lh | 9ed821d | 2023-04-07 01:36:19 -0700 | [diff] [blame] | 1 | /* Declarations for internal libc locale interfaces |
| 2 | Copyright (C) 1995-2015 Free Software Foundation, Inc. |
| 3 | This file is part of the GNU C Library. |
| 4 | |
| 5 | The GNU C Library is free software; you can redistribute it and/or |
| 6 | modify it under the terms of the GNU Lesser General Public |
| 7 | License as published by the Free Software Foundation; either |
| 8 | version 2.1 of the License, or (at your option) any later version. |
| 9 | |
| 10 | The GNU C Library is distributed in the hope that it will be useful, |
| 11 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 12 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 13 | Lesser General Public License for more details. |
| 14 | |
| 15 | You should have received a copy of the GNU Lesser General Public |
| 16 | License along with the GNU C Library; if not, see |
| 17 | <http://www.gnu.org/licenses/>. */ |
| 18 | |
| 19 | #ifndef _LOCALEINFO_H |
| 20 | #define _LOCALEINFO_H 1 |
| 21 | |
| 22 | #include <stddef.h> |
| 23 | #include <langinfo.h> |
| 24 | #include <limits.h> |
| 25 | #include <locale.h> |
| 26 | #include <time.h> |
| 27 | #include <stdint.h> |
| 28 | #include <sys/types.h> |
| 29 | |
| 30 | #include <intl/loadinfo.h> /* For loaded_l10nfile definition. */ |
| 31 | |
| 32 | /* Magic number at the beginning of a locale data file for CATEGORY. */ |
| 33 | #define LIMAGIC(category) \ |
| 34 | (category == LC_COLLATE \ |
| 35 | ? ((unsigned int) (0x20051014 ^ (category))) \ |
| 36 | : category == LC_CTYPE \ |
| 37 | ? ((unsigned int) (0x20090720 ^ (category))) \ |
| 38 | : ((unsigned int) (0x20031115 ^ (category)))) |
| 39 | |
| 40 | /* Two special weight constants for the collation data. */ |
| 41 | #define IGNORE_CHAR 2 |
| 42 | |
| 43 | /* We use a special value for the usage counter in `__locale_data' to |
| 44 | signal that this data must never be removed anymore. */ |
| 45 | #define MAX_USAGE_COUNT (UINT_MAX - 1) |
| 46 | #define UNDELETABLE UINT_MAX |
| 47 | |
| 48 | /* Structure describing locale data in core for a category. */ |
| 49 | struct __locale_data |
| 50 | { |
| 51 | const char *name; |
| 52 | const char *filedata; /* Region mapping the file data. */ |
| 53 | off_t filesize; /* Size of the file (and the region). */ |
| 54 | enum /* Flavor of storage used for those. */ |
| 55 | { |
| 56 | ld_malloced, /* Both are malloc'd. */ |
| 57 | ld_mapped, /* name is malloc'd, filedata mmap'd */ |
| 58 | ld_archive /* Both point into mmap'd archive regions. */ |
| 59 | } alloc; |
| 60 | |
| 61 | /* This provides a slot for category-specific code to cache data computed |
| 62 | about this locale. That code can set a cleanup function to deallocate |
| 63 | the data. */ |
| 64 | struct |
| 65 | { |
| 66 | void (*cleanup) (struct __locale_data *) internal_function; |
| 67 | union |
| 68 | { |
| 69 | void *data; |
| 70 | struct lc_time_data *time; |
| 71 | const struct gconv_fcts *ctype; |
| 72 | }; |
| 73 | } private; |
| 74 | |
| 75 | unsigned int usage_count; /* Counter for users. */ |
| 76 | |
| 77 | int use_translit; /* Nonzero if the mb*towv*() and wc*tomb() |
| 78 | functions should use transliteration. */ |
| 79 | |
| 80 | unsigned int nstrings; /* Number of strings below. */ |
| 81 | union locale_data_value |
| 82 | { |
| 83 | const uint32_t *wstr; |
| 84 | const char *string; |
| 85 | unsigned int word; /* Note endian issues vs 64-bit pointers. */ |
| 86 | } |
| 87 | values __flexarr; /* Items, usually pointers into `filedata'. */ |
| 88 | }; |
| 89 | |
| 90 | /* This alignment is used for 32-bit integers in locale files, both |
| 91 | those that are explicitly int32_t or uint32_t and those that are |
| 92 | wchar_t, regardless of the (possibly smaller) alignment required |
| 93 | for such integers on a particular host. */ |
| 94 | #define LOCFILE_ALIGN sizeof (int32_t) |
| 95 | #define LOCFILE_ALIGN_MASK (LOCFILE_ALIGN - 1) |
| 96 | #define LOCFILE_ALIGN_UP(x) (((x) + LOCFILE_ALIGN - 1) \ |
| 97 | & ~LOCFILE_ALIGN_MASK) |
| 98 | #define LOCFILE_ALIGNED_P(x) (((x) & LOCFILE_ALIGN_MASK) == 0) |
| 99 | |
| 100 | /* We know three kinds of collation sorting rules. */ |
| 101 | enum coll_sort_rule |
| 102 | { |
| 103 | illegal_0__, |
| 104 | sort_forward, |
| 105 | sort_backward, |
| 106 | illegal_3__, |
| 107 | sort_position, |
| 108 | sort_forward_position, |
| 109 | sort_backward_position, |
| 110 | sort_mask |
| 111 | }; |
| 112 | |
| 113 | /* Collation encoding type. */ |
| 114 | enum collation_encoding_type |
| 115 | { |
| 116 | __cet_other, |
| 117 | __cet_8bit, |
| 118 | __cet_utf8 |
| 119 | }; |
| 120 | |
| 121 | /* We can map the types of the entries into a few categories. */ |
| 122 | enum value_type |
| 123 | { |
| 124 | none, |
| 125 | string, |
| 126 | stringarray, |
| 127 | byte, |
| 128 | bytearray, |
| 129 | word, |
| 130 | stringlist, |
| 131 | wordarray, |
| 132 | wstring, |
| 133 | wstringarray, |
| 134 | wstringlist |
| 135 | }; |
| 136 | |
| 137 | |
| 138 | /* Definitions for `era' information from LC_TIME. */ |
| 139 | #define ERA_NAME_FORMAT_MEMBERS 4 |
| 140 | #define ERA_M_NAME 0 |
| 141 | #define ERA_M_FORMAT 1 |
| 142 | #define ERA_W_NAME 2 |
| 143 | #define ERA_W_FORMAT 3 |
| 144 | |
| 145 | |
| 146 | /* Structure to access `era' information from LC_TIME. */ |
| 147 | struct era_entry |
| 148 | { |
| 149 | uint32_t direction; /* Contains '+' or '-'. */ |
| 150 | int32_t offset; |
| 151 | int32_t start_date[3]; |
| 152 | int32_t stop_date[3]; |
| 153 | const char *era_name; |
| 154 | const char *era_format; |
| 155 | const wchar_t *era_wname; |
| 156 | const wchar_t *era_wformat; |
| 157 | int absolute_direction; |
| 158 | /* absolute direction: |
| 159 | +1 indicates that year number is higher in the future. (like A.D.) |
| 160 | -1 indicates that year number is higher in the past. (like B.C.) */ |
| 161 | }; |
| 162 | |
| 163 | /* Structure caching computed data about information from LC_TIME. |
| 164 | The `private.time' member of `struct __locale_data' points to this. */ |
| 165 | struct lc_time_data |
| 166 | { |
| 167 | struct era_entry *eras; |
| 168 | size_t num_eras; |
| 169 | int era_initialized; |
| 170 | |
| 171 | const char **alt_digits; |
| 172 | const wchar_t **walt_digits; |
| 173 | int alt_digits_initialized; |
| 174 | int walt_digits_initialized; |
| 175 | }; |
| 176 | |
| 177 | |
| 178 | /* LC_CTYPE specific: |
| 179 | Hardwired indices for standard wide character translation mappings. */ |
| 180 | enum |
| 181 | { |
| 182 | __TOW_toupper = 0, |
| 183 | __TOW_tolower = 1 |
| 184 | }; |
| 185 | |
| 186 | |
| 187 | /* LC_CTYPE specific: |
| 188 | Access a wide character class with a single character index. |
| 189 | _ISCTYPE (c, desc) = iswctype (btowc (c), desc). |
| 190 | c must be an `unsigned char'. desc must be a nonzero wctype_t. */ |
| 191 | #define _ISCTYPE(c, desc) \ |
| 192 | (((((const uint32_t *) (desc)) - 8)[(c) >> 5] >> ((c) & 0x1f)) & 1) |
| 193 | |
| 194 | /* Category name handling variables. */ |
| 195 | #define CATNAMEMF(line) CATNAMEMF1 (line) |
| 196 | #define CATNAMEMF1(line) str##line |
| 197 | extern const union catnamestr_t |
| 198 | { |
| 199 | struct |
| 200 | { |
| 201 | #define DEFINE_CATEGORY(category, category_name, items, a) \ |
| 202 | char CATNAMEMF (__LINE__)[sizeof (category_name)]; |
| 203 | #include "categories.def" |
| 204 | #undef DEFINE_CATEGORY |
| 205 | }; |
| 206 | char str[0]; |
| 207 | } _nl_category_names attribute_hidden; |
| 208 | extern const uint8_t _nl_category_name_idxs[__LC_LAST] attribute_hidden; |
| 209 | extern const uint8_t _nl_category_name_sizes[__LC_LAST] attribute_hidden; |
| 210 | |
| 211 | /* Name of the standard locales. */ |
| 212 | extern const char _nl_C_name[] attribute_hidden; |
| 213 | extern const char _nl_POSIX_name[] attribute_hidden; |
| 214 | |
| 215 | /* The standard codeset. */ |
| 216 | extern const char _nl_C_codeset[] attribute_hidden; |
| 217 | |
| 218 | /* This is the internal locale_t object that holds the global locale |
| 219 | controlled by calls to setlocale. A thread's TSD locale pointer |
| 220 | points to this when `uselocale (LC_GLOBAL_LOCALE)' is in effect. */ |
| 221 | extern struct __locale_struct _nl_global_locale attribute_hidden; |
| 222 | |
| 223 | /* This fetches the thread-local locale_t pointer, either one set with |
| 224 | uselocale or &_nl_global_locale. */ |
| 225 | #define _NL_CURRENT_LOCALE (__libc_tsd_get (__locale_t, LOCALE)) |
| 226 | #include <bits/libc-tsd.h> |
| 227 | __libc_tsd_define (extern, __locale_t, LOCALE) |
| 228 | |
| 229 | |
| 230 | /* For static linking it is desireable to avoid always linking in the code |
| 231 | and data for every category when we can tell at link time that they are |
| 232 | unused. We can manage this playing some tricks with weak references. |
| 233 | But with thread-local locale settings, it becomes quite ungainly unless |
| 234 | we can use __thread variables. So only in that case do we attempt this. */ |
| 235 | #ifndef SHARED |
| 236 | # include <tls.h> |
| 237 | # define NL_CURRENT_INDIRECT 1 |
| 238 | #endif |
| 239 | |
| 240 | #ifdef NL_CURRENT_INDIRECT |
| 241 | |
| 242 | /* For each category declare the thread-local variable for the current |
| 243 | locale data. This has an extra indirection so it points at the |
| 244 | __locales[CATEGORY] element in either _nl_global_locale or the current |
| 245 | locale object set by uselocale, which points at the actual data. The |
| 246 | reason for having these variables is so that references to particular |
| 247 | categories will link in the lc-CATEGORY.c module to define this symbol, |
| 248 | and we arrange that linking that module is what brings in all the code |
| 249 | associated with this category. */ |
| 250 | #define DEFINE_CATEGORY(category, category_name, items, a) \ |
| 251 | extern __thread struct __locale_data *const *_nl_current_##category \ |
| 252 | attribute_hidden attribute_tls_model_ie; |
| 253 | #include "categories.def" |
| 254 | #undef DEFINE_CATEGORY |
| 255 | |
| 256 | /* Return a pointer to the current `struct __locale_data' for CATEGORY. */ |
| 257 | #define _NL_CURRENT_DATA(category) (*_nl_current_##category) |
| 258 | |
| 259 | /* Extract the current CATEGORY locale's string for ITEM. */ |
| 260 | #define _NL_CURRENT(category, item) \ |
| 261 | ((*_nl_current_##category)->values[_NL_ITEM_INDEX (item)].string) |
| 262 | |
| 263 | /* Extract the current CATEGORY locale's string for ITEM. */ |
| 264 | #define _NL_CURRENT_WSTR(category, item) \ |
| 265 | ((wchar_t *) (*_nl_current_##category)->values[_NL_ITEM_INDEX (item)].wstr) |
| 266 | |
| 267 | /* Extract the current CATEGORY locale's word for ITEM. */ |
| 268 | #define _NL_CURRENT_WORD(category, item) \ |
| 269 | ((uint32_t) (*_nl_current_##category)->values[_NL_ITEM_INDEX (item)].word) |
| 270 | |
| 271 | /* This is used in lc-CATEGORY.c to define _nl_current_CATEGORY. */ |
| 272 | #define _NL_CURRENT_DEFINE(category) \ |
| 273 | __thread struct __locale_data *const *_nl_current_##category \ |
| 274 | attribute_hidden = &_nl_global_locale.__locales[category]; \ |
| 275 | asm (".globl " __SYMBOL_PREFIX "_nl_current_" #category "_used\n" \ |
| 276 | _NL_CURRENT_DEFINE_ABS (_nl_current_##category##_used, 1)); |
| 277 | #ifdef HAVE_ASM_SET_DIRECTIVE |
| 278 | # define _NL_CURRENT_DEFINE_ABS(sym, val) ".set " #sym ", " #val |
| 279 | #else |
| 280 | # define _NL_CURRENT_DEFINE_ABS(sym, val) #sym " = " #val |
| 281 | #endif |
| 282 | |
| 283 | #else |
| 284 | |
| 285 | /* All categories are always loaded in the shared library, so there is no |
| 286 | point in having lots of separate symbols for linking. */ |
| 287 | |
| 288 | /* Return a pointer to the current `struct __locale_data' for CATEGORY. */ |
| 289 | # define _NL_CURRENT_DATA(category) \ |
| 290 | (_NL_CURRENT_LOCALE->__locales[category]) |
| 291 | |
| 292 | /* Extract the current CATEGORY locale's string for ITEM. */ |
| 293 | # define _NL_CURRENT(category, item) \ |
| 294 | (_NL_CURRENT_DATA (category)->values[_NL_ITEM_INDEX (item)].string) |
| 295 | |
| 296 | /* Extract the current CATEGORY locale's string for ITEM. */ |
| 297 | # define _NL_CURRENT_WSTR(category, item) \ |
| 298 | ((wchar_t *) _NL_CURRENT_DATA (category)->values[_NL_ITEM_INDEX (item)].wstr) |
| 299 | |
| 300 | /* Extract the current CATEGORY locale's word for ITEM. */ |
| 301 | # define _NL_CURRENT_WORD(category, item) \ |
| 302 | ((uint32_t) _NL_CURRENT_DATA (category)->values[_NL_ITEM_INDEX (item)].word) |
| 303 | |
| 304 | /* This is used in lc-CATEGORY.c to define _nl_current_CATEGORY. */ |
| 305 | # define _NL_CURRENT_DEFINE(category) \ |
| 306 | /* No per-category variable here. */ |
| 307 | |
| 308 | #endif |
| 309 | |
| 310 | |
| 311 | /* Default search path if no LOCPATH environment variable. */ |
| 312 | extern const char _nl_default_locale_path[] attribute_hidden; |
| 313 | |
| 314 | /* Load the locale data for CATEGORY from the file specified by *NAME. |
| 315 | If *NAME is "", use environment variables as specified by POSIX, and |
| 316 | fill in *NAME with the actual name used. If LOCALE_PATH is not null, |
| 317 | those directories are searched for the locale files. If it's null, |
| 318 | the locale archive is checked first and then _nl_default_locale_path |
| 319 | is searched for locale files. */ |
| 320 | extern struct __locale_data *_nl_find_locale (const char *locale_path, |
| 321 | size_t locale_path_len, |
| 322 | int category, const char **name) |
| 323 | internal_function attribute_hidden; |
| 324 | |
| 325 | /* Try to load the file described by FILE. */ |
| 326 | extern void _nl_load_locale (struct loaded_l10nfile *file, int category) |
| 327 | internal_function attribute_hidden; |
| 328 | |
| 329 | /* Free all resource. */ |
| 330 | extern void _nl_unload_locale (struct __locale_data *locale) |
| 331 | internal_function attribute_hidden; |
| 332 | |
| 333 | /* Free the locale and give back all memory if the usage count is one. */ |
| 334 | extern void _nl_remove_locale (int locale, struct __locale_data *data) |
| 335 | internal_function attribute_hidden; |
| 336 | |
| 337 | /* Find the locale *NAMEP in the locale archive, and return the |
| 338 | internalized data structure for its CATEGORY data. If this locale has |
| 339 | already been loaded from the archive, just returns the existing data |
| 340 | structure. If successful, sets *NAMEP to point directly into the mapped |
| 341 | archive string table; that way, the next call can short-circuit strcmp. */ |
| 342 | extern struct __locale_data *_nl_load_locale_from_archive (int category, |
| 343 | const char **namep) |
| 344 | internal_function attribute_hidden; |
| 345 | |
| 346 | /* Subroutine of setlocale's __libc_subfreeres hook. */ |
| 347 | extern void _nl_archive_subfreeres (void) attribute_hidden; |
| 348 | |
| 349 | /* Subroutine of gconv-db's __libc_subfreeres hook. */ |
| 350 | extern void _nl_locale_subfreeres (void) attribute_hidden; |
| 351 | |
| 352 | /* Validate the contents of a locale file and set up the in-core |
| 353 | data structure to point into the data. This leaves the `alloc' |
| 354 | and `name' fields uninitialized, for the caller to fill in. |
| 355 | If any bogons are detected in the data, this will refuse to |
| 356 | intern it, and return a null pointer instead. */ |
| 357 | extern struct __locale_data *_nl_intern_locale_data (int category, |
| 358 | const void *data, |
| 359 | size_t datasize) |
| 360 | internal_function attribute_hidden; |
| 361 | |
| 362 | |
| 363 | /* Return `era' entry which corresponds to TP. Used in strftime. */ |
| 364 | extern struct era_entry *_nl_get_era_entry (const struct tm *tp, |
| 365 | struct __locale_data *lc_time) |
| 366 | internal_function attribute_hidden; |
| 367 | |
| 368 | /* Return `era' cnt'th entry . Used in strptime. */ |
| 369 | extern struct era_entry *_nl_select_era_entry (int cnt, |
| 370 | struct __locale_data *lc_time) |
| 371 | internal_function attribute_hidden; |
| 372 | |
| 373 | /* Return `alt_digit' which corresponds to NUMBER. Used in strftime. */ |
| 374 | extern const char *_nl_get_alt_digit (unsigned int number, |
| 375 | struct __locale_data *lc_time) |
| 376 | internal_function attribute_hidden; |
| 377 | |
| 378 | /* Similar, but now for wide characters. */ |
| 379 | extern const wchar_t *_nl_get_walt_digit (unsigned int number, |
| 380 | struct __locale_data *lc_time) |
| 381 | internal_function attribute_hidden; |
| 382 | |
| 383 | /* Parse string as alternative digit and return numeric value. */ |
| 384 | extern int _nl_parse_alt_digit (const char **strp, |
| 385 | struct __locale_data *lc_time) |
| 386 | internal_function attribute_hidden; |
| 387 | |
| 388 | /* Postload processing. */ |
| 389 | extern void _nl_postload_ctype (void); |
| 390 | |
| 391 | /* Functions used for the `private.cleanup' hook. */ |
| 392 | extern void _nl_cleanup_time (struct __locale_data *) |
| 393 | internal_function attribute_hidden; |
| 394 | |
| 395 | |
| 396 | #endif /* localeinfo.h */ |