iconv/gconv_conf.c

   1 /* Handle configuration data.
   2    Copyright (C) 1997-2003, 2005, 2006, 2011 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, write to the Free
  18    Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
  19    02111-1307 USA.  */
  20
  21 #include <assert.h>
  22 #include <ctype.h>
  23 #include <errno.h>
  24 #include <limits.h>
  25 #include <locale.h>
  26 #include <search.h>
  27 #include <stddef.h>
  28 #include <stdio.h>
  29 #include <stdio_ext.h>
  30 #include <stdlib.h>
  31 #include <string.h>
  32 #include <unistd.h>
  33 #include <sys/param.h>
  34
  35 #include <bits/libc-lock.h>
  36 #include <gconv_int.h>
  37
  38
  39 /* This is the default path where we look for module lists.  */
  40 static const char default_gconv_path[] = GCONV_PATH;
  41
  42 /* The path elements, as determined by the __gconv_get_path function.
  43    All path elements end in a slash.  */
  44 struct path_elem *__gconv_path_elem;
  45 /* Maximum length of a single path element in __gconv_path_elem.  */
  46 size_t __gconv_max_path_elem_len;
  47
  48 /* We use the following struct if we couldn't allocate memory.  */
  49 static const struct path_elem empty_path_elem = { NULL, 0 };
  50
  51 /* Name of the file containing the module information in the directories
  52    along the path.  */
  53 static const char gconv_conf_filename[] = "gconv-modules";
  54
  55 /* Filename extension for the modules.  */
  56 #ifndef MODULE_EXT
  57 # define MODULE_EXT ".so"
  58 #endif
  59 static const char gconv_module_ext[] = MODULE_EXT;
  60
  61 /* We have a few builtin transformations.  */
  62 static struct gconv_module builtin_modules[] =
  63 {
  64 #define BUILTIN_TRANSFORMATION(From, To, Cost, Name, Fct, BtowcFct, \
  65                                MinF, MaxF, MinT, MaxT) \
  66   {                                                                           \
  67     .from_string = From,                                                      \
  68     .to_string = To,                                                          \
  69     .cost_hi = Cost,                                                          \
  70     .cost_lo = INT_MAX,                                                       \
  71     .module_name = Name                                                       \
  72   },
  73 #define BUILTIN_ALIAS(From, To)
  74
  75 #include "gconv_builtin.h"
  76
  77 #undef BUILTIN_TRANSFORMATION
  78 #undef BUILTIN_ALIAS
  79 };
  80
  81 static const char builtin_aliases[] =
  82 {
  83 #define BUILTIN_TRANSFORMATION(From, To, Cost, Name, Fct, BtowcFct, \
  84                                MinF, MaxF, MinT, MaxT)
  85 #define BUILTIN_ALIAS(From, To) From "\0" To "\0"
  86
  87 #include "gconv_builtin.h"
  88
  89 #undef BUILTIN_TRANSFORMATION
  90 #undef BUILTIN_ALIAS
  91 };
  92
  93 #include <libio/libioP.h>
  94 #define __getdelim(line, len, c, fp) _IO_getdelim (line, len, c, fp)
  95
  96
  97 /* Value of the GCONV_PATH environment variable.  */
  98 const char *__gconv_path_envvar;
  99
 100
 101 /* Test whether there is already a matching module known.  */
 102 static int
 103 internal_function
 104 detect_conflict (const char *alias)
 105 {
 106   struct gconv_module *node = __gconv_modules_db;
 107
 108   while (node != NULL)
 109     {
 110       int cmpres = strcmp (alias, node->from_string);
 111
 112       if (cmpres == 0)
 113         /* We have a conflict.  */
 114         return 1;
 115       else if (cmpres < 0)
 116         node = node->left;
 117       else
 118         node = node->right;
 119     }
 120
 121   return node != NULL;
 122 }
 123
 124
 125 /* The actual code to add aliases.  */
 126 static void
 127 add_alias2 (const char *from, const char *to, const char *wp, void *modules)
 128 {
 129   /* Test whether this alias conflicts with any available module.  */
 130   if (detect_conflict (from))
 131     /* It does conflict, don't add the alias.  */
 132     return;
 133
 134   struct gconv_alias *new_alias = (struct gconv_alias *)
 135     malloc (sizeof (struct gconv_alias) + (wp - from));
 136   if (new_alias != NULL)
 137     {
 138       void **inserted;
 139
 140       new_alias->fromname = memcpy ((char *) new_alias
 141                                     + sizeof (struct gconv_alias),
 142                                     from, wp - from);
 143       new_alias->toname = new_alias->fromname + (to - from);
 144
 145       inserted = (void **) __tsearch (new_alias, &__gconv_alias_db,
 146                                       __gconv_alias_compare);
 147       if (inserted == NULL || *inserted != new_alias)
 148         /* Something went wrong, free this entry.  */
 149         free (new_alias);
 150     }
 151 }
 152
 153
 154 /* Add new alias.  */
 155 static void
 156 add_alias (char *rp, void *modules)
 157 {
 158   /* We now expect two more string.  The strings are normalized
 159      (converted to UPPER case) and strored in the alias database.  */
 160   char *from, *to, *wp;
 161
 162   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 163     ++rp;
 164   from = wp = rp;
 165   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 166     *wp++ = __toupper_l (*rp++, _nl_C_locobj_ptr);
 167   if (*rp == '\0')
 168     /* There is no `to' string on the line.  Ignore it.  */
 169     return;
 170   *wp++ = '\0';
 171   to = ++rp;
 172   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 173     ++rp;
 174   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 175     *wp++ = __toupper_l (*rp++, _nl_C_locobj_ptr);
 176   if (to == wp)
 177     /* No `to' string, ignore the line.  */
 178     return;
 179   *wp++ = '\0';
 180
 181   add_alias2 (from, to, wp, modules);
 182 }
 183
 184
 185 /* Insert a data structure for a new module in the search tree.  */
 186 static void
 187 internal_function
 188 insert_module (struct gconv_module *newp, int tobefreed)
 189 {
 190   struct gconv_module **rootp = &__gconv_modules_db;
 191
 192   while (*rootp != NULL)
 193     {
 194       struct gconv_module *root = *rootp;
 195       int cmpres;
 196
 197       cmpres = strcmp (newp->from_string, root->from_string);
 198       if (cmpres == 0)
 199         {
 200           /* Both strings are identical.  Insert the string at the
 201              end of the `same' list if it is not already there.  */
 202           while (strcmp (newp->from_string, root->from_string) != 0
 203                  || strcmp (newp->to_string, root->to_string) != 0)
 204             {
 205               rootp = &root->same;
 206               root = *rootp;
 207               if (root == NULL)
 208                 break;
 209             }
 210
 211           if (root != NULL)
 212             {
 213               /* This is a no new conversion.  But maybe the cost is
 214                  better.  */
 215               if (newp->cost_hi < root->cost_hi
 216                   || (newp->cost_hi == root->cost_hi
 217                       && newp->cost_lo < root->cost_lo))
 218                 {
 219                   newp->left = root->left;
 220                   newp->right = root->right;
 221                   newp->same = root->same;
 222                   *rootp = newp;
 223
 224                   free (root);
 225                 }
 226               else if (tobefreed)
 227                 free (newp);
 228               return;
 229             }
 230
 231           break;
 232         }
 233       else if (cmpres < 0)
 234         rootp = &root->left;
 235       else
 236         rootp = &root->right;
 237     }
 238
 239   /* Plug in the new node here.  */
 240   *rootp = newp;
 241 }
 242
 243
 244 /* Add new module.  */
 245 static void
 246 internal_function
 247 add_module (char *rp, const char *directory, size_t dir_len, void **modules,
 248             size_t *nmodules, int modcounter)
 249 {
 250   /* We expect now
 251      1. `from' name
 252      2. `to' name
 253      3. filename of the module
 254      4. an optional cost value
 255   */
 256   struct gconv_alias fake_alias;
 257   struct gconv_module *new_module;
 258   char *from, *to, *module, *wp;
 259   int need_ext;
 260   int cost_hi;
 261
 262   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 263     ++rp;
 264   from = rp;
 265   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 266     {
 267       *rp = __toupper_l (*rp, _nl_C_locobj_ptr);
 268       ++rp;
 269     }
 270   if (*rp == '\0')
 271     return;
 272   *rp++ = '\0';
 273   to = wp = rp;
 274   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 275     ++rp;
 276   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 277     *wp++ = __toupper_l (*rp++, _nl_C_locobj_ptr);
 278   if (*rp == '\0')
 279     return;
 280   *wp++ = '\0';
 281   do
 282     ++rp;
 283   while (__isspace_l (*rp, _nl_C_locobj_ptr));
 284   module = wp;
 285   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 286     *wp++ = *rp++;
 287   if (*rp == '\0')
 288     {
 289       /* There is no cost, use one by default.  */
 290       *wp++ = '\0';
 291       cost_hi = 1;
 292     }
 293   else
 294     {
 295       /* There might be a cost value.  */
 296       char *endp;
 297
 298       *wp++ = '\0';
 299       cost_hi = strtol (rp, &endp, 10);
 300       if (rp == endp || cost_hi < 1)
 301         /* No useful information.  */
 302         cost_hi = 1;
 303     }
 304
 305   if (module[0] == '\0')
 306     /* No module name given.  */
 307     return;
 308   if (module[0] == '/')
 309     dir_len = 0;
 310
 311   /* See whether we must add the ending.  */
 312   need_ext = 0;
 313   if (wp - module < (ptrdiff_t) sizeof (gconv_module_ext)
 314       || memcmp (wp - sizeof (gconv_module_ext), gconv_module_ext,
 315                  sizeof (gconv_module_ext)) != 0)
 316     /* We must add the module extension.  */
 317     need_ext = sizeof (gconv_module_ext) - 1;
 318
 319   /* See whether we have already an alias with this name defined.  */
 320   fake_alias.fromname = strndupa (from, to - from);
 321
 322   if (__tfind (&fake_alias, &__gconv_alias_db, __gconv_alias_compare) != NULL)
 323     /* This module duplicates an alias.  */
 324     return;
 325
 326   new_module = (struct gconv_module *) calloc (1,
 327                                                sizeof (struct gconv_module)
 328                                                + (wp - from)
 329                                                + dir_len + need_ext);
 330   if (new_module != NULL)
 331     {
 332       char *tmp;
 333
 334       new_module->from_string = tmp = (char *) (new_module + 1);
 335       tmp = __mempcpy (tmp, from, to - from);
 336
 337       new_module->to_string = tmp;
 338       tmp = __mempcpy (tmp, to, module - to);
 339
 340       new_module->cost_hi = cost_hi;
 341       new_module->cost_lo = modcounter;
 342
 343       new_module->module_name = tmp;
 344
 345       if (dir_len != 0)
 346         tmp = __mempcpy (tmp, directory, dir_len);
 347
 348       tmp = __mempcpy (tmp, module, wp - module);
 349
 350       if (need_ext)
 351         memcpy (tmp - 1, gconv_module_ext, sizeof (gconv_module_ext));
 352
 353       /* Now insert the new module data structure in our search tree.  */
 354       insert_module (new_module, 1);
 355     }
 356 }
 357
 358
 359 /* Read the next configuration file.  */
 360 static void
 361 internal_function
 362 read_conf_file (const char *filename, const char *directory, size_t dir_len,
 363                 void **modules, size_t *nmodules)
 364 {
 365   /* Note the file is opened with cancellation in the I/O functions
 366      disabled.  */
 367   FILE *fp = fopen (filename, "rce");
 368   char *line = NULL;
 369   size_t line_len = 0;
 370   static int modcounter;
 371
 372   /* Don't complain if a file is not present or readable, simply silently
 373      ignore it.  */
 374   if (fp == NULL)
 375     return;
 376
 377   /* No threads reading from this stream.  */
 378   __fsetlocking (fp, FSETLOCKING_BYCALLER);
 379
 380   /* Process the known entries of the file.  Comments start with `#' and
 381      end with the end of the line.  Empty lines are ignored.  */
 382   while (!feof_unlocked (fp))
 383     {
 384       char *rp, *endp, *word;
 385       ssize_t n = __getdelim (&line, &line_len, '\n', fp);
 386       if (n < 0)
 387         /* An error occurred.  */
 388         break;
 389
 390       rp = line;
 391       /* Terminate the line (excluding comments or newline) by an NUL byte
 392          to simplify the following code.  */
 393       endp = strchr (rp, '#');
 394       if (endp != NULL)
 395         *endp = '\0';
 396       else
 397         if (rp[n - 1] == '\n')
 398           rp[n - 1] = '\0';
 399
 400       while (__isspace_l (*rp, _nl_C_locobj_ptr))
 401         ++rp;
 402
 403       /* If this is an empty line go on with the next one.  */
 404       if (rp == endp)
 405         continue;
 406
 407       word = rp;
 408       while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 409         ++rp;
 410
 411       if (rp - word == sizeof ("alias") - 1
 412           && memcmp (word, "alias", sizeof ("alias") - 1) == 0)
 413         add_alias (rp, *modules);
 414       else if (rp - word == sizeof ("module") - 1
 415                && memcmp (word, "module", sizeof ("module") - 1) == 0)
 416         add_module (rp, directory, dir_len, modules, nmodules, modcounter++);
 417       /* else */
 418         /* Otherwise ignore the line.  */
 419     }
 420
 421   free (line);
 422
 423   fclose (fp);
 424 }
 425
 426
 427 /* Determine the directories we are looking for data in.  */
 428 void
 429 internal_function
 430 __gconv_get_path (void)
 431 {
 432   struct path_elem *result;
 433   __libc_lock_define_initialized (static, lock);
 434
 435   __libc_lock_lock (lock);
 436
 437   /* Make sure there wasn't a second thread doing it already.  */
 438   result = (struct path_elem *) __gconv_path_elem;
 439   if (result == NULL)
 440     {
 441       /* Determine the complete path first.  */
 442       char *gconv_path;
 443       size_t gconv_path_len;
 444       char *elem;
 445       char *oldp;
 446       char *cp;
 447       int nelems;
 448       char *cwd;
 449       size_t cwdlen;
 450
 451       if (__gconv_path_envvar == NULL)
 452         {
 453           /* No user-defined path.  Make a modifiable copy of the
 454              default path.  */
 455           gconv_path = strdupa (default_gconv_path);
 456           gconv_path_len = sizeof (default_gconv_path);
 457           cwd = NULL;
 458           cwdlen = 0;
 459         }
 460       else
 461         {
 462           /* Append the default path to the user-defined path.  */
 463           size_t user_len = strlen (__gconv_path_envvar);
 464
 465           gconv_path_len = user_len + 1 + sizeof (default_gconv_path);
 466           gconv_path = alloca (gconv_path_len);
 467           __mempcpy (__mempcpy (__mempcpy (gconv_path, __gconv_path_envvar,
 468                                            user_len),
 469                                 ":", 1),
 470                      default_gconv_path, sizeof (default_gconv_path));
 471           cwd = __getcwd (NULL, 0);
 472           cwdlen = strlen (cwd);
 473         }
 474       assert (default_gconv_path[0] == '/');
 475
 476       /* In a first pass we calculate the number of elements.  */
 477       oldp = NULL;
 478       cp = strchr (gconv_path, ':');
 479       nelems = 1;
 480       while (cp != NULL)
 481         {
 482           if (cp != oldp + 1)
 483             ++nelems;
 484           oldp = cp;
 485           cp =  strchr (cp + 1, ':');
 486         }
 487
 488       /* Allocate the memory for the result.  */
 489       result = (struct path_elem *) malloc ((nelems + 1)
 490                                             * sizeof (struct path_elem)
 491                                             + gconv_path_len + nelems
 492                                             + (nelems - 1) * (cwdlen + 1));
 493       if (result != NULL)
 494         {
 495           char *strspace = (char *) &result[nelems + 1];
 496           int n = 0;
 497
 498           /* Separate the individual parts.  */
 499           __gconv_max_path_elem_len = 0;
 500           elem = __strtok_r (gconv_path, ":", &gconv_path);
 501           assert (elem != NULL);
 502           do
 503             {
 504               result[n].name = strspace;
 505               if (elem[0] != '/')
 506                 {
 507                   assert (cwd != NULL);
 508                   strspace = __mempcpy (strspace, cwd, cwdlen);
 509                   *strspace++ = '/';
 510                 }
 511               strspace = __stpcpy (strspace, elem);
 512               if (strspace[-1] != '/')
 513                 *strspace++ = '/';
 514
 515               result[n].len = strspace - result[n].name;
 516               if (result[n].len > __gconv_max_path_elem_len)
 517                 __gconv_max_path_elem_len = result[n].len;
 518
 519               *strspace++ = '\0';
 520               ++n;
 521             }
 522           while ((elem = __strtok_r (NULL, ":", &gconv_path)) != NULL);
 523
 524           result[n].name = NULL;
 525           result[n].len = 0;
 526         }
 527
 528       __gconv_path_elem = result ?: (struct path_elem *) &empty_path_elem;
 529
 530       free (cwd);
 531     }
 532
 533   __libc_lock_unlock (lock);
 534 }
 535
 536
 537 /* Read all configuration files found in the user-specified and the default
 538    path.  */
 539 void
 540 attribute_hidden
 541 __gconv_read_conf (void)
 542 {
 543   void *modules = NULL;
 544   size_t nmodules = 0;
 545   int save_errno = errno;
 546   size_t cnt;
 547
 548   /* First see whether we should use the cache.  */
 549   if (__gconv_load_cache () == 0)
 550     {
 551       /* Yes, we are done.  */
 552       __set_errno (save_errno);
 553       return;
 554     }
 555
 556 #ifndef STATIC_GCONV
 557   /* Find out where we have to look.  */
 558   if (__gconv_path_elem == NULL)
 559     __gconv_get_path ();
 560
 561   for (cnt = 0; __gconv_path_elem[cnt].name != NULL; ++cnt)
 562     {
 563       const char *elem = __gconv_path_elem[cnt].name;
 564       size_t elem_len = __gconv_path_elem[cnt].len;
 565       char *filename;
 566
 567       /* No slash needs to be inserted between elem and gconv_conf_filename;
 568          elem already ends in a slash.  */
 569       filename = alloca (elem_len + sizeof (gconv_conf_filename));
 570       __mempcpy (__mempcpy (filename, elem, elem_len),
 571                  gconv_conf_filename, sizeof (gconv_conf_filename));
 572
 573       /* Read the next configuration file.  */
 574       read_conf_file (filename, elem, elem_len, &modules, &nmodules);
 575     }
 576 #endif
 577
 578   /* Add the internal modules.  */
 579   for (cnt = 0; cnt < sizeof (builtin_modules) / sizeof (builtin_modules[0]);
 580        ++cnt)
 581     {
 582       struct gconv_alias fake_alias;
 583
 584       fake_alias.fromname = (char *) builtin_modules[cnt].from_string;
 585
 586       if (__tfind (&fake_alias, &__gconv_alias_db, __gconv_alias_compare)
 587           != NULL)
 588         /* It'll conflict so don't add it.  */
 589         continue;
 590
 591       insert_module (&builtin_modules[cnt], 0);
 592     }
 593
 594   /* Add aliases for builtin conversions.  */
 595   const char *cp = builtin_aliases;
 596   do
 597     {
 598       const char *from = cp;
 599       const char *to = __rawmemchr (from, '\0') + 1;
 600       cp = __rawmemchr (to, '\0') + 1;
 601
 602       add_alias2 (from, to, cp, modules);
 603     }
 604   while (*cp != '\0');
 605
 606   /* Restore the error number.  */
 607   __set_errno (save_errno);
 608 }
 609
 610
 611
 612 /* Free all resources if necessary.  */
 613 libc_freeres_fn (free_mem)
 614 {
 615   if (__gconv_path_elem != NULL && __gconv_path_elem != &empty_path_elem)
 616     free ((void *) __gconv_path_elem);
 617 }