/
ParIterableLike.scala
1500 lines (1320 loc) · 65.9 KB
/
ParIterableLike.scala
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/* __ *\
** ________ ___ / / ___ Scala API **
** / __/ __// _ | / / / _ | (c) 2003-2013, LAMP/EPFL **
** __\ \/ /__/ __ |/ /__/ __ | http://scala-lang.org/ **
** /____/\___/_/ |_/____/_/ | | **
** |/ **
\* */
package scala
package collection.parallel
import scala.language.{ higherKinds, implicitConversions }
import scala.collection.mutable.Builder
import scala.collection.mutable.ArrayBuffer
import scala.collection.IterableLike
import scala.collection.Parallel
import scala.collection.Parallelizable
import scala.collection.CustomParallelizable
import scala.collection.generic._
import scala.collection.GenIterableLike
import scala.collection.GenIterable
import scala.collection.GenTraversableOnce
import scala.collection.GenTraversable
import immutable.HashMapCombiner
import scala.reflect.ClassTag
import scala.annotation.unchecked.uncheckedVariance
import scala.collection.parallel.ParallelCollectionImplicits._
/** A template trait for parallel collections of type `ParIterable[T]`.
*
* $paralleliterableinfo
*
* $sideeffects
*
* @tparam T the element type of the collection
* @tparam Repr the type of the actual collection containing the elements
*
* @define paralleliterableinfo
* This is a base trait for Scala parallel collections. It defines behaviour
* common to all parallel collections. Concrete parallel collections should
* inherit this trait and `ParIterable` if they want to define specific combiner
* factories.
*
* Parallel operations are implemented with divide and conquer style algorithms that
* parallelize well. The basic idea is to split the collection into smaller parts until
* they are small enough to be operated on sequentially.
*
* All of the parallel operations are implemented as tasks within this trait. Tasks rely
* on the concept of splitters, which extend iterators. Every parallel collection defines:
*
* {{{
* def splitter: IterableSplitter[T]
* }}}
*
* which returns an instance of `IterableSplitter[T]`, which is a subtype of `Splitter[T]`.
* Splitters have a method `remaining` to check the remaining number of elements,
* and method `split` which is defined by splitters. Method `split` divides the splitters
* iterate over into disjunct subsets:
*
* {{{
* def split: Seq[Splitter]
* }}}
*
* which splits the splitter into a sequence of disjunct subsplitters. This is typically a
* very fast operation which simply creates wrappers around the receiver collection.
* This can be repeated recursively.
*
* Tasks are scheduled for execution through a
* [[scala.collection.parallel.TaskSupport]] object, which can be changed
* through the `tasksupport` setter of the collection.
*
* Method `newCombiner` produces a new combiner. Combiners are an extension of builders.
* They provide a method `combine` which combines two combiners and returns a combiner
* containing elements of both combiners.
* This method can be implemented by aggressively copying all the elements into the new combiner
* or by lazily binding their results. It is recommended to avoid copying all of
* the elements for performance reasons, although that cost might be negligible depending on
* the use case. Standard parallel collection combiners avoid copying when merging results,
* relying either on a two-step lazy construction or specific data-structure properties.
*
* Methods:
*
* {{{
* def seq: Sequential
* def par: Repr
* }}}
*
* produce the sequential or parallel implementation of the collection, respectively.
* Method `par` just returns a reference to this parallel collection.
* Method `seq` is efficient - it will not copy the elements. Instead,
* it will create a sequential version of the collection using the same underlying data structure.
* Note that this is not the case for sequential collections in general - they may copy the elements
* and produce a different underlying data structure.
*
* The combination of methods `toMap`, `toSeq` or `toSet` along with `par` and `seq` is a flexible
* way to change between different collection types.
*
* Since this trait extends the `GenIterable` trait, methods like `size` must also
* be implemented in concrete collections, while `iterator` forwards to `splitter` by
* default.
*
* Each parallel collection is bound to a specific fork/join pool, on which dormant worker
* threads are kept. The fork/join pool contains other information such as the parallelism
* level, that is, the number of processors used. When a collection is created, it is assigned the
* default fork/join pool found in the `scala.parallel` package object.
*
* Parallel collections are not necessarily ordered in terms of the `foreach`
* operation (see `Traversable`). Parallel sequences have a well defined order for iterators - creating
* an iterator and traversing the elements linearly will always yield the same order.
* However, bulk operations such as `foreach`, `map` or `filter` always occur in undefined orders for all
* parallel collections.
*
* Existing parallel collection implementations provide strict parallel iterators. Strict parallel iterators are aware
* of the number of elements they have yet to traverse. It's also possible to provide non-strict parallel iterators,
* which do not know the number of elements remaining. To do this, the new collection implementation must override
* `isStrictSplitterCollection` to `false`. This will make some operations unavailable.
*
* To create a new parallel collection, extend the `ParIterable` trait, and implement `size`, `splitter`,
* `newCombiner` and `seq`. Having an implicit combiner factory requires extending this trait in addition, as
* well as providing a companion object, as with regular collections.
*
* Method `size` is implemented as a constant time operation for parallel collections, and parallel collection
* operations rely on this assumption.
*
* @author Aleksandar Prokopec
* @since 2.9
*
* @define sideeffects
* The higher-order functions passed to certain operations may contain side-effects. Since implementations
* of bulk operations may not be sequential, this means that side-effects may not be predictable and may
* produce data-races, deadlocks or invalidation of state if care is not taken. It is up to the programmer
* to either avoid using side-effects or to use some form of synchronization when accessing mutable data.
*
* @define pbfinfo
* An implicit value of class `CanCombineFrom` which determines the
* result class `That` from the current representation type `Repr` and
* and the new element type `B`. This builder factory can provide a parallel
* builder for the resulting collection.
*
* @define abortsignalling
* This method will use `abort` signalling capabilities. This means
* that splitters may send and read `abort` signals.
*
* @define indexsignalling
* This method will use `indexFlag` signalling capabilities. This means
* that splitters may set and read the `indexFlag` state.
* @define Coll `ParIterable`
* @define coll parallel iterable
*/
trait ParIterableLike[+T, +Repr <: ParIterable[T], +Sequential <: Iterable[T] with IterableLike[T, Sequential]]
extends GenIterableLike[T, Repr]
with CustomParallelizable[T, Repr]
with Parallel
with HasNewCombiner[T, Repr]
{
self: ParIterableLike[T, Repr, Sequential] =>
@transient
@volatile
private var _tasksupport = defaultTaskSupport
protected def initTaskSupport() {
_tasksupport = defaultTaskSupport
}
/** The task support object which is responsible for scheduling and
* load-balancing tasks to processors.
*
* @see [[scala.collection.parallel.TaskSupport]]
*/
def tasksupport = {
val ts = _tasksupport
if (ts eq null) {
_tasksupport = defaultTaskSupport
defaultTaskSupport
} else ts
}
/** Changes the task support object which is responsible for scheduling and
* load-balancing tasks to processors.
*
* A task support object can be changed in a parallel collection after it
* has been created, but only during a quiescent period, i.e. while there
* are no concurrent invocations to parallel collection methods.
*
* Here is a way to change the task support of a parallel collection:
*
* {{{
* import scala.collection.parallel._
* val pc = mutable.ParArray(1, 2, 3)
* pc.tasksupport = new ForkJoinTaskSupport(
* new java.util.concurrent.ForkJoinPool(2))
* }}}
*
* @see [[scala.collection.parallel.TaskSupport]]
*/
def tasksupport_=(ts: TaskSupport) = _tasksupport = ts
def seq: Sequential
def repr: Repr = this.asInstanceOf[Repr]
final def isTraversableAgain = true
def hasDefiniteSize = true
def isEmpty = size == 0
def nonEmpty = size != 0
def head = iterator.next()
def headOption = if (nonEmpty) Some(head) else None
def tail = drop(1)
def last = {
var lst = head
for (x <- this.seq) lst = x
lst
}
def lastOption = if (nonEmpty) Some(last) else None
def init = take(size - 1)
/** Creates a new parallel iterator used to traverse the elements of this parallel collection.
* This iterator is more specific than the iterator of the returned by `iterator`, and augmented
* with additional accessor and transformer methods.
*
* @return a parallel iterator
*/
protected[parallel] def splitter: IterableSplitter[T]
/** Creates a new split iterator used to traverse the elements of this collection.
*
* By default, this method is implemented in terms of the protected `splitter` method.
*
* @return a split iterator
*/
def iterator: Splitter[T] = splitter
override def par: Repr = repr
/** Denotes whether this parallel collection has strict splitters.
*
* This is true in general, and specific collection instances may choose to
* override this method. Such collections will fail to execute methods
* which rely on splitters being strict, i.e. returning a correct value
* in the `remaining` method.
*
* This method helps ensure that such failures occur on method invocations,
* rather than later on and in unpredictable ways.
*/
def isStrictSplitterCollection = true
/** The `newBuilder` operation returns a parallel builder assigned to this collection's fork/join pool.
* This method forwards the call to `newCombiner`.
*/
//protected[this] def newBuilder: scala.collection.mutable.Builder[T, Repr] = newCombiner
/** Optionally reuses an existing combiner for better performance. By default it doesn't - subclasses may override this behaviour.
* The provided combiner `oldc` that can potentially be reused will be either some combiner from the previous computational task, or `None` if there
* was no previous phase (in which case this method must return `newc`).
*
* @param oldc The combiner that is the result of the previous task, or `None` if there was no previous task.
* @param newc The new, empty combiner that can be used.
* @return Either `newc` or `oldc`.
*/
protected def reuse[S, That](oldc: Option[Combiner[S, That]], newc: Combiner[S, That]): Combiner[S, That] = newc
type SSCTask[R, Tp] = StrictSplitterCheckTask[R, Tp]
/* helper traits - to avoid structural invocations */
trait TaskOps[R, Tp] {
def mapResult[R1](mapping: R => R1): ResultMapping[R, Tp, R1]
// public method with inaccessible types in parameters
def compose[R3, R2, Tp2](t2: SSCTask[R2, Tp2])(resCombiner: (R, R2) => R3): SeqComposite[R, R2, R3, SSCTask[R, Tp], SSCTask[R2, Tp2]]
def parallel[R3, R2, Tp2](t2: SSCTask[R2, Tp2])(resCombiner: (R, R2) => R3): ParComposite[R, R2, R3, SSCTask[R, Tp], SSCTask[R2, Tp2]]
}
trait BuilderOps[Elem, To] {
trait Otherwise[Cmb] {
def otherwise(notbody: => Unit)(implicit t: ClassTag[Cmb]): Unit
}
def ifIs[Cmb](isbody: Cmb => Unit): Otherwise[Cmb]
def isCombiner: Boolean
def asCombiner: Combiner[Elem, To]
}
trait SignallingOps[PI <: DelegatedSignalling] {
def assign(cntx: Signalling): PI
}
/* convenience task operations wrapper */
protected implicit def task2ops[R, Tp](tsk: SSCTask[R, Tp]) = new TaskOps[R, Tp] {
def mapResult[R1](mapping: R => R1): ResultMapping[R, Tp, R1] = new ResultMapping[R, Tp, R1](tsk) {
def map(r: R): R1 = mapping(r)
}
def compose[R3, R2, Tp2](t2: SSCTask[R2, Tp2])(resCombiner: (R, R2) => R3) = new SeqComposite[R, R2, R3, SSCTask[R, Tp], SSCTask[R2, Tp2]](tsk, t2) {
def combineResults(fr: R, sr: R2): R3 = resCombiner(fr, sr)
}
def parallel[R3, R2, Tp2](t2: SSCTask[R2, Tp2])(resCombiner: (R, R2) => R3) = new ParComposite[R, R2, R3, SSCTask[R, Tp], SSCTask[R2, Tp2]](tsk, t2) {
def combineResults(fr: R, sr: R2): R3 = resCombiner(fr, sr)
}
}
protected def wrap[R](body: => R) = new NonDivisible[R] {
def leaf(prevr: Option[R]) = result = body
@volatile var result: R = null.asInstanceOf[R]
}
/* convenience signalling operations wrapper */
protected implicit def delegatedSignalling2ops[PI <: DelegatedSignalling](it: PI) = new SignallingOps[PI] {
def assign(cntx: Signalling): PI = {
it.signalDelegate = cntx
it
}
}
protected implicit def builder2ops[Elem, To](cb: Builder[Elem, To]) = new BuilderOps[Elem, To] {
def ifIs[Cmb](isbody: Cmb => Unit) = new Otherwise[Cmb] {
def otherwise(notbody: => Unit)(implicit t: ClassTag[Cmb]) {
if (cb.getClass == t.runtimeClass) isbody(cb.asInstanceOf[Cmb]) else notbody
}
}
def isCombiner = cb.isInstanceOf[Combiner[_, _]]
def asCombiner = cb.asInstanceOf[Combiner[Elem, To]]
}
protected[this] def bf2seq[S, That](bf: CanBuildFrom[Repr, S, That]) = new CanBuildFrom[Sequential, S, That] {
def apply(from: Sequential) = bf.apply(from.par.asInstanceOf[Repr]) // !!! we only use this on `this.seq`, and know that `this.seq.par.getClass == this.getClass`
def apply() = bf.apply()
}
protected[this] def sequentially[S, That <: Parallel](b: Sequential => Parallelizable[S, That]) = b(seq).par.asInstanceOf[Repr]
def mkString(start: String, sep: String, end: String): String = seq.mkString(start, sep, end)
def mkString(sep: String): String = seq.mkString("", sep, "")
def mkString: String = seq.mkString("")
override def toString = seq.mkString(stringPrefix + "(", ", ", ")")
def canEqual(other: Any) = true
/** Reduces the elements of this sequence using the specified associative binary operator.
*
* $undefinedorder
*
* Note this method has a different signature than the `reduceLeft`
* and `reduceRight` methods of the trait `Traversable`.
* The result of reducing may only be a supertype of this parallel collection's
* type parameter `T`.
*
* @tparam U A type parameter for the binary operator, a supertype of `T`.
* @param op A binary operator that must be associative.
* @return The result of applying reduce operator `op` between all the elements if the collection is nonempty.
* @throws UnsupportedOperationException
* if this $coll is empty.
*/
def reduce[U >: T](op: (U, U) => U): U = {
tasksupport.executeAndWaitResult(new Reduce(op, splitter) mapResult { _.get })
}
/** Optionally reduces the elements of this sequence using the specified associative binary operator.
*
* $undefinedorder
*
* Note this method has a different signature than the `reduceLeftOption`
* and `reduceRightOption` methods of the trait `Traversable`.
* The result of reducing may only be a supertype of this parallel collection's
* type parameter `T`.
*
* @tparam U A type parameter for the binary operator, a supertype of `T`.
* @param op A binary operator that must be associative.
* @return An option value containing result of applying reduce operator `op` between all
* the elements if the collection is nonempty, and `None` otherwise.
*/
def reduceOption[U >: T](op: (U, U) => U): Option[U] = if (isEmpty) None else Some(reduce(op))
/** Folds the elements of this sequence using the specified associative binary operator.
* The order in which the elements are reduced is unspecified and may be nondeterministic.
*
* Note this method has a different signature than the `foldLeft`
* and `foldRight` methods of the trait `Traversable`.
* The result of folding may only be a supertype of this parallel collection's
* type parameter `T`.
*
* @tparam U a type parameter for the binary operator, a supertype of `T`.
* @param z a neutral element for the fold operation, it may be added to the result
* an arbitrary number of times, not changing the result (e.g. `Nil` for list concatenation,
* 0 for addition, or 1 for multiplication)
* @param op a binary operator that must be associative
* @return the result of applying fold operator `op` between all the elements and `z`
*/
def fold[U >: T](z: U)(op: (U, U) => U): U = {
tasksupport.executeAndWaitResult(new Fold(z, op, splitter))
}
/** Aggregates the results of applying an operator to subsequent elements.
*
* This is a more general form of `fold` and `reduce`. It has similar semantics, but does
* not require the result to be a supertype of the element type. It traverses the elements in
* different partitions sequentially, using `seqop` to update the result, and then
* applies `combop` to results from different partitions. The implementation of this
* operation may operate on an arbitrary number of collection partitions, so `combop`
* may be invoked arbitrary number of times.
*
* For example, one might want to process some elements and then produce a `Set`. In this
* case, `seqop` would process an element and append it to the set, while `combop`
* would concatenate two sets from different partitions together. The initial value
* `z` would be an empty set.
*
* {{{
* pc.aggregate(Set[Int]())(_ += process(_), _ ++ _)
* }}}
*
* Another example is calculating geometric mean from a collection of doubles
* (one would typically require big doubles for this).
*
* @tparam S the type of accumulated results
* @param z the initial value for the accumulated result of the partition - this
* will typically be the neutral element for the `seqop` operator (e.g.
* `Nil` for list concatenation or `0` for summation) and may be evaluated
* more than once
* @param seqop an operator used to accumulate results within a partition
* @param combop an associative operator used to combine results from different partitions
*/
def aggregate[S](z: =>S)(seqop: (S, T) => S, combop: (S, S) => S): S = {
tasksupport.executeAndWaitResult(new Aggregate(() => z, seqop, combop, splitter))
}
def foldLeft[S](z: S)(op: (S, T) => S): S = seq.foldLeft(z)(op)
def foldRight[S](z: S)(op: (T, S) => S): S = seq.foldRight(z)(op)
def reduceLeft[U >: T](op: (U, T) => U): U = seq.reduceLeft(op)
def reduceRight[U >: T](op: (T, U) => U): U = seq.reduceRight(op)
def reduceLeftOption[U >: T](op: (U, T) => U): Option[U] = seq.reduceLeftOption(op)
def reduceRightOption[U >: T](op: (T, U) => U): Option[U] = seq.reduceRightOption(op)
/** Applies a function `f` to all the elements of $coll in an undefined order.
*
* @tparam U the result type of the function applied to each element, which is always discarded
* @param f function applied to each element
*/
def foreach[U](f: T => U) = {
tasksupport.executeAndWaitResult(new Foreach(f, splitter))
}
def count(p: T => Boolean): Int = {
tasksupport.executeAndWaitResult(new Count(p, splitter))
}
def sum[U >: T](implicit num: Numeric[U]): U = {
tasksupport.executeAndWaitResult(new Sum[U](num, splitter))
}
def product[U >: T](implicit num: Numeric[U]): U = {
tasksupport.executeAndWaitResult(new Product[U](num, splitter))
}
def min[U >: T](implicit ord: Ordering[U]): T = {
tasksupport.executeAndWaitResult(new Min(ord, splitter) mapResult { _.get }).asInstanceOf[T]
}
def max[U >: T](implicit ord: Ordering[U]): T = {
tasksupport.executeAndWaitResult(new Max(ord, splitter) mapResult { _.get }).asInstanceOf[T]
}
def maxBy[S](f: T => S)(implicit cmp: Ordering[S]): T = {
if (isEmpty) throw new UnsupportedOperationException("empty.maxBy")
reduce((x, y) => if (cmp.gteq(f(x), f(y))) x else y)
}
def minBy[S](f: T => S)(implicit cmp: Ordering[S]): T = {
if (isEmpty) throw new UnsupportedOperationException("empty.minBy")
reduce((x, y) => if (cmp.lteq(f(x), f(y))) x else y)
}
def map[S, That](f: T => S)(implicit bf: CanBuildFrom[Repr, S, That]): That = if (bf(repr).isCombiner) {
tasksupport.executeAndWaitResult(new Map[S, That](f, combinerFactory(() => bf(repr).asCombiner), splitter) mapResult { _.resultWithTaskSupport })
} else setTaskSupport(seq.map(f)(bf2seq(bf)), tasksupport)
/*bf ifParallel { pbf =>
tasksupport.executeAndWaitResult(new Map[S, That](f, pbf, splitter) mapResult { _.result })
} otherwise seq.map(f)(bf2seq(bf))*/
def collect[S, That](pf: PartialFunction[T, S])(implicit bf: CanBuildFrom[Repr, S, That]): That = if (bf(repr).isCombiner) {
tasksupport.executeAndWaitResult(new Collect[S, That](pf, combinerFactory(() => bf(repr).asCombiner), splitter) mapResult { _.resultWithTaskSupport })
} else setTaskSupport(seq.collect(pf)(bf2seq(bf)), tasksupport)
/*bf ifParallel { pbf =>
tasksupport.executeAndWaitResult(new Collect[S, That](pf, pbf, splitter) mapResult { _.result })
} otherwise seq.collect(pf)(bf2seq(bf))*/
def flatMap[S, That](f: T => GenTraversableOnce[S])(implicit bf: CanBuildFrom[Repr, S, That]): That = if (bf(repr).isCombiner) {
tasksupport.executeAndWaitResult(new FlatMap[S, That](f, combinerFactory(() => bf(repr).asCombiner), splitter) mapResult { _.resultWithTaskSupport })
} else setTaskSupport(seq.flatMap(f)(bf2seq(bf)), tasksupport)
/*bf ifParallel { pbf =>
tasksupport.executeAndWaitResult(new FlatMap[S, That](f, pbf, splitter) mapResult { _.result })
} otherwise seq.flatMap(f)(bf2seq(bf))*/
/** Tests whether a predicate holds for all elements of this $coll.
*
* $abortsignalling
*
* @param p a predicate used to test elements
* @return true if `p` holds for all elements, false otherwise
*/
def forall(@deprecatedName('pred) p: T => Boolean): Boolean = {
tasksupport.executeAndWaitResult(new Forall(p, splitter assign new DefaultSignalling with VolatileAbort))
}
/** Tests whether a predicate holds for some element of this $coll.
*
* $abortsignalling
*
* @param p a predicate used to test elements
* @return true if `p` holds for some element, false otherwise
*/
def exists(@deprecatedName('pred) p: T => Boolean): Boolean = {
tasksupport.executeAndWaitResult(new Exists(p, splitter assign new DefaultSignalling with VolatileAbort))
}
/** Finds some element in the collection for which the predicate holds, if such
* an element exists. The element may not necessarily be the first such element
* in the iteration order.
*
* If there are multiple elements obeying the predicate, the choice is nondeterministic.
*
* $abortsignalling
*
* @param p predicate used to test the elements
* @return an option value with the element if such an element exists, or `None` otherwise
*/
def find(@deprecatedName('pred) p: T => Boolean): Option[T] = {
tasksupport.executeAndWaitResult(new Find(p, splitter assign new DefaultSignalling with VolatileAbort))
}
/** Creates a combiner factory. Each combiner factory instance is used
* once per invocation of a parallel transformer method for a single
* collection.
*
* The default combiner factory creates a new combiner every time it
* is requested, unless the combiner is thread-safe as indicated by its
* `canBeShared` method. In this case, the method returns a factory which
* returns the same combiner each time. This is typically done for
* concurrent parallel collections, the combiners of which allow
* thread safe access.
*/
protected[this] def combinerFactory = {
val combiner = newCombiner
combiner.combinerTaskSupport = tasksupport
if (combiner.canBeShared) new CombinerFactory[T, Repr] {
val shared = combiner
def apply() = shared
def doesShareCombiners = true
} else new CombinerFactory[T, Repr] {
def apply() = newCombiner
def doesShareCombiners = false
}
}
protected[this] def combinerFactory[S, That](cbf: () => Combiner[S, That]) = {
val combiner = cbf()
combiner.combinerTaskSupport = tasksupport
if (combiner.canBeShared) new CombinerFactory[S, That] {
val shared = combiner
def apply() = shared
def doesShareCombiners = true
} else new CombinerFactory[S, That] {
def apply() = cbf()
def doesShareCombiners = false
}
}
def withFilter(pred: T => Boolean): Repr = filter(pred)
def filter(pred: T => Boolean): Repr = {
tasksupport.executeAndWaitResult(new Filter(pred, combinerFactory, splitter) mapResult { _.resultWithTaskSupport })
}
def filterNot(pred: T => Boolean): Repr = {
tasksupport.executeAndWaitResult(new FilterNot(pred, combinerFactory, splitter) mapResult { _.resultWithTaskSupport })
}
def ++[U >: T, That](that: GenTraversableOnce[U])(implicit bf: CanBuildFrom[Repr, U, That]): That = {
if (that.isParallel && bf.isParallel) {
// println("case both are parallel")
val other = that.asParIterable
val pbf = bf.asParallel
val cfactory = combinerFactory(() => pbf(repr))
val copythis = new Copy(cfactory, splitter)
val copythat = wrap {
val othtask = new other.Copy(cfactory, other.splitter)
tasksupport.executeAndWaitResult(othtask)
}
val task = (copythis parallel copythat) { _ combine _ } mapResult {
_.resultWithTaskSupport
}
tasksupport.executeAndWaitResult(task)
} else if (bf(repr).isCombiner) {
// println("case parallel builder, `that` not parallel")
val copythis = new Copy(combinerFactory(() => bf(repr).asCombiner), splitter)
val copythat = wrap {
val cb = bf(repr).asCombiner
for (elem <- that.seq) cb += elem
cb
}
tasksupport.executeAndWaitResult((copythis parallel copythat) { _ combine _ } mapResult { _.resultWithTaskSupport })
} else {
// println("case not a parallel builder")
val b = bf(repr)
this.splitter.copy2builder[U, That, Builder[U, That]](b)
for (elem <- that.seq) b += elem
setTaskSupport(b.result(), tasksupport)
}
}
def partition(pred: T => Boolean): (Repr, Repr) = {
tasksupport.executeAndWaitResult(
new Partition(pred, combinerFactory, combinerFactory, splitter) mapResult {
p => (p._1.resultWithTaskSupport, p._2.resultWithTaskSupport)
}
)
}
def groupBy[K](f: T => K): immutable.ParMap[K, Repr] = {
val r = tasksupport.executeAndWaitResult(new GroupBy(f, () => HashMapCombiner[K, T], splitter) mapResult {
rcb => rcb.groupByKey(() => combinerFactory())
})
setTaskSupport(r, tasksupport)
}
def take(n: Int): Repr = {
val actualn = if (size > n) n else size
if (actualn < MIN_FOR_COPY) take_sequential(actualn)
else tasksupport.executeAndWaitResult(new Take(actualn, combinerFactory, splitter) mapResult {
_.resultWithTaskSupport
})
}
private def take_sequential(n: Int) = {
val cb = newCombiner
cb.sizeHint(n)
val it = splitter
var left = n
while (left > 0) {
cb += it.next
left -= 1
}
cb.resultWithTaskSupport
}
def drop(n: Int): Repr = {
val actualn = if (size > n) n else size
if ((size - actualn) < MIN_FOR_COPY) drop_sequential(actualn)
else tasksupport.executeAndWaitResult(new Drop(actualn, combinerFactory, splitter) mapResult { _.resultWithTaskSupport })
}
private def drop_sequential(n: Int) = {
val it = splitter drop n
val cb = newCombiner
cb.sizeHint(size - n)
while (it.hasNext) cb += it.next
cb.resultWithTaskSupport
}
override def slice(unc_from: Int, unc_until: Int): Repr = {
val from = unc_from min size max 0
val until = unc_until min size max from
if ((until - from) <= MIN_FOR_COPY) slice_sequential(from, until)
else tasksupport.executeAndWaitResult(new Slice(from, until, combinerFactory, splitter) mapResult { _.resultWithTaskSupport })
}
private def slice_sequential(from: Int, until: Int): Repr = {
val cb = newCombiner
var left = until - from
val it = splitter drop from
while (left > 0) {
cb += it.next
left -= 1
}
cb.resultWithTaskSupport
}
def splitAt(n: Int): (Repr, Repr) = {
tasksupport.executeAndWaitResult(
new SplitAt(n, combinerFactory, combinerFactory, splitter) mapResult {
p => (p._1.resultWithTaskSupport, p._2.resultWithTaskSupport)
}
)
}
/** Computes a prefix scan of the elements of the collection.
*
* Note: The neutral element `z` may be applied more than once.
*
* @tparam U element type of the resulting collection
* @tparam That type of the resulting collection
* @param z neutral element for the operator `op`
* @param op the associative operator for the scan
* @param bf $bfinfo
* @return a collection containing the prefix scan of the elements in the original collection
*
* @usecase def scan(z: T)(op: (T, T) => T): $Coll[T]
* @inheritdoc
*
* @return a new $coll containing the prefix scan of the elements in this $coll
*/
def scan[U >: T, That](z: U)(op: (U, U) => U)(implicit bf: CanBuildFrom[Repr, U, That]): That = if (bf(repr).isCombiner) {
if (tasksupport.parallelismLevel > 1) {
if (size > 0) tasksupport.executeAndWaitResult(new CreateScanTree(0, size, z, op, splitter) mapResult {
tree => tasksupport.executeAndWaitResult(new FromScanTree(tree, z, op, combinerFactory(() => bf(repr).asCombiner)) mapResult {
cb => cb.resultWithTaskSupport
})
}) else setTaskSupport((bf(repr) += z).result(), tasksupport)
} else setTaskSupport(seq.scan(z)(op)(bf2seq(bf)), tasksupport)
} else setTaskSupport(seq.scan(z)(op)(bf2seq(bf)), tasksupport)
def scanLeft[S, That](z: S)(op: (S, T) => S)(implicit bf: CanBuildFrom[Repr, S, That]) = setTaskSupport(seq.scanLeft(z)(op)(bf2seq(bf)), tasksupport)
def scanRight[S, That](z: S)(op: (T, S) => S)(implicit bf: CanBuildFrom[Repr, S, That]) = setTaskSupport(seq.scanRight(z)(op)(bf2seq(bf)), tasksupport)
/** Takes the longest prefix of elements that satisfy the predicate.
*
* $indexsignalling
* The index flag is initially set to maximum integer value.
*
* @param pred the predicate used to test the elements
* @return the longest prefix of this $coll of elements that satisfy the predicate `pred`
*/
def takeWhile(pred: T => Boolean): Repr = {
val cbf = combinerFactory
if (cbf.doesShareCombiners) {
val parseqspan = toSeq.takeWhile(pred)
tasksupport.executeAndWaitResult(new Copy(combinerFactory, parseqspan.splitter) mapResult {
_.resultWithTaskSupport
})
} else {
val cntx = new DefaultSignalling with AtomicIndexFlag
cntx.setIndexFlag(Int.MaxValue)
tasksupport.executeAndWaitResult(new TakeWhile(0, pred, combinerFactory, splitter assign cntx) mapResult {
_._1.resultWithTaskSupport
})
}
}
/** Splits this $coll into a prefix/suffix pair according to a predicate.
*
* $indexsignalling
* The index flag is initially set to maximum integer value.
*
* @param pred the predicate used to test the elements
* @return a pair consisting of the longest prefix of the collection for which all
* the elements satisfy `pred`, and the rest of the collection
*/
def span(pred: T => Boolean): (Repr, Repr) = {
val cbf = combinerFactory
if (cbf.doesShareCombiners) {
val (xs, ys) = toSeq.span(pred)
val copyxs = new Copy(combinerFactory, xs.splitter) mapResult { _.resultWithTaskSupport }
val copyys = new Copy(combinerFactory, ys.splitter) mapResult { _.resultWithTaskSupport }
val copyall = (copyxs parallel copyys) {
(xr, yr) => (xr, yr)
}
tasksupport.executeAndWaitResult(copyall)
} else {
val cntx = new DefaultSignalling with AtomicIndexFlag
cntx.setIndexFlag(Int.MaxValue)
tasksupport.executeAndWaitResult(new Span(0, pred, combinerFactory, combinerFactory, splitter assign cntx) mapResult {
p => (p._1.resultWithTaskSupport, p._2.resultWithTaskSupport)
})
}
}
/** Drops all elements in the longest prefix of elements that satisfy the predicate,
* and returns a collection composed of the remaining elements.
*
* $indexsignalling
* The index flag is initially set to maximum integer value.
*
* @param pred the predicate used to test the elements
* @return a collection composed of all the elements after the longest prefix of elements
* in this $coll that satisfy the predicate `pred`
*/
def dropWhile(pred: T => Boolean): Repr = {
val cntx = new DefaultSignalling with AtomicIndexFlag
cntx.setIndexFlag(Int.MaxValue)
tasksupport.executeAndWaitResult(
new Span(0, pred, combinerFactory, combinerFactory, splitter assign cntx) mapResult {
_._2.resultWithTaskSupport
}
)
}
def copyToArray[U >: T](xs: Array[U]) = copyToArray(xs, 0)
def copyToArray[U >: T](xs: Array[U], start: Int) = copyToArray(xs, start, xs.length - start)
def copyToArray[U >: T](xs: Array[U], start: Int, len: Int) = if (len > 0) {
tasksupport.executeAndWaitResult(new CopyToArray(start, len, xs, splitter))
}
def sameElements[U >: T](that: GenIterable[U]) = seq.sameElements(that)
def zip[U >: T, S, That](that: GenIterable[S])(implicit bf: CanBuildFrom[Repr, (U, S), That]): That = if (bf(repr).isCombiner && that.isParSeq) {
val thatseq = that.asParSeq
tasksupport.executeAndWaitResult(new Zip(combinerFactory(() => bf(repr).asCombiner), splitter, thatseq.splitter) mapResult { _.resultWithTaskSupport })
} else setTaskSupport(seq.zip(that)(bf2seq(bf)), tasksupport)
def zipWithIndex[U >: T, That](implicit bf: CanBuildFrom[Repr, (U, Int), That]): That = this zip immutable.ParRange(0, size, 1, inclusive = false)
def zipAll[S, U >: T, That](that: GenIterable[S], thisElem: U, thatElem: S)(implicit bf: CanBuildFrom[Repr, (U, S), That]): That = if (bf(repr).isCombiner && that.isParSeq) {
val thatseq = that.asParSeq
tasksupport.executeAndWaitResult(
new ZipAll(size max thatseq.length, thisElem, thatElem, combinerFactory(() => bf(repr).asCombiner), splitter, thatseq.splitter) mapResult {
_.resultWithTaskSupport
}
)
} else setTaskSupport(seq.zipAll(that, thisElem, thatElem)(bf2seq(bf)), tasksupport)
protected def toParCollection[U >: T, That](cbf: () => Combiner[U, That]): That = {
tasksupport.executeAndWaitResult(new ToParCollection(combinerFactory(cbf), splitter) mapResult { _.resultWithTaskSupport })
}
protected def toParMap[K, V, That](cbf: () => Combiner[(K, V), That])(implicit ev: T <:< (K, V)): That = {
tasksupport.executeAndWaitResult(new ToParMap(combinerFactory(cbf), splitter)(ev) mapResult { _.resultWithTaskSupport })
}
@deprecated("use .seq.view instead", "2.11.0")
def view = seq.view
override def toArray[U >: T: ClassTag]: Array[U] = {
val arr = new Array[U](size)
copyToArray(arr)
arr
}
override def toList: List[T] = seq.toList
override def toIndexedSeq: scala.collection.immutable.IndexedSeq[T] = seq.toIndexedSeq
override def toStream: Stream[T] = seq.toStream
override def toIterator: Iterator[T] = splitter
// the methods below are overridden
override def toBuffer[U >: T]: scala.collection.mutable.Buffer[U] = seq.toBuffer // have additional, parallel buffers?
override def toTraversable: GenTraversable[T] = this.asInstanceOf[GenTraversable[T]]
override def toIterable: ParIterable[T] = this.asInstanceOf[ParIterable[T]]
override def toSeq: ParSeq[T] = toParCollection[T, ParSeq[T]](() => ParSeq.newCombiner[T])
override def toSet[U >: T]: immutable.ParSet[U] = toParCollection[U, immutable.ParSet[U]](() => immutable.ParSet.newCombiner[U])
override def toMap[K, V](implicit ev: T <:< (K, V)): immutable.ParMap[K, V] = toParMap[K, V, immutable.ParMap[K, V]](() => immutable.ParMap.newCombiner[K, V])
override def toVector: Vector[T] = to[Vector]
override def to[Col[_]](implicit cbf: CanBuildFrom[Nothing, T, Col[T @uncheckedVariance]]): Col[T @uncheckedVariance] = if (cbf().isCombiner) {
toParCollection[T, Col[T]](() => cbf().asCombiner)
} else seq.to(cbf)
/* tasks */
protected trait StrictSplitterCheckTask[R, Tp] extends Task[R, Tp] {
def requiresStrictSplitters = false
if (requiresStrictSplitters && !isStrictSplitterCollection)
throw new UnsupportedOperationException("This collection does not provide strict splitters.")
}
/** Standard accessor task that iterates over the elements of the collection.
*
* @tparam R type of the result of this method (`R` for result).
* @tparam Tp the representation type of the task at hand.
*/
protected trait Accessor[R, Tp]
extends StrictSplitterCheckTask[R, Tp] {
protected[this] val pit: IterableSplitter[T]
protected[this] def newSubtask(p: IterableSplitter[T]): Accessor[R, Tp]
def shouldSplitFurther = pit.shouldSplitFurther(self.repr, tasksupport.parallelismLevel)
def split = pit.splitWithSignalling.map(newSubtask(_)) // default split procedure
private[parallel] override def signalAbort = pit.abort()
override def toString = this.getClass.getSimpleName + "(" + pit.toString + ")(" + result + ")(supername: " + super.toString + ")"
}
protected[this] trait NonDivisibleTask[R, Tp] extends StrictSplitterCheckTask[R, Tp] {
def shouldSplitFurther = false
def split = throw new UnsupportedOperationException("Does not split.")
}
protected[this] trait NonDivisible[R] extends NonDivisibleTask[R, NonDivisible[R]]
protected[this] abstract class Composite[FR, SR, R, First <: StrictSplitterCheckTask[FR, _], Second <: StrictSplitterCheckTask[SR, _]]
(val ft: First, val st: Second)
extends NonDivisibleTask[R, Composite[FR, SR, R, First, Second]] {
def combineResults(fr: FR, sr: SR): R
@volatile var result: R = null.asInstanceOf[R]
private[parallel] override def signalAbort() {
ft.signalAbort()
st.signalAbort()
}
protected def mergeSubtasks() {
ft mergeThrowables st
if (throwable eq null) result = combineResults(ft.result, st.result)
}
override def requiresStrictSplitters = ft.requiresStrictSplitters || st.requiresStrictSplitters
}
/** Sequentially performs one task after another. */
protected[this] abstract class SeqComposite[FR, SR, R, First <: StrictSplitterCheckTask[FR, _], Second <: StrictSplitterCheckTask[SR, _]]
(f: First, s: Second)
extends Composite[FR, SR, R, First, Second](f, s) {
def leaf(prevr: Option[R]) = {
tasksupport.executeAndWaitResult(ft) : Any
tasksupport.executeAndWaitResult(st) : Any
mergeSubtasks()
}
}
/** Performs two tasks in parallel, and waits for both to finish. */
protected[this] abstract class ParComposite[FR, SR, R, First <: StrictSplitterCheckTask[FR, _], Second <: StrictSplitterCheckTask[SR, _]]
(f: First, s: Second)
extends Composite[FR, SR, R, First, Second](f, s) {
def leaf(prevr: Option[R]) = {
val ftfuture: () => Any = tasksupport.execute(ft)
tasksupport.executeAndWaitResult(st) : Any
ftfuture()
mergeSubtasks()
}
}
protected[this] abstract class ResultMapping[R, Tp, R1](val inner: StrictSplitterCheckTask[R, Tp])
extends NonDivisibleTask[R1, ResultMapping[R, Tp, R1]] {
@volatile var result: R1 = null.asInstanceOf[R1]
def map(r: R): R1
def leaf(prevr: Option[R1]) = {
val initialResult = tasksupport.executeAndWaitResult(inner)
result = map(initialResult)
}
private[parallel] override def signalAbort() {
inner.signalAbort()
}
override def requiresStrictSplitters = inner.requiresStrictSplitters
}
protected trait Transformer[R, Tp] extends Accessor[R, Tp]
protected[this] class Foreach[S](op: T => S, protected[this] val pit: IterableSplitter[T])
extends Accessor[Unit, Foreach[S]] {
@volatile var result: Unit = ()
def leaf(prevr: Option[Unit]) = pit.foreach(op)
protected[this] def newSubtask(p: IterableSplitter[T]) = new Foreach[S](op, p)
}
protected[this] class Count(pred: T => Boolean, protected[this] val pit: IterableSplitter[T])
extends Accessor[Int, Count] {
// val pittxt = pit.toString
@volatile var result: Int = 0
def leaf(prevr: Option[Int]) = result = pit.count(pred)
protected[this] def newSubtask(p: IterableSplitter[T]) = new Count(pred, p)
override def merge(that: Count) = result = result + that.result
// override def toString = "CountTask(" + pittxt + ")"
}
protected[this] class Reduce[U >: T](op: (U, U) => U, protected[this] val pit: IterableSplitter[T])
extends Accessor[Option[U], Reduce[U]] {
@volatile var result: Option[U] = None
def leaf(prevr: Option[Option[U]]) = if (pit.remaining > 0) result = Some(pit.reduce(op))
protected[this] def newSubtask(p: IterableSplitter[T]) = new Reduce(op, p)
override def merge(that: Reduce[U]) =
if (this.result == None) result = that.result
else if (that.result != None) result = Some(op(result.get, that.result.get))
override def requiresStrictSplitters = true
}
protected[this] class Fold[U >: T](z: U, op: (U, U) => U, protected[this] val pit: IterableSplitter[T])
extends Accessor[U, Fold[U]] {
@volatile var result: U = null.asInstanceOf[U]
def leaf(prevr: Option[U]) = result = pit.fold(z)(op)
protected[this] def newSubtask(p: IterableSplitter[T]) = new Fold(z, op, p)
override def merge(that: Fold[U]) = result = op(result, that.result)