locale.c 42 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480
  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, write to the Free
  15. * Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  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. #ifdef __UCLIBC_HAS_CTYPE_TABLES__
  56. #endif
  57. #ifdef __UCLIBC_MJN3_ONLY__
  58. #ifdef L_setlocale
  59. #warning TODO: Make the link_warning()s a config option?
  60. #endif
  61. #endif
  62. #undef link_warning
  63. #define link_warning(A,B)
  64. #undef __LOCALE_C_ONLY
  65. #ifndef __UCLIBC_HAS_LOCALE__
  66. #define __LOCALE_C_ONLY
  67. #endif /* __UCLIBC_HAS_LOCALE__ */
  68. #ifdef __LOCALE_C_ONLY
  69. #include <locale.h>
  70. #else /* __LOCALE_C_ONLY */
  71. #ifdef __UCLIBC_MJN3_ONLY__
  72. #ifdef L_setlocale
  73. #warning TODO: Fix the __CTYPE_HAS_8_BIT_LOCALES define at the top of the file.
  74. #warning TODO: Fix __WCHAR_ENABLED.
  75. #endif
  76. #endif
  77. /* Need to include this before locale.h and xlocale.h! */
  78. #include <bits/uClibc_locale.h>
  79. #undef CODESET_LIST
  80. #define CODESET_LIST (__locale_mmap->codeset_list)
  81. #ifdef __UCLIBC_HAS_XLOCALE__
  82. #include <xlocale.h>
  83. #include <locale.h>
  84. #else /* __UCLIBC_HAS_XLOCALE__ */
  85. /* We need this internally... */
  86. #define __UCLIBC_HAS_XLOCALE__ 1
  87. #include <xlocale.h>
  88. #include <locale.h>
  89. #undef __UCLIBC_HAS_XLOCALE__
  90. #endif /* __UCLIBC_HAS_XLOCALE__ */
  91. #include <wchar.h>
  92. #define LOCALE_NAMES (__locale_mmap->locale_names5)
  93. #define LOCALES (__locale_mmap->locales)
  94. #define LOCALE_AT_MODIFIERS (__locale_mmap->locale_at_modifiers)
  95. #define CATEGORY_NAMES (__locale_mmap->lc_names)
  96. #ifdef __UCLIBC_MJN3_ONLY__
  97. #warning REMINDER: redo the MAX_LOCALE_STR stuff...
  98. #endif
  99. #define MAX_LOCALE_STR 256 /* TODO: Only sufficient for current case. */
  100. #define MAX_LOCALE_CATEGORY_STR 32 /* TODO: Only sufficient for current case. */
  101. /* Note: Best if MAX_LOCALE_CATEGORY_STR is a power of 2. */
  102. extern int _locale_set_l(const unsigned char *p, __locale_t base) attribute_hidden;
  103. extern void _locale_init_l(__locale_t base) attribute_hidden;
  104. #endif /* __LOCALE_C_ONLY */
  105. #undef LOCALE_STRING_SIZE
  106. #define LOCALE_SELECTOR_SIZE (2 * __LC_ALL + 2)
  107. #ifdef __UCLIBC_MJN3_ONLY__
  108. #ifdef L_setlocale
  109. #warning TODO: Create a C locale selector string.
  110. #endif
  111. #endif
  112. #define C_LOCALE_SELECTOR "\x23\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80"
  113. #include <langinfo.h>
  114. #include <nl_types.h>
  115. /**********************************************************************/
  116. #ifdef L_setlocale
  117. #ifdef __LOCALE_C_ONLY
  118. link_warning(setlocale,"REMINDER: The 'setlocale' function supports only C|POSIX locales.")
  119. static const char C_string[] = "C";
  120. char *setlocale(int category, register const char *locale)
  121. {
  122. return ( (((unsigned int)(category)) <= LC_ALL)
  123. && ( (!locale) /* Request for locale category string. */
  124. || (!*locale) /* Implementation-defined default is C. */
  125. || ((*locale == 'C') && !locale[1])
  126. || (!strcmp(locale, "POSIX"))) )
  127. ? (char *) C_string /* Always in C/POSIX locale. */
  128. : NULL;
  129. }
  130. #else /* ---------------------------------------------- __LOCALE_C_ONLY */
  131. #ifdef __UCLIBC_HAS_THREADS__
  132. link_warning(setlocale,"REMINDER: The 'setlocale' function is _not_ threadsafe except for simple queries.")
  133. #endif
  134. #if !defined(__LOCALE_DATA_NUM_LOCALES) || (__LOCALE_DATA_NUM_LOCALES <= 1)
  135. #error locales enabled, but not data other than for C locale!
  136. #endif
  137. #ifdef __UCLIBC_MJN3_ONLY__
  138. #warning TODO: Move posix and utf8 strings.
  139. #endif
  140. static const char posix[] = "POSIX";
  141. static const char utf8[] = "UTF-8";
  142. #ifdef __UCLIBC_MJN3_ONLY__
  143. #warning TODO: Fix dimensions of hr_locale.
  144. #endif
  145. /* Individual category strings start at hr_locale + category * MAX_LOCALE_CATEGORY.
  146. * This holds for LC_ALL as well.
  147. */
  148. static char hr_locale[(MAX_LOCALE_CATEGORY_STR * LC_ALL) + MAX_LOCALE_STR];
  149. static void update_hr_locale(const unsigned char *spec)
  150. {
  151. const unsigned char *loc;
  152. const unsigned char *s;
  153. char *n;
  154. int i, category, done;
  155. done = category = 0;
  156. do {
  157. s = spec + 1;
  158. n = hr_locale + category * MAX_LOCALE_CATEGORY_STR;
  159. if (category == LC_ALL) {
  160. done = 1;
  161. for (i = 0 ; i < LC_ALL-1 ; i += 2) {
  162. if ((s[i] != s[i+2]) || (s[i+1] != s[i+3])) {
  163. goto SKIP;
  164. }
  165. }
  166. /* All categories the same, so simplify string by using a single
  167. * category. */
  168. category = LC_CTYPE;
  169. }
  170. SKIP:
  171. i = (category == LC_ALL) ? 0 : category;
  172. s += 2*i;
  173. do {
  174. if ((*s != 0xff) || (s[1] != 0xff)) {
  175. loc = LOCALES
  176. + __LOCALE_DATA_WIDTH_LOCALES * ((((int)(*s & 0x7f)) << 7)
  177. + (s[1] & 0x7f));
  178. if (category == LC_ALL) {
  179. /* CATEGORY_NAMES is unsigned char* */
  180. n = stpcpy(n, (char*) CATEGORY_NAMES + (int) CATEGORY_NAMES[i]);
  181. *n++ = '=';
  182. }
  183. if (*loc == 0) {
  184. *n++ = 'C';
  185. *n = 0;
  186. } else {
  187. char at = 0;
  188. memcpy(n, LOCALE_NAMES + 5*((*loc)-1), 5);
  189. if (n[2] != '_') {
  190. at = n[2];
  191. n[2] = '_';
  192. }
  193. n += 5;
  194. *n++ = '.';
  195. if (loc[2] == 2) {
  196. n = stpcpy(n, utf8);
  197. } else if (loc[2] >= 3) {
  198. n = stpcpy(n, (char*) CODESET_LIST + (int)(CODESET_LIST[loc[2] - 3]));
  199. }
  200. if (at) {
  201. const char *q;
  202. *n++ = '@';
  203. q = (char*) LOCALE_AT_MODIFIERS;
  204. do {
  205. if (q[1] == at) {
  206. n = stpcpy(n, q+2);
  207. break;
  208. }
  209. q += 2 + *q;
  210. } while (*q);
  211. }
  212. }
  213. *n++ = ';';
  214. }
  215. s += 2;
  216. } while (++i < category);
  217. *--n = 0; /* Remove trailing ';' and nul-terminate. */
  218. ++category;
  219. } while (!done);
  220. }
  221. char *setlocale(int category, const char *locale)
  222. {
  223. if (((unsigned int)(category)) > LC_ALL) {
  224. #if 0
  225. __set_errno(EINVAL); /* glibc sets errno -- SUSv3 doesn't say. */
  226. #endif
  227. return NULL; /* Illegal/unsupported category. */
  228. }
  229. if (locale != NULL) { /* Not just a query... */
  230. if (!newlocale((1 << category), locale, __global_locale)) {
  231. return NULL; /* Failed! */
  232. }
  233. update_hr_locale(__global_locale->cur_locale);
  234. }
  235. /* Either a query or a successful set, so return current locale string. */
  236. return hr_locale + (category * MAX_LOCALE_CATEGORY_STR);
  237. }
  238. #endif /* __LOCALE_C_ONLY */
  239. #endif
  240. /**********************************************************************/
  241. #ifdef L_localeconv
  242. /* Note: We assume here that the compiler does the sane thing regarding
  243. * placement of the fields in the struct. If necessary, we could ensure
  244. * this usings an array of offsets but at some size cost. */
  245. #ifdef __LOCALE_C_ONLY
  246. link_warning(localeconv,"REMINDER: The 'localeconv' function is hardwired for C/POSIX locale only.")
  247. static struct lconv the_lconv;
  248. static const char decpt[] = ".";
  249. struct lconv *localeconv(void)
  250. {
  251. register char *p = (char *)(&the_lconv);
  252. *((char **)p) = (char *) decpt;
  253. do {
  254. p += sizeof(char **);
  255. *((char **)p) = (char *) (decpt+1);
  256. } while (p < (char *) &the_lconv.negative_sign);
  257. p = (&the_lconv.int_frac_digits);
  258. do {
  259. *p = CHAR_MAX;
  260. ++p;
  261. } while (p <= &the_lconv.int_n_sign_posn);
  262. return &the_lconv;
  263. }
  264. #else /* __LOCALE_C_ONLY */
  265. static struct lconv the_lconv;
  266. struct lconv *localeconv(void)
  267. {
  268. register char *p = (char *) &the_lconv;
  269. register char **q = (char **) &(__UCLIBC_CURLOCALE->decimal_point);
  270. do {
  271. *((char **)p) = *q;
  272. p += sizeof(char **);
  273. ++q;
  274. } while (p < &the_lconv.int_frac_digits);
  275. do {
  276. *p = **q;
  277. ++p;
  278. ++q;
  279. } while (p <= &the_lconv.int_n_sign_posn);
  280. return &the_lconv;
  281. }
  282. #endif /* __LOCALE_C_ONLY */
  283. libc_hidden_def(localeconv)
  284. #endif
  285. /**********************************************************************/
  286. #if defined(L__locale_init) && !defined(__LOCALE_C_ONLY)
  287. #ifndef __UCLIBC_HAS_XLOCALE__
  288. #endif
  289. struct __uclibc_locale_struct __global_locale_data;
  290. __locale_t __global_locale = &__global_locale_data;
  291. #ifdef __UCLIBC_HAS_XLOCALE__
  292. __locale_t __curlocale_var = &__global_locale_data;
  293. #endif
  294. /*----------------------------------------------------------------------*/
  295. #ifdef __UCLIBC_MJN3_ONLY__
  296. #warning TODO: Move utf8 and ascii strings.
  297. #endif
  298. static const char utf8[] = "UTF-8";
  299. static const char ascii[] = "ASCII";
  300. typedef struct {
  301. uint16_t num_base;
  302. uint16_t num_der;
  303. uint16_t MAX_WEIGHTS;
  304. uint16_t num_index2weight;
  305. #define num_index2ruleidx num_index2weight
  306. uint16_t num_weightstr;
  307. uint16_t num_multistart;
  308. uint16_t num_override;
  309. uint16_t num_ruletable;
  310. } coldata_header_t;
  311. typedef struct {
  312. uint16_t num_weights;
  313. uint16_t num_starters;
  314. uint16_t ii_shift;
  315. uint16_t ti_shift;
  316. uint16_t ii_len;
  317. uint16_t ti_len;
  318. uint16_t max_weight;
  319. uint16_t num_col_base;
  320. uint16_t max_col_index;
  321. uint16_t undefined_idx;
  322. uint16_t range_low;
  323. uint16_t range_count;
  324. uint16_t range_base_weight;
  325. uint16_t range_rule_offset;
  326. uint16_t index2weight_offset;
  327. uint16_t index2ruleidx_offset;
  328. uint16_t multistart_offset;
  329. uint16_t wcs2colidt_offset_low;
  330. uint16_t wcs2colidt_offset_hi;
  331. } coldata_base_t;
  332. typedef struct {
  333. uint16_t base_idx;
  334. uint16_t undefined_idx;
  335. uint16_t overrides_offset;
  336. uint16_t multistart_offset;
  337. } coldata_der_t;
  338. static int init_cur_collate(int der_num, __collate_t *cur_collate)
  339. {
  340. const uint16_t *__locale_collate_tbl = __locale_mmap->collate_data;
  341. coldata_header_t *cdh;
  342. coldata_base_t *cdb;
  343. coldata_der_t *cdd;
  344. const uint16_t *p;
  345. size_t n;
  346. uint16_t i, w;
  347. #ifdef __UCLIBC_MJN3_ONLY__
  348. #warning kill of x86-specific asserts
  349. #endif
  350. #if 0
  351. assert(sizeof(coldata_base_t) == 19*2);
  352. assert(sizeof(coldata_der_t) == 4*2);
  353. assert(sizeof(coldata_header_t) == 8*2);
  354. #endif
  355. if (!der_num) { /* C locale... special */
  356. cur_collate->num_weights = 0;
  357. return 1;
  358. }
  359. --der_num;
  360. cdh = (coldata_header_t *) __locale_collate_tbl;
  361. #ifdef __UCLIBC_MJN3_ONLY__
  362. #warning CONSIDER: Should we assert here?
  363. #endif
  364. #if 0
  365. if (der_num >= cdh->num_der) {
  366. return 0;
  367. }
  368. #else
  369. assert((der_num < cdh->num_der));
  370. #endif
  371. cdd = (coldata_der_t *)(__locale_collate_tbl
  372. + (sizeof(coldata_header_t)
  373. + cdh->num_base * sizeof(coldata_base_t)
  374. + der_num * sizeof(coldata_der_t)
  375. )/2 );
  376. cdb = (coldata_base_t *)(__locale_collate_tbl
  377. + (sizeof(coldata_header_t)
  378. + cdd->base_idx * sizeof(coldata_base_t)
  379. )/2 );
  380. memcpy(cur_collate, cdb, offsetof(coldata_base_t,index2weight_offset));
  381. cur_collate->undefined_idx = cdd->undefined_idx;
  382. cur_collate->ti_mask = (1 << cur_collate->ti_shift)-1;
  383. cur_collate->ii_mask = (1 << cur_collate->ii_shift)-1;
  384. /* fflush(stdout); */
  385. /* fprintf(stderr,"base=%d num_col_base: %d %d\n", cdd->base_idx ,cur_collate->num_col_base, cdb->num_col_base); */
  386. n = (sizeof(coldata_header_t) + cdh->num_base * sizeof(coldata_base_t)
  387. + cdh->num_der * sizeof(coldata_der_t))/2;
  388. /* fprintf(stderr,"n = %d\n", n); */
  389. cur_collate->index2weight_tbl = __locale_collate_tbl + n + cdb->index2weight_offset;
  390. /* fprintf(stderr,"i2w = %d\n", n + cdb->index2weight_offset); */
  391. n += cdh->num_index2weight;
  392. cur_collate->index2ruleidx_tbl = __locale_collate_tbl + n + cdb->index2ruleidx_offset;
  393. /* fprintf(stderr,"i2r = %d\n", n + cdb->index2ruleidx_offset); */
  394. n += cdh->num_index2ruleidx;
  395. cur_collate->multistart_tbl = __locale_collate_tbl + n + cdd->multistart_offset;
  396. /* fprintf(stderr,"mts = %d\n", n + cdb->multistart_offset); */
  397. n += cdh->num_multistart;
  398. cur_collate->overrides_tbl = __locale_collate_tbl + n + cdd->overrides_offset;
  399. /* fprintf(stderr,"ovr = %d\n", n + cdd->overrides_offset); */
  400. n += cdh->num_override;
  401. cur_collate->ruletable = __locale_collate_tbl + n;
  402. /* fprintf(stderr, "rtb = %d\n", n); */
  403. n += cdh->num_ruletable;
  404. cur_collate->weightstr = __locale_collate_tbl + n;
  405. /* fprintf(stderr,"wts = %d\n", n); */
  406. n += cdh->num_weightstr;
  407. cur_collate->wcs2colidt_tbl = __locale_collate_tbl + n
  408. + (((unsigned long)(cdb->wcs2colidt_offset_hi)) << 16)
  409. + cdb->wcs2colidt_offset_low;
  410. /* fprintf(stderr,"wcs = %lu\n", n + (((unsigned long)(cdb->wcs2colidt_offset_hi)) << 16) */
  411. /* + cdb->wcs2colidt_offset_low); */
  412. cur_collate->MAX_WEIGHTS = cdh->MAX_WEIGHTS;
  413. #ifdef __UCLIBC_MJN3_ONLY__
  414. #warning CONSIDER: Fix the +1 by increasing max_col_index?
  415. #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
  416. #endif
  417. cur_collate->index2weight = calloc(2*cur_collate->max_col_index+2,
  418. sizeof(uint16_t));
  419. if (!cur_collate->index2weight) {
  420. return 0;
  421. }
  422. cur_collate->index2ruleidx = cur_collate->index2weight
  423. + cur_collate->max_col_index + 1;
  424. memcpy(cur_collate->index2weight, cur_collate->index2weight_tbl,
  425. cur_collate->num_col_base * sizeof(uint16_t));
  426. memcpy(cur_collate->index2ruleidx, cur_collate->index2ruleidx_tbl,
  427. cur_collate->num_col_base * sizeof(uint16_t));
  428. /* now do the overrides */
  429. p = cur_collate->overrides_tbl;
  430. while (*p > 1) {
  431. /* fprintf(stderr, "processing override -- count = %d\n", *p); */
  432. n = *p++;
  433. w = *p++;
  434. do {
  435. i = *p++;
  436. /* fprintf(stderr, " i=%d (%#x) w=%d *p=%d\n", i, i, w, *p); */
  437. cur_collate->index2weight[i-1] = w++;
  438. cur_collate->index2ruleidx[i-1] = *p++;
  439. } while (--n);
  440. }
  441. assert(*p == 1);
  442. while (*++p) {
  443. i = *p;
  444. /* fprintf(stderr, " i=%d (%#x) w=%d *p=%d\n", i, i, p[1], p[2]); */
  445. cur_collate->index2weight[i-1] = *++p;
  446. cur_collate->index2ruleidx[i-1] = *++p;
  447. }
  448. for (i=0 ; i < cur_collate->multistart_tbl[0] ; i++) {
  449. p = cur_collate->multistart_tbl;
  450. /* fprintf(stderr, "%2d of %2d: %d ", i, cur_collate->multistart_tbl[0], p[i]); */
  451. p += p[i];
  452. do {
  453. n = *p++;
  454. do {
  455. if (!*p) { /* found it */
  456. /* fprintf(stderr, "found: n=%d (%#lx) |%.*ls|\n", n, (int) *cs->s, n, cs->s); */
  457. /* fprintf(stderr, ": %d - single\n", n); */
  458. goto FOUND;
  459. }
  460. /* the lookup check here is safe since we're assured that *p is a valid colidex */
  461. /* fprintf(stderr, "lookup(%lc)==%d *p==%d\n", cs->s[n], lookup(cs->s[n]), (int) *p); */
  462. /* fprintf(stderr, ": %d - ", n); */
  463. do {
  464. /* fprintf(stderr, "%d|", *p); */
  465. } while (*p++);
  466. break;
  467. } while (1);
  468. } while (1);
  469. FOUND:
  470. continue;
  471. }
  472. return 1;
  473. }
  474. int attribute_hidden _locale_set_l(const unsigned char *p, __locale_t base)
  475. {
  476. const char **x;
  477. unsigned char *s = base->cur_locale + 1;
  478. const size_t *stp;
  479. const unsigned char *r;
  480. const uint16_t *io;
  481. const uint16_t *ii;
  482. const unsigned char *d;
  483. int row; /* locale row */
  484. int crow; /* category row */
  485. int len;
  486. int c;
  487. int i = 0;
  488. __collate_t newcol;
  489. ++p;
  490. newcol.index2weight = NULL;
  491. if ((p[2*LC_COLLATE] != s[2*LC_COLLATE])
  492. || (p[2*LC_COLLATE + 1] != s[2*LC_COLLATE + 1])
  493. ) {
  494. row = (((int)(*p & 0x7f)) << 7) + (p[1] & 0x7f);
  495. assert(row < __LOCALE_DATA_NUM_LOCALES);
  496. if (!init_cur_collate(__locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES
  497. * row + 3 + LC_COLLATE ],
  498. &newcol)
  499. ) {
  500. return 0; /* calloc failed. */
  501. }
  502. free(base->collate.index2weight);
  503. memcpy(&base->collate, &newcol, sizeof(__collate_t));
  504. }
  505. do {
  506. if ((*p != *s) || (p[1] != s[1])) {
  507. row = (((int)(*p & 0x7f)) << 7) + (p[1] & 0x7f);
  508. assert(row < __LOCALE_DATA_NUM_LOCALES);
  509. *s = *p;
  510. s[1] = p[1];
  511. if ((i != LC_COLLATE)
  512. && ((len = __locale_mmap->lc_common_item_offsets_LEN[i]) != 0)
  513. ) {
  514. crow = __locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES * row
  515. + 3 + i ]
  516. * len;
  517. x = (const char **)(((char *) base)
  518. + base->category_offsets[i]);
  519. stp = __locale_mmap->lc_common_tbl_offsets + 4*i;
  520. r = (const unsigned char *)( ((char *)__locale_mmap) + *stp );
  521. io = (const uint16_t *)( ((char *)__locale_mmap) + *++stp );
  522. ii = (const uint16_t *)( ((char *)__locale_mmap) + *++stp );
  523. d = (const unsigned char *)( ((char *)__locale_mmap) + *++stp );
  524. for (c = 0; c < len; c++) {
  525. x[c] = (char*)(d + ii[r[crow + c] + io[c]]);
  526. }
  527. }
  528. if (i == LC_CTYPE) {
  529. c = __locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES * row
  530. + 2 ]; /* codeset */
  531. if (c <= 2) {
  532. if (c == 2) {
  533. base->codeset = utf8;
  534. base->encoding = __ctype_encoding_utf8;
  535. /* TODO - fix for bcc */
  536. base->mb_cur_max = 6;
  537. } else {
  538. assert(c == 1);
  539. base->codeset = ascii;
  540. base->encoding = __ctype_encoding_7_bit;
  541. base->mb_cur_max = 1;
  542. }
  543. } else {
  544. const __codeset_8_bit_t *c8b;
  545. r = CODESET_LIST;
  546. c -= 3;
  547. base->codeset = (char *) (r + r[c]);
  548. base->encoding = __ctype_encoding_8_bit;
  549. #ifdef __UCLIBC_MJN3_ONLY__
  550. #warning REMINDER: update 8 bit mb_cur_max when translit implemented!
  551. #endif
  552. /* TODO - update when translit implemented! */
  553. base->mb_cur_max = 1;
  554. c8b = __locale_mmap->codeset_8_bit + c;
  555. #ifdef __CTYPE_HAS_8_BIT_LOCALES
  556. base->idx8ctype = c8b->idx8ctype;
  557. base->idx8uplow = c8b->idx8uplow;
  558. #ifdef __UCLIBC_HAS_WCHAR__
  559. base->idx8c2wc = c8b->idx8c2wc;
  560. base->idx8wc2c = c8b->idx8wc2c;
  561. /* translit */
  562. #endif /* __UCLIBC_HAS_WCHAR__ */
  563. /* What follows is fairly bloated, but it is just a hack
  564. * to get the 8-bit codeset ctype stuff functioning.
  565. * All of this will be replaced in the next generation
  566. * of locale support anyway... */
  567. memcpy(base->__ctype_b_data,
  568. __C_ctype_b - __UCLIBC_CTYPE_B_TBL_OFFSET,
  569. (256 + __UCLIBC_CTYPE_B_TBL_OFFSET)
  570. * sizeof(__ctype_mask_t));
  571. memcpy(base->__ctype_tolower_data,
  572. __C_ctype_tolower - __UCLIBC_CTYPE_TO_TBL_OFFSET,
  573. (256 + __UCLIBC_CTYPE_TO_TBL_OFFSET)
  574. * sizeof(__ctype_touplow_t));
  575. memcpy(base->__ctype_toupper_data,
  576. __C_ctype_toupper - __UCLIBC_CTYPE_TO_TBL_OFFSET,
  577. (256 + __UCLIBC_CTYPE_TO_TBL_OFFSET)
  578. * sizeof(__ctype_touplow_t));
  579. #define Cctype_TBL_MASK ((1 << __LOCALE_DATA_Cctype_IDX_SHIFT) - 1)
  580. #define Cctype_IDX_OFFSET (128 >> __LOCALE_DATA_Cctype_IDX_SHIFT)
  581. {
  582. int u;
  583. __ctype_mask_t m;
  584. for (u=0 ; u < 128 ; u++) {
  585. #ifdef __LOCALE_DATA_Cctype_PACKED
  586. c = base->tbl8ctype
  587. [ ((int)(c8b->idx8ctype
  588. [(u >> __LOCALE_DATA_Cctype_IDX_SHIFT) ])
  589. << (__LOCALE_DATA_Cctype_IDX_SHIFT - 1))
  590. + ((u & Cctype_TBL_MASK) >> 1)];
  591. c = (u & 1) ? (c >> 4) : (c & 0xf);
  592. #else
  593. c = base->tbl8ctype
  594. [ ((int)(c8b->idx8ctype
  595. [(u >> __LOCALE_DATA_Cctype_IDX_SHIFT) ])
  596. << __LOCALE_DATA_Cctype_IDX_SHIFT)
  597. + (u & Cctype_TBL_MASK) ];
  598. #endif
  599. m = base->code2flag[c];
  600. base->__ctype_b_data
  601. [128 + __UCLIBC_CTYPE_B_TBL_OFFSET + u]
  602. = m;
  603. #ifdef __UCLIBC_HAS_CTYPE_SIGNED__
  604. if (((signed char)(128 + u)) != -1) {
  605. base->__ctype_b_data[__UCLIBC_CTYPE_B_TBL_OFFSET
  606. + ((signed char)(128 + u))]
  607. = m;
  608. }
  609. #endif
  610. base->__ctype_tolower_data
  611. [128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
  612. = 128 + u;
  613. base->__ctype_toupper_data
  614. [128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
  615. = 128 + u;
  616. if (m & (_ISlower|_ISupper)) {
  617. c = base->tbl8uplow
  618. [ ((int)(c8b->idx8uplow
  619. [u >> __LOCALE_DATA_Cuplow_IDX_SHIFT])
  620. << __LOCALE_DATA_Cuplow_IDX_SHIFT)
  621. + ((128 + u)
  622. & ((1 << __LOCALE_DATA_Cuplow_IDX_SHIFT)
  623. - 1)) ];
  624. if (m & _ISlower) {
  625. base->__ctype_toupper_data
  626. [128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
  627. = (unsigned char)(128 + u + c);
  628. #ifdef __UCLIBC_HAS_CTYPE_SIGNED__
  629. if (((signed char)(128 + u)) != -1) {
  630. base->__ctype_toupper_data
  631. [__UCLIBC_CTYPE_TO_TBL_OFFSET
  632. + ((signed char)(128 + u))]
  633. = (unsigned char)(128 + u + c);
  634. }
  635. #endif
  636. } else {
  637. base->__ctype_tolower_data
  638. [128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
  639. = (unsigned char)(128 + u - c);
  640. #ifdef __UCLIBC_HAS_CTYPE_SIGNED__
  641. if (((signed char)(128 + u)) != -1) {
  642. base->__ctype_tolower_data
  643. [__UCLIBC_CTYPE_TO_TBL_OFFSET
  644. + ((signed char)(128 + u))]
  645. = (unsigned char)(128 + u - c);
  646. }
  647. #endif
  648. }
  649. }
  650. }
  651. }
  652. #ifdef __UCLIBC_HAS_XLOCALE__
  653. base->__ctype_b = base->__ctype_b_data
  654. + __UCLIBC_CTYPE_B_TBL_OFFSET;
  655. base->__ctype_tolower = base->__ctype_tolower_data
  656. + __UCLIBC_CTYPE_TO_TBL_OFFSET;
  657. base->__ctype_toupper = base->__ctype_toupper_data
  658. + __UCLIBC_CTYPE_TO_TBL_OFFSET;
  659. #else /* __UCLIBC_HAS_XLOCALE__ */
  660. __ctype_b = base->__ctype_b_data
  661. + __UCLIBC_CTYPE_B_TBL_OFFSET;
  662. __ctype_tolower = base->__ctype_tolower_data
  663. + __UCLIBC_CTYPE_TO_TBL_OFFSET;
  664. __ctype_toupper = base->__ctype_toupper_data
  665. + __UCLIBC_CTYPE_TO_TBL_OFFSET;
  666. #endif /* __UCLIBC_HAS_XLOCALE__ */
  667. #endif /* __CTYPE_HAS_8_BIT_LOCALES */
  668. }
  669. #ifdef __UCLIBC_MJN3_ONLY__
  670. #warning TODO: Put the outdigit string length in the locale_mmap object.
  671. #endif
  672. d = base->outdigit_length;
  673. x = &base->outdigit0_mb;
  674. for (c = 0 ; c < 10 ; c++) {
  675. ((unsigned char *)d)[c] = strlen(x[c]);
  676. assert(d[c] > 0);
  677. }
  678. } else if (i == LC_NUMERIC) {
  679. assert(LC_NUMERIC > LC_CTYPE); /* Need ctype initialized. */
  680. base->decimal_point_len
  681. = __locale_mbrtowc_l(&base->decimal_point_wc,
  682. base->decimal_point, base);
  683. assert(base->decimal_point_len > 0);
  684. assert(base->decimal_point[base->decimal_point_len] == 0);
  685. if (*base->grouping) {
  686. base->thousands_sep_len
  687. = __locale_mbrtowc_l(&base->thousands_sep_wc,
  688. base->thousands_sep, base);
  689. #if 1
  690. #ifdef __UCLIBC_MJN3_ONLY__
  691. #warning TODO: Remove hack involving grouping without a thousep char (bg_BG).
  692. #endif
  693. assert(base->thousands_sep_len >= 0);
  694. if (base->thousands_sep_len == 0) {
  695. base->grouping = base->thousands_sep; /* empty string */
  696. }
  697. assert(base->thousands_sep[base->thousands_sep_len] == 0);
  698. #else
  699. assert(base->thousands_sep_len > 0);
  700. assert(base->thousands_sep[base->thousands_sep_len] == 0);
  701. #endif
  702. }
  703. /* } else if (i == LC_COLLATE) { */
  704. /* init_cur_collate(__locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES */
  705. /* * row + 3 + i ], */
  706. /* &base->collate); */
  707. }
  708. }
  709. ++i;
  710. p += 2;
  711. s += 2;
  712. } while (i < LC_ALL);
  713. return 1;
  714. }
  715. static const uint16_t __code2flag[16] = {
  716. 0, /* unclassified = 0 */
  717. _ISprint|_ISgraph|_ISalnum|_ISalpha, /* alpha_nonupper_nonlower */
  718. _ISprint|_ISgraph|_ISalnum|_ISalpha|_ISlower, /* alpha_lower */
  719. _ISprint|_ISgraph|_ISalnum|_ISalpha|_ISlower|_ISupper, /* alpha_upper_lower */
  720. _ISprint|_ISgraph|_ISalnum|_ISalpha|_ISupper, /* alpha_upper */
  721. _ISprint|_ISgraph|_ISalnum|_ISdigit, /* digit */
  722. _ISprint|_ISgraph|_ISpunct, /* punct */
  723. _ISprint|_ISgraph, /* graph */
  724. _ISprint|_ISspace, /* print_space_nonblank */
  725. _ISprint|_ISspace|_ISblank, /* print_space_blank */
  726. _ISspace, /* space_nonblank_noncntrl */
  727. _ISspace|_ISblank, /* space_blank_noncntrl */
  728. _IScntrl|_ISspace, /* cntrl_space_nonblank */
  729. _IScntrl|_ISspace|_ISblank, /* cntrl_space_blank */
  730. _IScntrl /* cntrl_nonspace */
  731. };
  732. void attribute_hidden _locale_init_l(__locale_t base)
  733. {
  734. memset(base->cur_locale, 0, LOCALE_SELECTOR_SIZE);
  735. base->cur_locale[0] = '#';
  736. memcpy(base->category_item_count,
  737. __locale_mmap->lc_common_item_offsets_LEN,
  738. LC_ALL);
  739. ++base->category_item_count[0]; /* Increment for codeset entry. */
  740. base->category_offsets[0] = offsetof(struct __uclibc_locale_struct, outdigit0_mb);
  741. base->category_offsets[1] = offsetof(struct __uclibc_locale_struct, decimal_point);
  742. base->category_offsets[2] = offsetof(struct __uclibc_locale_struct, int_curr_symbol);
  743. base->category_offsets[3] = offsetof(struct __uclibc_locale_struct, abday_1);
  744. /* base->category_offsets[4] = offsetof(struct __uclibc_locale_struct, collate???); */
  745. base->category_offsets[5] = offsetof(struct __uclibc_locale_struct, yesexpr);
  746. #ifdef __CTYPE_HAS_8_BIT_LOCALES
  747. base->tbl8ctype
  748. = (const unsigned char *) &__locale_mmap->tbl8ctype;
  749. base->tbl8uplow
  750. = (const unsigned char *) &__locale_mmap->tbl8uplow;
  751. #ifdef __UCLIBC_HAS_WCHAR__
  752. base->tbl8c2wc
  753. = (const uint16_t *) &__locale_mmap->tbl8c2wc;
  754. base->tbl8wc2c
  755. = (const unsigned char *) &__locale_mmap->tbl8wc2c;
  756. /* translit */
  757. #endif /* __UCLIBC_HAS_WCHAR__ */
  758. #endif /* __CTYPE_HAS_8_BIT_LOCALES */
  759. #ifdef __UCLIBC_HAS_WCHAR__
  760. base->tblwctype
  761. = (const unsigned char *) &__locale_mmap->tblwctype;
  762. base->tblwuplow
  763. = (const unsigned char *) &__locale_mmap->tblwuplow;
  764. base->tblwuplow_diff
  765. = (const int16_t *) &__locale_mmap->tblwuplow_diff;
  766. /* base->tblwcomb */
  767. /* = (const unsigned char *) &__locale_mmap->tblwcomb; */
  768. /* width?? */
  769. #endif /* __UCLIBC_HAS_WCHAR__ */
  770. /* Initially, set things up to use the global C ctype tables.
  771. * This is correct for C (ASCII) and UTF-8 based locales (except tr_TR). */
  772. #ifdef __UCLIBC_HAS_XLOCALE__
  773. base->__ctype_b = __C_ctype_b;
  774. base->__ctype_tolower = __C_ctype_tolower;
  775. base->__ctype_toupper = __C_ctype_toupper;
  776. #else /* __UCLIBC_HAS_XLOCALE__ */
  777. __ctype_b = __C_ctype_b;
  778. __ctype_tolower = __C_ctype_tolower;
  779. __ctype_toupper = __C_ctype_toupper;
  780. #endif /* __UCLIBC_HAS_XLOCALE__ */
  781. #ifdef __UCLIBC_MJN3_ONLY__
  782. #warning TODO: Initialize code2flag correctly based on locale_mmap.
  783. #endif
  784. base->code2flag = __code2flag;
  785. _locale_set_l((unsigned char*) C_LOCALE_SELECTOR, base);
  786. }
  787. void _locale_init(void) attribute_hidden;
  788. void _locale_init(void)
  789. {
  790. /* TODO: mmap the locale file */
  791. /* TODO - ??? */
  792. _locale_init_l(__global_locale);
  793. }
  794. #endif
  795. /**********************************************************************/
  796. #if defined(L_nl_langinfo) || defined(L_nl_langinfo_l)
  797. #ifdef __LOCALE_C_ONLY
  798. /* We need to index 320 bytes of data, so you might initially think we
  799. * need to store the offsets in shorts. But since the offset of the
  800. * 64th item is 182, we'll store "offset - 2*64" for all items >= 64
  801. * and always calculate the data offset as "offset[i] + 2*(i & 64)".
  802. * This allows us to pack the data offsets in an unsigned char while
  803. * also avoiding an "if".
  804. *
  805. * Note: Category order is assumed to be:
  806. * ctype, numeric, monetary, time, collate, messages, all
  807. */
  808. #define C_LC_ALL 6
  809. /* Combine the data to avoid size penalty for seperate char arrays when
  810. * compiler aligns objects. The original code is left in as documentation. */
  811. #define cat_start nl_data
  812. #define C_locale_data (nl_data + C_LC_ALL + 1 + 90)
  813. static const unsigned char nl_data[C_LC_ALL + 1 + 90 + 320] = {
  814. /* static const char cat_start[LC_ALL + 1] = { */
  815. '\x00', '\x0b', '\x0e', '\x24', '\x56', '\x56', '\x5a',
  816. /* }; */
  817. /* static const char item_offset[90] = { */
  818. '\x00', '\x02', '\x04', '\x06', '\x08', '\x0a', '\x0c', '\x0e',
  819. '\x10', '\x12', '\x14', '\x1a', '\x1b', '\x1b', '\x1b', '\x1b',
  820. '\x1b', '\x1b', '\x1b', '\x1b', '\x1b', '\x1c', '\x1c', '\x1c',
  821. '\x1c', '\x1c', '\x1c', '\x1c', '\x1c', '\x1c', '\x1c', '\x1c',
  822. '\x1c', '\x1c', '\x1c', '\x1e', '\x20', '\x24', '\x28', '\x2c',
  823. '\x30', '\x34', '\x38', '\x3c', '\x43', '\x4a', '\x52', '\x5c',
  824. '\x65', '\x6c', '\x75', '\x79', '\x7d', '\x81', '\x85', '\x89',
  825. '\x8d', '\x91', '\x95', '\x99', '\x9d', '\xa1', '\xa5', '\xad',
  826. '\x36', '\x3c', '\x42', '\x46', '\x4b', '\x50', '\x57', '\x61',
  827. '\x69', '\x72', '\x7b', '\x7e', '\x81', '\x96', '\x9f', '\xa8',
  828. '\xb3', '\xb3', '\xb3', '\xb3', '\xb3', '\xb3', '\xb4', '\xba',
  829. '\xbf', '\xbf',
  830. /* }; */
  831. /* static const char C_locale_data[320] = { */
  832. '0', '\x00', '1', '\x00', '2', '\x00', '3', '\x00',
  833. '4', '\x00', '5', '\x00', '6', '\x00', '7', '\x00',
  834. '8', '\x00', '9', '\x00', 'A', 'S', 'C', 'I',
  835. 'I', '\x00', '.', '\x00', '\x7f', '\x00', '-', '\x00',
  836. 'S', 'u', 'n', '\x00', 'M', 'o', 'n', '\x00',
  837. 'T', 'u', 'e', '\x00', 'W', 'e', 'd', '\x00',
  838. 'T', 'h', 'u', '\x00', 'F', 'r', 'i', '\x00',
  839. 'S', 'a', 't', '\x00', 'S', 'u', 'n', 'd',
  840. 'a', 'y', '\x00', 'M', 'o', 'n', 'd', 'a',
  841. 'y', '\x00', 'T', 'u', 'e', 's', 'd', 'a',
  842. 'y', '\x00', 'W', 'e', 'd', 'n', 'e', 's',
  843. 'd', 'a', 'y', '\x00', 'T', 'h', 'u', 'r',
  844. 's', 'd', 'a', 'y', '\x00', 'F', 'r', 'i',
  845. 'd', 'a', 'y', '\x00', 'S', 'a', 't', 'u',
  846. 'r', 'd', 'a', 'y', '\x00', 'J', 'a', 'n',
  847. '\x00', 'F', 'e', 'b', '\x00', 'M', 'a', 'r',
  848. '\x00', 'A', 'p', 'r', '\x00', 'M', 'a', 'y',
  849. '\x00', 'J', 'u', 'n', '\x00', 'J', 'u', 'l',
  850. '\x00', 'A', 'u', 'g', '\x00', 'S', 'e', 'p',
  851. '\x00', 'O', 'c', 't', '\x00', 'N', 'o', 'v',
  852. '\x00', 'D', 'e', 'c', '\x00', 'J', 'a', 'n',
  853. 'u', 'a', 'r', 'y', '\x00', 'F', 'e', 'b',
  854. 'r', 'u', 'a', 'r', 'y', '\x00', 'M', 'a',
  855. 'r', 'c', 'h', '\x00', 'A', 'p', 'r', 'i',
  856. 'l', '\x00', 'M', 'a', 'y', '\x00', 'J', 'u',
  857. 'n', 'e', '\x00', 'J', 'u', 'l', 'y', '\x00',
  858. 'A', 'u', 'g', 'u', 's', 't', '\x00', 'S',
  859. 'e', 'p', 't', 'e', 'm', 'b', 'e', 'r',
  860. '\x00', 'O', 'c', 't', 'o', 'b', 'e', 'r',
  861. '\x00', 'N', 'o', 'v', 'e', 'm', 'b', 'e',
  862. 'r', '\x00', 'D', 'e', 'c', 'e', 'm', 'b',
  863. 'e', 'r', '\x00', 'A', 'M', '\x00', 'P', 'M',
  864. '\x00', '%', 'a', ' ', '%', 'b', ' ', '%',
  865. 'e', ' ', '%', 'H', ':', '%', 'M', ':',
  866. '%', 'S', ' ', '%', 'Y', '\x00', '%', 'm',
  867. '/', '%', 'd', '/', '%', 'y', '\x00', '%',
  868. 'H', ':', '%', 'M', ':', '%', 'S', '\x00',
  869. '%', 'I', ':', '%', 'M', ':', '%', 'S',
  870. ' ', '%', 'p', '\x00', '^', '[', 'y', 'Y',
  871. ']', '\x00', '^', '[', 'n', 'N', ']', '\x00',
  872. };
  873. char *nl_langinfo(nl_item item)
  874. {
  875. unsigned int c;
  876. unsigned int i;
  877. if ((c = _NL_ITEM_CATEGORY(item)) < C_LC_ALL) {
  878. if ((i = cat_start[c] + _NL_ITEM_INDEX(item)) < cat_start[c+1]) {
  879. /* return (char *) C_locale_data + item_offset[i] + (i & 64); */
  880. return (char *) C_locale_data + nl_data[C_LC_ALL+1+i] + 2*(i & 64);
  881. }
  882. }
  883. return (char *) cat_start; /* Conveniently, this is the empty string. */
  884. }
  885. libc_hidden_def(nl_langinfo)
  886. #else /* __LOCALE_C_ONLY */
  887. #if defined(__UCLIBC_HAS_XLOCALE__) && !defined(__UCLIBC_DO_XLOCALE)
  888. char *nl_langinfo(nl_item item)
  889. {
  890. return nl_langinfo_l(item, __UCLIBC_CURLOCALE);
  891. }
  892. libc_hidden_def(nl_langinfo)
  893. #else /* defined(__UCLIBC_HAS_XLOCALE__) && !defined(__UCLIBC_DO_XLOCALE) */
  894. libc_hidden_proto(__XL_NPP(nl_langinfo))
  895. static const char empty[] = "";
  896. char *__XL_NPP(nl_langinfo)(nl_item item __LOCALE_PARAM )
  897. {
  898. unsigned int c = _NL_ITEM_CATEGORY(item);
  899. unsigned int i = _NL_ITEM_INDEX(item);
  900. if ((c < LC_ALL) && (i < __LOCALE_PTR->category_item_count[c])) {
  901. return ((char **)(((char *) __LOCALE_PTR)
  902. + __LOCALE_PTR->category_offsets[c]))[i];
  903. }
  904. return (char *) empty;
  905. }
  906. libc_hidden_def(__XL_NPP(nl_langinfo))
  907. #endif /* defined(__UCLIBC_HAS_XLOCALE__) && !defined(__UCLIBC_DO_XLOCALE) */
  908. #endif /* __LOCALE_C_ONLY */
  909. #endif
  910. /**********************************************************************/
  911. #ifdef L_newlocale
  912. #ifdef __UCLIBC_MJN3_ONLY__
  913. #warning TODO: Move posix and utf8 strings.
  914. #endif
  915. static const char posix[] = "POSIX";
  916. static const char utf8[] = "UTF-8";
  917. static int find_locale(int category_mask, const char *p,
  918. unsigned char *new_locale)
  919. {
  920. int i;
  921. const unsigned char *s;
  922. uint16_t n;
  923. unsigned char lang_cult, codeset;
  924. #if defined(__LOCALE_DATA_AT_MODIFIERS_LENGTH) && 1
  925. /* Support standard locale handling for @-modifiers. */
  926. #ifdef __UCLIBC_MJN3_ONLY__
  927. #warning REMINDER: Fix buf size in find_locale.
  928. #endif
  929. char buf[18]; /* TODO: 7+{max codeset name length} */
  930. const char *q;
  931. if ((q = strchr(p,'@')) != NULL) {
  932. if ((((size_t)((q-p)-5)) > (sizeof(buf) - 5)) || (p[2] != '_')) {
  933. return 0;
  934. }
  935. /* locale name at least 5 chars long and 3rd char is '_' */
  936. s = LOCALE_AT_MODIFIERS;
  937. do {
  938. if (!strcmp((char*) (s + 2), q + 1)) {
  939. break;
  940. }
  941. s += 2 + *s; /* TODO - fix this throughout */
  942. } while (*s);
  943. if (!*s) {
  944. return 0;
  945. }
  946. assert(q - p < sizeof(buf));
  947. memcpy(buf, p, q-p);
  948. buf[q-p] = 0;
  949. buf[2] = s[1];
  950. p = buf;
  951. }
  952. #endif
  953. lang_cult = codeset = 0; /* Assume C and default codeset. */
  954. if (((*p == 'C') && !p[1]) || !strcmp(p, posix)) {
  955. goto FIND_LOCALE;
  956. }
  957. if ((strlen(p) > 5) && (p[5] == '.')) { /* Codeset in locale name? */
  958. /* TODO: maybe CODESET_LIST + *s ??? */
  959. /* 7bit is 1, UTF-8 is 2, 8-bit is >= 3 */
  960. codeset = 2;
  961. if (strcasecmp(utf8, p + 6) != 0) {/* TODO - fix! */
  962. s = CODESET_LIST;
  963. do {
  964. ++codeset; /* Increment codeset first. */
  965. if (!strcmp((char*) CODESET_LIST + *s, p + 6)) {
  966. goto FIND_LANG_CULT;
  967. }
  968. } while (*++s);
  969. return 0; /* No matching codeset! */
  970. }
  971. }
  972. FIND_LANG_CULT: /* Find language_culture number. */
  973. s = LOCALE_NAMES;
  974. do { /* TODO -- do a binary search? */
  975. /* TODO -- fix gen_mmap!*/
  976. ++lang_cult; /* Increment first since C/POSIX is 0. */
  977. if (!strncmp((char*) s, p, 5)) { /* Found a matching locale name; */
  978. goto FIND_LOCALE;
  979. }
  980. s += 5;
  981. } while (lang_cult < __LOCALE_DATA_NUM_LOCALE_NAMES);
  982. return 0; /* No matching language_culture! */
  983. FIND_LOCALE: /* Find locale row matching name and codeset */
  984. s = LOCALES;
  985. n = 0;
  986. do { /* TODO -- do a binary search? */
  987. if ((lang_cult == *s) && ((codeset == s[1]) || (codeset == s[2]))) {
  988. i = 1;
  989. s = new_locale + 1;
  990. do {
  991. if (category_mask & i) {
  992. /* Encode current locale row number. */
  993. ((unsigned char *) s)[0] = (n >> 7) | 0x80;
  994. ((unsigned char *) s)[1] = (n & 0x7f) | 0x80;
  995. }
  996. s += 2;
  997. i += i;
  998. } while (i < (1 << LC_ALL));
  999. return i; /* Return non-zero */
  1000. }
  1001. s += __LOCALE_DATA_WIDTH_LOCALES;
  1002. ++n;
  1003. } while (n <= __LOCALE_DATA_NUM_LOCALES); /* We started at 1!!! */
  1004. return 0; /* Unsupported locale. */
  1005. }
  1006. static unsigned char *composite_locale(int category_mask, const char *locale,
  1007. unsigned char *new_locale)
  1008. {
  1009. char buf[MAX_LOCALE_STR];
  1010. char *t;
  1011. char *e;
  1012. int c;
  1013. int component_mask;
  1014. if (!strchr(locale,'=')) {
  1015. if (!find_locale(category_mask, locale, new_locale)) {
  1016. return NULL;
  1017. }
  1018. return new_locale;
  1019. }
  1020. if (strlen(locale) >= sizeof(buf)) {
  1021. return NULL;
  1022. }
  1023. stpcpy(buf, locale);
  1024. component_mask = 0;
  1025. t = strtok_r(buf, "=", &e); /* This can't fail because of strchr test above. */
  1026. do {
  1027. c = 0;
  1028. /* CATEGORY_NAMES is unsigned char* */
  1029. while (strcmp((char*) CATEGORY_NAMES + (int) CATEGORY_NAMES[c], t)) {
  1030. if (++c == LC_ALL) { /* Unknown category name! */
  1031. return NULL;
  1032. }
  1033. }
  1034. t = strtok_r(NULL, ";", &e);
  1035. c = (1 << c);
  1036. if (component_mask & c) { /* Multiple components for one category. */
  1037. return NULL;
  1038. }
  1039. component_mask |= c;
  1040. if ((category_mask & c) && (!t || !find_locale(c, t, new_locale))) {
  1041. return NULL;
  1042. }
  1043. } while ((t = strtok_r(NULL, "=", &e)) != NULL);
  1044. if (category_mask & ~component_mask) { /* Category component(s) missing. */
  1045. return NULL;
  1046. }
  1047. return new_locale;
  1048. }
  1049. __locale_t newlocale(int category_mask, const char *locale, __locale_t base)
  1050. {
  1051. const char *p;
  1052. int i, j, k;
  1053. unsigned char new_selector[LOCALE_SELECTOR_SIZE];
  1054. if (category_mask == (1 << LC_ALL)) {
  1055. category_mask = LC_ALL_MASK;
  1056. }
  1057. if (!locale || ((unsigned)(category_mask) > LC_ALL_MASK)) {
  1058. INVALID:
  1059. __set_errno(EINVAL);
  1060. return NULL; /* No locale or illegal/unsupported category. */
  1061. }
  1062. #ifdef __UCLIBC_MJN3_ONLY__
  1063. #warning TODO: Rename cur_locale to locale_selector.
  1064. #endif
  1065. strcpy((char *) new_selector,
  1066. (base ? (char *) base->cur_locale : C_LOCALE_SELECTOR));
  1067. if (!locale[0]) { /* locale == "", so check environment. */
  1068. const char *envstr[4];
  1069. envstr[0] = "LC_ALL";
  1070. envstr[1] = NULL;
  1071. envstr[2] = "LANG";
  1072. envstr[3] = posix;
  1073. i = 1;
  1074. k = 0;
  1075. do {
  1076. if (category_mask & i) {
  1077. /* Note: SUSv3 doesn't define a fallback mechanism here.
  1078. * So, if LC_ALL is invalid, we do _not_ continue trying
  1079. * the other environment vars. */
  1080. envstr[1] = (char*) CATEGORY_NAMES + CATEGORY_NAMES[k];
  1081. j = 0;
  1082. while (1) {
  1083. p = envstr[j];
  1084. if (++j >= 4)
  1085. break; /* now p == "POSIX" */
  1086. p = getenv(p);
  1087. if (p && p[0])
  1088. break;
  1089. };
  1090. /* The user set something... is it valid? */
  1091. /* Note: Since we don't support user-supplied locales and
  1092. * alternate paths, we don't need to worry about special
  1093. * handling for suid/sgid apps. */
  1094. if (!find_locale(i, p, new_selector)) {
  1095. goto INVALID;
  1096. }
  1097. }
  1098. i += i;
  1099. } while (++k < LC_ALL);
  1100. } else if (!composite_locale(category_mask, locale, new_selector)) {
  1101. goto INVALID;
  1102. }
  1103. #ifdef __UCLIBC_MJN3_ONLY__
  1104. #warning TODO: Do a compatible codeset check!
  1105. #endif
  1106. /* If we get here, the new selector corresponds to a valid locale. */
  1107. #ifdef __UCLIBC_MJN3_ONLY__
  1108. #warning CONSIDER: Probably want a _locale_new func to allow for caching of locales.
  1109. #endif
  1110. #if 0
  1111. if (base) {
  1112. _locale_set_l(new_selector, base);
  1113. } else {
  1114. base = _locale_new(new_selector);
  1115. }
  1116. #else
  1117. if (!base) {
  1118. base = malloc(sizeof(struct __uclibc_locale_struct));
  1119. if (base == NULL)
  1120. return base;
  1121. _locale_init_l(base);
  1122. }
  1123. _locale_set_l(new_selector, base);
  1124. #endif
  1125. return base;
  1126. }
  1127. libc_hidden_def(newlocale)
  1128. #endif
  1129. /**********************************************************************/
  1130. #ifdef L_duplocale
  1131. #ifdef __UCLIBC_MJN3_ONLY__
  1132. #warning REMINDER: When we allocate ctype tables, remember to dup them.
  1133. #endif
  1134. __locale_t duplocale(__locale_t dataset)
  1135. {
  1136. __locale_t r;
  1137. uint16_t * i2w;
  1138. size_t n;
  1139. assert(dataset != LC_GLOBAL_LOCALE);
  1140. r = malloc(sizeof(struct __uclibc_locale_struct));
  1141. if (r != NULL) {
  1142. n = 2 * dataset->collate.max_col_index + 2;
  1143. i2w = calloc(n, sizeof(uint16_t));
  1144. if (i2w != NULL) {
  1145. memcpy(r, dataset, sizeof(struct __uclibc_locale_struct));
  1146. r->collate.index2weight = i2w;
  1147. memcpy(i2w, dataset->collate.index2weight, n * sizeof(uint16_t));
  1148. } else {
  1149. free(r);
  1150. r = NULL;
  1151. }
  1152. }
  1153. return r;
  1154. }
  1155. libc_hidden_def(duplocale)
  1156. #endif
  1157. /**********************************************************************/
  1158. #ifdef L_freelocale
  1159. #ifdef __UCLIBC_MJN3_ONLY__
  1160. #warning REMINDER: When we allocate ctype tables, remember to free them.
  1161. #endif
  1162. void freelocale(__locale_t dataset)
  1163. {
  1164. assert(dataset != __global_locale);
  1165. assert(dataset != LC_GLOBAL_LOCALE);
  1166. free(dataset->collate.index2weight); /* Free collation data. */
  1167. free(dataset); /* Free locale */
  1168. }
  1169. #endif
  1170. /**********************************************************************/
  1171. #ifdef L_uselocale
  1172. __locale_t uselocale(__locale_t dataset)
  1173. {
  1174. __locale_t old;
  1175. if (!dataset) {
  1176. old = __UCLIBC_CURLOCALE;
  1177. } else {
  1178. if (dataset == LC_GLOBAL_LOCALE) {
  1179. dataset = __global_locale;
  1180. }
  1181. #ifdef __UCLIBC_HAS_THREADS__
  1182. old = __curlocale_set(dataset);
  1183. #else
  1184. old = __curlocale_var;
  1185. __curlocale_var = dataset;
  1186. #endif
  1187. }
  1188. if (old == __global_locale) {
  1189. return LC_GLOBAL_LOCALE;
  1190. }
  1191. return old;
  1192. }
  1193. libc_hidden_def(uselocale)
  1194. #endif
  1195. /**********************************************************************/
  1196. #ifdef L___curlocale
  1197. #ifdef __UCLIBC_HAS_THREADS__
  1198. __locale_t weak_const_function __curlocale(void)
  1199. {
  1200. return __curlocale_var; /* This is overriden by the thread version. */
  1201. }
  1202. __locale_t weak_function __curlocale_set(__locale_t newloc)
  1203. {
  1204. __locale_t oldloc = __curlocale_var;
  1205. assert(newloc != LC_GLOBAL_LOCALE);
  1206. __curlocale_var = newloc;
  1207. return oldloc;
  1208. }
  1209. #endif
  1210. #endif
  1211. /**********************************************************************/
  1212. #ifdef L___locale_mbrtowc_l
  1213. /* NOTE: This returns an int... not size_t. Also, it is not a general
  1214. * routine. It is actually a very stripped-down version of mbrtowc
  1215. * that takes a __locale_t arg. This is used by strcoll and strxfrm.
  1216. * It is also used above to generate wchar_t versions of the decimal point
  1217. * and thousands seperator. */
  1218. #ifndef __CTYPE_HAS_UTF_8_LOCALES
  1219. #warning __CTYPE_HAS_UTF_8_LOCALES not set!
  1220. #endif
  1221. #ifndef __CTYPE_HAS_8_BIT_LOCALES
  1222. #warning __CTYPE_HAS_8_BIT_LOCALES not set!
  1223. #endif
  1224. #define Cc2wc_IDX_SHIFT __LOCALE_DATA_Cc2wc_IDX_SHIFT
  1225. #define Cc2wc_ROW_LEN __LOCALE_DATA_Cc2wc_ROW_LEN
  1226. extern size_t _wchar_utf8sntowcs(wchar_t *__restrict pwc, size_t wn,
  1227. const char **__restrict src, size_t n,
  1228. mbstate_t *ps, int allow_continuation) attribute_hidden;
  1229. int attribute_hidden __locale_mbrtowc_l(wchar_t *__restrict dst,
  1230. const char *__restrict src,
  1231. __locale_t loc )
  1232. {
  1233. #ifdef __CTYPE_HAS_UTF_8_LOCALES
  1234. if (loc->encoding == __ctype_encoding_utf8) {
  1235. mbstate_t ps;
  1236. const char *p = src;
  1237. size_t r;
  1238. ps.__mask = 0;
  1239. r = _wchar_utf8sntowcs(dst, 1, &p, SIZE_MAX, &ps, 1);
  1240. return (r == 1) ? (p-src) : r; /* Need to return 0 if nul char. */
  1241. }
  1242. #endif
  1243. #ifdef __CTYPE_HAS_8_BIT_LOCALES
  1244. assert((loc->encoding == __ctype_encoding_7_bit) || (loc->encoding == __ctype_encoding_8_bit));
  1245. #else
  1246. assert(loc->encoding == __ctype_encoding_7_bit);
  1247. #endif
  1248. if ((*dst = ((unsigned char)(*src))) < 0x80) { /* ASCII... */
  1249. return (*src != 0);
  1250. }
  1251. #ifdef __CTYPE_HAS_8_BIT_LOCALES
  1252. if (loc->encoding == __ctype_encoding_8_bit) {
  1253. wchar_t wc = *dst - 0x80;
  1254. *dst = loc->tbl8c2wc[
  1255. (loc->idx8c2wc[wc >> Cc2wc_IDX_SHIFT]
  1256. << Cc2wc_IDX_SHIFT) + (wc & (Cc2wc_ROW_LEN - 1))];
  1257. if (*dst) {
  1258. return 1;
  1259. }
  1260. }
  1261. #endif
  1262. return -1;
  1263. }
  1264. #endif
  1265. /**********************************************************************/