iconv/gconv_conf.c

   1 /* Handle configuration data.
   2    Copyright (C) 1997-2023 Free Software Foundation, Inc.
   3    This file is part of the GNU C Library.
   4
   5    The GNU C Library is free software; you can redistribute it and/or
   6    modify it under the terms of the GNU Lesser General Public
   7    License as published by the Free Software Foundation; either
   8    version 2.1 of the License, or (at your option) any later version.
   9
  10    The GNU C Library 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 GNU
  13    Lesser General Public License for more details.
  14
  15    You should have received a copy of the GNU Lesser General Public
  16    License along with the GNU C Library; if not, see
  17    <https://www.gnu.org/licenses/>.  */
  18
  19 #include <assert.h>
  20 #include <ctype.h>
  21 #include <errno.h>
  22 #include <limits.h>
  23 #include <locale.h>
  24 #include <search.h>
  25 #include <stddef.h>
  26 #include <stdio.h>
  27 #include <stdio_ext.h>
  28 #include <stdlib.h>
  29 #include <string.h>
  30 #include <unistd.h>
  31 #include <sys/param.h>
  32
  33 #include <libc-lock.h>
  34 #include <gconv_int.h>
  35 #include <gconv_parseconfdir.h>
  36
  37 /* This is the default path where we look for module lists.  */
  38 static const char default_gconv_path[] = GCONV_PATH;
  39
  40 /* Type to represent search path.  */
  41 struct path_elem
  42 {
  43   const char *name;
  44   size_t len;
  45 };
  46
  47 /* The path elements, as determined by the __gconv_get_path function.
  48    All path elements end in a slash.  */
  49 struct path_elem *__gconv_path_elem;
  50 /* Maximum length of a single path element in __gconv_path_elem.  */
  51 size_t __gconv_max_path_elem_len;
  52
  53 /* We use the following struct if we couldn't allocate memory.  */
  54 static const struct path_elem empty_path_elem = { NULL, 0 };
  55
  56 /* Filename extension for the modules.  */
  57 #ifndef MODULE_EXT
  58 # define MODULE_EXT ".so"
  59 #endif
  60 static const char gconv_module_ext[] = MODULE_EXT;
  61
  62 /* We have a few builtin transformations.  */
  63 static struct gconv_module builtin_modules[] =
  64 {
  65 #define BUILTIN_TRANSFORMATION(From, To, Cost, Name, Fct, BtowcFct, \
  66                                MinF, MaxF, MinT, MaxT) \
  67   {                                                                           \
  68     .from_string = From,                                                      \
  69     .to_string = To,                                                          \
  70     .cost_hi = Cost,                                                          \
  71     .cost_lo = INT_MAX,                                                       \
  72     .module_name = Name                                                       \
  73   },
  74 #define BUILTIN_ALIAS(From, To)
  75
  76 #include "gconv_builtin.h"
  77
  78 #undef BUILTIN_TRANSFORMATION
  79 #undef BUILTIN_ALIAS
  80 };
  81
  82 static const char builtin_aliases[] =
  83 {
  84 #define BUILTIN_TRANSFORMATION(From, To, Cost, Name, Fct, BtowcFct, \
  85                                MinF, MaxF, MinT, MaxT)
  86 #define BUILTIN_ALIAS(From, To) From "\0" To "\0"
  87
  88 #include "gconv_builtin.h"
  89
  90 #undef BUILTIN_TRANSFORMATION
  91 #undef BUILTIN_ALIAS
  92 };
  93
  94
  95 /* Value of the GCONV_PATH environment variable.  */
  96 const char *__gconv_path_envvar;
  97
  98
  99 /* Test whether there is already a matching module known.  */
 100 static int
 101 detect_conflict (const char *alias)
 102 {
 103   struct gconv_module *node = __gconv_modules_db;
 104
 105   while (node != NULL)
 106     {
 107       int cmpres = strcmp (alias, node->from_string);
 108
 109       if (cmpres == 0)
 110         /* We have a conflict.  */
 111         return 1;
 112       else if (cmpres < 0)
 113         node = node->left;
 114       else
 115         node = node->right;
 116     }
 117
 118   return node != NULL;
 119 }
 120
 121
 122 /* The actual code to add aliases.  */
 123 static void
 124 add_alias2 (const char *from, const char *to, const char *wp)
 125 {
 126   /* Test whether this alias conflicts with any available module.  */
 127   if (detect_conflict (from))
 128     /* It does conflict, don't add the alias.  */
 129     return;
 130
 131   struct gconv_alias *new_alias = (struct gconv_alias *)
 132     malloc (sizeof (struct gconv_alias) + (wp - from));
 133   if (new_alias != NULL)
 134     {
 135       void **inserted;
 136
 137       new_alias->fromname = memcpy ((char *) new_alias
 138                                     + sizeof (struct gconv_alias),
 139                                     from, wp - from);
 140       new_alias->toname = new_alias->fromname + (to - from);
 141
 142       inserted = (void **) __tsearch (new_alias, &__gconv_alias_db,
 143                                       __gconv_alias_compare);
 144       if (inserted == NULL || *inserted != new_alias)
 145         /* Something went wrong, free this entry.  */
 146         free (new_alias);
 147     }
 148 }
 149
 150
 151 /* Add new alias.  */
 152 static void
 153 add_alias (char *rp)
 154 {
 155   /* We now expect two more string.  The strings are normalized
 156      (converted to UPPER case) and strored in the alias database.  */
 157   char *from, *to, *wp;
 158
 159   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 160     ++rp;
 161   from = wp = rp;
 162   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 163     *wp++ = __toupper_l (*rp++, _nl_C_locobj_ptr);
 164   if (*rp == '\0')
 165     /* There is no `to' string on the line.  Ignore it.  */
 166     return;
 167   *wp++ = '\0';
 168   to = ++rp;
 169   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 170     ++rp;
 171   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 172     *wp++ = __toupper_l (*rp++, _nl_C_locobj_ptr);
 173   if (to == wp)
 174     /* No `to' string, ignore the line.  */
 175     return;
 176   *wp++ = '\0';
 177
 178   add_alias2 (from, to, wp);
 179 }
 180
 181
 182 /* Insert a data structure for a new module in the search tree.  */
 183 static void
 184 insert_module (struct gconv_module *newp, int tobefreed)
 185 {
 186   struct gconv_module **rootp = &__gconv_modules_db;
 187
 188   while (*rootp != NULL)
 189     {
 190       struct gconv_module *root = *rootp;
 191       int cmpres;
 192
 193       cmpres = strcmp (newp->from_string, root->from_string);
 194       if (cmpres == 0)
 195         {
 196           /* Both strings are identical.  Insert the string at the
 197              end of the `same' list if it is not already there.  */
 198           while (strcmp (newp->from_string, root->from_string) != 0
 199                  || strcmp (newp->to_string, root->to_string) != 0)
 200             {
 201               rootp = &root->same;
 202               root = *rootp;
 203               if (root == NULL)
 204                 break;
 205             }
 206
 207           if (root != NULL)
 208             {
 209               /* This is a no new conversion.  But maybe the cost is
 210                  better.  */
 211               if (newp->cost_hi < root->cost_hi
 212                   || (newp->cost_hi == root->cost_hi
 213                       && newp->cost_lo < root->cost_lo))
 214                 {
 215                   newp->left = root->left;
 216                   newp->right = root->right;
 217                   newp->same = root->same;
 218                   *rootp = newp;
 219
 220                   free (root);
 221                 }
 222               else if (tobefreed)
 223                 free (newp);
 224               return;
 225             }
 226
 227           break;
 228         }
 229       else if (cmpres < 0)
 230         rootp = &root->left;
 231       else
 232         rootp = &root->right;
 233     }
 234
 235   /* Plug in the new node here.  */
 236   *rootp = newp;
 237 }
 238
 239
 240 /* Add new module.  */
 241 static void
 242 add_module (char *rp, const char *directory, size_t dir_len, int modcounter)
 243 {
 244   /* We expect now
 245      1. `from' name
 246      2. `to' name
 247      3. filename of the module
 248      4. an optional cost value
 249   */
 250   struct gconv_alias fake_alias;
 251   struct gconv_module *new_module;
 252   char *from, *to, *module, *wp;
 253   int need_ext;
 254   int cost_hi;
 255
 256   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 257     ++rp;
 258   from = rp;
 259   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 260     {
 261       *rp = __toupper_l (*rp, _nl_C_locobj_ptr);
 262       ++rp;
 263     }
 264   if (*rp == '\0')
 265     return;
 266   *rp++ = '\0';
 267   to = wp = rp;
 268   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 269     ++rp;
 270   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 271     *wp++ = __toupper_l (*rp++, _nl_C_locobj_ptr);
 272   if (*rp == '\0')
 273     return;
 274   *wp++ = '\0';
 275   do
 276     ++rp;
 277   while (__isspace_l (*rp, _nl_C_locobj_ptr));
 278   module = wp;
 279   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 280     *wp++ = *rp++;
 281   if (*rp == '\0')
 282     {
 283       /* There is no cost, use one by default.  */
 284       *wp++ = '\0';
 285       cost_hi = 1;
 286     }
 287   else
 288     {
 289       /* There might be a cost value.  */
 290       char *endp;
 291
 292       *wp++ = '\0';
 293       cost_hi = strtol (rp, &endp, 10);
 294       if (rp == endp || cost_hi < 1)
 295         /* No useful information.  */
 296         cost_hi = 1;
 297     }
 298
 299   if (module[0] == '\0')
 300     /* No module name given.  */
 301     return;
 302   if (module[0] == '/')
 303     dir_len = 0;
 304
 305   /* See whether we must add the ending.  */
 306   need_ext = 0;
 307   if (wp - module < (ptrdiff_t) sizeof (gconv_module_ext)
 308       || memcmp (wp - sizeof (gconv_module_ext), gconv_module_ext,
 309                  sizeof (gconv_module_ext)) != 0)
 310     /* We must add the module extension.  */
 311     need_ext = sizeof (gconv_module_ext) - 1;
 312
 313   /* See whether we have already an alias with this name defined.  */
 314   fake_alias.fromname = strndupa (from, to - from);
 315
 316   if (__tfind (&fake_alias, &__gconv_alias_db, __gconv_alias_compare) != NULL)
 317     /* This module duplicates an alias.  */
 318     return;
 319
 320   new_module = (struct gconv_module *) calloc (1,
 321                                                sizeof (struct gconv_module)
 322                                                + (wp - from)
 323                                                + dir_len + need_ext);
 324   if (new_module != NULL)
 325     {
 326       char *tmp;
 327
 328       new_module->from_string = tmp = (char *) (new_module + 1);
 329       tmp = __mempcpy (tmp, from, to - from);
 330
 331       new_module->to_string = tmp;
 332       tmp = __mempcpy (tmp, to, module - to);
 333
 334       new_module->cost_hi = cost_hi;
 335       new_module->cost_lo = modcounter;
 336
 337       new_module->module_name = tmp;
 338
 339       if (dir_len != 0)
 340         tmp = __mempcpy (tmp, directory, dir_len);
 341
 342       tmp = __mempcpy (tmp, module, wp - module);
 343
 344       if (need_ext)
 345         memcpy (tmp - 1, gconv_module_ext, sizeof (gconv_module_ext));
 346
 347       /* Now insert the new module data structure in our search tree.  */
 348       insert_module (new_module, 1);
 349     }
 350 }
 351
 352
 353 /* Determine the directories we are looking for data in.  This function should
 354    only be called from __gconv_read_conf.  */
 355 static void
 356 __gconv_get_path (void)
 357 {
 358   struct path_elem *result;
 359
 360   /* This function is only ever called when __gconv_path_elem is NULL.  */
 361   result = __gconv_path_elem;
 362   assert (result == NULL);
 363
 364   /* Determine the complete path first.  */
 365   char *gconv_path;
 366   size_t gconv_path_len;
 367   char *elem;
 368   char *oldp;
 369   char *cp;
 370   int nelems;
 371   char *cwd;
 372   size_t cwdlen;
 373
 374   if (__gconv_path_envvar == NULL)
 375     {
 376       /* No user-defined path.  Make a modifiable copy of the
 377          default path.  */
 378       gconv_path = strdupa (default_gconv_path);
 379       gconv_path_len = sizeof (default_gconv_path);
 380       cwd = NULL;
 381       cwdlen = 0;
 382     }
 383   else
 384     {
 385       /* Append the default path to the user-defined path.  */
 386       size_t user_len = strlen (__gconv_path_envvar);
 387
 388       gconv_path_len = user_len + 1 + sizeof (default_gconv_path);
 389       gconv_path = alloca (gconv_path_len);
 390       __mempcpy (__mempcpy (__mempcpy (gconv_path, __gconv_path_envvar,
 391                                        user_len),
 392                             ":", 1),
 393                  default_gconv_path, sizeof (default_gconv_path));
 394       cwd = __getcwd (NULL, 0);
 395       cwdlen = __glibc_unlikely (cwd == NULL) ? 0 : strlen (cwd);
 396     }
 397   assert (default_gconv_path[0] == '/');
 398
 399   /* In a first pass we calculate the number of elements.  */
 400   oldp = NULL;
 401   cp = strchr (gconv_path, ':');
 402   nelems = 1;
 403   while (cp != NULL)
 404     {
 405       if (cp != oldp + 1)
 406         ++nelems;
 407       oldp = cp;
 408       cp = strchr (cp + 1, ':');
 409     }
 410
 411   /* Allocate the memory for the result.  */
 412   result = malloc ((nelems + 1)
 413                               * sizeof (struct path_elem)
 414                               + gconv_path_len + nelems
 415                               + (nelems - 1) * (cwdlen + 1));
 416   if (result != NULL)
 417     {
 418       char *strspace = (char *) &result[nelems + 1];
 419       int n = 0;
 420
 421       /* Separate the individual parts.  */
 422       __gconv_max_path_elem_len = 0;
 423       elem = __strtok_r (gconv_path, ":", &gconv_path);
 424       assert (elem != NULL);
 425       do
 426         {
 427           result[n].name = strspace;
 428           if (elem[0] != '/')
 429             {
 430               assert (cwd != NULL);
 431               strspace = __mempcpy (strspace, cwd, cwdlen);
 432               *strspace++ = '/';
 433             }
 434           strspace = __stpcpy (strspace, elem);
 435           if (strspace[-1] != '/')
 436             *strspace++ = '/';
 437
 438           result[n].len = strspace - result[n].name;
 439           if (result[n].len > __gconv_max_path_elem_len)
 440             __gconv_max_path_elem_len = result[n].len;
 441
 442           *strspace++ = '\0';
 443           ++n;
 444         }
 445       while ((elem = __strtok_r (NULL, ":", &gconv_path)) != NULL);
 446
 447       result[n].name = NULL;
 448       result[n].len = 0;
 449     }
 450
 451   __gconv_path_elem = result ?: (struct path_elem *) &empty_path_elem;
 452
 453   free (cwd);
 454 }
 455
 456
 457 /* Read all configuration files found in the user-specified and the default
 458    path.  This function should only be called once during the program's
 459    lifetime.  It disregards locking and synchronization because its only
 460    caller, __gconv_load_conf, handles this.  */
 461 static void
 462 __gconv_read_conf (void)
 463 {
 464   int save_errno = errno;
 465   size_t cnt;
 466
 467   /* First see whether we should use the cache.  */
 468   if (__gconv_load_cache () == 0)
 469     {
 470       /* Yes, we are done.  */
 471       __set_errno (save_errno);
 472       return;
 473     }
 474
 475 #ifndef STATIC_GCONV
 476   /* Find out where we have to look.  */
 477   __gconv_get_path ();
 478
 479   for (cnt = 0; __gconv_path_elem[cnt].name != NULL; ++cnt)
 480     gconv_parseconfdir (NULL, __gconv_path_elem[cnt].name,
 481                         __gconv_path_elem[cnt].len);
 482 #endif
 483
 484   /* Add the internal modules.  */
 485   for (cnt = 0; cnt < sizeof (builtin_modules) / sizeof (builtin_modules[0]);
 486        ++cnt)
 487     {
 488       struct gconv_alias fake_alias;
 489
 490       fake_alias.fromname = (char *) builtin_modules[cnt].from_string;
 491
 492       if (__tfind (&fake_alias, &__gconv_alias_db, __gconv_alias_compare)
 493           != NULL)
 494         /* It'll conflict so don't add it.  */
 495         continue;
 496
 497       insert_module (&builtin_modules[cnt], 0);
 498     }
 499
 500   /* Add aliases for builtin conversions.  */
 501   const char *cp = builtin_aliases;
 502   do
 503     {
 504       const char *from = cp;
 505       const char *to = strchr (from, '\0') + 1;
 506       cp = strchr (to, '\0') + 1;
 507
 508       add_alias2 (from, to, cp);
 509     }
 510   while (*cp != '\0');
 511
 512   /* Restore the error number.  */
 513   __set_errno (save_errno);
 514 }
 515
 516
 517 /* This "once" variable is used to do a one-time load of the configuration.  */
 518 __libc_once_define (static, once);
 519
 520
 521 /* Read all configuration files found in the user-specified and the default
 522    path, but do it only "once" using __gconv_read_conf to do the actual
 523    work.  This is the function that must be called when reading iconv
 524    configuration.  */
 525 void
 526 __gconv_load_conf (void)
 527 {
 528   __libc_once (once, __gconv_read_conf);
 529 }
 530
 531
 532 /* Free all resources if necessary.  */
 533 libc_freeres_fn (free_mem)
 534 {
 535   if (__gconv_path_elem != NULL && __gconv_path_elem != &empty_path_elem)
 536     free ((void *) __gconv_path_elem);
 537 }