iconv/gconv_conf.c

   1 /* Handle configuration data.
   2    Copyright (C) 1997,98,99,2000,2001,2002 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, 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 " " To,
  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 /* Add new alias.  */
 128 static inline void
 129 add_alias (char *rp, void *modules)
 130 {
 131   /* We now expect two more string.  The strings are normalized
 132      (converted to UPPER case) and strored in the alias database.  */
 133   struct gconv_alias *new_alias;
 134   char *from, *to, *wp;
 135
 136   while (__isspace_l (*rp, &_nl_C_locobj))
 137     ++rp;
 138   from = wp = rp;
 139   while (*rp != '\0' && !__isspace_l (*rp, &_nl_C_locobj))
 140     *wp++ = __toupper_l (*rp++, &_nl_C_locobj);
 141   if (*rp == '\0')
 142     /* There is no `to' string on the line.  Ignore it.  */
 143     return;
 144   *wp++ = '\0';
 145   to = ++rp;
 146   while (__isspace_l (*rp, &_nl_C_locobj))
 147     ++rp;
 148   while (*rp != '\0' && !__isspace_l (*rp, &_nl_C_locobj))
 149     *wp++ = __toupper_l (*rp++, &_nl_C_locobj);
 150   if (to == wp)
 151     /* No `to' string, ignore the line.  */
 152     return;
 153   *wp++ = '\0';
 154
 155   /* Test whether this alias conflicts with any available module.  */
 156   if (detect_conflict (from))
 157     /* It does conflict, don't add the alias.  */
 158     return;
 159
 160   new_alias = (struct gconv_alias *)
 161     malloc (sizeof (struct gconv_alias) + (wp - from));
 162   if (new_alias != NULL)
 163     {
 164       void **inserted;
 165
 166       new_alias->fromname = memcpy ((char *) new_alias
 167                                     + sizeof (struct gconv_alias),
 168                                     from, wp - from);
 169       new_alias->toname = new_alias->fromname + (to - from);
 170
 171       inserted = (void **) __tsearch (new_alias, &__gconv_alias_db,
 172                                       __gconv_alias_compare);
 173       if (inserted == NULL || *inserted != new_alias)
 174         /* Something went wrong, free this entry.  */
 175         free (new_alias);
 176     }
 177 }
 178
 179
 180 /* Insert a data structure for a new module in the search tree.  */
 181 static inline void
 182 internal_function
 183 insert_module (struct gconv_module *newp, int tobefreed)
 184 {
 185   struct gconv_module **rootp = &__gconv_modules_db;
 186
 187   while (*rootp != NULL)
 188     {
 189       struct gconv_module *root = *rootp;
 190       int cmpres;
 191
 192       cmpres = strcmp (newp->from_string, root->from_string);
 193       if (cmpres == 0)
 194         {
 195           /* Both strings are identical.  Insert the string at the
 196              end of the `same' list if it is not already there.  */
 197           while (strcmp (newp->from_string, root->from_string) != 0
 198                  || strcmp (newp->to_string, root->to_string) != 0)
 199             {
 200               rootp = &root->same;
 201               root = *rootp;
 202               if (root == NULL)
 203                 break;
 204             }
 205
 206           if (root != NULL)
 207             {
 208               /* This is a no new conversion.  But maybe the cost is
 209                  better.  */
 210               if (newp->cost_hi < root->cost_hi
 211                   || (newp->cost_hi == root->cost_hi
 212                       && newp->cost_lo < root->cost_lo))
 213                 {
 214                   newp->left = root->left;
 215                   newp->right = root->right;
 216                   newp->same = root->same;
 217                   *rootp = newp;
 218
 219                   free (root);
 220                 }
 221               else if (tobefreed)
 222                 free (newp);
 223               return;
 224             }
 225
 226           break;
 227         }
 228       else if (cmpres < 0)
 229         rootp = &root->left;
 230       else
 231         rootp = &root->right;
 232     }
 233
 234   /* Plug in the new node here.  */
 235   *rootp = newp;
 236 }
 237
 238
 239 /* Add new module.  */
 240 static void
 241 internal_function
 242 add_module (char *rp, const char *directory, size_t dir_len, void **modules,
 243             size_t *nmodules, int modcounter)
 244 {
 245   /* We expect now
 246      1. `from' name
 247      2. `to' name
 248      3. filename of the module
 249      4. an optional cost value
 250   */
 251   struct gconv_alias fake_alias;
 252   struct gconv_module *new_module;
 253   char *from, *to, *module, *wp;
 254   int need_ext;
 255   int cost_hi;
 256
 257   while (__isspace_l (*rp, &_nl_C_locobj))
 258     ++rp;
 259   from = rp;
 260   while (*rp != '\0' && !__isspace_l (*rp, &_nl_C_locobj))
 261     {
 262       *rp = __toupper_l (*rp, &_nl_C_locobj);
 263       ++rp;
 264     }
 265   if (*rp == '\0')
 266     return;
 267   *rp++ = '\0';
 268   to = wp = rp;
 269   while (__isspace_l (*rp, &_nl_C_locobj))
 270     ++rp;
 271   while (*rp != '\0' && !__isspace_l (*rp, &_nl_C_locobj))
 272     *wp++ = __toupper_l (*rp++, &_nl_C_locobj);
 273   if (*rp == '\0')
 274     return;
 275   *wp++ = '\0';
 276   do
 277     ++rp;
 278   while (__isspace_l (*rp, &_nl_C_locobj));
 279   module = wp;
 280   while (*rp != '\0' && !__isspace_l (*rp, &_nl_C_locobj))
 281     *wp++ = *rp++;
 282   if (*rp == '\0')
 283     {
 284       /* There is no cost, use one by default.  */
 285       *wp++ = '\0';
 286       cost_hi = 1;
 287     }
 288   else
 289     {
 290       /* There might be a cost value.  */
 291       char *endp;
 292
 293       *wp++ = '\0';
 294       cost_hi = strtol (rp, &endp, 10);
 295       if (rp == endp || cost_hi < 1)
 296         /* No useful information.  */
 297         cost_hi = 1;
 298     }
 299
 300   if (module[0] == '\0')
 301     /* No module name given.  */
 302     return;
 303   if (module[0] == '/')
 304     dir_len = 0;
 305
 306   /* See whether we must add the ending.  */
 307   need_ext = 0;
 308   if (wp - module < (ptrdiff_t) sizeof (gconv_module_ext)
 309       || memcmp (wp - sizeof (gconv_module_ext), gconv_module_ext,
 310                  sizeof (gconv_module_ext)) != 0)
 311     /* We must add the module extension.  */
 312     need_ext = sizeof (gconv_module_ext) - 1;
 313
 314   /* See whether we have already an alias with this name defined.  */
 315   fake_alias.fromname = strndupa (from, to - from);
 316
 317   if (__tfind (&fake_alias, &__gconv_alias_db, __gconv_alias_compare) != NULL)
 318     /* This module duplicates an alias.  */
 319     return;
 320
 321   new_module = (struct gconv_module *) calloc (1,
 322                                                sizeof (struct gconv_module)
 323                                                + (wp - from)
 324                                                + dir_len + need_ext);
 325   if (new_module != NULL)
 326     {
 327       char *tmp;
 328
 329       new_module->from_string = tmp = (char *) (new_module + 1);
 330       tmp = __mempcpy (tmp, from, to - from);
 331
 332       new_module->to_string = tmp;
 333       tmp = __mempcpy (tmp, to, module - to);
 334
 335       new_module->cost_hi = cost_hi;
 336       new_module->cost_lo = modcounter;
 337
 338       new_module->module_name = tmp;
 339
 340       if (dir_len != 0)
 341         tmp = __mempcpy (tmp, directory, dir_len);
 342
 343       tmp = __mempcpy (tmp, module, wp - module);
 344
 345       if (need_ext)
 346         memcpy (tmp - 1, gconv_module_ext, sizeof (gconv_module_ext));
 347
 348       /* Now insert the new module data structure in our search tree.  */
 349       insert_module (new_module, 1);
 350     }
 351 }
 352
 353
 354 /* Read the next configuration file.  */
 355 static void
 356 internal_function
 357 read_conf_file (const char *filename, const char *directory, size_t dir_len,
 358                 void **modules, size_t *nmodules)
 359 {
 360   FILE *fp = fopen (filename, "r");
 361   char *line = NULL;
 362   size_t line_len = 0;
 363   static int modcounter;
 364
 365   /* Don't complain if a file is not present or readable, simply silently
 366      ignore it.  */
 367   if (fp == NULL)
 368     return;
 369
 370   /* No threads reading from this stream.  */
 371   __fsetlocking (fp, FSETLOCKING_BYCALLER);
 372
 373   /* Process the known entries of the file.  Comments start with `#' and
 374      end with the end of the line.  Empty lines are ignored.  */
 375   while (!feof_unlocked (fp))
 376     {
 377       char *rp, *endp, *word;
 378       ssize_t n = __getdelim (&line, &line_len, '\n', fp);
 379       if (n < 0)
 380         /* An error occurred.  */
 381         break;
 382
 383       rp = line;
 384       /* Terminate the line (excluding comments or newline) by an NUL byte
 385          to simplify the following code.  */
 386       endp = strchr (rp, '#');
 387       if (endp != NULL)
 388         *endp = '\0';
 389       else
 390         if (rp[n - 1] == '\n')
 391           rp[n - 1] = '\0';
 392
 393       while (__isspace_l (*rp, &_nl_C_locobj))
 394         ++rp;
 395
 396       /* If this is an empty line go on with the next one.  */
 397       if (rp == endp)
 398         continue;
 399
 400       word = rp;
 401       while (*rp != '\0' && !__isspace_l (*rp, &_nl_C_locobj))
 402         ++rp;
 403
 404       if (rp - word == sizeof ("alias") - 1
 405           && memcmp (word, "alias", sizeof ("alias") - 1) == 0)
 406         add_alias (rp, *modules);
 407       else if (rp - word == sizeof ("module") - 1
 408                && memcmp (word, "module", sizeof ("module") - 1) == 0)
 409         add_module (rp, directory, dir_len, modules, nmodules, modcounter++);
 410       /* else */
 411         /* Otherwise ignore the line.  */
 412     }
 413
 414   free (line);
 415
 416   fclose (fp);
 417 }
 418
 419
 420 /* Determine the directories we are looking for data in.  */
 421 void
 422 internal_function
 423 __gconv_get_path (void)
 424 {
 425   struct path_elem *result;
 426   __libc_lock_define_initialized (static, lock);
 427
 428   __libc_lock_lock (lock);
 429
 430   /* Make sure there wasn't a second thread doing it already.  */
 431   result = (struct path_elem *) __gconv_path_elem;
 432   if (result == NULL)
 433     {
 434       /* Determine the complete path first.  */
 435       char *gconv_path;
 436       size_t gconv_path_len;
 437       char *elem;
 438       char *oldp;
 439       char *cp;
 440       int nelems;
 441       char *cwd;
 442       size_t cwdlen;
 443
 444       if (__gconv_path_envvar == NULL)
 445         {
 446           /* No user-defined path.  Make a modifiable copy of the
 447              default path.  */
 448           gconv_path = strdupa (default_gconv_path);
 449           gconv_path_len = sizeof (default_gconv_path);
 450           cwd = NULL;
 451           cwdlen = 0;
 452         }
 453       else
 454         {
 455           /* Append the default path to the user-defined path.  */
 456           size_t user_len = strlen (__gconv_path_envvar);
 457
 458           gconv_path_len = user_len + 1 + sizeof (default_gconv_path);
 459           gconv_path = alloca (gconv_path_len);
 460           __mempcpy (__mempcpy (__mempcpy (gconv_path, __gconv_path_envvar,
 461                                            user_len),
 462                                 ":", 1),
 463                      default_gconv_path, sizeof (default_gconv_path));
 464           cwd = __getcwd (NULL, 0);
 465           cwdlen = strlen (cwd);
 466         }
 467       assert (default_gconv_path[0] == '/');
 468
 469       /* In a first pass we calculate the number of elements.  */
 470       oldp = NULL;
 471       cp = strchr (gconv_path, ':');
 472       nelems = 1;
 473       while (cp != NULL)
 474         {
 475           if (cp != oldp + 1)
 476             ++nelems;
 477           oldp = cp;
 478           cp =  strchr (cp + 1, ':');
 479         }
 480
 481       /* Allocate the memory for the result.  */
 482       result = (struct path_elem *) malloc ((nelems + 1)
 483                                             * sizeof (struct path_elem)
 484                                             + gconv_path_len + nelems
 485                                             + (nelems - 1) * (cwdlen + 1));
 486       if (result != NULL)
 487         {
 488           char *strspace = (char *) &result[nelems + 1];
 489           int n = 0;
 490
 491           /* Separate the individual parts.  */
 492           __gconv_max_path_elem_len = 0;
 493           elem = __strtok_r (gconv_path, ":", &gconv_path);
 494           assert (elem != NULL);
 495           do
 496             {
 497               result[n].name = strspace;
 498               if (elem[0] != '/')
 499                 {
 500                   assert (cwd != NULL);
 501                   strspace = __mempcpy (strspace, cwd, cwdlen);
 502                   *strspace++ = '/';
 503                 }
 504               strspace = __stpcpy (strspace, elem);
 505               if (strspace[-1] != '/')
 506                 *strspace++ = '/';
 507
 508               result[n].len = strspace - result[n].name;
 509               if (result[n].len > __gconv_max_path_elem_len)
 510                 __gconv_max_path_elem_len = result[n].len;
 511
 512               *strspace++ = '\0';
 513               ++n;
 514             }
 515           while ((elem = __strtok_r (NULL, ":", &gconv_path)) != NULL);
 516
 517           result[n].name = NULL;
 518           result[n].len = 0;
 519         }
 520
 521       __gconv_path_elem = result ?: (struct path_elem *) &empty_path_elem;
 522
 523       if (cwd != NULL)
 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   cnt = sizeof (builtin_aliases) / sizeof (builtin_aliases[0]);
 590   while (cnt > 0)
 591     {
 592       char *copy = strdupa (builtin_aliases[--cnt]);
 593       add_alias (copy, modules);
 594     }
 595
 596   /* Restore the error number.  */
 597   __set_errno (save_errno);
 598 }
 599
 600
 601
 602 /* Free all resources if necessary.  */
 603 libc_freeres_fn (free_mem)
 604 {
 605   if (__gconv_path_elem != NULL && __gconv_path_elem != &empty_path_elem)
 606     free ((void *) __gconv_path_elem);
 607 }