-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathcspp_wbwi.cc
1341 lines (1332 loc) · 45 KB
/
cspp_wbwi.cc
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
// Copyright (c) 2021-present, Topling, Inc. All rights reserved.
// Created by leipeng, fully rewrite by leipeng 2021-05-12
#include "db/column_family.h"
#include "db/db_impl/db_impl.h"
#include "db/merge_context.h"
#include "db/merge_helper.h"
#include "options/db_options.h"
#include "rocksdb/comparator.h"
#include "rocksdb/iterator.h"
#include "util/cast_util.h"
#include "util/string_util.h"
#include <logging/logging.h>
#include <rocksdb/utilities/write_batch_with_index.h>
#include <utilities/write_batch_with_index/write_batch_with_index_internal.h>
#include <topling/side_plugin_factory.h>
#if defined(_MSC_VER)
#pragma warning(disable: 4245) // convert int to size_t in fsa of cspp
#endif
#include <terark/fsa/cspptrie.inl>
#include <terark/io/DataIO_Basic.hpp>
#include <terark/num_to_str.hpp>
const char* git_version_hash_info_cspp_wbwi();
namespace ROCKSDB_NAMESPACE {
using namespace terark;
// prepend bytewise(bigendian) cf_id on userkey
static fstring InitLookupKey(void* alloca_ptr, uint32_t cf_id, Slice userkey) {
fstring lookup_key((char*)alloca_ptr, 4 + userkey.size_);
aligned_save(alloca_ptr, BIG_ENDIAN_OF(cf_id));
memcpy((char*)alloca_ptr + 4, userkey.data_, userkey.size_);
return lookup_key;
}
#define DefineLookupKey(var, cf_id, userkey) \
const fstring var = InitLookupKey(alloca(4 + userkey.size_), cf_id, userkey)
struct CSPP_WBWIFactory;
struct CSPP_WBWI : public WriteBatchWithIndex {
using Elem = uint32_t;
struct VecNode {
uint32_t num;
uint32_t pos;
};
CSPP_WBWIFactory* m_fac;
mutable MainPatricia m_trie;
mutable Patricia::SingleWriterToken m_wtoken;
ReadableWriteBatch m_batch;
size_t m_max_cap;
bool m_overwrite_key;
uint32_t m_live_iter_num = 0;
size_t m_last_entry_offset = 0;
size_t m_last_sub_batch_offset = 0;
size_t m_sub_batch_cnt = 1;
const Comparator* m_default_cmp;
valvec<const Comparator*> m_cf_cmp{8, valvec_reserve()};
CSPP_WBWI(CSPP_WBWIFactory*, bool overwrite_key, const Comparator*, size_t);
~CSPP_WBWI() noexcept override;
void SetLastEntryOffset() {
m_last_entry_offset = m_batch.GetDataSize();
}
const Comparator* GetUserComparator(uint32_t cf_id) const final {
if (cf_id < m_cf_cmp.size() && m_cf_cmp[cf_id]) {
return m_cf_cmp[cf_id];
}
return m_default_cmp;
}
void AddOrUpdateIndexCFH(ColumnFamilyHandle* cfh, WriteType type) {
if (cfh) {
ROCKSDB_ASSERT_F(!cfh->GetComparator() ||
IsBytewiseComparator(cfh->GetComparator()),
"Name() = %s", cfh->GetComparator()->Name());
const uint32_t cf_id = cfh->GetID();
const Comparator*& cmp = m_cf_cmp.ensure_get(cf_id);
if (nullptr == cmp) {
cmp = cfh->GetComparator();
} else {
ROCKSDB_ASSERT_EQ(cmp, cfh->GetComparator());
}
AddOrUpdateIndex(cf_id, type);
} else {
AddOrUpdateIndex(0, type);
}
}
void AddOrUpdateIndex(uint32_t cf_id, WriteType type) {
size_t offset = m_last_entry_offset;
Slice raw_entry = Slice(m_batch.Data()).substr(offset), userkey;
bool success __attribute__((unused)) =
ReadKeyFromWriteBatchEntry(&raw_entry, &userkey, cf_id != 0);
assert(success);
DefineLookupKey(lookup_key, cf_id, userkey);
VecNode vn = {0,0};
if (m_trie.insert(lookup_key, &vn, &m_wtoken)) {
vn.num = 1;
vn.pos = uint32_t(m_trie.mem_alloc(sizeof(Elem)));
*(Elem*)m_trie.mem_get(vn.pos) = Elem(offset);
m_trie.mutable_value_of<VecNode>(m_wtoken) = vn;
}
else { // dup key, append on vector or overwirte last vector elem
vn = m_trie.value_of<VecNode>(m_wtoken);
auto vec = (Elem*)m_trie.mem_get(vn.pos);
if (LIKELY(m_last_sub_batch_offset <= vec[vn.num-1])) {
m_last_sub_batch_offset = m_last_entry_offset;
m_sub_batch_cnt++;
}
if (m_overwrite_key && kMergeRecord != type) {
vec[vn.num-1] = Elem(offset); // overwrite
}
else {
if (vn.num & (vn.num-1)) { // is not power of 2, has space
vec[vn.num] = Elem(offset);
m_trie.mutable_value_of<VecNode>(m_wtoken).num = vn.num + 1;
}
else {
size_t oldlen = sizeof(Elem) * vn.num;
size_t newlen = sizeof(Elem) * vn.num * 2;
size_t newpos = m_trie.mem_alloc3(vn.pos, oldlen, newlen);
vn.pos = (uint32_t)newpos;
vec = (Elem*)m_trie.mem_get(newpos);
vec[vn.num++] = Elem(offset);
m_trie.mutable_value_of<VecNode>(m_wtoken) = vn;
}
}
}
}
using WriteBatchBase::GetWriteBatch;
WriteBatch* GetWriteBatch() final { return &m_batch; }
void Clear() final {
m_batch.Clear();
ClearIndex();
}
void ClearIndex();
static WriteType WriteTypeOf(ValueType op) {
switch (op) {
case kTypeColumnFamilyValue:
case kTypeValue:
return kPutRecord;
case kTypeColumnFamilyDeletion:
case kTypeDeletion:
return kDeleteRecord;
case kTypeColumnFamilySingleDeletion:
case kTypeSingleDeletion:
return kSingleDeleteRecord;
case kTypeColumnFamilyMerge:
case kTypeMerge:
return kMergeRecord;
case kTypeLogData:
return kLogDataRecord;
case kTypeColumnFamilyWideColumnEntity:
case kTypeWideColumnEntity:
#if (ROCKSDB_MAJOR * 10000 + ROCKSDB_MINOR * 10 + ROCKSDB_PATCH) >= 80100
return kPutEntityRecord;
#else
ROCKSDB_DIE("Not Supported: must be version 8.10+");
#endif
break;
case kTypeBeginPrepareXID:
case kTypeBeginPersistedPrepareXID:
case kTypeBeginUnprepareXID:
case kTypeEndPrepareXID:
case kTypeCommitXID:
case kTypeCommitXIDAndTimestamp:
case kTypeRollbackXID:
case kTypeNoop:
//return kXIDRecord;
default:
return kUnknownRecord;
}
}
Status ReBuildIndex() {
Status s;
ClearIndex();
if (m_batch.Count() == 0) {
// Nothing to re-index
return s;
}
size_t offset = WriteBatchInternal::GetFirstOffset(&m_batch);
Slice input(m_batch.Data());
input.remove_prefix(offset);
size_t found = 0;
while (s.ok() && !input.empty()) {
Slice key, value, blob, xid;
uint32_t cf_id = 0; // default
char tag = 0;
m_last_entry_offset = input.data() - m_batch.Data().data();
s = ReadRecordFromWriteBatch(&input, &tag, &cf_id, &key,
&value, &blob, &xid);
if (!s.ok()) {
break;
}
switch (tag) {
case kTypeColumnFamilyValue:
case kTypeValue:
found++;
AddOrUpdateIndex(cf_id, kPutRecord);
break;
case kTypeColumnFamilyDeletion:
case kTypeDeletion:
found++;
AddOrUpdateIndex(cf_id, kDeleteRecord);
break;
case kTypeColumnFamilySingleDeletion:
case kTypeSingleDeletion:
found++;
AddOrUpdateIndex(cf_id, kSingleDeleteRecord);
break;
case kTypeColumnFamilyMerge:
case kTypeMerge:
found++;
AddOrUpdateIndex(cf_id, kMergeRecord);
break;
case kTypeLogData:
case kTypeBeginPrepareXID:
case kTypeBeginPersistedPrepareXID:
case kTypeBeginUnprepareXID:
case kTypeEndPrepareXID:
case kTypeCommitXID:
case kTypeCommitXIDAndTimestamp:
case kTypeRollbackXID:
case kTypeNoop:
break;
case kTypeColumnFamilyWideColumnEntity:
case kTypeWideColumnEntity:
#if (ROCKSDB_MAJOR * 10000 + ROCKSDB_MINOR * 10 + ROCKSDB_PATCH) >= 80100
found++;
AddOrUpdateIndex(cf_id, kPutEntityRecord);
#else
ROCKSDB_DIE("Not Supported: must be version 8.10+");
#endif
break;
default:
return Status::Corruption("unknown WriteBatch tag in ReBuildIndex",
std::to_string(size_t(tag)));
}
}
if (s.ok() && found != m_batch.Count()) {
s = Status::Corruption("WriteBatch has wrong count");
}
return s;
}
struct OneRecord {
ValueType tag;
WriteType type;
uint32_t cf_id;
Slice key, value, blob, xid;
const char* Read(const char* input);
};
void ReadRecord(size_t offset, OneRecord* p) const {
#if 0
OneRecord& r = *p;
Slice input = Slice(m_batch.Data()).substr(offset);
static_assert(sizeof(r.tag) == 1);
Status s = ReadRecordFromWriteBatch(&input, (char*)&r.tag, &r.cf_id, &r.key,
&r.value, &r.blob, &r.xid);
TERARK_VERIFY_S(s.ok(), "%s", s.ToString());
r.type = WriteTypeOf(r.tag);
#else
auto rec_end __attribute__((__unused__)) = p->Read(m_batch.Data().data() + offset);
ROCKSDB_ASSERT_LE(rec_end, Slice(m_batch.Data()).end());
#endif
}
#define CHECK_BATCH_SPACE_1(key) \
if (UNLIKELY(m_batch.GetDataSize() + key.size_ + 8192 > m_max_cap)) { \
char msg[1024]; \
auto len = snprintf(msg, sizeof(msg), \
"%s:%d: too large batch = %zd, " ROCKS_LOG_TOSTRING(key) " = %zd : %s", \
RocksLogShorterFileName(__FILE__), __LINE__, \
m_batch.GetDataSize(), key.size_, \
BOOST_CURRENT_FUNCTION); \
return Status::InvalidArgument(Slice(msg, len)); \
}
#define CHECK_BATCH_SPACE_2(key, value) \
if (UNLIKELY(m_batch.GetDataSize() + key.size_ + value.size_ + 8192 > m_max_cap)) { \
char msg[1024]; \
auto len = snprintf(msg, sizeof(msg), \
"%s:%d: too large batch = %zd, key = %zd, value = %zd : %s", \
RocksLogShorterFileName(__FILE__), __LINE__, \
m_batch.GetDataSize(), key.size_, value.size_, \
BOOST_CURRENT_FUNCTION); \
return Status::InvalidArgument(Slice(msg, len)); \
}
//~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
using WriteBatchWithIndex::Put;
Status Put(ColumnFamilyHandle* cfh, const Slice& key, const Slice& value) final {
CHECK_BATCH_SPACE_2(key, value);
SetLastEntryOffset();
auto s = m_batch.Put(cfh, key, value);
if (s.ok()) {
AddOrUpdateIndexCFH(cfh, kPutRecord);
}
return s;
}
Status Put(const Slice& key, const Slice& value) final {
CHECK_BATCH_SPACE_2(key, value);
SetLastEntryOffset();
auto s = m_batch.Put(key, value);
if (s.ok()) {
AddOrUpdateIndex(0, kPutRecord);
}
return s;
}
#if (ROCKSDB_MAJOR * 10000 + ROCKSDB_MINOR * 10 + ROCKSDB_PATCH) >= 80100
using WriteBatchWithIndex::PutEntity;
Status PutEntity(ColumnFamilyHandle* cfh, const Slice& key,
const WideColumns& columns) override {
struct dummy { size_t size_; } vsize{0}; // size_ only
for (auto& x : columns)
vsize.size_ += x.name().size() + x.value().size() + 10;
CHECK_BATCH_SPACE_2(key, vsize);
SetLastEntryOffset();
auto s = m_batch.PutEntity(cfh, key, columns);
if (s.ok()) {
AddOrUpdateIndexCFH(cfh, kPutEntityRecord);
}
return s;
}
#endif
using WriteBatchWithIndex::Delete;
Status Delete(ColumnFamilyHandle* cfh, const Slice& key) final {
CHECK_BATCH_SPACE_1(key);
SetLastEntryOffset();
auto s = m_batch.Delete(cfh, key);
if (s.ok()) {
AddOrUpdateIndexCFH(cfh, kDeleteRecord);
}
return s;
}
Status Delete(const Slice& key) final {
CHECK_BATCH_SPACE_1(key);
SetLastEntryOffset();
auto s = m_batch.Delete(key);
if (s.ok()) {
AddOrUpdateIndex(0, kDeleteRecord);
}
return s;
}
using WriteBatchWithIndex::SingleDelete;
Status SingleDelete(ColumnFamilyHandle* cfh, const Slice& key) final {
CHECK_BATCH_SPACE_1(key);
SetLastEntryOffset();
auto s = m_batch.SingleDelete(cfh, key);
if (s.ok()) {
AddOrUpdateIndexCFH(cfh, kSingleDeleteRecord);
}
return s;
}
Status SingleDelete(const Slice& key) final {
CHECK_BATCH_SPACE_1(key);
SetLastEntryOffset();
auto s = m_batch.SingleDelete(key);
if (s.ok()) {
AddOrUpdateIndex(0, kSingleDeleteRecord);
}
return s;
}
using WriteBatchWithIndex::Merge;
Status Merge(ColumnFamilyHandle* cfh, const Slice& key, const Slice& value) final {
CHECK_BATCH_SPACE_2(key, value);
SetLastEntryOffset();
auto s = m_batch.Merge(cfh, key, value);
if (s.ok()) {
AddOrUpdateIndexCFH(cfh, kMergeRecord);
}
return s;
}
Status Merge(const Slice& key, const Slice& value) final {
CHECK_BATCH_SPACE_2(key, value);
SetLastEntryOffset();
auto s = m_batch.Merge(key, value);
if (s.ok()) {
AddOrUpdateIndex(0, kMergeRecord);
}
return s;
}
Status PutLogData(const Slice& blob) final {
CHECK_BATCH_SPACE_1(blob);
return m_batch.PutLogData(blob);
}
void SetSavePoint() final { m_batch.SetSavePoint(); }
Status RollbackToSavePoint() final {
Status s = m_batch.RollbackToSavePoint();
if (s.ok()) {
m_sub_batch_cnt = 1;
m_last_sub_batch_offset = 0;
s = ReBuildIndex();
}
return s;
}
Status PopSavePoint() final { return m_batch.PopSavePoint(); }
void SetMaxBytes(size_t max_bytes) final { m_batch.SetMaxBytes(max_bytes); }
size_t GetDataSize() const final { return m_batch.GetDataSize(); }
size_t SubBatchCnt() final { return m_sub_batch_cnt; }
WBWIIterator::Result
FetchFromBatch(ColumnFamilyHandle* cfh, const Slice& userkey,
Slice* newest_put, MergeContext* mgcontext) {
if (0 == m_last_entry_offset) {
return WBWIIterator::kNotFound;
}
uint32_t cf_id = cfh ? cfh->GetID() : 0;
DefineLookupKey(lookup_key, cf_id, userkey);
// wtoken can also used for read
if (!m_trie.lookup(lookup_key, &m_wtoken)) {
return WBWIIterator::kNotFound;
}
auto vn = m_trie.value_of<VecNode>(m_wtoken);
auto vec = (const Elem*)m_trie.mem_get(vn.pos);
OneRecord rec;
for (size_t idx = vn.num; idx; ) {
idx--;
ReadRecord(vec[idx], &rec);
switch (rec.type) {
case kPutRecord:
*newest_put = rec.value;
return WBWIIterator::kFound;
#if (ROCKSDB_MAJOR * 10000 + ROCKSDB_MINOR * 10 + ROCKSDB_PATCH) >= 80100
case kPutEntityRecord:
*newest_put = rec.value;
return WBWIIterator::kFoundEntity;
#endif
case kDeleteRecord:
case kSingleDeleteRecord:
return WBWIIterator::kDeleted;
case kMergeRecord:
mgcontext->PushOperand(rec.value);
break;
case kLogDataRecord:
break; // ignore
case kXIDRecord:
break; // ignore
default:
return WBWIIterator::kError;
} // end switch statement
}
return WBWIIterator::kMergeInProgress;
}
Status GetFromBatch(ColumnFamilyHandle* cfh, const DBOptions& options,
const Slice& key, std::string* value) override {
MergeContext mgcontext;
Slice newest_put;
auto result = FetchFromBatch(cfh, key, &newest_put, &mgcontext);
Status st;
value->clear();
switch (result) {
case WBWIIterator::kFound:
if (mgcontext.GetNumOperands() > 0)
st = MergeKey(options, cfh, key, &newest_put, value, mgcontext);
else
value->assign(newest_put.data_, newest_put.size_);
break;
#if (ROCKSDB_MAJOR * 10000 + ROCKSDB_MINOR * 10 + ROCKSDB_PATCH) >= 80100
case WBWIIterator::kFoundEntity:
if (mgcontext.GetNumOperands() > 0)
st = WriteBatchWithIndexInternal::MergeKeyWithBaseValue(
cfh, key, MergeHelper::kWideBaseValue, newest_put,
mgcontext, value, static_cast<PinnableWideColumns*>(nullptr));
else {
Slice dv; // deserialized default column value
st = WideColumnSerialization::GetValueOfDefaultColumn(newest_put, dv);
if (st.ok())
value->assign(dv.data(), dv.size());
}
break;
#endif
case WBWIIterator::kError:
st = Status::Corruption("CSPP_WBWI::FetchFromBatch returned error");
break;
case WBWIIterator::kDeleted:
if (mgcontext.GetNumOperands() > 0)
st = MergeKey(options, cfh, key, nullptr, value, mgcontext);
else
st = Status::NotFound();
break;
case WBWIIterator::kMergeInProgress:
MergeKey(options, cfh, key, nullptr, value, mgcontext);
st = Status::MergeInProgress(); // rocksdb uint test assert this
break;
case WBWIIterator::kNotFound:
st = Status::NotFound();
break;
default:
ROCKSDB_DIE("Unexpected: result = %d", result);
}
return st;
}
// semantically same with WriteBatchWithIndexInternal::GetFromBatch
WBWIIterator::Result
GetFromBatchRaw(DB* db, ColumnFamilyHandle* cfh, const Slice& key,
MergeContext* mgcontext, std::string* value, Status* s)
override {
*s = Status::OK();
value->clear();
Slice newest_put;
auto result = FetchFromBatch(cfh, key, &newest_put, mgcontext);
switch (result) {
case WBWIIterator::kFound:
if (mgcontext->GetNumOperands() > 0) {
*s = MergeKey(db, cfh, key, &newest_put, value, *mgcontext);
if (!s->ok())
result = WBWIIterator::Result::kError;
}
else {
value->assign(newest_put.data_, newest_put.size_);
}
break;
#if (ROCKSDB_MAJOR * 10000 + ROCKSDB_MINOR * 10 + ROCKSDB_PATCH) >= 80100
case WBWIIterator::kFoundEntity:
if (mgcontext->GetNumOperands() > 0) {
*s = WriteBatchWithIndexInternal::MergeKeyWithBaseValue(
cfh, key, MergeHelper::kWideBaseValue, newest_put,
*mgcontext, value, static_cast<PinnableWideColumns*>(nullptr));
} else {
Slice dv; // deserialized default column value
*s = WideColumnSerialization::GetValueOfDefaultColumn(newest_put, dv);
if (s->ok())
value->assign(dv.data(), dv.size());
}
result = s->ok() ? WBWIIterator::kFound : WBWIIterator::kError;
break;
#endif
case WBWIIterator::kError:
*s = Status::Corruption("CSPP_WBWI::FetchFromBatch returned error");
break;
case WBWIIterator::kDeleted:
if (mgcontext->GetNumOperands() > 0) {
*s = MergeKey(db, cfh, key, nullptr, value, *mgcontext);
if (s->ok())
result = WBWIIterator::Result::kFound;
else
result = WBWIIterator::Result::kError;
}
break;
case WBWIIterator::kMergeInProgress:
case WBWIIterator::kNotFound:
break;
default:
ROCKSDB_DIE("Unexpected: result = %d", result);
}
return result;
}
Status GetFromBatchAndDB(DB* db, const ReadOptions& read_options,
ColumnFamilyHandle* cfh, const Slice& key,
PinnableSlice* pinnable_val, ReadCallback* callback)
override {
MergeContext mgcontext;
// Since the lifetime of the WriteBatch is the same as that of the transaction
// we cannot pin it as otherwise the returned value will not be available
// after the transaction finishes.
std::string& batch_value = *pinnable_val->GetSelf();
Slice newest_put;
auto result = FetchFromBatch(cfh, key, &newest_put, &mgcontext);
Status st;
switch (result) {
case WBWIIterator::kFound:
if (mgcontext.GetNumOperands() > 0) {
st = MergeKey(db, cfh, key, &newest_put, &batch_value, mgcontext);
if (!st.ok())
return st;
}
else {
batch_value.assign(newest_put.data_, newest_put.size_);
}
pinnable_val->PinSelf();
break;
#if (ROCKSDB_MAJOR * 10000 + ROCKSDB_MINOR * 10 + ROCKSDB_PATCH) >= 80100
case WBWIIterator::kFoundEntity:
if (mgcontext.GetNumOperands() > 0) {
st = WriteBatchWithIndexInternal::MergeKeyWithBaseValue(
cfh, key, MergeHelper::kWideBaseValue, newest_put,
mgcontext, &batch_value, static_cast<PinnableWideColumns*>(nullptr));
if (!st.ok())
return st;
}
else {
Slice dv; // deserialized default column value
st = WideColumnSerialization::GetValueOfDefaultColumn(newest_put, dv);
if (!st.ok())
return st;
batch_value.assign(dv.data(), dv.size());
}
pinnable_val->PinSelf();
break;
#endif
case WBWIIterator::kError:
st = Status::Corruption("CSPP_WBWI::FetchFromBatch returned error");
break;
case WBWIIterator::kDeleted:
if (mgcontext.GetNumOperands() > 0)
{
st = MergeKey(db, cfh, key, nullptr, &batch_value, mgcontext);
if (st.ok())
pinnable_val->PinSelf();
}
else
st = Status::NotFound();
break;
case WBWIIterator::kMergeInProgress:
case WBWIIterator::kNotFound:
{
// Did not find key in batch OR could not resolve Merges. Try DB.
DBImpl::GetImplOptions get_impl_options;
get_impl_options.column_family = cfh;
get_impl_options.value = pinnable_val;
get_impl_options.callback = callback;
#if (ROCKSDB_MAJOR * 10000 + ROCKSDB_MINOR * 10 + ROCKSDB_PATCH) >= 80100
PinnableWideColumns existing;
if (result == WBWIIteratorImpl::kMergeInProgress) {
get_impl_options.value = nullptr;
get_impl_options.columns = &existing;
}
#endif
auto root_db = static_cast<DBImpl*>(db->GetRootDB());
st = root_db->GetImpl(read_options, key, get_impl_options);
if (result == WBWIIterator::kMergeInProgress) {
if (st.ok() || st.IsNotFound()) { // DB Get Succeeded
// Merge result from DB with merges in Batch
#if (ROCKSDB_MAJOR * 10000 + ROCKSDB_MINOR * 10 + ROCKSDB_PATCH) >= 80100
MergeAcrossBatchAndDB(cfh, key, existing, mgcontext, pinnable_val, &st);
#else
std::string merge_result;
if (st.ok()) {
st = MergeKey(db, cfh, key, pinnable_val, &merge_result, mgcontext);
} else { // Key not present in db (s.IsNotFound())
st = MergeKey(db, cfh, key, nullptr, &merge_result, mgcontext);
}
if (st.ok()) {
pinnable_val->Reset();
pinnable_val->GetSelf()->assign(std::move(merge_result));
pinnable_val->PinSelf();
}
#endif
}
}
} // case
break;
default:
ROCKSDB_DIE("Unexpected: result = %d", result);
}
return st;
}
WBWIIterator* NewIterator(ColumnFamilyHandle* column_family) final;
WBWIIterator* NewIterator() final;
Iterator* NewIteratorWithBase(ColumnFamilyHandle*, Iterator* base,
const ReadOptions*) final;
Iterator* NewIteratorWithBase(Iterator* base) final; // default cf
struct Iter;
struct IterLinkNode {
IterLinkNode* m_prev;
IterLinkNode* m_next;
};
IterLinkNode m_head;
};
struct CSPP_WBWI::Iter : WBWIIterator, IterLinkNode, boost::noncopyable {
Patricia::Iterator* m_iter;
const Slice* m_lower_bound = nullptr;
const Slice* m_upper_bound = nullptr;
CSPP_WBWI* m_tab;
uint32_t m_cf_id = 0;
bool m_is_forward_cmp = false;
int m_idx = -1;
int m_num = 0;
size_t m_last_entry_offset;
const Elem* m_vec = nullptr;
OneRecord m_rec;
explicit Iter(CSPP_WBWI*, uint32_t cf_id, const Comparator* cmp);
~Iter() noexcept override;
uint32_t iter_cf_id() const {
auto bigendian_cf_id = *(const uint32_t*)m_iter->word().data();
return NATIVE_OF_BIG_ENDIAN(bigendian_cf_id);
}
void SetFirstEntry() {
auto vn = m_tab->m_trie.value_of<VecNode>(*m_iter);
m_idx = 0;
m_num = vn.num;
m_vec = (Elem*)m_tab->m_trie.mem_get(vn.pos);
m_tab->ReadRecord(m_vec[0], &m_rec);
assert(iter_cf_id() == m_cf_id);
assert(m_iter->word().substr(4) == m_rec.key);
}
void SetLastEntry() {
auto vn = m_tab->m_trie.value_of<VecNode>(*m_iter);
m_idx = vn.num - 1;
m_num = vn.num;
m_vec = (Elem*)m_tab->m_trie.mem_get(vn.pos);
m_tab->ReadRecord(m_vec[m_idx], &m_rec);
assert(iter_cf_id() == m_cf_id);
assert(m_iter->word().substr(4) == m_rec.key);
}
Status status() const final { return Status::OK(); }
bool Valid() const final { return m_idx >= 0; }
template<bool UpdateRecordCache>
terark_forceinline void CheckUpdates() {
TERARK_ASSERT_GE(m_idx, 0);
if (UNLIKELY(m_last_entry_offset != m_tab->m_last_entry_offset)) {
// CSPP_WBWI has changed
m_last_entry_offset = m_tab->m_last_entry_offset;
const fstring src = m_iter->word();
const fstring key((char*)memcpy(alloca(src.n), src.p, src.n), src.n);
ROCKSDB_VERIFY(m_iter->seek_lower_bound(key));
ROCKSDB_VERIFY(m_iter->word() == key);
auto vn = m_tab->m_trie.value_of<VecNode>(*m_iter);
ROCKSDB_ASSERT_LE(m_num, int(vn.num));
m_vec = (Elem*)m_tab->m_trie.mem_get(vn.pos);
m_num = vn.num;
if (UpdateRecordCache)
m_tab->ReadRecord(m_vec[m_idx], &m_rec);
}
}
WriteEntry Entry() const final {
TERARK_ASSERT_BT(m_idx, 0, m_num);
const_cast<Iter*>(this)->CheckUpdates<true>();
return {m_rec.type, m_rec.key, m_rec.value};
}
Slice user_key() const final {
TERARK_ASSERT_BT(m_idx, 0, m_num);
fstring k = m_iter->word();
assert(Slice(k.p + 4, k.n - 4) == Entry().key);
return Slice(k.p + 4, k.n - 4);
}
Slice user_key_no_assert() const {
fstring k = m_iter->word();
TERARK_ASSERT_GE(k.n, 4);
return Slice(k.p + 4, k.n - 4);
}
terark_forceinline bool CmpOrderNext() {
if (m_is_forward_cmp)
return m_iter->incr();
else
return m_iter->decr();
}
terark_forceinline bool CmpOrderPrev() {
if (m_is_forward_cmp)
return m_iter->decr();
else
return m_iter->incr();
}
terark_forceinline bool LT(Slice x, Slice y) const {
if (m_is_forward_cmp)
return x < y;
else
return y < x;
}
terark_forceinline bool GT(Slice x, Slice y) const {
if (m_is_forward_cmp)
return x > y;
else
return y > x;
}
terark_forceinline bool GE(Slice x, Slice y) const {
if (m_is_forward_cmp)
return x >= y;
else
return y >= x;
}
void Next() final {
TERARK_ASSERT_GE(m_idx, 0);
CheckUpdates<false>();
if (++m_idx == m_num) {
if (UNLIKELY(!CmpOrderNext() || iter_cf_id() != m_cf_id)) {
m_idx = -1;
return; // fail
}
if (m_upper_bound && GE(user_key_no_assert(), *m_upper_bound)) {
m_idx = -1;
return; // fail
}
auto vn = m_tab->m_trie.value_of<VecNode>(*m_iter);
m_idx = 0;
m_num = vn.num;
m_vec = (Elem*)m_tab->m_trie.mem_get(vn.pos);
}
m_tab->ReadRecord(m_vec[m_idx], &m_rec);
}
void Prev() final {
TERARK_ASSERT_GE(m_idx, 0);
CheckUpdates<false>();
if (m_idx-- == 0) {
if (UNLIKELY(!CmpOrderPrev() || iter_cf_id() != m_cf_id))
return; // fail
if (m_lower_bound && LT(user_key_no_assert(), *m_lower_bound))
return; // fail
auto vn = m_tab->m_trie.value_of<VecNode>(*m_iter);
m_idx = vn.num - 1;
m_num = vn.num;
m_vec = (Elem*)m_tab->m_trie.mem_get(vn.pos);
}
m_tab->ReadRecord(m_vec[m_idx], &m_rec);
}
struct ForwardLowerBound {
bool operator()(Patricia::Iterator* iter, fstring key) const
{ return iter->seek_lower_bound(key); }
};
struct ReverseLowerBound {
bool operator()(Patricia::Iterator* iter, fstring key) const
{ return iter->seek_rev_lower_bound(key); }
};
void Seek(const Slice& userkey) final {
if (m_is_forward_cmp)
SeekForward(userkey, ForwardLowerBound());
else
SeekForward(userkey, ReverseLowerBound());
}
template<class LowerBound>
void SeekForward(const Slice& userkey, LowerBound lower_bound_fn) {
m_idx = -1;
m_last_entry_offset = m_tab->m_last_entry_offset;
if (0 == m_last_entry_offset) return;
if (UNLIKELY(!m_iter)) {
m_iter = m_tab->m_trie.new_iter();
}
Slice seek_key = userkey;
if (m_lower_bound && LT(seek_key, *m_lower_bound)) {
seek_key = *m_lower_bound;
}
if (m_upper_bound && GE(seek_key, *m_upper_bound)) {
return; // fail
}
DefineLookupKey(lookup_key, m_cf_id, seek_key);
if (UNLIKELY(!lower_bound_fn(m_iter, lookup_key))) {
return; // fail
}
if (UNLIKELY(iter_cf_id() != m_cf_id)) {
return; // fail
}
if (m_upper_bound && GE(user_key_no_assert(), *m_upper_bound)) {
return; // fail
}
SetFirstEntry();
}
void SeekForPrev(const Slice& userkey) final {
if (m_is_forward_cmp)
SeekReverse(userkey, ReverseLowerBound());
else
SeekReverse(userkey, ForwardLowerBound());
}
template<class LowerBound>
void SeekReverse(const Slice& userkey, LowerBound lower_bound_fn) {
m_idx = -1;
m_last_entry_offset = m_tab->m_last_entry_offset;
if (0 == m_last_entry_offset) return;
if (UNLIKELY(!m_iter)) {
m_iter = m_tab->m_trie.new_iter();
}
Slice seek_key = userkey;
if (m_upper_bound && GT(seek_key, *m_upper_bound)) {
seek_key = *m_upper_bound;
}
if (m_lower_bound && LT(seek_key, *m_lower_bound)) {
return; // fail
}
DefineLookupKey(lookup_key, m_cf_id, seek_key);
if (UNLIKELY(!lower_bound_fn(m_iter, lookup_key))) {
return; // fail
}
if (UNLIKELY(iter_cf_id() != m_cf_id)) {
return; // fail
}
if (UNLIKELY(m_upper_bound && user_key_no_assert() == *m_upper_bound)) {
m_idx = 0;
if (!PrevKey())
return;
}
if (UNLIKELY(m_lower_bound && LT(user_key_no_assert(), *m_lower_bound))) {
m_idx = -1;
return; // fail
}
if (m_iter->word() == lookup_key)
SetFirstEntry();
else if (iter_cf_id() == m_cf_id)
SetLastEntry();
}
void SeekToFirst() final {
if (m_lower_bound) {
Seek(*m_lower_bound);
return;
}
if (m_is_forward_cmp ? SeekToFirstForward() : SeekToLastForward()) {
SetFirstEntry();
}
}
bool SeekToFirstForward() {
m_idx = -1;
m_last_entry_offset = m_tab->m_last_entry_offset;
if (0 == m_last_entry_offset) return false;
if (UNLIKELY(!m_iter)) {
m_iter = m_tab->m_trie.new_iter();
}
uint32_t big_cf_id = BIG_ENDIAN_OF(m_cf_id);
fstring lookup_key((char*)&big_cf_id, 4);
if (UNLIKELY(!m_iter->seek_lower_bound(lookup_key))) {
return false;
}
if (UNLIKELY(*(uint32_t*)m_iter->word().data() != big_cf_id)) {
return false;
}
return true;
}
void SeekToLast() final {
if (m_upper_bound) {
SeekForPrev(*m_upper_bound);
return;
}
if (m_is_forward_cmp ? SeekToLastForward() : SeekToFirstForward()) {
SetLastEntry();
}
}
bool SeekToLastForward() {
m_idx = -1;
m_last_entry_offset = m_tab->m_last_entry_offset;
if (0 == m_last_entry_offset) return false;
if (UNLIKELY(!m_iter)) {
m_iter = m_tab->m_trie.new_iter();
}
uint32_t big_next_cf_id = BIG_ENDIAN_OF(m_cf_id+1);
fstring lookup_key((char*)&big_next_cf_id, 4);
if (UNLIKELY(!m_iter->seek_rev_lower_bound(lookup_key))) {
return false;
}
if (UNLIKELY(*(uint32_t*)m_iter->word().data() == big_next_cf_id)) {
if (!m_iter->decr()) {
return false;
}
}
if (iter_cf_id() != m_cf_id) {
ROCKSDB_ASSERT_LT(iter_cf_id(), m_cf_id);
return false;
}
return true;
}
// Moves the iterator to first entry of the previous key.
bool PrevKey() final {
CheckUpdates<false>();
TERARK_ASSERT_GE(m_idx, 0);
if (UNLIKELY(!CmpOrderPrev() || iter_cf_id() != m_cf_id)) {
m_idx = -1;
return false; // fail
}
if (m_lower_bound && LT(user_key_no_assert(), *m_lower_bound)) {
m_idx = -1;
return false; // fail
}
SetFirstEntry();
return true;
}
// Moves the iterator to first entry of the next key.
bool NextKey() final {
CheckUpdates<false>();
TERARK_ASSERT_GE(m_idx, 0);
if (UNLIKELY(!CmpOrderNext() || iter_cf_id() != m_cf_id)) {
m_idx = -1;
return false; // fail
}
if (m_upper_bound && GE(user_key_no_assert(), *m_upper_bound)) {
m_idx = -1;
return false; // fail
}
SetFirstEntry();
return true;
}
bool EqualsKey(const Slice& key) const final {
TERARK_ASSERT_GE(m_idx, 0);
const_cast<Iter*>(this)->CheckUpdates<true>();
return m_rec.key == key;
}
terark_forceinline
Result FindLatestUpdateImpl(MergeContext* mgctx) {
Result result = WBWIIteratorImpl::kNotFound;
for (m_idx = m_num; m_idx > 0;) {
m_idx--;
m_tab->ReadRecord(m_vec[m_idx], &m_rec);
switch (m_rec.type) {
case kPutRecord:
return WBWIIteratorImpl::kFound;
#if (ROCKSDB_MAJOR * 10000 + ROCKSDB_MINOR * 10 + ROCKSDB_PATCH) >= 80100
case kPutEntityRecord:
//return WBWIIteratorImpl::kFoundEntity;
return WBWIIteratorImpl::kFound;
#endif
case kDeleteRecord:
return WBWIIteratorImpl::kDeleted;
case kSingleDeleteRecord:
return WBWIIteratorImpl::kDeleted;
case kMergeRecord:
result = WBWIIteratorImpl::kMergeInProgress;
mgctx->PushOperand(m_rec.value);
break;
case kLogDataRecord:
break; // ignore
case kXIDRecord:
break; // ignore
default:
return WBWIIteratorImpl::kError;
}
}
return result;
}
ROCKSDB_FLATTEN
Result FindLatestUpdate(const Slice& key, MergeContext* mgctx) final {
mgctx->Clear();
if (!Valid()) {