xref: /dragonfly/lib/libc/stdtime/strptime.c (revision 984263bc)
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  * Redistribution and use in source and binary forms, with or without
26  * modification, are permitted provided that the following conditions
27  * are met:
28  * 1. Redistributions of source code must retain the above copyright
29  *    notice, this list of conditions and the following disclaimer.
30  * 2. Redistributions in binary form must reproduce the above copyright
31  *    notice, this list of conditions and the following disclaimer
32  *    in the documentation and/or other materials provided with the
33  *    distribution.
34  * 3. All advertising materials mentioning features or use of this
35  *    software must display the following acknowledgement:
36  *      This product includes software developed by Powerdog Industries.
37  * 4. The name of Powerdog Industries may not be used to endorse or
38  *    promote products derived from this software without specific prior
39  *    written permission.
40  *
41  * THIS SOFTWARE IS PROVIDED BY POWERDOG INDUSTRIES ``AS IS'' AND ANY
42  * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
44  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE POWERDOG INDUSTRIES BE
45  * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
46  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
47  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
48  * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
49  * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
50  * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
51  * EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
52  */
53 
54 #ifdef LIBC_RCS
55 static const char rcsid[] =
56   "$FreeBSD: src/lib/libc/stdtime/strptime.c,v 1.17.2.3 2002/03/12 17:24:54 phantom Exp $";
57 #endif
58 
59 #ifndef lint
60 #ifndef NOID
61 static char copyright[] =
62 "@(#) Copyright (c) 1994 Powerdog Industries.  All rights reserved.";
63 static char sccsid[] = "@(#)strptime.c	0.1 (Powerdog) 94/03/27";
64 #endif /* !defined NOID */
65 #endif /* not lint */
66 
67 #include <time.h>
68 #include <ctype.h>
69 #include <limits.h>
70 #include <stdlib.h>
71 #include <string.h>
72 #ifdef	_THREAD_SAFE
73 #include <pthread.h>
74 #include "pthread_private.h"
75 #endif
76 #include "timelocal.h"
77 
78 static char * _strptime(const char *, const char *, struct tm *);
79 
80 #ifdef	_THREAD_SAFE
81 static struct pthread_mutex	_gotgmt_mutexd = PTHREAD_MUTEX_STATIC_INITIALIZER;
82 static pthread_mutex_t		gotgmt_mutex   = &_gotgmt_mutexd;
83 #endif
84 static int got_GMT;
85 
86 #define asizeof(a)	(sizeof (a) / sizeof ((a)[0]))
87 
88 static char *
89 _strptime(const char *buf, const char *fmt, struct tm *tm)
90 {
91 	char	c;
92 	const char *ptr;
93 	int	i,
94 		len;
95 	int Ealternative, Oalternative;
96 	struct lc_time_T *tptr = __get_current_time_locale();
97 
98 	ptr = fmt;
99 	while (*ptr != 0) {
100 		if (*buf == 0)
101 			break;
102 
103 		c = *ptr++;
104 
105 		if (c != '%') {
106 			if (isspace((unsigned char)c))
107 				while (*buf != 0 && isspace((unsigned char)*buf))
108 					buf++;
109 			else if (c != *buf++)
110 				return 0;
111 			continue;
112 		}
113 
114 		Ealternative = 0;
115 		Oalternative = 0;
116 label:
117 		c = *ptr++;
118 		switch (c) {
119 		case 0:
120 		case '%':
121 			if (*buf++ != '%')
122 				return 0;
123 			break;
124 
125 		case '+':
126 			buf = _strptime(buf, tptr->date_fmt, tm);
127 			if (buf == 0)
128 				return 0;
129 			break;
130 
131 		case 'C':
132 			if (!isdigit((unsigned char)*buf))
133 				return 0;
134 
135 			/* XXX This will break for 3-digit centuries. */
136 			len = 2;
137 			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
138 				i *= 10;
139 				i += *buf - '0';
140 				len--;
141 			}
142 			if (i < 19)
143 				return 0;
144 
145 			tm->tm_year = i * 100 - 1900;
146 			break;
147 
148 		case 'c':
149 			buf = _strptime(buf, tptr->c_fmt, tm);
150 			if (buf == 0)
151 				return 0;
152 			break;
153 
154 		case 'D':
155 			buf = _strptime(buf, "%m/%d/%y", tm);
156 			if (buf == 0)
157 				return 0;
158 			break;
159 
160 		case 'E':
161 			if (Ealternative || Oalternative)
162 				break;
163 			Ealternative++;
164 			goto label;
165 
166 		case 'O':
167 			if (Ealternative || Oalternative)
168 				break;
169 			Oalternative++;
170 			goto label;
171 
172 		case 'F':
173 			buf = _strptime(buf, "%Y-%m-%d", tm);
174 			if (buf == 0)
175 				return 0;
176 			break;
177 
178 		case 'R':
179 			buf = _strptime(buf, "%H:%M", tm);
180 			if (buf == 0)
181 				return 0;
182 			break;
183 
184 		case 'r':
185 			buf = _strptime(buf, tptr->ampm_fmt, tm);
186 			if (buf == 0)
187 				return 0;
188 			break;
189 
190 		case 'T':
191 			buf = _strptime(buf, "%H:%M:%S", tm);
192 			if (buf == 0)
193 				return 0;
194 			break;
195 
196 		case 'X':
197 			buf = _strptime(buf, tptr->X_fmt, tm);
198 			if (buf == 0)
199 				return 0;
200 			break;
201 
202 		case 'x':
203 			buf = _strptime(buf, tptr->x_fmt, tm);
204 			if (buf == 0)
205 				return 0;
206 			break;
207 
208 		case 'j':
209 			if (!isdigit((unsigned char)*buf))
210 				return 0;
211 
212 			len = 3;
213 			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
214 				i *= 10;
215 				i += *buf - '0';
216 				len--;
217 			}
218 			if (i < 1 || i > 366)
219 				return 0;
220 
221 			tm->tm_yday = i - 1;
222 			break;
223 
224 		case 'M':
225 		case 'S':
226 			if (*buf == 0 || isspace((unsigned char)*buf))
227 				break;
228 
229 			if (!isdigit((unsigned char)*buf))
230 				return 0;
231 
232 			len = 2;
233 			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
234 				i *= 10;
235 				i += *buf - '0';
236 				len--;
237 			}
238 
239 			if (c == 'M') {
240 				if (i > 59)
241 					return 0;
242 				tm->tm_min = i;
243 			} else {
244 				if (i > 60)
245 					return 0;
246 				tm->tm_sec = i;
247 			}
248 
249 			if (*buf != 0 && isspace((unsigned char)*buf))
250 				while (*ptr != 0 && !isspace((unsigned char)*ptr))
251 					ptr++;
252 			break;
253 
254 		case 'H':
255 		case 'I':
256 		case 'k':
257 		case 'l':
258 			/*
259 			 * Of these, %l is the only specifier explicitly
260 			 * documented as not being zero-padded.  However,
261 			 * there is no harm in allowing zero-padding.
262 			 *
263 			 * XXX The %l specifier may gobble one too many
264 			 * digits if used incorrectly.
265 			 */
266 			if (!isdigit((unsigned char)*buf))
267 				return 0;
268 
269 			len = 2;
270 			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
271 				i *= 10;
272 				i += *buf - '0';
273 				len--;
274 			}
275 			if (c == 'H' || c == 'k') {
276 				if (i > 23)
277 					return 0;
278 			} else if (i > 12)
279 				return 0;
280 
281 			tm->tm_hour = i;
282 
283 			if (*buf != 0 && isspace((unsigned char)*buf))
284 				while (*ptr != 0 && !isspace((unsigned char)*ptr))
285 					ptr++;
286 			break;
287 
288 		case 'p':
289 			/*
290 			 * XXX This is bogus if parsed before hour-related
291 			 * specifiers.
292 			 */
293 			len = strlen(tptr->am);
294 			if (strncasecmp(buf, tptr->am, len) == 0) {
295 				if (tm->tm_hour > 12)
296 					return 0;
297 				if (tm->tm_hour == 12)
298 					tm->tm_hour = 0;
299 				buf += len;
300 				break;
301 			}
302 
303 			len = strlen(tptr->pm);
304 			if (strncasecmp(buf, tptr->pm, len) == 0) {
305 				if (tm->tm_hour > 12)
306 					return 0;
307 				if (tm->tm_hour != 12)
308 					tm->tm_hour += 12;
309 				buf += len;
310 				break;
311 			}
312 
313 			return 0;
314 
315 		case 'A':
316 		case 'a':
317 			for (i = 0; i < asizeof(tptr->weekday); i++) {
318 				len = strlen(tptr->weekday[i]);
319 				if (strncasecmp(buf, tptr->weekday[i],
320 						len) == 0)
321 					break;
322 				len = strlen(tptr->wday[i]);
323 				if (strncasecmp(buf, tptr->wday[i],
324 						len) == 0)
325 					break;
326 			}
327 			if (i == asizeof(tptr->weekday))
328 				return 0;
329 
330 			tm->tm_wday = i;
331 			buf += len;
332 			break;
333 
334 		case 'U':
335 		case 'W':
336 			/*
337 			 * XXX This is bogus, as we can not assume any valid
338 			 * information present in the tm structure at this
339 			 * point to calculate a real value, so just check the
340 			 * range for now.
341 			 */
342 			if (!isdigit((unsigned char)*buf))
343 				return 0;
344 
345 			len = 2;
346 			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
347 				i *= 10;
348 				i += *buf - '0';
349 				len--;
350 			}
351 			if (i > 53)
352 				return 0;
353 
354 			if (*buf != 0 && isspace((unsigned char)*buf))
355 				while (*ptr != 0 && !isspace((unsigned char)*ptr))
356 					ptr++;
357 			break;
358 
359 		case 'w':
360 			if (!isdigit((unsigned char)*buf))
361 				return 0;
362 
363 			i = *buf - '0';
364 			if (i > 6)
365 				return 0;
366 
367 			tm->tm_wday = i;
368 
369 			if (*buf != 0 && isspace((unsigned char)*buf))
370 				while (*ptr != 0 && !isspace((unsigned char)*ptr))
371 					ptr++;
372 			break;
373 
374 		case 'd':
375 		case 'e':
376 			/*
377 			 * The %e specifier is explicitly documented as not
378 			 * being zero-padded but there is no harm in allowing
379 			 * such padding.
380 			 *
381 			 * XXX The %e specifier may gobble one too many
382 			 * digits if used incorrectly.
383 			 */
384 			if (!isdigit((unsigned char)*buf))
385 				return 0;
386 
387 			len = 2;
388 			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
389 				i *= 10;
390 				i += *buf - '0';
391 				len--;
392 			}
393 			if (i > 31)
394 				return 0;
395 
396 			tm->tm_mday = i;
397 
398 			if (*buf != 0 && isspace((unsigned char)*buf))
399 				while (*ptr != 0 && !isspace((unsigned char)*ptr))
400 					ptr++;
401 			break;
402 
403 		case 'B':
404 		case 'b':
405 		case 'h':
406 			for (i = 0; i < asizeof(tptr->month); i++) {
407 				if (Oalternative) {
408 					if (c == 'B') {
409 						len = strlen(tptr->alt_month[i]);
410 						if (strncasecmp(buf,
411 								tptr->alt_month[i],
412 								len) == 0)
413 							break;
414 					}
415 				} else {
416 					len = strlen(tptr->month[i]);
417 					if (strncasecmp(buf, tptr->month[i],
418 							len) == 0)
419 						break;
420 					len = strlen(tptr->mon[i]);
421 					if (strncasecmp(buf, tptr->mon[i],
422 							len) == 0)
423 						break;
424 				}
425 			}
426 			if (i == asizeof(tptr->month))
427 				return 0;
428 
429 			tm->tm_mon = i;
430 			buf += len;
431 			break;
432 
433 		case 'm':
434 			if (!isdigit((unsigned char)*buf))
435 				return 0;
436 
437 			len = 2;
438 			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
439 				i *= 10;
440 				i += *buf - '0';
441 				len--;
442 			}
443 			if (i < 1 || i > 12)
444 				return 0;
445 
446 			tm->tm_mon = i - 1;
447 
448 			if (*buf != 0 && isspace((unsigned char)*buf))
449 				while (*ptr != 0 && !isspace((unsigned char)*ptr))
450 					ptr++;
451 			break;
452 
453 		case 's':
454 			{
455 			char *cp;
456 			time_t t;
457 
458 			t = strtol(buf, &cp, 10);
459 			if (t == LONG_MAX)
460 				return 0;
461 			buf = cp;
462 			gmtime_r(&t, tm);
463 			got_GMT = 1;
464 			}
465 			break;
466 
467 		case 'Y':
468 		case 'y':
469 			if (*buf == 0 || isspace((unsigned char)*buf))
470 				break;
471 
472 			if (!isdigit((unsigned char)*buf))
473 				return 0;
474 
475 			len = (c == 'Y') ? 4 : 2;
476 			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
477 				i *= 10;
478 				i += *buf - '0';
479 				len--;
480 			}
481 			if (c == 'Y')
482 				i -= 1900;
483 			if (c == 'y' && i < 69)
484 				i += 100;
485 			if (i < 0)
486 				return 0;
487 
488 			tm->tm_year = i;
489 
490 			if (*buf != 0 && isspace((unsigned char)*buf))
491 				while (*ptr != 0 && !isspace((unsigned char)*ptr))
492 					ptr++;
493 			break;
494 
495 		case 'Z':
496 			{
497 			const char *cp;
498 			char *zonestr;
499 
500 			for (cp = buf; *cp && isupper((unsigned char)*cp); ++cp) {/*empty*/}
501 			if (cp - buf) {
502 				zonestr = alloca(cp - buf + 1);
503 				strncpy(zonestr, buf, cp - buf);
504 				zonestr[cp - buf] = '\0';
505 				tzset();
506 				if (0 == strcmp(zonestr, "GMT")) {
507 				    got_GMT = 1;
508 				} else if (0 == strcmp(zonestr, tzname[0])) {
509 				    tm->tm_isdst = 0;
510 				} else if (0 == strcmp(zonestr, tzname[1])) {
511 				    tm->tm_isdst = 1;
512 				} else {
513 				    return 0;
514 				}
515 				buf += cp - buf;
516 			}
517 			}
518 			break;
519 		}
520 	}
521 	return (char *)buf;
522 }
523 
524 
525 char *
526 strptime(const char *buf, const char *fmt, struct tm *tm)
527 {
528 	char *ret;
529 
530 #ifdef	_THREAD_SAFE
531 	pthread_mutex_lock(&gotgmt_mutex);
532 #endif
533 
534 	got_GMT = 0;
535 	ret = _strptime(buf, fmt, tm);
536 	if (ret && got_GMT) {
537 		time_t t = timegm(tm);
538 	    localtime_r(&t, tm);
539 		got_GMT = 0;
540 	}
541 
542 #ifdef	_THREAD_SAFE
543 	pthread_mutex_unlock(&gotgmt_mutex);
544 #endif
545 
546 	return ret;
547 }
548