-
-
Notifications
You must be signed in to change notification settings - Fork 492
/
sphinxrt.cpp
12764 lines (10473 loc) · 380 KB
/
sphinxrt.cpp
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) 2017-2018, Manticore Software LTD (http://manticoresearch.com)
// Copyright (c) 2001-2016, Andrew Aksyonoff
// Copyright (c) 2008-2016, Sphinx Technologies Inc
// All rights reserved
//
// This program is free software; you can redistribute it and/or modify
// it under the terms of the GNU General Public License. You should have
// received a copy of the GPL license along with this program; if you
// did not, you can find it at http://www.gnu.org/
//
#include "sphinx.h"
#include "sphinxint.h"
#include "sphinxrt.h"
#include "sphinxsearch.h"
#include "sphinxutils.h"
#include "sphinxjson.h"
#include "sphinxplugin.h"
#include "sphinxrlp.h"
#include "sphinxqcache.h"
#include <sys/stat.h>
#include <fcntl.h>
#if USE_WINDOWS
#include <io.h> // for open(), close()
#include <errno.h>
#else
#include <unistd.h>
#include <sys/time.h>
#endif
//////////////////////////////////////////////////////////////////////////
#define BINLOG_WRITE_BUFFER 256*1024
#define BINLOG_AUTO_FLUSH 1000000
#define RTDICT_CHECKPOINT_V3 1024
#define RTDICT_CHECKPOINT_V5 48
#define SPH_RT_DOUBLE_BUFFER_PERCENT 10
#define WORDID_MAX U64C(0xffffffffffffffff)
//////////////////////////////////////////////////////////////////////////
#ifndef NDEBUG
#define Verify(_expr) assert(_expr)
#else
#define Verify(_expr) _expr
#endif
//////////////////////////////////////////////////////////////////////////
// GLOBALS
//////////////////////////////////////////////////////////////////////////
/// publicly exposed binlog interface
ISphBinlog * g_pBinlog = NULL;
/// actual binlog implementation
class RtBinlog_c;
static RtBinlog_c * g_pRtBinlog = NULL;
/// protection from concurrent changes during binlog replay
static bool g_bRTChangesAllowed = false;
// optimize mode for disk chunks merge
static bool g_bProgressiveMerge = true;
static auto& g_bShutdown = sphGetShutdown();
//////////////////////////////////////////////////////////////////////////
// !COMMIT cleanup extern ref to sphinx.cpp
extern void sphSortDocinfos ( DWORD * pBuf, int iCount, int iStride );
// !COMMIT yes i am when debugging
#ifndef NDEBUG
#define PARANOID 1
#endif
//////////////////////////////////////////////////////////////////////////
// Variable Length Byte (VLB) encoding
// store int variable in as much bytes as actually needed to represent it
template < typename T, typename P >
static inline void ZipT ( CSphVector < BYTE, P > * pOut, T uValue )
{
do
{
BYTE bOut = (BYTE)( uValue & 0x7f );
uValue >>= 7;
if ( uValue )
bOut |= 0x80;
pOut->Add ( bOut );
} while ( uValue );
}
template < typename T >
static inline void ZipT ( BYTE * & pOut, T uValue )
{
do
{
BYTE bOut = (BYTE)( uValue & 0x7f );
uValue >>= 7;
if ( uValue )
bOut |= 0x80;
*pOut++ = bOut;
} while ( uValue );
}
#define SPH_MAX_KEYWORD_LEN (3*SPH_MAX_WORD_LEN+4)
STATIC_ASSERT ( SPH_MAX_KEYWORD_LEN<255, MAX_KEYWORD_LEN_SHOULD_FITS_BYTE );
// Variable Length Byte (VLB) decoding
template < typename T >
static inline const BYTE * UnzipT ( T * pValue, const BYTE * pIn )
{
T uValue = 0;
BYTE bIn;
int iOff = 0;
do
{
bIn = *pIn++;
uValue += ( T ( bIn & 0x7f ) ) << iOff;
iOff += 7;
} while ( bIn & 0x80 );
*pValue = uValue;
return pIn;
}
#define ZipDword ZipT<DWORD>
#define ZipQword ZipT<uint64_t>
#define UnzipDword UnzipT<DWORD>
#define UnzipQword UnzipT<uint64_t>
#define ZipDocid ZipQword
#define ZipWordid ZipQword
#define UnzipDocid UnzipQword
#define UnzipWordid UnzipQword
//////////////////////////////////////////////////////////////////////////
struct CmpHitPlain_fn
{
inline bool IsLess ( const CSphWordHit & a, const CSphWordHit & b ) const
{
return ( a.m_uWordID<b.m_uWordID ) ||
( a.m_uWordID==b.m_uWordID && a.m_uDocID<b.m_uDocID ) ||
( a.m_uWordID==b.m_uWordID && a.m_uDocID==b.m_uDocID && a.m_uWordPos<b.m_uWordPos );
}
};
struct CmpHitKeywords_fn
{
const BYTE * m_pBase;
explicit CmpHitKeywords_fn ( const BYTE * pBase ) : m_pBase ( pBase ) {}
inline bool IsLess ( const CSphWordHit & a, const CSphWordHit & b ) const
{
const BYTE * pPackedA = m_pBase + a.m_uWordID;
const BYTE * pPackedB = m_pBase + b.m_uWordID;
int iCmp = sphDictCmpStrictly ( (const char *)pPackedA+1, *pPackedA, (const char *)pPackedB+1, *pPackedB );
return ( iCmp<0 ) ||
( iCmp==0 && a.m_uDocID<b.m_uDocID ) ||
( iCmp==0 && a.m_uDocID==b.m_uDocID && a.m_uWordPos<b.m_uWordPos );
}
};
template < typename DOCID = SphDocID_t >
struct RtDoc_T
{
DOCID m_uDocID = 0; ///< my document id
DWORD m_uDocFields = 0; ///< fields mask
DWORD m_uHits = 0; ///< hit count
DWORD m_uHit = 0; ///< either index into segment hits, or the only hit itself (if hit count is 1)
};
template < typename WORDID=SphWordID_t >
struct RtWord_T
{
union
{
WORDID m_uWordID; ///< my keyword id
const BYTE * m_sWord = nullptr;
};
DWORD m_uDocs = 0; ///< document count (for stats and/or BM25)
DWORD m_uHits = 0; ///< hit count (for stats and/or BM25)
DWORD m_uDoc = 0; ///< index into segment docs
};
typedef RtDoc_T<> RtDoc_t;
typedef RtWord_T<> RtWord_t;
struct RtWordCheckpoint_t
{
union
{
SphWordID_t m_uWordID;
const char * m_sWord;
};
int m_iOffset;
};
// More than just sorted vector.
// OrderedHash is for fast (without sorting potentially big vector) inserts.
class CSphKilllist : public ISphNoncopyable
{
private:
static const int MAX_SMALL_SIZE = 512;
CSphVector<SphDocID_t> m_dLargeKlist;
CSphOrderedHash < bool, SphDocID_t, IdentityHash_fn, MAX_SMALL_SIZE > m_hSmallKlist;
CSphRwlock m_tLock;
public:
CSphKilllist()
{
m_tLock.Init();
}
virtual ~CSphKilllist()
{
m_tLock.Done();
}
void Flush ( CSphVector<SphDocID_t> & dKlist ) REQUIRES (!m_tLock)
{
{
CSphScopedRLock tRguard ( m_tLock );
bool bGotHash = (m_hSmallKlist.GetLength ()>0);
if ( !bGotHash )
NakedCopy ( dKlist );
if ( !bGotHash )
return;
}
CSphScopedWLock tWguard ( m_tLock );
NakedFlush ( nullptr, 0 );
NakedCopy ( dKlist );
}
inline void Add ( SphDocID_t * pDocs, int iCount ) REQUIRES (!m_tLock)
{
if ( !iCount )
return;
CSphScopedWLock tWlock ( m_tLock );
if ( m_hSmallKlist.GetLength()+iCount>=MAX_SMALL_SIZE )
{
NakedFlush ( pDocs, iCount );
} else
{
while ( iCount-- )
m_hSmallKlist.Add ( true, *pDocs++ );
}
}
bool Exists ( SphDocID_t uDoc ) REQUIRES (!m_tLock)
{
CSphScopedRLock tRguard ( m_tLock );
bool bGot = ( m_hSmallKlist.Exists ( uDoc ) || m_dLargeKlist.BinarySearch ( uDoc )!=NULL );
return bGot;
}
void Reset ( SphDocID_t * pDocs, int iCount ) REQUIRES ( !m_tLock )
{
m_tLock.WriteLock();
m_dLargeKlist.Reset();
m_hSmallKlist.Reset();
NakedFlush ( pDocs, iCount );
m_tLock.Unlock();
}
void LoadFromFile ( const char * sFilename ) REQUIRES ( !m_tLock );
void SaveToFile ( const char * sFilename ) REQUIRES ( !m_tLock );
private:
void NakedCopy ( CSphVector<SphDocID_t> & dKlist )
{
assert ( m_hSmallKlist.GetLength()==0 );
if ( !m_dLargeKlist.GetLength() )
return;
dKlist.Append ( m_dLargeKlist );
}
void NakedFlush ( SphDocID_t * pDocs, int iCount )
{
if ( m_hSmallKlist.GetLength()==0 && iCount==0 )
return;
m_dLargeKlist.Reserve ( m_dLargeKlist.GetLength()+m_hSmallKlist.GetLength()+iCount );
m_hSmallKlist.IterateStart();
while ( m_hSmallKlist.IterateNext() )
m_dLargeKlist.Add ( m_hSmallKlist.IterateGetKey() );
if ( pDocs && iCount )
m_dLargeKlist.Append ( pDocs, iCount );
m_dLargeKlist.Uniq();
m_hSmallKlist.Reset();
}
};
// is already id32<>id64 safe
void CSphKilllist::LoadFromFile ( const char * sFilename )
{
Reset ( NULL, 0 );
CSphString sName, sError;
sName.SetSprintf ( "%s.kill", sFilename );
if ( !sphIsReadable ( sName.cstr(), &sError ) )
return;
CSphAutoreader tKlistReader;
if ( !tKlistReader.Open ( sName, sError ) )
return;
// FIXME!!! got rid of locks here
m_tLock.WriteLock();
m_dLargeKlist.Resize ( tKlistReader.GetDword() );
SphDocID_t uLastDocID = 0;
ARRAY_FOREACH ( i, m_dLargeKlist )
{
uLastDocID += ( SphDocID_t ) tKlistReader.UnzipOffset();
m_dLargeKlist[i] = uLastDocID;
};
m_tLock.Unlock();
}
void CSphKilllist::SaveToFile ( const char * sFilename )
{
// FIXME!!! got rid of locks here
m_tLock.WriteLock();
NakedFlush ( nullptr, 0 );
CSphWriter tKlistWriter;
CSphString sName, sError;
sName.SetSprintf ( "%s.kill", sFilename );
tKlistWriter.OpenFile ( sName.cstr(), sError );
tKlistWriter.PutDword ( m_dLargeKlist.GetLength() );
SphDocID_t uLastDocID = 0;
ARRAY_FOREACH ( i, m_dLargeKlist )
{
tKlistWriter.ZipOffset ( m_dLargeKlist[i] - uLastDocID );
uLastDocID = ( SphDocID_t ) m_dLargeKlist[i];
};
m_tLock.Unlock();
tKlistWriter.CloseFile();
}
struct KlistRefcounted_t : ISphRefcountedMT
{
CSphFixedVector<SphDocID_t> m_dKilled { 0 };
private:
~KlistRefcounted_t() = default;
};
// this is what actually stores index data
// RAM chunk consists of such segments
struct RtSegment_t : ISphNoncopyable
{
protected:
static const int KLIST_ACCUM_THRESH = 32;
public:
static CSphAtomic m_iSegments; ///< age tag sequence generator
int m_iTag; ///< segment age tag
CSphTightVector<BYTE> m_dWords;
CSphVector<RtWordCheckpoint_t> m_dWordCheckpoints;
CSphTightVector<uint64_t> m_dInfixFilterCP;
CSphTightVector<BYTE> m_dDocs;
CSphTightVector<BYTE> m_dHits;
int m_iRows; ///< number of actually allocated rows
int m_iAliveRows; ///< number of alive (non-killed) rows
CSphTightVector<CSphRowitem> m_dRows; ///< row data storage
KlistRefcounted_t * m_pKlist;
bool m_bTlsKlist; ///< whether to apply TLS K-list during merge (must only be used by writer during Commit())
CSphTightVector<BYTE> m_dStrings; ///< strings storage
CSphTightVector<DWORD> m_dMvas; ///< MVAs storage
CSphVector<BYTE> m_dKeywordCheckpoints;
mutable CSphAtomic m_tRefCount;
RtSegment_t ()
{
m_iTag = m_iSegments.Inc();
m_iRows = 0;
m_iAliveRows = 0;
m_bTlsKlist = false;
m_dStrings.Add ( 0 ); // dummy zero offset
m_dMvas.Add ( 0 ); // dummy zero offset
m_pKlist = new KlistRefcounted_t();
}
~RtSegment_t ()
{
SafeRelease ( m_pKlist );
}
int64_t GetUsedRam () const
{
// FIXME! gonna break on vectors over 2GB
return
( (int64_t)m_dWords.GetLimit() )*sizeof(m_dWords[0]) +
( (int64_t)m_dDocs.GetLimit() )*sizeof(m_dDocs[0]) +
( (int64_t)m_dHits.GetLimit() )*sizeof(m_dHits[0]) +
( (int64_t)m_dStrings.GetLimit() )*sizeof(m_dStrings[0]) +
( (int64_t)m_dMvas.GetLimit() )*sizeof(m_dMvas[0]) +
( (int64_t)m_dKeywordCheckpoints.GetLimit() )*sizeof(m_dKeywordCheckpoints[0])+
( (int64_t)m_dRows.GetLimit() )*sizeof(m_dRows[0]) +
( (int64_t)m_dInfixFilterCP.GetLength()*sizeof(m_dInfixFilterCP[0]) );
}
int GetMergeFactor () const
{
return m_iRows;
}
int GetStride () const
{
return ( m_dRows.GetLength() / m_iRows );
}
const CSphFixedVector<SphDocID_t> & GetKlist() const { return m_pKlist->m_dKilled; }
const CSphRowitem * FindRow ( SphDocID_t uDocid ) const;
const CSphRowitem * FindAliveRow ( SphDocID_t uDocid ) const;
};
CSphAtomic RtSegment_t::m_iSegments { 0 };
const CSphRowitem * RtSegment_t::FindRow ( SphDocID_t uDocid ) const
{
// binary search through the rows
int iStride = GetStride();
SphDocID_t uL = DOCINFO2ID ( m_dRows.Begin() );
SphDocID_t uR = DOCINFO2ID ( &m_dRows[m_dRows.GetLength()-iStride] );
if ( uDocid==uL )
return m_dRows.Begin();
if ( uDocid==uR )
return &m_dRows[m_dRows.GetLength()-iStride];
if ( uDocid<uL || uDocid>uR )
return NULL;
int iL = 0;
int iR = m_iRows-1;
while ( iR-iL>1 )
{
int iM = iL + (iR-iL)/2;
SphDocID_t uM = DOCINFO2ID ( &m_dRows[iM*iStride] );
if ( uDocid==uM )
return &m_dRows[iM*iStride];
else if ( uDocid>uM )
iL = iM;
else
iR = iM;
}
return NULL;
}
const CSphRowitem * RtSegment_t::FindAliveRow ( SphDocID_t uDocid ) const
{
if ( m_pKlist->m_dKilled.BinarySearch ( uDocid ) )
return NULL;
return FindRow ( uDocid );
}
//////////////////////////////////////////////////////////////////////////
struct RtDocWriter_t
{
CSphTightVector<BYTE> * m_pDocs;
SphDocID_t m_uLastDocID;
explicit RtDocWriter_t ( RtSegment_t * pSeg )
: m_pDocs ( &pSeg->m_dDocs )
, m_uLastDocID ( 0 )
{}
void ZipDoc ( const RtDoc_t & tDoc )
{
CSphTightVector<BYTE> * pDocs = m_pDocs;
BYTE * pEnd = pDocs->AddN ( 12*sizeof(DWORD) );
const BYTE * pBegin = pDocs->Begin();
ZipDocid ( pEnd, tDoc.m_uDocID - m_uLastDocID );
m_uLastDocID = tDoc.m_uDocID;
ZipDword ( pEnd, tDoc.m_uDocFields );
ZipDword ( pEnd, tDoc.m_uHits );
if ( tDoc.m_uHits==1 )
{
ZipDword ( pEnd, tDoc.m_uHit & 0xffffffUL );
ZipDword ( pEnd, tDoc.m_uHit>>24 );
} else
ZipDword ( pEnd, tDoc.m_uHit );
pDocs->Resize ( pEnd-pBegin );
}
DWORD ZipDocPtr () const
{
return m_pDocs->GetLength();
}
void ZipRestart ()
{
m_uLastDocID = 0;
}
};
template < typename DOCID = SphDocID_t >
struct RtDocReader_T
{
typedef RtDoc_T<DOCID> RTDOC;
const BYTE * m_pDocs;
int m_iLeft;
RTDOC m_tDoc;
template < typename RTWORD >
RtDocReader_T ( const RtSegment_t * pSeg, const RTWORD & tWord )
{
m_pDocs = ( pSeg->m_dDocs.Begin() ? pSeg->m_dDocs.Begin() + tWord.m_uDoc : NULL );
m_iLeft = tWord.m_uDocs;
m_tDoc.m_uDocID = 0;
}
RtDocReader_T ()
{
m_pDocs = NULL;
m_iLeft = 0;
m_tDoc.m_uDocID = 0;
}
const RTDOC * UnzipDoc ()
{
if ( !m_iLeft || !m_pDocs )
return NULL;
const BYTE * pIn = m_pDocs;
SphDocID_t uDeltaID;
pIn = UnzipDocid ( &uDeltaID, pIn );
m_tDoc.m_uDocID += (DOCID) uDeltaID;
DWORD uField;
pIn = UnzipDword ( &uField, pIn );
m_tDoc.m_uDocFields = uField;
pIn = UnzipDword ( &m_tDoc.m_uHits, pIn );
if ( m_tDoc.m_uHits==1 )
{
DWORD a, b;
pIn = UnzipDword ( &a, pIn );
pIn = UnzipDword ( &b, pIn );
m_tDoc.m_uHit = a + ( b<<24 );
} else
pIn = UnzipDword ( &m_tDoc.m_uHit, pIn );
m_pDocs = pIn;
m_iLeft--;
return &m_tDoc;
}
};
typedef RtDocReader_T<> RtDocReader_t;
template < typename VECTOR >
static int sphPutBytes ( VECTOR * pOut, const void * pData, int iLen )
{
int iOff = pOut->GetLength();
pOut->Resize ( iOff + iLen );
memcpy ( pOut->Begin()+iOff, pData, iLen );
return iOff;
}
struct RtWordWriter_t
{
CSphTightVector<BYTE> * m_pWords;
CSphVector<RtWordCheckpoint_t> * m_pCheckpoints;
CSphVector<BYTE> * m_pKeywordCheckpoints;
CSphKeywordDeltaWriter m_tLastKeyword;
SphWordID_t m_uLastWordID;
DWORD m_uLastDoc;
int m_iWords;
bool m_bKeywordDict;
int m_iWordsCheckpoint;
RtWordWriter_t ( RtSegment_t * pSeg, bool bKeywordDict, int iWordsCheckpoint )
: m_pWords ( &pSeg->m_dWords )
, m_pCheckpoints ( &pSeg->m_dWordCheckpoints )
, m_pKeywordCheckpoints ( &pSeg->m_dKeywordCheckpoints )
, m_uLastWordID ( 0 )
, m_uLastDoc ( 0 )
, m_iWords ( 0 )
, m_bKeywordDict ( bKeywordDict )
, m_iWordsCheckpoint ( iWordsCheckpoint )
{
assert ( !m_pWords->GetLength() );
assert ( !m_pCheckpoints->GetLength() );
assert ( !m_pKeywordCheckpoints->GetLength() );
}
void ZipWord ( const RtWord_t & tWord )
{
CSphTightVector<BYTE> * pWords = m_pWords;
if ( ++m_iWords==m_iWordsCheckpoint )
{
RtWordCheckpoint_t & tCheckpoint = m_pCheckpoints->Add();
if ( !m_bKeywordDict )
{
tCheckpoint.m_uWordID = tWord.m_uWordID;
} else
{
int iLen = tWord.m_sWord[0];
assert ( iLen && iLen-1<SPH_MAX_KEYWORD_LEN );
tCheckpoint.m_uWordID = sphPutBytes ( m_pKeywordCheckpoints, tWord.m_sWord+1, iLen+1 );
m_pKeywordCheckpoints->Last() = '\0'; // checkpoint is NULL terminating string
// reset keywords delta encoding
m_tLastKeyword.Reset();
}
tCheckpoint.m_iOffset = pWords->GetLength();
m_uLastWordID = 0;
m_uLastDoc = 0;
m_iWords = 1;
}
if ( !m_bKeywordDict )
{
ZipWordid ( pWords, tWord.m_uWordID - m_uLastWordID );
} else
{
m_tLastKeyword.PutDelta ( *this, tWord.m_sWord+1, tWord.m_sWord[0] );
}
BYTE * pEnd = pWords->AddN ( 4*sizeof(DWORD) );
const BYTE * pBegin = pWords->Begin();
ZipDword ( pEnd, tWord.m_uDocs );
ZipDword ( pEnd, tWord.m_uHits );
ZipDword ( pEnd, tWord.m_uDoc - m_uLastDoc );
pWords->Resize ( pEnd-pBegin );
m_uLastWordID = tWord.m_uWordID;
m_uLastDoc = tWord.m_uDoc;
}
void PutBytes ( const BYTE * pData, int iLen )
{
sphPutBytes ( m_pWords, pData, iLen );
}
};
template < typename WORDID = SphWordID_t >
struct RtWordReader_T
{
typedef RtWord_T<WORDID> RTWORD;
BYTE m_tPackedWord[SPH_MAX_KEYWORD_LEN+1];
const BYTE * m_pCur;
const BYTE * m_pMax;
RTWORD m_tWord;
int m_iWords;
bool m_bWordDict;
int m_iWordsCheckpoint;
int m_iCheckpoint;
RtWordReader_T ( const RtSegment_t * pSeg, bool bWordDict, int iWordsCheckpoint )
: m_iWords ( 0 )
, m_bWordDict ( bWordDict )
, m_iWordsCheckpoint ( iWordsCheckpoint )
, m_iCheckpoint ( 0 )
{
m_tWord.m_uWordID = 0;
Reset ( pSeg );
if ( bWordDict )
m_tWord.m_sWord = m_tPackedWord;
}
void Reset ( const RtSegment_t * pSeg )
{
m_pCur = pSeg->m_dWords.Begin();
m_pMax = m_pCur + pSeg->m_dWords.GetLength();
m_tWord.m_uDoc = 0;
m_iWords = 0;
}
const RTWORD * UnzipWord ()
{
if ( ++m_iWords==m_iWordsCheckpoint )
{
m_tWord.m_uDoc = 0;
m_iWords = 1;
m_iCheckpoint++;
if ( !m_bWordDict )
m_tWord.m_uWordID = 0;
}
if ( m_pCur>=m_pMax )
return NULL;
const BYTE * pIn = m_pCur;
DWORD uDeltaDoc;
if ( m_bWordDict )
{
BYTE iMatch, iDelta, uPacked;
uPacked = *pIn++;
if ( uPacked & 0x80 )
{
iDelta = ( ( uPacked>>4 ) & 7 ) + 1;
iMatch = uPacked & 15;
} else
{
iDelta = uPacked & 127;
iMatch = *pIn++;
}
m_tPackedWord[0] = iMatch+iDelta;
memcpy ( m_tPackedWord+1+iMatch, pIn, iDelta );
m_tPackedWord[1+m_tPackedWord[0]] = 0;
pIn += iDelta;
} else
{
SphWordID_t uDeltaID;
pIn = UnzipWordid ( &uDeltaID, pIn );
m_tWord.m_uWordID += (WORDID) uDeltaID;
}
pIn = UnzipDword ( &m_tWord.m_uDocs, pIn );
pIn = UnzipDword ( &m_tWord.m_uHits, pIn );
pIn = UnzipDword ( &uDeltaDoc, pIn );
m_pCur = pIn;
m_tWord.m_uDoc += uDeltaDoc;
return &m_tWord;
}
};
typedef RtWordReader_T<SphWordID_t> RtWordReader_t;
struct RtHitWriter_t
{
CSphTightVector<BYTE> * m_pHits;
DWORD m_uLastHit;
explicit RtHitWriter_t ( RtSegment_t * pSeg )
: m_pHits ( &pSeg->m_dHits )
, m_uLastHit ( 0 )
{}
void ZipHit ( DWORD uValue )
{
ZipDword ( m_pHits, uValue - m_uLastHit );
m_uLastHit = uValue;
}
void ZipRestart ()
{
m_uLastHit = 0;
}
DWORD ZipHitPtr () const
{
return m_pHits->GetLength();
}
};
struct RtHitReader_t
{
const BYTE * m_pCur;
DWORD m_iLeft;
DWORD m_uLast;
RtHitReader_t ()
: m_pCur ( NULL )
, m_iLeft ( 0 )
, m_uLast ( 0 )
{}
template < typename RTDOC >
explicit RtHitReader_t ( const RtSegment_t * pSeg, const RTDOC * pDoc )
{
m_pCur = pSeg->m_dHits.Begin() + pDoc->m_uHit;
m_iLeft = pDoc->m_uHits;
m_uLast = 0;
}
DWORD UnzipHit ()
{
if ( !m_iLeft )
return 0;
DWORD uValue;
m_pCur = UnzipDword ( &uValue, m_pCur );
m_uLast += uValue;
m_iLeft--;
return m_uLast;
}
};
struct RtHitReader2_t : public RtHitReader_t
{
const BYTE * m_pBase;
RtHitReader2_t ()
: m_pBase ( NULL )
{}
void Seek ( SphOffset_t uOff, int iHits )
{
m_pCur = m_pBase + uOff;
m_iLeft = iHits;
m_uLast = 0;
}
};
//////////////////////////////////////////////////////////////////////////
/// forward ref
struct RtIndex_t;
/// indexing accumulator
class RtAccum_t : public ISphRtAccum
{
public:
int m_iAccumDocs;
CSphTightVector<CSphWordHit> m_dAccum;
CSphTightVector<CSphRowitem> m_dAccumRows;
CSphVector<SphDocID_t> m_dAccumKlist;
CSphTightVector<BYTE> m_dStrings;
CSphTightVector<DWORD> m_dMvas;
CSphVector<DWORD> m_dPerDocHitsCount;
bool m_bKeywordDict;
CSphDict * m_pDict;
private:
CSphDict * m_pRefDict;
CSphDict * m_pDictCloned;
ISphRtDictWraper * m_pDictRt;
bool m_bReplace; ///< insert or replace mode (affects CleanupDuplicates() behavior)
public:
explicit RtAccum_t ( bool bKeywordDict );
~RtAccum_t();
void SetupDict ( const ISphRtIndex * pIndex, CSphDict * pDict, bool bKeywordDict );
void ResetDict ();
void Sort ();
void AddDocument ( ISphHits * pHits, const CSphMatch & tDoc, bool bReplace, int iRowSize, const char ** ppStr, const CSphVector<DWORD> & dMvas );
RtSegment_t * CreateSegment ( int iRowSize, int iWordsCheckpoint );
void CleanupDuplicates ( int iRowSize );
void GrabLastWarning ( CSphString & sWarning );
void SetIndex ( ISphRtIndex * pIndex ) { m_pIndex = pIndex; }
};
/// TLS indexing accumulator (we disallow two uncommitted adds within one thread; and so need at most one)
static SphThreadKey_t g_tTlsAccumKey;
/// binlog file view of the index
/// everything that a given log file needs to know about an index
struct BinlogIndexInfo_t
{
CSphString m_sName; ///< index name
int64_t m_iMinTID; ///< min TID logged by this file
int64_t m_iMaxTID; ///< max TID logged by this file
int64_t m_iFlushedTID; ///< last flushed TID
int64_t m_tmMin; ///< min TID timestamp
int64_t m_tmMax; ///< max TID timestamp
CSphIndex * m_pIndex; ///< replay only; associated index (might be NULL if we don't serve it anymore!)
RtIndex_t * m_pRT; ///< replay only; RT index handle (might be NULL if N/A or non-RT)
int64_t m_iPreReplayTID; ///< replay only; index TID at the beginning of this file replay
BinlogIndexInfo_t ()
: m_iMinTID ( INT64_MAX )
, m_iMaxTID ( 0 )
, m_iFlushedTID ( 0 )
, m_tmMin ( INT64_MAX )
, m_tmMax ( 0 )
, m_pIndex ( NULL )
, m_pRT ( NULL )
, m_iPreReplayTID ( 0 )
{}
};
/// binlog file descriptor
/// file id (aka extension), plus a list of associated index infos
struct BinlogFileDesc_t
{
int m_iExt;
CSphVector<BinlogIndexInfo_t> m_dIndexInfos;
BinlogFileDesc_t ()
: m_iExt ( 0 )
{}
};
/// Bin Log Operation
enum Blop_e
{
BLOP_COMMIT = 1,
BLOP_UPDATE_ATTRS = 2,
BLOP_ADD_INDEX = 3,
BLOP_ADD_CACHE = 4,
BLOP_RECONFIGURE = 5,
BLOP_TOTAL
};
// forward declaration
class BufferReader_t;
class RtBinlog_c;
class BinlogWriter_c : public CSphWriter
{
public:
BinlogWriter_c ();
virtual ~BinlogWriter_c () {}
virtual void Flush ();
void Write ();
void Fsync ();
bool HasUnwrittenData () const { return m_iPoolUsed>0; }
bool HasUnsyncedData () const { return m_iLastFsyncPos!=m_iLastWritePos; }
void ResetCrc (); ///< restart checksumming
void WriteCrc (); ///< finalize and write current checksum to output stream
private:
int64_t m_iLastWritePos;
int64_t m_iLastFsyncPos;
int m_iLastCrcPos;
DWORD m_uCRC;
void HashCollected ();
};
class BinlogReader_c : public CSphAutoreader
{
public:
BinlogReader_c ();
void ResetCrc ();
bool CheckCrc ( const char * sOp, const char * sIndexName, int64_t iTid, int64_t iTxnPos );
private:
DWORD m_uCRC;
int m_iLastCrcPos;
virtual void UpdateCache ();
void HashCollected ();
};
class RtBinlog_c : public ISphBinlog
{
public:
RtBinlog_c ();
~RtBinlog_c ();
void BinlogCommit ( int64_t * pTID, const char * sIndexName, const RtSegment_t * pSeg, const CSphVector<SphDocID_t> & dKlist, bool bKeywordDict );
void BinlogUpdateAttributes ( int64_t * pTID, const char * sIndexName, const CSphAttrUpdate & tUpd );
void BinlogReconfigure ( int64_t * pTID, const char * sIndexName, const CSphReconfigureSetup & tSetup );
void NotifyIndexFlush ( const char * sIndexName, int64_t iTID, bool bShutdown );
void Configure ( const CSphConfigSection & hSearchd, bool bTestMode );
void Replay ( const SmallStringHash_T<CSphIndex*> & hIndexes, DWORD uReplayFlags, ProgressCallbackSimple_t * pfnProgressCallback );
void GetFlushInfo ( BinlogFlushInfo_t & tFlush );
bool IsActive () { return !m_bDisabled; }
void CheckPath ( const CSphConfigSection & hSearchd, bool bTestMode );
private:
static const DWORD BINLOG_VERSION = 6;
static const DWORD BINLOG_HEADER_MAGIC = 0x4c425053; /// magic 'SPBL' header that marks binlog file
static const DWORD BLOP_MAGIC = 0x214e5854; /// magic 'TXN!' header that marks binlog entry
static const DWORD BINLOG_META_MAGIC = 0x494c5053; /// magic 'SPLI' header that marks binlog meta