iconv/gconv_conf.c

   1 /* Handle configuration data.
   2    Copyright (C) 1997-2018 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    <http://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 /* The path elements, as determined by the __gconv_get_path function.
  42    All path elements end in a slash.  */
  43 struct path_elem *__gconv_path_elem;
  44 /* Maximum length of a single path element in __gconv_path_elem.  */
  45 size_t __gconv_max_path_elem_len;
  46
  47 /* We use the following struct if we couldn't allocate memory.  */
  48 static const struct path_elem empty_path_elem = { NULL, 0 };
  49
  50 /* Name of the file containing the module information in the directories
  51    along the path.  */
  52 static const char gconv_conf_filename[] = "gconv-modules";
  53
  54 /* Filename extension for the modules.  */
  55 #ifndef MODULE_EXT
  56 # define MODULE_EXT ".so"
  57 #endif
  58 static const char gconv_module_ext[] = MODULE_EXT;
  59
  60 /* We have a few builtin transformations.  */
  61 static struct gconv_module builtin_modules[] =
  62 {
  63 #define BUILTIN_TRANSFORMATION(From, To, Cost, Name, Fct, BtowcFct, \
  64                                MinF, MaxF, MinT, MaxT) \
  65   {                                                                           \
  66     .from_string = From,                                                      \
  67     .to_string = To,                                                          \
  68     .cost_hi = Cost,                                                          \
  69     .cost_lo = INT_MAX,                                                       \
  70     .module_name = Name                                                       \
  71   },
  72 #define BUILTIN_ALIAS(From, To)
  73
  74 #include "gconv_builtin.h"
  75
  76 #undef BUILTIN_TRANSFORMATION
  77 #undef BUILTIN_ALIAS
  78 };
  79
  80 static const char builtin_aliases[] =
  81 {
  82 #define BUILTIN_TRANSFORMATION(From, To, Cost, Name, Fct, BtowcFct, \
  83                                MinF, MaxF, MinT, MaxT)
  84 #define BUILTIN_ALIAS(From, To) From "\0" To "\0"
  85
  86 #include "gconv_builtin.h"
  87
  88 #undef BUILTIN_TRANSFORMATION
  89 #undef BUILTIN_ALIAS
  90 };
  91
  92 #include <libio/libioP.h>
  93 #define __getdelim(line, len, c, fp) _IO_getdelim (line, len, c, fp)
  94
  95
  96 /* Value of the GCONV_PATH environment variable.  */
  97 const char *__gconv_path_envvar;
  98
  99
 100 /* Test whether there is already a matching module known.  */
 101 static int
 102 detect_conflict (const char *alias)
 103 {
 104   struct gconv_module *node = __gconv_modules_db;
 105
 106   while (node != NULL)
 107     {
 108       int cmpres = strcmp (alias, node->from_string);
 109
 110       if (cmpres == 0)
 111         /* We have a conflict.  */
 112         return 1;
 113       else if (cmpres < 0)
 114         node = node->left;
 115       else
 116         node = node->right;
 117     }
 118
 119   return node != NULL;
 120 }
 121
 122
 123 /* The actual code to add aliases.  */
 124 static void
 125 add_alias2 (const char *from, const char *to, const char *wp, void *modules)
 126 {
 127   /* Test whether this alias conflicts with any available module.  */
 128   if (detect_conflict (from))
 129     /* It does conflict, don't add the alias.  */
 130     return;
 131
 132   struct gconv_alias *new_alias = (struct gconv_alias *)
 133     malloc (sizeof (struct gconv_alias) + (wp - from));
 134   if (new_alias != NULL)
 135     {
 136       void **inserted;
 137
 138       new_alias->fromname = memcpy ((char *) new_alias
 139                                     + sizeof (struct gconv_alias),
 140                                     from, wp - from);
 141       new_alias->toname = new_alias->fromname + (to - from);
 142
 143       inserted = (void **) __tsearch (new_alias, &__gconv_alias_db,
 144                                       __gconv_alias_compare);
 145       if (inserted == NULL || *inserted != new_alias)
 146         /* Something went wrong, free this entry.  */
 147         free (new_alias);
 148     }
 149 }
 150
 151
 152 /* Add new alias.  */
 153 static void
 154 add_alias (char *rp, void *modules)
 155 {
 156   /* We now expect two more string.  The strings are normalized
 157      (converted to UPPER case) and strored in the alias database.  */
 158   char *from, *to, *wp;
 159
 160   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 161     ++rp;
 162   from = wp = rp;
 163   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 164     *wp++ = __toupper_l (*rp++, _nl_C_locobj_ptr);
 165   if (*rp == '\0')
 166     /* There is no `to' string on the line.  Ignore it.  */
 167     return;
 168   *wp++ = '\0';
 169   to = ++rp;
 170   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 171     ++rp;
 172   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 173     *wp++ = __toupper_l (*rp++, _nl_C_locobj_ptr);
 174   if (to == wp)
 175     /* No `to' string, ignore the line.  */
 176     return;
 177   *wp++ = '\0';
 178
 179   add_alias2 (from, to, wp, modules);
 180 }
 181
 182
 183 /* Insert a data structure for a new module in the search tree.  */
 184 static void
 185 insert_module (struct gconv_module *newp, int tobefreed)
 186 {
 187   struct gconv_module **rootp = &__gconv_modules_db;
 188
 189   while (*rootp != NULL)
 190     {
 191       struct gconv_module *root = *rootp;
 192       int cmpres;
 193
 194       cmpres = strcmp (newp->from_string, root->from_string);
 195       if (cmpres == 0)
 196         {
 197           /* Both strings are identical.  Insert the string at the
 198              end of the `same' list if it is not already there.  */
 199           while (strcmp (newp->from_string, root->from_string) != 0
 200                  || strcmp (newp->to_string, root->to_string) != 0)
 201             {
 202               rootp = &root->same;
 203               root = *rootp;
 204               if (root == NULL)
 205                 break;
 206             }
 207
 208           if (root != NULL)
 209             {
 210               /* This is a no new conversion.  But maybe the cost is
 211                  better.  */
 212               if (newp->cost_hi < root->cost_hi
 213                   || (newp->cost_hi == root->cost_hi
 214                       && newp->cost_lo < root->cost_lo))
 215                 {
 216                   newp->left = root->left;
 217                   newp->right = root->right;
 218                   newp->same = root->same;
 219                   *rootp = newp;
 220
 221                   free (root);
 222                 }
 223               else if (tobefreed)
 224                 free (newp);
 225               return;
 226             }
 227
 228           break;
 229         }
 230       else if (cmpres < 0)
 231         rootp = &root->left;
 232       else
 233         rootp = &root->right;
 234     }
 235
 236   /* Plug in the new node here.  */
 237   *rootp = newp;
 238 }
 239
 240
 241 /* Add new module.  */
 242 static void
 243 add_module (char *rp, const char *directory, size_t dir_len, void **modules,
 244             size_t *nmodules, int modcounter)
 245 {
 246   /* We expect now
 247      1. `from' name
 248      2. `to' name
 249      3. filename of the module
 250      4. an optional cost value
 251   */
 252   struct gconv_alias fake_alias;
 253   struct gconv_module *new_module;
 254   char *from, *to, *module, *wp;
 255   int need_ext;
 256   int cost_hi;
 257
 258   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 259     ++rp;
 260   from = rp;
 261   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 262     {
 263       *rp = __toupper_l (*rp, _nl_C_locobj_ptr);
 264       ++rp;
 265     }
 266   if (*rp == '\0')
 267     return;
 268   *rp++ = '\0';
 269   to = wp = rp;
 270   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 271     ++rp;
 272   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 273     *wp++ = __toupper_l (*rp++, _nl_C_locobj_ptr);
 274   if (*rp == '\0')
 275     return;
 276   *wp++ = '\0';
 277   do
 278     ++rp;
 279   while (__isspace_l (*rp, _nl_C_locobj_ptr));
 280   module = wp;
 281   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 282     *wp++ = *rp++;
 283   if (*rp == '\0')
 284     {
 285       /* There is no cost, use one by default.  */
 286       *wp++ = '\0';
 287       cost_hi = 1;
 288     }
 289   else
 290     {
 291       /* There might be a cost value.  */
 292       char *endp;
 293
 294       *wp++ = '\0';
 295       cost_hi = strtol (rp, &endp, 10);
 296       if (rp == endp || cost_hi < 1)
 297         /* No useful information.  */
 298         cost_hi = 1;
 299     }
 300
 301   if (module[0] == '\0')
 302     /* No module name given.  */
 303     return;
 304   if (module[0] == '/')
 305     dir_len = 0;
 306
 307   /* See whether we must add the ending.  */
 308   need_ext = 0;
 309   if (wp - module < (ptrdiff_t) sizeof (gconv_module_ext)
 310       || memcmp (wp - sizeof (gconv_module_ext), gconv_module_ext,
 311                  sizeof (gconv_module_ext)) != 0)
 312     /* We must add the module extension.  */
 313     need_ext = sizeof (gconv_module_ext) - 1;
 314
 315   /* See whether we have already an alias with this name defined.  */
 316   fake_alias.fromname = strndupa (from, to - from);
 317
 318   if (__tfind (&fake_alias, &__gconv_alias_db, __gconv_alias_compare) != NULL)
 319     /* This module duplicates an alias.  */
 320     return;
 321
 322   new_module = (struct gconv_module *) calloc (1,
 323                                                sizeof (struct gconv_module)
 324                                                + (wp - from)
 325                                                + dir_len + need_ext);
 326   if (new_module != NULL)
 327     {
 328       char *tmp;
 329
 330       new_module->from_string = tmp = (char *) (new_module + 1);
 331       tmp = __mempcpy (tmp, from, to - from);
 332
 333       new_module->to_string = tmp;
 334       tmp = __mempcpy (tmp, to, module - to);
 335
 336       new_module->cost_hi = cost_hi;
 337       new_module->cost_lo = modcounter;
 338
 339       new_module->module_name = tmp;
 340
 341       if (dir_len != 0)
 342         tmp = __mempcpy (tmp, directory, dir_len);
 343
 344       tmp = __mempcpy (tmp, module, wp - module);
 345
 346       if (need_ext)
 347         memcpy (tmp - 1, gconv_module_ext, sizeof (gconv_module_ext));
 348
 349       /* Now insert the new module data structure in our search tree.  */
 350       insert_module (new_module, 1);
 351     }
 352 }
 353
 354
 355 /* Read the next configuration file.  */
 356 static void
 357 read_conf_file (const char *filename, const char *directory, size_t dir_len,
 358                 void **modules, size_t *nmodules)
 359 {
 360   /* Note the file is opened with cancellation in the I/O functions
 361      disabled.  */
 362   FILE *fp = fopen (filename, "rce");
 363   char *line = NULL;
 364   size_t line_len = 0;
 365   static int modcounter;
 366
 367   /* Don't complain if a file is not present or readable, simply silently
 368      ignore it.  */
 369   if (fp == NULL)
 370     return;
 371
 372   /* No threads reading from this stream.  */
 373   __fsetlocking (fp, FSETLOCKING_BYCALLER);
 374
 375   /* Process the known entries of the file.  Comments start with `#' and
 376      end with the end of the line.  Empty lines are ignored.  */
 377   while (!__feof_unlocked (fp))
 378     {
 379       char *rp, *endp, *word;
 380       ssize_t n = __getdelim (&line, &line_len, '\n', fp);
 381       if (n < 0)
 382         /* An error occurred.  */
 383         break;
 384
 385       rp = line;
 386       /* Terminate the line (excluding comments or newline) by an NUL byte
 387          to simplify the following code.  */
 388       endp = strchr (rp, '#');
 389       if (endp != NULL)
 390         *endp = '\0';
 391       else
 392         if (rp[n - 1] == '\n')
 393           rp[n - 1] = '\0';
 394
 395       while (__isspace_l (*rp, _nl_C_locobj_ptr))
 396         ++rp;
 397
 398       /* If this is an empty line go on with the next one.  */
 399       if (rp == endp)
 400         continue;
 401
 402       word = rp;
 403       while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 404         ++rp;
 405
 406       if (rp - word == sizeof ("alias") - 1
 407           && memcmp (word, "alias", sizeof ("alias") - 1) == 0)
 408         add_alias (rp, *modules);
 409       else if (rp - word == sizeof ("module") - 1
 410                && memcmp (word, "module", sizeof ("module") - 1) == 0)
 411         add_module (rp, directory, dir_len, modules, nmodules, modcounter++);
 412       /* else */
 413         /* Otherwise ignore the line.  */
 414     }
 415
 416   free (line);
 417
 418   fclose (fp);
 419 }
 420
 421
 422 /* Determine the directories we are looking for data in.  */
 423 void
 424 __gconv_get_path (void)
 425 {
 426   struct path_elem *result;
 427   __libc_lock_define_initialized (static, lock);
 428
 429   __libc_lock_lock (lock);
 430
 431   /* Make sure there wasn't a second thread doing it already.  */
 432   result = (struct path_elem *) __gconv_path_elem;
 433   if (result == NULL)
 434     {
 435       /* Determine the complete path first.  */
 436       char *gconv_path;
 437       size_t gconv_path_len;
 438       char *elem;
 439       char *oldp;
 440       char *cp;
 441       int nelems;
 442       char *cwd;
 443       size_t cwdlen;
 444
 445       if (__gconv_path_envvar == NULL)
 446         {
 447           /* No user-defined path.  Make a modifiable copy of the
 448              default path.  */
 449           gconv_path = strdupa (default_gconv_path);
 450           gconv_path_len = sizeof (default_gconv_path);
 451           cwd = NULL;
 452           cwdlen = 0;
 453         }
 454       else
 455         {
 456           /* Append the default path to the user-defined path.  */
 457           size_t user_len = strlen (__gconv_path_envvar);
 458
 459           gconv_path_len = user_len + 1 + sizeof (default_gconv_path);
 460           gconv_path = alloca (gconv_path_len);
 461           __mempcpy (__mempcpy (__mempcpy (gconv_path, __gconv_path_envvar,
 462                                            user_len),
 463                                 ":", 1),
 464                      default_gconv_path, sizeof (default_gconv_path));
 465           cwd = __getcwd (NULL, 0);
 466           cwdlen = __glibc_unlikely (cwd == NULL) ? 0 : strlen (cwd);
 467         }
 468       assert (default_gconv_path[0] == '/');
 469
 470       /* In a first pass we calculate the number of elements.  */
 471       oldp = NULL;
 472       cp = strchr (gconv_path, ':');
 473       nelems = 1;
 474       while (cp != NULL)
 475         {
 476           if (cp != oldp + 1)
 477             ++nelems;
 478           oldp = cp;
 479           cp =  strchr (cp + 1, ':');
 480         }
 481
 482       /* Allocate the memory for the result.  */
 483       result = (struct path_elem *) malloc ((nelems + 1)
 484                                             * sizeof (struct path_elem)
 485                                             + gconv_path_len + nelems
 486                                             + (nelems - 1) * (cwdlen + 1));
 487       if (result != NULL)
 488         {
 489           char *strspace = (char *) &result[nelems + 1];
 490           int n = 0;
 491
 492           /* Separate the individual parts.  */
 493           __gconv_max_path_elem_len = 0;
 494           elem = __strtok_r (gconv_path, ":", &gconv_path);
 495           assert (elem != NULL);
 496           do
 497             {
 498               result[n].name = strspace;
 499               if (elem[0] != '/')
 500                 {
 501                   assert (cwd != NULL);
 502                   strspace = __mempcpy (strspace, cwd, cwdlen);
 503                   *strspace++ = '/';
 504                 }
 505               strspace = __stpcpy (strspace, elem);
 506               if (strspace[-1] != '/')
 507                 *strspace++ = '/';
 508
 509               result[n].len = strspace - result[n].name;
 510               if (result[n].len > __gconv_max_path_elem_len)
 511                 __gconv_max_path_elem_len = result[n].len;
 512
 513               *strspace++ = '\0';
 514               ++n;
 515             }
 516           while ((elem = __strtok_r (NULL, ":", &gconv_path)) != NULL);
 517
 518           result[n].name = NULL;
 519           result[n].len = 0;
 520         }
 521
 522       __gconv_path_elem = result ?: (struct path_elem *) &empty_path_elem;
 523
 524       free (cwd);
 525     }
 526
 527   __libc_lock_unlock (lock);
 528 }
 529
 530
 531 /* Read all configuration files found in the user-specified and the default
 532    path.  */
 533 void
 534 attribute_hidden
 535 __gconv_read_conf (void)
 536 {
 537   void *modules = NULL;
 538   size_t nmodules = 0;
 539   int save_errno = errno;
 540   size_t cnt;
 541
 542   /* First see whether we should use the cache.  */
 543   if (__gconv_load_cache () == 0)
 544     {
 545       /* Yes, we are done.  */
 546       __set_errno (save_errno);
 547       return;
 548     }
 549
 550 #ifndef STATIC_GCONV
 551   /* Find out where we have to look.  */
 552   if (__gconv_path_elem == NULL)
 553     __gconv_get_path ();
 554
 555   for (cnt = 0; __gconv_path_elem[cnt].name != NULL; ++cnt)
 556     {
 557       const char *elem = __gconv_path_elem[cnt].name;
 558       size_t elem_len = __gconv_path_elem[cnt].len;
 559       char *filename;
 560
 561       /* No slash needs to be inserted between elem and gconv_conf_filename;
 562          elem already ends in a slash.  */
 563       filename = alloca (elem_len + sizeof (gconv_conf_filename));
 564       __mempcpy (__mempcpy (filename, elem, elem_len),
 565                  gconv_conf_filename, sizeof (gconv_conf_filename));
 566
 567       /* Read the next configuration file.  */
 568       read_conf_file (filename, elem, elem_len, &modules, &nmodules);
 569     }
 570 #endif
 571
 572   /* Add the internal modules.  */
 573   for (cnt = 0; cnt < sizeof (builtin_modules) / sizeof (builtin_modules[0]);
 574        ++cnt)
 575     {
 576       struct gconv_alias fake_alias;
 577
 578       fake_alias.fromname = (char *) builtin_modules[cnt].from_string;
 579
 580       if (__tfind (&fake_alias, &__gconv_alias_db, __gconv_alias_compare)
 581           != NULL)
 582         /* It'll conflict so don't add it.  */
 583         continue;
 584
 585       insert_module (&builtin_modules[cnt], 0);
 586     }
 587
 588   /* Add aliases for builtin conversions.  */
 589   const char *cp = builtin_aliases;
 590   do
 591     {
 592       const char *from = cp;
 593       const char *to = __rawmemchr (from, '\0') + 1;
 594       cp = __rawmemchr (to, '\0') + 1;
 595
 596       add_alias2 (from, to, cp, modules);
 597     }
 598   while (*cp != '\0');
 599
 600   /* Restore the error number.  */
 601   __set_errno (save_errno);
 602 }
 603
 604
 605
 606 /* Free all resources if necessary.  */
 607 libc_freeres_fn (free_mem)
 608 {
 609   if (__gconv_path_elem != NULL && __gconv_path_elem != &empty_path_elem)
 610     free ((void *) __gconv_path_elem);
 611 }