locale.c

   1 /* locale.c - Miscellaneous internationalization functions. */
   2
   3 /* Copyright (C) 1996-2009,2012,2016,2020 Free Software Foundation, Inc.
   4
   5    This file is part of GNU Bash, the Bourne Again SHell.
   6
   7    Bash is free software: you can redistribute it and/or modify
   8    it under the terms of the GNU General Public License as published by
   9    the Free Software Foundation, either version 3 of the License, or
  10    (at your option) any later version.
  11
  12    Bash is distributed in the hope that it will be useful,
  13    but WITHOUT ANY WARRANTY; without even the implied warranty of
  14    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15    GNU General Public License for more details.
  16
  17    You should have received a copy of the GNU General Public License
  18    along with Bash.  If not, see <http://www.gnu.org/licenses/>.
  19 */
  20
  21 #include "config.h"
  22
  23 #include "bashtypes.h"
  24
  25 #if defined (HAVE_UNISTD_H)
  26 #  include <unistd.h>
  27 #endif
  28
  29 #if HAVE_LANGINFO_CODESET
  30 #  include <langinfo.h>
  31 #endif
  32
  33 #include "bashintl.h"
  34 #include "bashansi.h"
  35 #include <stdio.h>
  36 #include "chartypes.h"
  37 #include <errno.h>
  38
  39 #include "shell.h"
  40 #include "input.h"      /* For bash_input */
  41
  42 #ifndef errno
  43 extern int errno;
  44 #endif
  45
  46 int locale_utf8locale;
  47 int locale_mb_cur_max;  /* value of MB_CUR_MAX for current locale (LC_CTYPE) */
  48 int locale_shiftstates = 0;
  49
  50 extern int dump_translatable_strings, dump_po_strings;
  51
  52 /* The current locale when the program begins */
  53 static char *default_locale;
  54
  55 /* The current domain for textdomain(3). */
  56 static char *default_domain;
  57 static char *default_dir;
  58
  59 /* tracks the value of LC_ALL; used to override values for other locale
  60    categories */
  61 static char *lc_all;
  62
  63 /* tracks the value of LC_ALL; used to provide defaults for locale
  64    categories */
  65 static char *lang;
  66
  67 /* Called to reset all of the locale variables to their appropriate values
  68    if (and only if) LC_ALL has not been assigned a value. */
  69 static int reset_locale_vars PARAMS((void));
  70
  71 static void locale_setblanks PARAMS((void));
  72 static int locale_isutf8 PARAMS((char *));
  73
  74 /* Set the value of default_locale and make the current locale the
  75    system default locale.  This should be called very early in main(). */
  76 void
  77 set_default_locale ()
  78 {
  79 #if defined (HAVE_SETLOCALE)
  80   default_locale = setlocale (LC_ALL, "");
  81   if (default_locale)
  82     default_locale = savestring (default_locale);
  83 #else
  84   default_locale = savestring ("C");
  85 #endif /* HAVE_SETLOCALE */
  86   bindtextdomain (PACKAGE, LOCALEDIR);
  87   textdomain (PACKAGE);
  88
  89   locale_mb_cur_max = MB_CUR_MAX;
  90   locale_utf8locale = locale_isutf8 (default_locale);
  91 #if defined (HANDLE_MULTIBYTE)
  92   locale_shiftstates = mblen ((char *)NULL, 0);
  93 #else
  94   local_shiftstates = 0;
  95 #endif
  96 }
  97
  98 /* Set default values for LC_CTYPE, LC_COLLATE, LC_MESSAGES, LC_NUMERIC and
  99    LC_TIME if they are not specified in the environment, but LC_ALL is.  This
 100    should be called from main() after parsing the environment. */
 101 void
 102 set_default_locale_vars ()
 103 {
 104   char *val;
 105
 106 #if defined (HAVE_SETLOCALE)
 107
 108 #  if defined (LC_CTYPE)
 109   val = get_string_value ("LC_CTYPE");
 110   if (val == 0 && lc_all && *lc_all)
 111     {
 112       setlocale (LC_CTYPE, lc_all);
 113       locale_setblanks ();
 114       locale_mb_cur_max = MB_CUR_MAX;
 115       locale_utf8locale = locale_isutf8 (lc_all);
 116
 117 #    if defined (HANDLE_MULTIBYTE)
 118       locale_shiftstates = mblen ((char *)NULL, 0);
 119 #    else
 120       local_shiftstates = 0;
 121 #    endif
 122
 123       u32reset ();
 124     }
 125 #  endif
 126
 127 #  if defined (LC_COLLATE)
 128   val = get_string_value ("LC_COLLATE");
 129   if (val == 0 && lc_all && *lc_all)
 130     setlocale (LC_COLLATE, lc_all);
 131 #  endif /* LC_COLLATE */
 132
 133 #  if defined (LC_MESSAGES)
 134   val = get_string_value ("LC_MESSAGES");
 135   if (val == 0 && lc_all && *lc_all)
 136     setlocale (LC_MESSAGES, lc_all);
 137 #  endif /* LC_MESSAGES */
 138
 139 #  if defined (LC_NUMERIC)
 140   val = get_string_value ("LC_NUMERIC");
 141   if (val == 0 && lc_all && *lc_all)
 142     setlocale (LC_NUMERIC, lc_all);
 143 #  endif /* LC_NUMERIC */
 144
 145 #  if defined (LC_TIME)
 146   val = get_string_value ("LC_TIME");
 147   if (val == 0 && lc_all && *lc_all)
 148     setlocale (LC_TIME, lc_all);
 149 #  endif /* LC_TIME */
 150
 151 #endif /* HAVE_SETLOCALE */
 152
 153   val = get_string_value ("TEXTDOMAIN");
 154   if (val && *val)
 155     {
 156       FREE (default_domain);
 157       default_domain = savestring (val);
 158       if (default_dir && *default_dir)
 159         bindtextdomain (default_domain, default_dir);
 160     }
 161
 162   val = get_string_value ("TEXTDOMAINDIR");
 163   if (val && *val)
 164     {
 165       FREE (default_dir);
 166       default_dir = savestring (val);
 167       if (default_domain && *default_domain)
 168         bindtextdomain (default_domain, default_dir);
 169     }
 170 }
 171
 172 /* Set one of the locale categories (specified by VAR) to VALUE.  Returns 1
 173   if successful, 0 otherwise. */
 174 int
 175 set_locale_var (var, value)
 176      char *var, *value;
 177 {
 178   int r;
 179   char *x;
 180
 181   x = "";
 182   errno = 0;
 183   if (var[0] == 'T' && var[10] == 0)            /* TEXTDOMAIN */
 184     {
 185       FREE (default_domain);
 186       default_domain = value ? savestring (value) : (char *)NULL;
 187       if (default_dir && *default_dir)
 188         bindtextdomain (default_domain, default_dir);
 189       return (1);
 190     }
 191   else if (var[0] == 'T')                       /* TEXTDOMAINDIR */
 192     {
 193       FREE (default_dir);
 194       default_dir = value ? savestring (value) : (char *)NULL;
 195       if (default_domain && *default_domain)
 196         bindtextdomain (default_domain, default_dir);
 197       return (1);
 198     }
 199
 200   /* var[0] == 'L' && var[1] == 'C' && var[2] == '_' */
 201
 202   else if (var[3] == 'A')                       /* LC_ALL */
 203     {
 204       FREE (lc_all);
 205       if (value)
 206         lc_all = savestring (value);
 207       else
 208         {
 209           lc_all = (char *)xmalloc (1);
 210           lc_all[0] = '\0';
 211         }
 212 #if defined (HAVE_SETLOCALE)
 213       r = *lc_all ? ((x = setlocale (LC_ALL, lc_all)) != 0) : reset_locale_vars ();
 214       if (x == 0)
 215         {
 216           if (errno == 0)
 217             internal_warning(_("setlocale: LC_ALL: cannot change locale (%s)"), lc_all);
 218           else
 219             internal_warning(_("setlocale: LC_ALL: cannot change locale (%s): %s"), lc_all, strerror (errno));
 220         }
 221       locale_setblanks ();
 222       locale_mb_cur_max = MB_CUR_MAX;
 223       /* if LC_ALL == "", reset_locale_vars has already called this */
 224       if (*lc_all && x)
 225         locale_utf8locale = locale_isutf8 (lc_all);
 226 #  if defined (HANDLE_MULTIBYTE)
 227       locale_shiftstates = mblen ((char *)NULL, 0);
 228 #  else
 229       local_shiftstates = 0;
 230 #  endif
 231       u32reset ();
 232       return r;
 233 #else
 234       return (1);
 235 #endif
 236     }
 237
 238 #if defined (HAVE_SETLOCALE)
 239   else if (var[3] == 'C' && var[4] == 'T')      /* LC_CTYPE */
 240     {
 241 #  if defined (LC_CTYPE)
 242       if (lc_all == 0 || *lc_all == '\0')
 243         {
 244           x = setlocale (LC_CTYPE, get_locale_var ("LC_CTYPE"));
 245           locale_setblanks ();
 246           locale_mb_cur_max = MB_CUR_MAX;
 247           /* if setlocale() returns NULL, the locale is not changed */
 248           if (x)
 249             locale_utf8locale = locale_isutf8 (x);
 250 #if defined (HANDLE_MULTIBYTE)
 251           locale_shiftstates = mblen ((char *)NULL, 0);
 252 #else
 253           local_shiftstates = 0;
 254 #endif
 255           u32reset ();
 256         }
 257 #  endif
 258     }
 259   else if (var[3] == 'C' && var[4] == 'O')      /* LC_COLLATE */
 260     {
 261 #  if defined (LC_COLLATE)
 262       if (lc_all == 0 || *lc_all == '\0')
 263         x = setlocale (LC_COLLATE, get_locale_var ("LC_COLLATE"));
 264 #  endif /* LC_COLLATE */
 265     }
 266   else if (var[3] == 'M' && var[4] == 'E')      /* LC_MESSAGES */
 267     {
 268 #  if defined (LC_MESSAGES)
 269       if (lc_all == 0 || *lc_all == '\0')
 270         x = setlocale (LC_MESSAGES, get_locale_var ("LC_MESSAGES"));
 271 #  endif /* LC_MESSAGES */
 272     }
 273   else if (var[3] == 'N' && var[4] == 'U')      /* LC_NUMERIC */
 274     {
 275 #  if defined (LC_NUMERIC)
 276       if (lc_all == 0 || *lc_all == '\0')
 277         x = setlocale (LC_NUMERIC, get_locale_var ("LC_NUMERIC"));
 278 #  endif /* LC_NUMERIC */
 279     }
 280   else if (var[3] == 'T' && var[4] == 'I')      /* LC_TIME */
 281     {
 282 #  if defined (LC_TIME)
 283       if (lc_all == 0 || *lc_all == '\0')
 284         x = setlocale (LC_TIME, get_locale_var ("LC_TIME"));
 285 #  endif /* LC_TIME */
 286     }
 287 #endif /* HAVE_SETLOCALE */
 288
 289   if (x == 0)
 290     {
 291       if (errno == 0)
 292         internal_warning(_("setlocale: %s: cannot change locale (%s)"), var, get_locale_var (var));
 293       else
 294         internal_warning(_("setlocale: %s: cannot change locale (%s): %s"), var, get_locale_var (var), strerror (errno));
 295     }
 296
 297   return (x != 0);
 298 }
 299
 300 /* Called when LANG is assigned a value.  Tracks value in `lang'.  Calls
 301    reset_locale_vars() to reset any default values if LC_ALL is unset or
 302    null. */
 303 int
 304 set_lang (var, value)
 305      char *var, *value;
 306 {
 307   FREE (lang);
 308   if (value)
 309     lang = savestring (value);
 310   else
 311     {
 312       lang = (char *)xmalloc (1);
 313       lang[0] = '\0';
 314     }
 315
 316   return ((lc_all == 0 || *lc_all == 0) ? reset_locale_vars () : 0);
 317 }
 318
 319 /* Set default values for LANG and LC_ALL.  Default values for all other
 320    locale-related variables depend on these. */
 321 void
 322 set_default_lang ()
 323 {
 324   char *v;
 325
 326   v = get_string_value ("LC_ALL");
 327   set_locale_var ("LC_ALL", v);
 328
 329   v = get_string_value ("LANG");
 330   set_lang ("LANG", v);
 331 }
 332
 333 /* Get the value of one of the locale variables (LC_MESSAGES, LC_CTYPE).
 334    The precedence is as POSIX.2 specifies:  LC_ALL has precedence over
 335    the specific locale variables, and LANG, if set, is used as the default. */
 336 char *
 337 get_locale_var (var)
 338      char *var;
 339 {
 340   char *locale;
 341
 342   locale = lc_all;
 343
 344   if (locale == 0 || *locale == 0)
 345     locale = get_string_value (var);    /* XXX - no mem leak */
 346   if (locale == 0 || *locale == 0)
 347     locale = lang;
 348   if (locale == 0 || *locale == 0)
 349 #if 0
 350     locale = default_locale;    /* system-dependent; not really portable.  should it be "C"? */
 351 #else
 352     locale = "";
 353 #endif
 354   return (locale);
 355 }
 356
 357 /* Called to reset all of the locale variables to their appropriate values
 358    if (and only if) LC_ALL has not been assigned a value.  DO NOT CALL THIS
 359    IF LC_ALL HAS BEEN ASSIGNED A VALUE. */
 360 static int
 361 reset_locale_vars ()
 362 {
 363   char *t, *x;
 364 #if defined (HAVE_SETLOCALE)
 365   if (lang == 0 || *lang == '\0')
 366     maybe_make_export_env ();           /* trust that this will change environment for setlocale */
 367   if (setlocale (LC_ALL, lang ? lang : "") == 0)
 368     return 0;
 369
 370   x = 0;
 371 #  if defined (LC_CTYPE)
 372   x = setlocale (LC_CTYPE, get_locale_var ("LC_CTYPE"));
 373 #  endif
 374 #  if defined (LC_COLLATE)
 375   t = setlocale (LC_COLLATE, get_locale_var ("LC_COLLATE"));
 376 #  endif
 377 #  if defined (LC_MESSAGES)
 378   t = setlocale (LC_MESSAGES, get_locale_var ("LC_MESSAGES"));
 379 #  endif
 380 #  if defined (LC_NUMERIC)
 381   t = setlocale (LC_NUMERIC, get_locale_var ("LC_NUMERIC"));
 382 #  endif
 383 #  if defined (LC_TIME)
 384   t = setlocale (LC_TIME, get_locale_var ("LC_TIME"));
 385 #  endif
 386
 387   locale_setblanks ();
 388   locale_mb_cur_max = MB_CUR_MAX;
 389   if (x)
 390     locale_utf8locale = locale_isutf8 (x);
 391 #  if defined (HANDLE_MULTIBYTE)
 392   locale_shiftstates = mblen ((char *)NULL, 0);
 393 #  else
 394   local_shiftstates = 0;
 395 #  endif
 396   u32reset ();
 397 #endif
 398   return 1;
 399 }
 400
 401 /* Translate the contents of STRING, a $"..." quoted string, according
 402    to the current locale.  In the `C' or `POSIX' locale, or if gettext()
 403    is not available, the passed string is returned unchanged.  The
 404    length of the translated string is returned in LENP, if non-null. */
 405 char *
 406 localetrans (string, len, lenp)
 407      char *string;
 408      int len, *lenp;
 409 {
 410   char *locale, *t;
 411   char *translated;
 412   int tlen;
 413
 414   /* Don't try to translate null strings. */
 415   if (string == 0 || *string == 0)
 416     {
 417       if (lenp)
 418         *lenp = 0;
 419       return ((char *)NULL);
 420     }
 421
 422   locale = get_locale_var ("LC_MESSAGES");
 423
 424   /* If we don't have setlocale() or the current locale is `C' or `POSIX',
 425      just return the string.  If we don't have gettext(), there's no use
 426      doing anything else. */
 427   if (locale == 0 || locale[0] == '\0' ||
 428       (locale[0] == 'C' && locale[1] == '\0') || STREQ (locale, "POSIX"))
 429     {
 430       t = (char *)xmalloc (len + 1);
 431       strcpy (t, string);
 432       if (lenp)
 433         *lenp = len;
 434       return (t);
 435     }
 436
 437   /* Now try to translate it. */
 438   if (default_domain && *default_domain)
 439     translated = dgettext (default_domain, string);
 440   else
 441     translated = string;
 442
 443   if (translated == string)     /* gettext returns its argument if untranslatable */
 444     {
 445       t = (char *)xmalloc (len + 1);
 446       strcpy (t, string);
 447       if (lenp)
 448         *lenp = len;
 449     }
 450   else
 451     {
 452       tlen = strlen (translated);
 453       t = (char *)xmalloc (tlen + 1);
 454       strcpy (t, translated);
 455       if (lenp)
 456         *lenp = tlen;
 457     }
 458   return (t);
 459 }
 460
 461 /* Change a bash string into a string suitable for inclusion in a `po' file.
 462    This backslash-escapes `"' and `\' and changes newlines into \\\n"\n". */
 463 char *
 464 mk_msgstr (string, foundnlp)
 465      char *string;
 466      int *foundnlp;
 467 {
 468   register int c, len;
 469   char *result, *r, *s;
 470
 471   for (len = 0, s = string; s && *s; s++)
 472     {
 473       len++;
 474       if (*s == '"' || *s == '\\')
 475         len++;
 476       else if (*s == '\n')
 477         len += 5;
 478     }
 479
 480   r = result = (char *)xmalloc (len + 3);
 481   *r++ = '"';
 482
 483   for (s = string; s && (c = *s); s++)
 484     {
 485       if (c == '\n')    /* <NL> -> \n"<NL>" */
 486         {
 487           *r++ = '\\';
 488           *r++ = 'n';
 489           *r++ = '"';
 490           *r++ = '\n';
 491           *r++ = '"';
 492           if (foundnlp)
 493             *foundnlp = 1;
 494           continue;
 495         }
 496       if (c == '"' || c == '\\')
 497         *r++ = '\\';
 498       *r++ = c;
 499     }
 500
 501   *r++ = '"';
 502   *r++ = '\0';
 503
 504   return result;
 505 }
 506
 507 /* $"..." -- Translate the portion of STRING between START and END
 508    according to current locale using gettext (if available) and return
 509    the result.  The caller will take care of leaving the quotes intact.
 510    The string will be left without the leading `$' by the caller.
 511    If translation is performed, the translated string will be double-quoted
 512    by the caller.  The length of the translated string is returned in LENP,
 513    if non-null. */
 514 char *
 515 localeexpand (string, start, end, lineno, lenp)
 516      char *string;
 517      int start, end, lineno, *lenp;
 518 {
 519   int len, tlen, foundnl;
 520   char *temp, *t, *t2;
 521
 522   temp = (char *)xmalloc (end - start + 1);
 523   for (tlen = 0, len = start; len < end; )
 524     temp[tlen++] = string[len++];
 525   temp[tlen] = '\0';
 526
 527   /* If we're just dumping translatable strings, don't do anything with the
 528      string itself, but if we're dumping in `po' file format, convert it into
 529      a form more palatable to gettext(3) and friends by quoting `"' and `\'
 530      with backslashes and converting <NL> into `\n"<NL>"'.  If we find a
 531      newline in TEMP, we first output a `msgid ""' line and then the
 532      translated string; otherwise we output the `msgid' and translated
 533      string all on one line. */
 534   if (dump_translatable_strings)
 535     {
 536       if (dump_po_strings)
 537         {
 538           foundnl = 0;
 539           t = mk_msgstr (temp, &foundnl);
 540           t2 = foundnl ? "\"\"\n" : "";
 541
 542           printf ("#: %s:%d\nmsgid %s%s\nmsgstr \"\"\n",
 543                         yy_input_name (), lineno, t2, t);
 544           free (t);
 545         }
 546       else
 547         printf ("\"%s\"\n", temp);
 548
 549       if (lenp)
 550         *lenp = tlen;
 551       return (temp);
 552     }
 553   else if (*temp)
 554     {
 555       t = localetrans (temp, tlen, &len);
 556       free (temp);
 557       if (lenp)
 558         *lenp = len;
 559       return (t);
 560     }
 561   else
 562     {
 563       if (lenp)
 564         *lenp = 0;
 565       return (temp);
 566     }
 567 }
 568
 569 /* Set every character in the <blank> character class to be a shell break
 570    character for the lexical analyzer when the locale changes. */
 571 static void
 572 locale_setblanks ()
 573 {
 574   int x;
 575
 576   for (x = 0; x < sh_syntabsiz; x++)
 577     {
 578       if (isblank ((unsigned char)x))
 579         sh_syntaxtab[x] |= CSHBRK|CBLANK;
 580       else if (member (x, shell_break_chars))
 581         {
 582           sh_syntaxtab[x] |= CSHBRK;
 583           sh_syntaxtab[x] &= ~CBLANK;
 584         }
 585       else
 586         sh_syntaxtab[x] &= ~(CSHBRK|CBLANK);
 587     }
 588 }
 589
 590 /* Parse a locale specification
 591      language[_territory][.codeset][@modifier][+special][,[sponsor][_revision]]
 592    and return TRUE if the codeset is UTF-8 or utf8 */
 593 static int
 594 locale_isutf8 (lspec)
 595      char *lspec;
 596 {
 597   char *cp, *encoding;
 598
 599 #if HAVE_LANGINFO_CODESET
 600   cp = nl_langinfo (CODESET);
 601   return (STREQ (cp, "UTF-8") || STREQ (cp, "utf8"));
 602 #elif HAVE_LOCALE_CHARSET
 603   cp = locale_charset ();
 604   return (STREQ (cp, "UTF-8") || STREQ (cp, "utf8"));
 605 #else
 606   /* Take a shot */
 607   for (cp = lspec; *cp && *cp != '@' && *cp != '+' && *cp != ','; cp++)
 608     {
 609       if (*cp == '.')
 610         {
 611           for (encoding = ++cp; *cp && *cp != '@' && *cp != '+' && *cp != ','; cp++)
 612             ;
 613           /* The encoding (codeset) is the substring between encoding and cp */
 614           if ((cp - encoding == 5 && STREQN (encoding, "UTF-8", 5)) ||
 615               (cp - encoding == 4 && STREQN (encoding, "utf8", 4)))
 616             return 1;
 617           else
 618             return 0;
 619         }
 620     }
 621   return 0;
 622 #endif
 623 }
 624
 625 #if defined (HAVE_LOCALECONV)
 626 int
 627 locale_decpoint ()
 628 {
 629   struct lconv *lv;
 630
 631   lv = localeconv ();
 632   return (lv && lv->decimal_point && lv->decimal_point[0]) ? lv->decimal_point[0] : '.';
 633 }
 634 #else
 635 #  undef locale_decpoint
 636 int
 637 locale_decpoint ()
 638 {
 639   return '.';
 640 }
 641 #endif