strftime.c

   1 /* -*- c-file-style: "linux" -*- */
   2
   3 /*
   4  * strftime.c
   5  *
   6  * Public-domain implementation of ANSI C library routine.
   7  *
   8  * It's written in old-style C for maximal portability.
   9  * However, since I'm used to prototypes, I've included them too.
  10  *
  11  * If you want stuff in the System V ascftime routine, add the SYSV_EXT define.
  12  * For extensions from SunOS, add SUNOS_EXT.
  13  * For stuff needed to implement the P1003.2 date command, add POSIX2_DATE.
  14  * For VMS dates, add VMS_EXT.
  15  * For a an RFC822 time format, add MAILHEADER_EXT.
  16  * For ISO week years, add ISO_DATE_EXT.
  17  * For complete POSIX semantics, add POSIX_SEMANTICS.
  18  *
  19  * The code for %c, %x, and %X now follows the 1003.2 specification for
  20  * the POSIX locale.
  21  * This version ignores LOCALE information.
  22  * It also doesn't worry about multi-byte characters.
  23  * So there.
  24  *
  25  * This file is also shipped with GAWK (GNU Awk), gawk specific bits of
  26  * code are included if GAWK is defined.
  27  *
  28  * Arnold Robbins
  29  * January, February, March, 1991
  30  * Updated March, April 1992
  31  * Updated April, 1993
  32  * Updated February, 1994
  33  * Updated May, 1994
  34  * Updated January, 1995
  35  * Updated September, 1995
  36  * Updated January, 1996
  37  *
  38  * Fixes from ado@elsie.nci.nih.gov
  39  * February 1991, May 1992
  40  * Fixes from Tor Lillqvist tml@tik.vtt.fi
  41  * May, 1993
  42  * Further fixes from ado@elsie.nci.nih.gov
  43  * February 1994
  44  * %z code from chip@chinacat.unicom.com
  45  * Applied September 1995
  46  * %V code fixed (again) and %G, %g added,
  47  * January 1996
  48  */
  49
  50 #include "ruby/internal/config.h"
  51
  52 #ifndef GAWK
  53 #include <stdio.h>
  54 #include <ctype.h>
  55 #include <string.h>
  56 #include <time.h>
  57 #include <sys/types.h>
  58 #include <errno.h>
  59 #endif
  60 #if defined(TM_IN_SYS_TIME) || !defined(GAWK)
  61 #include <sys/types.h>
  62 #ifdef HAVE_SYS_TIME_H
  63 #include <sys/time.h>
  64 #endif
  65 #endif
  66 #include <math.h>
  67
  68 #include "internal.h"
  69 #include "internal/encoding.h"
  70 #include "internal/string.h"
  71 #include "internal/vm.h"
  72 #include "ruby/encoding.h"
  73 #include "ruby/ruby.h"
  74 #include "ruby/util.h"
  75 #include "timev.h"
  76
  77 /* defaults: season to taste */
  78 #define SYSV_EXT        1       /* stuff in System V ascftime routine */
  79 #define SUNOS_EXT       1       /* stuff in SunOS strftime routine */
  80 #define POSIX2_DATE     1       /* stuff in Posix 1003.2 date command */
  81 #define VMS_EXT         1       /* include %v for VMS date format */
  82 #define MAILHEADER_EXT  1       /* add %z for HHMM format */
  83 #define ISO_DATE_EXT    1       /* %G and %g for year of ISO week */
  84
  85 #if defined(ISO_DATE_EXT)
  86 #if ! defined(POSIX2_DATE)
  87 #define POSIX2_DATE     1
  88 #endif
  89 #endif
  90
  91 #if defined(POSIX2_DATE)
  92 #if ! defined(SYSV_EXT)
  93 #define SYSV_EXT        1
  94 #endif
  95 #if ! defined(SUNOS_EXT)
  96 #define SUNOS_EXT       1
  97 #endif
  98 #endif
  99
 100 #if defined(POSIX2_DATE)
 101 #define adddecl(stuff)  stuff
 102 #else
 103 #define adddecl(stuff)
 104 #endif
 105
 106 #undef strchr   /* avoid AIX weirdness */
 107
 108 #if !defined __STDC__ && !defined _WIN32
 109 #define const   /**/
 110 static int weeknumber();
 111 adddecl(static int iso8601wknum();)
 112 static int weeknumber_v();
 113 adddecl(static int iso8601wknum_v();)
 114 #else
 115 static int weeknumber(const struct tm *timeptr, int firstweekday);
 116 adddecl(static int iso8601wknum(const struct tm *timeptr);)
 117 static int weeknumber_v(const struct vtm *vtm, int firstweekday);
 118 adddecl(static int iso8601wknum_v(const struct vtm *vtm);)
 119 #endif
 120
 121 #ifdef STDC_HEADERS
 122 #include <stdlib.h>
 123 #include <string.h>
 124 #else
 125 extern void *malloc();
 126 extern void *realloc();
 127 extern char *getenv();
 128 extern char *strchr();
 129 #endif
 130
 131 #define range(low, item, hi)    max((low), min((item), (hi)))
 132
 133 #undef min      /* just in case */
 134
 135 /* min --- return minimum of two numbers */
 136
 137 static inline int
 138 min(int a, int b)
 139 {
 140         return (a < b ? a : b);
 141 }
 142
 143 #undef max      /* also, just in case */
 144
 145 /* max --- return maximum of two numbers */
 146
 147 static inline int
 148 max(int a, int b)
 149 {
 150         return (a > b ? a : b);
 151 }
 152
 153 #ifdef NO_STRING_LITERAL_CONCATENATION
 154 #error No string literal concatenation
 155 #endif
 156
 157 #define add(x,y) (rb_funcall((x), '+', 1, (y)))
 158 #define sub(x,y) (rb_funcall((x), '-', 1, (y)))
 159 #define mul(x,y) (rb_funcall((x), '*', 1, (y)))
 160 #define quo(x,y) (rb_funcall((x), rb_intern("quo"), 1, (y)))
 161 #define div(x,y) (rb_funcall((x), rb_intern("div"), 1, (y)))
 162 #define mod(x,y) (rb_funcall((x), '%', 1, (y)))
 163
 164 /* strftime --- produce formatted time */
 165
 166 enum {LEFT, CHCASE, LOWER, UPPER};
 167 #define BIT_OF(n) (1U<<(n))
 168
 169 static char *
 170 resize_buffer(VALUE ftime, char *s, const char **start, const char **endp,
 171               ptrdiff_t n, size_t maxsize)
 172 {
 173         size_t len = s - *start;
 174         size_t need = len + n * 2;
 175         size_t nlen = rb_str_capacity(ftime);
 176         while (nlen < need) nlen <<= 1;
 177
 178         if (nlen < len || nlen > maxsize) {
 179                 return 0;
 180         }
 181         rb_str_set_len(ftime, len);
 182         rb_str_modify_expand(ftime, nlen-len);
 183         s = RSTRING_PTR(ftime);
 184         *endp = s + nlen;
 185         *start = s;
 186         return s += len;
 187 }
 188
 189 static void
 190 buffer_size_check(const char *s,
 191                   const char *format_end, size_t format_len,
 192                   rb_encoding *enc)
 193 {
 194         if (!s) {
 195                 const char *format = format_end-format_len;
 196                 VALUE fmt = rb_enc_str_new(format, format_len, enc);
 197                 rb_syserr_fail_str(ERANGE, fmt);
 198         }
 199 }
 200
 201 static char *
 202 case_conv(char *s, ptrdiff_t i, int flags)
 203 {
 204         switch (flags & (BIT_OF(UPPER)|BIT_OF(LOWER))) {
 205         case BIT_OF(UPPER):
 206                 do {
 207                         if (ISLOWER(*s)) *s = TOUPPER(*s);
 208                 } while (s++, --i);
 209                 break;
 210         case BIT_OF(LOWER):
 211                 do {
 212                         if (ISUPPER(*s)) *s = TOLOWER(*s);
 213                 } while (s++, --i);
 214                 break;
 215         default:
 216                 s += i;
 217                 break;
 218         }
 219         return s;
 220 }
 221
 222 static VALUE
 223 format_value(VALUE val, int base)
 224 {
 225         if (!RB_BIGNUM_TYPE_P(val))
 226                 val = rb_Integer(val);
 227         return rb_big2str(val, base);
 228 }
 229
 230 /*
 231  * enc is the encoding of the format. It is used as the encoding of resulted
 232  * string, but the name of the month and weekday are always US-ASCII. So it
 233  * is only used for the timezone name on Windows.
 234  */
 235 static VALUE
 236 rb_strftime_with_timespec(VALUE ftime, const char *format, size_t format_len,
 237                           rb_encoding *enc, VALUE time, const struct vtm *vtm,
 238                           VALUE timev, struct timespec *ts, int gmt, size_t maxsize)
 239 {
 240         size_t len = RSTRING_LEN(ftime);
 241         char *s = RSTRING_PTR(ftime);
 242         const char *start = s;
 243         const char *endp = start + rb_str_capacity(ftime);
 244         const char *const format_end = format + format_len;
 245         const char *sp, *tp;
 246 #define TBUFSIZE 100
 247         auto char tbuf[TBUFSIZE];
 248         long off;
 249         ptrdiff_t i;
 250         int w;
 251         long y;
 252         int precision, flags, colons;
 253         char padding;
 254 #ifdef MAILHEADER_EXT
 255         int sign;
 256 #endif
 257         VALUE zone = Qnil;
 258
 259         /* various tables, useful in North America */
 260         static const char days_l[][10] = {
 261                 "Sunday", "Monday", "Tuesday", "Wednesday",
 262                 "Thursday", "Friday", "Saturday",
 263         };
 264         static const char months_l[][10] = {
 265                 "January", "February", "March", "April",
 266                 "May", "June", "July", "August", "September",
 267                 "October", "November", "December",
 268         };
 269         static const char ampm[][3] = { "AM", "PM", };
 270
 271         if (format == NULL || format_len == 0 || vtm == NULL) {
 272                 goto err;
 273         }
 274
 275         if (enc &&
 276             (rb_is_usascii_enc(enc) ||
 277              rb_is_ascii8bit_enc(enc) ||
 278              rb_is_locale_enc(enc))) {
 279                 enc = NULL;
 280         }
 281
 282         s += len;
 283         for (; format < format_end; format++) {
 284 #define FLAG_FOUND() do { \
 285                         if (precision > 0) \
 286                                 goto unknown; \
 287                 } while (0)
 288 #define NEEDS(n) do { \
 289                         if (s >= endp || (n) >= endp - s - 1) { \
 290                                 s = resize_buffer(ftime, s, &start, &endp, (n), maxsize); \
 291                                 buffer_size_check(s, format_end, format_len, enc); \
 292                         } \
 293                 } while (0)
 294 #define FILL_PADDING(i) do { \
 295         if (!(flags & BIT_OF(LEFT)) && precision > (i)) { \
 296                 NEEDS(precision); \
 297                 memset(s, padding ? padding : ' ', precision - (i)); \
 298                 s += precision - (i); \
 299         } \
 300         else { \
 301                 NEEDS(i); \
 302         } \
 303 } while (0);
 304 #define FMT_PADDING(fmt, def_pad) \
 305                 (&"%*"fmt"\0""%0*"fmt[\
 306                         (padding == '0' || (!padding && (def_pad) == '0')) ? \
 307                         rb_strlen_lit("%*"fmt)+1 : 0])
 308 #define FMT_PRECISION(def_prec) \
 309                 ((flags & BIT_OF(LEFT)) ? (1) : \
 310                  (precision <= 0) ? (def_prec) : (precision))
 311 #define FMT(def_pad, def_prec, fmt, val) \
 312                 do { \
 313                         precision = FMT_PRECISION(def_prec); \
 314                         len = s - start; \
 315                         NEEDS(precision); \
 316                         rb_str_set_len(ftime, len); \
 317                         rb_str_catf(ftime, FMT_PADDING(fmt, def_pad), \
 318                                     precision, (val)); \
 319                         RSTRING_GETMEM(ftime, s, len); \
 320                         endp = (start = s) + rb_str_capacity(ftime); \
 321                         s += len; \
 322                 } while (0)
 323 #define STRFTIME(fmt) \
 324                 do { \
 325                         len = s - start; \
 326                         rb_str_set_len(ftime, len); \
 327                         if (!rb_strftime_with_timespec(ftime, (fmt), rb_strlen_lit(fmt), \
 328                                                        enc, time, vtm, timev, ts, gmt, maxsize)) \
 329                                 return 0; \
 330                         s = RSTRING_PTR(ftime); \
 331                         i = RSTRING_LEN(ftime) - len; \
 332                         endp = (start = s) + rb_str_capacity(ftime); \
 333                         s += len; \
 334                         if (i > 0) case_conv(s, i, flags); \
 335                         if (precision > i) {\
 336                                 s += i; \
 337                                 NEEDS(precision); \
 338                                 s -= i; \
 339                                 memmove(s + precision - i, s, i);\
 340                                 memset(s, padding ? padding : ' ', precision - i); \
 341                                 s += precision; \
 342                         } \
 343                         else s += i; \
 344                 } while (0)
 345 #define FMTV(def_pad, def_prec, fmt, val) \
 346                 do { \
 347                         VALUE tmp = (val); \
 348                         if (FIXNUM_P(tmp)) { \
 349                                 FMT((def_pad), (def_prec), "l"fmt, FIX2LONG(tmp)); \
 350                         } \
 351                         else { \
 352                                 const int base = ((fmt[0] == 'x') ? 16 : \
 353                                                   (fmt[0] == 'o') ? 8 : \
 354                                                   10); \
 355                                 precision = FMT_PRECISION(def_prec); \
 356                                 if (!padding) padding = (def_pad); \
 357                                 tmp = format_value(tmp, base); \
 358                                 i = RSTRING_LEN(tmp); \
 359                                 FILL_PADDING(i); \
 360                                 rb_str_set_len(ftime, s-start); \
 361                                 rb_str_append(ftime, tmp); \
 362                                 RSTRING_GETMEM(ftime, s, len); \
 363                                 endp = (start = s) + rb_str_capacity(ftime); \
 364                                 s += len; \
 365                         } \
 366                 } while (0)
 367
 368                 tp = memchr(format, '%', format_end - format);
 369                 if (!tp) tp = format_end;
 370                 NEEDS(tp - format);
 371                 memcpy(s, format, tp - format);
 372                 s += tp - format;
 373                 format = tp;
 374                 if (format == format_end) break;
 375
 376                 tp = tbuf;
 377                 sp = format;
 378                 precision = -1;
 379                 flags = 0;
 380                 padding = 0;
 381                 colons = 0;
 382         again:
 383                 if (++format >= format_end) goto unknown;
 384                 switch (*format) {
 385                 case '%':
 386                         FILL_PADDING(1);
 387                         *s++ = '%';
 388                         continue;
 389
 390                 case 'a':       /* abbreviated weekday name */
 391                         if (flags & BIT_OF(CHCASE)) {
 392                                 flags &= ~(BIT_OF(LOWER)|BIT_OF(CHCASE));
 393                                 flags |= BIT_OF(UPPER);
 394                         }
 395                         if (vtm->wday > 6)
 396                                 i = 1, tp = "?";
 397                         else
 398                                 i = 3, tp = days_l[vtm->wday];
 399                         break;
 400
 401                 case 'A':       /* full weekday name */
 402                         if (flags & BIT_OF(CHCASE)) {
 403                                 flags &= ~(BIT_OF(LOWER)|BIT_OF(CHCASE));
 404                                 flags |= BIT_OF(UPPER);
 405                         }
 406                         if (vtm->wday > 6)
 407                                 i = 1, tp = "?";
 408                         else
 409                                 i = strlen(tp = days_l[vtm->wday]);
 410                         break;
 411
 412 #ifdef SYSV_EXT
 413                 case 'h':       /* abbreviated month name */
 414 #endif
 415                 case 'b':       /* abbreviated month name */
 416                         if (flags & BIT_OF(CHCASE)) {
 417                                 flags &= ~(BIT_OF(LOWER)|BIT_OF(CHCASE));
 418                                 flags |= BIT_OF(UPPER);
 419                         }
 420                         if (vtm->mon < 1 || vtm->mon > 12)
 421                                 i = 1, tp = "?";
 422                         else
 423                                 i = 3, tp = months_l[vtm->mon-1];
 424                         break;
 425
 426                 case 'B':       /* full month name */
 427                         if (flags & BIT_OF(CHCASE)) {
 428                                 flags &= ~(BIT_OF(LOWER)|BIT_OF(CHCASE));
 429                                 flags |= BIT_OF(UPPER);
 430                         }
 431                         if (vtm->mon < 1 || vtm->mon > 12)
 432                                 i = 1, tp = "?";
 433                         else
 434                                 i = strlen(tp = months_l[vtm->mon-1]);
 435                         break;
 436
 437                 case 'c':       /* appropriate date and time representation */
 438                         STRFTIME("%a %b %e %H:%M:%S %Y");
 439                         continue;
 440
 441                 case 'd':       /* day of the month, 01 - 31 */
 442                         i = range(1, vtm->mday, 31);
 443                         FMT('0', 2, "d", (int)i);
 444                         continue;
 445
 446                 case 'H':       /* hour, 24-hour clock, 00 - 23 */
 447                         i = range(0, vtm->hour, 23);
 448                         FMT('0', 2, "d", (int)i);
 449                         continue;
 450
 451                 case 'I':       /* hour, 12-hour clock, 01 - 12 */
 452                         i = range(0, vtm->hour, 23);
 453                         if (i == 0)
 454                                 i = 12;
 455                         else if (i > 12)
 456                                 i -= 12;
 457                         FMT('0', 2, "d", (int)i);
 458                         continue;
 459
 460                 case 'j':       /* day of the year, 001 - 366 */
 461                         i = range(1, vtm->yday, 366);
 462                         FMT('0', 3, "d", (int)i);
 463                         continue;
 464
 465                 case 'm':       /* month, 01 - 12 */
 466                         i = range(1, vtm->mon, 12);
 467                         FMT('0', 2, "d", (int)i);
 468                         continue;
 469
 470                 case 'M':       /* minute, 00 - 59 */
 471                         i = range(0, vtm->min, 59);
 472                         FMT('0', 2, "d", (int)i);
 473                         continue;
 474
 475                 case 'p':       /* AM or PM based on 12-hour clock */
 476                 case 'P':       /* am or pm based on 12-hour clock */
 477                         if ((*format == 'p' && (flags & BIT_OF(CHCASE))) ||
 478                             (*format == 'P' && !(flags & (BIT_OF(CHCASE)|BIT_OF(UPPER))))) {
 479                                 flags &= ~(BIT_OF(UPPER)|BIT_OF(CHCASE));
 480                                 flags |= BIT_OF(LOWER);
 481                         }
 482                         i = range(0, vtm->hour, 23);
 483                         if (i < 12)
 484                                 tp = ampm[0];
 485                         else
 486                                 tp = ampm[1];
 487                         i = 2;
 488                         break;
 489
 490                 case 's':
 491                         if (ts) {
 492                                 time_t sec = ts->tv_sec;
 493                                 if (~(time_t)0 <= 0)
 494                                     FMT('0', 1, PRI_TIMET_PREFIX"d", sec);
 495                                 else
 496                                     FMT('0', 1, PRI_TIMET_PREFIX"u", sec);
 497                         }
 498                         else {
 499                                 VALUE sec = div(timev, INT2FIX(1));
 500                                 FMTV('0', 1, "d", sec);
 501                         }
 502                         continue;
 503
 504                 case 'S':       /* second, 00 - 60 */
 505                         i = range(0, vtm->sec, 60);
 506                         FMT('0', 2, "d", (int)i);
 507                         continue;
 508
 509                 case 'U':       /* week of year, Sunday is first day of week */
 510                         FMT('0', 2, "d", weeknumber_v(vtm, 0));
 511                         continue;
 512
 513                 case 'w':       /* weekday, Sunday == 0, 0 - 6 */
 514                         i = range(0, vtm->wday, 6);
 515                         FMT('0', 1, "d", (int)i);
 516                         continue;
 517
 518                 case 'W':       /* week of year, Monday is first day of week */
 519                         FMT('0', 2, "d", weeknumber_v(vtm, 1));
 520                         continue;
 521
 522                 case 'x':       /* appropriate date representation */
 523                         STRFTIME("%m/%d/%y");
 524                         continue;
 525
 526                 case 'X':       /* appropriate time representation */
 527                         STRFTIME("%H:%M:%S");
 528                         continue;
 529
 530                 case 'y':       /* year without a century, 00 - 99 */
 531                         i = NUM2INT(mod(vtm->year, INT2FIX(100)));
 532                         FMT('0', 2, "d", (int)i);
 533                         continue;
 534
 535                 case 'Y':       /* year with century */
 536                         if (FIXNUM_P(vtm->year)) {
 537                                 long y = FIX2LONG(vtm->year);
 538                                 FMT('0', 0 <= y ? 4 : 5, "ld", y);
 539                         }
 540                         else {
 541                                 FMTV('0', 4, "d", vtm->year);
 542                         }
 543                         continue;
 544
 545 #ifdef MAILHEADER_EXT
 546                 case 'z':       /* time zone offset east of GMT e.g. -0600 */
 547                         if (gmt) {
 548                                 off = 0;
 549                         }
 550                         else {
 551                                 off = NUM2LONG(rb_funcall(vtm->utc_offset, rb_intern("round"), 0));
 552                         }
 553                         if (off < 0 || (gmt && (flags & BIT_OF(LEFT)))) {
 554                                 off = -off;
 555                                 sign = -1;
 556                         }
 557                         else {
 558                                 sign = +1;
 559                         }
 560                         switch (colons) {
 561                         case 0: /* %z -> +hhmm */
 562                                 precision = precision <= 5 ? 2 : precision-3;
 563                                 NEEDS(precision + 3);
 564                                 break;
 565
 566                         case 1: /* %:z -> +hh:mm */
 567                                 precision = precision <= 6 ? 2 : precision-4;
 568                                 NEEDS(precision + 4);
 569                                 break;
 570
 571                         case 2: /* %::z -> +hh:mm:ss */
 572                                 precision = precision <= 9 ? 2 : precision-7;
 573                                 NEEDS(precision + 7);
 574                                 break;
 575
 576                         case 3: /* %:::z -> +hh[:mm[:ss]] */
 577                                 if (off % 3600 == 0) {
 578                                         precision = precision <= 3 ? 2 : precision-1;
 579                                         NEEDS(precision + 3);
 580                                 }
 581                                 else if (off % 60 == 0) {
 582                                         precision = precision <= 6 ? 2 : precision-4;
 583                                         NEEDS(precision + 4);
 584                                 }
 585                                 else {
 586                                         precision = precision <= 9 ? 2 : precision-7;
 587                                         NEEDS(precision + 9);
 588                                 }
 589                                 break;
 590
 591                         default:
 592                                 format--;
 593                                 goto unknown;
 594                         }
 595                         i = snprintf(s, endp - s, (padding == ' ' ? "%+*ld" : "%+.*ld"),
 596                                      precision + (padding == ' '), sign * (off / 3600));
 597                         if (i < 0) goto err;
 598                         if (sign < 0 && off < 3600) {
 599                                 *(padding == ' ' ? s + i - 2 : s) = '-';
 600                         }
 601                         s += i;
 602                         off = off % 3600;
 603                         if (colons == 3 && off == 0)
 604                                 continue;
 605                         if (1 <= colons)
 606                             *s++ = ':';
 607                         i = snprintf(s, endp - s, "%02d", (int)(off / 60));
 608                         if (i < 0) goto err;
 609                         s += i;
 610                         off = off % 60;
 611                         if (colons == 3 && off == 0)
 612                                 continue;
 613                         if (2 <= colons) {
 614                             *s++ = ':';
 615                             i = snprintf(s, endp - s, "%02d", (int)off);
 616                             if (i < 0) goto err;
 617                             s += i;
 618                         }
 619                         continue;
 620 #endif /* MAILHEADER_EXT */
 621
 622                 case 'Z':       /* time zone name or abbreviation */
 623                         if (flags & BIT_OF(CHCASE)) {
 624                                 flags &= ~(BIT_OF(UPPER)|BIT_OF(CHCASE));
 625                                 flags |= BIT_OF(LOWER);
 626                         }
 627                         if (gmt) {
 628                                 i = 3;
 629                                 tp = "UTC";
 630                                 break;
 631                         }
 632                         if (NIL_P(vtm->zone)) {
 633                             i = 0;
 634                         }
 635                         else {
 636                             if (NIL_P(zone)) {
 637                                 zone = rb_time_zone_abbreviation(vtm->zone, time);
 638                             }
 639                             tp = RSTRING_PTR(zone);
 640                             if (enc) {
 641                                 for (i = 0; i < TBUFSIZE && tp[i]; i++) {
 642                                     if ((unsigned char)tp[i] > 0x7F) {
 643                                         VALUE str = rb_str_conv_enc_opts(rb_str_new_cstr(tp), rb_locale_encoding(), enc, ECONV_UNDEF_REPLACE|ECONV_INVALID_REPLACE, Qnil);
 644                                         i = strlcpy(tbuf, RSTRING_PTR(str), TBUFSIZE);
 645                                         tp = tbuf;
 646                                         break;
 647                                     }
 648                                 }
 649                             }
 650                             else
 651                                 i = strlen(tp);
 652                         }
 653                         break;
 654
 655 #ifdef SYSV_EXT
 656                 case 'n':       /* same as \n */
 657                         FILL_PADDING(1);
 658                         *s++ = '\n';
 659                         continue;
 660
 661                 case 't':       /* same as \t */
 662                         FILL_PADDING(1);
 663                         *s++ = '\t';
 664                         continue;
 665
 666                 case 'D':       /* date as %m/%d/%y */
 667                         STRFTIME("%m/%d/%y");
 668                         continue;
 669
 670                 case 'e':       /* day of month, blank padded */
 671                         FMT(' ', 2, "d", range(1, vtm->mday, 31));
 672                         continue;
 673
 674                 case 'r':       /* time as %I:%M:%S %p */
 675                         STRFTIME("%I:%M:%S %p");
 676                         continue;
 677
 678                 case 'R':       /* time as %H:%M */
 679                         STRFTIME("%H:%M");
 680                         continue;
 681
 682                 case 'T':       /* time as %H:%M:%S */
 683                         STRFTIME("%H:%M:%S");
 684                         continue;
 685 #endif
 686
 687 #ifdef SUNOS_EXT
 688                 case 'k':       /* hour, 24-hour clock, blank pad */
 689                         i = range(0, vtm->hour, 23);
 690                         FMT(' ', 2, "d", (int)i);
 691                         continue;
 692
 693                 case 'l':       /* hour, 12-hour clock, 1 - 12, blank pad */
 694                         i = range(0, vtm->hour, 23);
 695                         if (i == 0)
 696                                 i = 12;
 697                         else if (i > 12)
 698                                 i -= 12;
 699                         FMT(' ', 2, "d", (int)i);
 700                         continue;
 701 #endif
 702
 703
 704 #ifdef VMS_EXT
 705                 case 'v':       /* date as dd-bbb-YYYY */
 706                         STRFTIME("%e-%^b-%4Y");
 707                         continue;
 708 #endif
 709
 710
 711 #ifdef POSIX2_DATE
 712                 case 'C':
 713                         FMTV('0', 2, "d", div(vtm->year, INT2FIX(100)));
 714                         continue;
 715
 716                 case 'E':
 717                         /* POSIX locale extensions, ignored for now */
 718                         if (!format[1] || !strchr("cCxXyY", format[1]))
 719                                 goto unknown;
 720                         goto again;
 721                 case 'O':
 722                         /* POSIX locale extensions, ignored for now */
 723                         if (!format[1] || !strchr("deHkIlmMSuUVwWy", format[1]))
 724                                 goto unknown;
 725                         goto again;
 726
 727                 case 'V':       /* week of year according ISO 8601 */
 728                         FMT('0', 2, "d", iso8601wknum_v(vtm));
 729                         continue;
 730
 731                 case 'u':
 732                 /* ISO 8601: Weekday as a decimal number [1 (Monday) - 7] */
 733                         FMT('0', 1, "d", vtm->wday == 0 ? 7 : vtm->wday);
 734                         continue;
 735 #endif  /* POSIX2_DATE */
 736
 737 #ifdef ISO_DATE_EXT
 738                 case 'G':
 739                 case 'g':
 740                         /*
 741                          * Year of ISO week.
 742                          *
 743                          * If it's December but the ISO week number is one,
 744                          * that week is in next year.
 745                          * If it's January but the ISO week number is 52 or
 746                          * 53, that week is in last year.
 747                          * Otherwise, it's this year.
 748                          */
 749                         {
 750                                 VALUE yv = vtm->year;
 751                                 w = iso8601wknum_v(vtm);
 752                                 if (vtm->mon == 12 && w == 1)
 753                                         yv = add(yv, INT2FIX(1));
 754                                 else if (vtm->mon == 1 && w >= 52)
 755                                         yv = sub(yv, INT2FIX(1));
 756
 757                                 if (*format == 'G') {
 758                                         if (FIXNUM_P(yv)) {
 759                                                 const long y = FIX2LONG(yv);
 760                                                 FMT('0', 0 <= y ? 4 : 5, "ld", y);
 761                                         }
 762                                         else {
 763                                                 FMTV('0', 4, "d", yv);
 764                                         }
 765                                 }
 766                                 else {
 767                                         yv = mod(yv, INT2FIX(100));
 768                                         y = FIX2LONG(yv);
 769                                         FMT('0', 2, "ld", y);
 770                                 }
 771                                 continue;
 772                         }
 773
 774 #endif /* ISO_DATE_EXT */
 775
 776
 777                 case 'L':
 778                         w = 3;
 779                         goto subsec;
 780
 781                 case 'N':
 782                         /*
 783                          * fractional second digits. default is 9 digits
 784                          * (nanosecond).
 785                          *
 786                          * %3N  millisecond (3 digits)
 787                          * %6N  microsecond (6 digits)
 788                          * %9N  nanosecond (9 digits)
 789                          */
 790                         w = 9;
 791                 subsec:
 792                         if (precision <= 0) {
 793                             precision = w;
 794                         }
 795                         NEEDS(precision);
 796
 797                         if (ts) {
 798                                 long subsec = ts->tv_nsec;
 799                                 if (9 < precision) {
 800                                         snprintf(s, endp - s, "%09ld", subsec);
 801                                         memset(s+9, '0', precision-9);
 802                                         s += precision;
 803                                 }
 804                                 else {
 805                                         int i;
 806                                         for (i = 0; i < 9-precision; i++)
 807                                                 subsec /= 10;
 808                                         snprintf(s, endp - s, "%0*ld", precision, subsec);
 809                                         s += precision;
 810                                 }
 811                         }
 812                         else {
 813                                 VALUE subsec = mod(timev, INT2FIX(1));
 814                                 int ww;
 815                                 long n;
 816
 817                                 ww = precision;
 818                                 while (9 <= ww) {
 819                                         subsec = mul(subsec, INT2FIX(1000000000));
 820                                         ww -= 9;
 821                                 }
 822                                 n = 1;
 823                                 for (; 0 < ww; ww--)
 824                                         n *= 10;
 825                                 if (n != 1)
 826                                         subsec = mul(subsec, INT2FIX(n));
 827                                 subsec = div(subsec, INT2FIX(1));
 828
 829                                 if (FIXNUM_P(subsec)) {
 830                                         (void)snprintf(s, endp - s, "%0*ld", precision, FIX2LONG(subsec));
 831                                         s += precision;
 832                                 }
 833                                 else {
 834                                         VALUE args[2], result;
 835                                         args[0] = INT2FIX(precision);
 836                                         args[1] = subsec;
 837                                         result = rb_str_format(2, args,
 838                                                       rb_fstring_lit("%0*d"));
 839                                         (void)strlcpy(s, StringValueCStr(result), endp-s);
 840                                         s += precision;
 841                                 }
 842                         }
 843                         continue;
 844
 845                 case 'F':       /*  Equivalent to %Y-%m-%d */
 846                         STRFTIME("%Y-%m-%d");
 847                         continue;
 848
 849                 case '-':
 850                         FLAG_FOUND();
 851                         flags |= BIT_OF(LEFT);
 852                         padding = precision = 0;
 853                         goto again;
 854
 855                 case '^':
 856                         FLAG_FOUND();
 857                         flags |= BIT_OF(UPPER);
 858                         goto again;
 859
 860                 case '#':
 861                         FLAG_FOUND();
 862                         flags |= BIT_OF(CHCASE);
 863                         goto again;
 864
 865                 case '_':
 866                         FLAG_FOUND();
 867                         padding = ' ';
 868                         goto again;
 869
 870                 case ':':
 871                         for (colons = 1; colons <= 3; ++colons) {
 872                                 if (format+colons >= format_end) goto unknown;
 873                                 if (format[colons] == 'z') break;
 874                                 if (format[colons] != ':') goto unknown;
 875                         }
 876                         format += colons - 1;
 877                         goto again;
 878
 879                 case '0':
 880                         padding = '0';
 881                 case '1':  case '2': case '3': case '4':
 882                 case '5': case '6':  case '7': case '8': case '9':
 883                         {
 884                                 size_t n;
 885                                 int ov;
 886                                 unsigned long u = ruby_scan_digits(format, format_end-format, 10, &n, &ov);
 887                                 if (ov || u > INT_MAX) goto unknown;
 888                                 precision = (int)u;
 889                                 format += n - 1;
 890                                 goto again;
 891                         }
 892
 893                 default:
 894                 unknown:
 895                         i = format - sp + 1;
 896                         tp = sp;
 897                         precision = -1;
 898                         flags = 0;
 899                         padding = 0;
 900                         colons = 0;
 901                         break;
 902                 }
 903                 if (i) {
 904                         FILL_PADDING(i);
 905                         memcpy(s, tp, i);
 906                         s = case_conv(s, i, flags);
 907                 }
 908         }
 909         if (format != format_end) {
 910                 return 0;
 911         }
 912         len = s - start;
 913         rb_str_set_len(ftime, len);
 914         rb_str_resize(ftime, len);
 915         return ftime;
 916
 917 err:
 918         return 0;
 919 }
 920
 921 static size_t
 922 strftime_size_limit(size_t format_len)
 923 {
 924         size_t limit = format_len * (1*1024*1024);
 925         if (limit < format_len) limit = format_len;
 926         else if (limit < 1024) limit = 1024;
 927         return limit;
 928 }
 929
 930 VALUE
 931 rb_strftime(const char *format, size_t format_len, rb_encoding *enc,
 932             VALUE time, const struct vtm *vtm, VALUE timev, int gmt)
 933 {
 934         VALUE result = rb_enc_str_new(0, 0, enc);
 935         ENC_CODERANGE_CLEAR(result);
 936         return rb_strftime_with_timespec(result, format, format_len, enc,
 937                                          time, vtm, timev, NULL, gmt,
 938                                          strftime_size_limit(format_len));
 939 }
 940
 941 VALUE
 942 rb_strftime_timespec(const char *format, size_t format_len, rb_encoding *enc,
 943                      VALUE time, const struct vtm *vtm, struct timespec *ts, int gmt)
 944 {
 945         VALUE result = rb_enc_str_new(0, 0, enc);
 946         ENC_CODERANGE_CLEAR(result);
 947         return rb_strftime_with_timespec(result, format, format_len, enc,
 948                                          time, vtm, Qnil, ts, gmt,
 949                                          strftime_size_limit(format_len));
 950 }
 951
 952 #if 0
 953 VALUE
 954 rb_strftime_limit(const char *format, size_t format_len, rb_encoding *enc,
 955                   VALUE time, const struct vtm *vtm, struct timespec *ts,
 956                   int gmt, size_t maxsize)
 957 {
 958         VALUE result = rb_enc_str_new(0, 0, enc);
 959         return rb_strftime_with_timespec(result, format, format_len, enc,
 960                                          time, vtm, Qnil, ts, gmt, maxsize);
 961 }
 962 #endif
 963
 964 /* isleap --- is a year a leap year? */
 965
 966 static int
 967 isleap(long year)
 968 {
 969         return ((year % 4 == 0 && year % 100 != 0) || year % 400 == 0);
 970 }
 971
 972
 973 static void
 974 vtm2tm_noyear(const struct vtm *vtm, struct tm *result)
 975 {
 976     struct tm tm;
 977
 978     /* for isleap() in iso8601wknum.  +100 is -1900 (mod 400). */
 979     tm.tm_year = FIX2INT(mod(vtm->year, INT2FIX(400))) + 100;
 980
 981     tm.tm_mon = vtm->mon-1;
 982     tm.tm_mday = vtm->mday;
 983     tm.tm_hour = vtm->hour;
 984     tm.tm_min = vtm->min;
 985     tm.tm_sec = vtm->sec;
 986     tm.tm_wday = vtm->wday;
 987     tm.tm_yday = vtm->yday-1;
 988     tm.tm_isdst = vtm->isdst;
 989 #if defined(HAVE_STRUCT_TM_TM_GMTOFF)
 990     tm.tm_gmtoff = 0;
 991 #endif
 992 #if defined(HAVE_TM_ZONE)
 993     tm.tm_zone = NULL;
 994 #endif
 995     *result = tm;
 996 }
 997
 998 #ifdef POSIX2_DATE
 999 /* iso8601wknum --- compute week number according to ISO 8601 */
1000
1001 static int
1002 iso8601wknum(const struct tm *timeptr)
1003 {
1004         /*
1005          * From 1003.2:
1006          *      If the week (Monday to Sunday) containing January 1
1007          *      has four or more days in the new year, then it is week 1;
1008          *      otherwise it is the highest numbered week of the previous
1009          *      year (52 or 53), and the next week is week 1.
1010          *
1011          * ADR: This means if Jan 1 was Monday through Thursday,
1012          *      it was week 1, otherwise week 52 or 53.
1013          *
1014          * XPG4 erroneously included POSIX.2 rationale text in the
1015          * main body of the standard. Thus it requires week 53.
1016          */
1017
1018         int weeknum, jan1day;
1019
1020         /* get week number, Monday as first day of the week */
1021         weeknum = weeknumber(timeptr, 1);
1022
1023         /*
1024          * With thanks and tip of the hatlo to tml@tik.vtt.fi
1025          *
1026          * What day of the week does January 1 fall on?
1027          * We know that
1028          *      (timeptr->tm_yday - jan1.tm_yday) MOD 7 ==
1029          *              (timeptr->tm_wday - jan1.tm_wday) MOD 7
1030          * and that
1031          *      jan1.tm_yday == 0
1032          * and that
1033          *      timeptr->tm_wday MOD 7 == timeptr->tm_wday
1034          * from which it follows that. . .
1035          */
1036         jan1day = timeptr->tm_wday - (timeptr->tm_yday % 7);
1037         if (jan1day < 0)
1038                 jan1day += 7;
1039
1040         /*
1041          * If Jan 1 was a Monday through Thursday, it was in
1042          * week 1.  Otherwise it was last year's highest week, which is
1043          * this year's week 0.
1044          *
1045          * What does that mean?
1046          * If Jan 1 was Monday, the week number is exactly right, it can
1047          *      never be 0.
1048          * If it was Tuesday through Thursday, the weeknumber is one
1049          *      less than it should be, so we add one.
1050          * Otherwise, Friday, Saturday or Sunday, the week number is
1051          * OK, but if it is 0, it needs to be 52 or 53.
1052          */
1053         switch (jan1day) {
1054         case 1:         /* Monday */
1055                 break;
1056         case 2:         /* Tuesday */
1057         case 3:         /* Wednesday */
1058         case 4:         /* Thursday */
1059                 weeknum++;
1060                 break;
1061         case 5:         /* Friday */
1062         case 6:         /* Saturday */
1063         case 0:         /* Sunday */
1064                 if (weeknum == 0) {
1065 #ifdef USE_BROKEN_XPG4
1066                         /* XPG4 (as of March 1994) says 53 unconditionally */
1067                         weeknum = 53;
1068 #else
1069                         /* get week number of last week of last year */
1070                         struct tm dec31ly;      /* 12/31 last year */
1071                         dec31ly = *timeptr;
1072                         dec31ly.tm_year--;
1073                         dec31ly.tm_mon = 11;
1074                         dec31ly.tm_mday = 31;
1075                         dec31ly.tm_wday = (jan1day == 0) ? 6 : jan1day - 1;
1076                         dec31ly.tm_yday = 364 + isleap(dec31ly.tm_year + 1900L);
1077                         weeknum = iso8601wknum(& dec31ly);
1078 #endif
1079                 }
1080                 break;
1081         }
1082
1083         if (timeptr->tm_mon == 11) {
1084                 /*
1085                  * The last week of the year
1086                  * can be in week 1 of next year.
1087                  * Sigh.
1088                  *
1089                  * This can only happen if
1090                  *      M   T  W
1091                  *      29  30 31
1092                  *      30  31
1093                  *      31
1094                  */
1095                 int wday, mday;
1096
1097                 wday = timeptr->tm_wday;
1098                 mday = timeptr->tm_mday;
1099                 if (   (wday == 1 && (mday >= 29 && mday <= 31))
1100                     || (wday == 2 && (mday == 30 || mday == 31))
1101                     || (wday == 3 &&  mday == 31))
1102                         weeknum = 1;
1103         }
1104
1105         return weeknum;
1106 }
1107
1108 static int
1109 iso8601wknum_v(const struct vtm *vtm)
1110 {
1111         struct tm tm;
1112         vtm2tm_noyear(vtm, &tm);
1113         return iso8601wknum(&tm);
1114 }
1115
1116 #endif
1117
1118 /* weeknumber --- figure how many weeks into the year */
1119
1120 /* With thanks and tip of the hatlo to ado@elsie.nci.nih.gov */
1121
1122 static int
1123 weeknumber(const struct tm *timeptr, int firstweekday)
1124 {
1125         int wday = timeptr->tm_wday;
1126         int ret;
1127
1128         if (firstweekday == 1) {
1129                 if (wday == 0)  /* sunday */
1130                         wday = 6;
1131                 else
1132                         wday--;
1133         }
1134         ret = ((timeptr->tm_yday + 7 - wday) / 7);
1135         if (ret < 0)
1136                 ret = 0;
1137         return ret;
1138 }
1139
1140 static int
1141 weeknumber_v(const struct vtm *vtm, int firstweekday)
1142 {
1143         struct tm tm;
1144         vtm2tm_noyear(vtm, &tm);
1145         return weeknumber(&tm, firstweekday);
1146 }
1147
1148 #if 0
1149 /* ADR --- I'm loathe to mess with ado's code ... */
1150
1151 Date:         Wed, 24 Apr 91 20:54:08 MDT
1152 From: Michal Jaegermann <audfax!emory!vm.ucs.UAlberta.CA!NTOMCZAK>
1153 To: arnold@audiofax.com
1154
1155 Hi Arnold,
1156 in a process of fixing of strftime() in libraries on Atari ST I grabbed
1157 some pieces of code from your own strftime.  When doing that it came
1158 to mind that your weeknumber() function compiles a little bit nicer
1159 in the following form:
1160 /*
1161  * firstweekday is 0 if starting in Sunday, non-zero if in Monday
1162  */
1163 {
1164     return (timeptr->tm_yday - timeptr->tm_wday +
1165             (firstweekday ? (timeptr->tm_wday ? 8 : 1) : 7)) / 7;
1166 }
1167 How nicer it depends on a compiler, of course, but always a tiny bit.
1168
1169    Cheers,
1170    Michal
1171    ntomczak@vm.ucs.ualberta.ca
1172 #endif
1173
1174 #ifdef  TEST_STRFTIME
1175
1176 /*
1177  * NAME:
1178  *      tst
1179  *
1180  * SYNOPSIS:
1181  *      tst
1182  *
1183  * DESCRIPTION:
1184  *      "tst" is a test driver for the function "strftime".
1185  *
1186  * OPTIONS:
1187  *      None.
1188  *
1189  * AUTHOR:
1190  *      Karl Vogel
1191  *      Control Data Systems, Inc.
1192  *      vogelke@c-17igp.wpafb.af.mil
1193  *
1194  * BUGS:
1195  *      None noticed yet.
1196  *
1197  * COMPILE:
1198  *      cc -o tst -DTEST_STRFTIME strftime.c
1199  */
1200
1201 /* ADR: I reformatted this to my liking, and deleted some unneeded code. */
1202
1203 #ifndef NULL
1204 #include        <stdio.h>
1205 #endif
1206 #include        <sys/time.h>
1207 #include        <string.h>
1208
1209 #define         MAXTIME         132
1210
1211 /*
1212  * Array of time formats.
1213  */
1214
1215 static char *array[] =
1216 {
1217         "(%%A)      full weekday name, var length (Sunday..Saturday)  %A",
1218         "(%%B)       full month name, var length (January..December)  %B",
1219         "(%%C)                                               Century  %C",
1220         "(%%D)                                       date (%%m/%%d/%%y)  %D",
1221         "(%%E)                           Locale extensions (ignored)  %E",
1222         "(%%H)                          hour (24-hour clock, 00..23)  %H",
1223         "(%%I)                          hour (12-hour clock, 01..12)  %I",
1224         "(%%M)                                       minute (00..59)  %M",
1225         "(%%O)                           Locale extensions (ignored)  %O",
1226         "(%%R)                                 time, 24-hour (%%H:%%M)  %R",
1227         "(%%S)                                       second (00..60)  %S",
1228         "(%%T)                              time, 24-hour (%%H:%%M:%%S)  %T",
1229         "(%%U)    week of year, Sunday as first day of week (00..53)  %U",
1230         "(%%V)                    week of year according to ISO 8601  %V",
1231         "(%%W)    week of year, Monday as first day of week (00..53)  %W",
1232         "(%%X)     appropriate locale time representation (%H:%M:%S)  %X",
1233         "(%%Y)                           year with century (1970...)  %Y",
1234         "(%%Z) timezone (EDT), or blank if timezone not determinable  %Z",
1235         "(%%a)          locale's abbreviated weekday name (Sun..Sat)  %a",
1236         "(%%b)            locale's abbreviated month name (Jan..Dec)  %b",
1237         "(%%c)           full date (Sat Nov  4 12:02:33 1989)%n%t%t%t  %c",
1238         "(%%d)                             day of the month (01..31)  %d",
1239         "(%%e)               day of the month, blank-padded ( 1..31)  %e",
1240         "(%%h)                                should be same as (%%b)  %h",
1241         "(%%j)                            day of the year (001..366)  %j",
1242         "(%%k)               hour, 24-hour clock, blank pad ( 0..23)  %k",
1243         "(%%l)               hour, 12-hour clock, blank pad ( 1..12)  %l",
1244         "(%%m)                                        month (01..12)  %m",
1245         "(%%p)              locale's AM or PM based on 12-hour clock  %p",
1246         "(%%r)                   time, 12-hour (same as %%I:%%M:%%S %%p)  %r",
1247         "(%%u) ISO 8601: Weekday as decimal number [1 (Monday) - 7]   %u",
1248         "(%%v)                                VMS date (dd-bbb-YYYY)  %v",
1249         "(%%w)                       day of week (0..6, Sunday == 0)  %w",
1250         "(%%x)                appropriate locale date representation  %x",
1251         "(%%y)                      last two digits of year (00..99)  %y",
1252         "(%%z)      timezone offset east of GMT as HHMM (e.g. -0500)  %z",
1253         (char *) NULL
1254 };
1255
1256 /* main routine. */
1257
1258 int
1259 main(int argc, char **argv)
1260 {
1261         long time();
1262
1263         char *next;
1264         char string[MAXTIME];
1265
1266         int k;
1267         int length;
1268
1269         struct tm *tm;
1270
1271         long clock;
1272
1273         /* Call the function. */
1274
1275         clock = time((long *) 0);
1276         tm = localtime(&clock);
1277
1278         for (k = 0; next = array[k]; k++) {
1279                 length = strftime(string, MAXTIME, next, tm);
1280                 printf("%s\n", string);
1281         }
1282
1283         exit(0);
1284 }
1285 #endif  /* TEST_STRFTIME */