iconv/gconv_conf.c

   1 /* Handle configuration data.
   2    Copyright (C) 1997-2021 Free Software Foundation, Inc.
   3    This file is part of the GNU C Library.
   4    Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997.
   5
   6    The GNU C Library is free software; you can redistribute it and/or
   7    modify it under the terms of the GNU Lesser General Public
   8    License as published by the Free Software Foundation; either
   9    version 2.1 of the License, or (at your option) any later version.
  10
  11    The GNU C Library is distributed in the hope that it will be useful,
  12    but WITHOUT ANY WARRANTY; without even the implied warranty of
  13    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  14    Lesser General Public License for more details.
  15
  16    You should have received a copy of the GNU Lesser General Public
  17    License along with the GNU C Library; if not, see
  18    <https://www.gnu.org/licenses/>.  */
  19
  20 #include <assert.h>
  21 #include <ctype.h>
  22 #include <errno.h>
  23 #include <limits.h>
  24 #include <locale.h>
  25 #include <search.h>
  26 #include <stddef.h>
  27 #include <stdio.h>
  28 #include <stdio_ext.h>
  29 #include <stdlib.h>
  30 #include <string.h>
  31 #include <unistd.h>
  32 #include <sys/param.h>
  33
  34 #include <libc-lock.h>
  35 #include <gconv_int.h>
  36
  37
  38 /* This is the default path where we look for module lists.  */
  39 static const char default_gconv_path[] = GCONV_PATH;
  40
  41 /* Type to represent search path.  */
  42 struct path_elem
  43 {
  44   const char *name;
  45   size_t len;
  46 };
  47
  48 /* The path elements, as determined by the __gconv_get_path function.
  49    All path elements end in a slash.  */
  50 struct path_elem *__gconv_path_elem;
  51 /* Maximum length of a single path element in __gconv_path_elem.  */
  52 size_t __gconv_max_path_elem_len;
  53
  54 /* We use the following struct if we couldn't allocate memory.  */
  55 static const struct path_elem empty_path_elem = { NULL, 0 };
  56
  57 /* Name of the file containing the module information in the directories
  58    along the path.  */
  59 static const char gconv_conf_filename[] = "gconv-modules";
  60
  61 /* Filename extension for the modules.  */
  62 #ifndef MODULE_EXT
  63 # define MODULE_EXT ".so"
  64 #endif
  65 static const char gconv_module_ext[] = MODULE_EXT;
  66
  67 /* We have a few builtin transformations.  */
  68 static struct gconv_module builtin_modules[] =
  69 {
  70 #define BUILTIN_TRANSFORMATION(From, To, Cost, Name, Fct, BtowcFct, \
  71                                MinF, MaxF, MinT, MaxT) \
  72   {                                                                           \
  73     .from_string = From,                                                      \
  74     .to_string = To,                                                          \
  75     .cost_hi = Cost,                                                          \
  76     .cost_lo = INT_MAX,                                                       \
  77     .module_name = Name                                                       \
  78   },
  79 #define BUILTIN_ALIAS(From, To)
  80
  81 #include "gconv_builtin.h"
  82
  83 #undef BUILTIN_TRANSFORMATION
  84 #undef BUILTIN_ALIAS
  85 };
  86
  87 static const char builtin_aliases[] =
  88 {
  89 #define BUILTIN_TRANSFORMATION(From, To, Cost, Name, Fct, BtowcFct, \
  90                                MinF, MaxF, MinT, MaxT)
  91 #define BUILTIN_ALIAS(From, To) From "\0" To "\0"
  92
  93 #include "gconv_builtin.h"
  94
  95 #undef BUILTIN_TRANSFORMATION
  96 #undef BUILTIN_ALIAS
  97 };
  98
  99 #include <libio/libioP.h>
 100 #define __getdelim(line, len, c, fp) _IO_getdelim (line, len, c, fp)
 101
 102
 103 /* Value of the GCONV_PATH environment variable.  */
 104 const char *__gconv_path_envvar;
 105
 106
 107 /* Test whether there is already a matching module known.  */
 108 static int
 109 detect_conflict (const char *alias)
 110 {
 111   struct gconv_module *node = __gconv_modules_db;
 112
 113   while (node != NULL)
 114     {
 115       int cmpres = strcmp (alias, node->from_string);
 116
 117       if (cmpres == 0)
 118         /* We have a conflict.  */
 119         return 1;
 120       else if (cmpres < 0)
 121         node = node->left;
 122       else
 123         node = node->right;
 124     }
 125
 126   return node != NULL;
 127 }
 128
 129
 130 /* The actual code to add aliases.  */
 131 static void
 132 add_alias2 (const char *from, const char *to, const char *wp, void *modules)
 133 {
 134   /* Test whether this alias conflicts with any available module.  */
 135   if (detect_conflict (from))
 136     /* It does conflict, don't add the alias.  */
 137     return;
 138
 139   struct gconv_alias *new_alias = (struct gconv_alias *)
 140     malloc (sizeof (struct gconv_alias) + (wp - from));
 141   if (new_alias != NULL)
 142     {
 143       void **inserted;
 144
 145       new_alias->fromname = memcpy ((char *) new_alias
 146                                     + sizeof (struct gconv_alias),
 147                                     from, wp - from);
 148       new_alias->toname = new_alias->fromname + (to - from);
 149
 150       inserted = (void **) __tsearch (new_alias, &__gconv_alias_db,
 151                                       __gconv_alias_compare);
 152       if (inserted == NULL || *inserted != new_alias)
 153         /* Something went wrong, free this entry.  */
 154         free (new_alias);
 155     }
 156 }
 157
 158
 159 /* Add new alias.  */
 160 static void
 161 add_alias (char *rp, void *modules)
 162 {
 163   /* We now expect two more string.  The strings are normalized
 164      (converted to UPPER case) and strored in the alias database.  */
 165   char *from, *to, *wp;
 166
 167   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 168     ++rp;
 169   from = wp = rp;
 170   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 171     *wp++ = __toupper_l (*rp++, _nl_C_locobj_ptr);
 172   if (*rp == '\0')
 173     /* There is no `to' string on the line.  Ignore it.  */
 174     return;
 175   *wp++ = '\0';
 176   to = ++rp;
 177   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 178     ++rp;
 179   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 180     *wp++ = __toupper_l (*rp++, _nl_C_locobj_ptr);
 181   if (to == wp)
 182     /* No `to' string, ignore the line.  */
 183     return;
 184   *wp++ = '\0';
 185
 186   add_alias2 (from, to, wp, modules);
 187 }
 188
 189
 190 /* Insert a data structure for a new module in the search tree.  */
 191 static void
 192 insert_module (struct gconv_module *newp, int tobefreed)
 193 {
 194   struct gconv_module **rootp = &__gconv_modules_db;
 195
 196   while (*rootp != NULL)
 197     {
 198       struct gconv_module *root = *rootp;
 199       int cmpres;
 200
 201       cmpres = strcmp (newp->from_string, root->from_string);
 202       if (cmpres == 0)
 203         {
 204           /* Both strings are identical.  Insert the string at the
 205              end of the `same' list if it is not already there.  */
 206           while (strcmp (newp->from_string, root->from_string) != 0
 207                  || strcmp (newp->to_string, root->to_string) != 0)
 208             {
 209               rootp = &root->same;
 210               root = *rootp;
 211               if (root == NULL)
 212                 break;
 213             }
 214
 215           if (root != NULL)
 216             {
 217               /* This is a no new conversion.  But maybe the cost is
 218                  better.  */
 219               if (newp->cost_hi < root->cost_hi
 220                   || (newp->cost_hi == root->cost_hi
 221                       && newp->cost_lo < root->cost_lo))
 222                 {
 223                   newp->left = root->left;
 224                   newp->right = root->right;
 225                   newp->same = root->same;
 226                   *rootp = newp;
 227
 228                   free (root);
 229                 }
 230               else if (tobefreed)
 231                 free (newp);
 232               return;
 233             }
 234
 235           break;
 236         }
 237       else if (cmpres < 0)
 238         rootp = &root->left;
 239       else
 240         rootp = &root->right;
 241     }
 242
 243   /* Plug in the new node here.  */
 244   *rootp = newp;
 245 }
 246
 247
 248 /* Add new module.  */
 249 static void
 250 add_module (char *rp, const char *directory, size_t dir_len, void **modules,
 251             size_t *nmodules, int modcounter)
 252 {
 253   /* We expect now
 254      1. `from' name
 255      2. `to' name
 256      3. filename of the module
 257      4. an optional cost value
 258   */
 259   struct gconv_alias fake_alias;
 260   struct gconv_module *new_module;
 261   char *from, *to, *module, *wp;
 262   int need_ext;
 263   int cost_hi;
 264
 265   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 266     ++rp;
 267   from = rp;
 268   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 269     {
 270       *rp = __toupper_l (*rp, _nl_C_locobj_ptr);
 271       ++rp;
 272     }
 273   if (*rp == '\0')
 274     return;
 275   *rp++ = '\0';
 276   to = wp = rp;
 277   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 278     ++rp;
 279   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 280     *wp++ = __toupper_l (*rp++, _nl_C_locobj_ptr);
 281   if (*rp == '\0')
 282     return;
 283   *wp++ = '\0';
 284   do
 285     ++rp;
 286   while (__isspace_l (*rp, _nl_C_locobj_ptr));
 287   module = wp;
 288   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 289     *wp++ = *rp++;
 290   if (*rp == '\0')
 291     {
 292       /* There is no cost, use one by default.  */
 293       *wp++ = '\0';
 294       cost_hi = 1;
 295     }
 296   else
 297     {
 298       /* There might be a cost value.  */
 299       char *endp;
 300
 301       *wp++ = '\0';
 302       cost_hi = strtol (rp, &endp, 10);
 303       if (rp == endp || cost_hi < 1)
 304         /* No useful information.  */
 305         cost_hi = 1;
 306     }
 307
 308   if (module[0] == '\0')
 309     /* No module name given.  */
 310     return;
 311   if (module[0] == '/')
 312     dir_len = 0;
 313
 314   /* See whether we must add the ending.  */
 315   need_ext = 0;
 316   if (wp - module < (ptrdiff_t) sizeof (gconv_module_ext)
 317       || memcmp (wp - sizeof (gconv_module_ext), gconv_module_ext,
 318                  sizeof (gconv_module_ext)) != 0)
 319     /* We must add the module extension.  */
 320     need_ext = sizeof (gconv_module_ext) - 1;
 321
 322   /* See whether we have already an alias with this name defined.  */
 323   fake_alias.fromname = strndupa (from, to - from);
 324
 325   if (__tfind (&fake_alias, &__gconv_alias_db, __gconv_alias_compare) != NULL)
 326     /* This module duplicates an alias.  */
 327     return;
 328
 329   new_module = (struct gconv_module *) calloc (1,
 330                                                sizeof (struct gconv_module)
 331                                                + (wp - from)
 332                                                + dir_len + need_ext);
 333   if (new_module != NULL)
 334     {
 335       char *tmp;
 336
 337       new_module->from_string = tmp = (char *) (new_module + 1);
 338       tmp = __mempcpy (tmp, from, to - from);
 339
 340       new_module->to_string = tmp;
 341       tmp = __mempcpy (tmp, to, module - to);
 342
 343       new_module->cost_hi = cost_hi;
 344       new_module->cost_lo = modcounter;
 345
 346       new_module->module_name = tmp;
 347
 348       if (dir_len != 0)
 349         tmp = __mempcpy (tmp, directory, dir_len);
 350
 351       tmp = __mempcpy (tmp, module, wp - module);
 352
 353       if (need_ext)
 354         memcpy (tmp - 1, gconv_module_ext, sizeof (gconv_module_ext));
 355
 356       /* Now insert the new module data structure in our search tree.  */
 357       insert_module (new_module, 1);
 358     }
 359 }
 360
 361
 362 /* Read the next configuration file.  */
 363 static void
 364 read_conf_file (const char *filename, const char *directory, size_t dir_len,
 365                 void **modules, size_t *nmodules)
 366 {
 367   /* Note the file is opened with cancellation in the I/O functions
 368      disabled.  */
 369   FILE *fp = fopen (filename, "rce");
 370   char *line = NULL;
 371   size_t line_len = 0;
 372   static int modcounter;
 373
 374   /* Don't complain if a file is not present or readable, simply silently
 375      ignore it.  */
 376   if (fp == NULL)
 377     return;
 378
 379   /* No threads reading from this stream.  */
 380   __fsetlocking (fp, FSETLOCKING_BYCALLER);
 381
 382   /* Process the known entries of the file.  Comments start with `#' and
 383      end with the end of the line.  Empty lines are ignored.  */
 384   while (!__feof_unlocked (fp))
 385     {
 386       char *rp, *endp, *word;
 387       ssize_t n = __getdelim (&line, &line_len, '\n', fp);
 388       if (n < 0)
 389         /* An error occurred.  */
 390         break;
 391
 392       rp = line;
 393       /* Terminate the line (excluding comments or newline) by an NUL byte
 394          to simplify the following code.  */
 395       endp = strchr (rp, '#');
 396       if (endp != NULL)
 397         *endp = '\0';
 398       else
 399         if (rp[n - 1] == '\n')
 400           rp[n - 1] = '\0';
 401
 402       while (__isspace_l (*rp, _nl_C_locobj_ptr))
 403         ++rp;
 404
 405       /* If this is an empty line go on with the next one.  */
 406       if (rp == endp)
 407         continue;
 408
 409       word = rp;
 410       while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 411         ++rp;
 412
 413       if (rp - word == sizeof ("alias") - 1
 414           && memcmp (word, "alias", sizeof ("alias") - 1) == 0)
 415         add_alias (rp, *modules);
 416       else if (rp - word == sizeof ("module") - 1
 417                && memcmp (word, "module", sizeof ("module") - 1) == 0)
 418         add_module (rp, directory, dir_len, modules, nmodules, modcounter++);
 419       /* else */
 420         /* Otherwise ignore the line.  */
 421     }
 422
 423   free (line);
 424
 425   fclose (fp);
 426 }
 427
 428
 429 /* Determine the directories we are looking for data in.  This function should
 430    only be called from __gconv_read_conf.  */
 431 static void
 432 __gconv_get_path (void)
 433 {
 434   struct path_elem *result;
 435
 436   /* This function is only ever called when __gconv_path_elem is NULL.  */
 437   result = __gconv_path_elem;
 438   assert (result == NULL);
 439
 440   /* Determine the complete path first.  */
 441   char *gconv_path;
 442   size_t gconv_path_len;
 443   char *elem;
 444   char *oldp;
 445   char *cp;
 446   int nelems;
 447   char *cwd;
 448   size_t cwdlen;
 449
 450   if (__gconv_path_envvar == NULL)
 451     {
 452       /* No user-defined path.  Make a modifiable copy of the
 453          default path.  */
 454       gconv_path = strdupa (default_gconv_path);
 455       gconv_path_len = sizeof (default_gconv_path);
 456       cwd = NULL;
 457       cwdlen = 0;
 458     }
 459   else
 460     {
 461       /* Append the default path to the user-defined path.  */
 462       size_t user_len = strlen (__gconv_path_envvar);
 463
 464       gconv_path_len = user_len + 1 + sizeof (default_gconv_path);
 465       gconv_path = alloca (gconv_path_len);
 466       __mempcpy (__mempcpy (__mempcpy (gconv_path, __gconv_path_envvar,
 467                                        user_len),
 468                             ":", 1),
 469                  default_gconv_path, sizeof (default_gconv_path));
 470       cwd = __getcwd (NULL, 0);
 471       cwdlen = __glibc_unlikely (cwd == NULL) ? 0 : strlen (cwd);
 472     }
 473   assert (default_gconv_path[0] == '/');
 474
 475   /* In a first pass we calculate the number of elements.  */
 476   oldp = NULL;
 477   cp = strchr (gconv_path, ':');
 478   nelems = 1;
 479   while (cp != NULL)
 480     {
 481       if (cp != oldp + 1)
 482         ++nelems;
 483       oldp = cp;
 484       cp = strchr (cp + 1, ':');
 485     }
 486
 487   /* Allocate the memory for the result.  */
 488   result = malloc ((nelems + 1)
 489                               * sizeof (struct path_elem)
 490                               + gconv_path_len + nelems
 491                               + (nelems - 1) * (cwdlen + 1));
 492   if (result != NULL)
 493     {
 494       char *strspace = (char *) &result[nelems + 1];
 495       int n = 0;
 496
 497       /* Separate the individual parts.  */
 498       __gconv_max_path_elem_len = 0;
 499       elem = __strtok_r (gconv_path, ":", &gconv_path);
 500       assert (elem != NULL);
 501       do
 502         {
 503           result[n].name = strspace;
 504           if (elem[0] != '/')
 505             {
 506               assert (cwd != NULL);
 507               strspace = __mempcpy (strspace, cwd, cwdlen);
 508               *strspace++ = '/';
 509             }
 510           strspace = __stpcpy (strspace, elem);
 511           if (strspace[-1] != '/')
 512             *strspace++ = '/';
 513
 514           result[n].len = strspace - result[n].name;
 515           if (result[n].len > __gconv_max_path_elem_len)
 516             __gconv_max_path_elem_len = result[n].len;
 517
 518           *strspace++ = '\0';
 519           ++n;
 520         }
 521       while ((elem = __strtok_r (NULL, ":", &gconv_path)) != NULL);
 522
 523       result[n].name = NULL;
 524       result[n].len = 0;
 525     }
 526
 527   __gconv_path_elem = result ?: (struct path_elem *) &empty_path_elem;
 528
 529   free (cwd);
 530 }
 531
 532
 533 /* Read all configuration files found in the user-specified and the default
 534    path.  This function should only be called once during the program's
 535    lifetime.  It disregards locking and synchronization because its only
 536    caller, __gconv_load_conf, handles this.  */
 537 static void
 538 __gconv_read_conf (void)
 539 {
 540   void *modules = NULL;
 541   size_t nmodules = 0;
 542   int save_errno = errno;
 543   size_t cnt;
 544
 545   /* First see whether we should use the cache.  */
 546   if (__gconv_load_cache () == 0)
 547     {
 548       /* Yes, we are done.  */
 549       __set_errno (save_errno);
 550       return;
 551     }
 552
 553 #ifndef STATIC_GCONV
 554   /* Find out where we have to look.  */
 555   __gconv_get_path ();
 556
 557   for (cnt = 0; __gconv_path_elem[cnt].name != NULL; ++cnt)
 558     {
 559       const char *elem = __gconv_path_elem[cnt].name;
 560       size_t elem_len = __gconv_path_elem[cnt].len;
 561       char *filename;
 562
 563       /* No slash needs to be inserted between elem and gconv_conf_filename;
 564          elem already ends in a slash.  */
 565       filename = alloca (elem_len + sizeof (gconv_conf_filename));
 566       __mempcpy (__mempcpy (filename, elem, elem_len),
 567                  gconv_conf_filename, sizeof (gconv_conf_filename));
 568
 569       /* Read the next configuration file.  */
 570       read_conf_file (filename, elem, elem_len, &modules, &nmodules);
 571     }
 572 #endif
 573
 574   /* Add the internal modules.  */
 575   for (cnt = 0; cnt < sizeof (builtin_modules) / sizeof (builtin_modules[0]);
 576        ++cnt)
 577     {
 578       struct gconv_alias fake_alias;
 579
 580       fake_alias.fromname = (char *) builtin_modules[cnt].from_string;
 581
 582       if (__tfind (&fake_alias, &__gconv_alias_db, __gconv_alias_compare)
 583           != NULL)
 584         /* It'll conflict so don't add it.  */
 585         continue;
 586
 587       insert_module (&builtin_modules[cnt], 0);
 588     }
 589
 590   /* Add aliases for builtin conversions.  */
 591   const char *cp = builtin_aliases;
 592   do
 593     {
 594       const char *from = cp;
 595       const char *to = __rawmemchr (from, '\0') + 1;
 596       cp = __rawmemchr (to, '\0') + 1;
 597
 598       add_alias2 (from, to, cp, modules);
 599     }
 600   while (*cp != '\0');
 601
 602   /* Restore the error number.  */
 603   __set_errno (save_errno);
 604 }
 605
 606
 607 /* This "once" variable is used to do a one-time load of the configuration.  */
 608 __libc_once_define (static, once);
 609
 610
 611 /* Read all configuration files found in the user-specified and the default
 612    path, but do it only "once" using __gconv_read_conf to do the actual
 613    work.  This is the function that must be called when reading iconv
 614    configuration.  */
 615 void
 616 __gconv_load_conf (void)
 617 {
 618   __libc_once (once, __gconv_read_conf);
 619 }
 620
 621
 622 /* Free all resources if necessary.  */
 623 libc_freeres_fn (free_mem)
 624 {
 625   if (__gconv_path_elem != NULL && __gconv_path_elem != &empty_path_elem)
 626     free ((void *) __gconv_path_elem);
 627 }