Coverage Report

Created: 2026-06-16 07:20

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/src/bzip2/blocksort.c
Line
Count
Source
1
2
/*-------------------------------------------------------------*/
3
/*--- Block sorting machinery                               ---*/
4
/*---                                           blocksort.c ---*/
5
/*-------------------------------------------------------------*/
6
7
/* ------------------------------------------------------------------
8
   This file is part of bzip2/libbzip2, a program and library for
9
   lossless, block-sorting data compression.
10
11
   bzip2/libbzip2 version 1.0.6 of 6 September 2010
12
   Copyright (C) 1996-2010 Julian Seward <jseward@acm.org>
13
14
   Please read the WARNING, DISCLAIMER and PATENTS sections in the 
15
   README file.
16
17
   This program is released under the terms of the license contained
18
   in the file LICENSE.
19
   ------------------------------------------------------------------ */
20
21
22
#include "bzlib_private.h"
23
24
/*---------------------------------------------*/
25
/*--- Fallback O(N log(N)^2) sorting        ---*/
26
/*--- algorithm, for repetitive blocks      ---*/
27
/*---------------------------------------------*/
28
29
/*---------------------------------------------*/
30
static 
31
__inline__
32
void fallbackSimpleSort ( UInt32* fmap, 
33
                          UInt32* eclass, 
34
                          Int32   lo, 
35
                          Int32   hi )
36
128M
{
37
128M
   Int32 i, j, tmp;
38
128M
   UInt32 ec_tmp;
39
40
128M
   if (lo == hi) return;
41
42
124M
   if (hi - lo > 3) {
43
111M
      for ( i = hi-4; i >= lo; i-- ) {
44
84.3M
         tmp = fmap[i];
45
84.3M
         ec_tmp = eclass[tmp];
46
113M
         for ( j = i+4; j <= hi && ec_tmp > eclass[fmap[j]]; j += 4 )
47
29.0M
            fmap[j-4] = fmap[j];
48
84.3M
         fmap[j-4] = tmp;
49
84.3M
      }
50
27.1M
   }
51
52
415M
   for ( i = hi-1; i >= lo; i-- ) {
53
290M
      tmp = fmap[i];
54
290M
      ec_tmp = eclass[tmp];
55
429M
      for ( j = i+1; j <= hi && ec_tmp > eclass[fmap[j]]; j++ )
56
138M
         fmap[j-1] = fmap[j];
57
290M
      fmap[j-1] = tmp;
58
290M
   }
59
124M
}
60
61
62
/*---------------------------------------------*/
63
#define fswap(zz1, zz2) \
64
1.73G
   { Int32 zztmp = zz1; zz1 = zz2; zz2 = zztmp; }
65
66
56.9M
#define fvswap(zzp1, zzp2, zzn)       \
67
56.9M
{                                     \
68
56.9M
   Int32 yyp1 = (zzp1);               \
69
56.9M
   Int32 yyp2 = (zzp2);               \
70
56.9M
   Int32 yyn  = (zzn);                \
71
299M
   while (yyn > 0) {                  \
72
242M
      fswap(fmap[yyp1], fmap[yyp2]);  \
73
242M
      yyp1++; yyp2++; yyn--;          \
74
242M
   }                                  \
75
56.9M
}
76
77
78
56.9M
#define fmin(a,b) ((a) < (b)) ? (a) : (b)
79
80
159M
#define fpush(lz,hz) { stackLo[sp] = lz; \
81
159M
                       stackHi[sp] = hz; \
82
159M
                       sp++; }
83
84
159M
#define fpop(lz,hz) { sp--;              \
85
159M
                      lz = stackLo[sp];  \
86
159M
                      hz = stackHi[sp]; }
87
88
159M
#define FALLBACK_QSORT_SMALL_THRESH 10
89
#define FALLBACK_QSORT_STACK_SIZE   100
90
91
92
static
93
void fallbackQSort3 ( UInt32* fmap, 
94
                      UInt32* eclass,
95
                      Int32   loSt, 
96
                      Int32   hiSt )
97
102M
{
98
102M
   Int32 unLo, unHi, ltLo, gtHi, n, m;
99
102M
   Int32 sp, lo, hi;
100
102M
   UInt32 med, r, r3;
101
102M
   Int32 stackLo[FALLBACK_QSORT_STACK_SIZE];
102
102M
   Int32 stackHi[FALLBACK_QSORT_STACK_SIZE];
103
104
102M
   r = 0;
105
106
102M
   sp = 0;
107
102M
   fpush ( loSt, hiSt );
108
109
261M
   while (sp > 0) {
110
111
159M
      AssertH ( sp < FALLBACK_QSORT_STACK_SIZE - 1, 1004 );
112
113
159M
      fpop ( lo, hi );
114
159M
      if (hi - lo < FALLBACK_QSORT_SMALL_THRESH) {
115
128M
         fallbackSimpleSort ( fmap, eclass, lo, hi );
116
128M
         continue;
117
128M
      }
118
119
      /* Random partitioning.  Median of 3 sometimes fails to
120
         avoid bad cases.  Median of 9 seems to help but 
121
         looks rather expensive.  This too seems to work but
122
         is cheaper.  Guidance for the magic constants 
123
         7621 and 32768 is taken from Sedgewick's algorithms
124
         book, chapter 35.
125
      */
126
30.8M
      r = ((r * 7621) + 1) % 32768;
127
30.8M
      r3 = r % 3;
128
30.8M
      if (r3 == 0) med = eclass[fmap[lo]]; else
129
26.9M
      if (r3 == 1) med = eclass[fmap[(lo+hi)>>1]]; else
130
10.0M
                   med = eclass[fmap[hi]];
131
132
30.8M
      unLo = ltLo = lo;
133
30.8M
      unHi = gtHi = hi;
134
135
165M
      while (1) {
136
1.67G
         while (1) {
137
1.67G
            if (unLo > unHi) break;
138
1.66G
            n = (Int32)eclass[fmap[unLo]] - (Int32)med;
139
1.66G
            if (n == 0) { 
140
1.07G
               fswap(fmap[unLo], fmap[ltLo]); 
141
1.07G
               ltLo++; unLo++; 
142
1.07G
               continue; 
143
1.07G
            };
144
584M
            if (n > 0) break;
145
435M
            unLo++;
146
435M
         }
147
881M
         while (1) {
148
881M
            if (unLo > unHi) break;
149
851M
            n = (Int32)eclass[fmap[unHi]] - (Int32)med;
150
851M
            if (n == 0) { 
151
276M
               fswap(fmap[unHi], fmap[gtHi]); 
152
276M
               gtHi--; unHi--; 
153
276M
               continue; 
154
574M
            };
155
574M
            if (n < 0) break;
156
439M
            unHi--;
157
439M
         }
158
165M
         if (unLo > unHi) break;
159
134M
         fswap(fmap[unLo], fmap[unHi]); unLo++; unHi--;
160
134M
      }
161
162
30.8M
      AssertD ( unHi == unLo-1, "fallbackQSort3(2)" );
163
164
30.8M
      if (gtHi < ltLo) continue;
165
166
28.4M
      n = fmin(ltLo-lo, unLo-ltLo); fvswap(lo, unLo-n, n);
167
28.4M
      m = fmin(hi-gtHi, gtHi-unHi); fvswap(unLo, hi-m+1, m);
168
169
28.4M
      n = lo + unLo - ltLo - 1;
170
28.4M
      m = hi - (gtHi - unHi) + 1;
171
172
28.4M
      if (n - lo > hi - m) {
173
14.3M
         fpush ( lo, n );
174
14.3M
         fpush ( m, hi );
175
14.3M
      } else {
176
14.1M
         fpush ( m, hi );
177
14.1M
         fpush ( lo, n );
178
14.1M
      }
179
28.4M
   }
180
102M
}
181
182
#undef fmin
183
#undef fpush
184
#undef fpop
185
#undef fswap
186
#undef fvswap
187
#undef FALLBACK_QSORT_SMALL_THRESH
188
#undef FALLBACK_QSORT_STACK_SIZE
189
190
191
/*---------------------------------------------*/
192
/* Pre:
193
      nblock > 0
194
      eclass exists for [0 .. nblock-1]
195
      ((UChar*)eclass) [0 .. nblock-1] holds block
196
      ptr exists for [0 .. nblock-1]
197
198
   Post:
199
      ((UChar*)eclass) [0 .. nblock-1] holds block
200
      All other areas of eclass destroyed
201
      fmap [0 .. nblock-1] holds sorted order
202
      bhtab [ 0 .. 2+(nblock/32) ] destroyed
203
*/
204
205
320M
#define       SET_BH(zz)  bhtab[(zz) >> 5] |= ((UInt32)1 << ((zz) & 31))
206
3.43M
#define     CLEAR_BH(zz)  bhtab[(zz) >> 5] &= ~((UInt32)1 << ((zz) & 31))
207
4.28G
#define     ISSET_BH(zz)  (bhtab[(zz) >> 5] & ((UInt32)1 << ((zz) & 31)))
208
74.2M
#define      WORD_BH(zz)  bhtab[(zz) >> 5]
209
585M
#define UNALIGNED_BH(zz)  ((zz) & 0x01f)
210
211
static
212
void fallbackSort ( UInt32* fmap, 
213
                    UInt32* eclass, 
214
                    UInt32* bhtab,
215
                    Int32   nblock,
216
                    Int32   verb )
217
107k
{
218
107k
   Int32 ftab[257];
219
107k
   Int32 ftabCopy[256];
220
107k
   Int32 H, i, j, k, l, r, cc, cc1;
221
107k
   Int32 nNotDone;
222
107k
   Int32 nBhtab;
223
107k
   UChar* eclass8 = (UChar*)eclass;
224
225
   /*--
226
      Initial 1-char radix sort to generate
227
      initial fmap and initial BH bits.
228
   --*/
229
107k
   if (verb >= 4)
230
0
      VPrintf0 ( "        bucket sorting ...\n" );
231
27.6M
   for (i = 0; i < 257;    i++) ftab[i] = 0;
232
199M
   for (i = 0; i < nblock; i++) ftab[eclass8[i]]++;
233
27.5M
   for (i = 0; i < 256;    i++) ftabCopy[i] = ftab[i];
234
27.5M
   for (i = 1; i < 257;    i++) ftab[i] += ftab[i-1];
235
236
199M
   for (i = 0; i < nblock; i++) {
237
199M
      j = eclass8[i];
238
199M
      k = ftab[j] - 1;
239
199M
      ftab[j] = k;
240
199M
      fmap[k] = i;
241
199M
   }
242
243
107k
   nBhtab = 2 + (nblock / 32);
244
6.52M
   for (i = 0; i < nBhtab; i++) bhtab[i] = 0;
245
27.5M
   for (i = 0; i < 256; i++) SET_BH(ftab[i]);
246
247
   /*--
248
      Inductively refine the buckets.  Kind-of an
249
      "exponential radix sort" (!), inspired by the
250
      Manber-Myers suffix array construction algorithm.
251
   --*/
252
253
   /*-- set sentinel bits for block-end detection --*/
254
3.53M
   for (i = 0; i < 32; i++) { 
255
3.43M
      SET_BH(nblock + 2*i);
256
3.43M
      CLEAR_BH(nblock + 2*i + 1);
257
3.43M
   }
258
259
   /*-- the log(N) loop --*/
260
107k
   H = 1;
261
677k
   while (1) {
262
263
677k
      if (verb >= 4) 
264
0
         VPrintf1 ( "        depth %6d has ", H );
265
266
677k
      j = 0;
267
2.35G
      for (i = 0; i < nblock; i++) {
268
2.35G
         if (ISSET_BH(i)) j = i;
269
2.35G
         k = fmap[i] - H; if (k < 0) k += nblock;
270
2.35G
         eclass[k] = j;
271
2.35G
      }
272
273
677k
      nNotDone = 0;
274
677k
      r = -1;
275
102M
      while (1) {
276
277
   /*-- find the next non-singleton bucket --*/
278
102M
         k = r + 1;
279
376M
         while (ISSET_BH(k) && UNALIGNED_BH(k)) k++;
280
102M
         if (ISSET_BH(k)) {
281
22.4M
            while (WORD_BH(k) == 0xffffffff) k += 32;
282
81.7M
            while (ISSET_BH(k)) k++;
283
11.4M
         }
284
102M
         l = k - 1;
285
102M
         if (l >= nblock) break;
286
391M
         while (!ISSET_BH(k) && UNALIGNED_BH(k)) k++;
287
102M
         if (!ISSET_BH(k)) {
288
51.8M
            while (WORD_BH(k) == 0x00000000) k += 32;
289
107M
            while (!ISSET_BH(k)) k++;
290
11.8M
         }
291
102M
         r = k - 1;
292
102M
         if (r >= nblock) break;
293
294
         /*-- now [l, r] bracket current bucket --*/
295
102M
         if (r > l) {
296
102M
            nNotDone += (r - l + 1);
297
102M
            fallbackQSort3 ( fmap, eclass, l, r );
298
299
            /*-- scan bucket and generate header bits-- */
300
102M
            cc = -1;
301
1.86G
            for (i = l; i <= r; i++) {
302
1.76G
               cc1 = eclass[fmap[i]];
303
1.76G
               if (cc != cc1) { SET_BH(i); cc = cc1; };
304
1.76G
            }
305
102M
         }
306
102M
      }
307
308
677k
      if (verb >= 4) 
309
0
         VPrintf1 ( "%6d unresolved strings\n", nNotDone );
310
311
677k
      H *= 2;
312
677k
      if (H > nblock || nNotDone == 0) break;
313
677k
   }
314
315
   /*-- 
316
      Reconstruct the original block in
317
      eclass8 [0 .. nblock-1], since the
318
      previous phase destroyed it.
319
   --*/
320
107k
   if (verb >= 4)
321
0
      VPrintf0 ( "        reconstructing block ...\n" );
322
107k
   j = 0;
323
199M
   for (i = 0; i < nblock; i++) {
324
218M
      while (ftabCopy[j] == 0) j++;
325
199M
      ftabCopy[j]--;
326
199M
      eclass8[fmap[i]] = (UChar)j;
327
199M
   }
328
107k
   AssertH ( j < 256, 1005 );
329
107k
}
330
331
#undef       SET_BH
332
#undef     CLEAR_BH
333
#undef     ISSET_BH
334
#undef      WORD_BH
335
#undef UNALIGNED_BH
336
337
338
/*---------------------------------------------*/
339
/*--- The main, O(N^2 log(N)) sorting       ---*/
340
/*--- algorithm.  Faster for "normal"       ---*/
341
/*--- non-repetitive blocks.                ---*/
342
/*---------------------------------------------*/
343
344
/*---------------------------------------------*/
345
static
346
__inline__
347
Bool mainGtU ( UInt32  i1, 
348
               UInt32  i2,
349
               UChar*  block, 
350
               UInt16* quadrant,
351
               UInt32  nblock,
352
               Int32*  budget )
353
108M
{
354
108M
   Int32  k;
355
108M
   UChar  c1, c2;
356
108M
   UInt16 s1, s2;
357
358
108M
   AssertD ( i1 != i2, "mainGtU" );
359
   /* 1 */
360
108M
   c1 = block[i1]; c2 = block[i2];
361
108M
   if (c1 != c2) return (c1 > c2);
362
105M
   i1++; i2++;
363
   /* 2 */
364
105M
   c1 = block[i1]; c2 = block[i2];
365
105M
   if (c1 != c2) return (c1 > c2);
366
103M
   i1++; i2++;
367
   /* 3 */
368
103M
   c1 = block[i1]; c2 = block[i2];
369
103M
   if (c1 != c2) return (c1 > c2);
370
102M
   i1++; i2++;
371
   /* 4 */
372
102M
   c1 = block[i1]; c2 = block[i2];
373
102M
   if (c1 != c2) return (c1 > c2);
374
100M
   i1++; i2++;
375
   /* 5 */
376
100M
   c1 = block[i1]; c2 = block[i2];
377
100M
   if (c1 != c2) return (c1 > c2);
378
98.9M
   i1++; i2++;
379
   /* 6 */
380
98.9M
   c1 = block[i1]; c2 = block[i2];
381
98.9M
   if (c1 != c2) return (c1 > c2);
382
97.3M
   i1++; i2++;
383
   /* 7 */
384
97.3M
   c1 = block[i1]; c2 = block[i2];
385
97.3M
   if (c1 != c2) return (c1 > c2);
386
95.8M
   i1++; i2++;
387
   /* 8 */
388
95.8M
   c1 = block[i1]; c2 = block[i2];
389
95.8M
   if (c1 != c2) return (c1 > c2);
390
94.4M
   i1++; i2++;
391
   /* 9 */
392
94.4M
   c1 = block[i1]; c2 = block[i2];
393
94.4M
   if (c1 != c2) return (c1 > c2);
394
93.1M
   i1++; i2++;
395
   /* 10 */
396
93.1M
   c1 = block[i1]; c2 = block[i2];
397
93.1M
   if (c1 != c2) return (c1 > c2);
398
91.6M
   i1++; i2++;
399
   /* 11 */
400
91.6M
   c1 = block[i1]; c2 = block[i2];
401
91.6M
   if (c1 != c2) return (c1 > c2);
402
90.4M
   i1++; i2++;
403
   /* 12 */
404
90.4M
   c1 = block[i1]; c2 = block[i2];
405
90.4M
   if (c1 != c2) return (c1 > c2);
406
89.3M
   i1++; i2++;
407
408
89.3M
   k = nblock + 8;
409
410
1.04G
   do {
411
      /* 1 */
412
1.04G
      c1 = block[i1]; c2 = block[i2];
413
1.04G
      if (c1 != c2) return (c1 > c2);
414
1.03G
      s1 = quadrant[i1]; s2 = quadrant[i2];
415
1.03G
      if (s1 != s2) return (s1 > s2);
416
1.03G
      i1++; i2++;
417
      /* 2 */
418
1.03G
      c1 = block[i1]; c2 = block[i2];
419
1.03G
      if (c1 != c2) return (c1 > c2);
420
1.02G
      s1 = quadrant[i1]; s2 = quadrant[i2];
421
1.02G
      if (s1 != s2) return (s1 > s2);
422
1.01G
      i1++; i2++;
423
      /* 3 */
424
1.01G
      c1 = block[i1]; c2 = block[i2];
425
1.01G
      if (c1 != c2) return (c1 > c2);
426
1.00G
      s1 = quadrant[i1]; s2 = quadrant[i2];
427
1.00G
      if (s1 != s2) return (s1 > s2);
428
1.00G
      i1++; i2++;
429
      /* 4 */
430
1.00G
      c1 = block[i1]; c2 = block[i2];
431
1.00G
      if (c1 != c2) return (c1 > c2);
432
999M
      s1 = quadrant[i1]; s2 = quadrant[i2];
433
999M
      if (s1 != s2) return (s1 > s2);
434
994M
      i1++; i2++;
435
      /* 5 */
436
994M
      c1 = block[i1]; c2 = block[i2];
437
994M
      if (c1 != c2) return (c1 > c2);
438
990M
      s1 = quadrant[i1]; s2 = quadrant[i2];
439
990M
      if (s1 != s2) return (s1 > s2);
440
986M
      i1++; i2++;
441
      /* 6 */
442
986M
      c1 = block[i1]; c2 = block[i2];
443
986M
      if (c1 != c2) return (c1 > c2);
444
981M
      s1 = quadrant[i1]; s2 = quadrant[i2];
445
981M
      if (s1 != s2) return (s1 > s2);
446
973M
      i1++; i2++;
447
      /* 7 */
448
973M
      c1 = block[i1]; c2 = block[i2];
449
973M
      if (c1 != c2) return (c1 > c2);
450
969M
      s1 = quadrant[i1]; s2 = quadrant[i2];
451
969M
      if (s1 != s2) return (s1 > s2);
452
966M
      i1++; i2++;
453
      /* 8 */
454
966M
      c1 = block[i1]; c2 = block[i2];
455
966M
      if (c1 != c2) return (c1 > c2);
456
962M
      s1 = quadrant[i1]; s2 = quadrant[i2];
457
962M
      if (s1 != s2) return (s1 > s2);
458
955M
      i1++; i2++;
459
460
955M
      if (i1 >= nblock) i1 -= nblock;
461
955M
      if (i2 >= nblock) i2 -= nblock;
462
463
955M
      k -= 8;
464
955M
      (*budget)--;
465
955M
   }
466
955M
      while (k >= 0);
467
468
30.4k
   return False;
469
89.3M
}
470
471
472
/*---------------------------------------------*/
473
/*--
474
   Knuth's increments seem to work better
475
   than Incerpi-Sedgewick here.  Possibly
476
   because the number of elems to sort is
477
   usually small, typically <= 20.
478
--*/
479
static
480
Int32 incs[14] = { 1, 4, 13, 40, 121, 364, 1093, 3280,
481
                   9841, 29524, 88573, 265720,
482
                   797161, 2391484 };
483
484
static
485
void mainSimpleSort ( UInt32* ptr,
486
                      UChar*  block,
487
                      UInt16* quadrant,
488
                      Int32   nblock,
489
                      Int32   lo, 
490
                      Int32   hi, 
491
                      Int32   d,
492
                      Int32*  budget )
493
925k
{
494
925k
   Int32 i, j, h, bigN, hp;
495
925k
   UInt32 v;
496
497
925k
   bigN = hi - lo + 1;
498
925k
   if (bigN < 2) return;
499
500
718k
   hp = 0;
501
1.99M
   while (incs[hp] < bigN) hp++;
502
718k
   hp--;
503
504
1.95M
   for (; hp >= 0; hp--) {
505
1.24M
      h = incs[hp];
506
507
1.24M
      i = lo + h;
508
17.9M
      while (True) {
509
510
         /*-- copy 1 --*/
511
17.9M
         if (i > hi) break;
512
17.5M
         v = ptr[i];
513
17.5M
         j = i;
514
36.3M
         while ( mainGtU ( 
515
36.3M
                    ptr[j-h]+d, v+d, block, quadrant, nblock, budget 
516
36.3M
                 ) ) {
517
21.5M
            ptr[j] = ptr[j-h];
518
21.5M
            j = j - h;
519
21.5M
            if (j <= (lo + h - 1)) break;
520
21.5M
         }
521
17.5M
         ptr[j] = v;
522
17.5M
         i++;
523
524
         /*-- copy 2 --*/
525
17.5M
         if (i > hi) break;
526
17.0M
         v = ptr[i];
527
17.0M
         j = i;
528
36.1M
         while ( mainGtU ( 
529
36.1M
                    ptr[j-h]+d, v+d, block, quadrant, nblock, budget 
530
36.1M
                 ) ) {
531
21.4M
            ptr[j] = ptr[j-h];
532
21.4M
            j = j - h;
533
21.4M
            if (j <= (lo + h - 1)) break;
534
21.4M
         }
535
17.0M
         ptr[j] = v;
536
17.0M
         i++;
537
538
         /*-- copy 3 --*/
539
17.0M
         if (i > hi) break;
540
16.6M
         v = ptr[i];
541
16.6M
         j = i;
542
35.8M
         while ( mainGtU ( 
543
35.8M
                    ptr[j-h]+d, v+d, block, quadrant, nblock, budget 
544
35.8M
                 ) ) {
545
21.4M
            ptr[j] = ptr[j-h];
546
21.4M
            j = j - h;
547
21.4M
            if (j <= (lo + h - 1)) break;
548
21.4M
         }
549
16.6M
         ptr[j] = v;
550
16.6M
         i++;
551
552
16.6M
         if (*budget < 0) return;
553
16.6M
      }
554
1.24M
   }
555
718k
}
556
557
558
/*---------------------------------------------*/
559
/*--
560
   The following is an implementation of
561
   an elegant 3-way quicksort for strings,
562
   described in a paper "Fast Algorithms for
563
   Sorting and Searching Strings", by Robert
564
   Sedgewick and Jon L. Bentley.
565
--*/
566
567
#define mswap(zz1, zz2) \
568
243M
   { Int32 zztmp = zz1; zz1 = zz2; zz2 = zztmp; }
569
570
413k
#define mvswap(zzp1, zzp2, zzn)       \
571
413k
{                                     \
572
413k
   Int32 yyp1 = (zzp1);               \
573
413k
   Int32 yyp2 = (zzp2);               \
574
413k
   Int32 yyn  = (zzn);                \
575
8.23M
   while (yyn > 0) {                  \
576
7.81M
      mswap(ptr[yyp1], ptr[yyp2]);    \
577
7.81M
      yyp1++; yyp2++; yyn--;          \
578
7.81M
   }                                  \
579
413k
}
580
581
static 
582
__inline__
583
UChar mmed3 ( UChar a, UChar b, UChar c )
584
971k
{
585
971k
   UChar t;
586
971k
   if (a > b) { t = a; a = b; b = t; };
587
971k
   if (b > c) { 
588
43.1k
      b = c;
589
43.1k
      if (a > b) b = a;
590
43.1k
   }
591
971k
   return b;
592
971k
}
593
594
413k
#define mmin(a,b) ((a) < (b)) ? (a) : (b)
595
596
1.89M
#define mpush(lz,hz,dz) { stackLo[sp] = lz; \
597
1.89M
                          stackHi[sp] = hz; \
598
1.89M
                          stackD [sp] = dz; \
599
1.89M
                          sp++; }
600
601
1.89M
#define mpop(lz,hz,dz) { sp--;             \
602
1.89M
                         lz = stackLo[sp]; \
603
1.89M
                         hz = stackHi[sp]; \
604
1.89M
                         dz = stackD [sp]; }
605
606
607
1.23M
#define mnextsize(az) (nextHi[az]-nextLo[az])
608
609
#define mnextswap(az,bz)                                        \
610
434k
   { Int32 tz;                                                  \
611
434k
     tz = nextLo[az]; nextLo[az] = nextLo[bz]; nextLo[bz] = tz; \
612
434k
     tz = nextHi[az]; nextHi[az] = nextHi[bz]; nextHi[bz] = tz; \
613
434k
     tz = nextD [az]; nextD [az] = nextD [bz]; nextD [bz] = tz; }
614
615
616
3.79M
#define MAIN_QSORT_SMALL_THRESH 20
617
1.06M
#define MAIN_QSORT_DEPTH_THRESH (BZ_N_RADIX + BZ_N_QSORT)
618
#define MAIN_QSORT_STACK_SIZE 100
619
620
static
621
void mainQSort3 ( UInt32* ptr,
622
                  UChar*  block,
623
                  UInt16* quadrant,
624
                  Int32   nblock,
625
                  Int32   loSt, 
626
                  Int32   hiSt, 
627
                  Int32   dSt,
628
                  Int32*  budget )
629
513k
{
630
513k
   Int32 unLo, unHi, ltLo, gtHi, n, m, med;
631
513k
   Int32 sp, lo, hi, d;
632
633
513k
   Int32 stackLo[MAIN_QSORT_STACK_SIZE];
634
513k
   Int32 stackHi[MAIN_QSORT_STACK_SIZE];
635
513k
   Int32 stackD [MAIN_QSORT_STACK_SIZE];
636
637
513k
   Int32 nextLo[3];
638
513k
   Int32 nextHi[3];
639
513k
   Int32 nextD [3];
640
641
513k
   sp = 0;
642
513k
   mpush ( loSt, hiSt, dSt );
643
644
2.40M
   while (sp > 0) {
645
646
1.89M
      AssertH ( sp < MAIN_QSORT_STACK_SIZE - 2, 1001 );
647
648
1.89M
      mpop ( lo, hi, d );
649
1.89M
      if (hi - lo < MAIN_QSORT_SMALL_THRESH || 
650
1.06M
          d > MAIN_QSORT_DEPTH_THRESH) {
651
925k
         mainSimpleSort ( ptr, block, quadrant, nblock, lo, hi, d, budget );
652
925k
         if (*budget < 0) return;
653
918k
         continue;
654
925k
      }
655
656
971k
      med = (Int32) 
657
971k
            mmed3 ( block[ptr[ lo         ]+d],
658
971k
                    block[ptr[ hi         ]+d],
659
971k
                    block[ptr[ (lo+hi)>>1 ]+d] );
660
661
971k
      unLo = ltLo = lo;
662
971k
      unHi = gtHi = hi;
663
664
1.39M
      while (True) {
665
215M
         while (True) {
666
215M
            if (unLo > unHi) break;
667
214M
            n = ((Int32)block[ptr[unLo]+d]) - med;
668
214M
            if (n == 0) { 
669
209M
               mswap(ptr[unLo], ptr[ltLo]); 
670
209M
               ltLo++; unLo++; continue; 
671
209M
            };
672
5.33M
            if (n >  0) break;
673
4.83M
            unLo++;
674
4.83M
         }
675
31.6M
         while (True) {
676
31.6M
            if (unLo > unHi) break;
677
30.7M
            n = ((Int32)block[ptr[unHi]+d]) - med;
678
30.7M
            if (n == 0) { 
679
25.9M
               mswap(ptr[unHi], ptr[gtHi]); 
680
25.9M
               gtHi--; unHi--; continue; 
681
25.9M
            };
682
4.80M
            if (n <  0) break;
683
4.38M
            unHi--;
684
4.38M
         }
685
1.39M
         if (unLo > unHi) break;
686
423k
         mswap(ptr[unLo], ptr[unHi]); unLo++; unHi--;
687
423k
      }
688
689
971k
      AssertD ( unHi == unLo-1, "mainQSort3(2)" );
690
691
971k
      if (gtHi < ltLo) {
692
764k
         mpush(lo, hi, d+1 );
693
764k
         continue;
694
764k
      }
695
696
206k
      n = mmin(ltLo-lo, unLo-ltLo); mvswap(lo, unLo-n, n);
697
206k
      m = mmin(hi-gtHi, gtHi-unHi); mvswap(unLo, hi-m+1, m);
698
699
206k
      n = lo + unLo - ltLo - 1;
700
206k
      m = hi - (gtHi - unHi) + 1;
701
702
206k
      nextLo[0] = lo;  nextHi[0] = n;   nextD[0] = d;
703
206k
      nextLo[1] = m;   nextHi[1] = hi;  nextD[1] = d;
704
206k
      nextLo[2] = n+1; nextHi[2] = m-1; nextD[2] = d+1;
705
706
206k
      if (mnextsize(0) < mnextsize(1)) mnextswap(0,1);
707
206k
      if (mnextsize(1) < mnextsize(2)) mnextswap(1,2);
708
206k
      if (mnextsize(0) < mnextsize(1)) mnextswap(0,1);
709
710
206k
      AssertD (mnextsize(0) >= mnextsize(1), "mainQSort3(8)" );
711
206k
      AssertD (mnextsize(1) >= mnextsize(2), "mainQSort3(9)" );
712
713
206k
      mpush (nextLo[0], nextHi[0], nextD[0]);
714
206k
      mpush (nextLo[1], nextHi[1], nextD[1]);
715
206k
      mpush (nextLo[2], nextHi[2], nextD[2]);
716
206k
   }
717
513k
}
718
719
#undef mswap
720
#undef mvswap
721
#undef mpush
722
#undef mpop
723
#undef mmin
724
#undef mnextsize
725
#undef mnextswap
726
#undef MAIN_QSORT_SMALL_THRESH
727
#undef MAIN_QSORT_DEPTH_THRESH
728
#undef MAIN_QSORT_STACK_SIZE
729
730
731
/*---------------------------------------------*/
732
/* Pre:
733
      nblock > N_OVERSHOOT
734
      block32 exists for [0 .. nblock-1 +N_OVERSHOOT]
735
      ((UChar*)block32) [0 .. nblock-1] holds block
736
      ptr exists for [0 .. nblock-1]
737
738
   Post:
739
      ((UChar*)block32) [0 .. nblock-1] holds block
740
      All other areas of block32 destroyed
741
      ftab [0 .. 65536 ] destroyed
742
      ptr [0 .. nblock-1] holds sorted order
743
      if (*budget < 0), sorting was abandoned
744
*/
745
746
23.2M
#define BIGFREQ(b) (ftab[((b)+1) << 8] - ftab[(b) << 8])
747
3.54G
#define SETMASK (1 << 21)
748
1.78G
#define CLEARMASK (~(SETMASK))
749
750
static
751
void mainSort ( UInt32* ptr, 
752
                UChar*  block,
753
                UInt16* quadrant, 
754
                UInt32* ftab,
755
                Int32   nblock,
756
                Int32   verb,
757
                Int32*  budget )
758
9.12k
{
759
9.12k
   Int32  i, j, k, ss, sb;
760
9.12k
   Int32  runningOrder[256];
761
9.12k
   Bool   bigDone[256];
762
9.12k
   Int32  copyStart[256];
763
9.12k
   Int32  copyEnd  [256];
764
9.12k
   UChar  c1;
765
9.12k
   Int32  numQSorted;
766
9.12k
   UInt16 s;
767
9.12k
   if (verb >= 4) VPrintf0 ( "        main sort initialise ...\n" );
768
769
   /*-- set up the 2-byte frequency table --*/
770
598M
   for (i = 65536; i >= 0; i--) ftab[i] = 0;
771
772
9.12k
   j = block[0] << 8;
773
9.12k
   i = nblock-1;
774
29.7M
   for (; i >= 3; i -= 4) {
775
29.7M
      quadrant[i] = 0;
776
29.7M
      j = (j >> 8) | ( ((UInt16)block[i]) << 8);
777
29.7M
      ftab[j]++;
778
29.7M
      quadrant[i-1] = 0;
779
29.7M
      j = (j >> 8) | ( ((UInt16)block[i-1]) << 8);
780
29.7M
      ftab[j]++;
781
29.7M
      quadrant[i-2] = 0;
782
29.7M
      j = (j >> 8) | ( ((UInt16)block[i-2]) << 8);
783
29.7M
      ftab[j]++;
784
29.7M
      quadrant[i-3] = 0;
785
29.7M
      j = (j >> 8) | ( ((UInt16)block[i-3]) << 8);
786
29.7M
      ftab[j]++;
787
29.7M
   }
788
20.7k
   for (; i >= 0; i--) {
789
11.6k
      quadrant[i] = 0;
790
11.6k
      j = (j >> 8) | ( ((UInt16)block[i]) << 8);
791
11.6k
      ftab[j]++;
792
11.6k
   }
793
794
   /*-- (emphasises close relationship of block & quadrant) --*/
795
319k
   for (i = 0; i < BZ_N_OVERSHOOT; i++) {
796
310k
      block   [nblock+i] = block[i];
797
310k
      quadrant[nblock+i] = 0;
798
310k
   }
799
800
9.12k
   if (verb >= 4) VPrintf0 ( "        bucket sorting ...\n" );
801
802
   /*-- Complete the initial radix sort --*/
803
598M
   for (i = 1; i <= 65536; i++) ftab[i] += ftab[i-1];
804
805
9.12k
   s = block[0] << 8;
806
9.12k
   i = nblock-1;
807
29.7M
   for (; i >= 3; i -= 4) {
808
29.7M
      s = (s >> 8) | (block[i] << 8);
809
29.7M
      j = ftab[s] -1;
810
29.7M
      ftab[s] = j;
811
29.7M
      ptr[j] = i;
812
29.7M
      s = (s >> 8) | (block[i-1] << 8);
813
29.7M
      j = ftab[s] -1;
814
29.7M
      ftab[s] = j;
815
29.7M
      ptr[j] = i-1;
816
29.7M
      s = (s >> 8) | (block[i-2] << 8);
817
29.7M
      j = ftab[s] -1;
818
29.7M
      ftab[s] = j;
819
29.7M
      ptr[j] = i-2;
820
29.7M
      s = (s >> 8) | (block[i-3] << 8);
821
29.7M
      j = ftab[s] -1;
822
29.7M
      ftab[s] = j;
823
29.7M
      ptr[j] = i-3;
824
29.7M
   }
825
20.7k
   for (; i >= 0; i--) {
826
11.6k
      s = (s >> 8) | (block[i] << 8);
827
11.6k
      j = ftab[s] -1;
828
11.6k
      ftab[s] = j;
829
11.6k
      ptr[j] = i;
830
11.6k
   }
831
832
   /*--
833
      Now ftab contains the first loc of every small bucket.
834
      Calculate the running order, from smallest to largest
835
      big bucket.
836
   --*/
837
2.34M
   for (i = 0; i <= 255; i++) {
838
2.33M
      bigDone     [i] = False;
839
2.33M
      runningOrder[i] = i;
840
2.33M
   }
841
842
9.12k
   {
843
9.12k
      Int32 vv;
844
9.12k
      Int32 h = 1;
845
45.6k
      do h = 3 * h + 1; while (h <= 256);
846
45.6k
      do {
847
45.6k
         h = h / 3;
848
10.0M
         for (i = h; i <= 255; i++) {
849
10.0M
            vv = runningOrder[i];
850
10.0M
            j = i;
851
11.6M
            while ( BIGFREQ(runningOrder[j-h]) > BIGFREQ(vv) ) {
852
1.75M
               runningOrder[j] = runningOrder[j-h];
853
1.75M
               j = j - h;
854
1.75M
               if (j <= (h - 1)) goto zero;
855
1.75M
            }
856
10.0M
            zero:
857
10.0M
            runningOrder[j] = vv;
858
10.0M
         }
859
45.6k
      } while (h != 1);
860
9.12k
   }
861
862
   /*--
863
      The main sorting loop.
864
   --*/
865
866
9.12k
   numQSorted = 0;
867
868
2.30M
   for (i = 0; i <= 255; i++) {
869
870
      /*--
871
         Process big buckets, starting with the least full.
872
         Basically this is a 3-step process in which we call
873
         mainQSort3 to sort the small buckets [ss, j], but
874
         also make a big effort to avoid the calls if we can.
875
      --*/
876
2.30M
      ss = runningOrder[i];
877
878
      /*--
879
         Step 1:
880
         Complete the big bucket [ss] by quicksorting
881
         any unsorted small buckets [ss, j], for j != ss.  
882
         Hopefully previous pointer-scanning phases have already
883
         completed many of the small buckets [ss, j], so
884
         we don't have to sort them at all.
885
      --*/
886
591M
      for (j = 0; j <= 255; j++) {
887
589M
         if (j != ss) {
888
586M
            sb = (ss << 8) + j;
889
586M
            if ( ! (ftab[sb] & SETMASK) ) {
890
297M
               Int32 lo = ftab[sb]   & CLEARMASK;
891
297M
               Int32 hi = (ftab[sb+1] & CLEARMASK) - 1;
892
297M
               if (hi > lo) {
893
513k
                  if (verb >= 4)
894
0
                     VPrintf4 ( "        qsort [0x%x, 0x%x]   "
895
513k
                                "done %d   this %d\n",
896
513k
                                ss, j, numQSorted, hi - lo + 1 );
897
513k
                  mainQSort3 ( 
898
513k
                     ptr, block, quadrant, nblock, 
899
513k
                     lo, hi, BZ_N_RADIX, budget 
900
513k
                  );   
901
513k
                  numQSorted += (hi - lo + 1);
902
513k
                  if (*budget < 0) return;
903
513k
               }
904
297M
            }
905
586M
            ftab[sb] |= SETMASK;
906
586M
         }
907
589M
      }
908
909
2.29M
      AssertH ( !bigDone[ss], 1006 );
910
911
      /*--
912
         Step 2:
913
         Now scan this big bucket [ss] so as to synthesise the
914
         sorted order for small buckets [t, ss] for all t,
915
         including, magically, the bucket [ss,ss] too.
916
         This will avoid doing Real Work in subsequent Step 1's.
917
      --*/
918
2.29M
      {
919
590M
         for (j = 0; j <= 255; j++) {
920
588M
            copyStart[j] =  ftab[(j << 8) + ss]     & CLEARMASK;
921
588M
            copyEnd  [j] = (ftab[(j << 8) + ss + 1] & CLEARMASK) - 1;
922
588M
         }
923
14.3M
         for (j = ftab[ss << 8] & CLEARMASK; j < copyStart[ss]; j++) {
924
12.0M
            k = ptr[j]-1; if (k < 0) k += nblock;
925
12.0M
            c1 = block[k];
926
12.0M
            if (!bigDone[c1])
927
7.47M
               ptr[ copyStart[c1]++ ] = k;
928
12.0M
         }
929
17.7M
         for (j = (ftab[(ss+1) << 8] & CLEARMASK) - 1; j > copyEnd[ss]; j--) {
930
15.4M
            k = ptr[j]-1; if (k < 0) k += nblock;
931
15.4M
            c1 = block[k];
932
15.4M
            if (!bigDone[c1]) 
933
9.36M
               ptr[ copyEnd[c1]-- ] = k;
934
15.4M
         }
935
2.29M
      }
936
937
2.29M
      AssertH ( (copyStart[ss]-1 == copyEnd[ss])
938
2.29M
                || 
939
                /* Extremely rare case missing in bzip2-1.0.0 and 1.0.1.
940
                   Necessity for this case is demonstrated by compressing 
941
                   a sequence of approximately 48.5 million of character 
942
                   251; 1.0.0/1.0.1 will then die here. */
943
2.29M
                (copyStart[ss] == 0 && copyEnd[ss] == nblock-1),
944
2.29M
                1007 )
945
946
590M
      for (j = 0; j <= 255; j++) ftab[(j << 8) + ss] |= SETMASK;
947
948
      /*--
949
         Step 3:
950
         The [ss] big bucket is now done.  Record this fact,
951
         and update the quadrant descriptors.  Remember to
952
         update quadrants in the overshoot area too, if
953
         necessary.  The "if (i < 255)" test merely skips
954
         this updating for the last bucket processed, since
955
         updating for the last bucket is pointless.
956
957
         The quadrant array provides a way to incrementally
958
         cache sort orderings, as they appear, so as to 
959
         make subsequent comparisons in fullGtU() complete
960
         faster.  For repetitive blocks this makes a big
961
         difference (but not big enough to be able to avoid
962
         the fallback sorting mechanism, exponential radix sort).
963
964
         The precise meaning is: at all times:
965
966
            for 0 <= i < nblock and 0 <= j <= nblock
967
968
            if block[i] != block[j], 
969
970
               then the relative values of quadrant[i] and 
971
                    quadrant[j] are meaningless.
972
973
               else {
974
                  if quadrant[i] < quadrant[j]
975
                     then the string starting at i lexicographically
976
                     precedes the string starting at j
977
978
                  else if quadrant[i] > quadrant[j]
979
                     then the string starting at j lexicographically
980
                     precedes the string starting at i
981
982
                  else
983
                     the relative ordering of the strings starting
984
                     at i and j has not yet been determined.
985
               }
986
      --*/
987
2.29M
      bigDone[ss] = True;
988
989
2.29M
      if (i < 255) {
990
2.29M
         Int32 bbStart  = ftab[ss << 8] & CLEARMASK;
991
2.29M
         Int32 bbSize   = (ftab[(ss+1) << 8] & CLEARMASK) - bbStart;
992
2.29M
         Int32 shifts   = 0;
993
994
2.29M
         while ((bbSize >> shifts) > 65534) shifts++;
995
996
21.1M
         for (j = bbSize-1; j >= 0; j--) {
997
18.8M
            Int32 a2update     = ptr[bbStart + j];
998
18.8M
            UInt16 qVal        = (UInt16)(j >> shifts);
999
18.8M
            quadrant[a2update] = qVal;
1000
18.8M
            if (a2update < BZ_N_OVERSHOOT)
1001
57.4k
               quadrant[a2update + nblock] = qVal;
1002
18.8M
         }
1003
2.29M
         AssertH ( ((bbSize-1) >> shifts) <= 65535, 1002 );
1004
2.29M
      }
1005
1006
2.29M
   }
1007
1008
2.12k
   if (verb >= 4)
1009
0
      VPrintf3 ( "        %d pointers, %d sorted, %d scanned\n",
1010
2.12k
                 nblock, numQSorted, nblock - numQSorted );
1011
2.12k
}
1012
1013
#undef BIGFREQ
1014
#undef SETMASK
1015
#undef CLEARMASK
1016
1017
1018
/*---------------------------------------------*/
1019
/* Pre:
1020
      nblock > 0
1021
      arr2 exists for [0 .. nblock-1 +N_OVERSHOOT]
1022
      ((UChar*)arr2)  [0 .. nblock-1] holds block
1023
      arr1 exists for [0 .. nblock-1]
1024
1025
   Post:
1026
      ((UChar*)arr2) [0 .. nblock-1] holds block
1027
      All other areas of block destroyed
1028
      ftab [ 0 .. 65536 ] destroyed
1029
      arr1 [0 .. nblock-1] holds sorted order
1030
*/
1031
void BZ2_blockSort ( EState* s )
1032
109k
{
1033
109k
   UInt32* ptr    = s->ptr; 
1034
109k
   UChar*  block  = s->block;
1035
109k
   UInt32* ftab   = s->ftab;
1036
109k
   Int32   nblock = s->nblock;
1037
109k
   Int32   verb   = s->verbosity;
1038
109k
   Int32   wfact  = s->workFactor;
1039
109k
   UInt16* quadrant;
1040
109k
   Int32   budget;
1041
109k
   Int32   budgetInit;
1042
109k
   Int32   i;
1043
1044
109k
   if (nblock < 10000) {
1045
100k
      fallbackSort ( s->arr1, s->arr2, ftab, nblock, verb );
1046
100k
   } else {
1047
      /* Calculate the location for quadrant, remembering to get
1048
         the alignment right.  Assumes that &(block[0]) is at least
1049
         2-byte aligned -- this should be ok since block is really
1050
         the first section of arr2.
1051
      */
1052
9.12k
      i = nblock+BZ_N_OVERSHOOT;
1053
9.12k
      if (i & 1) i++;
1054
9.12k
      quadrant = (UInt16*)(&(block[i]));
1055
1056
      /* (wfact-1) / 3 puts the default-factor-30
1057
         transition point at very roughly the same place as 
1058
         with v0.1 and v0.9.0.  
1059
         Not that it particularly matters any more, since the
1060
         resulting compressed stream is now the same regardless
1061
         of whether or not we use the main sort or fallback sort.
1062
      */
1063
9.12k
      if (wfact < 1  ) wfact = 1;
1064
9.12k
      if (wfact > 100) wfact = 100;
1065
9.12k
      budgetInit = nblock * ((wfact-1) / 3);
1066
9.12k
      budget = budgetInit;
1067
1068
9.12k
      mainSort ( ptr, block, quadrant, ftab, nblock, verb, &budget );
1069
9.12k
      if (verb >= 3) 
1070
0
         VPrintf3 ( "      %d work, %d block, ratio %5.2f\n",
1071
9.12k
                    budgetInit - budget,
1072
9.12k
                    nblock, 
1073
9.12k
                    (float)(budgetInit - budget) /
1074
9.12k
                    (float)(nblock==0 ? 1 : nblock) ); 
1075
9.12k
      if (budget < 0) {
1076
7.00k
         if (verb >= 2) 
1077
0
            VPrintf0 ( "    too repetitive; using fallback"
1078
7.00k
                       " sorting algorithm\n" );
1079
7.00k
         fallbackSort ( s->arr1, s->arr2, ftab, nblock, verb );
1080
7.00k
      }
1081
9.12k
   }
1082
1083
109k
   s->origPtr = -1;
1084
108M
   for (i = 0; i < s->nblock; i++)
1085
108M
      if (ptr[i] == 0)
1086
109k
         { s->origPtr = i; break; };
1087
1088
109k
   AssertH( s->origPtr != -1, 1003 );
1089
109k
}
1090
1091
1092
/*-------------------------------------------------------------*/
1093
/*--- end                                       blocksort.c ---*/
1094
/*-------------------------------------------------------------*/