summaryrefslogtreecommitdiff
path: root/db/key.cpp
blob: 47449986d211ae6e8a583bc6ac3d9d5ea9d96aae (plain)
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
// @file key.cpp

/**
*    Copyright (C) 2011 10gen Inc.
*
*    This program is free software: you can redistribute it and/or  modify
*    it under the terms of the GNU Affero General Public License, version 3,
*    as published by the Free Software Foundation.
*
*    This program is distributed in the hope that it will be useful,
*    but WITHOUT ANY WARRANTY; without even the implied warranty of
*    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
*    GNU Affero General Public License for more details.
*
*    You should have received a copy of the GNU Affero General Public License
*    along with this program.  If not, see <http://www.gnu.org/licenses/>.
*/

#include "pch.h"
#include "key.h"
#include "../util/unittest.h"

namespace mongo {

    extern const Ordering nullOrdering = Ordering::make(BSONObj());

    // KeyBson is for V0 (version #0) indexes

    int oldCompare(const BSONObj& l,const BSONObj& r, const Ordering &o);

    // "old" = pre signed dates & such; i.e. btree V0
    /* must be same canon type when called */
    int oldCompareElementValues(const BSONElement& l, const BSONElement& r) {
        dassert( l.canonicalType() == r.canonicalType() );
        int f;
        double x;

        switch ( l.type() ) {
        case EOO:
        case Undefined: // EOO and Undefined are same canonicalType
        case jstNULL:
        case MaxKey:
        case MinKey:
            return 0;
        case Bool:
            return *l.value() - *r.value();
        case Timestamp:
        case Date:
            // unsigned dates for old version
            if ( l.date() < r.date() )
                return -1;
            return l.date() == r.date() ? 0 : 1;
        case NumberLong:
            if( r.type() == NumberLong ) {
                long long L = l._numberLong();
                long long R = r._numberLong();
                if( L < R ) return -1;
                if( L == R ) return 0;
                return 1;
            }
            // else fall through
        case NumberInt:
        case NumberDouble: {
            double left = l.number();
            double right = r.number();
            bool lNan = !( left <= numeric_limits< double >::max() &&
                           left >= -numeric_limits< double >::max() );
            bool rNan = !( right <= numeric_limits< double >::max() &&
                           right >= -numeric_limits< double >::max() );
            if ( lNan ) {
                if ( rNan ) {
                    return 0;
                }
                else {
                    return -1;
                }
            }
            else if ( rNan ) {
                return 1;
            }
            x = left - right;
            if ( x < 0 ) return -1;
            return x == 0 ? 0 : 1;
        }
        case jstOID:
            return memcmp(l.value(), r.value(), 12);
        case Code:
        case Symbol:
        case String:
            // nulls not allowed in the middle of strings in the old version
            return strcmp(l.valuestr(), r.valuestr());
        case Object:
        case Array:
            return oldCompare(l.embeddedObject(), r.embeddedObject(), nullOrdering);
        case DBRef: {
            int lsz = l.valuesize();
            int rsz = r.valuesize();
            if ( lsz - rsz != 0 ) return lsz - rsz;
            return memcmp(l.value(), r.value(), lsz);
        }
        case BinData: {
            int lsz = l.objsize(); // our bin data size in bytes, not including the subtype byte
            int rsz = r.objsize();
            if ( lsz - rsz != 0 ) return lsz - rsz;
            return memcmp(l.value()+4, r.value()+4, lsz+1);
        }
        case RegEx: {
            int c = strcmp(l.regex(), r.regex());
            if ( c )
                return c;
            return strcmp(l.regexFlags(), r.regexFlags());
        }
        case CodeWScope : {
            f = l.canonicalType() - r.canonicalType();
            if ( f )
                return f;
            f = strcmp( l.codeWScopeCode() , r.codeWScopeCode() );
            if ( f )
                return f;
            f = strcmp( l.codeWScopeScopeData() , r.codeWScopeScopeData() );
            if ( f )
                return f;
            return 0;
        }
        default:
            out() << "oldCompareElementValues: bad type " << (int) l.type() << endl;
            assert(false);
        }
        return -1;
    }

    int oldElemCompare(const BSONElement&l , const BSONElement& r) { 
        int lt = (int) l.canonicalType();
        int rt = (int) r.canonicalType();
        int x = lt - rt;
        if( x )
            return x;
        return oldCompareElementValues(l, r);
    }

    // pre signed dates & such
    int oldCompare(const BSONObj& l,const BSONObj& r, const Ordering &o) {
        BSONObjIterator i(l);
        BSONObjIterator j(r);
        unsigned mask = 1;
        while ( 1 ) {
            // so far, equal...

            BSONElement l = i.next();
            BSONElement r = j.next();
            if ( l.eoo() )
                return r.eoo() ? 0 : -1;
            if ( r.eoo() )
                return 1;

            int x;
            {
                x = oldElemCompare(l, r);
                if( o.descending(mask) )
                    x = -x;
            }
            if ( x != 0 )
                return x;
            mask <<= 1;
        }
        return -1;
    }

    /* old style compares:
       - dates are unsigned 
       - strings no nulls
    */
    int KeyBson::woCompare(const KeyBson& r, const Ordering &o) const { 
        return oldCompare(_o, r._o, o); 
    }

    // woEqual could be made faster than woCompare but this is for backward compatibility so not worth a big effort
    bool KeyBson::woEqual(const KeyBson& r) const { 
        return oldCompare(_o, r._o, nullOrdering) == 0;
    }

    // [ ][HASMORE][x][y][canontype_4bits]
    enum CanonicalsEtc { 
        cminkey=1,
        cnull=2,
        cdouble=4,
        cstring=6,
        cbindata=7,
        coid=8,
        cfalse=10,
        ctrue=11,
        cdate=12,
        cmaxkey=14,
        cCANONTYPEMASK = 0xf,
        cY = 0x10,
        cint = cY | cdouble,
        cX = 0x20,
        clong = cX | cdouble,
        cHASMORE = 0x40,
        cNOTUSED = 0x80 // but see IsBSON sentinel - this bit not usable without great care
    };

    // bindata bson type
    const unsigned BinDataLenMask = 0xf0;  // lengths are powers of 2 of this value
    const unsigned BinDataTypeMask = 0x0f; // 0-7 as you would expect, 8-15 are 128+value.  see BinDataType.
    const int BinDataLenMax = 32;
    const int BinDataLengthToCode[] = { 
        0x00, 0x10, 0x20, 0x30, 0x40, 0x50, 0x60, 0x70, 
        0x80, -1/*9*/, 0x90/*10*/, -1/*11*/, 0xa0/*12*/, -1/*13*/, 0xb0/*14*/, -1/*15*/,
        0xc0/*16*/, -1, -1, -1, 0xd0/*20*/, -1, -1, -1, 
        0xe0/*24*/, -1, -1, -1, -1, -1, -1, -1, 
        0xf0/*32*/ 
    };
    const int BinDataCodeToLength[] = { 
        0, 1, 2, 3, 4, 5, 6, 7, 8, 10, 12, 14, 16, 20, 24, 32
    };

    int binDataCodeToLength(int codeByte) { 
        return BinDataCodeToLength[codeByte >> 4];
    }

    /** object cannot be represented in compact format.  so store in traditional bson format 
        with a leading sentinel byte IsBSON to indicate it's in that format.

        Given that the KeyV1Owned constructor already grabbed a bufbuilder, we reuse it here 
        so that we don't have to do an extra malloc.
    */
    void KeyV1Owned::traditional(const BSONObj& obj) { 
        b.reset();
        b.appendUChar(IsBSON);
        b.appendBuf(obj.objdata(), obj.objsize());
        _keyData = (const unsigned char *) b.buf();
    }

    KeyV1Owned::KeyV1Owned(const KeyV1& rhs) {
        b.appendBuf( rhs.data(), rhs.dataSize() );
        _keyData = (const unsigned char *) b.buf();
        dassert( b.len() == dataSize() ); // check datasize method is correct
        dassert( (*_keyData & cNOTUSED) == 0 );
    }

    // fromBSON to Key format
    KeyV1Owned::KeyV1Owned(const BSONObj& obj) {
        BSONObj::iterator i(obj);
        unsigned char bits = 0;
        while( 1 ) { 
            BSONElement e = i.next();
            if( i.more() )
                bits |= cHASMORE;
            switch( e.type() ) { 
            case MinKey:
                b.appendUChar(cminkey|bits);
                break;
            case jstNULL:
                b.appendUChar(cnull|bits);
                break;
            case MaxKey:
                b.appendUChar(cmaxkey|bits);
                break;
            case Bool:
                b.appendUChar( (e.boolean()?ctrue:cfalse) | bits );
                break;
            case jstOID:
                b.appendUChar(coid|bits);
                b.appendBuf(&e.__oid(), sizeof(OID));
                break;
            case BinData:
                {
                    int t = e.binDataType();
                    // 0-7 and 0x80 to 0x87 are supported by Key
                    if( (t & 0x78) == 0 && t != ByteArrayDeprecated ) {
                        int len;
                        const char * d = e.binData(len);
                        if( len <= BinDataLenMax ) {
                            int code = BinDataLengthToCode[len];
                            if( code >= 0 ) {
                                if( t >= 128 )
                                    t = (t-128) | 0x08;
                                dassert( (code&t) == 0 );
                                b.appendUChar( cbindata|bits );
                                b.appendUChar( code | t );
                                b.appendBuf(d, len);
                                break;
                            }
                        }
                    }
                    traditional(obj);
                    return;
                }
            case Date:
                b.appendUChar(cdate|bits);
                b.appendStruct(e.date());
                break;
            case String:
                {
                    b.appendUChar(cstring|bits);
                    // note we do not store the terminating null, to save space.
                    unsigned x = (unsigned) e.valuestrsize() - 1;
                    if( x > 255 ) { 
                        traditional(obj);
                        return;
                    }
                    b.appendUChar(x);
                    b.appendBuf(e.valuestr(), x);
                    break;
                }
            case NumberInt:
                b.appendUChar(cint|bits);
                b.appendNum((double) e._numberInt());
                break;
            case NumberLong:
                {
                    long long n = e._numberLong();
                    long long m = 2LL << 52;
                    DEV {
                        long long d = m-1;
                        assert( ((long long) ((double) -d)) == -d );
                    }
                    if( n >= m || n <= -m ) {
                        // can't represent exactly as a double
                        traditional(obj);
                        return;
                    }
                    b.appendUChar(clong|bits);
                    b.appendNum((double) n);
                    break;
                }
            case NumberDouble:
                {
                    double d = e._numberDouble();
                    if( isNaN(d) ) {
                        traditional(obj);
                        return;
                    }
                    b.appendUChar(cdouble|bits);
                    b.appendNum(d);
                    break;
                }
            default:
                // if other types involved, store as traditional BSON
                traditional(obj);
                return;
            }
            if( !i.more() )
                break;
            bits = 0;
        }
        _keyData = (const unsigned char *) b.buf();
        dassert( b.len() == dataSize() ); // check datasize method is correct
        dassert( (*_keyData & cNOTUSED) == 0 );
    }

    BSONObj KeyV1::toBson() const { 
        assert( _keyData != 0 );
        if( !isCompactFormat() )
            return bson();

        BSONObjBuilder b(512);
        const unsigned char *p = _keyData;
        while( 1 ) { 
            unsigned bits = *p++;

            switch( bits & 0x3f ) {
                case cminkey: b.appendMinKey(""); break;
                case cnull:   b.appendNull(""); break;
                case cfalse:  b.appendBool("", false); break;
                case ctrue:   b.appendBool("", true); break;
                case cmaxkey: 
                    b.appendMaxKey(""); 
                    break;
                case cstring:
                    {
                        unsigned sz = *p++;
                        // we build the element ourself as we have to null terminate it
                        BufBuilder &bb = b.bb();
                        bb.appendNum((char) String);
                        bb.appendUChar(0); // fieldname ""
                        bb.appendNum(sz+1);
                        bb.appendBuf(p, sz);
                        bb.appendUChar(0); // null char at end of string
                        p += sz;
                        break;
                    }
                case coid:
                    b.appendOID("", (OID *) p);
                    p += sizeof(OID);
                    break;
                case cbindata:
                    {
                        int len = binDataCodeToLength(*p);
                        int subtype = (*p) & BinDataTypeMask;
                        if( subtype & 0x8 ) { 
                            subtype = (subtype & 0x7) | 0x80;
                        }
                        b.appendBinData("", len, (BinDataType) subtype, ++p);
                        p += len;
                        break;
                    }
                case cdate:
                    b.appendDate("", (Date_t&) *p);
                    p += 8;
                    break;
                case cdouble:
                    b.append("", (double&) *p);
                    p += sizeof(double);
                    break;
                case cint:
                    b.append("", (int) ((double&) *p));
                    p += sizeof(double);
                    break;
                case clong:
                    b.append("", (long long) ((double&) *p));
                    p += sizeof(double);
                    break;
                default:
                    assert(false);
            }

            if( (bits & cHASMORE) == 0 )
                break;
        }
        return b.obj();
    }

    static int compare(const unsigned char *&l, const unsigned char *&r) { 
        int lt = (*l & cCANONTYPEMASK);
        int rt = (*r & cCANONTYPEMASK);
        int x = lt - rt;
        if( x ) 
            return x;

        l++; r++;

        // same type
        switch( lt ) { 
        case cdouble:
            {
                double L = *((double *) l);
                double R = *((double *) r);
                if( L < R )
                    return -1;
                if( L != R )
                    return 1;
                l += 8; r += 8;
                break;
            }
        case cstring:
            {
                int lsz = *l;
                int rsz = *r;
                int common = min(lsz, rsz);
                l++; r++; // skip the size byte
                // use memcmp as we (will) allow zeros in UTF8 strings
                int res = memcmp(l, r, common);
                if( res ) 
                    return res;
                // longer string is the greater one
                int diff = lsz-rsz;
                if( diff ) 
                    return diff;
                l += lsz; r += lsz;
                break;
            }
        case cbindata:
            {
                int L = *l;
                int R = *r;
                int llen = binDataCodeToLength(L);
                int diff = L-R; // checks length and subtype simultaneously
                if( diff ) {
                    // unfortunately nibbles are backwards to do subtype and len in one check (could bit swap...)
                    int rlen = binDataCodeToLength(R);
                    if( llen != rlen ) 
                        return llen - rlen;
                    return diff;
                }
                // same length, same type
                l++; r++;
                int res = memcmp(l, r, llen);
                if( res ) 
                    return res;
                l += llen; r += llen;
                break;
            }
        case cdate:
            {
                long long L = *((long long *) l);
                long long R = *((long long *) r);
                if( L < R )
                    return -1;
                if( L > R )
                    return 1;
                l += 8; r += 8;
                break;
            }
        case coid:
            {
                int res = memcmp(l, r, sizeof(OID));
                if( res ) 
                    return res;
                l += 12; r += 12;
                break;
            }
        default:
            // all the others are a match -- e.g. null == null
            ;
        }

        return 0;
    }

    // at least one of this and right are traditional BSON format
    int NOINLINE_DECL KeyV1::compareHybrid(const KeyV1& right, const Ordering& order) const { 
        BSONObj L = toBson();
        BSONObj R = right.toBson();
        return L.woCompare(R, order, /*considerfieldname*/false);
    }

    int KeyV1::woCompare(const KeyV1& right, const Ordering &order) const {
        const unsigned char *l = _keyData;
        const unsigned char *r = right._keyData;

        if( (*l|*r) == IsBSON ) // only can do this if cNOTUSED maintained
            return compareHybrid(right, order);

        unsigned mask = 1;
        while( 1 ) { 
            char lval = *l; 
            char rval = *r;
            {
                int x = compare(l, r); // updates l and r pointers
                if( x ) {
                    if( order.descending(mask) )
                        x = -x;
                    return x;
                }
            }

            {
                int x = ((int)(lval & cHASMORE)) - ((int)(rval & cHASMORE));
                if( x ) 
                    return x;
                if( (lval & cHASMORE) == 0 )
                    break;
            }

            mask <<= 1;
        }

        return 0;
    }

    static unsigned sizes[] = {
        0,
        1, //cminkey=1,
        1, //cnull=2,
        0,
        9, //cdouble=4,
        0,
        0, //cstring=6,
        0,
        13, //coid=8,
        0,
        1, //cfalse=10,
        1, //ctrue=11,
        9, //cdate=12,
        0,
        1, //cmaxkey=14,
        0
    };

    inline unsigned sizeOfElement(const unsigned char *p) { 
        unsigned type = *p & cCANONTYPEMASK;
        unsigned sz = sizes[type];
        if( sz == 0 ) {
            if( type == cstring ) { 
                sz = ((unsigned) p[1]) + 2;
            }
            else {
                assert( type == cbindata );
                sz = binDataCodeToLength(p[1]) + 2;
            }
        }
        return sz;
    }

    int KeyV1::dataSize() const { 
        const unsigned char *p = _keyData;
        if( !isCompactFormat() ) {
            return bson().objsize() + 1;
        }

        bool more;
        do { 
            unsigned z = sizeOfElement(p);
            more = (*p & cHASMORE) != 0;
            p += z;
        } while( more );
        return p - _keyData;
    }

    bool KeyV1::woEqual(const KeyV1& right) const {
        const unsigned char *l = _keyData;
        const unsigned char *r = right._keyData;

        if( (*l|*r) == IsBSON ) {
            return toBson().equal(right.toBson());
        }

        while( 1 ) { 
            char lval = *l; 
            char rval = *r;
            if( (lval&(cCANONTYPEMASK|cHASMORE)) != (rval&(cCANONTYPEMASK|cHASMORE)) )
                return false;
            l++; r++;
            switch( lval&cCANONTYPEMASK ) { 
            case coid:
                if( *((unsigned*) l) != *((unsigned*) r) )
                    return false;
                l += 4; r += 4;
            case cdate:
                if( *((unsigned long long *) l) != *((unsigned long long *) r) )
                    return false;
                l += 8; r += 8;
                break;
            case cdouble:
                if( *((double *) l) != *((double *) r) )
                    return false;
                l += 8; r += 8;
                break;
            case cstring:
                {
                    if( *l != *r ) 
                        return false; // not same length
                    unsigned sz = ((unsigned) *l) + 1;
                    if( memcmp(l, r, sz) )
                        return false;
                    l += sz; r += sz;
                    break;
                }
            case cbindata:
                {
                    if( *l != *r )
                        return false; // len or subtype mismatch
                    int len = binDataCodeToLength(*l) + 1;
                    if( memcmp(l, r, len) ) 
                        return false;
                    l += len; r += len;
                    break;
                }
            case cminkey:
            case cnull:
            case cfalse:
            case ctrue:
            case cmaxkey:
                break;
            default:
                assert(false);
            }
            if( (lval&cHASMORE) == 0 )
                break;
        }
        return true;
    }

    struct CmpUnitTest : public UnitTest {
        void run() {
            char a[2];
            char b[2];
            a[0] = -3;
            a[1] = 0;
            b[0] = 3;
            b[1] = 0;
            assert( strcmp(a,b)>0 && memcmp(a,b,2)>0 );
        }
    } cunittest;

}