locale.c 40 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387
  1. /* 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, see
  15. * <http://www.gnu.org/licenses/>.
  16. */
  17. /* Nov. 1, 2002
  18. * Reworked setlocale() return values and locale arg processing to
  19. * be more like glibc. Applications expecting to be able to
  20. * query locale settings should now work... at the cost of almost
  21. * doubling the size of the setlocale object code.
  22. * Fixed a bug in the internal fixed-size-string locale specifier code.
  23. *
  24. * Dec 20, 2002
  25. * Added in collation support and updated stub nl_langinfo.
  26. *
  27. * Aug 1, 2003
  28. * Added glibc-like extended locale stuff (newlocale, duplocale, etc).
  29. *
  30. * Aug 18, 2003
  31. * Bug in duplocale... collation data wasn't copied.
  32. * Bug in newlocale... translate 1<<LC_ALL to LC_ALL_MASK.
  33. * Bug in _wchar_utf8sntowcs... fix cut-n-paste error.
  34. *
  35. * Aug 31, 2003
  36. * Hack around bg_BG bug; grouping specified but no thousands separator.
  37. * Also, disable the locale link_warnings for now, as they generate a
  38. * lot of noise when using libstd++.
  39. */
  40. /* TODO:
  41. * Implement the shared mmap code so non-mmu platforms can use this.
  42. * Add some basic collate functionality similar to what the previous
  43. * locale support had (8-bit codesets only).
  44. */
  45. #define __CTYPE_HAS_8_BIT_LOCALES 1
  46. #include <string.h>
  47. #include <stdlib.h>
  48. #include <stddef.h>
  49. #include <limits.h>
  50. #include <stdint.h>
  51. #include <assert.h>
  52. #include <errno.h>
  53. #include <ctype.h>
  54. #include <stdio.h>
  55. #undef __LOCALE_C_ONLY
  56. #ifndef __UCLIBC_HAS_LOCALE__
  57. #define __LOCALE_C_ONLY
  58. #endif /* __UCLIBC_HAS_LOCALE__ */
  59. #ifdef __LOCALE_C_ONLY
  60. #include <locale.h>
  61. #else /* __LOCALE_C_ONLY */
  62. /* Need to include this before locale.h! */
  63. #include <bits/uClibc_locale.h>
  64. #undef CODESET_LIST
  65. #define CODESET_LIST (__locale_mmap->codeset_list)
  66. #ifdef __UCLIBC_HAS_XLOCALE__
  67. #include <locale.h>
  68. #else /* __UCLIBC_HAS_XLOCALE__ */
  69. /* We need this internally... */
  70. #define __UCLIBC_HAS_XLOCALE__ 1
  71. #include <locale.h>
  72. #undef __UCLIBC_HAS_XLOCALE__
  73. #endif /* __UCLIBC_HAS_XLOCALE__ */
  74. #include <wchar.h>
  75. #define LOCALE_NAMES (__locale_mmap->locale_names5)
  76. #define LOCALES (__locale_mmap->locales)
  77. #define LOCALE_AT_MODIFIERS (__locale_mmap->locale_at_modifiers)
  78. #define CATEGORY_NAMES (__locale_mmap->lc_names)
  79. #define MAX_LOCALE_STR 256 /* TODO: Only sufficient for current case. */
  80. #define MAX_LOCALE_CATEGORY_STR 32 /* TODO: Only sufficient for current case. */
  81. /* Note: Best if MAX_LOCALE_CATEGORY_STR is a power of 2. */
  82. extern int _locale_set_l(const unsigned char *p, __locale_t base) attribute_hidden;
  83. extern void _locale_init_l(__locale_t base) attribute_hidden;
  84. #endif /* __LOCALE_C_ONLY */
  85. #undef LOCALE_STRING_SIZE
  86. #define LOCALE_SELECTOR_SIZE (2 * __LC_ALL + 2)
  87. #define C_LOCALE_SELECTOR "\x23\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80"
  88. #include <langinfo.h>
  89. #include <nl_types.h>
  90. /**********************************************************************/
  91. #ifdef L_setlocale
  92. #ifdef __LOCALE_C_ONLY
  93. static const char C_string[] = "C";
  94. char *setlocale(int category, register const char *locale)
  95. {
  96. return ( (((unsigned int)(category)) <= LC_ALL)
  97. && ( (!locale) /* Request for locale category string. */
  98. || (!*locale) /* Implementation-defined default is C. */
  99. || ((*locale == 'C') && !locale[1])
  100. || (!strcmp(locale, "POSIX"))) )
  101. ? (char *) C_string /* Always in C/POSIX locale. */
  102. : NULL;
  103. }
  104. #else /* ---------------------------------------------- __LOCALE_C_ONLY */
  105. #if !defined(__LOCALE_DATA_NUM_LOCALES) || (__LOCALE_DATA_NUM_LOCALES <= 1)
  106. #error locales enabled, but not data other than for C locale!
  107. #endif
  108. static const char posix[] = "POSIX";
  109. static const char utf8[] = "UTF-8";
  110. /* Individual category strings start at hr_locale + category * MAX_LOCALE_CATEGORY.
  111. * This holds for LC_ALL as well.
  112. */
  113. static char hr_locale[(MAX_LOCALE_CATEGORY_STR * LC_ALL) + MAX_LOCALE_STR];
  114. static void update_hr_locale(const unsigned char *spec)
  115. {
  116. const unsigned char *loc;
  117. const unsigned char *s;
  118. char *n;
  119. int i, category, done;
  120. done = category = 0;
  121. do {
  122. s = spec + 1;
  123. n = hr_locale + category * MAX_LOCALE_CATEGORY_STR;
  124. if (category == LC_ALL) {
  125. done = 1;
  126. for (i = 0 ; i < LC_ALL-1 ; i += 2) {
  127. if ((s[i] != s[i+2]) || (s[i+1] != s[i+3])) {
  128. goto SKIP;
  129. }
  130. }
  131. /* All categories the same, so simplify string by using a single
  132. * category. */
  133. category = LC_CTYPE;
  134. }
  135. SKIP:
  136. i = (category == LC_ALL) ? 0 : category;
  137. s += 2*i;
  138. do {
  139. if ((*s != 0xff) || (s[1] != 0xff)) {
  140. loc = LOCALES
  141. + __LOCALE_DATA_WIDTH_LOCALES * ((((int)(*s & 0x7f)) << 7)
  142. + (s[1] & 0x7f));
  143. if (category == LC_ALL) {
  144. /* CATEGORY_NAMES is unsigned char* */
  145. n = stpcpy(n, (char*) CATEGORY_NAMES + (int) CATEGORY_NAMES[i]);
  146. *n++ = '=';
  147. }
  148. if (*loc == 0) {
  149. *n++ = 'C';
  150. *n = 0;
  151. } else {
  152. char at = 0;
  153. memcpy(n, LOCALE_NAMES + 5*((*loc)-1), 5);
  154. if (n[2] != '_') {
  155. at = n[2];
  156. n[2] = '_';
  157. }
  158. n += 5;
  159. *n++ = '.';
  160. if (loc[2] == 2) {
  161. n = stpcpy(n, utf8);
  162. } else if (loc[2] >= 3) {
  163. n = stpcpy(n, (char*) CODESET_LIST + (int)(CODESET_LIST[loc[2] - 3]));
  164. }
  165. if (at) {
  166. const char *q;
  167. *n++ = '@';
  168. q = (char*) LOCALE_AT_MODIFIERS;
  169. do {
  170. if (q[1] == at) {
  171. n = stpcpy(n, q+2);
  172. break;
  173. }
  174. q += 2 + *q;
  175. } while (*q);
  176. }
  177. }
  178. *n++ = ';';
  179. }
  180. s += 2;
  181. } while (++i < category);
  182. *--n = 0; /* Remove trailing ';' and nul-terminate. */
  183. ++category;
  184. } while (!done);
  185. }
  186. char *setlocale(int category, const char *locale)
  187. {
  188. if (((unsigned int)(category)) > LC_ALL) {
  189. #if 0
  190. __set_errno(EINVAL); /* glibc sets errno -- SUSv3 doesn't say. */
  191. #endif
  192. return NULL; /* Illegal/unsupported category. */
  193. }
  194. if (locale != NULL) { /* Not just a query... */
  195. if (!newlocale((1 << category), locale, __global_locale)) {
  196. return NULL; /* Failed! */
  197. }
  198. update_hr_locale(__global_locale->cur_locale);
  199. }
  200. /* Either a query or a successful set, so return current locale string. */
  201. return hr_locale + (category * MAX_LOCALE_CATEGORY_STR);
  202. }
  203. #endif /* __LOCALE_C_ONLY */
  204. #endif
  205. /**********************************************************************/
  206. #ifdef L_localeconv
  207. /* Note: We assume here that the compiler does the sane thing regarding
  208. * placement of the fields in the struct. If necessary, we could ensure
  209. * this usings an array of offsets but at some size cost. */
  210. #ifdef __LOCALE_C_ONLY
  211. static struct lconv the_lconv;
  212. static const char decpt[] = ".";
  213. struct lconv *localeconv(void)
  214. {
  215. register char *p = (char *)(&the_lconv);
  216. *((char **)p) = (char *) decpt;
  217. do {
  218. p += sizeof(char **);
  219. *((char **)p) = (char *) (decpt+1);
  220. } while (p < (char *) &the_lconv.negative_sign);
  221. p = (&the_lconv.int_frac_digits);
  222. do {
  223. *p = CHAR_MAX;
  224. ++p;
  225. } while (p <= &the_lconv.int_n_sign_posn);
  226. return &the_lconv;
  227. }
  228. #else /* __LOCALE_C_ONLY */
  229. static struct lconv the_lconv;
  230. struct lconv *localeconv(void)
  231. {
  232. register char *p = (char *) &the_lconv;
  233. register char **q = (char **) &(__UCLIBC_CURLOCALE->decimal_point);
  234. do {
  235. *((char **)p) = *q;
  236. p += sizeof(char **);
  237. ++q;
  238. } while (p < &the_lconv.int_frac_digits);
  239. do {
  240. *p = **q;
  241. ++p;
  242. ++q;
  243. } while (p <= &the_lconv.int_n_sign_posn);
  244. return &the_lconv;
  245. }
  246. #endif /* __LOCALE_C_ONLY */
  247. libc_hidden_def(localeconv)
  248. #endif
  249. /**********************************************************************/
  250. #if defined(L__locale_init) && !defined(__LOCALE_C_ONLY)
  251. struct __uclibc_locale_struct __global_locale_data;
  252. __locale_t __global_locale = &__global_locale_data;
  253. #ifdef __UCLIBC_HAS_XLOCALE__
  254. __locale_t __curlocale_var = &__global_locale_data;
  255. #endif
  256. /*----------------------------------------------------------------------*/
  257. static const char utf8[] = "UTF-8";
  258. static const char ascii[] = "ASCII";
  259. typedef struct {
  260. uint16_t num_base;
  261. uint16_t num_der;
  262. uint16_t MAX_WEIGHTS;
  263. uint16_t num_index2weight;
  264. #define num_index2ruleidx num_index2weight
  265. uint16_t num_weightstr;
  266. uint16_t num_multistart;
  267. uint16_t num_override;
  268. uint16_t num_ruletable;
  269. } coldata_header_t;
  270. typedef struct {
  271. uint16_t num_weights;
  272. uint16_t num_starters;
  273. uint16_t ii_shift;
  274. uint16_t ti_shift;
  275. uint16_t ii_len;
  276. uint16_t ti_len;
  277. uint16_t max_weight;
  278. uint16_t num_col_base;
  279. uint16_t max_col_index;
  280. uint16_t undefined_idx;
  281. uint16_t range_low;
  282. uint16_t range_count;
  283. uint16_t range_base_weight;
  284. uint16_t range_rule_offset;
  285. uint16_t index2weight_offset;
  286. uint16_t index2ruleidx_offset;
  287. uint16_t multistart_offset;
  288. uint16_t wcs2colidt_offset_low;
  289. uint16_t wcs2colidt_offset_hi;
  290. } coldata_base_t;
  291. typedef struct {
  292. uint16_t base_idx;
  293. uint16_t undefined_idx;
  294. uint16_t overrides_offset;
  295. uint16_t multistart_offset;
  296. } coldata_der_t;
  297. static int init_cur_collate(int der_num, __collate_t *cur_collate)
  298. {
  299. const uint16_t *__locale_collate_tbl = __locale_mmap->collate_data;
  300. coldata_header_t *cdh;
  301. coldata_base_t *cdb;
  302. coldata_der_t *cdd;
  303. const uint16_t *p;
  304. size_t n;
  305. uint16_t i, w;
  306. #if 0
  307. assert(sizeof(coldata_base_t) == 19*2);
  308. assert(sizeof(coldata_der_t) == 4*2);
  309. assert(sizeof(coldata_header_t) == 8*2);
  310. #endif
  311. if (!der_num) { /* C locale... special */
  312. cur_collate->num_weights = 0;
  313. return 1;
  314. }
  315. --der_num;
  316. cdh = (coldata_header_t *) __locale_collate_tbl;
  317. #if 0
  318. if (der_num >= cdh->num_der) {
  319. return 0;
  320. }
  321. #else
  322. assert((der_num < cdh->num_der));
  323. #endif
  324. cdd = (coldata_der_t *)(__locale_collate_tbl
  325. + (sizeof(coldata_header_t)
  326. + cdh->num_base * sizeof(coldata_base_t)
  327. + der_num * sizeof(coldata_der_t)
  328. )/2 );
  329. cdb = (coldata_base_t *)(__locale_collate_tbl
  330. + (sizeof(coldata_header_t)
  331. + cdd->base_idx * sizeof(coldata_base_t)
  332. )/2 );
  333. memcpy(cur_collate, cdb, offsetof(coldata_base_t,index2weight_offset));
  334. cur_collate->undefined_idx = cdd->undefined_idx;
  335. cur_collate->ti_mask = (1 << cur_collate->ti_shift)-1;
  336. cur_collate->ii_mask = (1 << cur_collate->ii_shift)-1;
  337. /* fflush(stdout); */
  338. /* fprintf(stderr,"base=%d num_col_base: %d %d\n", cdd->base_idx ,cur_collate->num_col_base, cdb->num_col_base); */
  339. n = (sizeof(coldata_header_t) + cdh->num_base * sizeof(coldata_base_t)
  340. + cdh->num_der * sizeof(coldata_der_t))/2;
  341. /* fprintf(stderr,"n = %d\n", n); */
  342. cur_collate->index2weight_tbl = __locale_collate_tbl + n + cdb->index2weight_offset;
  343. /* fprintf(stderr,"i2w = %d\n", n + cdb->index2weight_offset); */
  344. n += cdh->num_index2weight;
  345. cur_collate->index2ruleidx_tbl = __locale_collate_tbl + n + cdb->index2ruleidx_offset;
  346. /* fprintf(stderr,"i2r = %d\n", n + cdb->index2ruleidx_offset); */
  347. n += cdh->num_index2ruleidx;
  348. cur_collate->multistart_tbl = __locale_collate_tbl + n + cdd->multistart_offset;
  349. /* fprintf(stderr,"mts = %d\n", n + cdb->multistart_offset); */
  350. n += cdh->num_multistart;
  351. cur_collate->overrides_tbl = __locale_collate_tbl + n + cdd->overrides_offset;
  352. /* fprintf(stderr,"ovr = %d\n", n + cdd->overrides_offset); */
  353. n += cdh->num_override;
  354. cur_collate->ruletable = __locale_collate_tbl + n;
  355. /* fprintf(stderr, "rtb = %d\n", n); */
  356. n += cdh->num_ruletable;
  357. cur_collate->weightstr = __locale_collate_tbl + n;
  358. /* fprintf(stderr,"wts = %d\n", n); */
  359. n += cdh->num_weightstr;
  360. cur_collate->wcs2colidt_tbl = __locale_collate_tbl + n
  361. + (((unsigned long)(cdb->wcs2colidt_offset_hi)) << 16)
  362. + cdb->wcs2colidt_offset_low;
  363. /* fprintf(stderr,"wcs = %lu\n", n + (((unsigned long)(cdb->wcs2colidt_offset_hi)) << 16) */
  364. /* + cdb->wcs2colidt_offset_low); */
  365. cur_collate->MAX_WEIGHTS = cdh->MAX_WEIGHTS;
  366. cur_collate->index2weight = calloc(2*cur_collate->max_col_index+2,
  367. sizeof(uint16_t));
  368. if (!cur_collate->index2weight) {
  369. return 0;
  370. }
  371. cur_collate->index2ruleidx = cur_collate->index2weight
  372. + cur_collate->max_col_index + 1;
  373. memcpy(cur_collate->index2weight, cur_collate->index2weight_tbl,
  374. cur_collate->num_col_base * sizeof(uint16_t));
  375. memcpy(cur_collate->index2ruleidx, cur_collate->index2ruleidx_tbl,
  376. cur_collate->num_col_base * sizeof(uint16_t));
  377. /* now do the overrides */
  378. p = cur_collate->overrides_tbl;
  379. while (*p > 1) {
  380. /* fprintf(stderr, "processing override -- count = %d\n", *p); */
  381. n = *p++;
  382. w = *p++;
  383. do {
  384. i = *p++;
  385. /* fprintf(stderr, " i=%d (%#x) w=%d *p=%d\n", i, i, w, *p); */
  386. cur_collate->index2weight[i-1] = w++;
  387. cur_collate->index2ruleidx[i-1] = *p++;
  388. } while (--n);
  389. }
  390. assert(*p == 1);
  391. while (*++p) {
  392. i = *p;
  393. /* fprintf(stderr, " i=%d (%#x) w=%d *p=%d\n", i, i, p[1], p[2]); */
  394. cur_collate->index2weight[i-1] = *++p;
  395. cur_collate->index2ruleidx[i-1] = *++p;
  396. }
  397. for (i=0 ; i < cur_collate->multistart_tbl[0] ; i++) {
  398. p = cur_collate->multistart_tbl;
  399. /* fprintf(stderr, "%2d of %2d: %d ", i, cur_collate->multistart_tbl[0], p[i]); */
  400. p += p[i];
  401. do {
  402. n = *p++;
  403. do {
  404. if (!*p) { /* found it */
  405. /* fprintf(stderr, "found: n=%d (%#lx) |%.*ls|\n", n, (int) *cs->s, n, cs->s); */
  406. /* fprintf(stderr, ": %d - single\n", n); */
  407. goto FOUND;
  408. }
  409. /* the lookup check here is safe since we're assured that *p is a valid colidex */
  410. /* fprintf(stderr, "lookup(%lc)==%d *p==%d\n", cs->s[n], lookup(cs->s[n]), (int) *p); */
  411. /* fprintf(stderr, ": %d - ", n); */
  412. do {
  413. /* fprintf(stderr, "%d|", *p); */
  414. } while (*p++);
  415. break;
  416. } while (1);
  417. } while (1);
  418. FOUND:
  419. continue;
  420. }
  421. return 1;
  422. }
  423. int attribute_hidden _locale_set_l(const unsigned char *p, __locale_t base)
  424. {
  425. const char **x;
  426. unsigned char *s = base->cur_locale + 1;
  427. const size_t *stp;
  428. const unsigned char *r;
  429. const uint16_t *io;
  430. const uint16_t *ii;
  431. const unsigned char *d;
  432. int row; /* locale row */
  433. int crow; /* category row */
  434. int len;
  435. int c;
  436. int i = 0;
  437. __collate_t newcol;
  438. ++p;
  439. newcol.index2weight = NULL;
  440. if ((p[2*LC_COLLATE] != s[2*LC_COLLATE])
  441. || (p[2*LC_COLLATE + 1] != s[2*LC_COLLATE + 1])
  442. ) {
  443. row = (((int)(*p & 0x7f)) << 7) + (p[1] & 0x7f);
  444. assert(row < __LOCALE_DATA_NUM_LOCALES);
  445. if (!init_cur_collate(__locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES
  446. * row + 3 + LC_COLLATE ],
  447. &newcol)
  448. ) {
  449. return 0; /* calloc failed. */
  450. }
  451. free(base->collate.index2weight);
  452. memcpy(&base->collate, &newcol, sizeof(__collate_t));
  453. }
  454. do {
  455. if ((*p != *s) || (p[1] != s[1])) {
  456. row = (((int)(*p & 0x7f)) << 7) + (p[1] & 0x7f);
  457. assert(row < __LOCALE_DATA_NUM_LOCALES);
  458. *s = *p;
  459. s[1] = p[1];
  460. if ((i != LC_COLLATE)
  461. && ((len = __locale_mmap->lc_common_item_offsets_LEN[i]) != 0)
  462. ) {
  463. crow = __locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES * row
  464. + 3 + i ]
  465. * len;
  466. x = (const char **)(((char *) base)
  467. + base->category_offsets[i]);
  468. stp = __locale_mmap->lc_common_tbl_offsets + 4*i;
  469. r = (const unsigned char *)( ((char *)__locale_mmap) + *stp );
  470. io = (const uint16_t *)( ((char *)__locale_mmap) + *++stp );
  471. ii = (const uint16_t *)( ((char *)__locale_mmap) + *++stp );
  472. d = (const unsigned char *)( ((char *)__locale_mmap) + *++stp );
  473. for (c = 0; c < len; c++) {
  474. x[c] = (char*)(d + ii[r[crow + c] + io[c]]);
  475. }
  476. }
  477. if (i == LC_CTYPE) {
  478. c = __locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES * row
  479. + 2 ]; /* codeset */
  480. if (c <= 2) {
  481. if (c == 2) {
  482. base->codeset = utf8;
  483. base->encoding = __ctype_encoding_utf8;
  484. /* TODO - fix for bcc */
  485. base->mb_cur_max = 6;
  486. } else {
  487. assert(c == 1);
  488. base->codeset = ascii;
  489. base->encoding = __ctype_encoding_7_bit;
  490. base->mb_cur_max = 1;
  491. }
  492. } else {
  493. const __codeset_8_bit_t *c8b;
  494. r = CODESET_LIST;
  495. c -= 3;
  496. base->codeset = (char *) (r + r[c]);
  497. base->encoding = __ctype_encoding_8_bit;
  498. /* TODO - update when translit implemented! */
  499. base->mb_cur_max = 1;
  500. c8b = __locale_mmap->codeset_8_bit + c;
  501. #ifdef __CTYPE_HAS_8_BIT_LOCALES
  502. base->idx8ctype = c8b->idx8ctype;
  503. base->idx8uplow = c8b->idx8uplow;
  504. #ifdef __UCLIBC_HAS_WCHAR__
  505. base->idx8c2wc = c8b->idx8c2wc;
  506. base->idx8wc2c = c8b->idx8wc2c;
  507. /* translit */
  508. #endif /* __UCLIBC_HAS_WCHAR__ */
  509. /* What follows is fairly bloated, but it is just a hack
  510. * to get the 8-bit codeset ctype stuff functioning.
  511. * All of this will be replaced in the next generation
  512. * of locale support anyway... */
  513. memcpy(base->__ctype_b_data,
  514. __C_ctype_b - __UCLIBC_CTYPE_B_TBL_OFFSET,
  515. (256 + __UCLIBC_CTYPE_B_TBL_OFFSET)
  516. * sizeof(__ctype_mask_t));
  517. memcpy(base->__ctype_tolower_data,
  518. __C_ctype_tolower - __UCLIBC_CTYPE_TO_TBL_OFFSET,
  519. (256 + __UCLIBC_CTYPE_TO_TBL_OFFSET)
  520. * sizeof(__ctype_touplow_t));
  521. memcpy(base->__ctype_toupper_data,
  522. __C_ctype_toupper - __UCLIBC_CTYPE_TO_TBL_OFFSET,
  523. (256 + __UCLIBC_CTYPE_TO_TBL_OFFSET)
  524. * sizeof(__ctype_touplow_t));
  525. #define Cctype_TBL_MASK ((1 << __LOCALE_DATA_Cctype_IDX_SHIFT) - 1)
  526. #define Cctype_IDX_OFFSET (128 >> __LOCALE_DATA_Cctype_IDX_SHIFT)
  527. {
  528. int u;
  529. __ctype_mask_t m;
  530. for (u=0 ; u < 128 ; u++) {
  531. #ifdef __LOCALE_DATA_Cctype_PACKED
  532. c = base->tbl8ctype
  533. [ ((int)(c8b->idx8ctype
  534. [(u >> __LOCALE_DATA_Cctype_IDX_SHIFT) ])
  535. << (__LOCALE_DATA_Cctype_IDX_SHIFT - 1))
  536. + ((u & Cctype_TBL_MASK) >> 1)];
  537. c = (u & 1) ? (c >> 4) : (c & 0xf);
  538. #else
  539. c = base->tbl8ctype
  540. [ ((int)(c8b->idx8ctype
  541. [(u >> __LOCALE_DATA_Cctype_IDX_SHIFT) ])
  542. << __LOCALE_DATA_Cctype_IDX_SHIFT)
  543. + (u & Cctype_TBL_MASK) ];
  544. #endif
  545. m = base->code2flag[c];
  546. base->__ctype_b_data
  547. [128 + __UCLIBC_CTYPE_B_TBL_OFFSET + u]
  548. = m;
  549. #ifdef __UCLIBC_HAS_CTYPE_SIGNED__
  550. if (((signed char)(128 + u)) != -1) {
  551. base->__ctype_b_data[__UCLIBC_CTYPE_B_TBL_OFFSET
  552. + ((signed char)(128 + u))]
  553. = m;
  554. }
  555. #endif
  556. base->__ctype_tolower_data
  557. [128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
  558. = 128 + u;
  559. base->__ctype_toupper_data
  560. [128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
  561. = 128 + u;
  562. if (m & (_ISlower|_ISupper)) {
  563. c = base->tbl8uplow
  564. [ ((int)(c8b->idx8uplow
  565. [u >> __LOCALE_DATA_Cuplow_IDX_SHIFT])
  566. << __LOCALE_DATA_Cuplow_IDX_SHIFT)
  567. + ((128 + u)
  568. & ((1 << __LOCALE_DATA_Cuplow_IDX_SHIFT)
  569. - 1)) ];
  570. if (m & _ISlower) {
  571. base->__ctype_toupper_data
  572. [128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
  573. = (unsigned char)(128 + u + c);
  574. #ifdef __UCLIBC_HAS_CTYPE_SIGNED__
  575. if (((signed char)(128 + u)) != -1) {
  576. base->__ctype_toupper_data
  577. [__UCLIBC_CTYPE_TO_TBL_OFFSET
  578. + ((signed char)(128 + u))]
  579. = (unsigned char)(128 + u + c);
  580. }
  581. #endif
  582. } else {
  583. base->__ctype_tolower_data
  584. [128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
  585. = (unsigned char)(128 + u - c);
  586. #ifdef __UCLIBC_HAS_CTYPE_SIGNED__
  587. if (((signed char)(128 + u)) != -1) {
  588. base->__ctype_tolower_data
  589. [__UCLIBC_CTYPE_TO_TBL_OFFSET
  590. + ((signed char)(128 + u))]
  591. = (unsigned char)(128 + u - c);
  592. }
  593. #endif
  594. }
  595. }
  596. }
  597. }
  598. #ifdef __UCLIBC_HAS_XLOCALE__
  599. base->__ctype_b = base->__ctype_b_data
  600. + __UCLIBC_CTYPE_B_TBL_OFFSET;
  601. base->__ctype_tolower = base->__ctype_tolower_data
  602. + __UCLIBC_CTYPE_TO_TBL_OFFSET;
  603. base->__ctype_toupper = base->__ctype_toupper_data
  604. + __UCLIBC_CTYPE_TO_TBL_OFFSET;
  605. #else /* __UCLIBC_HAS_XLOCALE__ */
  606. __ctype_b = base->__ctype_b_data
  607. + __UCLIBC_CTYPE_B_TBL_OFFSET;
  608. __ctype_tolower = base->__ctype_tolower_data
  609. + __UCLIBC_CTYPE_TO_TBL_OFFSET;
  610. __ctype_toupper = base->__ctype_toupper_data
  611. + __UCLIBC_CTYPE_TO_TBL_OFFSET;
  612. #endif /* __UCLIBC_HAS_XLOCALE__ */
  613. #endif /* __CTYPE_HAS_8_BIT_LOCALES */
  614. }
  615. d = base->outdigit_length;
  616. x = &base->outdigit0_mb;
  617. for (c = 0 ; c < 10 ; c++) {
  618. ((unsigned char *)d)[c] = strlen(x[c]);
  619. assert(d[c] > 0);
  620. }
  621. } else if (i == LC_NUMERIC) {
  622. assert(LC_NUMERIC > LC_CTYPE); /* Need ctype initialized. */
  623. base->decimal_point_len
  624. = __locale_mbrtowc_l(&base->decimal_point_wc,
  625. base->decimal_point, base);
  626. assert(base->decimal_point_len > 0);
  627. assert(base->decimal_point[base->decimal_point_len] == 0);
  628. if (*base->grouping) {
  629. base->thousands_sep_len
  630. = __locale_mbrtowc_l(&base->thousands_sep_wc,
  631. base->thousands_sep, base);
  632. #if 1
  633. assert(base->thousands_sep_len >= 0);
  634. if (base->thousands_sep_len == 0) {
  635. base->grouping = base->thousands_sep; /* empty string */
  636. }
  637. assert(base->thousands_sep[base->thousands_sep_len] == 0);
  638. #else
  639. assert(base->thousands_sep_len > 0);
  640. assert(base->thousands_sep[base->thousands_sep_len] == 0);
  641. #endif
  642. }
  643. /* } else if (i == LC_COLLATE) { */
  644. /* init_cur_collate(__locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES */
  645. /* * row + 3 + i ], */
  646. /* &base->collate); */
  647. }
  648. }
  649. ++i;
  650. p += 2;
  651. s += 2;
  652. } while (i < LC_ALL);
  653. return 1;
  654. }
  655. static const uint16_t __code2flag[16] = {
  656. 0, /* unclassified = 0 */
  657. _ISprint|_ISgraph|_ISalnum|_ISalpha, /* alpha_nonupper_nonlower */
  658. _ISprint|_ISgraph|_ISalnum|_ISalpha|_ISlower, /* alpha_lower */
  659. _ISprint|_ISgraph|_ISalnum|_ISalpha|_ISlower|_ISupper, /* alpha_upper_lower */
  660. _ISprint|_ISgraph|_ISalnum|_ISalpha|_ISupper, /* alpha_upper */
  661. _ISprint|_ISgraph|_ISalnum|_ISdigit, /* digit */
  662. _ISprint|_ISgraph|_ISpunct, /* punct */
  663. _ISprint|_ISgraph, /* graph */
  664. _ISprint|_ISspace, /* print_space_nonblank */
  665. _ISprint|_ISspace|_ISblank, /* print_space_blank */
  666. _ISspace, /* space_nonblank_noncntrl */
  667. _ISspace|_ISblank, /* space_blank_noncntrl */
  668. _IScntrl|_ISspace, /* cntrl_space_nonblank */
  669. _IScntrl|_ISspace|_ISblank, /* cntrl_space_blank */
  670. _IScntrl /* cntrl_nonspace */
  671. };
  672. void attribute_hidden _locale_init_l(__locale_t base)
  673. {
  674. memset(base->cur_locale, 0, LOCALE_SELECTOR_SIZE);
  675. base->cur_locale[0] = '#';
  676. memcpy(base->category_item_count,
  677. __locale_mmap->lc_common_item_offsets_LEN,
  678. LC_ALL);
  679. ++base->category_item_count[0]; /* Increment for codeset entry. */
  680. base->category_offsets[0] = offsetof(struct __uclibc_locale_struct, outdigit0_mb);
  681. base->category_offsets[1] = offsetof(struct __uclibc_locale_struct, decimal_point);
  682. base->category_offsets[2] = offsetof(struct __uclibc_locale_struct, int_curr_symbol);
  683. base->category_offsets[3] = offsetof(struct __uclibc_locale_struct, abday_1);
  684. /* base->category_offsets[4] = offsetof(struct __uclibc_locale_struct, collate???); */
  685. base->category_offsets[5] = offsetof(struct __uclibc_locale_struct, yesexpr);
  686. #ifdef __CTYPE_HAS_8_BIT_LOCALES
  687. base->tbl8ctype
  688. = (const unsigned char *) &__locale_mmap->tbl8ctype;
  689. base->tbl8uplow
  690. = (const unsigned char *) &__locale_mmap->tbl8uplow;
  691. #ifdef __UCLIBC_HAS_WCHAR__
  692. base->tbl8c2wc
  693. = (const uint16_t *) &__locale_mmap->tbl8c2wc;
  694. base->tbl8wc2c
  695. = (const unsigned char *) &__locale_mmap->tbl8wc2c;
  696. /* translit */
  697. #endif /* __UCLIBC_HAS_WCHAR__ */
  698. #endif /* __CTYPE_HAS_8_BIT_LOCALES */
  699. #ifdef __UCLIBC_HAS_WCHAR__
  700. base->tblwctype
  701. = (const unsigned char *) &__locale_mmap->tblwctype;
  702. base->tblwuplow
  703. = (const unsigned char *) &__locale_mmap->tblwuplow;
  704. base->tblwuplow_diff
  705. = (const int16_t *) &__locale_mmap->tblwuplow_diff;
  706. /* base->tblwcomb */
  707. /* = (const unsigned char *) &__locale_mmap->tblwcomb; */
  708. /* width?? */
  709. #endif /* __UCLIBC_HAS_WCHAR__ */
  710. /* Initially, set things up to use the global C ctype tables.
  711. * This is correct for C (ASCII) and UTF-8 based locales (except tr_TR). */
  712. #ifdef __UCLIBC_HAS_XLOCALE__
  713. base->__ctype_b = __C_ctype_b;
  714. base->__ctype_tolower = __C_ctype_tolower;
  715. base->__ctype_toupper = __C_ctype_toupper;
  716. #else /* __UCLIBC_HAS_XLOCALE__ */
  717. __ctype_b = __C_ctype_b;
  718. __ctype_tolower = __C_ctype_tolower;
  719. __ctype_toupper = __C_ctype_toupper;
  720. #endif /* __UCLIBC_HAS_XLOCALE__ */
  721. base->code2flag = __code2flag;
  722. _locale_set_l((unsigned char*) C_LOCALE_SELECTOR, base);
  723. }
  724. void _locale_init(void)
  725. {
  726. /* TODO: mmap the locale file */
  727. /* TODO - ??? */
  728. _locale_init_l(__global_locale);
  729. }
  730. #endif
  731. /**********************************************************************/
  732. #if defined(L_nl_langinfo) || defined(L_nl_langinfo_l)
  733. #ifdef __LOCALE_C_ONLY
  734. /* We need to index 320 bytes of data, so you might initially think we
  735. * need to store the offsets in shorts. But since the offset of the
  736. * 64th item is 182, we'll store "offset - 2*64" for all items >= 64
  737. * and always calculate the data offset as "offset[i] + 2*(i & 64)".
  738. * This allows us to pack the data offsets in an unsigned char while
  739. * also avoiding an "if".
  740. *
  741. * Note: Category order is assumed to be:
  742. * ctype, numeric, monetary, time, collate, messages, all
  743. */
  744. #define C_LC_ALL 6
  745. /* Combine the data to avoid size penalty for seperate char arrays when
  746. * compiler aligns objects. The original code is left in as documentation. */
  747. #define cat_start nl_data
  748. #define C_locale_data (nl_data + C_LC_ALL + 1 + 90)
  749. static const unsigned char nl_data[C_LC_ALL + 1 + 90 + 320] = {
  750. /* static const char cat_start[LC_ALL + 1] = { */
  751. '\x00', '\x0b', '\x0e', '\x24', '\x56', '\x56', '\x5a',
  752. /* }; */
  753. /* static const char item_offset[90] = { */
  754. '\x00', '\x02', '\x04', '\x06', '\x08', '\x0a', '\x0c', '\x0e',
  755. '\x10', '\x12', '\x14', '\x1a', '\x1b', '\x1b', '\x1b', '\x1b',
  756. '\x1b', '\x1b', '\x1b', '\x1b', '\x1b', '\x1c', '\x1c', '\x1c',
  757. '\x1c', '\x1c', '\x1c', '\x1c', '\x1c', '\x1c', '\x1c', '\x1c',
  758. '\x1c', '\x1c', '\x1c', '\x1e', '\x20', '\x24', '\x28', '\x2c',
  759. '\x30', '\x34', '\x38', '\x3c', '\x43', '\x4a', '\x52', '\x5c',
  760. '\x65', '\x6c', '\x75', '\x79', '\x7d', '\x81', '\x85', '\x89',
  761. '\x8d', '\x91', '\x95', '\x99', '\x9d', '\xa1', '\xa5', '\xad',
  762. '\x36', '\x3c', '\x42', '\x46', '\x4b', '\x50', '\x57', '\x61',
  763. '\x69', '\x72', '\x7b', '\x7e', '\x81', '\x96', '\x9f', '\xa8',
  764. '\xb3', '\xb3', '\xb3', '\xb3', '\xb3', '\xb3', '\xb4', '\xba',
  765. '\xbf', '\xbf',
  766. /* }; */
  767. /* static const char C_locale_data[320] = { */
  768. '0', '\x00', '1', '\x00', '2', '\x00', '3', '\x00',
  769. '4', '\x00', '5', '\x00', '6', '\x00', '7', '\x00',
  770. '8', '\x00', '9', '\x00', 'A', 'S', 'C', 'I',
  771. 'I', '\x00', '.', '\x00', '\x7f', '\x00', '-', '\x00',
  772. 'S', 'u', 'n', '\x00', 'M', 'o', 'n', '\x00',
  773. 'T', 'u', 'e', '\x00', 'W', 'e', 'd', '\x00',
  774. 'T', 'h', 'u', '\x00', 'F', 'r', 'i', '\x00',
  775. 'S', 'a', 't', '\x00', 'S', 'u', 'n', 'd',
  776. 'a', 'y', '\x00', 'M', 'o', 'n', 'd', 'a',
  777. 'y', '\x00', 'T', 'u', 'e', 's', 'd', 'a',
  778. 'y', '\x00', 'W', 'e', 'd', 'n', 'e', 's',
  779. 'd', 'a', 'y', '\x00', 'T', 'h', 'u', 'r',
  780. 's', 'd', 'a', 'y', '\x00', 'F', 'r', 'i',
  781. 'd', 'a', 'y', '\x00', 'S', 'a', 't', 'u',
  782. 'r', 'd', 'a', 'y', '\x00', 'J', 'a', 'n',
  783. '\x00', 'F', 'e', 'b', '\x00', 'M', 'a', 'r',
  784. '\x00', 'A', 'p', 'r', '\x00', 'M', 'a', 'y',
  785. '\x00', 'J', 'u', 'n', '\x00', 'J', 'u', 'l',
  786. '\x00', 'A', 'u', 'g', '\x00', 'S', 'e', 'p',
  787. '\x00', 'O', 'c', 't', '\x00', 'N', 'o', 'v',
  788. '\x00', 'D', 'e', 'c', '\x00', 'J', 'a', 'n',
  789. 'u', 'a', 'r', 'y', '\x00', 'F', 'e', 'b',
  790. 'r', 'u', 'a', 'r', 'y', '\x00', 'M', 'a',
  791. 'r', 'c', 'h', '\x00', 'A', 'p', 'r', 'i',
  792. 'l', '\x00', 'M', 'a', 'y', '\x00', 'J', 'u',
  793. 'n', 'e', '\x00', 'J', 'u', 'l', 'y', '\x00',
  794. 'A', 'u', 'g', 'u', 's', 't', '\x00', 'S',
  795. 'e', 'p', 't', 'e', 'm', 'b', 'e', 'r',
  796. '\x00', 'O', 'c', 't', 'o', 'b', 'e', 'r',
  797. '\x00', 'N', 'o', 'v', 'e', 'm', 'b', 'e',
  798. 'r', '\x00', 'D', 'e', 'c', 'e', 'm', 'b',
  799. 'e', 'r', '\x00', 'A', 'M', '\x00', 'P', 'M',
  800. '\x00', '%', 'a', ' ', '%', 'b', ' ', '%',
  801. 'e', ' ', '%', 'H', ':', '%', 'M', ':',
  802. '%', 'S', ' ', '%', 'Y', '\x00', '%', 'm',
  803. '/', '%', 'd', '/', '%', 'y', '\x00', '%',
  804. 'H', ':', '%', 'M', ':', '%', 'S', '\x00',
  805. '%', 'I', ':', '%', 'M', ':', '%', 'S',
  806. ' ', '%', 'p', '\x00', '^', '[', 'y', 'Y',
  807. ']', '\x00', '^', '[', 'n', 'N', ']', '\x00',
  808. };
  809. char *nl_langinfo(nl_item item)
  810. {
  811. unsigned int c;
  812. unsigned int i;
  813. if ((c = _NL_ITEM_CATEGORY(item)) < C_LC_ALL) {
  814. if ((i = cat_start[c] + _NL_ITEM_INDEX(item)) < cat_start[c+1]) {
  815. /* return (char *) C_locale_data + item_offset[i] + (i & 64); */
  816. return (char *) C_locale_data + nl_data[C_LC_ALL+1+i] + 2*(i & 64);
  817. }
  818. }
  819. return (char *) cat_start; /* Conveniently, this is the empty string. */
  820. }
  821. libc_hidden_def(nl_langinfo)
  822. #else /* __LOCALE_C_ONLY */
  823. #if defined(__UCLIBC_HAS_XLOCALE__) && !defined(__UCLIBC_DO_XLOCALE)
  824. char *nl_langinfo(nl_item item)
  825. {
  826. return nl_langinfo_l(item, __UCLIBC_CURLOCALE);
  827. }
  828. libc_hidden_def(nl_langinfo)
  829. #else /* defined(__UCLIBC_HAS_XLOCALE__) && !defined(__UCLIBC_DO_XLOCALE) */
  830. libc_hidden_proto(__XL_NPP(nl_langinfo))
  831. static const char empty[] = "";
  832. char *__XL_NPP(nl_langinfo)(nl_item item __LOCALE_PARAM )
  833. {
  834. unsigned int c = _NL_ITEM_CATEGORY(item);
  835. unsigned int i = _NL_ITEM_INDEX(item);
  836. if ((c < LC_ALL) && (i < __LOCALE_PTR->category_item_count[c])) {
  837. return ((char **)(((char *) __LOCALE_PTR)
  838. + __LOCALE_PTR->category_offsets[c]))[i];
  839. }
  840. return (char *) empty;
  841. }
  842. libc_hidden_def(__XL_NPP(nl_langinfo))
  843. #endif /* defined(__UCLIBC_HAS_XLOCALE__) && !defined(__UCLIBC_DO_XLOCALE) */
  844. #endif /* __LOCALE_C_ONLY */
  845. #endif
  846. /**********************************************************************/
  847. #ifdef L_newlocale
  848. static const char posix[] = "POSIX";
  849. static const char utf8[] = "UTF-8";
  850. static int find_locale(int category_mask, const char *p,
  851. unsigned char *new_locale)
  852. {
  853. int i;
  854. const unsigned char *s;
  855. uint16_t n;
  856. unsigned char lang_cult, codeset;
  857. #if defined(__LOCALE_DATA_AT_MODIFIERS_LENGTH) && 1
  858. /* Support standard locale handling for @-modifiers. */
  859. char buf[18]; /* TODO: 7+{max codeset name length} */
  860. const char *q;
  861. if ((q = strchr(p,'@')) != NULL) {
  862. if ((((size_t)((q-p)-5)) > (sizeof(buf) - 5)) || (p[2] != '_')) {
  863. return 0;
  864. }
  865. /* locale name at least 5 chars long and 3rd char is '_' */
  866. s = LOCALE_AT_MODIFIERS;
  867. do {
  868. if (!strcmp((char*) (s + 2), q + 1)) {
  869. break;
  870. }
  871. s += 2 + *s; /* TODO - fix this throughout */
  872. } while (*s);
  873. if (!*s) {
  874. return 0;
  875. }
  876. assert(q - p < sizeof(buf));
  877. memcpy(buf, p, q-p);
  878. buf[q-p] = 0;
  879. buf[2] = s[1];
  880. p = buf;
  881. }
  882. #endif
  883. lang_cult = codeset = 0; /* Assume C and default codeset. */
  884. if (((*p == 'C') && !p[1]) || !strcmp(p, posix)) {
  885. goto FIND_LOCALE;
  886. }
  887. if ((strlen(p) > 5) && (p[5] == '.')) { /* Codeset in locale name? */
  888. /* TODO: maybe CODESET_LIST + *s ??? */
  889. /* 7bit is 1, UTF-8 is 2, 8-bit is >= 3 */
  890. codeset = 2;
  891. if (strcasecmp(utf8, p + 6) != 0) {/* TODO - fix! */
  892. s = CODESET_LIST;
  893. do {
  894. ++codeset; /* Increment codeset first. */
  895. if (!strcmp((char*) CODESET_LIST + *s, p + 6)) {
  896. goto FIND_LANG_CULT;
  897. }
  898. } while (*++s);
  899. return 0; /* No matching codeset! */
  900. }
  901. }
  902. FIND_LANG_CULT: /* Find language_culture number. */
  903. s = LOCALE_NAMES;
  904. do { /* TODO -- do a binary search? */
  905. /* TODO -- fix gen_mmap!*/
  906. ++lang_cult; /* Increment first since C/POSIX is 0. */
  907. if (!strncmp((char*) s, p, 5)) { /* Found a matching locale name; */
  908. goto FIND_LOCALE;
  909. }
  910. s += 5;
  911. } while (lang_cult < __LOCALE_DATA_NUM_LOCALE_NAMES);
  912. return 0; /* No matching language_culture! */
  913. FIND_LOCALE: /* Find locale row matching name and codeset */
  914. s = LOCALES;
  915. n = 0;
  916. do { /* TODO -- do a binary search? */
  917. if ((lang_cult == *s) && ((codeset == s[1]) || (codeset == s[2]))) {
  918. i = 1;
  919. s = new_locale + 1;
  920. do {
  921. if (category_mask & i) {
  922. /* Encode current locale row number. */
  923. ((unsigned char *) s)[0] = (n >> 7) | 0x80;
  924. ((unsigned char *) s)[1] = (n & 0x7f) | 0x80;
  925. }
  926. s += 2;
  927. i += i;
  928. } while (i < (1 << LC_ALL));
  929. return i; /* Return non-zero */
  930. }
  931. s += __LOCALE_DATA_WIDTH_LOCALES;
  932. ++n;
  933. } while (n <= __LOCALE_DATA_NUM_LOCALES); /* We started at 1!!! */
  934. return 0; /* Unsupported locale. */
  935. }
  936. static unsigned char *composite_locale(int category_mask, const char *locale,
  937. unsigned char *new_locale)
  938. {
  939. char buf[MAX_LOCALE_STR];
  940. char *t;
  941. char *e;
  942. int c;
  943. int component_mask;
  944. if (!strchr(locale,'=')) {
  945. if (!find_locale(category_mask, locale, new_locale)) {
  946. return NULL;
  947. }
  948. return new_locale;
  949. }
  950. if (strlen(locale) >= sizeof(buf)) {
  951. return NULL;
  952. }
  953. stpcpy(buf, locale);
  954. component_mask = 0;
  955. t = strtok_r(buf, "=", &e); /* This can't fail because of strchr test above. */
  956. do {
  957. c = 0;
  958. /* CATEGORY_NAMES is unsigned char* */
  959. while (strcmp((char*) CATEGORY_NAMES + (int) CATEGORY_NAMES[c], t)) {
  960. if (++c == LC_ALL) { /* Unknown category name! */
  961. return NULL;
  962. }
  963. }
  964. t = strtok_r(NULL, ";", &e);
  965. c = (1 << c);
  966. if (component_mask & c) { /* Multiple components for one category. */
  967. return NULL;
  968. }
  969. component_mask |= c;
  970. if ((category_mask & c) && (!t || !find_locale(c, t, new_locale))) {
  971. return NULL;
  972. }
  973. } while ((t = strtok_r(NULL, "=", &e)) != NULL);
  974. if (category_mask & ~component_mask) { /* Category component(s) missing. */
  975. return NULL;
  976. }
  977. return new_locale;
  978. }
  979. __locale_t newlocale(int category_mask, const char *locale, __locale_t base)
  980. {
  981. const char *p;
  982. int i, j, k;
  983. unsigned char new_selector[LOCALE_SELECTOR_SIZE];
  984. if (category_mask == (1 << LC_ALL)) {
  985. category_mask = LC_ALL_MASK;
  986. }
  987. if (!locale || ((unsigned)(category_mask) > LC_ALL_MASK)) {
  988. INVALID:
  989. __set_errno(EINVAL);
  990. return NULL; /* No locale or illegal/unsupported category. */
  991. }
  992. strcpy((char *) new_selector,
  993. (base ? (char *) base->cur_locale : C_LOCALE_SELECTOR));
  994. if (!locale[0]) { /* locale == "", so check environment. */
  995. const char *envstr[4];
  996. envstr[0] = "LC_ALL";
  997. envstr[1] = NULL;
  998. envstr[2] = "LANG";
  999. envstr[3] = posix;
  1000. i = 1;
  1001. k = 0;
  1002. do {
  1003. if (category_mask & i) {
  1004. /* Note: SUSv3 doesn't define a fallback mechanism here.
  1005. * So, if LC_ALL is invalid, we do _not_ continue trying
  1006. * the other environment vars. */
  1007. envstr[1] = (char*) CATEGORY_NAMES + CATEGORY_NAMES[k];
  1008. j = 0;
  1009. while (1) {
  1010. p = envstr[j];
  1011. if (++j >= 4)
  1012. break; /* now p == "POSIX" */
  1013. p = getenv(p);
  1014. if (p && p[0])
  1015. break;
  1016. };
  1017. /* The user set something... is it valid? */
  1018. /* Note: Since we don't support user-supplied locales and
  1019. * alternate paths, we don't need to worry about special
  1020. * handling for suid/sgid apps. */
  1021. if (!find_locale(i, p, new_selector)) {
  1022. goto INVALID;
  1023. }
  1024. }
  1025. i += i;
  1026. } while (++k < LC_ALL);
  1027. } else if (!composite_locale(category_mask, locale, new_selector)) {
  1028. goto INVALID;
  1029. }
  1030. /* If we get here, the new selector corresponds to a valid locale. */
  1031. #if 0
  1032. if (base) {
  1033. _locale_set_l(new_selector, base);
  1034. } else {
  1035. base = _locale_new(new_selector);
  1036. }
  1037. #else
  1038. if (!base) {
  1039. base = calloc(1, sizeof(struct __uclibc_locale_struct));
  1040. if (base == NULL)
  1041. return base;
  1042. _locale_init_l(base);
  1043. }
  1044. _locale_set_l(new_selector, base);
  1045. #endif
  1046. return base;
  1047. }
  1048. #ifdef __UCLIBC_HAS_XLOCALE__
  1049. libc_hidden_def(newlocale)
  1050. #endif
  1051. #endif
  1052. /**********************************************************************/
  1053. #ifdef L_duplocale
  1054. __locale_t duplocale(__locale_t dataset)
  1055. {
  1056. __locale_t r;
  1057. uint16_t * i2w;
  1058. size_t n;
  1059. assert(dataset != LC_GLOBAL_LOCALE);
  1060. r = malloc(sizeof(struct __uclibc_locale_struct));
  1061. if (r != NULL) {
  1062. n = 2 * dataset->collate.max_col_index + 2;
  1063. i2w = calloc(n, sizeof(uint16_t));
  1064. if (i2w != NULL) {
  1065. memcpy(r, dataset, sizeof(struct __uclibc_locale_struct));
  1066. r->collate.index2weight = i2w;
  1067. memcpy(i2w, dataset->collate.index2weight, n * sizeof(uint16_t));
  1068. } else {
  1069. free(r);
  1070. r = NULL;
  1071. }
  1072. }
  1073. return r;
  1074. }
  1075. #endif
  1076. /**********************************************************************/
  1077. #ifdef L_freelocale
  1078. void freelocale(__locale_t dataset)
  1079. {
  1080. assert(dataset != __global_locale);
  1081. assert(dataset != LC_GLOBAL_LOCALE);
  1082. free(dataset->collate.index2weight); /* Free collation data. */
  1083. free(dataset); /* Free locale */
  1084. }
  1085. #endif
  1086. /**********************************************************************/
  1087. #ifdef L_uselocale
  1088. __locale_t uselocale(__locale_t dataset)
  1089. {
  1090. __locale_t old;
  1091. if (!dataset) {
  1092. old = __UCLIBC_CURLOCALE;
  1093. } else {
  1094. if (dataset == LC_GLOBAL_LOCALE) {
  1095. dataset = __global_locale;
  1096. }
  1097. #ifdef __UCLIBC_HAS_THREADS__
  1098. old = __curlocale_set(dataset);
  1099. #else
  1100. old = __curlocale_var;
  1101. __curlocale_var = dataset;
  1102. #endif
  1103. }
  1104. if (old == __global_locale) {
  1105. return LC_GLOBAL_LOCALE;
  1106. }
  1107. return old;
  1108. }
  1109. libc_hidden_def(uselocale)
  1110. #endif
  1111. /**********************************************************************/
  1112. #ifdef L___curlocale
  1113. #ifdef __UCLIBC_HAS_THREADS__
  1114. __locale_t weak_const_function __curlocale(void)
  1115. {
  1116. return __curlocale_var; /* This is overriden by the thread version. */
  1117. }
  1118. libc_hidden_weak(__curlocale)
  1119. __locale_t weak_function __curlocale_set(__locale_t newloc)
  1120. {
  1121. __locale_t oldloc = __curlocale_var;
  1122. assert(newloc != LC_GLOBAL_LOCALE);
  1123. __curlocale_var = newloc;
  1124. return oldloc;
  1125. }
  1126. libc_hidden_weak(__curlocale_set)
  1127. #endif
  1128. #endif
  1129. /**********************************************************************/
  1130. #ifdef L___locale_mbrtowc_l
  1131. /* NOTE: This returns an int... not size_t. Also, it is not a general
  1132. * routine. It is actually a very stripped-down version of mbrtowc
  1133. * that takes a __locale_t arg. This is used by strcoll and strxfrm.
  1134. * It is also used above to generate wchar_t versions of the decimal point
  1135. * and thousands seperator. */
  1136. #ifndef __CTYPE_HAS_UTF_8_LOCALES
  1137. #warning __CTYPE_HAS_UTF_8_LOCALES not set!
  1138. #endif
  1139. #ifndef __CTYPE_HAS_8_BIT_LOCALES
  1140. #warning __CTYPE_HAS_8_BIT_LOCALES not set!
  1141. #endif
  1142. #define Cc2wc_IDX_SHIFT __LOCALE_DATA_Cc2wc_IDX_SHIFT
  1143. #define Cc2wc_ROW_LEN __LOCALE_DATA_Cc2wc_ROW_LEN
  1144. extern size_t _wchar_utf8sntowcs(wchar_t *__restrict pwc, size_t wn,
  1145. const char **__restrict src, size_t n,
  1146. mbstate_t *ps, int allow_continuation) attribute_hidden;
  1147. int attribute_hidden __locale_mbrtowc_l(wchar_t *__restrict dst,
  1148. const char *__restrict src,
  1149. __locale_t loc )
  1150. {
  1151. #ifdef __CTYPE_HAS_UTF_8_LOCALES
  1152. if (loc->encoding == __ctype_encoding_utf8) {
  1153. mbstate_t ps;
  1154. const char *p = src;
  1155. size_t r;
  1156. ps.__mask = 0;
  1157. r = _wchar_utf8sntowcs(dst, 1, &p, SIZE_MAX, &ps, 1);
  1158. return (r == 1) ? (p-src) : r; /* Need to return 0 if nul char. */
  1159. }
  1160. #endif
  1161. #ifdef __CTYPE_HAS_8_BIT_LOCALES
  1162. assert((loc->encoding == __ctype_encoding_7_bit) || (loc->encoding == __ctype_encoding_8_bit));
  1163. #else
  1164. assert(loc->encoding == __ctype_encoding_7_bit);
  1165. #endif
  1166. if ((*dst = ((unsigned char)(*src))) < 0x80) { /* ASCII... */
  1167. return (*src != 0);
  1168. }
  1169. #ifdef __CTYPE_HAS_8_BIT_LOCALES
  1170. if (loc->encoding == __ctype_encoding_8_bit) {
  1171. wchar_t wc = *dst - 0x80;
  1172. *dst = loc->tbl8c2wc[
  1173. (loc->idx8c2wc[wc >> Cc2wc_IDX_SHIFT]
  1174. << Cc2wc_IDX_SHIFT) + (wc & (Cc2wc_ROW_LEN - 1))];
  1175. if (*dst) {
  1176. return 1;
  1177. }
  1178. }
  1179. #endif
  1180. return -1;
  1181. }
  1182. #endif
  1183. /**********************************************************************/