This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Integrate mainline
[perl5.git] / numeric.c
CommitLineData
98994639
HS
1/* numeric.c
2 *
3 * Copyright (c) 2001, Larry Wall
4 *
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
7 *
8 */
9
10/*
11 * "That only makes eleven (plus one mislaid) and not fourteen, unless
12 * wizards count differently to other people."
13 */
14
15#include "EXTERN.h"
16#define PERL_IN_NUMERIC_C
17#include "perl.h"
18
19U32
20Perl_cast_ulong(pTHX_ NV f)
21{
22 if (f < 0.0)
23 return f < I32_MIN ? (U32) I32_MIN : (U32)(I32) f;
24 if (f < U32_MAX_P1) {
25#if CASTFLAGS & 2
26 if (f < U32_MAX_P1_HALF)
27 return (U32) f;
28 f -= U32_MAX_P1_HALF;
29 return ((U32) f) | (1 + U32_MAX >> 1);
30#else
31 return (U32) f;
32#endif
33 }
34 return f > 0 ? U32_MAX : 0 /* NaN */;
35}
36
37I32
38Perl_cast_i32(pTHX_ NV f)
39{
40 if (f < I32_MAX_P1)
41 return f < I32_MIN ? I32_MIN : (I32) f;
42 if (f < U32_MAX_P1) {
43#if CASTFLAGS & 2
44 if (f < U32_MAX_P1_HALF)
45 return (I32)(U32) f;
46 f -= U32_MAX_P1_HALF;
47 return (I32)(((U32) f) | (1 + U32_MAX >> 1));
48#else
49 return (I32)(U32) f;
50#endif
51 }
52 return f > 0 ? (I32)U32_MAX : 0 /* NaN */;
53}
54
55IV
56Perl_cast_iv(pTHX_ NV f)
57{
58 if (f < IV_MAX_P1)
59 return f < IV_MIN ? IV_MIN : (IV) f;
60 if (f < UV_MAX_P1) {
61#if CASTFLAGS & 2
62 /* For future flexibility allowing for sizeof(UV) >= sizeof(IV) */
63 if (f < UV_MAX_P1_HALF)
64 return (IV)(UV) f;
65 f -= UV_MAX_P1_HALF;
66 return (IV)(((UV) f) | (1 + UV_MAX >> 1));
67#else
68 return (IV)(UV) f;
69#endif
70 }
71 return f > 0 ? (IV)UV_MAX : 0 /* NaN */;
72}
73
74UV
75Perl_cast_uv(pTHX_ NV f)
76{
77 if (f < 0.0)
78 return f < IV_MIN ? (UV) IV_MIN : (UV)(IV) f;
79 if (f < UV_MAX_P1) {
80#if CASTFLAGS & 2
81 if (f < UV_MAX_P1_HALF)
82 return (UV) f;
83 f -= UV_MAX_P1_HALF;
84 return ((UV) f) | (1 + UV_MAX >> 1);
85#else
86 return (UV) f;
87#endif
88 }
89 return f > 0 ? UV_MAX : 0 /* NaN */;
90}
91
92#if defined(HUGE_VAL) || (defined(USE_LONG_DOUBLE) && defined(HUGE_VALL))
93/*
94 * This hack is to force load of "huge" support from libm.a
95 * So it is in perl for (say) POSIX to use.
96 * Needed for SunOS with Sun's 'acc' for example.
97 */
98NV
99Perl_huge(void)
100{
101# if defined(USE_LONG_DOUBLE) && defined(HUGE_VALL)
102 return HUGE_VALL;
103# endif
104 return HUGE_VAL;
105}
106#endif
107
53305cf1
NC
108/*
109=for apidoc grok_bin
98994639 110
53305cf1
NC
111converts a string representing a binary number to numeric form.
112
113On entry I<start> and I<*len> give the string to scan, I<*flags> gives
114conversion flags, and I<result> should be NULL or a pointer to an NV.
115The scan stops at the end of the string, or the first invalid character.
116On return I<*len> is set to the length scanned string, and I<*flags> gives
117output flags.
118
119If the value is <= UV_MAX it is returned as a UV, the output flags are clear,
120and nothing is written to I<*result>. If the value is > UV_MAX C<grok_bin>
121returns UV_MAX, sets C<PERL_SCAN_GREATER_THAN_UV_MAX> in the output flags,
122and writes the value to I<*result> (or the value is discarded if I<result>
123is NULL).
124
d1be9408 125The hex number may optionally be prefixed with "0b" or "b" unless
a4c04bdc
NC
126C<PERL_SCAN_DISALLOW_PREFIX> is set in I<*flags> on entry. If
127C<PERL_SCAN_ALLOW_UNDERSCORES> is set in I<*flags> then the binary
53305cf1
NC
128number may use '_' characters to separate digits.
129
130=cut
131 */
132
133UV
134Perl_grok_bin(pTHX_ char *start, STRLEN *len_p, I32 *flags, NV *result) {
135 const char *s = start;
136 STRLEN len = *len_p;
137 UV value = 0;
138 NV value_nv = 0;
139
140 const UV max_div_2 = UV_MAX / 2;
141 bool allow_underscores = *flags & PERL_SCAN_ALLOW_UNDERSCORES;
142 bool overflowed = FALSE;
143
a4c04bdc
NC
144 if (!(*flags & PERL_SCAN_DISALLOW_PREFIX)) {
145 /* strip off leading b or 0b.
146 for compatibility silently suffer "b" and "0b" as valid binary
147 numbers. */
148 if (len >= 1) {
149 if (s[0] == 'b') {
150 s++;
151 len--;
152 }
153 else if (len >= 2 && s[0] == '0' && s[1] == 'b') {
154 s+=2;
155 len-=2;
156 }
157 }
53305cf1
NC
158 }
159
160 for (; len-- && *s; s++) {
161 char bit = *s;
162 if (bit == '0' || bit == '1') {
163 /* Write it in this wonky order with a goto to attempt to get the
164 compiler to make the common case integer-only loop pretty tight.
165 With gcc seems to be much straighter code than old scan_bin. */
166 redo:
167 if (!overflowed) {
168 if (value <= max_div_2) {
169 value = (value << 1) | (bit - '0');
170 continue;
171 }
172 /* Bah. We're just overflowed. */
173 if (ckWARN_d(WARN_OVERFLOW))
174 Perl_warner(aTHX_ WARN_OVERFLOW,
175 "Integer overflow in binary number");
176 overflowed = TRUE;
177 value_nv = (NV) value;
178 }
179 value_nv *= 2.0;
98994639 180 /* If an NV has not enough bits in its mantissa to
d1be9408 181 * represent a UV this summing of small low-order numbers
98994639
HS
182 * is a waste of time (because the NV cannot preserve
183 * the low-order bits anyway): we could just remember when
53305cf1 184 * did we overflow and in the end just multiply value_nv by the
98994639 185 * right amount. */
53305cf1
NC
186 value_nv += (NV)(bit - '0');
187 continue;
188 }
189 if (bit == '_' && len && allow_underscores && (bit = s[1])
190 && (bit == '0' || bit == '1'))
98994639
HS
191 {
192 --len;
193 ++s;
53305cf1 194 goto redo;
98994639 195 }
53305cf1
NC
196 if (ckWARN(WARN_DIGIT))
197 Perl_warner(aTHX_ WARN_DIGIT,
198 "Illegal binary digit '%c' ignored", *s);
199 break;
98994639 200 }
53305cf1
NC
201
202 if ( ( overflowed && value_nv > 4294967295.0)
98994639 203#if UVSIZE > 4
53305cf1 204 || (!overflowed && value > 0xffffffff )
98994639
HS
205#endif
206 ) {
207 if (ckWARN(WARN_PORTABLE))
208 Perl_warner(aTHX_ WARN_PORTABLE,
53305cf1
NC
209 "Binary number > 0b11111111111111111111111111111111 non-portable");
210 }
211 *len_p = s - start;
212 if (!overflowed) {
213 *flags = 0;
214 return value;
98994639 215 }
53305cf1
NC
216 *flags = PERL_SCAN_GREATER_THAN_UV_MAX;
217 if (result)
218 *result = value_nv;
219 return UV_MAX;
98994639
HS
220}
221
53305cf1
NC
222/*
223=for apidoc grok_hex
224
225converts a string representing a hex number to numeric form.
226
227On entry I<start> and I<*len> give the string to scan, I<*flags> gives
228conversion flags, and I<result> should be NULL or a pointer to an NV.
229The scan stops at the end of the string, or the first non-hex-digit character.
230On return I<*len> is set to the length scanned string, and I<*flags> gives
231output flags.
232
233If the value is <= UV_MAX it is returned as a UV, the output flags are clear,
234and nothing is written to I<*result>. If the value is > UV_MAX C<grok_hex>
235returns UV_MAX, sets C<PERL_SCAN_GREATER_THAN_UV_MAX> in the output flags,
236and writes the value to I<*result> (or the value is discarded if I<result>
237is NULL).
238
d1be9408 239The hex number may optionally be prefixed with "0x" or "x" unless
a4c04bdc
NC
240C<PERL_SCAN_DISALLOW_PREFIX> is set in I<*flags> on entry. If
241C<PERL_SCAN_ALLOW_UNDERSCORES> is set in I<*flags> then the hex
53305cf1
NC
242number may use '_' characters to separate digits.
243
244=cut
245 */
246
247UV
248Perl_grok_hex(pTHX_ char *start, STRLEN *len_p, I32 *flags, NV *result) {
249 const char *s = start;
250 STRLEN len = *len_p;
251 UV value = 0;
252 NV value_nv = 0;
253
254 const UV max_div_16 = UV_MAX / 16;
255 bool allow_underscores = *flags & PERL_SCAN_ALLOW_UNDERSCORES;
256 bool overflowed = FALSE;
257 const char *hexdigit;
98994639 258
a4c04bdc
NC
259 if (!(*flags & PERL_SCAN_DISALLOW_PREFIX)) {
260 /* strip off leading x or 0x.
261 for compatibility silently suffer "x" and "0x" as valid hex numbers.
262 */
263 if (len >= 1) {
264 if (s[0] == 'x') {
265 s++;
266 len--;
267 }
268 else if (len >= 2 && s[0] == '0' && s[1] == 'x') {
269 s+=2;
270 len-=2;
271 }
272 }
98994639
HS
273 }
274
275 for (; len-- && *s; s++) {
276 hexdigit = strchr((char *) PL_hexdigit, *s);
53305cf1
NC
277 if (hexdigit) {
278 /* Write it in this wonky order with a goto to attempt to get the
279 compiler to make the common case integer-only loop pretty tight.
280 With gcc seems to be much straighter code than old scan_hex. */
281 redo:
282 if (!overflowed) {
283 if (value <= max_div_16) {
284 value = (value << 4) | ((hexdigit - PL_hexdigit) & 15);
285 continue;
286 }
287 /* Bah. We're just overflowed. */
288 if (ckWARN_d(WARN_OVERFLOW))
289 Perl_warner(aTHX_ WARN_OVERFLOW,
290 "Integer overflow in hexadecimal number");
291 overflowed = TRUE;
292 value_nv = (NV) value;
293 }
294 value_nv *= 16.0;
295 /* If an NV has not enough bits in its mantissa to
d1be9408 296 * represent a UV this summing of small low-order numbers
53305cf1
NC
297 * is a waste of time (because the NV cannot preserve
298 * the low-order bits anyway): we could just remember when
299 * did we overflow and in the end just multiply value_nv by the
300 * right amount of 16-tuples. */
301 value_nv += (NV)((hexdigit - PL_hexdigit) & 15);
302 continue;
303 }
304 if (*s == '_' && len && allow_underscores && s[1]
98994639
HS
305 && (hexdigit = strchr((char *) PL_hexdigit, s[1])))
306 {
307 --len;
308 ++s;
53305cf1 309 goto redo;
98994639 310 }
53305cf1
NC
311 if (ckWARN(WARN_DIGIT))
312 Perl_warner(aTHX_ WARN_DIGIT,
313 "Illegal hexadecimal digit '%c' ignored", *s);
314 break;
315 }
316
317 if ( ( overflowed && value_nv > 4294967295.0)
318#if UVSIZE > 4
319 || (!overflowed && value > 0xffffffff )
320#endif
321 ) {
322 if (ckWARN(WARN_PORTABLE))
323 Perl_warner(aTHX_ WARN_PORTABLE,
324 "Hexadecimal number > 0xffffffff non-portable");
325 }
326 *len_p = s - start;
327 if (!overflowed) {
328 *flags = 0;
329 return value;
330 }
331 *flags = PERL_SCAN_GREATER_THAN_UV_MAX;
332 if (result)
333 *result = value_nv;
334 return UV_MAX;
335}
336
337/*
338=for apidoc grok_oct
339
340
341=cut
342 */
343
344UV
345Perl_grok_oct(pTHX_ char *start, STRLEN *len_p, I32 *flags, NV *result) {
346 const char *s = start;
347 STRLEN len = *len_p;
348 UV value = 0;
349 NV value_nv = 0;
350
351 const UV max_div_8 = UV_MAX / 8;
352 bool allow_underscores = *flags & PERL_SCAN_ALLOW_UNDERSCORES;
353 bool overflowed = FALSE;
354
355 for (; len-- && *s; s++) {
356 /* gcc 2.95 optimiser not smart enough to figure that this subtraction
357 out front allows slicker code. */
358 int digit = *s - '0';
359 if (digit >= 0 && digit <= 7) {
360 /* Write it in this wonky order with a goto to attempt to get the
361 compiler to make the common case integer-only loop pretty tight.
362 */
363 redo:
364 if (!overflowed) {
365 if (value <= max_div_8) {
366 value = (value << 3) | digit;
367 continue;
368 }
369 /* Bah. We're just overflowed. */
370 if (ckWARN_d(WARN_OVERFLOW))
371 Perl_warner(aTHX_ WARN_OVERFLOW,
372 "Integer overflow in octal number");
373 overflowed = TRUE;
374 value_nv = (NV) value;
375 }
376 value_nv *= 8.0;
98994639 377 /* If an NV has not enough bits in its mantissa to
d1be9408 378 * represent a UV this summing of small low-order numbers
98994639
HS
379 * is a waste of time (because the NV cannot preserve
380 * the low-order bits anyway): we could just remember when
53305cf1
NC
381 * did we overflow and in the end just multiply value_nv by the
382 * right amount of 8-tuples. */
383 value_nv += (NV)digit;
384 continue;
385 }
386 if (digit == ('_' - '0') && len && allow_underscores
387 && (digit = s[1] - '0') && (digit >= 0 && digit <= 7))
388 {
389 --len;
390 ++s;
391 goto redo;
392 }
393 /* Allow \octal to work the DWIM way (that is, stop scanning
394 * as soon as non-octal characters are seen, complain only iff
395 * someone seems to want to use the digits eight and nine). */
396 if (digit == 8 || digit == 9) {
397 if (ckWARN(WARN_DIGIT))
398 Perl_warner(aTHX_ WARN_DIGIT,
399 "Illegal octal digit '%c' ignored", *s);
400 }
401 break;
98994639 402 }
53305cf1
NC
403
404 if ( ( overflowed && value_nv > 4294967295.0)
98994639 405#if UVSIZE > 4
53305cf1 406 || (!overflowed && value > 0xffffffff )
98994639
HS
407#endif
408 ) {
409 if (ckWARN(WARN_PORTABLE))
410 Perl_warner(aTHX_ WARN_PORTABLE,
53305cf1
NC
411 "Octal number > 037777777777 non-portable");
412 }
413 *len_p = s - start;
414 if (!overflowed) {
415 *flags = 0;
416 return value;
98994639 417 }
53305cf1
NC
418 *flags = PERL_SCAN_GREATER_THAN_UV_MAX;
419 if (result)
420 *result = value_nv;
421 return UV_MAX;
422}
423
424/*
425=for apidoc scan_bin
426
427For backwards compatibility. Use C<grok_bin> instead.
428
429=for apidoc scan_hex
430
431For backwards compatibility. Use C<grok_hex> instead.
432
433=for apidoc scan_oct
434
435For backwards compatibility. Use C<grok_oct> instead.
436
437=cut
438 */
439
440NV
441Perl_scan_bin(pTHX_ char *start, STRLEN len, STRLEN *retlen)
442{
443 NV rnv;
444 I32 flags = *retlen ? PERL_SCAN_ALLOW_UNDERSCORES : 0;
445 UV ruv = grok_bin (start, &len, &flags, &rnv);
446
447 *retlen = len;
448 return (flags & PERL_SCAN_GREATER_THAN_UV_MAX) ? rnv : (NV)ruv;
449}
450
451NV
452Perl_scan_oct(pTHX_ char *start, STRLEN len, STRLEN *retlen)
453{
454 NV rnv;
455 I32 flags = *retlen ? PERL_SCAN_ALLOW_UNDERSCORES : 0;
456 UV ruv = grok_oct (start, &len, &flags, &rnv);
457
458 *retlen = len;
459 return (flags & PERL_SCAN_GREATER_THAN_UV_MAX) ? rnv : (NV)ruv;
460}
461
462NV
463Perl_scan_hex(pTHX_ char *start, STRLEN len, STRLEN *retlen)
464{
465 NV rnv;
466 I32 flags = *retlen ? PERL_SCAN_ALLOW_UNDERSCORES : 0;
467 UV ruv = grok_hex (start, &len, &flags, &rnv);
468
469 *retlen = len;
470 return (flags & PERL_SCAN_GREATER_THAN_UV_MAX) ? rnv : (NV)ruv;
98994639
HS
471}
472
473/*
474=for apidoc grok_numeric_radix
475
476Scan and skip for a numeric decimal separator (radix).
477
478=cut
479 */
480bool
481Perl_grok_numeric_radix(pTHX_ const char **sp, const char *send)
482{
483#ifdef USE_LOCALE_NUMERIC
484 if (PL_numeric_radix_sv && IN_LOCALE) {
485 STRLEN len;
486 char* radix = SvPV(PL_numeric_radix_sv, len);
487 if (*sp + len <= send && memEQ(*sp, radix, len)) {
488 *sp += len;
489 return TRUE;
490 }
491 }
492 /* always try "." if numeric radix didn't match because
493 * we may have data from different locales mixed */
494#endif
495 if (*sp < send && **sp == '.') {
496 ++*sp;
497 return TRUE;
498 }
499 return FALSE;
500}
501
502/*
503=for apidoc grok_number
504
505Recognise (or not) a number. The type of the number is returned
506(0 if unrecognised), otherwise it is a bit-ORed combination of
507IS_NUMBER_IN_UV, IS_NUMBER_GREATER_THAN_UV_MAX, IS_NUMBER_NOT_INT,
aa8b85de 508IS_NUMBER_NEG, IS_NUMBER_INFINITY, IS_NUMBER_NAN (defined in perl.h).
60939fb8
NC
509
510If the value of the number can fit an in UV, it is returned in the *valuep
511IS_NUMBER_IN_UV will be set to indicate that *valuep is valid, IS_NUMBER_IN_UV
512will never be set unless *valuep is valid, but *valuep may have been assigned
513to during processing even though IS_NUMBER_IN_UV is not set on return.
514If valuep is NULL, IS_NUMBER_IN_UV will be set for the same cases as when
515valuep is non-NULL, but no actual assignment (or SEGV) will occur.
516
517IS_NUMBER_NOT_INT will be set with IS_NUMBER_IN_UV if trailing decimals were
518seen (in which case *valuep gives the true value truncated to an integer), and
519IS_NUMBER_NEG if the number is negative (in which case *valuep holds the
520absolute value). IS_NUMBER_IN_UV is not set if e notation was used or the
521number is larger than a UV.
98994639
HS
522
523=cut
524 */
525int
526Perl_grok_number(pTHX_ const char *pv, STRLEN len, UV *valuep)
527{
60939fb8
NC
528 const char *s = pv;
529 const char *send = pv + len;
530 const UV max_div_10 = UV_MAX / 10;
531 const char max_mod_10 = UV_MAX % 10;
532 int numtype = 0;
533 int sawinf = 0;
aa8b85de 534 int sawnan = 0;
60939fb8
NC
535
536 while (s < send && isSPACE(*s))
537 s++;
538 if (s == send) {
539 return 0;
540 } else if (*s == '-') {
541 s++;
542 numtype = IS_NUMBER_NEG;
543 }
544 else if (*s == '+')
545 s++;
546
547 if (s == send)
548 return 0;
549
550 /* next must be digit or the radix separator or beginning of infinity */
551 if (isDIGIT(*s)) {
552 /* UVs are at least 32 bits, so the first 9 decimal digits cannot
553 overflow. */
554 UV value = *s - '0';
555 /* This construction seems to be more optimiser friendly.
556 (without it gcc does the isDIGIT test and the *s - '0' separately)
557 With it gcc on arm is managing 6 instructions (6 cycles) per digit.
558 In theory the optimiser could deduce how far to unroll the loop
559 before checking for overflow. */
58bb9ec3
NC
560 if (++s < send) {
561 int digit = *s - '0';
60939fb8
NC
562 if (digit >= 0 && digit <= 9) {
563 value = value * 10 + digit;
58bb9ec3
NC
564 if (++s < send) {
565 digit = *s - '0';
60939fb8
NC
566 if (digit >= 0 && digit <= 9) {
567 value = value * 10 + digit;
58bb9ec3
NC
568 if (++s < send) {
569 digit = *s - '0';
60939fb8
NC
570 if (digit >= 0 && digit <= 9) {
571 value = value * 10 + digit;
58bb9ec3
NC
572 if (++s < send) {
573 digit = *s - '0';
60939fb8
NC
574 if (digit >= 0 && digit <= 9) {
575 value = value * 10 + digit;
58bb9ec3
NC
576 if (++s < send) {
577 digit = *s - '0';
60939fb8
NC
578 if (digit >= 0 && digit <= 9) {
579 value = value * 10 + digit;
58bb9ec3
NC
580 if (++s < send) {
581 digit = *s - '0';
60939fb8
NC
582 if (digit >= 0 && digit <= 9) {
583 value = value * 10 + digit;
58bb9ec3
NC
584 if (++s < send) {
585 digit = *s - '0';
60939fb8
NC
586 if (digit >= 0 && digit <= 9) {
587 value = value * 10 + digit;
58bb9ec3
NC
588 if (++s < send) {
589 digit = *s - '0';
60939fb8
NC
590 if (digit >= 0 && digit <= 9) {
591 value = value * 10 + digit;
58bb9ec3 592 if (++s < send) {
60939fb8
NC
593 /* Now got 9 digits, so need to check
594 each time for overflow. */
58bb9ec3 595 digit = *s - '0';
60939fb8
NC
596 while (digit >= 0 && digit <= 9
597 && (value < max_div_10
598 || (value == max_div_10
599 && digit <= max_mod_10))) {
600 value = value * 10 + digit;
58bb9ec3
NC
601 if (++s < send)
602 digit = *s - '0';
60939fb8
NC
603 else
604 break;
605 }
606 if (digit >= 0 && digit <= 9
51bd16da 607 && (s < send)) {
60939fb8
NC
608 /* value overflowed.
609 skip the remaining digits, don't
610 worry about setting *valuep. */
611 do {
612 s++;
613 } while (s < send && isDIGIT(*s));
614 numtype |=
615 IS_NUMBER_GREATER_THAN_UV_MAX;
616 goto skip_value;
617 }
618 }
619 }
98994639 620 }
60939fb8
NC
621 }
622 }
623 }
624 }
625 }
626 }
627 }
628 }
629 }
630 }
631 }
98994639 632 }
60939fb8 633 }
98994639 634 }
60939fb8
NC
635 numtype |= IS_NUMBER_IN_UV;
636 if (valuep)
637 *valuep = value;
638
639 skip_value:
640 if (GROK_NUMERIC_RADIX(&s, send)) {
641 numtype |= IS_NUMBER_NOT_INT;
642 while (s < send && isDIGIT(*s)) /* optional digits after the radix */
643 s++;
98994639 644 }
60939fb8
NC
645 }
646 else if (GROK_NUMERIC_RADIX(&s, send)) {
647 numtype |= IS_NUMBER_NOT_INT | IS_NUMBER_IN_UV; /* valuep assigned below */
648 /* no digits before the radix means we need digits after it */
649 if (s < send && isDIGIT(*s)) {
650 do {
651 s++;
652 } while (s < send && isDIGIT(*s));
653 if (valuep) {
654 /* integer approximation is valid - it's 0. */
655 *valuep = 0;
656 }
98994639 657 }
60939fb8
NC
658 else
659 return 0;
660 } else if (*s == 'I' || *s == 'i') {
661 s++; if (s == send || (*s != 'N' && *s != 'n')) return 0;
662 s++; if (s == send || (*s != 'F' && *s != 'f')) return 0;
663 s++; if (s < send && (*s == 'I' || *s == 'i')) {
664 s++; if (s == send || (*s != 'N' && *s != 'n')) return 0;
665 s++; if (s == send || (*s != 'I' && *s != 'i')) return 0;
666 s++; if (s == send || (*s != 'T' && *s != 't')) return 0;
667 s++; if (s == send || (*s != 'Y' && *s != 'y')) return 0;
668 s++;
98994639 669 }
60939fb8 670 sawinf = 1;
aa8b85de
JH
671 } else if (*s == 'N' || *s == 'n') {
672 /* XXX TODO: There are signaling NaNs and quiet NaNs. */
673 s++; if (s == send || (*s != 'A' && *s != 'a')) return 0;
674 s++; if (s == send || (*s != 'N' && *s != 'n')) return 0;
675 s++;
676 sawnan = 1;
677 } else
98994639 678 return 0;
60939fb8
NC
679
680 if (sawinf) {
681 numtype &= IS_NUMBER_NEG; /* Keep track of sign */
682 numtype |= IS_NUMBER_INFINITY | IS_NUMBER_NOT_INT;
aa8b85de
JH
683 } else if (sawnan) {
684 numtype &= IS_NUMBER_NEG; /* Keep track of sign */
685 numtype |= IS_NUMBER_NAN | IS_NUMBER_NOT_INT;
60939fb8
NC
686 } else if (s < send) {
687 /* we can have an optional exponent part */
688 if (*s == 'e' || *s == 'E') {
689 /* The only flag we keep is sign. Blow away any "it's UV" */
690 numtype &= IS_NUMBER_NEG;
691 numtype |= IS_NUMBER_NOT_INT;
692 s++;
693 if (s < send && (*s == '-' || *s == '+'))
694 s++;
695 if (s < send && isDIGIT(*s)) {
696 do {
697 s++;
698 } while (s < send && isDIGIT(*s));
699 }
700 else
701 return 0;
702 }
703 }
704 while (s < send && isSPACE(*s))
705 s++;
706 if (s >= send)
aa8b85de 707 return numtype;
60939fb8
NC
708 if (len == 10 && memEQ(pv, "0 but true", 10)) {
709 if (valuep)
710 *valuep = 0;
711 return IS_NUMBER_IN_UV;
712 }
713 return 0;
98994639
HS
714}
715
716NV
717S_mulexp10(NV value, I32 exponent)
718{
719 NV result = 1.0;
720 NV power = 10.0;
721 bool negative = 0;
722 I32 bit;
723
724 if (exponent == 0)
725 return value;
726 else if (exponent < 0) {
727 negative = 1;
728 exponent = -exponent;
729 }
87032ba1 730
24866caa 731 /* On OpenVMS VAX we by default use the D_FLOAT double format,
67597c89 732 * and that format does not have *easy* capabilities [1] for
24866caa
CB
733 * overflowing doubles 'silently' as IEEE fp does. We also need
734 * to support G_FLOAT on both VAX and Alpha, and though the exponent
735 * range is much larger than D_FLOAT it still doesn't do silent
736 * overflow. Therefore we need to detect early whether we would
737 * overflow (this is the behaviour of the native string-to-float
738 * conversion routines, and therefore of native applications, too).
67597c89 739 *
24866caa
CB
740 * [1] Trying to establish a condition handler to trap floating point
741 * exceptions is not a good idea. */
742#if defined(VMS) && !defined(__IEEE_FP) && defined(NV_MAX_10_EXP)
67597c89 743 if (!negative &&
24866caa 744 (log10(value) + exponent) >= (NV_MAX_10_EXP))
67597c89 745 return NV_MAX;
67597c89 746#endif
87032ba1
JH
747
748 /* In UNICOS and in certain Cray models (such as T90) there is no
749 * IEEE fp, and no way at all from C to catch fp overflows gracefully.
750 * There is something you can do if you are willing to use some
751 * inline assembler: the instruction is called DFI-- but that will
752 * disable *all* floating point interrupts, a little bit too large
753 * a hammer. Therefore we need to catch potential overflows before
754 * it's too late. */
755#if defined(_UNICOS) && defined(NV_MAX_10_EXP)
756 if (!negative &&
757 (log10(value) + exponent) >= NV_MAX_10_EXP)
758 return NV_MAX;
759#endif
760
98994639
HS
761 for (bit = 1; exponent; bit <<= 1) {
762 if (exponent & bit) {
763 exponent ^= bit;
764 result *= power;
765 }
7014c407 766 /* Floating point exceptions are supposed to be turned off. */
98994639
HS
767 power *= power;
768 }
769 return negative ? value / result : value * result;
770}
771
772NV
773Perl_my_atof(pTHX_ const char* s)
774{
775 NV x = 0.0;
776#ifdef USE_LOCALE_NUMERIC
777 if (PL_numeric_local && IN_LOCALE) {
778 NV y;
779
780 /* Scan the number twice; once using locale and once without;
781 * choose the larger result (in absolute value). */
782 Perl_atof2(aTHX_ s, &x);
783 SET_NUMERIC_STANDARD();
784 Perl_atof2(aTHX_ s, &y);
785 SET_NUMERIC_LOCAL();
786 if ((y < 0.0 && y < x) || (y > 0.0 && y > x))
787 return y;
788 }
789 else
790 Perl_atof2(aTHX_ s, &x);
791#else
792 Perl_atof2(aTHX_ s, &x);
793#endif
794 return x;
795}
796
797char*
798Perl_my_atof2(pTHX_ const char* orig, NV* value)
799{
800 NV result = 0.0;
801 bool negative = 0;
802 char* s = (char*)orig;
803 char* send = s + strlen(orig) - 1;
804 bool seendigit = 0;
805 I32 expextra = 0;
806 I32 exponent = 0;
807 I32 i;
808/* this is arbitrary */
809#define PARTLIM 6
810/* we want the largest integers we can usefully use */
811#if defined(HAS_QUAD) && defined(USE_64_BIT_INT)
812# define PARTSIZE ((int)TYPE_DIGITS(U64)-1)
813 U64 part[PARTLIM];
814#else
815# define PARTSIZE ((int)TYPE_DIGITS(U32)-1)
816 U32 part[PARTLIM];
817#endif
818 I32 ipart = 0; /* index into part[] */
819 I32 offcount; /* number of digits in least significant part */
820
96a05aee
HS
821 /* leading whitespace */
822 while (isSPACE(*s))
823 ++s;
824
98994639
HS
825 /* sign */
826 switch (*s) {
827 case '-':
828 negative = 1;
829 /* fall through */
830 case '+':
831 ++s;
832 }
833
834 part[0] = offcount = 0;
835 if (isDIGIT(*s)) {
836 seendigit = 1; /* get this over with */
837
838 /* skip leading zeros */
839 while (*s == '0')
840 ++s;
841 }
842
843 /* integer digits */
844 while (isDIGIT(*s)) {
845 if (++offcount > PARTSIZE) {
846 if (++ipart < PARTLIM) {
847 part[ipart] = 0;
848 offcount = 1; /* ++0 */
849 }
850 else {
851 /* limits of precision reached */
852 --ipart;
853 --offcount;
854 if (*s >= '5')
855 ++part[ipart];
856 while (isDIGIT(*s)) {
857 ++expextra;
858 ++s;
859 }
860 /* warn of loss of precision? */
861 break;
862 }
863 }
864 part[ipart] = part[ipart] * 10 + (*s++ - '0');
865 }
866
867 /* decimal point */
868 if (GROK_NUMERIC_RADIX((const char **)&s, send)) {
869 if (isDIGIT(*s))
870 seendigit = 1; /* get this over with */
871
872 /* decimal digits */
873 while (isDIGIT(*s)) {
874 if (++offcount > PARTSIZE) {
875 if (++ipart < PARTLIM) {
876 part[ipart] = 0;
877 offcount = 1; /* ++0 */
878 }
879 else {
880 /* limits of precision reached */
881 --ipart;
882 --offcount;
883 if (*s >= '5')
884 ++part[ipart];
885 while (isDIGIT(*s))
886 ++s;
887 /* warn of loss of precision? */
888 break;
889 }
890 }
891 --expextra;
892 part[ipart] = part[ipart] * 10 + (*s++ - '0');
893 }
894 }
895
896 /* combine components of mantissa */
897 for (i = 0; i <= ipart; ++i)
898 result += S_mulexp10((NV)part[ipart - i],
899 i ? offcount + (i - 1) * PARTSIZE : 0);
900
901 if (seendigit && (*s == 'e' || *s == 'E')) {
902 bool expnegative = 0;
903
904 ++s;
905 switch (*s) {
906 case '-':
907 expnegative = 1;
908 /* fall through */
909 case '+':
910 ++s;
911 }
912 while (isDIGIT(*s))
913 exponent = exponent * 10 + (*s++ - '0');
914 if (expnegative)
915 exponent = -exponent;
916 }
917
918 /* now apply the exponent */
919 exponent += expextra;
920 result = S_mulexp10(result, exponent);
921
922 /* now apply the sign */
923 if (negative)
924 result = -result;
925 *value = result;
926 return s;
927}
928