1: /***************************************************************************
2: * _ _ ____ _
3: * Project ___| | | | _ \| |
4: * / __| | | | |_) | |
5: * | (__| |_| | _ <| |___
6: * \___|\___/|_| \_\_____|
7: *
8: * Copyright (C) 1998 - 2019, Daniel Stenberg, <daniel@haxx.se>, et al.
9: *
10: * This software is licensed as described in the file COPYING, which
11: * you should have received as part of this distribution. The terms
12: * are also available at https://curl.haxx.se/docs/copyright.html.
13: *
14: * You may opt to use, copy, modify, merge, publish, distribute and/or sell
15: * copies of the Software, and permit persons to whom the Software is
16: * furnished to do so, under the terms of the COPYING file.
17: *
18: * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
19: * KIND, either express or implied.
20: *
21: ***************************************************************************/
22: /*
23: A brief summary of the date string formats this parser groks:
24:
25: RFC 2616 3.3.1
26:
27: Sun, 06 Nov 1994 08:49:37 GMT ; RFC 822, updated by RFC 1123
28: Sunday, 06-Nov-94 08:49:37 GMT ; RFC 850, obsoleted by RFC 1036
29: Sun Nov 6 08:49:37 1994 ; ANSI C's asctime() format
30:
31: we support dates without week day name:
32:
33: 06 Nov 1994 08:49:37 GMT
34: 06-Nov-94 08:49:37 GMT
35: Nov 6 08:49:37 1994
36:
37: without the time zone:
38:
39: 06 Nov 1994 08:49:37
40: 06-Nov-94 08:49:37
41:
42: weird order:
43:
44: 1994 Nov 6 08:49:37 (GNU date fails)
45: GMT 08:49:37 06-Nov-94 Sunday
46: 94 6 Nov 08:49:37 (GNU date fails)
47:
48: time left out:
49:
50: 1994 Nov 6
51: 06-Nov-94
52: Sun Nov 6 94
53:
54: unusual separators:
55:
56: 1994.Nov.6
57: Sun/Nov/6/94/GMT
58:
59: commonly used time zone names:
60:
61: Sun, 06 Nov 1994 08:49:37 CET
62: 06 Nov 1994 08:49:37 EST
63:
64: time zones specified using RFC822 style:
65:
66: Sun, 12 Sep 2004 15:05:58 -0700
67: Sat, 11 Sep 2004 21:32:11 +0200
68:
69: compact numerical date strings:
70:
71: 20040912 15:05:58 -0700
72: 20040911 +0200
73:
74: */
75:
76: #include "curl_setup.h"
77:
78: #include <limits.h>
79:
80: #include <curl/curl.h>
81: #include "strcase.h"
82: #include "warnless.h"
83: #include "parsedate.h"
84:
85: /*
86: * parsedate()
87: *
88: * Returns:
89: *
90: * PARSEDATE_OK - a fine conversion
91: * PARSEDATE_FAIL - failed to convert
92: * PARSEDATE_LATER - time overflow at the far end of time_t
93: * PARSEDATE_SOONER - time underflow at the low end of time_t
94: */
95:
96: static int parsedate(const char *date, time_t *output);
97:
98: #define PARSEDATE_OK 0
99: #define PARSEDATE_FAIL -1
100: #define PARSEDATE_LATER 1
101: #define PARSEDATE_SOONER 2
102:
103: #if !defined(CURL_DISABLE_PARSEDATE) || !defined(CURL_DISABLE_FTP) || \
104: !defined(CURL_DISABLE_FILE)
105: /* These names are also used by FTP and FILE code */
106: const char * const Curl_wkday[] =
107: {"Mon", "Tue", "Wed", "Thu", "Fri", "Sat", "Sun"};
108: const char * const Curl_month[]=
109: { "Jan", "Feb", "Mar", "Apr", "May", "Jun",
110: "Jul", "Aug", "Sep", "Oct", "Nov", "Dec" };
111: #endif
112:
113: #ifndef CURL_DISABLE_PARSEDATE
114: static const char * const weekday[] =
115: { "Monday", "Tuesday", "Wednesday", "Thursday",
116: "Friday", "Saturday", "Sunday" };
117:
118: struct tzinfo {
119: char name[5];
120: int offset; /* +/- in minutes */
121: };
122:
123: /* Here's a bunch of frequently used time zone names. These were supported
124: by the old getdate parser. */
125: #define tDAYZONE -60 /* offset for daylight savings time */
126: static const struct tzinfo tz[]= {
127: {"GMT", 0}, /* Greenwich Mean */
128: {"UT", 0}, /* Universal Time */
129: {"UTC", 0}, /* Universal (Coordinated) */
130: {"WET", 0}, /* Western European */
131: {"BST", 0 tDAYZONE}, /* British Summer */
132: {"WAT", 60}, /* West Africa */
133: {"AST", 240}, /* Atlantic Standard */
134: {"ADT", 240 tDAYZONE}, /* Atlantic Daylight */
135: {"EST", 300}, /* Eastern Standard */
136: {"EDT", 300 tDAYZONE}, /* Eastern Daylight */
137: {"CST", 360}, /* Central Standard */
138: {"CDT", 360 tDAYZONE}, /* Central Daylight */
139: {"MST", 420}, /* Mountain Standard */
140: {"MDT", 420 tDAYZONE}, /* Mountain Daylight */
141: {"PST", 480}, /* Pacific Standard */
142: {"PDT", 480 tDAYZONE}, /* Pacific Daylight */
143: {"YST", 540}, /* Yukon Standard */
144: {"YDT", 540 tDAYZONE}, /* Yukon Daylight */
145: {"HST", 600}, /* Hawaii Standard */
146: {"HDT", 600 tDAYZONE}, /* Hawaii Daylight */
147: {"CAT", 600}, /* Central Alaska */
148: {"AHST", 600}, /* Alaska-Hawaii Standard */
149: {"NT", 660}, /* Nome */
150: {"IDLW", 720}, /* International Date Line West */
151: {"CET", -60}, /* Central European */
152: {"MET", -60}, /* Middle European */
153: {"MEWT", -60}, /* Middle European Winter */
154: {"MEST", -60 tDAYZONE}, /* Middle European Summer */
155: {"CEST", -60 tDAYZONE}, /* Central European Summer */
156: {"MESZ", -60 tDAYZONE}, /* Middle European Summer */
157: {"FWT", -60}, /* French Winter */
158: {"FST", -60 tDAYZONE}, /* French Summer */
159: {"EET", -120}, /* Eastern Europe, USSR Zone 1 */
160: {"WAST", -420}, /* West Australian Standard */
161: {"WADT", -420 tDAYZONE}, /* West Australian Daylight */
162: {"CCT", -480}, /* China Coast, USSR Zone 7 */
163: {"JST", -540}, /* Japan Standard, USSR Zone 8 */
164: {"EAST", -600}, /* Eastern Australian Standard */
165: {"EADT", -600 tDAYZONE}, /* Eastern Australian Daylight */
166: {"GST", -600}, /* Guam Standard, USSR Zone 9 */
167: {"NZT", -720}, /* New Zealand */
168: {"NZST", -720}, /* New Zealand Standard */
169: {"NZDT", -720 tDAYZONE}, /* New Zealand Daylight */
170: {"IDLE", -720}, /* International Date Line East */
171: /* Next up: Military timezone names. RFC822 allowed these, but (as noted in
172: RFC 1123) had their signs wrong. Here we use the correct signs to match
173: actual military usage.
174: */
175: {"A", 1 * 60}, /* Alpha */
176: {"B", 2 * 60}, /* Bravo */
177: {"C", 3 * 60}, /* Charlie */
178: {"D", 4 * 60}, /* Delta */
179: {"E", 5 * 60}, /* Echo */
180: {"F", 6 * 60}, /* Foxtrot */
181: {"G", 7 * 60}, /* Golf */
182: {"H", 8 * 60}, /* Hotel */
183: {"I", 9 * 60}, /* India */
184: /* "J", Juliet is not used as a timezone, to indicate the observer's local
185: time */
186: {"K", 10 * 60}, /* Kilo */
187: {"L", 11 * 60}, /* Lima */
188: {"M", 12 * 60}, /* Mike */
189: {"N", -1 * 60}, /* November */
190: {"O", -2 * 60}, /* Oscar */
191: {"P", -3 * 60}, /* Papa */
192: {"Q", -4 * 60}, /* Quebec */
193: {"R", -5 * 60}, /* Romeo */
194: {"S", -6 * 60}, /* Sierra */
195: {"T", -7 * 60}, /* Tango */
196: {"U", -8 * 60}, /* Uniform */
197: {"V", -9 * 60}, /* Victor */
198: {"W", -10 * 60}, /* Whiskey */
199: {"X", -11 * 60}, /* X-ray */
200: {"Y", -12 * 60}, /* Yankee */
201: {"Z", 0}, /* Zulu, zero meridian, a.k.a. UTC */
202: };
203:
204: /* returns:
205: -1 no day
206: 0 monday - 6 sunday
207: */
208:
209: static int checkday(const char *check, size_t len)
210: {
211: int i;
212: const char * const *what;
213: bool found = FALSE;
214: if(len > 3)
215: what = &weekday[0];
216: else
217: what = &Curl_wkday[0];
218: for(i = 0; i<7; i++) {
219: if(strcasecompare(check, what[0])) {
220: found = TRUE;
221: break;
222: }
223: what++;
224: }
225: return found?i:-1;
226: }
227:
228: static int checkmonth(const char *check)
229: {
230: int i;
231: const char * const *what;
232: bool found = FALSE;
233:
234: what = &Curl_month[0];
235: for(i = 0; i<12; i++) {
236: if(strcasecompare(check, what[0])) {
237: found = TRUE;
238: break;
239: }
240: what++;
241: }
242: return found?i:-1; /* return the offset or -1, no real offset is -1 */
243: }
244:
245: /* return the time zone offset between GMT and the input one, in number
246: of seconds or -1 if the timezone wasn't found/legal */
247:
248: static int checktz(const char *check)
249: {
250: unsigned int i;
251: const struct tzinfo *what;
252: bool found = FALSE;
253:
254: what = tz;
255: for(i = 0; i< sizeof(tz)/sizeof(tz[0]); i++) {
256: if(strcasecompare(check, what->name)) {
257: found = TRUE;
258: break;
259: }
260: what++;
261: }
262: return found?what->offset*60:-1;
263: }
264:
265: static void skip(const char **date)
266: {
267: /* skip everything that aren't letters or digits */
268: while(**date && !ISALNUM(**date))
269: (*date)++;
270: }
271:
272: enum assume {
273: DATE_MDAY,
274: DATE_YEAR,
275: DATE_TIME
276: };
277:
278: /* this is a clone of 'struct tm' but with all fields we don't need or use
279: cut out */
280: struct my_tm {
281: int tm_sec;
282: int tm_min;
283: int tm_hour;
284: int tm_mday;
285: int tm_mon;
286: int tm_year; /* full year */
287: };
288:
289: /* struct tm to time since epoch in GMT time zone.
290: * This is similar to the standard mktime function but for GMT only, and
291: * doesn't suffer from the various bugs and portability problems that
292: * some systems' implementations have.
293: *
294: * Returns 0 on success, otherwise non-zero.
295: */
296: static void my_timegm(struct my_tm *tm, time_t *t)
297: {
298: static const int month_days_cumulative [12] =
299: { 0, 31, 59, 90, 120, 151, 181, 212, 243, 273, 304, 334 };
300: int month, year, leap_days;
301:
302: year = tm->tm_year;
303: month = tm->tm_mon;
304: if(month < 0) {
305: year += (11 - month) / 12;
306: month = 11 - (11 - month) % 12;
307: }
308: else if(month >= 12) {
309: year -= month / 12;
310: month = month % 12;
311: }
312:
313: leap_days = year - (tm->tm_mon <= 1);
314: leap_days = ((leap_days / 4) - (leap_days / 100) + (leap_days / 400)
315: - (1969 / 4) + (1969 / 100) - (1969 / 400));
316:
317: *t = ((((time_t) (year - 1970) * 365
318: + leap_days + month_days_cumulative[month] + tm->tm_mday - 1) * 24
319: + tm->tm_hour) * 60 + tm->tm_min) * 60 + tm->tm_sec;
320: }
321:
322: /*
323: * parsedate()
324: *
325: * Returns:
326: *
327: * PARSEDATE_OK - a fine conversion
328: * PARSEDATE_FAIL - failed to convert
329: * PARSEDATE_LATER - time overflow at the far end of time_t
330: * PARSEDATE_SOONER - time underflow at the low end of time_t
331: */
332:
333: static int parsedate(const char *date, time_t *output)
334: {
335: time_t t = 0;
336: int wdaynum = -1; /* day of the week number, 0-6 (mon-sun) */
337: int monnum = -1; /* month of the year number, 0-11 */
338: int mdaynum = -1; /* day of month, 1 - 31 */
339: int hournum = -1;
340: int minnum = -1;
341: int secnum = -1;
342: int yearnum = -1;
343: int tzoff = -1;
344: struct my_tm tm;
345: enum assume dignext = DATE_MDAY;
346: const char *indate = date; /* save the original pointer */
347: int part = 0; /* max 6 parts */
348:
349: while(*date && (part < 6)) {
350: bool found = FALSE;
351:
352: skip(&date);
353:
354: if(ISALPHA(*date)) {
355: /* a name coming up */
356: char buf[32]="";
357: size_t len;
358: if(sscanf(date, "%31[ABCDEFGHIJKLMNOPQRSTUVWXYZ"
359: "abcdefghijklmnopqrstuvwxyz]", buf))
360: len = strlen(buf);
361: else
362: len = 0;
363:
364: if(wdaynum == -1) {
365: wdaynum = checkday(buf, len);
366: if(wdaynum != -1)
367: found = TRUE;
368: }
369: if(!found && (monnum == -1)) {
370: monnum = checkmonth(buf);
371: if(monnum != -1)
372: found = TRUE;
373: }
374:
375: if(!found && (tzoff == -1)) {
376: /* this just must be a time zone string */
377: tzoff = checktz(buf);
378: if(tzoff != -1)
379: found = TRUE;
380: }
381:
382: if(!found)
383: return PARSEDATE_FAIL; /* bad string */
384:
385: date += len;
386: }
387: else if(ISDIGIT(*date)) {
388: /* a digit */
389: int val;
390: char *end;
391: int len = 0;
392: if((secnum == -1) &&
393: (3 == sscanf(date, "%02d:%02d:%02d%n",
394: &hournum, &minnum, &secnum, &len))) {
395: /* time stamp! */
396: date += len;
397: }
398: else if((secnum == -1) &&
399: (2 == sscanf(date, "%02d:%02d%n", &hournum, &minnum, &len))) {
400: /* time stamp without seconds */
401: date += len;
402: secnum = 0;
403: }
404: else {
405: long lval;
406: int error;
407: int old_errno;
408:
409: old_errno = errno;
410: errno = 0;
411: lval = strtol(date, &end, 10);
412: error = errno;
413: if(errno != old_errno)
414: errno = old_errno;
415:
416: if(error)
417: return PARSEDATE_FAIL;
418:
419: #if LONG_MAX != INT_MAX
420: if((lval > (long)INT_MAX) || (lval < (long)INT_MIN))
421: return PARSEDATE_FAIL;
422: #endif
423:
424: val = curlx_sltosi(lval);
425:
426: if((tzoff == -1) &&
427: ((end - date) == 4) &&
428: (val <= 1400) &&
429: (indate< date) &&
430: ((date[-1] == '+' || date[-1] == '-'))) {
431: /* four digits and a value less than or equal to 1400 (to take into
432: account all sorts of funny time zone diffs) and it is preceded
433: with a plus or minus. This is a time zone indication. 1400 is
434: picked since +1300 is frequently used and +1400 is mentioned as
435: an edge number in the document "ISO C 200X Proposal: Timezone
436: Functions" at http://david.tribble.com/text/c0xtimezone.html If
437: anyone has a more authoritative source for the exact maximum time
438: zone offsets, please speak up! */
439: found = TRUE;
440: tzoff = (val/100 * 60 + val%100)*60;
441:
442: /* the + and - prefix indicates the local time compared to GMT,
443: this we need their reversed math to get what we want */
444: tzoff = date[-1]=='+'?-tzoff:tzoff;
445: }
446:
447: if(((end - date) == 8) &&
448: (yearnum == -1) &&
449: (monnum == -1) &&
450: (mdaynum == -1)) {
451: /* 8 digits, no year, month or day yet. This is YYYYMMDD */
452: found = TRUE;
453: yearnum = val/10000;
454: monnum = (val%10000)/100-1; /* month is 0 - 11 */
455: mdaynum = val%100;
456: }
457:
458: if(!found && (dignext == DATE_MDAY) && (mdaynum == -1)) {
459: if((val > 0) && (val<32)) {
460: mdaynum = val;
461: found = TRUE;
462: }
463: dignext = DATE_YEAR;
464: }
465:
466: if(!found && (dignext == DATE_YEAR) && (yearnum == -1)) {
467: yearnum = val;
468: found = TRUE;
469: if(yearnum < 100) {
470: if(yearnum > 70)
471: yearnum += 1900;
472: else
473: yearnum += 2000;
474: }
475: if(mdaynum == -1)
476: dignext = DATE_MDAY;
477: }
478:
479: if(!found)
480: return PARSEDATE_FAIL;
481:
482: date = end;
483: }
484: }
485:
486: part++;
487: }
488:
489: if(-1 == secnum)
490: secnum = minnum = hournum = 0; /* no time, make it zero */
491:
492: if((-1 == mdaynum) ||
493: (-1 == monnum) ||
494: (-1 == yearnum))
495: /* lacks vital info, fail */
496: return PARSEDATE_FAIL;
497:
498: #ifdef HAVE_TIME_T_UNSIGNED
499: if(yearnum < 1970) {
500: /* only positive numbers cannot return earlier */
501: *output = TIME_T_MIN;
502: return PARSEDATE_SOONER;
503: }
504: #endif
505:
506: #if (SIZEOF_TIME_T < 5)
507:
508: #ifdef HAVE_TIME_T_UNSIGNED
509: /* an unsigned 32 bit time_t can only hold dates to 2106 */
510: if(yearnum > 2105) {
511: *output = TIME_T_MAX;
512: return PARSEDATE_LATER;
513: }
514: #else
515: /* a signed 32 bit time_t can only hold dates to the beginning of 2038 */
516: if(yearnum > 2037) {
517: *output = TIME_T_MAX;
518: return PARSEDATE_LATER;
519: }
520: if(yearnum < 1903) {
521: *output = TIME_T_MIN;
522: return PARSEDATE_SOONER;
523: }
524: #endif
525:
526: #else
527: /* The Gregorian calendar was introduced 1582 */
528: if(yearnum < 1583)
529: return PARSEDATE_FAIL;
530: #endif
531:
532: if((mdaynum > 31) || (monnum > 11) ||
533: (hournum > 23) || (minnum > 59) || (secnum > 60))
534: return PARSEDATE_FAIL; /* clearly an illegal date */
535:
536: tm.tm_sec = secnum;
537: tm.tm_min = minnum;
538: tm.tm_hour = hournum;
539: tm.tm_mday = mdaynum;
540: tm.tm_mon = monnum;
541: tm.tm_year = yearnum;
542:
543: /* my_timegm() returns a time_t. time_t is often 32 bits, sometimes even on
544: architectures that feature 64 bit 'long' but ultimately time_t is the
545: correct data type to use.
546: */
547: my_timegm(&tm, &t);
548:
549: /* Add the time zone diff between local time zone and GMT. */
550: if(tzoff == -1)
551: tzoff = 0;
552:
553: if((tzoff > 0) && (t > TIME_T_MAX - tzoff)) {
554: *output = TIME_T_MAX;
555: return PARSEDATE_LATER; /* time_t overflow */
556: }
557:
558: t += tzoff;
559:
560: *output = t;
561:
562: return PARSEDATE_OK;
563: }
564: #else
565: /* disabled */
566: static int parsedate(const char *date, time_t *output)
567: {
568: (void)date;
569: *output = 0;
570: return PARSEDATE_OK; /* a lie */
571: }
572: #endif
573:
574: time_t curl_getdate(const char *p, const time_t *now)
575: {
576: time_t parsed = -1;
577: int rc = parsedate(p, &parsed);
578: (void)now; /* legacy argument from the past that we ignore */
579:
580: if(rc == PARSEDATE_OK) {
581: if(parsed == -1)
582: /* avoid returning -1 for a working scenario */
583: parsed++;
584: return parsed;
585: }
586: /* everything else is fail */
587: return -1;
588: }
589:
590: /* Curl_getdate_capped() differs from curl_getdate() in that this will return
591: TIME_T_MAX in case the parsed time value was too big, instead of an
592: error. */
593:
594: time_t Curl_getdate_capped(const char *p)
595: {
596: time_t parsed = -1;
597: int rc = parsedate(p, &parsed);
598:
599: switch(rc) {
600: case PARSEDATE_OK:
601: if(parsed == -1)
602: /* avoid returning -1 for a working scenario */
603: parsed++;
604: return parsed;
605: case PARSEDATE_LATER:
606: /* this returns the maximum time value */
607: return parsed;
608: default:
609: return -1; /* everything else is fail */
610: }
611: /* UNREACHABLE */
612: }
613:
614: /*
615: * Curl_gmtime() is a gmtime() replacement for portability. Do not use the
616: * gmtime_r() or gmtime() functions anywhere else but here.
617: *
618: */
619:
620: CURLcode Curl_gmtime(time_t intime, struct tm *store)
621: {
622: const struct tm *tm;
623: #ifdef HAVE_GMTIME_R
624: /* thread-safe version */
625: tm = (struct tm *)gmtime_r(&intime, store);
626: #else
627: tm = gmtime(&intime);
628: if(tm)
629: *store = *tm; /* copy the pointed struct to the local copy */
630: #endif
631:
632: if(!tm)
633: return CURLE_BAD_FUNCTION_ARGUMENT;
634: return CURLE_OK;
635: }
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>
![[BACK]](/icons/cvsweb/back.gif){kind=icon}
Return to parsedate.c CVS log ![[TXT]](/icons/cvsweb/text.gif){kind=icon}
![[DIR]](/icons/cvsweb/dir.gif){kind=icon}
Up to [ELWIX - Embedded LightWeight unIX -] / embedaddon / curl / lib