strptime.c revision 72183 
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
55static const char rcsid[] =
56  "$FreeBSD: head/lib/libc/stdtime/strptime.c 72183 2001-02-08 20:55:38Z ache $";
57#endif
58
59#ifndef lint
60#ifndef NOID
61static char copyright[] =
62"@(#) Copyright (c) 1994 Powerdog Industries.  All rights reserved.";
63static 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
76static char * _strptime(const char *, const char *, struct tm *);
77
78static pthread_mutex_t	gotgmt_mutex = PTHREAD_MUTEX_INITIALIZER;
79static int got_GMT;
80
81#define asizeof(a)	(sizeof (a) / sizeof ((a)[0]))
82
83static 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;
111label:
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			/* NOTE: c_fmt is hardcoded in timelocal.c */
145			buf = _strptime(buf, tptr->c_fmt, tm);
146			if (buf == 0)
147				return 0;
148			break;
149
150		case 'D':
151			buf = _strptime(buf, "%m/%d/%y", tm);
152			if (buf == 0)
153				return 0;
154			break;
155
156		case 'E':
157			if (Ealternative || Oalternative)
158				break;
159			Ealternative++;
160			goto label;
161
162		case 'O':
163			if (Ealternative || Oalternative)
164				break;
165			Oalternative++;
166			goto label;
167
168		case 'F':
169		case 'f':
170			if (!Ealternative)
171				break;
172			buf = _strptime(buf, (c == 'f') ? tptr->Ef_fmt : tptr->EF_fmt, tm);
173			if (buf == 0)
174				return 0;
175			break;
176
177		case 'R':
178			buf = _strptime(buf, "%H:%M", tm);
179			if (buf == 0)
180				return 0;
181			break;
182
183		case 'r':
184			buf = _strptime(buf, "%I:%M:%S %p", tm);
185			if (buf == 0)
186				return 0;
187			break;
188
189		case 'T':
190			buf = _strptime(buf, "%H:%M:%S", tm);
191			if (buf == 0)
192				return 0;
193			break;
194
195		case 'X':
196			buf = _strptime(buf, tptr->X_fmt, tm);
197			if (buf == 0)
198				return 0;
199			break;
200
201		case 'x':
202			buf = _strptime(buf, tptr->x_fmt, tm);
203			if (buf == 0)
204				return 0;
205			break;
206
207		case 'j':
208			if (!isdigit((unsigned char)*buf))
209				return 0;
210
211			len = 3;
212			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
213				i *= 10;
214				i += *buf - '0';
215				len--;
216			}
217			if (i < 1 || i > 366)
218				return 0;
219
220			tm->tm_yday = i - 1;
221			break;
222
223		case 'M':
224		case 'S':
225			if (*buf == 0 || isspace((unsigned char)*buf))
226				break;
227
228			if (!isdigit((unsigned char)*buf))
229				return 0;
230
231			len = 2;
232			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
233				i *= 10;
234				i += *buf - '0';
235				len--;
236			}
237
238			if (c == 'M') {
239				if (i > 59)
240					return 0;
241				tm->tm_min = i;
242			} else {
243				if (i > 60)
244					return 0;
245				tm->tm_sec = i;
246			}
247
248			if (*buf != 0 && isspace((unsigned char)*buf))
249				while (*ptr != 0 && !isspace((unsigned char)*ptr))
250					ptr++;
251			break;
252
253		case 'H':
254		case 'I':
255		case 'k':
256		case 'l':
257			/*
258			 * Of these, %l is the only specifier explicitly
259			 * documented as not being zero-padded.  However,
260			 * there is no harm in allowing zero-padding.
261			 *
262			 * XXX The %l specifier may gobble one too many
263			 * digits if used incorrectly.
264			 */
265			if (!isdigit((unsigned char)*buf))
266				return 0;
267
268			len = 2;
269			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); 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 && isspace((unsigned char)*buf))
283				while (*ptr != 0 && !isspace((unsigned char)*ptr))
284					ptr++;
285			break;
286
287		case 'p':
288			/*
289			 * XXX This is bogus if parsed before hour-related
290			 * specifiers.
291			 */
292			len = strlen(tptr->am);
293			if (strncasecmp(buf, tptr->am, len) == 0) {
294				if (tm->tm_hour > 12)
295					return 0;
296				if (tm->tm_hour == 12)
297					tm->tm_hour = 0;
298				buf += len;
299				break;
300			}
301
302			len = strlen(tptr->pm);
303			if (strncasecmp(buf, tptr->pm, len) == 0) {
304				if (tm->tm_hour > 12)
305					return 0;
306				if (tm->tm_hour != 12)
307					tm->tm_hour += 12;
308				buf += len;
309				break;
310			}
311
312			return 0;
313
314		case 'A':
315		case 'a':
316			for (i = 0; i < asizeof(tptr->weekday); i++) {
317				if (c == 'A') {
318					len = strlen(tptr->weekday[i]);
319					if (strncasecmp(buf,
320							tptr->weekday[i],
321							len) == 0)
322						break;
323				} else {
324					len = strlen(tptr->wday[i]);
325					if (strncasecmp(buf,
326							tptr->wday[i],
327							len) == 0)
328						break;
329				}
330			}
331			if (i == asizeof(tptr->weekday))
332				return 0;
333
334			tm->tm_wday = i;
335			buf += len;
336			break;
337
338		case 'U':
339		case 'W':
340			/*
341			 * XXX This is bogus, as we can not assume any valid
342			 * information present in the tm structure at this
343			 * point to calculate a real value, so just check the
344			 * range for now.
345			 */
346			if (!isdigit((unsigned char)*buf))
347				return 0;
348
349			len = 2;
350			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
351				i *= 10;
352				i += *buf - '0';
353				len--;
354			}
355			if (i > 53)
356				return 0;
357
358			if (*buf != 0 && isspace((unsigned char)*buf))
359				while (*ptr != 0 && !isspace((unsigned char)*ptr))
360					ptr++;
361			break;
362
363		case 'w':
364			if (!isdigit((unsigned char)*buf))
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 && isspace((unsigned char)*buf))
374				while (*ptr != 0 && !isspace((unsigned char)*ptr))
375					ptr++;
376			break;
377
378		case 'd':
379		case 'e':
380			/*
381			 * The %e specifier is explicitly documented as not
382			 * being zero-padded but there is no harm in allowing
383			 * such padding.
384			 *
385			 * XXX The %e specifier may gobble one too many
386			 * digits if used incorrectly.
387			 */
388			if (!isdigit((unsigned char)*buf))
389				return 0;
390
391			len = 2;
392			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
393				i *= 10;
394				i += *buf - '0';
395				len--;
396			}
397			if (i > 31)
398				return 0;
399
400			tm->tm_mday = i;
401
402			if (*buf != 0 && isspace((unsigned char)*buf))
403				while (*ptr != 0 && !isspace((unsigned char)*ptr))
404					ptr++;
405			break;
406
407		case 'B':
408		case 'b':
409		case 'h':
410			for (i = 0; i < asizeof(tptr->month); i++) {
411				if (Oalternative) {
412					if (c == 'B') {
413						len = strlen(tptr->alt_month[i]);
414						if (strncasecmp(buf,
415								tptr->alt_month[i],
416								len) == 0)
417							break;
418					}
419				} else {
420					if (c == 'B') {
421						len = strlen(tptr->month[i]);
422						if (strncasecmp(buf,
423								tptr->month[i],
424								len) == 0)
425							break;
426					} else {
427						len = strlen(tptr->mon[i]);
428						if (strncasecmp(buf,
429								tptr->mon[i],
430								len) == 0)
431							break;
432					}
433				}
434			}
435			if (i == asizeof(tptr->month))
436				return 0;
437
438			tm->tm_mon = i;
439			buf += len;
440			break;
441
442		case 'm':
443			if (!isdigit((unsigned char)*buf))
444				return 0;
445
446			len = 2;
447			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
448				i *= 10;
449				i += *buf - '0';
450				len--;
451			}
452			if (i < 1 || i > 12)
453				return 0;
454
455			tm->tm_mon = i - 1;
456
457			if (*buf != 0 && isspace((unsigned char)*buf))
458				while (*ptr != 0 && !isspace((unsigned char)*ptr))
459					ptr++;
460			break;
461
462		case 'Y':
463		case 'y':
464			if (*buf == 0 || isspace((unsigned char)*buf))
465				break;
466
467			if (!isdigit((unsigned char)*buf))
468				return 0;
469
470			len = (c == 'Y') ? 4 : 2;
471			for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) {
472				i *= 10;
473				i += *buf - '0';
474				len--;
475			}
476			if (c == 'Y')
477				i -= 1900;
478			if (c == 'y' && i < 69)
479				i += 100;
480			if (i < 0)
481				return 0;
482
483			tm->tm_year = i;
484
485			if (*buf != 0 && isspace((unsigned char)*buf))
486				while (*ptr != 0 && !isspace((unsigned char)*ptr))
487					ptr++;
488			break;
489
490		case 'Z':
491			{
492			const char *cp;
493			char *zonestr;
494
495			for (cp = buf; *cp && isupper((unsigned char)*cp); ++cp) {/*empty*/}
496			if (cp - buf) {
497				zonestr = alloca(cp - buf + 1);
498				strncpy(zonestr, buf, cp - buf);
499				zonestr[cp - buf] = '\0';
500				tzset();
501				if (0 == strcmp(zonestr, "GMT")) {
502				    got_GMT = 1;
503				} else if (0 == strcmp(zonestr, tzname[0])) {
504				    tm->tm_isdst = 0;
505				} else if (0 == strcmp(zonestr, tzname[1])) {
506				    tm->tm_isdst = 1;
507				} else {
508				    return 0;
509				}
510				buf += cp - buf;
511			}
512			}
513			break;
514		}
515	}
516	return (char *)buf;
517}
518
519
520char *
521strptime(const char *buf, const char *fmt, struct tm *tm)
522{
523	char *ret;
524
525	if (__isthreaded)
526		_pthread_mutex_lock(&gotgmt_mutex);
527
528	got_GMT = 0;
529	ret = _strptime(buf, fmt, tm);
530	if (ret && got_GMT) {
531		time_t t = timegm(tm);
532		localtime_r(&t, tm);
533		got_GMT = 0;
534	}
535
536	if (__isthreaded)
537		_pthread_mutex_unlock(&gotgmt_mutex);
538
539	return ret;
540}
541