src/basic/locale-util.c

   1 /* SPDX-License-Identifier: LGPL-2.1-or-later */
   2
   3 #include <errno.h>
   4 #include <fcntl.h>
   5 #include <langinfo.h>
   6 #include <libintl.h>
   7 #include <stddef.h>
   8 #include <stdint.h>
   9 #include <stdlib.h>
  10 #include <sys/mman.h>
  11 #include <sys/stat.h>
  12
  13 #include "constants.h"
  14 #include "dirent-util.h"
  15 #include "env-util.h"
  16 #include "fd-util.h"
  17 #include "fileio.h"
  18 #include "hashmap.h"
  19 #include "locale-util.h"
  20 #include "path-util.h"
  21 #include "set.h"
  22 #include "string-table.h"
  23 #include "string-util.h"
  24 #include "strv.h"
  25 #include "utf8.h"
  26
  27 static char *normalize_locale(const char *name) {
  28         const char *e;
  29
  30         /* Locale names are weird: glibc has some magic rules when looking for the charset name on disk: it
  31          * lowercases everything, and removes most special chars. This means the official .UTF-8 suffix
  32          * becomes .utf8 when looking things up on disk. When enumerating locales, let's do the reverse
  33          * operation, and go back to ".UTF-8" which appears to be the more commonly accepted name. We only do
  34          * that for UTF-8 however, since it's kinda the only charset that matters. */
  35
  36         e = endswith(name, ".utf8");
  37         if (e) {
  38                 _cleanup_free_ char *prefix = NULL;
  39
  40                 prefix = strndup(name, e - name);
  41                 if (!prefix)
  42                         return NULL;
  43
  44                 return strjoin(prefix, ".UTF-8");
  45         }
  46
  47         e = strstr(name, ".utf8@");
  48         if (e) {
  49                 _cleanup_free_ char *prefix = NULL;
  50
  51                 prefix = strndup(name, e - name);
  52                 if (!prefix)
  53                         return NULL;
  54
  55                 return strjoin(prefix, ".UTF-8@", e + 6);
  56         }
  57
  58         return strdup(name);
  59 }
  60
  61 static int add_locales_from_archive(Set *locales) {
  62         /* Stolen from glibc... */
  63
  64         struct locarhead {
  65                 uint32_t magic;
  66                 /* Serial number.  */
  67                 uint32_t serial;
  68                 /* Name hash table.  */
  69                 uint32_t namehash_offset;
  70                 uint32_t namehash_used;
  71                 uint32_t namehash_size;
  72                 /* String table.  */
  73                 uint32_t string_offset;
  74                 uint32_t string_used;
  75                 uint32_t string_size;
  76                 /* Table with locale records.  */
  77                 uint32_t locrectab_offset;
  78                 uint32_t locrectab_used;
  79                 uint32_t locrectab_size;
  80                 /* MD5 sum hash table.  */
  81                 uint32_t sumhash_offset;
  82                 uint32_t sumhash_used;
  83                 uint32_t sumhash_size;
  84         };
  85
  86         struct namehashent {
  87                 /* Hash value of the name.  */
  88                 uint32_t hashval;
  89                 /* Offset of the name in the string table.  */
  90                 uint32_t name_offset;
  91                 /* Offset of the locale record.  */
  92                 uint32_t locrec_offset;
  93         };
  94
  95         const struct locarhead *h;
  96         const struct namehashent *e;
  97         const void *p = MAP_FAILED;
  98         _cleanup_close_ int fd = -EBADF;
  99         size_t sz = 0;
 100         struct stat st;
 101         int r;
 102
 103         fd = open("/usr/lib/locale/locale-archive", O_RDONLY|O_NOCTTY|O_CLOEXEC);
 104         if (fd < 0)
 105                 return errno == ENOENT ? 0 : -errno;
 106
 107         if (fstat(fd, &st) < 0)
 108                 return -errno;
 109
 110         if (!S_ISREG(st.st_mode))
 111                 return -EBADMSG;
 112
 113         if (st.st_size < (off_t) sizeof(struct locarhead))
 114                 return -EBADMSG;
 115
 116         if (file_offset_beyond_memory_size(st.st_size))
 117                 return -EFBIG;
 118
 119         p = mmap(NULL, st.st_size, PROT_READ, MAP_SHARED, fd, 0);
 120         if (p == MAP_FAILED)
 121                 return -errno;
 122
 123         h = (const struct locarhead *) p;
 124         if (h->magic != 0xde020109 ||
 125             h->namehash_offset + h->namehash_size > st.st_size ||
 126             h->string_offset + h->string_size > st.st_size ||
 127             h->locrectab_offset + h->locrectab_size > st.st_size ||
 128             h->sumhash_offset + h->sumhash_size > st.st_size) {
 129                 r = -EBADMSG;
 130                 goto finish;
 131         }
 132
 133         e = (const struct namehashent*) ((const uint8_t*) p + h->namehash_offset);
 134         for (size_t i = 0; i < h->namehash_size; i++) {
 135                 char *z;
 136
 137                 if (e[i].locrec_offset == 0)
 138                         continue;
 139
 140                 if (!utf8_is_valid((char*) p + e[i].name_offset))
 141                         continue;
 142
 143                 z = normalize_locale((char*) p + e[i].name_offset);
 144                 if (!z) {
 145                         r = -ENOMEM;
 146                         goto finish;
 147                 }
 148
 149                 r = set_consume(locales, z);
 150                 if (r < 0)
 151                         goto finish;
 152         }
 153
 154         r = 0;
 155
 156  finish:
 157         if (p != MAP_FAILED)
 158                 munmap((void*) p, sz);
 159
 160         return r;
 161 }
 162
 163 static int add_locales_from_libdir(Set *locales) {
 164         _cleanup_closedir_ DIR *dir = NULL;
 165         int r;
 166
 167         dir = opendir("/usr/lib/locale");
 168         if (!dir)
 169                 return errno == ENOENT ? 0 : -errno;
 170
 171         FOREACH_DIRENT(de, dir, return -errno) {
 172                 char *z;
 173
 174                 if (de->d_type != DT_DIR)
 175                         continue;
 176
 177                 z = normalize_locale(de->d_name);
 178                 if (!z)
 179                         return -ENOMEM;
 180
 181                 r = set_consume(locales, z);
 182                 if (r < 0 && r != -EEXIST)
 183                         return r;
 184         }
 185
 186         return 0;
 187 }
 188
 189 int get_locales(char ***ret) {
 190         _cleanup_set_free_free_ Set *locales = NULL;
 191         _cleanup_strv_free_ char **l = NULL;
 192         int r;
 193
 194         locales = set_new(&string_hash_ops);
 195         if (!locales)
 196                 return -ENOMEM;
 197
 198         r = add_locales_from_archive(locales);
 199         if (r < 0 && r != -ENOENT)
 200                 return r;
 201
 202         r = add_locales_from_libdir(locales);
 203         if (r < 0)
 204                 return r;
 205
 206         char *locale;
 207         SET_FOREACH(locale, locales) {
 208                 r = locale_is_installed(locale);
 209                 if (r < 0)
 210                         return r;
 211                 if (r == 0)
 212                         free(set_remove(locales, locale));
 213         }
 214
 215         l = set_get_strv(locales);
 216         if (!l)
 217                 return -ENOMEM;
 218
 219         /* Now, all elements are owned by strv 'l'. Hence, do not call set_free_free(). */
 220         locales = set_free(locales);
 221
 222         r = getenv_bool("SYSTEMD_LIST_NON_UTF8_LOCALES");
 223         if (r == -ENXIO || r == 0) {
 224                 char **a, **b;
 225
 226                 /* Filter out non-UTF-8 locales, because it's 2019, by default */
 227                 for (a = b = l; *a; a++) {
 228
 229                         if (endswith(*a, "UTF-8") ||
 230                             strstr(*a, ".UTF-8@"))
 231                                 *(b++) = *a;
 232                         else
 233                                 free(*a);
 234                 }
 235
 236                 *b = NULL;
 237
 238         } else if (r < 0)
 239                 log_debug_errno(r, "Failed to parse $SYSTEMD_LIST_NON_UTF8_LOCALES as boolean");
 240
 241         strv_sort(l);
 242
 243         *ret = TAKE_PTR(l);
 244
 245         return 0;
 246 }
 247
 248 bool locale_is_valid(const char *name) {
 249
 250         if (isempty(name))
 251                 return false;
 252
 253         if (strlen(name) >= 128)
 254                 return false;
 255
 256         if (!utf8_is_valid(name))
 257                 return false;
 258
 259         if (!filename_is_valid(name))
 260                 return false;
 261
 262         if (!string_is_safe(name))
 263                 return false;
 264
 265         return true;
 266 }
 267
 268 int locale_is_installed(const char *name) {
 269         if (!locale_is_valid(name))
 270                 return false;
 271
 272         if (STR_IN_SET(name, "C", "POSIX")) /* These ones are always OK */
 273                 return true;
 274
 275         _cleanup_(freelocalep) locale_t loc =
 276                 newlocale(LC_ALL_MASK, name, 0);
 277         if (loc == (locale_t) 0)
 278                 return errno == ENOMEM ? -ENOMEM : false;
 279
 280         return true;
 281 }
 282
 283 void init_gettext(void) {
 284         setlocale(LC_ALL, "");
 285         textdomain(GETTEXT_PACKAGE);
 286 }
 287
 288 bool is_locale_utf8(void) {
 289         static int cached_answer = -1;
 290         const char *set;
 291         int r;
 292
 293         /* Note that we default to 'true' here, since today UTF8 is
 294          * pretty much supported everywhere. */
 295
 296         if (cached_answer >= 0)
 297                 goto out;
 298
 299         r = getenv_bool_secure("SYSTEMD_UTF8");
 300         if (r >= 0) {
 301                 cached_answer = r;
 302                 goto out;
 303         } else if (r != -ENXIO)
 304                 log_debug_errno(r, "Failed to parse $SYSTEMD_UTF8, ignoring: %m");
 305
 306         if (!setlocale(LC_ALL, "")) {
 307                 cached_answer = true;
 308                 goto out;
 309         }
 310
 311         set = nl_langinfo(CODESET);
 312         if (!set) {
 313                 cached_answer = true;
 314                 goto out;
 315         }
 316
 317         if (streq(set, "UTF-8")) {
 318                 cached_answer = true;
 319                 goto out;
 320         }
 321
 322         /* For LC_CTYPE=="C" return true, because CTYPE is effectively
 323          * unset and everything can do to UTF-8 nowadays. */
 324         set = setlocale(LC_CTYPE, NULL);
 325         if (!set) {
 326                 cached_answer = true;
 327                 goto out;
 328         }
 329
 330         /* Check result, but ignore the result if C was set
 331          * explicitly. */
 332         cached_answer =
 333                 STR_IN_SET(set, "C", "POSIX") &&
 334                 !getenv("LC_ALL") &&
 335                 !getenv("LC_CTYPE") &&
 336                 !getenv("LANG");
 337
 338 out:
 339         return (bool) cached_answer;
 340 }
 341
 342 void locale_variables_free(char *l[_VARIABLE_LC_MAX]) {
 343         if (!l)
 344                 return;
 345
 346         for (LocaleVariable i = 0; i < _VARIABLE_LC_MAX; i++)
 347                 l[i] = mfree(l[i]);
 348 }
 349
 350 void locale_variables_simplify(char *l[_VARIABLE_LC_MAX]) {
 351         assert(l);
 352
 353         for (LocaleVariable p = 0; p < _VARIABLE_LC_MAX; p++) {
 354                 if (p == VARIABLE_LANG)
 355                         continue;
 356                 if (isempty(l[p]) || streq_ptr(l[VARIABLE_LANG], l[p]))
 357                         l[p] = mfree(l[p]);
 358         }
 359 }
 360
 361 static const char * const locale_variable_table[_VARIABLE_LC_MAX] = {
 362         [VARIABLE_LANG]              = "LANG",
 363         [VARIABLE_LANGUAGE]          = "LANGUAGE",
 364         [VARIABLE_LC_CTYPE]          = "LC_CTYPE",
 365         [VARIABLE_LC_NUMERIC]        = "LC_NUMERIC",
 366         [VARIABLE_LC_TIME]           = "LC_TIME",
 367         [VARIABLE_LC_COLLATE]        = "LC_COLLATE",
 368         [VARIABLE_LC_MONETARY]       = "LC_MONETARY",
 369         [VARIABLE_LC_MESSAGES]       = "LC_MESSAGES",
 370         [VARIABLE_LC_PAPER]          = "LC_PAPER",
 371         [VARIABLE_LC_NAME]           = "LC_NAME",
 372         [VARIABLE_LC_ADDRESS]        = "LC_ADDRESS",
 373         [VARIABLE_LC_TELEPHONE]      = "LC_TELEPHONE",
 374         [VARIABLE_LC_MEASUREMENT]    = "LC_MEASUREMENT",
 375         [VARIABLE_LC_IDENTIFICATION] = "LC_IDENTIFICATION"
 376 };
 377
 378 DEFINE_STRING_TABLE_LOOKUP(locale_variable, LocaleVariable);