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[1], 510 (a, t) -> { a[0] += mapper.applyAsDouble(t); }, 511 (a, b) -> { a[0] += b[0]; return a; }, 512 a -> a[0], CH_NOID); 513 } 514 515 /** 516 * Returns a {@code Collector} that produces the arithmetic mean of an integer-valued 517 * function applied to the input elements. If no elements are present, 518 * the result is 0. 519 * 520 * @param <T> the type of the input elements 521 * @param mapper a function extracting the property to be summed 522 * @return a {@code Collector} that produces the sum of a derived property 523 */ 524 public static <T> Collector<T, ?, Double> 525 averagingInt(ToIntFunction<? super T> mapper) { 526 return new CollectorImpl<>( 527 () -> new long[2], 528 (a, t) -> { a[0] += mapper.applyAsInt(t); a[1]++; }, 529 (a, b) -> { a[0] += b[0]; a[1] += b[1]; return a; }, 530 a -> (a[1] == 0) ? 0.0d : (double) a[0] / a[1], CH_NOID); 531 } 532 533 /** 534 * Returns a {@code Collector} that produces the arithmetic mean of a long-valued 535 * function applied to the input elements. If no elements are present, 536 * the result is 0. 537 * 538 * @param <T> the type of the input elements 539 * @param mapper a function extracting the property to be summed 540 * @return a {@code Collector} that produces the sum of a derived property 541 */ 542 public static <T> Collector<T, ?, Double> 543 averagingLong(ToLongFunction<? super T> mapper) { 544 return new CollectorImpl<>( 545 () -> new long[2], 546 (a, t) -> { a[0] += mapper.applyAsLong(t); a[1]++; }, 547 (a, b) -> { a[0] += b[0]; a[1] += b[1]; return a; }, 548 a -> (a[1] == 0) ? 0.0d : (double) a[0] / a[1], CH_NOID); 549 } 550 551 /** 552 * Returns a {@code Collector} that produces the arithmetic mean of a double-valued 553 * function applied to the input elements. If no elements are present, 554 * the result is 0. 555 * 556 * <p>The average returned can vary depending upon the order in which 557 * values are recorded, due to accumulated rounding error in 558 * addition of values of differing magnitudes. Values sorted by increasing 559 * absolute magnitude tend to yield more accurate results. If any recorded 560 * value is a {@code NaN} or the sum is at any point a {@code NaN} then the 561 * average will be {@code NaN}. 562 * 563 * @param <T> the type of the input elements 564 * @param mapper a function extracting the property to be summed 565 * @return a {@code Collector} that produces the sum of a derived property 566 */ 567 public static <T> Collector<T, ?, Double> 568 averagingDouble(ToDoubleFunction<? super T> mapper) { 569 return new CollectorImpl<>( 570 () -> new double[2], 571 (a, t) -> { a[0] += mapper.applyAsDouble(t); a[1]++; }, 572 (a, b) -> { a[0] += b[0]; a[1] += b[1]; return a; }, 573 a -> (a[1] == 0) ? 0.0d : a[0] / a[1], CH_NOID); 574 } 575 576 /** 577 * Returns a {@code Collector} which performs a reduction of its 578 * input elements under a specified {@code BinaryOperator} using the 579 * provided identity. 580 * 581 * @apiNote 582 * The {@code reducing()} collectors are most useful when used in a 583 * multi-level reduction, downstream of {@code groupingBy} or 584 * {@code partitioningBy}. To perform a simple reduction on a stream, 585 * use {@link Stream#reduce(Object, BinaryOperator)}} instead. 586 * 587 * @param <T> element type for the input and output of the reduction 588 * @param identity the identity value for the reduction (also, the value 589 * that is returned when there are no input elements) 590 * @param op a {@code BinaryOperator<T>} used to reduce the input elements 591 * @return a {@code Collector} which implements the reduction operation 592 * 593 * @see #reducing(BinaryOperator) 594 * @see #reducing(Object, Function, BinaryOperator) 595 */ 596 public static <T> Collector<T, ?, T> 597 reducing(T identity, BinaryOperator<T> op) { 598 return new CollectorImpl<>( 599 boxSupplier(identity), 600 (a, t) -> { a[0] = op.apply(a[0], t); }, 601 (a, b) -> { a[0] = op.apply(a[0], b[0]); return a; }, 602 a -> a[0], 603 CH_NOID); 604 } 605 606 @SuppressWarnings("unchecked") 607 private static <T> Supplier<T[]> boxSupplier(T identity) { 608 return () -> (T[]) new Object[] { identity }; 609 } 610 611 /** 612 * Returns a {@code Collector} which performs a reduction of its 613 * input elements under a specified {@code BinaryOperator}. The result 614 * is described as an {@code Optional<T>}. 615 * 616 * @apiNote 617 * The {@code reducing()} collectors are most useful when used in a 618 * multi-level reduction, downstream of {@code groupingBy} or 619 * {@code partitioningBy}. To perform a simple reduction on a stream, 620 * use {@link Stream#reduce(BinaryOperator)} instead. 621 * 622 * <p>For example, given a stream of {@code Person}, to calculate tallest 623 * person in each city: 624 * <pre>{@code 625 * Comparator<Person> byHeight = Comparator.comparing(Person::getHeight); 626 * Map<City, Person> tallestByCity 627 * = people.stream().collect(groupingBy(Person::getCity, reducing(BinaryOperator.maxBy(byHeight)))); 628 * }</pre> 629 * 630 * @param <T> element type for the input and output of the reduction 631 * @param op a {@code BinaryOperator<T>} used to reduce the input elements 632 * @return a {@code Collector} which implements the reduction operation 633 * 634 * @see #reducing(Object, BinaryOperator) 635 * @see #reducing(Object, Function, BinaryOperator) 636 */ 637 public static <T> Collector<T, ?, Optional<T>> 638 reducing(BinaryOperator<T> op) { 639 class OptionalBox implements Consumer<T> { 640 T value = null; 641 boolean present = false; 642 643 @Override 644 public void accept(T t) { 645 if (present) { 646 value = op.apply(value, t); 647 } 648 else { 649 value = t; 650 present = true; 651 } 652 } 653 } 654 655 return new CollectorImpl<T, OptionalBox, Optional<T>>( 656 OptionalBox::new, OptionalBox::accept, 657 (a, b) -> { if (b.present) a.accept(b.value); return a; }, 658 a -> Optional.ofNullable(a.value), CH_NOID); 659 } 660 661 /** 662 * Returns a {@code Collector} which performs a reduction of its 663 * input elements under a specified mapping function and 664 * {@code BinaryOperator}. This is a generalization of 665 * {@link #reducing(Object, BinaryOperator)} which allows a transformation 666 * of the elements before reduction. 667 * 668 * @apiNote 669 * The {@code reducing()} collectors are most useful when used in a 670 * multi-level reduction, downstream of {@code groupingBy} or 671 * {@code partitioningBy}. To perform a simple map-reduce on a stream, 672 * use {@link Stream#map(Function)} and {@link Stream#reduce(Object, BinaryOperator)} 673 * instead. 674 * 675 * <p>For example, given a stream of {@code Person}, to calculate the longest 676 * last name of residents in each city: 677 * <pre>{@code 678 * Comparator<String> byLength = Comparator.comparing(String::length); 679 * Map<City, String> longestLastNameByCity 680 * = people.stream().collect(groupingBy(Person::getCity, 681 * reducing(Person::getLastName, BinaryOperator.maxBy(byLength)))); 682 * }</pre> 683 * 684 * @param <T> the type of the input elements 685 * @param <U> the type of the mapped values 686 * @param identity the identity value for the reduction (also, the value 687 * that is returned when there are no input elements) 688 * @param mapper a mapping function to apply to each input value 689 * @param op a {@code BinaryOperator<U>} used to reduce the mapped values 690 * @return a {@code Collector} implementing the map-reduce operation 691 * 692 * @see #reducing(Object, BinaryOperator) 693 * @see #reducing(BinaryOperator) 694 */ 695 public static <T, U> 696 Collector<T, ?, U> reducing(U identity, 697 Function<? super T, ? extends U> mapper, 698 BinaryOperator<U> op) { 699 return new CollectorImpl<>( 700 boxSupplier(identity), 701 (a, t) -> { a[0] = op.apply(a[0], mapper.apply(t)); }, 702 (a, b) -> { a[0] = op.apply(a[0], b[0]); return a; }, 703 a -> a[0], CH_NOID); 704 } 705 706 /** 707 * Returns a {@code Collector} implementing a "group by" operation on 708 * input elements of type {@code T}, grouping elements according to a 709 * classification function, and returning the results in a {@code Map}. 710 * 711 * <p>The classification function maps elements to some key type {@code K}. 712 * The collector produces a {@code Map<K, List<T>>} whose keys are the 713 * values resulting from applying the classification function to the input 714 * elements, and whose corresponding values are {@code List}s containing the 715 * input elements which map to the associated key under the classification 716 * function. 717 * 718 * <p>There are no guarantees on the type, mutability, serializability, or 719 * thread-safety of the {@code Map} or {@code List} objects returned. 720 * @implSpec 721 * This produces a result similar to: 722 * <pre>{@code 723 * groupingBy(classifier, toList()); 724 * }</pre> 725 * 726 * @implNote 727 * The returned {@code Collector} is not concurrent. For parallel stream 728 * pipelines, the {@code combiner} function operates by merging the keys 729 * from one map into another, which can be an expensive operation. If 730 * preservation of the order in which elements appear in the resulting {@code Map} 731 * collector is not required, using {@link #groupingByConcurrent(Function)} 732 * may offer better parallel performance. 733 * 734 * @param <T> the type of the input elements 735 * @param <K> the type of the keys 736 * @param classifier the classifier function mapping input elements to keys 737 * @return a {@code Collector} implementing the group-by operation 738 * 739 * @see #groupingBy(Function, Collector) 740 * @see #groupingBy(Function, Supplier, Collector) 741 * @see #groupingByConcurrent(Function) 742 */ 743 public static <T, K> Collector<T, ?, Map<K, List<T>>> 744 groupingBy(Function<? super T, ? extends K> classifier) { 745 return groupingBy(classifier, toList()); 746 } 747 748 /** 749 * Returns a {@code Collector} implementing a cascaded "group by" operation 750 * on input elements of type {@code T}, grouping elements according to a 751 * classification function, and then performing a reduction operation on 752 * the values associated with a given key using the specified downstream 753 * {@code Collector}. 754 * 755 * <p>The classification function maps elements to some key type {@code K}. 756 * The downstream collector operates on elements of type {@code T} and 757 * produces a result of type {@code D}. The resulting collector produces a 758 * {@code Map<K, D>}. 759 * 760 * <p>There are no guarantees on the type, mutability, 761 * serializability, or thread-safety of the {@code Map} returned. 762 * 763 * <p>For example, to compute the set of last names of people in each city: 764 * <pre>{@code 765 * Map<City, Set<String>> namesByCity 766 * = people.stream().collect(groupingBy(Person::getCity, 767 * mapping(Person::getLastName, toSet()))); 768 * }</pre> 769 * 770 * @implNote 771 * The returned {@code Collector} is not concurrent. For parallel stream 772 * pipelines, the {@code combiner} function operates by merging the keys 773 * from one map into another, which can be an expensive operation. If 774 * preservation of the order in which elements are presented to the downstream 775 * collector is not required, using {@link #groupingByConcurrent(Function, Collector)} 776 * may offer better parallel performance. 777 * 778 * @param <T> the type of the input elements 779 * @param <K> the type of the keys 780 * @param <A> the intermediate accumulation type of the downstream collector 781 * @param <D> the result type of the downstream reduction 782 * @param classifier a classifier function mapping input elements to keys 783 * @param downstream a {@code Collector} implementing the downstream reduction 784 * @return a {@code Collector} implementing the cascaded group-by operation 785 * @see #groupingBy(Function) 786 * 787 * @see #groupingBy(Function, Supplier, Collector) 788 * @see #groupingByConcurrent(Function, Collector) 789 */ 790 public static <T, K, A, D> 791 Collector<T, ?, Map<K, D>> groupingBy(Function<? super T, ? extends K> classifier, 792 Collector<? super T, A, D> downstream) { 793 return groupingBy(classifier, HashMap::new, downstream); 794 } 795 796 /** 797 * Returns a {@code Collector} implementing a cascaded "group by" operation 798 * on input elements of type {@code T}, grouping elements according to a 799 * classification function, and then performing a reduction operation on 800 * the values associated with a given key using the specified downstream 801 * {@code Collector}. The {@code Map} produced by the Collector is created 802 * with the supplied factory function. 803 * 804 * <p>The classification function maps elements to some key type {@code K}. 805 * The downstream collector operates on elements of type {@code T} and 806 * produces a result of type {@code D}. The resulting collector produces a 807 * {@code Map<K, D>}. 808 * 809 * <p>For example, to compute the set of last names of people in each city, 810 * where the city names are sorted: 811 * <pre>{@code 812 * Map<City, Set<String>> namesByCity 813 * = people.stream().collect(groupingBy(Person::getCity, TreeMap::new, 814 * mapping(Person::getLastName, toSet()))); 815 * }</pre> 816 * 817 * @implNote 818 * The returned {@code Collector} is not concurrent. For parallel stream 819 * pipelines, the {@code combiner} function operates by merging the keys 820 * from one map into another, which can be an expensive operation. If 821 * preservation of the order in which elements are presented to the downstream 822 * collector is not required, using {@link #groupingByConcurrent(Function, Supplier, Collector)} 823 * may offer better parallel performance. 824 * 825 * @param <T> the type of the input elements 826 * @param <K> the type of the keys 827 * @param <A> the intermediate accumulation type of the downstream collector 828 * @param <D> the result type of the downstream reduction 829 * @param <M> the type of the resulting {@code Map} 830 * @param classifier a classifier function mapping input elements to keys 831 * @param downstream a {@code Collector} implementing the downstream reduction 832 * @param mapFactory a function which, when called, produces a new empty 833 * {@code Map} of the desired type 834 * @return a {@code Collector} implementing the cascaded group-by operation 835 * 836 * @see #groupingBy(Function, Collector) 837 * @see #groupingBy(Function) 838 * @see #groupingByConcurrent(Function, Supplier, Collector) 839 */ 840 public static <T, K, D, A, M extends Map<K, D>> 841 Collector<T, ?, M> groupingBy(Function<? super T, ? extends K> classifier, 842 Supplier<M> mapFactory, 843 Collector<? super T, A, D> downstream) { 844 Supplier<A> downstreamSupplier = downstream.supplier(); 845 BiConsumer<A, ? super T> downstreamAccumulator = downstream.accumulator(); 846 BiConsumer<Map<K, A>, T> accumulator = (m, t) -> { 847 K key = Objects.requireNonNull(classifier.apply(t), "element cannot be mapped to a null key"); 848 A container = m.computeIfAbsent(key, k -> downstreamSupplier.get()); 849 downstreamAccumulator.accept(container, t); 850 }; 851 BinaryOperator<Map<K, A>> merger = Collectors.<K, A, Map<K, A>>mapMerger(downstream.combiner()); 852 @SuppressWarnings("unchecked") 853 Supplier<Map<K, A>> mangledFactory = (Supplier<Map<K, A>>) mapFactory; 854 855 if (downstream.characteristics().contains(Collector.Characteristics.IDENTITY_FINISH)) { 856 return new CollectorImpl<>(mangledFactory, accumulator, merger, CH_ID); 857 } 858 else { 859 @SuppressWarnings("unchecked") 860 Function<A, A> downstreamFinisher = (Function<A, A>) downstream.finisher(); 861 Function<Map<K, A>, M> finisher = intermediate -> { 862 intermediate.replaceAll((k, v) -> downstreamFinisher.apply(v)); 863 @SuppressWarnings("unchecked") 864 M castResult = (M) intermediate; 865 return castResult; 866 }; 867 return new CollectorImpl<>(mangledFactory, accumulator, merger, finisher, CH_NOID); 868 } 869 } 870 871 /** 872 * Returns a concurrent {@code Collector} implementing a "group by" 873 * operation on input elements of type {@code T}, grouping elements 874 * according to a classification function. 875 * 876 * <p>This is a {@link Collector.Characteristics#CONCURRENT concurrent} and 877 * {@link Collector.Characteristics#UNORDERED unordered} Collector. 878 * 879 * <p>The classification function maps elements to some key type {@code K}. 880 * The collector produces a {@code ConcurrentMap<K, List<T>>} whose keys are the 881 * values resulting from applying the classification function to the input 882 * elements, and whose corresponding values are {@code List}s containing the 883 * input elements which map to the associated key under the classification 884 * function. 885 * 886 * <p>There are no guarantees on the type, mutability, or serializability 887 * of the {@code Map} or {@code List} objects returned, or of the 888 * thread-safety of the {@code List} objects returned. 889 * @implSpec 890 * This produces a result similar to: 891 * <pre>{@code 892 * groupingByConcurrent(classifier, toList()); 893 * }</pre> 894 * 895 * @param <T> the type of the input elements 896 * @param <K> the type of the keys 897 * @param classifier a classifier function mapping input elements to keys 898 * @return a concurrent, unordered {@code Collector} implementing the group-by operation 899 * 900 * @see #groupingBy(Function) 901 * @see #groupingByConcurrent(Function, Collector) 902 * @see #groupingByConcurrent(Function, Supplier, Collector) 903 */ 904 public static <T, K> 905 Collector<T, ?, ConcurrentMap<K, List<T>>> 906 groupingByConcurrent(Function<? super T, ? extends K> classifier) { 907 return groupingByConcurrent(classifier, ConcurrentHashMap::new, toList()); 908 } 909 910 /** 911 * Returns a concurrent {@code Collector} implementing a cascaded "group by" 912 * operation on input elements of type {@code T}, grouping elements 913 * according to a classification function, and then performing a reduction 914 * operation on the values associated with a given key using the specified 915 * downstream {@code Collector}. 916 * 917 * <p>This is a {@link Collector.Characteristics#CONCURRENT concurrent} and 918 * {@link Collector.Characteristics#UNORDERED unordered} Collector. 919 * 920 * <p>The classification function maps elements to some key type {@code K}. 921 * The downstream collector operates on elements of type {@code T} and 922 * produces a result of type {@code D}. The resulting collector produces a 923 * {@code Map<K, D>}. 924 * 925 * <p>For example, to compute the set of last names of people in each city, 926 * where the city names are sorted: 927 * <pre>{@code 928 * ConcurrentMap<City, Set<String>> namesByCity 929 * = people.stream().collect(groupingByConcurrent(Person::getCity, 930 * mapping(Person::getLastName, toSet()))); 931 * }</pre> 932 * 933 * @param <T> the type of the input elements 934 * @param <K> the type of the keys 935 * @param <A> the intermediate accumulation type of the downstream collector 936 * @param <D> the result type of the downstream reduction 937 * @param classifier a classifier function mapping input elements to keys 938 * @param downstream a {@code Collector} implementing the downstream reduction 939 * @return a concurrent, unordered {@code Collector} implementing the cascaded group-by operation 940 * 941 * @see #groupingBy(Function, Collector) 942 * @see #groupingByConcurrent(Function) 943 * @see #groupingByConcurrent(Function, Supplier, Collector) 944 */ 945 public static <T, K, A, D> 946 Collector<T, ?, ConcurrentMap<K, D>> groupingByConcurrent(Function<? super T, ? extends K> classifier, 947 Collector<? super T, A, D> downstream) { 948 return groupingByConcurrent(classifier, ConcurrentHashMap::new, downstream); 949 } 950 951 /** 952 * Returns a concurrent {@code Collector} implementing a cascaded "group by" 953 * operation on input elements of type {@code T}, grouping elements 954 * according to a classification function, and then performing a reduction 955 * operation on the values associated with a given key using the specified 956 * downstream {@code Collector}. The {@code ConcurrentMap} produced by the 957 * Collector is created with the supplied factory function. 958 * 959 * <p>This is a {@link Collector.Characteristics#CONCURRENT concurrent} and 960 * {@link Collector.Characteristics#UNORDERED unordered} Collector. 961 * 962 * <p>The classification function maps elements to some key type {@code K}. 963 * The downstream collector operates on elements of type {@code T} and 964 * produces a result of type {@code D}. The resulting collector produces a 965 * {@code Map<K, D>}. 966 * 967 * <p>For example, to compute the set of last names of people in each city, 968 * where the city names are sorted: 969 * <pre>{@code 970 * ConcurrentMap<City, Set<String>> namesByCity 971 * = people.stream().collect(groupingBy(Person::getCity, ConcurrentSkipListMap::new, 972 * mapping(Person::getLastName, toSet()))); 973 * }</pre> 974 * 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 <M> the type of the resulting {@code ConcurrentMap} 981 * @param classifier a classifier function mapping input elements to keys 982 * @param downstream a {@code Collector} implementing the downstream reduction 983 * @param mapFactory a function which, when called, produces a new empty 984 * {@code ConcurrentMap} of the desired type 985 * @return a concurrent, unordered {@code Collector} implementing the cascaded group-by operation 986 * 987 * @see #groupingByConcurrent(Function) 988 * @see #groupingByConcurrent(Function, Collector) 989 * @see #groupingBy(Function, Supplier, Collector) 990 */ 991 public static <T, K, A, D, M extends ConcurrentMap<K, D>> 992 Collector<T, ?, M> groupingByConcurrent(Function<? super T, ? extends K> classifier, 993 Supplier<M> mapFactory, 994 Collector<? super T, A, D> downstream) { 995 Supplier<A> downstreamSupplier = downstream.supplier(); 996 BiConsumer<A, ? super T> downstreamAccumulator = downstream.accumulator(); 997 BinaryOperator<ConcurrentMap<K, A>> merger = Collectors.<K, A, ConcurrentMap<K, A>>mapMerger(downstream.combiner()); 998 @SuppressWarnings("unchecked") 999 Supplier<ConcurrentMap<K, A>> mangledFactory = (Supplier<ConcurrentMap<K, A>>) mapFactory; 1000 BiConsumer<ConcurrentMap<K, A>, T> accumulator; 1001 if (downstream.characteristics().contains(Collector.Characteristics.CONCURRENT)) { 1002 accumulator = (m, t) -> { 1003 K key = Objects.requireNonNull(classifier.apply(t), "element cannot be mapped to a null key"); 1004 A resultContainer = m.computeIfAbsent(key, k -> downstreamSupplier.get()); 1005 downstreamAccumulator.accept(resultContainer, t); 1006 }; 1007 } 1008 else { 1009 accumulator = (m, t) -> { 1010 K key = Objects.requireNonNull(classifier.apply(t), "element cannot be mapped to a null key"); 1011 A resultContainer = m.computeIfAbsent(key, k -> downstreamSupplier.get()); 1012 synchronized (resultContainer) { 1013 downstreamAccumulator.accept(resultContainer, t); 1014 } 1015 }; 1016 } 1017 1018 if (downstream.characteristics().contains(Collector.Characteristics.IDENTITY_FINISH)) { 1019 return new CollectorImpl<>(mangledFactory, accumulator, merger, CH_CONCURRENT_ID); 1020 } 1021 else { 1022 @SuppressWarnings("unchecked") 1023 Function<A, A> downstreamFinisher = (Function<A, A>) downstream.finisher(); 1024 Function<ConcurrentMap<K, A>, M> finisher = intermediate -> { 1025 intermediate.replaceAll((k, v) -> downstreamFinisher.apply(v)); 1026 @SuppressWarnings("unchecked") 1027 M castResult = (M) intermediate; 1028 return castResult; 1029 }; 1030 return new CollectorImpl<>(mangledFactory, accumulator, merger, finisher, CH_CONCURRENT_NOID); 1031 } 1032 } 1033 1034 /** 1035 * Returns a {@code Collector} which partitions the input elements according 1036 * to a {@code Predicate}, and organizes them into a 1037 * {@code Map<Boolean, List<T>>}. 1038 * 1039 * There are no guarantees on the type, mutability, 1040 * serializability, or thread-safety of the {@code Map} returned. 1041 * 1042 * @param <T> the type of the input elements 1043 * @param predicate a predicate used for classifying input elements 1044 * @return a {@code Collector} implementing the partitioning operation 1045 * 1046 * @see #partitioningBy(Predicate, Collector) 1047 */ 1048 public static <T> 1049 Collector<T, ?, Map<Boolean, List<T>>> partitioningBy(Predicate<? super T> predicate) { 1050 return partitioningBy(predicate, toList()); 1051 } 1052 1053 /** 1054 * Returns a {@code Collector} which partitions the input elements according 1055 * to a {@code Predicate}, reduces the values in each partition according to 1056 * another {@code Collector}, and organizes them into a 1057 * {@code Map<Boolean, D>} whose values are the result of the downstream 1058 * reduction. 1059 * 1060 * <p>There are no guarantees on the type, mutability, 1061 * serializability, or thread-safety of the {@code Map} returned. 1062 * 1063 * @param <T> the type of the input elements 1064 * @param <A> the intermediate accumulation type of the downstream collector 1065 * @param <D> the result type of the downstream reduction 1066 * @param predicate a predicate used for classifying input elements 1067 * @param downstream a {@code Collector} implementing the downstream 1068 * reduction 1069 * @return a {@code Collector} implementing the cascaded partitioning 1070 * operation 1071 * 1072 * @see #partitioningBy(Predicate) 1073 */ 1074 public static <T, D, A> 1075 Collector<T, ?, Map<Boolean, D>> partitioningBy(Predicate<? super T> predicate, 1076 Collector<? super T, A, D> downstream) { 1077 BiConsumer<A, ? super T> downstreamAccumulator = downstream.accumulator(); 1078 BiConsumer<Partition<A>, T> accumulator = (result, t) -> 1079 downstreamAccumulator.accept(predicate.test(t) ? result.forTrue : result.forFalse, t); 1080 BinaryOperator<A> op = downstream.combiner(); 1081 BinaryOperator<Partition<A>> merger = (left, right) -> 1082 new Partition<>(op.apply(left.forTrue, right.forTrue), 1083 op.apply(left.forFalse, right.forFalse)); 1084 Supplier<Partition<A>> supplier = () -> 1085 new Partition<>(downstream.supplier().get(), 1086 downstream.supplier().get()); 1087 if (downstream.characteristics().contains(Collector.Characteristics.IDENTITY_FINISH)) { 1088 return new CollectorImpl<>(supplier, accumulator, merger, CH_ID); 1089 } 1090 else { 1091 Function<Partition<A>, Map<Boolean, D>> finisher = par -> 1092 new Partition<>(downstream.finisher().apply(par.forTrue), 1093 downstream.finisher().apply(par.forFalse)); 1094 return new CollectorImpl<>(supplier, accumulator, merger, finisher, CH_NOID); 1095 } 1096 } 1097 1098 /** 1099 * Returns a {@code Collector} that accumulates elements into a 1100 * {@code Map} whose keys and values are the result of applying the provided 1101 * mapping functions to the input elements. 1102 * 1103 * <p>If the mapped keys contains duplicates (according to 1104 * {@link Object#equals(Object)}), an {@code IllegalStateException} is 1105 * thrown when the collection operation is performed. If the mapped keys 1106 * may have duplicates, use {@link #toMap(Function, Function, BinaryOperator)} 1107 * instead. 1108 * 1109 * @apiNote 1110 * It is common for either the key or the value to be the input elements. 1111 * In this case, the utility method 1112 * {@link java.util.function.Function#identity()} may be helpful. 1113 * For example, the following produces a {@code Map} mapping 1114 * students to their grade point average: 1115 * <pre>{@code 1116 * Map<Student, Double> studentToGPA 1117 * students.stream().collect(toMap(Functions.identity(), 1118 * student -> computeGPA(student))); 1119 * }</pre> 1120 * And the following produces a {@code Map} mapping a unique identifier to 1121 * students: 1122 * <pre>{@code 1123 * Map<String, Student> studentIdToStudent 1124 * students.stream().collect(toMap(Student::getId, 1125 * Functions.identity()); 1126 * }</pre> 1127 * 1128 * @implNote 1129 * The returned {@code Collector} is not concurrent. For parallel stream 1130 * pipelines, the {@code combiner} function operates by merging the keys 1131 * from one map into another, which can be an expensive operation. If it is 1132 * not required that results are inserted into the {@code Map} in encounter 1133 * order, using {@link #toConcurrentMap(Function, Function)} 1134 * may offer better parallel performance. 1135 * 1136 * @param <T> the type of the input elements 1137 * @param <K> the output type of the key mapping function 1138 * @param <U> the output type of the value mapping function 1139 * @param keyMapper a mapping function to produce keys 1140 * @param valueMapper a mapping function to produce values 1141 * @return a {@code Collector} which collects elements into a {@code Map} 1142 * whose keys and values are the result of applying mapping functions to 1143 * the input elements 1144 * 1145 * @see #toMap(Function, Function, BinaryOperator) 1146 * @see #toMap(Function, Function, BinaryOperator, Supplier) 1147 * @see #toConcurrentMap(Function, Function) 1148 */ 1149 public static <T, K, U> 1150 Collector<T, ?, Map<K,U>> toMap(Function<? super T, ? extends K> keyMapper, 1151 Function<? super T, ? extends U> valueMapper) { 1152 return toMap(keyMapper, valueMapper, throwingMerger(), HashMap::new); 1153 } 1154 1155 /** 1156 * Returns a {@code Collector} that accumulates elements into a 1157 * {@code Map} whose keys and values are the result of applying the provided 1158 * mapping functions to the input elements. 1159 * 1160 * <p>If the mapped 1161 * keys contains duplicates (according to {@link Object#equals(Object)}), 1162 * the value mapping function is applied to each equal element, and the 1163 * results are merged using the provided merging function. 1164 * 1165 * @apiNote 1166 * There are multiple ways to deal with collisions between multiple elements 1167 * mapping to the same key. The other forms of {@code toMap} simply use 1168 * a merge function that throws unconditionally, but you can easily write 1169 * more flexible merge policies. For example, if you have a stream 1170 * of {@code Person}, and you want to produce a "phone book" mapping name to 1171 * address, but it is possible that two persons have the same name, you can 1172 * do as follows to gracefully deals with these collisions, and produce a 1173 * {@code Map} mapping names to a concatenated list of addresses: 1174 * <pre>{@code 1175 * Map<String, String> phoneBook 1176 * people.stream().collect(toMap(Person::getName, 1177 * Person::getAddress, 1178 * (s, a) -> s + ", " + a)); 1179 * }</pre> 1180 * 1181 * @implNote 1182 * The returned {@code Collector} is not concurrent. For parallel stream 1183 * pipelines, the {@code combiner} function operates by merging the keys 1184 * from one map into another, which can be an expensive operation. If it is 1185 * not required that results are merged into the {@code Map} in encounter 1186 * order, using {@link #toConcurrentMap(Function, Function, BinaryOperator)} 1187 * may offer better parallel performance. 1188 * 1189 * @param <T> the type of the input elements 1190 * @param <K> the output type of the key mapping function 1191 * @param <U> the output type of the value mapping function 1192 * @param keyMapper a mapping function to produce keys 1193 * @param valueMapper a mapping function to produce values 1194 * @param mergeFunction a merge function, used to resolve collisions between 1195 * values associated with the same key, as supplied 1196 * to {@link Map#merge(Object, Object, BiFunction)} 1197 * @return a {@code Collector} which collects elements into a {@code Map} 1198 * whose keys are the result of applying a key mapping function to the input 1199 * elements, and whose values are the result of applying a value mapping 1200 * function to all input elements equal to the key and combining them 1201 * using the merge function 1202 * 1203 * @see #toMap(Function, Function) 1204 * @see #toMap(Function, Function, BinaryOperator, Supplier) 1205 * @see #toConcurrentMap(Function, Function, BinaryOperator) 1206 */ 1207 public static <T, K, U> 1208 Collector<T, ?, Map<K,U>> toMap(Function<? super T, ? extends K> keyMapper, 1209 Function<? super T, ? extends U> valueMapper, 1210 BinaryOperator<U> mergeFunction) { 1211 return toMap(keyMapper, valueMapper, mergeFunction, HashMap::new); 1212 } 1213 1214 /** 1215 * Returns a {@code Collector} that accumulates elements into a 1216 * {@code Map} whose keys and values are the result of applying the provided 1217 * mapping functions to the input elements. 1218 * 1219 * <p>If the mapped 1220 * keys contains duplicates (according to {@link Object#equals(Object)}), 1221 * the value mapping function is applied to each equal element, and the 1222 * results are merged using the provided merging function. The {@code Map} 1223 * is created by a provided supplier function. 1224 * 1225 * @implNote 1226 * The returned {@code Collector} is not concurrent. For parallel stream 1227 * pipelines, the {@code combiner} function operates by merging the keys 1228 * from one map into another, which can be an expensive operation. If it is 1229 * not required that results are merged into the {@code Map} in encounter 1230 * order, using {@link #toConcurrentMap(Function, Function, BinaryOperator, Supplier)} 1231 * may offer better parallel performance. 1232 * 1233 * @param <T> the type of the input elements 1234 * @param <K> the output type of the key mapping function 1235 * @param <U> the output type of the value mapping function 1236 * @param <M> the type of the resulting {@code Map} 1237 * @param keyMapper a mapping function to produce keys 1238 * @param valueMapper a mapping function to produce values 1239 * @param mergeFunction a merge function, used to resolve collisions between 1240 * values associated with the same key, as supplied 1241 * to {@link Map#merge(Object, Object, BiFunction)} 1242 * @param mapSupplier a function which returns a new, empty {@code Map} into 1243 * which the results will be inserted 1244 * @return a {@code Collector} which collects elements into a {@code Map} 1245 * whose keys are the result of applying a key mapping function to the input 1246 * elements, and whose values are the result of applying a value mapping 1247 * function to all input elements equal to the key and combining them 1248 * using the merge function 1249 * 1250 * @see #toMap(Function, Function) 1251 * @see #toMap(Function, Function, BinaryOperator) 1252 * @see #toConcurrentMap(Function, Function, BinaryOperator, Supplier) 1253 */ 1254 public static <T, K, U, M extends Map<K, U>> 1255 Collector<T, ?, M> toMap(Function<? super T, ? extends K> keyMapper, 1256 Function<? super T, ? extends U> valueMapper, 1257 BinaryOperator<U> mergeFunction, 1258 Supplier<M> mapSupplier) { 1259 BiConsumer<M, T> accumulator 1260 = (map, element) -> map.merge(keyMapper.apply(element), 1261 valueMapper.apply(element), mergeFunction); 1262 return new CollectorImpl<>(mapSupplier, accumulator, mapMerger(mergeFunction), CH_ID); 1263 } 1264 1265 /** 1266 * Returns a concurrent {@code Collector} that accumulates elements into a 1267 * {@code ConcurrentMap} whose keys and values are the result of applying 1268 * the provided mapping functions to the input elements. 1269 * 1270 * <p>If the mapped keys contains duplicates (according to 1271 * {@link Object#equals(Object)}), an {@code IllegalStateException} is 1272 * thrown when the collection operation is performed. If the mapped keys 1273 * may have duplicates, use 1274 * {@link #toConcurrentMap(Function, Function, BinaryOperator)} instead. 1275 * 1276 * @apiNote 1277 * It is common for either the key or the value to be the input elements. 1278 * In this case, the utility method 1279 * {@link java.util.function.Function#identity()} may be helpful. 1280 * For example, the following produces a {@code Map} mapping 1281 * students to their grade point average: 1282 * <pre>{@code 1283 * Map<Student, Double> studentToGPA 1284 * students.stream().collect(toMap(Functions.identity(), 1285 * student -> computeGPA(student))); 1286 * }</pre> 1287 * And the following produces a {@code Map} mapping a unique identifier to 1288 * students: 1289 * <pre>{@code 1290 * Map<String, Student> studentIdToStudent 1291 * students.stream().collect(toConcurrentMap(Student::getId, 1292 * Functions.identity()); 1293 * }</pre> 1294 * 1295 * <p>This is a {@link Collector.Characteristics#CONCURRENT concurrent} and 1296 * {@link Collector.Characteristics#UNORDERED unordered} Collector. 1297 * 1298 * @param <T> the type of the input elements 1299 * @param <K> the output type of the key mapping function 1300 * @param <U> the output type of the value mapping function 1301 * @param keyMapper the mapping function to produce keys 1302 * @param valueMapper the mapping function to produce values 1303 * @return a concurrent, unordered {@code Collector} which collects elements into a 1304 * {@code ConcurrentMap} whose keys are the result of applying a key mapping 1305 * function to the input elements, and whose values are the result of 1306 * applying a value mapping function to the input elements 1307 * 1308 * @see #toMap(Function, Function) 1309 * @see #toConcurrentMap(Function, Function, BinaryOperator) 1310 * @see #toConcurrentMap(Function, Function, BinaryOperator, Supplier) 1311 */ 1312 public static <T, K, U> 1313 Collector<T, ?, ConcurrentMap<K,U>> toConcurrentMap(Function<? super T, ? extends K> keyMapper, 1314 Function<? super T, ? extends U> valueMapper) { 1315 return toConcurrentMap(keyMapper, valueMapper, throwingMerger(), ConcurrentHashMap::new); 1316 } 1317 1318 /** 1319 * Returns a concurrent {@code Collector} that accumulates elements into a 1320 * {@code ConcurrentMap} whose keys and values are the result of applying 1321 * the provided mapping functions to the input elements. 1322 * 1323 * <p>If the mapped keys contains duplicates (according to {@link Object#equals(Object)}), 1324 * the value mapping function is applied to each equal element, and the 1325 * results are merged using the provided merging function. 1326 * 1327 * @apiNote 1328 * There are multiple ways to deal with collisions between multiple elements 1329 * mapping to the same key. The other forms of {@code toConcurrentMap} simply use 1330 * a merge function that throws unconditionally, but you can easily write 1331 * more flexible merge policies. For example, if you have a stream 1332 * of {@code Person}, and you want to produce a "phone book" mapping name to 1333 * address, but it is possible that two persons have the same name, you can 1334 * do as follows to gracefully deals with these collisions, and produce a 1335 * {@code Map} mapping names to a concatenated list of addresses: 1336 * <pre>{@code 1337 * Map<String, String> phoneBook 1338 * people.stream().collect(toConcurrentMap(Person::getName, 1339 * Person::getAddress, 1340 * (s, a) -> s + ", " + a)); 1341 * }</pre> 1342 * 1343 * <p>This is a {@link Collector.Characteristics#CONCURRENT concurrent} and 1344 * {@link Collector.Characteristics#UNORDERED unordered} Collector. 1345 * 1346 * @param <T> the type of the input elements 1347 * @param <K> the output type of the key mapping function 1348 * @param <U> the output type of the value mapping function 1349 * @param keyMapper a mapping function to produce keys 1350 * @param valueMapper a mapping function to produce values 1351 * @param mergeFunction a merge function, used to resolve collisions between 1352 * values associated with the same key, as supplied 1353 * to {@link Map#merge(Object, Object, BiFunction)} 1354 * @return a concurrent, unordered {@code Collector} which collects elements into a 1355 * {@code ConcurrentMap} whose keys are the result of applying a key mapping 1356 * function to the input elements, and whose values are the result of 1357 * applying a value mapping function to all input elements equal to the key 1358 * and combining them using the merge function 1359 * 1360 * @see #toConcurrentMap(Function, Function) 1361 * @see #toConcurrentMap(Function, Function, BinaryOperator, Supplier) 1362 * @see #toMap(Function, Function, BinaryOperator) 1363 */ 1364 public static <T, K, U> 1365 Collector<T, ?, ConcurrentMap<K,U>> 1366 toConcurrentMap(Function<? super T, ? extends K> keyMapper, 1367 Function<? super T, ? extends U> valueMapper, 1368 BinaryOperator<U> mergeFunction) { 1369 return toConcurrentMap(keyMapper, valueMapper, mergeFunction, ConcurrentHashMap::new); 1370 } 1371 1372 /** 1373 * Returns a concurrent {@code Collector} that accumulates elements into a 1374 * {@code ConcurrentMap} whose keys and values are the result of applying 1375 * the provided mapping functions to the input elements. 1376 * 1377 * <p>If the mapped keys contains duplicates (according to {@link Object#equals(Object)}), 1378 * the value mapping function is applied to each equal element, and the 1379 * results are merged using the provided merging function. The 1380 * {@code ConcurrentMap} is created by a provided supplier function. 1381 * 1382 * <p>This is a {@link Collector.Characteristics#CONCURRENT concurrent} and 1383 * {@link Collector.Characteristics#UNORDERED unordered} Collector. 1384 * 1385 * @param <T> the type of the input elements 1386 * @param <K> the output type of the key mapping function 1387 * @param <U> the output type of the value mapping function 1388 * @param <M> the type of the resulting {@code ConcurrentMap} 1389 * @param keyMapper a mapping function to produce keys 1390 * @param valueMapper a mapping function to produce values 1391 * @param mergeFunction a merge function, used to resolve collisions between 1392 * values associated with the same key, as supplied 1393 * to {@link Map#merge(Object, Object, BiFunction)} 1394 * @param mapSupplier a function which returns a new, empty {@code Map} into 1395 * which the results will be inserted 1396 * @return a concurrent, unordered {@code Collector} which collects elements into a 1397 * {@code ConcurrentMap} whose keys are the result of applying a key mapping 1398 * function to the input elements, and whose values are the result of 1399 * applying a value mapping function to all input elements equal to the key 1400 * and combining them using the merge function 1401 * 1402 * @see #toConcurrentMap(Function, Function) 1403 * @see #toConcurrentMap(Function, Function, BinaryOperator) 1404 * @see #toMap(Function, Function, BinaryOperator, Supplier) 1405 */ 1406 public static <T, K, U, M extends ConcurrentMap<K, U>> 1407 Collector<T, ?, M> toConcurrentMap(Function<? super T, ? extends K> keyMapper, 1408 Function<? super T, ? extends U> valueMapper, 1409 BinaryOperator<U> mergeFunction, 1410 Supplier<M> mapSupplier) { 1411 BiConsumer<M, T> accumulator 1412 = (map, element) -> map.merge(keyMapper.apply(element), 1413 valueMapper.apply(element), mergeFunction); 1414 return new CollectorImpl<>(mapSupplier, accumulator, mapMerger(mergeFunction), CH_CONCURRENT_ID); 1415 } 1416 1417 /** 1418 * Returns a {@code Collector} which applies an {@code int}-producing 1419 * mapping function to each input element, and returns summary statistics 1420 * for the resulting values. 1421 * 1422 * @param <T> the type of the input elements 1423 * @param mapper a mapping function to apply to each element 1424 * @return a {@code Collector} implementing the summary-statistics reduction 1425 * 1426 * @see #summarizingDouble(ToDoubleFunction) 1427 * @see #summarizingLong(ToLongFunction) 1428 */ 1429 public static <T> 1430 Collector<T, ?, IntSummaryStatistics> summarizingInt(ToIntFunction<? super T> mapper) { 1431 return new CollectorImpl<T, IntSummaryStatistics, IntSummaryStatistics>( 1432 IntSummaryStatistics::new, 1433 (r, t) -> r.accept(mapper.applyAsInt(t)), 1434 (l, r) -> { l.combine(r); return l; }, CH_ID); 1435 } 1436 1437 /** 1438 * Returns a {@code Collector} which applies an {@code long}-producing 1439 * mapping function to each input element, and returns summary statistics 1440 * for the resulting values. 1441 * 1442 * @param <T> the type of the input elements 1443 * @param mapper the mapping function to apply to each element 1444 * @return a {@code Collector} implementing the summary-statistics reduction 1445 * 1446 * @see #summarizingDouble(ToDoubleFunction) 1447 * @see #summarizingInt(ToIntFunction) 1448 */ 1449 public static <T> 1450 Collector<T, ?, LongSummaryStatistics> summarizingLong(ToLongFunction<? super T> mapper) { 1451 return new CollectorImpl<T, LongSummaryStatistics, LongSummaryStatistics>( 1452 LongSummaryStatistics::new, 1453 (r, t) -> r.accept(mapper.applyAsLong(t)), 1454 (l, r) -> { l.combine(r); return l; }, CH_ID); 1455 } 1456 1457 /** 1458 * Returns a {@code Collector} which applies an {@code double}-producing 1459 * mapping function to each input element, and returns summary statistics 1460 * for the resulting values. 1461 * 1462 * @param <T> the type of the input elements 1463 * @param mapper a mapping function to apply to each element 1464 * @return a {@code Collector} implementing the summary-statistics reduction 1465 * 1466 * @see #summarizingLong(ToLongFunction) 1467 * @see #summarizingInt(ToIntFunction) 1468 */ 1469 public static <T> 1470 Collector<T, ?, DoubleSummaryStatistics> summarizingDouble(ToDoubleFunction<? super T> mapper) { 1471 return new CollectorImpl<T, DoubleSummaryStatistics, DoubleSummaryStatistics>( 1472 DoubleSummaryStatistics::new, 1473 (r, t) -> r.accept(mapper.applyAsDouble(t)), 1474 (l, r) -> { l.combine(r); return l; }, CH_ID); 1475 } 1476 1477 /** 1478 * Implementation class used by partitioningBy. 1479 */ 1480 private static final class Partition<T> 1481 extends AbstractMap<Boolean, T> 1482 implements Map<Boolean, T> { 1483 final T forTrue; 1484 final T forFalse; 1485 1486 Partition(T forTrue, T forFalse) { 1487 this.forTrue = forTrue; 1488 this.forFalse = forFalse; 1489 } 1490 1491 @Override 1492 public Set<Map.Entry<Boolean, T>> entrySet() { 1493 return new AbstractSet<Map.Entry<Boolean, T>>() { 1494 @Override 1495 public Iterator<Map.Entry<Boolean, T>> iterator() { 1496 Map.Entry<Boolean, T> falseEntry = new SimpleImmutableEntry<>(false, forFalse); 1497 Map.Entry<Boolean, T> trueEntry = new SimpleImmutableEntry<>(true, forTrue); 1498 return Arrays.asList(falseEntry, trueEntry).iterator(); 1499 } 1500 1501 @Override 1502 public int size() { 1503 return 2; 1504 } 1505 }; 1506 } 1507 } 1508 }