Commit | Line | Data |
---|---|---|
16433e2b | 1 | #ifdef __cplusplus |
c944940b | 2 | extern "C" { |
16433e2b | 3 | #endif |
d93e3b8c | 4 | #define PERL_NO_GET_CONTEXT |
16433e2b SP |
5 | #include "EXTERN.h" |
6 | #include "perl.h" | |
7 | #include "XSUB.h" | |
8 | #include <time.h> | |
9 | #ifdef __cplusplus | |
10 | } | |
11 | #endif | |
12 | ||
13 | /* XXX struct tm on some systems (SunOS4/BSD) contains extra (non POSIX) | |
d4d72fe6 | 14 | * fields for which we don't have Configure support prior to Perl 5.8.0: |
16433e2b SP |
15 | * char *tm_zone; -- abbreviation of timezone name |
16 | * long tm_gmtoff; -- offset from GMT in seconds | |
17 | * To workaround core dumps from the uninitialised tm_zone we get the | |
18 | * system to give us a reasonable struct to copy. This fix means that | |
19 | * strftime uses the tm_zone and tm_gmtoff values returned by | |
20 | * localtime(time()). That should give the desired result most of the | |
21 | * time. But probably not always! | |
22 | * | |
d4d72fe6 CB |
23 | * This is a vestigial workaround for Perls prior to 5.8.0. We now |
24 | * rely on the initialization (still likely a workaround) in util.c. | |
16433e2b | 25 | */ |
d4d72fe6 CB |
26 | #if !defined(PERL_VERSION) || PERL_VERSION < 8 |
27 | ||
036055ae | 28 | #if defined(HAS_GNULIBC) |
16433e2b SP |
29 | # ifndef STRUCT_TM_HASZONE |
30 | # define STRUCT_TM_HASZONE | |
31 | # else | |
32 | # define USE_TM_GMTOFF | |
33 | # endif | |
34 | #endif | |
35 | ||
d4d72fe6 CB |
36 | #endif /* end of pre-5.8 */ |
37 | ||
16433e2b SP |
38 | #define DAYS_PER_YEAR 365 |
39 | #define DAYS_PER_QYEAR (4*DAYS_PER_YEAR+1) | |
40 | #define DAYS_PER_CENT (25*DAYS_PER_QYEAR-1) | |
41 | #define DAYS_PER_QCENT (4*DAYS_PER_CENT+1) | |
42 | #define SECS_PER_HOUR (60*60) | |
43 | #define SECS_PER_DAY (24*SECS_PER_HOUR) | |
44 | /* parentheses deliberately absent on these two, otherwise they don't work */ | |
45 | #define MONTH_TO_DAYS 153/5 | |
46 | #define DAYS_TO_MONTH 5/153 | |
47 | /* offset to bias by March (month 4) 1st between month/mday & year finding */ | |
48 | #define YEAR_ADJUST (4*MONTH_TO_DAYS+1) | |
49 | /* as used here, the algorithm leaves Sunday as day 1 unless we adjust it */ | |
50 | #define WEEKDAY_BIAS 6 /* (1+6)%7 makes Sunday 0 again */ | |
51 | ||
d4d72fe6 CB |
52 | #if !defined(PERL_VERSION) || PERL_VERSION < 8 |
53 | ||
16433e2b SP |
54 | #ifdef STRUCT_TM_HASZONE |
55 | static void | |
56 | my_init_tm(struct tm *ptm) /* see mktime, strftime and asctime */ | |
57 | { | |
58 | Time_t now; | |
59 | (void)time(&now); | |
60 | Copy(localtime(&now), ptm, 1, struct tm); | |
61 | } | |
62 | ||
63 | #else | |
64 | # define my_init_tm(ptm) | |
65 | #endif | |
66 | ||
d4d72fe6 CB |
67 | #else |
68 | /* use core version from util.c in 5.8.0 and later */ | |
69 | # define my_init_tm init_tm | |
12bbe8b8 | 70 | #endif |
d4d72fe6 | 71 | |
6e073399 SH |
72 | #ifdef WIN32 |
73 | ||
74 | /* | |
75 | * (1) The CRT maintains its own copy of the environment, separate from | |
76 | * the Win32API copy. | |
77 | * | |
78 | * (2) CRT getenv() retrieves from this copy. CRT putenv() updates this | |
79 | * copy, and then calls SetEnvironmentVariableA() to update the Win32API | |
80 | * copy. | |
81 | * | |
82 | * (3) win32_getenv() and win32_putenv() call GetEnvironmentVariableA() and | |
83 | * SetEnvironmentVariableA() directly, bypassing the CRT copy of the | |
84 | * environment. | |
85 | * | |
86 | * (4) The CRT strftime() "%Z" implementation calls __tzset(). That | |
87 | * calls CRT tzset(), but only the first time it is called, and in turn | |
88 | * that uses CRT getenv("TZ") to retrieve the timezone info from the CRT | |
89 | * local copy of the environment and hence gets the original setting as | |
90 | * perl never updates the CRT copy when assigning to $ENV{TZ}. | |
91 | * | |
92 | * Therefore, we need to retrieve the value of $ENV{TZ} and call CRT | |
12016aad SH |
93 | * putenv() to update the CRT copy of the environment (if it is different) |
94 | * whenever we're about to call tzset(). | |
95 | * | |
96 | * In addition to all that, when perl is built with PERL_IMPLICIT_SYS | |
97 | * defined: | |
98 | * | |
99 | * (a) Each interpreter has its own copy of the environment inside the | |
100 | * perlhost structure. That allows applications that host multiple | |
101 | * independent Perl interpreters to isolate environment changes from | |
102 | * each other. (This is similar to how the perlhost mechanism keeps a | |
103 | * separate working directory for each Perl interpreter, so that calling | |
104 | * chdir() will not affect other interpreters.) | |
105 | * | |
106 | * (b) Only the first Perl interpreter instantiated within a process will | |
107 | * "write through" environment changes to the process environment. | |
108 | * | |
109 | * (c) Even the primary Perl interpreter won't update the CRT copy of the | |
110 | * the environment, only the Win32API copy (it calls win32_putenv()). | |
111 | * | |
112 | * As with CPerlHost::Getenv() and CPerlHost::Putenv() themselves, it makes | |
113 | * sense to only update the process environment when inside the main | |
114 | * interpreter, but we don't have access to CPerlHost's m_bTopLevel member | |
115 | * from here so we'll just have to check PL_curinterp instead. | |
116 | * | |
117 | * Therefore, we can simply #undef getenv() and putenv() so that those names | |
118 | * always refer to the CRT functions, and explicitly call win32_getenv() to | |
119 | * access perl's %ENV. | |
120 | * | |
121 | * We also #undef malloc() and free() to be sure we are using the CRT | |
122 | * functions otherwise under PERL_IMPLICIT_SYS they are redefined to calls | |
123 | * into VMem::Malloc() and VMem::Free() and all allocations will be freed | |
124 | * when the Perl interpreter is being destroyed so we'd end up with a pointer | |
125 | * into deallocated memory in environ[] if a program embedding a Perl | |
126 | * interpreter continues to operate even after the main Perl interpreter has | |
127 | * been destroyed. | |
128 | * | |
129 | * Note that we don't free() the malloc()ed memory unless and until we call | |
130 | * malloc() again ourselves because the CRT putenv() function simply puts its | |
131 | * pointer argument into the environ[] arrary (it doesn't make a copy of it) | |
132 | * so this memory must otherwise be leaked. | |
6e073399 SH |
133 | */ |
134 | ||
6e073399 | 135 | #undef getenv |
6e073399 | 136 | #undef putenv |
d93e3b8c CBW |
137 | # ifdef UNDER_CE |
138 | # define getenv xcegetenv | |
139 | # define putenv xceputenv | |
140 | # endif | |
12016aad SH |
141 | #undef malloc |
142 | #undef free | |
6e073399 SH |
143 | |
144 | static void | |
145 | fix_win32_tzenv(void) | |
146 | { | |
12016aad SH |
147 | static char* oldenv = NULL; |
148 | char* newenv; | |
149 | const char* perl_tz_env = win32_getenv("TZ"); | |
150 | const char* crt_tz_env = getenv("TZ"); | |
151 | if (perl_tz_env == NULL) | |
152 | perl_tz_env = ""; | |
153 | if (crt_tz_env == NULL) | |
154 | crt_tz_env = ""; | |
155 | if (strcmp(perl_tz_env, crt_tz_env) != 0) { | |
156 | newenv = (char*)malloc((strlen(perl_tz_env) + 4) * sizeof(char)); | |
157 | if (newenv != NULL) { | |
158 | sprintf(newenv, "TZ=%s", perl_tz_env); | |
159 | putenv(newenv); | |
160 | if (oldenv != NULL) | |
161 | free(oldenv); | |
162 | oldenv = newenv; | |
163 | } | |
6e073399 | 164 | } |
6e073399 SH |
165 | } |
166 | ||
167 | #endif | |
168 | ||
81ab4c44 SH |
169 | /* |
170 | * my_tzset - wrapper to tzset() with a fix to make it work (better) on Win32. | |
171 | * This code is duplicated in the POSIX module, so any changes made here | |
172 | * should be made there too. | |
173 | */ | |
6e073399 | 174 | static void |
12016aad | 175 | my_tzset(pTHX) |
6e073399 SH |
176 | { |
177 | #ifdef WIN32 | |
12016aad SH |
178 | #if defined(USE_ITHREADS) && defined(PERL_IMPLICIT_SYS) |
179 | if (PL_curinterp == aTHX) | |
180 | #endif | |
181 | fix_win32_tzenv(); | |
6e073399 SH |
182 | #endif |
183 | tzset(); | |
184 | } | |
185 | ||
16433e2b SP |
186 | /* |
187 | * my_mini_mktime - normalise struct tm values without the localtime() | |
124e6c84 RGS |
188 | * semantics (and overhead) of mktime(). Stolen shamelessly from Perl's |
189 | * Perl_mini_mktime() in util.c - for details on the algorithm, see that | |
190 | * file. | |
16433e2b SP |
191 | */ |
192 | static void | |
193 | my_mini_mktime(struct tm *ptm) | |
194 | { | |
195 | int yearday; | |
196 | int secs; | |
197 | int month, mday, year, jday; | |
198 | int odd_cent, odd_year; | |
199 | ||
16433e2b SP |
200 | year = 1900 + ptm->tm_year; |
201 | month = ptm->tm_mon; | |
202 | mday = ptm->tm_mday; | |
203 | /* allow given yday with no month & mday to dominate the result */ | |
204 | if (ptm->tm_yday >= 0 && mday <= 0 && month <= 0) { | |
205 | month = 0; | |
206 | mday = 0; | |
207 | jday = 1 + ptm->tm_yday; | |
208 | } | |
209 | else { | |
210 | jday = 0; | |
211 | } | |
212 | if (month >= 2) | |
213 | month+=2; | |
214 | else | |
215 | month+=14, year--; | |
216 | ||
217 | yearday = DAYS_PER_YEAR * year + year/4 - year/100 + year/400; | |
218 | yearday += month*MONTH_TO_DAYS + mday + jday; | |
219 | /* | |
220 | * Note that we don't know when leap-seconds were or will be, | |
221 | * so we have to trust the user if we get something which looks | |
222 | * like a sensible leap-second. Wild values for seconds will | |
223 | * be rationalised, however. | |
224 | */ | |
225 | if ((unsigned) ptm->tm_sec <= 60) { | |
226 | secs = 0; | |
227 | } | |
228 | else { | |
229 | secs = ptm->tm_sec; | |
230 | ptm->tm_sec = 0; | |
231 | } | |
232 | secs += 60 * ptm->tm_min; | |
233 | secs += SECS_PER_HOUR * ptm->tm_hour; | |
234 | if (secs < 0) { | |
235 | if (secs-(secs/SECS_PER_DAY*SECS_PER_DAY) < 0) { | |
236 | /* got negative remainder, but need positive time */ | |
237 | /* back off an extra day to compensate */ | |
238 | yearday += (secs/SECS_PER_DAY)-1; | |
239 | secs -= SECS_PER_DAY * (secs/SECS_PER_DAY - 1); | |
240 | } | |
241 | else { | |
242 | yearday += (secs/SECS_PER_DAY); | |
243 | secs -= SECS_PER_DAY * (secs/SECS_PER_DAY); | |
244 | } | |
245 | } | |
246 | else if (secs >= SECS_PER_DAY) { | |
247 | yearday += (secs/SECS_PER_DAY); | |
248 | secs %= SECS_PER_DAY; | |
249 | } | |
250 | ptm->tm_hour = secs/SECS_PER_HOUR; | |
251 | secs %= SECS_PER_HOUR; | |
252 | ptm->tm_min = secs/60; | |
253 | secs %= 60; | |
254 | ptm->tm_sec += secs; | |
255 | /* done with time of day effects */ | |
256 | /* | |
257 | * The algorithm for yearday has (so far) left it high by 428. | |
258 | * To avoid mistaking a legitimate Feb 29 as Mar 1, we need to | |
259 | * bias it by 123 while trying to figure out what year it | |
260 | * really represents. Even with this tweak, the reverse | |
261 | * translation fails for years before A.D. 0001. | |
262 | * It would still fail for Feb 29, but we catch that one below. | |
263 | */ | |
264 | jday = yearday; /* save for later fixup vis-a-vis Jan 1 */ | |
265 | yearday -= YEAR_ADJUST; | |
266 | year = (yearday / DAYS_PER_QCENT) * 400; | |
267 | yearday %= DAYS_PER_QCENT; | |
268 | odd_cent = yearday / DAYS_PER_CENT; | |
269 | year += odd_cent * 100; | |
270 | yearday %= DAYS_PER_CENT; | |
271 | year += (yearday / DAYS_PER_QYEAR) * 4; | |
272 | yearday %= DAYS_PER_QYEAR; | |
273 | odd_year = yearday / DAYS_PER_YEAR; | |
274 | year += odd_year; | |
275 | yearday %= DAYS_PER_YEAR; | |
276 | if (!yearday && (odd_cent==4 || odd_year==4)) { /* catch Feb 29 */ | |
277 | month = 1; | |
278 | yearday = 29; | |
279 | } | |
280 | else { | |
281 | yearday += YEAR_ADJUST; /* recover March 1st crock */ | |
282 | month = yearday*DAYS_TO_MONTH; | |
283 | yearday -= month*MONTH_TO_DAYS; | |
284 | /* recover other leap-year adjustment */ | |
285 | if (month > 13) { | |
286 | month-=14; | |
287 | year++; | |
288 | } | |
289 | else { | |
290 | month-=2; | |
291 | } | |
292 | } | |
293 | ptm->tm_year = year - 1900; | |
294 | if (yearday) { | |
295 | ptm->tm_mday = yearday; | |
296 | ptm->tm_mon = month; | |
297 | } | |
298 | else { | |
299 | ptm->tm_mday = 31; | |
300 | ptm->tm_mon = month - 1; | |
301 | } | |
302 | /* re-build yearday based on Jan 1 to get tm_yday */ | |
303 | year--; | |
304 | yearday = year*DAYS_PER_YEAR + year/4 - year/100 + year/400; | |
305 | yearday += 14*MONTH_TO_DAYS + 1; | |
306 | ptm->tm_yday = jday - yearday; | |
307 | /* fix tm_wday if not overridden by caller */ | |
308 | ptm->tm_wday = (jday + WEEKDAY_BIAS) % 7; | |
309 | } | |
310 | ||
12016aad | 311 | # if defined(WIN32) || (defined(__QNX__) && defined(__WATCOMC__)) |
90d55c29 | 312 | # define strncasecmp(x,y,n) strnicmp(x,y,n) |
12016aad | 313 | # endif |
be8a15fc | 314 | |
12bbe8b8 | 315 | /* strptime.c 0.1 (Powerdog) 94/03/27 */ |
124e6c84 RGS |
316 | /* strptime copied from freebsd with the following copyright: */ |
317 | /* | |
318 | * Copyright (c) 1994 Powerdog Industries. All rights reserved. | |
319 | * | |
320 | * Redistribution and use in source and binary forms, with or without | |
321 | * modification, are permitted provided that the following conditions | |
322 | * are met: | |
12bbe8b8 | 323 | * |
124e6c84 RGS |
324 | * 1. Redistributions of source code must retain the above copyright |
325 | * notice, this list of conditions and the following disclaimer. | |
12bbe8b8 | 326 | * |
124e6c84 RGS |
327 | * 2. Redistributions in binary form must reproduce the above copyright |
328 | * notice, this list of conditions and the following disclaimer | |
329 | * in the documentation and/or other materials provided with the | |
330 | * distribution. | |
124e6c84 RGS |
331 | * |
332 | * THIS SOFTWARE IS PROVIDED BY POWERDOG INDUSTRIES ``AS IS'' AND ANY | |
333 | * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE | |
334 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR | |
335 | * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE POWERDOG INDUSTRIES BE | |
336 | * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR | |
337 | * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF | |
338 | * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR | |
339 | * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, | |
340 | * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE | |
341 | * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, | |
342 | * EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. | |
12bbe8b8 CBW |
343 | * |
344 | * The views and conclusions contained in the software and documentation | |
345 | * are those of the authors and should not be interpreted as representing | |
346 | * official policies, either expressed or implied, of Powerdog Industries. | |
124e6c84 | 347 | */ |
be8a15fc | 348 | |
16433e2b SP |
349 | #include <time.h> |
350 | #include <ctype.h> | |
351 | #include <string.h> | |
e9f284c9 MB |
352 | static char * _strptime(pTHX_ const char *, const char *, struct tm *, |
353 | int *got_GMT); | |
16433e2b SP |
354 | |
355 | #define asizeof(a) (sizeof (a) / sizeof ((a)[0])) | |
356 | ||
357 | struct lc_time_T { | |
358 | const char * mon[12]; | |
359 | const char * month[12]; | |
360 | const char * wday[7]; | |
361 | const char * weekday[7]; | |
12bbe8b8 | 362 | const char * X_fmt; |
16433e2b SP |
363 | const char * x_fmt; |
364 | const char * c_fmt; | |
365 | const char * am; | |
366 | const char * pm; | |
367 | const char * date_fmt; | |
368 | const char * alt_month[12]; | |
369 | const char * Ef_fmt; | |
370 | const char * EF_fmt; | |
371 | }; | |
372 | ||
373 | struct lc_time_T _time_localebuf; | |
374 | int _time_using_locale; | |
375 | ||
376 | const struct lc_time_T _C_time_locale = { | |
377 | { | |
378 | "Jan", "Feb", "Mar", "Apr", "May", "Jun", | |
379 | "Jul", "Aug", "Sep", "Oct", "Nov", "Dec" | |
380 | }, { | |
381 | "January", "February", "March", "April", "May", "June", | |
382 | "July", "August", "September", "October", "November", "December" | |
383 | }, { | |
384 | "Sun", "Mon", "Tue", "Wed", | |
385 | "Thu", "Fri", "Sat" | |
386 | }, { | |
387 | "Sunday", "Monday", "Tuesday", "Wednesday", | |
388 | "Thursday", "Friday", "Saturday" | |
389 | }, | |
390 | ||
391 | /* X_fmt */ | |
392 | "%H:%M:%S", | |
393 | ||
394 | /* | |
395 | ** x_fmt | |
396 | ** Since the C language standard calls for | |
397 | ** "date, using locale's date format," anything goes. | |
398 | ** Using just numbers (as here) makes Quakers happier; | |
399 | ** it's also compatible with SVR4. | |
400 | */ | |
401 | "%m/%d/%y", | |
402 | ||
403 | /* | |
404 | ** c_fmt (ctime-compatible) | |
405 | ** Not used, just compatibility placeholder. | |
406 | */ | |
407 | NULL, | |
408 | ||
409 | /* am */ | |
410 | "AM", | |
411 | ||
412 | /* pm */ | |
413 | "PM", | |
414 | ||
415 | /* date_fmt */ | |
416 | "%a %Ef %X %Z %Y", | |
417 | ||
418 | { | |
419 | "January", "February", "March", "April", "May", "June", | |
420 | "July", "August", "September", "October", "November", "December" | |
421 | }, | |
422 | ||
423 | /* Ef_fmt | |
424 | ** To determine short months / day order | |
425 | */ | |
426 | "%b %e", | |
427 | ||
428 | /* EF_fmt | |
429 | ** To determine long months / day order | |
430 | */ | |
431 | "%B %e" | |
432 | }; | |
433 | ||
434 | #define Locale (&_C_time_locale) | |
435 | ||
436 | static char * | |
e9f284c9 | 437 | _strptime(pTHX_ const char *buf, const char *fmt, struct tm *tm, int *got_GMT) |
16433e2b SP |
438 | { |
439 | char c; | |
440 | const char *ptr; | |
441 | int i, | |
442 | len; | |
443 | int Ealternative, Oalternative; | |
444 | ||
90d55c29 CBW |
445 | /* There seems to be a slightly improved version at |
446 | * http://www.opensource.apple.com/source/Libc/Libc-583/stdtime/strptime-fbsd.c | |
447 | * which we may end up borrowing more from | |
448 | */ | |
16433e2b SP |
449 | ptr = fmt; |
450 | while (*ptr != 0) { | |
451 | if (*buf == 0) | |
452 | break; | |
453 | ||
454 | c = *ptr++; | |
90d55c29 | 455 | |
16433e2b SP |
456 | if (c != '%') { |
457 | if (isspace((unsigned char)c)) | |
458 | while (*buf != 0 && isspace((unsigned char)*buf)) | |
459 | buf++; | |
460 | else if (c != *buf++) | |
461 | return 0; | |
462 | continue; | |
463 | } | |
464 | ||
465 | Ealternative = 0; | |
466 | Oalternative = 0; | |
467 | label: | |
468 | c = *ptr++; | |
469 | switch (c) { | |
470 | case 0: | |
471 | case '%': | |
472 | if (*buf++ != '%') | |
473 | return 0; | |
474 | break; | |
475 | ||
476 | case '+': | |
e9f284c9 | 477 | buf = _strptime(aTHX_ buf, Locale->date_fmt, tm, got_GMT); |
16433e2b SP |
478 | if (buf == 0) |
479 | return 0; | |
480 | break; | |
481 | ||
482 | case 'C': | |
483 | if (!isdigit((unsigned char)*buf)) | |
484 | return 0; | |
485 | ||
486 | /* XXX This will break for 3-digit centuries. */ | |
487 | len = 2; | |
488 | for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) { | |
489 | i *= 10; | |
490 | i += *buf - '0'; | |
491 | len--; | |
492 | } | |
493 | if (i < 19) | |
494 | return 0; | |
495 | ||
496 | tm->tm_year = i * 100 - 1900; | |
497 | break; | |
498 | ||
499 | case 'c': | |
500 | /* NOTE: c_fmt is intentionally ignored */ | |
e9f284c9 | 501 | buf = _strptime(aTHX_ buf, "%a %Ef %T %Y", tm, got_GMT); |
16433e2b SP |
502 | if (buf == 0) |
503 | return 0; | |
504 | break; | |
505 | ||
506 | case 'D': | |
e9f284c9 | 507 | buf = _strptime(aTHX_ buf, "%m/%d/%y", tm, got_GMT); |
16433e2b SP |
508 | if (buf == 0) |
509 | return 0; | |
510 | break; | |
511 | ||
512 | case 'E': | |
513 | if (Ealternative || Oalternative) | |
514 | break; | |
515 | Ealternative++; | |
516 | goto label; | |
517 | ||
518 | case 'O': | |
519 | if (Ealternative || Oalternative) | |
520 | break; | |
521 | Oalternative++; | |
522 | goto label; | |
523 | ||
524 | case 'F': | |
525 | case 'f': | |
526 | if (!Ealternative) | |
527 | break; | |
e9f284c9 | 528 | buf = _strptime(aTHX_ buf, (c == 'f') ? Locale->Ef_fmt : Locale->EF_fmt, tm, got_GMT); |
16433e2b SP |
529 | if (buf == 0) |
530 | return 0; | |
531 | break; | |
532 | ||
533 | case 'R': | |
e9f284c9 | 534 | buf = _strptime(aTHX_ buf, "%H:%M", tm, got_GMT); |
16433e2b SP |
535 | if (buf == 0) |
536 | return 0; | |
537 | break; | |
538 | ||
539 | case 'r': | |
e9f284c9 | 540 | buf = _strptime(aTHX_ buf, "%I:%M:%S %p", tm, got_GMT); |
16433e2b SP |
541 | if (buf == 0) |
542 | return 0; | |
543 | break; | |
544 | ||
90d55c29 CBW |
545 | case 'n': /* whitespace */ |
546 | case 't': | |
547 | if (!isspace((unsigned char)*buf)) | |
548 | return 0; | |
549 | while (isspace((unsigned char)*buf)) | |
550 | buf++; | |
551 | break; | |
552 | ||
16433e2b | 553 | case 'T': |
e9f284c9 | 554 | buf = _strptime(aTHX_ buf, "%H:%M:%S", tm, got_GMT); |
16433e2b SP |
555 | if (buf == 0) |
556 | return 0; | |
557 | break; | |
558 | ||
559 | case 'X': | |
e9f284c9 | 560 | buf = _strptime(aTHX_ buf, Locale->X_fmt, tm, got_GMT); |
16433e2b SP |
561 | if (buf == 0) |
562 | return 0; | |
563 | break; | |
564 | ||
565 | case 'x': | |
e9f284c9 | 566 | buf = _strptime(aTHX_ buf, Locale->x_fmt, tm, got_GMT); |
16433e2b SP |
567 | if (buf == 0) |
568 | return 0; | |
569 | break; | |
570 | ||
571 | case 'j': | |
572 | if (!isdigit((unsigned char)*buf)) | |
573 | return 0; | |
574 | ||
575 | len = 3; | |
576 | for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) { | |
577 | i *= 10; | |
578 | i += *buf - '0'; | |
579 | len--; | |
580 | } | |
581 | if (i < 1 || i > 366) | |
582 | return 0; | |
583 | ||
584 | tm->tm_yday = i - 1; | |
9bc7f50b | 585 | tm->tm_mday = 0; |
16433e2b SP |
586 | break; |
587 | ||
588 | case 'M': | |
589 | case 'S': | |
590 | if (*buf == 0 || isspace((unsigned char)*buf)) | |
591 | break; | |
592 | ||
593 | if (!isdigit((unsigned char)*buf)) | |
594 | return 0; | |
595 | ||
596 | len = 2; | |
597 | for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) { | |
598 | i *= 10; | |
599 | i += *buf - '0'; | |
600 | len--; | |
601 | } | |
602 | ||
603 | if (c == 'M') { | |
604 | if (i > 59) | |
605 | return 0; | |
606 | tm->tm_min = i; | |
607 | } else { | |
608 | if (i > 60) | |
609 | return 0; | |
610 | tm->tm_sec = i; | |
611 | } | |
612 | ||
613 | if (*buf != 0 && isspace((unsigned char)*buf)) | |
614 | while (*ptr != 0 && !isspace((unsigned char)*ptr)) | |
615 | ptr++; | |
616 | break; | |
617 | ||
618 | case 'H': | |
619 | case 'I': | |
620 | case 'k': | |
621 | case 'l': | |
622 | /* | |
623 | * Of these, %l is the only specifier explicitly | |
624 | * documented as not being zero-padded. However, | |
625 | * there is no harm in allowing zero-padding. | |
626 | * | |
627 | * XXX The %l specifier may gobble one too many | |
628 | * digits if used incorrectly. | |
629 | */ | |
90d55c29 | 630 | if (!isdigit((unsigned char)*buf)) |
16433e2b SP |
631 | return 0; |
632 | ||
633 | len = 2; | |
634 | for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) { | |
635 | i *= 10; | |
636 | i += *buf - '0'; | |
637 | len--; | |
638 | } | |
639 | if (c == 'H' || c == 'k') { | |
640 | if (i > 23) | |
641 | return 0; | |
642 | } else if (i > 12) | |
643 | return 0; | |
644 | ||
645 | tm->tm_hour = i; | |
646 | ||
647 | if (*buf != 0 && isspace((unsigned char)*buf)) | |
648 | while (*ptr != 0 && !isspace((unsigned char)*ptr)) | |
649 | ptr++; | |
650 | break; | |
651 | ||
652 | case 'p': | |
653 | /* | |
654 | * XXX This is bogus if parsed before hour-related | |
655 | * specifiers. | |
656 | */ | |
90d55c29 | 657 | len = strlen(Locale->am); |
16433e2b SP |
658 | if (strncasecmp(buf, Locale->am, len) == 0) { |
659 | if (tm->tm_hour > 12) | |
660 | return 0; | |
661 | if (tm->tm_hour == 12) | |
662 | tm->tm_hour = 0; | |
663 | buf += len; | |
664 | break; | |
665 | } | |
666 | ||
667 | len = strlen(Locale->pm); | |
668 | if (strncasecmp(buf, Locale->pm, len) == 0) { | |
669 | if (tm->tm_hour > 12) | |
670 | return 0; | |
671 | if (tm->tm_hour != 12) | |
672 | tm->tm_hour += 12; | |
673 | buf += len; | |
674 | break; | |
675 | } | |
676 | ||
677 | return 0; | |
678 | ||
679 | case 'A': | |
680 | case 'a': | |
12bbe8b8 | 681 | for (i = 0; i < (int)asizeof(Locale->weekday); i++) { |
16433e2b SP |
682 | if (c == 'A') { |
683 | len = strlen(Locale->weekday[i]); | |
684 | if (strncasecmp(buf, | |
685 | Locale->weekday[i], | |
686 | len) == 0) | |
687 | break; | |
688 | } else { | |
689 | len = strlen(Locale->wday[i]); | |
690 | if (strncasecmp(buf, | |
691 | Locale->wday[i], | |
692 | len) == 0) | |
693 | break; | |
694 | } | |
695 | } | |
12bbe8b8 | 696 | if (i == (int)asizeof(Locale->weekday)) |
16433e2b SP |
697 | return 0; |
698 | ||
699 | tm->tm_wday = i; | |
700 | buf += len; | |
701 | break; | |
702 | ||
703 | case 'U': | |
704 | case 'W': | |
705 | /* | |
706 | * XXX This is bogus, as we can not assume any valid | |
707 | * information present in the tm structure at this | |
708 | * point to calculate a real value, so just check the | |
709 | * range for now. | |
710 | */ | |
90d55c29 | 711 | if (!isdigit((unsigned char)*buf)) |
16433e2b SP |
712 | return 0; |
713 | ||
714 | len = 2; | |
715 | for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) { | |
716 | i *= 10; | |
717 | i += *buf - '0'; | |
718 | len--; | |
719 | } | |
720 | if (i > 53) | |
721 | return 0; | |
722 | ||
723 | if (*buf != 0 && isspace((unsigned char)*buf)) | |
724 | while (*ptr != 0 && !isspace((unsigned char)*ptr)) | |
725 | ptr++; | |
726 | break; | |
727 | ||
728 | case 'w': | |
729 | if (!isdigit((unsigned char)*buf)) | |
730 | return 0; | |
731 | ||
732 | i = *buf - '0'; | |
733 | if (i > 6) | |
734 | return 0; | |
735 | ||
736 | tm->tm_wday = i; | |
737 | ||
738 | if (*buf != 0 && isspace((unsigned char)*buf)) | |
739 | while (*ptr != 0 && !isspace((unsigned char)*ptr)) | |
740 | ptr++; | |
741 | break; | |
742 | ||
743 | case 'd': | |
744 | case 'e': | |
745 | /* | |
746 | * The %e specifier is explicitly documented as not | |
747 | * being zero-padded but there is no harm in allowing | |
748 | * such padding. | |
749 | * | |
750 | * XXX The %e specifier may gobble one too many | |
751 | * digits if used incorrectly. | |
752 | */ | |
753 | if (!isdigit((unsigned char)*buf)) | |
754 | return 0; | |
755 | ||
756 | len = 2; | |
757 | for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) { | |
758 | i *= 10; | |
759 | i += *buf - '0'; | |
760 | len--; | |
761 | } | |
762 | if (i > 31) | |
763 | return 0; | |
764 | ||
765 | tm->tm_mday = i; | |
766 | ||
767 | if (*buf != 0 && isspace((unsigned char)*buf)) | |
768 | while (*ptr != 0 && !isspace((unsigned char)*ptr)) | |
769 | ptr++; | |
770 | break; | |
771 | ||
772 | case 'B': | |
773 | case 'b': | |
774 | case 'h': | |
12bbe8b8 | 775 | for (i = 0; i < (int)asizeof(Locale->month); i++) { |
16433e2b SP |
776 | if (Oalternative) { |
777 | if (c == 'B') { | |
778 | len = strlen(Locale->alt_month[i]); | |
779 | if (strncasecmp(buf, | |
780 | Locale->alt_month[i], | |
781 | len) == 0) | |
782 | break; | |
783 | } | |
784 | } else { | |
785 | if (c == 'B') { | |
786 | len = strlen(Locale->month[i]); | |
787 | if (strncasecmp(buf, | |
788 | Locale->month[i], | |
789 | len) == 0) | |
790 | break; | |
791 | } else { | |
792 | len = strlen(Locale->mon[i]); | |
793 | if (strncasecmp(buf, | |
794 | Locale->mon[i], | |
795 | len) == 0) | |
796 | break; | |
797 | } | |
798 | } | |
799 | } | |
12bbe8b8 | 800 | if (i == (int)asizeof(Locale->month)) |
16433e2b SP |
801 | return 0; |
802 | ||
803 | tm->tm_mon = i; | |
804 | buf += len; | |
805 | break; | |
806 | ||
807 | case 'm': | |
808 | if (!isdigit((unsigned char)*buf)) | |
809 | return 0; | |
810 | ||
811 | len = 2; | |
812 | for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) { | |
813 | i *= 10; | |
814 | i += *buf - '0'; | |
815 | len--; | |
816 | } | |
817 | if (i < 1 || i > 12) | |
818 | return 0; | |
819 | ||
820 | tm->tm_mon = i - 1; | |
821 | ||
822 | if (*buf != 0 && isspace((unsigned char)*buf)) | |
823 | while (*ptr != 0 && !isspace((unsigned char)*ptr)) | |
824 | ptr++; | |
825 | break; | |
826 | ||
90d55c29 CBW |
827 | case 's': |
828 | { | |
829 | char *cp; | |
830 | int sverrno; | |
831 | long n; | |
832 | time_t t; | |
833 | struct tm mytm; | |
834 | ||
835 | sverrno = errno; | |
836 | errno = 0; | |
837 | n = strtol(buf, &cp, 10); | |
838 | if (errno == ERANGE || (long)(t = n) != n) { | |
839 | errno = sverrno; | |
840 | return 0; | |
841 | } | |
842 | errno = sverrno; | |
843 | buf = cp; | |
844 | memset(&mytm, 0, sizeof(mytm)); | |
845 | my_init_tm(&mytm); /* XXX workaround - see my_init_tm() above */ | |
846 | mytm = *gmtime(&t); | |
847 | tm->tm_sec = mytm.tm_sec; | |
848 | tm->tm_min = mytm.tm_min; | |
849 | tm->tm_hour = mytm.tm_hour; | |
850 | tm->tm_mday = mytm.tm_mday; | |
851 | tm->tm_mon = mytm.tm_mon; | |
852 | tm->tm_year = mytm.tm_year; | |
853 | tm->tm_wday = mytm.tm_wday; | |
854 | tm->tm_yday = mytm.tm_yday; | |
855 | tm->tm_isdst = mytm.tm_isdst; | |
856 | } | |
857 | break; | |
858 | ||
16433e2b SP |
859 | case 'Y': |
860 | case 'y': | |
861 | if (*buf == 0 || isspace((unsigned char)*buf)) | |
862 | break; | |
863 | ||
864 | if (!isdigit((unsigned char)*buf)) | |
865 | return 0; | |
866 | ||
867 | len = (c == 'Y') ? 4 : 2; | |
868 | for (i = 0; len && *buf != 0 && isdigit((unsigned char)*buf); buf++) { | |
869 | i *= 10; | |
870 | i += *buf - '0'; | |
871 | len--; | |
872 | } | |
873 | if (c == 'Y') | |
874 | i -= 1900; | |
875 | if (c == 'y' && i < 69) | |
876 | i += 100; | |
877 | if (i < 0) | |
878 | return 0; | |
879 | ||
880 | tm->tm_year = i; | |
881 | ||
882 | if (*buf != 0 && isspace((unsigned char)*buf)) | |
883 | while (*ptr != 0 && !isspace((unsigned char)*ptr)) | |
884 | ptr++; | |
885 | break; | |
886 | ||
887 | case 'Z': | |
888 | { | |
889 | const char *cp; | |
890 | char *zonestr; | |
891 | ||
12bbe8b8 | 892 | for (cp = buf; *cp && isupper((unsigned char)*cp); ++cp) |
16433e2b SP |
893 | {/*empty*/} |
894 | if (cp - buf) { | |
8a0cff69 | 895 | zonestr = (char *)malloc(cp - buf + 1); |
ded2eedc CB |
896 | if (!zonestr) { |
897 | errno = ENOMEM; | |
898 | return 0; | |
899 | } | |
16433e2b SP |
900 | strncpy(zonestr, buf, cp - buf); |
901 | zonestr[cp - buf] = '\0'; | |
12016aad | 902 | my_tzset(aTHX); |
16433e2b | 903 | if (0 == strcmp(zonestr, "GMT")) { |
e9f284c9 | 904 | *got_GMT = 1; |
16433e2b | 905 | } |
ded2eedc | 906 | free(zonestr); |
e9f284c9 | 907 | if (!*got_GMT) return 0; |
16433e2b SP |
908 | buf += cp - buf; |
909 | } | |
910 | } | |
911 | break; | |
90d55c29 CBW |
912 | |
913 | case 'z': | |
914 | { | |
915 | int sign = 1; | |
916 | ||
917 | if (*buf != '+') { | |
918 | if (*buf == '-') | |
919 | sign = -1; | |
920 | else | |
921 | return 0; | |
922 | } | |
923 | ||
924 | buf++; | |
925 | i = 0; | |
926 | for (len = 4; len > 0; len--) { | |
927 | if (isdigit((int)*buf)) { | |
928 | i *= 10; | |
929 | i += *buf - '0'; | |
930 | buf++; | |
931 | } else | |
932 | return 0; | |
933 | } | |
934 | ||
935 | tm->tm_hour -= sign * (i / 100); | |
936 | tm->tm_min -= sign * (i % 100); | |
e9f284c9 | 937 | *got_GMT = 1; |
90d55c29 CBW |
938 | } |
939 | break; | |
16433e2b SP |
940 | } |
941 | } | |
942 | return (char *)buf; | |
943 | } | |
944 | ||
d93e3b8c CBW |
945 | /* Saves alot of machine code. |
946 | Takes a (auto) SP, which may or may not have been PUSHed before, puts | |
947 | tm struct members on Perl stack, then returns new, advanced, SP to caller. | |
948 | Assign the return of push_common_tm to your SP, so you can continue to PUSH | |
949 | or do a PUTBACK and return eventually. | |
950 | !!!! push_common_tm does not touch PL_stack_sp !!!! | |
951 | !!!! do not use PUTBACK then SPAGAIN semantics around push_common_tm !!!! | |
952 | !!!! You must mortalize whatever push_common_tm put on stack yourself to | |
953 | avoid leaking !!!! | |
954 | */ | |
955 | SV ** | |
956 | push_common_tm(pTHX_ SV ** SP, struct tm *mytm) | |
957 | { | |
958 | PUSHs(newSViv(mytm->tm_sec)); | |
959 | PUSHs(newSViv(mytm->tm_min)); | |
960 | PUSHs(newSViv(mytm->tm_hour)); | |
961 | PUSHs(newSViv(mytm->tm_mday)); | |
962 | PUSHs(newSViv(mytm->tm_mon)); | |
963 | PUSHs(newSViv(mytm->tm_year)); | |
964 | PUSHs(newSViv(mytm->tm_wday)); | |
965 | PUSHs(newSViv(mytm->tm_yday)); | |
12bbe8b8 | 966 | PUSHs(newSViv(mytm->tm_isdst)); |
d93e3b8c CBW |
967 | return SP; |
968 | } | |
969 | ||
970 | /* specialized common end of 2 XSUBs | |
971 | SV ** SP -- pass your (auto) SP, which has not been PUSHed before, but was | |
972 | reset to 0 (PPCODE only or SP -= items or XSprePUSH) | |
973 | tm *mytm -- a tm *, will be proprocessed with my_mini_mktime | |
974 | return -- none, after calling return_11part_tm, you must call "return;" | |
975 | no exceptions | |
976 | */ | |
977 | void | |
978 | return_11part_tm(pTHX_ SV ** SP, struct tm *mytm) | |
979 | { | |
980 | my_mini_mktime(mytm); | |
981 | ||
982 | /* warn("tm: %d-%d-%d %d:%d:%d\n", mytm.tm_year, mytm.tm_mon, mytm.tm_mday, mytm.tm_hour, mytm.tm_min, mytm.tm_sec); */ | |
d93e3b8c CBW |
983 | EXTEND(SP, 11); |
984 | SP = push_common_tm(aTHX_ SP, mytm); | |
d93e3b8c CBW |
985 | /* epoch */ |
986 | PUSHs(newSViv(0)); | |
987 | /* islocal */ | |
988 | PUSHs(newSViv(0)); | |
989 | PUTBACK; | |
990 | { | |
991 | SV ** endsp = SP; /* the SV * under SP needs to be mortaled */ | |
992 | SP -= (11 - 1); /* subtract 0 based count of SVs to mortal */ | |
993 | /* mortal target of SP, then increment before function call | |
994 | so SP is already calculated before next comparison to not stall CPU */ | |
995 | do { | |
996 | sv_2mortal(*SP++); | |
997 | } while(SP <= endsp); | |
998 | } | |
999 | return; | |
1000 | } | |
1001 | ||
16433e2b SP |
1002 | MODULE = Time::Piece PACKAGE = Time::Piece |
1003 | ||
1004 | PROTOTYPES: ENABLE | |
1005 | ||
9331e88f | 1006 | void |
12bbe8b8 CBW |
1007 | _strftime(fmt, epoch, islocal = 1) |
1008 | char * fmt | |
1009 | time_t epoch | |
1010 | int islocal | |
16433e2b SP |
1011 | CODE: |
1012 | { | |
1013 | char tmpbuf[128]; | |
1014 | struct tm mytm; | |
1015 | int len; | |
12bbe8b8 CBW |
1016 | |
1017 | if(islocal == 1) | |
1018 | mytm = *localtime(&epoch); | |
1019 | else | |
1020 | mytm = *gmtime(&epoch); | |
1021 | ||
16433e2b SP |
1022 | len = strftime(tmpbuf, sizeof tmpbuf, fmt, &mytm); |
1023 | /* | |
12bbe8b8 | 1024 | ** The following is needed to handle to the situation where |
16433e2b SP |
1025 | ** tmpbuf overflows. Basically we want to allocate a buffer |
1026 | ** and try repeatedly. The reason why it is so complicated | |
1027 | ** is that getting a return value of 0 from strftime can indicate | |
1028 | ** one of the following: | |
1029 | ** 1. buffer overflowed, | |
1030 | ** 2. illegal conversion specifier, or | |
1031 | ** 3. the format string specifies nothing to be returned(not | |
1032 | ** an error). This could be because format is an empty string | |
1033 | ** or it specifies %p that yields an empty string in some locale. | |
1034 | ** If there is a better way to make it portable, go ahead by | |
1035 | ** all means. | |
1036 | */ | |
1037 | if ((len > 0 && len < sizeof(tmpbuf)) || (len == 0 && *fmt == '\0')) | |
1038 | ST(0) = sv_2mortal(newSVpv(tmpbuf, len)); | |
1039 | else { | |
1040 | /* Possibly buf overflowed - try again with a bigger buf */ | |
1041 | int fmtlen = strlen(fmt); | |
1042 | int bufsize = fmtlen + sizeof(tmpbuf); | |
1043 | char* buf; | |
1044 | int buflen; | |
1045 | ||
1046 | New(0, buf, bufsize, char); | |
1047 | while (buf) { | |
1048 | buflen = strftime(buf, bufsize, fmt, &mytm); | |
1049 | if (buflen > 0 && buflen < bufsize) | |
1050 | break; | |
1051 | /* heuristic to prevent out-of-memory errors */ | |
1052 | if (bufsize > 100*fmtlen) { | |
1053 | Safefree(buf); | |
1054 | buf = NULL; | |
1055 | break; | |
1056 | } | |
1057 | bufsize *= 2; | |
1058 | Renew(buf, bufsize, char); | |
1059 | } | |
1060 | if (buf) { | |
1061 | ST(0) = sv_2mortal(newSVpv(buf, buflen)); | |
1062 | Safefree(buf); | |
1063 | } | |
1064 | else | |
1065 | ST(0) = sv_2mortal(newSVpv(tmpbuf, len)); | |
1066 | } | |
1067 | } | |
1068 | ||
1069 | void | |
1070 | _tzset() | |
1071 | PPCODE: | |
d93e3b8c | 1072 | PUTBACK; /* makes rest of this function tailcall friendly */ |
12016aad | 1073 | my_tzset(aTHX); |
d93e3b8c | 1074 | return; /* skip XSUBPP's PUTBACK */ |
16433e2b SP |
1075 | |
1076 | void | |
1077 | _strptime ( string, format ) | |
1078 | char * string | |
1079 | char * format | |
1080 | PREINIT: | |
16433e2b SP |
1081 | struct tm mytm; |
1082 | time_t t; | |
1083 | char * remainder; | |
933a2256 | 1084 | int got_GMT; |
16433e2b SP |
1085 | PPCODE: |
1086 | t = 0; | |
1087 | mytm = *gmtime(&t); | |
12bbe8b8 | 1088 | mytm.tm_isdst = -1; /* -1 means we don't know */ |
933a2256 CBW |
1089 | got_GMT = 0; |
1090 | ||
1091 | remainder = (char *)_strptime(aTHX_ string, format, &mytm, &got_GMT); | |
16433e2b | 1092 | if (remainder == NULL) { |
90d55c29 | 1093 | croak("Error parsing time"); |
16433e2b | 1094 | } |
16433e2b SP |
1095 | if (*remainder != '\0') { |
1096 | warn("garbage at end of string in strptime: %s", remainder); | |
1097 | } | |
16433e2b | 1098 | |
d93e3b8c CBW |
1099 | return_11part_tm(aTHX_ SP, &mytm); |
1100 | return; | |
3df1a9e2 GA |
1101 | |
1102 | void | |
1103 | _mini_mktime(int sec, int min, int hour, int mday, int mon, int year) | |
1104 | PREINIT: | |
1105 | struct tm mytm; | |
1106 | time_t t; | |
1107 | PPCODE: | |
1108 | t = 0; | |
1109 | mytm = *gmtime(&t); | |
1110 | ||
1111 | mytm.tm_sec = sec; | |
1112 | mytm.tm_min = min; | |
1113 | mytm.tm_hour = hour; | |
1114 | mytm.tm_mday = mday; | |
1115 | mytm.tm_mon = mon; | |
1116 | mytm.tm_year = year; | |
3df1a9e2 | 1117 | |
d93e3b8c CBW |
1118 | return_11part_tm(aTHX_ SP, &mytm); |
1119 | return; | |
90d55c29 CBW |
1120 | |
1121 | void | |
1122 | _crt_localtime(time_t sec) | |
d93e3b8c CBW |
1123 | ALIAS: |
1124 | _crt_gmtime = 1 | |
90d55c29 CBW |
1125 | PREINIT: |
1126 | struct tm mytm; | |
1127 | PPCODE: | |
d93e3b8c CBW |
1128 | if(ix) mytm = *gmtime(&sec); |
1129 | else mytm = *localtime(&sec); | |
90d55c29 | 1130 | /* Need to get: $s,$n,$h,$d,$m,$y */ |
12bbe8b8 | 1131 | |
90d55c29 | 1132 | EXTEND(SP, 9); |
d93e3b8c CBW |
1133 | SP = push_common_tm(aTHX_ SP, &mytm); |
1134 | PUSHs(newSViv(mytm.tm_isdst)); | |
1135 | PUTBACK; | |
1136 | { | |
1137 | SV ** endsp = SP; /* the SV * under SP needs to be mortaled */ | |
1138 | SP -= (9 - 1); /* subtract 0 based count of SVs to mortal */ | |
1139 | /* mortal target of SP, then increment before function call | |
1140 | so SP is already calculated before next comparison to not stall CPU */ | |
1141 | do { | |
1142 | sv_2mortal(*SP++); | |
1143 | } while(SP <= endsp); | |
1144 | } | |
1145 | return; |