locale.c 43 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484
  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. libc_hidden_proto(memcpy)
  57. libc_hidden_proto(memset)
  58. libc_hidden_proto(stpcpy)
  59. libc_hidden_proto(strtok_r)
  60. libc_hidden_proto(strlen)
  61. libc_hidden_proto(strcmp)
  62. libc_hidden_proto(strcpy)
  63. libc_hidden_proto(strncmp)
  64. libc_hidden_proto(strchr)
  65. libc_hidden_proto(getenv)
  66. /*libc_hidden_proto(fflush)*/
  67. #ifdef __UCLIBC_MJN3_ONLY__
  68. #ifdef L_setlocale
  69. #warning TODO: Make the link_warning()s a config option?
  70. #endif
  71. #endif
  72. #undef link_warning
  73. #define link_warning(A,B)
  74. #undef __LOCALE_C_ONLY
  75. #ifndef __UCLIBC_HAS_LOCALE__
  76. #define __LOCALE_C_ONLY
  77. #endif /* __UCLIBC_HAS_LOCALE__ */
  78. #ifdef __LOCALE_C_ONLY
  79. #include <locale.h>
  80. #else /* __LOCALE_C_ONLY */
  81. #ifdef __UCLIBC_MJN3_ONLY__
  82. #ifdef L_setlocale
  83. #warning TODO: Fix the __CTYPE_HAS_8_BIT_LOCALES define at the top of the file.
  84. #warning TODO: Fix __WCHAR_ENABLED.
  85. #endif
  86. #endif
  87. /* Need to include this before locale.h and xlocale.h! */
  88. #include <bits/uClibc_locale.h>
  89. #undef CODESET_LIST
  90. #define CODESET_LIST (__locale_mmap->codeset_list)
  91. #ifdef __UCLIBC_HAS_XLOCALE__
  92. #include <xlocale.h>
  93. #include <locale.h>
  94. #else /* __UCLIBC_HAS_XLOCALE__ */
  95. /* We need this internally... */
  96. #define __UCLIBC_HAS_XLOCALE__ 1
  97. #include <xlocale.h>
  98. #include <locale.h>
  99. #undef __UCLIBC_HAS_XLOCALE__
  100. #endif /* __UCLIBC_HAS_XLOCALE__ */
  101. #include <wchar.h>
  102. #define LOCALE_NAMES (__locale_mmap->locale_names5)
  103. #define LOCALES (__locale_mmap->locales)
  104. #define LOCALE_AT_MODIFIERS (__locale_mmap->locale_at_modifiers)
  105. #define CATEGORY_NAMES (__locale_mmap->lc_names)
  106. #ifdef __UCLIBC_MJN3_ONLY__
  107. #warning REMINDER: redo the MAX_LOCALE_STR stuff...
  108. #endif
  109. #define MAX_LOCALE_STR 256 /* TODO: Only sufficient for current case. */
  110. #define MAX_LOCALE_CATEGORY_STR 32 /* TODO: Only sufficient for current case. */
  111. /* Note: Best if MAX_LOCALE_CATEGORY_STR is a power of 2. */
  112. extern int _locale_set_l(const unsigned char *p, __locale_t base) attribute_hidden;
  113. extern void _locale_init_l(__locale_t base) attribute_hidden;
  114. #endif /* __LOCALE_C_ONLY */
  115. #undef LOCALE_STRING_SIZE
  116. #define LOCALE_SELECTOR_SIZE (2 * __LC_ALL + 2)
  117. #ifdef __UCLIBC_MJN3_ONLY__
  118. #ifdef L_setlocale
  119. #warning TODO: Create a C locale selector string.
  120. #endif
  121. #endif
  122. #define C_LOCALE_SELECTOR "\x23\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80\x80"
  123. #include <langinfo.h>
  124. #include <nl_types.h>
  125. /**********************************************************************/
  126. #ifdef L_setlocale
  127. #ifdef __LOCALE_C_ONLY
  128. link_warning(setlocale,"REMINDER: The 'setlocale' function supports only C|POSIX locales.")
  129. static const char C_string[] = "C";
  130. char *setlocale(int category, register const char *locale)
  131. {
  132. return ( (((unsigned int)(category)) <= LC_ALL)
  133. && ( (!locale) /* Request for locale category string. */
  134. || (!*locale) /* Implementation-defined default is C. */
  135. || ((*locale == 'C') && !locale[1])
  136. || (!strcmp(locale, "POSIX"))) )
  137. ? (char *) C_string /* Always in C/POSIX locale. */
  138. : NULL;
  139. }
  140. #else /* ---------------------------------------------- __LOCALE_C_ONLY */
  141. #ifdef __UCLIBC_HAS_THREADS__
  142. link_warning(setlocale,"REMINDER: The 'setlocale' function is _not_ threadsafe except for simple queries.")
  143. #endif
  144. #if !defined(__LOCALE_DATA_NUM_LOCALES) || (__LOCALE_DATA_NUM_LOCALES <= 1)
  145. #error locales enabled, but not data other than for C locale!
  146. #endif
  147. #ifdef __UCLIBC_MJN3_ONLY__
  148. #warning TODO: Move posix and utf8 strings.
  149. #endif
  150. static const char posix[] = "POSIX";
  151. static const char utf8[] = "UTF-8";
  152. #ifdef __UCLIBC_MJN3_ONLY__
  153. #warning TODO: Fix dimensions of hr_locale.
  154. #endif
  155. /* Individual category strings start at hr_locale + category * MAX_LOCALE_CATEGORY.
  156. * This holds for LC_ALL as well.
  157. */
  158. static char hr_locale[(MAX_LOCALE_CATEGORY_STR * LC_ALL) + MAX_LOCALE_STR];
  159. libc_hidden_proto(newlocale)
  160. static void update_hr_locale(const unsigned char *spec)
  161. {
  162. const unsigned char *loc;
  163. const unsigned char *s;
  164. char *n;
  165. int i, category, done;
  166. done = category = 0;
  167. do {
  168. s = spec + 1;
  169. n = hr_locale + category * MAX_LOCALE_CATEGORY_STR;
  170. if (category == LC_ALL) {
  171. done = 1;
  172. for (i = 0 ; i < LC_ALL-1 ; i += 2) {
  173. if ((s[i] != s[i+2]) || (s[i+1] != s[i+3])) {
  174. goto SKIP;
  175. }
  176. }
  177. /* All categories the same, so simplify string by using a single
  178. * category. */
  179. category = LC_CTYPE;
  180. }
  181. SKIP:
  182. i = (category == LC_ALL) ? 0 : category;
  183. s += 2*i;
  184. do {
  185. if ((*s != 0xff) || (s[1] != 0xff)) {
  186. loc = LOCALES
  187. + __LOCALE_DATA_WIDTH_LOCALES * ((((int)(*s & 0x7f)) << 7)
  188. + (s[1] & 0x7f));
  189. if (category == LC_ALL) {
  190. n = stpcpy(n, CATEGORY_NAMES + (int) CATEGORY_NAMES[i]);
  191. *n++ = '=';
  192. }
  193. if (*loc == 0) {
  194. *n++ = 'C';
  195. *n = 0;
  196. } else {
  197. char at = 0;
  198. memcpy(n, LOCALE_NAMES + 5*((*loc)-1), 5);
  199. if (n[2] != '_') {
  200. at = n[2];
  201. n[2] = '_';
  202. }
  203. n += 5;
  204. *n++ = '.';
  205. if (loc[2] == 2) {
  206. n = stpcpy(n, utf8);
  207. } else if (loc[2] >= 3) {
  208. n = stpcpy(n, CODESET_LIST + (int)(CODESET_LIST[loc[2] - 3]));
  209. }
  210. if (at) {
  211. const char *q;
  212. *n++ = '@';
  213. q = LOCALE_AT_MODIFIERS;
  214. do {
  215. if (q[1] == at) {
  216. n = stpcpy(n, q+2);
  217. break;
  218. }
  219. q += 2 + *q;
  220. } while (*q);
  221. }
  222. }
  223. *n++ = ';';
  224. }
  225. s += 2;
  226. } while (++i < category);
  227. *--n = 0; /* Remove trailing ';' and nul-terminate. */
  228. ++category;
  229. } while (!done);
  230. }
  231. char *setlocale(int category, const char *locale)
  232. {
  233. if (((unsigned int)(category)) > LC_ALL) {
  234. #if 0
  235. __set_errno(EINVAL); /* glibc sets errno -- SUSv3 doesn't say. */
  236. #endif
  237. return NULL; /* Illegal/unsupported category. */
  238. }
  239. if (locale != NULL) { /* Not just a query... */
  240. if (!newlocale((1 << category), locale, __global_locale)) {
  241. return NULL; /* Failed! */
  242. }
  243. update_hr_locale(__global_locale->cur_locale);
  244. }
  245. /* Either a query or a successful set, so return current locale string. */
  246. return hr_locale + (category * MAX_LOCALE_CATEGORY_STR);
  247. }
  248. #endif /* __LOCALE_C_ONLY */
  249. #endif
  250. /**********************************************************************/
  251. #ifdef L_localeconv
  252. /* Note: We assume here that the compiler does the sane thing regarding
  253. * placement of the fields in the struct. If necessary, we could ensure
  254. * this usings an array of offsets but at some size cost. */
  255. #ifdef __LOCALE_C_ONLY
  256. link_warning(localeconv,"REMINDER: The 'localeconv' function is hardwired for C/POSIX locale only.")
  257. static struct lconv the_lconv;
  258. static const char decpt[] = ".";
  259. struct lconv *localeconv(void)
  260. {
  261. register char *p = (char *)(&the_lconv);
  262. *((char **)p) = (char *) decpt;
  263. do {
  264. p += sizeof(char **);
  265. *((char **)p) = (char *) (decpt+1);
  266. } while (p < (char *) &the_lconv.negative_sign);
  267. p = (&the_lconv.int_frac_digits);
  268. do {
  269. *p = CHAR_MAX;
  270. ++p;
  271. } while (p <= &the_lconv.int_n_sign_posn);
  272. return &the_lconv;
  273. }
  274. #else /* __LOCALE_C_ONLY */
  275. static struct lconv the_lconv;
  276. struct lconv *localeconv(void)
  277. {
  278. register char *p = (char *) &the_lconv;
  279. register char **q = (char **) &(__UCLIBC_CURLOCALE_DATA).decimal_point;
  280. do {
  281. *((char **)p) = *q;
  282. p += sizeof(char **);
  283. ++q;
  284. } while (p < &the_lconv.int_frac_digits);
  285. do {
  286. *p = **q;
  287. ++p;
  288. ++q;
  289. } while (p <= &the_lconv.int_n_sign_posn);
  290. return &the_lconv;
  291. }
  292. #endif /* __LOCALE_C_ONLY */
  293. #endif
  294. /**********************************************************************/
  295. #if defined(L__locale_init) && !defined(__LOCALE_C_ONLY)
  296. __uclibc_locale_t __global_locale_data;
  297. __locale_t __global_locale = &__global_locale_data;
  298. #ifdef __UCLIBC_HAS_XLOCALE__
  299. __locale_t __curlocale_var = &__global_locale_data;
  300. #endif
  301. /*----------------------------------------------------------------------*/
  302. #ifdef __UCLIBC_MJN3_ONLY__
  303. #warning TODO: Move utf8 and ascii strings.
  304. #endif
  305. static const char utf8[] = "UTF-8";
  306. static const char ascii[] = "ASCII";
  307. typedef struct {
  308. uint16_t num_base;
  309. uint16_t num_der;
  310. uint16_t MAX_WEIGHTS;
  311. uint16_t num_index2weight;
  312. #define num_index2ruleidx num_index2weight
  313. uint16_t num_weightstr;
  314. uint16_t num_multistart;
  315. uint16_t num_override;
  316. uint16_t num_ruletable;
  317. } coldata_header_t;
  318. typedef struct {
  319. uint16_t num_weights;
  320. uint16_t num_starters;
  321. uint16_t ii_shift;
  322. uint16_t ti_shift;
  323. uint16_t ii_len;
  324. uint16_t ti_len;
  325. uint16_t max_weight;
  326. uint16_t num_col_base;
  327. uint16_t max_col_index;
  328. uint16_t undefined_idx;
  329. uint16_t range_low;
  330. uint16_t range_count;
  331. uint16_t range_base_weight;
  332. uint16_t range_rule_offset;
  333. uint16_t index2weight_offset;
  334. uint16_t index2ruleidx_offset;
  335. uint16_t multistart_offset;
  336. uint16_t wcs2colidt_offset_low;
  337. uint16_t wcs2colidt_offset_hi;
  338. } coldata_base_t;
  339. typedef struct {
  340. uint16_t base_idx;
  341. uint16_t undefined_idx;
  342. uint16_t overrides_offset;
  343. uint16_t multistart_offset;
  344. } coldata_der_t;
  345. static int init_cur_collate(int der_num, __collate_t *cur_collate)
  346. {
  347. const uint16_t *__locale_collate_tbl = __locale_mmap->collate_data;
  348. coldata_header_t *cdh;
  349. coldata_base_t *cdb;
  350. coldata_der_t *cdd;
  351. const uint16_t *p;
  352. size_t n;
  353. uint16_t i, w;
  354. #ifdef __UCLIBC_MJN3_ONLY__
  355. #warning kill of x86-specific asserts
  356. #endif
  357. #if 0
  358. assert(sizeof(coldata_base_t) == 19*2);
  359. assert(sizeof(coldata_der_t) == 4*2);
  360. assert(sizeof(coldata_header_t) == 8*2);
  361. #endif
  362. if (!der_num) { /* C locale... special */
  363. cur_collate->num_weights = 0;
  364. return 1;
  365. }
  366. --der_num;
  367. cdh = (coldata_header_t *) __locale_collate_tbl;
  368. #ifdef __UCLIBC_MJN3_ONLY__
  369. #warning CONSIDER: Should we assert here?
  370. #endif
  371. #if 0
  372. if (der_num >= cdh->num_der) {
  373. return 0;
  374. }
  375. #else
  376. assert((der_num < cdh->num_der));
  377. #endif
  378. cdd = (coldata_der_t *)(__locale_collate_tbl
  379. + (sizeof(coldata_header_t)
  380. + cdh->num_base * sizeof(coldata_base_t)
  381. + der_num * sizeof(coldata_der_t)
  382. )/2 );
  383. cdb = (coldata_base_t *)(__locale_collate_tbl
  384. + (sizeof(coldata_header_t)
  385. + cdd->base_idx * sizeof(coldata_base_t)
  386. )/2 );
  387. memcpy(cur_collate, cdb, offsetof(coldata_base_t,index2weight_offset));
  388. cur_collate->undefined_idx = cdd->undefined_idx;
  389. cur_collate->ti_mask = (1 << cur_collate->ti_shift)-1;
  390. cur_collate->ii_mask = (1 << cur_collate->ii_shift)-1;
  391. /* fflush(stdout); */
  392. /* fprintf(stderr,"base=%d num_col_base: %d %d\n", cdd->base_idx ,cur_collate->num_col_base, cdb->num_col_base); */
  393. n = (sizeof(coldata_header_t) + cdh->num_base * sizeof(coldata_base_t)
  394. + cdh->num_der * sizeof(coldata_der_t))/2;
  395. /* fprintf(stderr,"n = %d\n", n); */
  396. cur_collate->index2weight_tbl = __locale_collate_tbl + n + cdb->index2weight_offset;
  397. /* fprintf(stderr,"i2w = %d\n", n + cdb->index2weight_offset); */
  398. n += cdh->num_index2weight;
  399. cur_collate->index2ruleidx_tbl = __locale_collate_tbl + n + cdb->index2ruleidx_offset;
  400. /* fprintf(stderr,"i2r = %d\n", n + cdb->index2ruleidx_offset); */
  401. n += cdh->num_index2ruleidx;
  402. cur_collate->multistart_tbl = __locale_collate_tbl + n + cdd->multistart_offset;
  403. /* fprintf(stderr,"mts = %d\n", n + cdb->multistart_offset); */
  404. n += cdh->num_multistart;
  405. cur_collate->overrides_tbl = __locale_collate_tbl + n + cdd->overrides_offset;
  406. /* fprintf(stderr,"ovr = %d\n", n + cdd->overrides_offset); */
  407. n += cdh->num_override;
  408. cur_collate->ruletable = __locale_collate_tbl + n;
  409. /* fprintf(stderr, "rtb = %d\n", n); */
  410. n += cdh->num_ruletable;
  411. cur_collate->weightstr = __locale_collate_tbl + n;
  412. /* fprintf(stderr,"wts = %d\n", n); */
  413. n += cdh->num_weightstr;
  414. cur_collate->wcs2colidt_tbl = __locale_collate_tbl + n
  415. + (((unsigned long)(cdb->wcs2colidt_offset_hi)) << 16)
  416. + cdb->wcs2colidt_offset_low;
  417. /* fprintf(stderr,"wcs = %lu\n", n + (((unsigned long)(cdb->wcs2colidt_offset_hi)) << 16) */
  418. /* + cdb->wcs2colidt_offset_low); */
  419. cur_collate->MAX_WEIGHTS = cdh->MAX_WEIGHTS;
  420. #ifdef __UCLIBC_MJN3_ONLY__
  421. #warning CONSIDER: Fix the +1 by increasing max_col_index?
  422. #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
  423. #endif
  424. cur_collate->index2weight = calloc(2*cur_collate->max_col_index+2,
  425. sizeof(uint16_t));
  426. if (!cur_collate->index2weight) {
  427. return 0;
  428. }
  429. cur_collate->index2ruleidx = cur_collate->index2weight
  430. + cur_collate->max_col_index + 1;
  431. memcpy(cur_collate->index2weight, cur_collate->index2weight_tbl,
  432. cur_collate->num_col_base * sizeof(uint16_t));
  433. memcpy(cur_collate->index2ruleidx, cur_collate->index2ruleidx_tbl,
  434. cur_collate->num_col_base * sizeof(uint16_t));
  435. /* now do the overrides */
  436. p = cur_collate->overrides_tbl;
  437. while (*p > 1) {
  438. /* fprintf(stderr, "processing override -- count = %d\n", *p); */
  439. n = *p++;
  440. w = *p++;
  441. do {
  442. i = *p++;
  443. /* fprintf(stderr, " i=%d (%#x) w=%d *p=%d\n", i, i, w, *p); */
  444. cur_collate->index2weight[i-1] = w++;
  445. cur_collate->index2ruleidx[i-1] = *p++;
  446. } while (--n);
  447. }
  448. assert(*p == 1);
  449. while (*++p) {
  450. i = *p;
  451. /* fprintf(stderr, " i=%d (%#x) w=%d *p=%d\n", i, i, p[1], p[2]); */
  452. cur_collate->index2weight[i-1] = *++p;
  453. cur_collate->index2ruleidx[i-1] = *++p;
  454. }
  455. for (i=0 ; i < cur_collate->multistart_tbl[0] ; i++) {
  456. p = cur_collate->multistart_tbl;
  457. /* fprintf(stderr, "%2d of %2d: %d ", i, cur_collate->multistart_tbl[0], p[i]); */
  458. p += p[i];
  459. do {
  460. n = *p++;
  461. do {
  462. if (!*p) { /* found it */
  463. /* fprintf(stderr, "found: n=%d (%#lx) |%.*ls|\n", n, (int) *cs->s, n, cs->s); */
  464. /* fprintf(stderr, ": %d - single\n", n); */
  465. goto FOUND;
  466. }
  467. /* the lookup check here is safe since we're assured that *p is a valid colidex */
  468. /* fprintf(stderr, "lookup(%lc)==%d *p==%d\n", cs->s[n], lookup(cs->s[n]), (int) *p); */
  469. /* fprintf(stderr, ": %d - ", n); */
  470. do {
  471. /* fprintf(stderr, "%d|", *p); */
  472. } while (*p++);
  473. break;
  474. } while (1);
  475. } while (1);
  476. FOUND:
  477. continue;
  478. }
  479. return 1;
  480. }
  481. int attribute_hidden _locale_set_l(const unsigned char *p, __locale_t base)
  482. {
  483. const char **x;
  484. unsigned char *s = base->cur_locale + 1;
  485. const size_t *stp;
  486. const unsigned char *r;
  487. const uint16_t *io;
  488. const uint16_t *ii;
  489. const unsigned char *d;
  490. int row; /* locale row */
  491. int crow; /* category row */
  492. int len;
  493. int c;
  494. int i = 0;
  495. __collate_t newcol;
  496. ++p;
  497. newcol.index2weight = NULL;
  498. if ((p[2*LC_COLLATE] != s[2*LC_COLLATE])
  499. || (p[2*LC_COLLATE + 1] != s[2*LC_COLLATE + 1])
  500. ) {
  501. row = (((int)(*p & 0x7f)) << 7) + (p[1] & 0x7f);
  502. assert(row < __LOCALE_DATA_NUM_LOCALES);
  503. if (!init_cur_collate(__locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES
  504. * row + 3 + LC_COLLATE ],
  505. &newcol)
  506. ) {
  507. return 0; /* calloc failed. */
  508. }
  509. free(base->collate.index2weight);
  510. memcpy(&base->collate, &newcol, sizeof(__collate_t));
  511. }
  512. do {
  513. if ((*p != *s) || (p[1] != s[1])) {
  514. row = (((int)(*p & 0x7f)) << 7) + (p[1] & 0x7f);
  515. assert(row < __LOCALE_DATA_NUM_LOCALES);
  516. *s = *p;
  517. s[1] = p[1];
  518. if ((i != LC_COLLATE)
  519. && ((len = __locale_mmap->lc_common_item_offsets_LEN[i]) != 0)
  520. ) {
  521. crow = __locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES * row
  522. + 3 + i ]
  523. * len;
  524. x = (const char **)(((char *) base)
  525. + base->category_offsets[i]);
  526. stp = __locale_mmap->lc_common_tbl_offsets + 4*i;
  527. r = (const unsigned char *)( ((char *)__locale_mmap) + *stp );
  528. io = (const uint16_t *)( ((char *)__locale_mmap) + *++stp );
  529. ii = (const uint16_t *)( ((char *)__locale_mmap) + *++stp );
  530. d = (const unsigned char *)( ((char *)__locale_mmap) + *++stp );
  531. for (c=0 ; c < len ; c++) {
  532. *(x + c) = d + ii[ r[crow + c] + io[c] ];
  533. }
  534. }
  535. if (i == LC_CTYPE) {
  536. c = __locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES * row
  537. + 2 ]; /* codeset */
  538. if (c <= 2) {
  539. if (c == 2) {
  540. base->codeset = utf8;
  541. base->encoding = __ctype_encoding_utf8;
  542. /* TODO - fix for bcc */
  543. base->mb_cur_max = 6;
  544. } else {
  545. assert(c==1);
  546. base->codeset = ascii;
  547. base->encoding = __ctype_encoding_7_bit;
  548. base->mb_cur_max = 1;
  549. }
  550. } else {
  551. const __codeset_8_bit_t *c8b;
  552. r = CODESET_LIST;
  553. base->codeset = r + r[c -= 3];
  554. base->encoding = __ctype_encoding_8_bit;
  555. #ifdef __UCLIBC_MJN3_ONLY__
  556. #warning REMINDER: update 8 bit mb_cur_max when translit implemented!
  557. #endif
  558. /* TODO - update when translit implemented! */
  559. base->mb_cur_max = 1;
  560. c8b = __locale_mmap->codeset_8_bit + c;
  561. #ifdef __CTYPE_HAS_8_BIT_LOCALES
  562. base->idx8ctype = c8b->idx8ctype;
  563. base->idx8uplow = c8b->idx8uplow;
  564. #ifdef __UCLIBC_HAS_WCHAR__
  565. base->idx8c2wc = c8b->idx8c2wc;
  566. base->idx8wc2c = c8b->idx8wc2c;
  567. /* translit */
  568. #endif /* __UCLIBC_HAS_WCHAR__ */
  569. /* What follows is fairly bloated, but it is just a hack
  570. * to get the 8-bit codeset ctype stuff functioning.
  571. * All of this will be replaced in the next generation
  572. * of locale support anyway... */
  573. memcpy(base->__ctype_b_data,
  574. __C_ctype_b - __UCLIBC_CTYPE_B_TBL_OFFSET,
  575. (256 + __UCLIBC_CTYPE_B_TBL_OFFSET)
  576. * sizeof(__ctype_mask_t));
  577. memcpy(base->__ctype_tolower_data,
  578. __C_ctype_tolower - __UCLIBC_CTYPE_TO_TBL_OFFSET,
  579. (256 + __UCLIBC_CTYPE_TO_TBL_OFFSET)
  580. * sizeof(__ctype_touplow_t));
  581. memcpy(base->__ctype_toupper_data,
  582. __C_ctype_toupper - __UCLIBC_CTYPE_TO_TBL_OFFSET,
  583. (256 + __UCLIBC_CTYPE_TO_TBL_OFFSET)
  584. * sizeof(__ctype_touplow_t));
  585. #define Cctype_TBL_MASK ((1 << __LOCALE_DATA_Cctype_IDX_SHIFT) - 1)
  586. #define Cctype_IDX_OFFSET (128 >> __LOCALE_DATA_Cctype_IDX_SHIFT)
  587. {
  588. int u;
  589. __ctype_mask_t m;
  590. for (u=0 ; u < 128 ; u++) {
  591. #ifdef __LOCALE_DATA_Cctype_PACKED
  592. c = base->tbl8ctype
  593. [ ((int)(c8b->idx8ctype
  594. [(u >> __LOCALE_DATA_Cctype_IDX_SHIFT) ])
  595. << (__LOCALE_DATA_Cctype_IDX_SHIFT - 1))
  596. + ((u & Cctype_TBL_MASK) >> 1)];
  597. c = (u & 1) ? (c >> 4) : (c & 0xf);
  598. #else
  599. c = base->tbl8ctype
  600. [ ((int)(c8b->idx8ctype
  601. [(u >> __LOCALE_DATA_Cctype_IDX_SHIFT) ])
  602. << __LOCALE_DATA_Cctype_IDX_SHIFT)
  603. + (u & Cctype_TBL_MASK) ];
  604. #endif
  605. m = base->code2flag[c];
  606. base->__ctype_b_data
  607. [128 + __UCLIBC_CTYPE_B_TBL_OFFSET + u]
  608. = m;
  609. #ifdef __UCLIBC_HAS_CTYPE_SIGNED__
  610. if (((signed char)(128 + u)) != -1) {
  611. base->__ctype_b_data[__UCLIBC_CTYPE_B_TBL_OFFSET
  612. + ((signed char)(128 + u))]
  613. = m;
  614. }
  615. #endif
  616. base->__ctype_tolower_data
  617. [128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
  618. = 128 + u;
  619. base->__ctype_toupper_data
  620. [128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
  621. = 128 + u;
  622. if (m & (_ISlower|_ISupper)) {
  623. c = base->tbl8uplow
  624. [ ((int)(c8b->idx8uplow
  625. [u >> __LOCALE_DATA_Cuplow_IDX_SHIFT])
  626. << __LOCALE_DATA_Cuplow_IDX_SHIFT)
  627. + ((128 + u)
  628. & ((1 << __LOCALE_DATA_Cuplow_IDX_SHIFT)
  629. - 1)) ];
  630. if (m & _ISlower) {
  631. base->__ctype_toupper_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_toupper_data
  637. [__UCLIBC_CTYPE_TO_TBL_OFFSET
  638. + ((signed char)(128 + u))]
  639. = (unsigned char)(128 + u + c);
  640. }
  641. #endif
  642. } else {
  643. base->__ctype_tolower_data
  644. [128 + __UCLIBC_CTYPE_TO_TBL_OFFSET + u]
  645. = (unsigned char)(128 + u - c);
  646. #ifdef __UCLIBC_HAS_CTYPE_SIGNED__
  647. if (((signed char)(128 + u)) != -1) {
  648. base->__ctype_tolower_data
  649. [__UCLIBC_CTYPE_TO_TBL_OFFSET
  650. + ((signed char)(128 + u))]
  651. = (unsigned char)(128 + u - c);
  652. }
  653. #endif
  654. }
  655. }
  656. }
  657. }
  658. #ifdef __UCLIBC_HAS_XLOCALE__
  659. base->__ctype_b = base->__ctype_b_data
  660. + __UCLIBC_CTYPE_B_TBL_OFFSET;
  661. base->__ctype_tolower = base->__ctype_tolower_data
  662. + __UCLIBC_CTYPE_TO_TBL_OFFSET;
  663. base->__ctype_toupper = base->__ctype_toupper_data
  664. + __UCLIBC_CTYPE_TO_TBL_OFFSET;
  665. #else /* __UCLIBC_HAS_XLOCALE__ */
  666. __ctype_b = base->__ctype_b_data
  667. + __UCLIBC_CTYPE_B_TBL_OFFSET;
  668. __ctype_tolower = base->__ctype_tolower_data
  669. + __UCLIBC_CTYPE_TO_TBL_OFFSET;
  670. __ctype_toupper = base->__ctype_toupper_data
  671. + __UCLIBC_CTYPE_TO_TBL_OFFSET;
  672. #endif /* __UCLIBC_HAS_XLOCALE__ */
  673. #endif /* __CTYPE_HAS_8_BIT_LOCALES */
  674. }
  675. #ifdef __UCLIBC_MJN3_ONLY__
  676. #warning TODO: Put the outdigit string length in the locale_mmap object.
  677. #endif
  678. d = base->outdigit_length;
  679. x = &base->outdigit0_mb;
  680. for (c = 0 ; c < 10 ; c++) {
  681. ((unsigned char *)d)[c] = strlen(x[c]);
  682. assert(d[c] > 0);
  683. }
  684. } else if (i == LC_NUMERIC) {
  685. assert(LC_NUMERIC > LC_CTYPE); /* Need ctype initialized. */
  686. base->decimal_point_len
  687. = __locale_mbrtowc_l(&base->decimal_point_wc,
  688. base->decimal_point, base);
  689. assert(base->decimal_point_len > 0);
  690. assert(base->decimal_point[base->decimal_point_len] == 0);
  691. if (*base->grouping) {
  692. base->thousands_sep_len
  693. = __locale_mbrtowc_l(&base->thousands_sep_wc,
  694. base->thousands_sep, base);
  695. #if 1
  696. #ifdef __UCLIBC_MJN3_ONLY__
  697. #warning TODO: Remove hack involving grouping without a thousep char (bg_BG).
  698. #endif
  699. assert(base->thousands_sep_len >= 0);
  700. if (base->thousands_sep_len == 0) {
  701. base->grouping = base->thousands_sep; /* empty string */
  702. }
  703. assert(base->thousands_sep[base->thousands_sep_len] == 0);
  704. #else
  705. assert(base->thousands_sep_len > 0);
  706. assert(base->thousands_sep[base->thousands_sep_len] == 0);
  707. #endif
  708. }
  709. /* } else if (i == LC_COLLATE) { */
  710. /* init_cur_collate(__locale_mmap->locales[ __LOCALE_DATA_WIDTH_LOCALES */
  711. /* * row + 3 + i ], */
  712. /* &base->collate); */
  713. }
  714. }
  715. ++i;
  716. p += 2;
  717. s += 2;
  718. } while (i < LC_ALL);
  719. return 1;
  720. }
  721. static const uint16_t __code2flag[16] = {
  722. 0, /* unclassified = 0 */
  723. _ISprint|_ISgraph|_ISalnum|_ISalpha, /* alpha_nonupper_nonlower */
  724. _ISprint|_ISgraph|_ISalnum|_ISalpha|_ISlower, /* alpha_lower */
  725. _ISprint|_ISgraph|_ISalnum|_ISalpha|_ISlower|_ISupper, /* alpha_upper_lower */
  726. _ISprint|_ISgraph|_ISalnum|_ISalpha|_ISupper, /* alpha_upper */
  727. _ISprint|_ISgraph|_ISalnum|_ISdigit, /* digit */
  728. _ISprint|_ISgraph|_ISpunct, /* punct */
  729. _ISprint|_ISgraph, /* graph */
  730. _ISprint|_ISspace, /* print_space_nonblank */
  731. _ISprint|_ISspace|_ISblank, /* print_space_blank */
  732. _ISspace, /* space_nonblank_noncntrl */
  733. _ISspace|_ISblank, /* space_blank_noncntrl */
  734. _IScntrl|_ISspace, /* cntrl_space_nonblank */
  735. _IScntrl|_ISspace|_ISblank, /* cntrl_space_blank */
  736. _IScntrl /* cntrl_nonspace */
  737. };
  738. void attribute_hidden _locale_init_l(__locale_t base)
  739. {
  740. memset(base->cur_locale, 0, LOCALE_SELECTOR_SIZE);
  741. base->cur_locale[0] = '#';
  742. memcpy(base->category_item_count,
  743. __locale_mmap->lc_common_item_offsets_LEN,
  744. LC_ALL);
  745. ++base->category_item_count[0]; /* Increment for codeset entry. */
  746. base->category_offsets[0] = offsetof(__uclibc_locale_t, outdigit0_mb);
  747. base->category_offsets[1] = offsetof(__uclibc_locale_t, decimal_point);
  748. base->category_offsets[2] = offsetof(__uclibc_locale_t, int_curr_symbol);
  749. base->category_offsets[3] = offsetof(__uclibc_locale_t, abday_1);
  750. /* base->category_offsets[4] = offsetof(__uclibc_locale_t, collate???); */
  751. base->category_offsets[5] = offsetof(__uclibc_locale_t, yesexpr);
  752. #ifdef __CTYPE_HAS_8_BIT_LOCALES
  753. base->tbl8ctype
  754. = (const unsigned char *) &__locale_mmap->tbl8ctype;
  755. base->tbl8uplow
  756. = (const unsigned char *) &__locale_mmap->tbl8uplow;
  757. #ifdef __UCLIBC_HAS_WCHAR__
  758. base->tbl8c2wc
  759. = (const uint16_t *) &__locale_mmap->tbl8c2wc;
  760. base->tbl8wc2c
  761. = (const unsigned char *) &__locale_mmap->tbl8wc2c;
  762. /* translit */
  763. #endif /* __UCLIBC_HAS_WCHAR__ */
  764. #endif /* __CTYPE_HAS_8_BIT_LOCALES */
  765. #ifdef __UCLIBC_HAS_WCHAR__
  766. base->tblwctype
  767. = (const unsigned char *) &__locale_mmap->tblwctype;
  768. base->tblwuplow
  769. = (const unsigned char *) &__locale_mmap->tblwuplow;
  770. base->tblwuplow_diff
  771. = (const uint16_t *) &__locale_mmap->tblwuplow_diff;
  772. /* base->tblwcomb */
  773. /* = (const unsigned char *) &__locale_mmap->tblwcomb; */
  774. /* width?? */
  775. #endif /* __UCLIBC_HAS_WCHAR__ */
  776. /* Initially, set things up to use the global C ctype tables.
  777. * This is correct for C (ASCII) and UTF-8 based locales (except tr_TR). */
  778. #ifdef __UCLIBC_HAS_XLOCALE__
  779. base->__ctype_b = __C_ctype_b;
  780. base->__ctype_tolower = __C_ctype_tolower;
  781. base->__ctype_toupper = __C_ctype_toupper;
  782. #else /* __UCLIBC_HAS_XLOCALE__ */
  783. __ctype_b = __C_ctype_b;
  784. __ctype_tolower = __C_ctype_tolower;
  785. __ctype_toupper = __C_ctype_toupper;
  786. #endif /* __UCLIBC_HAS_XLOCALE__ */
  787. #ifdef __UCLIBC_MJN3_ONLY__
  788. #warning TODO: Initialize code2flag correctly based on locale_mmap.
  789. #endif
  790. base->code2flag = __code2flag;
  791. _locale_set_l(C_LOCALE_SELECTOR, base);
  792. }
  793. void attribute_hidden _locale_init(void)
  794. {
  795. /* TODO: mmap the locale file */
  796. /* TODO - ??? */
  797. _locale_init_l(__global_locale);
  798. }
  799. #endif
  800. /**********************************************************************/
  801. #if defined(L_nl_langinfo) || defined(L_nl_langinfo_l)
  802. #ifdef __LOCALE_C_ONLY
  803. /* We need to index 320 bytes of data, so you might initially think we
  804. * need to store the offsets in shorts. But since the offset of the
  805. * 64th item is 182, we'll store "offset - 2*64" for all items >= 64
  806. * and always calculate the data offset as "offset[i] + 2*(i & 64)".
  807. * This allows us to pack the data offsets in an unsigned char while
  808. * also avoiding an "if".
  809. *
  810. * Note: Category order is assumed to be:
  811. * ctype, numeric, monetary, time, collate, messages, all
  812. */
  813. #define C_LC_ALL 6
  814. /* Combine the data to avoid size penalty for seperate char arrays when
  815. * compiler aligns objects. The original code is left in as documentation. */
  816. #define cat_start nl_data
  817. #define C_locale_data (nl_data + C_LC_ALL + 1 + 90)
  818. static const unsigned char nl_data[C_LC_ALL + 1 + 90 + 320] = {
  819. /* static const char cat_start[LC_ALL + 1] = { */
  820. '\x00', '\x0b', '\x0e', '\x24', '\x56', '\x56', '\x5a',
  821. /* }; */
  822. /* static const char item_offset[90] = { */
  823. '\x00', '\x02', '\x04', '\x06', '\x08', '\x0a', '\x0c', '\x0e',
  824. '\x10', '\x12', '\x14', '\x1a', '\x1b', '\x1b', '\x1b', '\x1b',
  825. '\x1b', '\x1b', '\x1b', '\x1b', '\x1b', '\x1c', '\x1c', '\x1c',
  826. '\x1c', '\x1c', '\x1c', '\x1c', '\x1c', '\x1c', '\x1c', '\x1c',
  827. '\x1c', '\x1c', '\x1c', '\x1e', '\x20', '\x24', '\x28', '\x2c',
  828. '\x30', '\x34', '\x38', '\x3c', '\x43', '\x4a', '\x52', '\x5c',
  829. '\x65', '\x6c', '\x75', '\x79', '\x7d', '\x81', '\x85', '\x89',
  830. '\x8d', '\x91', '\x95', '\x99', '\x9d', '\xa1', '\xa5', '\xad',
  831. '\x36', '\x3c', '\x42', '\x46', '\x4b', '\x50', '\x57', '\x61',
  832. '\x69', '\x72', '\x7b', '\x7e', '\x81', '\x96', '\x9f', '\xa8',
  833. '\xb3', '\xb3', '\xb3', '\xb3', '\xb3', '\xb3', '\xb4', '\xba',
  834. '\xbf', '\xbf',
  835. /* }; */
  836. /* static const char C_locale_data[320] = { */
  837. '0', '\x00', '1', '\x00', '2', '\x00', '3', '\x00',
  838. '4', '\x00', '5', '\x00', '6', '\x00', '7', '\x00',
  839. '8', '\x00', '9', '\x00', 'A', 'S', 'C', 'I',
  840. 'I', '\x00', '.', '\x00', '\x7f', '\x00', '-', '\x00',
  841. 'S', 'u', 'n', '\x00', 'M', 'o', 'n', '\x00',
  842. 'T', 'u', 'e', '\x00', 'W', 'e', 'd', '\x00',
  843. 'T', 'h', 'u', '\x00', 'F', 'r', 'i', '\x00',
  844. 'S', 'a', 't', '\x00', 'S', 'u', 'n', 'd',
  845. 'a', 'y', '\x00', 'M', 'o', 'n', 'd', 'a',
  846. 'y', '\x00', 'T', 'u', 'e', 's', 'd', 'a',
  847. 'y', '\x00', 'W', 'e', 'd', 'n', 'e', 's',
  848. 'd', 'a', 'y', '\x00', 'T', 'h', 'u', 'r',
  849. 's', 'd', 'a', 'y', '\x00', 'F', 'r', 'i',
  850. 'd', 'a', 'y', '\x00', 'S', 'a', 't', 'u',
  851. 'r', 'd', 'a', 'y', '\x00', 'J', 'a', 'n',
  852. '\x00', 'F', 'e', 'b', '\x00', 'M', 'a', 'r',
  853. '\x00', 'A', 'p', 'r', '\x00', 'M', 'a', 'y',
  854. '\x00', 'J', 'u', 'n', '\x00', 'J', 'u', 'l',
  855. '\x00', 'A', 'u', 'g', '\x00', 'S', 'e', 'p',
  856. '\x00', 'O', 'c', 't', '\x00', 'N', 'o', 'v',
  857. '\x00', 'D', 'e', 'c', '\x00', 'J', 'a', 'n',
  858. 'u', 'a', 'r', 'y', '\x00', 'F', 'e', 'b',
  859. 'r', 'u', 'a', 'r', 'y', '\x00', 'M', 'a',
  860. 'r', 'c', 'h', '\x00', 'A', 'p', 'r', 'i',
  861. 'l', '\x00', 'M', 'a', 'y', '\x00', 'J', 'u',
  862. 'n', 'e', '\x00', 'J', 'u', 'l', 'y', '\x00',
  863. 'A', 'u', 'g', 'u', 's', 't', '\x00', 'S',
  864. 'e', 'p', 't', 'e', 'm', 'b', 'e', 'r',
  865. '\x00', 'O', 'c', 't', 'o', 'b', 'e', 'r',
  866. '\x00', 'N', 'o', 'v', 'e', 'm', 'b', 'e',
  867. 'r', '\x00', 'D', 'e', 'c', 'e', 'm', 'b',
  868. 'e', 'r', '\x00', 'A', 'M', '\x00', 'P', 'M',
  869. '\x00', '%', 'a', ' ', '%', 'b', ' ', '%',
  870. 'e', ' ', '%', 'H', ':', '%', 'M', ':',
  871. '%', 'S', ' ', '%', 'Y', '\x00', '%', 'm',
  872. '/', '%', 'd', '/', '%', 'y', '\x00', '%',
  873. 'H', ':', '%', 'M', ':', '%', 'S', '\x00',
  874. '%', 'I', ':', '%', 'M', ':', '%', 'S',
  875. ' ', '%', 'p', '\x00', '^', '[', 'y', 'Y',
  876. ']', '\x00', '^', '[', 'n', 'N', ']', '\x00',
  877. };
  878. libc_hidden_proto(nl_langinfo)
  879. char *nl_langinfo(nl_item item)
  880. {
  881. unsigned int c;
  882. unsigned int i;
  883. if ((c = _NL_ITEM_CATEGORY(item)) < C_LC_ALL) {
  884. if ((i = cat_start[c] + _NL_ITEM_INDEX(item)) < cat_start[c+1]) {
  885. /* return (char *) C_locale_data + item_offset[i] + (i & 64); */
  886. return (char *) C_locale_data + nl_data[C_LC_ALL+1+i] + 2*(i & 64);
  887. }
  888. }
  889. return (char *) cat_start; /* Conveniently, this is the empty string. */
  890. }
  891. libc_hidden_def(nl_langinfo)
  892. #else /* __LOCALE_C_ONLY */
  893. #if defined(__UCLIBC_HAS_XLOCALE__) && !defined(__UCLIBC_DO_XLOCALE)
  894. libc_hidden_proto(nl_langinfo)
  895. libc_hidden_proto(nl_langinfo_l)
  896. char *nl_langinfo(nl_item item)
  897. {
  898. return nl_langinfo_l(item, __UCLIBC_CURLOCALE);
  899. }
  900. libc_hidden_def(nl_langinfo)
  901. #else /* defined(__UCLIBC_HAS_XLOCALE__) && !defined(__UCLIBC_DO_XLOCALE) */
  902. libc_hidden_proto(__XL_NPP(nl_langinfo))
  903. static const char empty[] = "";
  904. char *__XL_NPP(nl_langinfo)(nl_item item __LOCALE_PARAM )
  905. {
  906. unsigned int c = _NL_ITEM_CATEGORY(item);
  907. unsigned int i = _NL_ITEM_INDEX(item);
  908. if ((c < LC_ALL) && (i < __LOCALE_PTR->category_item_count[c])) {
  909. return ((char **)(((char *) __LOCALE_PTR)
  910. + __LOCALE_PTR->category_offsets[c]))[i];
  911. }
  912. return (char *) empty;
  913. }
  914. libc_hidden_def(__XL_NPP(nl_langinfo))
  915. #endif /* defined(__UCLIBC_HAS_XLOCALE__) && !defined(__UCLIBC_DO_XLOCALE) */
  916. #endif /* __LOCALE_C_ONLY */
  917. #endif
  918. /**********************************************************************/
  919. #ifdef L_newlocale
  920. libc_hidden_proto(newlocale)
  921. #ifdef __UCLIBC_MJN3_ONLY__
  922. #warning TODO: Move posix and utf8 strings.
  923. #endif
  924. static const char posix[] = "POSIX";
  925. static const char utf8[] = "UTF-8";
  926. static int find_locale(int category_mask, const char *p,
  927. unsigned char *new_locale)
  928. {
  929. int i;
  930. const unsigned char *s;
  931. uint16_t n;
  932. unsigned char lang_cult, codeset;
  933. #if defined(__LOCALE_DATA_AT_MODIFIERS_LENGTH) && 1
  934. /* Support standard locale handling for @-modifiers. */
  935. #ifdef __UCLIBC_MJN3_ONLY__
  936. #warning REMINDER: Fix buf size in find_locale.
  937. #endif
  938. char buf[18]; /* TODO: 7+{max codeset name length} */
  939. const char *q;
  940. if ((q = strchr(p,'@')) != NULL) {
  941. if ((((size_t)((q-p)-5)) > (sizeof(buf) - 5)) || (p[2] != '_')) {
  942. return 0;
  943. }
  944. /* locale name at least 5 chars long and 3rd char is '_' */
  945. s = LOCALE_AT_MODIFIERS;
  946. do {
  947. if (!strcmp(s+2, q+1)) {
  948. break;
  949. }
  950. s += 2 + *s; /* TODO - fix this throughout */
  951. } while (*s);
  952. if (!*s) {
  953. return 0;
  954. }
  955. assert(q - p < sizeof(buf));
  956. memcpy(buf, p, q-p);
  957. buf[q-p] = 0;
  958. buf[2] = s[1];
  959. p = buf;
  960. }
  961. #endif
  962. lang_cult = codeset = 0; /* Assume C and default codeset. */
  963. if (((*p == 'C') && !p[1]) || !strcmp(p, posix)) {
  964. goto FIND_LOCALE;
  965. }
  966. if ((strlen(p) > 5) && (p[5] == '.')) { /* Codeset in locale name? */
  967. /* TODO: maybe CODESET_LIST + *s ??? */
  968. /* 7bit is 1, UTF-8 is 2, 8-bit is >= 3 */
  969. codeset = 2;
  970. if (strcmp(utf8,p+6) != 0) {/* TODO - fix! */
  971. s = CODESET_LIST;
  972. do {
  973. ++codeset; /* Increment codeset first. */
  974. if (!strcmp(CODESET_LIST+*s, p+6)) {
  975. goto FIND_LANG_CULT;
  976. }
  977. } while (*++s);
  978. return 0; /* No matching codeset! */
  979. }
  980. }
  981. FIND_LANG_CULT: /* Find language_culture number. */
  982. s = LOCALE_NAMES;
  983. do { /* TODO -- do a binary search? */
  984. /* TODO -- fix gen_mmap!*/
  985. ++lang_cult; /* Increment first since C/POSIX is 0. */
  986. if (!strncmp(s,p,5)) { /* Found a matching locale name; */
  987. goto FIND_LOCALE;
  988. }
  989. s += 5;
  990. } while (lang_cult < __LOCALE_DATA_NUM_LOCALE_NAMES);
  991. return 0; /* No matching language_culture! */
  992. FIND_LOCALE: /* Find locale row matching name and codeset */
  993. s = LOCALES;
  994. n = 0;
  995. do { /* TODO -- do a binary search? */
  996. if ((lang_cult == *s) && ((codeset == s[1]) || (codeset == s[2]))) {
  997. i = 1;
  998. s = new_locale + 1;
  999. do {
  1000. if (category_mask & i) {
  1001. /* Encode current locale row number. */
  1002. ((unsigned char *) s)[0] = (n >> 7) | 0x80;
  1003. ((unsigned char *) s)[1] = (n & 0x7f) | 0x80;
  1004. }
  1005. s += 2;
  1006. i += i;
  1007. } while (i < (1 << LC_ALL));
  1008. return i; /* Return non-zero */
  1009. }
  1010. s += __LOCALE_DATA_WIDTH_LOCALES;
  1011. ++n;
  1012. } while (n <= __LOCALE_DATA_NUM_LOCALES); /* We started at 1!!! */
  1013. return 0; /* Unsupported locale. */
  1014. }
  1015. static unsigned char *composite_locale(int category_mask, const char *locale,
  1016. unsigned char *new_locale)
  1017. {
  1018. char buf[MAX_LOCALE_STR];
  1019. char *t;
  1020. char *e;
  1021. int c;
  1022. int component_mask;
  1023. if (!strchr(locale,'=')) {
  1024. if (!find_locale(category_mask, locale, new_locale)) {
  1025. return NULL;
  1026. }
  1027. return new_locale;
  1028. }
  1029. if (strlen(locale) >= sizeof(buf)) {
  1030. return NULL;
  1031. }
  1032. stpcpy(buf, locale);
  1033. component_mask = 0;
  1034. t = strtok_r(buf, "=", &e); /* This can't fail because of strchr test above. */
  1035. do {
  1036. c = 0;
  1037. while (strcmp(CATEGORY_NAMES + (int) CATEGORY_NAMES[c], t)) {
  1038. if (++c == LC_ALL) { /* Unknown category name! */
  1039. return NULL;
  1040. }
  1041. }
  1042. t = strtok_r(NULL, ";", &e);
  1043. c = (1 << c);
  1044. if (component_mask & c) { /* Multiple components for one category. */
  1045. return NULL;
  1046. }
  1047. component_mask |= c;
  1048. if ((category_mask & c) && (!t || !find_locale(c, t, new_locale))) {
  1049. return NULL;
  1050. }
  1051. } while ((t = strtok_r(NULL, "=", &e)) != NULL);
  1052. if (category_mask & ~component_mask) { /* Category component(s) missing. */
  1053. return NULL;
  1054. }
  1055. return new_locale;
  1056. }
  1057. __locale_t newlocale(int category_mask, const char *locale, __locale_t base)
  1058. {
  1059. const unsigned char *p;
  1060. int i, j, k;
  1061. unsigned char new_selector[LOCALE_SELECTOR_SIZE];
  1062. if (category_mask == (1 << LC_ALL)) {
  1063. category_mask = LC_ALL_MASK;
  1064. }
  1065. if (!locale || (((unsigned int)(category_mask)) > LC_ALL_MASK)) {
  1066. INVALID:
  1067. __set_errno(EINVAL);
  1068. return NULL; /* No locale or illegal/unsupported category. */
  1069. }
  1070. #ifdef __UCLIBC_MJN3_ONLY__
  1071. #warning TODO: Rename cur_locale to locale_selector.
  1072. #endif
  1073. strcpy((char *) new_selector,
  1074. (base ? (char *) base->cur_locale : C_LOCALE_SELECTOR));
  1075. if (!*locale) { /* locale == "", so check environment. */
  1076. #ifndef __UCLIBC_HAS_THREADS__
  1077. static /* If no threads, then envstr can be static. */
  1078. #endif /* __UCLIBC_HAS_THREADS__ */
  1079. const char *envstr[4] = { "LC_ALL", NULL, "LANG", posix };
  1080. i = 1;
  1081. k = 0;
  1082. do {
  1083. if (category_mask & i) {
  1084. /* Note: SUSv3 doesn't define a fallback mechanism here.
  1085. * So, if LC_ALL is invalid, we do _not_ continue trying
  1086. * the other environment vars. */
  1087. envstr[1] = CATEGORY_NAMES + CATEGORY_NAMES[k];
  1088. j = 0;
  1089. do {
  1090. p = envstr[j];
  1091. } while ((++j < 4) && (!(p = getenv(p)) || !*p));
  1092. /* The user set something... is it valid? */
  1093. /* Note: Since we don't support user-supplied locales and
  1094. * alternate paths, we don't need to worry about special
  1095. * handling for suid/sgid apps. */
  1096. if (!find_locale(i, p, new_selector)) {
  1097. goto INVALID;
  1098. }
  1099. }
  1100. i += i;
  1101. } while (++k < LC_ALL);
  1102. } else if (!composite_locale(category_mask, locale, new_selector)) {
  1103. goto INVALID;
  1104. }
  1105. #ifdef __UCLIBC_MJN3_ONLY__
  1106. #warning TODO: Do a compatible codeset check!
  1107. #endif
  1108. /* If we get here, the new selector corresponds to a valid locale. */
  1109. #ifdef __UCLIBC_MJN3_ONLY__
  1110. #warning CONSIDER: Probably want a _locale_new func to allow for caching of locales.
  1111. #endif
  1112. #if 0
  1113. if (base) {
  1114. _locale_set_l(new_selector, base);
  1115. } else {
  1116. base = _locale_new(new_selector);
  1117. }
  1118. #else
  1119. if (!base) {
  1120. if ((base = malloc(sizeof(__uclibc_locale_t))) == NULL) {
  1121. return base;
  1122. }
  1123. _locale_init_l(base);
  1124. }
  1125. _locale_set_l(new_selector, base);
  1126. #endif
  1127. return base;
  1128. }
  1129. libc_hidden_def(newlocale)
  1130. #endif
  1131. /**********************************************************************/
  1132. #ifdef L_duplocale
  1133. libc_hidden_proto(duplocale)
  1134. #ifdef __UCLIBC_MJN3_ONLY__
  1135. #warning REMINDER: When we allocate ctype tables, remember to dup them.
  1136. #endif
  1137. __locale_t duplocale(__locale_t dataset)
  1138. {
  1139. __locale_t r;
  1140. uint16_t * i2w;
  1141. size_t n;
  1142. assert(dataset != LC_GLOBAL_LOCALE);
  1143. if ((r = malloc(sizeof(__uclibc_locale_t))) != NULL) {
  1144. n = 2*dataset->collate.max_col_index+2;
  1145. if ((i2w = calloc(n, sizeof(uint16_t)))
  1146. != NULL
  1147. ) {
  1148. memcpy(r, dataset, sizeof(__uclibc_locale_t));
  1149. r->collate.index2weight = i2w;
  1150. memcpy(i2w, dataset->collate.index2weight, n * sizeof(uint16_t));
  1151. } else {
  1152. free(r);
  1153. r = NULL;
  1154. }
  1155. }
  1156. return r;
  1157. }
  1158. libc_hidden_def(duplocale)
  1159. #endif
  1160. /**********************************************************************/
  1161. #ifdef L_freelocale
  1162. #ifdef __UCLIBC_MJN3_ONLY__
  1163. #warning REMINDER: When we allocate ctype tables, remember to free them.
  1164. #endif
  1165. void freelocale(__locale_t dataset)
  1166. {
  1167. assert(dataset != __global_locale);
  1168. assert(dataset != LC_GLOBAL_LOCALE);
  1169. free(dataset->collate.index2weight); /* Free collation data. */
  1170. free(dataset); /* Free locale */
  1171. }
  1172. #endif
  1173. /**********************************************************************/
  1174. #ifdef L_uselocale
  1175. libc_hidden_proto(uselocale)
  1176. __locale_t uselocale(__locale_t dataset)
  1177. {
  1178. __locale_t old;
  1179. if (!dataset) {
  1180. old = __UCLIBC_CURLOCALE;
  1181. } else {
  1182. if (dataset == LC_GLOBAL_LOCALE) {
  1183. dataset = __global_locale;
  1184. }
  1185. #ifdef __UCLIBC_HAS_THREADS__
  1186. old = __curlocale_set(dataset);
  1187. #else
  1188. old = __curlocale_var;
  1189. __curlocale_var = dataset;
  1190. #endif
  1191. }
  1192. if (old == __global_locale) {
  1193. return LC_GLOBAL_LOCALE;
  1194. }
  1195. return old;
  1196. }
  1197. libc_hidden_def(uselocale)
  1198. #endif
  1199. /**********************************************************************/
  1200. #ifdef L___curlocale
  1201. #ifdef __UCLIBC_HAS_THREADS__
  1202. __locale_t weak_const_function __curlocale(void)
  1203. {
  1204. return __curlocale_var; /* This is overriden by the thread version. */
  1205. }
  1206. __locale_t weak_function __curlocale_set(__locale_t newloc)
  1207. {
  1208. __locale_t oldloc = __curlocale_var;
  1209. assert(newloc != LC_GLOBAL_LOCALE);
  1210. __curlocale_var = newloc;
  1211. return oldloc;
  1212. }
  1213. #endif
  1214. #endif
  1215. /**********************************************************************/
  1216. #ifdef L___locale_mbrtowc_l
  1217. /* NOTE: This returns an int... not size_t. Also, it is not a general
  1218. * routine. It is actually a very stripped-down version of mbrtowc
  1219. * that takes a __locale_t arg. This is used by strcoll and strxfrm.
  1220. * It is also used above to generate wchar_t versions of the decimal point
  1221. * and thousands seperator. */
  1222. #ifndef __CTYPE_HAS_UTF_8_LOCALES
  1223. #warning __CTYPE_HAS_UTF_8_LOCALES not set!
  1224. #endif
  1225. #ifndef __CTYPE_HAS_8_BIT_LOCALES
  1226. #warning __CTYPE_HAS_8_BIT_LOCALES not set!
  1227. #endif
  1228. #define Cc2wc_IDX_SHIFT __LOCALE_DATA_Cc2wc_IDX_SHIFT
  1229. #define Cc2wc_ROW_LEN __LOCALE_DATA_Cc2wc_ROW_LEN
  1230. extern size_t _wchar_utf8sntowcs(wchar_t *__restrict pwc, size_t wn,
  1231. const char **__restrict src, size_t n,
  1232. mbstate_t *ps, int allow_continuation) attribute_hidden;
  1233. int attribute_hidden __locale_mbrtowc_l(wchar_t *__restrict dst,
  1234. const char *__restrict src,
  1235. __locale_t loc )
  1236. {
  1237. #ifdef __CTYPE_HAS_UTF_8_LOCALES
  1238. if (loc->encoding == __ctype_encoding_utf8) {
  1239. mbstate_t ps;
  1240. const char *p = src;
  1241. size_t r;
  1242. ps.__mask = 0;
  1243. r = _wchar_utf8sntowcs(dst, 1, &p, SIZE_MAX, &ps, 1);
  1244. return (r == 1) ? (p-src) : r; /* Need to return 0 if nul char. */
  1245. }
  1246. #endif
  1247. #ifdef __CTYPE_HAS_8_BIT_LOCALES
  1248. assert((loc->encoding == __ctype_encoding_7_bit) || (loc->encoding == __ctype_encoding_8_bit));
  1249. #else
  1250. assert(loc->encoding == __ctype_encoding_7_bit);
  1251. #endif
  1252. if ((*dst = ((unsigned char)(*src))) < 0x80) { /* ASCII... */
  1253. return (*src != 0);
  1254. }
  1255. #ifdef __CTYPE_HAS_8_BIT_LOCALES
  1256. if (loc->encoding == __ctype_encoding_8_bit) {
  1257. wchar_t wc = *dst - 0x80;
  1258. *dst = loc->tbl8c2wc[
  1259. (loc->idx8c2wc[wc >> Cc2wc_IDX_SHIFT]
  1260. << Cc2wc_IDX_SHIFT) + (wc & (Cc2wc_ROW_LEN - 1))];
  1261. if (*dst) {
  1262. return 1;
  1263. }
  1264. }
  1265. #endif
  1266. return -1;
  1267. }
  1268. #endif
  1269. /**********************************************************************/