intl/dcigettext.c

   1 /* Implementation of the internal dcigettext function.
   2    Copyright (C) 1995-1999, 2000-2003 Free Software Foundation, Inc.
   3
   4    This program is free software; you can redistribute it and/or modify it
   5    under the terms of the GNU Library General Public License as published
   6    by the Free Software Foundation; either version 2, or (at your option)
   7    any later version.
   8
   9    This program is distributed in the hope that it will be useful,
  10    but WITHOUT ANY WARRANTY; without even the implied warranty of
  11    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  12    Library General Public License for more details.
  13
  14    You should have received a copy of the GNU Library General Public
  15    License along with this program; if not, write to the Free Software
  16    Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307,
  17    USA.  */
  18
  19 /* Tell glibc's <string.h> to provide a prototype for mempcpy().
  20    This must come before <config.h> because <config.h> may include
  21    <features.h>, and once <features.h> has been included, it's too late.  */
  22 #ifndef _GNU_SOURCE
  23 # define _GNU_SOURCE    1
  24 #endif
  25
  26 #ifdef HAVE_CONFIG_H
  27 # include <config.h>
  28 #endif
  29
  30 #include <sys/types.h>
  31
  32 #ifdef __GNUC__
  33 # define alloca __builtin_alloca
  34 # define HAVE_ALLOCA 1
  35 #else
  36 # ifdef _MSC_VER
  37 #  include <malloc.h>
  38 #  define alloca _alloca
  39 # else
  40 #  if defined HAVE_ALLOCA_H || defined _LIBC
  41 #   include <alloca.h>
  42 #  else
  43 #   ifdef _AIX
  44  #pragma alloca
  45 #   else
  46 #    ifndef alloca
  47 char *alloca ();
  48 #    endif
  49 #   endif
  50 #  endif
  51 # endif
  52 #endif
  53
  54 #include <errno.h>
  55 #ifndef errno
  56 extern int errno;
  57 #endif
  58 #ifndef __set_errno
  59 # define __set_errno(val) errno = (val)
  60 #endif
  61
  62 #include <stddef.h>
  63 #include <stdlib.h>
  64 #include <string.h>
  65
  66 #if defined HAVE_UNISTD_H || defined _LIBC
  67 # include <unistd.h>
  68 #endif
  69
  70 #include <locale.h>
  71
  72 #ifdef _LIBC
  73   /* Guess whether integer division by zero raises signal SIGFPE.
  74      Set to 1 only if you know for sure.  In case of doubt, set to 0.  */
  75 # if defined __alpha__ || defined __arm__ || defined __i386__ \
  76      || defined __m68k__ || defined __s390__
  77 #  define INTDIV0_RAISES_SIGFPE 1
  78 # else
  79 #  define INTDIV0_RAISES_SIGFPE 0
  80 # endif
  81 #endif
  82 #if !INTDIV0_RAISES_SIGFPE
  83 # include <signal.h>
  84 #endif
  85
  86 #if defined HAVE_SYS_PARAM_H || defined _LIBC
  87 # include <sys/param.h>
  88 #endif
  89
  90 #include "gettextP.h"
  91 #include "plural-exp.h"
  92 #ifdef _LIBC
  93 # include <libintl.h>
  94 #else
  95 # include "libgnuintl.h"
  96 #endif
  97 #include "hash-string.h"
  98
  99 /* Thread safetyness.  */
 100 #ifdef _LIBC
 101 # include <bits/libc-lock.h>
 102 #else
 103 /* Provide dummy implementation if this is outside glibc.  */
 104 # define __libc_lock_define_initialized(CLASS, NAME)
 105 # define __libc_lock_lock(NAME)
 106 # define __libc_lock_unlock(NAME)
 107 # define __libc_rwlock_define_initialized(CLASS, NAME)
 108 # define __libc_rwlock_rdlock(NAME)
 109 # define __libc_rwlock_unlock(NAME)
 110 #endif
 111
 112 /* Alignment of types.  */
 113 #if defined __GNUC__ && __GNUC__ >= 2
 114 # define alignof(TYPE) __alignof__ (TYPE)
 115 #else
 116 # define alignof(TYPE) \
 117     ((int) &((struct { char dummy1; TYPE dummy2; } *) 0)->dummy2)
 118 #endif
 119
 120 /* The internal variables in the standalone libintl.a must have different
 121    names than the internal variables in GNU libc, otherwise programs
 122    using libintl.a cannot be linked statically.  */
 123 #if !defined _LIBC
 124 # define _nl_default_default_domain libintl_nl_default_default_domain
 125 # define _nl_current_default_domain libintl_nl_current_default_domain
 126 # define _nl_default_dirname libintl_nl_default_dirname
 127 # define _nl_domain_bindings libintl_nl_domain_bindings
 128 #endif
 129
 130 /* Some compilers, like SunOS4 cc, don't have offsetof in <stddef.h>.  */
 131 #ifndef offsetof
 132 # define offsetof(type,ident) ((size_t)&(((type*)0)->ident))
 133 #endif
 134
 135 /* @@ end of prolog @@ */
 136
 137 #ifdef _LIBC
 138 /* Rename the non ANSI C functions.  This is required by the standard
 139    because some ANSI C functions will require linking with this object
 140    file and the name space must not be polluted.  */
 141 # define getcwd __getcwd
 142 # ifndef stpcpy
 143 #  define stpcpy __stpcpy
 144 # endif
 145 # define tfind __tfind
 146 #else
 147 # if !defined HAVE_GETCWD
 148 char *getwd ();
 149 #  define getcwd(buf, max) getwd (buf)
 150 # else
 151 char *getcwd ();
 152 # endif
 153 # ifndef HAVE_STPCPY
 154 static char *stpcpy PARAMS ((char *dest, const char *src));
 155 # endif
 156 # ifndef HAVE_MEMPCPY
 157 static void *mempcpy PARAMS ((void *dest, const void *src, size_t n));
 158 # endif
 159 #endif
 160
 161 /* Amount to increase buffer size by in each try.  */
 162 #define PATH_INCR 32
 163
 164 /* The following is from pathmax.h.  */
 165 /* Non-POSIX BSD systems might have gcc's limits.h, which doesn't define
 166    PATH_MAX but might cause redefinition warnings when sys/param.h is
 167    later included (as on MORE/BSD 4.3).  */
 168 #if defined _POSIX_VERSION || (defined HAVE_LIMITS_H && !defined __GNUC__)
 169 # include <limits.h>
 170 #endif
 171
 172 #ifndef _POSIX_PATH_MAX
 173 # define _POSIX_PATH_MAX 255
 174 #endif
 175
 176 #if !defined PATH_MAX && defined _PC_PATH_MAX
 177 # define PATH_MAX (pathconf ("/", _PC_PATH_MAX) < 1 ? 1024 : pathconf ("/", _PC_PATH_MAX))
 178 #endif
 179
 180 /* Don't include sys/param.h if it already has been.  */
 181 #if defined HAVE_SYS_PARAM_H && !defined PATH_MAX && !defined MAXPATHLEN
 182 # include <sys/param.h>
 183 #endif
 184
 185 #if !defined PATH_MAX && defined MAXPATHLEN
 186 # define PATH_MAX MAXPATHLEN
 187 #endif
 188
 189 #ifndef PATH_MAX
 190 # define PATH_MAX _POSIX_PATH_MAX
 191 #endif
 192
 193 /* Pathname support.
 194    ISSLASH(C)           tests whether C is a directory separator character.
 195    IS_ABSOLUTE_PATH(P)  tests whether P is an absolute path.  If it is not,
 196                         it may be concatenated to a directory pathname.
 197    IS_PATH_WITH_DIR(P)  tests whether P contains a directory specification.
 198  */
 199 #if defined _WIN32 || defined __WIN32__ || defined __EMX__ || defined __DJGPP__
 200   /* Win32, OS/2, DOS */
 201 # define ISSLASH(C) ((C) == '/' || (C) == '\\')
 202 # define HAS_DEVICE(P) \
 203     ((((P)[0] >= 'A' && (P)[0] <= 'Z') || ((P)[0] >= 'a' && (P)[0] <= 'z')) \
 204      && (P)[1] == ':')
 205 # define IS_ABSOLUTE_PATH(P) (ISSLASH ((P)[0]) || HAS_DEVICE (P))
 206 # define IS_PATH_WITH_DIR(P) \
 207     (strchr (P, '/') != NULL || strchr (P, '\\') != NULL || HAS_DEVICE (P))
 208 #else
 209   /* Unix */
 210 # define ISSLASH(C) ((C) == '/')
 211 # define IS_ABSOLUTE_PATH(P) ISSLASH ((P)[0])
 212 # define IS_PATH_WITH_DIR(P) (strchr (P, '/') != NULL)
 213 #endif
 214
 215 /* This is the type used for the search tree where known translations
 216    are stored.  */
 217 struct known_translation_t
 218 {
 219   /* Domain in which to search.  */
 220   char *domainname;
 221
 222   /* The category.  */
 223   int category;
 224
 225   /* State of the catalog counter at the point the string was found.  */
 226   int counter;
 227
 228   /* Catalog where the string was found.  */
 229   struct loaded_l10nfile *domain;
 230
 231   /* And finally the translation.  */
 232   const char *translation;
 233   size_t translation_length;
 234
 235   /* Pointer to the string in question.  */
 236   char msgid[ZERO];
 237 };
 238
 239 /* Root of the search tree with known translations.  We can use this
 240    only if the system provides the `tsearch' function family.  */
 241 #if defined HAVE_TSEARCH || defined _LIBC
 242 # include <search.h>
 243
 244 static void *root;
 245
 246 # ifdef _LIBC
 247 #  define tsearch __tsearch
 248 # endif
 249
 250 /* Function to compare two entries in the table of known translations.  */
 251 static int transcmp PARAMS ((const void *p1, const void *p2));
 252 static int
 253 transcmp (p1, p2)
 254      const void *p1;
 255      const void *p2;
 256 {
 257   const struct known_translation_t *s1;
 258   const struct known_translation_t *s2;
 259   int result;
 260
 261   s1 = (const struct known_translation_t *) p1;
 262   s2 = (const struct known_translation_t *) p2;
 263
 264   result = strcmp (s1->msgid, s2->msgid);
 265   if (result == 0)
 266     {
 267       result = strcmp (s1->domainname, s2->domainname);
 268       if (result == 0)
 269         /* We compare the category last (though this is the cheapest
 270            operation) since it is hopefully always the same (namely
 271            LC_MESSAGES).  */
 272         result = s1->category - s2->category;
 273     }
 274
 275   return result;
 276 }
 277 #endif
 278
 279 #ifndef INTVARDEF
 280 # define INTVARDEF(name)
 281 #endif
 282 #ifndef INTUSE
 283 # define INTUSE(name) name
 284 #endif
 285
 286 /* Name of the default domain used for gettext(3) prior any call to
 287    textdomain(3).  The default value for this is "messages".  */
 288 const char _nl_default_default_domain[] attribute_hidden = "messages";
 289
 290 /* Value used as the default domain for gettext(3).  */
 291 const char *_nl_current_default_domain attribute_hidden
 292      = _nl_default_default_domain;
 293
 294 /* Contains the default location of the message catalogs.  */
 295 #if defined __EMX__
 296 extern const char _nl_default_dirname[];
 297 #else
 298 const char _nl_default_dirname[] = LOCALEDIR;
 299 INTVARDEF (_nl_default_dirname)
 300 #endif
 301
 302 /* List with bindings of specific domains created by bindtextdomain()
 303    calls.  */
 304 struct binding *_nl_domain_bindings;
 305
 306 /* Prototypes for local functions.  */
 307 static char *plural_lookup PARAMS ((struct loaded_l10nfile *domain,
 308                                     unsigned long int n,
 309                                     const char *translation,
 310                                     size_t translation_len))
 311      internal_function;
 312 static const char *guess_category_value PARAMS ((int category,
 313                                                  const char *categoryname))
 314      internal_function;
 315 #ifdef _LIBC
 316 # include "../locale/localeinfo.h"
 317 # define category_to_name(category)     _nl_category_names[category]
 318 #else
 319 static const char *category_to_name PARAMS ((int category)) internal_function;
 320 #endif
 321
 322
 323 /* For those loosing systems which don't have `alloca' we have to add
 324    some additional code emulating it.  */
 325 #ifdef HAVE_ALLOCA
 326 /* Nothing has to be done.  */
 327 # define freea(p) /* nothing */
 328 # define ADD_BLOCK(list, address) /* nothing */
 329 # define FREE_BLOCKS(list) /* nothing */
 330 #else
 331 struct block_list
 332 {
 333   void *address;
 334   struct block_list *next;
 335 };
 336 # define ADD_BLOCK(list, addr)                                                \
 337   do {                                                                        \
 338     struct block_list *newp = (struct block_list *) malloc (sizeof (*newp));  \
 339     /* If we cannot get a free block we cannot add the new element to         \
 340        the list.  */                                                          \
 341     if (newp != NULL) {                                                       \
 342       newp->address = (addr);                                                 \
 343       newp->next = (list);                                                    \
 344       (list) = newp;                                                          \
 345     }                                                                         \
 346   } while (0)
 347 # define FREE_BLOCKS(list)                                                    \
 348   do {                                                                        \
 349     while (list != NULL) {                                                    \
 350       struct block_list *old = list;                                          \
 351       list = list->next;                                                      \
 352       free (old->address);                                                    \
 353       free (old);                                                             \
 354     }                                                                         \
 355   } while (0)
 356 # undef alloca
 357 # define alloca(size) (malloc (size))
 358 # define freea(p) free (p)
 359 #endif  /* have alloca */
 360
 361
 362 #ifdef _LIBC
 363 /* List of blocks allocated for translations.  */
 364 typedef struct transmem_list
 365 {
 366   struct transmem_list *next;
 367   char data[ZERO];
 368 } transmem_block_t;
 369 static struct transmem_list *transmem_list;
 370 #else
 371 typedef unsigned char transmem_block_t;
 372 #endif
 373
 374
 375 /* Names for the libintl functions are a problem.  They must not clash
 376    with existing names and they should follow ANSI C.  But this source
 377    code is also used in GNU C Library where the names have a __
 378    prefix.  So we have to make a difference here.  */
 379 #ifdef _LIBC
 380 # define DCIGETTEXT __dcigettext
 381 #else
 382 # define DCIGETTEXT libintl_dcigettext
 383 #endif
 384
 385 /* Lock variable to protect the global data in the gettext implementation.  */
 386 #ifdef _LIBC
 387 __libc_rwlock_define_initialized (, _nl_state_lock attribute_hidden)
 388 #endif
 389
 390 /* Checking whether the binaries runs SUID must be done and glibc provides
 391    easier methods therefore we make a difference here.  */
 392 #ifdef _LIBC
 393 # define ENABLE_SECURE __libc_enable_secure
 394 # define DETERMINE_SECURE
 395 #else
 396 # ifndef HAVE_GETUID
 397 #  define getuid() 0
 398 # endif
 399 # ifndef HAVE_GETGID
 400 #  define getgid() 0
 401 # endif
 402 # ifndef HAVE_GETEUID
 403 #  define geteuid() getuid()
 404 # endif
 405 # ifndef HAVE_GETEGID
 406 #  define getegid() getgid()
 407 # endif
 408 static int enable_secure;
 409 # define ENABLE_SECURE (enable_secure == 1)
 410 # define DETERMINE_SECURE \
 411   if (enable_secure == 0)                                                     \
 412     {                                                                         \
 413       if (getuid () != geteuid () || getgid () != getegid ())                 \
 414         enable_secure = 1;                                                    \
 415       else                                                                    \
 416         enable_secure = -1;                                                   \
 417     }
 418 #endif
 419
 420 /* Get the function to evaluate the plural expression.  */
 421 #include "eval-plural.h"
 422
 423 /* Look up MSGID in the DOMAINNAME message catalog for the current
 424    CATEGORY locale and, if PLURAL is nonzero, search over string
 425    depending on the plural form determined by N.  */
 426 char *
 427 DCIGETTEXT (domainname, msgid1, msgid2, plural, n, category)
 428      const char *domainname;
 429      const char *msgid1;
 430      const char *msgid2;
 431      int plural;
 432      unsigned long int n;
 433      int category;
 434 {
 435 #ifndef HAVE_ALLOCA
 436   struct block_list *block_list = NULL;
 437 #endif
 438   struct loaded_l10nfile *domain;
 439   struct binding *binding;
 440   const char *categoryname;
 441   const char *categoryvalue;
 442   char *dirname, *xdomainname;
 443   char *single_locale;
 444   char *retval;
 445   size_t retlen;
 446   int saved_errno;
 447 #if defined HAVE_TSEARCH || defined _LIBC
 448   struct known_translation_t *search;
 449   struct known_translation_t **foundp = NULL;
 450   size_t msgid_len;
 451 #endif
 452   size_t domainname_len;
 453
 454   /* If no real MSGID is given return NULL.  */
 455   if (msgid1 == NULL)
 456     return NULL;
 457
 458 #ifdef _LIBC
 459   if (category < 0 || category >= __LC_LAST || category == LC_ALL)
 460     /* Bogus.  */
 461     return (plural == 0
 462             ? (char *) msgid1
 463             /* Use the Germanic plural rule.  */
 464             : n == 1 ? (char *) msgid1 : (char *) msgid2);
 465 #endif
 466
 467   __libc_rwlock_rdlock (_nl_state_lock);
 468
 469   /* If DOMAINNAME is NULL, we are interested in the default domain.  If
 470      CATEGORY is not LC_MESSAGES this might not make much sense but the
 471      definition left this undefined.  */
 472   if (domainname == NULL)
 473     domainname = _nl_current_default_domain;
 474
 475   /* OS/2 specific: backward compatibility with older libintl versions  */
 476 #ifdef LC_MESSAGES_COMPAT
 477   if (category == LC_MESSAGES_COMPAT)
 478     category = LC_MESSAGES;
 479 #endif
 480
 481 #if defined HAVE_TSEARCH || defined _LIBC
 482   msgid_len = strlen (msgid1) + 1;
 483
 484   /* Try to find the translation among those which we found at
 485      some time.  */
 486   search = (struct known_translation_t *)
 487            alloca (offsetof (struct known_translation_t, msgid) + msgid_len);
 488   memcpy (search->msgid, msgid1, msgid_len);
 489   search->domainname = (char *) domainname;
 490   search->category = category;
 491
 492   foundp = (struct known_translation_t **) tfind (search, &root, transcmp);
 493   freea (search);
 494   if (foundp != NULL && (*foundp)->counter == _nl_msg_cat_cntr)
 495     {
 496       /* Now deal with plural.  */
 497       if (plural)
 498         retval = plural_lookup ((*foundp)->domain, n, (*foundp)->translation,
 499                                 (*foundp)->translation_length);
 500       else
 501         retval = (char *) (*foundp)->translation;
 502
 503       __libc_rwlock_unlock (_nl_state_lock);
 504       return retval;
 505     }
 506 #endif
 507
 508   /* Preserve the `errno' value.  */
 509   saved_errno = errno;
 510
 511   /* See whether this is a SUID binary or not.  */
 512   DETERMINE_SECURE;
 513
 514   /* First find matching binding.  */
 515   for (binding = _nl_domain_bindings; binding != NULL; binding = binding->next)
 516     {
 517       int compare = strcmp (domainname, binding->domainname);
 518       if (compare == 0)
 519         /* We found it!  */
 520         break;
 521       if (compare < 0)
 522         {
 523           /* It is not in the list.  */
 524           binding = NULL;
 525           break;
 526         }
 527     }
 528
 529   if (binding == NULL)
 530     dirname = (char *) INTUSE(_nl_default_dirname);
 531   else if (IS_ABSOLUTE_PATH (binding->dirname))
 532     dirname = binding->dirname;
 533   else
 534     {
 535       /* We have a relative path.  Make it absolute now.  */
 536       size_t dirname_len = strlen (binding->dirname) + 1;
 537       size_t path_max;
 538       char *ret;
 539
 540       path_max = (unsigned int) PATH_MAX;
 541       path_max += 2;            /* The getcwd docs say to do this.  */
 542
 543       for (;;)
 544         {
 545           dirname = (char *) alloca (path_max + dirname_len);
 546           ADD_BLOCK (block_list, dirname);
 547
 548           __set_errno (0);
 549           ret = getcwd (dirname, path_max);
 550           if (ret != NULL || errno != ERANGE)
 551             break;
 552
 553           path_max += path_max / 2;
 554           path_max += PATH_INCR;
 555         }
 556
 557       if (ret == NULL)
 558         /* We cannot get the current working directory.  Don't signal an
 559            error but simply return the default string.  */
 560         goto return_untranslated;
 561
 562       stpcpy (stpcpy (strchr (dirname, '\0'), "/"), binding->dirname);
 563     }
 564
 565   /* Now determine the symbolic name of CATEGORY and its value.  */
 566   categoryname = category_to_name (category);
 567   categoryvalue = guess_category_value (category, categoryname);
 568
 569   domainname_len = strlen (domainname);
 570   xdomainname = (char *) alloca (strlen (categoryname)
 571                                  + domainname_len + 5);
 572   ADD_BLOCK (block_list, xdomainname);
 573
 574   stpcpy (mempcpy (stpcpy (stpcpy (xdomainname, categoryname), "/"),
 575                   domainname, domainname_len),
 576           ".mo");
 577
 578   /* Creating working area.  */
 579   single_locale = (char *) alloca (strlen (categoryvalue) + 1);
 580   ADD_BLOCK (block_list, single_locale);
 581
 582
 583   /* Search for the given string.  This is a loop because we perhaps
 584      got an ordered list of languages to consider for the translation.  */
 585   while (1)
 586     {
 587       /* Make CATEGORYVALUE point to the next element of the list.  */
 588       while (categoryvalue[0] != '\0' && categoryvalue[0] == ':')
 589         ++categoryvalue;
 590       if (categoryvalue[0] == '\0')
 591         {
 592           /* The whole contents of CATEGORYVALUE has been searched but
 593              no valid entry has been found.  We solve this situation
 594              by implicitly appending a "C" entry, i.e. no translation
 595              will take place.  */
 596           single_locale[0] = 'C';
 597           single_locale[1] = '\0';
 598         }
 599       else
 600         {
 601           char *cp = single_locale;
 602           while (categoryvalue[0] != '\0' && categoryvalue[0] != ':')
 603             *cp++ = *categoryvalue++;
 604           *cp = '\0';
 605
 606           /* When this is a SUID binary we must not allow accessing files
 607              outside the dedicated directories.  */
 608           if (ENABLE_SECURE && IS_PATH_WITH_DIR (single_locale))
 609             /* Ingore this entry.  */
 610             continue;
 611         }
 612
 613       /* If the current locale value is C (or POSIX) we don't load a
 614          domain.  Return the MSGID.  */
 615       if (strcmp (single_locale, "C") == 0
 616           || strcmp (single_locale, "POSIX") == 0)
 617         break;
 618
 619       /* Find structure describing the message catalog matching the
 620          DOMAINNAME and CATEGORY.  */
 621       domain = _nl_find_domain (dirname, single_locale, xdomainname, binding);
 622
 623       if (domain != NULL)
 624         {
 625           retval = _nl_find_msg (domain, binding, msgid1, &retlen);
 626
 627           if (retval == NULL)
 628             {
 629               int cnt;
 630
 631               for (cnt = 0; domain->successor[cnt] != NULL; ++cnt)
 632                 {
 633                   retval = _nl_find_msg (domain->successor[cnt], binding,
 634                                          msgid1, &retlen);
 635
 636                   if (retval != NULL)
 637                     {
 638                       domain = domain->successor[cnt];
 639                       break;
 640                     }
 641                 }
 642             }
 643
 644           if (retval != NULL)
 645             {
 646               /* Found the translation of MSGID1 in domain DOMAIN:
 647                  starting at RETVAL, RETLEN bytes.  */
 648               FREE_BLOCKS (block_list);
 649 #if defined HAVE_TSEARCH || defined _LIBC
 650               if (foundp == NULL)
 651                 {
 652                   /* Create a new entry and add it to the search tree.  */
 653                   struct known_translation_t *newp;
 654
 655                   newp = (struct known_translation_t *)
 656                     malloc (offsetof (struct known_translation_t, msgid)
 657                             + msgid_len + domainname_len + 1);
 658                   if (newp != NULL)
 659                     {
 660                       newp->domainname =
 661                         mempcpy (newp->msgid, msgid1, msgid_len);
 662                       memcpy (newp->domainname, domainname, domainname_len + 1);
 663                       newp->category = category;
 664                       newp->counter = _nl_msg_cat_cntr;
 665                       newp->domain = domain;
 666                       newp->translation = retval;
 667                       newp->translation_length = retlen;
 668
 669                       /* Insert the entry in the search tree.  */
 670                       foundp = (struct known_translation_t **)
 671                         tsearch (newp, &root, transcmp);
 672                       if (foundp == NULL
 673                           || __builtin_expect (*foundp != newp, 0))
 674                         /* The insert failed.  */
 675                         free (newp);
 676                     }
 677                 }
 678               else
 679                 {
 680                   /* We can update the existing entry.  */
 681                   (*foundp)->counter = _nl_msg_cat_cntr;
 682                   (*foundp)->domain = domain;
 683                   (*foundp)->translation = retval;
 684                   (*foundp)->translation_length = retlen;
 685                 }
 686 #endif
 687               __set_errno (saved_errno);
 688
 689               /* Now deal with plural.  */
 690               if (plural)
 691                 retval = plural_lookup (domain, n, retval, retlen);
 692
 693               __libc_rwlock_unlock (_nl_state_lock);
 694               return retval;
 695             }
 696         }
 697     }
 698
 699  return_untranslated:
 700   /* Return the untranslated MSGID.  */
 701   FREE_BLOCKS (block_list);
 702   __libc_rwlock_unlock (_nl_state_lock);
 703 #ifndef _LIBC
 704   if (!ENABLE_SECURE)
 705     {
 706       extern void _nl_log_untranslated PARAMS ((const char *logfilename,
 707                                                 const char *domainname,
 708                                                 const char *msgid1,
 709                                                 const char *msgid2,
 710                                                 int plural));
 711       const char *logfilename = getenv ("GETTEXT_LOG_UNTRANSLATED");
 712
 713       if (logfilename != NULL && logfilename[0] != '\0')
 714         _nl_log_untranslated (logfilename, domainname, msgid1, msgid2, plural);
 715     }
 716 #endif
 717   __set_errno (saved_errno);
 718   return (plural == 0
 719           ? (char *) msgid1
 720           /* Use the Germanic plural rule.  */
 721           : n == 1 ? (char *) msgid1 : (char *) msgid2);
 722 }
 723
 724
 725 char *
 726 internal_function
 727 _nl_find_msg (domain_file, domainbinding, msgid, lengthp)
 728      struct loaded_l10nfile *domain_file;
 729      struct binding *domainbinding;
 730      const char *msgid;
 731      size_t *lengthp;
 732 {
 733   struct loaded_domain *domain;
 734   nls_uint32 nstrings;
 735   size_t act;
 736   char *result;
 737   size_t resultlen;
 738
 739   if (domain_file->decided == 0)
 740     _nl_load_domain (domain_file, domainbinding);
 741
 742   if (domain_file->data == NULL)
 743     return NULL;
 744
 745   domain = (struct loaded_domain *) domain_file->data;
 746
 747   nstrings = domain->nstrings;
 748
 749   /* Locate the MSGID and its translation.  */
 750   if (domain->hash_tab != NULL)
 751     {
 752       /* Use the hashing table.  */
 753       nls_uint32 len = strlen (msgid);
 754       nls_uint32 hash_val = hash_string (msgid);
 755       nls_uint32 idx = hash_val % domain->hash_size;
 756       nls_uint32 incr = 1 + (hash_val % (domain->hash_size - 2));
 757
 758       while (1)
 759         {
 760           nls_uint32 nstr =
 761             W (domain->must_swap_hash_tab, domain->hash_tab[idx]);
 762
 763           if (nstr == 0)
 764             /* Hash table entry is empty.  */
 765             return NULL;
 766
 767           nstr--;
 768
 769           /* Compare msgid with the original string at index nstr.
 770              We compare the lengths with >=, not ==, because plural entries
 771              are represented by strings with an embedded NUL.  */
 772           if (nstr < nstrings
 773               ? W (domain->must_swap, domain->orig_tab[nstr].length) >= len
 774                 && (strcmp (msgid,
 775                             domain->data + W (domain->must_swap,
 776                                               domain->orig_tab[nstr].offset))
 777                     == 0)
 778               : domain->orig_sysdep_tab[nstr - nstrings].length > len
 779                 && (strcmp (msgid,
 780                             domain->orig_sysdep_tab[nstr - nstrings].pointer)
 781                     == 0))
 782             {
 783               act = nstr;
 784               goto found;
 785             }
 786
 787           if (idx >= domain->hash_size - incr)
 788             idx -= domain->hash_size - incr;
 789           else
 790             idx += incr;
 791         }
 792       /* NOTREACHED */
 793     }
 794   else
 795     {
 796       /* Try the default method:  binary search in the sorted array of
 797          messages.  */
 798       size_t top, bottom;
 799
 800       bottom = 0;
 801       top = nstrings;
 802       while (bottom < top)
 803         {
 804           int cmp_val;
 805
 806           act = (bottom + top) / 2;
 807           cmp_val = strcmp (msgid, (domain->data
 808                                     + W (domain->must_swap,
 809                                          domain->orig_tab[act].offset)));
 810           if (cmp_val < 0)
 811             top = act;
 812           else if (cmp_val > 0)
 813             bottom = act + 1;
 814           else
 815             goto found;
 816         }
 817       /* No translation was found.  */
 818       return NULL;
 819     }
 820
 821  found:
 822   /* The translation was found at index ACT.  If we have to convert the
 823      string to use a different character set, this is the time.  */
 824   if (act < nstrings)
 825     {
 826       result = (char *)
 827         (domain->data + W (domain->must_swap, domain->trans_tab[act].offset));
 828       resultlen = W (domain->must_swap, domain->trans_tab[act].length) + 1;
 829     }
 830   else
 831     {
 832       result = (char *) domain->trans_sysdep_tab[act - nstrings].pointer;
 833       resultlen = domain->trans_sysdep_tab[act - nstrings].length;
 834     }
 835
 836 #if defined _LIBC || HAVE_ICONV
 837   if (domain->codeset_cntr
 838       != (domainbinding != NULL ? domainbinding->codeset_cntr : 0))
 839     {
 840       /* The domain's codeset has changed through bind_textdomain_codeset()
 841          since the message catalog was initialized or last accessed.  We
 842          have to reinitialize the converter.  */
 843       _nl_free_domain_conv (domain);
 844       _nl_init_domain_conv (domain_file, domain, domainbinding);
 845     }
 846
 847   if (
 848 # ifdef _LIBC
 849       domain->conv != (__gconv_t) -1
 850 # else
 851 #  if HAVE_ICONV
 852       domain->conv != (iconv_t) -1
 853 #  endif
 854 # endif
 855       )
 856     {
 857       /* We are supposed to do a conversion.  First allocate an
 858          appropriate table with the same structure as the table
 859          of translations in the file, where we can put the pointers
 860          to the converted strings in.
 861          There is a slight complication with plural entries.  They
 862          are represented by consecutive NUL terminated strings.  We
 863          handle this case by converting RESULTLEN bytes, including
 864          NULs.  */
 865
 866       if (domain->conv_tab == NULL
 867           && ((domain->conv_tab =
 868                  (char **) calloc (nstrings + domain->n_sysdep_strings,
 869                                    sizeof (char *)))
 870               == NULL))
 871         /* Mark that we didn't succeed allocating a table.  */
 872         domain->conv_tab = (char **) -1;
 873
 874       if (__builtin_expect (domain->conv_tab == (char **) -1, 0))
 875         /* Nothing we can do, no more memory.  */
 876         goto converted;
 877
 878       if (domain->conv_tab[act] == NULL)
 879         {
 880           /* We haven't used this string so far, so it is not
 881              translated yet.  Do this now.  */
 882           /* We use a bit more efficient memory handling.
 883              We allocate always larger blocks which get used over
 884              time.  This is faster than many small allocations.   */
 885           __libc_lock_define_initialized (static, lock)
 886 # define INITIAL_BLOCK_SIZE     4080
 887           static unsigned char *freemem;
 888           static size_t freemem_size;
 889
 890           const unsigned char *inbuf;
 891           unsigned char *outbuf;
 892           int malloc_count;
 893 # ifndef _LIBC
 894           transmem_block_t *transmem_list = NULL;
 895 # endif
 896
 897           __libc_lock_lock (lock);
 898
 899           inbuf = (const unsigned char *) result;
 900           outbuf = freemem + sizeof (size_t);
 901
 902           malloc_count = 0;
 903           while (1)
 904             {
 905               transmem_block_t *newmem;
 906 # ifdef _LIBC
 907               size_t non_reversible;
 908               int res;
 909
 910               if (freemem_size < sizeof (size_t))
 911                 goto resize_freemem;
 912
 913               res = __gconv (domain->conv,
 914                              &inbuf, inbuf + resultlen,
 915                              &outbuf,
 916                              outbuf + freemem_size - sizeof (size_t),
 917                              &non_reversible);
 918
 919               if (res == __GCONV_OK || res == __GCONV_EMPTY_INPUT)
 920                 break;
 921
 922               if (res != __GCONV_FULL_OUTPUT)
 923                 {
 924                   __libc_lock_unlock (lock);
 925                   goto converted;
 926                 }
 927
 928               inbuf = result;
 929 # else
 930 #  if HAVE_ICONV
 931               const char *inptr = (const char *) inbuf;
 932               size_t inleft = resultlen;
 933               char *outptr = (char *) outbuf;
 934               size_t outleft;
 935
 936               if (freemem_size < sizeof (size_t))
 937                 goto resize_freemem;
 938
 939               outleft = freemem_size - sizeof (size_t);
 940               if (iconv (domain->conv,
 941                          (ICONV_CONST char **) &inptr, &inleft,
 942                          &outptr, &outleft)
 943                   != (size_t) (-1))
 944                 {
 945                   outbuf = (unsigned char *) outptr;
 946                   break;
 947                 }
 948               if (errno != E2BIG)
 949                 {
 950                   __libc_lock_unlock (lock);
 951                   goto converted;
 952                 }
 953 #  endif
 954 # endif
 955
 956             resize_freemem:
 957               /* We must allocate a new buffer or resize the old one.  */
 958               if (malloc_count > 0)
 959                 {
 960                   ++malloc_count;
 961                   freemem_size = malloc_count * INITIAL_BLOCK_SIZE;
 962                   newmem = (transmem_block_t *) realloc (transmem_list,
 963                                                          freemem_size);
 964 # ifdef _LIBC
 965                   if (newmem != NULL)
 966                     transmem_list = transmem_list->next;
 967                   else
 968                     {
 969                       struct transmem_list *old = transmem_list;
 970
 971                       transmem_list = transmem_list->next;
 972                       free (old);
 973                     }
 974 # endif
 975                 }
 976               else
 977                 {
 978                   malloc_count = 1;
 979                   freemem_size = INITIAL_BLOCK_SIZE;
 980                   newmem = (transmem_block_t *) malloc (freemem_size);
 981                 }
 982               if (__builtin_expect (newmem == NULL, 0))
 983                 {
 984                   freemem = NULL;
 985                   freemem_size = 0;
 986                   __libc_lock_unlock (lock);
 987                   goto converted;
 988                 }
 989
 990 # ifdef _LIBC
 991               /* Add the block to the list of blocks we have to free
 992                  at some point.  */
 993               newmem->next = transmem_list;
 994               transmem_list = newmem;
 995
 996               freemem = newmem->data;
 997               freemem_size -= offsetof (struct transmem_list, data);
 998 # else
 999               transmem_list = newmem;
1000               freemem = newmem;
1001 # endif
1002
1003               outbuf = freemem + sizeof (size_t);
1004             }
1005
1006           /* We have now in our buffer a converted string.  Put this
1007              into the table of conversions.  */
1008           *(size_t *) freemem = outbuf - freemem - sizeof (size_t);
1009           domain->conv_tab[act] = (char *) freemem;
1010           /* Shrink freemem, but keep it aligned.  */
1011           freemem_size -= outbuf - freemem;
1012           freemem = outbuf;
1013           freemem += freemem_size & (alignof (size_t) - 1);
1014           freemem_size = freemem_size & ~ (alignof (size_t) - 1);
1015
1016           __libc_lock_unlock (lock);
1017         }
1018
1019       /* Now domain->conv_tab[act] contains the translation of all
1020          the plural variants.  */
1021       result = domain->conv_tab[act] + sizeof (size_t);
1022       resultlen = *(size_t *) domain->conv_tab[act];
1023     }
1024
1025  converted:
1026   /* The result string is converted.  */
1027
1028 #endif /* _LIBC || HAVE_ICONV */
1029
1030   *lengthp = resultlen;
1031   return result;
1032 }
1033
1034
1035 /* Look up a plural variant.  */
1036 static char *
1037 internal_function
1038 plural_lookup (domain, n, translation, translation_len)
1039      struct loaded_l10nfile *domain;
1040      unsigned long int n;
1041      const char *translation;
1042      size_t translation_len;
1043 {
1044   struct loaded_domain *domaindata = (struct loaded_domain *) domain->data;
1045   unsigned long int index;
1046   const char *p;
1047
1048   index = plural_eval (domaindata->plural, n);
1049   if (index >= domaindata->nplurals)
1050     /* This should never happen.  It means the plural expression and the
1051        given maximum value do not match.  */
1052     index = 0;
1053
1054   /* Skip INDEX strings at TRANSLATION.  */
1055   p = translation;
1056   while (index-- > 0)
1057     {
1058 #ifdef _LIBC
1059       p = __rawmemchr (p, '\0');
1060 #else
1061       p = strchr (p, '\0');
1062 #endif
1063       /* And skip over the NUL byte.  */
1064       p++;
1065
1066       if (p >= translation + translation_len)
1067         /* This should never happen.  It means the plural expression
1068            evaluated to a value larger than the number of variants
1069            available for MSGID1.  */
1070         return (char *) translation;
1071     }
1072   return (char *) p;
1073 }
1074
1075 #ifndef _LIBC
1076 /* Return string representation of locale CATEGORY.  */
1077 static const char *
1078 internal_function
1079 category_to_name (category)
1080      int category;
1081 {
1082   const char *retval;
1083
1084   switch (category)
1085   {
1086 #ifdef LC_COLLATE
1087   case LC_COLLATE:
1088     retval = "LC_COLLATE";
1089     break;
1090 #endif
1091 #ifdef LC_CTYPE
1092   case LC_CTYPE:
1093     retval = "LC_CTYPE";
1094     break;
1095 #endif
1096 #ifdef LC_MONETARY
1097   case LC_MONETARY:
1098     retval = "LC_MONETARY";
1099     break;
1100 #endif
1101 #ifdef LC_NUMERIC
1102   case LC_NUMERIC:
1103     retval = "LC_NUMERIC";
1104     break;
1105 #endif
1106 #ifdef LC_TIME
1107   case LC_TIME:
1108     retval = "LC_TIME";
1109     break;
1110 #endif
1111 #ifdef LC_MESSAGES
1112   case LC_MESSAGES:
1113     retval = "LC_MESSAGES";
1114     break;
1115 #endif
1116 #ifdef LC_RESPONSE
1117   case LC_RESPONSE:
1118     retval = "LC_RESPONSE";
1119     break;
1120 #endif
1121 #ifdef LC_ALL
1122   case LC_ALL:
1123     /* This might not make sense but is perhaps better than any other
1124        value.  */
1125     retval = "LC_ALL";
1126     break;
1127 #endif
1128   default:
1129     /* If you have a better idea for a default value let me know.  */
1130     retval = "LC_XXX";
1131   }
1132
1133   return retval;
1134 }
1135 #endif
1136
1137 /* Guess value of current locale from value of the environment variables.  */
1138 static const char *
1139 internal_function
1140 guess_category_value (category, categoryname)
1141      int category;
1142      const char *categoryname;
1143 {
1144   const char *language;
1145   const char *retval;
1146
1147   /* The highest priority value is the `LANGUAGE' environment
1148      variable.  But we don't use the value if the currently selected
1149      locale is the C locale.  This is a GNU extension.  */
1150   language = getenv ("LANGUAGE");
1151   if (language != NULL && language[0] == '\0')
1152     language = NULL;
1153
1154   /* We have to proceed with the POSIX methods of looking to `LC_ALL',
1155      `LC_xxx', and `LANG'.  On some systems this can be done by the
1156      `setlocale' function itself.  */
1157 #ifdef _LIBC
1158   retval = __current_locale_name (category);
1159 #else
1160   retval = _nl_locale_name (category, categoryname);
1161 #endif
1162
1163   /* Ignore LANGUAGE if the locale is set to "C" because
1164      1. "C" locale usually uses the ASCII encoding, and most international
1165         messages use non-ASCII characters. These characters get displayed
1166         as question marks (if using glibc's iconv()) or as invalid 8-bit
1167         characters (because other iconv()s refuse to convert most non-ASCII
1168         characters to ASCII). In any case, the output is ugly.
1169      2. The precise output of some programs in the "C" locale is specified
1170         by POSIX and should not depend on environment variables like
1171         "LANGUAGE".  We allow such programs to use gettext().  */
1172   return language != NULL && strcmp (retval, "C") != 0 ? language : retval;
1173 }
1174
1175 /* @@ begin of epilog @@ */
1176
1177 /* We don't want libintl.a to depend on any other library.  So we
1178    avoid the non-standard function stpcpy.  In GNU C Library this
1179    function is available, though.  Also allow the symbol HAVE_STPCPY
1180    to be defined.  */
1181 #if !_LIBC && !HAVE_STPCPY
1182 static char *
1183 stpcpy (dest, src)
1184      char *dest;
1185      const char *src;
1186 {
1187   while ((*dest++ = *src++) != '\0')
1188     /* Do nothing. */ ;
1189   return dest - 1;
1190 }
1191 #endif
1192
1193 #if !_LIBC && !HAVE_MEMPCPY
1194 static void *
1195 mempcpy (dest, src, n)
1196      void *dest;
1197      const void *src;
1198      size_t n;
1199 {
1200   return (void *) ((char *) memcpy (dest, src, n) + n);
1201 }
1202 #endif
1203
1204
1205 #ifdef _LIBC
1206 /* If we want to free all resources we have to do some work at
1207    program's end.  */
1208 libc_freeres_fn (free_mem)
1209 {
1210   void *old;
1211
1212   while (_nl_domain_bindings != NULL)
1213     {
1214       struct binding *oldp = _nl_domain_bindings;
1215       _nl_domain_bindings = _nl_domain_bindings->next;
1216       if (oldp->dirname != INTUSE(_nl_default_dirname))
1217         /* Yes, this is a pointer comparison.  */
1218         free (oldp->dirname);
1219       free (oldp->codeset);
1220       free (oldp);
1221     }
1222
1223   if (_nl_current_default_domain != _nl_default_default_domain)
1224     /* Yes, again a pointer comparison.  */
1225     free ((char *) _nl_current_default_domain);
1226
1227   /* Remove the search tree with the known translations.  */
1228   __tdestroy (root, free);
1229   root = NULL;
1230
1231   while (transmem_list != NULL)
1232     {
1233       old = transmem_list;
1234       transmem_list = transmem_list->next;
1235       free (old);
1236     }
1237 }
1238 #endif