1 |
1325 |
phoenix |
/* Copyright (C) 2002 Manuel Novoa III
|
2 |
|
|
*
|
3 |
|
|
* This library is free software; you can redistribute it and/or
|
4 |
|
|
* modify it under the terms of the GNU Library General Public
|
5 |
|
|
* License as published by the Free Software Foundation; either
|
6 |
|
|
* version 2 of the License, or (at your option) any later version.
|
7 |
|
|
*
|
8 |
|
|
* This library is distributed in the hope that it will be useful,
|
9 |
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
10 |
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
11 |
|
|
* Library General Public License for more details.
|
12 |
|
|
*
|
13 |
|
|
* You should have received a copy of the GNU Library General Public
|
14 |
|
|
* License along with this library; if not, write to the Free
|
15 |
|
|
* Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
16 |
|
|
*/
|
17 |
|
|
|
18 |
|
|
/* Nov. 1, 2002
|
19 |
|
|
* Reworked setlocale() return values and locale arg processing to
|
20 |
|
|
* be more like glibc. Applications expecting to be able to
|
21 |
|
|
* query locale settings should now work... at the cost of almost
|
22 |
|
|
* doubling the size of the setlocale object code.
|
23 |
|
|
* Fixed a bug in the internal fixed-size-string locale specifier code.
|
24 |
|
|
*
|
25 |
|
|
* Dec 20, 2002
|
26 |
|
|
* Added in collation support and updated stub nl_langinfo.
|
27 |
|
|
*
|
28 |
|
|
* Aug 1, 2003
|
29 |
|
|
* Added glibc-like extended locale stuff (newlocale, duplocale, etc).
|
30 |
|
|
*
|
31 |
|
|
* Aug 18, 2003
|
32 |
|
|
* Bug in duplocale... collation data wasn't copied.
|
33 |
|
|
* Bug in newlocale... translate 1<<LC_ALL to LC_ALL_MASK.
|
34 |
|
|
* Bug in _wchar_utf8sntowcs... fix cut-n-paste error.
|
35 |
|
|
*
|
36 |
|
|
* Aug 31, 2003
|
37 |
|
|
* Hack around bg_BG bug; grouping specified but no thousands separator.
|
38 |
|
|
* Also, disable the locale link_warnings for now, as they generate a
|
39 |
|
|
* lot of noise when using libstd++.
|
40 |
|
|
*/
|
41 |
|
|
|
42 |
|
|
|
43 |
|
|
/* TODO:
|
44 |
|
|
* Implement the shared mmap code so non-mmu platforms can use this.
|
45 |
|
|
* Add some basic collate functionality similar to what the previous
|
46 |
|
|
* locale support had (8-bit codesets only).
|
47 |
|
|
*/
|
48 |
|
|
|
49 |
|
|
#define _GNU_SOURCE
|
50 |
|
|
|
51 |
|
|
#define __CTYPE_HAS_8_BIT_LOCALES 1
|
52 |
|
|
|
53 |
|
|
|
54 |
|
|
#include <string.h>
|
55 |
|
|
#include <stdlib.h>
|
56 |
|
|
#include <stddef.h>
|
57 |
|
|
#include <limits.h>
|
58 |
|
|
#include <stdint.h>
|
59 |
|
|
#include <assert.h>
|
60 |
|
|
#include <errno.h>
|
61 |
|
|
#include <ctype.h>
|
62 |
|
|
#include <stdio.h>
|
63 |
|
|
|
64 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
65 |
|
|
#ifdef L_setlocale
|
66 |
|
|
#warning TODO: Make the link_warning()s a config option?
|
67 |
|
|
#endif
|
68 |
|
|
#endif
|
69 |
|
|
#undef link_warning
|
70 |
|
|
#define link_warning(A,B)
|
71 |
|
|
|
72 |
|
|
#undef __LOCALE_C_ONLY
|
73 |
|
|
#ifndef __UCLIBC_HAS_LOCALE__
|
74 |
|
|
#define __LOCALE_C_ONLY
|
75 |
|
|
#endif /* __UCLIBC_HAS_LOCALE__ */
|
76 |
|
|
|
77 |
|
|
|
78 |
|
|
#ifdef __LOCALE_C_ONLY
|
79 |
|
|
|
80 |
|
|
#include <locale.h>
|
81 |
|
|
|
82 |
|
|
#else /* __LOCALE_C_ONLY */
|
83 |
|
|
|
84 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
85 |
|
|
#ifdef L_setlocale
|
86 |
|
|
#warning TODO: Fix the __CTYPE_HAS_8_BIT_LOCALES define at the top of the file.
|
87 |
|
|
#warning TODO: Fix __WCHAR_ENABLED.
|
88 |
|
|
#endif
|
89 |
|
|
#endif
|
90 |
|
|
|
91 |
|
|
/* Need to include this before locale.h and xlocale.h! */
|
92 |
|
|
#include <bits/uClibc_locale.h>
|
93 |
|
|
|
94 |
|
|
#undef CODESET_LIST
|
95 |
|
|
#define CODESET_LIST (__locale_mmap->codeset_list)
|
96 |
|
|
|
97 |
|
|
#ifdef __UCLIBC_HAS_XLOCALE__
|
98 |
|
|
#include <xlocale.h>
|
99 |
|
|
#include <locale.h>
|
100 |
|
|
#else /* __UCLIBC_HAS_XLOCALE__ */
|
101 |
|
|
/* We need this internally... */
|
102 |
|
|
#define __UCLIBC_HAS_XLOCALE__ 1
|
103 |
|
|
#include <xlocale.h>
|
104 |
|
|
#include <locale.h>
|
105 |
|
|
#undef __UCLIBC_HAS_XLOCALE__
|
106 |
|
|
#endif /* __UCLIBC_HAS_XLOCALE__ */
|
107 |
|
|
|
108 |
|
|
#include <wchar.h>
|
109 |
|
|
|
110 |
|
|
#define LOCALE_NAMES (__locale_mmap->locale_names5)
|
111 |
|
|
#define LOCALES (__locale_mmap->locales)
|
112 |
|
|
#define LOCALE_AT_MODIFIERS (__locale_mmap->locale_at_modifiers)
|
113 |
|
|
#define CATEGORY_NAMES (__locale_mmap->lc_names)
|
114 |
|
|
|
115 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
116 |
|
|
#warning REMINDER: redo the MAX_LOCALE_STR stuff...
|
117 |
|
|
#endif
|
118 |
|
|
#define MAX_LOCALE_STR 256 /* TODO: Only sufficient for current case. */
|
119 |
|
|
#define MAX_LOCALE_CATEGORY_STR 32 /* TODO: Only sufficient for current case. */
|
120 |
|
|
/* Note: Best if MAX_LOCALE_CATEGORY_STR is a power of 2. */
|
121 |
|
|
|
122 |
|
|
extern int _locale_set_l(const unsigned char *p, __locale_t base);
|
123 |
|
|
extern void _locale_init_l(__locale_t base);
|
124 |
|
|
|
125 |
|
|
#endif /* __LOCALE_C_ONLY */
|
126 |
|
|
|
127 |
|
|
#undef LOCALE_STRING_SIZE
|
128 |
|
|
#define LOCALE_SELECTOR_SIZE (2 * __LC_ALL + 2)
|
129 |
|
|
|
130 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
131 |
|
|
#ifdef L_setlocale
|
132 |
|
|
#warning TODO: Create a C locale selector string.
|
133 |
|
|
#endif
|
134 |
|
|
#endif
|
135 |
|
|
#define C_LOCALE_SELECTOR "\x23\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80"
|
136 |
|
|
|
137 |
|
|
|
138 |
|
|
#include <langinfo.h>
|
139 |
|
|
#include <nl_types.h>
|
140 |
|
|
|
141 |
|
|
/**********************************************************************/
|
142 |
|
|
#ifdef L_setlocale
|
143 |
|
|
|
144 |
|
|
#ifdef __LOCALE_C_ONLY
|
145 |
|
|
|
146 |
|
|
link_warning(setlocale,"REMINDER: The 'setlocale' function supports only C|POSIX locales.")
|
147 |
|
|
|
148 |
|
|
static const char C_string[] = "C";
|
149 |
|
|
|
150 |
|
|
char *setlocale(int category, register const char *locale)
|
151 |
|
|
{
|
152 |
|
|
return ( (((unsigned int)(category)) <= LC_ALL)
|
153 |
|
|
&& ( (!locale) /* Request for locale category string. */
|
154 |
|
|
|| (!*locale) /* Implementation-defined default is C. */
|
155 |
|
|
|| ((*locale == 'C') && !locale[1])
|
156 |
|
|
|| (!strcmp(locale, "POSIX"))) )
|
157 |
|
|
? (char *) C_string /* Always in C/POSIX locale. */
|
158 |
|
|
: NULL;
|
159 |
|
|
}
|
160 |
|
|
|
161 |
|
|
#else /* ---------------------------------------------- __LOCALE_C_ONLY */
|
162 |
|
|
|
163 |
|
|
#ifdef __UCLIBC_HAS_THREADS__
|
164 |
|
|
link_warning(setlocale,"REMINDER: The 'setlocale' function is _not_ threadsafe except for simple queries.")
|
165 |
|
|
#endif
|
166 |
|
|
|
167 |
|
|
#if !defined(__LOCALE_DATA_NUM_LOCALES) || (__LOCALE_DATA_NUM_LOCALES <= 1)
|
168 |
|
|
#error locales enabled, but not data other than for C locale!
|
169 |
|
|
#endif
|
170 |
|
|
|
171 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
172 |
|
|
#warning TODO: Move posix and utf8 strings.
|
173 |
|
|
#endif
|
174 |
|
|
static const char posix[] = "POSIX";
|
175 |
|
|
static const char utf8[] = "UTF-8";
|
176 |
|
|
|
177 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
178 |
|
|
#warning TODO: Fix dimensions of hr_locale.
|
179 |
|
|
#endif
|
180 |
|
|
/* Individual category strings start at hr_locale + category * MAX_LOCALE_CATEGORY.
|
181 |
|
|
* This holds for LC_ALL as well.
|
182 |
|
|
*/
|
183 |
|
|
static char hr_locale[(MAX_LOCALE_CATEGORY_STR * LC_ALL) + MAX_LOCALE_STR];
|
184 |
|
|
|
185 |
|
|
static void update_hr_locale(const unsigned char *spec)
|
186 |
|
|
{
|
187 |
|
|
const unsigned char *loc;
|
188 |
|
|
const unsigned char *s;
|
189 |
|
|
char *n;
|
190 |
|
|
int i, category, done;
|
191 |
|
|
|
192 |
|
|
done = category = 0;
|
193 |
|
|
do {
|
194 |
|
|
s = spec + 1;
|
195 |
|
|
n = hr_locale + category * MAX_LOCALE_CATEGORY_STR;
|
196 |
|
|
|
197 |
|
|
if (category == LC_ALL) {
|
198 |
|
|
done = 1;
|
199 |
|
|
for (i = 0 ; i < LC_ALL-1 ; i += 2) {
|
200 |
|
|
if ((s[i] != s[i+2]) || (s[i+1] != s[i+3])) {
|
201 |
|
|
goto SKIP;
|
202 |
|
|
}
|
203 |
|
|
}
|
204 |
|
|
/* All categories the same, so simplify string by using a single
|
205 |
|
|
* category. */
|
206 |
|
|
category = LC_CTYPE;
|
207 |
|
|
}
|
208 |
|
|
|
209 |
|
|
SKIP:
|
210 |
|
|
i = (category == LC_ALL) ? 0 : category;
|
211 |
|
|
s += 2*i;
|
212 |
|
|
|
213 |
|
|
do {
|
214 |
|
|
if ((*s != 0xff) || (s[1] != 0xff)) {
|
215 |
|
|
loc = LOCALES
|
216 |
|
|
+ __LOCALE_DATA_WIDTH_LOCALES * ((((int)(*s & 0x7f)) << 7)
|
217 |
|
|
+ (s[1] & 0x7f));
|
218 |
|
|
if (category == LC_ALL) {
|
219 |
|
|
n = stpcpy(n, CATEGORY_NAMES + (int) CATEGORY_NAMES[i]);
|
220 |
|
|
*n++ = '=';
|
221 |
|
|
}
|
222 |
|
|
if (*loc == 0) {
|
223 |
|
|
*n++ = 'C';
|
224 |
|
|
*n = 0;
|
225 |
|
|
} else {
|
226 |
|
|
char at = 0;
|
227 |
|
|
memcpy(n, LOCALE_NAMES + 5*((*loc)-1), 5);
|
228 |
|
|
if (n[2] != '_') {
|
229 |
|
|
at = n[2];
|
230 |
|
|
n[2] = '_';
|
231 |
|
|
}
|
232 |
|
|
n += 5;
|
233 |
|
|
*n++ = '.';
|
234 |
|
|
if (loc[2] == 2) {
|
235 |
|
|
n = stpcpy(n, utf8);
|
236 |
|
|
} else if (loc[2] >= 3) {
|
237 |
|
|
n = stpcpy(n, CODESET_LIST + (int)(CODESET_LIST[loc[2] - 3]));
|
238 |
|
|
}
|
239 |
|
|
if (at) {
|
240 |
|
|
const char *q;
|
241 |
|
|
*n++ = '@';
|
242 |
|
|
q = LOCALE_AT_MODIFIERS;
|
243 |
|
|
do {
|
244 |
|
|
if (q[1] == at) {
|
245 |
|
|
n = stpcpy(n, q+2);
|
246 |
|
|
break;
|
247 |
|
|
}
|
248 |
|
|
q += 2 + *q;
|
249 |
|
|
} while (*q);
|
250 |
|
|
}
|
251 |
|
|
}
|
252 |
|
|
*n++ = ';';
|
253 |
|
|
}
|
254 |
|
|
s += 2;
|
255 |
|
|
} while (++i < category);
|
256 |
|
|
*--n = 0; /* Remove trailing ';' and nul-terminate. */
|
257 |
|
|
|
258 |
|
|
++category;
|
259 |
|
|
} while (!done);
|
260 |
|
|
}
|
261 |
|
|
|
262 |
|
|
char *setlocale(int category, const char *locale)
|
263 |
|
|
{
|
264 |
|
|
if (((unsigned int)(category)) > LC_ALL) {
|
265 |
|
|
#if 0
|
266 |
|
|
__set_errno(EINVAL); /* glibc sets errno -- SUSv3 doesn't say. */
|
267 |
|
|
#endif
|
268 |
|
|
return NULL; /* Illegal/unsupported category. */
|
269 |
|
|
}
|
270 |
|
|
|
271 |
|
|
if (locale != NULL) { /* Not just a query... */
|
272 |
|
|
if (!__newlocale((1 << category), locale, __global_locale)) {
|
273 |
|
|
return NULL; /* Failed! */
|
274 |
|
|
}
|
275 |
|
|
update_hr_locale(__global_locale->cur_locale);
|
276 |
|
|
}
|
277 |
|
|
|
278 |
|
|
/* Either a query or a successful set, so return current locale string. */
|
279 |
|
|
return hr_locale + (category * MAX_LOCALE_CATEGORY_STR);
|
280 |
|
|
}
|
281 |
|
|
|
282 |
|
|
#endif /* __LOCALE_C_ONLY */
|
283 |
|
|
|
284 |
|
|
#endif
|
285 |
|
|
/**********************************************************************/
|
286 |
|
|
#ifdef L_localeconv
|
287 |
|
|
|
288 |
|
|
/* Note: We assume here that the compiler does the sane thing regarding
|
289 |
|
|
* placement of the fields in the struct. If necessary, we could ensure
|
290 |
|
|
* this usings an array of offsets but at some size cost. */
|
291 |
|
|
|
292 |
|
|
#ifdef __LOCALE_C_ONLY
|
293 |
|
|
|
294 |
|
|
link_warning(localeconv,"REMINDER: The 'localeconv' function is hardwired for C/POSIX locale only.")
|
295 |
|
|
|
296 |
|
|
static struct lconv the_lconv;
|
297 |
|
|
|
298 |
|
|
static const char decpt[] = ".";
|
299 |
|
|
|
300 |
|
|
struct lconv *localeconv(void)
|
301 |
|
|
{
|
302 |
|
|
register char *p = (char *)(&the_lconv);
|
303 |
|
|
|
304 |
|
|
*((char **)p) = (char *) decpt;
|
305 |
|
|
do {
|
306 |
|
|
p += sizeof(char **);
|
307 |
|
|
*((char **)p) = (char *) (decpt+1);
|
308 |
|
|
} while (p < (char *) &the_lconv.negative_sign);
|
309 |
|
|
|
310 |
|
|
p = (&the_lconv.int_frac_digits);
|
311 |
|
|
do {
|
312 |
|
|
*p = CHAR_MAX;
|
313 |
|
|
++p;
|
314 |
|
|
} while (p <= &the_lconv.int_n_sign_posn);
|
315 |
|
|
|
316 |
|
|
return &the_lconv;
|
317 |
|
|
}
|
318 |
|
|
|
319 |
|
|
#else /* __LOCALE_C_ONLY */
|
320 |
|
|
|
321 |
|
|
static struct lconv the_lconv;
|
322 |
|
|
|
323 |
|
|
struct lconv *localeconv(void)
|
324 |
|
|
{
|
325 |
|
|
register char *p = (char *) &the_lconv;
|
326 |
|
|
register char **q = (char **) &(__UCLIBC_CURLOCALE_DATA).decimal_point;
|
327 |
|
|
|
328 |
|
|
do {
|
329 |
|
|
*((char **)p) = *q;
|
330 |
|
|
p += sizeof(char **);
|
331 |
|
|
++q;
|
332 |
|
|
} while (p < &the_lconv.int_frac_digits);
|
333 |
|
|
|
334 |
|
|
do {
|
335 |
|
|
*p = **q;
|
336 |
|
|
++p;
|
337 |
|
|
++q;
|
338 |
|
|
} while (p <= &the_lconv.int_n_sign_posn);
|
339 |
|
|
|
340 |
|
|
return &the_lconv;
|
341 |
|
|
}
|
342 |
|
|
|
343 |
|
|
#endif /* __LOCALE_C_ONLY */
|
344 |
|
|
|
345 |
|
|
#endif
|
346 |
|
|
/**********************************************************************/
|
347 |
|
|
#if defined(L__locale_init) && !defined(__LOCALE_C_ONLY)
|
348 |
|
|
|
349 |
|
|
static __uclibc_locale_t __global_locale_data;
|
350 |
|
|
|
351 |
|
|
__locale_t __global_locale = &__global_locale_data;
|
352 |
|
|
|
353 |
|
|
#ifdef __UCLIBC_HAS_XLOCALE__
|
354 |
|
|
__locale_t __curlocale_var = &__global_locale_data;
|
355 |
|
|
#endif
|
356 |
|
|
|
357 |
|
|
/*----------------------------------------------------------------------*/
|
358 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
359 |
|
|
#warning TODO: Move utf8 and ascii strings.
|
360 |
|
|
#endif
|
361 |
|
|
static const char utf8[] = "UTF-8";
|
362 |
|
|
static const char ascii[] = "ASCII";
|
363 |
|
|
|
364 |
|
|
typedef struct {
|
365 |
|
|
uint16_t num_base;
|
366 |
|
|
uint16_t num_der;
|
367 |
|
|
uint16_t MAX_WEIGHTS;
|
368 |
|
|
uint16_t num_index2weight;
|
369 |
|
|
#define num_index2ruleidx num_index2weight
|
370 |
|
|
uint16_t num_weightstr;
|
371 |
|
|
uint16_t num_multistart;
|
372 |
|
|
uint16_t num_override;
|
373 |
|
|
uint16_t num_ruletable;
|
374 |
|
|
} coldata_header_t;
|
375 |
|
|
|
376 |
|
|
typedef struct {
|
377 |
|
|
uint16_t num_weights;
|
378 |
|
|
uint16_t num_starters;
|
379 |
|
|
uint16_t ii_shift;
|
380 |
|
|
uint16_t ti_shift;
|
381 |
|
|
uint16_t ii_len;
|
382 |
|
|
uint16_t ti_len;
|
383 |
|
|
uint16_t max_weight;
|
384 |
|
|
uint16_t num_col_base;
|
385 |
|
|
uint16_t max_col_index;
|
386 |
|
|
uint16_t undefined_idx;
|
387 |
|
|
uint16_t range_low;
|
388 |
|
|
uint16_t range_count;
|
389 |
|
|
uint16_t range_base_weight;
|
390 |
|
|
uint16_t range_rule_offset;
|
391 |
|
|
|
392 |
|
|
uint16_t index2weight_offset;
|
393 |
|
|
uint16_t index2ruleidx_offset;
|
394 |
|
|
uint16_t multistart_offset;
|
395 |
|
|
uint16_t wcs2colidt_offset_low;
|
396 |
|
|
uint16_t wcs2colidt_offset_hi;
|
397 |
|
|
} coldata_base_t;
|
398 |
|
|
|
399 |
|
|
typedef struct {
|
400 |
|
|
uint16_t base_idx;
|
401 |
|
|
uint16_t undefined_idx;
|
402 |
|
|
uint16_t overrides_offset;
|
403 |
|
|
uint16_t multistart_offset;
|
404 |
|
|
} coldata_der_t;
|
405 |
|
|
|
406 |
|
|
static int init_cur_collate(int der_num, __collate_t *cur_collate)
|
407 |
|
|
{
|
408 |
|
|
const uint16_t *__locale_collate_tbl = __locale_mmap->collate_data;
|
409 |
|
|
coldata_header_t *cdh;
|
410 |
|
|
coldata_base_t *cdb;
|
411 |
|
|
coldata_der_t *cdd;
|
412 |
|
|
const uint16_t *p;
|
413 |
|
|
size_t n;
|
414 |
|
|
uint16_t i, w;
|
415 |
|
|
|
416 |
|
|
assert(sizeof(coldata_base_t) == 19*2);
|
417 |
|
|
assert(sizeof(coldata_der_t) == 4*2);
|
418 |
|
|
assert(sizeof(coldata_header_t) == 8*2);
|
419 |
|
|
|
420 |
|
|
if (!der_num) { /* C locale... special */
|
421 |
|
|
cur_collate->num_weights = 0;
|
422 |
|
|
return 1;
|
423 |
|
|
}
|
424 |
|
|
|
425 |
|
|
--der_num;
|
426 |
|
|
|
427 |
|
|
cdh = (coldata_header_t *) __locale_collate_tbl;
|
428 |
|
|
|
429 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
430 |
|
|
#warning CONSIDER: Should we assert here?
|
431 |
|
|
#endif
|
432 |
|
|
#if 0
|
433 |
|
|
if (der_num >= cdh->num_der) {
|
434 |
|
|
return 0;
|
435 |
|
|
}
|
436 |
|
|
#else
|
437 |
|
|
assert((der_num < cdh->num_der));
|
438 |
|
|
#endif
|
439 |
|
|
|
440 |
|
|
cdd = (coldata_der_t *)(__locale_collate_tbl
|
441 |
|
|
+ (sizeof(coldata_header_t)
|
442 |
|
|
+ cdh->num_base * sizeof(coldata_base_t)
|
443 |
|
|
+ der_num * sizeof(coldata_der_t)
|
444 |
|
|
)/2 );
|
445 |
|
|
|
446 |
|
|
cdb = (coldata_base_t *)(__locale_collate_tbl
|
447 |
|
|
+ (sizeof(coldata_header_t)
|
448 |
|
|
+ cdd->base_idx * sizeof(coldata_base_t)
|
449 |
|
|
)/2 );
|
450 |
|
|
|
451 |
|
|
memcpy(cur_collate, cdb, offsetof(coldata_base_t,index2weight_offset));
|
452 |
|
|
cur_collate->undefined_idx = cdd->undefined_idx;
|
453 |
|
|
|
454 |
|
|
cur_collate->ti_mask = (1 << cur_collate->ti_shift)-1;
|
455 |
|
|
cur_collate->ii_mask = (1 << cur_collate->ii_shift)-1;
|
456 |
|
|
|
457 |
|
|
/* fflush(stdout); */
|
458 |
|
|
/* fprintf(stderr,"base=%d num_col_base: %d %d\n", cdd->base_idx ,cur_collate->num_col_base, cdb->num_col_base); */
|
459 |
|
|
|
460 |
|
|
n = (sizeof(coldata_header_t) + cdh->num_base * sizeof(coldata_base_t)
|
461 |
|
|
+ cdh->num_der * sizeof(coldata_der_t))/2;
|
462 |
|
|
|
463 |
|
|
/* fprintf(stderr,"n = %d\n", n); */
|
464 |
|
|
cur_collate->index2weight_tbl = __locale_collate_tbl + n + cdb->index2weight_offset;
|
465 |
|
|
/* fprintf(stderr,"i2w = %d\n", n + cdb->index2weight_offset); */
|
466 |
|
|
n += cdh->num_index2weight;
|
467 |
|
|
cur_collate->index2ruleidx_tbl = __locale_collate_tbl + n + cdb->index2ruleidx_offset;
|
468 |
|
|
/* fprintf(stderr,"i2r = %d\n", n + cdb->index2ruleidx_offset); */
|
469 |
|
|
n += cdh->num_index2ruleidx;
|
470 |
|
|
cur_collate->multistart_tbl = __locale_collate_tbl + n + cdd->multistart_offset;
|
471 |
|
|
/* fprintf(stderr,"mts = %d\n", n + cdb->multistart_offset); */
|
472 |
|
|
n += cdh->num_multistart;
|
473 |
|
|
cur_collate->overrides_tbl = __locale_collate_tbl + n + cdd->overrides_offset;
|
474 |
|
|
/* fprintf(stderr,"ovr = %d\n", n + cdd->overrides_offset); */
|
475 |
|
|
n += cdh->num_override;
|
476 |
|
|
cur_collate->ruletable = __locale_collate_tbl + n;
|
477 |
|
|
/* fprintf(stderr, "rtb = %d\n", n); */
|
478 |
|
|
n += cdh->num_ruletable;
|
479 |
|
|
cur_collate->weightstr = __locale_collate_tbl + n;
|
480 |
|
|
/* fprintf(stderr,"wts = %d\n", n); */
|
481 |
|
|
n += cdh->num_weightstr;
|
482 |
|
|
cur_collate->wcs2colidt_tbl = __locale_collate_tbl + n
|
483 |
|
|
+ (((unsigned long)(cdb->wcs2colidt_offset_hi)) << 16)
|
484 |
|
|
+ cdb->wcs2colidt_offset_low;
|
485 |
|
|
/* fprintf(stderr,"wcs = %lu\n", n + (((unsigned long)(cdb->wcs2colidt_offset_hi)) << 16) */
|
486 |
|
|
/* + cdb->wcs2colidt_offset_low); */
|
487 |
|
|
|
488 |
|
|
cur_collate->MAX_WEIGHTS = cdh->MAX_WEIGHTS;
|
489 |
|
|
|
490 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
491 |
|
|
#warning CONSIDER: Fix the +1 by increasing max_col_index?
|
492 |
|
|
#warning CONSIDER: Since this collate info is dependent only on LC_COLLATE ll_cc and not on codeset, we could just globally allocate this for each in a table
|
493 |
|
|
#endif
|
494 |
|
|
|
495 |
|
|
cur_collate->index2weight = calloc(2*cur_collate->max_col_index+2,
|
496 |
|
|
sizeof(uint16_t));
|
497 |
|
|
if (!cur_collate->index2weight) {
|
498 |
|
|
return 0;
|
499 |
|
|
}
|
500 |
|
|
cur_collate->index2ruleidx = cur_collate->index2weight
|
501 |
|
|
+ cur_collate->max_col_index + 1;
|
502 |
|
|
|
503 |
|
|
memcpy(cur_collate->index2weight, cur_collate->index2weight_tbl,
|
504 |
|
|
cur_collate->num_col_base * sizeof(uint16_t));
|
505 |
|
|
memcpy(cur_collate->index2ruleidx, cur_collate->index2ruleidx_tbl,
|
506 |
|
|
cur_collate->num_col_base * sizeof(uint16_t));
|
507 |
|
|
|
508 |
|
|
/* now do the overrides */
|
509 |
|
|
p = cur_collate->overrides_tbl;
|
510 |
|
|
while (*p > 1) {
|
511 |
|
|
/* fprintf(stderr, "processing override -- count = %d\n", *p); */
|
512 |
|
|
n = *p++;
|
513 |
|
|
w = *p++;
|
514 |
|
|
do {
|
515 |
|
|
i = *p++;
|
516 |
|
|
/* fprintf(stderr, " i=%d (%#x) w=%d *p=%d\n", i, i, w, *p); */
|
517 |
|
|
cur_collate->index2weight[i-1] = w++;
|
518 |
|
|
cur_collate->index2ruleidx[i-1] = *p++;
|
519 |
|
|
} while (--n);
|
520 |
|
|
}
|
521 |
|
|
assert(*p == 1);
|
522 |
|
|
while (*++p) {
|
523 |
|
|
i = *p;
|
524 |
|
|
/* fprintf(stderr, " i=%d (%#x) w=%d *p=%d\n", i, i, p[1], p[2]); */
|
525 |
|
|
cur_collate->index2weight[i-1] = *++p;
|
526 |
|
|
cur_collate->index2ruleidx[i-1] = *++p;
|
527 |
|
|
}
|
528 |
|
|
|
529 |
|
|
|
530 |
|
|
for (i=0 ; i < cur_collate->multistart_tbl[0] ; i++) {
|
531 |
|
|
p = cur_collate->multistart_tbl;
|
532 |
|
|
/* fprintf(stderr, "%2d of %2d: %d ", i, cur_collate->multistart_tbl[0], p[i]); */
|
533 |
|
|
p += p[i];
|
534 |
|
|
|
535 |
|
|
do {
|
536 |
|
|
n = *p++;
|
537 |
|
|
do {
|
538 |
|
|
if (!*p) { /* found it */
|
539 |
|
|
/* fprintf(stderr, "found: n=%d (%#lx) |%.*ls|\n", n, (int) *cs->s, n, cs->s); */
|
540 |
|
|
/* fprintf(stderr, ": %d - single\n", n); */
|
541 |
|
|
goto FOUND;
|
542 |
|
|
}
|
543 |
|
|
/* the lookup check here is safe since we're assured that *p is a valid colidex */
|
544 |
|
|
/* fprintf(stderr, "lookup(%lc)==%d *p==%d\n", cs->s[n], lookup(cs->s[n]), (int) *p); */
|
545 |
|
|
/* fprintf(stderr, ": %d - ", n); */
|
546 |
|
|
do {
|
547 |
|
|
/* fprintf(stderr, "%d|", *p); */
|
548 |
|
|
} while (*p++);
|
549 |
|
|
break;
|
550 |
|
|
} while (1);
|
551 |
|
|
} while (1);
|
552 |
|
|
FOUND:
|
553 |
|
|
continue;
|
554 |
|
|
}
|
555 |
|
|
|
556 |
|
|
return 1;
|
557 |
|
|
}
|
558 |
|
|
|
559 |
|
|
int _locale_set_l(const unsigned char *p, __locale_t base)
|
560 |
|
|
{
|
561 |
|
|
const char **x;
|
562 |
|
|
unsigned char *s = base->cur_locale + 1;
|
563 |
|
|
const size_t *stp;
|
564 |
|
|
const unsigned char *r;
|
565 |
|
|
const uint16_t *io;
|
566 |
|
|
const uint16_t *ii;
|
567 |
|
|
const unsigned char *d;
|
568 |
|
|
int row; /* locale row */
|
569 |
|
|
int crow; /* category row */
|
570 |
|
|
int len;
|
571 |
|
|
int c;
|
572 |
|
|
int i = 0;
|
573 |
|
|
__collate_t newcol;
|
574 |
|
|
|
575 |
|
|
++p;
|
576 |
|
|
|
577 |
|
|
newcol.index2weight = NULL;
|
578 |
|
|
if ((p[2*LC_COLLATE] != s[2*LC_COLLATE])
|
579 |
|
|
|| (p[2*LC_COLLATE + 1] != s[2*LC_COLLATE + 1])
|
580 |
|
|
) {
|
581 |
|
|
row = (((int)(*p & 0x7f)) << 7) + (p[1] & 0x7f);
|
582 |
|
|
assert(row < __LOCALE_DATA_NUM_LOCALES);
|
583 |
|
|
if (!init_cur_collate(__locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES
|
584 |
|
|
* row + 3 + LC_COLLATE ],
|
585 |
|
|
&newcol)
|
586 |
|
|
) {
|
587 |
|
|
return 0; /* calloc failed. */
|
588 |
|
|
}
|
589 |
|
|
free(base->collate.index2weight);
|
590 |
|
|
memcpy(&base->collate, &newcol, sizeof(__collate_t));
|
591 |
|
|
}
|
592 |
|
|
|
593 |
|
|
do {
|
594 |
|
|
if ((*p != *s) || (p[1] != s[1])) {
|
595 |
|
|
row = (((int)(*p & 0x7f)) << 7) + (p[1] & 0x7f);
|
596 |
|
|
assert(row < __LOCALE_DATA_NUM_LOCALES);
|
597 |
|
|
|
598 |
|
|
*s = *p;
|
599 |
|
|
s[1] = p[1];
|
600 |
|
|
|
601 |
|
|
if ((i != LC_COLLATE)
|
602 |
|
|
&& ((len = __locale_mmap->lc_common_item_offsets_LEN[i]) != 0)
|
603 |
|
|
) {
|
604 |
|
|
crow = __locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES * row
|
605 |
|
|
+ 3 + i ]
|
606 |
|
|
* len;
|
607 |
|
|
|
608 |
|
|
x = (const char **)(((char *) base)
|
609 |
|
|
+ base->category_offsets[i]);
|
610 |
|
|
|
611 |
|
|
stp = __locale_mmap->lc_common_tbl_offsets + 4*i;
|
612 |
|
|
r = (const unsigned char *)( ((char *)__locale_mmap) + *stp );
|
613 |
|
|
io = (const uint16_t *)( ((char *)__locale_mmap) + *++stp );
|
614 |
|
|
ii = (const uint16_t *)( ((char *)__locale_mmap) + *++stp );
|
615 |
|
|
d = (const unsigned char *)( ((char *)__locale_mmap) + *++stp );
|
616 |
|
|
for (c=0 ; c < len ; c++) {
|
617 |
|
|
*(x + c) = d + ii[ r[crow + c] + io[c] ];
|
618 |
|
|
}
|
619 |
|
|
}
|
620 |
|
|
if (i == LC_CTYPE) {
|
621 |
|
|
c = __locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES * row
|
622 |
|
|
+ 2 ]; /* codeset */
|
623 |
|
|
if (c <= 2) {
|
624 |
|
|
if (c == 2) {
|
625 |
|
|
base->codeset = utf8;
|
626 |
|
|
base->encoding = __ctype_encoding_utf8;
|
627 |
|
|
/* TODO - fix for bcc */
|
628 |
|
|
base->mb_cur_max = 6;
|
629 |
|
|
} else {
|
630 |
|
|
assert(c==1);
|
631 |
|
|
base->codeset = ascii;
|
632 |
|
|
base->encoding = __ctype_encoding_7_bit;
|
633 |
|
|
base->mb_cur_max = 1;
|
634 |
|
|
}
|
635 |
|
|
} else {
|
636 |
|
|
const __codeset_8_bit_t *c8b;
|
637 |
|
|
r = CODESET_LIST;
|
638 |
|
|
base->codeset = r + r[c -= 3];
|
639 |
|
|
base->encoding = __ctype_encoding_8_bit;
|
640 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
641 |
|
|
#warning REMINDER: update 8 bit mb_cur_max when translit implemented!
|
642 |
|
|
#endif
|
643 |
|
|
/* TODO - update when translit implemented! */
|
644 |
|
|
base->mb_cur_max = 1;
|
645 |
|
|
c8b = __locale_mmap->codeset_8_bit + c;
|
646 |
|
|
#ifdef __CTYPE_HAS_8_BIT_LOCALES
|
647 |
|
|
base->idx8ctype = c8b->idx8ctype;
|
648 |
|
|
base->idx8uplow = c8b->idx8uplow;
|
649 |
|
|
#ifdef __UCLIBC_HAS_WCHAR__
|
650 |
|
|
base->idx8c2wc = c8b->idx8c2wc;
|
651 |
|
|
base->idx8wc2c = c8b->idx8wc2c;
|
652 |
|
|
/* translit */
|
653 |
|
|
#endif /* __UCLIBC_HAS_WCHAR__ */
|
654 |
|
|
|
655 |
|
|
/* What follows is fairly bloated, but it is just a hack
|
656 |
|
|
* to get the 8-bit codeset ctype stuff functioning.
|
657 |
|
|
* All of this will be replaced in the next generation
|
658 |
|
|
* of locale support anyway... */
|
659 |
|
|
|
660 |
|
|
memcpy(base->__ctype_b_data,
|
661 |
|
|
__C_ctype_b - __UCLIBC_CTYPE_B_TBL_OFFSET,
|
662 |
|
|
(256 + __UCLIBC_CTYPE_B_TBL_OFFSET)
|
663 |
|
|
* sizeof(__ctype_mask_t));
|
664 |
|
|
memcpy(base->__ctype_tolower_data,
|
665 |
|
|
__C_ctype_tolower - __UCLIBC_CTYPE_TO_TBL_OFFSET,
|
666 |
|
|
(256 + __UCLIBC_CTYPE_TO_TBL_OFFSET)
|
667 |
|
|
* sizeof(__ctype_touplow_t));
|
668 |
|
|
memcpy(base->__ctype_toupper_data,
|
669 |
|
|
__C_ctype_toupper - __UCLIBC_CTYPE_TO_TBL_OFFSET,
|
670 |
|
|
(256 + __UCLIBC_CTYPE_TO_TBL_OFFSET)
|
671 |
|
|
* sizeof(__ctype_touplow_t));
|
672 |
|
|
|
673 |
|
|
#define Cctype_TBL_MASK ((1 << __LOCALE_DATA_Cctype_IDX_SHIFT) - 1)
|
674 |
|
|
#define Cctype_IDX_OFFSET (128 >> __LOCALE_DATA_Cctype_IDX_SHIFT)
|
675 |
|
|
|
676 |
|
|
{
|
677 |
|
|
int u;
|
678 |
|
|
__ctype_mask_t m;
|
679 |
|
|
|
680 |
|
|
for (u=0 ; u < 128 ; u++) {
|
681 |
|
|
#ifdef __LOCALE_DATA_Cctype_PACKED
|
682 |
|
|
c = base->tbl8ctype
|
683 |
|
|
[ ((int)(c8b->idx8ctype
|
684 |
|
|
[(u >> __LOCALE_DATA_Cctype_IDX_SHIFT) ])
|
685 |
|
|
<< (__LOCALE_DATA_Cctype_IDX_SHIFT - 1))
|
686 |
|
|
+ ((u & Cctype_TBL_MASK) >> 1)];
|
687 |
|
|
c = (u & 1) ? (c >> 4) : (c & 0xf);
|
688 |
|
|
#else
|
689 |
|
|
c = base->tbl8ctype
|
690 |
|
|
[ ((int)(c8b->idx8ctype
|
691 |
|
|
[(u >> __LOCALE_DATA_Cctype_IDX_SHIFT) ])
|
692 |
|
|
<< __LOCALE_DATA_Cctype_IDX_SHIFT)
|
693 |
|
|
+ (u & Cctype_TBL_MASK) ];
|
694 |
|
|
#endif
|
695 |
|
|
|
696 |
|
|
m = base->code2flag[c];
|
697 |
|
|
|
698 |
|
|
base->__ctype_b_data
|
699 |
|
|
[128 + __UCLIBC_CTYPE_B_TBL_OFFSET + u]
|
700 |
|
|
= m;
|
701 |
|
|
|
702 |
|
|
#ifdef __UCLIBC_HAS_CTYPE_SIGNED__
|
703 |
|
|
if (((signed char)(128 + u)) != -1) {
|
704 |
|
|
base->__ctype_b_data[__UCLIBC_CTYPE_B_TBL_OFFSET
|
705 |
|
|
+ ((signed char)(128 + u))]
|
706 |
|
|
= m;
|
707 |
|
|
}
|
708 |
|
|
#endif
|
709 |
|
|
|
710 |
|
|
base->__ctype_tolower_data
|
711 |
|
|
[128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
|
712 |
|
|
= 128 + u;
|
713 |
|
|
base->__ctype_toupper_data
|
714 |
|
|
[128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
|
715 |
|
|
= 128 + u;
|
716 |
|
|
|
717 |
|
|
if (m & (_ISlower|_ISupper)) {
|
718 |
|
|
c = base->tbl8uplow
|
719 |
|
|
[ ((int)(c8b->idx8uplow
|
720 |
|
|
[u >> __LOCALE_DATA_Cuplow_IDX_SHIFT])
|
721 |
|
|
<< __LOCALE_DATA_Cuplow_IDX_SHIFT)
|
722 |
|
|
+ ((128 + u)
|
723 |
|
|
& ((1 << __LOCALE_DATA_Cuplow_IDX_SHIFT)
|
724 |
|
|
- 1)) ];
|
725 |
|
|
if (m & _ISlower) {
|
726 |
|
|
base->__ctype_toupper_data
|
727 |
|
|
[128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
|
728 |
|
|
= (unsigned char)(128 + u + c);
|
729 |
|
|
#ifdef __UCLIBC_HAS_CTYPE_SIGNED__
|
730 |
|
|
if (((signed char)(128 + u)) != -1) {
|
731 |
|
|
base->__ctype_toupper_data
|
732 |
|
|
[__UCLIBC_CTYPE_TO_TBL_OFFSET
|
733 |
|
|
+ ((signed char)(128 + u))]
|
734 |
|
|
= (unsigned char)(128 + u + c);
|
735 |
|
|
}
|
736 |
|
|
#endif
|
737 |
|
|
} else {
|
738 |
|
|
base->__ctype_tolower_data
|
739 |
|
|
[128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
|
740 |
|
|
= (unsigned char)(128 + u - c);
|
741 |
|
|
#ifdef __UCLIBC_HAS_CTYPE_SIGNED__
|
742 |
|
|
if (((signed char)(128 + u)) != -1) {
|
743 |
|
|
base->__ctype_tolower_data
|
744 |
|
|
[__UCLIBC_CTYPE_TO_TBL_OFFSET
|
745 |
|
|
+ ((signed char)(128 + u))]
|
746 |
|
|
= (unsigned char)(128 + u - c);
|
747 |
|
|
}
|
748 |
|
|
#endif
|
749 |
|
|
}
|
750 |
|
|
}
|
751 |
|
|
}
|
752 |
|
|
}
|
753 |
|
|
|
754 |
|
|
#ifdef __UCLIBC_HAS_XLOCALE__
|
755 |
|
|
base->__ctype_b = base->__ctype_b_data
|
756 |
|
|
+ __UCLIBC_CTYPE_B_TBL_OFFSET;
|
757 |
|
|
base->__ctype_tolower = base->__ctype_tolower_data
|
758 |
|
|
+ __UCLIBC_CTYPE_TO_TBL_OFFSET;
|
759 |
|
|
base->__ctype_toupper = base->__ctype_toupper_data
|
760 |
|
|
+ __UCLIBC_CTYPE_TO_TBL_OFFSET;
|
761 |
|
|
#else /* __UCLIBC_HAS_XLOCALE__ */
|
762 |
|
|
__ctype_b = base->__ctype_b_data
|
763 |
|
|
+ __UCLIBC_CTYPE_B_TBL_OFFSET;
|
764 |
|
|
__ctype_tolower = base->__ctype_tolower_data
|
765 |
|
|
+ __UCLIBC_CTYPE_TO_TBL_OFFSET;
|
766 |
|
|
__ctype_toupper = base->__ctype_toupper_data
|
767 |
|
|
+ __UCLIBC_CTYPE_TO_TBL_OFFSET;
|
768 |
|
|
#endif /* __UCLIBC_HAS_XLOCALE__ */
|
769 |
|
|
|
770 |
|
|
#endif /* __CTYPE_HAS_8_BIT_LOCALES */
|
771 |
|
|
}
|
772 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
773 |
|
|
#warning TODO: Put the outdigit string length in the locale_mmap object.
|
774 |
|
|
#endif
|
775 |
|
|
d = base->outdigit_length;
|
776 |
|
|
x = &base->outdigit0_mb;
|
777 |
|
|
for (c = 0 ; c < 10 ; c++) {
|
778 |
|
|
((unsigned char *)d)[c] = strlen(x[c]);
|
779 |
|
|
assert(d[c] > 0);
|
780 |
|
|
}
|
781 |
|
|
} else if (i == LC_NUMERIC) {
|
782 |
|
|
assert(LC_NUMERIC > LC_CTYPE); /* Need ctype initialized. */
|
783 |
|
|
|
784 |
|
|
base->decimal_point_len
|
785 |
|
|
= __locale_mbrtowc_l(&base->decimal_point_wc,
|
786 |
|
|
base->decimal_point, base);
|
787 |
|
|
assert(base->decimal_point_len > 0);
|
788 |
|
|
assert(base->decimal_point[base->decimal_point_len] == 0);
|
789 |
|
|
|
790 |
|
|
if (*base->grouping) {
|
791 |
|
|
base->thousands_sep_len
|
792 |
|
|
= __locale_mbrtowc_l(&base->thousands_sep_wc,
|
793 |
|
|
base->thousands_sep, base);
|
794 |
|
|
#if 1
|
795 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
796 |
|
|
#warning TODO: Remove hack involving grouping without a thousep char (bg_BG).
|
797 |
|
|
#endif
|
798 |
|
|
assert(base->thousands_sep_len >= 0);
|
799 |
|
|
if (base->thousands_sep_len == 0) {
|
800 |
|
|
base->grouping = base->thousands_sep; /* empty string */
|
801 |
|
|
}
|
802 |
|
|
assert(base->thousands_sep[base->thousands_sep_len] == 0);
|
803 |
|
|
#else
|
804 |
|
|
assert(base->thousands_sep_len > 0);
|
805 |
|
|
assert(base->thousands_sep[base->thousands_sep_len] == 0);
|
806 |
|
|
#endif
|
807 |
|
|
}
|
808 |
|
|
|
809 |
|
|
/* } else if (i == LC_COLLATE) { */
|
810 |
|
|
/* init_cur_collate(__locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES */
|
811 |
|
|
/* * row + 3 + i ], */
|
812 |
|
|
/* &base->collate); */
|
813 |
|
|
}
|
814 |
|
|
}
|
815 |
|
|
++i;
|
816 |
|
|
p += 2;
|
817 |
|
|
s += 2;
|
818 |
|
|
} while (i < LC_ALL);
|
819 |
|
|
|
820 |
|
|
return 1;
|
821 |
|
|
}
|
822 |
|
|
|
823 |
|
|
static const uint16_t __code2flag[16] = {
|
824 |
|
|
0, /* unclassified = 0 */
|
825 |
|
|
_ISprint|_ISgraph|_ISalnum|_ISalpha, /* alpha_nonupper_nonlower */
|
826 |
|
|
_ISprint|_ISgraph|_ISalnum|_ISalpha|_ISlower, /* alpha_lower */
|
827 |
|
|
_ISprint|_ISgraph|_ISalnum|_ISalpha|_ISlower|_ISupper, /* alpha_upper_lower */
|
828 |
|
|
_ISprint|_ISgraph|_ISalnum|_ISalpha|_ISupper, /* alpha_upper */
|
829 |
|
|
_ISprint|_ISgraph|_ISalnum|_ISdigit, /* digit */
|
830 |
|
|
_ISprint|_ISgraph|_ISpunct, /* punct */
|
831 |
|
|
_ISprint|_ISgraph, /* graph */
|
832 |
|
|
_ISprint|_ISspace, /* print_space_nonblank */
|
833 |
|
|
_ISprint|_ISspace|_ISblank, /* print_space_blank */
|
834 |
|
|
_ISspace, /* space_nonblank_noncntrl */
|
835 |
|
|
_ISspace|_ISblank, /* space_blank_noncntrl */
|
836 |
|
|
_IScntrl|_ISspace, /* cntrl_space_nonblank */
|
837 |
|
|
_IScntrl|_ISspace|_ISblank, /* cntrl_space_blank */
|
838 |
|
|
_IScntrl /* cntrl_nonspace */
|
839 |
|
|
};
|
840 |
|
|
|
841 |
|
|
void _locale_init_l(__locale_t base)
|
842 |
|
|
{
|
843 |
|
|
memset(base->cur_locale, 0, LOCALE_SELECTOR_SIZE);
|
844 |
|
|
base->cur_locale[0] = '#';
|
845 |
|
|
|
846 |
|
|
memcpy(base->category_item_count,
|
847 |
|
|
__locale_mmap->lc_common_item_offsets_LEN,
|
848 |
|
|
LC_ALL);
|
849 |
|
|
|
850 |
|
|
++base->category_item_count[0]; /* Increment for codeset entry. */
|
851 |
|
|
base->category_offsets[0] = offsetof(__uclibc_locale_t, outdigit0_mb);
|
852 |
|
|
base->category_offsets[1] = offsetof(__uclibc_locale_t, decimal_point);
|
853 |
|
|
base->category_offsets[2] = offsetof(__uclibc_locale_t, int_curr_symbol);
|
854 |
|
|
base->category_offsets[3] = offsetof(__uclibc_locale_t, abday_1);
|
855 |
|
|
/* base->category_offsets[4] = offsetof(__uclibc_locale_t, collate???); */
|
856 |
|
|
base->category_offsets[5] = offsetof(__uclibc_locale_t, yesexpr);
|
857 |
|
|
|
858 |
|
|
#ifdef __CTYPE_HAS_8_BIT_LOCALES
|
859 |
|
|
base->tbl8ctype
|
860 |
|
|
= (const unsigned char *) &__locale_mmap->tbl8ctype;
|
861 |
|
|
base->tbl8uplow
|
862 |
|
|
= (const unsigned char *) &__locale_mmap->tbl8uplow;
|
863 |
|
|
#ifdef __UCLIBC_HAS_WCHAR__
|
864 |
|
|
base->tbl8c2wc
|
865 |
|
|
= (const uint16_t *) &__locale_mmap->tbl8c2wc;
|
866 |
|
|
base->tbl8wc2c
|
867 |
|
|
= (const unsigned char *) &__locale_mmap->tbl8wc2c;
|
868 |
|
|
/* translit */
|
869 |
|
|
#endif /* __UCLIBC_HAS_WCHAR__ */
|
870 |
|
|
#endif /* __CTYPE_HAS_8_BIT_LOCALES */
|
871 |
|
|
#ifdef __UCLIBC_HAS_WCHAR__
|
872 |
|
|
base->tblwctype
|
873 |
|
|
= (const unsigned char *) &__locale_mmap->tblwctype;
|
874 |
|
|
base->tblwuplow
|
875 |
|
|
= (const unsigned char *) &__locale_mmap->tblwuplow;
|
876 |
|
|
base->tblwuplow_diff
|
877 |
|
|
= (const uint16_t *) &__locale_mmap->tblwuplow_diff;
|
878 |
|
|
/* base->tblwcomb */
|
879 |
|
|
/* = (const unsigned char *) &__locale_mmap->tblwcomb; */
|
880 |
|
|
/* width?? */
|
881 |
|
|
#endif /* __UCLIBC_HAS_WCHAR__ */
|
882 |
|
|
|
883 |
|
|
/* Initially, set things up to use the global C ctype tables.
|
884 |
|
|
* This is correct for C (ASCII) and UTF-8 based locales (except tr_TR). */
|
885 |
|
|
#ifdef __UCLIBC_HAS_XLOCALE__
|
886 |
|
|
base->__ctype_b = __C_ctype_b;
|
887 |
|
|
base->__ctype_tolower = __C_ctype_tolower;
|
888 |
|
|
base->__ctype_toupper = __C_ctype_toupper;
|
889 |
|
|
#else /* __UCLIBC_HAS_XLOCALE__ */
|
890 |
|
|
__ctype_b = __C_ctype_b;
|
891 |
|
|
__ctype_tolower = __C_ctype_tolower;
|
892 |
|
|
__ctype_toupper = __C_ctype_toupper;
|
893 |
|
|
#endif /* __UCLIBC_HAS_XLOCALE__ */
|
894 |
|
|
|
895 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
896 |
|
|
#warning TODO: Initialize code2flag correctly based on locale_mmap.
|
897 |
|
|
#endif
|
898 |
|
|
base->code2flag = __code2flag;
|
899 |
|
|
|
900 |
|
|
|
901 |
|
|
_locale_set_l(C_LOCALE_SELECTOR, base);
|
902 |
|
|
}
|
903 |
|
|
|
904 |
|
|
void _locale_init(void)
|
905 |
|
|
{
|
906 |
|
|
/* TODO: mmap the locale file */
|
907 |
|
|
|
908 |
|
|
/* TODO - ??? */
|
909 |
|
|
_locale_init_l(__global_locale);
|
910 |
|
|
}
|
911 |
|
|
|
912 |
|
|
#endif
|
913 |
|
|
/**********************************************************************/
|
914 |
|
|
#if defined(L_nl_langinfo) || defined(L_nl_langinfo_l)
|
915 |
|
|
|
916 |
|
|
#ifdef __LOCALE_C_ONLY
|
917 |
|
|
|
918 |
|
|
/* We need to index 320 bytes of data, so you might initially think we
|
919 |
|
|
* need to store the offsets in shorts. But since the offset of the
|
920 |
|
|
* 64th item is 182, we'll store "offset - 2*64" for all items >= 64
|
921 |
|
|
* and always calculate the data offset as "offset[i] + 2*(i & 64)".
|
922 |
|
|
* This allows us to pack the data offsets in an unsigned char while
|
923 |
|
|
* also avoiding an "if".
|
924 |
|
|
*
|
925 |
|
|
* Note: Category order is assumed to be:
|
926 |
|
|
* ctype, numeric, monetary, time, collate, messages, all
|
927 |
|
|
*/
|
928 |
|
|
|
929 |
|
|
#define C_LC_ALL 6
|
930 |
|
|
|
931 |
|
|
/* Combine the data to avoid size penalty for seperate char arrays when
|
932 |
|
|
* compiler aligns objects. The original code is left in as documentation. */
|
933 |
|
|
#define cat_start nl_data
|
934 |
|
|
#define C_locale_data (nl_data + C_LC_ALL + 1 + 90)
|
935 |
|
|
|
936 |
|
|
static const unsigned char nl_data[C_LC_ALL + 1 + 90 + 320] = {
|
937 |
|
|
/* static const char cat_start[LC_ALL + 1] = { */
|
938 |
|
|
'\x00', '\x0b', '\x0e', '\x24', '\x56', '\x56', '\x5a',
|
939 |
|
|
/* }; */
|
940 |
|
|
/* static const char item_offset[90] = { */
|
941 |
|
|
'\x00', '\x02', '\x04', '\x06', '\x08', '\x0a', '\x0c', '\x0e',
|
942 |
|
|
'\x10', '\x12', '\x14', '\x1a', '\x1b', '\x1b', '\x1b', '\x1b',
|
943 |
|
|
'\x1b', '\x1b', '\x1b', '\x1b', '\x1b', '\x1c', '\x1c', '\x1c',
|
944 |
|
|
'\x1c', '\x1c', '\x1c', '\x1c', '\x1c', '\x1c', '\x1c', '\x1c',
|
945 |
|
|
'\x1c', '\x1c', '\x1c', '\x1e', '\x20', '\x24', '\x28', '\x2c',
|
946 |
|
|
'\x30', '\x34', '\x38', '\x3c', '\x43', '\x4a', '\x52', '\x5c',
|
947 |
|
|
'\x65', '\x6c', '\x75', '\x79', '\x7d', '\x81', '\x85', '\x89',
|
948 |
|
|
'\x8d', '\x91', '\x95', '\x99', '\x9d', '\xa1', '\xa5', '\xad',
|
949 |
|
|
'\x36', '\x3c', '\x42', '\x46', '\x4b', '\x50', '\x57', '\x61',
|
950 |
|
|
'\x69', '\x72', '\x7b', '\x7e', '\x81', '\x96', '\x9f', '\xa8',
|
951 |
|
|
'\xb3', '\xb3', '\xb3', '\xb3', '\xb3', '\xb3', '\xb4', '\xba',
|
952 |
|
|
'\xbf', '\xbf',
|
953 |
|
|
/* }; */
|
954 |
|
|
/* static const char C_locale_data[320] = { */
|
955 |
|
|
'0', '\x00', '1', '\x00', '2', '\x00', '3', '\x00',
|
956 |
|
|
'4', '\x00', '5', '\x00', '6', '\x00', '7', '\x00',
|
957 |
|
|
'8', '\x00', '9', '\x00', 'A', 'S', 'C', 'I',
|
958 |
|
|
'I', '\x00', '.', '\x00', '\x7f', '\x00', '-', '\x00',
|
959 |
|
|
'S', 'u', 'n', '\x00', 'M', 'o', 'n', '\x00',
|
960 |
|
|
'T', 'u', 'e', '\x00', 'W', 'e', 'd', '\x00',
|
961 |
|
|
'T', 'h', 'u', '\x00', 'F', 'r', 'i', '\x00',
|
962 |
|
|
'S', 'a', 't', '\x00', 'S', 'u', 'n', 'd',
|
963 |
|
|
'a', 'y', '\x00', 'M', 'o', 'n', 'd', 'a',
|
964 |
|
|
'y', '\x00', 'T', 'u', 'e', 's', 'd', 'a',
|
965 |
|
|
'y', '\x00', 'W', 'e', 'd', 'n', 'e', 's',
|
966 |
|
|
'd', 'a', 'y', '\x00', 'T', 'h', 'u', 'r',
|
967 |
|
|
's', 'd', 'a', 'y', '\x00', 'F', 'r', 'i',
|
968 |
|
|
'd', 'a', 'y', '\x00', 'S', 'a', 't', 'u',
|
969 |
|
|
'r', 'd', 'a', 'y', '\x00', 'J', 'a', 'n',
|
970 |
|
|
'\x00', 'F', 'e', 'b', '\x00', 'M', 'a', 'r',
|
971 |
|
|
'\x00', 'A', 'p', 'r', '\x00', 'M', 'a', 'y',
|
972 |
|
|
'\x00', 'J', 'u', 'n', '\x00', 'J', 'u', 'l',
|
973 |
|
|
'\x00', 'A', 'u', 'g', '\x00', 'S', 'e', 'p',
|
974 |
|
|
'\x00', 'O', 'c', 't', '\x00', 'N', 'o', 'v',
|
975 |
|
|
'\x00', 'D', 'e', 'c', '\x00', 'J', 'a', 'n',
|
976 |
|
|
'u', 'a', 'r', 'y', '\x00', 'F', 'e', 'b',
|
977 |
|
|
'r', 'u', 'a', 'r', 'y', '\x00', 'M', 'a',
|
978 |
|
|
'r', 'c', 'h', '\x00', 'A', 'p', 'r', 'i',
|
979 |
|
|
'l', '\x00', 'M', 'a', 'y', '\x00', 'J', 'u',
|
980 |
|
|
'n', 'e', '\x00', 'J', 'u', 'l', 'y', '\x00',
|
981 |
|
|
'A', 'u', 'g', 'u', 's', 't', '\x00', 'S',
|
982 |
|
|
'e', 'p', 't', 'e', 'm', 'b', 'e', 'r',
|
983 |
|
|
'\x00', 'O', 'c', 't', 'o', 'b', 'e', 'r',
|
984 |
|
|
'\x00', 'N', 'o', 'v', 'e', 'm', 'b', 'e',
|
985 |
|
|
'r', '\x00', 'D', 'e', 'c', 'e', 'm', 'b',
|
986 |
|
|
'e', 'r', '\x00', 'A', 'M', '\x00', 'P', 'M',
|
987 |
|
|
'\x00', '%', 'a', ' ', '%', 'b', ' ', '%',
|
988 |
|
|
'e', ' ', '%', 'H', ':', '%', 'M', ':',
|
989 |
|
|
'%', 'S', ' ', '%', 'Y', '\x00', '%', 'm',
|
990 |
|
|
'/', '%', 'd', '/', '%', 'y', '\x00', '%',
|
991 |
|
|
'H', ':', '%', 'M', ':', '%', 'S', '\x00',
|
992 |
|
|
'%', 'I', ':', '%', 'M', ':', '%', 'S',
|
993 |
|
|
' ', '%', 'p', '\x00', '^', '[', 'y', 'Y',
|
994 |
|
|
']', '\x00', '^', '[', 'n', 'N', ']', '\x00',
|
995 |
|
|
};
|
996 |
|
|
|
997 |
|
|
char *nl_langinfo(nl_item item)
|
998 |
|
|
{
|
999 |
|
|
unsigned int c;
|
1000 |
|
|
unsigned int i;
|
1001 |
|
|
|
1002 |
|
|
if ((c = _NL_ITEM_CATEGORY(item)) < C_LC_ALL) {
|
1003 |
|
|
if ((i = cat_start[c] + _NL_ITEM_INDEX(item)) < cat_start[c+1]) {
|
1004 |
|
|
/* return (char *) C_locale_data + item_offset[i] + (i & 64); */
|
1005 |
|
|
return (char *) C_locale_data + nl_data[C_LC_ALL+1+i] + 2*(i & 64);
|
1006 |
|
|
}
|
1007 |
|
|
}
|
1008 |
|
|
return (char *) cat_start; /* Conveniently, this is the empty string. */
|
1009 |
|
|
}
|
1010 |
|
|
|
1011 |
|
|
#else /* __LOCALE_C_ONLY */
|
1012 |
|
|
|
1013 |
|
|
#if defined(__UCLIBC_HAS_XLOCALE__) && !defined(__UCLIBC_DO_XLOCALE)
|
1014 |
|
|
|
1015 |
|
|
char *nl_langinfo(nl_item item)
|
1016 |
|
|
{
|
1017 |
|
|
return __nl_langinfo_l(item, __UCLIBC_CURLOCALE);
|
1018 |
|
|
}
|
1019 |
|
|
|
1020 |
|
|
#else /* defined(__UCLIBC_HAS_XLOCALE__) && !defined(__UCLIBC_DO_XLOCALE) */
|
1021 |
|
|
|
1022 |
|
|
static const char empty[] = "";
|
1023 |
|
|
|
1024 |
|
|
char *__XL(nl_langinfo)(nl_item item __LOCALE_PARAM )
|
1025 |
|
|
{
|
1026 |
|
|
unsigned int c = _NL_ITEM_CATEGORY(item);
|
1027 |
|
|
unsigned int i = _NL_ITEM_INDEX(item);
|
1028 |
|
|
|
1029 |
|
|
if ((c < LC_ALL) && (i < __LOCALE_PTR->category_item_count[c])) {
|
1030 |
|
|
return ((char **)(((char *) __LOCALE_PTR)
|
1031 |
|
|
+ __LOCALE_PTR->category_offsets[c]))[i];
|
1032 |
|
|
}
|
1033 |
|
|
|
1034 |
|
|
return (char *) empty;
|
1035 |
|
|
}
|
1036 |
|
|
|
1037 |
|
|
#endif /* defined(__UCLIBC_HAS_XLOCALE__) && !defined(__UCLIBC_DO_XLOCALE) */
|
1038 |
|
|
|
1039 |
|
|
#endif /* __LOCALE_C_ONLY */
|
1040 |
|
|
|
1041 |
|
|
#endif
|
1042 |
|
|
/**********************************************************************/
|
1043 |
|
|
#ifdef L_newlocale
|
1044 |
|
|
|
1045 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
1046 |
|
|
#warning TODO: Move posix and utf8 strings.
|
1047 |
|
|
#endif
|
1048 |
|
|
static const char posix[] = "POSIX";
|
1049 |
|
|
static const char utf8[] = "UTF-8";
|
1050 |
|
|
|
1051 |
|
|
static int find_locale(int category_mask, const char *p,
|
1052 |
|
|
unsigned char *new_locale)
|
1053 |
|
|
{
|
1054 |
|
|
int i;
|
1055 |
|
|
const unsigned char *s;
|
1056 |
|
|
uint16_t n;
|
1057 |
|
|
unsigned char lang_cult, codeset;
|
1058 |
|
|
|
1059 |
|
|
#if defined(__LOCALE_DATA_AT_MODIFIERS_LENGTH) && 1
|
1060 |
|
|
/* Support standard locale handling for @-modifiers. */
|
1061 |
|
|
|
1062 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
1063 |
|
|
#warning REMINDER: Fix buf size in find_locale.
|
1064 |
|
|
#endif
|
1065 |
|
|
char buf[18]; /* TODO: 7+{max codeset name length} */
|
1066 |
|
|
const char *q;
|
1067 |
|
|
|
1068 |
|
|
if ((q = strchr(p,'@')) != NULL) {
|
1069 |
|
|
if ((((size_t)((q-p)-5)) > (sizeof(buf) - 5)) || (p[2] != '_')) {
|
1070 |
|
|
return 0;
|
1071 |
|
|
}
|
1072 |
|
|
/* locale name at least 5 chars long and 3rd char is '_' */
|
1073 |
|
|
s = LOCALE_AT_MODIFIERS;
|
1074 |
|
|
do {
|
1075 |
|
|
if (!strcmp(s+2, q+1)) {
|
1076 |
|
|
break;
|
1077 |
|
|
}
|
1078 |
|
|
s += 2 + *s; /* TODO - fix this throughout */
|
1079 |
|
|
} while (*s);
|
1080 |
|
|
if (!*s) {
|
1081 |
|
|
return 0;
|
1082 |
|
|
}
|
1083 |
|
|
assert(q - p < sizeof(buf));
|
1084 |
|
|
memcpy(buf, p, q-p);
|
1085 |
|
|
buf[q-p] = 0;
|
1086 |
|
|
buf[2] = s[1];
|
1087 |
|
|
p = buf;
|
1088 |
|
|
}
|
1089 |
|
|
#endif
|
1090 |
|
|
|
1091 |
|
|
lang_cult = codeset = 0; /* Assume C and default codeset. */
|
1092 |
|
|
if (((*p == 'C') && !p[1]) || !strcmp(p, posix)) {
|
1093 |
|
|
goto FIND_LOCALE;
|
1094 |
|
|
}
|
1095 |
|
|
|
1096 |
|
|
if ((strlen(p) > 5) && (p[5] == '.')) { /* Codeset in locale name? */
|
1097 |
|
|
/* TODO: maybe CODESET_LIST + *s ??? */
|
1098 |
|
|
/* 7bit is 1, UTF-8 is 2, 8-bit is >= 3 */
|
1099 |
|
|
codeset = 2;
|
1100 |
|
|
if (strcmp(utf8,p+6) != 0) {/* TODO - fix! */
|
1101 |
|
|
s = CODESET_LIST;
|
1102 |
|
|
do {
|
1103 |
|
|
++codeset; /* Increment codeset first. */
|
1104 |
|
|
if (!strcmp(CODESET_LIST+*s, p+6)) {
|
1105 |
|
|
goto FIND_LANG_CULT;
|
1106 |
|
|
}
|
1107 |
|
|
} while (*++s);
|
1108 |
|
|
return 0; /* No matching codeset! */
|
1109 |
|
|
}
|
1110 |
|
|
}
|
1111 |
|
|
|
1112 |
|
|
FIND_LANG_CULT: /* Find language_culture number. */
|
1113 |
|
|
s = LOCALE_NAMES;
|
1114 |
|
|
do { /* TODO -- do a binary search? */
|
1115 |
|
|
/* TODO -- fix gen_mmap!*/
|
1116 |
|
|
++lang_cult; /* Increment first since C/POSIX is 0. */
|
1117 |
|
|
if (!strncmp(s,p,5)) { /* Found a matching locale name; */
|
1118 |
|
|
goto FIND_LOCALE;
|
1119 |
|
|
}
|
1120 |
|
|
s += 5;
|
1121 |
|
|
} while (lang_cult < __LOCALE_DATA_NUM_LOCALE_NAMES);
|
1122 |
|
|
return 0; /* No matching language_culture! */
|
1123 |
|
|
|
1124 |
|
|
FIND_LOCALE: /* Find locale row matching name and codeset */
|
1125 |
|
|
s = LOCALES;
|
1126 |
|
|
n = 0;
|
1127 |
|
|
do { /* TODO -- do a binary search? */
|
1128 |
|
|
if ((lang_cult == *s) && ((codeset == s[1]) || (codeset == s[2]))) {
|
1129 |
|
|
i = 1;
|
1130 |
|
|
s = new_locale + 1;
|
1131 |
|
|
do {
|
1132 |
|
|
if (category_mask & i) {
|
1133 |
|
|
/* Encode current locale row number. */
|
1134 |
|
|
((unsigned char *) s)[0] = (n >> 7) | 0x80;
|
1135 |
|
|
((unsigned char *) s)[1] = (n & 0x7f) | 0x80;
|
1136 |
|
|
}
|
1137 |
|
|
s += 2;
|
1138 |
|
|
i += i;
|
1139 |
|
|
} while (i < (1 << LC_ALL));
|
1140 |
|
|
|
1141 |
|
|
return i; /* Return non-zero */
|
1142 |
|
|
}
|
1143 |
|
|
s += __LOCALE_DATA_WIDTH_LOCALES;
|
1144 |
|
|
++n;
|
1145 |
|
|
} while (n <= __LOCALE_DATA_NUM_LOCALES); /* We started at 1!!! */
|
1146 |
|
|
|
1147 |
|
|
return 0; /* Unsupported locale. */
|
1148 |
|
|
}
|
1149 |
|
|
|
1150 |
|
|
static unsigned char *composite_locale(int category_mask, const char *locale,
|
1151 |
|
|
unsigned char *new_locale)
|
1152 |
|
|
{
|
1153 |
|
|
char buf[MAX_LOCALE_STR];
|
1154 |
|
|
char *t;
|
1155 |
|
|
char *e;
|
1156 |
|
|
int c;
|
1157 |
|
|
int component_mask;
|
1158 |
|
|
|
1159 |
|
|
if (!strchr(locale,'=')) {
|
1160 |
|
|
if (!find_locale(category_mask, locale, new_locale)) {
|
1161 |
|
|
return NULL;
|
1162 |
|
|
}
|
1163 |
|
|
return new_locale;
|
1164 |
|
|
}
|
1165 |
|
|
|
1166 |
|
|
if (strlen(locale) >= sizeof(buf)) {
|
1167 |
|
|
return NULL;
|
1168 |
|
|
}
|
1169 |
|
|
stpcpy(buf, locale);
|
1170 |
|
|
|
1171 |
|
|
component_mask = 0;
|
1172 |
|
|
t = strtok_r(buf, "=", &e); /* This can't fail because of strchr test above. */
|
1173 |
|
|
do {
|
1174 |
|
|
c = 0;
|
1175 |
|
|
while (strcmp(CATEGORY_NAMES + (int) CATEGORY_NAMES[c], t)) {
|
1176 |
|
|
if (++c == LC_ALL) { /* Unknown category name! */
|
1177 |
|
|
return NULL;
|
1178 |
|
|
}
|
1179 |
|
|
}
|
1180 |
|
|
t = strtok_r(NULL, ";", &e);
|
1181 |
|
|
c = (1 << c);
|
1182 |
|
|
if (component_mask & c) { /* Multiple components for one category. */
|
1183 |
|
|
return NULL;
|
1184 |
|
|
}
|
1185 |
|
|
component_mask |= c;
|
1186 |
|
|
if ((category_mask & c) && (!t || !find_locale(c, t, new_locale))) {
|
1187 |
|
|
return NULL;
|
1188 |
|
|
}
|
1189 |
|
|
} while ((t = strtok_r(NULL, "=", &e)) != NULL);
|
1190 |
|
|
|
1191 |
|
|
if (category_mask & ~component_mask) { /* Category component(s) missing. */
|
1192 |
|
|
return NULL;
|
1193 |
|
|
}
|
1194 |
|
|
|
1195 |
|
|
return new_locale;
|
1196 |
|
|
}
|
1197 |
|
|
|
1198 |
|
|
__locale_t __newlocale(int category_mask, const char *locale, __locale_t base)
|
1199 |
|
|
{
|
1200 |
|
|
const unsigned char *p;
|
1201 |
|
|
int i, j, k;
|
1202 |
|
|
unsigned char new_selector[LOCALE_SELECTOR_SIZE];
|
1203 |
|
|
|
1204 |
|
|
if (category_mask == (1 << LC_ALL)) {
|
1205 |
|
|
category_mask = LC_ALL_MASK;
|
1206 |
|
|
}
|
1207 |
|
|
|
1208 |
|
|
if (!locale || (((unsigned int)(category_mask)) > LC_ALL_MASK)) {
|
1209 |
|
|
INVALID:
|
1210 |
|
|
__set_errno(EINVAL);
|
1211 |
|
|
return NULL; /* No locale or illegal/unsupported category. */
|
1212 |
|
|
}
|
1213 |
|
|
|
1214 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
1215 |
|
|
#warning TODO: Rename cur_locale to locale_selector.
|
1216 |
|
|
#endif
|
1217 |
|
|
strcpy((char *) new_selector,
|
1218 |
|
|
(base ? (char *) base->cur_locale : C_LOCALE_SELECTOR));
|
1219 |
|
|
|
1220 |
|
|
if (!*locale) { /* locale == "", so check environment. */
|
1221 |
|
|
#ifndef __UCLIBC_HAS_THREADS__
|
1222 |
|
|
static /* If no threads, then envstr can be static. */
|
1223 |
|
|
#endif /* __UCLIBC_HAS_THREADS__ */
|
1224 |
|
|
const char *envstr[4] = { "LC_ALL", NULL, "LANG", posix };
|
1225 |
|
|
|
1226 |
|
|
i = 1;
|
1227 |
|
|
k = 0;
|
1228 |
|
|
do {
|
1229 |
|
|
if (category_mask & i) {
|
1230 |
|
|
/* Note: SUSv3 doesn't define a fallback mechanism here.
|
1231 |
|
|
* So, if LC_ALL is invalid, we do _not_ continue trying
|
1232 |
|
|
* the other environment vars. */
|
1233 |
|
|
envstr[1] = CATEGORY_NAMES + CATEGORY_NAMES[k];
|
1234 |
|
|
j = 0;
|
1235 |
|
|
do {
|
1236 |
|
|
p = envstr[j];
|
1237 |
|
|
} while ((++j < 4) && (!(p = getenv(p)) || !*p));
|
1238 |
|
|
|
1239 |
|
|
|
1240 |
|
|
/* The user set something... is it valid? */
|
1241 |
|
|
/* Note: Since we don't support user-supplied locales and
|
1242 |
|
|
* alternate paths, we don't need to worry about special
|
1243 |
|
|
* handling for suid/sgid apps. */
|
1244 |
|
|
if (!find_locale(i, p, new_selector)) {
|
1245 |
|
|
goto INVALID;
|
1246 |
|
|
}
|
1247 |
|
|
}
|
1248 |
|
|
i += i;
|
1249 |
|
|
} while (++k < LC_ALL);
|
1250 |
|
|
} else if (!composite_locale(category_mask, locale, new_selector)) {
|
1251 |
|
|
goto INVALID;
|
1252 |
|
|
}
|
1253 |
|
|
|
1254 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
1255 |
|
|
#warning TODO: Do a compatible codeset check!
|
1256 |
|
|
#endif
|
1257 |
|
|
|
1258 |
|
|
/* If we get here, the new selector corresponds to a valid locale. */
|
1259 |
|
|
|
1260 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
1261 |
|
|
#warning CONSIDER: Probably want a _locale_new func to allow for caching of locales.
|
1262 |
|
|
#endif
|
1263 |
|
|
#if 0
|
1264 |
|
|
if (base) {
|
1265 |
|
|
_locale_set_l(new_selector, base);
|
1266 |
|
|
} else {
|
1267 |
|
|
base = _locale_new(new_selector);
|
1268 |
|
|
}
|
1269 |
|
|
#else
|
1270 |
|
|
if (!base) {
|
1271 |
|
|
if ((base = malloc(sizeof(__uclibc_locale_t))) == NULL) {
|
1272 |
|
|
return base;
|
1273 |
|
|
}
|
1274 |
|
|
_locale_init_l(base);
|
1275 |
|
|
}
|
1276 |
|
|
|
1277 |
|
|
_locale_set_l(new_selector, base);
|
1278 |
|
|
#endif
|
1279 |
|
|
|
1280 |
|
|
return base;
|
1281 |
|
|
}
|
1282 |
|
|
|
1283 |
|
|
weak_alias(__newlocale, newlocale)
|
1284 |
|
|
|
1285 |
|
|
#endif
|
1286 |
|
|
/**********************************************************************/
|
1287 |
|
|
#ifdef L_duplocale
|
1288 |
|
|
|
1289 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
1290 |
|
|
#warning REMINDER: When we allocate ctype tables, remember to dup them.
|
1291 |
|
|
#endif
|
1292 |
|
|
|
1293 |
|
|
__locale_t __duplocale(__locale_t dataset)
|
1294 |
|
|
{
|
1295 |
|
|
__locale_t r;
|
1296 |
|
|
uint16_t * i2w;
|
1297 |
|
|
size_t n;
|
1298 |
|
|
|
1299 |
|
|
assert(dataset != LC_GLOBAL_LOCALE);
|
1300 |
|
|
|
1301 |
|
|
if ((r = malloc(sizeof(__uclibc_locale_t))) != NULL) {
|
1302 |
|
|
n = 2*dataset->collate.max_col_index+2;
|
1303 |
|
|
if ((i2w = calloc(n, sizeof(uint16_t)))
|
1304 |
|
|
!= NULL
|
1305 |
|
|
) {
|
1306 |
|
|
memcpy(r, dataset, sizeof(__uclibc_locale_t));
|
1307 |
|
|
r->collate.index2weight = i2w;
|
1308 |
|
|
memcpy(i2w, dataset->collate.index2weight, n * sizeof(uint16_t));
|
1309 |
|
|
} else {
|
1310 |
|
|
free(r);
|
1311 |
|
|
r = NULL;
|
1312 |
|
|
}
|
1313 |
|
|
}
|
1314 |
|
|
return r;
|
1315 |
|
|
}
|
1316 |
|
|
|
1317 |
|
|
weak_alias(__duplocale, duplocale)
|
1318 |
|
|
|
1319 |
|
|
#endif
|
1320 |
|
|
/**********************************************************************/
|
1321 |
|
|
#ifdef L_freelocale
|
1322 |
|
|
|
1323 |
|
|
#ifdef __UCLIBC_MJN3_ONLY__
|
1324 |
|
|
#warning REMINDER: When we allocate ctype tables, remember to free them.
|
1325 |
|
|
#endif
|
1326 |
|
|
|
1327 |
|
|
void __freelocale(__locale_t dataset)
|
1328 |
|
|
{
|
1329 |
|
|
assert(dataset != __global_locale);
|
1330 |
|
|
assert(dataset != LC_GLOBAL_LOCALE);
|
1331 |
|
|
|
1332 |
|
|
free(dataset->collate.index2weight); /* Free collation data. */
|
1333 |
|
|
free(dataset); /* Free locale */
|
1334 |
|
|
}
|
1335 |
|
|
|
1336 |
|
|
weak_alias(__freelocale, freelocale)
|
1337 |
|
|
|
1338 |
|
|
#endif
|
1339 |
|
|
/**********************************************************************/
|
1340 |
|
|
#ifdef L_uselocale
|
1341 |
|
|
|
1342 |
|
|
__locale_t __uselocale(__locale_t dataset)
|
1343 |
|
|
{
|
1344 |
|
|
__locale_t old;
|
1345 |
|
|
|
1346 |
|
|
if (!dataset) {
|
1347 |
|
|
old = __UCLIBC_CURLOCALE;
|
1348 |
|
|
} else {
|
1349 |
|
|
if (dataset == LC_GLOBAL_LOCALE) {
|
1350 |
|
|
dataset = __global_locale;
|
1351 |
|
|
}
|
1352 |
|
|
#ifdef __UCLIBC_HAS_THREADS__
|
1353 |
|
|
old = __curlocale_set(dataset);
|
1354 |
|
|
#else
|
1355 |
|
|
old = __curlocale_var;
|
1356 |
|
|
__curlocale_var = dataset;
|
1357 |
|
|
#endif
|
1358 |
|
|
}
|
1359 |
|
|
|
1360 |
|
|
if (old == __global_locale) {
|
1361 |
|
|
return LC_GLOBAL_LOCALE;
|
1362 |
|
|
}
|
1363 |
|
|
return old;
|
1364 |
|
|
}
|
1365 |
|
|
|
1366 |
|
|
weak_alias(__uselocale, uselocale)
|
1367 |
|
|
|
1368 |
|
|
#endif
|
1369 |
|
|
/**********************************************************************/
|
1370 |
|
|
#ifdef L___curlocale
|
1371 |
|
|
|
1372 |
|
|
#ifdef __UCLIBC_HAS_THREADS__
|
1373 |
|
|
|
1374 |
|
|
__locale_t weak_const_function __curlocale(void)
|
1375 |
|
|
{
|
1376 |
|
|
return __curlocale_var; /* This is overriden by the thread version. */
|
1377 |
|
|
}
|
1378 |
|
|
|
1379 |
|
|
__locale_t weak_function __curlocale_set(__locale_t newloc)
|
1380 |
|
|
{
|
1381 |
|
|
__locale_t oldloc = __curlocale_var;
|
1382 |
|
|
assert(newloc != LC_GLOBAL_LOCALE);
|
1383 |
|
|
__curlocale_var = newloc;
|
1384 |
|
|
return oldloc;
|
1385 |
|
|
}
|
1386 |
|
|
|
1387 |
|
|
#endif
|
1388 |
|
|
|
1389 |
|
|
#endif
|
1390 |
|
|
/**********************************************************************/
|
1391 |
|
|
#ifdef L___locale_mbrtowc_l
|
1392 |
|
|
|
1393 |
|
|
/* NOTE: This returns an int... not size_t. Also, it is not a general
|
1394 |
|
|
* routine. It is actually a very stripped-down version of mbrtowc
|
1395 |
|
|
* that takes a __locale_t arg. This is used by strcoll and strxfrm.
|
1396 |
|
|
* It is also used above to generate wchar_t versions of the decimal point
|
1397 |
|
|
* and thousands seperator. */
|
1398 |
|
|
|
1399 |
|
|
|
1400 |
|
|
#ifndef __CTYPE_HAS_UTF_8_LOCALES
|
1401 |
|
|
#warning __CTYPE_HAS_UTF_8_LOCALES not set!
|
1402 |
|
|
#endif
|
1403 |
|
|
#ifndef __CTYPE_HAS_8_BIT_LOCALES
|
1404 |
|
|
#warning __CTYPE_HAS_8_BIT_LOCALES not set!
|
1405 |
|
|
#endif
|
1406 |
|
|
|
1407 |
|
|
#define Cc2wc_IDX_SHIFT __LOCALE_DATA_Cc2wc_IDX_SHIFT
|
1408 |
|
|
#define Cc2wc_ROW_LEN __LOCALE_DATA_Cc2wc_ROW_LEN
|
1409 |
|
|
|
1410 |
|
|
extern size_t _wchar_utf8sntowcs(wchar_t *__restrict pwc, size_t wn,
|
1411 |
|
|
const char **__restrict src, size_t n,
|
1412 |
|
|
mbstate_t *ps, int allow_continuation);
|
1413 |
|
|
|
1414 |
|
|
int __locale_mbrtowc_l(wchar_t *__restrict dst,
|
1415 |
|
|
const char *__restrict src,
|
1416 |
|
|
__locale_t loc )
|
1417 |
|
|
{
|
1418 |
|
|
#ifdef __CTYPE_HAS_UTF_8_LOCALES
|
1419 |
|
|
if (loc->encoding == __ctype_encoding_utf8) {
|
1420 |
|
|
mbstate_t ps;
|
1421 |
|
|
const char *p = src;
|
1422 |
|
|
size_t r;
|
1423 |
|
|
ps.mask = 0;
|
1424 |
|
|
r = _wchar_utf8sntowcs(dst, 1, &p, SIZE_MAX, &ps, 1);
|
1425 |
|
|
return (r == 1) ? (p-src) : r; /* Need to return 0 if nul char. */
|
1426 |
|
|
}
|
1427 |
|
|
#endif
|
1428 |
|
|
|
1429 |
|
|
#ifdef __CTYPE_HAS_8_BIT_LOCALES
|
1430 |
|
|
assert((loc->encoding == __ctype_encoding_7_bit) || (loc->encoding == __ctype_encoding_8_bit));
|
1431 |
|
|
#else
|
1432 |
|
|
assert(loc->encoding == __ctype_encoding_7_bit);
|
1433 |
|
|
#endif
|
1434 |
|
|
|
1435 |
|
|
if ((*dst = ((unsigned char)(*src))) < 0x80) { /* ASCII... */
|
1436 |
|
|
return (*src != 0);
|
1437 |
|
|
}
|
1438 |
|
|
|
1439 |
|
|
#ifdef __CTYPE_HAS_8_BIT_LOCALES
|
1440 |
|
|
if (loc->encoding == __ctype_encoding_8_bit) {
|
1441 |
|
|
wchar_t wc = *dst - 0x80;
|
1442 |
|
|
*dst = loc->tbl8c2wc[
|
1443 |
|
|
(loc->idx8c2wc[wc >> Cc2wc_IDX_SHIFT]
|
1444 |
|
|
<< Cc2wc_IDX_SHIFT) + (wc & (Cc2wc_ROW_LEN - 1))];
|
1445 |
|
|
if (*dst) {
|
1446 |
|
|
return 1;
|
1447 |
|
|
}
|
1448 |
|
|
}
|
1449 |
|
|
#endif
|
1450 |
|
|
|
1451 |
|
|
return -1;
|
1452 |
|
|
}
|
1453 |
|
|
|
1454 |
|
|
#endif
|
1455 |
|
|
/**********************************************************************/
|