gen_wc8bit.c 18 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717
  1. /*
  2. * Copyright (C) 2000-2006 Erik Andersen <andersen@uclibc.org>
  3. *
  4. * Licensed under the LGPL v2.1, see the file COPYING.LIB in this tarball.
  5. */
  6. #define _GNU_SOURCE
  7. #include <stdio.h>
  8. #include <stdlib.h>
  9. #include <string.h>
  10. #include <locale.h>
  11. #include <stddef.h>
  12. #include <wctype.h>
  13. #include <limits.h>
  14. #ifndef _CTYPE_H
  15. #define _CTYPE_H
  16. #endif
  17. #ifndef _WCTYPE_H
  18. #define _WCTYPE_H
  19. #endif
  20. #include "include/bits/uClibc_ctype.h"
  21. /* TODO: maybe support -v like gen_wctype.c */
  22. #define verbose_msg(msg...) if (verbose) fprintf(stderr, msg)
  23. /* #define CTYPE_PACKED */
  24. #define UPLOW_IDX_SHIFT 3
  25. /* best if 2 unpacked or 3 packed */
  26. #define CTYPE_IDX_SHIFT 3
  27. /* 3 or 4 are very similar */
  28. #define C2WC_IDX_SHIFT 3
  29. #define CTYPE_IDX_LEN (128 >> (CTYPE_IDX_SHIFT))
  30. #define UPLOW_IDX_LEN (128 >> (UPLOW_IDX_SHIFT))
  31. #define C2WC_IDX_LEN (128 >> (C2WC_IDX_SHIFT))
  32. /* #ifdef CTYPE_PACKED */
  33. /* #define CTYPE_ROW_LEN (1 << ((CTYPE_IDX_SHIFT)-1)) */
  34. /* #else */
  35. #define CTYPE_ROW_LEN (1 << (CTYPE_IDX_SHIFT))
  36. /* #endif */
  37. #define UPLOW_ROW_LEN (1 << (UPLOW_IDX_SHIFT))
  38. #define C2WC_ROW_LEN (1 << (C2WC_IDX_SHIFT))
  39. #define MAX_WCHAR (0x2600-1)
  40. static unsigned char ctype_tbl[256 * CTYPE_ROW_LEN];
  41. static unsigned char uplow_tbl[256 * UPLOW_ROW_LEN];
  42. #ifdef DO_WIDE_CHAR
  43. static unsigned short c2wc_tbl[256 * C2WC_ROW_LEN];
  44. #endif
  45. static unsigned char tt[MAX_WCHAR+1];
  46. static unsigned char ti[MAX_WCHAR+1];
  47. static unsigned char xi[MAX_WCHAR+1];
  48. static int n_ctype_rows;
  49. static int n_uplow_rows;
  50. #ifdef DO_WIDE_CHAR
  51. static int n_c2wc_rows;
  52. #endif
  53. static int tt_num;
  54. static int ti_num;
  55. #define RANGE MAX_WCHAR
  56. #define TT_SHIFT 4
  57. #define TI_SHIFT 4
  58. #define II_LEN ((MAX_WCHAR+1) >> (TT_SHIFT+TI_SHIFT))
  59. typedef struct {
  60. unsigned long c2w[256];
  61. unsigned char w2c[MAX_WCHAR];
  62. unsigned char ii[II_LEN];
  63. unsigned char ctype_idx[CTYPE_IDX_LEN];
  64. unsigned char uplow_idx[UPLOW_IDX_LEN];
  65. unsigned char c2wc_idx[C2WC_IDX_LEN];
  66. } charset_data;
  67. int main(int argc, char **argv)
  68. {
  69. FILE *fp;
  70. charset_data csd[30];
  71. unsigned long max_wchar;
  72. unsigned char *p;
  73. int numsets;
  74. int i;
  75. int j;
  76. char buf[80];
  77. unsigned char row[256];
  78. #ifdef DO_WIDE_CHAR
  79. unsigned short wrow[256];
  80. #endif
  81. char codeset_list[500];
  82. char codeset_index[30];
  83. int codeset_list_end = 0;
  84. int total_size = 0;
  85. if (!setlocale(LC_CTYPE, "en_US.UTF-8")) {
  86. /* Silly foreigners disabling en_US locales */
  87. FILE *fp = popen("locale -a", "r");
  88. if (!fp)
  89. goto locale_failure;
  90. while (!feof(fp)) {
  91. char buf[256];
  92. size_t len;
  93. if (fgets(buf, sizeof(buf) - 10, fp) == NULL)
  94. goto locale_failure;
  95. len = strlen(buf);
  96. if (len > 0 && buf[len - 1] == '\n')
  97. buf[--len] = '\0';
  98. if (len < 5 || strcasecmp(&buf[len-5], ".UTF8") != 0)
  99. strcat(buf, ".UTF8");
  100. if (setlocale(LC_CTYPE, buf))
  101. goto locale_success;
  102. }
  103. locale_failure:
  104. printf("could not find a UTF8 locale ... please enable en_US.UTF-8\n");
  105. return EXIT_FAILURE;
  106. locale_success:
  107. pclose(fp);
  108. }
  109. if (argc == 1) {
  110. printf("#undef __CTYPE_HAS_8_BIT_LOCALES\n\n");
  111. printf("#define __LOCALE_DATA_NUM_CODESETS\t\t0\n");
  112. printf("#define __LOCALE_DATA_CODESET_LIST\t\t\"\"\n");
  113. } else {
  114. printf("#define __CTYPE_HAS_8_BIT_LOCALES\t\t1\n\n");
  115. }
  116. printf("#define __LOCALE_DATA_Cctype_IDX_SHIFT\t%d\n", CTYPE_IDX_SHIFT);
  117. printf("#define __LOCALE_DATA_Cctype_IDX_LEN\t\t%d\n", CTYPE_IDX_LEN);
  118. #ifdef CTYPE_PACKED
  119. printf("#define __LOCALE_DATA_Cctype_ROW_LEN\t\t%d\n", CTYPE_ROW_LEN >> 1);
  120. printf("#define __LOCALE_DATA_Cctype_PACKED\t\t1\n");
  121. #else
  122. printf("#define __LOCALE_DATA_Cctype_ROW_LEN\t\t%d\n", CTYPE_ROW_LEN);
  123. printf("#undef __LOCALE_DATA_Cctype_PACKED\n");
  124. #endif
  125. printf("\n#define __LOCALE_DATA_Cuplow_IDX_SHIFT\t%d\n", UPLOW_IDX_SHIFT);
  126. printf("#define __LOCALE_DATA_Cuplow_IDX_LEN\t\t%d\n", UPLOW_IDX_LEN);
  127. printf("#define __LOCALE_DATA_Cuplow_ROW_LEN\t\t%d\n", UPLOW_ROW_LEN);
  128. #ifdef DO_WIDE_CHAR
  129. printf("\n#define __LOCALE_DATA_Cc2wc_IDX_LEN\t\t%d\n", C2WC_IDX_LEN);
  130. printf("#define __LOCALE_DATA_Cc2wc_IDX_SHIFT\t\t%d\n", C2WC_IDX_SHIFT);
  131. printf("#define __LOCALE_DATA_Cc2wc_ROW_LEN\t\t%d\n", C2WC_ROW_LEN);
  132. #endif
  133. printf("\ntypedef struct {\n");
  134. printf("\tunsigned char idx8ctype[%d];\n", CTYPE_IDX_LEN);
  135. printf("\tunsigned char idx8uplow[%d];\n", UPLOW_IDX_LEN);
  136. #ifdef DO_WIDE_CHAR
  137. printf("\tunsigned char idx8c2wc[%d];\n", C2WC_IDX_LEN);
  138. printf("\tunsigned char idx8wc2c[%d];\n", II_LEN);
  139. #endif
  140. #ifndef __metag__
  141. printf("} __codeset_8_bit_t;\n\n");
  142. #else
  143. printf("} __attribute__((__packed__)) __codeset_8_bit_t;\n\n");
  144. #endif /* __metag__ */
  145. printf("#ifdef WANT_DATA\n\n");
  146. printf("static const __codeset_8_bit_t codeset_8_bit[%d] = {\n", argc-1);
  147. max_wchar = 0x7f;
  148. numsets = 0;
  149. codeset_index[0] = 0;
  150. while (--argc) {
  151. if (!(fp = fopen(*++argv,"r"))) {
  152. fprintf(stderr, "cannot open file \"%s\"\n", *argv);
  153. return EXIT_FAILURE;
  154. }
  155. fprintf(stderr, "processing %s... ", *argv);
  156. {
  157. char *s0;
  158. char *s1;
  159. int n;
  160. s0 = strrchr(*argv, '/');
  161. if (!s0) {
  162. s0 = *argv;
  163. } else {
  164. ++s0;
  165. }
  166. s1 = strrchr(s0, '.');
  167. if (!s1) {
  168. n = strlen(s0);
  169. } else {
  170. n = s1 - s0;
  171. }
  172. /* if ((numsets == 0) && strncmp("ASCII", s0, n)) { */
  173. /* printf("error - first codeset isn't ASCII!\n"); */
  174. /* return EXIT_FAILURE; */
  175. /* } */
  176. if (numsets >= sizeof(codeset_index)) {
  177. fprintf(stderr, "error - too many codesets!\n");
  178. return EXIT_FAILURE;
  179. }
  180. if (codeset_list_end + n + 1 + numsets + 1 + 1 >= 256) {
  181. fprintf(stderr, "error - codeset list to big!\n");
  182. return EXIT_FAILURE;
  183. }
  184. codeset_index[numsets+1] = codeset_index[numsets] + n+1;
  185. strncpy(codeset_list + codeset_list_end, s0, n);
  186. codeset_list_end += (n+1);
  187. codeset_list[codeset_list_end - 1] = 0;
  188. printf("\t{ /* %.*s */", n, s0);
  189. }
  190. memset(&csd[numsets], 0, sizeof(charset_data));
  191. memset(xi, 0, sizeof(xi));
  192. {
  193. unsigned long c, wc;
  194. int lines;
  195. lines = 0;
  196. while (fgets(buf,sizeof(buf),fp)) {
  197. if ((2 != sscanf(buf, "{ %lx , %lx", &c, &wc))
  198. || (c >= 256) || (wc > MAX_WCHAR)) {
  199. fprintf(stderr, "error: scanf failure! \"%s\"\n", buf);
  200. return EXIT_FAILURE;
  201. }
  202. /* don't put in w2c... dynamicly build tt instead. */
  203. if (c <= 0x7f) { /* check the 7bit entries but don't store */
  204. if (c != wc) {
  205. fprintf(stderr, "error: c != wc in %s\n", buf);
  206. return EXIT_FAILURE;
  207. }
  208. csd[numsets].c2w[c] = wc;
  209. csd[numsets].w2c[wc] = 0; /* ignore */
  210. if (wc > max_wchar) {
  211. max_wchar = wc;
  212. }
  213. } else {
  214. csd[numsets].c2w[c] = wc;
  215. csd[numsets].w2c[wc] = c;
  216. if (wc > max_wchar) {
  217. max_wchar = wc;
  218. }
  219. }
  220. ++lines;
  221. }
  222. fprintf(stderr, "%d lines ", lines);
  223. for (i = 0 ; i <= MAX_WCHAR ; i += (1 << TT_SHIFT)) {
  224. p = &csd[numsets].w2c[i];
  225. for (j = 0 ; j < tt_num ; j++) {
  226. if (!memcmp(p, &tt[j << TT_SHIFT], (1 << TT_SHIFT))) {
  227. break;
  228. }
  229. }
  230. if (j == tt_num) { /* new entry */
  231. memcpy(&tt[j << TT_SHIFT], p, (1 << TT_SHIFT));
  232. ++tt_num;
  233. }
  234. xi[i >> TT_SHIFT] = j;
  235. }
  236. for (i = 0 ; i <= (MAX_WCHAR >> TT_SHIFT) ; i += (1 << TI_SHIFT)) {
  237. p = &xi[i];
  238. for (j = 0 ; j < ti_num ; j++) {
  239. if (!memcmp(p, &ti[j << TI_SHIFT], (1 << TI_SHIFT))) {
  240. break;
  241. }
  242. }
  243. if (j == ti_num) { /* new entry */
  244. memcpy(&ti[j << TI_SHIFT], p, (1 << TI_SHIFT));
  245. ++ti_num;
  246. }
  247. csd[numsets].ii[i >> TI_SHIFT] = j;
  248. /* fprintf(stderr, "%d ", i >> TI_SHIFT); */
  249. }
  250. #if 1
  251. printf("\n\t\t/* idx8ctype data */\n\t\t{");
  252. for (i = 128 ; i < 256 ; i++) {
  253. wchar_t c;
  254. unsigned int d;
  255. /* if (!(i & 0x7)) { */
  256. /* printf("\n"); */
  257. /* } */
  258. c = csd[numsets].c2w[i];
  259. if (c == 0) { /* non-existant char in codeset */
  260. d = __CTYPE_unclassified;
  261. } else if (iswdigit(c)) {
  262. d = __CTYPE_digit;
  263. } else if (iswalpha(c)) {
  264. d = __CTYPE_alpha_nonupper_nonlower;
  265. if (iswlower(c)) {
  266. d = __CTYPE_alpha_lower;
  267. if (iswupper(c)) {
  268. d = __CTYPE_alpha_upper_lower;
  269. }
  270. } else if (iswupper(c)) {
  271. d = __CTYPE_alpha_upper;
  272. }
  273. } else if (iswpunct(c)) {
  274. d = __CTYPE_punct;
  275. } else if (iswgraph(c)) {
  276. d = __CTYPE_graph;
  277. } else if (iswprint(c)) {
  278. d = __CTYPE_print_space_nonblank;
  279. if (iswblank(c)) {
  280. d = __CTYPE_print_space_blank;
  281. }
  282. } else if (iswspace(c) && !iswcntrl(c)) {
  283. d = __CTYPE_space_nonblank_noncntrl;
  284. if (iswblank(c)) {
  285. d = __CTYPE_space_blank_noncntrl;
  286. }
  287. } else if (iswcntrl(c)) {
  288. d = __CTYPE_cntrl_nonspace;
  289. if (iswspace(c)) {
  290. d = __CTYPE_cntrl_space_nonblank;
  291. if (iswblank(c)) {
  292. d = __CTYPE_cntrl_space_blank;
  293. }
  294. }
  295. } else {
  296. d = __CTYPE_unclassified;
  297. }
  298. #if 1
  299. row[i & (CTYPE_ROW_LEN-1)] = d;
  300. if ((i & (CTYPE_ROW_LEN-1)) == (CTYPE_ROW_LEN-1)) {
  301. p = ctype_tbl;
  302. for (j=0 ; j < n_ctype_rows ; j++) {
  303. if (!memcmp(p, row, CTYPE_ROW_LEN)) {
  304. break;
  305. }
  306. p += CTYPE_ROW_LEN;
  307. }
  308. if (j == n_ctype_rows) { /* new entry */
  309. if (++n_ctype_rows > 256) {
  310. fprintf(stderr, "error -- to many ctype rows!\n");
  311. return EXIT_FAILURE;
  312. }
  313. memcpy(p, row, CTYPE_ROW_LEN);
  314. }
  315. csd[numsets].ctype_idx[i >> CTYPE_IDX_SHIFT] = j;
  316. if (!((i >> CTYPE_IDX_SHIFT) & 0x7)
  317. && (i != (127 + CTYPE_ROW_LEN))
  318. ) {
  319. printf("\n\t\t ");
  320. }
  321. printf(" %#4x,", j);
  322. }
  323. #else
  324. printf(" %#4x,", d);
  325. #endif
  326. }
  327. #endif
  328. printf(" }");
  329. #if 1
  330. printf(",\n\t\t/* idx8uplow data */\n\t\t{");
  331. for (i = 128 ; i < 256 ; i++) {
  332. wchar_t c, u, l;
  333. /* if (!(i & 0x7)) { */
  334. /* printf("\n"); */
  335. /* } */
  336. c = csd[numsets].c2w[i];
  337. if ((c != 0) || 1) {
  338. u = towupper(c);
  339. l = towlower(c);
  340. if (u >= 0x80) u = csd[numsets].w2c[u];
  341. if (l >= 0x80) l = csd[numsets].w2c[l];
  342. if (u == 0) u = i; /* upper is missing, so ignore */
  343. if (l == 0) l = i; /* lower is missing, so ignore */
  344. #if 1
  345. /* store as unsigned char and let overflow handle it. */
  346. /* if ((((u-i) < CHAR_MIN) || ((u-i) > CHAR_MAX)) */
  347. /* || (((i-l) < CHAR_MIN) || ((i-l) > CHAR_MAX)) */
  348. /* ) { */
  349. /* fprintf(stderr, "error - uplow diff out of range! %d %ld %ld\n", */
  350. /* i, u, l); */
  351. /* return EXIT_FAILURE; */
  352. /* } */
  353. row[i & (UPLOW_ROW_LEN-1)] = ((l==i) ? (u-i) : (i-l));
  354. if ((i & (UPLOW_ROW_LEN-1)) == (UPLOW_ROW_LEN-1)) {
  355. p = uplow_tbl;
  356. for (j=0 ; j < n_uplow_rows ; j++) {
  357. if (!memcmp(p, row, UPLOW_ROW_LEN)) {
  358. break;
  359. }
  360. p += UPLOW_ROW_LEN;
  361. }
  362. if (j == n_uplow_rows) { /* new entry */
  363. if (++n_uplow_rows > 256) {
  364. fprintf(stderr, "error -- to many uplow rows!\n");
  365. return EXIT_FAILURE;
  366. }
  367. memcpy(p, row, UPLOW_ROW_LEN);
  368. }
  369. csd[numsets].uplow_idx[i >> UPLOW_IDX_SHIFT] = j;
  370. if (!((i >> UPLOW_IDX_SHIFT) & 0x7)
  371. && (i != (127 + UPLOW_ROW_LEN))
  372. ) {
  373. printf("\n\t\t ");
  374. }
  375. printf(" %#4x,", j);
  376. }
  377. #elif 0
  378. if (!(i & 0x7) && i) {
  379. printf("\n");
  380. }
  381. printf(" %4ld,", (l==i) ? (u-i) : (i-l));
  382. /* printf(" %4ld,", (l==i) ? u : l); */
  383. #else
  384. if ((u != i) || (l != i)) {
  385. #if 0
  386. printf(" %#08lx, %#08lx, %#08lx, %#08lx, %#08lx, %#08lx, \n",
  387. (unsigned long) i,
  388. (unsigned long) c,
  389. (unsigned long) l,
  390. (unsigned long) towlower(c),
  391. (unsigned long) u,
  392. (unsigned long) towupper(c));
  393. #else
  394. printf(" %#08lx, %8ld, %d, %8ld, %d, %#08lx\n",
  395. (unsigned long) i,
  396. (long) (l - i),
  397. iswupper(c),
  398. (long) (i - u),
  399. iswlower(c),
  400. (unsigned long) c);
  401. #endif
  402. }
  403. #endif
  404. }
  405. }
  406. printf(" }");
  407. #endif
  408. #ifndef DO_WIDE_CHAR
  409. printf("\n");
  410. #else /* DO_WIDE_CHAR */
  411. #if 1
  412. printf(",\n\t\t/* idx8c2wc data */\n\t\t{");
  413. for (i = 128 ; i < 256 ; i++) {
  414. #if 1
  415. wrow[i & (C2WC_ROW_LEN-1)] = csd[numsets].c2w[i];
  416. if ((i & (C2WC_ROW_LEN-1)) == (C2WC_ROW_LEN-1)) {
  417. p = (unsigned char *) c2wc_tbl;
  418. for (j=0 ; j < n_c2wc_rows ; j++) {
  419. if (!memcmp(p, (char *) wrow, 2*C2WC_ROW_LEN)) {
  420. break;
  421. }
  422. p += 2*C2WC_ROW_LEN;
  423. }
  424. if (j == n_c2wc_rows) { /* new entry */
  425. if (++n_c2wc_rows > 256) {
  426. fprintf(stderr, "error -- to many c2wc rows!\n");
  427. return EXIT_FAILURE;
  428. }
  429. memcpy(p, (char *) wrow, 2*C2WC_ROW_LEN);
  430. }
  431. csd[numsets].c2wc_idx[i >> C2WC_IDX_SHIFT] = j;
  432. if (!((i >> C2WC_IDX_SHIFT) & 0x7)
  433. && (i != (127 + C2WC_ROW_LEN))
  434. ) {
  435. printf("\n\t\t ");
  436. }
  437. printf(" %#4x,", j);
  438. }
  439. #else
  440. if (!(i & 0x7) && i) {
  441. printf("\n");
  442. }
  443. printf(" %#6lx,", csd[numsets].c2w[i]);
  444. #endif
  445. }
  446. printf(" },\n");
  447. #endif
  448. #if 1
  449. /* fprintf(stderr, "\nII_LEN = %d\n", II_LEN); */
  450. printf("\t\t/* idx8wc2c data */\n\t\t{");
  451. for (i = 0 ; i < II_LEN ; i++) {
  452. if (!(i & 0x7) && i) {
  453. printf("\n\t\t ");
  454. }
  455. printf(" %#4x,", csd[numsets].ii[i]);
  456. }
  457. printf(" }\n");
  458. #endif
  459. #endif /* DO_WIDE_CHAR */
  460. printf("\t},\n");
  461. }
  462. ++numsets;
  463. fprintf(stderr, "done\n");
  464. }
  465. printf("};\n");
  466. printf("\n#endif /* WANT_DATA */\n");
  467. #ifdef DO_WIDE_CHAR
  468. printf("\n");
  469. printf("#define __LOCALE_DATA_Cwc2c_DOMAIN_MAX\t%#x\n", RANGE);
  470. printf("#define __LOCALE_DATA_Cwc2c_TI_SHIFT\t\t%d\n", TI_SHIFT);
  471. printf("#define __LOCALE_DATA_Cwc2c_TT_SHIFT\t\t%d\n", TT_SHIFT);
  472. printf("#define __LOCALE_DATA_Cwc2c_II_LEN\t\t%d\n", II_LEN);
  473. printf("#define __LOCALE_DATA_Cwc2c_TI_LEN\t\t%d\n", ti_num << TI_SHIFT);
  474. printf("#define __LOCALE_DATA_Cwc2c_TT_LEN\t\t%d\n", tt_num << TT_SHIFT);
  475. printf("\n");
  476. printf("\n#define __LOCALE_DATA_Cwc2c_TBL_LEN\t\t%d\n",
  477. (ti_num << TI_SHIFT) + (tt_num << TT_SHIFT));
  478. printf("#ifdef WANT_DATA\n\n");
  479. printf("static const unsigned char __LOCALE_DATA_Cwc2c_data[%d] = {\n",
  480. (ti_num << TI_SHIFT) + (tt_num << TT_SHIFT));
  481. printf("\t/* ti_table */\n\t");
  482. for (i=0 ; i < ti_num << TI_SHIFT ; i++) {
  483. if (!(i & 7) && i) {
  484. printf("\n\t");
  485. }
  486. printf(" %#4x,", ti[i]);
  487. }
  488. printf("\n");
  489. printf("\t/* tt_table */\n\t");
  490. for (i=0 ; i < tt_num << TT_SHIFT ; i++) {
  491. if (!(i & 7) && i) {
  492. printf("\n\t");
  493. }
  494. printf(" %#4x,", tt[i]);
  495. }
  496. printf("\n};\n");
  497. printf("\n#endif /* WANT_DATA */\n");
  498. #endif /* DO_WIDE_CHAR */
  499. printf("\n#define __LOCALE_DATA_Cuplow_TBL_LEN\t\t%d\n",
  500. n_uplow_rows * UPLOW_ROW_LEN);
  501. printf("\n#ifdef WANT_DATA\n\n");
  502. printf("\nstatic const unsigned char __LOCALE_DATA_Cuplow_data[%d] = {\n",
  503. n_uplow_rows * UPLOW_ROW_LEN);
  504. p = uplow_tbl;
  505. for (j=0 ; j < n_uplow_rows ; j++) {
  506. printf("\t");
  507. for (i=0 ; i < UPLOW_ROW_LEN ; i++) {
  508. printf(" %#4x,", (unsigned int)((unsigned char) p[i]));
  509. }
  510. printf("\n");
  511. p += UPLOW_ROW_LEN;
  512. }
  513. printf("};\n");
  514. printf("\n#endif /* WANT_DATA */\n");
  515. printf("\n#define __LOCALE_DATA_Cctype_TBL_LEN\t\t%d\n",
  516. #ifdef CTYPE_PACKED
  517. n_ctype_rows * CTYPE_ROW_LEN / 2
  518. #else
  519. n_ctype_rows * CTYPE_ROW_LEN
  520. #endif
  521. );
  522. printf("\n#ifdef WANT_DATA\n\n");
  523. printf("\nstatic const unsigned char __LOCALE_DATA_Cctype_data[%d] = {\n",
  524. #ifdef CTYPE_PACKED
  525. n_ctype_rows * CTYPE_ROW_LEN / 2
  526. #else
  527. n_ctype_rows * CTYPE_ROW_LEN
  528. #endif
  529. );
  530. p = ctype_tbl;
  531. for (j=0 ; j < n_ctype_rows ; j++) {
  532. printf("\t");
  533. for (i=0 ; i < CTYPE_ROW_LEN ; i++) {
  534. #ifdef CTYPE_PACKED
  535. printf(" %#4x,", (unsigned int)(p[i] + (p[i+1] << 4)));
  536. ++i;
  537. #else
  538. printf(" %#4x,", (unsigned int)p[i]);
  539. #endif
  540. }
  541. printf("\n");
  542. p += CTYPE_ROW_LEN;
  543. }
  544. printf("};\n");
  545. printf("\n#endif /* WANT_DATA */\n");
  546. #ifdef DO_WIDE_CHAR
  547. printf("\n#define __LOCALE_DATA_Cc2wc_TBL_LEN\t\t%d\n",
  548. n_c2wc_rows * C2WC_ROW_LEN);
  549. printf("\n#ifdef WANT_DATA\n\n");
  550. printf("\nstatic const unsigned short __LOCALE_DATA_Cc2wc_data[%d] = {\n",
  551. n_c2wc_rows * C2WC_ROW_LEN);
  552. p = (unsigned char *) c2wc_tbl;
  553. for (j=0 ; j < n_c2wc_rows ; j++) {
  554. printf("\t");
  555. for (i=0 ; i < C2WC_ROW_LEN ; i++) {
  556. printf(" %#6x,", (unsigned int)(((unsigned short *)p)[i]));
  557. }
  558. printf("\n");
  559. p += 2*C2WC_ROW_LEN;
  560. }
  561. printf("};\n");
  562. printf("\n#endif /* WANT_DATA */\n");
  563. #endif /* DO_WIDE_CHAR */
  564. printf("\n\n");
  565. printf("#define __LOCALE_DATA_NUM_CODESETS\t\t%d\n", numsets);
  566. printf("#define __LOCALE_DATA_CODESET_LIST \\\n\t\"");
  567. for (i=0 ; i < numsets ; i++) {
  568. printf("\\x%02x", numsets + 1 + (unsigned char) codeset_index[i]);
  569. if (((i & 7) == 7) && (i + 1 < numsets)) {
  570. printf("\" \\\n\t\"");
  571. }
  572. }
  573. printf("\" \\\n\t\"\\0\"");
  574. for (i=0 ; i < numsets ; i++) {
  575. printf(" \\\n\t\"%s\\0\"",
  576. codeset_list + ((unsigned char)codeset_index[i]));
  577. }
  578. printf("\n\n");
  579. for (i=0 ; i < numsets ; i++) {
  580. char buf[30];
  581. char *z;
  582. strcpy(buf, codeset_list + ((unsigned char)codeset_index[i]));
  583. for (z=buf ; *z ; z++) {
  584. if (*z == '-') {
  585. *z = '_';
  586. }
  587. }
  588. printf("#define __CTYPE_HAS_CODESET_%s\n", buf);
  589. }
  590. #ifdef DO_WIDE_CHAR
  591. printf("#define __CTYPE_HAS_CODESET_UTF_8\n");
  592. #endif /* DO_WIDE_CHAR */
  593. #if 0
  594. printf("\n#endif /* __CTYPE_HAS_8_BIT_LOCALES */\n\n");
  595. #endif
  596. total_size = 0;
  597. #ifdef DO_WIDE_CHAR
  598. fprintf(stderr, "tt_num = %d ti_num = %d\n", tt_num, ti_num);
  599. fprintf(stderr, "max_wchar = %#lx\n", max_wchar);
  600. fprintf(stderr, "size is %d * %d + %d * %d + %d * %d = %d\n",
  601. tt_num, 1 << TT_SHIFT, ti_num, 1 << TI_SHIFT,
  602. ((MAX_WCHAR >> (TT_SHIFT + TI_SHIFT)) + 1), numsets,
  603. j = tt_num * (1 << TT_SHIFT) + ti_num * (1 << TI_SHIFT)
  604. + ((MAX_WCHAR >> (TT_SHIFT + TI_SHIFT)) + 1) * numsets);
  605. total_size += j;
  606. #endif /* DO_WIDE_CHAR */
  607. #ifdef CTYPE_PACKED
  608. i = 2;
  609. #else
  610. i = 1;
  611. #endif
  612. fprintf(stderr, "ctype - CTYPE_IDX_SHIFT = %d -- %d * %d + %d * %d = %d\n",
  613. CTYPE_IDX_SHIFT, numsets, CTYPE_IDX_LEN, n_ctype_rows, CTYPE_ROW_LEN / i,
  614. j = numsets * CTYPE_IDX_LEN + n_ctype_rows * CTYPE_ROW_LEN / i);
  615. total_size += j;
  616. fprintf(stderr, "uplow - UPLOW_IDX_SHIFT = %d -- %d * %d + %d * %d = %d\n",
  617. UPLOW_IDX_SHIFT, numsets, UPLOW_IDX_LEN, n_uplow_rows, UPLOW_ROW_LEN,
  618. j = numsets * UPLOW_IDX_LEN + n_uplow_rows * UPLOW_ROW_LEN);
  619. total_size += j;
  620. #ifdef DO_WIDE_CHAR
  621. fprintf(stderr, "c2wc - C2WC_IDX_SHIFT = %d -- %d * %d + 2 * %d * %d = %d\n",
  622. C2WC_IDX_SHIFT, numsets, C2WC_IDX_LEN, n_c2wc_rows, C2WC_ROW_LEN,
  623. j = numsets * C2WC_IDX_LEN + 2 * n_c2wc_rows * C2WC_ROW_LEN);
  624. total_size += j;
  625. #endif /* DO_WIDE_CHAR */
  626. fprintf(stderr, "total size = %d\n", total_size);
  627. /* for (i=0 ; i < numsets ; i++) { */
  628. /* printf("codeset_index[i] = %d codeset_list[ci[i]] = \"%s\"\n", */
  629. /* (unsigned char) codeset_index[i], */
  630. /* codeset_list + ((unsigned char)codeset_index[i])); */
  631. /* } */
  632. return EXIT_SUCCESS;
  633. }