locale.c 42 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460
  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 _GNU_SOURCE
  46. #define __CTYPE_HAS_8_BIT_LOCALES 1
  47. #include <string.h>
  48. #include <stdlib.h>
  49. #include <stddef.h>
  50. #include <limits.h>
  51. #include <stdint.h>
  52. #include <assert.h>
  53. #include <errno.h>
  54. #include <ctype.h>
  55. #include <stdio.h>
  56. #ifdef __UCLIBC_MJN3_ONLY__
  57. #ifdef L_setlocale
  58. #warning TODO: Make the link_warning()s a config option?
  59. #endif
  60. #endif
  61. #undef link_warning
  62. #define link_warning(A,B)
  63. #undef __LOCALE_C_ONLY
  64. #ifndef __UCLIBC_HAS_LOCALE__
  65. #define __LOCALE_C_ONLY
  66. #endif /* __UCLIBC_HAS_LOCALE__ */
  67. #ifdef __LOCALE_C_ONLY
  68. #include <locale.h>
  69. #else /* __LOCALE_C_ONLY */
  70. #ifdef __UCLIBC_MJN3_ONLY__
  71. #ifdef L_setlocale
  72. #warning TODO: Fix the __CTYPE_HAS_8_BIT_LOCALES define at the top of the file.
  73. #warning TODO: Fix __WCHAR_ENABLED.
  74. #endif
  75. #endif
  76. /* Need to include this before locale.h and xlocale.h! */
  77. #include <bits/uClibc_locale.h>
  78. #undef CODESET_LIST
  79. #define CODESET_LIST (__locale_mmap->codeset_list)
  80. #ifdef __UCLIBC_HAS_XLOCALE__
  81. #include <xlocale.h>
  82. #include <locale.h>
  83. #else /* __UCLIBC_HAS_XLOCALE__ */
  84. /* We need this internally... */
  85. #define __UCLIBC_HAS_XLOCALE__ 1
  86. #include <xlocale.h>
  87. #include <locale.h>
  88. #undef __UCLIBC_HAS_XLOCALE__
  89. #endif /* __UCLIBC_HAS_XLOCALE__ */
  90. #include <wchar.h>
  91. #define LOCALE_NAMES (__locale_mmap->locale_names5)
  92. #define LOCALES (__locale_mmap->locales)
  93. #define LOCALE_AT_MODIFIERS (__locale_mmap->locale_at_modifiers)
  94. #define CATEGORY_NAMES (__locale_mmap->lc_names)
  95. #ifdef __UCLIBC_MJN3_ONLY__
  96. #warning REMINDER: redo the MAX_LOCALE_STR stuff...
  97. #endif
  98. #define MAX_LOCALE_STR 256 /* TODO: Only sufficient for current case. */
  99. #define MAX_LOCALE_CATEGORY_STR 32 /* TODO: Only sufficient for current case. */
  100. /* Note: Best if MAX_LOCALE_CATEGORY_STR is a power of 2. */
  101. extern int _locale_set_l(const unsigned char *p, __locale_t base);
  102. extern void _locale_init_l(__locale_t base);
  103. #endif /* __LOCALE_C_ONLY */
  104. #undef LOCALE_STRING_SIZE
  105. #define LOCALE_SELECTOR_SIZE (2 * __LC_ALL + 2)
  106. #ifdef __UCLIBC_MJN3_ONLY__
  107. #ifdef L_setlocale
  108. #warning TODO: Create a C locale selector string.
  109. #endif
  110. #endif
  111. #define C_LOCALE_SELECTOR "\x23\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80"
  112. #include <langinfo.h>
  113. #include <nl_types.h>
  114. /**********************************************************************/
  115. #ifdef L_setlocale
  116. #ifdef __LOCALE_C_ONLY
  117. link_warning(setlocale,"REMINDER: The 'setlocale' function supports only C|POSIX locales.")
  118. static const char C_string[] = "C";
  119. char *setlocale(int category, register const char *locale)
  120. {
  121. return ( (((unsigned int)(category)) <= LC_ALL)
  122. && ( (!locale) /* Request for locale category string. */
  123. || (!*locale) /* Implementation-defined default is C. */
  124. || ((*locale == 'C') && !locale[1])
  125. || (!strcmp(locale, "POSIX"))) )
  126. ? (char *) C_string /* Always in C/POSIX locale. */
  127. : NULL;
  128. }
  129. #else /* ---------------------------------------------- __LOCALE_C_ONLY */
  130. #ifdef __UCLIBC_HAS_THREADS__
  131. link_warning(setlocale,"REMINDER: The 'setlocale' function is _not_ threadsafe except for simple queries.")
  132. #endif
  133. #if !defined(__LOCALE_DATA_NUM_LOCALES) || (__LOCALE_DATA_NUM_LOCALES <= 1)
  134. #error locales enabled, but not data other than for C locale!
  135. #endif
  136. #ifdef __UCLIBC_MJN3_ONLY__
  137. #warning TODO: Move posix and utf8 strings.
  138. #endif
  139. static const char posix[] = "POSIX";
  140. static const char utf8[] = "UTF-8";
  141. #ifdef __UCLIBC_MJN3_ONLY__
  142. #warning TODO: Fix dimensions of hr_locale.
  143. #endif
  144. /* Individual category strings start at hr_locale + category * MAX_LOCALE_CATEGORY.
  145. * This holds for LC_ALL as well.
  146. */
  147. static char hr_locale[(MAX_LOCALE_CATEGORY_STR * LC_ALL) + MAX_LOCALE_STR];
  148. static void update_hr_locale(const unsigned char *spec)
  149. {
  150. const unsigned char *loc;
  151. const unsigned char *s;
  152. char *n;
  153. int i, category, done;
  154. done = category = 0;
  155. do {
  156. s = spec + 1;
  157. n = hr_locale + category * MAX_LOCALE_CATEGORY_STR;
  158. if (category == LC_ALL) {
  159. done = 1;
  160. for (i = 0 ; i < LC_ALL-1 ; i += 2) {
  161. if ((s[i] != s[i+2]) || (s[i+1] != s[i+3])) {
  162. goto SKIP;
  163. }
  164. }
  165. /* All categories the same, so simplify string by using a single
  166. * category. */
  167. category = LC_CTYPE;
  168. }
  169. SKIP:
  170. i = (category == LC_ALL) ? 0 : category;
  171. s += 2*i;
  172. do {
  173. if ((*s != 0xff) || (s[1] != 0xff)) {
  174. loc = LOCALES
  175. + __LOCALE_DATA_WIDTH_LOCALES * ((((int)(*s & 0x7f)) << 7)
  176. + (s[1] & 0x7f));
  177. if (category == LC_ALL) {
  178. n = stpcpy(n, CATEGORY_NAMES + (int) CATEGORY_NAMES[i]);
  179. *n++ = '=';
  180. }
  181. if (*loc == 0) {
  182. *n++ = 'C';
  183. *n = 0;
  184. } else {
  185. char at = 0;
  186. memcpy(n, LOCALE_NAMES + 5*((*loc)-1), 5);
  187. if (n[2] != '_') {
  188. at = n[2];
  189. n[2] = '_';
  190. }
  191. n += 5;
  192. *n++ = '.';
  193. if (loc[2] == 2) {
  194. n = stpcpy(n, utf8);
  195. } else if (loc[2] >= 3) {
  196. n = stpcpy(n, CODESET_LIST + (int)(CODESET_LIST[loc[2] - 3]));
  197. }
  198. if (at) {
  199. const char *q;
  200. *n++ = '@';
  201. q = LOCALE_AT_MODIFIERS;
  202. do {
  203. if (q[1] == at) {
  204. n = stpcpy(n, q+2);
  205. break;
  206. }
  207. q += 2 + *q;
  208. } while (*q);
  209. }
  210. }
  211. *n++ = ';';
  212. }
  213. s += 2;
  214. } while (++i < category);
  215. *--n = 0; /* Remove trailing ';' and nul-terminate. */
  216. ++category;
  217. } while (!done);
  218. }
  219. char *setlocale(int category, const char *locale)
  220. {
  221. if (((unsigned int)(category)) > LC_ALL) {
  222. #if 0
  223. __set_errno(EINVAL); /* glibc sets errno -- SUSv3 doesn't say. */
  224. #endif
  225. return NULL; /* Illegal/unsupported category. */
  226. }
  227. if (locale != NULL) { /* Not just a query... */
  228. if (!__newlocale((1 << category), locale, __global_locale)) {
  229. return NULL; /* Failed! */
  230. }
  231. update_hr_locale(__global_locale->cur_locale);
  232. }
  233. /* Either a query or a successful set, so return current locale string. */
  234. return hr_locale + (category * MAX_LOCALE_CATEGORY_STR);
  235. }
  236. #endif /* __LOCALE_C_ONLY */
  237. #endif
  238. /**********************************************************************/
  239. #ifdef L_localeconv
  240. /* Note: We assume here that the compiler does the sane thing regarding
  241. * placement of the fields in the struct. If necessary, we could ensure
  242. * this usings an array of offsets but at some size cost. */
  243. #ifdef __LOCALE_C_ONLY
  244. link_warning(localeconv,"REMINDER: The 'localeconv' function is hardwired for C/POSIX locale only.")
  245. static struct lconv the_lconv;
  246. static const char decpt[] = ".";
  247. struct lconv *localeconv(void)
  248. {
  249. register char *p = (char *)(&the_lconv);
  250. *((char **)p) = (char *) decpt;
  251. do {
  252. p += sizeof(char **);
  253. *((char **)p) = (char *) (decpt+1);
  254. } while (p < (char *) &the_lconv.negative_sign);
  255. p = (&the_lconv.int_frac_digits);
  256. do {
  257. *p = CHAR_MAX;
  258. ++p;
  259. } while (p <= &the_lconv.int_n_sign_posn);
  260. return &the_lconv;
  261. }
  262. #else /* __LOCALE_C_ONLY */
  263. static struct lconv the_lconv;
  264. struct lconv *localeconv(void)
  265. {
  266. register char *p = (char *) &the_lconv;
  267. register char **q = (char **) &(__UCLIBC_CURLOCALE_DATA).decimal_point;
  268. do {
  269. *((char **)p) = *q;
  270. p += sizeof(char **);
  271. ++q;
  272. } while (p < &the_lconv.int_frac_digits);
  273. do {
  274. *p = **q;
  275. ++p;
  276. ++q;
  277. } while (p <= &the_lconv.int_n_sign_posn);
  278. return &the_lconv;
  279. }
  280. #endif /* __LOCALE_C_ONLY */
  281. #endif
  282. /**********************************************************************/
  283. #if defined(L__locale_init) && !defined(__LOCALE_C_ONLY)
  284. __uclibc_locale_t __global_locale_data;
  285. __locale_t __global_locale = &__global_locale_data;
  286. #ifdef __UCLIBC_HAS_XLOCALE__
  287. __locale_t __curlocale_var = &__global_locale_data;
  288. #endif
  289. /*----------------------------------------------------------------------*/
  290. #ifdef __UCLIBC_MJN3_ONLY__
  291. #warning TODO: Move utf8 and ascii strings.
  292. #endif
  293. static const char utf8[] = "UTF-8";
  294. static const char ascii[] = "ASCII";
  295. typedef struct {
  296. uint16_t num_base;
  297. uint16_t num_der;
  298. uint16_t MAX_WEIGHTS;
  299. uint16_t num_index2weight;
  300. #define num_index2ruleidx num_index2weight
  301. uint16_t num_weightstr;
  302. uint16_t num_multistart;
  303. uint16_t num_override;
  304. uint16_t num_ruletable;
  305. } coldata_header_t;
  306. typedef struct {
  307. uint16_t num_weights;
  308. uint16_t num_starters;
  309. uint16_t ii_shift;
  310. uint16_t ti_shift;
  311. uint16_t ii_len;
  312. uint16_t ti_len;
  313. uint16_t max_weight;
  314. uint16_t num_col_base;
  315. uint16_t max_col_index;
  316. uint16_t undefined_idx;
  317. uint16_t range_low;
  318. uint16_t range_count;
  319. uint16_t range_base_weight;
  320. uint16_t range_rule_offset;
  321. uint16_t index2weight_offset;
  322. uint16_t index2ruleidx_offset;
  323. uint16_t multistart_offset;
  324. uint16_t wcs2colidt_offset_low;
  325. uint16_t wcs2colidt_offset_hi;
  326. } coldata_base_t;
  327. typedef struct {
  328. uint16_t base_idx;
  329. uint16_t undefined_idx;
  330. uint16_t overrides_offset;
  331. uint16_t multistart_offset;
  332. } coldata_der_t;
  333. static int init_cur_collate(int der_num, __collate_t *cur_collate)
  334. {
  335. const uint16_t *__locale_collate_tbl = __locale_mmap->collate_data;
  336. coldata_header_t *cdh;
  337. coldata_base_t *cdb;
  338. coldata_der_t *cdd;
  339. const uint16_t *p;
  340. size_t n;
  341. uint16_t i, w;
  342. #ifdef __UCLIBC_MJN3_ONLY__
  343. #warning kill of x86-specific asserts
  344. #endif
  345. #if 0
  346. assert(sizeof(coldata_base_t) == 19*2);
  347. assert(sizeof(coldata_der_t) == 4*2);
  348. assert(sizeof(coldata_header_t) == 8*2);
  349. #endif
  350. if (!der_num) { /* C locale... special */
  351. cur_collate->num_weights = 0;
  352. return 1;
  353. }
  354. --der_num;
  355. cdh = (coldata_header_t *) __locale_collate_tbl;
  356. #ifdef __UCLIBC_MJN3_ONLY__
  357. #warning CONSIDER: Should we assert here?
  358. #endif
  359. #if 0
  360. if (der_num >= cdh->num_der) {
  361. return 0;
  362. }
  363. #else
  364. assert((der_num < cdh->num_der));
  365. #endif
  366. cdd = (coldata_der_t *)(__locale_collate_tbl
  367. + (sizeof(coldata_header_t)
  368. + cdh->num_base * sizeof(coldata_base_t)
  369. + der_num * sizeof(coldata_der_t)
  370. )/2 );
  371. cdb = (coldata_base_t *)(__locale_collate_tbl
  372. + (sizeof(coldata_header_t)
  373. + cdd->base_idx * sizeof(coldata_base_t)
  374. )/2 );
  375. memcpy(cur_collate, cdb, offsetof(coldata_base_t,index2weight_offset));
  376. cur_collate->undefined_idx = cdd->undefined_idx;
  377. cur_collate->ti_mask = (1 << cur_collate->ti_shift)-1;
  378. cur_collate->ii_mask = (1 << cur_collate->ii_shift)-1;
  379. /* fflush(stdout); */
  380. /* fprintf(stderr,"base=%d num_col_base: %d %d\n", cdd->base_idx ,cur_collate->num_col_base, cdb->num_col_base); */
  381. n = (sizeof(coldata_header_t) + cdh->num_base * sizeof(coldata_base_t)
  382. + cdh->num_der * sizeof(coldata_der_t))/2;
  383. /* fprintf(stderr,"n = %d\n", n); */
  384. cur_collate->index2weight_tbl = __locale_collate_tbl + n + cdb->index2weight_offset;
  385. /* fprintf(stderr,"i2w = %d\n", n + cdb->index2weight_offset); */
  386. n += cdh->num_index2weight;
  387. cur_collate->index2ruleidx_tbl = __locale_collate_tbl + n + cdb->index2ruleidx_offset;
  388. /* fprintf(stderr,"i2r = %d\n", n + cdb->index2ruleidx_offset); */
  389. n += cdh->num_index2ruleidx;
  390. cur_collate->multistart_tbl = __locale_collate_tbl + n + cdd->multistart_offset;
  391. /* fprintf(stderr,"mts = %d\n", n + cdb->multistart_offset); */
  392. n += cdh->num_multistart;
  393. cur_collate->overrides_tbl = __locale_collate_tbl + n + cdd->overrides_offset;
  394. /* fprintf(stderr,"ovr = %d\n", n + cdd->overrides_offset); */
  395. n += cdh->num_override;
  396. cur_collate->ruletable = __locale_collate_tbl + n;
  397. /* fprintf(stderr, "rtb = %d\n", n); */
  398. n += cdh->num_ruletable;
  399. cur_collate->weightstr = __locale_collate_tbl + n;
  400. /* fprintf(stderr,"wts = %d\n", n); */
  401. n += cdh->num_weightstr;
  402. cur_collate->wcs2colidt_tbl = __locale_collate_tbl + n
  403. + (((unsigned long)(cdb->wcs2colidt_offset_hi)) << 16)
  404. + cdb->wcs2colidt_offset_low;
  405. /* fprintf(stderr,"wcs = %lu\n", n + (((unsigned long)(cdb->wcs2colidt_offset_hi)) << 16) */
  406. /* + cdb->wcs2colidt_offset_low); */
  407. cur_collate->MAX_WEIGHTS = cdh->MAX_WEIGHTS;
  408. #ifdef __UCLIBC_MJN3_ONLY__
  409. #warning CONSIDER: Fix the +1 by increasing max_col_index?
  410. #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
  411. #endif
  412. cur_collate->index2weight = calloc(2*cur_collate->max_col_index+2,
  413. sizeof(uint16_t));
  414. if (!cur_collate->index2weight) {
  415. return 0;
  416. }
  417. cur_collate->index2ruleidx = cur_collate->index2weight
  418. + cur_collate->max_col_index + 1;
  419. memcpy(cur_collate->index2weight, cur_collate->index2weight_tbl,
  420. cur_collate->num_col_base * sizeof(uint16_t));
  421. memcpy(cur_collate->index2ruleidx, cur_collate->index2ruleidx_tbl,
  422. cur_collate->num_col_base * sizeof(uint16_t));
  423. /* now do the overrides */
  424. p = cur_collate->overrides_tbl;
  425. while (*p > 1) {
  426. /* fprintf(stderr, "processing override -- count = %d\n", *p); */
  427. n = *p++;
  428. w = *p++;
  429. do {
  430. i = *p++;
  431. /* fprintf(stderr, " i=%d (%#x) w=%d *p=%d\n", i, i, w, *p); */
  432. cur_collate->index2weight[i-1] = w++;
  433. cur_collate->index2ruleidx[i-1] = *p++;
  434. } while (--n);
  435. }
  436. assert(*p == 1);
  437. while (*++p) {
  438. i = *p;
  439. /* fprintf(stderr, " i=%d (%#x) w=%d *p=%d\n", i, i, p[1], p[2]); */
  440. cur_collate->index2weight[i-1] = *++p;
  441. cur_collate->index2ruleidx[i-1] = *++p;
  442. }
  443. for (i=0 ; i < cur_collate->multistart_tbl[0] ; i++) {
  444. p = cur_collate->multistart_tbl;
  445. /* fprintf(stderr, "%2d of %2d: %d ", i, cur_collate->multistart_tbl[0], p[i]); */
  446. p += p[i];
  447. do {
  448. n = *p++;
  449. do {
  450. if (!*p) { /* found it */
  451. /* fprintf(stderr, "found: n=%d (%#lx) |%.*ls|\n", n, (int) *cs->s, n, cs->s); */
  452. /* fprintf(stderr, ": %d - single\n", n); */
  453. goto FOUND;
  454. }
  455. /* the lookup check here is safe since we're assured that *p is a valid colidex */
  456. /* fprintf(stderr, "lookup(%lc)==%d *p==%d\n", cs->s[n], lookup(cs->s[n]), (int) *p); */
  457. /* fprintf(stderr, ": %d - ", n); */
  458. do {
  459. /* fprintf(stderr, "%d|", *p); */
  460. } while (*p++);
  461. break;
  462. } while (1);
  463. } while (1);
  464. FOUND:
  465. continue;
  466. }
  467. return 1;
  468. }
  469. int _locale_set_l(const unsigned char *p, __locale_t base)
  470. {
  471. const char **x;
  472. unsigned char *s = base->cur_locale + 1;
  473. const size_t *stp;
  474. const unsigned char *r;
  475. const uint16_t *io;
  476. const uint16_t *ii;
  477. const unsigned char *d;
  478. int row; /* locale row */
  479. int crow; /* category row */
  480. int len;
  481. int c;
  482. int i = 0;
  483. __collate_t newcol;
  484. ++p;
  485. newcol.index2weight = NULL;
  486. if ((p[2*LC_COLLATE] != s[2*LC_COLLATE])
  487. || (p[2*LC_COLLATE + 1] != s[2*LC_COLLATE + 1])
  488. ) {
  489. row = (((int)(*p & 0x7f)) << 7) + (p[1] & 0x7f);
  490. assert(row < __LOCALE_DATA_NUM_LOCALES);
  491. if (!init_cur_collate(__locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES
  492. * row + 3 + LC_COLLATE ],
  493. &newcol)
  494. ) {
  495. return 0; /* calloc failed. */
  496. }
  497. free(base->collate.index2weight);
  498. memcpy(&base->collate, &newcol, sizeof(__collate_t));
  499. }
  500. do {
  501. if ((*p != *s) || (p[1] != s[1])) {
  502. row = (((int)(*p & 0x7f)) << 7) + (p[1] & 0x7f);
  503. assert(row < __LOCALE_DATA_NUM_LOCALES);
  504. *s = *p;
  505. s[1] = p[1];
  506. if ((i != LC_COLLATE)
  507. && ((len = __locale_mmap->lc_common_item_offsets_LEN[i]) != 0)
  508. ) {
  509. crow = __locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES * row
  510. + 3 + i ]
  511. * len;
  512. x = (const char **)(((char *) base)
  513. + base->category_offsets[i]);
  514. stp = __locale_mmap->lc_common_tbl_offsets + 4*i;
  515. r = (const unsigned char *)( ((char *)__locale_mmap) + *stp );
  516. io = (const uint16_t *)( ((char *)__locale_mmap) + *++stp );
  517. ii = (const uint16_t *)( ((char *)__locale_mmap) + *++stp );
  518. d = (const unsigned char *)( ((char *)__locale_mmap) + *++stp );
  519. for (c=0 ; c < len ; c++) {
  520. *(x + c) = d + ii[ r[crow + c] + io[c] ];
  521. }
  522. }
  523. if (i == LC_CTYPE) {
  524. c = __locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES * row
  525. + 2 ]; /* codeset */
  526. if (c <= 2) {
  527. if (c == 2) {
  528. base->codeset = utf8;
  529. base->encoding = __ctype_encoding_utf8;
  530. /* TODO - fix for bcc */
  531. base->mb_cur_max = 6;
  532. } else {
  533. assert(c==1);
  534. base->codeset = ascii;
  535. base->encoding = __ctype_encoding_7_bit;
  536. base->mb_cur_max = 1;
  537. }
  538. } else {
  539. const __codeset_8_bit_t *c8b;
  540. r = CODESET_LIST;
  541. base->codeset = r + r[c -= 3];
  542. base->encoding = __ctype_encoding_8_bit;
  543. #ifdef __UCLIBC_MJN3_ONLY__
  544. #warning REMINDER: update 8 bit mb_cur_max when translit implemented!
  545. #endif
  546. /* TODO - update when translit implemented! */
  547. base->mb_cur_max = 1;
  548. c8b = __locale_mmap->codeset_8_bit + c;
  549. #ifdef __CTYPE_HAS_8_BIT_LOCALES
  550. base->idx8ctype = c8b->idx8ctype;
  551. base->idx8uplow = c8b->idx8uplow;
  552. #ifdef __UCLIBC_HAS_WCHAR__
  553. base->idx8c2wc = c8b->idx8c2wc;
  554. base->idx8wc2c = c8b->idx8wc2c;
  555. /* translit */
  556. #endif /* __UCLIBC_HAS_WCHAR__ */
  557. /* What follows is fairly bloated, but it is just a hack
  558. * to get the 8-bit codeset ctype stuff functioning.
  559. * All of this will be replaced in the next generation
  560. * of locale support anyway... */
  561. memcpy(base->__ctype_b_data,
  562. __C_ctype_b - __UCLIBC_CTYPE_B_TBL_OFFSET,
  563. (256 + __UCLIBC_CTYPE_B_TBL_OFFSET)
  564. * sizeof(__ctype_mask_t));
  565. memcpy(base->__ctype_tolower_data,
  566. __C_ctype_tolower - __UCLIBC_CTYPE_TO_TBL_OFFSET,
  567. (256 + __UCLIBC_CTYPE_TO_TBL_OFFSET)
  568. * sizeof(__ctype_touplow_t));
  569. memcpy(base->__ctype_toupper_data,
  570. __C_ctype_toupper - __UCLIBC_CTYPE_TO_TBL_OFFSET,
  571. (256 + __UCLIBC_CTYPE_TO_TBL_OFFSET)
  572. * sizeof(__ctype_touplow_t));
  573. #define Cctype_TBL_MASK ((1 << __LOCALE_DATA_Cctype_IDX_SHIFT) - 1)
  574. #define Cctype_IDX_OFFSET (128 >> __LOCALE_DATA_Cctype_IDX_SHIFT)
  575. {
  576. int u;
  577. __ctype_mask_t m;
  578. for (u=0 ; u < 128 ; u++) {
  579. #ifdef __LOCALE_DATA_Cctype_PACKED
  580. c = base->tbl8ctype
  581. [ ((int)(c8b->idx8ctype
  582. [(u >> __LOCALE_DATA_Cctype_IDX_SHIFT) ])
  583. << (__LOCALE_DATA_Cctype_IDX_SHIFT - 1))
  584. + ((u & Cctype_TBL_MASK) >> 1)];
  585. c = (u & 1) ? (c >> 4) : (c & 0xf);
  586. #else
  587. c = base->tbl8ctype
  588. [ ((int)(c8b->idx8ctype
  589. [(u >> __LOCALE_DATA_Cctype_IDX_SHIFT) ])
  590. << __LOCALE_DATA_Cctype_IDX_SHIFT)
  591. + (u & Cctype_TBL_MASK) ];
  592. #endif
  593. m = base->code2flag[c];
  594. base->__ctype_b_data
  595. [128 + __UCLIBC_CTYPE_B_TBL_OFFSET + u]
  596. = m;
  597. #ifdef __UCLIBC_HAS_CTYPE_SIGNED__
  598. if (((signed char)(128 + u)) != -1) {
  599. base->__ctype_b_data[__UCLIBC_CTYPE_B_TBL_OFFSET
  600. + ((signed char)(128 + u))]
  601. = m;
  602. }
  603. #endif
  604. base->__ctype_tolower_data
  605. [128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
  606. = 128 + u;
  607. base->__ctype_toupper_data
  608. [128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
  609. = 128 + u;
  610. if (m & (_ISlower|_ISupper)) {
  611. c = base->tbl8uplow
  612. [ ((int)(c8b->idx8uplow
  613. [u >> __LOCALE_DATA_Cuplow_IDX_SHIFT])
  614. << __LOCALE_DATA_Cuplow_IDX_SHIFT)
  615. + ((128 + u)
  616. & ((1 << __LOCALE_DATA_Cuplow_IDX_SHIFT)
  617. - 1)) ];
  618. if (m & _ISlower) {
  619. base->__ctype_toupper_data
  620. [128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
  621. = (unsigned char)(128 + u + c);
  622. #ifdef __UCLIBC_HAS_CTYPE_SIGNED__
  623. if (((signed char)(128 + u)) != -1) {
  624. base->__ctype_toupper_data
  625. [__UCLIBC_CTYPE_TO_TBL_OFFSET
  626. + ((signed char)(128 + u))]
  627. = (unsigned char)(128 + u + c);
  628. }
  629. #endif
  630. } else {
  631. base->__ctype_tolower_data
  632. [128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
  633. = (unsigned char)(128 + u - c);
  634. #ifdef __UCLIBC_HAS_CTYPE_SIGNED__
  635. if (((signed char)(128 + u)) != -1) {
  636. base->__ctype_tolower_data
  637. [__UCLIBC_CTYPE_TO_TBL_OFFSET
  638. + ((signed char)(128 + u))]
  639. = (unsigned char)(128 + u - c);
  640. }
  641. #endif
  642. }
  643. }
  644. }
  645. }
  646. #ifdef __UCLIBC_HAS_XLOCALE__
  647. base->__ctype_b = base->__ctype_b_data
  648. + __UCLIBC_CTYPE_B_TBL_OFFSET;
  649. base->__ctype_tolower = base->__ctype_tolower_data
  650. + __UCLIBC_CTYPE_TO_TBL_OFFSET;
  651. base->__ctype_toupper = base->__ctype_toupper_data
  652. + __UCLIBC_CTYPE_TO_TBL_OFFSET;
  653. #else /* __UCLIBC_HAS_XLOCALE__ */
  654. __ctype_b = base->__ctype_b_data
  655. + __UCLIBC_CTYPE_B_TBL_OFFSET;
  656. __ctype_tolower = base->__ctype_tolower_data
  657. + __UCLIBC_CTYPE_TO_TBL_OFFSET;
  658. __ctype_toupper = base->__ctype_toupper_data
  659. + __UCLIBC_CTYPE_TO_TBL_OFFSET;
  660. #endif /* __UCLIBC_HAS_XLOCALE__ */
  661. #endif /* __CTYPE_HAS_8_BIT_LOCALES */
  662. }
  663. #ifdef __UCLIBC_MJN3_ONLY__
  664. #warning TODO: Put the outdigit string length in the locale_mmap object.
  665. #endif
  666. d = base->outdigit_length;
  667. x = &base->outdigit0_mb;
  668. for (c = 0 ; c < 10 ; c++) {
  669. ((unsigned char *)d)[c] = strlen(x[c]);
  670. assert(d[c] > 0);
  671. }
  672. } else if (i == LC_NUMERIC) {
  673. assert(LC_NUMERIC > LC_CTYPE); /* Need ctype initialized. */
  674. base->decimal_point_len
  675. = __locale_mbrtowc_l(&base->decimal_point_wc,
  676. base->decimal_point, base);
  677. assert(base->decimal_point_len > 0);
  678. assert(base->decimal_point[base->decimal_point_len] == 0);
  679. if (*base->grouping) {
  680. base->thousands_sep_len
  681. = __locale_mbrtowc_l(&base->thousands_sep_wc,
  682. base->thousands_sep, base);
  683. #if 1
  684. #ifdef __UCLIBC_MJN3_ONLY__
  685. #warning TODO: Remove hack involving grouping without a thousep char (bg_BG).
  686. #endif
  687. assert(base->thousands_sep_len >= 0);
  688. if (base->thousands_sep_len == 0) {
  689. base->grouping = base->thousands_sep; /* empty string */
  690. }
  691. assert(base->thousands_sep[base->thousands_sep_len] == 0);
  692. #else
  693. assert(base->thousands_sep_len > 0);
  694. assert(base->thousands_sep[base->thousands_sep_len] == 0);
  695. #endif
  696. }
  697. /* } else if (i == LC_COLLATE) { */
  698. /* init_cur_collate(__locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES */
  699. /* * row + 3 + i ], */
  700. /* &base->collate); */
  701. }
  702. }
  703. ++i;
  704. p += 2;
  705. s += 2;
  706. } while (i < LC_ALL);
  707. return 1;
  708. }
  709. static const uint16_t __code2flag[16] = {
  710. 0, /* unclassified = 0 */
  711. _ISprint|_ISgraph|_ISalnum|_ISalpha, /* alpha_nonupper_nonlower */
  712. _ISprint|_ISgraph|_ISalnum|_ISalpha|_ISlower, /* alpha_lower */
  713. _ISprint|_ISgraph|_ISalnum|_ISalpha|_ISlower|_ISupper, /* alpha_upper_lower */
  714. _ISprint|_ISgraph|_ISalnum|_ISalpha|_ISupper, /* alpha_upper */
  715. _ISprint|_ISgraph|_ISalnum|_ISdigit, /* digit */
  716. _ISprint|_ISgraph|_ISpunct, /* punct */
  717. _ISprint|_ISgraph, /* graph */
  718. _ISprint|_ISspace, /* print_space_nonblank */
  719. _ISprint|_ISspace|_ISblank, /* print_space_blank */
  720. _ISspace, /* space_nonblank_noncntrl */
  721. _ISspace|_ISblank, /* space_blank_noncntrl */
  722. _IScntrl|_ISspace, /* cntrl_space_nonblank */
  723. _IScntrl|_ISspace|_ISblank, /* cntrl_space_blank */
  724. _IScntrl /* cntrl_nonspace */
  725. };
  726. void _locale_init_l(__locale_t base)
  727. {
  728. memset(base->cur_locale, 0, LOCALE_SELECTOR_SIZE);
  729. base->cur_locale[0] = '#';
  730. memcpy(base->category_item_count,
  731. __locale_mmap->lc_common_item_offsets_LEN,
  732. LC_ALL);
  733. ++base->category_item_count[0]; /* Increment for codeset entry. */
  734. base->category_offsets[0] = offsetof(__uclibc_locale_t, outdigit0_mb);
  735. base->category_offsets[1] = offsetof(__uclibc_locale_t, decimal_point);
  736. base->category_offsets[2] = offsetof(__uclibc_locale_t, int_curr_symbol);
  737. base->category_offsets[3] = offsetof(__uclibc_locale_t, abday_1);
  738. /* base->category_offsets[4] = offsetof(__uclibc_locale_t, collate???); */
  739. base->category_offsets[5] = offsetof(__uclibc_locale_t, yesexpr);
  740. #ifdef __CTYPE_HAS_8_BIT_LOCALES
  741. base->tbl8ctype
  742. = (const unsigned char *) &__locale_mmap->tbl8ctype;
  743. base->tbl8uplow
  744. = (const unsigned char *) &__locale_mmap->tbl8uplow;
  745. #ifdef __UCLIBC_HAS_WCHAR__
  746. base->tbl8c2wc
  747. = (const uint16_t *) &__locale_mmap->tbl8c2wc;
  748. base->tbl8wc2c
  749. = (const unsigned char *) &__locale_mmap->tbl8wc2c;
  750. /* translit */
  751. #endif /* __UCLIBC_HAS_WCHAR__ */
  752. #endif /* __CTYPE_HAS_8_BIT_LOCALES */
  753. #ifdef __UCLIBC_HAS_WCHAR__
  754. base->tblwctype
  755. = (const unsigned char *) &__locale_mmap->tblwctype;
  756. base->tblwuplow
  757. = (const unsigned char *) &__locale_mmap->tblwuplow;
  758. base->tblwuplow_diff
  759. = (const uint16_t *) &__locale_mmap->tblwuplow_diff;
  760. /* base->tblwcomb */
  761. /* = (const unsigned char *) &__locale_mmap->tblwcomb; */
  762. /* width?? */
  763. #endif /* __UCLIBC_HAS_WCHAR__ */
  764. /* Initially, set things up to use the global C ctype tables.
  765. * This is correct for C (ASCII) and UTF-8 based locales (except tr_TR). */
  766. #ifdef __UCLIBC_HAS_XLOCALE__
  767. base->__ctype_b = __C_ctype_b;
  768. base->__ctype_tolower = __C_ctype_tolower;
  769. base->__ctype_toupper = __C_ctype_toupper;
  770. #else /* __UCLIBC_HAS_XLOCALE__ */
  771. __ctype_b = __C_ctype_b;
  772. __ctype_tolower = __C_ctype_tolower;
  773. __ctype_toupper = __C_ctype_toupper;
  774. #endif /* __UCLIBC_HAS_XLOCALE__ */
  775. #ifdef __UCLIBC_MJN3_ONLY__
  776. #warning TODO: Initialize code2flag correctly based on locale_mmap.
  777. #endif
  778. base->code2flag = __code2flag;
  779. _locale_set_l(C_LOCALE_SELECTOR, base);
  780. }
  781. void _locale_init(void)
  782. {
  783. /* TODO: mmap the locale file */
  784. /* TODO - ??? */
  785. _locale_init_l(__global_locale);
  786. }
  787. #endif
  788. /**********************************************************************/
  789. #if defined(L_nl_langinfo) || defined(L_nl_langinfo_l)
  790. #ifdef __LOCALE_C_ONLY
  791. /* We need to index 320 bytes of data, so you might initially think we
  792. * need to store the offsets in shorts. But since the offset of the
  793. * 64th item is 182, we'll store "offset - 2*64" for all items >= 64
  794. * and always calculate the data offset as "offset[i] + 2*(i & 64)".
  795. * This allows us to pack the data offsets in an unsigned char while
  796. * also avoiding an "if".
  797. *
  798. * Note: Category order is assumed to be:
  799. * ctype, numeric, monetary, time, collate, messages, all
  800. */
  801. #define C_LC_ALL 6
  802. /* Combine the data to avoid size penalty for seperate char arrays when
  803. * compiler aligns objects. The original code is left in as documentation. */
  804. #define cat_start nl_data
  805. #define C_locale_data (nl_data + C_LC_ALL + 1 + 90)
  806. static const unsigned char nl_data[C_LC_ALL + 1 + 90 + 320] = {
  807. /* static const char cat_start[LC_ALL + 1] = { */
  808. '\x00', '\x0b', '\x0e', '\x24', '\x56', '\x56', '\x5a',
  809. /* }; */
  810. /* static const char item_offset[90] = { */
  811. '\x00', '\x02', '\x04', '\x06', '\x08', '\x0a', '\x0c', '\x0e',
  812. '\x10', '\x12', '\x14', '\x1a', '\x1b', '\x1b', '\x1b', '\x1b',
  813. '\x1b', '\x1b', '\x1b', '\x1b', '\x1b', '\x1c', '\x1c', '\x1c',
  814. '\x1c', '\x1c', '\x1c', '\x1c', '\x1c', '\x1c', '\x1c', '\x1c',
  815. '\x1c', '\x1c', '\x1c', '\x1e', '\x20', '\x24', '\x28', '\x2c',
  816. '\x30', '\x34', '\x38', '\x3c', '\x43', '\x4a', '\x52', '\x5c',
  817. '\x65', '\x6c', '\x75', '\x79', '\x7d', '\x81', '\x85', '\x89',
  818. '\x8d', '\x91', '\x95', '\x99', '\x9d', '\xa1', '\xa5', '\xad',
  819. '\x36', '\x3c', '\x42', '\x46', '\x4b', '\x50', '\x57', '\x61',
  820. '\x69', '\x72', '\x7b', '\x7e', '\x81', '\x96', '\x9f', '\xa8',
  821. '\xb3', '\xb3', '\xb3', '\xb3', '\xb3', '\xb3', '\xb4', '\xba',
  822. '\xbf', '\xbf',
  823. /* }; */
  824. /* static const char C_locale_data[320] = { */
  825. '0', '\x00', '1', '\x00', '2', '\x00', '3', '\x00',
  826. '4', '\x00', '5', '\x00', '6', '\x00', '7', '\x00',
  827. '8', '\x00', '9', '\x00', 'A', 'S', 'C', 'I',
  828. 'I', '\x00', '.', '\x00', '\x7f', '\x00', '-', '\x00',
  829. 'S', 'u', 'n', '\x00', 'M', 'o', 'n', '\x00',
  830. 'T', 'u', 'e', '\x00', 'W', 'e', 'd', '\x00',
  831. 'T', 'h', 'u', '\x00', 'F', 'r', 'i', '\x00',
  832. 'S', 'a', 't', '\x00', 'S', 'u', 'n', 'd',
  833. 'a', 'y', '\x00', 'M', 'o', 'n', 'd', 'a',
  834. 'y', '\x00', 'T', 'u', 'e', 's', 'd', 'a',
  835. 'y', '\x00', 'W', 'e', 'd', 'n', 'e', 's',
  836. 'd', 'a', 'y', '\x00', 'T', 'h', 'u', 'r',
  837. 's', 'd', 'a', 'y', '\x00', 'F', 'r', 'i',
  838. 'd', 'a', 'y', '\x00', 'S', 'a', 't', 'u',
  839. 'r', 'd', 'a', 'y', '\x00', 'J', 'a', 'n',
  840. '\x00', 'F', 'e', 'b', '\x00', 'M', 'a', 'r',
  841. '\x00', 'A', 'p', 'r', '\x00', 'M', 'a', 'y',
  842. '\x00', 'J', 'u', 'n', '\x00', 'J', 'u', 'l',
  843. '\x00', 'A', 'u', 'g', '\x00', 'S', 'e', 'p',
  844. '\x00', 'O', 'c', 't', '\x00', 'N', 'o', 'v',
  845. '\x00', 'D', 'e', 'c', '\x00', 'J', 'a', 'n',
  846. 'u', 'a', 'r', 'y', '\x00', 'F', 'e', 'b',
  847. 'r', 'u', 'a', 'r', 'y', '\x00', 'M', 'a',
  848. 'r', 'c', 'h', '\x00', 'A', 'p', 'r', 'i',
  849. 'l', '\x00', 'M', 'a', 'y', '\x00', 'J', 'u',
  850. 'n', 'e', '\x00', 'J', 'u', 'l', 'y', '\x00',
  851. 'A', 'u', 'g', 'u', 's', 't', '\x00', 'S',
  852. 'e', 'p', 't', 'e', 'm', 'b', 'e', 'r',
  853. '\x00', 'O', 'c', 't', 'o', 'b', 'e', 'r',
  854. '\x00', 'N', 'o', 'v', 'e', 'm', 'b', 'e',
  855. 'r', '\x00', 'D', 'e', 'c', 'e', 'm', 'b',
  856. 'e', 'r', '\x00', 'A', 'M', '\x00', 'P', 'M',
  857. '\x00', '%', 'a', ' ', '%', 'b', ' ', '%',
  858. 'e', ' ', '%', 'H', ':', '%', 'M', ':',
  859. '%', 'S', ' ', '%', 'Y', '\x00', '%', 'm',
  860. '/', '%', 'd', '/', '%', 'y', '\x00', '%',
  861. 'H', ':', '%', 'M', ':', '%', 'S', '\x00',
  862. '%', 'I', ':', '%', 'M', ':', '%', 'S',
  863. ' ', '%', 'p', '\x00', '^', '[', 'y', 'Y',
  864. ']', '\x00', '^', '[', 'n', 'N', ']', '\x00',
  865. };
  866. char *nl_langinfo(nl_item item)
  867. {
  868. unsigned int c;
  869. unsigned int i;
  870. if ((c = _NL_ITEM_CATEGORY(item)) < C_LC_ALL) {
  871. if ((i = cat_start[c] + _NL_ITEM_INDEX(item)) < cat_start[c+1]) {
  872. /* return (char *) C_locale_data + item_offset[i] + (i & 64); */
  873. return (char *) C_locale_data + nl_data[C_LC_ALL+1+i] + 2*(i & 64);
  874. }
  875. }
  876. return (char *) cat_start; /* Conveniently, this is the empty string. */
  877. }
  878. #else /* __LOCALE_C_ONLY */
  879. #if defined(__UCLIBC_HAS_XLOCALE__) && !defined(__UCLIBC_DO_XLOCALE)
  880. char *nl_langinfo(nl_item item)
  881. {
  882. return __nl_langinfo_l(item, __UCLIBC_CURLOCALE);
  883. }
  884. #else /* defined(__UCLIBC_HAS_XLOCALE__) && !defined(__UCLIBC_DO_XLOCALE) */
  885. static const char empty[] = "";
  886. char *__XL(nl_langinfo)(nl_item item __LOCALE_PARAM )
  887. {
  888. unsigned int c = _NL_ITEM_CATEGORY(item);
  889. unsigned int i = _NL_ITEM_INDEX(item);
  890. if ((c < LC_ALL) && (i < __LOCALE_PTR->category_item_count[c])) {
  891. return ((char **)(((char *) __LOCALE_PTR)
  892. + __LOCALE_PTR->category_offsets[c]))[i];
  893. }
  894. return (char *) empty;
  895. }
  896. #endif /* defined(__UCLIBC_HAS_XLOCALE__) && !defined(__UCLIBC_DO_XLOCALE) */
  897. #endif /* __LOCALE_C_ONLY */
  898. #endif
  899. /**********************************************************************/
  900. #ifdef L_newlocale
  901. #ifdef __UCLIBC_MJN3_ONLY__
  902. #warning TODO: Move posix and utf8 strings.
  903. #endif
  904. static const char posix[] = "POSIX";
  905. static const char utf8[] = "UTF-8";
  906. static int find_locale(int category_mask, const char *p,
  907. unsigned char *new_locale)
  908. {
  909. int i;
  910. const unsigned char *s;
  911. uint16_t n;
  912. unsigned char lang_cult, codeset;
  913. #if defined(__LOCALE_DATA_AT_MODIFIERS_LENGTH) && 1
  914. /* Support standard locale handling for @-modifiers. */
  915. #ifdef __UCLIBC_MJN3_ONLY__
  916. #warning REMINDER: Fix buf size in find_locale.
  917. #endif
  918. char buf[18]; /* TODO: 7+{max codeset name length} */
  919. const char *q;
  920. if ((q = strchr(p,'@')) != NULL) {
  921. if ((((size_t)((q-p)-5)) > (sizeof(buf) - 5)) || (p[2] != '_')) {
  922. return 0;
  923. }
  924. /* locale name at least 5 chars long and 3rd char is '_' */
  925. s = LOCALE_AT_MODIFIERS;
  926. do {
  927. if (!strcmp(s+2, q+1)) {
  928. break;
  929. }
  930. s += 2 + *s; /* TODO - fix this throughout */
  931. } while (*s);
  932. if (!*s) {
  933. return 0;
  934. }
  935. assert(q - p < sizeof(buf));
  936. memcpy(buf, p, q-p);
  937. buf[q-p] = 0;
  938. buf[2] = s[1];
  939. p = buf;
  940. }
  941. #endif
  942. lang_cult = codeset = 0; /* Assume C and default codeset. */
  943. if (((*p == 'C') && !p[1]) || !strcmp(p, posix)) {
  944. goto FIND_LOCALE;
  945. }
  946. if ((strlen(p) > 5) && (p[5] == '.')) { /* Codeset in locale name? */
  947. /* TODO: maybe CODESET_LIST + *s ??? */
  948. /* 7bit is 1, UTF-8 is 2, 8-bit is >= 3 */
  949. codeset = 2;
  950. if (strcmp(utf8,p+6) != 0) {/* TODO - fix! */
  951. s = CODESET_LIST;
  952. do {
  953. ++codeset; /* Increment codeset first. */
  954. if (!strcmp(CODESET_LIST+*s, p+6)) {
  955. goto FIND_LANG_CULT;
  956. }
  957. } while (*++s);
  958. return 0; /* No matching codeset! */
  959. }
  960. }
  961. FIND_LANG_CULT: /* Find language_culture number. */
  962. s = LOCALE_NAMES;
  963. do { /* TODO -- do a binary search? */
  964. /* TODO -- fix gen_mmap!*/
  965. ++lang_cult; /* Increment first since C/POSIX is 0. */
  966. if (!strncmp(s,p,5)) { /* Found a matching locale name; */
  967. goto FIND_LOCALE;
  968. }
  969. s += 5;
  970. } while (lang_cult < __LOCALE_DATA_NUM_LOCALE_NAMES);
  971. return 0; /* No matching language_culture! */
  972. FIND_LOCALE: /* Find locale row matching name and codeset */
  973. s = LOCALES;
  974. n = 0;
  975. do { /* TODO -- do a binary search? */
  976. if ((lang_cult == *s) && ((codeset == s[1]) || (codeset == s[2]))) {
  977. i = 1;
  978. s = new_locale + 1;
  979. do {
  980. if (category_mask & i) {
  981. /* Encode current locale row number. */
  982. ((unsigned char *) s)[0] = (n >> 7) | 0x80;
  983. ((unsigned char *) s)[1] = (n & 0x7f) | 0x80;
  984. }
  985. s += 2;
  986. i += i;
  987. } while (i < (1 << LC_ALL));
  988. return i; /* Return non-zero */
  989. }
  990. s += __LOCALE_DATA_WIDTH_LOCALES;
  991. ++n;
  992. } while (n <= __LOCALE_DATA_NUM_LOCALES); /* We started at 1!!! */
  993. return 0; /* Unsupported locale. */
  994. }
  995. static unsigned char *composite_locale(int category_mask, const char *locale,
  996. unsigned char *new_locale)
  997. {
  998. char buf[MAX_LOCALE_STR];
  999. char *t;
  1000. char *e;
  1001. int c;
  1002. int component_mask;
  1003. if (!strchr(locale,'=')) {
  1004. if (!find_locale(category_mask, locale, new_locale)) {
  1005. return NULL;
  1006. }
  1007. return new_locale;
  1008. }
  1009. if (strlen(locale) >= sizeof(buf)) {
  1010. return NULL;
  1011. }
  1012. stpcpy(buf, locale);
  1013. component_mask = 0;
  1014. t = strtok_r(buf, "=", &e); /* This can't fail because of strchr test above. */
  1015. do {
  1016. c = 0;
  1017. while (strcmp(CATEGORY_NAMES + (int) CATEGORY_NAMES[c], t)) {
  1018. if (++c == LC_ALL) { /* Unknown category name! */
  1019. return NULL;
  1020. }
  1021. }
  1022. t = strtok_r(NULL, ";", &e);
  1023. c = (1 << c);
  1024. if (component_mask & c) { /* Multiple components for one category. */
  1025. return NULL;
  1026. }
  1027. component_mask |= c;
  1028. if ((category_mask & c) && (!t || !find_locale(c, t, new_locale))) {
  1029. return NULL;
  1030. }
  1031. } while ((t = strtok_r(NULL, "=", &e)) != NULL);
  1032. if (category_mask & ~component_mask) { /* Category component(s) missing. */
  1033. return NULL;
  1034. }
  1035. return new_locale;
  1036. }
  1037. __locale_t __newlocale(int category_mask, const char *locale, __locale_t base)
  1038. {
  1039. const unsigned char *p;
  1040. int i, j, k;
  1041. unsigned char new_selector[LOCALE_SELECTOR_SIZE];
  1042. if (category_mask == (1 << LC_ALL)) {
  1043. category_mask = LC_ALL_MASK;
  1044. }
  1045. if (!locale || (((unsigned int)(category_mask)) > LC_ALL_MASK)) {
  1046. INVALID:
  1047. __set_errno(EINVAL);
  1048. return NULL; /* No locale or illegal/unsupported category. */
  1049. }
  1050. #ifdef __UCLIBC_MJN3_ONLY__
  1051. #warning TODO: Rename cur_locale to locale_selector.
  1052. #endif
  1053. strcpy((char *) new_selector,
  1054. (base ? (char *) base->cur_locale : C_LOCALE_SELECTOR));
  1055. if (!*locale) { /* locale == "", so check environment. */
  1056. #ifndef __UCLIBC_HAS_THREADS__
  1057. static /* If no threads, then envstr can be static. */
  1058. #endif /* __UCLIBC_HAS_THREADS__ */
  1059. const char *envstr[4] = { "LC_ALL", NULL, "LANG", posix };
  1060. i = 1;
  1061. k = 0;
  1062. do {
  1063. if (category_mask & i) {
  1064. /* Note: SUSv3 doesn't define a fallback mechanism here.
  1065. * So, if LC_ALL is invalid, we do _not_ continue trying
  1066. * the other environment vars. */
  1067. envstr[1] = CATEGORY_NAMES + CATEGORY_NAMES[k];
  1068. j = 0;
  1069. do {
  1070. p = envstr[j];
  1071. } while ((++j < 4) && (!(p = getenv(p)) || !*p));
  1072. /* The user set something... is it valid? */
  1073. /* Note: Since we don't support user-supplied locales and
  1074. * alternate paths, we don't need to worry about special
  1075. * handling for suid/sgid apps. */
  1076. if (!find_locale(i, p, new_selector)) {
  1077. goto INVALID;
  1078. }
  1079. }
  1080. i += i;
  1081. } while (++k < LC_ALL);
  1082. } else if (!composite_locale(category_mask, locale, new_selector)) {
  1083. goto INVALID;
  1084. }
  1085. #ifdef __UCLIBC_MJN3_ONLY__
  1086. #warning TODO: Do a compatible codeset check!
  1087. #endif
  1088. /* If we get here, the new selector corresponds to a valid locale. */
  1089. #ifdef __UCLIBC_MJN3_ONLY__
  1090. #warning CONSIDER: Probably want a _locale_new func to allow for caching of locales.
  1091. #endif
  1092. #if 0
  1093. if (base) {
  1094. _locale_set_l(new_selector, base);
  1095. } else {
  1096. base = _locale_new(new_selector);
  1097. }
  1098. #else
  1099. if (!base) {
  1100. if ((base = malloc(sizeof(__uclibc_locale_t))) == NULL) {
  1101. return base;
  1102. }
  1103. _locale_init_l(base);
  1104. }
  1105. _locale_set_l(new_selector, base);
  1106. #endif
  1107. return base;
  1108. }
  1109. weak_alias(__newlocale, newlocale)
  1110. #endif
  1111. /**********************************************************************/
  1112. #ifdef L_duplocale
  1113. #ifdef __UCLIBC_MJN3_ONLY__
  1114. #warning REMINDER: When we allocate ctype tables, remember to dup them.
  1115. #endif
  1116. __locale_t __duplocale(__locale_t dataset)
  1117. {
  1118. __locale_t r;
  1119. uint16_t * i2w;
  1120. size_t n;
  1121. assert(dataset != LC_GLOBAL_LOCALE);
  1122. if ((r = malloc(sizeof(__uclibc_locale_t))) != NULL) {
  1123. n = 2*dataset->collate.max_col_index+2;
  1124. if ((i2w = calloc(n, sizeof(uint16_t)))
  1125. != NULL
  1126. ) {
  1127. memcpy(r, dataset, sizeof(__uclibc_locale_t));
  1128. r->collate.index2weight = i2w;
  1129. memcpy(i2w, dataset->collate.index2weight, n * sizeof(uint16_t));
  1130. } else {
  1131. free(r);
  1132. r = NULL;
  1133. }
  1134. }
  1135. return r;
  1136. }
  1137. weak_alias(__duplocale, duplocale)
  1138. #endif
  1139. /**********************************************************************/
  1140. #ifdef L_freelocale
  1141. #ifdef __UCLIBC_MJN3_ONLY__
  1142. #warning REMINDER: When we allocate ctype tables, remember to free them.
  1143. #endif
  1144. void __freelocale(__locale_t dataset)
  1145. {
  1146. assert(dataset != __global_locale);
  1147. assert(dataset != LC_GLOBAL_LOCALE);
  1148. free(dataset->collate.index2weight); /* Free collation data. */
  1149. free(dataset); /* Free locale */
  1150. }
  1151. weak_alias(__freelocale, freelocale)
  1152. #endif
  1153. /**********************************************************************/
  1154. #ifdef L_uselocale
  1155. __locale_t __uselocale(__locale_t dataset)
  1156. {
  1157. __locale_t old;
  1158. if (!dataset) {
  1159. old = __UCLIBC_CURLOCALE;
  1160. } else {
  1161. if (dataset == LC_GLOBAL_LOCALE) {
  1162. dataset = __global_locale;
  1163. }
  1164. #ifdef __UCLIBC_HAS_THREADS__
  1165. old = __curlocale_set(dataset);
  1166. #else
  1167. old = __curlocale_var;
  1168. __curlocale_var = dataset;
  1169. #endif
  1170. }
  1171. if (old == __global_locale) {
  1172. return LC_GLOBAL_LOCALE;
  1173. }
  1174. return old;
  1175. }
  1176. weak_alias(__uselocale, uselocale)
  1177. #endif
  1178. /**********************************************************************/
  1179. #ifdef L___curlocale
  1180. #ifdef __UCLIBC_HAS_THREADS__
  1181. __locale_t weak_const_function __curlocale(void)
  1182. {
  1183. return __curlocale_var; /* This is overriden by the thread version. */
  1184. }
  1185. __locale_t weak_function __curlocale_set(__locale_t newloc)
  1186. {
  1187. __locale_t oldloc = __curlocale_var;
  1188. assert(newloc != LC_GLOBAL_LOCALE);
  1189. __curlocale_var = newloc;
  1190. return oldloc;
  1191. }
  1192. #endif
  1193. #endif
  1194. /**********************************************************************/
  1195. #ifdef L___locale_mbrtowc_l
  1196. /* NOTE: This returns an int... not size_t. Also, it is not a general
  1197. * routine. It is actually a very stripped-down version of mbrtowc
  1198. * that takes a __locale_t arg. This is used by strcoll and strxfrm.
  1199. * It is also used above to generate wchar_t versions of the decimal point
  1200. * and thousands seperator. */
  1201. #ifndef __CTYPE_HAS_UTF_8_LOCALES
  1202. #warning __CTYPE_HAS_UTF_8_LOCALES not set!
  1203. #endif
  1204. #ifndef __CTYPE_HAS_8_BIT_LOCALES
  1205. #warning __CTYPE_HAS_8_BIT_LOCALES not set!
  1206. #endif
  1207. #define Cc2wc_IDX_SHIFT __LOCALE_DATA_Cc2wc_IDX_SHIFT
  1208. #define Cc2wc_ROW_LEN __LOCALE_DATA_Cc2wc_ROW_LEN
  1209. extern size_t _wchar_utf8sntowcs(wchar_t *__restrict pwc, size_t wn,
  1210. const char **__restrict src, size_t n,
  1211. mbstate_t *ps, int allow_continuation);
  1212. int __locale_mbrtowc_l(wchar_t *__restrict dst,
  1213. const char *__restrict src,
  1214. __locale_t loc )
  1215. {
  1216. #ifdef __CTYPE_HAS_UTF_8_LOCALES
  1217. if (loc->encoding == __ctype_encoding_utf8) {
  1218. mbstate_t ps;
  1219. const char *p = src;
  1220. size_t r;
  1221. ps.__mask = 0;
  1222. r = _wchar_utf8sntowcs(dst, 1, &p, SIZE_MAX, &ps, 1);
  1223. return (r == 1) ? (p-src) : r; /* Need to return 0 if nul char. */
  1224. }
  1225. #endif
  1226. #ifdef __CTYPE_HAS_8_BIT_LOCALES
  1227. assert((loc->encoding == __ctype_encoding_7_bit) || (loc->encoding == __ctype_encoding_8_bit));
  1228. #else
  1229. assert(loc->encoding == __ctype_encoding_7_bit);
  1230. #endif
  1231. if ((*dst = ((unsigned char)(*src))) < 0x80) { /* ASCII... */
  1232. return (*src != 0);
  1233. }
  1234. #ifdef __CTYPE_HAS_8_BIT_LOCALES
  1235. if (loc->encoding == __ctype_encoding_8_bit) {
  1236. wchar_t wc = *dst - 0x80;
  1237. *dst = loc->tbl8c2wc[
  1238. (loc->idx8c2wc[wc >> Cc2wc_IDX_SHIFT]
  1239. << Cc2wc_IDX_SHIFT) + (wc & (Cc2wc_ROW_LEN - 1))];
  1240. if (*dst) {
  1241. return 1;
  1242. }
  1243. }
  1244. #endif
  1245. return -1;
  1246. }
  1247. #endif
  1248. /**********************************************************************/