This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
buildperl.pl: Add ability to build in parallel
[perl5.git] / numeric.c
... / ...
CommitLineData
1/* numeric.c
2 *
3 * Copyright (C) 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001,
4 * 2002, 2003, 2004, 2005, 2006, 2007, 2008 by Larry Wall and others
5 *
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8 *
9 */
10
11/*
12 * "That only makes eleven (plus one mislaid) and not fourteen,
13 * unless wizards count differently to other people." --Beorn
14 *
15 * [p.115 of _The Hobbit_: "Queer Lodgings"]
16 */
17
18/*
19=head1 Numeric functions
20
21=cut
22
23This file contains all the stuff needed by perl for manipulating numeric
24values, including such things as replacements for the OS's atof() function
25
26*/
27
28#include "EXTERN.h"
29#define PERL_IN_NUMERIC_C
30#include "perl.h"
31
32#ifdef Perl_strtod
33
34PERL_STATIC_INLINE NV
35S_strtod(pTHX_ const char * const s, char ** e)
36{
37 DECLARATION_FOR_LC_NUMERIC_MANIPULATION;
38 NV result;
39
40 STORE_LC_NUMERIC_SET_TO_NEEDED();
41
42# ifdef USE_QUADMATH
43
44 result = strtoflt128(s, e);
45
46# elif defined(HAS_STRTOLD) && defined(HAS_LONG_DOUBLE) \
47 && defined(USE_LONG_DOUBLE)
48# if defined(__MINGW64_VERSION_MAJOR)
49 /***********************************************
50 We are unable to use strtold because of
51 https://sourceforge.net/p/mingw-w64/bugs/711/
52 &
53 https://sourceforge.net/p/mingw-w64/bugs/725/
54
55 but __mingw_strtold is fine.
56 ***********************************************/
57
58 result = __mingw_strtold(s, e);
59
60# else
61
62 result = strtold(s, e);
63
64# endif
65# elif defined(HAS_STRTOD)
66
67 result = strtod(s, e);
68
69# else
70# error No strtod() equivalent found
71# endif
72
73 RESTORE_LC_NUMERIC();
74
75 return result;
76}
77
78#endif /* #ifdef Perl_strtod */
79
80/*
81
82=for apidoc my_strtod
83
84This function is equivalent to the libc strtod() function, and is available
85even on platforms that lack plain strtod(). Its return value is the best
86available precision depending on platform capabilities and F<Configure>
87options.
88
89It properly handles the locale radix character, meaning it expects a dot except
90when called from within the scope of S<C<use locale>>, in which case the radix
91character should be that specified by the current locale.
92
93The synonym Strtod() may be used instead.
94
95=cut
96
97*/
98
99NV
100Perl_my_strtod(const char * const s, char **e)
101{
102 dTHX;
103
104 PERL_ARGS_ASSERT_MY_STRTOD;
105
106#ifdef Perl_strtod
107
108 return S_strtod(aTHX_ s, e);
109
110#else
111
112 {
113 NV result;
114 char ** end_ptr = NULL;
115
116 *end_ptr = my_atof2(s, &result);
117 if (e) {
118 *e = *end_ptr;
119 }
120
121 if (! *end_ptr) {
122 result = 0.0;
123 }
124
125 return result;
126 }
127
128#endif
129
130}
131
132
133U32
134Perl_cast_ulong(NV f)
135{
136 if (f < 0.0)
137 return f < I32_MIN ? (U32) I32_MIN : (U32)(I32) f;
138 if (f < U32_MAX_P1) {
139#if CASTFLAGS & 2
140 if (f < U32_MAX_P1_HALF)
141 return (U32) f;
142 f -= U32_MAX_P1_HALF;
143 return ((U32) f) | (1 + (U32_MAX >> 1));
144#else
145 return (U32) f;
146#endif
147 }
148 return f > 0 ? U32_MAX : 0 /* NaN */;
149}
150
151I32
152Perl_cast_i32(NV f)
153{
154 if (f < I32_MAX_P1)
155 return f < I32_MIN ? I32_MIN : (I32) f;
156 if (f < U32_MAX_P1) {
157#if CASTFLAGS & 2
158 if (f < U32_MAX_P1_HALF)
159 return (I32)(U32) f;
160 f -= U32_MAX_P1_HALF;
161 return (I32)(((U32) f) | (1 + (U32_MAX >> 1)));
162#else
163 return (I32)(U32) f;
164#endif
165 }
166 return f > 0 ? (I32)U32_MAX : 0 /* NaN */;
167}
168
169IV
170Perl_cast_iv(NV f)
171{
172 if (f < IV_MAX_P1)
173 return f < IV_MIN ? IV_MIN : (IV) f;
174 if (f < UV_MAX_P1) {
175#if CASTFLAGS & 2
176 /* For future flexibility allowing for sizeof(UV) >= sizeof(IV) */
177 if (f < UV_MAX_P1_HALF)
178 return (IV)(UV) f;
179 f -= UV_MAX_P1_HALF;
180 return (IV)(((UV) f) | (1 + (UV_MAX >> 1)));
181#else
182 return (IV)(UV) f;
183#endif
184 }
185 return f > 0 ? (IV)UV_MAX : 0 /* NaN */;
186}
187
188UV
189Perl_cast_uv(NV f)
190{
191 if (f < 0.0)
192 return f < IV_MIN ? (UV) IV_MIN : (UV)(IV) f;
193 if (f < UV_MAX_P1) {
194#if CASTFLAGS & 2
195 if (f < UV_MAX_P1_HALF)
196 return (UV) f;
197 f -= UV_MAX_P1_HALF;
198 return ((UV) f) | (1 + (UV_MAX >> 1));
199#else
200 return (UV) f;
201#endif
202 }
203 return f > 0 ? UV_MAX : 0 /* NaN */;
204}
205
206/*
207=for apidoc grok_bin
208
209converts a string representing a binary number to numeric form.
210
211On entry C<start> and C<*len> give the string to scan, C<*flags> gives
212conversion flags, and C<result> should be C<NULL> or a pointer to an NV.
213The scan stops at the end of the string, or the first invalid character.
214Unless C<PERL_SCAN_SILENT_ILLDIGIT> is set in C<*flags>, encountering an
215invalid character will also trigger a warning.
216On return C<*len> is set to the length of the scanned string,
217and C<*flags> gives output flags.
218
219If the value is <= C<UV_MAX> it is returned as a UV, the output flags are clear,
220and nothing is written to C<*result>. If the value is > C<UV_MAX>, C<grok_bin>
221returns C<UV_MAX>, sets C<PERL_SCAN_GREATER_THAN_UV_MAX> in the output flags,
222and writes the value to C<*result> (or the value is discarded if C<result>
223is NULL).
224
225The binary number may optionally be prefixed with C<"0b"> or C<"b"> unless
226C<PERL_SCAN_DISALLOW_PREFIX> is set in C<*flags> on entry. If
227C<PERL_SCAN_ALLOW_UNDERSCORES> is set in C<*flags> then the binary
228number may use C<"_"> characters to separate digits.
229
230=for apidoc Amnh||PERL_SCAN_ALLOW_UNDERSCORES
231=for apidoc Amnh||PERL_SCAN_DISALLOW_PREFIX
232=for apidoc Amnh||PERL_SCAN_GREATER_THAN_UV_MAX
233=for apidoc Amnh||PERL_SCAN_SILENT_ILLDIGIT
234=for apidoc Amnh||PERL_SCAN_TRAILING
235
236=cut
237
238Not documented yet because experimental is C<PERL_SCAN_SILENT_NON_PORTABLE
239which suppresses any message for non-portable numbers that are still valid
240on this platform.
241 */
242
243UV
244Perl_grok_bin(pTHX_ const char *start, STRLEN *len_p, I32 *flags, NV *result)
245{
246 const char *s = start;
247 STRLEN len = *len_p;
248 UV value = 0;
249 NV value_nv = 0;
250
251 const UV max_div_2 = UV_MAX / 2;
252 const bool allow_underscores = cBOOL(*flags & PERL_SCAN_ALLOW_UNDERSCORES);
253 bool overflowed = FALSE;
254 char bit;
255
256 PERL_ARGS_ASSERT_GROK_BIN;
257
258 if (!(*flags & PERL_SCAN_DISALLOW_PREFIX)) {
259 /* strip off leading b or 0b.
260 for compatibility silently suffer "b" and "0b" as valid binary
261 numbers. */
262 if (len >= 1) {
263 if (isALPHA_FOLD_EQ(s[0], 'b')) {
264 s++;
265 len--;
266 }
267 else if (len >= 2 && s[0] == '0' && (isALPHA_FOLD_EQ(s[1], 'b'))) {
268 s+=2;
269 len-=2;
270 }
271 }
272 }
273
274 for (; len-- && (bit = *s); s++) {
275 if (bit == '0' || bit == '1') {
276 /* Write it in this wonky order with a goto to attempt to get the
277 compiler to make the common case integer-only loop pretty tight.
278 With gcc seems to be much straighter code than old scan_bin. */
279 redo:
280 if (!overflowed) {
281 if (value <= max_div_2) {
282 value = (value << 1) | (bit - '0');
283 continue;
284 }
285 /* Bah. We're just overflowed. */
286 /* diag_listed_as: Integer overflow in %s number */
287 Perl_ck_warner_d(aTHX_ packWARN(WARN_OVERFLOW),
288 "Integer overflow in binary number");
289 overflowed = TRUE;
290 value_nv = (NV) value;
291 }
292 value_nv *= 2.0;
293 /* If an NV has not enough bits in its mantissa to
294 * represent a UV this summing of small low-order numbers
295 * is a waste of time (because the NV cannot preserve
296 * the low-order bits anyway): we could just remember when
297 * did we overflow and in the end just multiply value_nv by the
298 * right amount. */
299 value_nv += (NV)(bit - '0');
300 continue;
301 }
302 if (bit == '_' && len && allow_underscores && (bit = s[1])
303 && (bit == '0' || bit == '1'))
304 {
305 --len;
306 ++s;
307 goto redo;
308 }
309 if (!(*flags & PERL_SCAN_SILENT_ILLDIGIT))
310 Perl_ck_warner(aTHX_ packWARN(WARN_DIGIT),
311 "Illegal binary digit '%c' ignored", *s);
312 break;
313 }
314
315 if ( ( overflowed && value_nv > 4294967295.0)
316#if UVSIZE > 4
317 || (!overflowed && value > 0xffffffff
318 && ! (*flags & PERL_SCAN_SILENT_NON_PORTABLE))
319#endif
320 ) {
321 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
322 "Binary number > 0b11111111111111111111111111111111 non-portable");
323 }
324 *len_p = s - start;
325 if (!overflowed) {
326 *flags = 0;
327 return value;
328 }
329 *flags = PERL_SCAN_GREATER_THAN_UV_MAX;
330 if (result)
331 *result = value_nv;
332 return UV_MAX;
333}
334
335/*
336=for apidoc grok_hex
337
338converts a string representing a hex number to numeric form.
339
340On entry C<start> and C<*len_p> give the string to scan, C<*flags> gives
341conversion flags, and C<result> should be C<NULL> or a pointer to an NV.
342The scan stops at the end of the string, or the first invalid character.
343Unless C<PERL_SCAN_SILENT_ILLDIGIT> is set in C<*flags>, encountering an
344invalid character will also trigger a warning.
345On return C<*len> is set to the length of the scanned string,
346and C<*flags> gives output flags.
347
348If the value is <= C<UV_MAX> it is returned as a UV, the output flags are clear,
349and nothing is written to C<*result>. If the value is > C<UV_MAX>, C<grok_hex>
350returns C<UV_MAX>, sets C<PERL_SCAN_GREATER_THAN_UV_MAX> in the output flags,
351and writes the value to C<*result> (or the value is discarded if C<result>
352is C<NULL>).
353
354The hex number may optionally be prefixed with C<"0x"> or C<"x"> unless
355C<PERL_SCAN_DISALLOW_PREFIX> is set in C<*flags> on entry. If
356C<PERL_SCAN_ALLOW_UNDERSCORES> is set in C<*flags> then the hex
357number may use C<"_"> characters to separate digits.
358
359=cut
360
361Not documented yet because experimental is C<PERL_SCAN_SILENT_NON_PORTABLE
362which suppresses any message for non-portable numbers, but which are valid
363on this platform.
364 */
365
366UV
367Perl_grok_hex(pTHX_ const char *start, STRLEN *len_p, I32 *flags, NV *result)
368{
369 const char *s = start;
370 STRLEN len = *len_p;
371 UV value = 0;
372 NV value_nv = 0;
373 const UV max_div_16 = UV_MAX / 16;
374 const bool allow_underscores = cBOOL(*flags & PERL_SCAN_ALLOW_UNDERSCORES);
375 bool overflowed = FALSE;
376
377 PERL_ARGS_ASSERT_GROK_HEX;
378
379 if (!(*flags & PERL_SCAN_DISALLOW_PREFIX)) {
380 /* strip off leading x or 0x.
381 for compatibility silently suffer "x" and "0x" as valid hex numbers.
382 */
383 if (len >= 1) {
384 if (isALPHA_FOLD_EQ(s[0], 'x')) {
385 s++;
386 len--;
387 }
388 else if (len >= 2 && s[0] == '0' && (isALPHA_FOLD_EQ(s[1], 'x'))) {
389 s+=2;
390 len-=2;
391 }
392 }
393 }
394
395 for (; len-- && *s; s++) {
396 if (isXDIGIT(*s)) {
397 /* Write it in this wonky order with a goto to attempt to get the
398 compiler to make the common case integer-only loop pretty tight.
399 With gcc seems to be much straighter code than old scan_hex. */
400 redo:
401 if (!overflowed) {
402 if (value <= max_div_16) {
403 value = (value << 4) | XDIGIT_VALUE(*s);
404 continue;
405 }
406 /* Bah. We're just overflowed. */
407 /* diag_listed_as: Integer overflow in %s number */
408 Perl_ck_warner_d(aTHX_ packWARN(WARN_OVERFLOW),
409 "Integer overflow in hexadecimal number");
410 overflowed = TRUE;
411 value_nv = (NV) value;
412 }
413 value_nv *= 16.0;
414 /* If an NV has not enough bits in its mantissa to
415 * represent a UV this summing of small low-order numbers
416 * is a waste of time (because the NV cannot preserve
417 * the low-order bits anyway): we could just remember when
418 * did we overflow and in the end just multiply value_nv by the
419 * right amount of 16-tuples. */
420 value_nv += (NV) XDIGIT_VALUE(*s);
421 continue;
422 }
423 if (*s == '_' && len && allow_underscores && s[1]
424 && isXDIGIT(s[1]))
425 {
426 --len;
427 ++s;
428 goto redo;
429 }
430 if (!(*flags & PERL_SCAN_SILENT_ILLDIGIT))
431 Perl_ck_warner(aTHX_ packWARN(WARN_DIGIT),
432 "Illegal hexadecimal digit '%c' ignored", *s);
433 break;
434 }
435
436 if ( ( overflowed && value_nv > 4294967295.0)
437#if UVSIZE > 4
438 || (!overflowed && value > 0xffffffff
439 && ! (*flags & PERL_SCAN_SILENT_NON_PORTABLE))
440#endif
441 ) {
442 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
443 "Hexadecimal number > 0xffffffff non-portable");
444 }
445 *len_p = s - start;
446 if (!overflowed) {
447 *flags = 0;
448 return value;
449 }
450 *flags = PERL_SCAN_GREATER_THAN_UV_MAX;
451 if (result)
452 *result = value_nv;
453 return UV_MAX;
454}
455
456/*
457=for apidoc grok_oct
458
459converts a string representing an octal number to numeric form.
460
461On entry C<start> and C<*len> give the string to scan, C<*flags> gives
462conversion flags, and C<result> should be C<NULL> or a pointer to an NV.
463The scan stops at the end of the string, or the first invalid character.
464Unless C<PERL_SCAN_SILENT_ILLDIGIT> is set in C<*flags>, encountering an
4658 or 9 will also trigger a warning.
466On return C<*len> is set to the length of the scanned string,
467and C<*flags> gives output flags.
468
469If the value is <= C<UV_MAX> it is returned as a UV, the output flags are clear,
470and nothing is written to C<*result>. If the value is > C<UV_MAX>, C<grok_oct>
471returns C<UV_MAX>, sets C<PERL_SCAN_GREATER_THAN_UV_MAX> in the output flags,
472and writes the value to C<*result> (or the value is discarded if C<result>
473is C<NULL>).
474
475If C<PERL_SCAN_ALLOW_UNDERSCORES> is set in C<*flags> then the octal
476number may use C<"_"> characters to separate digits.
477
478=cut
479
480Not documented yet because experimental is C<PERL_SCAN_SILENT_NON_PORTABLE>
481which suppresses any message for non-portable numbers, but which are valid
482on this platform.
483 */
484
485UV
486Perl_grok_oct(pTHX_ const char *start, STRLEN *len_p, I32 *flags, NV *result)
487{
488 const char *s = start;
489 STRLEN len = *len_p;
490 UV value = 0;
491 NV value_nv = 0;
492 const UV max_div_8 = UV_MAX / 8;
493 const bool allow_underscores = cBOOL(*flags & PERL_SCAN_ALLOW_UNDERSCORES);
494 bool overflowed = FALSE;
495
496 PERL_ARGS_ASSERT_GROK_OCT;
497
498 for (; len-- && *s; s++) {
499 if (isOCTAL(*s)) {
500 /* Write it in this wonky order with a goto to attempt to get the
501 compiler to make the common case integer-only loop pretty tight.
502 */
503 redo:
504 if (!overflowed) {
505 if (value <= max_div_8) {
506 value = (value << 3) | OCTAL_VALUE(*s);
507 continue;
508 }
509 /* Bah. We're just overflowed. */
510 /* diag_listed_as: Integer overflow in %s number */
511 Perl_ck_warner_d(aTHX_ packWARN(WARN_OVERFLOW),
512 "Integer overflow in octal number");
513 overflowed = TRUE;
514 value_nv = (NV) value;
515 }
516 value_nv *= 8.0;
517 /* If an NV has not enough bits in its mantissa to
518 * represent a UV this summing of small low-order numbers
519 * is a waste of time (because the NV cannot preserve
520 * the low-order bits anyway): we could just remember when
521 * did we overflow and in the end just multiply value_nv by the
522 * right amount of 8-tuples. */
523 value_nv += (NV) OCTAL_VALUE(*s);
524 continue;
525 }
526 if (*s == '_' && len && allow_underscores && isOCTAL(s[1])) {
527 --len;
528 ++s;
529 goto redo;
530 }
531 /* Allow \octal to work the DWIM way (that is, stop scanning
532 * as soon as non-octal characters are seen, complain only if
533 * someone seems to want to use the digits eight and nine. Since we
534 * know it is not octal, then if isDIGIT, must be an 8 or 9). */
535 if (isDIGIT(*s)) {
536 if (!(*flags & PERL_SCAN_SILENT_ILLDIGIT))
537 Perl_ck_warner(aTHX_ packWARN(WARN_DIGIT),
538 "Illegal octal digit '%c' ignored", *s);
539 }
540 break;
541 }
542
543 if ( ( overflowed && value_nv > 4294967295.0)
544#if UVSIZE > 4
545 || (!overflowed && value > 0xffffffff
546 && ! (*flags & PERL_SCAN_SILENT_NON_PORTABLE))
547#endif
548 ) {
549 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
550 "Octal number > 037777777777 non-portable");
551 }
552 *len_p = s - start;
553 if (!overflowed) {
554 *flags = 0;
555 return value;
556 }
557 *flags = PERL_SCAN_GREATER_THAN_UV_MAX;
558 if (result)
559 *result = value_nv;
560 return UV_MAX;
561}
562
563/*
564=for apidoc scan_bin
565
566For backwards compatibility. Use C<grok_bin> instead.
567
568=for apidoc scan_hex
569
570For backwards compatibility. Use C<grok_hex> instead.
571
572=for apidoc scan_oct
573
574For backwards compatibility. Use C<grok_oct> instead.
575
576=cut
577 */
578
579NV
580Perl_scan_bin(pTHX_ const char *start, STRLEN len, STRLEN *retlen)
581{
582 NV rnv;
583 I32 flags = *retlen ? PERL_SCAN_ALLOW_UNDERSCORES : 0;
584 const UV ruv = grok_bin (start, &len, &flags, &rnv);
585
586 PERL_ARGS_ASSERT_SCAN_BIN;
587
588 *retlen = len;
589 return (flags & PERL_SCAN_GREATER_THAN_UV_MAX) ? rnv : (NV)ruv;
590}
591
592NV
593Perl_scan_oct(pTHX_ const char *start, STRLEN len, STRLEN *retlen)
594{
595 NV rnv;
596 I32 flags = *retlen ? PERL_SCAN_ALLOW_UNDERSCORES : 0;
597 const UV ruv = grok_oct (start, &len, &flags, &rnv);
598
599 PERL_ARGS_ASSERT_SCAN_OCT;
600
601 *retlen = len;
602 return (flags & PERL_SCAN_GREATER_THAN_UV_MAX) ? rnv : (NV)ruv;
603}
604
605NV
606Perl_scan_hex(pTHX_ const char *start, STRLEN len, STRLEN *retlen)
607{
608 NV rnv;
609 I32 flags = *retlen ? PERL_SCAN_ALLOW_UNDERSCORES : 0;
610 const UV ruv = grok_hex (start, &len, &flags, &rnv);
611
612 PERL_ARGS_ASSERT_SCAN_HEX;
613
614 *retlen = len;
615 return (flags & PERL_SCAN_GREATER_THAN_UV_MAX) ? rnv : (NV)ruv;
616}
617
618/*
619=for apidoc grok_numeric_radix
620
621Scan and skip for a numeric decimal separator (radix).
622
623=cut
624 */
625bool
626Perl_grok_numeric_radix(pTHX_ const char **sp, const char *send)
627{
628 PERL_ARGS_ASSERT_GROK_NUMERIC_RADIX;
629
630#ifdef USE_LOCALE_NUMERIC
631
632 if (IN_LC(LC_NUMERIC)) {
633 STRLEN len;
634 char * radix;
635 bool matches_radix = FALSE;
636 DECLARATION_FOR_LC_NUMERIC_MANIPULATION;
637
638 STORE_LC_NUMERIC_FORCE_TO_UNDERLYING();
639
640 radix = SvPV(PL_numeric_radix_sv, len);
641 radix = savepvn(radix, len);
642
643 RESTORE_LC_NUMERIC();
644
645 if (*sp + len <= send) {
646 matches_radix = memEQ(*sp, radix, len);
647 }
648
649 Safefree(radix);
650
651 if (matches_radix) {
652 *sp += len;
653 return TRUE;
654 }
655 }
656
657#endif
658
659 /* always try "." if numeric radix didn't match because
660 * we may have data from different locales mixed */
661 if (*sp < send && **sp == '.') {
662 ++*sp;
663 return TRUE;
664 }
665
666 return FALSE;
667}
668
669/*
670=for apidoc grok_infnan
671
672Helper for C<grok_number()>, accepts various ways of spelling "infinity"
673or "not a number", and returns one of the following flag combinations:
674
675 IS_NUMBER_INFINITY
676 IS_NUMBER_NAN
677 IS_NUMBER_INFINITY | IS_NUMBER_NEG
678 IS_NUMBER_NAN | IS_NUMBER_NEG
679 0
680
681possibly |-ed with C<IS_NUMBER_TRAILING>.
682
683If an infinity or a not-a-number is recognized, C<*sp> will point to
684one byte past the end of the recognized string. If the recognition fails,
685zero is returned, and C<*sp> will not move.
686
687=for apidoc Amn|bool|IS_NUMBER_GREATER_THAN_UV_MAX
688=for apidoc Amn|bool|IS_NUMBER_INFINITY
689=for apidoc Amn|bool|IS_NUMBER_IN_UV
690=for apidoc Amn|bool|IS_NUMBER_NAN
691=for apidoc Amn|bool|IS_NUMBER_NEG
692=for apidoc Amn|bool|IS_NUMBER_NOT_INT
693
694=cut
695*/
696
697int
698Perl_grok_infnan(pTHX_ const char** sp, const char* send)
699{
700 const char* s = *sp;
701 int flags = 0;
702#if defined(NV_INF) || defined(NV_NAN)
703 bool odh = FALSE; /* one-dot-hash: 1.#INF */
704
705 PERL_ARGS_ASSERT_GROK_INFNAN;
706
707 if (*s == '+') {
708 s++; if (s == send) return 0;
709 }
710 else if (*s == '-') {
711 flags |= IS_NUMBER_NEG; /* Yes, -NaN happens. Incorrect but happens. */
712 s++; if (s == send) return 0;
713 }
714
715 if (*s == '1') {
716 /* Visual C: 1.#SNAN, -1.#QNAN, 1#INF, 1.#IND (maybe also 1.#NAN)
717 * Let's keep the dot optional. */
718 s++; if (s == send) return 0;
719 if (*s == '.') {
720 s++; if (s == send) return 0;
721 }
722 if (*s == '#') {
723 s++; if (s == send) return 0;
724 } else
725 return 0;
726 odh = TRUE;
727 }
728
729 if (isALPHA_FOLD_EQ(*s, 'I')) {
730 /* INF or IND (1.#IND is "indeterminate", a certain type of NAN) */
731
732 s++; if (s == send || isALPHA_FOLD_NE(*s, 'N')) return 0;
733 s++; if (s == send) return 0;
734 if (isALPHA_FOLD_EQ(*s, 'F')) {
735 s++;
736 if (s < send && (isALPHA_FOLD_EQ(*s, 'I'))) {
737 int fail =
738 flags | IS_NUMBER_INFINITY | IS_NUMBER_NOT_INT | IS_NUMBER_TRAILING;
739 s++; if (s == send || isALPHA_FOLD_NE(*s, 'N')) return fail;
740 s++; if (s == send || isALPHA_FOLD_NE(*s, 'I')) return fail;
741 s++; if (s == send || isALPHA_FOLD_NE(*s, 'T')) return fail;
742 s++; if (s == send || isALPHA_FOLD_NE(*s, 'Y')) return fail;
743 s++;
744 } else if (odh) {
745 while (*s == '0') { /* 1.#INF00 */
746 s++;
747 }
748 }
749 while (s < send && isSPACE(*s))
750 s++;
751 if (s < send && *s) {
752 flags |= IS_NUMBER_TRAILING;
753 }
754 flags |= IS_NUMBER_INFINITY | IS_NUMBER_NOT_INT;
755 }
756 else if (isALPHA_FOLD_EQ(*s, 'D') && odh) { /* 1.#IND */
757 s++;
758 flags |= IS_NUMBER_NAN | IS_NUMBER_NOT_INT;
759 while (*s == '0') { /* 1.#IND00 */
760 s++;
761 }
762 if (*s) {
763 flags |= IS_NUMBER_TRAILING;
764 }
765 } else
766 return 0;
767 }
768 else {
769 /* Maybe NAN of some sort */
770
771 if (isALPHA_FOLD_EQ(*s, 'S') || isALPHA_FOLD_EQ(*s, 'Q')) {
772 /* snan, qNaN */
773 /* XXX do something with the snan/qnan difference */
774 s++; if (s == send) return 0;
775 }
776
777 if (isALPHA_FOLD_EQ(*s, 'N')) {
778 s++; if (s == send || isALPHA_FOLD_NE(*s, 'A')) return 0;
779 s++; if (s == send || isALPHA_FOLD_NE(*s, 'N')) return 0;
780 s++;
781
782 flags |= IS_NUMBER_NAN | IS_NUMBER_NOT_INT;
783
784 /* NaN can be followed by various stuff (NaNQ, NaNS), but
785 * there are also multiple different NaN values, and some
786 * implementations output the "payload" values,
787 * e.g. NaN123, NAN(abc), while some legacy implementations
788 * have weird stuff like NaN%. */
789 if (isALPHA_FOLD_EQ(*s, 'q') ||
790 isALPHA_FOLD_EQ(*s, 's')) {
791 /* "nanq" or "nans" are ok, though generating
792 * these portably is tricky. */
793 s++;
794 }
795 if (*s == '(') {
796 /* C99 style "nan(123)" or Perlish equivalent "nan($uv)". */
797 const char *t;
798 s++;
799 if (s == send) {
800 return flags | IS_NUMBER_TRAILING;
801 }
802 t = s + 1;
803 while (t < send && *t && *t != ')') {
804 t++;
805 }
806 if (t == send) {
807 return flags | IS_NUMBER_TRAILING;
808 }
809 if (*t == ')') {
810 int nantype;
811 UV nanval;
812 if (s[0] == '0' && s + 2 < t &&
813 isALPHA_FOLD_EQ(s[1], 'x') &&
814 isXDIGIT(s[2])) {
815 STRLEN len = t - s;
816 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES;
817 nanval = grok_hex(s, &len, &flags, NULL);
818 if ((flags & PERL_SCAN_GREATER_THAN_UV_MAX)) {
819 nantype = 0;
820 } else {
821 nantype = IS_NUMBER_IN_UV;
822 }
823 s += len;
824 } else if (s[0] == '0' && s + 2 < t &&
825 isALPHA_FOLD_EQ(s[1], 'b') &&
826 (s[2] == '0' || s[2] == '1')) {
827 STRLEN len = t - s;
828 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES;
829 nanval = grok_bin(s, &len, &flags, NULL);
830 if ((flags & PERL_SCAN_GREATER_THAN_UV_MAX)) {
831 nantype = 0;
832 } else {
833 nantype = IS_NUMBER_IN_UV;
834 }
835 s += len;
836 } else {
837 const char *u;
838 nantype =
839 grok_number_flags(s, t - s, &nanval,
840 PERL_SCAN_TRAILING |
841 PERL_SCAN_ALLOW_UNDERSCORES);
842 /* Unfortunately grok_number_flags() doesn't
843 * tell how far we got and the ')' will always
844 * be "trailing", so we need to double-check
845 * whether we had something dubious. */
846 for (u = s; u < t; u++) {
847 if (!isDIGIT(*u)) {
848 flags |= IS_NUMBER_TRAILING;
849 break;
850 }
851 }
852 s = u;
853 }
854
855 /* XXX Doesn't do octal: nan("0123").
856 * Probably not a big loss. */
857
858 if ((nantype & IS_NUMBER_NOT_INT) ||
859 !(nantype && IS_NUMBER_IN_UV)) {
860 /* XXX the nanval is currently unused, that is,
861 * not inserted as the NaN payload of the NV.
862 * But the above code already parses the C99
863 * nan(...) format. See below, and see also
864 * the nan() in POSIX.xs.
865 *
866 * Certain configuration combinations where
867 * NVSIZE is greater than UVSIZE mean that
868 * a single UV cannot contain all the possible
869 * NaN payload bits. There would need to be
870 * some more generic syntax than "nan($uv)".
871 *
872 * Issues to keep in mind:
873 *
874 * (1) In most common cases there would
875 * not be an integral number of bytes that
876 * could be set, only a certain number of bits.
877 * For example for the common case of
878 * NVSIZE == UVSIZE == 8 there is room for 52
879 * bits in the payload, but the most significant
880 * bit is commonly reserved for the
881 * signaling/quiet bit, leaving 51 bits.
882 * Furthermore, the C99 nan() is supposed
883 * to generate quiet NaNs, so it is doubtful
884 * whether it should be able to generate
885 * signaling NaNs. For the x86 80-bit doubles
886 * (if building a long double Perl) there would
887 * be 62 bits (s/q bit being the 63rd).
888 *
889 * (2) Endianness of the payload bits. If the
890 * payload is specified as an UV, the low-order
891 * bits of the UV are naturally little-endianed
892 * (rightmost) bits of the payload. The endianness
893 * of UVs and NVs can be different. */
894 return 0;
895 }
896 if (s < t) {
897 flags |= IS_NUMBER_TRAILING;
898 }
899 } else {
900 /* Looked like nan(...), but no close paren. */
901 flags |= IS_NUMBER_TRAILING;
902 }
903 } else {
904 while (s < send && isSPACE(*s))
905 s++;
906 if (s < send && *s) {
907 /* Note that we here implicitly accept (parse as
908 * "nan", but with warnings) also any other weird
909 * trailing stuff for "nan". In the above we just
910 * check that if we got the C99-style "nan(...)",
911 * the "..." looks sane.
912 * If in future we accept more ways of specifying
913 * the nan payload, the accepting would happen around
914 * here. */
915 flags |= IS_NUMBER_TRAILING;
916 }
917 }
918 s = send;
919 }
920 else
921 return 0;
922 }
923
924 while (s < send && isSPACE(*s))
925 s++;
926
927#else
928 PERL_UNUSED_ARG(send);
929#endif /* #if defined(NV_INF) || defined(NV_NAN) */
930 *sp = s;
931 return flags;
932}
933
934/*
935=for apidoc grok_number_flags
936
937Recognise (or not) a number. The type of the number is returned
938(0 if unrecognised), otherwise it is a bit-ORed combination of
939C<IS_NUMBER_IN_UV>, C<IS_NUMBER_GREATER_THAN_UV_MAX>, C<IS_NUMBER_NOT_INT>,
940C<IS_NUMBER_NEG>, C<IS_NUMBER_INFINITY>, C<IS_NUMBER_NAN> (defined in perl.h).
941
942If the value of the number can fit in a UV, it is returned in C<*valuep>.
943C<IS_NUMBER_IN_UV> will be set to indicate that C<*valuep> is valid, C<IS_NUMBER_IN_UV>
944will never be set unless C<*valuep> is valid, but C<*valuep> may have been assigned
945to during processing even though C<IS_NUMBER_IN_UV> is not set on return.
946If C<valuep> is C<NULL>, C<IS_NUMBER_IN_UV> will be set for the same cases as when
947C<valuep> is non-C<NULL>, but no actual assignment (or SEGV) will occur.
948
949C<IS_NUMBER_NOT_INT> will be set with C<IS_NUMBER_IN_UV> if trailing decimals were
950seen (in which case C<*valuep> gives the true value truncated to an integer), and
951C<IS_NUMBER_NEG> if the number is negative (in which case C<*valuep> holds the
952absolute value). C<IS_NUMBER_IN_UV> is not set if e notation was used or the
953number is larger than a UV.
954
955C<flags> allows only C<PERL_SCAN_TRAILING>, which allows for trailing
956non-numeric text on an otherwise successful I<grok>, setting
957C<IS_NUMBER_TRAILING> on the result.
958
959=for apidoc grok_number
960
961Identical to C<grok_number_flags()> with C<flags> set to zero.
962
963=cut
964 */
965int
966Perl_grok_number(pTHX_ const char *pv, STRLEN len, UV *valuep)
967{
968 PERL_ARGS_ASSERT_GROK_NUMBER;
969
970 return grok_number_flags(pv, len, valuep, 0);
971}
972
973static const UV uv_max_div_10 = UV_MAX / 10;
974static const U8 uv_max_mod_10 = UV_MAX % 10;
975
976int
977Perl_grok_number_flags(pTHX_ const char *pv, STRLEN len, UV *valuep, U32 flags)
978{
979 const char *s = pv;
980 const char * const send = pv + len;
981 const char *d;
982 int numtype = 0;
983
984 PERL_ARGS_ASSERT_GROK_NUMBER_FLAGS;
985
986 while (s < send && isSPACE(*s))
987 s++;
988 if (s == send) {
989 return 0;
990 } else if (*s == '-') {
991 s++;
992 numtype = IS_NUMBER_NEG;
993 }
994 else if (*s == '+')
995 s++;
996
997 if (s == send)
998 return 0;
999
1000 /* The first digit (after optional sign): note that might
1001 * also point to "infinity" or "nan", or "1.#INF". */
1002 d = s;
1003
1004 /* next must be digit or the radix separator or beginning of infinity/nan */
1005 if (isDIGIT(*s)) {
1006 /* UVs are at least 32 bits, so the first 9 decimal digits cannot
1007 overflow. */
1008 UV value = *s - '0';
1009 /* This construction seems to be more optimiser friendly.
1010 (without it gcc does the isDIGIT test and the *s - '0' separately)
1011 With it gcc on arm is managing 6 instructions (6 cycles) per digit.
1012 In theory the optimiser could deduce how far to unroll the loop
1013 before checking for overflow. */
1014 if (++s < send) {
1015 int digit = *s - '0';
1016 if (inRANGE(digit, 0, 9)) {
1017 value = value * 10 + digit;
1018 if (++s < send) {
1019 digit = *s - '0';
1020 if (inRANGE(digit, 0, 9)) {
1021 value = value * 10 + digit;
1022 if (++s < send) {
1023 digit = *s - '0';
1024 if (inRANGE(digit, 0, 9)) {
1025 value = value * 10 + digit;
1026 if (++s < send) {
1027 digit = *s - '0';
1028 if (inRANGE(digit, 0, 9)) {
1029 value = value * 10 + digit;
1030 if (++s < send) {
1031 digit = *s - '0';
1032 if (inRANGE(digit, 0, 9)) {
1033 value = value * 10 + digit;
1034 if (++s < send) {
1035 digit = *s - '0';
1036 if (inRANGE(digit, 0, 9)) {
1037 value = value * 10 + digit;
1038 if (++s < send) {
1039 digit = *s - '0';
1040 if (inRANGE(digit, 0, 9)) {
1041 value = value * 10 + digit;
1042 if (++s < send) {
1043 digit = *s - '0';
1044 if (inRANGE(digit, 0, 9)) {
1045 value = value * 10 + digit;
1046 if (++s < send) {
1047 /* Now got 9 digits, so need to check
1048 each time for overflow. */
1049 digit = *s - '0';
1050 while ( inRANGE(digit, 0, 9)
1051 && (value < uv_max_div_10
1052 || (value == uv_max_div_10
1053 && digit <= uv_max_mod_10))) {
1054 value = value * 10 + digit;
1055 if (++s < send)
1056 digit = *s - '0';
1057 else
1058 break;
1059 }
1060 if (inRANGE(digit, 0, 9)
1061 && (s < send)) {
1062 /* value overflowed.
1063 skip the remaining digits, don't
1064 worry about setting *valuep. */
1065 do {
1066 s++;
1067 } while (s < send && isDIGIT(*s));
1068 numtype |=
1069 IS_NUMBER_GREATER_THAN_UV_MAX;
1070 goto skip_value;
1071 }
1072 }
1073 }
1074 }
1075 }
1076 }
1077 }
1078 }
1079 }
1080 }
1081 }
1082 }
1083 }
1084 }
1085 }
1086 }
1087 }
1088 }
1089 numtype |= IS_NUMBER_IN_UV;
1090 if (valuep)
1091 *valuep = value;
1092
1093 skip_value:
1094 if (GROK_NUMERIC_RADIX(&s, send)) {
1095 numtype |= IS_NUMBER_NOT_INT;
1096 while (s < send && isDIGIT(*s)) /* optional digits after the radix */
1097 s++;
1098 }
1099 }
1100 else if (GROK_NUMERIC_RADIX(&s, send)) {
1101 numtype |= IS_NUMBER_NOT_INT | IS_NUMBER_IN_UV; /* valuep assigned below */
1102 /* no digits before the radix means we need digits after it */
1103 if (s < send && isDIGIT(*s)) {
1104 do {
1105 s++;
1106 } while (s < send && isDIGIT(*s));
1107 if (valuep) {
1108 /* integer approximation is valid - it's 0. */
1109 *valuep = 0;
1110 }
1111 }
1112 else
1113 return 0;
1114 }
1115
1116 if (s > d && s < send) {
1117 /* we can have an optional exponent part */
1118 if (isALPHA_FOLD_EQ(*s, 'e')) {
1119 s++;
1120 if (s < send && (*s == '-' || *s == '+'))
1121 s++;
1122 if (s < send && isDIGIT(*s)) {
1123 do {
1124 s++;
1125 } while (s < send && isDIGIT(*s));
1126 }
1127 else if (flags & PERL_SCAN_TRAILING)
1128 return numtype | IS_NUMBER_TRAILING;
1129 else
1130 return 0;
1131
1132 /* The only flag we keep is sign. Blow away any "it's UV" */
1133 numtype &= IS_NUMBER_NEG;
1134 numtype |= IS_NUMBER_NOT_INT;
1135 }
1136 }
1137 while (s < send && isSPACE(*s))
1138 s++;
1139 if (s >= send)
1140 return numtype;
1141 if (memEQs(pv, len, "0 but true")) {
1142 if (valuep)
1143 *valuep = 0;
1144 return IS_NUMBER_IN_UV;
1145 }
1146 /* We could be e.g. at "Inf" or "NaN", or at the "#" of "1.#INF". */
1147 if ((s + 2 < send) && strchr("inqs#", toFOLD(*s))) {
1148 /* Really detect inf/nan. Start at d, not s, since the above
1149 * code might have already consumed the "1." or "1". */
1150 const int infnan = Perl_grok_infnan(aTHX_ &d, send);
1151 if ((infnan & IS_NUMBER_INFINITY)) {
1152 return (numtype | infnan); /* Keep sign for infinity. */
1153 }
1154 else if ((infnan & IS_NUMBER_NAN)) {
1155 return (numtype | infnan) & ~IS_NUMBER_NEG; /* Clear sign for nan. */
1156 }
1157 }
1158 else if (flags & PERL_SCAN_TRAILING) {
1159 return numtype | IS_NUMBER_TRAILING;
1160 }
1161
1162 return 0;
1163}
1164
1165/*
1166=for apidoc grok_atoUV
1167
1168parse a string, looking for a decimal unsigned integer.
1169
1170On entry, C<pv> points to the beginning of the string;
1171C<valptr> points to a UV that will receive the converted value, if found;
1172C<endptr> is either NULL or points to a variable that points to one byte
1173beyond the point in C<pv> that this routine should examine.
1174If C<endptr> is NULL, C<pv> is assumed to be NUL-terminated.
1175
1176Returns FALSE if C<pv> doesn't represent a valid unsigned integer value (with
1177no leading zeros). Otherwise it returns TRUE, and sets C<*valptr> to that
1178value.
1179
1180If you constrain the portion of C<pv> that is looked at by this function (by
1181passing a non-NULL C<endptr>), and if the intial bytes of that portion form a
1182valid value, it will return TRUE, setting C<*endptr> to the byte following the
1183final digit of the value. But if there is no constraint at what's looked at,
1184all of C<pv> must be valid in order for TRUE to be returned.
1185
1186The only characters this accepts are the decimal digits '0'..'9'.
1187
1188As opposed to L<atoi(3)> or L<strtol(3)>, C<grok_atoUV> does NOT allow optional
1189leading whitespace, nor negative inputs. If such features are required, the
1190calling code needs to explicitly implement those.
1191
1192Note that this function returns FALSE for inputs that would overflow a UV,
1193or have leading zeros. Thus a single C<0> is accepted, but not C<00> nor
1194C<01>, C<002>, I<etc>.
1195
1196Background: C<atoi> has severe problems with illegal inputs, it cannot be
1197used for incremental parsing, and therefore should be avoided
1198C<atoi> and C<strtol> are also affected by locale settings, which can also be
1199seen as a bug (global state controlled by user environment).
1200
1201=cut
1202
1203*/
1204
1205bool
1206Perl_grok_atoUV(const char *pv, UV *valptr, const char** endptr)
1207{
1208 const char* s = pv;
1209 const char** eptr;
1210 const char* end2; /* Used in case endptr is NULL. */
1211 UV val = 0; /* The parsed value. */
1212
1213 PERL_ARGS_ASSERT_GROK_ATOUV;
1214
1215 if (endptr) {
1216 eptr = endptr;
1217 }
1218 else {
1219 end2 = s + strlen(s);
1220 eptr = &end2;
1221 }
1222
1223 if ( *eptr <= s
1224 || ! isDIGIT(*s))
1225 {
1226 return FALSE;
1227 }
1228
1229 /* Single-digit inputs are quite common. */
1230 val = *s++ - '0';
1231 if (s < *eptr && isDIGIT(*s)) {
1232 /* Fail on extra leading zeros. */
1233 if (val == 0)
1234 return FALSE;
1235 while (s < *eptr && isDIGIT(*s)) {
1236 /* This could be unrolled like in grok_number(), but
1237 * the expected uses of this are not speed-needy, and
1238 * unlikely to need full 64-bitness. */
1239 const U8 digit = *s++ - '0';
1240 if (val < uv_max_div_10 ||
1241 (val == uv_max_div_10 && digit <= uv_max_mod_10)) {
1242 val = val * 10 + digit;
1243 } else {
1244 return FALSE;
1245 }
1246 }
1247 }
1248
1249 if (endptr == NULL) {
1250 if (*s) {
1251 return FALSE; /* If endptr is NULL, no trailing non-digits allowed. */
1252 }
1253 }
1254 else {
1255 *endptr = s;
1256 }
1257
1258 *valptr = val;
1259 return TRUE;
1260}
1261
1262#ifndef Perl_strtod
1263STATIC NV
1264S_mulexp10(NV value, I32 exponent)
1265{
1266 NV result = 1.0;
1267 NV power = 10.0;
1268 bool negative = 0;
1269 I32 bit;
1270
1271 if (exponent == 0)
1272 return value;
1273 if (value == 0)
1274 return (NV)0;
1275
1276 /* On OpenVMS VAX we by default use the D_FLOAT double format,
1277 * and that format does not have *easy* capabilities [1] for
1278 * overflowing doubles 'silently' as IEEE fp does. We also need
1279 * to support G_FLOAT on both VAX and Alpha, and though the exponent
1280 * range is much larger than D_FLOAT it still doesn't do silent
1281 * overflow. Therefore we need to detect early whether we would
1282 * overflow (this is the behaviour of the native string-to-float
1283 * conversion routines, and therefore of native applications, too).
1284 *
1285 * [1] Trying to establish a condition handler to trap floating point
1286 * exceptions is not a good idea. */
1287
1288 /* In UNICOS and in certain Cray models (such as T90) there is no
1289 * IEEE fp, and no way at all from C to catch fp overflows gracefully.
1290 * There is something you can do if you are willing to use some
1291 * inline assembler: the instruction is called DFI-- but that will
1292 * disable *all* floating point interrupts, a little bit too large
1293 * a hammer. Therefore we need to catch potential overflows before
1294 * it's too late. */
1295
1296#if ((defined(VMS) && !defined(_IEEE_FP)) || defined(_UNICOS) || defined(DOUBLE_IS_VAX_FLOAT)) && defined(NV_MAX_10_EXP)
1297 STMT_START {
1298 const NV exp_v = log10(value);
1299 if (exponent >= NV_MAX_10_EXP || exponent + exp_v >= NV_MAX_10_EXP)
1300 return NV_MAX;
1301 if (exponent < 0) {
1302 if (-(exponent + exp_v) >= NV_MAX_10_EXP)
1303 return 0.0;
1304 while (-exponent >= NV_MAX_10_EXP) {
1305 /* combination does not overflow, but 10^(-exponent) does */
1306 value /= 10;
1307 ++exponent;
1308 }
1309 }
1310 } STMT_END;
1311#endif
1312
1313 if (exponent < 0) {
1314 negative = 1;
1315 exponent = -exponent;
1316#ifdef NV_MAX_10_EXP
1317 /* for something like 1234 x 10^-309, the action of calculating
1318 * the intermediate value 10^309 then returning 1234 / (10^309)
1319 * will fail, since 10^309 becomes infinity. In this case try to
1320 * refactor it as 123 / (10^308) etc.
1321 */
1322 while (value && exponent > NV_MAX_10_EXP) {
1323 exponent--;
1324 value /= 10;
1325 }
1326 if (value == 0.0)
1327 return value;
1328#endif
1329 }
1330#if defined(__osf__)
1331 /* Even with cc -ieee + ieee_set_fp_control(IEEE_TRAP_ENABLE_INV)
1332 * Tru64 fp behavior on inf/nan is somewhat broken. Another way
1333 * to do this would be ieee_set_fp_control(IEEE_TRAP_ENABLE_OVF)
1334 * but that breaks another set of infnan.t tests. */
1335# define FP_OVERFLOWS_TO_ZERO
1336#endif
1337 for (bit = 1; exponent; bit <<= 1) {
1338 if (exponent & bit) {
1339 exponent ^= bit;
1340 result *= power;
1341#ifdef FP_OVERFLOWS_TO_ZERO
1342 if (result == 0)
1343# ifdef NV_INF
1344 return value < 0 ? -NV_INF : NV_INF;
1345# else
1346 return value < 0 ? -FLT_MAX : FLT_MAX;
1347# endif
1348#endif
1349 /* Floating point exceptions are supposed to be turned off,
1350 * but if we're obviously done, don't risk another iteration.
1351 */
1352 if (exponent == 0) break;
1353 }
1354 power *= power;
1355 }
1356 return negative ? value / result : value * result;
1357}
1358#endif /* #ifndef Perl_strtod */
1359
1360#ifdef Perl_strtod
1361# define ATOF(s, x) my_atof2(s, &x)
1362#else
1363# define ATOF(s, x) Perl_atof2(s, x)
1364#endif
1365
1366NV
1367Perl_my_atof(pTHX_ const char* s)
1368{
1369 /* 's' must be NUL terminated */
1370
1371 NV x = 0.0;
1372
1373 PERL_ARGS_ASSERT_MY_ATOF;
1374
1375#if ! defined(USE_LOCALE_NUMERIC)
1376
1377 ATOF(s, x);
1378
1379#else
1380
1381 {
1382 DECLARATION_FOR_LC_NUMERIC_MANIPULATION;
1383 STORE_LC_NUMERIC_SET_TO_NEEDED();
1384 if (! (PL_numeric_radix_sv && IN_LC(LC_NUMERIC))) {
1385 ATOF(s,x);
1386 }
1387 else {
1388
1389 /* Look through the string for the first thing that looks like a
1390 * decimal point: either the value in the current locale or the
1391 * standard fallback of '.'. The one which appears earliest in the
1392 * input string is the one that we should have atof look for. Note
1393 * that we have to determine this beforehand because on some
1394 * systems, Perl_atof2 is just a wrapper around the system's atof.
1395 * */
1396 const char * const standard_pos = strchr(s, '.');
1397 const char * const local_pos
1398 = strstr(s, SvPV_nolen(PL_numeric_radix_sv));
1399 const bool use_standard_radix
1400 = standard_pos && (!local_pos || standard_pos < local_pos);
1401
1402 if (use_standard_radix) {
1403 SET_NUMERIC_STANDARD();
1404 LOCK_LC_NUMERIC_STANDARD();
1405 }
1406
1407 ATOF(s,x);
1408
1409 if (use_standard_radix) {
1410 UNLOCK_LC_NUMERIC_STANDARD();
1411 SET_NUMERIC_UNDERLYING();
1412 }
1413 }
1414 RESTORE_LC_NUMERIC();
1415 }
1416
1417#endif
1418
1419 return x;
1420}
1421
1422#if defined(NV_INF) || defined(NV_NAN)
1423
1424#ifdef USING_MSVC6
1425# pragma warning(push)
1426# pragma warning(disable:4756;disable:4056)
1427#endif
1428static char*
1429S_my_atof_infnan(pTHX_ const char* s, bool negative, const char* send, NV* value)
1430{
1431 const char *p0 = negative ? s - 1 : s;
1432 const char *p = p0;
1433 const int infnan = grok_infnan(&p, send);
1434 if (infnan && p != p0) {
1435 /* If we can generate inf/nan directly, let's do so. */
1436#ifdef NV_INF
1437 if ((infnan & IS_NUMBER_INFINITY)) {
1438 *value = (infnan & IS_NUMBER_NEG) ? -NV_INF: NV_INF;
1439 return (char*)p;
1440 }
1441#endif
1442#ifdef NV_NAN
1443 if ((infnan & IS_NUMBER_NAN)) {
1444 *value = NV_NAN;
1445 return (char*)p;
1446 }
1447#endif
1448#ifdef Perl_strtod
1449 /* If still here, we didn't have either NV_INF or NV_NAN,
1450 * and can try falling back to native strtod/strtold.
1451 *
1452 * The native interface might not recognize all the possible
1453 * inf/nan strings Perl recognizes. What we can try
1454 * is to try faking the input. We will try inf/-inf/nan
1455 * as the most promising/portable input. */
1456 {
1457 const char* fake = "silence compiler warning";
1458 char* endp;
1459 NV nv;
1460#ifdef NV_INF
1461 if ((infnan & IS_NUMBER_INFINITY)) {
1462 fake = ((infnan & IS_NUMBER_NEG)) ? "-inf" : "inf";
1463 }
1464#endif
1465#ifdef NV_NAN
1466 if ((infnan & IS_NUMBER_NAN)) {
1467 fake = "nan";
1468 }
1469#endif
1470 assert(strNE(fake, "silence compiler warning"));
1471 nv = S_strtod(aTHX_ fake, &endp);
1472 if (fake != endp) {
1473#ifdef NV_INF
1474 if ((infnan & IS_NUMBER_INFINITY)) {
1475# ifdef Perl_isinf
1476 if (Perl_isinf(nv))
1477 *value = nv;
1478# else
1479 /* last resort, may generate SIGFPE */
1480 *value = Perl_exp((NV)1e9);
1481 if ((infnan & IS_NUMBER_NEG))
1482 *value = -*value;
1483# endif
1484 return (char*)p; /* p, not endp */
1485 }
1486#endif
1487#ifdef NV_NAN
1488 if ((infnan & IS_NUMBER_NAN)) {
1489# ifdef Perl_isnan
1490 if (Perl_isnan(nv))
1491 *value = nv;
1492# else
1493 /* last resort, may generate SIGFPE */
1494 *value = Perl_log((NV)-1.0);
1495# endif
1496 return (char*)p; /* p, not endp */
1497#endif
1498 }
1499 }
1500 }
1501#endif /* #ifdef Perl_strtod */
1502 }
1503 return NULL;
1504}
1505#ifdef USING_MSVC6
1506# pragma warning(pop)
1507#endif
1508
1509#endif /* if defined(NV_INF) || defined(NV_NAN) */
1510
1511char*
1512Perl_my_atof2(pTHX_ const char* orig, NV* value)
1513{
1514 PERL_ARGS_ASSERT_MY_ATOF2;
1515 return my_atof3(orig, value, 0);
1516}
1517
1518char*
1519Perl_my_atof3(pTHX_ const char* orig, NV* value, const STRLEN len)
1520{
1521 const char* s = orig;
1522 NV result[3] = {0.0, 0.0, 0.0};
1523#if defined(USE_PERL_ATOF) || defined(Perl_strtod)
1524 const char* send = s + ((len != 0)
1525 ? len
1526 : strlen(orig)); /* one past the last */
1527 bool negative = 0;
1528#endif
1529#if defined(USE_PERL_ATOF) && !defined(Perl_strtod)
1530 UV accumulator[2] = {0,0}; /* before/after dp */
1531 bool seen_digit = 0;
1532 I32 exp_adjust[2] = {0,0};
1533 I32 exp_acc[2] = {-1, -1};
1534 /* the current exponent adjust for the accumulators */
1535 I32 exponent = 0;
1536 I32 seen_dp = 0;
1537 I32 digit = 0;
1538 I32 old_digit = 0;
1539 I32 sig_digits = 0; /* noof significant digits seen so far */
1540#endif
1541
1542#if defined(USE_PERL_ATOF) || defined(Perl_strtod)
1543 PERL_ARGS_ASSERT_MY_ATOF3;
1544
1545 /* leading whitespace */
1546 while (s < send && isSPACE(*s))
1547 ++s;
1548
1549 /* sign */
1550 switch (*s) {
1551 case '-':
1552 negative = 1;
1553 /* FALLTHROUGH */
1554 case '+':
1555 ++s;
1556 }
1557#endif
1558
1559#ifdef Perl_strtod
1560 {
1561 char* endp;
1562 char* copy = NULL;
1563
1564 if ((endp = S_my_atof_infnan(aTHX_ s, negative, send, value)))
1565 return endp;
1566
1567 /* strtold() accepts 0x-prefixed hex and in POSIX implementations,
1568 0b-prefixed binary numbers, which is backward incompatible
1569 */
1570 if ((len == 0 || len >= 2) && *s == '0' &&
1571 (isALPHA_FOLD_EQ(s[1], 'x') || isALPHA_FOLD_EQ(s[1], 'b'))) {
1572 *value = 0;
1573 return (char *)s+1;
1574 }
1575
1576 /* If the length is passed in, the input string isn't NUL-terminated,
1577 * and in it turns out the function below assumes it is; therefore we
1578 * create a copy and NUL-terminate that */
1579 if (len) {
1580 Newx(copy, len + 1, char);
1581 Copy(orig, copy, len, char);
1582 copy[len] = '\0';
1583 s = copy + (s - orig);
1584 }
1585
1586 result[2] = S_strtod(aTHX_ s, &endp);
1587
1588 /* If we created a copy, 'endp' is in terms of that. Convert back to
1589 * the original */
1590 if (copy) {
1591 s = (s - copy) + (char *) orig;
1592 endp = (endp - copy) + (char *) orig;
1593 Safefree(copy);
1594 }
1595
1596 if (s != endp) {
1597 *value = negative ? -result[2] : result[2];
1598 return endp;
1599 }
1600 return NULL;
1601 }
1602#elif defined(USE_PERL_ATOF)
1603
1604/* There is no point in processing more significant digits
1605 * than the NV can hold. Note that NV_DIG is a lower-bound value,
1606 * while we need an upper-bound value. We add 2 to account for this;
1607 * since it will have been conservative on both the first and last digit.
1608 * For example a 32-bit mantissa with an exponent of 4 would have
1609 * exact values in the set
1610 * 4
1611 * 8
1612 * ..
1613 * 17179869172
1614 * 17179869176
1615 * 17179869180
1616 *
1617 * where for the purposes of calculating NV_DIG we would have to discount
1618 * both the first and last digit, since neither can hold all values from
1619 * 0..9; but for calculating the value we must examine those two digits.
1620 */
1621#ifdef MAX_SIG_DIG_PLUS
1622 /* It is not necessarily the case that adding 2 to NV_DIG gets all the
1623 possible digits in a NV, especially if NVs are not IEEE compliant
1624 (e.g., long doubles on IRIX) - Allen <allens@cpan.org> */
1625# define MAX_SIG_DIGITS (NV_DIG+MAX_SIG_DIG_PLUS)
1626#else
1627# define MAX_SIG_DIGITS (NV_DIG+2)
1628#endif
1629
1630/* the max number we can accumulate in a UV, and still safely do 10*N+9 */
1631#define MAX_ACCUMULATE ( (UV) ((UV_MAX - 9)/10))
1632
1633#if defined(NV_INF) || defined(NV_NAN)
1634 {
1635 char* endp;
1636 if ((endp = S_my_atof_infnan(aTHX_ s, negative, send, value)))
1637 return endp;
1638 }
1639#endif
1640
1641 /* we accumulate digits into an integer; when this becomes too
1642 * large, we add the total to NV and start again */
1643
1644 while (s < send) {
1645 if (isDIGIT(*s)) {
1646 seen_digit = 1;
1647 old_digit = digit;
1648 digit = *s++ - '0';
1649 if (seen_dp)
1650 exp_adjust[1]++;
1651
1652 /* don't start counting until we see the first significant
1653 * digit, eg the 5 in 0.00005... */
1654 if (!sig_digits && digit == 0)
1655 continue;
1656
1657 if (++sig_digits > MAX_SIG_DIGITS) {
1658 /* limits of precision reached */
1659 if (digit > 5) {
1660 ++accumulator[seen_dp];
1661 } else if (digit == 5) {
1662 if (old_digit % 2) { /* round to even - Allen */
1663 ++accumulator[seen_dp];
1664 }
1665 }
1666 if (seen_dp) {
1667 exp_adjust[1]--;
1668 } else {
1669 exp_adjust[0]++;
1670 }
1671 /* skip remaining digits */
1672 while (s < send && isDIGIT(*s)) {
1673 ++s;
1674 if (! seen_dp) {
1675 exp_adjust[0]++;
1676 }
1677 }
1678 /* warn of loss of precision? */
1679 }
1680 else {
1681 if (accumulator[seen_dp] > MAX_ACCUMULATE) {
1682 /* add accumulator to result and start again */
1683 result[seen_dp] = S_mulexp10(result[seen_dp],
1684 exp_acc[seen_dp])
1685 + (NV)accumulator[seen_dp];
1686 accumulator[seen_dp] = 0;
1687 exp_acc[seen_dp] = 0;
1688 }
1689 accumulator[seen_dp] = accumulator[seen_dp] * 10 + digit;
1690 ++exp_acc[seen_dp];
1691 }
1692 }
1693 else if (!seen_dp && GROK_NUMERIC_RADIX(&s, send)) {
1694 seen_dp = 1;
1695 if (sig_digits > MAX_SIG_DIGITS) {
1696 while (s < send && isDIGIT(*s)) {
1697 ++s;
1698 }
1699 break;
1700 }
1701 }
1702 else {
1703 break;
1704 }
1705 }
1706
1707 result[0] = S_mulexp10(result[0], exp_acc[0]) + (NV)accumulator[0];
1708 if (seen_dp) {
1709 result[1] = S_mulexp10(result[1], exp_acc[1]) + (NV)accumulator[1];
1710 }
1711
1712 if (s < send && seen_digit && (isALPHA_FOLD_EQ(*s, 'e'))) {
1713 bool expnegative = 0;
1714
1715 ++s;
1716 switch (*s) {
1717 case '-':
1718 expnegative = 1;
1719 /* FALLTHROUGH */
1720 case '+':
1721 ++s;
1722 }
1723 while (s < send && isDIGIT(*s))
1724 exponent = exponent * 10 + (*s++ - '0');
1725 if (expnegative)
1726 exponent = -exponent;
1727 }
1728
1729 /* now apply the exponent */
1730
1731 if (seen_dp) {
1732 result[2] = S_mulexp10(result[0],exponent+exp_adjust[0])
1733 + S_mulexp10(result[1],exponent-exp_adjust[1]);
1734 } else {
1735 result[2] = S_mulexp10(result[0],exponent+exp_adjust[0]);
1736 }
1737
1738 /* now apply the sign */
1739 if (negative)
1740 result[2] = -result[2];
1741#endif /* USE_PERL_ATOF */
1742 *value = result[2];
1743 return (char *)s;
1744}
1745
1746/*
1747=for apidoc isinfnan
1748
1749C<Perl_isinfnan()> is utility function that returns true if the NV
1750argument is either an infinity or a C<NaN>, false otherwise. To test
1751in more detail, use C<Perl_isinf()> and C<Perl_isnan()>.
1752
1753This is also the logical inverse of Perl_isfinite().
1754
1755=cut
1756*/
1757bool
1758Perl_isinfnan(NV nv)
1759{
1760 PERL_UNUSED_ARG(nv);
1761#ifdef Perl_isinf
1762 if (Perl_isinf(nv))
1763 return TRUE;
1764#endif
1765#ifdef Perl_isnan
1766 if (Perl_isnan(nv))
1767 return TRUE;
1768#endif
1769 return FALSE;
1770}
1771
1772/*
1773=for apidoc
1774
1775Checks whether the argument would be either an infinity or C<NaN> when used
1776as a number, but is careful not to trigger non-numeric or uninitialized
1777warnings. it assumes the caller has done C<SvGETMAGIC(sv)> already.
1778
1779=cut
1780*/
1781
1782bool
1783Perl_isinfnansv(pTHX_ SV *sv)
1784{
1785 PERL_ARGS_ASSERT_ISINFNANSV;
1786 if (!SvOK(sv))
1787 return FALSE;
1788 if (SvNOKp(sv))
1789 return Perl_isinfnan(SvNVX(sv));
1790 if (SvIOKp(sv))
1791 return FALSE;
1792 {
1793 STRLEN len;
1794 const char *s = SvPV_nomg_const(sv, len);
1795 return cBOOL(grok_infnan(&s, s+len));
1796 }
1797}
1798
1799#ifndef HAS_MODFL
1800/* C99 has truncl, pre-C99 Solaris had aintl. We can use either with
1801 * copysignl to emulate modfl, which is in some platforms missing or
1802 * broken. */
1803# if defined(HAS_TRUNCL) && defined(HAS_COPYSIGNL)
1804long double
1805Perl_my_modfl(long double x, long double *ip)
1806{
1807 *ip = truncl(x);
1808 return (x == *ip ? copysignl(0.0L, x) : x - *ip);
1809}
1810# elif defined(HAS_AINTL) && defined(HAS_COPYSIGNL)
1811long double
1812Perl_my_modfl(long double x, long double *ip)
1813{
1814 *ip = aintl(x);
1815 return (x == *ip ? copysignl(0.0L, x) : x - *ip);
1816}
1817# endif
1818#endif
1819
1820/* Similarly, with ilogbl and scalbnl we can emulate frexpl. */
1821#if ! defined(HAS_FREXPL) && defined(HAS_ILOGBL) && defined(HAS_SCALBNL)
1822long double
1823Perl_my_frexpl(long double x, int *e) {
1824 *e = x == 0.0L ? 0 : ilogbl(x) + 1;
1825 return (scalbnl(x, -*e));
1826}
1827#endif
1828
1829/*
1830=for apidoc Perl_signbit
1831
1832Return a non-zero integer if the sign bit on an NV is set, and 0 if
1833it is not.
1834
1835If F<Configure> detects this system has a C<signbit()> that will work with
1836our NVs, then we just use it via the C<#define> in F<perl.h>. Otherwise,
1837fall back on this implementation. The main use of this function
1838is catching C<-0.0>.
1839
1840C<Configure> notes: This function is called C<'Perl_signbit'> instead of a
1841plain C<'signbit'> because it is easy to imagine a system having a C<signbit()>
1842function or macro that doesn't happen to work with our particular choice
1843of NVs. We shouldn't just re-C<#define> C<signbit> as C<Perl_signbit> and expect
1844the standard system headers to be happy. Also, this is a no-context
1845function (no C<pTHX_>) because C<Perl_signbit()> is usually re-C<#defined> in
1846F<perl.h> as a simple macro call to the system's C<signbit()>.
1847Users should just always call C<Perl_signbit()>.
1848
1849=cut
1850*/
1851#if !defined(HAS_SIGNBIT)
1852int
1853Perl_signbit(NV x) {
1854# ifdef Perl_fp_class_nzero
1855 return Perl_fp_class_nzero(x);
1856 /* Try finding the high byte, and assume it's highest bit
1857 * is the sign. This assumption is probably wrong somewhere. */
1858# elif defined(USE_LONG_DOUBLE) && LONG_DOUBLEKIND == LONG_DOUBLE_IS_X86_80_BIT_LITTLE_ENDIAN
1859 return (((unsigned char *)&x)[9] & 0x80);
1860# elif defined(NV_LITTLE_ENDIAN)
1861 /* Note that NVSIZE is sizeof(NV), which would make the below be
1862 * wrong if the end bytes are unused, which happens with the x86
1863 * 80-bit long doubles, which is why take care of that above. */
1864 return (((unsigned char *)&x)[NVSIZE - 1] & 0x80);
1865# elif defined(NV_BIG_ENDIAN)
1866 return (((unsigned char *)&x)[0] & 0x80);
1867# else
1868 /* This last resort fallback is wrong for the negative zero. */
1869 return (x < 0.0) ? 1 : 0;
1870# endif
1871}
1872#endif
1873
1874/*
1875 * ex: set ts=8 sts=4 sw=4 et:
1876 */