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 #include "archive_platform.h"
31
32 #include <ctype.h>
33 #include <stdio.h>
34 #include <stdlib.h>
35 #include <string.h>
36 #include <time.h>
37
38 #include "archive.h"
39 #include "archive_integer.h"
40
41 /* Basic time units. */
42 #define EPOCH 1970
43 #define MINUTE (60L)
44 #define HOUR (60L * MINUTE)
45 #define DAY (24L * HOUR)
46
47 /* Daylight-savings mode: on, off, or not yet known. */
48 enum DSTMODE { DSTon, DSToff, DSTmaybe };
49 /* Meridian: am or pm. */
50 enum { tAM, tPM };
51 /* Token types returned by nexttoken() */
52 enum { tAGO = 260, tDAY, tDAYZONE, tAMPM, tMONTH, tMONTH_UNIT, tSEC_UNIT,
53 tUNUMBER, tZONE, tDST, tERROR };
54 struct token { int token; time_t value; };
55
56 /*
57 * Parser state.
58 */
59 struct gdstate {
60 struct token *tokenp; /* Pointer to next token. */
61 /* HaveXxxx counts how many of this kind of phrase we've seen;
62 * it's a fatal error to have more than one time, zone, day,
63 * or date phrase. */
64 int HaveYear;
65 int HaveMonth;
66 int HaveDay;
67 int HaveWeekDay; /* Day of week */
68 int HaveTime; /* Hour/minute/second */
69 int HaveZone; /* timezone and/or DST info */
70 int HaveRel; /* time offset; we can have more than one */
71 /* Absolute time values. */
72 time_t Timezone; /* Seconds offset from GMT */
73 time_t Day;
74 time_t Hour;
75 time_t Minutes;
76 time_t Month;
77 time_t Seconds;
78 time_t Year;
79 /* DST selection */
80 enum DSTMODE DSTmode;
81 /* Day of week accounting, e.g., "3rd Tuesday" */
82 time_t DayOrdinal; /* "3" in "3rd Tuesday" */
83 time_t DayNumber; /* "Tuesday" in "3rd Tuesday" */
84 /* Relative time values: hour/day/week offsets are measured in
85 * seconds, month/year are counted in months. */
86 time_t RelMonth;
87 time_t RelSeconds;
88 };
89
90 /*
91 * A series of functions that recognize certain common time phrases.
92 * Each function returns 1 if it managed to make sense of some of the
93 * tokens, zero otherwise.
94 */
95
96 /*
97 * hour:minute or hour:minute:second with optional AM, PM, or numeric
98 * timezone offset
99 */
100 static int
timephrase(struct gdstate * gds)101 timephrase(struct gdstate *gds)
102 {
103 if (gds->tokenp[0].token == tUNUMBER
104 && gds->tokenp[1].token == ':'
105 && gds->tokenp[2].token == tUNUMBER
106 && gds->tokenp[3].token == ':'
107 && gds->tokenp[4].token == tUNUMBER) {
108 /* "12:14:18" or "22:08:07" */
109 ++gds->HaveTime;
110 gds->Hour = gds->tokenp[0].value;
111 gds->Minutes = gds->tokenp[2].value;
112 gds->Seconds = gds->tokenp[4].value;
113 gds->tokenp += 5;
114 }
115 else if (gds->tokenp[0].token == tUNUMBER
116 && gds->tokenp[1].token == ':'
117 && gds->tokenp[2].token == tUNUMBER) {
118 /* "12:14" or "22:08" */
119 ++gds->HaveTime;
120 gds->Hour = gds->tokenp[0].value;
121 gds->Minutes = gds->tokenp[2].value;
122 gds->Seconds = 0;
123 gds->tokenp += 3;
124 }
125 else if (gds->tokenp[0].token == tUNUMBER
126 && gds->tokenp[1].token == tAMPM) {
127 /* "7" is a time if it's followed by "am" or "pm" */
128 ++gds->HaveTime;
129 gds->Hour = gds->tokenp[0].value;
130 gds->Minutes = gds->Seconds = 0;
131 /* We'll handle the AM/PM below. */
132 gds->tokenp += 1;
133 } else {
134 /* We can't handle this. */
135 return 0;
136 }
137
138 if (gds->tokenp[0].token == tAMPM) {
139 /* "7:12pm", "12:20:13am" */
140 if (gds->Hour == 12)
141 gds->Hour = 0;
142 if (gds->tokenp[0].value == tPM)
143 gds->Hour += 12;
144 gds->tokenp += 1;
145 }
146 if (gds->tokenp[0].token == '+'
147 && gds->tokenp[1].token == tUNUMBER) {
148 /* "7:14+0700" */
149 gds->HaveZone++;
150 gds->DSTmode = DSToff;
151 gds->Timezone = - ((gds->tokenp[1].value / 100) * HOUR
152 + (gds->tokenp[1].value % 100) * MINUTE);
153 gds->tokenp += 2;
154 }
155 if (gds->tokenp[0].token == '-'
156 && gds->tokenp[1].token == tUNUMBER) {
157 /* "19:14:12-0530" */
158 gds->HaveZone++;
159 gds->DSTmode = DSToff;
160 gds->Timezone = + ((gds->tokenp[1].value / 100) * HOUR
161 + (gds->tokenp[1].value % 100) * MINUTE);
162 gds->tokenp += 2;
163 }
164 return 1;
165 }
166
167 /*
168 * Timezone name, possibly including DST.
169 */
170 static int
zonephrase(struct gdstate * gds)171 zonephrase(struct gdstate *gds)
172 {
173 if (gds->tokenp[0].token == tZONE
174 && gds->tokenp[1].token == tDST) {
175 gds->HaveZone++;
176 gds->Timezone = gds->tokenp[0].value;
177 gds->DSTmode = DSTon;
178 gds->tokenp += 1;
179 return 1;
180 }
181
182 if (gds->tokenp[0].token == tZONE) {
183 gds->HaveZone++;
184 gds->Timezone = gds->tokenp[0].value;
185 gds->DSTmode = DSToff;
186 gds->tokenp += 1;
187 return 1;
188 }
189
190 if (gds->tokenp[0].token == tDAYZONE) {
191 gds->HaveZone++;
192 gds->Timezone = gds->tokenp[0].value;
193 gds->DSTmode = DSTon;
194 gds->tokenp += 1;
195 return 1;
196 }
197 return 0;
198 }
199
200 /*
201 * Year/month/day in various combinations.
202 */
203 static int
datephrase(struct gdstate * gds)204 datephrase(struct gdstate *gds)
205 {
206 if (gds->tokenp[0].token == tUNUMBER
207 && gds->tokenp[1].token == '/'
208 && gds->tokenp[2].token == tUNUMBER
209 && gds->tokenp[3].token == '/'
210 && gds->tokenp[4].token == tUNUMBER) {
211 gds->HaveYear++;
212 gds->HaveMonth++;
213 gds->HaveDay++;
214 if (gds->tokenp[0].value >= 13) {
215 /* First number is big: 2004/01/29, 99/02/17 */
216 gds->Year = gds->tokenp[0].value;
217 gds->Month = gds->tokenp[2].value;
218 gds->Day = gds->tokenp[4].value;
219 } else if ((gds->tokenp[4].value >= 13)
220 || (gds->tokenp[2].value >= 13)) {
221 /* Last number is big: 01/07/98 */
222 /* Middle number is big: 01/29/04 */
223 gds->Month = gds->tokenp[0].value;
224 gds->Day = gds->tokenp[2].value;
225 gds->Year = gds->tokenp[4].value;
226 } else {
227 /* No significant clues: 02/03/04 */
228 gds->Month = gds->tokenp[0].value;
229 gds->Day = gds->tokenp[2].value;
230 gds->Year = gds->tokenp[4].value;
231 }
232 gds->tokenp += 5;
233 return 1;
234 }
235
236 if (gds->tokenp[0].token == tUNUMBER
237 && gds->tokenp[1].token == '/'
238 && gds->tokenp[2].token == tUNUMBER) {
239 /* "1/15" */
240 gds->HaveMonth++;
241 gds->HaveDay++;
242 gds->Month = gds->tokenp[0].value;
243 gds->Day = gds->tokenp[2].value;
244 gds->tokenp += 3;
245 return 1;
246 }
247
248 if (gds->tokenp[0].token == tUNUMBER
249 && gds->tokenp[1].token == '-'
250 && gds->tokenp[2].token == tUNUMBER
251 && gds->tokenp[3].token == '-'
252 && gds->tokenp[4].token == tUNUMBER) {
253 /* ISO 8601 format. yyyy-mm-dd. */
254 gds->HaveYear++;
255 gds->HaveMonth++;
256 gds->HaveDay++;
257 gds->Year = gds->tokenp[0].value;
258 gds->Month = gds->tokenp[2].value;
259 gds->Day = gds->tokenp[4].value;
260 gds->tokenp += 5;
261 return 1;
262 }
263
264 if (gds->tokenp[0].token == tUNUMBER
265 && gds->tokenp[1].token == '-'
266 && gds->tokenp[2].token == tMONTH
267 && gds->tokenp[3].token == '-'
268 && gds->tokenp[4].token == tUNUMBER) {
269 gds->HaveYear++;
270 gds->HaveMonth++;
271 gds->HaveDay++;
272 if (gds->tokenp[0].value > 31) {
273 /* e.g. 1992-Jun-17 */
274 gds->Year = gds->tokenp[0].value;
275 gds->Month = gds->tokenp[2].value;
276 gds->Day = gds->tokenp[4].value;
277 } else {
278 /* e.g. 17-JUN-1992. */
279 gds->Day = gds->tokenp[0].value;
280 gds->Month = gds->tokenp[2].value;
281 gds->Year = gds->tokenp[4].value;
282 }
283 gds->tokenp += 5;
284 return 1;
285 }
286
287 if (gds->tokenp[0].token == tMONTH
288 && gds->tokenp[1].token == tUNUMBER
289 && gds->tokenp[2].token == ','
290 && gds->tokenp[3].token == tUNUMBER) {
291 /* "June 17, 2001" */
292 gds->HaveYear++;
293 gds->HaveMonth++;
294 gds->HaveDay++;
295 gds->Month = gds->tokenp[0].value;
296 gds->Day = gds->tokenp[1].value;
297 gds->Year = gds->tokenp[3].value;
298 gds->tokenp += 4;
299 return 1;
300 }
301
302 if (gds->tokenp[0].token == tMONTH
303 && gds->tokenp[1].token == tUNUMBER) {
304 /* "May 3" */
305 gds->HaveMonth++;
306 gds->HaveDay++;
307 gds->Month = gds->tokenp[0].value;
308 gds->Day = gds->tokenp[1].value;
309 gds->tokenp += 2;
310 return 1;
311 }
312
313 if (gds->tokenp[0].token == tUNUMBER
314 && gds->tokenp[1].token == tMONTH
315 && gds->tokenp[2].token == tUNUMBER) {
316 /* "12 Sept 1997" */
317 gds->HaveYear++;
318 gds->HaveMonth++;
319 gds->HaveDay++;
320 gds->Day = gds->tokenp[0].value;
321 gds->Month = gds->tokenp[1].value;
322 gds->Year = gds->tokenp[2].value;
323 gds->tokenp += 3;
324 return 1;
325 }
326
327 if (gds->tokenp[0].token == tUNUMBER
328 && gds->tokenp[1].token == tMONTH) {
329 /* "12 Sept" */
330 gds->HaveMonth++;
331 gds->HaveDay++;
332 gds->Day = gds->tokenp[0].value;
333 gds->Month = gds->tokenp[1].value;
334 gds->tokenp += 2;
335 return 1;
336 }
337
338 return 0;
339 }
340
341 /*
342 * Relative time phrase: "tomorrow", "yesterday", "+1 hour", etc.
343 */
344 static int
relunitphrase(struct gdstate * gds)345 relunitphrase(struct gdstate *gds)
346 {
347 if (gds->tokenp[0].token == '-'
348 && gds->tokenp[1].token == tUNUMBER
349 && gds->tokenp[2].token == tSEC_UNIT) {
350 /* "-3 hours" */
351 gds->HaveRel++;
352 gds->RelSeconds -= gds->tokenp[1].value * gds->tokenp[2].value;
353 gds->tokenp += 3;
354 return 1;
355 }
356 if (gds->tokenp[0].token == '+'
357 && gds->tokenp[1].token == tUNUMBER
358 && gds->tokenp[2].token == tSEC_UNIT) {
359 /* "+1 minute" */
360 gds->HaveRel++;
361 gds->RelSeconds += gds->tokenp[1].value * gds->tokenp[2].value;
362 gds->tokenp += 3;
363 return 1;
364 }
365 if (gds->tokenp[0].token == tUNUMBER
366 && gds->tokenp[1].token == tSEC_UNIT) {
367 /* "1 day" */
368 gds->HaveRel++;
369 gds->RelSeconds += gds->tokenp[0].value * gds->tokenp[1].value;
370 gds->tokenp += 2;
371 return 1;
372 }
373 if (gds->tokenp[0].token == '-'
374 && gds->tokenp[1].token == tUNUMBER
375 && gds->tokenp[2].token == tMONTH_UNIT) {
376 /* "-3 months" */
377 gds->HaveRel++;
378 gds->RelMonth -= gds->tokenp[1].value * gds->tokenp[2].value;
379 gds->tokenp += 3;
380 return 1;
381 }
382 if (gds->tokenp[0].token == '+'
383 && gds->tokenp[1].token == tUNUMBER
384 && gds->tokenp[2].token == tMONTH_UNIT) {
385 /* "+5 years" */
386 gds->HaveRel++;
387 gds->RelMonth += gds->tokenp[1].value * gds->tokenp[2].value;
388 gds->tokenp += 3;
389 return 1;
390 }
391 if (gds->tokenp[0].token == tUNUMBER
392 && gds->tokenp[1].token == tMONTH_UNIT) {
393 /* "2 years" */
394 gds->HaveRel++;
395 gds->RelMonth += gds->tokenp[0].value * gds->tokenp[1].value;
396 gds->tokenp += 2;
397 return 1;
398 }
399 if (gds->tokenp[0].token == tSEC_UNIT) {
400 /* "now", "tomorrow" */
401 gds->HaveRel++;
402 gds->RelSeconds += gds->tokenp[0].value;
403 gds->tokenp += 1;
404 return 1;
405 }
406 if (gds->tokenp[0].token == tMONTH_UNIT) {
407 /* "month" */
408 gds->HaveRel++;
409 gds->RelMonth += gds->tokenp[0].value;
410 gds->tokenp += 1;
411 return 1;
412 }
413 return 0;
414 }
415
416 /*
417 * Day of the week specification.
418 */
419 static int
dayphrase(struct gdstate * gds)420 dayphrase(struct gdstate *gds)
421 {
422 if (gds->tokenp[0].token == tDAY) {
423 /* "tues", "wednesday," */
424 gds->HaveWeekDay++;
425 gds->DayOrdinal = 1;
426 gds->DayNumber = gds->tokenp[0].value;
427 gds->tokenp += 1;
428 if (gds->tokenp[0].token == ',')
429 gds->tokenp += 1;
430 return 1;
431 }
432 if (gds->tokenp[0].token == tUNUMBER
433 && gds->tokenp[1].token == tDAY) {
434 /* "second tues" "3 wed" */
435 gds->HaveWeekDay++;
436 gds->DayOrdinal = gds->tokenp[0].value;
437 gds->DayNumber = gds->tokenp[1].value;
438 gds->tokenp += 2;
439 return 1;
440 }
441 return 0;
442 }
443
444 /*
445 * Try to match a phrase using one of the above functions.
446 * This layer also deals with a couple of generic issues.
447 */
448 static int
phrase(struct gdstate * gds)449 phrase(struct gdstate *gds)
450 {
451 if (timephrase(gds))
452 return 1;
453 if (zonephrase(gds))
454 return 1;
455 if (datephrase(gds))
456 return 1;
457 if (dayphrase(gds))
458 return 1;
459 if (relunitphrase(gds)) {
460 if (gds->tokenp[0].token == tAGO) {
461 gds->RelSeconds = -gds->RelSeconds;
462 gds->RelMonth = -gds->RelMonth;
463 gds->tokenp += 1;
464 }
465 return 1;
466 }
467
468 /* Bare numbers sometimes have meaning. */
469 if (gds->tokenp[0].token == tUNUMBER) {
470 if (gds->HaveTime && !gds->HaveYear && !gds->HaveRel) {
471 gds->HaveYear++;
472 gds->Year = gds->tokenp[0].value;
473 gds->tokenp += 1;
474 return 1;
475 }
476
477 if(gds->tokenp[0].value > 10000) {
478 /* "20040301" */
479 gds->HaveYear++;
480 gds->HaveMonth++;
481 gds->HaveDay++;
482 gds->Day= (gds->tokenp[0].value)%100;
483 gds->Month= (gds->tokenp[0].value/100)%100;
484 gds->Year = gds->tokenp[0].value/10000;
485 gds->tokenp += 1;
486 return 1;
487 }
488
489 if (gds->tokenp[0].value < 24) {
490 gds->HaveTime++;
491 gds->Hour = gds->tokenp[0].value;
492 gds->Minutes = 0;
493 gds->Seconds = 0;
494 gds->tokenp += 1;
495 return 1;
496 }
497
498 if ((gds->tokenp[0].value / 100 < 24)
499 && (gds->tokenp[0].value % 100 < 60)) {
500 /* "513" is same as "5:13" */
501 gds->Hour = gds->tokenp[0].value / 100;
502 gds->Minutes = gds->tokenp[0].value % 100;
503 gds->Seconds = 0;
504 gds->tokenp += 1;
505 return 1;
506 }
507 }
508
509 return 0;
510 }
511
512 /*
513 * A dictionary of time words.
514 */
515 static struct LEXICON {
516 size_t abbrev;
517 const char *name;
518 int type;
519 time_t value;
520 } const TimeWords[] = {
521 /* am/pm */
522 { 0, "am", tAMPM, tAM },
523 { 0, "pm", tAMPM, tPM },
524
525 /* Month names. */
526 { 3, "january", tMONTH, 1 },
527 { 3, "february", tMONTH, 2 },
528 { 3, "march", tMONTH, 3 },
529 { 3, "april", tMONTH, 4 },
530 { 3, "may", tMONTH, 5 },
531 { 3, "june", tMONTH, 6 },
532 { 3, "july", tMONTH, 7 },
533 { 3, "august", tMONTH, 8 },
534 { 3, "september", tMONTH, 9 },
535 { 3, "october", tMONTH, 10 },
536 { 3, "november", tMONTH, 11 },
537 { 3, "december", tMONTH, 12 },
538
539 /* Days of the week. */
540 { 2, "sunday", tDAY, 0 },
541 { 3, "monday", tDAY, 1 },
542 { 2, "tuesday", tDAY, 2 },
543 { 3, "wednesday", tDAY, 3 },
544 { 2, "thursday", tDAY, 4 },
545 { 2, "friday", tDAY, 5 },
546 { 2, "saturday", tDAY, 6 },
547
548 /* Timezones: Offsets are in seconds. */
549 { 0, "gmt", tZONE, 0*HOUR }, /* Greenwich Mean */
550 { 0, "ut", tZONE, 0*HOUR }, /* Universal (Coordinated) */
551 { 0, "utc", tZONE, 0*HOUR },
552 { 0, "wet", tZONE, 0*HOUR }, /* Western European */
553 { 0, "bst", tDAYZONE, 0*HOUR }, /* British Summer */
554 { 0, "wat", tZONE, 1*HOUR }, /* West Africa */
555 { 0, "at", tZONE, 2*HOUR }, /* Azores */
556 /* { 0, "bst", tZONE, 3*HOUR }, */ /* Brazil Standard: Conflict */
557 /* { 0, "gst", tZONE, 3*HOUR }, */ /* Greenland Standard: Conflict*/
558 { 0, "nft", tZONE, 3*HOUR+30*MINUTE }, /* Newfoundland */
559 { 0, "nst", tZONE, 3*HOUR+30*MINUTE }, /* Newfoundland Standard */
560 { 0, "ndt", tDAYZONE, 3*HOUR+30*MINUTE }, /* Newfoundland Daylight */
561 { 0, "ast", tZONE, 4*HOUR }, /* Atlantic Standard */
562 { 0, "adt", tDAYZONE, 4*HOUR }, /* Atlantic Daylight */
563 { 0, "est", tZONE, 5*HOUR }, /* Eastern Standard */
564 { 0, "edt", tDAYZONE, 5*HOUR }, /* Eastern Daylight */
565 { 0, "cst", tZONE, 6*HOUR }, /* Central Standard */
566 { 0, "cdt", tDAYZONE, 6*HOUR }, /* Central Daylight */
567 { 0, "mst", tZONE, 7*HOUR }, /* Mountain Standard */
568 { 0, "mdt", tDAYZONE, 7*HOUR }, /* Mountain Daylight */
569 { 0, "pst", tZONE, 8*HOUR }, /* Pacific Standard */
570 { 0, "pdt", tDAYZONE, 8*HOUR }, /* Pacific Daylight */
571 { 0, "yst", tZONE, 9*HOUR }, /* Yukon Standard */
572 { 0, "ydt", tDAYZONE, 9*HOUR }, /* Yukon Daylight */
573 { 0, "hst", tZONE, 10*HOUR }, /* Hawaii Standard */
574 { 0, "hdt", tDAYZONE, 10*HOUR }, /* Hawaii Daylight */
575 { 0, "cat", tZONE, 10*HOUR }, /* Central Alaska */
576 { 0, "ahst", tZONE, 10*HOUR }, /* Alaska-Hawaii Standard */
577 { 0, "nt", tZONE, 11*HOUR }, /* Nome */
578 { 0, "idlw", tZONE, 12*HOUR }, /* Intl Date Line West */
579 { 0, "cet", tZONE, -1*HOUR }, /* Central European */
580 { 0, "met", tZONE, -1*HOUR }, /* Middle European */
581 { 0, "mewt", tZONE, -1*HOUR }, /* Middle European Winter */
582 { 0, "mest", tDAYZONE, -1*HOUR }, /* Middle European Summer */
583 { 0, "swt", tZONE, -1*HOUR }, /* Swedish Winter */
584 { 0, "sst", tDAYZONE, -1*HOUR }, /* Swedish Summer */
585 { 0, "fwt", tZONE, -1*HOUR }, /* French Winter */
586 { 0, "fst", tDAYZONE, -1*HOUR }, /* French Summer */
587 { 0, "eet", tZONE, -2*HOUR }, /* Eastern Eur, USSR Zone 1 */
588 { 0, "bt", tZONE, -3*HOUR }, /* Baghdad, USSR Zone 2 */
589 { 0, "it", tZONE, -3*HOUR-30*MINUTE },/* Iran */
590 { 0, "zp4", tZONE, -4*HOUR }, /* USSR Zone 3 */
591 { 0, "zp5", tZONE, -5*HOUR }, /* USSR Zone 4 */
592 { 0, "ist", tZONE, -5*HOUR-30*MINUTE },/* Indian Standard */
593 { 0, "zp6", tZONE, -6*HOUR }, /* USSR Zone 5 */
594 /* { 0, "nst", tZONE, -6.5*HOUR }, */ /* North Sumatra: Conflict */
595 /* { 0, "sst", tZONE, -7*HOUR }, */ /* So Sumatra, USSR 6: Conflict */
596 { 0, "wast", tZONE, -7*HOUR }, /* West Australian Standard */
597 { 0, "wadt", tDAYZONE, -7*HOUR }, /* West Australian Daylight */
598 { 0, "jt", tZONE, -7*HOUR-30*MINUTE },/* Java (3pm in Cronusland!)*/
599 { 0, "cct", tZONE, -8*HOUR }, /* China Coast, USSR Zone 7 */
600 { 0, "jst", tZONE, -9*HOUR }, /* Japan Std, USSR Zone 8 */
601 { 0, "cast", tZONE, -9*HOUR-30*MINUTE },/* Ctrl Australian Std */
602 { 0, "cadt", tDAYZONE, -9*HOUR-30*MINUTE },/* Ctrl Australian Daylt */
603 { 0, "east", tZONE, -10*HOUR }, /* Eastern Australian Std */
604 { 0, "eadt", tDAYZONE, -10*HOUR }, /* Eastern Australian Daylt */
605 { 0, "gst", tZONE, -10*HOUR }, /* Guam Std, USSR Zone 9 */
606 { 0, "nzt", tZONE, -12*HOUR }, /* New Zealand */
607 { 0, "nzst", tZONE, -12*HOUR }, /* New Zealand Standard */
608 { 0, "nzdt", tDAYZONE, -12*HOUR }, /* New Zealand Daylight */
609 { 0, "idle", tZONE, -12*HOUR }, /* Intl Date Line East */
610
611 { 0, "dst", tDST, 0 },
612
613 /* Time units. */
614 { 4, "years", tMONTH_UNIT, 12 },
615 { 5, "months", tMONTH_UNIT, 1 },
616 { 9, "fortnights", tSEC_UNIT, 14 * DAY },
617 { 4, "weeks", tSEC_UNIT, 7 * DAY },
618 { 3, "days", tSEC_UNIT, DAY },
619 { 4, "hours", tSEC_UNIT, HOUR },
620 { 3, "minutes", tSEC_UNIT, MINUTE },
621 { 3, "seconds", tSEC_UNIT, 1 },
622
623 /* Relative-time words. */
624 { 0, "tomorrow", tSEC_UNIT, DAY },
625 { 0, "yesterday", tSEC_UNIT, -DAY },
626 { 0, "today", tSEC_UNIT, 0 },
627 { 0, "now", tSEC_UNIT, 0 },
628 { 0, "last", tUNUMBER, -1 },
629 { 0, "this", tSEC_UNIT, 0 },
630 { 0, "next", tUNUMBER, 2 },
631 { 0, "first", tUNUMBER, 1 },
632 { 0, "1st", tUNUMBER, 1 },
633 /* { 0, "second", tUNUMBER, 2 }, */
634 { 0, "2nd", tUNUMBER, 2 },
635 { 0, "third", tUNUMBER, 3 },
636 { 0, "3rd", tUNUMBER, 3 },
637 { 0, "fourth", tUNUMBER, 4 },
638 { 0, "4th", tUNUMBER, 4 },
639 { 0, "fifth", tUNUMBER, 5 },
640 { 0, "5th", tUNUMBER, 5 },
641 { 0, "sixth", tUNUMBER, 6 },
642 { 0, "seventh", tUNUMBER, 7 },
643 { 0, "eighth", tUNUMBER, 8 },
644 { 0, "ninth", tUNUMBER, 9 },
645 { 0, "tenth", tUNUMBER, 10 },
646 { 0, "eleventh", tUNUMBER, 11 },
647 { 0, "twelfth", tUNUMBER, 12 },
648 { 0, "ago", tAGO, 1 },
649
650 /* Military timezones. */
651 { 0, "a", tZONE, 1*HOUR },
652 { 0, "b", tZONE, 2*HOUR },
653 { 0, "c", tZONE, 3*HOUR },
654 { 0, "d", tZONE, 4*HOUR },
655 { 0, "e", tZONE, 5*HOUR },
656 { 0, "f", tZONE, 6*HOUR },
657 { 0, "g", tZONE, 7*HOUR },
658 { 0, "h", tZONE, 8*HOUR },
659 { 0, "i", tZONE, 9*HOUR },
660 { 0, "k", tZONE, 10*HOUR },
661 { 0, "l", tZONE, 11*HOUR },
662 { 0, "m", tZONE, 12*HOUR },
663 { 0, "n", tZONE, -1*HOUR },
664 { 0, "o", tZONE, -2*HOUR },
665 { 0, "p", tZONE, -3*HOUR },
666 { 0, "q", tZONE, -4*HOUR },
667 { 0, "r", tZONE, -5*HOUR },
668 { 0, "s", tZONE, -6*HOUR },
669 { 0, "t", tZONE, -7*HOUR },
670 { 0, "u", tZONE, -8*HOUR },
671 { 0, "v", tZONE, -9*HOUR },
672 { 0, "w", tZONE, -10*HOUR },
673 { 0, "x", tZONE, -11*HOUR },
674 { 0, "y", tZONE, -12*HOUR },
675 { 0, "z", tZONE, 0*HOUR },
676
677 /* End of table. */
678 { 0, NULL, 0, 0 }
679 };
680
681 /*
682 * Year is either:
683 * = A number from 0 to 99, which means a year from 1970 to 2069, or
684 * = The actual year (>=100).
685 */
686 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)687 Convert(time_t Month, time_t Day, time_t Year,
688 time_t Hours, time_t Minutes, time_t Seconds,
689 time_t Timezone, enum DSTMODE DSTmode)
690 {
691 signed char DaysInMonth[12] = {
692 31, 0, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31
693 };
694 time_t Julian;
695 int i;
696 struct tm *ltime;
697 #if defined(HAVE_LOCALTIME_R) || defined(HAVE_LOCALTIME_S)
698 struct tm tmbuf;
699 #endif
700
701 if (Year < 69)
702 Year += 2000;
703 else if (Year < 100)
704 Year += 1900;
705 DaysInMonth[1] = Year % 4 == 0 && (Year % 100 != 0 || Year % 400 == 0)
706 ? 29 : 28;
707 if (Year < EPOCH || (sizeof(time_t) <= 4 && Year >= 2038)
708 || Month < 1 || Month > 12
709 /* Lint fluff: "conversion from long may lose accuracy" */
710 || Day < 1 || Day > DaysInMonth[(int)--Month]
711 || Hours < 0 || Hours > 23
712 || Minutes < 0 || Minutes > 59
713 || Seconds < 0 || Seconds > 59)
714 return -1;
715
716 Julian = Day - 1;
717 for (i = 0; i < Month; i++)
718 Julian += DaysInMonth[i];
719 for (i = EPOCH; i < Year; i++)
720 Julian += 365 + (i % 4 == 0);
721 Julian *= DAY;
722 Julian += Timezone;
723 Julian += Hours * HOUR + Minutes * MINUTE + Seconds;
724 #if defined(HAVE_LOCALTIME_S)
725 ltime = localtime_s(&tmbuf, &Julian) ? NULL : &tmbuf;
726 #elif defined(HAVE_LOCALTIME_R)
727 ltime = localtime_r(&Julian, &tmbuf);
728 #else
729 ltime = localtime(&Julian);
730 #endif
731 if (DSTmode == DSTon
732 || (DSTmode == DSTmaybe && ltime->tm_isdst))
733 Julian -= HOUR;
734 return Julian;
735 }
736
737 static time_t
DSTcorrect(time_t Start,time_t Future)738 DSTcorrect(time_t Start, time_t Future)
739 {
740 time_t StartDay;
741 time_t FutureDay;
742 struct tm *ltime;
743 #if defined(HAVE_LOCALTIME_R) || defined(HAVE_LOCALTIME_S)
744 struct tm tmbuf;
745 #endif
746 #if defined(HAVE_LOCALTIME_S)
747 ltime = localtime_s(&tmbuf, &Start) ? NULL : &tmbuf;
748 #elif defined(HAVE_LOCALTIME_R)
749 ltime = localtime_r(&Start, &tmbuf);
750 #else
751 ltime = localtime(&Start);
752 #endif
753 StartDay = (ltime->tm_hour + 1) % 24;
754 #if defined(HAVE_LOCALTIME_S)
755 ltime = localtime_s(&tmbuf, &Future) ? NULL : &tmbuf;
756 #elif defined(HAVE_LOCALTIME_R)
757 ltime = localtime_r(&Future, &tmbuf);
758 #else
759 ltime = localtime(&Future);
760 #endif
761 FutureDay = (ltime->tm_hour + 1) % 24;
762 return (Future - Start) + (StartDay - FutureDay) * HOUR;
763 }
764
765
766 static time_t
RelativeDate(time_t Start,time_t zone,int dstmode,time_t DayOrdinal,time_t DayNumber)767 RelativeDate(time_t Start, time_t zone, int dstmode,
768 time_t DayOrdinal, time_t DayNumber)
769 {
770 struct tm *tm;
771 time_t t, now;
772 #if defined(HAVE_GMTIME_R) || defined(HAVE_GMTIME_S)
773 struct tm tmbuf;
774 #endif
775
776 t = Start - zone;
777 #if defined(HAVE_GMTIME_S)
778 tm = gmtime_s(&tmbuf, &t) ? NULL : &tmbuf;
779 #elif defined(HAVE_GMTIME_R)
780 tm = gmtime_r(&t, &tmbuf);
781 #else
782 tm = gmtime(&t);
783 #endif
784 now = Start;
785 now += DAY * ((DayNumber - tm->tm_wday + 7) % 7);
786 now += 7 * DAY * (DayOrdinal <= 0 ? DayOrdinal : DayOrdinal - 1);
787 if (dstmode == DSTmaybe)
788 return DSTcorrect(Start, now);
789 return now - Start;
790 }
791
792
793 static time_t
RelativeMonth(time_t Start,time_t Timezone,time_t RelMonth)794 RelativeMonth(time_t Start, time_t Timezone, time_t RelMonth)
795 {
796 struct tm *tm;
797 time_t Month;
798 time_t Year;
799 #if defined(HAVE_LOCALTIME_R) || defined(HAVE_LOCALTIME_S)
800 struct tm tmbuf;
801 #endif
802
803 if (RelMonth == 0)
804 return 0;
805 #if defined(HAVE_LOCALTIME_S)
806 tm = localtime_s(&tmbuf, &Start) ? NULL : &tmbuf;
807 #elif defined(HAVE_LOCALTIME_R)
808 tm = localtime_r(&Start, &tmbuf);
809 #else
810 tm = localtime(&Start);
811 #endif
812 Month = 12 * (tm->tm_year + 1900) + tm->tm_mon + RelMonth;
813 Year = Month / 12;
814 Month = Month % 12 + 1;
815 return DSTcorrect(Start,
816 Convert(Month, (time_t)tm->tm_mday, Year,
817 (time_t)tm->tm_hour, (time_t)tm->tm_min, (time_t)tm->tm_sec,
818 Timezone, DSTmaybe));
819 }
820
821 /*
822 * Parses and consumes an unsigned 64-bit number.
823 * Returns UINT64_MAX if the number overflows.
824 */
825 static uint64_t
consume_unsigned_number(const char ** in)826 consume_unsigned_number(const char **in) {
827 uint64_t value = 0;
828 unsigned char c;
829
830 /* Get the first character, abort if it's not a digit */
831 c = (unsigned char)(**in);
832 if (c < '0' || c > '9') {
833 return UINT64_MAX;
834 }
835
836 /* Fold digits into the value, abort on overflow */
837 while (c >= '0' && c <= '9') {
838 unsigned char digit = c - '0';
839
840 /* Return error if the result would overflow UINT64_MAX */
841 if (archive_ckd_mul_u64(&value, value, 10) ||
842 archive_ckd_add_u64(&value, value, digit)) {
843 return UINT64_MAX;
844 }
845 (*in)++;
846 c = (unsigned char)(**in);
847 }
848 return value;
849 }
850
851 /*
852 * Tokenizer.
853 */
854 static int
nexttoken(const char ** in,time_t * value)855 nexttoken(const char **in, time_t *value)
856 {
857 char c;
858 char buff[64];
859
860 for ( ; ; ) {
861 while (isspace((unsigned char)**in))
862 ++*in;
863
864 /* Skip parenthesized comments. */
865 if (**in == '(') {
866 int Count = 0;
867 do {
868 c = *(*in)++;
869 if (c == '\0')
870 return c;
871 if (c == '(')
872 Count++;
873 else if (c == ')')
874 Count--;
875 } while (Count > 0);
876 continue;
877 }
878
879 /* Try the next token in the word table first. */
880 /* This allows us to match "2nd", for example. */
881 {
882 const char *src = *in;
883 const struct LEXICON *tp;
884 unsigned i = 0;
885
886 /* Force to lowercase and strip '.' characters. */
887 while (*src != '\0'
888 && (isalnum((unsigned char)*src) || *src == '.')
889 && i < sizeof(buff)-1) {
890 if (*src != '.') {
891 if (isupper((unsigned char)*src))
892 buff[i++] = (char)tolower(
893 (unsigned char)*src);
894 else
895 buff[i++] = *src;
896 }
897 src++;
898 }
899 buff[i] = '\0';
900
901 /*
902 * Find the first match. If the word can be
903 * abbreviated, make sure we match at least
904 * the minimum abbreviation.
905 */
906 for (tp = TimeWords; tp->name; tp++) {
907 size_t abbrev = tp->abbrev;
908 if (abbrev == 0)
909 abbrev = strlen(tp->name);
910 if (strlen(buff) >= abbrev
911 && strncmp(tp->name, buff, strlen(buff))
912 == 0) {
913 /* Skip over token. */
914 *in = src;
915 /* Return the match. */
916 *value = tp->value;
917 return tp->type;
918 }
919 }
920 }
921
922 /*
923 * Not in the word table. If it starts with a digit,
924 * it must be a number. Note: Because '-' and '+' have
925 * other special meanings, I don't deal with signed
926 * numbers here.
927 */
928 if (isdigit((unsigned char)(**in))) {
929 uint64_t val = consume_unsigned_number(in);
930 if (val > 9999) {
931 return (tERROR);
932 } else {
933 *value = val;
934 return (tUNUMBER);
935 }
936 }
937
938 return *(*in)++;
939 }
940 }
941
942 #define TM_YEAR_ORIGIN 1900
943
944 /* Yield A - B, measured in seconds. */
945 static long
difftm(struct tm * a,struct tm * b)946 difftm (struct tm *a, struct tm *b)
947 {
948 int ay = a->tm_year + (TM_YEAR_ORIGIN - 1);
949 int by = b->tm_year + (TM_YEAR_ORIGIN - 1);
950 long days = (
951 /* difference in day of year */
952 a->tm_yday - b->tm_yday
953 /* + intervening leap days */
954 + ((ay >> 2) - (by >> 2))
955 - (ay/100 - by/100)
956 + ((ay/100 >> 2) - (by/100 >> 2))
957 /* + difference in years * 365 */
958 + (long)(ay-by) * 365
959 );
960 return (days * DAY + (a->tm_hour - b->tm_hour) * HOUR
961 + (a->tm_min - b->tm_min) * MINUTE
962 + (a->tm_sec - b->tm_sec));
963 }
964
965 /*
966 * Parses a Unix epoch timestamp (seconds).
967 * This supports a subset of what GNU tar accepts from black box testing,
968 * but covers common use cases.
969 */
970 static time_t
parse_unix_epoch(const char * p)971 parse_unix_epoch(const char *p)
972 {
973 uint64_t val;
974 time_t epoch;
975
976 /* may begin with + */
977 if (*p == '+') {
978 p++;
979 }
980
981 /* followed by some number */
982 val = consume_unsigned_number(&p);
983 /* Truncate to time_t */
984 epoch = (time_t)val;
985 /* If truncated value is different, then
986 * the value is too large for `time_t`. */
987 if (epoch < 0 || (uint64_t)epoch != val) {
988 return (time_t)-1;
989 }
990 /* If there's any more characters, fail. */
991 if (*p != '\0') {
992 return (time_t)-1;
993 }
994
995 return epoch;
996 }
997
998 /*
999 *
1000 * The public function.
1001 *
1002 * TODO: tokens[] array should be dynamically sized.
1003 */
1004 time_t
archive_parse_date(time_t now,const char * p)1005 archive_parse_date(time_t now, const char *p)
1006 {
1007 struct token tokens[256];
1008 struct gdstate _gds;
1009 struct token *lasttoken;
1010 struct gdstate *gds;
1011 struct tm local, *tm;
1012 struct tm gmt, *gmt_ptr;
1013 time_t Start;
1014 time_t tod;
1015 long tzone;
1016
1017 /*
1018 * @-prefixed Unix epoch timestamps (seconds)
1019 * Skip the complex tokenizer - We do not want to accept strings like "@tenth"
1020 */
1021 if (*p == '@')
1022 return parse_unix_epoch(p + 1);
1023
1024 /* Clear out the parsed token array. */
1025 memset(tokens, 0, sizeof(tokens));
1026 /* Initialize the parser state. */
1027 memset(&_gds, 0, sizeof(_gds));
1028 gds = &_gds;
1029
1030 /* Look up the current time. */
1031 #if defined(HAVE_LOCALTIME_S)
1032 tm = localtime_s(&local, &now) ? NULL : &local;
1033 #elif defined(HAVE_LOCALTIME_R)
1034 tm = localtime_r(&now, &local);
1035 #else
1036 memset(&local, 0, sizeof(local));
1037 tm = localtime(&now);
1038 #endif
1039 if (tm == NULL)
1040 return -1;
1041 #if !defined(HAVE_LOCALTIME_R) && !defined(HAVE_LOCALTIME_S)
1042 local = *tm;
1043 #endif
1044
1045 /* Look up UTC if we can and use that to determine the current
1046 * timezone offset. */
1047 #if defined(HAVE_GMTIME_S)
1048 gmt_ptr = gmtime_s(&gmt, &now) ? NULL : &gmt;
1049 #elif defined(HAVE_GMTIME_R)
1050 gmt_ptr = gmtime_r(&now, &gmt);
1051 #else
1052 memset(&gmt, 0, sizeof(gmt));
1053 gmt_ptr = gmtime(&now);
1054 if (gmt_ptr != NULL) {
1055 /* Copy, in case localtime and gmtime use the same buffer. */
1056 gmt = *gmt_ptr;
1057 }
1058 #endif
1059 if (gmt_ptr != NULL)
1060 tzone = difftm (&gmt, &local);
1061 else
1062 /* This system doesn't understand timezones; fake it. */
1063 tzone = 0;
1064 if(local.tm_isdst)
1065 tzone += HOUR;
1066
1067 /* Tokenize the input string. */
1068 lasttoken = tokens;
1069 while ((lasttoken->token = nexttoken(&p, &lasttoken->value)) != 0) {
1070 ++lasttoken;
1071 if (lasttoken > tokens + 255)
1072 return -1;
1073 }
1074 gds->tokenp = tokens;
1075
1076 /* Match phrases until we run out of input tokens. */
1077 while (gds->tokenp < lasttoken) {
1078 if (!phrase(gds))
1079 return -1;
1080 }
1081
1082 /* Use current local timezone if none was specified. */
1083 if (!gds->HaveZone) {
1084 gds->Timezone = tzone;
1085 gds->DSTmode = DSTmaybe;
1086 }
1087
1088 /* If a timezone was specified, use that for generating the default
1089 * time components instead of the local timezone. */
1090 if (gds->HaveZone && gmt_ptr != NULL) {
1091 now -= gds->Timezone;
1092 #if defined(HAVE_GMTIME_S)
1093 gmt_ptr = gmtime_s(&gmt, &now) ? NULL : &gmt;
1094 #elif defined(HAVE_GMTIME_R)
1095 gmt_ptr = gmtime_r(&now, &gmt);
1096 #else
1097 gmt_ptr = gmtime(&now);
1098 #endif
1099 if (gmt_ptr != NULL)
1100 local = *gmt_ptr;
1101 now += gds->Timezone;
1102 }
1103
1104 if (!gds->HaveYear)
1105 gds->Year = local.tm_year + 1900;
1106 if (!gds->HaveMonth)
1107 gds->Month = local.tm_mon + 1;
1108 if (!gds->HaveDay)
1109 gds->Day = local.tm_mday;
1110 /* Note: No default for hour/min/sec; a specifier that just
1111 * gives date always refers to 00:00 on that date. */
1112
1113 /* If we saw more than one time, timezone, weekday, year, month,
1114 * or day, then give up. */
1115 if (gds->HaveTime > 1 || gds->HaveZone > 1 || gds->HaveWeekDay > 1
1116 || gds->HaveYear > 1 || gds->HaveMonth > 1 || gds->HaveDay > 1)
1117 return -1;
1118
1119 /* Compute an absolute time based on whatever absolute information
1120 * we collected. */
1121 if (gds->HaveYear || gds->HaveMonth || gds->HaveDay
1122 || gds->HaveTime || gds->HaveWeekDay) {
1123 Start = Convert(gds->Month, gds->Day, gds->Year,
1124 gds->Hour, gds->Minutes, gds->Seconds,
1125 gds->Timezone, gds->DSTmode);
1126 if (Start < 0)
1127 return -1;
1128 } else {
1129 Start = now;
1130 if (!gds->HaveRel)
1131 Start -= local.tm_hour * HOUR + local.tm_min * MINUTE
1132 + local.tm_sec;
1133 }
1134
1135 /* Add the relative offset. */
1136 Start += gds->RelSeconds;
1137 Start += RelativeMonth(Start, gds->Timezone, gds->RelMonth);
1138
1139 /* Adjust for day-of-week offsets. */
1140 if (gds->HaveWeekDay
1141 && !(gds->HaveYear || gds->HaveMonth || gds->HaveDay)) {
1142 tod = RelativeDate(Start, gds->Timezone,
1143 gds->DSTmode, gds->DayOrdinal, gds->DayNumber);
1144 Start += tod;
1145 }
1146
1147 /* -1 is an error indicator, so return 0 instead of -1 if
1148 * that's the actual time. */
1149 return Start == -1 ? 0 : Start;
1150 }
1151
1152
1153 #if defined(TEST)
1154
1155 /* ARGSUSED */
1156 int
main(int argc,char ** argv)1157 main(int argc, char **argv)
1158 {
1159 time_t d;
1160 time_t now = time(NULL);
1161
1162 while (*++argv != NULL) {
1163 (void)printf("Input: %s\n", *argv);
1164 d = get_date(now, *argv);
1165 if (d == -1)
1166 (void)printf("Bad format - couldn't convert.\n");
1167 else
1168 (void)printf("Output: %s\n", ctime(&d));
1169 }
1170 exit(0);
1171 /* NOTREACHED */
1172 }
1173 #endif /* defined(TEST) */
1174