src/basic/locale-util.c

   1 /* SPDX-License-Identifier: LGPL-2.1+ */
   2 /***
   3   This file is part of systemd.
   4
   5   Copyright 2014 Lennart Poettering
   6 ***/
   7
   8 #include <dirent.h>
   9 #include <errno.h>
  10 #include <fcntl.h>
  11 #include <ftw.h>
  12 #include <langinfo.h>
  13 #include <libintl.h>
  14 #include <locale.h>
  15 #include <stddef.h>
  16 #include <stdint.h>
  17 #include <stdlib.h>
  18 #include <string.h>
  19 #include <sys/mman.h>
  20 #include <sys/stat.h>
  21
  22 #include "def.h"
  23 #include "dirent-util.h"
  24 #include "fd-util.h"
  25 #include "hashmap.h"
  26 #include "locale-util.h"
  27 #include "path-util.h"
  28 #include "set.h"
  29 #include "string-table.h"
  30 #include "string-util.h"
  31 #include "strv.h"
  32 #include "utf8.h"
  33
  34 static int add_locales_from_archive(Set *locales) {
  35         /* Stolen from glibc... */
  36
  37         struct locarhead {
  38                 uint32_t magic;
  39                 /* Serial number.  */
  40                 uint32_t serial;
  41                 /* Name hash table.  */
  42                 uint32_t namehash_offset;
  43                 uint32_t namehash_used;
  44                 uint32_t namehash_size;
  45                 /* String table.  */
  46                 uint32_t string_offset;
  47                 uint32_t string_used;
  48                 uint32_t string_size;
  49                 /* Table with locale records.  */
  50                 uint32_t locrectab_offset;
  51                 uint32_t locrectab_used;
  52                 uint32_t locrectab_size;
  53                 /* MD5 sum hash table.  */
  54                 uint32_t sumhash_offset;
  55                 uint32_t sumhash_used;
  56                 uint32_t sumhash_size;
  57         };
  58
  59         struct namehashent {
  60                 /* Hash value of the name.  */
  61                 uint32_t hashval;
  62                 /* Offset of the name in the string table.  */
  63                 uint32_t name_offset;
  64                 /* Offset of the locale record.  */
  65                 uint32_t locrec_offset;
  66         };
  67
  68         const struct locarhead *h;
  69         const struct namehashent *e;
  70         const void *p = MAP_FAILED;
  71         _cleanup_close_ int fd = -1;
  72         size_t sz = 0;
  73         struct stat st;
  74         size_t i;
  75         int r;
  76
  77         fd = open("/usr/lib/locale/locale-archive", O_RDONLY|O_NOCTTY|O_CLOEXEC);
  78         if (fd < 0)
  79                 return errno == ENOENT ? 0 : -errno;
  80
  81         if (fstat(fd, &st) < 0)
  82                 return -errno;
  83
  84         if (!S_ISREG(st.st_mode))
  85                 return -EBADMSG;
  86
  87         if (st.st_size < (off_t) sizeof(struct locarhead))
  88                 return -EBADMSG;
  89
  90         p = mmap(NULL, st.st_size, PROT_READ, MAP_SHARED, fd, 0);
  91         if (p == MAP_FAILED)
  92                 return -errno;
  93
  94         h = (const struct locarhead *) p;
  95         if (h->magic != 0xde020109 ||
  96             h->namehash_offset + h->namehash_size > st.st_size ||
  97             h->string_offset + h->string_size > st.st_size ||
  98             h->locrectab_offset + h->locrectab_size > st.st_size ||
  99             h->sumhash_offset + h->sumhash_size > st.st_size) {
 100                 r = -EBADMSG;
 101                 goto finish;
 102         }
 103
 104         e = (const struct namehashent*) ((const uint8_t*) p + h->namehash_offset);
 105         for (i = 0; i < h->namehash_size; i++) {
 106                 char *z;
 107
 108                 if (e[i].locrec_offset == 0)
 109                         continue;
 110
 111                 if (!utf8_is_valid((char*) p + e[i].name_offset))
 112                         continue;
 113
 114                 z = strdup((char*) p + e[i].name_offset);
 115                 if (!z) {
 116                         r = -ENOMEM;
 117                         goto finish;
 118                 }
 119
 120                 r = set_consume(locales, z);
 121                 if (r < 0)
 122                         goto finish;
 123         }
 124
 125         r = 0;
 126
 127  finish:
 128         if (p != MAP_FAILED)
 129                 munmap((void*) p, sz);
 130
 131         return r;
 132 }
 133
 134 static int add_locales_from_libdir (Set *locales) {
 135         _cleanup_closedir_ DIR *dir = NULL;
 136         struct dirent *entry;
 137         int r;
 138
 139         dir = opendir("/usr/lib/locale");
 140         if (!dir)
 141                 return errno == ENOENT ? 0 : -errno;
 142
 143         FOREACH_DIRENT(entry, dir, return -errno) {
 144                 char *z;
 145
 146                 dirent_ensure_type(dir, entry);
 147
 148                 if (entry->d_type != DT_DIR)
 149                         continue;
 150
 151                 z = strdup(entry->d_name);
 152                 if (!z)
 153                         return -ENOMEM;
 154
 155                 r = set_consume(locales, z);
 156                 if (r < 0 && r != -EEXIST)
 157                         return r;
 158         }
 159
 160         return 0;
 161 }
 162
 163 int get_locales(char ***ret) {
 164         _cleanup_set_free_ Set *locales = NULL;
 165         _cleanup_strv_free_ char **l = NULL;
 166         int r;
 167
 168         locales = set_new(&string_hash_ops);
 169         if (!locales)
 170                 return -ENOMEM;
 171
 172         r = add_locales_from_archive(locales);
 173         if (r < 0 && r != -ENOENT)
 174                 return r;
 175
 176         r = add_locales_from_libdir(locales);
 177         if (r < 0)
 178                 return r;
 179
 180         l = set_get_strv(locales);
 181         if (!l)
 182                 return -ENOMEM;
 183
 184         strv_sort(l);
 185
 186         *ret = TAKE_PTR(l);
 187
 188         return 0;
 189 }
 190
 191 bool locale_is_valid(const char *name) {
 192
 193         if (isempty(name))
 194                 return false;
 195
 196         if (strlen(name) >= 128)
 197                 return false;
 198
 199         if (!utf8_is_valid(name))
 200                 return false;
 201
 202         if (!filename_is_valid(name))
 203                 return false;
 204
 205         if (!string_is_safe(name))
 206                 return false;
 207
 208         return true;
 209 }
 210
 211 void init_gettext(void) {
 212         setlocale(LC_ALL, "");
 213         textdomain(GETTEXT_PACKAGE);
 214 }
 215
 216 bool is_locale_utf8(void) {
 217         const char *set;
 218         static int cached_answer = -1;
 219
 220         /* Note that we default to 'true' here, since today UTF8 is
 221          * pretty much supported everywhere. */
 222
 223         if (cached_answer >= 0)
 224                 goto out;
 225
 226         if (!setlocale(LC_ALL, "")) {
 227                 cached_answer = true;
 228                 goto out;
 229         }
 230
 231         set = nl_langinfo(CODESET);
 232         if (!set) {
 233                 cached_answer = true;
 234                 goto out;
 235         }
 236
 237         if (streq(set, "UTF-8")) {
 238                 cached_answer = true;
 239                 goto out;
 240         }
 241
 242         /* For LC_CTYPE=="C" return true, because CTYPE is effectly
 243          * unset and everything can do to UTF-8 nowadays. */
 244         set = setlocale(LC_CTYPE, NULL);
 245         if (!set) {
 246                 cached_answer = true;
 247                 goto out;
 248         }
 249
 250         /* Check result, but ignore the result if C was set
 251          * explicitly. */
 252         cached_answer =
 253                 STR_IN_SET(set, "C", "POSIX") &&
 254                 !getenv("LC_ALL") &&
 255                 !getenv("LC_CTYPE") &&
 256                 !getenv("LANG");
 257
 258 out:
 259         return (bool) cached_answer;
 260 }
 261
 262 static thread_local Set *keymaps = NULL;
 263
 264 static int nftw_cb(
 265                 const char *fpath,
 266                 const struct stat *sb,
 267                 int tflag,
 268                 struct FTW *ftwbuf) {
 269
 270         char *p, *e;
 271         int r;
 272
 273         if (tflag != FTW_F)
 274                 return 0;
 275
 276         if (!endswith(fpath, ".map") &&
 277             !endswith(fpath, ".map.gz"))
 278                 return 0;
 279
 280         p = strdup(basename(fpath));
 281         if (!p)
 282                 return FTW_STOP;
 283
 284         e = endswith(p, ".map");
 285         if (e)
 286                 *e = 0;
 287
 288         e = endswith(p, ".map.gz");
 289         if (e)
 290                 *e = 0;
 291
 292         r = set_consume(keymaps, p);
 293         if (r < 0 && r != -EEXIST)
 294                 return r;
 295
 296         return 0;
 297 }
 298
 299 int get_keymaps(char ***ret) {
 300         _cleanup_strv_free_ char **l = NULL;
 301         const char *dir;
 302         int r;
 303
 304         keymaps = set_new(&string_hash_ops);
 305         if (!keymaps)
 306                 return -ENOMEM;
 307
 308         NULSTR_FOREACH(dir, KBD_KEYMAP_DIRS) {
 309                 r = nftw(dir, nftw_cb, 20, FTW_PHYS|FTW_ACTIONRETVAL);
 310
 311                 if (r == FTW_STOP)
 312                         log_debug("Directory not found %s", dir);
 313                 else if (r < 0)
 314                         log_debug_errno(r, "Can't add keymap: %m");
 315         }
 316
 317         l = set_get_strv(keymaps);
 318         if (!l) {
 319                 set_free_free(keymaps);
 320                 return -ENOMEM;
 321         }
 322
 323         set_free(keymaps);
 324
 325         if (strv_isempty(l))
 326                 return -ENOENT;
 327
 328         strv_sort(l);
 329
 330         *ret = TAKE_PTR(l);
 331
 332         return 0;
 333 }
 334
 335 bool keymap_is_valid(const char *name) {
 336
 337         if (isempty(name))
 338                 return false;
 339
 340         if (strlen(name) >= 128)
 341                 return false;
 342
 343         if (!utf8_is_valid(name))
 344                 return false;
 345
 346         if (!filename_is_valid(name))
 347                 return false;
 348
 349         if (!string_is_safe(name))
 350                 return false;
 351
 352         return true;
 353 }
 354
 355 const char *special_glyph(SpecialGlyph code) {
 356
 357         /* A list of a number of interesting unicode glyphs we can use to decorate our output. It's probably wise to be
 358          * conservative here, and primarily stick to the glyphs defined in the eurlatgr font, so that display still
 359          * works reasonably well on the Linux console. For details see:
 360          *
 361          * http://git.altlinux.org/people/legion/packages/kbd.git?p=kbd.git;a=blob;f=data/consolefonts/README.eurlatgr
 362          */
 363
 364         static const char* const draw_table[2][_SPECIAL_GLYPH_MAX] = {
 365                 /* ASCII fallback */
 366                 [false] = {
 367                         [TREE_VERTICAL]      = "| ",
 368                         [TREE_BRANCH]        = "|-",
 369                         [TREE_RIGHT]         = "`-",
 370                         [TREE_SPACE]         = "  ",
 371                         [TRIANGULAR_BULLET]  = ">",
 372                         [BLACK_CIRCLE]       = "*",
 373                         [ARROW]              = "->",
 374                         [MDASH]              = "-",
 375                         [ELLIPSIS]           = "..."
 376                 },
 377
 378                 /* UTF-8 */
 379                 [true] = {
 380                         [TREE_VERTICAL]      = "\342\224\202 ",            /* │  */
 381                         [TREE_BRANCH]        = "\342\224\234\342\224\200", /* ├─ */
 382                         [TREE_RIGHT]         = "\342\224\224\342\224\200", /* └─ */
 383                         [TREE_SPACE]         = "  ",                       /*    */
 384                         [TRIANGULAR_BULLET]  = "\342\200\243",             /* ‣ */
 385                         [BLACK_CIRCLE]       = "\342\227\217",             /* ● */
 386                         [ARROW]              = "\342\206\222",             /* → */
 387                         [MDASH]              = "\342\200\223",             /* – */
 388                         [ELLIPSIS]           = "\342\200\246",             /* … */
 389                 },
 390         };
 391
 392         return draw_table[is_locale_utf8()][code];
 393 }
 394
 395 static const char * const locale_variable_table[_VARIABLE_LC_MAX] = {
 396         [VARIABLE_LANG] = "LANG",
 397         [VARIABLE_LANGUAGE] = "LANGUAGE",
 398         [VARIABLE_LC_CTYPE] = "LC_CTYPE",
 399         [VARIABLE_LC_NUMERIC] = "LC_NUMERIC",
 400         [VARIABLE_LC_TIME] = "LC_TIME",
 401         [VARIABLE_LC_COLLATE] = "LC_COLLATE",
 402         [VARIABLE_LC_MONETARY] = "LC_MONETARY",
 403         [VARIABLE_LC_MESSAGES] = "LC_MESSAGES",
 404         [VARIABLE_LC_PAPER] = "LC_PAPER",
 405         [VARIABLE_LC_NAME] = "LC_NAME",
 406         [VARIABLE_LC_ADDRESS] = "LC_ADDRESS",
 407         [VARIABLE_LC_TELEPHONE] = "LC_TELEPHONE",
 408         [VARIABLE_LC_MEASUREMENT] = "LC_MEASUREMENT",
 409         [VARIABLE_LC_IDENTIFICATION] = "LC_IDENTIFICATION"
 410 };
 411
 412 DEFINE_STRING_TABLE_LOOKUP(locale_variable, LocaleVariable);