lib/nl_langinfo.c

   1 /* nl_langinfo() replacement: query locale dependent information.
   2
   3    Copyright (C) 2007-2020 Free Software Foundation, Inc.
   4
   5    This program is free software: you can redistribute it and/or modify
   6    it under the terms of the GNU General Public License as published by
   7    the Free Software Foundation; either version 3 of the License, or
   8    (at your option) any later version.
   9
  10    This program is distributed in the hope that it will be useful,
  11    but WITHOUT ANY WARRANTY; without even the implied warranty of
  12    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13    GNU General Public License for more details.
  14
  15    You should have received a copy of the GNU General Public License
  16    along with this program.  If not, see <https://www.gnu.org/licenses/>.  */
  17
  18 #include <config.h>
  19
  20 /* Specification.  */
  21 #include <langinfo.h>
  22
  23 #include <locale.h>
  24 #include <string.h>
  25 #if defined _WIN32 && ! defined __CYGWIN__
  26 # define WIN32_LEAN_AND_MEAN  /* avoid including junk */
  27 # include <windows.h>
  28 # include <stdio.h>
  29 #endif
  30
  31 /* nl_langinfo() must be multithread-safe.  To achieve this without using
  32    thread-local storage:
  33      1. We use a specific static buffer for each possible argument.
  34         So that different threads can call nl_langinfo with different arguments,
  35         without interfering.
  36      2. We use a simple strcpy or memcpy to fill this static buffer.  Filling it
  37         through, for example, strcpy + strcat would not be guaranteed to leave
  38         the buffer's contents intact if another thread is currently accessing
  39         it.  If necessary, the contents is first assembled in a stack-allocated
  40         buffer.  */
  41
  42 #if !REPLACE_NL_LANGINFO || GNULIB_defined_CODESET
  43 /* Return the codeset of the current locale, if this is easily deducible.
  44    Otherwise, return "".  */
  45 static char *
  46 ctype_codeset (void)
  47 {
  48   static char result[2 + 10 + 1];
  49   char buf[2 + 10 + 1];
  50   char locale[SETLOCALE_NULL_MAX];
  51   char *codeset;
  52   size_t codesetlen;
  53
  54   if (setlocale_null_r (LC_CTYPE, locale, sizeof (locale)))
  55     locale[0] = '\0';
  56
  57   codeset = buf;
  58   codeset[0] = '\0';
  59
  60   if (locale && locale[0])
  61     {
  62       /* If the locale name contains an encoding after the dot, return it.  */
  63       char *dot = strchr (locale, '.');
  64
  65       if (dot)
  66         {
  67           /* Look for the possible @... trailer and remove it, if any.  */
  68           char *codeset_start = dot + 1;
  69           char const *modifier = strchr (codeset_start, '@');
  70
  71           if (! modifier)
  72             codeset = codeset_start;
  73           else
  74             {
  75               codesetlen = modifier - codeset_start;
  76               if (codesetlen < sizeof buf)
  77                 {
  78                   codeset = memcpy (buf, codeset_start, codesetlen);
  79                   codeset[codesetlen] = '\0';
  80                 }
  81             }
  82         }
  83     }
  84
  85 # if defined _WIN32 && ! defined __CYGWIN__
  86   /* If setlocale is successful, it returns the number of the
  87      codepage, as a string.  Otherwise, fall back on Windows API
  88      GetACP, which returns the locale's codepage as a number (although
  89      this doesn't change according to what the 'setlocale' call specified).
  90      Either way, prepend "CP" to make it a valid codeset name.  */
  91   codesetlen = strlen (codeset);
  92   if (0 < codesetlen && codesetlen < sizeof buf - 2)
  93     memmove (buf + 2, codeset, codesetlen + 1);
  94   else
  95     sprintf (buf + 2, "%u", GetACP ());
  96   /* For a locale name such as "French_France.65001", in Windows 10,
  97      setlocale now returns "French_France.utf8" instead.  */
  98   if (strcmp (buf + 2, "65001") == 0 || strcmp (buf + 2, "utf8") == 0)
  99     return (char *) "UTF-8";
 100   else
 101     {
 102       memcpy (buf, "CP", 2);
 103       strcpy (result, buf);
 104       return result;
 105     }
 106 # else
 107   strcpy (result, codeset);
 108   return result;
 109 #endif
 110 }
 111 #endif
 112
 113
 114 #if REPLACE_NL_LANGINFO
 115
 116 /* Override nl_langinfo with support for added nl_item values.  */
 117
 118 # undef nl_langinfo
 119
 120 char *
 121 rpl_nl_langinfo (nl_item item)
 122 {
 123   switch (item)
 124     {
 125 # if GNULIB_defined_CODESET
 126     case CODESET:
 127       return ctype_codeset ();
 128 # endif
 129 # if GNULIB_defined_T_FMT_AMPM
 130     case T_FMT_AMPM:
 131       return (char *) "%I:%M:%S %p";
 132 # endif
 133 # if GNULIB_defined_ALTMON
 134     case ALTMON_1:
 135     case ALTMON_2:
 136     case ALTMON_3:
 137     case ALTMON_4:
 138     case ALTMON_5:
 139     case ALTMON_6:
 140     case ALTMON_7:
 141     case ALTMON_8:
 142     case ALTMON_9:
 143     case ALTMON_10:
 144     case ALTMON_11:
 145     case ALTMON_12:
 146       /* We don't ship the appropriate localizations with gnulib.  Therefore,
 147          treat ALTMON_i like MON_i.  */
 148       item = item - ALTMON_1 + MON_1;
 149       break;
 150 # endif
 151 # if GNULIB_defined_ERA
 152     case ERA:
 153       /* The format is not standardized.  In glibc it is a sequence of strings
 154          of the form "direction:offset:start_date:end_date:era_name:era_format"
 155          with an empty string at the end.  */
 156       return (char *) "";
 157     case ERA_D_FMT:
 158       /* The %Ex conversion in strftime behaves like %x if the locale does not
 159          have an alternative time format.  */
 160       item = D_FMT;
 161       break;
 162     case ERA_D_T_FMT:
 163       /* The %Ec conversion in strftime behaves like %c if the locale does not
 164          have an alternative time format.  */
 165       item = D_T_FMT;
 166       break;
 167     case ERA_T_FMT:
 168       /* The %EX conversion in strftime behaves like %X if the locale does not
 169          have an alternative time format.  */
 170       item = T_FMT;
 171       break;
 172     case ALT_DIGITS:
 173       /* The format is not standardized.  In glibc it is a sequence of 10
 174          strings, appended in memory.  */
 175       return (char *) "\0\0\0\0\0\0\0\0\0\0";
 176 # endif
 177 # if GNULIB_defined_YESEXPR || !FUNC_NL_LANGINFO_YESEXPR_WORKS
 178     case YESEXPR:
 179       return (char *) "^[yY]";
 180     case NOEXPR:
 181       return (char *) "^[nN]";
 182 # endif
 183     default:
 184       break;
 185     }
 186   return nl_langinfo (item);
 187 }
 188
 189 #else
 190
 191 /* Provide nl_langinfo from scratch, either for native MS-Windows, or
 192    for old Unix platforms without locales, such as Linux libc5 or
 193    BeOS.  */
 194
 195 # include <time.h>
 196
 197 char *
 198 nl_langinfo (nl_item item)
 199 {
 200   char buf[100];
 201   struct tm tmm = { 0 };
 202
 203   switch (item)
 204     {
 205     /* nl_langinfo items of the LC_CTYPE category */
 206     case CODESET:
 207       {
 208         char *codeset = ctype_codeset ();
 209         if (*codeset)
 210           return codeset;
 211       }
 212 # ifdef __BEOS__
 213       return (char *) "UTF-8";
 214 # else
 215       return (char *) "ISO-8859-1";
 216 # endif
 217     /* nl_langinfo items of the LC_NUMERIC category */
 218     case RADIXCHAR:
 219       return localeconv () ->decimal_point;
 220     case THOUSEP:
 221       return localeconv () ->thousands_sep;
 222 # ifdef GROUPING
 223     case GROUPING:
 224       return localeconv () ->grouping;
 225 # endif
 226     /* nl_langinfo items of the LC_TIME category.
 227        TODO: Really use the locale.  */
 228     case D_T_FMT:
 229     case ERA_D_T_FMT:
 230       return (char *) "%a %b %e %H:%M:%S %Y";
 231     case D_FMT:
 232     case ERA_D_FMT:
 233       return (char *) "%m/%d/%y";
 234     case T_FMT:
 235     case ERA_T_FMT:
 236       return (char *) "%H:%M:%S";
 237     case T_FMT_AMPM:
 238       return (char *) "%I:%M:%S %p";
 239     case AM_STR:
 240       {
 241         static char result[80];
 242         if (!strftime (buf, sizeof result, "%p", &tmm))
 243           return (char *) "AM";
 244         strcpy (result, buf);
 245         return result;
 246       }
 247     case PM_STR:
 248       {
 249         static char result[80];
 250         tmm.tm_hour = 12;
 251         if (!strftime (buf, sizeof result, "%p", &tmm))
 252           return (char *) "PM";
 253         strcpy (result, buf);
 254         return result;
 255       }
 256     case DAY_1:
 257     case DAY_2:
 258     case DAY_3:
 259     case DAY_4:
 260     case DAY_5:
 261     case DAY_6:
 262     case DAY_7:
 263       {
 264         static char result[7][50];
 265         static char const days[][sizeof "Wednesday"] = {
 266           "Sunday", "Monday", "Tuesday", "Wednesday", "Thursday",
 267           "Friday", "Saturday"
 268         };
 269         tmm.tm_wday = item - DAY_1;
 270         if (!strftime (buf, sizeof result[0], "%A", &tmm))
 271           return (char *) days[item - DAY_1];
 272         strcpy (result[item - DAY_1], buf);
 273         return result[item - DAY_1];
 274       }
 275     case ABDAY_1:
 276     case ABDAY_2:
 277     case ABDAY_3:
 278     case ABDAY_4:
 279     case ABDAY_5:
 280     case ABDAY_6:
 281     case ABDAY_7:
 282       {
 283         static char result[7][30];
 284         static char const abdays[][sizeof "Sun"] = {
 285           "Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"
 286         };
 287         tmm.tm_wday = item - ABDAY_1;
 288         if (!strftime (buf, sizeof result[0], "%a", &tmm))
 289           return (char *) abdays[item - ABDAY_1];
 290         strcpy (result[item - ABDAY_1], buf);
 291         return result[item - ABDAY_1];
 292       }
 293     {
 294       static char const months[][sizeof "September"] = {
 295         "January", "February", "March", "April", "May", "June", "July",
 296         "September", "October", "November", "December"
 297       };
 298       case MON_1:
 299       case MON_2:
 300       case MON_3:
 301       case MON_4:
 302       case MON_5:
 303       case MON_6:
 304       case MON_7:
 305       case MON_8:
 306       case MON_9:
 307       case MON_10:
 308       case MON_11:
 309       case MON_12:
 310         {
 311           static char result[12][50];
 312           tmm.tm_mon = item - MON_1;
 313           if (!strftime (buf, sizeof result[0], "%B", &tmm))
 314             return (char *) months[item - MON_1];
 315           strcpy (result[item - MON_1], buf);
 316           return result[item - MON_1];
 317         }
 318       case ALTMON_1:
 319       case ALTMON_2:
 320       case ALTMON_3:
 321       case ALTMON_4:
 322       case ALTMON_5:
 323       case ALTMON_6:
 324       case ALTMON_7:
 325       case ALTMON_8:
 326       case ALTMON_9:
 327       case ALTMON_10:
 328       case ALTMON_11:
 329       case ALTMON_12:
 330         {
 331           static char result[12][50];
 332           tmm.tm_mon = item - ALTMON_1;
 333           /* The platforms without nl_langinfo() don't support strftime with
 334              %OB.  We don't even need to try.  */
 335           #if 0
 336           if (!strftime (buf, sizeof result[0], "%OB", &tmm))
 337           #endif
 338             if (!strftime (buf, sizeof result[0], "%B", &tmm))
 339               return (char *) months[item - ALTMON_1];
 340           strcpy (result[item - ALTMON_1], buf);
 341           return result[item - ALTMON_1];
 342         }
 343     }
 344     case ABMON_1:
 345     case ABMON_2:
 346     case ABMON_3:
 347     case ABMON_4:
 348     case ABMON_5:
 349     case ABMON_6:
 350     case ABMON_7:
 351     case ABMON_8:
 352     case ABMON_9:
 353     case ABMON_10:
 354     case ABMON_11:
 355     case ABMON_12:
 356       {
 357         static char result[12][30];
 358         static char const abmonths[][sizeof "Jan"] = {
 359           "Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul",
 360           "Sep", "Oct", "Nov", "Dec"
 361         };
 362         tmm.tm_mon = item - ABMON_1;
 363         if (!strftime (buf, sizeof result[0], "%b", &tmm))
 364           return (char *) abmonths[item - ABMON_1];
 365         strcpy (result[item - ABMON_1], buf);
 366         return result[item - ABMON_1];
 367       }
 368     case ERA:
 369       return (char *) "";
 370     case ALT_DIGITS:
 371       return (char *) "\0\0\0\0\0\0\0\0\0\0";
 372     /* nl_langinfo items of the LC_MONETARY category.  */
 373     case CRNCYSTR:
 374       return localeconv () ->currency_symbol;
 375 # ifdef INT_CURR_SYMBOL
 376     case INT_CURR_SYMBOL:
 377       return localeconv () ->int_curr_symbol;
 378     case MON_DECIMAL_POINT:
 379       return localeconv () ->mon_decimal_point;
 380     case MON_THOUSANDS_SEP:
 381       return localeconv () ->mon_thousands_sep;
 382     case MON_GROUPING:
 383       return localeconv () ->mon_grouping;
 384     case POSITIVE_SIGN:
 385       return localeconv () ->positive_sign;
 386     case NEGATIVE_SIGN:
 387       return localeconv () ->negative_sign;
 388     case FRAC_DIGITS:
 389       return & localeconv () ->frac_digits;
 390     case INT_FRAC_DIGITS:
 391       return & localeconv () ->int_frac_digits;
 392     case P_CS_PRECEDES:
 393       return & localeconv () ->p_cs_precedes;
 394     case N_CS_PRECEDES:
 395       return & localeconv () ->n_cs_precedes;
 396     case P_SEP_BY_SPACE:
 397       return & localeconv () ->p_sep_by_space;
 398     case N_SEP_BY_SPACE:
 399       return & localeconv () ->n_sep_by_space;
 400     case P_SIGN_POSN:
 401       return & localeconv () ->p_sign_posn;
 402     case N_SIGN_POSN:
 403       return & localeconv () ->n_sign_posn;
 404 # endif
 405     /* nl_langinfo items of the LC_MESSAGES category
 406        TODO: Really use the locale. */
 407     case YESEXPR:
 408       return (char *) "^[yY]";
 409     case NOEXPR:
 410       return (char *) "^[nN]";
 411     default:
 412       return (char *) "";
 413     }
 414 }
 415
 416 #endif