xref: /dragonfly/lib/libc/stdtime/strptime.c (revision cfd1aba3)
1 /*
2  * Powerdog Industries kindly requests feedback from anyone modifying
3  * this function:
4  *
5  * Date: Thu, 05 Jun 1997 23:17:17 -0400
6  * From: Kevin Ruddy <kevin.ruddy@powerdog.com>
7  * To: James FitzGibbon <james@nexis.net>
8  * Subject: Re: Use of your strptime(3) code (fwd)
9  *
10  * The reason for the "no mod" clause was so that modifications would
11  * come back and we could integrate them and reissue so that a wider
12  * audience could use it (thereby spreading the wealth).  This has
13  * made it possible to get strptime to work on many operating systems.
14  * I'm not sure why that's "plain unacceptable" to the FreeBSD team.
15  *
16  * Anyway, you can change it to "with or without modification" as
17  * you see fit.  Enjoy.
18  *
19  * Kevin Ruddy
20  * Powerdog Industries, Inc.
21  */
22 /*
23  * Copyright (c) 1994 Powerdog Industries.  All rights reserved.
24  *
25  * Copyright (c) 2011 The FreeBSD Foundation
26  * All rights reserved.
27  * Portions of this software were developed by David Chisnall
28  * under sponsorship from the FreeBSD Foundation.
29  *
30  * Redistribution and use in source and binary forms, with or without
31  * modification, are permitted provided that the following conditions
32  * are met:
33  * 1. Redistributions of source code must retain the above copyright
34  *    notice, this list of conditions and the following disclaimer.
35  * 2. Redistributions in binary form must reproduce the above copyright
36  *    notice, this list of conditions and the following disclaimer
37  *    in the documentation and/or other materials provided with the
38  *    distribution.
39  * 3. All advertising materials mentioning features or use of this
40  *    software must display the following acknowledgement:
41  *      This product includes software developed by Powerdog Industries.
42  * 4. The name of Powerdog Industries may not be used to endorse or
43  *    promote products derived from this software without specific prior
44  *    written permission.
45  *
46  * THIS SOFTWARE IS PROVIDED BY POWERDOG INDUSTRIES ``AS IS'' AND ANY
47  * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
48  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
49  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE POWERDOG INDUSTRIES BE
50  * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
51  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
52  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
53  * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
54  * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
55  * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
56  * EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
57  *
58  * @(#)strptime.c	0.1 (Powerdog) 94/03/27
59  * @(#) Copyright (c) 1994 Powerdog Industries.  All rights reserved.
60  * $FreeBSD: head/lib/libc/stdtime/strptime.c 227753 2011-11-20 14:45:42Z theraven $
61  */
62 
63 #include "namespace.h"
64 #include <time.h>
65 #include <ctype.h>
66 #include <errno.h>
67 #include <stdlib.h>
68 #include <string.h>
69 #include <pthread.h>
70 #include "un-namespace.h"
71 #include "libc_private.h"
72 #include "timelocal.h"
73 
74 static char * _strptime(const char *, const char *, struct tm *, int *, locale_t);
75 
76 #define asizeof(a)	(sizeof (a) / sizeof ((a)[0]))
77 
78 static char *
79 _strptime(const char *buf, const char *fmt, struct tm *tm, int *GMTp,
80 		locale_t locale)
81 {
82 	char	c;
83 	const char *ptr;
84 	int	i,
85 		len;
86 	int Ealternative, Oalternative;
87 	struct lc_time_T *tptr = __get_current_time_locale(locale);
88 
89 	ptr = fmt;
90 	while (*ptr != 0) {
91 		if (*buf == 0)
92 			break;
93 
94 		c = *ptr++;
95 
96 		if (c != '%') {
97 			if (isspace_l((unsigned char)c, locale))
98 				while (*buf != 0 &&
99 				       isspace_l((unsigned char)*buf, locale))
100 					buf++;
101 			else if (c != *buf++)
102 				return 0;
103 			continue;
104 		}
105 
106 		Ealternative = 0;
107 		Oalternative = 0;
108 label:
109 		c = *ptr++;
110 		switch (c) {
111 		case 0:
112 		case '%':
113 			if (*buf++ != '%')
114 				return 0;
115 			break;
116 
117 		case '+':
118 			buf = _strptime(buf, tptr->date_fmt, tm, GMTp, locale);
119 			if (buf == 0)
120 				return 0;
121 			break;
122 
123 		case 'C':
124 			if (!isdigit_l((unsigned char)*buf, locale))
125 				return 0;
126 
127 			/* XXX This will break for 3-digit centuries. */
128 			len = 2;
129 			for (i = 0; len && *buf != 0 &&
130 			     isdigit_l((unsigned char)*buf, locale); buf++) {
131 				i *= 10;
132 				i += *buf - '0';
133 				len--;
134 			}
135 			if (i < 19)
136 				return 0;
137 
138 			tm->tm_year = i * 100 - 1900;
139 			break;
140 
141 		case 'c':
142 			buf = _strptime(buf, tptr->c_fmt, tm, GMTp, locale);
143 			if (buf == 0)
144 				return 0;
145 			break;
146 
147 		case 'D':
148 			buf = _strptime(buf, "%m/%d/%y", tm, GMTp, locale);
149 			if (buf == 0)
150 				return 0;
151 			break;
152 
153 		case 'E':
154 			if (Ealternative || Oalternative)
155 				break;
156 			Ealternative++;
157 			goto label;
158 
159 		case 'O':
160 			if (Ealternative || Oalternative)
161 				break;
162 			Oalternative++;
163 			goto label;
164 
165 		case 'F':
166 			buf = _strptime(buf, "%Y-%m-%d", tm, GMTp, locale);
167 			if (buf == 0)
168 				return 0;
169 			break;
170 
171 		case 'R':
172 			buf = _strptime(buf, "%H:%M", tm, GMTp, locale);
173 			if (buf == 0)
174 				return 0;
175 			break;
176 
177 		case 'r':
178 			buf = _strptime(buf, tptr->ampm_fmt, tm, GMTp, locale);
179 			if (buf == 0)
180 				return 0;
181 			break;
182 
183 		case 'T':
184 			buf = _strptime(buf, "%H:%M:%S", tm, GMTp, locale);
185 			if (buf == 0)
186 				return 0;
187 			break;
188 
189 		case 'X':
190 			buf = _strptime(buf, tptr->X_fmt, tm, GMTp, locale);
191 			if (buf == 0)
192 				return 0;
193 			break;
194 
195 		case 'x':
196 			buf = _strptime(buf, tptr->x_fmt, tm, GMTp, locale);
197 			if (buf == 0)
198 				return 0;
199 			break;
200 
201 		case 'j':
202 			if (!isdigit_l((unsigned char)*buf, locale))
203 				return 0;
204 
205 			len = 3;
206 			for (i = 0; len && *buf != 0 &&
207 			     isdigit_l((unsigned char)*buf, locale); buf++){
208 				i *= 10;
209 				i += *buf - '0';
210 				len--;
211 			}
212 			if (i < 1 || i > 366)
213 				return 0;
214 
215 			tm->tm_yday = i - 1;
216 			break;
217 
218 		case 'M':
219 		case 'S':
220 			if (*buf == 0 ||
221 				isspace_l((unsigned char)*buf, locale))
222 				break;
223 
224 			if (!isdigit_l((unsigned char)*buf, locale))
225 				return 0;
226 
227 			len = 2;
228 			for (i = 0; len && *buf != 0 &&
229 				isdigit_l((unsigned char)*buf, locale); buf++){
230 				i *= 10;
231 				i += *buf - '0';
232 				len--;
233 			}
234 
235 			if (c == 'M') {
236 				if (i > 59)
237 					return 0;
238 				tm->tm_min = i;
239 			} else {
240 				if (i > 60)
241 					return 0;
242 				tm->tm_sec = i;
243 			}
244 
245 			if (*buf != 0 &&
246 				isspace_l((unsigned char)*buf, locale))
247 				while (*ptr != 0 &&
248 				       !isspace_l((unsigned char)*ptr, locale))
249 					ptr++;
250 			break;
251 
252 		case 'H':
253 		case 'I':
254 		case 'k':
255 		case 'l':
256 			/*
257 			 * Of these, %l is the only specifier explicitly
258 			 * documented as not being zero-padded.  However,
259 			 * there is no harm in allowing zero-padding.
260 			 *
261 			 * XXX The %l specifier may gobble one too many
262 			 * digits if used incorrectly.
263 			 */
264 			if (!isdigit_l((unsigned char)*buf, locale))
265 				return 0;
266 
267 			len = 2;
268 			for (i = 0; len && *buf != 0 &&
269 			     isdigit_l((unsigned char)*buf, locale); buf++) {
270 				i *= 10;
271 				i += *buf - '0';
272 				len--;
273 			}
274 			if (c == 'H' || c == 'k') {
275 				if (i > 23)
276 					return 0;
277 			} else if (i > 12)
278 				return 0;
279 
280 			tm->tm_hour = i;
281 
282 			if (*buf != 0 &&
283 			    isspace_l((unsigned char)*buf, locale))
284 				while (*ptr != 0 &&
285 				       !isspace_l((unsigned char)*ptr, locale))
286 					ptr++;
287 			break;
288 
289 		case 'p':
290 			/*
291 			 * XXX This is bogus if parsed before hour-related
292 			 * specifiers.
293 			 */
294 			len = strlen(tptr->am);
295 			if (strncasecmp_l(buf, tptr->am, len, locale) == 0) {
296 				if (tm->tm_hour > 12)
297 					return 0;
298 				if (tm->tm_hour == 12)
299 					tm->tm_hour = 0;
300 				buf += len;
301 				break;
302 			}
303 
304 			len = strlen(tptr->pm);
305 			if (strncasecmp_l(buf, tptr->pm, len, locale) == 0) {
306 				if (tm->tm_hour > 12)
307 					return 0;
308 				if (tm->tm_hour != 12)
309 					tm->tm_hour += 12;
310 				buf += len;
311 				break;
312 			}
313 
314 			return 0;
315 
316 		case 'A':
317 		case 'a':
318 			for (i = 0; i < asizeof(tptr->weekday); i++) {
319 				len = strlen(tptr->weekday[i]);
320 				if (strncasecmp_l(buf, tptr->weekday[i],
321 						len, locale) == 0)
322 					break;
323 				len = strlen(tptr->wday[i]);
324 				if (strncasecmp_l(buf, tptr->wday[i],
325 						len, locale) == 0)
326 					break;
327 			}
328 			if (i == asizeof(tptr->weekday))
329 				return 0;
330 
331 			tm->tm_wday = i;
332 			buf += len;
333 			break;
334 
335 		case 'U':
336 		case 'W':
337 			/*
338 			 * XXX This is bogus, as we can not assume any valid
339 			 * information present in the tm structure at this
340 			 * point to calculate a real value, so just check the
341 			 * range for now.
342 			 */
343 			if (!isdigit_l((unsigned char)*buf, locale))
344 				return 0;
345 
346 			len = 2;
347 			for (i = 0; len && *buf != 0 &&
348 			     isdigit_l((unsigned char)*buf, locale); buf++) {
349 				i *= 10;
350 				i += *buf - '0';
351 				len--;
352 			}
353 			if (i > 53)
354 				return 0;
355 
356 			if (*buf != 0 &&
357 			    isspace_l((unsigned char)*buf, locale))
358 				while (*ptr != 0 &&
359 				       !isspace_l((unsigned char)*ptr, locale))
360 					ptr++;
361 			break;
362 
363 		case 'w':
364 			if (!isdigit_l((unsigned char)*buf, locale))
365 				return 0;
366 
367 			i = *buf - '0';
368 			if (i > 6)
369 				return 0;
370 
371 			tm->tm_wday = i;
372 
373 			if (*buf != 0 &&
374 			    isspace_l((unsigned char)*buf, locale))
375 				while (*ptr != 0 &&
376 				       !isspace_l((unsigned char)*ptr, locale))
377 					ptr++;
378 			break;
379 
380 		case 'd':
381 		case 'e':
382 			/*
383 			 * The %e specifier is explicitly documented as not
384 			 * being zero-padded but there is no harm in allowing
385 			 * such padding.
386 			 *
387 			 * XXX The %e specifier may gobble one too many
388 			 * digits if used incorrectly.
389 			 */
390 			if (!isdigit_l((unsigned char)*buf, locale))
391 				return 0;
392 
393 			len = 2;
394 			for (i = 0; len && *buf != 0 &&
395 			     isdigit_l((unsigned char)*buf, locale); buf++) {
396 				i *= 10;
397 				i += *buf - '0';
398 				len--;
399 			}
400 			if (i > 31)
401 				return 0;
402 
403 			tm->tm_mday = i;
404 
405 			if (*buf != 0 &&
406 			    isspace_l((unsigned char)*buf, locale))
407 				while (*ptr != 0 &&
408 				       !isspace_l((unsigned char)*ptr, locale))
409 					ptr++;
410 			break;
411 
412 		case 'B':
413 		case 'b':
414 		case 'h':
415 			for (i = 0; i < asizeof(tptr->month); i++) {
416 				if (Oalternative) {
417 					if (c == 'B') {
418 						len = strlen(tptr->alt_month[i]);
419 						if (strncasecmp_l(buf,
420 								tptr->alt_month[i],
421 								len, locale) == 0)
422 							break;
423 					}
424 				} else {
425 					len = strlen(tptr->month[i]);
426 					if (strncasecmp_l(buf, tptr->month[i],
427 							len, locale) == 0)
428 						break;
429 				}
430 			}
431 			/*
432 			 * Try the abbreviated month name if the full name
433 			 * wasn't found and Oalternative was not requested.
434 			 */
435 			if (i == asizeof(tptr->month) && !Oalternative) {
436 				for (i = 0; i < asizeof(tptr->month); i++) {
437 					len = strlen(tptr->mon[i]);
438 					if (strncasecmp_l(buf, tptr->mon[i],
439 							len, locale) == 0)
440 						break;
441 				}
442 			}
443 			if (i == asizeof(tptr->month))
444 				return 0;
445 
446 			tm->tm_mon = i;
447 			buf += len;
448 			break;
449 
450 		case 'm':
451 			if (!isdigit_l((unsigned char)*buf, locale))
452 				return 0;
453 
454 			len = 2;
455 			for (i = 0; len && *buf != 0 &&
456 			     isdigit_l((unsigned char)*buf, locale); buf++) {
457 				i *= 10;
458 				i += *buf - '0';
459 				len--;
460 			}
461 			if (i < 1 || i > 12)
462 				return 0;
463 
464 			tm->tm_mon = i - 1;
465 
466 			if (*buf != 0 &&
467 			    isspace_l((unsigned char)*buf, locale))
468 				while (*ptr != 0 &&
469 				       !isspace_l((unsigned char)*ptr, locale))
470 					ptr++;
471 			break;
472 
473 		case 's':
474 			{
475 			char *cp;
476 			int sverrno;
477 			long n;
478 			time_t t;
479 
480 			sverrno = errno;
481 			errno = 0;
482 			n = strtol_l(buf, &cp, 10, locale);
483 			if (errno == ERANGE || (long)(t = n) != n) {
484 				errno = sverrno;
485 				return 0;
486 			}
487 			errno = sverrno;
488 			buf = cp;
489 			gmtime_r(&t, tm);
490 			*GMTp = 1;
491 			}
492 			break;
493 
494 		case 'Y':
495 		case 'y':
496 			if (*buf == 0 ||
497 			    isspace_l((unsigned char)*buf, locale))
498 				break;
499 
500 			if (!isdigit_l((unsigned char)*buf, locale))
501 				return 0;
502 
503 			len = (c == 'Y') ? 4 : 2;
504 			for (i = 0; len && *buf != 0 &&
505 			     isdigit_l((unsigned char)*buf, locale); buf++) {
506 				i *= 10;
507 				i += *buf - '0';
508 				len--;
509 			}
510 			if (c == 'Y')
511 				i -= 1900;
512 			if (c == 'y' && i < 69)
513 				i += 100;
514 			if (i < 0)
515 				return 0;
516 
517 			tm->tm_year = i;
518 
519 			if (*buf != 0 &&
520 			    isspace_l((unsigned char)*buf, locale))
521 				while (*ptr != 0 &&
522 				       !isspace_l((unsigned char)*ptr, locale))
523 					ptr++;
524 			break;
525 
526 		case 'Z':
527 			{
528 			const char *cp;
529 			char *zonestr;
530 
531 			for (cp = buf; *cp &&
532 			     isupper_l((unsigned char)*cp, locale); ++cp) {
533 				/*empty*/}
534 			if (cp - buf) {
535 				zonestr = alloca(cp - buf + 1);
536 				strncpy(zonestr, buf, cp - buf);
537 				zonestr[cp - buf] = '\0';
538 				tzset();
539 				if (0 == strcmp(zonestr, "GMT")) {
540 				    *GMTp = 1;
541 				} else if (0 == strcmp(zonestr, tzname[0])) {
542 				    tm->tm_isdst = 0;
543 				} else if (0 == strcmp(zonestr, tzname[1])) {
544 				    tm->tm_isdst = 1;
545 				} else {
546 				    return 0;
547 				}
548 				buf += cp - buf;
549 			}
550 			}
551 			break;
552 
553 		case 'z':
554 			{
555 			int sign = 1;
556 
557 			if (*buf != '+') {
558 				if (*buf == '-')
559 					sign = -1;
560 				else
561 					return 0;
562 			}
563 
564 			buf++;
565 			i = 0;
566 			for (len = 4; len > 0; len--) {
567 				if (isdigit_l((unsigned char)*buf, locale)) {
568 					i *= 10;
569 					i += *buf - '0';
570 					buf++;
571 				} else
572 					return 0;
573 			}
574 
575 			tm->tm_hour -= sign * (i / 100);
576 			tm->tm_min  -= sign * (i % 100);
577 			*GMTp = 1;
578 			}
579 			break;
580 		}
581 	}
582 	return (char *)buf;
583 }
584 
585 
586 char *
587 strptime_l(const char * __restrict buf, const char * __restrict fmt,
588     struct tm * __restrict tm, locale_t loc)
589 {
590 	char *ret;
591 	int gmt;
592 	FIX_LOCALE(loc);
593 
594 	gmt = 0;
595 	ret = _strptime(buf, fmt, tm, &gmt, loc);
596 	if (ret && gmt) {
597 		time_t t = timegm(tm);
598 		localtime_r(&t, tm);
599 	}
600 
601 	return (ret);
602 }
603 char *
604 strptime(const char * __restrict buf, const char * __restrict fmt,
605     struct tm * __restrict tm)
606 {
607 	return strptime_l(buf, fmt, tm, __get_locale());
608 }
609