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 /*
509 * In the arrays allocated for the collect operation, index 0
510 * holds the high-order bits of the running sum and index 1
511 * holds the low-order bits of the sum computed via
512 * compensated summation.
513 */
514 return new CollectorImpl<>(
515 () -> new double[2],
516 (a, t) -> { sumWithCompensation(a, mapper.applyAsDouble(t)); },
517 (a, b) -> { sumWithCompensation(a, b[0]); return sumWithCompensation(a, b[1]); },
518 // Better error bounds to add both terms as the final sum
519 a -> a[0] + a[1],
520 CH_NOID);
521 }
522
523 /**
524 * Incorporate a new double value using Kahan summation /
525 * compensation summation.
526 *
527 * High-order bits of the sum are in intermediateSum[0], low-order
528 * bits of the sum are in intermediateSum[1], any additional
529 * elements are application-specific.
530 *
531 * @param intermediateSum the high-order and low-order words of the intermediate sum
532 * @param value the name value to be included in the running sum
533 */
534 static double[] sumWithCompensation(double[] intermediateSum, double value) {
535 double tmp = value - intermediateSum[1];
536 double sum = intermediateSum[0];
537 double velvel = sum + tmp; // Little wolf of rounding error
538 intermediateSum[1] = (velvel - sum) - tmp;
539 intermediateSum[0] = velvel;
540 return intermediateSum;
541 }
542
543
544 /**
545 * Returns a {@code Collector} that produces the arithmetic mean of an integer-valued
546 * function applied to the input elements. If no elements are present,
547 * the result is 0.
548 *
549 * @param <T> the type of the input elements
550 * @param mapper a function extracting the property to be summed
551 * @return a {@code Collector} that produces the sum of a derived property
552 */
553 public static <T> Collector<T, ?, Double>
554 averagingInt(ToIntFunction<? super T> mapper) {
555 return new CollectorImpl<>(
556 () -> new long[2],
557 (a, t) -> { a[0] += mapper.applyAsInt(t); a[1]++; },
558 (a, b) -> { a[0] += b[0]; a[1] += b[1]; return a; },
559 a -> (a[1] == 0) ? 0.0d : (double) a[0] / a[1], CH_NOID);
560 }
561
562 /**
563 * Returns a {@code Collector} that produces the arithmetic mean of a long-valued
564 * function applied to the input elements. If no elements are present,
565 * the result is 0.
566 *
567 * @param <T> the type of the input elements
568 * @param mapper a function extracting the property to be summed
569 * @return a {@code Collector} that produces the sum of a derived property
570 */
571 public static <T> Collector<T, ?, Double>
572 averagingLong(ToLongFunction<? super T> mapper) {
573 return new CollectorImpl<>(
574 () -> new long[2],
575 (a, t) -> { a[0] += mapper.applyAsLong(t); a[1]++; },
576 (a, b) -> { a[0] += b[0]; a[1] += b[1]; return a; },
577 a -> (a[1] == 0) ? 0.0d : (double) a[0] / a[1], CH_NOID);
578 }
579
580 /**
581 * Returns a {@code Collector} that produces the arithmetic mean of a double-valued
582 * function applied to the input elements. If no elements are present,
583 * the result is 0.
584 *
585 * <p>The average returned can vary depending upon the order in which
586 * values are recorded, due to accumulated rounding error in
587 * addition of values of differing magnitudes. Values sorted by increasing
588 * absolute magnitude tend to yield more accurate results. If any recorded
589 * value is a {@code NaN} or the sum is at any point a {@code NaN} then the
590 * average will be {@code NaN}.
591 *
592 * @implNote The {@code double} format can represent all
593 * consecutive integers in the range -2<sup>53</sup> to
594 * 2<sup>53</sup>. If the pipeline has more than 2<sup>53</sup>
595 * values, the divisor in the average computation will saturate at
596 * 2<sup>53</sup>, leading to additional numerical errors.
597 *
598 * @param <T> the type of the input elements
599 * @param mapper a function extracting the property to be summed
600 * @return a {@code Collector} that produces the sum of a derived property
601 */
602 public static <T> Collector<T, ?, Double>
603 averagingDouble(ToDoubleFunction<? super T> mapper) {
604 /*
605 * In the arrays allocated for the collect operation, index 0
606 * holds the high-order bits of the running sum, index 1 holds
607 * the low-order bits of the sum computed via compensated
608 * summation, and index 2 holds the number of values seen.
609 */
610 return new CollectorImpl<>(
611 () -> new double[3],
612 (a, t) -> { sumWithCompensation(a, mapper.applyAsDouble(t)); a[2]++; },
613 (a, b) -> { sumWithCompensation(a, b[0]); sumWithCompensation(a, b[1]); a[2] += b[2]; return a; },
614 // Better error bounds to add both terms as the final sum to compute average
615 a -> (a[2] == 0) ? 0.0d : ((a[0] + a[1]) / a[2]),
616 CH_NOID);
617 }
618
619 /**
620 * Returns a {@code Collector} which performs a reduction of its
621 * input elements under a specified {@code BinaryOperator} using the
622 * provided identity.
623 *
624 * @apiNote
625 * The {@code reducing()} collectors are most useful when used in a
626 * multi-level reduction, downstream of {@code groupingBy} or
627 * {@code partitioningBy}. To perform a simple reduction on a stream,
628 * use {@link Stream#reduce(Object, BinaryOperator)}} instead.
629 *
630 * @param <T> element type for the input and output of the reduction
631 * @param identity the identity value for the reduction (also, the value
632 * that is returned when there are no input elements)
633 * @param op a {@code BinaryOperator<T>} used to reduce the input elements
634 * @return a {@code Collector} which implements the reduction operation
635 *
636 * @see #reducing(BinaryOperator)
637 * @see #reducing(Object, Function, BinaryOperator)
638 */
639 public static <T> Collector<T, ?, T>
640 reducing(T identity, BinaryOperator<T> op) {
641 return new CollectorImpl<>(
642 boxSupplier(identity),
643 (a, t) -> { a[0] = op.apply(a[0], t); },
644 (a, b) -> { a[0] = op.apply(a[0], b[0]); return a; },
645 a -> a[0],
646 CH_NOID);
647 }
648
649 @SuppressWarnings("unchecked")
650 private static <T> Supplier<T[]> boxSupplier(T identity) {
651 return () -> (T[]) new Object[] { identity };
652 }
653
654 /**
655 * Returns a {@code Collector} which performs a reduction of its
656 * input elements under a specified {@code BinaryOperator}. The result
657 * is described as an {@code Optional<T>}.
658 *
659 * @apiNote
660 * The {@code reducing()} collectors are most useful when used in a
661 * multi-level reduction, downstream of {@code groupingBy} or
662 * {@code partitioningBy}. To perform a simple reduction on a stream,
663 * use {@link Stream#reduce(BinaryOperator)} instead.
664 *
665 * <p>For example, given a stream of {@code Person}, to calculate tallest
666 * person in each city:
667 * <pre>{@code
668 * Comparator<Person> byHeight = Comparator.comparing(Person::getHeight);
669 * Map<City, Person> tallestByCity
670 * = people.stream().collect(groupingBy(Person::getCity, reducing(BinaryOperator.maxBy(byHeight))));
671 * }</pre>
672 *
673 * @param <T> element type for the input and output of the reduction
674 * @param op a {@code BinaryOperator<T>} used to reduce the input elements
675 * @return a {@code Collector} which implements the reduction operation
676 *
677 * @see #reducing(Object, BinaryOperator)
678 * @see #reducing(Object, Function, BinaryOperator)
679 */
680 public static <T> Collector<T, ?, Optional<T>>
681 reducing(BinaryOperator<T> op) {
682 class OptionalBox implements Consumer<T> {
683 T value = null;
684 boolean present = false;
685
686 @Override
687 public void accept(T t) {
688 if (present) {
689 value = op.apply(value, t);
690 }
691 else {
692 value = t;
693 present = true;
694 }
695 }
696 }
697
698 return new CollectorImpl<T, OptionalBox, Optional<T>>(
699 OptionalBox::new, OptionalBox::accept,
700 (a, b) -> { if (b.present) a.accept(b.value); return a; },
701 a -> Optional.ofNullable(a.value), CH_NOID);
702 }
703
704 /**
705 * Returns a {@code Collector} which performs a reduction of its
706 * input elements under a specified mapping function and
707 * {@code BinaryOperator}. This is a generalization of
708 * {@link #reducing(Object, BinaryOperator)} which allows a transformation
709 * of the elements before reduction.
710 *
711 * @apiNote
712 * The {@code reducing()} collectors are most useful when used in a
713 * multi-level reduction, downstream of {@code groupingBy} or
714 * {@code partitioningBy}. To perform a simple map-reduce on a stream,
715 * use {@link Stream#map(Function)} and {@link Stream#reduce(Object, BinaryOperator)}
716 * instead.
717 *
718 * <p>For example, given a stream of {@code Person}, to calculate the longest
719 * last name of residents in each city:
720 * <pre>{@code
721 * Comparator<String> byLength = Comparator.comparing(String::length);
722 * Map<City, String> longestLastNameByCity
723 * = people.stream().collect(groupingBy(Person::getCity,
724 * reducing(Person::getLastName, BinaryOperator.maxBy(byLength))));
725 * }</pre>
726 *
727 * @param <T> the type of the input elements
728 * @param <U> the type of the mapped values
729 * @param identity the identity value for the reduction (also, the value
730 * that is returned when there are no input elements)
731 * @param mapper a mapping function to apply to each input value
732 * @param op a {@code BinaryOperator<U>} used to reduce the mapped values
733 * @return a {@code Collector} implementing the map-reduce operation
734 *
735 * @see #reducing(Object, BinaryOperator)
736 * @see #reducing(BinaryOperator)
737 */
738 public static <T, U>
739 Collector<T, ?, U> reducing(U identity,
740 Function<? super T, ? extends U> mapper,
741 BinaryOperator<U> op) {
742 return new CollectorImpl<>(
743 boxSupplier(identity),
744 (a, t) -> { a[0] = op.apply(a[0], mapper.apply(t)); },
745 (a, b) -> { a[0] = op.apply(a[0], b[0]); return a; },
746 a -> a[0], CH_NOID);
747 }
748
749 /**
750 * Returns a {@code Collector} implementing a "group by" operation on
751 * input elements of type {@code T}, grouping elements according to a
752 * classification function, and returning the results in a {@code Map}.
753 *
754 * <p>The classification function maps elements to some key type {@code K}.
755 * The collector produces a {@code Map<K, List<T>>} whose keys are the
756 * values resulting from applying the classification function to the input
757 * elements, and whose corresponding values are {@code List}s containing the
758 * input elements which map to the associated key under the classification
759 * function.
760 *
761 * <p>There are no guarantees on the type, mutability, serializability, or
762 * thread-safety of the {@code Map} or {@code List} objects returned.
763 * @implSpec
764 * This produces a result similar to:
765 * <pre>{@code
766 * groupingBy(classifier, toList());
767 * }</pre>
768 *
769 * @implNote
770 * The returned {@code Collector} is not concurrent. For parallel stream
771 * pipelines, the {@code combiner} function operates by merging the keys
772 * from one map into another, which can be an expensive operation. If
773 * preservation of the order in which elements appear in the resulting {@code Map}
774 * collector is not required, using {@link #groupingByConcurrent(Function)}
775 * may offer better parallel performance.
776 *
777 * @param <T> the type of the input elements
778 * @param <K> the type of the keys
779 * @param classifier the classifier function mapping input elements to keys
780 * @return a {@code Collector} implementing the group-by operation
781 *
782 * @see #groupingBy(Function, Collector)
783 * @see #groupingBy(Function, Supplier, Collector)
784 * @see #groupingByConcurrent(Function)
785 */
786 public static <T, K> Collector<T, ?, Map<K, List<T>>>
787 groupingBy(Function<? super T, ? extends K> classifier) {
788 return groupingBy(classifier, toList());
789 }
790
791 /**
792 * Returns a {@code Collector} implementing a cascaded "group by" operation
793 * on input elements of type {@code T}, grouping elements according to a
794 * classification function, and then performing a reduction operation on
795 * the values associated with a given key using the specified downstream
796 * {@code Collector}.
797 *
798 * <p>The classification function maps elements to some key type {@code K}.
799 * The downstream collector operates on elements of type {@code T} and
800 * produces a result of type {@code D}. The resulting collector produces a
801 * {@code Map<K, D>}.
802 *
803 * <p>There are no guarantees on the type, mutability,
804 * serializability, or thread-safety of the {@code Map} returned.
805 *
806 * <p>For example, to compute the set of last names of people in each city:
807 * <pre>{@code
808 * Map<City, Set<String>> namesByCity
809 * = people.stream().collect(groupingBy(Person::getCity,
810 * mapping(Person::getLastName, toSet())));
811 * }</pre>
812 *
813 * @implNote
814 * The returned {@code Collector} is not concurrent. For parallel stream
815 * pipelines, the {@code combiner} function operates by merging the keys
816 * from one map into another, which can be an expensive operation. If
817 * preservation of the order in which elements are presented to the downstream
818 * collector is not required, using {@link #groupingByConcurrent(Function, Collector)}
819 * may offer better parallel performance.
820 *
821 * @param <T> the type of the input elements
822 * @param <K> the type of the keys
823 * @param <A> the intermediate accumulation type of the downstream collector
824 * @param <D> the result type of the downstream reduction
825 * @param classifier a classifier function mapping input elements to keys
826 * @param downstream a {@code Collector} implementing the downstream reduction
827 * @return a {@code Collector} implementing the cascaded group-by operation
828 * @see #groupingBy(Function)
829 *
830 * @see #groupingBy(Function, Supplier, Collector)
831 * @see #groupingByConcurrent(Function, Collector)
832 */
833 public static <T, K, A, D>
834 Collector<T, ?, Map<K, D>> groupingBy(Function<? super T, ? extends K> classifier,
835 Collector<? super T, A, D> downstream) {
836 return groupingBy(classifier, HashMap::new, downstream);
837 }
838
839 /**
840 * Returns a {@code Collector} implementing a cascaded "group by" operation
841 * on input elements of type {@code T}, grouping elements according to a
842 * classification function, and then performing a reduction operation on
843 * the values associated with a given key using the specified downstream
844 * {@code Collector}. The {@code Map} produced by the Collector is created
845 * with the supplied factory function.
846 *
847 * <p>The classification function maps elements to some key type {@code K}.
848 * The downstream collector operates on elements of type {@code T} and
849 * produces a result of type {@code D}. The resulting collector produces a
850 * {@code Map<K, D>}.
851 *
852 * <p>For example, to compute the set of last names of people in each city,
853 * where the city names are sorted:
854 * <pre>{@code
855 * Map<City, Set<String>> namesByCity
856 * = people.stream().collect(groupingBy(Person::getCity, TreeMap::new,
857 * mapping(Person::getLastName, toSet())));
858 * }</pre>
859 *
860 * @implNote
861 * The returned {@code Collector} is not concurrent. For parallel stream
862 * pipelines, the {@code combiner} function operates by merging the keys
863 * from one map into another, which can be an expensive operation. If
864 * preservation of the order in which elements are presented to the downstream
865 * collector is not required, using {@link #groupingByConcurrent(Function, Supplier, Collector)}
866 * may offer better parallel performance.
867 *
868 * @param <T> the type of the input elements
869 * @param <K> the type of the keys
870 * @param <A> the intermediate accumulation type of the downstream collector
871 * @param <D> the result type of the downstream reduction
872 * @param <M> the type of the resulting {@code Map}
873 * @param classifier a classifier function mapping input elements to keys
874 * @param downstream a {@code Collector} implementing the downstream reduction
875 * @param mapFactory a function which, when called, produces a new empty
876 * {@code Map} of the desired type
877 * @return a {@code Collector} implementing the cascaded group-by operation
878 *
879 * @see #groupingBy(Function, Collector)
880 * @see #groupingBy(Function)
881 * @see #groupingByConcurrent(Function, Supplier, Collector)
882 */
883 public static <T, K, D, A, M extends Map<K, D>>
884 Collector<T, ?, M> groupingBy(Function<? super T, ? extends K> classifier,
885 Supplier<M> mapFactory,
886 Collector<? super T, A, D> downstream) {
887 Supplier<A> downstreamSupplier = downstream.supplier();
888 BiConsumer<A, ? super T> downstreamAccumulator = downstream.accumulator();
889 BiConsumer<Map<K, A>, T> accumulator = (m, t) -> {
890 K key = Objects.requireNonNull(classifier.apply(t), "element cannot be mapped to a null key");
891 A container = m.computeIfAbsent(key, k -> downstreamSupplier.get());
892 downstreamAccumulator.accept(container, t);
893 };
894 BinaryOperator<Map<K, A>> merger = Collectors.<K, A, Map<K, A>>mapMerger(downstream.combiner());
895 @SuppressWarnings("unchecked")
896 Supplier<Map<K, A>> mangledFactory = (Supplier<Map<K, A>>) mapFactory;
897
898 if (downstream.characteristics().contains(Collector.Characteristics.IDENTITY_FINISH)) {
899 return new CollectorImpl<>(mangledFactory, accumulator, merger, CH_ID);
900 }
901 else {
902 @SuppressWarnings("unchecked")
903 Function<A, A> downstreamFinisher = (Function<A, A>) downstream.finisher();
904 Function<Map<K, A>, M> finisher = intermediate -> {
905 intermediate.replaceAll((k, v) -> downstreamFinisher.apply(v));
906 @SuppressWarnings("unchecked")
907 M castResult = (M) intermediate;
908 return castResult;
909 };
910 return new CollectorImpl<>(mangledFactory, accumulator, merger, finisher, CH_NOID);
911 }
912 }
913
914 /**
915 * Returns a concurrent {@code Collector} implementing a "group by"
916 * operation on input elements of type {@code T}, grouping elements
917 * according to a classification function.
918 *
919 * <p>This is a {@link Collector.Characteristics#CONCURRENT concurrent} and
920 * {@link Collector.Characteristics#UNORDERED unordered} Collector.
921 *
922 * <p>The classification function maps elements to some key type {@code K}.
923 * The collector produces a {@code ConcurrentMap<K, List<T>>} whose keys are the
924 * values resulting from applying the classification function to the input
925 * elements, and whose corresponding values are {@code List}s containing the
926 * input elements which map to the associated key under the classification
927 * function.
928 *
929 * <p>There are no guarantees on the type, mutability, or serializability
930 * of the {@code Map} or {@code List} objects returned, or of the
931 * thread-safety of the {@code List} objects returned.
932 * @implSpec
933 * This produces a result similar to:
934 * <pre>{@code
935 * groupingByConcurrent(classifier, toList());
936 * }</pre>
937 *
938 * @param <T> the type of the input elements
939 * @param <K> the type of the keys
940 * @param classifier a classifier function mapping input elements to keys
941 * @return a concurrent, unordered {@code Collector} implementing the group-by operation
942 *
943 * @see #groupingBy(Function)
944 * @see #groupingByConcurrent(Function, Collector)
945 * @see #groupingByConcurrent(Function, Supplier, Collector)
946 */
947 public static <T, K>
948 Collector<T, ?, ConcurrentMap<K, List<T>>>
949 groupingByConcurrent(Function<? super T, ? extends K> classifier) {
950 return groupingByConcurrent(classifier, ConcurrentHashMap::new, toList());
951 }
952
953 /**
954 * Returns a concurrent {@code Collector} implementing a cascaded "group by"
955 * operation on input elements of type {@code T}, grouping elements
956 * according to a classification function, and then performing a reduction
957 * operation on the values associated with a given key using the specified
958 * downstream {@code Collector}.
959 *
960 * <p>This is a {@link Collector.Characteristics#CONCURRENT concurrent} and
961 * {@link Collector.Characteristics#UNORDERED unordered} Collector.
962 *
963 * <p>The classification function maps elements to some key type {@code K}.
964 * The downstream collector operates on elements of type {@code T} and
965 * produces a result of type {@code D}. The resulting collector produces a
966 * {@code Map<K, D>}.
967 *
968 * <p>For example, to compute the set of last names of people in each city,
969 * where the city names are sorted:
970 * <pre>{@code
971 * ConcurrentMap<City, Set<String>> namesByCity
972 * = people.stream().collect(groupingByConcurrent(Person::getCity,
973 * mapping(Person::getLastName, toSet())));
974 * }</pre>
975 *
976 * @param <T> the type of the input elements
977 * @param <K> the type of the keys
978 * @param <A> the intermediate accumulation type of the downstream collector
979 * @param <D> the result type of the downstream reduction
980 * @param classifier a classifier function mapping input elements to keys
981 * @param downstream a {@code Collector} implementing the downstream reduction
982 * @return a concurrent, unordered {@code Collector} implementing the cascaded group-by operation
983 *
984 * @see #groupingBy(Function, Collector)
985 * @see #groupingByConcurrent(Function)
986 * @see #groupingByConcurrent(Function, Supplier, Collector)
987 */
988 public static <T, K, A, D>
989 Collector<T, ?, ConcurrentMap<K, D>> groupingByConcurrent(Function<? super T, ? extends K> classifier,
990 Collector<? super T, A, D> downstream) {
991 return groupingByConcurrent(classifier, ConcurrentHashMap::new, downstream);
992 }
993
994 /**
995 * Returns a concurrent {@code Collector} implementing a cascaded "group by"
996 * operation on input elements of type {@code T}, grouping elements
997 * according to a classification function, and then performing a reduction
998 * operation on the values associated with a given key using the specified
999 * downstream {@code Collector}. The {@code ConcurrentMap} produced by the
1000 * Collector is created with the supplied factory function.
1001 *
1002 * <p>This is a {@link Collector.Characteristics#CONCURRENT concurrent} and
1003 * {@link Collector.Characteristics#UNORDERED unordered} Collector.
1004 *
1005 * <p>The classification function maps elements to some key type {@code K}.
1006 * The downstream collector operates on elements of type {@code T} and
1007 * produces a result of type {@code D}. The resulting collector produces a
1008 * {@code Map<K, D>}.
1009 *
1010 * <p>For example, to compute the set of last names of people in each city,
1011 * where the city names are sorted:
1012 * <pre>{@code
1013 * ConcurrentMap<City, Set<String>> namesByCity
1014 * = people.stream().collect(groupingBy(Person::getCity, ConcurrentSkipListMap::new,
1015 * mapping(Person::getLastName, toSet())));
1016 * }</pre>
1017 *
1018 *
1019 * @param <T> the type of the input elements
1020 * @param <K> the type of the keys
1021 * @param <A> the intermediate accumulation type of the downstream collector
1022 * @param <D> the result type of the downstream reduction
1023 * @param <M> the type of the resulting {@code ConcurrentMap}
1024 * @param classifier a classifier function mapping input elements to keys
1025 * @param downstream a {@code Collector} implementing the downstream reduction
1026 * @param mapFactory a function which, when called, produces a new empty
1027 * {@code ConcurrentMap} of the desired type
1028 * @return a concurrent, unordered {@code Collector} implementing the cascaded group-by operation
1029 *
1030 * @see #groupingByConcurrent(Function)
1031 * @see #groupingByConcurrent(Function, Collector)
1032 * @see #groupingBy(Function, Supplier, Collector)
1033 */
1034 public static <T, K, A, D, M extends ConcurrentMap<K, D>>
1035 Collector<T, ?, M> groupingByConcurrent(Function<? super T, ? extends K> classifier,
1036 Supplier<M> mapFactory,
1037 Collector<? super T, A, D> downstream) {
1038 Supplier<A> downstreamSupplier = downstream.supplier();
1039 BiConsumer<A, ? super T> downstreamAccumulator = downstream.accumulator();
1040 BinaryOperator<ConcurrentMap<K, A>> merger = Collectors.<K, A, ConcurrentMap<K, A>>mapMerger(downstream.combiner());
1041 @SuppressWarnings("unchecked")
1042 Supplier<ConcurrentMap<K, A>> mangledFactory = (Supplier<ConcurrentMap<K, A>>) mapFactory;
1043 BiConsumer<ConcurrentMap<K, A>, T> accumulator;
1044 if (downstream.characteristics().contains(Collector.Characteristics.CONCURRENT)) {
1045 accumulator = (m, t) -> {
1046 K key = Objects.requireNonNull(classifier.apply(t), "element cannot be mapped to a null key");
1047 A resultContainer = m.computeIfAbsent(key, k -> downstreamSupplier.get());
1048 downstreamAccumulator.accept(resultContainer, t);
1049 };
1050 }
1051 else {
1052 accumulator = (m, t) -> {
1053 K key = Objects.requireNonNull(classifier.apply(t), "element cannot be mapped to a null key");
1054 A resultContainer = m.computeIfAbsent(key, k -> downstreamSupplier.get());
1055 synchronized (resultContainer) {
1056 downstreamAccumulator.accept(resultContainer, t);
1057 }
1058 };
1059 }
1060
1061 if (downstream.characteristics().contains(Collector.Characteristics.IDENTITY_FINISH)) {
1062 return new CollectorImpl<>(mangledFactory, accumulator, merger, CH_CONCURRENT_ID);
1063 }
1064 else {
1065 @SuppressWarnings("unchecked")
1066 Function<A, A> downstreamFinisher = (Function<A, A>) downstream.finisher();
1067 Function<ConcurrentMap<K, A>, M> finisher = intermediate -> {
1068 intermediate.replaceAll((k, v) -> downstreamFinisher.apply(v));
1069 @SuppressWarnings("unchecked")
1070 M castResult = (M) intermediate;
1071 return castResult;
1072 };
1073 return new CollectorImpl<>(mangledFactory, accumulator, merger, finisher, CH_CONCURRENT_NOID);
1074 }
1075 }
1076
1077 /**
1078 * Returns a {@code Collector} which partitions the input elements according
1079 * to a {@code Predicate}, and organizes them into a
1080 * {@code Map<Boolean, List<T>>}.
1081 *
1082 * There are no guarantees on the type, mutability,
1083 * serializability, or thread-safety of the {@code Map} returned.
1084 *
1085 * @param <T> the type of the input elements
1086 * @param predicate a predicate used for classifying input elements
1087 * @return a {@code Collector} implementing the partitioning operation
1088 *
1089 * @see #partitioningBy(Predicate, Collector)
1090 */
1091 public static <T>
1092 Collector<T, ?, Map<Boolean, List<T>>> partitioningBy(Predicate<? super T> predicate) {
1093 return partitioningBy(predicate, toList());
1094 }
1095
1096 /**
1097 * Returns a {@code Collector} which partitions the input elements according
1098 * to a {@code Predicate}, reduces the values in each partition according to
1099 * another {@code Collector}, and organizes them into a
1100 * {@code Map<Boolean, D>} whose values are the result of the downstream
1101 * reduction.
1102 *
1103 * <p>There are no guarantees on the type, mutability,
1104 * serializability, or thread-safety of the {@code Map} returned.
1105 *
1106 * @param <T> the type of the input elements
1107 * @param <A> the intermediate accumulation type of the downstream collector
1108 * @param <D> the result type of the downstream reduction
1109 * @param predicate a predicate used for classifying input elements
1110 * @param downstream a {@code Collector} implementing the downstream
1111 * reduction
1112 * @return a {@code Collector} implementing the cascaded partitioning
1113 * operation
1114 *
1115 * @see #partitioningBy(Predicate)
1116 */
1117 public static <T, D, A>
1118 Collector<T, ?, Map<Boolean, D>> partitioningBy(Predicate<? super T> predicate,
1119 Collector<? super T, A, D> downstream) {
1120 BiConsumer<A, ? super T> downstreamAccumulator = downstream.accumulator();
1121 BiConsumer<Partition<A>, T> accumulator = (result, t) ->
1122 downstreamAccumulator.accept(predicate.test(t) ? result.forTrue : result.forFalse, t);
1123 BinaryOperator<A> op = downstream.combiner();
1124 BinaryOperator<Partition<A>> merger = (left, right) ->
1125 new Partition<>(op.apply(left.forTrue, right.forTrue),
1126 op.apply(left.forFalse, right.forFalse));
1127 Supplier<Partition<A>> supplier = () ->
1128 new Partition<>(downstream.supplier().get(),
1129 downstream.supplier().get());
1130 if (downstream.characteristics().contains(Collector.Characteristics.IDENTITY_FINISH)) {
1131 return new CollectorImpl<>(supplier, accumulator, merger, CH_ID);
1132 }
1133 else {
1134 Function<Partition<A>, Map<Boolean, D>> finisher = par ->
1135 new Partition<>(downstream.finisher().apply(par.forTrue),
1136 downstream.finisher().apply(par.forFalse));
1137 return new CollectorImpl<>(supplier, accumulator, merger, finisher, CH_NOID);
1138 }
1139 }
1140
1141 /**
1142 * Returns a {@code Collector} that accumulates elements into a
1143 * {@code Map} whose keys and values are the result of applying the provided
1144 * mapping functions to the input elements.
1145 *
1146 * <p>If the mapped keys contains duplicates (according to
1147 * {@link Object#equals(Object)}), an {@code IllegalStateException} is
1148 * thrown when the collection operation is performed. If the mapped keys
1149 * may have duplicates, use {@link #toMap(Function, Function, BinaryOperator)}
1150 * instead.
1151 *
1152 * @apiNote
1153 * It is common for either the key or the value to be the input elements.
1154 * In this case, the utility method
1155 * {@link java.util.function.Function#identity()} may be helpful.
1156 * For example, the following produces a {@code Map} mapping
1157 * students to their grade point average:
1158 * <pre>{@code
1159 * Map<Student, Double> studentToGPA
1160 * students.stream().collect(toMap(Functions.identity(),
1161 * student -> computeGPA(student)));
1162 * }</pre>
1163 * And the following produces a {@code Map} mapping a unique identifier to
1164 * students:
1165 * <pre>{@code
1166 * Map<String, Student> studentIdToStudent
1167 * students.stream().collect(toMap(Student::getId,
1168 * Functions.identity());
1169 * }</pre>
1170 *
1171 * @implNote
1172 * The returned {@code Collector} is not concurrent. For parallel stream
1173 * pipelines, the {@code combiner} function operates by merging the keys
1174 * from one map into another, which can be an expensive operation. If it is
1175 * not required that results are inserted into the {@code Map} in encounter
1176 * order, using {@link #toConcurrentMap(Function, Function)}
1177 * may offer better parallel performance.
1178 *
1179 * @param <T> the type of the input elements
1180 * @param <K> the output type of the key mapping function
1181 * @param <U> the output type of the value mapping function
1182 * @param keyMapper a mapping function to produce keys
1183 * @param valueMapper a mapping function to produce values
1184 * @return a {@code Collector} which collects elements into a {@code Map}
1185 * whose keys and values are the result of applying mapping functions to
1186 * the input elements
1187 *
1188 * @see #toMap(Function, Function, BinaryOperator)
1189 * @see #toMap(Function, Function, BinaryOperator, Supplier)
1190 * @see #toConcurrentMap(Function, Function)
1191 */
1192 public static <T, K, U>
1193 Collector<T, ?, Map<K,U>> toMap(Function<? super T, ? extends K> keyMapper,
1194 Function<? super T, ? extends U> valueMapper) {
1195 return toMap(keyMapper, valueMapper, throwingMerger(), HashMap::new);
1196 }
1197
1198 /**
1199 * Returns a {@code Collector} that accumulates elements into a
1200 * {@code Map} whose keys and values are the result of applying the provided
1201 * mapping functions to the input elements.
1202 *
1203 * <p>If the mapped
1204 * keys contains duplicates (according to {@link Object#equals(Object)}),
1205 * the value mapping function is applied to each equal element, and the
1206 * results are merged using the provided merging function.
1207 *
1208 * @apiNote
1209 * There are multiple ways to deal with collisions between multiple elements
1210 * mapping to the same key. The other forms of {@code toMap} simply use
1211 * a merge function that throws unconditionally, but you can easily write
1212 * more flexible merge policies. For example, if you have a stream
1213 * of {@code Person}, and you want to produce a "phone book" mapping name to
1214 * address, but it is possible that two persons have the same name, you can
1215 * do as follows to gracefully deals with these collisions, and produce a
1216 * {@code Map} mapping names to a concatenated list of addresses:
1217 * <pre>{@code
1218 * Map<String, String> phoneBook
1219 * people.stream().collect(toMap(Person::getName,
1220 * Person::getAddress,
1221 * (s, a) -> s + ", " + a));
1222 * }</pre>
1223 *
1224 * @implNote
1225 * The returned {@code Collector} is not concurrent. For parallel stream
1226 * pipelines, the {@code combiner} function operates by merging the keys
1227 * from one map into another, which can be an expensive operation. If it is
1228 * not required that results are merged into the {@code Map} in encounter
1229 * order, using {@link #toConcurrentMap(Function, Function, BinaryOperator)}
1230 * may offer better parallel performance.
1231 *
1232 * @param <T> the type of the input elements
1233 * @param <K> the output type of the key mapping function
1234 * @param <U> the output type of the value mapping function
1235 * @param keyMapper a mapping function to produce keys
1236 * @param valueMapper a mapping function to produce values
1237 * @param mergeFunction a merge function, used to resolve collisions between
1238 * values associated with the same key, as supplied
1239 * to {@link Map#merge(Object, Object, BiFunction)}
1240 * @return a {@code Collector} which collects elements into a {@code Map}
1241 * whose keys are the result of applying a key mapping function to the input
1242 * elements, and whose values are the result of applying a value mapping
1243 * function to all input elements equal to the key and combining them
1244 * using the merge function
1245 *
1246 * @see #toMap(Function, Function)
1247 * @see #toMap(Function, Function, BinaryOperator, Supplier)
1248 * @see #toConcurrentMap(Function, Function, BinaryOperator)
1249 */
1250 public static <T, K, U>
1251 Collector<T, ?, Map<K,U>> toMap(Function<? super T, ? extends K> keyMapper,
1252 Function<? super T, ? extends U> valueMapper,
1253 BinaryOperator<U> mergeFunction) {
1254 return toMap(keyMapper, valueMapper, mergeFunction, HashMap::new);
1255 }
1256
1257 /**
1258 * Returns a {@code Collector} that accumulates elements into a
1259 * {@code Map} whose keys and values are the result of applying the provided
1260 * mapping functions to the input elements.
1261 *
1262 * <p>If the mapped
1263 * keys contains duplicates (according to {@link Object#equals(Object)}),
1264 * the value mapping function is applied to each equal element, and the
1265 * results are merged using the provided merging function. The {@code Map}
1266 * is created by a provided supplier function.
1267 *
1268 * @implNote
1269 * The returned {@code Collector} is not concurrent. For parallel stream
1270 * pipelines, the {@code combiner} function operates by merging the keys
1271 * from one map into another, which can be an expensive operation. If it is
1272 * not required that results are merged into the {@code Map} in encounter
1273 * order, using {@link #toConcurrentMap(Function, Function, BinaryOperator, Supplier)}
1274 * may offer better parallel performance.
1275 *
1276 * @param <T> the type of the input elements
1277 * @param <K> the output type of the key mapping function
1278 * @param <U> the output type of the value mapping function
1279 * @param <M> the type of the resulting {@code Map}
1280 * @param keyMapper a mapping function to produce keys
1281 * @param valueMapper a mapping function to produce values
1282 * @param mergeFunction a merge function, used to resolve collisions between
1283 * values associated with the same key, as supplied
1284 * to {@link Map#merge(Object, Object, BiFunction)}
1285 * @param mapSupplier a function which returns a new, empty {@code Map} into
1286 * which the results will be inserted
1287 * @return a {@code Collector} which collects elements into a {@code Map}
1288 * whose keys are the result of applying a key mapping function to the input
1289 * elements, and whose values are the result of applying a value mapping
1290 * function to all input elements equal to the key and combining them
1291 * using the merge function
1292 *
1293 * @see #toMap(Function, Function)
1294 * @see #toMap(Function, Function, BinaryOperator)
1295 * @see #toConcurrentMap(Function, Function, BinaryOperator, Supplier)
1296 */
1297 public static <T, K, U, M extends Map<K, U>>
1298 Collector<T, ?, M> toMap(Function<? super T, ? extends K> keyMapper,
1299 Function<? super T, ? extends U> valueMapper,
1300 BinaryOperator<U> mergeFunction,
1301 Supplier<M> mapSupplier) {
1302 BiConsumer<M, T> accumulator
1303 = (map, element) -> map.merge(keyMapper.apply(element),
1304 valueMapper.apply(element), mergeFunction);
1305 return new CollectorImpl<>(mapSupplier, accumulator, mapMerger(mergeFunction), CH_ID);
1306 }
1307
1308 /**
1309 * Returns a concurrent {@code Collector} that accumulates elements into a
1310 * {@code ConcurrentMap} whose keys and values are the result of applying
1311 * the provided mapping functions to the input elements.
1312 *
1313 * <p>If the mapped keys contains duplicates (according to
1314 * {@link Object#equals(Object)}), an {@code IllegalStateException} is
1315 * thrown when the collection operation is performed. If the mapped keys
1316 * may have duplicates, use
1317 * {@link #toConcurrentMap(Function, Function, BinaryOperator)} instead.
1318 *
1319 * @apiNote
1320 * It is common for either the key or the value to be the input elements.
1321 * In this case, the utility method
1322 * {@link java.util.function.Function#identity()} may be helpful.
1323 * For example, the following produces a {@code Map} mapping
1324 * students to their grade point average:
1325 * <pre>{@code
1326 * Map<Student, Double> studentToGPA
1327 * students.stream().collect(toMap(Functions.identity(),
1328 * student -> computeGPA(student)));
1329 * }</pre>
1330 * And the following produces a {@code Map} mapping a unique identifier to
1331 * students:
1332 * <pre>{@code
1333 * Map<String, Student> studentIdToStudent
1334 * students.stream().collect(toConcurrentMap(Student::getId,
1335 * Functions.identity());
1336 * }</pre>
1337 *
1338 * <p>This is a {@link Collector.Characteristics#CONCURRENT concurrent} and
1339 * {@link Collector.Characteristics#UNORDERED unordered} Collector.
1340 *
1341 * @param <T> the type of the input elements
1342 * @param <K> the output type of the key mapping function
1343 * @param <U> the output type of the value mapping function
1344 * @param keyMapper the mapping function to produce keys
1345 * @param valueMapper the mapping function to produce values
1346 * @return a concurrent, unordered {@code Collector} which collects elements into a
1347 * {@code ConcurrentMap} whose keys are the result of applying a key mapping
1348 * function to the input elements, and whose values are the result of
1349 * applying a value mapping function to the input elements
1350 *
1351 * @see #toMap(Function, Function)
1352 * @see #toConcurrentMap(Function, Function, BinaryOperator)
1353 * @see #toConcurrentMap(Function, Function, BinaryOperator, Supplier)
1354 */
1355 public static <T, K, U>
1356 Collector<T, ?, ConcurrentMap<K,U>> toConcurrentMap(Function<? super T, ? extends K> keyMapper,
1357 Function<? super T, ? extends U> valueMapper) {
1358 return toConcurrentMap(keyMapper, valueMapper, throwingMerger(), ConcurrentHashMap::new);
1359 }
1360
1361 /**
1362 * Returns a concurrent {@code Collector} that accumulates elements into a
1363 * {@code ConcurrentMap} whose keys and values are the result of applying
1364 * the provided mapping functions to the input elements.
1365 *
1366 * <p>If the mapped keys contains duplicates (according to {@link Object#equals(Object)}),
1367 * the value mapping function is applied to each equal element, and the
1368 * results are merged using the provided merging function.
1369 *
1370 * @apiNote
1371 * There are multiple ways to deal with collisions between multiple elements
1372 * mapping to the same key. The other forms of {@code toConcurrentMap} simply use
1373 * a merge function that throws unconditionally, but you can easily write
1374 * more flexible merge policies. For example, if you have a stream
1375 * of {@code Person}, and you want to produce a "phone book" mapping name to
1376 * address, but it is possible that two persons have the same name, you can
1377 * do as follows to gracefully deals with these collisions, and produce a
1378 * {@code Map} mapping names to a concatenated list of addresses:
1379 * <pre>{@code
1380 * Map<String, String> phoneBook
1381 * people.stream().collect(toConcurrentMap(Person::getName,
1382 * Person::getAddress,
1383 * (s, a) -> s + ", " + a));
1384 * }</pre>
1385 *
1386 * <p>This is a {@link Collector.Characteristics#CONCURRENT concurrent} and
1387 * {@link Collector.Characteristics#UNORDERED unordered} Collector.
1388 *
1389 * @param <T> the type of the input elements
1390 * @param <K> the output type of the key mapping function
1391 * @param <U> the output type of the value mapping function
1392 * @param keyMapper a mapping function to produce keys
1393 * @param valueMapper a mapping function to produce values
1394 * @param mergeFunction a merge function, used to resolve collisions between
1395 * values associated with the same key, as supplied
1396 * to {@link Map#merge(Object, Object, BiFunction)}
1397 * @return a concurrent, unordered {@code Collector} which collects elements into a
1398 * {@code ConcurrentMap} whose keys are the result of applying a key mapping
1399 * function to the input elements, and whose values are the result of
1400 * applying a value mapping function to all input elements equal to the key
1401 * and combining them using the merge function
1402 *
1403 * @see #toConcurrentMap(Function, Function)
1404 * @see #toConcurrentMap(Function, Function, BinaryOperator, Supplier)
1405 * @see #toMap(Function, Function, BinaryOperator)
1406 */
1407 public static <T, K, U>
1408 Collector<T, ?, ConcurrentMap<K,U>>
1409 toConcurrentMap(Function<? super T, ? extends K> keyMapper,
1410 Function<? super T, ? extends U> valueMapper,
1411 BinaryOperator<U> mergeFunction) {
1412 return toConcurrentMap(keyMapper, valueMapper, mergeFunction, ConcurrentHashMap::new);
1413 }
1414
1415 /**
1416 * Returns a concurrent {@code Collector} that accumulates elements into a
1417 * {@code ConcurrentMap} whose keys and values are the result of applying
1418 * the provided mapping functions to the input elements.
1419 *
1420 * <p>If the mapped keys contains duplicates (according to {@link Object#equals(Object)}),
1421 * the value mapping function is applied to each equal element, and the
1422 * results are merged using the provided merging function. The
1423 * {@code ConcurrentMap} is created by a provided supplier function.
1424 *
1425 * <p>This is a {@link Collector.Characteristics#CONCURRENT concurrent} and
1426 * {@link Collector.Characteristics#UNORDERED unordered} Collector.
1427 *
1428 * @param <T> the type of the input elements
1429 * @param <K> the output type of the key mapping function
1430 * @param <U> the output type of the value mapping function
1431 * @param <M> the type of the resulting {@code ConcurrentMap}
1432 * @param keyMapper a mapping function to produce keys
1433 * @param valueMapper a mapping function to produce values
1434 * @param mergeFunction a merge function, used to resolve collisions between
1435 * values associated with the same key, as supplied
1436 * to {@link Map#merge(Object, Object, BiFunction)}
1437 * @param mapSupplier a function which returns a new, empty {@code Map} into
1438 * which the results will be inserted
1439 * @return a concurrent, unordered {@code Collector} which collects elements into a
1440 * {@code ConcurrentMap} whose keys are the result of applying a key mapping
1441 * function to the input elements, and whose values are the result of
1442 * applying a value mapping function to all input elements equal to the key
1443 * and combining them using the merge function
1444 *
1445 * @see #toConcurrentMap(Function, Function)
1446 * @see #toConcurrentMap(Function, Function, BinaryOperator)
1447 * @see #toMap(Function, Function, BinaryOperator, Supplier)
1448 */
1449 public static <T, K, U, M extends ConcurrentMap<K, U>>
1450 Collector<T, ?, M> toConcurrentMap(Function<? super T, ? extends K> keyMapper,
1451 Function<? super T, ? extends U> valueMapper,
1452 BinaryOperator<U> mergeFunction,
1453 Supplier<M> mapSupplier) {
1454 BiConsumer<M, T> accumulator
1455 = (map, element) -> map.merge(keyMapper.apply(element),
1456 valueMapper.apply(element), mergeFunction);
1457 return new CollectorImpl<>(mapSupplier, accumulator, mapMerger(mergeFunction), CH_CONCURRENT_ID);
1458 }
1459
1460 /**
1461 * Returns a {@code Collector} which applies an {@code int}-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 a mapping function to apply to each element
1467 * @return a {@code Collector} implementing the summary-statistics reduction
1468 *
1469 * @see #summarizingDouble(ToDoubleFunction)
1470 * @see #summarizingLong(ToLongFunction)
1471 */
1472 public static <T>
1473 Collector<T, ?, IntSummaryStatistics> summarizingInt(ToIntFunction<? super T> mapper) {
1474 return new CollectorImpl<T, IntSummaryStatistics, IntSummaryStatistics>(
1475 IntSummaryStatistics::new,
1476 (r, t) -> r.accept(mapper.applyAsInt(t)),
1477 (l, r) -> { l.combine(r); return l; }, CH_ID);
1478 }
1479
1480 /**
1481 * Returns a {@code Collector} which applies an {@code long}-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 the mapping function to apply to each element
1487 * @return a {@code Collector} implementing the summary-statistics reduction
1488 *
1489 * @see #summarizingDouble(ToDoubleFunction)
1490 * @see #summarizingInt(ToIntFunction)
1491 */
1492 public static <T>
1493 Collector<T, ?, LongSummaryStatistics> summarizingLong(ToLongFunction<? super T> mapper) {
1494 return new CollectorImpl<T, LongSummaryStatistics, LongSummaryStatistics>(
1495 LongSummaryStatistics::new,
1496 (r, t) -> r.accept(mapper.applyAsLong(t)),
1497 (l, r) -> { l.combine(r); return l; }, CH_ID);
1498 }
1499
1500 /**
1501 * Returns a {@code Collector} which applies an {@code double}-producing
1502 * mapping function to each input element, and returns summary statistics
1503 * for the resulting values.
1504 *
1505 * @param <T> the type of the input elements
1506 * @param mapper a mapping function to apply to each element
1507 * @return a {@code Collector} implementing the summary-statistics reduction
1508 *
1509 * @see #summarizingLong(ToLongFunction)
1510 * @see #summarizingInt(ToIntFunction)
1511 */
1512 public static <T>
1513 Collector<T, ?, DoubleSummaryStatistics> summarizingDouble(ToDoubleFunction<? super T> mapper) {
1514 return new CollectorImpl<T, DoubleSummaryStatistics, DoubleSummaryStatistics>(
1515 DoubleSummaryStatistics::new,
1516 (r, t) -> r.accept(mapper.applyAsDouble(t)),
1517 (l, r) -> { l.combine(r); return l; }, CH_ID);
1518 }
1519
1520 /**
1521 * Implementation class used by partitioningBy.
1522 */
1523 private static final class Partition<T>
1524 extends AbstractMap<Boolean, T>
1525 implements Map<Boolean, T> {
1526 final T forTrue;
1527 final T forFalse;
1528
1529 Partition(T forTrue, T forFalse) {
1530 this.forTrue = forTrue;
1531 this.forFalse = forFalse;
1532 }
1533
1534 @Override
1535 public Set<Map.Entry<Boolean, T>> entrySet() {
1536 return new AbstractSet<Map.Entry<Boolean, T>>() {
1537 @Override
1538 public Iterator<Map.Entry<Boolean, T>> iterator() {
1539 Map.Entry<Boolean, T> falseEntry = new SimpleImmutableEntry<>(false, forFalse);
1540 Map.Entry<Boolean, T> trueEntry = new SimpleImmutableEntry<>(true, forTrue);
1541 return Arrays.asList(falseEntry, trueEntry).iterator();
1542 }
1543
1544 @Override
1545 public int size() {
1546 return 2;
1547 }
1548 };
1549 }
1550 }
1551 }