1 /* $NetBSD: strptime.c,v 1.49 2015/10/09 17:21:45 christos Exp $ */
2
3 /*-
4 * Copyright (c) 1997, 1998, 2005, 2008 The NetBSD Foundation, Inc.
5 * All rights reserved.
6 *
7 * This code was contributed to The NetBSD Foundation by Klaus Klein.
8 * Heavily optimised by David Laight
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
20 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
21 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
22 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
23 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
24 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
25 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
26 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
27 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
28 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
29 * POSSIBILITY OF SUCH DAMAGE.
30 */
31
32 #include <sys/cdefs.h>
33 #if defined(LIBC_SCCS) && !defined(lint)
34 __RCSID("$NetBSD: strptime.c,v 1.49 2015/10/09 17:21:45 christos Exp $");
35 #endif
36
37 #include "namespace.h"
38 #include <sys/localedef.h>
39 #include <sys/types.h>
40 #include <ctype.h>
41 #include <locale.h>
42 #include <string.h>
43 #include <time.h>
44 #include <tzfile.h>
45 #include "private.h"
46 #include "setlocale_local.h"
47
48 #ifdef __weak_alias
49 __weak_alias(strptime,_strptime)
50 __weak_alias(strptime_l, _strptime_l)
51 #endif
52
53 static const u_char *conv_num(const unsigned char *, int *, uint, uint);
54 static const u_char *find_string(const u_char *, int *, const char * const *,
55 const char * const *, int);
56
57 #define _TIME_LOCALE(loc) \
58 ((_TimeLocale *)((loc)->part_impl[(size_t)LC_TIME]))
59
60 /*
61 * We do not implement alternate representations. However, we always
62 * check whether a given modifier is allowed for a certain conversion.
63 */
64 #define ALT_E 0x01
65 #define ALT_O 0x02
66 #define LEGAL_ALT(x) { if (alt_format & ~(x)) return NULL; }
67
68 #define S_YEAR (1 << 0)
69 #define S_MON (1 << 1)
70 #define S_YDAY (1 << 2)
71 #define S_MDAY (1 << 3)
72 #define S_WDAY (1 << 4)
73 #define S_HOUR (1 << 5)
74
75 #define HAVE_MDAY(s) (s & S_MDAY)
76 #define HAVE_MON(s) (s & S_MON)
77 #define HAVE_WDAY(s) (s & S_WDAY)
78 #define HAVE_YDAY(s) (s & S_YDAY)
79 #define HAVE_YEAR(s) (s & S_YEAR)
80 #define HAVE_HOUR(s) (s & S_HOUR)
81
82 static char gmt[] = { "GMT" };
83 static char utc[] = { "UTC" };
84 /* RFC-822/RFC-2822 */
85 static const char * const nast[5] = {
86 "EST", "CST", "MST", "PST", "\0\0\0"
87 };
88 static const char * const nadt[5] = {
89 "EDT", "CDT", "MDT", "PDT", "\0\0\0"
90 };
91
92 /*
93 * Table to determine the ordinal date for the start of a month.
94 * Ref: http://en.wikipedia.org/wiki/ISO_week_date
95 */
96 static const int start_of_month[2][13] = {
97 /* non-leap year */
98 { 0, 31, 59, 90, 120, 151, 181, 212, 243, 273, 304, 334, 365 },
99 /* leap year */
100 { 0, 31, 60, 91, 121, 152, 182, 213, 244, 274, 305, 335, 366 }
101 };
102
103 /*
104 * Calculate the week day of the first day of a year. Valid for
105 * the Gregorian calendar, which began Sept 14, 1752 in the UK
106 * and its colonies. Ref:
107 * http://en.wikipedia.org/wiki/Determination_of_the_day_of_the_week
108 */
109
110 static int
first_wday_of(int yr)111 first_wday_of(int yr)
112 {
113 return ((2 * (3 - (yr / 100) % 4)) + (yr % 100) + ((yr % 100) / 4) +
114 (isleap(yr) ? 6 : 0) + 1) % 7;
115 }
116
117 char *
strptime(const char * buf,const char * fmt,struct tm * tm)118 strptime(const char *buf, const char *fmt, struct tm *tm)
119 {
120 return strptime_l(buf, fmt, tm, _current_locale());
121 }
122
123 char *
strptime_l(const char * buf,const char * fmt,struct tm * tm,locale_t loc)124 strptime_l(const char *buf, const char *fmt, struct tm *tm, locale_t loc)
125 {
126 unsigned char c;
127 const unsigned char *bp, *ep;
128 int alt_format, i, split_year = 0, neg = 0, state = 0,
129 day_offset = -1, week_offset = 0, offs;
130 const char *new_fmt;
131
132 bp = (const u_char *)buf;
133
134 while (bp != NULL && (c = *fmt++) != '\0') {
135 /* Clear `alternate' modifier prior to new conversion. */
136 alt_format = 0;
137 i = 0;
138
139 /* Eat up white-space. */
140 if (isspace(c)) {
141 while (isspace(*bp))
142 bp++;
143 continue;
144 }
145
146 if (c != '%')
147 goto literal;
148
149
150 again: switch (c = *fmt++) {
151 case '%': /* "%%" is converted to "%". */
152 literal:
153 if (c != *bp++)
154 return NULL;
155 LEGAL_ALT(0);
156 continue;
157
158 /*
159 * "Alternative" modifiers. Just set the appropriate flag
160 * and start over again.
161 */
162 case 'E': /* "%E?" alternative conversion modifier. */
163 LEGAL_ALT(0);
164 alt_format |= ALT_E;
165 goto again;
166
167 case 'O': /* "%O?" alternative conversion modifier. */
168 LEGAL_ALT(0);
169 alt_format |= ALT_O;
170 goto again;
171
172 /*
173 * "Complex" conversion rules, implemented through recursion.
174 */
175 case 'c': /* Date and time, using the locale's format. */
176 new_fmt = _TIME_LOCALE(loc)->d_t_fmt;
177 state |= S_WDAY | S_MON | S_MDAY | S_YEAR;
178 goto recurse;
179
180 case 'D': /* The date as "%m/%d/%y". */
181 new_fmt = "%m/%d/%y";
182 LEGAL_ALT(0);
183 state |= S_MON | S_MDAY | S_YEAR;
184 goto recurse;
185
186 case 'F': /* The date as "%Y-%m-%d". */
187 new_fmt = "%Y-%m-%d";
188 LEGAL_ALT(0);
189 state |= S_MON | S_MDAY | S_YEAR;
190 goto recurse;
191
192 case 'R': /* The time as "%H:%M". */
193 new_fmt = "%H:%M";
194 LEGAL_ALT(0);
195 goto recurse;
196
197 case 'r': /* The time in 12-hour clock representation. */
198 new_fmt = _TIME_LOCALE(loc)->t_fmt_ampm;
199 LEGAL_ALT(0);
200 goto recurse;
201
202 case 'T': /* The time as "%H:%M:%S". */
203 new_fmt = "%H:%M:%S";
204 LEGAL_ALT(0);
205 goto recurse;
206
207 case 'X': /* The time, using the locale's format. */
208 new_fmt = _TIME_LOCALE(loc)->t_fmt;
209 goto recurse;
210
211 case 'x': /* The date, using the locale's format. */
212 new_fmt = _TIME_LOCALE(loc)->d_fmt;
213 state |= S_MON | S_MDAY | S_YEAR;
214 recurse:
215 bp = (const u_char *)strptime((const char *)bp,
216 new_fmt, tm);
217 LEGAL_ALT(ALT_E);
218 continue;
219
220 /*
221 * "Elementary" conversion rules.
222 */
223 case 'A': /* The day of week, using the locale's form. */
224 case 'a':
225 bp = find_string(bp, &tm->tm_wday,
226 _TIME_LOCALE(loc)->day, _TIME_LOCALE(loc)->abday, 7);
227 LEGAL_ALT(0);
228 state |= S_WDAY;
229 continue;
230
231 case 'B': /* The month, using the locale's form. */
232 case 'b':
233 case 'h':
234 bp = find_string(bp, &tm->tm_mon,
235 _TIME_LOCALE(loc)->mon, _TIME_LOCALE(loc)->abmon,
236 12);
237 LEGAL_ALT(0);
238 state |= S_MON;
239 continue;
240
241 case 'C': /* The century number. */
242 i = 20;
243 bp = conv_num(bp, &i, 0, 99);
244
245 i = i * 100 - TM_YEAR_BASE;
246 if (split_year)
247 i += tm->tm_year % 100;
248 split_year = 1;
249 tm->tm_year = i;
250 LEGAL_ALT(ALT_E);
251 state |= S_YEAR;
252 continue;
253
254 case 'd': /* The day of month. */
255 case 'e':
256 bp = conv_num(bp, &tm->tm_mday, 1, 31);
257 LEGAL_ALT(ALT_O);
258 state |= S_MDAY;
259 continue;
260
261 case 'k': /* The hour (24-hour clock representation). */
262 LEGAL_ALT(0);
263 /* FALLTHROUGH */
264 case 'H':
265 bp = conv_num(bp, &tm->tm_hour, 0, 23);
266 LEGAL_ALT(ALT_O);
267 state |= S_HOUR;
268 continue;
269
270 case 'l': /* The hour (12-hour clock representation). */
271 LEGAL_ALT(0);
272 /* FALLTHROUGH */
273 case 'I':
274 bp = conv_num(bp, &tm->tm_hour, 1, 12);
275 if (tm->tm_hour == 12)
276 tm->tm_hour = 0;
277 LEGAL_ALT(ALT_O);
278 state |= S_HOUR;
279 continue;
280
281 case 'j': /* The day of year. */
282 i = 1;
283 bp = conv_num(bp, &i, 1, 366);
284 tm->tm_yday = i - 1;
285 LEGAL_ALT(0);
286 state |= S_YDAY;
287 continue;
288
289 case 'M': /* The minute. */
290 bp = conv_num(bp, &tm->tm_min, 0, 59);
291 LEGAL_ALT(ALT_O);
292 continue;
293
294 case 'm': /* The month. */
295 i = 1;
296 bp = conv_num(bp, &i, 1, 12);
297 tm->tm_mon = i - 1;
298 LEGAL_ALT(ALT_O);
299 state |= S_MON;
300 continue;
301
302 case 'p': /* The locale's equivalent of AM/PM. */
303 bp = find_string(bp, &i, _TIME_LOCALE(loc)->am_pm,
304 NULL, 2);
305 if (HAVE_HOUR(state) && tm->tm_hour > 11)
306 return NULL;
307 tm->tm_hour += i * 12;
308 LEGAL_ALT(0);
309 continue;
310
311 case 'S': /* The seconds. */
312 bp = conv_num(bp, &tm->tm_sec, 0, 61);
313 LEGAL_ALT(ALT_O);
314 continue;
315
316 #ifndef TIME_MAX
317 #define TIME_MAX INT64_MAX
318 #endif
319 case 's': /* seconds since the epoch */
320 {
321 time_t sse = 0;
322 uint64_t rulim = TIME_MAX;
323
324 if (*bp < '0' || *bp > '9') {
325 bp = NULL;
326 continue;
327 }
328
329 do {
330 sse *= 10;
331 sse += *bp++ - '0';
332 rulim /= 10;
333 } while ((sse * 10 <= TIME_MAX) &&
334 rulim && *bp >= '0' && *bp <= '9');
335
336 if (sse < 0 || (uint64_t)sse > TIME_MAX) {
337 bp = NULL;
338 continue;
339 }
340
341 if (localtime_r(&sse, tm) == NULL)
342 bp = NULL;
343 else
344 state |= S_YDAY | S_WDAY |
345 S_MON | S_MDAY | S_YEAR;
346 }
347 continue;
348
349 case 'U': /* The week of year, beginning on sunday. */
350 case 'W': /* The week of year, beginning on monday. */
351 /*
352 * XXX This is bogus, as we can not assume any valid
353 * information present in the tm structure at this
354 * point to calculate a real value, so just check the
355 * range for now.
356 */
357 bp = conv_num(bp, &i, 0, 53);
358 LEGAL_ALT(ALT_O);
359 if (c == 'U')
360 day_offset = TM_SUNDAY;
361 else
362 day_offset = TM_MONDAY;
363 week_offset = i;
364 continue;
365
366 case 'w': /* The day of week, beginning on sunday. */
367 bp = conv_num(bp, &tm->tm_wday, 0, 6);
368 LEGAL_ALT(ALT_O);
369 state |= S_WDAY;
370 continue;
371
372 case 'u': /* The day of week, monday = 1. */
373 bp = conv_num(bp, &i, 1, 7);
374 tm->tm_wday = i % 7;
375 LEGAL_ALT(ALT_O);
376 state |= S_WDAY;
377 continue;
378
379 case 'g': /* The year corresponding to the ISO week
380 * number but without the century.
381 */
382 bp = conv_num(bp, &i, 0, 99);
383 continue;
384
385 case 'G': /* The year corresponding to the ISO week
386 * number with century.
387 */
388 do
389 bp++;
390 while (isdigit(*bp));
391 continue;
392
393 case 'V': /* The ISO 8601:1988 week number as decimal */
394 bp = conv_num(bp, &i, 0, 53);
395 continue;
396
397 case 'Y': /* The year. */
398 i = TM_YEAR_BASE; /* just for data sanity... */
399 bp = conv_num(bp, &i, 0, 9999);
400 tm->tm_year = i - TM_YEAR_BASE;
401 LEGAL_ALT(ALT_E);
402 state |= S_YEAR;
403 continue;
404
405 case 'y': /* The year within 100 years of the epoch. */
406 /* LEGAL_ALT(ALT_E | ALT_O); */
407 bp = conv_num(bp, &i, 0, 99);
408
409 if (split_year)
410 /* preserve century */
411 i += (tm->tm_year / 100) * 100;
412 else {
413 split_year = 1;
414 if (i <= 68)
415 i = i + 2000 - TM_YEAR_BASE;
416 else
417 i = i + 1900 - TM_YEAR_BASE;
418 }
419 tm->tm_year = i;
420 state |= S_YEAR;
421 continue;
422
423 case 'Z':
424 tzset();
425 if (strncmp((const char *)bp, gmt, 3) == 0 ||
426 strncmp((const char *)bp, utc, 3) == 0) {
427 tm->tm_isdst = 0;
428 #ifdef TM_GMTOFF
429 tm->TM_GMTOFF = 0;
430 #endif
431 #ifdef TM_ZONE
432 tm->TM_ZONE = gmt;
433 #endif
434 bp += 3;
435 } else {
436 ep = find_string(bp, &i,
437 (const char * const *)tzname,
438 NULL, 2);
439 if (ep != NULL) {
440 tm->tm_isdst = i;
441 #ifdef TM_GMTOFF
442 tm->TM_GMTOFF = -(timezone);
443 #endif
444 #ifdef TM_ZONE
445 tm->TM_ZONE = tzname[i];
446 #endif
447 }
448 bp = ep;
449 }
450 continue;
451
452 case 'z':
453 /*
454 * We recognize all ISO 8601 formats:
455 * Z = Zulu time/UTC
456 * [+-]hhmm
457 * [+-]hh:mm
458 * [+-]hh
459 * We recognize all RFC-822/RFC-2822 formats:
460 * UT|GMT
461 * North American : UTC offsets
462 * E[DS]T = Eastern : -4 | -5
463 * C[DS]T = Central : -5 | -6
464 * M[DS]T = Mountain: -6 | -7
465 * P[DS]T = Pacific : -7 | -8
466 * Military
467 * [A-IL-M] = -1 ... -9 (J not used)
468 * [N-Y] = +1 ... +12
469 */
470 while (isspace(*bp))
471 bp++;
472
473 switch (*bp++) {
474 case 'G':
475 if (*bp++ != 'M')
476 return NULL;
477 /*FALLTHROUGH*/
478 case 'U':
479 if (*bp++ != 'T')
480 return NULL;
481 /*FALLTHROUGH*/
482 case 'Z':
483 tm->tm_isdst = 0;
484 #ifdef TM_GMTOFF
485 tm->TM_GMTOFF = 0;
486 #endif
487 #ifdef TM_ZONE
488 tm->TM_ZONE = utc;
489 #endif
490 continue;
491 case '+':
492 neg = 0;
493 break;
494 case '-':
495 neg = 1;
496 break;
497 default:
498 --bp;
499 ep = find_string(bp, &i, nast, NULL, 4);
500 if (ep != NULL) {
501 #ifdef TM_GMTOFF
502 tm->TM_GMTOFF = -5 - i;
503 #endif
504 #ifdef TM_ZONE
505 tm->TM_ZONE = __UNCONST(nast[i]);
506 #endif
507 bp = ep;
508 continue;
509 }
510 ep = find_string(bp, &i, nadt, NULL, 4);
511 if (ep != NULL) {
512 tm->tm_isdst = 1;
513 #ifdef TM_GMTOFF
514 tm->TM_GMTOFF = -4 - i;
515 #endif
516 #ifdef TM_ZONE
517 tm->TM_ZONE = __UNCONST(nadt[i]);
518 #endif
519 bp = ep;
520 continue;
521 }
522
523 if ((*bp >= 'A' && *bp <= 'I') ||
524 (*bp >= 'L' && *bp <= 'Y')) {
525 #ifdef TM_GMTOFF
526 /* Argh! No 'J'! */
527 if (*bp >= 'A' && *bp <= 'I')
528 tm->TM_GMTOFF =
529 ('A' - 1) - (int)*bp;
530 else if (*bp >= 'L' && *bp <= 'M')
531 tm->TM_GMTOFF = 'A' - (int)*bp;
532 else if (*bp >= 'N' && *bp <= 'Y')
533 tm->TM_GMTOFF = (int)*bp - 'M';
534 #endif
535 #ifdef TM_ZONE
536 tm->TM_ZONE = utc; /* XXX */
537 #endif
538 bp++;
539 continue;
540 }
541 return NULL;
542 }
543 offs = 0;
544 for (i = 0; i < 4; ) {
545 if (isdigit(*bp)) {
546 offs = offs * 10 + (*bp++ - '0');
547 i++;
548 continue;
549 }
550 if (i == 2 && *bp == ':') {
551 bp++;
552 continue;
553 }
554 break;
555 }
556 switch (i) {
557 case 2:
558 offs *= 100;
559 break;
560 case 4:
561 i = offs % 100;
562 if (i >= 60)
563 return NULL;
564 /* Convert minutes into decimal */
565 offs = (offs / 100) * 100 + (i * 50) / 30;
566 break;
567 default:
568 return NULL;
569 }
570 if (neg)
571 offs = -offs;
572 tm->tm_isdst = 0; /* XXX */
573 #ifdef TM_GMTOFF
574 tm->TM_GMTOFF = offs;
575 #endif
576 #ifdef TM_ZONE
577 tm->TM_ZONE = utc; /* XXX */
578 #endif
579 continue;
580
581 /*
582 * Miscellaneous conversions.
583 */
584 case 'n': /* Any kind of white-space. */
585 case 't':
586 while (isspace(*bp))
587 bp++;
588 LEGAL_ALT(0);
589 continue;
590
591
592 default: /* Unknown/unsupported conversion. */
593 return NULL;
594 }
595 }
596
597 if (!HAVE_YDAY(state) && HAVE_YEAR(state)) {
598 if (HAVE_MON(state) && HAVE_MDAY(state)) {
599 /* calculate day of year (ordinal date) */
600 tm->tm_yday = start_of_month[isleap_sum(tm->tm_year,
601 TM_YEAR_BASE)][tm->tm_mon] + (tm->tm_mday - 1);
602 state |= S_YDAY;
603 } else if (day_offset != -1) {
604 /*
605 * Set the date to the first Sunday (or Monday)
606 * of the specified week of the year.
607 */
608 if (!HAVE_WDAY(state)) {
609 tm->tm_wday = day_offset;
610 state |= S_WDAY;
611 }
612 tm->tm_yday = (7 -
613 first_wday_of(tm->tm_year + TM_YEAR_BASE) +
614 day_offset) % 7 + (week_offset - 1) * 7 +
615 tm->tm_wday - day_offset;
616 state |= S_YDAY;
617 }
618 }
619
620 if (HAVE_YDAY(state) && HAVE_YEAR(state)) {
621 int isleap;
622
623 if (!HAVE_MON(state)) {
624 /* calculate month of day of year */
625 i = 0;
626 isleap = isleap_sum(tm->tm_year, TM_YEAR_BASE);
627 while (tm->tm_yday >= start_of_month[isleap][i])
628 i++;
629 if (i > 12) {
630 i = 1;
631 tm->tm_yday -= start_of_month[isleap][12];
632 tm->tm_year++;
633 }
634 tm->tm_mon = i - 1;
635 state |= S_MON;
636 }
637
638 if (!HAVE_MDAY(state)) {
639 /* calculate day of month */
640 isleap = isleap_sum(tm->tm_year, TM_YEAR_BASE);
641 tm->tm_mday = tm->tm_yday -
642 start_of_month[isleap][tm->tm_mon] + 1;
643 state |= S_MDAY;
644 }
645
646 if (!HAVE_WDAY(state)) {
647 /* calculate day of week */
648 i = 0;
649 week_offset = first_wday_of(tm->tm_year);
650 while (i++ <= tm->tm_yday) {
651 if (week_offset++ >= 6)
652 week_offset = 0;
653 }
654 tm->tm_wday = week_offset;
655 state |= S_WDAY;
656 }
657 }
658
659 return __UNCONST(bp);
660 }
661
662
663 static const u_char *
conv_num(const unsigned char * buf,int * dest,uint llim,uint ulim)664 conv_num(const unsigned char *buf, int *dest, uint llim, uint ulim)
665 {
666 uint result = 0;
667 unsigned char ch;
668
669 /* The limit also determines the number of valid digits. */
670 uint rulim = ulim;
671
672 ch = *buf;
673 if (ch < '0' || ch > '9')
674 return NULL;
675
676 do {
677 result *= 10;
678 result += ch - '0';
679 rulim /= 10;
680 ch = *++buf;
681 } while ((result * 10 <= ulim) && rulim && ch >= '0' && ch <= '9');
682
683 if (result < llim || result > ulim)
684 return NULL;
685
686 *dest = result;
687 return buf;
688 }
689
690 static const u_char *
find_string(const u_char * bp,int * tgt,const char * const * n1,const char * const * n2,int c)691 find_string(const u_char *bp, int *tgt, const char * const *n1,
692 const char * const *n2, int c)
693 {
694 int i;
695 size_t len;
696
697 /* check full name - then abbreviated ones */
698 for (; n1 != NULL; n1 = n2, n2 = NULL) {
699 for (i = 0; i < c; i++, n1++) {
700 len = strlen(*n1);
701 if (strncasecmp(*n1, (const char *)bp, len) == 0) {
702 *tgt = i;
703 return bp + len;
704 }
705 }
706 }
707
708 /* Nothing matched */
709 return NULL;
710 }
711