Coverage Report

Created: 2025-11-30 06:38

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/src/cpython/Python/dtoa.c
Line
Count
Source
1
/****************************************************************
2
 *
3
 * The author of this software is David M. Gay.
4
 *
5
 * Copyright (c) 1991, 2000, 2001 by Lucent Technologies.
6
 *
7
 * Permission to use, copy, modify, and distribute this software for any
8
 * purpose without fee is hereby granted, provided that this entire notice
9
 * is included in all copies of any software which is or includes a copy
10
 * or modification of this software and in all copies of the supporting
11
 * documentation for such software.
12
 *
13
 * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED
14
 * WARRANTY.  IN PARTICULAR, NEITHER THE AUTHOR NOR LUCENT MAKES ANY
15
 * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY
16
 * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE.
17
 *
18
 ***************************************************************/
19
20
/****************************************************************
21
 * This is dtoa.c by David M. Gay, downloaded from
22
 * http://www.netlib.org/fp/dtoa.c on April 15, 2009 and modified for
23
 * inclusion into the Python core by Mark E. T. Dickinson and Eric V. Smith.
24
 *
25
 * Please remember to check http://www.netlib.org/fp regularly (and especially
26
 * before any Python release) for bugfixes and updates.
27
 *
28
 * The major modifications from Gay's original code are as follows:
29
 *
30
 *  0. The original code has been specialized to Python's needs by removing
31
 *     many of the #ifdef'd sections.  In particular, code to support VAX and
32
 *     IBM floating-point formats, hex NaNs, hex floats, locale-aware
33
 *     treatment of the decimal point, and setting of the inexact flag have
34
 *     been removed.
35
 *
36
 *  1. We use PyMem_Malloc and PyMem_Free in place of malloc and free.
37
 *
38
 *  2. The public functions strtod, dtoa and freedtoa all now have
39
 *     a _Py_dg_ prefix.
40
 *
41
 *  3. Instead of assuming that PyMem_Malloc always succeeds, we thread
42
 *     PyMem_Malloc failures through the code.  The functions
43
 *
44
 *       Balloc, multadd, s2b, i2b, mult, pow5mult, lshift, diff, d2b
45
 *
46
 *     of return type *Bigint all return NULL to indicate a malloc failure.
47
 *     Similarly, rv_alloc and nrv_alloc (return type char *) return NULL on
48
 *     failure.  bigcomp now has return type int (it used to be void) and
49
 *     returns -1 on failure and 0 otherwise.  _Py_dg_dtoa returns NULL
50
 *     on failure.  _Py_dg_strtod indicates failure due to malloc failure
51
 *     by returning -1.0, setting errno=ENOMEM and *se to s00.
52
 *
53
 *  4. The static variable dtoa_result has been removed.  Callers of
54
 *     _Py_dg_dtoa are expected to call _Py_dg_freedtoa to free
55
 *     the memory allocated by _Py_dg_dtoa.
56
 *
57
 *  5. The code has been reformatted to better fit with Python's
58
 *     C style guide (PEP 7).
59
 *
60
 *  6. A bug in the memory allocation has been fixed: to avoid FREEing memory
61
 *     that hasn't been MALLOC'ed, private_mem should only be used when k <=
62
 *     Kmax.
63
 *
64
 *  7. _Py_dg_strtod has been modified so that it doesn't accept strings with
65
 *     leading whitespace.
66
 *
67
 *  8. A corner case where _Py_dg_dtoa didn't strip trailing zeros has been
68
 *     fixed. (bugs.python.org/issue40780)
69
 *
70
 ***************************************************************/
71
72
/* Please send bug reports for the original dtoa.c code to David M. Gay (dmg
73
 * at acm dot org, with " at " changed at "@" and " dot " changed to ".").
74
 * Please report bugs for this modified version using the Python issue tracker
75
 * as detailed at (https://devguide.python.org/triage/issue-tracker/). */
76
77
/* On a machine with IEEE extended-precision registers, it is
78
 * necessary to specify double-precision (53-bit) rounding precision
79
 * before invoking strtod or dtoa.  If the machine uses (the equivalent
80
 * of) Intel 80x87 arithmetic, the call
81
 *      _control87(PC_53, MCW_PC);
82
 * does this with many compilers.  Whether this or another call is
83
 * appropriate depends on the compiler; for this to work, it may be
84
 * necessary to #include "float.h" or another system-dependent header
85
 * file.
86
 */
87
88
/* strtod for IEEE-, VAX-, and IBM-arithmetic machines.
89
 *
90
 * This strtod returns a nearest machine number to the input decimal
91
 * string (or sets errno to ERANGE).  With IEEE arithmetic, ties are
92
 * broken by the IEEE round-even rule.  Otherwise ties are broken by
93
 * biased rounding (add half and chop).
94
 *
95
 * Inspired loosely by William D. Clinger's paper "How to Read Floating
96
 * Point Numbers Accurately" [Proc. ACM SIGPLAN '90, pp. 92-101].
97
 *
98
 * Modifications:
99
 *
100
 *      1. We only require IEEE, IBM, or VAX double-precision
101
 *              arithmetic (not IEEE double-extended).
102
 *      2. We get by with floating-point arithmetic in a case that
103
 *              Clinger missed -- when we're computing d * 10^n
104
 *              for a small integer d and the integer n is not too
105
 *              much larger than 22 (the maximum integer k for which
106
 *              we can represent 10^k exactly), we may be able to
107
 *              compute (d*10^k) * 10^(e-k) with just one roundoff.
108
 *      3. Rather than a bit-at-a-time adjustment of the binary
109
 *              result in the hard case, we use floating-point
110
 *              arithmetic to determine the adjustment to within
111
 *              one bit; only in really hard cases do we need to
112
 *              compute a second residual.
113
 *      4. Because of 3., we don't need a large table of powers of 10
114
 *              for ten-to-e (just some small tables, e.g. of 10^k
115
 *              for 0 <= k <= 22).
116
 */
117
118
/* Linking of Python's #defines to Gay's #defines starts here. */
119
120
#include "Python.h"
121
#include "pycore_dtoa.h"          // _PY_SHORT_FLOAT_REPR
122
#include "pycore_interp_structs.h"// struct Bigint
123
#include "pycore_pystate.h"       // _PyInterpreterState_GET()
124
#include <stdlib.h>               // exit()
125
126
127
/* if _PY_SHORT_FLOAT_REPR == 0, then don't even try to compile
128
   the following code */
129
#if _PY_SHORT_FLOAT_REPR == 1
130
131
#include "float.h"
132
133
56
#define MALLOC PyMem_Malloc
134
0
#define FREE PyMem_Free
135
136
/* This code should also work for ARM mixed-endian format on little-endian
137
   machines, where doubles have byte order 45670123 (in increasing address
138
   order, 0 being the least significant byte). */
139
#ifdef DOUBLE_IS_LITTLE_ENDIAN_IEEE754
140
#  define IEEE_8087
141
#endif
142
#if defined(DOUBLE_IS_BIG_ENDIAN_IEEE754) ||  \
143
  defined(DOUBLE_IS_ARM_MIXED_ENDIAN_IEEE754)
144
#  define IEEE_MC68k
145
#endif
146
#if defined(IEEE_8087) + defined(IEEE_MC68k) != 1
147
#error "Exactly one of IEEE_8087 or IEEE_MC68k should be defined."
148
#endif
149
150
/* The code below assumes that the endianness of integers matches the
151
   endianness of the two 32-bit words of a double.  Check this. */
152
#if defined(WORDS_BIGENDIAN) && (defined(DOUBLE_IS_LITTLE_ENDIAN_IEEE754) || \
153
                                 defined(DOUBLE_IS_ARM_MIXED_ENDIAN_IEEE754))
154
#error "doubles and ints have incompatible endianness"
155
#endif
156
157
#if !defined(WORDS_BIGENDIAN) && defined(DOUBLE_IS_BIG_ENDIAN_IEEE754)
158
#error "doubles and ints have incompatible endianness"
159
#endif
160
161
162
typedef uint32_t ULong;
163
typedef int32_t Long;
164
typedef uint64_t ULLong;
165
166
#undef DEBUG
167
#ifdef Py_DEBUG
168
#define DEBUG
169
#endif
170
171
/* End Python #define linking */
172
173
#ifdef DEBUG
174
#define Bug(x) {fprintf(stderr, "%s\n", x); exit(1);}
175
#endif
176
177
typedef union { double d; ULong L[2]; } U;
178
179
#ifdef IEEE_8087
180
1.96M
#define word0(x) (x)->L[1]
181
1.33M
#define word1(x) (x)->L[0]
182
#else
183
#define word0(x) (x)->L[0]
184
#define word1(x) (x)->L[1]
185
#endif
186
5.15M
#define dval(x) (x)->d
187
188
#ifndef STRTOD_DIGLIM
189
104k
#define STRTOD_DIGLIM 40
190
#endif
191
192
/* maximum permitted exponent value for strtod; exponents larger than
193
   MAX_ABS_EXP in absolute value get truncated to +-MAX_ABS_EXP.  MAX_ABS_EXP
194
   should fit into an int. */
195
#ifndef MAX_ABS_EXP
196
735k
#define MAX_ABS_EXP 1100000000U
197
#endif
198
/* Bound on length of pieces of input strings in _Py_dg_strtod; specifically,
199
   this is used to bound the total number of digits ignoring leading zeros and
200
   the number of digits that follow the decimal point.  Ideally, MAX_DIGITS
201
   should satisfy MAX_DIGITS + 400 < MAX_ABS_EXP; that ensures that the
202
   exponent clipping in _Py_dg_strtod can't affect the value of the output. */
203
#ifndef MAX_DIGITS
204
2.48M
#define MAX_DIGITS 1000000000U
205
#endif
206
207
/* Guard against trying to use the above values on unusual platforms with ints
208
 * of width less than 32 bits. */
209
#if MAX_ABS_EXP > INT_MAX
210
#error "MAX_ABS_EXP should fit in an int"
211
#endif
212
#if MAX_DIGITS > INT_MAX
213
#error "MAX_DIGITS should fit in an int"
214
#endif
215
216
/* The following definition of Storeinc is appropriate for MIPS processors.
217
 * An alternative that might be better on some machines is
218
 * #define Storeinc(a,b,c) (*a++ = b << 16 | c & 0xffff)
219
 */
220
#if defined(IEEE_8087)
221
#define Storeinc(a,b,c) (((unsigned short *)a)[1] = (unsigned short)b,  \
222
                         ((unsigned short *)a)[0] = (unsigned short)c, a++)
223
#else
224
#define Storeinc(a,b,c) (((unsigned short *)a)[0] = (unsigned short)b,  \
225
                         ((unsigned short *)a)[1] = (unsigned short)c, a++)
226
#endif
227
228
/* #define P DBL_MANT_DIG */
229
/* Ten_pmax = floor(P*log(2)/log(5)) */
230
/* Bletch = (highest power of 2 < DBL_MAX_10_EXP) / 16 */
231
/* Quick_max = floor((P-1)*log(FLT_RADIX)/log(10) - 1) */
232
/* Int_max = floor(P*log(FLT_RADIX)/log(10) - 1) */
233
234
221k
#define Exp_shift  20
235
89.2k
#define Exp_shift1 20
236
638k
#define Exp_msk1    0x100000
237
#define Exp_msk11   0x100000
238
1.18M
#define Exp_mask  0x7ff00000
239
497k
#define P 53
240
#define Nbits 53
241
261k
#define Bias 1023
242
#define Emax 1023
243
#define Emin (-1022)
244
399k
#define Etiny (-1074)  /* smallest denormal is 2**Etiny */
245
122k
#define Exp_1  0x3ff00000
246
40.4k
#define Exp_11 0x3ff00000
247
276k
#define Ebits 11
248
201k
#define Frac_mask  0xfffff
249
42.1k
#define Frac_mask1 0xfffff
250
1.45M
#define Ten_pmax 22
251
86
#define Bletch 0x10
252
75.8k
#define Bndry_mask  0xfffff
253
6.51k
#define Bndry_mask1 0xfffff
254
64.8k
#define Sign_bit 0x80000000
255
6.20k
#define Log2P 1
256
#define Tiny0 0
257
35.9k
#define Tiny1 1
258
44.7k
#define Quick_max 14
259
26.6k
#define Int_max 14
260
261
#ifndef Flt_Rounds
262
#ifdef FLT_ROUNDS
263
777k
#define Flt_Rounds FLT_ROUNDS
264
#else
265
#define Flt_Rounds 1
266
#endif
267
#endif /*Flt_Rounds*/
268
269
#define Rounding Flt_Rounds
270
271
2.67k
#define Big0 (Frac_mask1 | Exp_msk1*(DBL_MAX_EXP+Bias-1))
272
1.69k
#define Big1 0xffffffff
273
274
/* Bits of the representation of positive infinity. */
275
276
#define POSINF_WORD0 0x7ff00000
277
#define POSINF_WORD1 0
278
279
/* struct BCinfo is used to pass information from _Py_dg_strtod to bigcomp */
280
281
typedef struct BCinfo BCinfo;
282
struct
283
BCinfo {
284
    int e0, nd, nd0, scale;
285
};
286
287
18.7M
#define FFFFFFFF 0xffffffffUL
288
289
/* struct Bigint is used to represent arbitrary-precision integers.  These
290
   integers are stored in sign-magnitude format, with the magnitude stored as
291
   an array of base 2**32 digits.  Bigints are always normalized: if x is a
292
   Bigint then x->wds >= 1, and either x->wds == 1 or x[wds-1] is nonzero.
293
294
   The Bigint fields are as follows:
295
296
     - next is a header used by Balloc and Bfree to keep track of lists
297
         of freed Bigints;  it's also used for the linked list of
298
         powers of 5 of the form 5**2**i used by pow5mult.
299
     - k indicates which pool this Bigint was allocated from
300
     - maxwds is the maximum number of words space was allocated for
301
       (usually maxwds == 2**k)
302
     - sign is 1 for negative Bigints, 0 for positive.  The sign is unused
303
       (ignored on inputs, set to 0 on outputs) in almost all operations
304
       involving Bigints: a notable exception is the diff function, which
305
       ignores signs on inputs but sets the sign of the output correctly.
306
     - wds is the actual number of significant words
307
     - x contains the vector of words (digits) for this Bigint, from least
308
       significant (x[0]) to most significant (x[wds-1]).
309
*/
310
311
// struct Bigint is defined in pycore_dtoa.h.
312
typedef struct Bigint Bigint;
313
314
#if !defined(Py_GIL_DISABLED) && !defined(Py_USING_MEMORY_DEBUGGER)
315
316
/* Memory management: memory is allocated from, and returned to, Kmax+1 pools
317
   of memory, where pool k (0 <= k <= Kmax) is for Bigints b with b->maxwds ==
318
   1 << k.  These pools are maintained as linked lists, with freelist[k]
319
   pointing to the head of the list for pool k.
320
321
   On allocation, if there's no free slot in the appropriate pool, MALLOC is
322
   called to get more memory.  This memory is not returned to the system until
323
   Python quits.  There's also a private memory pool that's allocated from
324
   in preference to using MALLOC.
325
326
   For Bigints with more than (1 << Kmax) digits (which implies at least 1233
327
   decimal digits), memory is directly allocated using MALLOC, and freed using
328
   FREE.
329
330
   XXX: it would be easy to bypass this memory-management system and
331
   translate each call to Balloc into a call to PyMem_Malloc, and each
332
   Bfree to PyMem_Free.  Investigate whether this has any significant
333
   performance on impact. */
334
335
7.63M
#define freelist interp->dtoa.freelist
336
362
#define private_mem interp->dtoa.preallocated
337
974
#define pmem_next interp->dtoa.preallocated_next
338
339
/* Allocate space for a Bigint with up to 1<<k digits */
340
341
static Bigint *
342
Balloc(int k)
343
1.90M
{
344
1.90M
    int x;
345
1.90M
    Bigint *rv;
346
1.90M
    unsigned int len;
347
1.90M
    PyInterpreterState *interp = _PyInterpreterState_GET();
348
349
1.90M
    if (k <= Bigint_Kmax && (rv = freelist[k]))
350
1.90M
        freelist[k] = rv->next;
351
362
    else {
352
362
        x = 1 << k;
353
362
        len = (sizeof(Bigint) + (x-1)*sizeof(ULong) + sizeof(double) - 1)
354
362
            /sizeof(double);
355
362
        if (k <= Bigint_Kmax &&
356
362
            pmem_next - private_mem + len <= (Py_ssize_t)Bigint_PREALLOC_SIZE
357
362
        ) {
358
306
            rv = (Bigint*)pmem_next;
359
306
            pmem_next += len;
360
306
        }
361
56
        else {
362
56
            rv = (Bigint*)MALLOC(len*sizeof(double));
363
56
            if (rv == NULL)
364
0
                return NULL;
365
56
        }
366
362
        rv->k = k;
367
362
        rv->maxwds = x;
368
362
    }
369
1.90M
    rv->sign = rv->wds = 0;
370
1.90M
    return rv;
371
1.90M
}
372
373
/* Free a Bigint allocated with Balloc */
374
375
static void
376
Bfree(Bigint *v)
377
5.53M
{
378
5.53M
    if (v) {
379
1.90M
        if (v->k > Bigint_Kmax)
380
0
            FREE((void*)v);
381
1.90M
        else {
382
1.90M
            PyInterpreterState *interp = _PyInterpreterState_GET();
383
1.90M
            v->next = freelist[v->k];
384
1.90M
            freelist[v->k] = v;
385
1.90M
        }
386
1.90M
    }
387
5.53M
}
388
389
#undef pmem_next
390
#undef private_mem
391
#undef freelist
392
393
#else
394
395
/* Alternative versions of Balloc and Bfree that use PyMem_Malloc and
396
   PyMem_Free directly in place of the custom memory allocation scheme above.
397
   These are provided for the benefit of memory debugging tools like
398
   Valgrind. */
399
400
/* Allocate space for a Bigint with up to 1<<k digits */
401
402
static Bigint *
403
Balloc(int k)
404
{
405
    int x;
406
    Bigint *rv;
407
    unsigned int len;
408
409
    x = 1 << k;
410
    len = (sizeof(Bigint) + (x-1)*sizeof(ULong) + sizeof(double) - 1)
411
        /sizeof(double);
412
413
    rv = (Bigint*)MALLOC(len*sizeof(double));
414
    if (rv == NULL)
415
        return NULL;
416
417
    rv->k = k;
418
    rv->maxwds = x;
419
    rv->sign = rv->wds = 0;
420
    return rv;
421
}
422
423
/* Free a Bigint allocated with Balloc */
424
425
static void
426
Bfree(Bigint *v)
427
{
428
    if (v) {
429
        FREE((void*)v);
430
    }
431
}
432
433
#endif /* !defined(Py_GIL_DISABLED) && !defined(Py_USING_MEMORY_DEBUGGER) */
434
435
140k
#define Bcopy(x,y) memcpy((char *)&x->sign, (char *)&y->sign,   \
436
140k
                          y->wds*sizeof(Long) + 2*sizeof(int))
437
438
/* Multiply a Bigint b by m and add a.  Either modifies b in place and returns
439
   a pointer to the modified b, or Bfrees b and returns a pointer to a copy.
440
   On failure, return NULL.  In this case, b will have been already freed. */
441
442
static Bigint *
443
multadd(Bigint *b, int m, int a)       /* multiply by m and add a */
444
1.22M
{
445
1.22M
    int i, wds;
446
1.22M
    ULong *x;
447
1.22M
    ULLong carry, y;
448
1.22M
    Bigint *b1;
449
450
1.22M
    wds = b->wds;
451
1.22M
    x = b->x;
452
1.22M
    i = 0;
453
1.22M
    carry = a;
454
3.84M
    do {
455
3.84M
        y = *x * (ULLong)m + carry;
456
3.84M
        carry = y >> 32;
457
3.84M
        *x++ = (ULong)(y & FFFFFFFF);
458
3.84M
    }
459
3.84M
    while(++i < wds);
460
1.22M
    if (carry) {
461
84.1k
        if (wds >= b->maxwds) {
462
5.26k
            b1 = Balloc(b->k+1);
463
5.26k
            if (b1 == NULL){
464
0
                Bfree(b);
465
0
                return NULL;
466
0
            }
467
5.26k
            Bcopy(b1, b);
468
5.26k
            Bfree(b);
469
5.26k
            b = b1;
470
5.26k
        }
471
84.1k
        b->x[wds++] = (ULong)carry;
472
84.1k
        b->wds = wds;
473
84.1k
    }
474
1.22M
    return b;
475
1.22M
}
476
477
/* convert a string s containing nd decimal digits (possibly containing a
478
   decimal separator at position nd0, which is ignored) to a Bigint.  This
479
   function carries on where the parsing code in _Py_dg_strtod leaves off: on
480
   entry, y9 contains the result of converting the first 9 digits.  Returns
481
   NULL on failure. */
482
483
static Bigint *
484
s2b(const char *s, int nd0, int nd, ULong y9)
485
104k
{
486
104k
    Bigint *b;
487
104k
    int i, k;
488
104k
    Long x, y;
489
490
104k
    x = (nd + 8) / 9;
491
164k
    for(k = 0, y = 1; x > y; y <<= 1, k++) ;
492
104k
    b = Balloc(k);
493
104k
    if (b == NULL)
494
0
        return NULL;
495
104k
    b->x[0] = y9;
496
104k
    b->wds = 1;
497
498
104k
    if (nd <= 9)
499
58.8k
      return b;
500
501
45.8k
    s += 9;
502
452k
    for (i = 9; i < nd0; i++) {
503
406k
        b = multadd(b, 10, *s++ - '0');
504
406k
        if (b == NULL)
505
0
            return NULL;
506
406k
    }
507
45.8k
    s++;
508
171k
    for(; i < nd; i++) {
509
125k
        b = multadd(b, 10, *s++ - '0');
510
125k
        if (b == NULL)
511
0
            return NULL;
512
125k
    }
513
45.8k
    return b;
514
45.8k
}
515
516
/* count leading 0 bits in the 32-bit integer x. */
517
518
static int
519
hi0bits(ULong x)
520
162k
{
521
162k
    int k = 0;
522
523
162k
    if (!(x & 0xffff0000)) {
524
88.5k
        k = 16;
525
88.5k
        x <<= 16;
526
88.5k
    }
527
162k
    if (!(x & 0xff000000)) {
528
91.9k
        k += 8;
529
91.9k
        x <<= 8;
530
91.9k
    }
531
162k
    if (!(x & 0xf0000000)) {
532
96.5k
        k += 4;
533
96.5k
        x <<= 4;
534
96.5k
    }
535
162k
    if (!(x & 0xc0000000)) {
536
92.7k
        k += 2;
537
92.7k
        x <<= 2;
538
92.7k
    }
539
162k
    if (!(x & 0x80000000)) {
540
90.2k
        k++;
541
90.2k
        if (!(x & 0x40000000))
542
0
            return 32;
543
90.2k
    }
544
162k
    return k;
545
162k
}
546
547
/* count trailing 0 bits in the 32-bit integer y, and shift y right by that
548
   number of bits. */
549
550
static int
551
lo0bits(ULong *y)
552
44.6k
{
553
44.6k
    int k;
554
44.6k
    ULong x = *y;
555
556
44.6k
    if (x & 7) {
557
25.1k
        if (x & 1)
558
13.8k
            return 0;
559
11.3k
        if (x & 2) {
560
7.09k
            *y = x >> 1;
561
7.09k
            return 1;
562
7.09k
        }
563
4.24k
        *y = x >> 2;
564
4.24k
        return 2;
565
11.3k
    }
566
19.4k
    k = 0;
567
19.4k
    if (!(x & 0xffff)) {
568
8.01k
        k = 16;
569
8.01k
        x >>= 16;
570
8.01k
    }
571
19.4k
    if (!(x & 0xff)) {
572
4.31k
        k += 8;
573
4.31k
        x >>= 8;
574
4.31k
    }
575
19.4k
    if (!(x & 0xf)) {
576
10.6k
        k += 4;
577
10.6k
        x >>= 4;
578
10.6k
    }
579
19.4k
    if (!(x & 0x3)) {
580
9.96k
        k += 2;
581
9.96k
        x >>= 2;
582
9.96k
    }
583
19.4k
    if (!(x & 1)) {
584
13.2k
        k++;
585
13.2k
        x >>= 1;
586
13.2k
        if (!x)
587
0
            return 32;
588
13.2k
    }
589
19.4k
    *y = x;
590
19.4k
    return k;
591
19.4k
}
592
593
/* convert a small nonnegative integer to a Bigint */
594
595
static Bigint *
596
i2b(int i)
597
221k
{
598
221k
    Bigint *b;
599
600
221k
    b = Balloc(1);
601
221k
    if (b == NULL)
602
0
        return NULL;
603
221k
    b->x[0] = i;
604
221k
    b->wds = 1;
605
221k
    return b;
606
221k
}
607
608
/* multiply two Bigints.  Returns a new Bigint, or NULL on failure.  Ignores
609
   the signs of a and b. */
610
611
static Bigint *
612
mult(Bigint *a, Bigint *b)
613
511k
{
614
511k
    Bigint *c;
615
511k
    int k, wa, wb, wc;
616
511k
    ULong *x, *xa, *xae, *xb, *xbe, *xc, *xc0;
617
511k
    ULong y;
618
511k
    ULLong carry, z;
619
620
511k
    if ((!a->x[0] && a->wds == 1) || (!b->x[0] && b->wds == 1)) {
621
5.16k
        c = Balloc(0);
622
5.16k
        if (c == NULL)
623
0
            return NULL;
624
5.16k
        c->wds = 1;
625
5.16k
        c->x[0] = 0;
626
5.16k
        return c;
627
5.16k
    }
628
629
506k
    if (a->wds < b->wds) {
630
234k
        c = a;
631
234k
        a = b;
632
234k
        b = c;
633
234k
    }
634
506k
    k = a->k;
635
506k
    wa = a->wds;
636
506k
    wb = b->wds;
637
506k
    wc = wa + wb;
638
506k
    if (wc > a->maxwds)
639
242k
        k++;
640
506k
    c = Balloc(k);
641
506k
    if (c == NULL)
642
0
        return NULL;
643
4.26M
    for(x = c->x, xa = x + wc; x < xa; x++)
644
3.75M
        *x = 0;
645
506k
    xa = a->x;
646
506k
    xae = xa + wa;
647
506k
    xb = b->x;
648
506k
    xbe = xb + wb;
649
506k
    xc0 = c->x;
650
1.54M
    for(; xb < xbe; xc0++) {
651
1.03M
        if ((y = *xb++)) {
652
1.03M
            x = xa;
653
1.03M
            xc = xc0;
654
1.03M
            carry = 0;
655
9.23M
            do {
656
9.23M
                z = *x++ * (ULLong)y + *xc + carry;
657
9.23M
                carry = z >> 32;
658
9.23M
                *xc++ = (ULong)(z & FFFFFFFF);
659
9.23M
            }
660
9.23M
            while(x < xae);
661
1.03M
            *xc = (ULong)carry;
662
1.03M
        }
663
1.03M
    }
664
861k
    for(xc0 = c->x, xc = xc0 + wc; wc > 0 && !*--xc; --wc) ;
665
506k
    c->wds = wc;
666
506k
    return c;
667
506k
}
668
669
#ifndef Py_USING_MEMORY_DEBUGGER
670
671
/* multiply the Bigint b by 5**k.  Returns a pointer to the result, or NULL on
672
   failure; if the returned pointer is distinct from b then the original
673
   Bigint b will have been Bfree'd.   Ignores the sign of b. */
674
675
static Bigint *
676
pow5mult(Bigint *b, int k)
677
179k
{
678
179k
    Bigint *b1, *p5, **p5s;
679
179k
    int i;
680
179k
    static const int p05[3] = { 5, 25, 125 };
681
682
    // For double-to-string conversion, the maximum value of k is limited by
683
    // DBL_MAX_10_EXP (308), the maximum decimal base-10 exponent for binary64.
684
    // For string-to-double conversion, the extreme case is constrained by our
685
    // hardcoded exponent limit before we underflow of -512, adjusted by
686
    // STRTOD_DIGLIM-DBL_DIG-1, giving a maximum of k=535.
687
179k
    assert(0 <= k && k < 1024);
688
689
179k
    if ((i = k & 3)) {
690
126k
        b = multadd(b, p05[i-1], 0);
691
126k
        if (b == NULL)
692
0
            return NULL;
693
126k
    }
694
695
179k
    if (!(k >>= 2))
696
11.1k
        return b;
697
168k
    PyInterpreterState *interp = _PyInterpreterState_GET();
698
168k
    p5s = interp->dtoa.p5s;
699
746k
    for(;;) {
700
746k
        assert(p5s != interp->dtoa.p5s + Bigint_Pow5size);
701
746k
        p5 = *p5s;
702
746k
        p5s++;
703
746k
        if (k & 1) {
704
439k
            b1 = mult(b, p5);
705
439k
            Bfree(b);
706
439k
            b = b1;
707
439k
            if (b == NULL)
708
0
                return NULL;
709
439k
        }
710
746k
        if (!(k >>= 1))
711
168k
            break;
712
746k
    }
713
168k
    return b;
714
168k
}
715
716
#else
717
718
/* Version of pow5mult that doesn't cache powers of 5. Provided for
719
   the benefit of memory debugging tools like Valgrind. */
720
721
static Bigint *
722
pow5mult(Bigint *b, int k)
723
{
724
    Bigint *b1, *p5, *p51;
725
    int i;
726
    static const int p05[3] = { 5, 25, 125 };
727
728
    if ((i = k & 3)) {
729
        b = multadd(b, p05[i-1], 0);
730
        if (b == NULL)
731
            return NULL;
732
    }
733
734
    if (!(k >>= 2))
735
        return b;
736
    p5 = i2b(625);
737
    if (p5 == NULL) {
738
        Bfree(b);
739
        return NULL;
740
    }
741
742
    for(;;) {
743
        if (k & 1) {
744
            b1 = mult(b, p5);
745
            Bfree(b);
746
            b = b1;
747
            if (b == NULL) {
748
                Bfree(p5);
749
                return NULL;
750
            }
751
        }
752
        if (!(k >>= 1))
753
            break;
754
        p51 = mult(p5, p5);
755
        Bfree(p5);
756
        p5 = p51;
757
        if (p5 == NULL) {
758
            Bfree(b);
759
            return NULL;
760
        }
761
    }
762
    Bfree(p5);
763
    return b;
764
}
765
766
#endif /* Py_USING_MEMORY_DEBUGGER */
767
768
/* shift a Bigint b left by k bits.  Return a pointer to the shifted result,
769
   or NULL on failure.  If the returned pointer is distinct from b then the
770
   original b will have been Bfree'd.   Ignores the sign of b. */
771
772
static Bigint *
773
lshift(Bigint *b, int k)
774
432k
{
775
432k
    int i, k1, n, n1;
776
432k
    Bigint *b1;
777
432k
    ULong *x, *x1, *xe, z;
778
779
432k
    if (!k || (!b->x[0] && b->wds == 1))
780
5.61k
        return b;
781
782
427k
    n = k >> 5;
783
427k
    k1 = b->k;
784
427k
    n1 = n + b->wds + 1;
785
992k
    for(i = b->maxwds; n1 > i; i <<= 1)
786
565k
        k1++;
787
427k
    b1 = Balloc(k1);
788
427k
    if (b1 == NULL) {
789
0
        Bfree(b);
790
0
        return NULL;
791
0
    }
792
427k
    x1 = b1->x;
793
2.04M
    for(i = 0; i < n; i++)
794
1.61M
        *x1++ = 0;
795
427k
    x = b->x;
796
427k
    xe = x + b->wds;
797
427k
    if (k &= 0x1f) {
798
424k
        k1 = 32 - k;
799
424k
        z = 0;
800
1.73M
        do {
801
1.73M
            *x1++ = *x << k | z;
802
1.73M
            z = *x++ >> k1;
803
1.73M
        }
804
1.73M
        while(x < xe);
805
424k
        if ((*x1 = z))
806
70.7k
            ++n1;
807
424k
    }
808
2.92k
    else do
809
6.13k
             *x1++ = *x++;
810
6.13k
        while(x < xe);
811
427k
    b1->wds = n1 - 1;
812
427k
    Bfree(b);
813
427k
    return b1;
814
427k
}
815
816
/* Do a three-way compare of a and b, returning -1 if a < b, 0 if a == b and
817
   1 if a > b.  Ignores signs of a and b. */
818
819
static int
820
cmp(Bigint *a, Bigint *b)
821
1.14M
{
822
1.14M
    ULong *xa, *xa0, *xb, *xb0;
823
1.14M
    int i, j;
824
825
1.14M
    i = a->wds;
826
1.14M
    j = b->wds;
827
#ifdef DEBUG
828
    if (i > 1 && !a->x[i-1])
829
        Bug("cmp called with a->x[a->wds-1] == 0");
830
    if (j > 1 && !b->x[j-1])
831
        Bug("cmp called with b->x[b->wds-1] == 0");
832
#endif
833
1.14M
    if (i -= j)
834
176k
        return i;
835
970k
    xa0 = a->x;
836
970k
    xa = xa0 + j;
837
970k
    xb0 = b->x;
838
970k
    xb = xb0 + j;
839
1.19M
    for(;;) {
840
1.19M
        if (*--xa != *--xb)
841
953k
            return *xa < *xb ? -1 : 1;
842
238k
        if (xa <= xa0)
843
16.5k
            break;
844
238k
    }
845
16.5k
    return 0;
846
970k
}
847
848
/* Take the difference of Bigints a and b, returning a new Bigint.  Returns
849
   NULL on failure.  The signs of a and b are ignored, but the sign of the
850
   result is set appropriately. */
851
852
static Bigint *
853
diff(Bigint *a, Bigint *b)
854
251k
{
855
251k
    Bigint *c;
856
251k
    int i, wa, wb;
857
251k
    ULong *xa, *xae, *xb, *xbe, *xc;
858
251k
    ULLong borrow, y;
859
860
251k
    i = cmp(a,b);
861
251k
    if (!i) {
862
2.86k
        c = Balloc(0);
863
2.86k
        if (c == NULL)
864
0
            return NULL;
865
2.86k
        c->wds = 1;
866
2.86k
        c->x[0] = 0;
867
2.86k
        return c;
868
2.86k
    }
869
248k
    if (i < 0) {
870
69.5k
        c = a;
871
69.5k
        a = b;
872
69.5k
        b = c;
873
69.5k
        i = 1;
874
69.5k
    }
875
178k
    else
876
178k
        i = 0;
877
248k
    c = Balloc(a->k);
878
248k
    if (c == NULL)
879
0
        return NULL;
880
248k
    c->sign = i;
881
248k
    wa = a->wds;
882
248k
    xa = a->x;
883
248k
    xae = xa + wa;
884
248k
    wb = b->wds;
885
248k
    xb = b->x;
886
248k
    xbe = xb + wb;
887
248k
    xc = c->x;
888
248k
    borrow = 0;
889
1.76M
    do {
890
1.76M
        y = (ULLong)*xa++ - *xb++ - borrow;
891
1.76M
        borrow = y >> 32 & (ULong)1;
892
1.76M
        *xc++ = (ULong)(y & FFFFFFFF);
893
1.76M
    }
894
1.76M
    while(xb < xbe);
895
484k
    while(xa < xae) {
896
236k
        y = *xa++ - borrow;
897
236k
        borrow = y >> 32 & (ULong)1;
898
236k
        *xc++ = (ULong)(y & FFFFFFFF);
899
236k
    }
900
434k
    while(!*--xc)
901
186k
        wa--;
902
248k
    c->wds = wa;
903
248k
    return c;
904
248k
}
905
906
/* Given a positive normal double x, return the difference between x and the
907
   next double up.  Doesn't give correct results for subnormals. */
908
909
static double
910
ulp(U *x)
911
55.0k
{
912
55.0k
    Long L;
913
55.0k
    U u;
914
915
55.0k
    L = (word0(x) & Exp_mask) - (P-1)*Exp_msk1;
916
55.0k
    word0(&u) = L;
917
55.0k
    word1(&u) = 0;
918
55.0k
    return dval(&u);
919
55.0k
}
920
921
/* Convert a Bigint to a double plus an exponent */
922
923
static double
924
b2d(Bigint *a, int *e)
925
104k
{
926
104k
    ULong *xa, *xa0, w, y, z;
927
104k
    int k;
928
104k
    U d;
929
930
104k
    xa0 = a->x;
931
104k
    xa = xa0 + a->wds;
932
104k
    y = *--xa;
933
#ifdef DEBUG
934
    if (!y) Bug("zero y in b2d");
935
#endif
936
104k
    k = hi0bits(y);
937
104k
    *e = 32 - k;
938
104k
    if (k < Ebits) {
939
34.0k
        word0(&d) = Exp_1 | y >> (Ebits - k);
940
34.0k
        w = xa > xa0 ? *--xa : 0;
941
34.0k
        word1(&d) = y << ((32-Ebits) + k) | w >> (Ebits - k);
942
34.0k
        goto ret_d;
943
34.0k
    }
944
70.1k
    z = xa > xa0 ? *--xa : 0;
945
70.1k
    if (k -= Ebits) {
946
66.5k
        word0(&d) = Exp_1 | y << k | z >> (32 - k);
947
66.5k
        y = xa > xa0 ? *--xa : 0;
948
66.5k
        word1(&d) = z << k | y >> (32 - k);
949
66.5k
    }
950
3.64k
    else {
951
3.64k
        word0(&d) = Exp_1 | y;
952
3.64k
        word1(&d) = z;
953
3.64k
    }
954
104k
  ret_d:
955
104k
    return dval(&d);
956
70.1k
}
957
958
/* Convert a scaled double to a Bigint plus an exponent.  Similar to d2b,
959
   except that it accepts the scale parameter used in _Py_dg_strtod (which
960
   should be either 0 or 2*P), and the normalization for the return value is
961
   different (see below).  On input, d should be finite and nonnegative, and d
962
   / 2**scale should be exactly representable as an IEEE 754 double.
963
964
   Returns a Bigint b and an integer e such that
965
966
     dval(d) / 2**scale = b * 2**e.
967
968
   Unlike d2b, b is not necessarily odd: b and e are normalized so
969
   that either 2**(P-1) <= b < 2**P and e >= Etiny, or b < 2**P
970
   and e == Etiny.  This applies equally to an input of 0.0: in that
971
   case the return values are b = 0 and e = Etiny.
972
973
   The above normalization ensures that for all possible inputs d,
974
   2**e gives ulp(d/2**scale).
975
976
   Returns NULL on failure.
977
*/
978
979
static Bigint *
980
sd2b(U *d, int scale, int *e)
981
156k
{
982
156k
    Bigint *b;
983
984
156k
    b = Balloc(1);
985
156k
    if (b == NULL)
986
0
        return NULL;
987
988
    /* First construct b and e assuming that scale == 0. */
989
156k
    b->wds = 2;
990
156k
    b->x[0] = word1(d);
991
156k
    b->x[1] = word0(d) & Frac_mask;
992
156k
    *e = Etiny - 1 + (int)((word0(d) & Exp_mask) >> Exp_shift);
993
156k
    if (*e < Etiny)
994
5.61k
        *e = Etiny;
995
151k
    else
996
151k
        b->x[1] |= Exp_msk1;
997
998
    /* Now adjust for scale, provided that b != 0. */
999
156k
    if (scale && (b->x[0] || b->x[1])) {
1000
29.0k
        *e -= scale;
1001
29.0k
        if (*e < Etiny) {
1002
25.7k
            scale = Etiny - *e;
1003
25.7k
            *e = Etiny;
1004
            /* We can't shift more than P-1 bits without shifting out a 1. */
1005
25.7k
            assert(0 < scale && scale <= P - 1);
1006
25.7k
            if (scale >= 32) {
1007
                /* The bits shifted out should all be zero. */
1008
13.0k
                assert(b->x[0] == 0);
1009
13.0k
                b->x[0] = b->x[1];
1010
13.0k
                b->x[1] = 0;
1011
13.0k
                scale -= 32;
1012
13.0k
            }
1013
25.7k
            if (scale) {
1014
                /* The bits shifted out should all be zero. */
1015
24.2k
                assert(b->x[0] << (32 - scale) == 0);
1016
24.2k
                b->x[0] = (b->x[0] >> scale) | (b->x[1] << (32 - scale));
1017
24.2k
                b->x[1] >>= scale;
1018
24.2k
            }
1019
25.7k
        }
1020
29.0k
    }
1021
    /* Ensure b is normalized. */
1022
156k
    if (!b->x[1])
1023
22.4k
        b->wds = 1;
1024
1025
156k
    return b;
1026
156k
}
1027
1028
/* Convert a double to a Bigint plus an exponent.  Return NULL on failure.
1029
1030
   Given a finite nonzero double d, return an odd Bigint b and exponent *e
1031
   such that fabs(d) = b * 2**e.  On return, *bbits gives the number of
1032
   significant bits of b; that is, 2**(*bbits-1) <= b < 2**(*bbits).
1033
1034
   If d is zero, then b == 0, *e == -1010, *bbits = 0.
1035
 */
1036
1037
static Bigint *
1038
d2b(U *d, int *e, int *bits)
1039
44.6k
{
1040
44.6k
    Bigint *b;
1041
44.6k
    int de, k;
1042
44.6k
    ULong *x, y, z;
1043
44.6k
    int i;
1044
1045
44.6k
    b = Balloc(1);
1046
44.6k
    if (b == NULL)
1047
0
        return NULL;
1048
44.6k
    x = b->x;
1049
1050
44.6k
    z = word0(d) & Frac_mask;
1051
44.6k
    word0(d) &= 0x7fffffff;   /* clear sign bit, which we ignore */
1052
44.6k
    if ((de = (int)(word0(d) >> Exp_shift)))
1053
40.4k
        z |= Exp_msk1;
1054
44.6k
    if ((y = word1(d))) {
1055
32.5k
        if ((k = lo0bits(&y))) {
1056
19.3k
            x[0] = y | z << (32 - k);
1057
19.3k
            z >>= k;
1058
19.3k
        }
1059
13.1k
        else
1060
13.1k
            x[0] = y;
1061
32.5k
        i =
1062
32.5k
            b->wds = (x[1] = z) ? 2 : 1;
1063
32.5k
    }
1064
12.0k
    else {
1065
12.0k
        k = lo0bits(&z);
1066
12.0k
        x[0] = z;
1067
12.0k
        i =
1068
12.0k
            b->wds = 1;
1069
12.0k
        k += 32;
1070
12.0k
    }
1071
44.6k
    if (de) {
1072
40.4k
        *e = de - Bias - (P-1) + k;
1073
40.4k
        *bits = P - k;
1074
40.4k
    }
1075
4.13k
    else {
1076
4.13k
        *e = de - Bias - (P-1) + 1 + k;
1077
4.13k
        *bits = 32*i - hi0bits(x[i-1]);
1078
4.13k
    }
1079
44.6k
    return b;
1080
44.6k
}
1081
1082
/* Compute the ratio of two Bigints, as a double.  The result may have an
1083
   error of up to 2.5 ulps. */
1084
1085
static double
1086
ratio(Bigint *a, Bigint *b)
1087
52.1k
{
1088
52.1k
    U da, db;
1089
52.1k
    int k, ka, kb;
1090
1091
52.1k
    dval(&da) = b2d(a, &ka);
1092
52.1k
    dval(&db) = b2d(b, &kb);
1093
52.1k
    k = ka - kb + 32*(a->wds - b->wds);
1094
52.1k
    if (k > 0)
1095
28.4k
        word0(&da) += k*Exp_msk1;
1096
23.6k
    else {
1097
23.6k
        k = -k;
1098
23.6k
        word0(&db) += k*Exp_msk1;
1099
23.6k
    }
1100
52.1k
    return dval(&da) / dval(&db);
1101
52.1k
}
1102
1103
static const double
1104
tens[] = {
1105
    1e0, 1e1, 1e2, 1e3, 1e4, 1e5, 1e6, 1e7, 1e8, 1e9,
1106
    1e10, 1e11, 1e12, 1e13, 1e14, 1e15, 1e16, 1e17, 1e18, 1e19,
1107
    1e20, 1e21, 1e22
1108
};
1109
1110
static const double
1111
bigtens[] = { 1e16, 1e32, 1e64, 1e128, 1e256 };
1112
static const double tinytens[] = { 1e-16, 1e-32, 1e-64, 1e-128,
1113
                                   9007199254740992.*9007199254740992.e-256
1114
                                   /* = 2^106 * 1e-256 */
1115
};
1116
/* The factor of 2^53 in tinytens[4] helps us avoid setting the underflow */
1117
/* flag unnecessarily.  It leads to a song and dance at the end of strtod. */
1118
23.3k
#define Scale_Bit 0x10
1119
24.3k
#define n_bigtens 5
1120
1121
#define ULbits 32
1122
#define kshift 5
1123
54.4k
#define kmask 31
1124
1125
1126
static int
1127
dshift(Bigint *b, int p2)
1128
54.4k
{
1129
54.4k
    int rv = hi0bits(b->x[b->wds-1]) - 4;
1130
54.4k
    if (p2 > 0)
1131
24.6k
        rv -= p2;
1132
54.4k
    return rv & kmask;
1133
54.4k
}
1134
1135
/* special case of Bigint division.  The quotient is always in the range 0 <=
1136
   quotient < 10, and on entry the divisor S is normalized so that its top 4
1137
   bits (28--31) are zero and bit 27 is set. */
1138
1139
static int
1140
quorem(Bigint *b, Bigint *S)
1141
507k
{
1142
507k
    int n;
1143
507k
    ULong *bx, *bxe, q, *sx, *sxe;
1144
507k
    ULLong borrow, carry, y, ys;
1145
1146
507k
    n = S->wds;
1147
#ifdef DEBUG
1148
    /*debug*/ if (b->wds > n)
1149
        /*debug*/       Bug("oversize b in quorem");
1150
#endif
1151
507k
    if (b->wds < n)
1152
28.9k
        return 0;
1153
478k
    sx = S->x;
1154
478k
    sxe = sx + --n;
1155
478k
    bx = b->x;
1156
478k
    bxe = bx + n;
1157
478k
    q = *bxe / (*sxe + 1);      /* ensure q <= true quotient */
1158
#ifdef DEBUG
1159
    /*debug*/ if (q > 9)
1160
        /*debug*/       Bug("oversized quotient in quorem");
1161
#endif
1162
478k
    if (q) {
1163
338k
        borrow = 0;
1164
338k
        carry = 0;
1165
1.66M
        do {
1166
1.66M
            ys = *sx++ * (ULLong)q + carry;
1167
1.66M
            carry = ys >> 32;
1168
1.66M
            y = *bx - (ys & FFFFFFFF) - borrow;
1169
1.66M
            borrow = y >> 32 & (ULong)1;
1170
1.66M
            *bx++ = (ULong)(y & FFFFFFFF);
1171
1.66M
        }
1172
1.66M
        while(sx <= sxe);
1173
338k
        if (!*bxe) {
1174
956
            bx = b->x;
1175
956
            while(--bxe > bx && !*bxe)
1176
0
                --n;
1177
956
            b->wds = n;
1178
956
        }
1179
338k
    }
1180
478k
    if (cmp(b, S) >= 0) {
1181
24.9k
        q++;
1182
24.9k
        borrow = 0;
1183
24.9k
        carry = 0;
1184
24.9k
        bx = b->x;
1185
24.9k
        sx = S->x;
1186
140k
        do {
1187
140k
            ys = *sx++ + carry;
1188
140k
            carry = ys >> 32;
1189
140k
            y = *bx - (ys & FFFFFFFF) - borrow;
1190
140k
            borrow = y >> 32 & (ULong)1;
1191
140k
            *bx++ = (ULong)(y & FFFFFFFF);
1192
140k
        }
1193
140k
        while(sx <= sxe);
1194
24.9k
        bx = b->x;
1195
24.9k
        bxe = bx + n;
1196
24.9k
        if (!*bxe) {
1197
23.6k
            while(--bxe > bx && !*bxe)
1198
1.58k
                --n;
1199
22.0k
            b->wds = n;
1200
22.0k
        }
1201
24.9k
    }
1202
478k
    return q;
1203
507k
}
1204
1205
/* sulp(x) is a version of ulp(x) that takes bc.scale into account.
1206
1207
   Assuming that x is finite and nonnegative (positive zero is fine
1208
   here) and x / 2^bc.scale is exactly representable as a double,
1209
   sulp(x) is equivalent to 2^bc.scale * ulp(x / 2^bc.scale). */
1210
1211
static double
1212
sulp(U *x, BCinfo *bc)
1213
3.31k
{
1214
3.31k
    U u;
1215
1216
3.31k
    if (bc->scale && 2*P + 1 > (int)((word0(x) & Exp_mask) >> Exp_shift)) {
1217
        /* rv/2^bc->scale is subnormal */
1218
399
        word0(&u) = (P+2)*Exp_msk1;
1219
399
        word1(&u) = 0;
1220
399
        return u.d;
1221
399
    }
1222
2.91k
    else {
1223
2.91k
        assert(word0(x) || word1(x)); /* x != 0.0 */
1224
2.91k
        return ulp(x);
1225
2.91k
    }
1226
3.31k
}
1227
1228
/* The bigcomp function handles some hard cases for strtod, for inputs
1229
   with more than STRTOD_DIGLIM digits.  It's called once an initial
1230
   estimate for the double corresponding to the input string has
1231
   already been obtained by the code in _Py_dg_strtod.
1232
1233
   The bigcomp function is only called after _Py_dg_strtod has found a
1234
   double value rv such that either rv or rv + 1ulp represents the
1235
   correctly rounded value corresponding to the original string.  It
1236
   determines which of these two values is the correct one by
1237
   computing the decimal digits of rv + 0.5ulp and comparing them with
1238
   the corresponding digits of s0.
1239
1240
   In the following, write dv for the absolute value of the number represented
1241
   by the input string.
1242
1243
   Inputs:
1244
1245
     s0 points to the first significant digit of the input string.
1246
1247
     rv is a (possibly scaled) estimate for the closest double value to the
1248
        value represented by the original input to _Py_dg_strtod.  If
1249
        bc->scale is nonzero, then rv/2^(bc->scale) is the approximation to
1250
        the input value.
1251
1252
     bc is a struct containing information gathered during the parsing and
1253
        estimation steps of _Py_dg_strtod.  Description of fields follows:
1254
1255
        bc->e0 gives the exponent of the input value, such that dv = (integer
1256
           given by the bd->nd digits of s0) * 10**e0
1257
1258
        bc->nd gives the total number of significant digits of s0.  It will
1259
           be at least 1.
1260
1261
        bc->nd0 gives the number of significant digits of s0 before the
1262
           decimal separator.  If there's no decimal separator, bc->nd0 ==
1263
           bc->nd.
1264
1265
        bc->scale is the value used to scale rv to avoid doing arithmetic with
1266
           subnormal values.  It's either 0 or 2*P (=106).
1267
1268
   Outputs:
1269
1270
     On successful exit, rv/2^(bc->scale) is the closest double to dv.
1271
1272
     Returns 0 on success, -1 on failure (e.g., due to a failed malloc call). */
1273
1274
static int
1275
bigcomp(U *rv, const char *s0, BCinfo *bc)
1276
22.2k
{
1277
22.2k
    Bigint *b, *d;
1278
22.2k
    int b2, d2, dd, i, nd, nd0, odd, p2, p5;
1279
1280
22.2k
    nd = bc->nd;
1281
22.2k
    nd0 = bc->nd0;
1282
22.2k
    p5 = nd + bc->e0;
1283
22.2k
    b = sd2b(rv, bc->scale, &p2);
1284
22.2k
    if (b == NULL)
1285
0
        return -1;
1286
1287
    /* record whether the lsb of rv/2^(bc->scale) is odd:  in the exact halfway
1288
       case, this is used for round to even. */
1289
22.2k
    odd = b->x[0] & 1;
1290
1291
    /* left shift b by 1 bit and or a 1 into the least significant bit;
1292
       this gives us b * 2**p2 = rv/2^(bc->scale) + 0.5 ulp. */
1293
22.2k
    b = lshift(b, 1);
1294
22.2k
    if (b == NULL)
1295
0
        return -1;
1296
22.2k
    b->x[0] |= 1;
1297
22.2k
    p2--;
1298
1299
22.2k
    p2 -= p5;
1300
22.2k
    d = i2b(1);
1301
22.2k
    if (d == NULL) {
1302
0
        Bfree(b);
1303
0
        return -1;
1304
0
    }
1305
    /* Arrange for convenient computation of quotients:
1306
     * shift left if necessary so divisor has 4 leading 0 bits.
1307
     */
1308
22.2k
    if (p5 > 0) {
1309
19.8k
        d = pow5mult(d, p5);
1310
19.8k
        if (d == NULL) {
1311
0
            Bfree(b);
1312
0
            return -1;
1313
0
        }
1314
19.8k
    }
1315
2.36k
    else if (p5 < 0) {
1316
1.60k
        b = pow5mult(b, -p5);
1317
1.60k
        if (b == NULL) {
1318
0
            Bfree(d);
1319
0
            return -1;
1320
0
        }
1321
1.60k
    }
1322
22.2k
    if (p2 > 0) {
1323
18.1k
        b2 = p2;
1324
18.1k
        d2 = 0;
1325
18.1k
    }
1326
4.07k
    else {
1327
4.07k
        b2 = 0;
1328
4.07k
        d2 = -p2;
1329
4.07k
    }
1330
22.2k
    i = dshift(d, d2);
1331
22.2k
    if ((b2 += i) > 0) {
1332
21.9k
        b = lshift(b, b2);
1333
21.9k
        if (b == NULL) {
1334
0
            Bfree(d);
1335
0
            return -1;
1336
0
        }
1337
21.9k
    }
1338
22.2k
    if ((d2 += i) > 0) {
1339
21.5k
        d = lshift(d, d2);
1340
21.5k
        if (d == NULL) {
1341
0
            Bfree(b);
1342
0
            return -1;
1343
0
        }
1344
21.5k
    }
1345
1346
    /* Compare s0 with b/d: set dd to -1, 0, or 1 according as s0 < b/d, s0 ==
1347
     * b/d, or s0 > b/d.  Here the digits of s0 are thought of as representing
1348
     * a number in the range [0.1, 1). */
1349
22.2k
    if (cmp(b, d) >= 0)
1350
        /* b/d >= 1 */
1351
890
        dd = -1;
1352
21.3k
    else {
1353
21.3k
        i = 0;
1354
390k
        for(;;) {
1355
390k
            b = multadd(b, 10, 0);
1356
390k
            if (b == NULL) {
1357
0
                Bfree(d);
1358
0
                return -1;
1359
0
            }
1360
390k
            dd = s0[i < nd0 ? i : i+1] - '0' - quorem(b, d);
1361
390k
            i++;
1362
1363
390k
            if (dd)
1364
19.7k
                break;
1365
370k
            if (!b->x[0] && b->wds == 1) {
1366
                /* b/d == 0 */
1367
865
                dd = i < nd;
1368
865
                break;
1369
865
            }
1370
369k
            if (!(i < nd)) {
1371
                /* b/d != 0, but digits of s0 exhausted */
1372
703
                dd = -1;
1373
703
                break;
1374
703
            }
1375
369k
        }
1376
21.3k
    }
1377
22.2k
    Bfree(b);
1378
22.2k
    Bfree(d);
1379
22.2k
    if (dd > 0 || (dd == 0 && odd))
1380
1.83k
        dval(rv) += sulp(rv, bc);
1381
22.2k
    return 0;
1382
22.2k
}
1383
1384
1385
double
1386
_Py_dg_strtod(const char *s00, char **se)
1387
829k
{
1388
829k
    int bb2, bb5, bbe, bd2, bd5, bs2, c, dsign, e, e1, error;
1389
829k
    int esign, i, j, k, lz, nd, nd0, odd, sign;
1390
829k
    const char *s, *s0, *s1;
1391
829k
    double aadj, aadj1;
1392
829k
    U aadj2, adj, rv, rv0;
1393
829k
    ULong y, z, abs_exp;
1394
829k
    Long L;
1395
829k
    BCinfo bc;
1396
829k
    Bigint *bb = NULL, *bd = NULL, *bd0 = NULL, *bs = NULL, *delta = NULL;
1397
829k
    size_t ndigits, fraclen;
1398
829k
    double result;
1399
1400
829k
    dval(&rv) = 0.;
1401
1402
    /* Start parsing. */
1403
829k
    c = *(s = s00);
1404
1405
    /* Parse optional sign, if present. */
1406
829k
    sign = 0;
1407
829k
    switch (c) {
1408
660k
    case '-':
1409
660k
        sign = 1;
1410
660k
        _Py_FALLTHROUGH;
1411
660k
    case '+':
1412
660k
        c = *++s;
1413
829k
    }
1414
1415
    /* Skip leading zeros: lz is true iff there were leading zeros. */
1416
829k
    s1 = s;
1417
851k
    while (c == '0')
1418
21.4k
        c = *++s;
1419
829k
    lz = s != s1;
1420
1421
    /* Point s0 at the first nonzero digit (if any).  fraclen will be the
1422
       number of digits between the decimal point and the end of the
1423
       digit string.  ndigits will be the total number of digits ignoring
1424
       leading zeros. */
1425
829k
    s0 = s1 = s;
1426
5.12M
    while ('0' <= c && c <= '9')
1427
4.29M
        c = *++s;
1428
829k
    ndigits = s - s1;
1429
829k
    fraclen = 0;
1430
1431
    /* Parse decimal point and following digits. */
1432
829k
    if (c == '.') {
1433
91.8k
        c = *++s;
1434
91.8k
        if (!ndigits) {
1435
23.8k
            s1 = s;
1436
2.46M
            while (c == '0')
1437
2.43M
                c = *++s;
1438
23.8k
            lz = lz || s != s1;
1439
23.8k
            fraclen += (s - s1);
1440
23.8k
            s0 = s;
1441
23.8k
        }
1442
91.8k
        s1 = s;
1443
22.2M
        while ('0' <= c && c <= '9')
1444
22.1M
            c = *++s;
1445
91.8k
        ndigits += s - s1;
1446
91.8k
        fraclen += s - s1;
1447
91.8k
    }
1448
1449
    /* Now lz is true if and only if there were leading zero digits, and
1450
       ndigits gives the total number of digits ignoring leading zeros.  A
1451
       valid input must have at least one digit. */
1452
829k
    if (!ndigits && !lz) {
1453
19
        if (se)
1454
19
            *se = (char *)s00;
1455
19
        goto parse_error;
1456
19
    }
1457
1458
    /* Range check ndigits and fraclen to make sure that they, and values
1459
       computed with them, can safely fit in an int. */
1460
829k
    if (ndigits > MAX_DIGITS || fraclen > MAX_DIGITS) {
1461
0
        if (se)
1462
0
            *se = (char *)s00;
1463
0
        goto parse_error;
1464
0
    }
1465
829k
    nd = (int)ndigits;
1466
829k
    nd0 = (int)ndigits - (int)fraclen;
1467
1468
    /* Parse exponent. */
1469
829k
    e = 0;
1470
829k
    if (c == 'e' || c == 'E') {
1471
735k
        s00 = s;
1472
735k
        c = *++s;
1473
1474
        /* Exponent sign. */
1475
735k
        esign = 0;
1476
735k
        switch (c) {
1477
27.8k
        case '-':
1478
27.8k
            esign = 1;
1479
27.8k
            _Py_FALLTHROUGH;
1480
40.0k
        case '+':
1481
40.0k
            c = *++s;
1482
735k
        }
1483
1484
        /* Skip zeros.  lz is true iff there are leading zeros. */
1485
735k
        s1 = s;
1486
916k
        while (c == '0')
1487
181k
            c = *++s;
1488
735k
        lz = s != s1;
1489
1490
        /* Get absolute value of the exponent. */
1491
735k
        s1 = s;
1492
735k
        abs_exp = 0;
1493
5.85M
        while ('0' <= c && c <= '9') {
1494
5.11M
            abs_exp = 10*abs_exp + (c - '0');
1495
5.11M
            c = *++s;
1496
5.11M
        }
1497
1498
        /* abs_exp will be correct modulo 2**32.  But 10**9 < 2**32, so if
1499
           there are at most 9 significant exponent digits then overflow is
1500
           impossible. */
1501
735k
        if (s - s1 > 9 || abs_exp > MAX_ABS_EXP)
1502
4.57k
            e = (int)MAX_ABS_EXP;
1503
730k
        else
1504
730k
            e = (int)abs_exp;
1505
735k
        if (esign)
1506
27.8k
            e = -e;
1507
1508
        /* A valid exponent must have at least one digit. */
1509
735k
        if (s == s1 && !lz)
1510
0
            s = s00;
1511
735k
    }
1512
1513
    /* Adjust exponent to take into account position of the point. */
1514
829k
    e -= nd - nd0;
1515
829k
    if (nd0 <= 0)
1516
28.0k
        nd0 = nd;
1517
1518
    /* Finished parsing.  Set se to indicate how far we parsed */
1519
829k
    if (se)
1520
829k
        *se = (char *)s;
1521
1522
    /* If all digits were zero, exit with return value +-0.0.  Otherwise,
1523
       strip trailing zeros: scan back until we hit a nonzero digit. */
1524
829k
    if (!nd)
1525
11.0k
        goto ret;
1526
5.83M
    for (i = nd; i > 0; ) {
1527
5.83M
        --i;
1528
5.83M
        if (s0[i < nd0 ? i : i+1] != '0') {
1529
818k
            ++i;
1530
818k
            break;
1531
818k
        }
1532
5.83M
    }
1533
818k
    e += nd - i;
1534
818k
    nd = i;
1535
818k
    if (nd0 > nd)
1536
10.9k
        nd0 = nd;
1537
1538
    /* Summary of parsing results.  After parsing, and dealing with zero
1539
     * inputs, we have values s0, nd0, nd, e, sign, where:
1540
     *
1541
     *  - s0 points to the first significant digit of the input string
1542
     *
1543
     *  - nd is the total number of significant digits (here, and
1544
     *    below, 'significant digits' means the set of digits of the
1545
     *    significand of the input that remain after ignoring leading
1546
     *    and trailing zeros).
1547
     *
1548
     *  - nd0 indicates the position of the decimal point, if present; it
1549
     *    satisfies 1 <= nd0 <= nd.  The nd significant digits are in
1550
     *    s0[0:nd0] and s0[nd0+1:nd+1] using the usual Python half-open slice
1551
     *    notation.  (If nd0 < nd, then s0[nd0] contains a '.'  character; if
1552
     *    nd0 == nd, then s0[nd0] could be any non-digit character.)
1553
     *
1554
     *  - e is the adjusted exponent: the absolute value of the number
1555
     *    represented by the original input string is n * 10**e, where
1556
     *    n is the integer represented by the concatenation of
1557
     *    s0[0:nd0] and s0[nd0+1:nd+1]
1558
     *
1559
     *  - sign gives the sign of the input:  1 for negative, 0 for positive
1560
     *
1561
     *  - the first and last significant digits are nonzero
1562
     */
1563
1564
    /* put first DBL_DIG+1 digits into integer y and z.
1565
     *
1566
     *  - y contains the value represented by the first min(9, nd)
1567
     *    significant digits
1568
     *
1569
     *  - if nd > 9, z contains the value represented by significant digits
1570
     *    with indices in [9, min(16, nd)).  So y * 10**(min(16, nd) - 9) + z
1571
     *    gives the value represented by the first min(16, nd) sig. digits.
1572
     */
1573
1574
818k
    bc.e0 = e1 = e;
1575
818k
    y = z = 0;
1576
2.69M
    for (i = 0; i < nd; i++) {
1577
1.92M
        if (i < 9)
1578
1.47M
            y = 10*y + s0[i < nd0 ? i : i+1] - '0';
1579
458k
        else if (i < DBL_DIG+1)
1580
409k
            z = 10*z + s0[i < nd0 ? i : i+1] - '0';
1581
49.1k
        else
1582
49.1k
            break;
1583
1.92M
    }
1584
1585
818k
    k = nd < DBL_DIG + 1 ? nd : DBL_DIG + 1;
1586
818k
    dval(&rv) = y;
1587
818k
    if (k > 9) {
1588
66.0k
        dval(&rv) = tens[k - 9] * dval(&rv) + z;
1589
66.0k
    }
1590
818k
    if (nd <= DBL_DIG
1591
762k
        && Flt_Rounds == 1
1592
818k
        ) {
1593
762k
        if (!e)
1594
12.3k
            goto ret;
1595
750k
        if (e > 0) {
1596
700k
            if (e <= Ten_pmax) {
1597
25.1k
                dval(&rv) *= tens[e];
1598
25.1k
                goto ret;
1599
25.1k
            }
1600
675k
            i = DBL_DIG - nd;
1601
675k
            if (e <= Ten_pmax + i) {
1602
                /* A fancier test would sometimes let us do
1603
                 * this for larger i values.
1604
                 */
1605
2.87k
                e -= i;
1606
2.87k
                dval(&rv) *= tens[i];
1607
2.87k
                dval(&rv) *= tens[e];
1608
2.87k
                goto ret;
1609
2.87k
            }
1610
675k
        }
1611
50.0k
        else if (e >= -Ten_pmax) {
1612
30.8k
            dval(&rv) /= tens[-e];
1613
30.8k
            goto ret;
1614
30.8k
        }
1615
750k
    }
1616
747k
    e1 += nd - k;
1617
1618
747k
    bc.scale = 0;
1619
1620
    /* Get starting approximation = rv * 10**e1 */
1621
1622
747k
    if (e1 > 0) {
1623
710k
        if ((i = e1 & 15))
1624
695k
            dval(&rv) *= tens[i];
1625
710k
        if (e1 &= ~15) {
1626
695k
            if (e1 > DBL_MAX_10_EXP)
1627
641k
                goto ovfl;
1628
54.2k
            e1 >>= 4;
1629
128k
            for(j = 0; e1 > 1; j++, e1 >>= 1)
1630
74.0k
                if (e1 & 1)
1631
32.8k
                    dval(&rv) *= bigtens[j];
1632
            /* The last multiplication could overflow. */
1633
54.2k
            word0(&rv) -= P*Exp_msk1;
1634
54.2k
            dval(&rv) *= bigtens[j];
1635
54.2k
            if ((z = word0(&rv) & Exp_mask)
1636
54.2k
                > Exp_msk1*(DBL_MAX_EXP+Bias-P))
1637
610
                goto ovfl;
1638
53.6k
            if (z > Exp_msk1*(DBL_MAX_EXP+Bias-1-P)) {
1639
                /* set to largest number */
1640
                /* (Can't trust DBL_MAX) */
1641
457
                word0(&rv) = Big0;
1642
457
                word1(&rv) = Big1;
1643
457
            }
1644
53.2k
            else
1645
53.2k
                word0(&rv) += P*Exp_msk1;
1646
53.6k
        }
1647
710k
    }
1648
37.0k
    else if (e1 < 0) {
1649
        /* The input decimal value lies in [10**e1, 10**(e1+16)).
1650
1651
           If e1 <= -512, underflow immediately.
1652
           If e1 <= -256, set bc.scale to 2*P.
1653
1654
           So for input value < 1e-256, bc.scale is always set;
1655
           for input value >= 1e-240, bc.scale is never set.
1656
           For input values in [1e-256, 1e-240), bc.scale may or may
1657
           not be set. */
1658
1659
33.7k
        e1 = -e1;
1660
33.7k
        if ((i = e1 & 15))
1661
29.7k
            dval(&rv) /= tens[i];
1662
33.7k
        if (e1 >>= 4) {
1663
24.3k
            if (e1 >= 1 << n_bigtens)
1664
1.02k
                goto undfl;
1665
23.3k
            if (e1 & Scale_Bit)
1666
18.3k
                bc.scale = 2*P;
1667
122k
            for(j = 0; e1 > 0; j++, e1 >>= 1)
1668
99.4k
                if (e1 & 1)
1669
58.9k
                    dval(&rv) *= tinytens[j];
1670
23.3k
            if (bc.scale && (j = 2*P + 1 - ((word0(&rv) & Exp_mask)
1671
18.3k
                                            >> Exp_shift)) > 0) {
1672
                /* scaled rv is denormal; clear j low bits */
1673
16.9k
                if (j >= 32) {
1674
10.2k
                    word1(&rv) = 0;
1675
10.2k
                    if (j >= 53)
1676
5.68k
                        word0(&rv) = (P+2)*Exp_msk1;
1677
4.56k
                    else
1678
4.56k
                        word0(&rv) &= 0xffffffff << (j-32);
1679
10.2k
                }
1680
6.70k
                else
1681
6.70k
                    word1(&rv) &= 0xffffffff << j;
1682
16.9k
            }
1683
23.3k
            if (!dval(&rv))
1684
0
                goto undfl;
1685
23.3k
        }
1686
33.7k
    }
1687
1688
    /* Now the hard part -- adjusting rv to the correct value.*/
1689
1690
    /* Put digits into bd: true value = bd * 10^e */
1691
1692
104k
    bc.nd = nd;
1693
104k
    bc.nd0 = nd0;       /* Only needed if nd > STRTOD_DIGLIM, but done here */
1694
                        /* to silence an erroneous warning about bc.nd0 */
1695
                        /* possibly not being initialized. */
1696
104k
    if (nd > STRTOD_DIGLIM) {
1697
        /* ASSERT(STRTOD_DIGLIM >= 18); 18 == one more than the */
1698
        /* minimum number of decimal digits to distinguish double values */
1699
        /* in IEEE arithmetic. */
1700
1701
        /* Truncate input to 18 significant digits, then discard any trailing
1702
           zeros on the result by updating nd, nd0, e and y suitably. (There's
1703
           no need to update z; it's not reused beyond this point.) */
1704
177k
        for (i = 18; i > 0; ) {
1705
            /* scan back until we hit a nonzero digit.  significant digit 'i'
1706
            is s0[i] if i < nd0, s0[i+1] if i >= nd0. */
1707
177k
            --i;
1708
177k
            if (s0[i < nd0 ? i : i+1] != '0') {
1709
29.1k
                ++i;
1710
29.1k
                break;
1711
29.1k
            }
1712
177k
        }
1713
29.1k
        e += nd - i;
1714
29.1k
        nd = i;
1715
29.1k
        if (nd0 > nd)
1716
26.0k
            nd0 = nd;
1717
29.1k
        if (nd < 9) { /* must recompute y */
1718
10.9k
            y = 0;
1719
62.1k
            for(i = 0; i < nd0; ++i)
1720
51.1k
                y = 10*y + s0[i] - '0';
1721
18.7k
            for(; i < nd; ++i)
1722
7.72k
                y = 10*y + s0[i+1] - '0';
1723
10.9k
        }
1724
29.1k
    }
1725
104k
    bd0 = s2b(s0, nd0, nd, y);
1726
104k
    if (bd0 == NULL)
1727
0
        goto failed_malloc;
1728
1729
    /* Notation for the comments below.  Write:
1730
1731
         - dv for the absolute value of the number represented by the original
1732
           decimal input string.
1733
1734
         - if we've truncated dv, write tdv for the truncated value.
1735
           Otherwise, set tdv == dv.
1736
1737
         - srv for the quantity rv/2^bc.scale; so srv is the current binary
1738
           approximation to tdv (and dv).  It should be exactly representable
1739
           in an IEEE 754 double.
1740
    */
1741
1742
134k
    for(;;) {
1743
1744
        /* This is the main correction loop for _Py_dg_strtod.
1745
1746
           We've got a decimal value tdv, and a floating-point approximation
1747
           srv=rv/2^bc.scale to tdv.  The aim is to determine whether srv is
1748
           close enough (i.e., within 0.5 ulps) to tdv, and to compute a new
1749
           approximation if not.
1750
1751
           To determine whether srv is close enough to tdv, compute integers
1752
           bd, bb and bs proportional to tdv, srv and 0.5 ulp(srv)
1753
           respectively, and then use integer arithmetic to determine whether
1754
           |tdv - srv| is less than, equal to, or greater than 0.5 ulp(srv).
1755
        */
1756
1757
134k
        bd = Balloc(bd0->k);
1758
134k
        if (bd == NULL) {
1759
0
            goto failed_malloc;
1760
0
        }
1761
134k
        Bcopy(bd, bd0);
1762
134k
        bb = sd2b(&rv, bc.scale, &bbe);   /* srv = bb * 2^bbe */
1763
134k
        if (bb == NULL) {
1764
0
            goto failed_malloc;
1765
0
        }
1766
        /* Record whether lsb of bb is odd, in case we need this
1767
           for the round-to-even step later. */
1768
134k
        odd = bb->x[0] & 1;
1769
1770
        /* tdv = bd * 10**e;  srv = bb * 2**bbe */
1771
134k
        bs = i2b(1);
1772
134k
        if (bs == NULL) {
1773
0
            goto failed_malloc;
1774
0
        }
1775
1776
134k
        if (e >= 0) {
1777
75.8k
            bb2 = bb5 = 0;
1778
75.8k
            bd2 = bd5 = e;
1779
75.8k
        }
1780
58.5k
        else {
1781
58.5k
            bb2 = bb5 = -e;
1782
58.5k
            bd2 = bd5 = 0;
1783
58.5k
        }
1784
134k
        if (bbe >= 0)
1785
84.1k
            bb2 += bbe;
1786
50.2k
        else
1787
50.2k
            bd2 -= bbe;
1788
134k
        bs2 = bb2;
1789
134k
        bb2++;
1790
134k
        bd2++;
1791
1792
        /* At this stage bd5 - bb5 == e == bd2 - bb2 + bbe, bb2 - bs2 == 1,
1793
           and bs == 1, so:
1794
1795
              tdv == bd * 10**e = bd * 2**(bbe - bb2 + bd2) * 5**(bd5 - bb5)
1796
              srv == bb * 2**bbe = bb * 2**(bbe - bb2 + bb2)
1797
              0.5 ulp(srv) == 2**(bbe-1) = bs * 2**(bbe - bb2 + bs2)
1798
1799
           It follows that:
1800
1801
              M * tdv = bd * 2**bd2 * 5**bd5
1802
              M * srv = bb * 2**bb2 * 5**bb5
1803
              M * 0.5 ulp(srv) = bs * 2**bs2 * 5**bb5
1804
1805
           for some constant M.  (Actually, M == 2**(bb2 - bbe) * 5**bb5, but
1806
           this fact is not needed below.)
1807
        */
1808
1809
        /* Remove factor of 2**i, where i = min(bb2, bd2, bs2). */
1810
134k
        i = bb2 < bd2 ? bb2 : bd2;
1811
134k
        if (i > bs2)
1812
49.2k
            i = bs2;
1813
134k
        if (i > 0) {
1814
133k
            bb2 -= i;
1815
133k
            bd2 -= i;
1816
133k
            bs2 -= i;
1817
133k
        }
1818
1819
        /* Scale bb, bd, bs by the appropriate powers of 2 and 5. */
1820
134k
        if (bb5 > 0) {
1821
58.5k
            bs = pow5mult(bs, bb5);
1822
58.5k
            if (bs == NULL) {
1823
0
                goto failed_malloc;
1824
0
            }
1825
58.5k
            Bigint *bb1 = mult(bs, bb);
1826
58.5k
            Bfree(bb);
1827
58.5k
            bb = bb1;
1828
58.5k
            if (bb == NULL) {
1829
0
                goto failed_malloc;
1830
0
            }
1831
58.5k
        }
1832
134k
        if (bb2 > 0) {
1833
134k
            bb = lshift(bb, bb2);
1834
134k
            if (bb == NULL) {
1835
0
                goto failed_malloc;
1836
0
            }
1837
134k
        }
1838
134k
        if (bd5 > 0) {
1839
69.7k
            bd = pow5mult(bd, bd5);
1840
69.7k
            if (bd == NULL) {
1841
0
                goto failed_malloc;
1842
0
            }
1843
69.7k
        }
1844
134k
        if (bd2 > 0) {
1845
49.2k
            bd = lshift(bd, bd2);
1846
49.2k
            if (bd == NULL) {
1847
0
                goto failed_malloc;
1848
0
            }
1849
49.2k
        }
1850
134k
        if (bs2 > 0) {
1851
81.3k
            bs = lshift(bs, bs2);
1852
81.3k
            if (bs == NULL) {
1853
0
                goto failed_malloc;
1854
0
            }
1855
81.3k
        }
1856
1857
        /* Now bd, bb and bs are scaled versions of tdv, srv and 0.5 ulp(srv),
1858
           respectively.  Compute the difference |tdv - srv|, and compare
1859
           with 0.5 ulp(srv). */
1860
1861
134k
        delta = diff(bb, bd);
1862
134k
        if (delta == NULL) {
1863
0
            goto failed_malloc;
1864
0
        }
1865
134k
        dsign = delta->sign;
1866
134k
        delta->sign = 0;
1867
134k
        i = cmp(delta, bs);
1868
134k
        if (bc.nd > nd && i <= 0) {
1869
29.1k
            if (dsign)
1870
21.4k
                break;  /* Must use bigcomp(). */
1871
1872
            /* Here rv overestimates the truncated decimal value by at most
1873
               0.5 ulp(rv).  Hence rv either overestimates the true decimal
1874
               value by <= 0.5 ulp(rv), or underestimates it by some small
1875
               amount (< 0.1 ulp(rv)); either way, rv is within 0.5 ulps of
1876
               the true decimal value, so it's possible to exit.
1877
1878
               Exception: if scaled rv is a normal exact power of 2, but not
1879
               DBL_MIN, then rv - 0.5 ulp(rv) takes us all the way down to the
1880
               next double, so the correctly rounded result is either rv - 0.5
1881
               ulp(rv) or rv; in this case, use bigcomp to distinguish. */
1882
1883
7.79k
            if (!word1(&rv) && !(word0(&rv) & Bndry_mask)) {
1884
                /* rv can't be 0, since it's an overestimate for some
1885
                   nonzero value.  So rv is a normal power of 2. */
1886
1.11k
                j = (int)(word0(&rv) & Exp_mask) >> Exp_shift;
1887
                /* rv / 2^bc.scale = 2^(j - 1023 - bc.scale); use bigcomp if
1888
                   rv / 2^bc.scale >= 2^-1021. */
1889
1.11k
                if (j - bc.scale >= 2) {
1890
827
                    dval(&rv) -= 0.5 * sulp(&rv, &bc);
1891
827
                    break; /* Use bigcomp. */
1892
827
                }
1893
1.11k
            }
1894
1895
6.96k
            {
1896
6.96k
                bc.nd = nd;
1897
6.96k
                i = -1; /* Discarded digits make delta smaller. */
1898
6.96k
            }
1899
6.96k
        }
1900
1901
112k
        if (i < 0) {
1902
            /* Error is less than half an ulp -- check for
1903
             * special case of mantissa a power of two.
1904
             */
1905
56.6k
            if (dsign || word1(&rv) || word0(&rv) & Bndry_mask
1906
4.71k
                || (word0(&rv) & Exp_mask) <= (2*P+1)*Exp_msk1
1907
56.6k
                ) {
1908
53.3k
                break;
1909
53.3k
            }
1910
3.26k
            if (!delta->x[0] && delta->wds <= 1) {
1911
                /* exact result */
1912
433
                break;
1913
433
            }
1914
2.83k
            delta = lshift(delta,Log2P);
1915
2.83k
            if (delta == NULL) {
1916
0
                goto failed_malloc;
1917
0
            }
1918
2.83k
            if (cmp(delta, bs) > 0)
1919
1.00k
                goto drop_down;
1920
1.82k
            break;
1921
2.83k
        }
1922
55.5k
        if (i == 0) {
1923
            /* exactly half-way between */
1924
3.39k
            if (dsign) {
1925
1.83k
                if ((word0(&rv) & Bndry_mask1) == Bndry_mask1
1926
825
                    &&  word1(&rv) == (
1927
825
                        (bc.scale &&
1928
0
                         (y = word0(&rv) & Exp_mask) <= 2*P*Exp_msk1) ?
1929
0
                        (0xffffffff & (0xffffffff << (2*P+1-(y>>Exp_shift)))) :
1930
825
                        0xffffffff)) {
1931
                    /*boundary case -- increment exponent*/
1932
462
                    word0(&rv) = (word0(&rv) & Exp_mask)
1933
462
                        + Exp_msk1
1934
462
                        ;
1935
462
                    word1(&rv) = 0;
1936
                    /* dsign = 0; */
1937
462
                    break;
1938
462
                }
1939
1.83k
            }
1940
1.56k
            else if (!(word0(&rv) & Bndry_mask) && !word1(&rv)) {
1941
1.00k
              drop_down:
1942
                /* boundary case -- decrement exponent */
1943
1.00k
                if (bc.scale) {
1944
0
                    L = word0(&rv) & Exp_mask;
1945
0
                    if (L <= (2*P+1)*Exp_msk1) {
1946
0
                        if (L > (P+2)*Exp_msk1)
1947
                            /* round even ==> */
1948
                            /* accept rv */
1949
0
                            break;
1950
                        /* rv = smallest denormal */
1951
0
                        if (bc.nd > nd)
1952
0
                            break;
1953
0
                        goto undfl;
1954
0
                    }
1955
0
                }
1956
1.00k
                L = (word0(&rv) & Exp_mask) - Exp_msk1;
1957
1.00k
                word0(&rv) = L | Bndry_mask1;
1958
1.00k
                word1(&rv) = 0xffffffff;
1959
1.00k
                break;
1960
1.00k
            }
1961
2.93k
            if (!odd)
1962
2.28k
                break;
1963
655
            if (dsign)
1964
337
                dval(&rv) += sulp(&rv, &bc);
1965
318
            else {
1966
318
                dval(&rv) -= sulp(&rv, &bc);
1967
318
                if (!dval(&rv)) {
1968
0
                    if (bc.nd >nd)
1969
0
                        break;
1970
0
                    goto undfl;
1971
0
                }
1972
318
            }
1973
            /* dsign = 1 - dsign; */
1974
655
            break;
1975
655
        }
1976
52.1k
        if ((aadj = ratio(delta, bs)) <= 2.) {
1977
37.7k
            if (dsign)
1978
14.5k
                aadj = aadj1 = 1.;
1979
23.1k
            else if (word1(&rv) || word0(&rv) & Bndry_mask) {
1980
17.9k
                if (word1(&rv) == Tiny1 && !word0(&rv)) {
1981
0
                    if (bc.nd >nd)
1982
0
                        break;
1983
0
                    goto undfl;
1984
0
                }
1985
17.9k
                aadj = 1.;
1986
17.9k
                aadj1 = -1.;
1987
17.9k
            }
1988
5.16k
            else {
1989
                /* special case -- power of FLT_RADIX to be */
1990
                /* rounded down... */
1991
1992
5.16k
                if (aadj < 2./FLT_RADIX)
1993
0
                    aadj = 1./FLT_RADIX;
1994
5.16k
                else
1995
5.16k
                    aadj *= 0.5;
1996
5.16k
                aadj1 = -aadj;
1997
5.16k
            }
1998
37.7k
        }
1999
14.4k
        else {
2000
14.4k
            aadj *= 0.5;
2001
14.4k
            aadj1 = dsign ? aadj : -aadj;
2002
14.4k
            if (Flt_Rounds == 0)
2003
0
                aadj1 += 0.5;
2004
14.4k
        }
2005
52.1k
        y = word0(&rv) & Exp_mask;
2006
2007
        /* Check for overflow */
2008
2009
52.1k
        if (y == Exp_msk1*(DBL_MAX_EXP+Bias-1)) {
2010
2.13k
            dval(&rv0) = dval(&rv);
2011
2.13k
            word0(&rv) -= P*Exp_msk1;
2012
2.13k
            adj.d = aadj1 * ulp(&rv);
2013
2.13k
            dval(&rv) += adj.d;
2014
2.13k
            if ((word0(&rv) & Exp_mask) >=
2015
2.13k
                Exp_msk1*(DBL_MAX_EXP+Bias-P)) {
2016
979
                if (word0(&rv0) == Big0 && word1(&rv0) == Big1) {
2017
718
                    goto ovfl;
2018
718
                }
2019
261
                word0(&rv) = Big0;
2020
261
                word1(&rv) = Big1;
2021
261
                goto cont;
2022
979
            }
2023
1.15k
            else
2024
1.15k
                word0(&rv) += P*Exp_msk1;
2025
2.13k
        }
2026
50.0k
        else {
2027
50.0k
            if (bc.scale && y <= 2*P*Exp_msk1) {
2028
14.0k
                if (aadj <= 0x7fffffff) {
2029
14.0k
                    if ((z = (ULong)aadj) <= 0)
2030
740
                        z = 1;
2031
14.0k
                    aadj = z;
2032
14.0k
                    aadj1 = dsign ? aadj : -aadj;
2033
14.0k
                }
2034
14.0k
                dval(&aadj2) = aadj1;
2035
14.0k
                word0(&aadj2) += (2*P+1)*Exp_msk1 - y;
2036
14.0k
                aadj1 = dval(&aadj2);
2037
14.0k
            }
2038
50.0k
            adj.d = aadj1 * ulp(&rv);
2039
50.0k
            dval(&rv) += adj.d;
2040
50.0k
        }
2041
51.1k
        z = word0(&rv) & Exp_mask;
2042
51.1k
        if (bc.nd == nd) {
2043
38.2k
            if (!bc.scale)
2044
24.1k
                if (y == z) {
2045
                    /* Can we stop now? */
2046
22.5k
                    L = (Long)aadj;
2047
22.5k
                    aadj -= L;
2048
                    /* The tolerances below are conservative. */
2049
22.5k
                    if (dsign || word1(&rv) || word0(&rv) & Bndry_mask) {
2050
22.4k
                        if (aadj < .4999999 || aadj > .5000001)
2051
21.6k
                            break;
2052
22.4k
                    }
2053
15
                    else if (aadj < .4999999/FLT_RADIX)
2054
15
                        break;
2055
22.5k
                }
2056
38.2k
        }
2057
29.7k
      cont:
2058
29.7k
        Bfree(bb); bb = NULL;
2059
29.7k
        Bfree(bd); bd = NULL;
2060
29.7k
        Bfree(bs); bs = NULL;
2061
29.7k
        Bfree(delta); delta = NULL;
2062
29.7k
    }
2063
103k
    if (bc.nd > nd) {
2064
22.2k
        error = bigcomp(&rv, s0, &bc);
2065
22.2k
        if (error)
2066
0
            goto failed_malloc;
2067
22.2k
    }
2068
2069
103k
    if (bc.scale) {
2070
18.3k
        word0(&rv0) = Exp_1 - 2*P*Exp_msk1;
2071
18.3k
        word1(&rv0) = 0;
2072
18.3k
        dval(&rv) *= dval(&rv0);
2073
18.3k
    }
2074
2075
186k
  ret:
2076
186k
    result = sign ? -dval(&rv) : dval(&rv);
2077
186k
    goto done;
2078
2079
19
  parse_error:
2080
19
    result = 0.0;
2081
19
    goto done;
2082
2083
0
  failed_malloc:
2084
0
    errno = ENOMEM;
2085
0
    result = -1.0;
2086
0
    goto done;
2087
2088
1.02k
  undfl:
2089
1.02k
    result = sign ? -0.0 : 0.0;
2090
1.02k
    goto done;
2091
2092
642k
  ovfl:
2093
642k
    errno = ERANGE;
2094
    /* Can't trust HUGE_VAL */
2095
642k
    word0(&rv) = Exp_mask;
2096
642k
    word1(&rv) = 0;
2097
642k
    result = sign ? -dval(&rv) : dval(&rv);
2098
642k
    goto done;
2099
2100
829k
  done:
2101
829k
    Bfree(bb);
2102
829k
    Bfree(bd);
2103
829k
    Bfree(bs);
2104
829k
    Bfree(bd0);
2105
829k
    Bfree(delta);
2106
829k
    return result;
2107
2108
103k
}
2109
2110
static char *
2111
rv_alloc(int i)
2112
50.2k
{
2113
50.2k
    int j, k, *r;
2114
2115
50.2k
    j = sizeof(ULong);
2116
50.2k
    for(k = 0;
2117
50.2k
        sizeof(Bigint) - sizeof(ULong) - sizeof(int) + j <= (unsigned)i;
2118
50.2k
        j <<= 1)
2119
0
        k++;
2120
50.2k
    r = (int*)Balloc(k);
2121
50.2k
    if (r == NULL)
2122
0
        return NULL;
2123
50.2k
    *r = k;
2124
50.2k
    return (char *)(r+1);
2125
50.2k
}
2126
2127
static char *
2128
nrv_alloc(const char *s, char **rve, int n)
2129
5.61k
{
2130
5.61k
    char *rv, *t;
2131
2132
5.61k
    rv = rv_alloc(n);
2133
5.61k
    if (rv == NULL)
2134
0
        return NULL;
2135
5.61k
    t = rv;
2136
14.5k
    while((*t = *s++)) t++;
2137
5.61k
    if (rve)
2138
5.61k
        *rve = t;
2139
5.61k
    return rv;
2140
5.61k
}
2141
2142
/* freedtoa(s) must be used to free values s returned by dtoa
2143
 * when MULTIPLE_THREADS is #defined.  It should be used in all cases,
2144
 * but for consistency with earlier versions of dtoa, it is optional
2145
 * when MULTIPLE_THREADS is not defined.
2146
 */
2147
2148
void
2149
_Py_dg_freedtoa(char *s)
2150
50.2k
{
2151
50.2k
    Bigint *b = (Bigint *)((int *)s - 1);
2152
50.2k
    b->maxwds = 1 << (b->k = *(int*)b);
2153
50.2k
    Bfree(b);
2154
50.2k
}
2155
2156
/* dtoa for IEEE arithmetic (dmg): convert double to ASCII string.
2157
 *
2158
 * Inspired by "How to Print Floating-Point Numbers Accurately" by
2159
 * Guy L. Steele, Jr. and Jon L. White [Proc. ACM SIGPLAN '90, pp. 112-126].
2160
 *
2161
 * Modifications:
2162
 *      1. Rather than iterating, we use a simple numeric overestimate
2163
 *         to determine k = floor(log10(d)).  We scale relevant
2164
 *         quantities using O(log2(k)) rather than O(k) multiplications.
2165
 *      2. For some modes > 2 (corresponding to ecvt and fcvt), we don't
2166
 *         try to generate digits strictly left to right.  Instead, we
2167
 *         compute with fewer bits and propagate the carry if necessary
2168
 *         when rounding the final digit up.  This is often faster.
2169
 *      3. Under the assumption that input will be rounded nearest,
2170
 *         mode 0 renders 1e23 as 1e23 rather than 9.999999999999999e22.
2171
 *         That is, we allow equality in stopping tests when the
2172
 *         round-nearest rule will give the same floating-point value
2173
 *         as would satisfaction of the stopping test with strict
2174
 *         inequality.
2175
 *      4. We remove common factors of powers of 2 from relevant
2176
 *         quantities.
2177
 *      5. When converting floating-point integers less than 1e16,
2178
 *         we use floating-point arithmetic rather than resorting
2179
 *         to multiple-precision integers.
2180
 *      6. When asked to produce fewer than 15 digits, we first try
2181
 *         to get by with floating-point arithmetic; we resort to
2182
 *         multiple-precision integer arithmetic only if we cannot
2183
 *         guarantee that the floating-point calculation has given
2184
 *         the correctly rounded result.  For k requested digits and
2185
 *         "uniformly" distributed input, the probability is
2186
 *         something like 10^(k-15) that we must resort to the Long
2187
 *         calculation.
2188
 */
2189
2190
/* Additional notes (METD): (1) returns NULL on failure.  (2) to avoid memory
2191
   leakage, a successful call to _Py_dg_dtoa should always be matched by a
2192
   call to _Py_dg_freedtoa. */
2193
2194
char *
2195
_Py_dg_dtoa(double dd, int mode, int ndigits,
2196
            int *decpt, int *sign, char **rve)
2197
50.2k
{
2198
    /*  Arguments ndigits, decpt, sign are similar to those
2199
        of ecvt and fcvt; trailing zeros are suppressed from
2200
        the returned string.  If not null, *rve is set to point
2201
        to the end of the return value.  If d is +-Infinity or NaN,
2202
        then *decpt is set to 9999.
2203
2204
        mode:
2205
        0 ==> shortest string that yields d when read in
2206
        and rounded to nearest.
2207
        1 ==> like 0, but with Steele & White stopping rule;
2208
        e.g. with IEEE P754 arithmetic , mode 0 gives
2209
        1e23 whereas mode 1 gives 9.999999999999999e22.
2210
        2 ==> max(1,ndigits) significant digits.  This gives a
2211
        return value similar to that of ecvt, except
2212
        that trailing zeros are suppressed.
2213
        3 ==> through ndigits past the decimal point.  This
2214
        gives a return value similar to that from fcvt,
2215
        except that trailing zeros are suppressed, and
2216
        ndigits can be negative.
2217
        4,5 ==> similar to 2 and 3, respectively, but (in
2218
        round-nearest mode) with the tests of mode 0 to
2219
        possibly return a shorter string that rounds to d.
2220
        With IEEE arithmetic and compilation with
2221
        -DHonor_FLT_ROUNDS, modes 4 and 5 behave the same
2222
        as modes 2 and 3 when FLT_ROUNDS != 1.
2223
        6-9 ==> Debugging modes similar to mode - 4:  don't try
2224
        fast floating-point estimate (if applicable).
2225
2226
        Values of mode other than 0-9 are treated as mode 0.
2227
2228
        Sufficient space is allocated to the return value
2229
        to hold the suppressed trailing zeros.
2230
    */
2231
2232
50.2k
    int bbits, b2, b5, be, dig, i, ieps, ilim, ilim0, ilim1,
2233
50.2k
        j, j1, k, k0, k_check, leftright, m2, m5, s2, s5,
2234
50.2k
        spec_case, try_quick;
2235
50.2k
    Long L;
2236
50.2k
    int denorm;
2237
50.2k
    ULong x;
2238
50.2k
    Bigint *b, *b1, *delta, *mlo, *mhi, *S;
2239
50.2k
    U d2, eps, u;
2240
50.2k
    double ds;
2241
50.2k
    char *s, *s0;
2242
2243
    /* set pointers to NULL, to silence gcc compiler warnings and make
2244
       cleanup easier on error */
2245
50.2k
    mlo = mhi = S = 0;
2246
50.2k
    s0 = 0;
2247
2248
50.2k
    u.d = dd;
2249
50.2k
    if (word0(&u) & Sign_bit) {
2250
        /* set sign for everything, including 0's and NaNs */
2251
14.5k
        *sign = 1;
2252
14.5k
        word0(&u) &= ~Sign_bit; /* clear sign bit */
2253
14.5k
    }
2254
35.6k
    else
2255
35.6k
        *sign = 0;
2256
2257
    /* quick return for Infinities, NaNs and zeros */
2258
50.2k
    if ((word0(&u) & Exp_mask) == Exp_mask)
2259
478
    {
2260
        /* Infinity or NaN */
2261
478
        *decpt = 9999;
2262
478
        if (!word1(&u) && !(word0(&u) & 0xfffff))
2263
478
            return nrv_alloc("Infinity", rve, 8);
2264
0
        return nrv_alloc("NaN", rve, 3);
2265
478
    }
2266
49.7k
    if (!dval(&u)) {
2267
5.13k
        *decpt = 1;
2268
5.13k
        return nrv_alloc("0", rve, 1);
2269
5.13k
    }
2270
2271
    /* compute k = floor(log10(d)).  The computation may leave k
2272
       one too large, but should never leave k too small. */
2273
44.6k
    b = d2b(&u, &be, &bbits);
2274
44.6k
    if (b == NULL)
2275
0
        goto failed_malloc;
2276
44.6k
    if ((i = (int)(word0(&u) >> Exp_shift1 & (Exp_mask>>Exp_shift1)))) {
2277
40.4k
        dval(&d2) = dval(&u);
2278
40.4k
        word0(&d2) &= Frac_mask1;
2279
40.4k
        word0(&d2) |= Exp_11;
2280
2281
        /* log(x)       ~=~ log(1.5) + (x-1.5)/1.5
2282
         * log10(x)      =  log(x) / log(10)
2283
         *              ~=~ log(1.5)/log(10) + (x-1.5)/(1.5*log(10))
2284
         * log10(d) = (i-Bias)*log(2)/log(10) + log10(d2)
2285
         *
2286
         * This suggests computing an approximation k to log10(d) by
2287
         *
2288
         * k = (i - Bias)*0.301029995663981
2289
         *      + ( (d2-1.5)*0.289529654602168 + 0.176091259055681 );
2290
         *
2291
         * We want k to be too large rather than too small.
2292
         * The error in the first-order Taylor series approximation
2293
         * is in our favor, so we just round up the constant enough
2294
         * to compensate for any error in the multiplication of
2295
         * (i - Bias) by 0.301029995663981; since |i - Bias| <= 1077,
2296
         * and 1077 * 0.30103 * 2^-52 ~=~ 7.2e-14,
2297
         * adding 1e-13 to the constant term more than suffices.
2298
         * Hence we adjust the constant term to 0.1760912590558.
2299
         * (We could get a more accurate k by invoking log10,
2300
         *  but this is probably not worthwhile.)
2301
         */
2302
2303
40.4k
        i -= Bias;
2304
40.4k
        denorm = 0;
2305
40.4k
    }
2306
4.13k
    else {
2307
        /* d is denormalized */
2308
2309
4.13k
        i = bbits + be + (Bias + (P-1) - 1);
2310
4.13k
        x = i > 32  ? word0(&u) << (64 - i) | word1(&u) >> (i - 32)
2311
4.13k
            : word1(&u) << (32 - i);
2312
4.13k
        dval(&d2) = x;
2313
4.13k
        word0(&d2) -= 31*Exp_msk1; /* adjust exponent */
2314
4.13k
        i -= (Bias + (P-1) - 1) + 1;
2315
4.13k
        denorm = 1;
2316
4.13k
    }
2317
44.6k
    ds = (dval(&d2)-1.5)*0.289529654602168 + 0.1760912590558 +
2318
44.6k
        i*0.301029995663981;
2319
44.6k
    k = (int)ds;
2320
44.6k
    if (ds < 0. && ds != k)
2321
12.8k
        k--;    /* want k = floor(ds) */
2322
44.6k
    k_check = 1;
2323
44.6k
    if (k >= 0 && k <= Ten_pmax) {
2324
20.6k
        if (dval(&u) < tens[k])
2325
2.29k
            k--;
2326
20.6k
        k_check = 0;
2327
20.6k
    }
2328
44.6k
    j = bbits - i - 1;
2329
44.6k
    if (j >= 0) {
2330
19.8k
        b2 = 0;
2331
19.8k
        s2 = j;
2332
19.8k
    }
2333
24.7k
    else {
2334
24.7k
        b2 = -j;
2335
24.7k
        s2 = 0;
2336
24.7k
    }
2337
44.6k
    if (k >= 0) {
2338
31.0k
        b5 = 0;
2339
31.0k
        s5 = k;
2340
31.0k
        s2 += k;
2341
31.0k
    }
2342
13.5k
    else {
2343
13.5k
        b2 -= k;
2344
13.5k
        b5 = -k;
2345
13.5k
        s5 = 0;
2346
13.5k
    }
2347
44.6k
    if (mode < 0 || mode > 9)
2348
0
        mode = 0;
2349
2350
44.6k
    try_quick = 1;
2351
2352
44.6k
    if (mode > 5) {
2353
0
        mode -= 4;
2354
0
        try_quick = 0;
2355
0
    }
2356
44.6k
    leftright = 1;
2357
44.6k
    ilim = ilim1 = -1;  /* Values for cases 0 and 1; done here to */
2358
    /* silence erroneous "gcc -Wall" warning. */
2359
44.6k
    switch(mode) {
2360
44.5k
    case 0:
2361
44.5k
    case 1:
2362
44.5k
        i = 18;
2363
44.5k
        ndigits = 0;
2364
44.5k
        break;
2365
0
    case 2:
2366
0
        leftright = 0;
2367
0
        _Py_FALLTHROUGH;
2368
0
    case 4:
2369
0
        if (ndigits <= 0)
2370
0
            ndigits = 1;
2371
0
        ilim = ilim1 = i = ndigits;
2372
0
        break;
2373
107
    case 3:
2374
107
        leftright = 0;
2375
107
        _Py_FALLTHROUGH;
2376
107
    case 5:
2377
107
        i = ndigits + k + 1;
2378
107
        ilim = i;
2379
107
        ilim1 = i - 1;
2380
107
        if (i <= 0)
2381
0
            i = 1;
2382
44.6k
    }
2383
44.6k
    s0 = rv_alloc(i);
2384
44.6k
    if (s0 == NULL)
2385
0
        goto failed_malloc;
2386
44.6k
    s = s0;
2387
2388
2389
44.6k
    if (ilim >= 0 && ilim <= Quick_max && try_quick) {
2390
2391
        /* Try to get by with floating-point arithmetic. */
2392
2393
107
        i = 0;
2394
107
        dval(&d2) = dval(&u);
2395
107
        k0 = k;
2396
107
        ilim0 = ilim;
2397
107
        ieps = 2; /* conservative */
2398
107
        if (k > 0) {
2399
86
            ds = tens[k&0xf];
2400
86
            j = k >> 4;
2401
86
            if (j & Bletch) {
2402
                /* prevent overflows */
2403
0
                j &= Bletch - 1;
2404
0
                dval(&u) /= bigtens[n_bigtens-1];
2405
0
                ieps++;
2406
0
            }
2407
86
            for(; j; j >>= 1, i++)
2408
0
                if (j & 1) {
2409
0
                    ieps++;
2410
0
                    ds *= bigtens[i];
2411
0
                }
2412
86
            dval(&u) /= ds;
2413
86
        }
2414
21
        else if ((j1 = -k)) {
2415
0
            dval(&u) *= tens[j1 & 0xf];
2416
0
            for(j = j1 >> 4; j; j >>= 1, i++)
2417
0
                if (j & 1) {
2418
0
                    ieps++;
2419
0
                    dval(&u) *= bigtens[i];
2420
0
                }
2421
0
        }
2422
107
        if (k_check && dval(&u) < 1. && ilim > 0) {
2423
0
            if (ilim1 <= 0)
2424
0
                goto fast_failed;
2425
0
            ilim = ilim1;
2426
0
            k--;
2427
0
            dval(&u) *= 10.;
2428
0
            ieps++;
2429
0
        }
2430
107
        dval(&eps) = ieps*dval(&u) + 7.;
2431
107
        word0(&eps) -= (P-1)*Exp_msk1;
2432
107
        if (ilim == 0) {
2433
0
            S = mhi = 0;
2434
0
            dval(&u) -= 5.;
2435
0
            if (dval(&u) > dval(&eps))
2436
0
                goto one_digit;
2437
0
            if (dval(&u) < -dval(&eps))
2438
0
                goto no_digits;
2439
0
            goto fast_failed;
2440
0
        }
2441
107
        if (leftright) {
2442
            /* Use Steele & White method of only
2443
             * generating digits needed.
2444
             */
2445
0
            dval(&eps) = 0.5/tens[ilim-1] - dval(&eps);
2446
0
            for(i = 0;;) {
2447
0
                L = (Long)dval(&u);
2448
0
                dval(&u) -= L;
2449
0
                *s++ = '0' + (int)L;
2450
0
                if (dval(&u) < dval(&eps))
2451
0
                    goto ret1;
2452
0
                if (1. - dval(&u) < dval(&eps))
2453
0
                    goto bump_up;
2454
0
                if (++i >= ilim)
2455
0
                    break;
2456
0
                dval(&eps) *= 10.;
2457
0
                dval(&u) *= 10.;
2458
0
            }
2459
0
        }
2460
107
        else {
2461
            /* Generate ilim digits, then fix them up. */
2462
107
            dval(&eps) *= tens[ilim-1];
2463
288
            for(i = 1;; i++, dval(&u) *= 10.) {
2464
288
                L = (Long)(dval(&u));
2465
288
                if (!(dval(&u) -= L))
2466
17
                    ilim = i;
2467
288
                *s++ = '0' + (int)L;
2468
288
                if (i == ilim) {
2469
107
                    if (dval(&u) > 0.5 + dval(&eps))
2470
53
                        goto bump_up;
2471
54
                    else if (dval(&u) < 0.5 - dval(&eps)) {
2472
58
                        while(*--s == '0');
2473
54
                        s++;
2474
54
                        goto ret1;
2475
54
                    }
2476
0
                    break;
2477
107
                }
2478
288
            }
2479
107
        }
2480
0
      fast_failed:
2481
0
        s = s0;
2482
0
        dval(&u) = dval(&d2);
2483
0
        k = k0;
2484
0
        ilim = ilim0;
2485
0
    }
2486
2487
    /* Do we have a "small" integer? */
2488
2489
44.5k
    if (be >= 0 && k <= Int_max) {
2490
        /* Yes. */
2491
12.2k
        ds = tens[k];
2492
12.2k
        if (ndigits < 0 && ilim <= 0) {
2493
0
            S = mhi = 0;
2494
0
            if (ilim < 0 || dval(&u) <= 5*ds)
2495
0
                goto no_digits;
2496
0
            goto one_digit;
2497
0
        }
2498
18.5k
        for(i = 1;; i++, dval(&u) *= 10.) {
2499
18.5k
            L = (Long)(dval(&u) / ds);
2500
18.5k
            dval(&u) -= L*ds;
2501
18.5k
            *s++ = '0' + (int)L;
2502
18.5k
            if (!dval(&u)) {
2503
12.2k
                break;
2504
12.2k
            }
2505
6.25k
            if (i == ilim) {
2506
0
                dval(&u) += dval(&u);
2507
0
                if (dval(&u) > ds || (dval(&u) == ds && L & 1)) {
2508
53
                  bump_up:
2509
55
                    while(*--s == '9')
2510
2
                        if (s == s0) {
2511
0
                            k++;
2512
0
                            *s = '0';
2513
0
                            break;
2514
0
                        }
2515
53
                    ++*s++;
2516
53
                }
2517
0
                else {
2518
                    /* Strip trailing zeros. This branch was missing from the
2519
                       original dtoa.c, leading to surplus trailing zeros in
2520
                       some cases. See bugs.python.org/issue40780. */
2521
0
                    while (s > s0 && s[-1] == '0') {
2522
0
                        --s;
2523
0
                    }
2524
0
                }
2525
53
                break;
2526
0
            }
2527
6.25k
        }
2528
12.3k
        goto ret1;
2529
12.2k
    }
2530
2531
32.2k
    m2 = b2;
2532
32.2k
    m5 = b5;
2533
32.2k
    if (leftright) {
2534
32.2k
        i =
2535
32.2k
            denorm ? be + (Bias + (P-1) - 1 + 1) :
2536
32.2k
            1 + P - bbits;
2537
32.2k
        b2 += i;
2538
32.2k
        s2 += i;
2539
32.2k
        mhi = i2b(1);
2540
32.2k
        if (mhi == NULL)
2541
0
            goto failed_malloc;
2542
32.2k
    }
2543
32.2k
    if (m2 > 0 && s2 > 0) {
2544
27.8k
        i = m2 < s2 ? m2 : s2;
2545
27.8k
        b2 -= i;
2546
27.8k
        m2 -= i;
2547
27.8k
        s2 -= i;
2548
27.8k
    }
2549
32.2k
    if (b5 > 0) {
2550
13.5k
        if (leftright) {
2551
13.5k
            if (m5 > 0) {
2552
13.5k
                mhi = pow5mult(mhi, m5);
2553
13.5k
                if (mhi == NULL)
2554
0
                    goto failed_malloc;
2555
13.5k
                b1 = mult(mhi, b);
2556
13.5k
                Bfree(b);
2557
13.5k
                b = b1;
2558
13.5k
                if (b == NULL)
2559
0
                    goto failed_malloc;
2560
13.5k
            }
2561
13.5k
            if ((j = b5 - m5)) {
2562
0
                b = pow5mult(b, j);
2563
0
                if (b == NULL)
2564
0
                    goto failed_malloc;
2565
0
            }
2566
13.5k
        }
2567
0
        else {
2568
0
            b = pow5mult(b, b5);
2569
0
            if (b == NULL)
2570
0
                goto failed_malloc;
2571
0
        }
2572
13.5k
    }
2573
32.2k
    S = i2b(1);
2574
32.2k
    if (S == NULL)
2575
0
        goto failed_malloc;
2576
32.2k
    if (s5 > 0) {
2577
15.9k
        S = pow5mult(S, s5);
2578
15.9k
        if (S == NULL)
2579
0
            goto failed_malloc;
2580
15.9k
    }
2581
2582
    /* Check for special case that d is a normalized power of 2. */
2583
2584
32.2k
    spec_case = 0;
2585
32.2k
    if ((mode < 2 || leftright)
2586
32.2k
        ) {
2587
32.2k
        if (!word1(&u) && !(word0(&u) & Bndry_mask)
2588
1.38k
            && word0(&u) & (Exp_mask & ~Exp_msk1)
2589
32.2k
            ) {
2590
            /* The special case */
2591
1.12k
            b2 += Log2P;
2592
1.12k
            s2 += Log2P;
2593
1.12k
            spec_case = 1;
2594
1.12k
        }
2595
32.2k
    }
2596
2597
    /* Arrange for convenient computation of quotients:
2598
     * shift left if necessary so divisor has 4 leading 0 bits.
2599
     *
2600
     * Perhaps we should just compute leading 28 bits of S once
2601
     * and for all and pass them and a shift to quorem, so it
2602
     * can do shifts and ors to compute the numerator for q.
2603
     */
2604
32.2k
#define iInc 28
2605
32.2k
    i = dshift(S, s2);
2606
32.2k
    b2 += i;
2607
32.2k
    m2 += i;
2608
32.2k
    s2 += i;
2609
32.2k
    if (b2 > 0) {
2610
32.2k
        b = lshift(b, b2);
2611
32.2k
        if (b == NULL)
2612
0
            goto failed_malloc;
2613
32.2k
    }
2614
32.2k
    if (s2 > 0) {
2615
31.5k
        S = lshift(S, s2);
2616
31.5k
        if (S == NULL)
2617
0
            goto failed_malloc;
2618
31.5k
    }
2619
32.2k
    if (k_check) {
2620
23.9k
        if (cmp(b,S) < 0) {
2621
2.28k
            k--;
2622
2.28k
            b = multadd(b, 10, 0);      /* we botched the k estimate */
2623
2.28k
            if (b == NULL)
2624
0
                goto failed_malloc;
2625
2.28k
            if (leftright) {
2626
2.28k
                mhi = multadd(mhi, 10, 0);
2627
2.28k
                if (mhi == NULL)
2628
0
                    goto failed_malloc;
2629
2.28k
            }
2630
2.28k
            ilim = ilim1;
2631
2.28k
        }
2632
23.9k
    }
2633
32.2k
    if (ilim <= 0 && (mode == 3 || mode == 5)) {
2634
0
        if (ilim < 0) {
2635
            /* no digits, fcvt style */
2636
0
          no_digits:
2637
0
            k = -1 - ndigits;
2638
0
            goto ret;
2639
0
        }
2640
0
        else {
2641
0
            S = multadd(S, 5, 0);
2642
0
            if (S == NULL)
2643
0
                goto failed_malloc;
2644
0
            if (cmp(b, S) <= 0)
2645
0
                goto no_digits;
2646
0
        }
2647
0
      one_digit:
2648
0
        *s++ = '1';
2649
0
        k++;
2650
0
        goto ret;
2651
0
    }
2652
32.2k
    if (leftright) {
2653
32.2k
        if (m2 > 0) {
2654
31.4k
            mhi = lshift(mhi, m2);
2655
31.4k
            if (mhi == NULL)
2656
0
                goto failed_malloc;
2657
31.4k
        }
2658
2659
        /* Compute mlo -- check for special case
2660
         * that d is a normalized power of 2.
2661
         */
2662
2663
32.2k
        mlo = mhi;
2664
32.2k
        if (spec_case) {
2665
1.12k
            mhi = Balloc(mhi->k);
2666
1.12k
            if (mhi == NULL)
2667
0
                goto failed_malloc;
2668
1.12k
            Bcopy(mhi, mlo);
2669
1.12k
            mhi = lshift(mhi, Log2P);
2670
1.12k
            if (mhi == NULL)
2671
0
                goto failed_malloc;
2672
1.12k
        }
2673
2674
116k
        for(i = 1;;i++) {
2675
116k
            dig = quorem(b,S) + '0';
2676
            /* Do we yet have the shortest decimal string
2677
             * that will round to d?
2678
             */
2679
116k
            j = cmp(b, mlo);
2680
116k
            delta = diff(S, mhi);
2681
116k
            if (delta == NULL)
2682
0
                goto failed_malloc;
2683
116k
            j1 = delta->sign ? 1 : cmp(b, delta);
2684
116k
            Bfree(delta);
2685
116k
            if (j1 == 0 && mode != 1 && !(word1(&u) & 1)
2686
116k
                ) {
2687
1.88k
                if (dig == '9')
2688
396
                    goto round_9_up;
2689
1.49k
                if (j > 0)
2690
720
                    dig++;
2691
1.49k
                *s++ = dig;
2692
1.49k
                goto ret;
2693
1.88k
            }
2694
114k
            if (j < 0 || (j == 0 && mode != 1
2695
1.80k
                          && !(word1(&u) & 1)
2696
96.8k
                    )) {
2697
18.8k
                if (!b->x[0] && b->wds <= 1) {
2698
2.84k
                    goto accept_dig;
2699
2.84k
                }
2700
16.0k
                if (j1 > 0) {
2701
3.03k
                    b = lshift(b, 1);
2702
3.03k
                    if (b == NULL)
2703
0
                        goto failed_malloc;
2704
3.03k
                    j1 = cmp(b, S);
2705
3.03k
                    if ((j1 > 0 || (j1 == 0 && dig & 1))
2706
1.82k
                        && dig++ == '9')
2707
301
                        goto round_9_up;
2708
3.03k
                }
2709
18.5k
              accept_dig:
2710
18.5k
                *s++ = dig;
2711
18.5k
                goto ret;
2712
16.0k
            }
2713
96.0k
            if (j1 > 0) {
2714
11.4k
                if (dig == '9') { /* possible if i == 1 */
2715
2.22k
                  round_9_up:
2716
2.22k
                    *s++ = '9';
2717
2.22k
                    goto roundoff;
2718
1.52k
                }
2719
9.93k
                *s++ = dig + 1;
2720
9.93k
                goto ret;
2721
11.4k
            }
2722
84.5k
            *s++ = dig;
2723
84.5k
            if (i == ilim)
2724
0
                break;
2725
84.5k
            b = multadd(b, 10, 0);
2726
84.5k
            if (b == NULL)
2727
0
                goto failed_malloc;
2728
84.5k
            if (mlo == mhi) {
2729
81.6k
                mlo = mhi = multadd(mhi, 10, 0);
2730
81.6k
                if (mlo == NULL)
2731
0
                    goto failed_malloc;
2732
81.6k
            }
2733
2.91k
            else {
2734
2.91k
                mlo = multadd(mlo, 10, 0);
2735
2.91k
                if (mlo == NULL)
2736
0
                    goto failed_malloc;
2737
2.91k
                mhi = multadd(mhi, 10, 0);
2738
2.91k
                if (mhi == NULL)
2739
0
                    goto failed_malloc;
2740
2.91k
            }
2741
84.5k
        }
2742
32.2k
    }
2743
0
    else
2744
0
        for(i = 1;; i++) {
2745
0
            *s++ = dig = quorem(b,S) + '0';
2746
0
            if (!b->x[0] && b->wds <= 1) {
2747
0
                goto ret;
2748
0
            }
2749
0
            if (i >= ilim)
2750
0
                break;
2751
0
            b = multadd(b, 10, 0);
2752
0
            if (b == NULL)
2753
0
                goto failed_malloc;
2754
0
        }
2755
2756
    /* Round off last digit */
2757
2758
0
    b = lshift(b, 1);
2759
0
    if (b == NULL)
2760
0
        goto failed_malloc;
2761
0
    j = cmp(b, S);
2762
0
    if (j > 0 || (j == 0 && dig & 1)) {
2763
2.22k
      roundoff:
2764
2.22k
        while(*--s == '9')
2765
2.22k
            if (s == s0) {
2766
2.22k
                k++;
2767
2.22k
                *s++ = '1';
2768
2.22k
                goto ret;
2769
2.22k
            }
2770
0
        ++*s++;
2771
0
    }
2772
0
    else {
2773
0
        while(*--s == '0');
2774
0
        s++;
2775
0
    }
2776
32.2k
  ret:
2777
32.2k
    Bfree(S);
2778
32.2k
    if (mhi) {
2779
32.2k
        if (mlo && mlo != mhi)
2780
1.12k
            Bfree(mlo);
2781
32.2k
        Bfree(mhi);
2782
32.2k
    }
2783
44.6k
  ret1:
2784
44.6k
    Bfree(b);
2785
44.6k
    *s = 0;
2786
44.6k
    *decpt = k + 1;
2787
44.6k
    if (rve)
2788
44.6k
        *rve = s;
2789
44.6k
    return s0;
2790
0
  failed_malloc:
2791
0
    if (S)
2792
0
        Bfree(S);
2793
0
    if (mlo && mlo != mhi)
2794
0
        Bfree(mlo);
2795
0
    if (mhi)
2796
0
        Bfree(mhi);
2797
0
    if (b)
2798
0
        Bfree(b);
2799
0
    if (s0)
2800
0
        _Py_dg_freedtoa(s0);
2801
0
    return NULL;
2802
32.2k
}
2803
2804
#endif  // _PY_SHORT_FLOAT_REPR == 1
2805
2806
PyStatus
2807
_PyDtoa_Init(PyInterpreterState *interp)
2808
28
{
2809
28
#if _PY_SHORT_FLOAT_REPR == 1 && !defined(Py_USING_MEMORY_DEBUGGER)
2810
28
    Bigint **p5s = interp->dtoa.p5s;
2811
2812
    // 5**4 = 625
2813
28
    Bigint *p5 = i2b(625);
2814
28
    if (p5 == NULL) {
2815
0
        return PyStatus_NoMemory();
2816
0
    }
2817
28
    p5s[0] = p5;
2818
2819
    // compute 5**8, 5**16, 5**32, ..., 5**512
2820
224
    for (Py_ssize_t i = 1; i < Bigint_Pow5size; i++) {
2821
196
        p5 = mult(p5, p5);
2822
196
        if (p5 == NULL) {
2823
0
            return PyStatus_NoMemory();
2824
0
        }
2825
196
        p5s[i] = p5;
2826
196
    }
2827
2828
28
#endif
2829
28
    return PyStatus_Ok();
2830
28
}
2831
2832
void
2833
_PyDtoa_Fini(PyInterpreterState *interp)
2834
0
{
2835
0
#if _PY_SHORT_FLOAT_REPR == 1 && !defined(Py_USING_MEMORY_DEBUGGER)
2836
0
    Bigint **p5s = interp->dtoa.p5s;
2837
0
    for (Py_ssize_t i = 0; i < Bigint_Pow5size; i++) {
2838
0
        Bigint *p5 = p5s[i];
2839
        p5s[i] = NULL;
2840
0
        Bfree(p5);
2841
0
    }
2842
0
#endif
2843
0
}