src/basic/locale-util.c

   1 /* SPDX-License-Identifier: LGPL-2.1+ */
   2
   3 #include <dirent.h>
   4 #include <errno.h>
   5 #include <fcntl.h>
   6 #include <ftw.h>
   7 #include <langinfo.h>
   8 #include <libintl.h>
   9 #include <locale.h>
  10 #include <stddef.h>
  11 #include <stdint.h>
  12 #include <stdlib.h>
  13 #include <string.h>
  14 #include <sys/mman.h>
  15 #include <sys/stat.h>
  16
  17 #include "def.h"
  18 #include "dirent-util.h"
  19 #include "fd-util.h"
  20 #include "hashmap.h"
  21 #include "locale-util.h"
  22 #include "path-util.h"
  23 #include "set.h"
  24 #include "string-table.h"
  25 #include "string-util.h"
  26 #include "strv.h"
  27 #include "utf8.h"
  28
  29 static int add_locales_from_archive(Set *locales) {
  30         /* Stolen from glibc... */
  31
  32         struct locarhead {
  33                 uint32_t magic;
  34                 /* Serial number.  */
  35                 uint32_t serial;
  36                 /* Name hash table.  */
  37                 uint32_t namehash_offset;
  38                 uint32_t namehash_used;
  39                 uint32_t namehash_size;
  40                 /* String table.  */
  41                 uint32_t string_offset;
  42                 uint32_t string_used;
  43                 uint32_t string_size;
  44                 /* Table with locale records.  */
  45                 uint32_t locrectab_offset;
  46                 uint32_t locrectab_used;
  47                 uint32_t locrectab_size;
  48                 /* MD5 sum hash table.  */
  49                 uint32_t sumhash_offset;
  50                 uint32_t sumhash_used;
  51                 uint32_t sumhash_size;
  52         };
  53
  54         struct namehashent {
  55                 /* Hash value of the name.  */
  56                 uint32_t hashval;
  57                 /* Offset of the name in the string table.  */
  58                 uint32_t name_offset;
  59                 /* Offset of the locale record.  */
  60                 uint32_t locrec_offset;
  61         };
  62
  63         const struct locarhead *h;
  64         const struct namehashent *e;
  65         const void *p = MAP_FAILED;
  66         _cleanup_close_ int fd = -1;
  67         size_t sz = 0;
  68         struct stat st;
  69         size_t i;
  70         int r;
  71
  72         fd = open("/usr/lib/locale/locale-archive", O_RDONLY|O_NOCTTY|O_CLOEXEC);
  73         if (fd < 0)
  74                 return errno == ENOENT ? 0 : -errno;
  75
  76         if (fstat(fd, &st) < 0)
  77                 return -errno;
  78
  79         if (!S_ISREG(st.st_mode))
  80                 return -EBADMSG;
  81
  82         if (st.st_size < (off_t) sizeof(struct locarhead))
  83                 return -EBADMSG;
  84
  85         p = mmap(NULL, st.st_size, PROT_READ, MAP_SHARED, fd, 0);
  86         if (p == MAP_FAILED)
  87                 return -errno;
  88
  89         h = (const struct locarhead *) p;
  90         if (h->magic != 0xde020109 ||
  91             h->namehash_offset + h->namehash_size > st.st_size ||
  92             h->string_offset + h->string_size > st.st_size ||
  93             h->locrectab_offset + h->locrectab_size > st.st_size ||
  94             h->sumhash_offset + h->sumhash_size > st.st_size) {
  95                 r = -EBADMSG;
  96                 goto finish;
  97         }
  98
  99         e = (const struct namehashent*) ((const uint8_t*) p + h->namehash_offset);
 100         for (i = 0; i < h->namehash_size; i++) {
 101                 char *z;
 102
 103                 if (e[i].locrec_offset == 0)
 104                         continue;
 105
 106                 if (!utf8_is_valid((char*) p + e[i].name_offset))
 107                         continue;
 108
 109                 z = strdup((char*) p + e[i].name_offset);
 110                 if (!z) {
 111                         r = -ENOMEM;
 112                         goto finish;
 113                 }
 114
 115                 r = set_consume(locales, z);
 116                 if (r < 0)
 117                         goto finish;
 118         }
 119
 120         r = 0;
 121
 122  finish:
 123         if (p != MAP_FAILED)
 124                 munmap((void*) p, sz);
 125
 126         return r;
 127 }
 128
 129 static int add_locales_from_libdir (Set *locales) {
 130         _cleanup_closedir_ DIR *dir = NULL;
 131         struct dirent *entry;
 132         int r;
 133
 134         dir = opendir("/usr/lib/locale");
 135         if (!dir)
 136                 return errno == ENOENT ? 0 : -errno;
 137
 138         FOREACH_DIRENT(entry, dir, return -errno) {
 139                 char *z;
 140
 141                 dirent_ensure_type(dir, entry);
 142
 143                 if (entry->d_type != DT_DIR)
 144                         continue;
 145
 146                 z = strdup(entry->d_name);
 147                 if (!z)
 148                         return -ENOMEM;
 149
 150                 r = set_consume(locales, z);
 151                 if (r < 0 && r != -EEXIST)
 152                         return r;
 153         }
 154
 155         return 0;
 156 }
 157
 158 int get_locales(char ***ret) {
 159         _cleanup_set_free_ Set *locales = NULL;
 160         _cleanup_strv_free_ char **l = NULL;
 161         int r;
 162
 163         locales = set_new(&string_hash_ops);
 164         if (!locales)
 165                 return -ENOMEM;
 166
 167         r = add_locales_from_archive(locales);
 168         if (r < 0 && r != -ENOENT)
 169                 return r;
 170
 171         r = add_locales_from_libdir(locales);
 172         if (r < 0)
 173                 return r;
 174
 175         l = set_get_strv(locales);
 176         if (!l)
 177                 return -ENOMEM;
 178
 179         strv_sort(l);
 180
 181         *ret = TAKE_PTR(l);
 182
 183         return 0;
 184 }
 185
 186 bool locale_is_valid(const char *name) {
 187
 188         if (isempty(name))
 189                 return false;
 190
 191         if (strlen(name) >= 128)
 192                 return false;
 193
 194         if (!utf8_is_valid(name))
 195                 return false;
 196
 197         if (!filename_is_valid(name))
 198                 return false;
 199
 200         if (!string_is_safe(name))
 201                 return false;
 202
 203         return true;
 204 }
 205
 206 void init_gettext(void) {
 207         setlocale(LC_ALL, "");
 208         textdomain(GETTEXT_PACKAGE);
 209 }
 210
 211 bool is_locale_utf8(void) {
 212         const char *set;
 213         static int cached_answer = -1;
 214
 215         /* Note that we default to 'true' here, since today UTF8 is
 216          * pretty much supported everywhere. */
 217
 218         if (cached_answer >= 0)
 219                 goto out;
 220
 221         if (!setlocale(LC_ALL, "")) {
 222                 cached_answer = true;
 223                 goto out;
 224         }
 225
 226         set = nl_langinfo(CODESET);
 227         if (!set) {
 228                 cached_answer = true;
 229                 goto out;
 230         }
 231
 232         if (streq(set, "UTF-8")) {
 233                 cached_answer = true;
 234                 goto out;
 235         }
 236
 237         /* For LC_CTYPE=="C" return true, because CTYPE is effectly
 238          * unset and everything can do to UTF-8 nowadays. */
 239         set = setlocale(LC_CTYPE, NULL);
 240         if (!set) {
 241                 cached_answer = true;
 242                 goto out;
 243         }
 244
 245         /* Check result, but ignore the result if C was set
 246          * explicitly. */
 247         cached_answer =
 248                 STR_IN_SET(set, "C", "POSIX") &&
 249                 !getenv("LC_ALL") &&
 250                 !getenv("LC_CTYPE") &&
 251                 !getenv("LANG");
 252
 253 out:
 254         return (bool) cached_answer;
 255 }
 256
 257 static thread_local Set *keymaps = NULL;
 258
 259 static int nftw_cb(
 260                 const char *fpath,
 261                 const struct stat *sb,
 262                 int tflag,
 263                 struct FTW *ftwbuf) {
 264
 265         char *p, *e;
 266         int r;
 267
 268         if (tflag != FTW_F)
 269                 return 0;
 270
 271         if (!endswith(fpath, ".map") &&
 272             !endswith(fpath, ".map.gz"))
 273                 return 0;
 274
 275         p = strdup(basename(fpath));
 276         if (!p)
 277                 return FTW_STOP;
 278
 279         e = endswith(p, ".map");
 280         if (e)
 281                 *e = 0;
 282
 283         e = endswith(p, ".map.gz");
 284         if (e)
 285                 *e = 0;
 286
 287         r = set_consume(keymaps, p);
 288         if (r < 0 && r != -EEXIST)
 289                 return r;
 290
 291         return 0;
 292 }
 293
 294 int get_keymaps(char ***ret) {
 295         _cleanup_strv_free_ char **l = NULL;
 296         const char *dir;
 297         int r;
 298
 299         keymaps = set_new(&string_hash_ops);
 300         if (!keymaps)
 301                 return -ENOMEM;
 302
 303         NULSTR_FOREACH(dir, KBD_KEYMAP_DIRS) {
 304                 r = nftw(dir, nftw_cb, 20, FTW_PHYS|FTW_ACTIONRETVAL);
 305
 306                 if (r == FTW_STOP)
 307                         log_debug("Directory not found %s", dir);
 308                 else if (r < 0)
 309                         log_debug_errno(r, "Can't add keymap: %m");
 310         }
 311
 312         l = set_get_strv(keymaps);
 313         if (!l) {
 314                 set_free_free(keymaps);
 315                 return -ENOMEM;
 316         }
 317
 318         set_free(keymaps);
 319
 320         if (strv_isempty(l))
 321                 return -ENOENT;
 322
 323         strv_sort(l);
 324
 325         *ret = TAKE_PTR(l);
 326
 327         return 0;
 328 }
 329
 330 bool keymap_is_valid(const char *name) {
 331
 332         if (isempty(name))
 333                 return false;
 334
 335         if (strlen(name) >= 128)
 336                 return false;
 337
 338         if (!utf8_is_valid(name))
 339                 return false;
 340
 341         if (!filename_is_valid(name))
 342                 return false;
 343
 344         if (!string_is_safe(name))
 345                 return false;
 346
 347         return true;
 348 }
 349
 350 const char *special_glyph(SpecialGlyph code) {
 351
 352         /* A list of a number of interesting unicode glyphs we can use to decorate our output. It's probably wise to be
 353          * conservative here, and primarily stick to the glyphs defined in the eurlatgr font, so that display still
 354          * works reasonably well on the Linux console. For details see:
 355          *
 356          * http://git.altlinux.org/people/legion/packages/kbd.git?p=kbd.git;a=blob;f=data/consolefonts/README.eurlatgr
 357          */
 358
 359         static const char* const draw_table[2][_SPECIAL_GLYPH_MAX] = {
 360                 /* ASCII fallback */
 361                 [false] = {
 362                         [TREE_VERTICAL]      = "| ",
 363                         [TREE_BRANCH]        = "|-",
 364                         [TREE_RIGHT]         = "`-",
 365                         [TREE_SPACE]         = "  ",
 366                         [TRIANGULAR_BULLET]  = ">",
 367                         [BLACK_CIRCLE]       = "*",
 368                         [ARROW]              = "->",
 369                         [MDASH]              = "-",
 370                         [ELLIPSIS]           = "..."
 371                 },
 372
 373                 /* UTF-8 */
 374                 [true] = {
 375                         [TREE_VERTICAL]      = "\342\224\202 ",            /* │  */
 376                         [TREE_BRANCH]        = "\342\224\234\342\224\200", /* ├─ */
 377                         [TREE_RIGHT]         = "\342\224\224\342\224\200", /* └─ */
 378                         [TREE_SPACE]         = "  ",                       /*    */
 379                         [TRIANGULAR_BULLET]  = "\342\200\243",             /* ‣ */
 380                         [BLACK_CIRCLE]       = "\342\227\217",             /* ● */
 381                         [ARROW]              = "\342\206\222",             /* → */
 382                         [MDASH]              = "\342\200\223",             /* – */
 383                         [ELLIPSIS]           = "\342\200\246",             /* … */
 384                 },
 385         };
 386
 387         return draw_table[is_locale_utf8()][code];
 388 }
 389
 390 static const char * const locale_variable_table[_VARIABLE_LC_MAX] = {
 391         [VARIABLE_LANG] = "LANG",
 392         [VARIABLE_LANGUAGE] = "LANGUAGE",
 393         [VARIABLE_LC_CTYPE] = "LC_CTYPE",
 394         [VARIABLE_LC_NUMERIC] = "LC_NUMERIC",
 395         [VARIABLE_LC_TIME] = "LC_TIME",
 396         [VARIABLE_LC_COLLATE] = "LC_COLLATE",
 397         [VARIABLE_LC_MONETARY] = "LC_MONETARY",
 398         [VARIABLE_LC_MESSAGES] = "LC_MESSAGES",
 399         [VARIABLE_LC_PAPER] = "LC_PAPER",
 400         [VARIABLE_LC_NAME] = "LC_NAME",
 401         [VARIABLE_LC_ADDRESS] = "LC_ADDRESS",
 402         [VARIABLE_LC_TELEPHONE] = "LC_TELEPHONE",
 403         [VARIABLE_LC_MEASUREMENT] = "LC_MEASUREMENT",
 404         [VARIABLE_LC_IDENTIFICATION] = "LC_IDENTIFICATION"
 405 };
 406
 407 DEFINE_STRING_TABLE_LOOKUP(locale_variable, LocaleVariable);