them
1224 * using the merge function
1225 *
1226 * @see #toMap(Function, Function)
1227 * @see #toMap(Function, Function, BinaryOperator, Supplier)
1228 * @see #toConcurrentMap(Function, Function, BinaryOperator)
1229 */
1230 public static <T, K, U>
1231 Collector<T, ?, Map<K,U>> toMap(Function<? super T, ? extends K> keyMapper,
1232 Function<? super T, ? extends U> valueMapper,
1233 BinaryOperator<U> mergeFunction) {
1234 return toMap(keyMapper, valueMapper, mergeFunction, HashMap::new);
1235 }
1236
1237 /**
1238 * Returns a {@code Collector} that accumulates elements into a
1239 * {@code Map} whose keys and values are the result of applying the provided
1240 * mapping functions to the input elements.
1241 *
1242 * <p>If the mapped
1243 * keys contains duplicates (according to {@link Object#equals(Object)}),
1244 * the value mapping function is applied to each equal element, and the
1245 * results are merged using the provided merging function. The {@code Map}
1246 * is created by a provided supplier function.
1247 *
1248 * @implNote
1249 * The returned {@code Collector} is not concurrent. For parallel stream
1250 * pipelines, the {@code combiner} function operates by merging the keys
1251 * from one map into another, which can be an expensive operation. If it is
1252 * not required that results are merged into the {@code Map} in encounter
1253 * order, using {@link #toConcurrentMap(Function, Function, BinaryOperator, Supplier)}
1254 * may offer better parallel performance.
1255 *
1256 * @param <T> the type of the input elements
1257 * @param <K> the output type of the key mapping function
1258 * @param <U> the output type of the value mapping function
1259 * @param <M> the type of the resulting {@code Map}
1260 * @param keyMapper a mapping function to produce keys
1261 * @param valueMapper a mapping function to produce values
1262 * @param mergeFunction a merge function, used to resolve collisions between
1263 * values associated with the same key, as supplied
1264 * to {@link Map#merge(Object, Object, BiFunction)}
1265 * @param mapSupplier a function which returns a new, empty {@code Map} into
1266 * which the results will be inserted
1267 * @return a {@code Collector} which collects elements into a {@code Map}
1268 * whose keys are the result of applying a key mapping function to the input
1269 * elements, and whose values are the result of applying a value mapping
1270 * function to all input elements equal to the key and combining them
1271 * using the merge function
1272 *
1273 * @see #toMap(Function, Function)
1274 * @see #toMap(Function, Function, BinaryOperator)
1275 * @see #toConcurrentMap(Function, Function, BinaryOperator, Supplier)
1276 */
1277 public static <T, K, U, M extends Map<K, U>>
1278 Collector<T, ?, M> toMap(Function<? super T, ? extends K> keyMapper,
1279 Function<? super T, ? extends U> valueMapper,
1280 BinaryOperator<U> mergeFunction,
1281 Supplier<M> mapSupplier) {
1282 BiConsumer<M, T> accumulator
1283 = (map, element) -> map.merge(keyMapper.apply(element),
1284 valueMapper.apply(element), mergeFunction);
1285 return new CollectorImpl<>(mapSupplier, accumulator, mapMerger(mergeFunction), CH_ID);
1286 }
1287
1288 /**
1289 * Returns a concurrent {@code Collector} that accumulates elements into a
1290 * {@code ConcurrentMap} whose keys and values are the result of applying
1291 * the provided mapping functions to the input elements.
1292 *
1293 * <p>If the mapped keys contains duplicates (according to
1294 * {@link Object#equals(Object)}), an {@code IllegalStateException} is
1295 * thrown when the collection operation is performed. If the mapped keys
1296 * may have duplicates, use
1297 * {@link #toConcurrentMap(Function, Function, BinaryOperator)} instead.
1298 *
1299 * @apiNote
1300 * It is common for either the key or the value to be the input elements.
1301 * In this case, the utility method
1302 * {@link java.util.function.Function#identity()} may be helpful.
1303 * For example, the following produces a {@code Map} mapping
1304 * students to their grade point average:
1305 * <pre>{@code
1306 * Map<Student, Double> studentToGPA
1307 * students.stream().collect(toMap(Functions.identity(),
1308 * student -> computeGPA(student)));
1309 * }</pre>
1310 * And the following produces a {@code Map} mapping a unique identifier to
1311 * students:
1312 * <pre>{@code
1313 * Map<String, Student> studentIdToStudent
1314 * students.stream().collect(toConcurrentMap(Student::getId,
1315 * Functions.identity());
1316 * }</pre>
1317 *
1318 * <p>This is a {@link Collector.Characteristics#CONCURRENT concurrent} and
1319 * {@link Collector.Characteristics#UNORDERED unordered} Collector.
1320 *
1321 * @param <T> the type of the input elements
1322 * @param <K> the output type of the key mapping function
1323 * @param <U> the output type of the value mapping function
1324 * @param keyMapper the mapping function to produce keys
1325 * @param valueMapper the mapping function to produce values
1326 * @return a concurrent, unordered {@code Collector} which collects elements into a
1327 * {@code ConcurrentMap} whose keys are the result of applying a key mapping
1328 * function to the input elements, and whose values are the result of
1329 * applying a value mapping function to the input elements
1330 *
1331 * @see #toMap(Function, Function)
1332 * @see #toConcurrentMap(Function, Function, BinaryOperator)
1333 * @see #toConcurrentMap(Function, Function, BinaryOperator, Supplier)
1334 */
1335 public static <T, K, U>
1336 Collector<T, ?, ConcurrentMap<K,U>> toConcurrentMap(Function<? super T, ? extends K> keyMapper,
1337 Function<? super T, ? extends U> valueMapper) {
1338 return toConcurrentMap(keyMapper, valueMapper, throwingMerger(), ConcurrentHashMap::new);
1339 }
1340
1341 /**
1342 * Returns a concurrent {@code Collector} that accumulates elements into a
1343 * {@code ConcurrentMap} whose keys and values are the result of applying
1344 * the provided mapping functions to the input elements.
1345 *
1346 * <p>If the mapped keys contains duplicates (according to {@link Object#equals(Object)}),
1347 * the value mapping function is applied to each equal element, and the
1348 * results are merged using the provided merging function.
1349 *
1350 * @apiNote
1351 * There are multiple ways to deal with collisions between multiple elements
1352 * mapping to the same key. The other forms of {@code toConcurrentMap} simply use
1353 * a merge function that throws unconditionally, but you can easily write
1354 * more flexible merge policies. For example, if you have a stream
1355 * of {@code Person}, and you want to produce a "phone book" mapping name to
1356 * address, but it is possible that two persons have the same name, you can
1357 * do as follows to gracefully deals with these collisions, and produce a
1358 * {@code Map} mapping names to a concatenated list of addresses:
1359 * <pre>{@code
1360 * Map<String, String> phoneBook
1361 * people.stream().collect(toConcurrentMap(Person::getName,
1362 * Person::getAddress,
1363 * (s, a) -> s + ", " + a));
1364 * }</pre>
1365 *
1366 * <p>This is a {@link Collector.Characteristics#CONCURRENT concurrent} and
1367 * {@link Collector.Characteristics#UNORDERED unordered} Collector.
1368 *
1369 * @param <T> the type of the input elements
1370 * @param <K> the output type of the key mapping function
1371 * @param <U> the output type of the value mapping function
1372 * @param keyMapper a mapping function to produce keys
1373 * @param valueMapper a mapping function to produce values
1374 * @param mergeFunction a merge function, used to resolve collisions between
1375 * values associated with the same key, as supplied
1376 * to {@link Map#merge(Object, Object, BiFunction)}
1377 * @return a concurrent, unordered {@code Collector} which collects elements into a
1378 * {@code ConcurrentMap} whose keys are the result of applying a key mapping
1379 * function to the input elements, and whose values are the result of
1380 * applying a value mapping function to all input elements equal to the key
1381 * and combining them using the merge function
1382 *
1383 * @see #toConcurrentMap(Function, Function)
1384 * @see #toConcurrentMap(Function, Function, BinaryOperator, Supplier)
1385 * @see #toMap(Function, Function, BinaryOperator)
1386 */
1387 public static <T, K, U>
1388 Collector<T, ?, ConcurrentMap<K,U>>
1389 toConcurrentMap(Function<? super T, ? extends K> keyMapper,
1390 Function<? super T, ? extends U> valueMapper,
1391 BinaryOperator<U> mergeFunction) {
1392 return toConcurrentMap(keyMapper, valueMapper, mergeFunction, ConcurrentHashMap::new);
1393 }
1394
1395 /**
1396 * Returns a concurrent {@code Collector} that accumulates elements into a
1397 * {@code ConcurrentMap} whose keys and values are the result of applying
1398 * the provided mapping functions to the input elements.
1399 *
1400 * <p>If the mapped keys contains duplicates (according to {@link Object#equals(Object)}),
1401 * the value mapping function is applied to each equal element, and the
1402 * results are merged using the provided merging function. The
1403 * {@code ConcurrentMap} is created by a provided supplier function.
1404 *
1405 * <p>This is a {@link Collector.Characteristics#CONCURRENT concurrent} and
1406 * {@link Collector.Characteristics#UNORDERED unordered} Collector.
1407 *
1408 * @param <T> the type of the input elements
1409 * @param <K> the output type of the key mapping function
1410 * @param <U> the output type of the value mapping function
1411 * @param <M> the type of the resulting {@code ConcurrentMap}
1412 * @param keyMapper a mapping function to produce keys
1413 * @param valueMapper a mapping function to produce values
1414 * @param mergeFunction a merge function, used to resolve collisions between
1415 * values associated with the same key, as supplied
1416 * to {@link Map#merge(Object, Object, BiFunction)}
1417 * @param mapSupplier a function which returns a new, empty {@code Map} into
1418 * which the results will be inserted
1419 * @return a concurrent, unordered {@code Collector} which collects elements into a
1420 * {@code ConcurrentMap} whose keys are the result of applying a key mapping
1421 * function to the input elements, and whose values are the result of
1422 * applying a value mapping function to all input elements equal to the key
1423 * and combining them using the merge function
1424 *
1425 * @see #toConcurrentMap(Function, Function)
1426 * @see #toConcurrentMap(Function, Function, BinaryOperator)
1427 * @see #toMap(Function, Function, BinaryOperator, Supplier)
1428 */
1429 public static <T, K, U, M extends ConcurrentMap<K, U>>
1430 Collector<T, ?, M> toConcurrentMap(Function<? super T, ? extends K> keyMapper,
1431 Function<? super T, ? extends U> valueMapper,
1432 BinaryOperator<U> mergeFunction,
1433 Supplier<M> mapSupplier) {
1434 BiConsumer<M, T> accumulator
1435 = (map, element) -> map.merge(keyMapper.apply(element),
1436 valueMapper.apply(element), mergeFunction);
1437 return new CollectorImpl<>(mapSupplier, accumulator, mapMerger(mergeFunction), CH_CONCURRENT_ID);
1438 }
1439
1440 /**
1441 * Returns a {@code Collector} which applies an {@code int}-producing
1442 * mapping function to each input element, and returns summary statistics
1443 * for the resulting values.
1444 *
1445 * @param <T> the type of the input elements
1446 * @param mapper a mapping function to apply to each element
1447 * @return a {@code Collector} implementing the summary-statistics reduction
1448 *
1449 * @see #summarizingDouble(ToDoubleFunction)
1450 * @see #summarizingLong(ToLongFunction)
1451 */
1452 public static <T>
1453 Collector<T, ?, IntSummaryStatistics> summarizingInt(ToIntFunction<? super T> mapper) {
1454 return new CollectorImpl<T, IntSummaryStatistics, IntSummaryStatistics>(
1455 IntSummaryStatistics::new,
1456 (r, t) -> r.accept(mapper.applyAsInt(t)),
1457 (l, r) -> { l.combine(r); return l; }, CH_ID);
1458 }
1459
1460 /**
1461 * Returns a {@code Collector} which applies an {@code long}-producing
1462 * mapping function to each input element, and returns summary statistics
1463 * for the resulting values.
1464 *
1465 * @param <T> the type of the input elements
1466 * @param mapper the mapping function to apply to each element
1467 * @return a {@code Collector} implementing the summary-statistics reduction
1468 *
1469 * @see #summarizingDouble(ToDoubleFunction)
1470 * @see #summarizingInt(ToIntFunction)
1471 */
1472 public static <T>
1473 Collector<T, ?, LongSummaryStatistics> summarizingLong(ToLongFunction<? super T> mapper) {
1474 return new CollectorImpl<T, LongSummaryStatistics, LongSummaryStatistics>(
1475 LongSummaryStatistics::new,
1476 (r, t) -> r.accept(mapper.applyAsLong(t)),
1477 (l, r) -> { l.combine(r); return l; }, CH_ID);
1478 }
1479
1480 /**
1481 * Returns a {@code Collector} which applies an {@code double}-producing
1482 * mapping function to each input element, and returns summary statistics
1483 * for the resulting values.
1484 *
1485 * @param <T> the type of the input elements
1486 * @param mapper a mapping function to apply to each element
1487 * @return a {@code Collector} implementing the summary-statistics reduction
1488 *
1489 * @see #summarizingLong(ToLongFunction)
1490 * @see #summarizingInt(ToIntFunction)
1491 */
1492 public static <T>
1493 Collector<T, ?, DoubleSummaryStatistics> summarizingDouble(ToDoubleFunction<? super T> mapper) {
1494 return new CollectorImpl<T, DoubleSummaryStatistics, DoubleSummaryStatistics>(
1495 DoubleSummaryStatistics::new,
1496 (r, t) -> r.accept(mapper.applyAsDouble(t)),
1497 (l, r) -> { l.combine(r); return l; }, CH_ID);
1498 }
1499
1500 /**
1501 * Implementation class used by partitioningBy.
1502 */
1503 private static final class Partition<T>
1504 extends AbstractMap<Boolean, T>
1505 implements Map<Boolean, T> {
1506 final T forTrue;
1507 final T forFalse;
1508
1509 Partition(T forTrue, T forFalse) {
1510 this.forTrue = forTrue;
1511 this.forFalse = forFalse;
1512 }
1513
1514 @Override
1515 public Set<Map.Entry<Boolean, T>> entrySet() {
1516 return new AbstractSet<Map.Entry<Boolean, T>>() {
1517 @Override
1518 public Iterator<Map.Entry<Boolean, T>> iterator() {
1519 Map.Entry<Boolean, T> falseEntry = new SimpleImmutableEntry<>(false, forFalse);
1520 Map.Entry<Boolean, T> trueEntry = new SimpleImmutableEntry<>(true, forTrue);
1521 return Arrays.asList(falseEntry, trueEntry).iterator();
1522 }
1523
1524 @Override
1525 public int size() {
1526 return 2;
1527 }
1528 };
1529 }
1530 }
1531 }