1 /*
2 * This code is in the public domain and has no copyright.
3 *
4 * This is a plain C recursive-descent translation of an old
5 * public-domain YACC grammar that has been used for parsing dates in
6 * very many open-source projects.
7 *
8 * Since the original authors were generous enough to donate their
9 * work to the public domain, I feel compelled to match their
10 * generosity.
11 *
12 * Tim Kientzle, February 2009.
13 */
14
15 /*
16 * Header comment from original getdate.y:
17 */
18
19 /*
20 ** Originally written by Steven M. Bellovin <smb@research.att.com> while
21 ** at the University of North Carolina at Chapel Hill. Later tweaked by
22 ** a couple of people on Usenet. Completely overhauled by Rich $alz
23 ** <rsalz@bbn.com> and Jim Berets <jberets@bbn.com> in August, 1990;
24 **
25 ** This grammar has 10 shift/reduce conflicts.
26 **
27 ** This code is in the public domain and has no copyright.
28 */
29
30 #ifndef CM_GET_DATE
31 #include "archive_platform.h"
32 #endif
33 #ifdef __FreeBSD__
34 #include <sys/cdefs.h>
35 __FBSDID("$FreeBSD$");
36 #endif
37
38 #include <ctype.h>
39 #include <stdio.h>
40 #include <stdlib.h>
41 #include <string.h>
42 #include <time.h>
43
44 #define __LIBARCHIVE_BUILD 1
45 #include "archive_getdate.h"
46
47 /* Basic time units. */
48 #define EPOCH 1970
49 #define MINUTE (60L)
50 #define HOUR (60L * MINUTE)
51 #define DAY (24L * HOUR)
52
53 /* Daylight-savings mode: on, off, or not yet known. */
54 enum DSTMODE { DSTon, DSToff, DSTmaybe };
55 /* Meridian: am or pm. */
56 enum { tAM, tPM };
57 /* Token types returned by nexttoken() */
58 enum { tAGO = 260, tDAY, tDAYZONE, tAMPM, tMONTH, tMONTH_UNIT, tSEC_UNIT,
59 tUNUMBER, tZONE, tDST };
60 struct token { int token; time_t value; };
61
62 /*
63 * Parser state.
64 */
65 struct gdstate {
66 struct token *tokenp; /* Pointer to next token. */
67 /* HaveXxxx counts how many of this kind of phrase we've seen;
68 * it's a fatal error to have more than one time, zone, day,
69 * or date phrase. */
70 int HaveYear;
71 int HaveMonth;
72 int HaveDay;
73 int HaveWeekDay; /* Day of week */
74 int HaveTime; /* Hour/minute/second */
75 int HaveZone; /* timezone and/or DST info */
76 int HaveRel; /* time offset; we can have more than one */
77 /* Absolute time values. */
78 time_t Timezone; /* Seconds offset from GMT */
79 time_t Day;
80 time_t Hour;
81 time_t Minutes;
82 time_t Month;
83 time_t Seconds;
84 time_t Year;
85 /* DST selection */
86 enum DSTMODE DSTmode;
87 /* Day of week accounting, e.g., "3rd Tuesday" */
88 time_t DayOrdinal; /* "3" in "3rd Tuesday" */
89 time_t DayNumber; /* "Tuesday" in "3rd Tuesday" */
90 /* Relative time values: hour/day/week offsets are measured in
91 * seconds, month/year are counted in months. */
92 time_t RelMonth;
93 time_t RelSeconds;
94 };
95
96 /*
97 * A series of functions that recognize certain common time phrases.
98 * Each function returns 1 if it managed to make sense of some of the
99 * tokens, zero otherwise.
100 */
101
102 /*
103 * hour:minute or hour:minute:second with optional AM, PM, or numeric
104 * timezone offset
105 */
106 static int
timephrase(struct gdstate * gds)107 timephrase(struct gdstate *gds)
108 {
109 if (gds->tokenp[0].token == tUNUMBER
110 && gds->tokenp[1].token == ':'
111 && gds->tokenp[2].token == tUNUMBER
112 && gds->tokenp[3].token == ':'
113 && gds->tokenp[4].token == tUNUMBER) {
114 /* "12:14:18" or "22:08:07" */
115 ++gds->HaveTime;
116 gds->Hour = gds->tokenp[0].value;
117 gds->Minutes = gds->tokenp[2].value;
118 gds->Seconds = gds->tokenp[4].value;
119 gds->tokenp += 5;
120 }
121 else if (gds->tokenp[0].token == tUNUMBER
122 && gds->tokenp[1].token == ':'
123 && gds->tokenp[2].token == tUNUMBER) {
124 /* "12:14" or "22:08" */
125 ++gds->HaveTime;
126 gds->Hour = gds->tokenp[0].value;
127 gds->Minutes = gds->tokenp[2].value;
128 gds->Seconds = 0;
129 gds->tokenp += 3;
130 }
131 else if (gds->tokenp[0].token == tUNUMBER
132 && gds->tokenp[1].token == tAMPM) {
133 /* "7" is a time if it's followed by "am" or "pm" */
134 ++gds->HaveTime;
135 gds->Hour = gds->tokenp[0].value;
136 gds->Minutes = gds->Seconds = 0;
137 /* We'll handle the AM/PM below. */
138 gds->tokenp += 1;
139 } else {
140 /* We can't handle this. */
141 return 0;
142 }
143
144 if (gds->tokenp[0].token == tAMPM) {
145 /* "7:12pm", "12:20:13am" */
146 if (gds->Hour == 12)
147 gds->Hour = 0;
148 if (gds->tokenp[0].value == tPM)
149 gds->Hour += 12;
150 gds->tokenp += 1;
151 }
152 if (gds->tokenp[0].token == '+'
153 && gds->tokenp[1].token == tUNUMBER) {
154 /* "7:14+0700" */
155 gds->HaveZone++;
156 gds->DSTmode = DSToff;
157 gds->Timezone = - ((gds->tokenp[1].value / 100) * HOUR
158 + (gds->tokenp[1].value % 100) * MINUTE);
159 gds->tokenp += 2;
160 }
161 if (gds->tokenp[0].token == '-'
162 && gds->tokenp[1].token == tUNUMBER) {
163 /* "19:14:12-0530" */
164 gds->HaveZone++;
165 gds->DSTmode = DSToff;
166 gds->Timezone = + ((gds->tokenp[1].value / 100) * HOUR
167 + (gds->tokenp[1].value % 100) * MINUTE);
168 gds->tokenp += 2;
169 }
170 return 1;
171 }
172
173 /*
174 * Timezone name, possibly including DST.
175 */
176 static int
zonephrase(struct gdstate * gds)177 zonephrase(struct gdstate *gds)
178 {
179 if (gds->tokenp[0].token == tZONE
180 && gds->tokenp[1].token == tDST) {
181 gds->HaveZone++;
182 gds->Timezone = gds->tokenp[0].value;
183 gds->DSTmode = DSTon;
184 gds->tokenp += 1;
185 return 1;
186 }
187
188 if (gds->tokenp[0].token == tZONE) {
189 gds->HaveZone++;
190 gds->Timezone = gds->tokenp[0].value;
191 gds->DSTmode = DSToff;
192 gds->tokenp += 1;
193 return 1;
194 }
195
196 if (gds->tokenp[0].token == tDAYZONE) {
197 gds->HaveZone++;
198 gds->Timezone = gds->tokenp[0].value;
199 gds->DSTmode = DSTon;
200 gds->tokenp += 1;
201 return 1;
202 }
203 return 0;
204 }
205
206 /*
207 * Year/month/day in various combinations.
208 */
209 static int
datephrase(struct gdstate * gds)210 datephrase(struct gdstate *gds)
211 {
212 if (gds->tokenp[0].token == tUNUMBER
213 && gds->tokenp[1].token == '/'
214 && gds->tokenp[2].token == tUNUMBER
215 && gds->tokenp[3].token == '/'
216 && gds->tokenp[4].token == tUNUMBER) {
217 gds->HaveYear++;
218 gds->HaveMonth++;
219 gds->HaveDay++;
220 if (gds->tokenp[0].value >= 13) {
221 /* First number is big: 2004/01/29, 99/02/17 */
222 gds->Year = gds->tokenp[0].value;
223 gds->Month = gds->tokenp[2].value;
224 gds->Day = gds->tokenp[4].value;
225 } else if ((gds->tokenp[4].value >= 13)
226 || (gds->tokenp[2].value >= 13)) {
227 /* Last number is big: 01/07/98 */
228 /* Middle number is big: 01/29/04 */
229 gds->Month = gds->tokenp[0].value;
230 gds->Day = gds->tokenp[2].value;
231 gds->Year = gds->tokenp[4].value;
232 } else {
233 /* No significant clues: 02/03/04 */
234 gds->Month = gds->tokenp[0].value;
235 gds->Day = gds->tokenp[2].value;
236 gds->Year = gds->tokenp[4].value;
237 }
238 gds->tokenp += 5;
239 return 1;
240 }
241
242 if (gds->tokenp[0].token == tUNUMBER
243 && gds->tokenp[1].token == '/'
244 && gds->tokenp[2].token == tUNUMBER) {
245 /* "1/15" */
246 gds->HaveMonth++;
247 gds->HaveDay++;
248 gds->Month = gds->tokenp[0].value;
249 gds->Day = gds->tokenp[2].value;
250 gds->tokenp += 3;
251 return 1;
252 }
253
254 if (gds->tokenp[0].token == tUNUMBER
255 && gds->tokenp[1].token == '-'
256 && gds->tokenp[2].token == tUNUMBER
257 && gds->tokenp[3].token == '-'
258 && gds->tokenp[4].token == tUNUMBER) {
259 /* ISO 8601 format. yyyy-mm-dd. */
260 gds->HaveYear++;
261 gds->HaveMonth++;
262 gds->HaveDay++;
263 gds->Year = gds->tokenp[0].value;
264 gds->Month = gds->tokenp[2].value;
265 gds->Day = gds->tokenp[4].value;
266 gds->tokenp += 5;
267 return 1;
268 }
269
270 if (gds->tokenp[0].token == tUNUMBER
271 && gds->tokenp[1].token == '-'
272 && gds->tokenp[2].token == tMONTH
273 && gds->tokenp[3].token == '-'
274 && gds->tokenp[4].token == tUNUMBER) {
275 gds->HaveYear++;
276 gds->HaveMonth++;
277 gds->HaveDay++;
278 if (gds->tokenp[0].value > 31) {
279 /* e.g. 1992-Jun-17 */
280 gds->Year = gds->tokenp[0].value;
281 gds->Month = gds->tokenp[2].value;
282 gds->Day = gds->tokenp[4].value;
283 } else {
284 /* e.g. 17-JUN-1992. */
285 gds->Day = gds->tokenp[0].value;
286 gds->Month = gds->tokenp[2].value;
287 gds->Year = gds->tokenp[4].value;
288 }
289 gds->tokenp += 5;
290 return 1;
291 }
292
293 if (gds->tokenp[0].token == tMONTH
294 && gds->tokenp[1].token == tUNUMBER
295 && gds->tokenp[2].token == ','
296 && gds->tokenp[3].token == tUNUMBER) {
297 /* "June 17, 2001" */
298 gds->HaveYear++;
299 gds->HaveMonth++;
300 gds->HaveDay++;
301 gds->Month = gds->tokenp[0].value;
302 gds->Day = gds->tokenp[1].value;
303 gds->Year = gds->tokenp[3].value;
304 gds->tokenp += 4;
305 return 1;
306 }
307
308 if (gds->tokenp[0].token == tMONTH
309 && gds->tokenp[1].token == tUNUMBER) {
310 /* "May 3" */
311 gds->HaveMonth++;
312 gds->HaveDay++;
313 gds->Month = gds->tokenp[0].value;
314 gds->Day = gds->tokenp[1].value;
315 gds->tokenp += 2;
316 return 1;
317 }
318
319 if (gds->tokenp[0].token == tUNUMBER
320 && gds->tokenp[1].token == tMONTH
321 && gds->tokenp[2].token == tUNUMBER) {
322 /* "12 Sept 1997" */
323 gds->HaveYear++;
324 gds->HaveMonth++;
325 gds->HaveDay++;
326 gds->Day = gds->tokenp[0].value;
327 gds->Month = gds->tokenp[1].value;
328 gds->Year = gds->tokenp[2].value;
329 gds->tokenp += 3;
330 return 1;
331 }
332
333 if (gds->tokenp[0].token == tUNUMBER
334 && gds->tokenp[1].token == tMONTH) {
335 /* "12 Sept" */
336 gds->HaveMonth++;
337 gds->HaveDay++;
338 gds->Day = gds->tokenp[0].value;
339 gds->Month = gds->tokenp[1].value;
340 gds->tokenp += 2;
341 return 1;
342 }
343
344 return 0;
345 }
346
347 /*
348 * Relative time phrase: "tomorrow", "yesterday", "+1 hour", etc.
349 */
350 static int
relunitphrase(struct gdstate * gds)351 relunitphrase(struct gdstate *gds)
352 {
353 if (gds->tokenp[0].token == '-'
354 && gds->tokenp[1].token == tUNUMBER
355 && gds->tokenp[2].token == tSEC_UNIT) {
356 /* "-3 hours" */
357 gds->HaveRel++;
358 gds->RelSeconds -= gds->tokenp[1].value * gds->tokenp[2].value;
359 gds->tokenp += 3;
360 return 1;
361 }
362 if (gds->tokenp[0].token == '+'
363 && gds->tokenp[1].token == tUNUMBER
364 && gds->tokenp[2].token == tSEC_UNIT) {
365 /* "+1 minute" */
366 gds->HaveRel++;
367 gds->RelSeconds += gds->tokenp[1].value * gds->tokenp[2].value;
368 gds->tokenp += 3;
369 return 1;
370 }
371 if (gds->tokenp[0].token == tUNUMBER
372 && gds->tokenp[1].token == tSEC_UNIT) {
373 /* "1 day" */
374 gds->HaveRel++;
375 gds->RelSeconds += gds->tokenp[0].value * gds->tokenp[1].value;
376 gds->tokenp += 2;
377 return 1;
378 }
379 if (gds->tokenp[0].token == '-'
380 && gds->tokenp[1].token == tUNUMBER
381 && gds->tokenp[2].token == tMONTH_UNIT) {
382 /* "-3 months" */
383 gds->HaveRel++;
384 gds->RelMonth -= gds->tokenp[1].value * gds->tokenp[2].value;
385 gds->tokenp += 3;
386 return 1;
387 }
388 if (gds->tokenp[0].token == '+'
389 && gds->tokenp[1].token == tUNUMBER
390 && gds->tokenp[2].token == tMONTH_UNIT) {
391 /* "+5 years" */
392 gds->HaveRel++;
393 gds->RelMonth += gds->tokenp[1].value * gds->tokenp[2].value;
394 gds->tokenp += 3;
395 return 1;
396 }
397 if (gds->tokenp[0].token == tUNUMBER
398 && gds->tokenp[1].token == tMONTH_UNIT) {
399 /* "2 years" */
400 gds->HaveRel++;
401 gds->RelMonth += gds->tokenp[0].value * gds->tokenp[1].value;
402 gds->tokenp += 2;
403 return 1;
404 }
405 if (gds->tokenp[0].token == tSEC_UNIT) {
406 /* "now", "tomorrow" */
407 gds->HaveRel++;
408 gds->RelSeconds += gds->tokenp[0].value;
409 gds->tokenp += 1;
410 return 1;
411 }
412 if (gds->tokenp[0].token == tMONTH_UNIT) {
413 /* "month" */
414 gds->HaveRel++;
415 gds->RelMonth += gds->tokenp[0].value;
416 gds->tokenp += 1;
417 return 1;
418 }
419 return 0;
420 }
421
422 /*
423 * Day of the week specification.
424 */
425 static int
dayphrase(struct gdstate * gds)426 dayphrase(struct gdstate *gds)
427 {
428 if (gds->tokenp[0].token == tDAY) {
429 /* "tues", "wednesday," */
430 gds->HaveWeekDay++;
431 gds->DayOrdinal = 1;
432 gds->DayNumber = gds->tokenp[0].value;
433 gds->tokenp += 1;
434 if (gds->tokenp[0].token == ',')
435 gds->tokenp += 1;
436 return 1;
437 }
438 if (gds->tokenp[0].token == tUNUMBER
439 && gds->tokenp[1].token == tDAY) {
440 /* "second tues" "3 wed" */
441 gds->HaveWeekDay++;
442 gds->DayOrdinal = gds->tokenp[0].value;
443 gds->DayNumber = gds->tokenp[1].value;
444 gds->tokenp += 2;
445 return 1;
446 }
447 return 0;
448 }
449
450 /*
451 * Try to match a phrase using one of the above functions.
452 * This layer also deals with a couple of generic issues.
453 */
454 static int
phrase(struct gdstate * gds)455 phrase(struct gdstate *gds)
456 {
457 if (timephrase(gds))
458 return 1;
459 if (zonephrase(gds))
460 return 1;
461 if (datephrase(gds))
462 return 1;
463 if (dayphrase(gds))
464 return 1;
465 if (relunitphrase(gds)) {
466 if (gds->tokenp[0].token == tAGO) {
467 gds->RelSeconds = -gds->RelSeconds;
468 gds->RelMonth = -gds->RelMonth;
469 gds->tokenp += 1;
470 }
471 return 1;
472 }
473
474 /* Bare numbers sometimes have meaning. */
475 if (gds->tokenp[0].token == tUNUMBER) {
476 if (gds->HaveTime && !gds->HaveYear && !gds->HaveRel) {
477 gds->HaveYear++;
478 gds->Year = gds->tokenp[0].value;
479 gds->tokenp += 1;
480 return 1;
481 }
482
483 if(gds->tokenp[0].value > 10000) {
484 /* "20040301" */
485 gds->HaveYear++;
486 gds->HaveMonth++;
487 gds->HaveDay++;
488 gds->Day= (gds->tokenp[0].value)%100;
489 gds->Month= (gds->tokenp[0].value/100)%100;
490 gds->Year = gds->tokenp[0].value/10000;
491 gds->tokenp += 1;
492 return 1;
493 }
494
495 if (gds->tokenp[0].value < 24) {
496 gds->HaveTime++;
497 gds->Hour = gds->tokenp[0].value;
498 gds->Minutes = 0;
499 gds->Seconds = 0;
500 gds->tokenp += 1;
501 return 1;
502 }
503
504 if ((gds->tokenp[0].value / 100 < 24)
505 && (gds->tokenp[0].value % 100 < 60)) {
506 /* "513" is same as "5:13" */
507 gds->Hour = gds->tokenp[0].value / 100;
508 gds->Minutes = gds->tokenp[0].value % 100;
509 gds->Seconds = 0;
510 gds->tokenp += 1;
511 return 1;
512 }
513 }
514
515 return 0;
516 }
517
518 /*
519 * A dictionary of time words.
520 */
521 static struct LEXICON {
522 size_t abbrev;
523 const char *name;
524 int type;
525 time_t value;
526 } const TimeWords[] = {
527 /* am/pm */
528 { 0, "am", tAMPM, tAM },
529 { 0, "pm", tAMPM, tPM },
530
531 /* Month names. */
532 { 3, "january", tMONTH, 1 },
533 { 3, "february", tMONTH, 2 },
534 { 3, "march", tMONTH, 3 },
535 { 3, "april", tMONTH, 4 },
536 { 3, "may", tMONTH, 5 },
537 { 3, "june", tMONTH, 6 },
538 { 3, "july", tMONTH, 7 },
539 { 3, "august", tMONTH, 8 },
540 { 3, "september", tMONTH, 9 },
541 { 3, "october", tMONTH, 10 },
542 { 3, "november", tMONTH, 11 },
543 { 3, "december", tMONTH, 12 },
544
545 /* Days of the week. */
546 { 2, "sunday", tDAY, 0 },
547 { 3, "monday", tDAY, 1 },
548 { 2, "tuesday", tDAY, 2 },
549 { 3, "wednesday", tDAY, 3 },
550 { 2, "thursday", tDAY, 4 },
551 { 2, "friday", tDAY, 5 },
552 { 2, "saturday", tDAY, 6 },
553
554 /* Timezones: Offsets are in seconds. */
555 { 0, "gmt", tZONE, 0*HOUR }, /* Greenwich Mean */
556 { 0, "ut", tZONE, 0*HOUR }, /* Universal (Coordinated) */
557 { 0, "utc", tZONE, 0*HOUR },
558 { 0, "wet", tZONE, 0*HOUR }, /* Western European */
559 { 0, "bst", tDAYZONE, 0*HOUR }, /* British Summer */
560 { 0, "wat", tZONE, 1*HOUR }, /* West Africa */
561 { 0, "at", tZONE, 2*HOUR }, /* Azores */
562 /* { 0, "bst", tZONE, 3*HOUR }, */ /* Brazil Standard: Conflict */
563 /* { 0, "gst", tZONE, 3*HOUR }, */ /* Greenland Standard: Conflict*/
564 { 0, "nft", tZONE, 3*HOUR+30*MINUTE }, /* Newfoundland */
565 { 0, "nst", tZONE, 3*HOUR+30*MINUTE }, /* Newfoundland Standard */
566 { 0, "ndt", tDAYZONE, 3*HOUR+30*MINUTE }, /* Newfoundland Daylight */
567 { 0, "ast", tZONE, 4*HOUR }, /* Atlantic Standard */
568 { 0, "adt", tDAYZONE, 4*HOUR }, /* Atlantic Daylight */
569 { 0, "est", tZONE, 5*HOUR }, /* Eastern Standard */
570 { 0, "edt", tDAYZONE, 5*HOUR }, /* Eastern Daylight */
571 { 0, "cst", tZONE, 6*HOUR }, /* Central Standard */
572 { 0, "cdt", tDAYZONE, 6*HOUR }, /* Central Daylight */
573 { 0, "mst", tZONE, 7*HOUR }, /* Mountain Standard */
574 { 0, "mdt", tDAYZONE, 7*HOUR }, /* Mountain Daylight */
575 { 0, "pst", tZONE, 8*HOUR }, /* Pacific Standard */
576 { 0, "pdt", tDAYZONE, 8*HOUR }, /* Pacific Daylight */
577 { 0, "yst", tZONE, 9*HOUR }, /* Yukon Standard */
578 { 0, "ydt", tDAYZONE, 9*HOUR }, /* Yukon Daylight */
579 { 0, "hst", tZONE, 10*HOUR }, /* Hawaii Standard */
580 { 0, "hdt", tDAYZONE, 10*HOUR }, /* Hawaii Daylight */
581 { 0, "cat", tZONE, 10*HOUR }, /* Central Alaska */
582 { 0, "ahst", tZONE, 10*HOUR }, /* Alaska-Hawaii Standard */
583 { 0, "nt", tZONE, 11*HOUR }, /* Nome */
584 { 0, "idlw", tZONE, 12*HOUR }, /* Intl Date Line West */
585 { 0, "cet", tZONE, -1*HOUR }, /* Central European */
586 { 0, "met", tZONE, -1*HOUR }, /* Middle European */
587 { 0, "mewt", tZONE, -1*HOUR }, /* Middle European Winter */
588 { 0, "mest", tDAYZONE, -1*HOUR }, /* Middle European Summer */
589 { 0, "swt", tZONE, -1*HOUR }, /* Swedish Winter */
590 { 0, "sst", tDAYZONE, -1*HOUR }, /* Swedish Summer */
591 { 0, "fwt", tZONE, -1*HOUR }, /* French Winter */
592 { 0, "fst", tDAYZONE, -1*HOUR }, /* French Summer */
593 { 0, "eet", tZONE, -2*HOUR }, /* Eastern Eur, USSR Zone 1 */
594 { 0, "bt", tZONE, -3*HOUR }, /* Baghdad, USSR Zone 2 */
595 { 0, "it", tZONE, -3*HOUR-30*MINUTE },/* Iran */
596 { 0, "zp4", tZONE, -4*HOUR }, /* USSR Zone 3 */
597 { 0, "zp5", tZONE, -5*HOUR }, /* USSR Zone 4 */
598 { 0, "ist", tZONE, -5*HOUR-30*MINUTE },/* Indian Standard */
599 { 0, "zp6", tZONE, -6*HOUR }, /* USSR Zone 5 */
600 /* { 0, "nst", tZONE, -6.5*HOUR }, */ /* North Sumatra: Conflict */
601 /* { 0, "sst", tZONE, -7*HOUR }, */ /* So Sumatra, USSR 6: Conflict */
602 { 0, "wast", tZONE, -7*HOUR }, /* West Australian Standard */
603 { 0, "wadt", tDAYZONE, -7*HOUR }, /* West Australian Daylight */
604 { 0, "jt", tZONE, -7*HOUR-30*MINUTE },/* Java (3pm in Cronusland!)*/
605 { 0, "cct", tZONE, -8*HOUR }, /* China Coast, USSR Zone 7 */
606 { 0, "jst", tZONE, -9*HOUR }, /* Japan Std, USSR Zone 8 */
607 { 0, "cast", tZONE, -9*HOUR-30*MINUTE },/* Ctrl Australian Std */
608 { 0, "cadt", tDAYZONE, -9*HOUR-30*MINUTE },/* Ctrl Australian Daylt */
609 { 0, "east", tZONE, -10*HOUR }, /* Eastern Australian Std */
610 { 0, "eadt", tDAYZONE, -10*HOUR }, /* Eastern Australian Daylt */
611 { 0, "gst", tZONE, -10*HOUR }, /* Guam Std, USSR Zone 9 */
612 { 0, "nzt", tZONE, -12*HOUR }, /* New Zealand */
613 { 0, "nzst", tZONE, -12*HOUR }, /* New Zealand Standard */
614 { 0, "nzdt", tDAYZONE, -12*HOUR }, /* New Zealand Daylight */
615 { 0, "idle", tZONE, -12*HOUR }, /* Intl Date Line East */
616
617 { 0, "dst", tDST, 0 },
618
619 /* Time units. */
620 { 4, "years", tMONTH_UNIT, 12 },
621 { 5, "months", tMONTH_UNIT, 1 },
622 { 9, "fortnights", tSEC_UNIT, 14 * DAY },
623 { 4, "weeks", tSEC_UNIT, 7 * DAY },
624 { 3, "days", tSEC_UNIT, DAY },
625 { 4, "hours", tSEC_UNIT, HOUR },
626 { 3, "minutes", tSEC_UNIT, MINUTE },
627 { 3, "seconds", tSEC_UNIT, 1 },
628
629 /* Relative-time words. */
630 { 0, "tomorrow", tSEC_UNIT, DAY },
631 { 0, "yesterday", tSEC_UNIT, -DAY },
632 { 0, "today", tSEC_UNIT, 0 },
633 { 0, "now", tSEC_UNIT, 0 },
634 { 0, "last", tUNUMBER, -1 },
635 { 0, "this", tSEC_UNIT, 0 },
636 { 0, "next", tUNUMBER, 2 },
637 { 0, "first", tUNUMBER, 1 },
638 { 0, "1st", tUNUMBER, 1 },
639 /* { 0, "second", tUNUMBER, 2 }, */
640 { 0, "2nd", tUNUMBER, 2 },
641 { 0, "third", tUNUMBER, 3 },
642 { 0, "3rd", tUNUMBER, 3 },
643 { 0, "fourth", tUNUMBER, 4 },
644 { 0, "4th", tUNUMBER, 4 },
645 { 0, "fifth", tUNUMBER, 5 },
646 { 0, "5th", tUNUMBER, 5 },
647 { 0, "sixth", tUNUMBER, 6 },
648 { 0, "seventh", tUNUMBER, 7 },
649 { 0, "eighth", tUNUMBER, 8 },
650 { 0, "ninth", tUNUMBER, 9 },
651 { 0, "tenth", tUNUMBER, 10 },
652 { 0, "eleventh", tUNUMBER, 11 },
653 { 0, "twelfth", tUNUMBER, 12 },
654 { 0, "ago", tAGO, 1 },
655
656 /* Military timezones. */
657 { 0, "a", tZONE, 1*HOUR },
658 { 0, "b", tZONE, 2*HOUR },
659 { 0, "c", tZONE, 3*HOUR },
660 { 0, "d", tZONE, 4*HOUR },
661 { 0, "e", tZONE, 5*HOUR },
662 { 0, "f", tZONE, 6*HOUR },
663 { 0, "g", tZONE, 7*HOUR },
664 { 0, "h", tZONE, 8*HOUR },
665 { 0, "i", tZONE, 9*HOUR },
666 { 0, "k", tZONE, 10*HOUR },
667 { 0, "l", tZONE, 11*HOUR },
668 { 0, "m", tZONE, 12*HOUR },
669 { 0, "n", tZONE, -1*HOUR },
670 { 0, "o", tZONE, -2*HOUR },
671 { 0, "p", tZONE, -3*HOUR },
672 { 0, "q", tZONE, -4*HOUR },
673 { 0, "r", tZONE, -5*HOUR },
674 { 0, "s", tZONE, -6*HOUR },
675 { 0, "t", tZONE, -7*HOUR },
676 { 0, "u", tZONE, -8*HOUR },
677 { 0, "v", tZONE, -9*HOUR },
678 { 0, "w", tZONE, -10*HOUR },
679 { 0, "x", tZONE, -11*HOUR },
680 { 0, "y", tZONE, -12*HOUR },
681 { 0, "z", tZONE, 0*HOUR },
682
683 /* End of table. */
684 { 0, NULL, 0, 0 }
685 };
686
687 /*
688 * Year is either:
689 * = A number from 0 to 99, which means a year from 1970 to 2069, or
690 * = The actual year (>=100).
691 */
692 static time_t
Convert(time_t Month,time_t Day,time_t Year,time_t Hours,time_t Minutes,time_t Seconds,time_t Timezone,enum DSTMODE DSTmode)693 Convert(time_t Month, time_t Day, time_t Year,
694 time_t Hours, time_t Minutes, time_t Seconds,
695 time_t Timezone, enum DSTMODE DSTmode)
696 {
697 signed char DaysInMonth[12] = {
698 31, 0, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31
699 };
700 time_t Julian;
701 int i;
702 struct tm *ltime;
703 #if defined(HAVE_LOCALTIME_R) || defined(HAVE__LOCALTIME64_S)
704 struct tm tmbuf;
705 #endif
706 #if defined(HAVE__LOCALTIME64_S)
707 errno_t terr;
708 __time64_t tmptime;
709 #endif
710
711 if (Year < 69)
712 Year += 2000;
713 else if (Year < 100)
714 Year += 1900;
715 DaysInMonth[1] = Year % 4 == 0 && (Year % 100 != 0 || Year % 400 == 0)
716 ? 29 : 28;
717 /* Checking for 2038 bogusly assumes that time_t is 32 bits. But
718 I'm too lazy to try to check for time_t overflow in another way. */
719 if (Year < EPOCH || Year > 2038
720 || Month < 1 || Month > 12
721 /* Lint fluff: "conversion from long may lose accuracy" */
722 || Day < 1 || Day > DaysInMonth[(int)--Month]
723 || Hours < 0 || Hours > 23
724 || Minutes < 0 || Minutes > 59
725 || Seconds < 0 || Seconds > 59)
726 return -1;
727
728 Julian = Day - 1;
729 for (i = 0; i < Month; i++)
730 Julian += DaysInMonth[i];
731 for (i = EPOCH; i < Year; i++)
732 Julian += 365 + (i % 4 == 0);
733 Julian *= DAY;
734 Julian += Timezone;
735 Julian += Hours * HOUR + Minutes * MINUTE + Seconds;
736 #if defined(HAVE_LOCALTIME_R)
737 ltime = localtime_r(&Julian, &tmbuf);
738 #elif defined(HAVE__LOCALTIME64_S)
739 tmptime = Julian;
740 terr = _localtime64_s(&tmbuf, &tmptime);
741 if (terr)
742 ltime = NULL;
743 else
744 ltime = &tmbuf;
745 #else
746 ltime = localtime(&Julian);
747 #endif
748 if (DSTmode == DSTon
749 || (DSTmode == DSTmaybe && ltime->tm_isdst))
750 Julian -= HOUR;
751 return Julian;
752 }
753
754 static time_t
DSTcorrect(time_t Start,time_t Future)755 DSTcorrect(time_t Start, time_t Future)
756 {
757 time_t StartDay;
758 time_t FutureDay;
759 struct tm *ltime;
760 #if defined(HAVE_LOCALTIME_R) || defined(HAVE__LOCALTIME64_S)
761 struct tm tmbuf;
762 #endif
763 #if defined(HAVE__LOCALTIME64_S)
764 errno_t terr;
765 __time64_t tmptime;
766 #endif
767
768 #if defined(HAVE_LOCALTIME_R)
769 ltime = localtime_r(&Start, &tmbuf);
770 #elif defined(HAVE__LOCALTIME64_S)
771 tmptime = Start;
772 terr = _localtime64_s(&tmbuf, &tmptime);
773 if (terr)
774 ltime = NULL;
775 else
776 ltime = &tmbuf;
777 #else
778 ltime = localtime(&Start);
779 #endif
780 StartDay = (ltime->tm_hour + 1) % 24;
781 #if defined(HAVE_LOCALTIME_R)
782 ltime = localtime_r(&Future, &tmbuf);
783 #elif defined(HAVE__LOCALTIME64_S)
784 tmptime = Future;
785 terr = _localtime64_s(&tmbuf, &tmptime);
786 if (terr)
787 ltime = NULL;
788 else
789 ltime = &tmbuf;
790 #else
791 ltime = localtime(&Future);
792 #endif
793 FutureDay = (ltime->tm_hour + 1) % 24;
794 return (Future - Start) + (StartDay - FutureDay) * HOUR;
795 }
796
797
798 static time_t
RelativeDate(time_t Start,time_t zone,int dstmode,time_t DayOrdinal,time_t DayNumber)799 RelativeDate(time_t Start, time_t zone, int dstmode,
800 time_t DayOrdinal, time_t DayNumber)
801 {
802 struct tm *tm;
803 time_t t, now;
804 #if defined(HAVE_GMTIME_R) || defined(HAVE__GMTIME64_S)
805 struct tm tmbuf;
806 #endif
807 #if defined(HAVE__GMTIME64_S)
808 errno_t terr;
809 __time64_t tmptime;
810 #endif
811
812 t = Start - zone;
813 #if defined(HAVE_GMTIME_R)
814 tm = gmtime_r(&t, &tmbuf);
815 #elif defined(HAVE__GMTIME64_S)
816 tmptime = t;
817 terr = _gmtime64_s(&tmbuf, &tmptime);
818 if (terr)
819 tm = NULL;
820 else
821 tm = &tmbuf;
822 #else
823 tm = gmtime(&t);
824 #endif
825 now = Start;
826 now += DAY * ((DayNumber - tm->tm_wday + 7) % 7);
827 now += 7 * DAY * (DayOrdinal <= 0 ? DayOrdinal : DayOrdinal - 1);
828 if (dstmode == DSTmaybe)
829 return DSTcorrect(Start, now);
830 return now - Start;
831 }
832
833
834 static time_t
RelativeMonth(time_t Start,time_t Timezone,time_t RelMonth)835 RelativeMonth(time_t Start, time_t Timezone, time_t RelMonth)
836 {
837 struct tm *tm;
838 time_t Month;
839 time_t Year;
840 #if defined(HAVE_LOCALTIME_R) || defined(HAVE__LOCALTIME64_S)
841 struct tm tmbuf;
842 #endif
843 #if defined(HAVE__LOCALTIME64_S)
844 errno_t terr;
845 __time64_t tmptime;
846 #endif
847
848 if (RelMonth == 0)
849 return 0;
850 #if defined(HAVE_LOCALTIME_R)
851 tm = localtime_r(&Start, &tmbuf);
852 #elif defined(HAVE__LOCALTIME64_S)
853 tmptime = Start;
854 terr = _localtime64_s(&tmbuf, &tmptime);
855 if (terr)
856 tm = NULL;
857 else
858 tm = &tmbuf;
859 #else
860 tm = localtime(&Start);
861 #endif
862 Month = 12 * (tm->tm_year + 1900) + tm->tm_mon + RelMonth;
863 Year = Month / 12;
864 Month = Month % 12 + 1;
865 return DSTcorrect(Start,
866 Convert(Month, (time_t)tm->tm_mday, Year,
867 (time_t)tm->tm_hour, (time_t)tm->tm_min, (time_t)tm->tm_sec,
868 Timezone, DSTmaybe));
869 }
870
871 /*
872 * Tokenizer.
873 */
874 static int
nexttoken(const char ** in,time_t * value)875 nexttoken(const char **in, time_t *value)
876 {
877 char c;
878 char buff[64];
879
880 for ( ; ; ) {
881 while (isspace((unsigned char)**in))
882 ++*in;
883
884 /* Skip parenthesized comments. */
885 if (**in == '(') {
886 int Count = 0;
887 do {
888 c = *(*in)++;
889 if (c == '\0')
890 return c;
891 if (c == '(')
892 Count++;
893 else if (c == ')')
894 Count--;
895 } while (Count > 0);
896 continue;
897 }
898
899 /* Try the next token in the word table first. */
900 /* This allows us to match "2nd", for example. */
901 {
902 const char *src = *in;
903 const struct LEXICON *tp;
904 unsigned i = 0;
905
906 /* Force to lowercase and strip '.' characters. */
907 while (*src != '\0'
908 && (isalnum((unsigned char)*src) || *src == '.')
909 && i < sizeof(buff)-1) {
910 if (*src != '.') {
911 if (isupper((unsigned char)*src))
912 buff[i++] = tolower((unsigned char)*src);
913 else
914 buff[i++] = *src;
915 }
916 src++;
917 }
918 buff[i] = '\0';
919
920 /*
921 * Find the first match. If the word can be
922 * abbreviated, make sure we match at least
923 * the minimum abbreviation.
924 */
925 for (tp = TimeWords; tp->name; tp++) {
926 size_t abbrev = tp->abbrev;
927 if (abbrev == 0)
928 abbrev = strlen(tp->name);
929 if (strlen(buff) >= abbrev
930 && strncmp(tp->name, buff, strlen(buff))
931 == 0) {
932 /* Skip over token. */
933 *in = src;
934 /* Return the match. */
935 *value = tp->value;
936 return tp->type;
937 }
938 }
939 }
940
941 /*
942 * Not in the word table, maybe it's a number. Note:
943 * Because '-' and '+' have other special meanings, I
944 * don't deal with signed numbers here.
945 */
946 if (isdigit((unsigned char)(c = **in))) {
947 for (*value = 0; isdigit((unsigned char)(c = *(*in)++)); )
948 *value = 10 * *value + c - '0';
949 (*in)--;
950 return (tUNUMBER);
951 }
952
953 return *(*in)++;
954 }
955 }
956
957 #define TM_YEAR_ORIGIN 1900
958
959 /* Yield A - B, measured in seconds. */
960 static long
difftm(struct tm * a,struct tm * b)961 difftm (struct tm *a, struct tm *b)
962 {
963 int ay = a->tm_year + (TM_YEAR_ORIGIN - 1);
964 int by = b->tm_year + (TM_YEAR_ORIGIN - 1);
965 int days = (
966 /* difference in day of year */
967 a->tm_yday - b->tm_yday
968 /* + intervening leap days */
969 + ((ay >> 2) - (by >> 2))
970 - (ay/100 - by/100)
971 + ((ay/100 >> 2) - (by/100 >> 2))
972 /* + difference in years * 365 */
973 + (long)(ay-by) * 365
974 );
975 return (days * DAY + (a->tm_hour - b->tm_hour) * HOUR
976 + (a->tm_min - b->tm_min) * MINUTE
977 + (a->tm_sec - b->tm_sec));
978 }
979
980 /*
981 *
982 * The public function.
983 *
984 * TODO: tokens[] array should be dynamically sized.
985 */
986 time_t
__archive_get_date(time_t now,const char * p)987 __archive_get_date(time_t now, const char *p)
988 {
989 struct token tokens[256];
990 struct gdstate _gds;
991 struct token *lasttoken;
992 struct gdstate *gds;
993 struct tm local, *tm;
994 struct tm gmt, *gmt_ptr;
995 time_t Start;
996 time_t tod;
997 long tzone;
998 #if defined(HAVE__LOCALTIME64_S) || defined(HAVE__GMTIME64_S)
999 errno_t terr;
1000 __time64_t tmptime;
1001 #endif
1002
1003 /* Clear out the parsed token array. */
1004 memset(tokens, 0, sizeof(tokens));
1005 /* Initialize the parser state. */
1006 memset(&_gds, 0, sizeof(_gds));
1007 gds = &_gds;
1008
1009 /* Look up the current time. */
1010 #if defined(HAVE_LOCALTIME_R)
1011 tm = localtime_r(&now, &local);
1012 #elif defined(HAVE__LOCALTIME64_S)
1013 tmptime = now;
1014 terr = _localtime64_s(&local, &tmptime);
1015 if (terr)
1016 tm = NULL;
1017 else
1018 tm = &local;
1019 #else
1020 memset(&local, 0, sizeof(local));
1021 tm = localtime(&now);
1022 #endif
1023 if (tm == NULL)
1024 return -1;
1025 #if !defined(HAVE_LOCALTIME_R) && !defined(HAVE__LOCALTIME64_S)
1026 local = *tm;
1027 #endif
1028
1029 /* Look up UTC if we can and use that to determine the current
1030 * timezone offset. */
1031 #if defined(HAVE_GMTIME_R)
1032 gmt_ptr = gmtime_r(&now, &gmt);
1033 #elif defined(HAVE__GMTIME64_S)
1034 tmptime = now;
1035 terr = _gmtime64_s(&gmt, &tmptime);
1036 if (terr)
1037 gmt_ptr = NULL;
1038 else
1039 gmt_ptr = &gmt;
1040 #else
1041 memset(&gmt, 0, sizeof(gmt));
1042 gmt_ptr = gmtime(&now);
1043 if (gmt_ptr != NULL) {
1044 /* Copy, in case localtime and gmtime use the same buffer. */
1045 gmt = *gmt_ptr;
1046 }
1047 #endif
1048 if (gmt_ptr != NULL)
1049 tzone = difftm (&gmt, &local);
1050 else
1051 /* This system doesn't understand timezones; fake it. */
1052 tzone = 0;
1053 if(local.tm_isdst)
1054 tzone += HOUR;
1055
1056 /* Tokenize the input string. */
1057 lasttoken = tokens;
1058 while ((lasttoken->token = nexttoken(&p, &lasttoken->value)) != 0) {
1059 ++lasttoken;
1060 if (lasttoken > tokens + 255)
1061 return -1;
1062 }
1063 gds->tokenp = tokens;
1064
1065 /* Match phrases until we run out of input tokens. */
1066 while (gds->tokenp < lasttoken) {
1067 if (!phrase(gds))
1068 return -1;
1069 }
1070
1071 /* Use current local timezone if none was specified. */
1072 if (!gds->HaveZone) {
1073 gds->Timezone = tzone;
1074 gds->DSTmode = DSTmaybe;
1075 }
1076
1077 /* If a timezone was specified, use that for generating the default
1078 * time components instead of the local timezone. */
1079 if (gds->HaveZone && gmt_ptr != NULL) {
1080 now -= gds->Timezone;
1081 #if defined(HAVE_GMTIME_R)
1082 gmt_ptr = gmtime_r(&now, &gmt);
1083 #elif defined(HAVE__GMTIME64_S)
1084 tmptime = now;
1085 terr = _gmtime64_s(&gmt, &tmptime);
1086 if (terr)
1087 gmt_ptr = NULL;
1088 else
1089 gmt_ptr = &gmt;
1090 #else
1091 gmt_ptr = gmtime(&now);
1092 #endif
1093 if (gmt_ptr != NULL)
1094 local = *gmt_ptr;
1095 now += gds->Timezone;
1096 }
1097
1098 if (!gds->HaveYear)
1099 gds->Year = local.tm_year + 1900;
1100 if (!gds->HaveMonth)
1101 gds->Month = local.tm_mon + 1;
1102 if (!gds->HaveDay)
1103 gds->Day = local.tm_mday;
1104 /* Note: No default for hour/min/sec; a specifier that just
1105 * gives date always refers to 00:00 on that date. */
1106
1107 /* If we saw more than one time, timezone, weekday, year, month,
1108 * or day, then give up. */
1109 if (gds->HaveTime > 1 || gds->HaveZone > 1 || gds->HaveWeekDay > 1
1110 || gds->HaveYear > 1 || gds->HaveMonth > 1 || gds->HaveDay > 1)
1111 return -1;
1112
1113 /* Compute an absolute time based on whatever absolute information
1114 * we collected. */
1115 if (gds->HaveYear || gds->HaveMonth || gds->HaveDay
1116 || gds->HaveTime || gds->HaveWeekDay) {
1117 Start = Convert(gds->Month, gds->Day, gds->Year,
1118 gds->Hour, gds->Minutes, gds->Seconds,
1119 gds->Timezone, gds->DSTmode);
1120 if (Start < 0)
1121 return -1;
1122 } else {
1123 Start = now;
1124 if (!gds->HaveRel)
1125 Start -= local.tm_hour * HOUR + local.tm_min * MINUTE
1126 + local.tm_sec;
1127 }
1128
1129 /* Add the relative offset. */
1130 Start += gds->RelSeconds;
1131 Start += RelativeMonth(Start, gds->Timezone, gds->RelMonth);
1132
1133 /* Adjust for day-of-week offsets. */
1134 if (gds->HaveWeekDay
1135 && !(gds->HaveYear || gds->HaveMonth || gds->HaveDay)) {
1136 tod = RelativeDate(Start, gds->Timezone,
1137 gds->DSTmode, gds->DayOrdinal, gds->DayNumber);
1138 Start += tod;
1139 }
1140
1141 /* -1 is an error indicator, so return 0 instead of -1 if
1142 * that's the actual time. */
1143 return Start == -1 ? 0 : Start;
1144 }
1145
1146
1147 #if defined(TEST)
1148
1149 /* ARGSUSED */
1150 int
main(int argc,char ** argv)1151 main(int argc, char **argv)
1152 {
1153 time_t d;
1154 time_t now = time(NULL);
1155
1156 while (*++argv != NULL) {
1157 (void)printf("Input: %s\n", *argv);
1158 d = get_date(now, *argv);
1159 if (d == -1)
1160 (void)printf("Bad format - couldn't convert.\n");
1161 else
1162 (void)printf("Output: %s\n", ctime(&d));
1163 }
1164 exit(0);
1165 /* NOTREACHED */
1166 }
1167 #endif /* defined(TEST) */
1168