iconv/gconv_conf.c

   1 /* Handle configuration data.
   2    Copyright (C) 1997-2003, 2005, 2006 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 #ifdef USE_IN_LIBIO
  94 # include <libio/libioP.h>
  95 # define __getdelim(line, len, c, fp) _IO_getdelim (line, len, c, fp)
  96 #endif
  97
  98
  99 /* Value of the GCONV_PATH environment variable.  */
 100 const char *__gconv_path_envvar;
 101
 102
 103 /* Test whether there is already a matching module known.  */
 104 static int
 105 internal_function
 106 detect_conflict (const char *alias)
 107 {
 108   struct gconv_module *node = __gconv_modules_db;
 109
 110   while (node != NULL)
 111     {
 112       int cmpres = strcmp (alias, node->from_string);
 113
 114       if (cmpres == 0)
 115         /* We have a conflict.  */
 116         return 1;
 117       else if (cmpres < 0)
 118         node = node->left;
 119       else
 120         node = node->right;
 121     }
 122
 123   return node != NULL;
 124 }
 125
 126
 127 /* The actual code to add aliases.  */
 128 static void
 129 add_alias2 (const char *from, const char *to, const char *wp, void *modules)
 130 {
 131   /* Test whether this alias conflicts with any available module.  */
 132   if (detect_conflict (from))
 133     /* It does conflict, don't add the alias.  */
 134     return;
 135
 136   struct gconv_alias *new_alias = (struct gconv_alias *)
 137     malloc (sizeof (struct gconv_alias) + (wp - from));
 138   if (new_alias != NULL)
 139     {
 140       void **inserted;
 141
 142       new_alias->fromname = memcpy ((char *) new_alias
 143                                     + sizeof (struct gconv_alias),
 144                                     from, wp - from);
 145       new_alias->toname = new_alias->fromname + (to - from);
 146
 147       inserted = (void **) __tsearch (new_alias, &__gconv_alias_db,
 148                                       __gconv_alias_compare);
 149       if (inserted == NULL || *inserted != new_alias)
 150         /* Something went wrong, free this entry.  */
 151         free (new_alias);
 152     }
 153 }
 154
 155
 156 /* Add new alias.  */
 157 static void
 158 add_alias (char *rp, void *modules)
 159 {
 160   /* We now expect two more string.  The strings are normalized
 161      (converted to UPPER case) and strored in the alias database.  */
 162   char *from, *to, *wp;
 163
 164   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 165     ++rp;
 166   from = wp = rp;
 167   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 168     *wp++ = __toupper_l (*rp++, _nl_C_locobj_ptr);
 169   if (*rp == '\0')
 170     /* There is no `to' string on the line.  Ignore it.  */
 171     return;
 172   *wp++ = '\0';
 173   to = ++rp;
 174   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 175     ++rp;
 176   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 177     *wp++ = __toupper_l (*rp++, _nl_C_locobj_ptr);
 178   if (to == wp)
 179     /* No `to' string, ignore the line.  */
 180     return;
 181   *wp++ = '\0';
 182
 183   add_alias2 (from, to, wp, modules);
 184 }
 185
 186
 187 /* Insert a data structure for a new module in the search tree.  */
 188 static void
 189 internal_function
 190 insert_module (struct gconv_module *newp, int tobefreed)
 191 {
 192   struct gconv_module **rootp = &__gconv_modules_db;
 193
 194   while (*rootp != NULL)
 195     {
 196       struct gconv_module *root = *rootp;
 197       int cmpres;
 198
 199       cmpres = strcmp (newp->from_string, root->from_string);
 200       if (cmpres == 0)
 201         {
 202           /* Both strings are identical.  Insert the string at the
 203              end of the `same' list if it is not already there.  */
 204           while (strcmp (newp->from_string, root->from_string) != 0
 205                  || strcmp (newp->to_string, root->to_string) != 0)
 206             {
 207               rootp = &root->same;
 208               root = *rootp;
 209               if (root == NULL)
 210                 break;
 211             }
 212
 213           if (root != NULL)
 214             {
 215               /* This is a no new conversion.  But maybe the cost is
 216                  better.  */
 217               if (newp->cost_hi < root->cost_hi
 218                   || (newp->cost_hi == root->cost_hi
 219                       && newp->cost_lo < root->cost_lo))
 220                 {
 221                   newp->left = root->left;
 222                   newp->right = root->right;
 223                   newp->same = root->same;
 224                   *rootp = newp;
 225
 226                   free (root);
 227                 }
 228               else if (tobefreed)
 229                 free (newp);
 230               return;
 231             }
 232
 233           break;
 234         }
 235       else if (cmpres < 0)
 236         rootp = &root->left;
 237       else
 238         rootp = &root->right;
 239     }
 240
 241   /* Plug in the new node here.  */
 242   *rootp = newp;
 243 }
 244
 245
 246 /* Add new module.  */
 247 static void
 248 internal_function
 249 add_module (char *rp, const char *directory, size_t dir_len, void **modules,
 250             size_t *nmodules, int modcounter)
 251 {
 252   /* We expect now
 253      1. `from' name
 254      2. `to' name
 255      3. filename of the module
 256      4. an optional cost value
 257   */
 258   struct gconv_alias fake_alias;
 259   struct gconv_module *new_module;
 260   char *from, *to, *module, *wp;
 261   int need_ext;
 262   int cost_hi;
 263
 264   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 265     ++rp;
 266   from = rp;
 267   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 268     {
 269       *rp = __toupper_l (*rp, _nl_C_locobj_ptr);
 270       ++rp;
 271     }
 272   if (*rp == '\0')
 273     return;
 274   *rp++ = '\0';
 275   to = wp = rp;
 276   while (__isspace_l (*rp, _nl_C_locobj_ptr))
 277     ++rp;
 278   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 279     *wp++ = __toupper_l (*rp++, _nl_C_locobj_ptr);
 280   if (*rp == '\0')
 281     return;
 282   *wp++ = '\0';
 283   do
 284     ++rp;
 285   while (__isspace_l (*rp, _nl_C_locobj_ptr));
 286   module = wp;
 287   while (*rp != '\0' && !__isspace_l (*rp, _nl_C_locobj_ptr))
 288     *wp++ = *rp++;
 289   if (*rp == '\0')
 290     {
 291       /* There is no cost, use one by default.  */
 292       *wp++ = '\0';
 293       cost_hi = 1;
 294     }
 295   else
 296     {
 297       /* There might be a cost value.  */
 298       char *endp;
 299
 300       *wp++ = '\0';
 301       cost_hi = strtol (rp, &endp, 10);
 302       if (rp == endp || cost_hi < 1)
 303         /* No useful information.  */
 304         cost_hi = 1;
 305     }
 306
 307   if (module[0] == '\0')
 308     /* No module name given.  */
 309     return;
 310   if (module[0] == '/')
 311     dir_len = 0;
 312
 313   /* See whether we must add the ending.  */
 314   need_ext = 0;
 315   if (wp - module < (ptrdiff_t) sizeof (gconv_module_ext)
 316       || memcmp (wp - sizeof (gconv_module_ext), gconv_module_ext,
 317                  sizeof (gconv_module_ext)) != 0)
 318     /* We must add the module extension.  */
 319     need_ext = sizeof (gconv_module_ext) - 1;
 320
 321   /* See whether we have already an alias with this name defined.  */
 322   fake_alias.fromname = strndupa (from, to - from);
 323
 324   if (__tfind (&fake_alias, &__gconv_alias_db, __gconv_alias_compare) != NULL)
 325     /* This module duplicates an alias.  */
 326     return;
 327
 328   new_module = (struct gconv_module *) calloc (1,
 329                                                sizeof (struct gconv_module)
 330                                                + (wp - from)
 331                                                + dir_len + need_ext);
 332   if (new_module != NULL)
 333     {
 334       char *tmp;
 335
 336       new_module->from_string = tmp = (char *) (new_module + 1);
 337       tmp = __mempcpy (tmp, from, to - from);
 338
 339       new_module->to_string = tmp;
 340       tmp = __mempcpy (tmp, to, module - to);
 341
 342       new_module->cost_hi = cost_hi;
 343       new_module->cost_lo = modcounter;
 344
 345       new_module->module_name = tmp;
 346
 347       if (dir_len != 0)
 348         tmp = __mempcpy (tmp, directory, dir_len);
 349
 350       tmp = __mempcpy (tmp, module, wp - module);
 351
 352       if (need_ext)
 353         memcpy (tmp - 1, gconv_module_ext, sizeof (gconv_module_ext));
 354
 355       /* Now insert the new module data structure in our search tree.  */
 356       insert_module (new_module, 1);
 357     }
 358 }
 359
 360
 361 /* Read the next configuration file.  */
 362 static void
 363 internal_function
 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, "rc");
 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.  */
 430 void
 431 internal_function
 432 __gconv_get_path (void)
 433 {
 434   struct path_elem *result;
 435   __libc_lock_define_initialized (static, lock);
 436
 437   __libc_lock_lock (lock);
 438
 439   /* Make sure there wasn't a second thread doing it already.  */
 440   result = (struct path_elem *) __gconv_path_elem;
 441   if (result == NULL)
 442     {
 443       /* Determine the complete path first.  */
 444       char *gconv_path;
 445       size_t gconv_path_len;
 446       char *elem;
 447       char *oldp;
 448       char *cp;
 449       int nelems;
 450       char *cwd;
 451       size_t cwdlen;
 452
 453       if (__gconv_path_envvar == NULL)
 454         {
 455           /* No user-defined path.  Make a modifiable copy of the
 456              default path.  */
 457           gconv_path = strdupa (default_gconv_path);
 458           gconv_path_len = sizeof (default_gconv_path);
 459           cwd = NULL;
 460           cwdlen = 0;
 461         }
 462       else
 463         {
 464           /* Append the default path to the user-defined path.  */
 465           size_t user_len = strlen (__gconv_path_envvar);
 466
 467           gconv_path_len = user_len + 1 + sizeof (default_gconv_path);
 468           gconv_path = alloca (gconv_path_len);
 469           __mempcpy (__mempcpy (__mempcpy (gconv_path, __gconv_path_envvar,
 470                                            user_len),
 471                                 ":", 1),
 472                      default_gconv_path, sizeof (default_gconv_path));
 473           cwd = __getcwd (NULL, 0);
 474           cwdlen = strlen (cwd);
 475         }
 476       assert (default_gconv_path[0] == '/');
 477
 478       /* In a first pass we calculate the number of elements.  */
 479       oldp = NULL;
 480       cp = strchr (gconv_path, ':');
 481       nelems = 1;
 482       while (cp != NULL)
 483         {
 484           if (cp != oldp + 1)
 485             ++nelems;
 486           oldp = cp;
 487           cp =  strchr (cp + 1, ':');
 488         }
 489
 490       /* Allocate the memory for the result.  */
 491       result = (struct path_elem *) malloc ((nelems + 1)
 492                                             * sizeof (struct path_elem)
 493                                             + gconv_path_len + nelems
 494                                             + (nelems - 1) * (cwdlen + 1));
 495       if (result != NULL)
 496         {
 497           char *strspace = (char *) &result[nelems + 1];
 498           int n = 0;
 499
 500           /* Separate the individual parts.  */
 501           __gconv_max_path_elem_len = 0;
 502           elem = __strtok_r (gconv_path, ":", &gconv_path);
 503           assert (elem != NULL);
 504           do
 505             {
 506               result[n].name = strspace;
 507               if (elem[0] != '/')
 508                 {
 509                   assert (cwd != NULL);
 510                   strspace = __mempcpy (strspace, cwd, cwdlen);
 511                   *strspace++ = '/';
 512                 }
 513               strspace = __stpcpy (strspace, elem);
 514               if (strspace[-1] != '/')
 515                 *strspace++ = '/';
 516
 517               result[n].len = strspace - result[n].name;
 518               if (result[n].len > __gconv_max_path_elem_len)
 519                 __gconv_max_path_elem_len = result[n].len;
 520
 521               *strspace++ = '\0';
 522               ++n;
 523             }
 524           while ((elem = __strtok_r (NULL, ":", &gconv_path)) != NULL);
 525
 526           result[n].name = NULL;
 527           result[n].len = 0;
 528         }
 529
 530       __gconv_path_elem = result ?: (struct path_elem *) &empty_path_elem;
 531
 532       if (cwd != NULL)
 533         free (cwd);
 534     }
 535
 536   __libc_lock_unlock (lock);
 537 }
 538
 539
 540 /* Read all configuration files found in the user-specified and the default
 541    path.  */
 542 void
 543 attribute_hidden
 544 __gconv_read_conf (void)
 545 {
 546   void *modules = NULL;
 547   size_t nmodules = 0;
 548   int save_errno = errno;
 549   size_t cnt;
 550
 551   /* First see whether we should use the cache.  */
 552   if (__gconv_load_cache () == 0)
 553     {
 554       /* Yes, we are done.  */
 555       __set_errno (save_errno);
 556       return;
 557     }
 558
 559 #ifndef STATIC_GCONV
 560   /* Find out where we have to look.  */
 561   if (__gconv_path_elem == NULL)
 562     __gconv_get_path ();
 563
 564   for (cnt = 0; __gconv_path_elem[cnt].name != NULL; ++cnt)
 565     {
 566       const char *elem = __gconv_path_elem[cnt].name;
 567       size_t elem_len = __gconv_path_elem[cnt].len;
 568       char *filename;
 569
 570       /* No slash needs to be inserted between elem and gconv_conf_filename;
 571          elem already ends in a slash.  */
 572       filename = alloca (elem_len + sizeof (gconv_conf_filename));
 573       __mempcpy (__mempcpy (filename, elem, elem_len),
 574                  gconv_conf_filename, sizeof (gconv_conf_filename));
 575
 576       /* Read the next configuration file.  */
 577       read_conf_file (filename, elem, elem_len, &modules, &nmodules);
 578     }
 579 #endif
 580
 581   /* Add the internal modules.  */
 582   for (cnt = 0; cnt < sizeof (builtin_modules) / sizeof (builtin_modules[0]);
 583        ++cnt)
 584     {
 585       struct gconv_alias fake_alias;
 586
 587       fake_alias.fromname = (char *) builtin_modules[cnt].from_string;
 588
 589       if (__tfind (&fake_alias, &__gconv_alias_db, __gconv_alias_compare)
 590           != NULL)
 591         /* It'll conflict so don't add it.  */
 592         continue;
 593
 594       insert_module (&builtin_modules[cnt], 0);
 595     }
 596
 597   /* Add aliases for builtin conversions.  */
 598   const char *cp = builtin_aliases;
 599   do
 600     {
 601       const char *from = cp;
 602       const char *to = __rawmemchr (from, '\0') + 1;
 603       cp = __rawmemchr (to, '\0') + 1;
 604
 605       add_alias2 (from, to, cp, modules);
 606     }
 607   while (*cp != '\0');
 608
 609   /* Restore the error number.  */
 610   __set_errno (save_errno);
 611 }
 612
 613
 614
 615 /* Free all resources if necessary.  */
 616 libc_freeres_fn (free_mem)
 617 {
 618   if (__gconv_path_elem != NULL && __gconv_path_elem != &empty_path_elem)
 619     free ((void *) __gconv_path_elem);
 620 }