iconv/gconv_conf.c

   1 /* Handle configuration data.
   2    Copyright (C) 1997,98,99,2000,2001 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;
  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, MinF, MaxF, \
  65                                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
  78 #undef BUILTIN_TRANSFORMATION
  79 #undef BUILTIN_ALIAS
  80
  81 static const char *builtin_aliases[] =
  82 {
  83 #define BUILTIN_TRANSFORMATION(From, To, Cost, Name, Fct, MinF, MaxF, \
  84                                MinT, MaxT)
  85 #define BUILTIN_ALIAS(From, To) From " " To,
  86
  87 #include "gconv_builtin.h"
  88 };
  89
  90 #ifdef USE_IN_LIBIO
  91 # include <libio/libioP.h>
  92 # define __getdelim(line, len, c, fp) _IO_getdelim (line, len, c, fp)
  93 #endif
  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 internal_function
 103 detect_conflict (const char *alias)
 104 {
 105   struct gconv_module *node = __gconv_modules_db;
 106
 107   while (node != NULL)
 108     {
 109       int cmpres = strcmp (alias, node->from_string);
 110
 111       if (cmpres == 0)
 112         /* We have a conflict.  */
 113         return 1;
 114       else if (cmpres < 0)
 115         node = node->left;
 116       else
 117         node = node->right;
 118     }
 119
 120   return node != NULL;
 121 }
 122
 123
 124 /* Add new alias.  */
 125 static inline void
 126 add_alias (char *rp, void *modules)
 127 {
 128   /* We now expect two more string.  The strings are normalized
 129      (converted to UPPER case) and strored in the alias database.  */
 130   struct gconv_alias *new_alias;
 131   char *from, *to, *wp;
 132
 133   while (__isspace_l (*rp, &_nl_C_locobj))
 134     ++rp;
 135   from = wp = rp;
 136   while (*rp != '\0' && !__isspace_l (*rp, &_nl_C_locobj))
 137     *wp++ = __toupper_l (*rp++, &_nl_C_locobj);
 138   if (*rp == '\0')
 139     /* There is no `to' string on the line.  Ignore it.  */
 140     return;
 141   *wp++ = '\0';
 142   to = ++rp;
 143   while (__isspace_l (*rp, &_nl_C_locobj))
 144     ++rp;
 145   while (*rp != '\0' && !__isspace_l (*rp, &_nl_C_locobj))
 146     *wp++ = __toupper_l (*rp++, &_nl_C_locobj);
 147   if (to == wp)
 148     /* No `to' string, ignore the line.  */
 149     return;
 150   *wp++ = '\0';
 151
 152   /* Test whether this alias conflicts with any available module.  */
 153   if (detect_conflict (from))
 154     /* It does conflict, don't add the alias.  */
 155     return;
 156
 157   new_alias = (struct gconv_alias *)
 158     malloc (sizeof (struct gconv_alias) + (wp - from));
 159   if (new_alias != NULL)
 160     {
 161       void **inserted;
 162
 163       new_alias->fromname = memcpy ((char *) new_alias
 164                                     + sizeof (struct gconv_alias),
 165                                     from, wp - from);
 166       new_alias->toname = new_alias->fromname + (to - from);
 167
 168       inserted = (void **) __tsearch (new_alias, &__gconv_alias_db,
 169                                       __gconv_alias_compare);
 170       if (inserted == NULL || *inserted != new_alias)
 171         /* Something went wrong, free this entry.  */
 172         free (new_alias);
 173     }
 174 }
 175
 176
 177 /* Insert a data structure for a new module in the search tree.  */
 178 static inline void
 179 internal_function
 180 insert_module (struct gconv_module *newp, int tobefreed)
 181 {
 182   struct gconv_module **rootp = &__gconv_modules_db;
 183
 184   while (*rootp != NULL)
 185     {
 186       struct gconv_module *root = *rootp;
 187       int cmpres;
 188
 189       cmpres = strcmp (newp->from_string, root->from_string);
 190       if (cmpres == 0)
 191         {
 192           /* Both strings are identical.  Insert the string at the
 193              end of the `same' list if it is not already there.  */
 194           while (strcmp (newp->from_string, root->from_string) != 0
 195                  || strcmp (newp->to_string, root->to_string) != 0)
 196             {
 197               rootp = &root->same;
 198               root = *rootp;
 199               if (root == NULL)
 200                 break;
 201             }
 202
 203           if (root != NULL)
 204             {
 205               /* This is a no new conversion.  But maybe the cost is
 206                  better.  */
 207               if (newp->cost_hi < root->cost_hi
 208                   || (newp->cost_hi == root->cost_hi
 209                       && newp->cost_lo < root->cost_lo))
 210                 {
 211                   newp->left = root->left;
 212                   newp->right = root->right;
 213                   newp->same = root->same;
 214                   *rootp = newp;
 215
 216                   free (root);
 217                 }
 218               else if (tobefreed)
 219                 free (newp);
 220               return;
 221             }
 222
 223           break;
 224         }
 225       else if (cmpres < 0)
 226         rootp = &root->left;
 227       else
 228         rootp = &root->right;
 229     }
 230
 231   /* Plug in the new node here.  */
 232   *rootp = newp;
 233 }
 234
 235
 236 /* Add new module.  */
 237 static void
 238 internal_function
 239 add_module (char *rp, const char *directory, size_t dir_len, void **modules,
 240             size_t *nmodules, int modcounter)
 241 {
 242   /* We expect now
 243      1. `from' name
 244      2. `to' name
 245      3. filename of the module
 246      4. an optional cost value
 247   */
 248   struct gconv_alias fake_alias;
 249   struct gconv_module *new_module;
 250   char *from, *to, *module, *wp;
 251   int need_ext;
 252   int cost_hi;
 253
 254   while (__isspace_l (*rp, &_nl_C_locobj))
 255     ++rp;
 256   from = rp;
 257   while (*rp != '\0' && !__isspace_l (*rp, &_nl_C_locobj))
 258     {
 259       *rp = __toupper_l (*rp, &_nl_C_locobj);
 260       ++rp;
 261     }
 262   if (*rp == '\0')
 263     return;
 264   *rp++ = '\0';
 265   to = wp = rp;
 266   while (__isspace_l (*rp, &_nl_C_locobj))
 267     ++rp;
 268   while (*rp != '\0' && !__isspace_l (*rp, &_nl_C_locobj))
 269     *wp++ = __toupper_l (*rp++, &_nl_C_locobj);
 270   if (*rp == '\0')
 271     return;
 272   *wp++ = '\0';
 273   do
 274     ++rp;
 275   while (__isspace_l (*rp, &_nl_C_locobj));
 276   module = wp;
 277   while (*rp != '\0' && !__isspace_l (*rp, &_nl_C_locobj))
 278     *wp++ = *rp++;
 279   if (*rp == '\0')
 280     {
 281       /* There is no cost, use one by default.  */
 282       *wp++ = '\0';
 283       cost_hi = 1;
 284     }
 285   else
 286     {
 287       /* There might be a cost value.  */
 288       char *endp;
 289
 290       *wp++ = '\0';
 291       cost_hi = strtol (rp, &endp, 10);
 292       if (rp == endp || cost_hi < 1)
 293         /* No useful information.  */
 294         cost_hi = 1;
 295     }
 296
 297   if (module[0] == '\0')
 298     /* No module name given.  */
 299     return;
 300   if (module[0] == '/')
 301     dir_len = 0;
 302
 303   /* See whether we must add the ending.  */
 304   need_ext = 0;
 305   if (wp - module < (ptrdiff_t) sizeof (gconv_module_ext)
 306       || memcmp (wp - sizeof (gconv_module_ext), gconv_module_ext,
 307                  sizeof (gconv_module_ext)) != 0)
 308     /* We must add the module extension.  */
 309     need_ext = sizeof (gconv_module_ext) - 1;
 310
 311   /* See whether we have already an alias with this name defined.  */
 312   fake_alias.fromname = strndupa (from, to - from);
 313
 314   if (__tfind (&fake_alias, &__gconv_alias_db, __gconv_alias_compare) != NULL)
 315     /* This module duplicates an alias.  */
 316     return;
 317
 318   new_module = (struct gconv_module *) calloc (1,
 319                                                sizeof (struct gconv_module)
 320                                                + (wp - from)
 321                                                + dir_len + need_ext);
 322   if (new_module != NULL)
 323     {
 324       char *tmp;
 325
 326       new_module->from_string = tmp = (char *) (new_module + 1);
 327       tmp = __mempcpy (tmp, from, to - from);
 328
 329       new_module->to_string = tmp;
 330       tmp = __mempcpy (tmp, to, module - to);
 331
 332       new_module->cost_hi = cost_hi;
 333       new_module->cost_lo = modcounter;
 334
 335       new_module->module_name = tmp;
 336
 337       if (dir_len != 0)
 338         tmp = __mempcpy (tmp, directory, dir_len);
 339
 340       tmp = __mempcpy (tmp, module, wp - module);
 341
 342       if (need_ext)
 343         memcpy (tmp - 1, gconv_module_ext, sizeof (gconv_module_ext));
 344
 345       /* Now insert the new module data structure in our search tree.  */
 346       insert_module (new_module, 1);
 347     }
 348 }
 349
 350
 351 /* Read the next configuration file.  */
 352 static void
 353 internal_function
 354 read_conf_file (const char *filename, const char *directory, size_t dir_len,
 355                 void **modules, size_t *nmodules)
 356 {
 357   FILE *fp = fopen (filename, "r");
 358   char *line = NULL;
 359   size_t line_len = 0;
 360   static int modcounter;
 361
 362   /* Don't complain if a file is not present or readable, simply silently
 363      ignore it.  */
 364   if (fp == NULL)
 365     return;
 366
 367   /* No threads reading from this stream.  */
 368   __fsetlocking (fp, FSETLOCKING_BYCALLER);
 369
 370   /* Process the known entries of the file.  Comments start with `#' and
 371      end with the end of the line.  Empty lines are ignored.  */
 372   while (!feof_unlocked (fp))
 373     {
 374       char *rp, *endp, *word;
 375       ssize_t n = __getdelim (&line, &line_len, '\n', fp);
 376       if (n < 0)
 377         /* An error occurred.  */
 378         break;
 379
 380       rp = line;
 381       /* Terminate the line (excluding comments or newline) by an NUL byte
 382          to simplify the following code.  */
 383       endp = strchr (rp, '#');
 384       if (endp != NULL)
 385         *endp = '\0';
 386       else
 387         if (rp[n - 1] == '\n')
 388           rp[n - 1] = '\0';
 389
 390       while (__isspace_l (*rp, &_nl_C_locobj))
 391         ++rp;
 392
 393       /* If this is an empty line go on with the next one.  */
 394       if (rp == endp)
 395         continue;
 396
 397       word = rp;
 398       while (*rp != '\0' && !__isspace_l (*rp, &_nl_C_locobj))
 399         ++rp;
 400
 401       if (rp - word == sizeof ("alias") - 1
 402           && memcmp (word, "alias", sizeof ("alias") - 1) == 0)
 403         add_alias (rp, *modules);
 404       else if (rp - word == sizeof ("module") - 1
 405                && memcmp (word, "module", sizeof ("module") - 1) == 0)
 406         add_module (rp, directory, dir_len, modules, nmodules, modcounter++);
 407       /* else */
 408         /* Otherwise ignore the line.  */
 409     }
 410
 411   free (line);
 412
 413   fclose (fp);
 414 }
 415
 416
 417 /* Determine the directories we are looking for data in.  */
 418 void
 419 __gconv_get_path (void)
 420 {
 421   struct path_elem *result;
 422   __libc_lock_define_initialized (static, lock);
 423
 424   __libc_lock_lock (lock);
 425
 426   /* Make sure there wasn't a second thread doing it already.  */
 427   result = (struct path_elem *) __gconv_path_elem;
 428   if (result == NULL)
 429     {
 430       /* Determine the complete path first.  */
 431       char *gconv_path;
 432       size_t gconv_path_len;
 433       char *elem;
 434       char *oldp;
 435       char *cp;
 436       int nelems;
 437       char *cwd;
 438       size_t cwdlen;
 439
 440       if (__gconv_path_envvar == NULL)
 441         {
 442           /* No user-defined path.  Make a modifiable copy of the
 443              default path.  */
 444           gconv_path = strdupa (default_gconv_path);
 445           gconv_path_len = sizeof (default_gconv_path);
 446           cwd = NULL;
 447           cwdlen = 0;
 448         }
 449       else
 450         {
 451           /* Append the default path to the user-defined path.  */
 452           size_t user_len = strlen (__gconv_path_envvar);
 453
 454           gconv_path_len = user_len + 1 + sizeof (default_gconv_path);
 455           gconv_path = alloca (gconv_path_len);
 456           __mempcpy (__mempcpy (__mempcpy (gconv_path, __gconv_path_envvar,
 457                                            user_len),
 458                                 ":", 1),
 459                      default_gconv_path, sizeof (default_gconv_path));
 460           cwd = __getcwd (NULL, 0);
 461           cwdlen = strlen (cwd);
 462         }
 463       assert (default_gconv_path[0] == '/');
 464
 465       /* In a first pass we calculate the number of elements.  */
 466       oldp = NULL;
 467       cp = strchr (gconv_path, ':');
 468       nelems = 1;
 469       while (cp != NULL)
 470         {
 471           if (cp != oldp + 1)
 472             ++nelems;
 473           oldp = cp;
 474           cp =  strchr (cp + 1, ':');
 475         }
 476
 477       /* Allocate the memory for the result.  */
 478       result = (struct path_elem *) malloc ((nelems + 1)
 479                                             * sizeof (struct path_elem)
 480                                             + gconv_path_len + nelems
 481                                             + (nelems - 1) * (cwdlen + 1));
 482       if (result != NULL)
 483         {
 484           char *strspace = (char *) &result[nelems + 1];
 485           int n = 0;
 486
 487           /* Separate the individual parts.  */
 488           __gconv_max_path_elem_len = 0;
 489           elem = __strtok_r (gconv_path, ":", &gconv_path);
 490           assert (elem != NULL);
 491           do
 492             {
 493               result[n].name = strspace;
 494               if (elem[0] != '/')
 495                 {
 496                   assert (cwd != NULL);
 497                   strspace = __mempcpy (strspace, cwd, cwdlen);
 498                   *strspace++ = '/';
 499                 }
 500               strspace = __stpcpy (strspace, elem);
 501               if (strspace[-1] != '/')
 502                 *strspace++ = '/';
 503
 504               result[n].len = strspace - result[n].name;
 505               if (result[n].len > __gconv_max_path_elem_len)
 506                 __gconv_max_path_elem_len = result[n].len;
 507
 508               *strspace++ = '\0';
 509               ++n;
 510             }
 511           while ((elem = __strtok_r (NULL, ":", &gconv_path)) != NULL);
 512
 513           result[n].name = NULL;
 514           result[n].len = 0;
 515         }
 516
 517       __gconv_path_elem = result ?: (struct path_elem *) &empty_path_elem;
 518
 519       if (cwd != NULL)
 520         free (cwd);
 521     }
 522
 523   __libc_lock_unlock (lock);
 524 }
 525
 526
 527 /* Read all configuration files found in the user-specified and the default
 528    path.  */
 529 void
 530 __gconv_read_conf (void)
 531 {
 532   void *modules = NULL;
 533   size_t nmodules = 0;
 534   int save_errno = errno;
 535   size_t cnt;
 536
 537   /* First see whether we should use the cache.  */
 538   if (__gconv_load_cache () == 0)
 539     {
 540       /* Yes, we are done.  */
 541       __set_errno (save_errno);
 542       return;
 543     }
 544
 545 #ifndef STATIC_GCONV
 546   /* Find out where we have to look.  */
 547   if (__gconv_path_elem == NULL)
 548     __gconv_get_path ();
 549
 550   for (cnt = 0; __gconv_path_elem[cnt].name != NULL; ++cnt)
 551     {
 552       const char *elem = __gconv_path_elem[cnt].name;
 553       size_t elem_len = __gconv_path_elem[cnt].len;
 554       char *filename;
 555
 556       /* No slash needs to be inserted between elem and gconv_conf_filename;
 557          elem already ends in a slash.  */
 558       filename = alloca (elem_len + sizeof (gconv_conf_filename));
 559       __mempcpy (__mempcpy (filename, elem, elem_len),
 560                  gconv_conf_filename, sizeof (gconv_conf_filename));
 561
 562       /* Read the next configuration file.  */
 563       read_conf_file (filename, elem, elem_len, &modules, &nmodules);
 564     }
 565 #endif
 566
 567   /* Add the internal modules.  */
 568   for (cnt = 0; cnt < sizeof (builtin_modules) / sizeof (builtin_modules[0]);
 569        ++cnt)
 570     {
 571       struct gconv_alias fake_alias;
 572
 573       fake_alias.fromname = (char *) builtin_modules[cnt].from_string;
 574
 575       if (__tfind (&fake_alias, &__gconv_alias_db, __gconv_alias_compare)
 576           != NULL)
 577         /* It'll conflict so don't add it.  */
 578         continue;
 579
 580       insert_module (&builtin_modules[cnt], 0);
 581     }
 582
 583   /* Add aliases for builtin conversions.  */
 584   cnt = sizeof (builtin_aliases) / sizeof (builtin_aliases[0]);
 585   while (cnt > 0)
 586     {
 587       char *copy = strdupa (builtin_aliases[--cnt]);
 588       add_alias (copy, modules);
 589     }
 590
 591   /* Restore the error number.  */
 592   __set_errno (save_errno);
 593 }
 594
 595
 596
 597 /* Free all resources if necessary.  */
 598 static void __attribute__ ((unused))
 599 free_mem (void)
 600 {
 601   if (__gconv_path_elem != NULL && __gconv_path_elem != &empty_path_elem)
 602     free ((void *) __gconv_path_elem);
 603 }
 604
 605 text_set_element (__libc_subfreeres, free_mem);