aboutsummaryrefslogtreecommitdiffstats
path: root/contrib/libs/icu/common/uhash.cpp
blob: 4d92dfa93fe1cd162f60d93d37ed33c8c53715b0 (plain) (blame)
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
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
******************************************************************************
*   Copyright (C) 1997-2016, International Business Machines
*   Corporation and others.  All Rights Reserved.
******************************************************************************
*   Date        Name        Description
*   03/22/00    aliu        Adapted from original C++ ICU Hashtable.
*   07/06/01    aliu        Modified to support int32_t keys on
*                           platforms with sizeof(void*) < 32.
******************************************************************************
*/

#include "uhash.h"
#include "unicode/ustring.h"
#include "cstring.h"
#include "cmemory.h"
#include "uassert.h"
#include "ustr_imp.h"

/* This hashtable is implemented as a double hash.  All elements are
 * stored in a single array with no secondary storage for collision
 * resolution (no linked list, etc.).  When there is a hash collision
 * (when two unequal keys have the same hashcode) we resolve this by
 * using a secondary hash.  The secondary hash is an increment
 * computed as a hash function (a different one) of the primary
 * hashcode.  This increment is added to the initial hash value to
 * obtain further slots assigned to the same hash code.  For this to
 * work, the length of the array and the increment must be relatively
 * prime.  The easiest way to achieve this is to have the length of
 * the array be prime, and the increment be any value from
 * 1..length-1.
 *
 * Hashcodes are 32-bit integers.  We make sure all hashcodes are
 * non-negative by masking off the top bit.  This has two effects: (1)
 * modulo arithmetic is simplified.  If we allowed negative hashcodes,
 * then when we computed hashcode % length, we could get a negative
 * result, which we would then have to adjust back into range.  It's
 * simpler to just make hashcodes non-negative. (2) It makes it easy
 * to check for empty vs. occupied slots in the table.  We just mark
 * empty or deleted slots with a negative hashcode.
 *
 * The central function is _uhash_find().  This function looks for a
 * slot matching the given key and hashcode.  If one is found, it
 * returns a pointer to that slot.  If the table is full, and no match
 * is found, it returns nullptr -- in theory.  This would make the code
 * more complicated, since all callers of _uhash_find() would then
 * have to check for a nullptr result.  To keep this from happening, we
 * don't allow the table to fill.  When there is only one
 * empty/deleted slot left, uhash_put() will refuse to increase the
 * count, and fail.  This simplifies the code.  In practice, one will
 * seldom encounter this using default UHashtables.  However, if a
 * hashtable is set to a U_FIXED resize policy, or if memory is
 * exhausted, then the table may fill.
 *
 * High and low water ratios control rehashing.  They establish levels
 * of fullness (from 0 to 1) outside of which the data array is
 * reallocated and repopulated.  Setting the low water ratio to zero
 * means the table will never shrink.  Setting the high water ratio to
 * one means the table will never grow.  The ratios should be
 * coordinated with the ratio between successive elements of the
 * PRIMES table, so that when the primeIndex is incremented or
 * decremented during rehashing, it brings the ratio of count / length
 * back into the desired range (between low and high water ratios).
 */

/********************************************************************
 * PRIVATE Constants, Macros
 ********************************************************************/

/* This is a list of non-consecutive primes chosen such that
 * PRIMES[i+1] ~ 2*PRIMES[i].  (Currently, the ratio ranges from 1.81
 * to 2.18; the inverse ratio ranges from 0.459 to 0.552.)  If this
 * ratio is changed, the low and high water ratios should also be
 * adjusted to suit.
 *
 * These prime numbers were also chosen so that they are the largest
 * prime number while being less than a power of two.
 */
static const int32_t PRIMES[] = {
    7, 13, 31, 61, 127, 251, 509, 1021, 2039, 4093, 8191, 16381, 32749,
    65521, 131071, 262139, 524287, 1048573, 2097143, 4194301, 8388593,
    16777213, 33554393, 67108859, 134217689, 268435399, 536870909,
    1073741789, 2147483647 /*, 4294967291 */
};

#define PRIMES_LENGTH UPRV_LENGTHOF(PRIMES)
#define DEFAULT_PRIME_INDEX 4

/* These ratios are tuned to the PRIMES array such that a resize
 * places the table back into the zone of non-resizing.  That is,
 * after a call to _uhash_rehash(), a subsequent call to
 * _uhash_rehash() should do nothing (should not churn).  This is only
 * a potential problem with U_GROW_AND_SHRINK.
 */
static const float RESIZE_POLICY_RATIO_TABLE[6] = {
    /* low, high water ratio */
    0.0F, 0.5F, /* U_GROW: Grow on demand, do not shrink */
    0.1F, 0.5F, /* U_GROW_AND_SHRINK: Grow and shrink on demand */
    0.0F, 1.0F  /* U_FIXED: Never change size */
};

/*
  Invariants for hashcode values:

  * DELETED < 0
  * EMPTY < 0
  * Real hashes >= 0

  Hashcodes may not start out this way, but internally they are
  adjusted so that they are always positive.  We assume 32-bit
  hashcodes; adjust these constants for other hashcode sizes.
*/
#define HASH_DELETED    ((int32_t) 0x80000000)
#define HASH_EMPTY      ((int32_t) HASH_DELETED + 1)

#define IS_EMPTY_OR_DELETED(x) ((x) < 0)

/* This macro expects a UHashTok.pointer as its keypointer and
   valuepointer parameters */
#define HASH_DELETE_KEY_VALUE(hash, keypointer, valuepointer) UPRV_BLOCK_MACRO_BEGIN { \
    if (hash->keyDeleter != nullptr && keypointer != nullptr) { \
        (*hash->keyDeleter)(keypointer); \
    } \
    if (hash->valueDeleter != nullptr && valuepointer != nullptr) { \
        (*hash->valueDeleter)(valuepointer); \
    } \
} UPRV_BLOCK_MACRO_END

/*
 * Constants for hinting whether a key or value is an integer
 * or a pointer.  If a hint bit is zero, then the associated
 * token is assumed to be an integer.
 */
#define HINT_BOTH_INTEGERS (0)
#define HINT_KEY_POINTER   (1)
#define HINT_VALUE_POINTER (2)
#define HINT_ALLOW_ZERO    (4)

/********************************************************************
 * PRIVATE Implementation
 ********************************************************************/

static UHashTok
_uhash_setElement(UHashtable *hash, UHashElement* e,
                  int32_t hashcode,
                  UHashTok key, UHashTok value, int8_t hint) {

    UHashTok oldValue = e->value;
    if (hash->keyDeleter != nullptr && e->key.pointer != nullptr &&
        e->key.pointer != key.pointer) { /* Avoid double deletion */
        (*hash->keyDeleter)(e->key.pointer);
    }
    if (hash->valueDeleter != nullptr) {
        if (oldValue.pointer != nullptr &&
            oldValue.pointer != value.pointer) { /* Avoid double deletion */
            (*hash->valueDeleter)(oldValue.pointer);
        }
        oldValue.pointer = nullptr;
    }
    /* Compilers should copy the UHashTok union correctly, but even if
     * they do, memory heap tools (e.g. BoundsChecker) can get
     * confused when a pointer is cloaked in a union and then copied.
     * TO ALLEVIATE THIS, we use hints (based on what API the user is
     * calling) to copy pointers when we know the user thinks
     * something is a pointer. */
    if (hint & HINT_KEY_POINTER) {
        e->key.pointer = key.pointer;
    } else {
        e->key = key;
    }
    if (hint & HINT_VALUE_POINTER) {
        e->value.pointer = value.pointer;
    } else {
        e->value = value;
    }
    e->hashcode = hashcode;
    return oldValue;
}

/**
 * Assumes that the given element is not empty or deleted.
 */
static UHashTok
_uhash_internalRemoveElement(UHashtable *hash, UHashElement* e) {
    UHashTok empty;
    U_ASSERT(!IS_EMPTY_OR_DELETED(e->hashcode));
    --hash->count;
    empty.pointer = nullptr; empty.integer = 0;
    return _uhash_setElement(hash, e, HASH_DELETED, empty, empty, 0);
}

static void
_uhash_internalSetResizePolicy(UHashtable *hash, enum UHashResizePolicy policy) {
    U_ASSERT(hash != nullptr);
    U_ASSERT(((int32_t)policy) >= 0);
    U_ASSERT(((int32_t)policy) < 3);
    hash->lowWaterRatio  = RESIZE_POLICY_RATIO_TABLE[policy * 2];
    hash->highWaterRatio = RESIZE_POLICY_RATIO_TABLE[policy * 2 + 1];
}

/**
 * Allocate internal data array of a size determined by the given
 * prime index.  If the index is out of range it is pinned into range.
 * If the allocation fails the status is set to
 * U_MEMORY_ALLOCATION_ERROR and all array storage is freed.  In
 * either case the previous array pointer is overwritten.
 *
 * Caller must ensure primeIndex is in range 0..PRIME_LENGTH-1.
 */
static void
_uhash_allocate(UHashtable *hash,
                int32_t primeIndex,
                UErrorCode *status) {

    UHashElement *p, *limit;
    UHashTok emptytok;

    if (U_FAILURE(*status)) return;

    U_ASSERT(primeIndex >= 0 && primeIndex < PRIMES_LENGTH);

    hash->primeIndex = static_cast<int8_t>(primeIndex);
    hash->length = PRIMES[primeIndex];

    p = hash->elements = (UHashElement*)
        uprv_malloc(sizeof(UHashElement) * hash->length);

    if (hash->elements == nullptr) {
        *status = U_MEMORY_ALLOCATION_ERROR;
        return;
    }

    emptytok.pointer = nullptr; /* Only one of these two is needed */
    emptytok.integer = 0;    /* but we don't know which one. */

    limit = p + hash->length;
    while (p < limit) {
        p->key = emptytok;
        p->value = emptytok;
        p->hashcode = HASH_EMPTY;
        ++p;
    }

    hash->count = 0;
    hash->lowWaterMark = (int32_t)(hash->length * hash->lowWaterRatio);
    hash->highWaterMark = (int32_t)(hash->length * hash->highWaterRatio);
}

static UHashtable*
_uhash_init(UHashtable *result,
              UHashFunction *keyHash,
              UKeyComparator *keyComp,
              UValueComparator *valueComp,
              int32_t primeIndex,
              UErrorCode *status)
{
    if (U_FAILURE(*status)) return nullptr;
    U_ASSERT(keyHash != nullptr);
    U_ASSERT(keyComp != nullptr);

    result->keyHasher       = keyHash;
    result->keyComparator   = keyComp;
    result->valueComparator = valueComp;
    result->keyDeleter      = nullptr;
    result->valueDeleter    = nullptr;
    result->allocated       = false;
    _uhash_internalSetResizePolicy(result, U_GROW);

    _uhash_allocate(result, primeIndex, status);

    if (U_FAILURE(*status)) {
        return nullptr;
    }

    return result;
}

static UHashtable*
_uhash_create(UHashFunction *keyHash,
              UKeyComparator *keyComp,
              UValueComparator *valueComp,
              int32_t primeIndex,
              UErrorCode *status) {
    UHashtable *result;

    if (U_FAILURE(*status)) return nullptr;

    result = (UHashtable*) uprv_malloc(sizeof(UHashtable));
    if (result == nullptr) {
        *status = U_MEMORY_ALLOCATION_ERROR;
        return nullptr;
    }

    _uhash_init(result, keyHash, keyComp, valueComp, primeIndex, status);
    result->allocated       = true;

    if (U_FAILURE(*status)) {
        uprv_free(result);
        return nullptr;
    }

    return result;
}

/**
 * Look for a key in the table, or if no such key exists, the first
 * empty slot matching the given hashcode.  Keys are compared using
 * the keyComparator function.
 *
 * First find the start position, which is the hashcode modulo
 * the length.  Test it to see if it is:
 *
 * a. identical:  First check the hash values for a quick check,
 *    then compare keys for equality using keyComparator.
 * b. deleted
 * c. empty
 *
 * Stop if it is identical or empty, otherwise continue by adding a
 * "jump" value (moduloing by the length again to keep it within
 * range) and retesting.  For efficiency, there need enough empty
 * values so that the searches stop within a reasonable amount of time.
 * This can be changed by changing the high/low water marks.
 *
 * In theory, this function can return nullptr, if it is full (no empty
 * or deleted slots) and if no matching key is found.  In practice, we
 * prevent this elsewhere (in uhash_put) by making sure the last slot
 * in the table is never filled.
 *
 * The size of the table should be prime for this algorithm to work;
 * otherwise we are not guaranteed that the jump value (the secondary
 * hash) is relatively prime to the table length.
 */
static UHashElement*
_uhash_find(const UHashtable *hash, UHashTok key,
            int32_t hashcode) {

    int32_t firstDeleted = -1;  /* assume invalid index */
    int32_t theIndex, startIndex;
    int32_t jump = 0; /* lazy evaluate */
    int32_t tableHash;
    UHashElement *elements = hash->elements;

    hashcode &= 0x7FFFFFFF; /* must be positive */
    startIndex = theIndex = (hashcode ^ 0x4000000) % hash->length;

    do {
        tableHash = elements[theIndex].hashcode;
        if (tableHash == hashcode) {          /* quick check */
            if ((*hash->keyComparator)(key, elements[theIndex].key)) {
                return &(elements[theIndex]);
            }
        } else if (!IS_EMPTY_OR_DELETED(tableHash)) {
            /* We have hit a slot which contains a key-value pair,
             * but for which the hash code does not match.  Keep
             * looking.
             */
        } else if (tableHash == HASH_EMPTY) { /* empty, end o' the line */
            break;
        } else if (firstDeleted < 0) { /* remember first deleted */
            firstDeleted = theIndex;
        }
        if (jump == 0) { /* lazy compute jump */
            /* The jump value must be relatively prime to the table
             * length.  As long as the length is prime, then any value
             * 1..length-1 will be relatively prime to it.
             */
            jump = (hashcode % (hash->length - 1)) + 1;
        }
        theIndex = (theIndex + jump) % hash->length;
    } while (theIndex != startIndex);

    if (firstDeleted >= 0) {
        theIndex = firstDeleted; /* reset if had deleted slot */
    } else if (tableHash != HASH_EMPTY) {
        /* We get to this point if the hashtable is full (no empty or
         * deleted slots), and we've failed to find a match.  THIS
         * WILL NEVER HAPPEN as long as uhash_put() makes sure that
         * count is always < length.
         */
        UPRV_UNREACHABLE_EXIT;
    }
    return &(elements[theIndex]);
}

/**
 * Attempt to grow or shrink the data arrays in order to make the
 * count fit between the high and low water marks.  hash_put() and
 * hash_remove() call this method when the count exceeds the high or
 * low water marks.  This method may do nothing, if memory allocation
 * fails, or if the count is already in range, or if the length is
 * already at the low or high limit.  In any case, upon return the
 * arrays will be valid.
 */
static void
_uhash_rehash(UHashtable *hash, UErrorCode *status) {

    UHashElement *old = hash->elements;
    int32_t oldLength = hash->length;
    int32_t newPrimeIndex = hash->primeIndex;
    int32_t i;

    if (hash->count > hash->highWaterMark) {
        if (++newPrimeIndex >= PRIMES_LENGTH) {
            return;
        }
    } else if (hash->count < hash->lowWaterMark) {
        if (--newPrimeIndex < 0) {
            return;
        }
    } else {
        return;
    }

    _uhash_allocate(hash, newPrimeIndex, status);

    if (U_FAILURE(*status)) {
        hash->elements = old;
        hash->length = oldLength;
        return;
    }

    for (i = oldLength - 1; i >= 0; --i) {
        if (!IS_EMPTY_OR_DELETED(old[i].hashcode)) {
            UHashElement *e = _uhash_find(hash, old[i].key, old[i].hashcode);
            U_ASSERT(e != nullptr);
            U_ASSERT(e->hashcode == HASH_EMPTY);
            e->key = old[i].key;
            e->value = old[i].value;
            e->hashcode = old[i].hashcode;
            ++hash->count;
        }
    }

    uprv_free(old);
}

static UHashTok
_uhash_remove(UHashtable *hash,
              UHashTok key) {
    /* First find the position of the key in the table.  If the object
     * has not been removed already, remove it.  If the user wanted
     * keys deleted, then delete it also.  We have to put a special
     * hashcode in that position that means that something has been
     * deleted, since when we do a find, we have to continue PAST any
     * deleted values.
     */
    UHashTok result;
    UHashElement* e = _uhash_find(hash, key, hash->keyHasher(key));
    U_ASSERT(e != nullptr);
    result.pointer = nullptr;
    result.integer = 0;
    if (!IS_EMPTY_OR_DELETED(e->hashcode)) {
        result = _uhash_internalRemoveElement(hash, e);
        if (hash->count < hash->lowWaterMark) {
            UErrorCode status = U_ZERO_ERROR;
            _uhash_rehash(hash, &status);
        }
    }
    return result;
}

static UHashTok
_uhash_put(UHashtable *hash,
           UHashTok key,
           UHashTok value,
           int8_t hint,
           UErrorCode *status) {

    /* Put finds the position in the table for the new value.  If the
     * key is already in the table, it is deleted, if there is a
     * non-nullptr keyDeleter.  Then the key, the hash and the value are
     * all put at the position in their respective arrays.
     */
    int32_t hashcode;
    UHashElement* e;
    UHashTok emptytok;

    if (U_FAILURE(*status)) {
        goto err;
    }
    U_ASSERT(hash != nullptr);
    if ((hint & HINT_VALUE_POINTER) ?
            value.pointer == nullptr :
            value.integer == 0 && (hint & HINT_ALLOW_ZERO) == 0) {
        /* Disallow storage of nullptr values, since nullptr is returned by
         * get() to indicate an absent key.  Storing nullptr == removing.
         */
        return _uhash_remove(hash, key);
    }
    if (hash->count > hash->highWaterMark) {
        _uhash_rehash(hash, status);
        if (U_FAILURE(*status)) {
            goto err;
        }
    }

    hashcode = (*hash->keyHasher)(key);
    e = _uhash_find(hash, key, hashcode);
    U_ASSERT(e != nullptr);

    if (IS_EMPTY_OR_DELETED(e->hashcode)) {
        /* Important: We must never actually fill the table up.  If we
         * do so, then _uhash_find() will return nullptr, and we'll have
         * to check for nullptr after every call to _uhash_find().  To
         * avoid this we make sure there is always at least one empty
         * or deleted slot in the table.  This only is a problem if we
         * are out of memory and rehash isn't working.
         */
        ++hash->count;
        if (hash->count == hash->length) {
            /* Don't allow count to reach length */
            --hash->count;
            *status = U_MEMORY_ALLOCATION_ERROR;
            goto err;
        }
    }

    /* We must in all cases handle storage properly.  If there was an
     * old key, then it must be deleted (if the deleter != nullptr).
     * Make hashcodes stored in table positive.
     */
    return _uhash_setElement(hash, e, hashcode & 0x7FFFFFFF, key, value, hint);

 err:
    /* If the deleters are non-nullptr, this method adopts its key and/or
     * value arguments, and we must be sure to delete the key and/or
     * value in all cases, even upon failure.
     */
    HASH_DELETE_KEY_VALUE(hash, key.pointer, value.pointer);
    emptytok.pointer = nullptr; emptytok.integer = 0;
    return emptytok;
}


/********************************************************************
 * PUBLIC API
 ********************************************************************/

U_CAPI UHashtable* U_EXPORT2
uhash_open(UHashFunction *keyHash,
           UKeyComparator *keyComp,
           UValueComparator *valueComp,
           UErrorCode *status) {

    return _uhash_create(keyHash, keyComp, valueComp, DEFAULT_PRIME_INDEX, status);
}

U_CAPI UHashtable* U_EXPORT2
uhash_openSize(UHashFunction *keyHash,
               UKeyComparator *keyComp,
               UValueComparator *valueComp,
               int32_t size,
               UErrorCode *status) {

    /* Find the smallest index i for which PRIMES[i] >= size. */
    int32_t i = 0;
    while (i<(PRIMES_LENGTH-1) && PRIMES[i]<size) {
        ++i;
    }

    return _uhash_create(keyHash, keyComp, valueComp, i, status);
}

U_CAPI UHashtable* U_EXPORT2
uhash_init(UHashtable *fillinResult,
           UHashFunction *keyHash,
           UKeyComparator *keyComp,
           UValueComparator *valueComp,
           UErrorCode *status) {

    return _uhash_init(fillinResult, keyHash, keyComp, valueComp, DEFAULT_PRIME_INDEX, status);
}

U_CAPI UHashtable* U_EXPORT2
uhash_initSize(UHashtable *fillinResult,
               UHashFunction *keyHash,
               UKeyComparator *keyComp,
               UValueComparator *valueComp,
               int32_t size,
               UErrorCode *status) {

    // Find the smallest index i for which PRIMES[i] >= size.
    int32_t i = 0;
    while (i<(PRIMES_LENGTH-1) && PRIMES[i]<size) {
        ++i;
    }
    return _uhash_init(fillinResult, keyHash, keyComp, valueComp, i, status);
}

U_CAPI void U_EXPORT2
uhash_close(UHashtable *hash) {
    if (hash == nullptr) {
        return;
    }
    if (hash->elements != nullptr) {
        if (hash->keyDeleter != nullptr || hash->valueDeleter != nullptr) {
            int32_t pos=UHASH_FIRST;
            UHashElement *e;
            while ((e = (UHashElement*) uhash_nextElement(hash, &pos)) != nullptr) {
                HASH_DELETE_KEY_VALUE(hash, e->key.pointer, e->value.pointer);
            }
        }
        uprv_free(hash->elements);
        hash->elements = nullptr;
    }
    if (hash->allocated) {
        uprv_free(hash);
    }
}

U_CAPI UHashFunction *U_EXPORT2
uhash_setKeyHasher(UHashtable *hash, UHashFunction *fn) {
    UHashFunction *result = hash->keyHasher;
    hash->keyHasher = fn;
    return result;
}

U_CAPI UKeyComparator *U_EXPORT2
uhash_setKeyComparator(UHashtable *hash, UKeyComparator *fn) {
    UKeyComparator *result = hash->keyComparator;
    hash->keyComparator = fn;
    return result;
}
U_CAPI UValueComparator *U_EXPORT2
uhash_setValueComparator(UHashtable *hash, UValueComparator *fn){
    UValueComparator *result = hash->valueComparator;
    hash->valueComparator = fn;
    return result;
}

U_CAPI UObjectDeleter *U_EXPORT2
uhash_setKeyDeleter(UHashtable *hash, UObjectDeleter *fn) {
    UObjectDeleter *result = hash->keyDeleter;
    hash->keyDeleter = fn;
    return result;
}

U_CAPI UObjectDeleter *U_EXPORT2
uhash_setValueDeleter(UHashtable *hash, UObjectDeleter *fn) {
    UObjectDeleter *result = hash->valueDeleter;
    hash->valueDeleter = fn;
    return result;
}

U_CAPI void U_EXPORT2
uhash_setResizePolicy(UHashtable *hash, enum UHashResizePolicy policy) {
    UErrorCode status = U_ZERO_ERROR;
    _uhash_internalSetResizePolicy(hash, policy);
    hash->lowWaterMark  = (int32_t)(hash->length * hash->lowWaterRatio);
    hash->highWaterMark = (int32_t)(hash->length * hash->highWaterRatio);
    _uhash_rehash(hash, &status);
}

U_CAPI int32_t U_EXPORT2
uhash_count(const UHashtable *hash) {
    return hash->count;
}

U_CAPI void* U_EXPORT2
uhash_get(const UHashtable *hash,
          const void* key) {
    UHashTok keyholder;
    keyholder.pointer = (void*) key;
    return _uhash_find(hash, keyholder, hash->keyHasher(keyholder))->value.pointer;
}

U_CAPI void* U_EXPORT2
uhash_iget(const UHashtable *hash,
           int32_t key) {
    UHashTok keyholder;
    keyholder.integer = key;
    return _uhash_find(hash, keyholder, hash->keyHasher(keyholder))->value.pointer;
}

U_CAPI int32_t U_EXPORT2
uhash_geti(const UHashtable *hash,
           const void* key) {
    UHashTok keyholder;
    keyholder.pointer = (void*) key;
    return _uhash_find(hash, keyholder, hash->keyHasher(keyholder))->value.integer;
}

U_CAPI int32_t U_EXPORT2
uhash_igeti(const UHashtable *hash,
           int32_t key) {
    UHashTok keyholder;
    keyholder.integer = key;
    return _uhash_find(hash, keyholder, hash->keyHasher(keyholder))->value.integer;
}

U_CAPI int32_t U_EXPORT2
uhash_getiAndFound(const UHashtable *hash,
                   const void *key,
                   UBool *found) {
    UHashTok keyholder;
    keyholder.pointer = (void *)key;
    const UHashElement *e = _uhash_find(hash, keyholder, hash->keyHasher(keyholder));
    *found = !IS_EMPTY_OR_DELETED(e->hashcode);
    return e->value.integer;
}

U_CAPI int32_t U_EXPORT2
uhash_igetiAndFound(const UHashtable *hash,
                    int32_t key,
                    UBool *found) {
    UHashTok keyholder;
    keyholder.integer = key;
    const UHashElement *e = _uhash_find(hash, keyholder, hash->keyHasher(keyholder));
    *found = !IS_EMPTY_OR_DELETED(e->hashcode);
    return e->value.integer;
}

U_CAPI void* U_EXPORT2
uhash_put(UHashtable *hash,
          void* key,
          void* value,
          UErrorCode *status) {
    UHashTok keyholder, valueholder;
    keyholder.pointer = key;
    valueholder.pointer = value;
    return _uhash_put(hash, keyholder, valueholder,
                      HINT_KEY_POINTER | HINT_VALUE_POINTER,
                      status).pointer;
}

U_CAPI void* U_EXPORT2
uhash_iput(UHashtable *hash,
           int32_t key,
           void* value,
           UErrorCode *status) {
    UHashTok keyholder, valueholder;
    keyholder.integer = key;
    valueholder.pointer = value;
    return _uhash_put(hash, keyholder, valueholder,
                      HINT_VALUE_POINTER,
                      status).pointer;
}

U_CAPI int32_t U_EXPORT2
uhash_puti(UHashtable *hash,
           void* key,
           int32_t value,
           UErrorCode *status) {
    UHashTok keyholder, valueholder;
    keyholder.pointer = key;
    valueholder.integer = value;
    return _uhash_put(hash, keyholder, valueholder,
                      HINT_KEY_POINTER,
                      status).integer;
}


U_CAPI int32_t U_EXPORT2
uhash_iputi(UHashtable *hash,
           int32_t key,
           int32_t value,
           UErrorCode *status) {
    UHashTok keyholder, valueholder;
    keyholder.integer = key;
    valueholder.integer = value;
    return _uhash_put(hash, keyholder, valueholder,
                      HINT_BOTH_INTEGERS,
                      status).integer;
}

U_CAPI int32_t U_EXPORT2
uhash_putiAllowZero(UHashtable *hash,
                    void *key,
                    int32_t value,
                    UErrorCode *status) {
    UHashTok keyholder, valueholder;
    keyholder.pointer = key;
    valueholder.integer = value;
    return _uhash_put(hash, keyholder, valueholder,
                      HINT_KEY_POINTER | HINT_ALLOW_ZERO,
                      status).integer;
}


U_CAPI int32_t U_EXPORT2
uhash_iputiAllowZero(UHashtable *hash,
                     int32_t key,
                     int32_t value,
                     UErrorCode *status) {
    UHashTok keyholder, valueholder;
    keyholder.integer = key;
    valueholder.integer = value;
    return _uhash_put(hash, keyholder, valueholder,
                      HINT_BOTH_INTEGERS | HINT_ALLOW_ZERO,
                      status).integer;
}

U_CAPI void* U_EXPORT2
uhash_remove(UHashtable *hash,
             const void* key) {
    UHashTok keyholder;
    keyholder.pointer = (void*) key;
    return _uhash_remove(hash, keyholder).pointer;
}

U_CAPI void* U_EXPORT2
uhash_iremove(UHashtable *hash,
              int32_t key) {
    UHashTok keyholder;
    keyholder.integer = key;
    return _uhash_remove(hash, keyholder).pointer;
}

U_CAPI int32_t U_EXPORT2
uhash_removei(UHashtable *hash,
              const void* key) {
    UHashTok keyholder;
    keyholder.pointer = (void*) key;
    return _uhash_remove(hash, keyholder).integer;
}

U_CAPI int32_t U_EXPORT2
uhash_iremovei(UHashtable *hash,
               int32_t key) {
    UHashTok keyholder;
    keyholder.integer = key;
    return _uhash_remove(hash, keyholder).integer;
}

U_CAPI void U_EXPORT2
uhash_removeAll(UHashtable *hash) {
    int32_t pos = UHASH_FIRST;
    const UHashElement *e;
    U_ASSERT(hash != nullptr);
    if (hash->count != 0) {
        while ((e = uhash_nextElement(hash, &pos)) != nullptr) {
            uhash_removeElement(hash, e);
        }
    }
    U_ASSERT(hash->count == 0);
}

U_CAPI UBool U_EXPORT2
uhash_containsKey(const UHashtable *hash, const void *key) {
    UHashTok keyholder;
    keyholder.pointer = (void *)key;
    const UHashElement *e = _uhash_find(hash, keyholder, hash->keyHasher(keyholder));
    return !IS_EMPTY_OR_DELETED(e->hashcode);
}

/**
 * Returns true if the UHashtable contains an item with this integer key.
 *
 * @param hash The target UHashtable.
 * @param key An integer key stored in a hashtable
 * @return true if the key is found.
 */
U_CAPI UBool U_EXPORT2
uhash_icontainsKey(const UHashtable *hash, int32_t key) {
    UHashTok keyholder;
    keyholder.integer = key;
    const UHashElement *e = _uhash_find(hash, keyholder, hash->keyHasher(keyholder));
    return !IS_EMPTY_OR_DELETED(e->hashcode);
}

U_CAPI const UHashElement* U_EXPORT2
uhash_find(const UHashtable *hash, const void* key) {
    UHashTok keyholder;
    const UHashElement *e;
    keyholder.pointer = (void*) key;
    e = _uhash_find(hash, keyholder, hash->keyHasher(keyholder));
    return IS_EMPTY_OR_DELETED(e->hashcode) ? nullptr : e;
}

U_CAPI const UHashElement* U_EXPORT2
uhash_nextElement(const UHashtable *hash, int32_t *pos) {
    /* Walk through the array until we find an element that is not
     * EMPTY and not DELETED.
     */
    int32_t i;
    U_ASSERT(hash != nullptr);
    for (i = *pos + 1; i < hash->length; ++i) {
        if (!IS_EMPTY_OR_DELETED(hash->elements[i].hashcode)) {
            *pos = i;
            return &(hash->elements[i]);
        }
    }

    /* No more elements */
    return nullptr;
}

U_CAPI void* U_EXPORT2
uhash_removeElement(UHashtable *hash, const UHashElement* e) {
    U_ASSERT(hash != nullptr);
    U_ASSERT(e != nullptr);
    if (!IS_EMPTY_OR_DELETED(e->hashcode)) {
        UHashElement *nce = (UHashElement *)e;
        return _uhash_internalRemoveElement(hash, nce).pointer;
    }
    return nullptr;
}

/********************************************************************
 * UHashTok convenience
 ********************************************************************/

/**
 * Return a UHashTok for an integer.
 */
/*U_CAPI UHashTok U_EXPORT2
uhash_toki(int32_t i) {
    UHashTok tok;
    tok.integer = i;
    return tok;
}*/

/**
 * Return a UHashTok for a pointer.
 */
/*U_CAPI UHashTok U_EXPORT2
uhash_tokp(void* p) {
    UHashTok tok;
    tok.pointer = p;
    return tok;
}*/

/********************************************************************
 * PUBLIC Key Hash Functions
 ********************************************************************/

U_CAPI int32_t U_EXPORT2
uhash_hashUChars(const UHashTok key) {
    const char16_t *s = (const char16_t *)key.pointer;
    return s == nullptr ? 0 : ustr_hashUCharsN(s, u_strlen(s));
}

U_CAPI int32_t U_EXPORT2
uhash_hashChars(const UHashTok key) {
    const char *s = (const char *)key.pointer;
    return s == nullptr ? 0 : static_cast<int32_t>(ustr_hashCharsN(s, static_cast<int32_t>(uprv_strlen(s))));
}

U_CAPI int32_t U_EXPORT2
uhash_hashIChars(const UHashTok key) {
    const char *s = (const char *)key.pointer;
    return s == nullptr ? 0 : ustr_hashICharsN(s, static_cast<int32_t>(uprv_strlen(s)));
}

U_CAPI UBool U_EXPORT2
uhash_equals(const UHashtable* hash1, const UHashtable* hash2){
    int32_t count1, count2, pos, i;

    if(hash1==hash2){
        return true;
    }

    /*
     * Make sure that we are comparing 2 valid hashes of the same type
     * with valid comparison functions.
     * Without valid comparison functions, a binary comparison
     * of the hash values will yield random results on machines
     * with 64-bit pointers and 32-bit integer hashes.
     * A valueComparator is normally optional.
     */
    if (hash1==nullptr || hash2==nullptr ||
        hash1->keyComparator != hash2->keyComparator ||
        hash1->valueComparator != hash2->valueComparator ||
        hash1->valueComparator == nullptr)
    {
        /*
        Normally we would return an error here about incompatible hash tables,
        but we return false instead.
        */
        return false;
    }

    count1 = uhash_count(hash1);
    count2 = uhash_count(hash2);
    if(count1!=count2){
        return false;
    }

    pos=UHASH_FIRST;
    for(i=0; i<count1; i++){
        const UHashElement* elem1 = uhash_nextElement(hash1, &pos);
        const UHashTok key1 = elem1->key;
        const UHashTok val1 = elem1->value;
        /* here the keys are not compared, instead the key form hash1 is used to fetch
         * value from hash2. If the hashes are equal then then both hashes should
         * contain equal values for the same key!
         */
        const UHashElement* elem2 = _uhash_find(hash2, key1, hash2->keyHasher(key1));
        const UHashTok val2 = elem2->value;
        if(hash1->valueComparator(val1, val2)==false){
            return false;
        }
    }
    return true;
}

/********************************************************************
 * PUBLIC Comparator Functions
 ********************************************************************/

U_CAPI UBool U_EXPORT2
uhash_compareUChars(const UHashTok key1, const UHashTok key2) {
    const char16_t *p1 = (const char16_t*) key1.pointer;
    const char16_t *p2 = (const char16_t*) key2.pointer;
    if (p1 == p2) {
        return true;
    }
    if (p1 == nullptr || p2 == nullptr) {
        return false;
    }
    while (*p1 != 0 && *p1 == *p2) {
        ++p1;
        ++p2;
    }
    return (UBool)(*p1 == *p2);
}

U_CAPI UBool U_EXPORT2
uhash_compareChars(const UHashTok key1, const UHashTok key2) {
    const char *p1 = (const char*) key1.pointer;
    const char *p2 = (const char*) key2.pointer;
    if (p1 == p2) {
        return true;
    }
    if (p1 == nullptr || p2 == nullptr) {
        return false;
    }
    while (*p1 != 0 && *p1 == *p2) {
        ++p1;
        ++p2;
    }
    return (UBool)(*p1 == *p2);
}

U_CAPI UBool U_EXPORT2
uhash_compareIChars(const UHashTok key1, const UHashTok key2) {
    const char *p1 = (const char*) key1.pointer;
    const char *p2 = (const char*) key2.pointer;
    if (p1 == p2) {
        return true;
    }
    if (p1 == nullptr || p2 == nullptr) {
        return false;
    }
    while (*p1 != 0 && uprv_tolower(*p1) == uprv_tolower(*p2)) {
        ++p1;
        ++p2;
    }
    return (UBool)(*p1 == *p2);
}

/********************************************************************
 * PUBLIC int32_t Support Functions
 ********************************************************************/

U_CAPI int32_t U_EXPORT2
uhash_hashLong(const UHashTok key) {
    return key.integer;
}

U_CAPI UBool U_EXPORT2
uhash_compareLong(const UHashTok key1, const UHashTok key2) {
    return (UBool)(key1.integer == key2.integer);
}