1 /***************************************************************************
2 * _ _ ____ _
3 * Project ___| | | | _ \| |
4 * / __| | | | |_) | |
5 * | (__| |_| | _ <| |___
6 * \___|\___/|_| \_\_____|
7 *
8 * Copyright (C) 1998 - 2008, 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 http://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 * $Id: parsedate.c,v 1.36 2008-10-23 11:49:19 bagder Exp $
22 ***************************************************************************/
23 /*
24 A brief summary of the date string formats this parser groks:
25
26 RFC 2616 3.3.1
27
28 Sun, 06 Nov 1994 08:49:37 GMT ; RFC 822, updated by RFC 1123
29 Sunday, 06-Nov-94 08:49:37 GMT ; RFC 850, obsoleted by RFC 1036
30 Sun Nov 6 08:49:37 1994 ; ANSI C's asctime() format
31
32 we support dates without week day name:
33
34 06 Nov 1994 08:49:37 GMT
35 06-Nov-94 08:49:37 GMT
36 Nov 6 08:49:37 1994
37
38 without the time zone:
39
40 06 Nov 1994 08:49:37
41 06-Nov-94 08:49:37
42
43 weird order:
44
45 1994 Nov 6 08:49:37 (GNU date fails)
46 GMT 08:49:37 06-Nov-94 Sunday
47 94 6 Nov 08:49:37 (GNU date fails)
48
49 time left out:
50
51 1994 Nov 6
52 06-Nov-94
53 Sun Nov 6 94
54
55 unusual separators:
56
57 1994.Nov.6
58 Sun/Nov/6/94/GMT
59
60 commonly used time zone names:
61
62 Sun, 06 Nov 1994 08:49:37 CET
63 06 Nov 1994 08:49:37 EST
64
65 time zones specified using RFC822 style:
66
67 Sun, 12 Sep 2004 15:05:58 -0700
68 Sat, 11 Sep 2004 21:32:11 +0200
69
70 compact numerical date strings:
71
72 20040912 15:05:58 -0700
73 20040911 +0200
74
75 */
76 #include "setup.h"
77 #include <stdio.h>
78 #include <ctype.h>
79 #include <string.h>
80
81 #ifdef HAVE_STDLIB_H
82 #include <stdlib.h> /* for strtol() */
83 #endif
84
85 #include <curl/curl.h>
86 #include "rawstr.h"
87 #include "parsedate.h"
88
89 const char * const Curl_wkday[] =
90 {"Mon", "Tue", "Wed", "Thu", "Fri", "Sat", "Sun"};
91 static const char * const weekday[] =
92 { "Monday", "Tuesday", "Wednesday", "Thursday",
93 "Friday", "Saturday", "Sunday" };
94 const char * const Curl_month[]=
95 { "Jan", "Feb", "Mar", "Apr", "May", "Jun",
96 "Jul", "Aug", "Sep", "Oct", "Nov", "Dec" };
97
98 struct tzinfo {
99 char name[5];
100 int offset; /* +/- in minutes */
101 };
102
103 /* Here's a bunch of frequently used time zone names. These were supported
104 by the old getdate parser. */
105 #define tDAYZONE -60 /* offset for daylight savings time */
106 static const struct tzinfo tz[]= {
107 {"GMT", 0}, /* Greenwich Mean */
108 {"UTC", 0}, /* Universal (Coordinated) */
109 {"WET", 0}, /* Western European */
110 {"BST", 0 tDAYZONE}, /* British Summer */
111 {"WAT", 60}, /* West Africa */
112 {"AST", 240}, /* Atlantic Standard */
113 {"ADT", 240 tDAYZONE}, /* Atlantic Daylight */
114 {"EST", 300}, /* Eastern Standard */
115 {"EDT", 300 tDAYZONE}, /* Eastern Daylight */
116 {"CST", 360}, /* Central Standard */
117 {"CDT", 360 tDAYZONE}, /* Central Daylight */
118 {"MST", 420}, /* Mountain Standard */
119 {"MDT", 420 tDAYZONE}, /* Mountain Daylight */
120 {"PST", 480}, /* Pacific Standard */
121 {"PDT", 480 tDAYZONE}, /* Pacific Daylight */
122 {"YST", 540}, /* Yukon Standard */
123 {"YDT", 540 tDAYZONE}, /* Yukon Daylight */
124 {"HST", 600}, /* Hawaii Standard */
125 {"HDT", 600 tDAYZONE}, /* Hawaii Daylight */
126 {"CAT", 600}, /* Central Alaska */
127 {"AHST", 600}, /* Alaska-Hawaii Standard */
128 {"NT", 660}, /* Nome */
129 {"IDLW", 720}, /* International Date Line West */
130 {"CET", -60}, /* Central European */
131 {"MET", -60}, /* Middle European */
132 {"MEWT", -60}, /* Middle European Winter */
133 {"MEST", -60 tDAYZONE}, /* Middle European Summer */
134 {"CEST", -60 tDAYZONE}, /* Central European Summer */
135 {"MESZ", -60 tDAYZONE}, /* Middle European Summer */
136 {"FWT", -60}, /* French Winter */
137 {"FST", -60 tDAYZONE}, /* French Summer */
138 {"EET", -120}, /* Eastern Europe, USSR Zone 1 */
139 {"WAST", -420}, /* West Australian Standard */
140 {"WADT", -420 tDAYZONE}, /* West Australian Daylight */
141 {"CCT", -480}, /* China Coast, USSR Zone 7 */
142 {"JST", -540}, /* Japan Standard, USSR Zone 8 */
143 {"EAST", -600}, /* Eastern Australian Standard */
144 {"EADT", -600 tDAYZONE}, /* Eastern Australian Daylight */
145 {"GST", -600}, /* Guam Standard, USSR Zone 9 */
146 {"NZT", -720}, /* New Zealand */
147 {"NZST", -720}, /* New Zealand Standard */
148 {"NZDT", -720 tDAYZONE}, /* New Zealand Daylight */
149 {"IDLE", -720}, /* International Date Line East */
150 };
151
152 /* returns:
153 -1 no day
154 0 monday - 6 sunday
155 */
156
checkday(const char * check,size_t len)157 static int checkday(const char *check, size_t len)
158 {
159 int i;
160 const char * const *what;
161 bool found= FALSE;
162 if(len > 3)
163 what = &weekday[0];
164 else
165 what = &Curl_wkday[0];
166 for(i=0; i<7; i++) {
167 if(Curl_raw_equal(check, what[0])) {
168 found=TRUE;
169 break;
170 }
171 what++;
172 }
173 return found?i:-1;
174 }
175
checkmonth(const char * check)176 static int checkmonth(const char *check)
177 {
178 int i;
179 const char * const *what;
180 bool found= FALSE;
181
182 what = &Curl_month[0];
183 for(i=0; i<12; i++) {
184 if(Curl_raw_equal(check, what[0])) {
185 found=TRUE;
186 break;
187 }
188 what++;
189 }
190 return found?i:-1; /* return the offset or -1, no real offset is -1 */
191 }
192
193 /* return the time zone offset between GMT and the input one, in number
194 of seconds or -1 if the timezone wasn't found/legal */
195
checktz(const char * check)196 static int checktz(const char *check)
197 {
198 unsigned int i;
199 const struct tzinfo *what;
200 bool found= FALSE;
201
202 what = tz;
203 for(i=0; i< sizeof(tz)/sizeof(tz[0]); i++) {
204 if(Curl_raw_equal(check, what->name)) {
205 found=TRUE;
206 break;
207 }
208 what++;
209 }
210 return found?what->offset*60:-1;
211 }
212
skip(const char ** date)213 static void skip(const char **date)
214 {
215 /* skip everything that aren't letters or digits */
216 while(**date && !ISALNUM(**date))
217 (*date)++;
218 }
219
220 enum assume {
221 DATE_MDAY,
222 DATE_YEAR,
223 DATE_TIME
224 };
225
226 /* this is a clone of 'struct tm' but with all fields we don't need or use
227 cut out */
228 struct my_tm {
229 int tm_sec;
230 int tm_min;
231 int tm_hour;
232 int tm_mday;
233 int tm_mon;
234 int tm_year;
235 };
236
237 /* struct tm to time since epoch in GMT time zone.
238 * This is similar to the standard mktime function but for GMT only, and
239 * doesn't suffer from the various bugs and portability problems that
240 * some systems' implementations have.
241 */
my_timegm(struct my_tm * tm)242 static time_t my_timegm(struct my_tm *tm)
243 {
244 static const int month_days_cumulative [12] =
245 { 0, 31, 59, 90, 120, 151, 181, 212, 243, 273, 304, 334 };
246 int month, year, leap_days;
247
248 if(tm->tm_year < 70)
249 /* we don't support years before 1970 as they will cause this function
250 to return a negative value */
251 return -1;
252
253 year = tm->tm_year + 1900;
254 month = tm->tm_mon;
255 if (month < 0) {
256 year += (11 - month) / 12;
257 month = 11 - (11 - month) % 12;
258 }
259 else if (month >= 12) {
260 year -= month / 12;
261 month = month % 12;
262 }
263
264 leap_days = year - (tm->tm_mon <= 1);
265 leap_days = ((leap_days / 4) - (leap_days / 100) + (leap_days / 400)
266 - (1969 / 4) + (1969 / 100) - (1969 / 400));
267
268 return ((((time_t) (year - 1970) * 365
269 + leap_days + month_days_cumulative [month] + tm->tm_mday - 1) * 24
270 + tm->tm_hour) * 60 + tm->tm_min) * 60 + tm->tm_sec;
271 }
272
parsedate(const char * date)273 static time_t parsedate(const char *date)
274 {
275 time_t t = 0;
276 int wdaynum=-1; /* day of the week number, 0-6 (mon-sun) */
277 int monnum=-1; /* month of the year number, 0-11 */
278 int mdaynum=-1; /* day of month, 1 - 31 */
279 int hournum=-1;
280 int minnum=-1;
281 int secnum=-1;
282 int yearnum=-1;
283 int tzoff=-1;
284 struct my_tm tm;
285 enum assume dignext = DATE_MDAY;
286 const char *indate = date; /* save the original pointer */
287 int part = 0; /* max 6 parts */
288
289 while(*date && (part < 6)) {
290 bool found=FALSE;
291
292 skip(&date);
293
294 if(ISALPHA(*date)) {
295 /* a name coming up */
296 char buf[32]="";
297 size_t len;
298 sscanf(date, "%31[ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz]",
299 buf);
300 len = strlen(buf);
301
302 if(wdaynum == -1) {
303 wdaynum = checkday(buf, len);
304 if(wdaynum != -1)
305 found = TRUE;
306 }
307 if(!found && (monnum == -1)) {
308 monnum = checkmonth(buf);
309 if(monnum != -1)
310 found = TRUE;
311 }
312
313 if(!found && (tzoff == -1)) {
314 /* this just must be a time zone string */
315 tzoff = checktz(buf);
316 if(tzoff != -1)
317 found = TRUE;
318 }
319
320 if(!found)
321 return -1; /* bad string */
322
323 date += len;
324 }
325 else if(ISDIGIT(*date)) {
326 /* a digit */
327 int val;
328 char *end;
329 if((secnum == -1) &&
330 (3 == sscanf(date, "%02d:%02d:%02d", &hournum, &minnum, &secnum))) {
331 /* time stamp! */
332 date += 8;
333 found = TRUE;
334 }
335 else {
336 val = (int)strtol(date, &end, 10);
337
338 if((tzoff == -1) &&
339 ((end - date) == 4) &&
340 (val <= 1400) &&
341 (indate< date) &&
342 ((date[-1] == '+' || date[-1] == '-'))) {
343 /* four digits and a value less than or equal to 1400 (to take into
344 account all sorts of funny time zone diffs) and it is preceeded
345 with a plus or minus. This is a time zone indication. 1400 is
346 picked since +1300 is frequently used and +1400 is mentioned as
347 an edge number in the document "ISO C 200X Proposal: Timezone
348 Functions" at http://david.tribble.com/text/c0xtimezone.html If
349 anyone has a more authoritative source for the exact maximum time
350 zone offsets, please speak up! */
351 found = TRUE;
352 tzoff = (val/100 * 60 + val%100)*60;
353
354 /* the + and - prefix indicates the local time compared to GMT,
355 this we need ther reversed math to get what we want */
356 tzoff = date[-1]=='+'?-tzoff:tzoff;
357 }
358
359 if(((end - date) == 8) &&
360 (yearnum == -1) &&
361 (monnum == -1) &&
362 (mdaynum == -1)) {
363 /* 8 digits, no year, month or day yet. This is YYYYMMDD */
364 found = TRUE;
365 yearnum = val/10000;
366 monnum = (val%10000)/100-1; /* month is 0 - 11 */
367 mdaynum = val%100;
368 }
369
370 if(!found && (dignext == DATE_MDAY) && (mdaynum == -1)) {
371 if((val > 0) && (val<32)) {
372 mdaynum = val;
373 found = TRUE;
374 }
375 dignext = DATE_YEAR;
376 }
377
378 if(!found && (dignext == DATE_YEAR) && (yearnum == -1)) {
379 yearnum = val;
380 found = TRUE;
381 if(yearnum < 1900) {
382 if(yearnum > 70)
383 yearnum += 1900;
384 else
385 yearnum += 2000;
386 }
387 if(mdaynum == -1)
388 dignext = DATE_MDAY;
389 }
390
391 if(!found)
392 return -1;
393
394 date = end;
395 }
396 }
397
398 part++;
399 }
400
401 if(-1 == secnum)
402 secnum = minnum = hournum = 0; /* no time, make it zero */
403
404 if((-1 == mdaynum) ||
405 (-1 == monnum) ||
406 (-1 == yearnum))
407 /* lacks vital info, fail */
408 return -1;
409
410 #if SIZEOF_TIME_T < 5
411 /* 32 bit time_t can only hold dates to the beginning of 2038 */
412 if(yearnum > 2037)
413 return 0x7fffffff;
414 #endif
415
416 tm.tm_sec = secnum;
417 tm.tm_min = minnum;
418 tm.tm_hour = hournum;
419 tm.tm_mday = mdaynum;
420 tm.tm_mon = monnum;
421 tm.tm_year = yearnum - 1900;
422
423 /* my_timegm() returns a time_t. time_t is often 32 bits, even on many
424 architectures that feature 64 bit 'long'.
425
426 Some systems have 64 bit time_t and deal with years beyond 2038. However,
427 even on some of the systems with 64 bit time_t mktime() returns -1 for
428 dates beyond 03:14:07 UTC, January 19, 2038. (Such as AIX 5100-06)
429 */
430 t = my_timegm(&tm);
431
432 /* time zone adjust (cast t to int to compare to negative one) */
433 if(-1 != (int)t) {
434
435 /* Add the time zone diff between local time zone and GMT. */
436 long delta = (long)(tzoff!=-1?tzoff:0);
437
438 if((delta>0) && (t + delta < t))
439 return -1; /* time_t overflow */
440
441 t += delta;
442 }
443
444 return t;
445 }
446
curl_getdate(const char * p,const time_t * now)447 time_t curl_getdate(const char *p, const time_t *now)
448 {
449 (void)now;
450 return parsedate(p);
451 }
452