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