xref: /freebsd/lib/libc/stdtime/strptime.c (revision 1b6c76a2fe091c74f08427e6c870851025a9cf67)
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   "$FreeBSD$";
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 "namespace.h"
68 #include <time.h>
69 #include <ctype.h>
70 #include <string.h>
71 #include <pthread.h>
72 #include "un-namespace.h"
73 #include "libc_private.h"
74 #include "timelocal.h"
75 
76 static char * _strptime(const char *, const char *, struct tm *);
77 
78 static pthread_mutex_t	gotgmt_mutex = PTHREAD_MUTEX_INITIALIZER;
79 static int got_GMT;
80 
81 #define asizeof(a)	(sizeof (a) / sizeof ((a)[0]))
82 
83 static char *
84 _strptime(const char *buf, const char *fmt, struct tm *tm)
85 {
86 	char	c;
87 	const char *ptr;
88 	int	i,
89 		len;
90 	int Ealternative, Oalternative;
91 	struct lc_time_T *tptr = __get_current_time_locale();
92 
93 	ptr = fmt;
94 	while (*ptr != 0) {
95 		if (*buf == 0)
96 			break;
97 
98 		c = *ptr++;
99 
100 		if (c != '%') {
101 			if (isspace((unsigned char)c))
102 				while (*buf != 0 && isspace((unsigned char)*buf))
103 					buf++;
104 			else if (c != *buf++)
105 				return 0;
106 			continue;
107 		}
108 
109 		Ealternative = 0;
110 		Oalternative = 0;
111 label:
112 		c = *ptr++;
113 		switch (c) {
114 		case 0:
115 		case '%':
116 			if (*buf++ != '%')
117 				return 0;
118 			break;
119 
120 		case '+':
121 			buf = _strptime(buf, tptr->date_fmt, tm);
122 			if (buf == 0)
123 				return 0;
124 			break;
125 
126 		case 'C':
127 			if (!isdigit((unsigned char)*buf))
128 				return 0;
129 
130 			/* XXX This will break for 3-digit centuries. */
131 			len = 2;
132 			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
133 				i *= 10;
134 				i += *buf - '0';
135 				len--;
136 			}
137 			if (i < 19)
138 				return 0;
139 
140 			tm->tm_year = i * 100 - 1900;
141 			break;
142 
143 		case 'c':
144 			buf = _strptime(buf, tptr->c_fmt, tm);
145 			if (buf == 0)
146 				return 0;
147 			break;
148 
149 		case 'D':
150 			buf = _strptime(buf, "%m/%d/%y", tm);
151 			if (buf == 0)
152 				return 0;
153 			break;
154 
155 		case 'E':
156 			if (Ealternative || Oalternative)
157 				break;
158 			Ealternative++;
159 			goto label;
160 
161 		case 'O':
162 			if (Ealternative || Oalternative)
163 				break;
164 			Oalternative++;
165 			goto label;
166 
167 		case 'F':
168 			buf = _strptime(buf, "%Y-%m-%d", tm);
169 			if (buf == 0)
170 				return 0;
171 			break;
172 
173 		case 'R':
174 			buf = _strptime(buf, "%H:%M", tm);
175 			if (buf == 0)
176 				return 0;
177 			break;
178 
179 		case 'r':
180 			buf = _strptime(buf, tptr->ampm_fmt, tm);
181 			if (buf == 0)
182 				return 0;
183 			break;
184 
185 		case 'T':
186 			buf = _strptime(buf, "%H:%M:%S", tm);
187 			if (buf == 0)
188 				return 0;
189 			break;
190 
191 		case 'X':
192 			buf = _strptime(buf, tptr->X_fmt, tm);
193 			if (buf == 0)
194 				return 0;
195 			break;
196 
197 		case 'x':
198 			buf = _strptime(buf, tptr->x_fmt, tm);
199 			if (buf == 0)
200 				return 0;
201 			break;
202 
203 		case 'j':
204 			if (!isdigit((unsigned char)*buf))
205 				return 0;
206 
207 			len = 3;
208 			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
209 				i *= 10;
210 				i += *buf - '0';
211 				len--;
212 			}
213 			if (i < 1 || i > 366)
214 				return 0;
215 
216 			tm->tm_yday = i - 1;
217 			break;
218 
219 		case 'M':
220 		case 'S':
221 			if (*buf == 0 || isspace((unsigned char)*buf))
222 				break;
223 
224 			if (!isdigit((unsigned char)*buf))
225 				return 0;
226 
227 			len = 2;
228 			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
229 				i *= 10;
230 				i += *buf - '0';
231 				len--;
232 			}
233 
234 			if (c == 'M') {
235 				if (i > 59)
236 					return 0;
237 				tm->tm_min = i;
238 			} else {
239 				if (i > 60)
240 					return 0;
241 				tm->tm_sec = i;
242 			}
243 
244 			if (*buf != 0 && isspace((unsigned char)*buf))
245 				while (*ptr != 0 && !isspace((unsigned char)*ptr))
246 					ptr++;
247 			break;
248 
249 		case 'H':
250 		case 'I':
251 		case 'k':
252 		case 'l':
253 			/*
254 			 * Of these, %l is the only specifier explicitly
255 			 * documented as not being zero-padded.  However,
256 			 * there is no harm in allowing zero-padding.
257 			 *
258 			 * XXX The %l specifier may gobble one too many
259 			 * digits if used incorrectly.
260 			 */
261 			if (!isdigit((unsigned char)*buf))
262 				return 0;
263 
264 			len = 2;
265 			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
266 				i *= 10;
267 				i += *buf - '0';
268 				len--;
269 			}
270 			if (c == 'H' || c == 'k') {
271 				if (i > 23)
272 					return 0;
273 			} else if (i > 12)
274 				return 0;
275 
276 			tm->tm_hour = i;
277 
278 			if (*buf != 0 && isspace((unsigned char)*buf))
279 				while (*ptr != 0 && !isspace((unsigned char)*ptr))
280 					ptr++;
281 			break;
282 
283 		case 'p':
284 			/*
285 			 * XXX This is bogus if parsed before hour-related
286 			 * specifiers.
287 			 */
288 			len = strlen(tptr->am);
289 			if (strncasecmp(buf, tptr->am, len) == 0) {
290 				if (tm->tm_hour > 12)
291 					return 0;
292 				if (tm->tm_hour == 12)
293 					tm->tm_hour = 0;
294 				buf += len;
295 				break;
296 			}
297 
298 			len = strlen(tptr->pm);
299 			if (strncasecmp(buf, tptr->pm, len) == 0) {
300 				if (tm->tm_hour > 12)
301 					return 0;
302 				if (tm->tm_hour != 12)
303 					tm->tm_hour += 12;
304 				buf += len;
305 				break;
306 			}
307 
308 			return 0;
309 
310 		case 'A':
311 		case 'a':
312 			for (i = 0; i < asizeof(tptr->weekday); i++) {
313 				len = strlen(tptr->weekday[i]);
314 				if (strncasecmp(buf, tptr->weekday[i],
315 						len) == 0)
316 					break;
317 				len = strlen(tptr->wday[i]);
318 				if (strncasecmp(buf, tptr->wday[i],
319 						len) == 0)
320 					break;
321 			}
322 			if (i == asizeof(tptr->weekday))
323 				return 0;
324 
325 			tm->tm_wday = i;
326 			buf += len;
327 			break;
328 
329 		case 'U':
330 		case 'W':
331 			/*
332 			 * XXX This is bogus, as we can not assume any valid
333 			 * information present in the tm structure at this
334 			 * point to calculate a real value, so just check the
335 			 * range for now.
336 			 */
337 			if (!isdigit((unsigned char)*buf))
338 				return 0;
339 
340 			len = 2;
341 			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
342 				i *= 10;
343 				i += *buf - '0';
344 				len--;
345 			}
346 			if (i > 53)
347 				return 0;
348 
349 			if (*buf != 0 && isspace((unsigned char)*buf))
350 				while (*ptr != 0 && !isspace((unsigned char)*ptr))
351 					ptr++;
352 			break;
353 
354 		case 'w':
355 			if (!isdigit((unsigned char)*buf))
356 				return 0;
357 
358 			i = *buf - '0';
359 			if (i > 6)
360 				return 0;
361 
362 			tm->tm_wday = i;
363 
364 			if (*buf != 0 && isspace((unsigned char)*buf))
365 				while (*ptr != 0 && !isspace((unsigned char)*ptr))
366 					ptr++;
367 			break;
368 
369 		case 'd':
370 		case 'e':
371 			/*
372 			 * The %e specifier is explicitly documented as not
373 			 * being zero-padded but there is no harm in allowing
374 			 * such padding.
375 			 *
376 			 * XXX The %e specifier may gobble one too many
377 			 * digits if used incorrectly.
378 			 */
379 			if (!isdigit((unsigned char)*buf))
380 				return 0;
381 
382 			len = 2;
383 			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
384 				i *= 10;
385 				i += *buf - '0';
386 				len--;
387 			}
388 			if (i > 31)
389 				return 0;
390 
391 			tm->tm_mday = i;
392 
393 			if (*buf != 0 && isspace((unsigned char)*buf))
394 				while (*ptr != 0 && !isspace((unsigned char)*ptr))
395 					ptr++;
396 			break;
397 
398 		case 'B':
399 		case 'b':
400 		case 'h':
401 			for (i = 0; i < asizeof(tptr->month); i++) {
402 				if (Oalternative) {
403 					if (c == 'B') {
404 						len = strlen(tptr->alt_month[i]);
405 						if (strncasecmp(buf,
406 								tptr->alt_month[i],
407 								len) == 0)
408 							break;
409 					}
410 				} else {
411 					len = strlen(tptr->month[i]);
412 					if (strncasecmp(buf, tptr->month[i],
413 							len) == 0)
414 						break;
415 					len = strlen(tptr->mon[i]);
416 					if (strncasecmp(buf, tptr->mon[i],
417 							len) == 0)
418 						break;
419 				}
420 			}
421 			if (i == asizeof(tptr->month))
422 				return 0;
423 
424 			tm->tm_mon = i;
425 			buf += len;
426 			break;
427 
428 		case 'm':
429 			if (!isdigit((unsigned char)*buf))
430 				return 0;
431 
432 			len = 2;
433 			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
434 				i *= 10;
435 				i += *buf - '0';
436 				len--;
437 			}
438 			if (i < 1 || i > 12)
439 				return 0;
440 
441 			tm->tm_mon = i - 1;
442 
443 			if (*buf != 0 && isspace((unsigned char)*buf))
444 				while (*ptr != 0 && !isspace((unsigned char)*ptr))
445 					ptr++;
446 			break;
447 
448 		case 'Y':
449 		case 'y':
450 			if (*buf == 0 || isspace((unsigned char)*buf))
451 				break;
452 
453 			if (!isdigit((unsigned char)*buf))
454 				return 0;
455 
456 			len = (c == 'Y') ? 4 : 2;
457 			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
458 				i *= 10;
459 				i += *buf - '0';
460 				len--;
461 			}
462 			if (c == 'Y')
463 				i -= 1900;
464 			if (c == 'y' && i < 69)
465 				i += 100;
466 			if (i < 0)
467 				return 0;
468 
469 			tm->tm_year = i;
470 
471 			if (*buf != 0 && isspace((unsigned char)*buf))
472 				while (*ptr != 0 && !isspace((unsigned char)*ptr))
473 					ptr++;
474 			break;
475 
476 		case 'Z':
477 			{
478 			const char *cp;
479 			char *zonestr;
480 
481 			for (cp = buf; *cp && isupper((unsigned char)*cp); ++cp) {/*empty*/}
482 			if (cp - buf) {
483 				zonestr = alloca(cp - buf + 1);
484 				strncpy(zonestr, buf, cp - buf);
485 				zonestr[cp - buf] = '\0';
486 				tzset();
487 				if (0 == strcmp(zonestr, "GMT")) {
488 				    got_GMT = 1;
489 				} else if (0 == strcmp(zonestr, tzname[0])) {
490 				    tm->tm_isdst = 0;
491 				} else if (0 == strcmp(zonestr, tzname[1])) {
492 				    tm->tm_isdst = 1;
493 				} else {
494 				    return 0;
495 				}
496 				buf += cp - buf;
497 			}
498 			}
499 			break;
500 		}
501 	}
502 	return (char *)buf;
503 }
504 
505 
506 char *
507 strptime(const char *buf, const char *fmt, struct tm *tm)
508 {
509 	char *ret;
510 
511 	if (__isthreaded)
512 		_pthread_mutex_lock(&gotgmt_mutex);
513 
514 	got_GMT = 0;
515 	ret = _strptime(buf, fmt, tm);
516 	if (ret && got_GMT) {
517 		time_t t = timegm(tm);
518 		localtime_r(&t, tm);
519 		got_GMT = 0;
520 	}
521 
522 	if (__isthreaded)
523 		_pthread_mutex_unlock(&gotgmt_mutex);
524 
525 	return ret;
526 }
527