| 1 |
1325 |
phoenix |
#define _GNU_SOURCE
|
| 2 |
|
|
#include <stdio.h>
|
| 3 |
|
|
#include <stdlib.h>
|
| 4 |
|
|
#include <string.h>
|
| 5 |
|
|
#include <locale.h>
|
| 6 |
|
|
#include <stddef.h>
|
| 7 |
|
|
#include <wctype.h>
|
| 8 |
|
|
#include <limits.h>
|
| 9 |
|
|
|
| 10 |
|
|
#ifndef _CTYPE_H
|
| 11 |
|
|
#define _CTYPE_H
|
| 12 |
|
|
#endif
|
| 13 |
|
|
#ifndef _WCTYPE_H
|
| 14 |
|
|
#define _WCTYPE_H
|
| 15 |
|
|
#endif
|
| 16 |
|
|
#include "../../libc/sysdeps/linux/common/bits/uClibc_ctype.h"
|
| 17 |
|
|
|
| 18 |
|
|
|
| 19 |
|
|
/* #define CTYPE_PACKED */
|
| 20 |
|
|
#define UPLOW_IDX_SHIFT 3
|
| 21 |
|
|
/* best if 2 unpacked or 3 packed */
|
| 22 |
|
|
#define CTYPE_IDX_SHIFT 3
|
| 23 |
|
|
/* 3 or 4 are very similar */
|
| 24 |
|
|
#define C2WC_IDX_SHIFT 3
|
| 25 |
|
|
|
| 26 |
|
|
#define CTYPE_IDX_LEN (128 >> (CTYPE_IDX_SHIFT))
|
| 27 |
|
|
#define UPLOW_IDX_LEN (128 >> (UPLOW_IDX_SHIFT))
|
| 28 |
|
|
#define C2WC_IDX_LEN (128 >> (C2WC_IDX_SHIFT))
|
| 29 |
|
|
|
| 30 |
|
|
/* #ifdef CTYPE_PACKED */
|
| 31 |
|
|
/* #define CTYPE_ROW_LEN (1 << ((CTYPE_IDX_SHIFT)-1)) */
|
| 32 |
|
|
/* #else */
|
| 33 |
|
|
#define CTYPE_ROW_LEN (1 << (CTYPE_IDX_SHIFT))
|
| 34 |
|
|
/* #endif */
|
| 35 |
|
|
#define UPLOW_ROW_LEN (1 << (UPLOW_IDX_SHIFT))
|
| 36 |
|
|
#define C2WC_ROW_LEN (1 << (C2WC_IDX_SHIFT))
|
| 37 |
|
|
|
| 38 |
|
|
|
| 39 |
|
|
|
| 40 |
|
|
#define MAX_WCHAR (0x2600-1)
|
| 41 |
|
|
|
| 42 |
|
|
static unsigned char ctype_tbl[256 * CTYPE_ROW_LEN];
|
| 43 |
|
|
static unsigned char uplow_tbl[256 * UPLOW_ROW_LEN];
|
| 44 |
|
|
#ifdef DO_WIDE_CHAR
|
| 45 |
|
|
static unsigned short c2wc_tbl[256 * C2WC_ROW_LEN];
|
| 46 |
|
|
#endif
|
| 47 |
|
|
static unsigned char tt[MAX_WCHAR+1];
|
| 48 |
|
|
static unsigned char ti[MAX_WCHAR+1];
|
| 49 |
|
|
static unsigned char xi[MAX_WCHAR+1];
|
| 50 |
|
|
|
| 51 |
|
|
static int n_ctype_rows;
|
| 52 |
|
|
static int n_uplow_rows;
|
| 53 |
|
|
#ifdef DO_WIDE_CHAR
|
| 54 |
|
|
static int n_c2wc_rows;
|
| 55 |
|
|
#endif
|
| 56 |
|
|
static int tt_num;
|
| 57 |
|
|
static int ti_num;
|
| 58 |
|
|
|
| 59 |
|
|
#define RANGE MAX_WCHAR
|
| 60 |
|
|
|
| 61 |
|
|
#define TT_SHIFT 4
|
| 62 |
|
|
#define TI_SHIFT 4
|
| 63 |
|
|
|
| 64 |
|
|
#define II_LEN ((MAX_WCHAR+1) >> (TT_SHIFT+TI_SHIFT))
|
| 65 |
|
|
|
| 66 |
|
|
typedef struct {
|
| 67 |
|
|
unsigned long c2w[256];
|
| 68 |
|
|
unsigned char w2c[MAX_WCHAR];
|
| 69 |
|
|
unsigned char ii[II_LEN];
|
| 70 |
|
|
unsigned char ctype_idx[CTYPE_IDX_LEN];
|
| 71 |
|
|
unsigned char uplow_idx[UPLOW_IDX_LEN];
|
| 72 |
|
|
unsigned char c2wc_idx[C2WC_IDX_LEN];
|
| 73 |
|
|
} charset_data;
|
| 74 |
|
|
|
| 75 |
|
|
int main(int argc, char **argv)
|
| 76 |
|
|
{
|
| 77 |
|
|
FILE *fp;
|
| 78 |
|
|
FILE *out;
|
| 79 |
|
|
charset_data csd[30];
|
| 80 |
|
|
unsigned long max_wchar;
|
| 81 |
|
|
unsigned char *p;
|
| 82 |
|
|
int numsets;
|
| 83 |
|
|
int i;
|
| 84 |
|
|
int j;
|
| 85 |
|
|
char buf[80];
|
| 86 |
|
|
unsigned char row[256];
|
| 87 |
|
|
#ifdef DO_WIDE_CHAR
|
| 88 |
|
|
unsigned short wrow[256];
|
| 89 |
|
|
#endif
|
| 90 |
|
|
char codeset_list[500];
|
| 91 |
|
|
char codeset_index[30];
|
| 92 |
|
|
int codeset_list_end = 0;
|
| 93 |
|
|
int total_size = 0;
|
| 94 |
|
|
|
| 95 |
|
|
if (!setlocale(LC_CTYPE, "en_US.UTF-8")) {
|
| 96 |
|
|
printf("setlocale(LC_CTYPE,\"en_US.UTF-8\") failed!\n");
|
| 97 |
|
|
return EXIT_FAILURE;
|
| 98 |
|
|
}
|
| 99 |
|
|
|
| 100 |
|
|
if (!(out = fopen("c8tables.h","w"))) {
|
| 101 |
|
|
printf("error: couldn't open file \"c8tables.h\"\n");
|
| 102 |
|
|
return EXIT_FAILURE;
|
| 103 |
|
|
}
|
| 104 |
|
|
|
| 105 |
|
|
#if 0
|
| 106 |
|
|
if (argc == 1) {
|
| 107 |
|
|
/* User requested 8-bit codesets, but didn't list any... */
|
| 108 |
|
|
/* Allow to build, just so this feature can be left on in config. */
|
| 109 |
|
|
fprintf(out, "#ifdef __CTYPE_HAS_8_BIT_LOCALES\n");
|
| 110 |
|
|
fprintf(out, "#warning ignoring 8 bit codesets request"
|
| 111 |
|
|
" as no codesets specified.\n");
|
| 112 |
|
|
fprintf(out, "#endif\n");
|
| 113 |
|
|
fprintf(out, "#undef __CTYPE_HAS_8_BIT_LOCALES\n\n");
|
| 114 |
|
|
|
| 115 |
|
|
fprintf(out, "#define __LOCALE_DATA_NUM_CODESETS\t\t0\n");
|
| 116 |
|
|
fprintf(out, "#define __LOCALE_DATA_CODESET_LIST\t\t\"\"\n");
|
| 117 |
|
|
fclose(out);
|
| 118 |
|
|
return EXIT_SUCCESS;
|
| 119 |
|
|
}
|
| 120 |
|
|
|
| 121 |
|
|
/* fprintf(out, "#define __CTYPE_HAS_8_BIT_LOCALES\t1\n\n"); */
|
| 122 |
|
|
fprintf(out, "#ifdef __CTYPE_HAS_8_BIT_LOCALES\n\n");
|
| 123 |
|
|
#endif
|
| 124 |
|
|
|
| 125 |
|
|
if (argc == 1) {
|
| 126 |
|
|
fprintf(out, "#undef __CTYPE_HAS_8_BIT_LOCALES\n\n");
|
| 127 |
|
|
|
| 128 |
|
|
fprintf(out, "#define __LOCALE_DATA_NUM_CODESETS\t\t0\n");
|
| 129 |
|
|
fprintf(out, "#define __LOCALE_DATA_CODESET_LIST\t\t\"\"\n");
|
| 130 |
|
|
} else {
|
| 131 |
|
|
fprintf(out, "#define __CTYPE_HAS_8_BIT_LOCALES\t\t1\n\n");
|
| 132 |
|
|
}
|
| 133 |
|
|
|
| 134 |
|
|
fprintf(out, "#define __LOCALE_DATA_Cctype_IDX_SHIFT\t%d\n", CTYPE_IDX_SHIFT);
|
| 135 |
|
|
fprintf(out, "#define __LOCALE_DATA_Cctype_IDX_LEN\t\t%d\n", CTYPE_IDX_LEN);
|
| 136 |
|
|
#ifdef CTYPE_PACKED
|
| 137 |
|
|
fprintf(out, "#define __LOCALE_DATA_Cctype_ROW_LEN\t\t%d\n", CTYPE_ROW_LEN >> 1);
|
| 138 |
|
|
fprintf(out, "#define __LOCALE_DATA_Cctype_PACKED\t\t1\n");
|
| 139 |
|
|
#else
|
| 140 |
|
|
fprintf(out, "#define __LOCALE_DATA_Cctype_ROW_LEN\t\t%d\n", CTYPE_ROW_LEN);
|
| 141 |
|
|
fprintf(out, "#undef __LOCALE_DATA_Cctype_PACKED\n");
|
| 142 |
|
|
#endif
|
| 143 |
|
|
|
| 144 |
|
|
fprintf(out, "\n#define __LOCALE_DATA_Cuplow_IDX_SHIFT\t%d\n", UPLOW_IDX_SHIFT);
|
| 145 |
|
|
fprintf(out, "#define __LOCALE_DATA_Cuplow_IDX_LEN\t\t%d\n", UPLOW_IDX_LEN);
|
| 146 |
|
|
fprintf(out, "#define __LOCALE_DATA_Cuplow_ROW_LEN\t\t%d\n", UPLOW_ROW_LEN);
|
| 147 |
|
|
|
| 148 |
|
|
#ifdef DO_WIDE_CHAR
|
| 149 |
|
|
fprintf(out, "\n#define __LOCALE_DATA_Cc2wc_IDX_LEN\t\t%d\n", C2WC_IDX_LEN);
|
| 150 |
|
|
fprintf(out, "#define __LOCALE_DATA_Cc2wc_IDX_SHIFT\t\t%d\n", C2WC_IDX_SHIFT);
|
| 151 |
|
|
fprintf(out, "#define __LOCALE_DATA_Cc2wc_ROW_LEN\t\t%d\n", C2WC_ROW_LEN);
|
| 152 |
|
|
#endif
|
| 153 |
|
|
|
| 154 |
|
|
fprintf(out, "\ntypedef struct {\n");
|
| 155 |
|
|
fprintf(out, "\tunsigned char idx8ctype[%d];\n", CTYPE_IDX_LEN);
|
| 156 |
|
|
fprintf(out, "\tunsigned char idx8uplow[%d];\n", UPLOW_IDX_LEN);
|
| 157 |
|
|
#ifdef DO_WIDE_CHAR
|
| 158 |
|
|
fprintf(out, "\tunsigned char idx8c2wc[%d];\n", C2WC_IDX_LEN);
|
| 159 |
|
|
fprintf(out, "\tunsigned char idx8wc2c[%d];\n", II_LEN);
|
| 160 |
|
|
#endif
|
| 161 |
|
|
fprintf(out, "} __codeset_8_bit_t;\n\n");
|
| 162 |
|
|
|
| 163 |
|
|
fprintf(out, "#ifdef WANT_DATA\n\n");
|
| 164 |
|
|
fprintf(out, "static const __codeset_8_bit_t codeset_8_bit[%d] = {\n", argc-1);
|
| 165 |
|
|
|
| 166 |
|
|
max_wchar = 0x7f;
|
| 167 |
|
|
numsets = 0;
|
| 168 |
|
|
codeset_index[0] = 0;
|
| 169 |
|
|
while (--argc) {
|
| 170 |
|
|
if (!(fp = fopen(*++argv,"r"))) {
|
| 171 |
|
|
printf("error: couldn't open file \"%s\"\n", *argv);
|
| 172 |
|
|
return EXIT_FAILURE;
|
| 173 |
|
|
}
|
| 174 |
|
|
printf("processing %s... ", *argv);
|
| 175 |
|
|
|
| 176 |
|
|
{
|
| 177 |
|
|
char *s0;
|
| 178 |
|
|
char *s1;
|
| 179 |
|
|
int n;
|
| 180 |
|
|
|
| 181 |
|
|
s0 = strrchr(*argv, '/');
|
| 182 |
|
|
if (!s0) {
|
| 183 |
|
|
s0 = *argv;
|
| 184 |
|
|
} else {
|
| 185 |
|
|
++s0;
|
| 186 |
|
|
}
|
| 187 |
|
|
s1 = strchr(s0, '.');
|
| 188 |
|
|
if (!s1) {
|
| 189 |
|
|
n = strlen(s0);
|
| 190 |
|
|
} else {
|
| 191 |
|
|
n = s1 - s0;
|
| 192 |
|
|
}
|
| 193 |
|
|
|
| 194 |
|
|
/* if ((numsets == 0) && strncmp("ASCII", s0, n)) { */
|
| 195 |
|
|
/* printf("error - first codeset isn't ASCII!\n"); */
|
| 196 |
|
|
/* return EXIT_FAILURE; */
|
| 197 |
|
|
/* } */
|
| 198 |
|
|
|
| 199 |
|
|
if (numsets >= sizeof(codeset_index)) {
|
| 200 |
|
|
printf("error - too many codesets!\n");
|
| 201 |
|
|
return EXIT_FAILURE;
|
| 202 |
|
|
}
|
| 203 |
|
|
|
| 204 |
|
|
if (codeset_list_end + n + 1 + numsets + 1 + 1 >= 256) {
|
| 205 |
|
|
printf("error - codeset list to big!\n");
|
| 206 |
|
|
return EXIT_FAILURE;
|
| 207 |
|
|
}
|
| 208 |
|
|
|
| 209 |
|
|
codeset_index[numsets+1] = codeset_index[numsets] + n+1;
|
| 210 |
|
|
strncpy(codeset_list + codeset_list_end, s0, n);
|
| 211 |
|
|
codeset_list_end += (n+1);
|
| 212 |
|
|
codeset_list[codeset_list_end - 1] = 0;
|
| 213 |
|
|
|
| 214 |
|
|
fprintf(out, "\t{ /* %.*s */", n, s0);
|
| 215 |
|
|
}
|
| 216 |
|
|
|
| 217 |
|
|
memset(&csd[numsets],sizeof(charset_data),0);
|
| 218 |
|
|
memset(xi, sizeof(xi), 0);
|
| 219 |
|
|
{
|
| 220 |
|
|
unsigned long c, wc;
|
| 221 |
|
|
int lines;
|
| 222 |
|
|
lines = 0;
|
| 223 |
|
|
while (fgets(buf,sizeof(buf),fp)) {
|
| 224 |
|
|
if ((2 != sscanf(buf, "{ %lx , %lx", &c, &wc))
|
| 225 |
|
|
|| (c >= 256) || (wc > MAX_WCHAR)) {
|
| 226 |
|
|
printf("error: scanf failure! \"%s\"\n", buf);
|
| 227 |
|
|
return EXIT_FAILURE;
|
| 228 |
|
|
}
|
| 229 |
|
|
|
| 230 |
|
|
/* don't put in w2c... dynamicly build tt instead. */
|
| 231 |
|
|
|
| 232 |
|
|
if (c <= 0x7f) { /* check the 7bit entries but don't store */
|
| 233 |
|
|
if (c != wc) {
|
| 234 |
|
|
printf("error: c != wc in %s\n", buf);
|
| 235 |
|
|
return EXIT_FAILURE;
|
| 236 |
|
|
}
|
| 237 |
|
|
csd[numsets].c2w[c] = wc;
|
| 238 |
|
|
csd[numsets].w2c[wc] = 0; /* ignore */
|
| 239 |
|
|
if (wc > max_wchar) {
|
| 240 |
|
|
max_wchar = wc;
|
| 241 |
|
|
}
|
| 242 |
|
|
} else {
|
| 243 |
|
|
csd[numsets].c2w[c] = wc;
|
| 244 |
|
|
csd[numsets].w2c[wc] = c;
|
| 245 |
|
|
if (wc > max_wchar) {
|
| 246 |
|
|
max_wchar = wc;
|
| 247 |
|
|
}
|
| 248 |
|
|
}
|
| 249 |
|
|
++lines;
|
| 250 |
|
|
}
|
| 251 |
|
|
printf("%d lines ", lines);
|
| 252 |
|
|
|
| 253 |
|
|
for (i = 0 ; i <= MAX_WCHAR ; i += (1 << TT_SHIFT)) {
|
| 254 |
|
|
p = &csd[numsets].w2c[i];
|
| 255 |
|
|
for (j = 0 ; j < tt_num ; j++) {
|
| 256 |
|
|
if (!memcmp(p, &tt[j << TT_SHIFT], (1 << TT_SHIFT))) {
|
| 257 |
|
|
break;
|
| 258 |
|
|
}
|
| 259 |
|
|
}
|
| 260 |
|
|
if (j == tt_num) { /* new entry */
|
| 261 |
|
|
memcpy(&tt[j << TT_SHIFT], p, (1 << TT_SHIFT));
|
| 262 |
|
|
++tt_num;
|
| 263 |
|
|
}
|
| 264 |
|
|
xi[i >> TT_SHIFT] = j;
|
| 265 |
|
|
}
|
| 266 |
|
|
|
| 267 |
|
|
for (i = 0 ; i <= (MAX_WCHAR >> TT_SHIFT) ; i += (1 << TI_SHIFT)) {
|
| 268 |
|
|
p = &xi[i];
|
| 269 |
|
|
for (j = 0 ; j < ti_num ; j++) {
|
| 270 |
|
|
if (!memcmp(p, &ti[j << TI_SHIFT], (1 << TI_SHIFT))) {
|
| 271 |
|
|
break;
|
| 272 |
|
|
}
|
| 273 |
|
|
}
|
| 274 |
|
|
if (j == ti_num) { /* new entry */
|
| 275 |
|
|
memcpy(&ti[j << TI_SHIFT], p, (1 << TI_SHIFT));
|
| 276 |
|
|
++ti_num;
|
| 277 |
|
|
}
|
| 278 |
|
|
csd[numsets].ii[i >> TI_SHIFT] = j;
|
| 279 |
|
|
/* printf("%d ", i >> TI_SHIFT); */
|
| 280 |
|
|
}
|
| 281 |
|
|
|
| 282 |
|
|
#if 1
|
| 283 |
|
|
fprintf(out, "\n\t\t/* idx8ctype data */\n\t\t{");
|
| 284 |
|
|
for (i = 128 ; i < 256 ; i++) {
|
| 285 |
|
|
wchar_t c;
|
| 286 |
|
|
unsigned int d;
|
| 287 |
|
|
|
| 288 |
|
|
/* if (!(i & 0x7)) { */
|
| 289 |
|
|
/* fprintf(out, "\n"); */
|
| 290 |
|
|
/* } */
|
| 291 |
|
|
|
| 292 |
|
|
c = csd[numsets].c2w[i];
|
| 293 |
|
|
|
| 294 |
|
|
if (c == 0) { /* non-existant char in codeset */
|
| 295 |
|
|
d = __CTYPE_unclassified;
|
| 296 |
|
|
} else if (iswdigit(c)) {
|
| 297 |
|
|
d = __CTYPE_digit;
|
| 298 |
|
|
} else if (iswalpha(c)) {
|
| 299 |
|
|
d = __CTYPE_alpha_nonupper_nonlower;
|
| 300 |
|
|
if (iswlower(c)) {
|
| 301 |
|
|
d = __CTYPE_alpha_lower;
|
| 302 |
|
|
if (iswupper(c)) {
|
| 303 |
|
|
d = __CTYPE_alpha_upper_lower;
|
| 304 |
|
|
}
|
| 305 |
|
|
} else if (iswupper(c)) {
|
| 306 |
|
|
d = __CTYPE_alpha_upper;
|
| 307 |
|
|
}
|
| 308 |
|
|
} else if (iswpunct(c)) {
|
| 309 |
|
|
d = __CTYPE_punct;
|
| 310 |
|
|
} else if (iswgraph(c)) {
|
| 311 |
|
|
d = __CTYPE_graph;
|
| 312 |
|
|
} else if (iswprint(c)) {
|
| 313 |
|
|
d = __CTYPE_print_space_nonblank;
|
| 314 |
|
|
if (iswblank(c)) {
|
| 315 |
|
|
d = __CTYPE_print_space_blank;
|
| 316 |
|
|
}
|
| 317 |
|
|
} else if (iswspace(c) && !iswcntrl(c)) {
|
| 318 |
|
|
d = __CTYPE_space_nonblank_noncntrl;
|
| 319 |
|
|
if (iswblank(c)) {
|
| 320 |
|
|
d = __CTYPE_space_blank_noncntrl;
|
| 321 |
|
|
}
|
| 322 |
|
|
} else if (iswcntrl(c)) {
|
| 323 |
|
|
d = __CTYPE_cntrl_nonspace;
|
| 324 |
|
|
if (iswspace(c)) {
|
| 325 |
|
|
d = __CTYPE_cntrl_space_nonblank;
|
| 326 |
|
|
if (iswblank(c)) {
|
| 327 |
|
|
d = __CTYPE_cntrl_space_blank;
|
| 328 |
|
|
}
|
| 329 |
|
|
}
|
| 330 |
|
|
} else {
|
| 331 |
|
|
d = __CTYPE_unclassified;
|
| 332 |
|
|
}
|
| 333 |
|
|
|
| 334 |
|
|
#if 1
|
| 335 |
|
|
row[i & (CTYPE_ROW_LEN-1)] = d;
|
| 336 |
|
|
if ((i & (CTYPE_ROW_LEN-1)) == (CTYPE_ROW_LEN-1)) {
|
| 337 |
|
|
p = ctype_tbl;
|
| 338 |
|
|
for (j=0 ; j < n_ctype_rows ; j++) {
|
| 339 |
|
|
if (!memcmp(p, row, CTYPE_ROW_LEN)) {
|
| 340 |
|
|
break;
|
| 341 |
|
|
}
|
| 342 |
|
|
p += CTYPE_ROW_LEN;
|
| 343 |
|
|
}
|
| 344 |
|
|
if (j == n_ctype_rows) { /* new entry */
|
| 345 |
|
|
if (++n_ctype_rows > 256) {
|
| 346 |
|
|
printf("error -- to many ctype rows!\n");
|
| 347 |
|
|
return EXIT_FAILURE;
|
| 348 |
|
|
}
|
| 349 |
|
|
memcpy(p, row, CTYPE_ROW_LEN);
|
| 350 |
|
|
}
|
| 351 |
|
|
csd[numsets].ctype_idx[i >> CTYPE_IDX_SHIFT] = j;
|
| 352 |
|
|
if (!((i >> CTYPE_IDX_SHIFT) & 0x7)
|
| 353 |
|
|
&& (i != (127 + CTYPE_ROW_LEN))
|
| 354 |
|
|
) {
|
| 355 |
|
|
fprintf(out, "\n\t\t ");
|
| 356 |
|
|
}
|
| 357 |
|
|
fprintf(out, " %#4x,", j);
|
| 358 |
|
|
}
|
| 359 |
|
|
#else
|
| 360 |
|
|
fprintf(out, " %#4x,", d);
|
| 361 |
|
|
#endif
|
| 362 |
|
|
}
|
| 363 |
|
|
#endif
|
| 364 |
|
|
fprintf(out, " }");
|
| 365 |
|
|
|
| 366 |
|
|
#if 1
|
| 367 |
|
|
fprintf(out, ",\n\t\t/* idx8uplow data */\n\t\t{");
|
| 368 |
|
|
for (i = 128 ; i < 256 ; i++) {
|
| 369 |
|
|
wchar_t c, u, l;
|
| 370 |
|
|
/* if (!(i & 0x7)) { */
|
| 371 |
|
|
/* fprintf(out, "\n"); */
|
| 372 |
|
|
/* } */
|
| 373 |
|
|
c = csd[numsets].c2w[i];
|
| 374 |
|
|
if ((c != 0) || 1) {
|
| 375 |
|
|
u = towupper(c);
|
| 376 |
|
|
l = towlower(c);
|
| 377 |
|
|
|
| 378 |
|
|
if (u >= 0x80) u = csd[numsets].w2c[u];
|
| 379 |
|
|
if (l >= 0x80) l = csd[numsets].w2c[l];
|
| 380 |
|
|
|
| 381 |
|
|
if (u == 0) u = i; /* upper is missing, so ignore */
|
| 382 |
|
|
if (l == 0) l = i; /* lower is missing, so ignore */
|
| 383 |
|
|
|
| 384 |
|
|
#if 1
|
| 385 |
|
|
/* store as unsigned char and let overflow handle it. */
|
| 386 |
|
|
/* if ((((u-i) < CHAR_MIN) || ((u-i) > CHAR_MAX)) */
|
| 387 |
|
|
/* || (((i-l) < CHAR_MIN) || ((i-l) > CHAR_MAX)) */
|
| 388 |
|
|
/* ) { */
|
| 389 |
|
|
/* printf("error - uplow diff out of range! %d %ld %ld\n", */
|
| 390 |
|
|
/* i, u, l); */
|
| 391 |
|
|
/* return EXIT_FAILURE; */
|
| 392 |
|
|
/* } */
|
| 393 |
|
|
|
| 394 |
|
|
row[i & (UPLOW_ROW_LEN-1)] = ((l==i) ? (u-i) : (i-l));
|
| 395 |
|
|
if ((i & (UPLOW_ROW_LEN-1)) == (UPLOW_ROW_LEN-1)) {
|
| 396 |
|
|
p = uplow_tbl;
|
| 397 |
|
|
for (j=0 ; j < n_uplow_rows ; j++) {
|
| 398 |
|
|
if (!memcmp(p, row, UPLOW_ROW_LEN)) {
|
| 399 |
|
|
break;
|
| 400 |
|
|
}
|
| 401 |
|
|
p += UPLOW_ROW_LEN;
|
| 402 |
|
|
}
|
| 403 |
|
|
if (j == n_uplow_rows) { /* new entry */
|
| 404 |
|
|
if (++n_uplow_rows > 256) {
|
| 405 |
|
|
printf("error -- to many uplow rows!\n");
|
| 406 |
|
|
return EXIT_FAILURE;
|
| 407 |
|
|
}
|
| 408 |
|
|
memcpy(p, row, UPLOW_ROW_LEN);
|
| 409 |
|
|
}
|
| 410 |
|
|
csd[numsets].uplow_idx[i >> UPLOW_IDX_SHIFT] = j;
|
| 411 |
|
|
if (!((i >> UPLOW_IDX_SHIFT) & 0x7)
|
| 412 |
|
|
&& (i != (127 + UPLOW_ROW_LEN))
|
| 413 |
|
|
) {
|
| 414 |
|
|
fprintf(out, "\n\t\t ");
|
| 415 |
|
|
}
|
| 416 |
|
|
fprintf(out, " %#4x,", j);
|
| 417 |
|
|
}
|
| 418 |
|
|
|
| 419 |
|
|
#elif 0
|
| 420 |
|
|
if (!(i & 0x7) && i) {
|
| 421 |
|
|
fprintf(out, "\n");
|
| 422 |
|
|
}
|
| 423 |
|
|
fprintf(out, " %4ld,", (l==i) ? (u-i) : (i-l));
|
| 424 |
|
|
/* fprintf(out, " %4ld,", (l==i) ? u : l); */
|
| 425 |
|
|
#else
|
| 426 |
|
|
if ((u != i) || (l != i)) {
|
| 427 |
|
|
#if 0
|
| 428 |
|
|
fprintf(out, " %#08lx, %#08lx, %#08lx, %#08lx, %#08lx, %#08lx, \n",
|
| 429 |
|
|
(unsigned long) i,
|
| 430 |
|
|
(unsigned long) c,
|
| 431 |
|
|
(unsigned long) l,
|
| 432 |
|
|
(unsigned long) towlower(c),
|
| 433 |
|
|
(unsigned long) u,
|
| 434 |
|
|
(unsigned long) towupper(c));
|
| 435 |
|
|
|
| 436 |
|
|
#else
|
| 437 |
|
|
fprintf(out, " %#08lx, %8ld, %d, %8ld, %d, %#08lx\n",
|
| 438 |
|
|
(unsigned long) i,
|
| 439 |
|
|
(long) (l - i),
|
| 440 |
|
|
iswupper(c),
|
| 441 |
|
|
(long) (i - u),
|
| 442 |
|
|
iswlower(c),
|
| 443 |
|
|
(unsigned long) c);
|
| 444 |
|
|
#endif
|
| 445 |
|
|
}
|
| 446 |
|
|
#endif
|
| 447 |
|
|
}
|
| 448 |
|
|
}
|
| 449 |
|
|
fprintf(out, " }");
|
| 450 |
|
|
#endif
|
| 451 |
|
|
|
| 452 |
|
|
#ifndef DO_WIDE_CHAR
|
| 453 |
|
|
fprintf(out,"\n");
|
| 454 |
|
|
#else /* DO_WIDE_CHAR */
|
| 455 |
|
|
|
| 456 |
|
|
#if 1
|
| 457 |
|
|
fprintf(out, ",\n\t\t/* idx8c2wc data */\n\t\t{");
|
| 458 |
|
|
for (i = 128 ; i < 256 ; i++) {
|
| 459 |
|
|
#if 1
|
| 460 |
|
|
wrow[i & (C2WC_ROW_LEN-1)] = csd[numsets].c2w[i];
|
| 461 |
|
|
if ((i & (C2WC_ROW_LEN-1)) == (C2WC_ROW_LEN-1)) {
|
| 462 |
|
|
p = (char *) c2wc_tbl;
|
| 463 |
|
|
for (j=0 ; j < n_c2wc_rows ; j++) {
|
| 464 |
|
|
if (!memcmp(p, (char *) wrow, 2*C2WC_ROW_LEN)) {
|
| 465 |
|
|
break;
|
| 466 |
|
|
}
|
| 467 |
|
|
p += 2*C2WC_ROW_LEN;
|
| 468 |
|
|
}
|
| 469 |
|
|
if (j == n_c2wc_rows) { /* new entry */
|
| 470 |
|
|
if (++n_c2wc_rows > 256) {
|
| 471 |
|
|
printf("error -- to many c2wc rows!\n");
|
| 472 |
|
|
return EXIT_FAILURE;
|
| 473 |
|
|
}
|
| 474 |
|
|
memcpy(p, (char *) wrow, 2*C2WC_ROW_LEN);
|
| 475 |
|
|
}
|
| 476 |
|
|
csd[numsets].c2wc_idx[i >> C2WC_IDX_SHIFT] = j;
|
| 477 |
|
|
if (!((i >> C2WC_IDX_SHIFT) & 0x7)
|
| 478 |
|
|
&& (i != (127 + C2WC_ROW_LEN))
|
| 479 |
|
|
) {
|
| 480 |
|
|
fprintf(out, "\n\t\t ");
|
| 481 |
|
|
}
|
| 482 |
|
|
fprintf(out, " %#4x,", j);
|
| 483 |
|
|
}
|
| 484 |
|
|
#else
|
| 485 |
|
|
if (!(i & 0x7) && i) {
|
| 486 |
|
|
fprintf(out, "\n");
|
| 487 |
|
|
}
|
| 488 |
|
|
fprintf(out, " %#6lx,", csd[numsets].c2w[i]);
|
| 489 |
|
|
#endif
|
| 490 |
|
|
}
|
| 491 |
|
|
fprintf(out, " },\n");
|
| 492 |
|
|
#endif
|
| 493 |
|
|
|
| 494 |
|
|
#if 1
|
| 495 |
|
|
/* fprintf(out, "\nII_LEN = %d\n", II_LEN); */
|
| 496 |
|
|
fprintf(out, "\t\t/* idx8wc2c data */\n\t\t{");
|
| 497 |
|
|
for (i = 0 ; i < II_LEN ; i++) {
|
| 498 |
|
|
if (!(i & 0x7) && i) {
|
| 499 |
|
|
fprintf(out, "\n\t\t ");
|
| 500 |
|
|
}
|
| 501 |
|
|
fprintf(out, " %#4x,", csd[numsets].ii[i]);
|
| 502 |
|
|
}
|
| 503 |
|
|
fprintf(out, " }\n");
|
| 504 |
|
|
#endif
|
| 505 |
|
|
|
| 506 |
|
|
#endif /* DO_WIDE_CHAR */
|
| 507 |
|
|
fprintf(out, "\t},\n");
|
| 508 |
|
|
|
| 509 |
|
|
}
|
| 510 |
|
|
++numsets;
|
| 511 |
|
|
printf("done\n");
|
| 512 |
|
|
}
|
| 513 |
|
|
fprintf(out, "};\n");
|
| 514 |
|
|
fprintf(out, "\n#endif /* WANT_DATA */\n");
|
| 515 |
|
|
|
| 516 |
|
|
#ifdef DO_WIDE_CHAR
|
| 517 |
|
|
fprintf(out, "\n");
|
| 518 |
|
|
fprintf(out, "#define __LOCALE_DATA_Cwc2c_DOMAIN_MAX\t%#x\n", RANGE);
|
| 519 |
|
|
fprintf(out, "#define __LOCALE_DATA_Cwc2c_TI_SHIFT\t\t%d\n", TI_SHIFT);
|
| 520 |
|
|
fprintf(out, "#define __LOCALE_DATA_Cwc2c_TT_SHIFT\t\t%d\n", TT_SHIFT);
|
| 521 |
|
|
fprintf(out, "#define __LOCALE_DATA_Cwc2c_II_LEN\t\t%d\n", II_LEN);
|
| 522 |
|
|
fprintf(out, "#define __LOCALE_DATA_Cwc2c_TI_LEN\t\t%d\n", ti_num << TI_SHIFT);
|
| 523 |
|
|
fprintf(out, "#define __LOCALE_DATA_Cwc2c_TT_LEN\t\t%d\n", tt_num << TT_SHIFT);
|
| 524 |
|
|
fprintf(out, "\n");
|
| 525 |
|
|
|
| 526 |
|
|
fprintf(out, "\n#define __LOCALE_DATA_Cwc2c_TBL_LEN\t\t%d\n",
|
| 527 |
|
|
(ti_num << TI_SHIFT) + (tt_num << TT_SHIFT));
|
| 528 |
|
|
|
| 529 |
|
|
fprintf(out, "#ifdef WANT_DATA\n\n");
|
| 530 |
|
|
fprintf(out, "static const unsigned char __LOCALE_DATA_Cwc2c_data[%d] = {\n",
|
| 531 |
|
|
(ti_num << TI_SHIFT) + (tt_num << TT_SHIFT));
|
| 532 |
|
|
fprintf(out, "\t/* ti_table */\n\t");
|
| 533 |
|
|
for (i=0 ; i < ti_num << TI_SHIFT ; i++) {
|
| 534 |
|
|
if (!(i & 7) && i) {
|
| 535 |
|
|
fprintf(out, "\n\t");
|
| 536 |
|
|
}
|
| 537 |
|
|
fprintf(out, " %#4x,", ti[i]);
|
| 538 |
|
|
}
|
| 539 |
|
|
fprintf(out, "\n");
|
| 540 |
|
|
fprintf(out, "\t/* tt_table */\n\t");
|
| 541 |
|
|
for (i=0 ; i < tt_num << TT_SHIFT ; i++) {
|
| 542 |
|
|
if (!(i & 7) && i) {
|
| 543 |
|
|
fprintf(out, "\n\t");
|
| 544 |
|
|
}
|
| 545 |
|
|
fprintf(out, " %#4x,", tt[i]);
|
| 546 |
|
|
}
|
| 547 |
|
|
fprintf(out, "\n};\n");
|
| 548 |
|
|
|
| 549 |
|
|
fprintf(out, "\n#endif /* WANT_DATA */\n");
|
| 550 |
|
|
#endif /* DO_WIDE_CHAR */
|
| 551 |
|
|
|
| 552 |
|
|
fprintf(out, "\n#define __LOCALE_DATA_Cuplow_TBL_LEN\t\t%d\n",
|
| 553 |
|
|
n_uplow_rows * UPLOW_ROW_LEN);
|
| 554 |
|
|
fprintf(out, "\n#ifdef WANT_DATA\n\n");
|
| 555 |
|
|
|
| 556 |
|
|
fprintf(out, "\nstatic const unsigned char __LOCALE_DATA_Cuplow_data[%d] = {\n",
|
| 557 |
|
|
n_uplow_rows * UPLOW_ROW_LEN);
|
| 558 |
|
|
p = uplow_tbl;
|
| 559 |
|
|
for (j=0 ; j < n_uplow_rows ; j++) {
|
| 560 |
|
|
fprintf(out, "\t");
|
| 561 |
|
|
for (i=0 ; i < UPLOW_ROW_LEN ; i++) {
|
| 562 |
|
|
fprintf(out, " %#4x,", (unsigned int)((unsigned char) p[i]));
|
| 563 |
|
|
}
|
| 564 |
|
|
fprintf(out, "\n");
|
| 565 |
|
|
p += UPLOW_ROW_LEN;
|
| 566 |
|
|
}
|
| 567 |
|
|
fprintf(out, "};\n");
|
| 568 |
|
|
|
| 569 |
|
|
fprintf(out, "\n#endif /* WANT_DATA */\n");
|
| 570 |
|
|
fprintf(out, "\n#define __LOCALE_DATA_Cctype_TBL_LEN\t\t%d\n",
|
| 571 |
|
|
#ifdef CTYPE_PACKED
|
| 572 |
|
|
n_ctype_rows * CTYPE_ROW_LEN / 2
|
| 573 |
|
|
#else
|
| 574 |
|
|
n_ctype_rows * CTYPE_ROW_LEN
|
| 575 |
|
|
#endif
|
| 576 |
|
|
);
|
| 577 |
|
|
fprintf(out, "\n#ifdef WANT_DATA\n\n");
|
| 578 |
|
|
|
| 579 |
|
|
|
| 580 |
|
|
fprintf(out, "\nstatic const unsigned char __LOCALE_DATA_Cctype_data[%d] = {\n",
|
| 581 |
|
|
#ifdef CTYPE_PACKED
|
| 582 |
|
|
n_ctype_rows * CTYPE_ROW_LEN / 2
|
| 583 |
|
|
#else
|
| 584 |
|
|
n_ctype_rows * CTYPE_ROW_LEN
|
| 585 |
|
|
#endif
|
| 586 |
|
|
);
|
| 587 |
|
|
p = ctype_tbl;
|
| 588 |
|
|
for (j=0 ; j < n_ctype_rows ; j++) {
|
| 589 |
|
|
fprintf(out, "\t");
|
| 590 |
|
|
for (i=0 ; i < CTYPE_ROW_LEN ; i++) {
|
| 591 |
|
|
#ifdef CTYPE_PACKED
|
| 592 |
|
|
fprintf(out, " %#4x,", (unsigned int)(p[i] + (p[i+1] << 4)));
|
| 593 |
|
|
++i;
|
| 594 |
|
|
#else
|
| 595 |
|
|
fprintf(out, " %#4x,", (unsigned int)p[i]);
|
| 596 |
|
|
#endif
|
| 597 |
|
|
}
|
| 598 |
|
|
fprintf(out, "\n");
|
| 599 |
|
|
p += CTYPE_ROW_LEN;
|
| 600 |
|
|
}
|
| 601 |
|
|
fprintf(out, "};\n");
|
| 602 |
|
|
|
| 603 |
|
|
fprintf(out, "\n#endif /* WANT_DATA */\n");
|
| 604 |
|
|
|
| 605 |
|
|
#ifdef DO_WIDE_CHAR
|
| 606 |
|
|
|
| 607 |
|
|
fprintf(out, "\n#define __LOCALE_DATA_Cc2wc_TBL_LEN\t\t%d\n",
|
| 608 |
|
|
n_c2wc_rows * C2WC_ROW_LEN);
|
| 609 |
|
|
fprintf(out, "\n#ifdef WANT_DATA\n\n");
|
| 610 |
|
|
|
| 611 |
|
|
fprintf(out, "\nstatic const unsigned short __LOCALE_DATA_Cc2wc_data[%d] = {\n",
|
| 612 |
|
|
n_c2wc_rows * C2WC_ROW_LEN);
|
| 613 |
|
|
p = (char *) c2wc_tbl;
|
| 614 |
|
|
for (j=0 ; j < n_c2wc_rows ; j++) {
|
| 615 |
|
|
fprintf(out, "\t");
|
| 616 |
|
|
for (i=0 ; i < C2WC_ROW_LEN ; i++) {
|
| 617 |
|
|
fprintf(out, " %#6x,", (unsigned int)(((unsigned short *)p)[i]));
|
| 618 |
|
|
}
|
| 619 |
|
|
fprintf(out, "\n");
|
| 620 |
|
|
p += 2*C2WC_ROW_LEN;
|
| 621 |
|
|
}
|
| 622 |
|
|
fprintf(out, "};\n");
|
| 623 |
|
|
fprintf(out, "\n#endif /* WANT_DATA */\n");
|
| 624 |
|
|
#endif /* DO_WIDE_CHAR */
|
| 625 |
|
|
fprintf(out, "\n\n");
|
| 626 |
|
|
|
| 627 |
|
|
fprintf(out, "#define __LOCALE_DATA_NUM_CODESETS\t\t%d\n", numsets);
|
| 628 |
|
|
fprintf(out, "#define __LOCALE_DATA_CODESET_LIST \\\n\t\"");
|
| 629 |
|
|
for (i=0 ; i < numsets ; i++) {
|
| 630 |
|
|
fprintf(out, "\\x%02x", numsets + 1 + (unsigned char) codeset_index[i]);
|
| 631 |
|
|
if (((i & 7) == 7) && (i + 1 < numsets)) {
|
| 632 |
|
|
fprintf(out, "\" \\\n\t\"");
|
| 633 |
|
|
}
|
| 634 |
|
|
}
|
| 635 |
|
|
fprintf(out, "\" \\\n\t\"\\0\"");
|
| 636 |
|
|
for (i=0 ; i < numsets ; i++) {
|
| 637 |
|
|
fprintf(out, " \\\n\t\"%s\\0\"",
|
| 638 |
|
|
codeset_list + ((unsigned char)codeset_index[i]));
|
| 639 |
|
|
}
|
| 640 |
|
|
|
| 641 |
|
|
fprintf(out, "\n\n");
|
| 642 |
|
|
for (i=0 ; i < numsets ; i++) {
|
| 643 |
|
|
char buf[30];
|
| 644 |
|
|
char *z;
|
| 645 |
|
|
strcpy(buf, codeset_list + ((unsigned char)codeset_index[i]));
|
| 646 |
|
|
for (z=buf ; *z ; z++) {
|
| 647 |
|
|
if (*z == '-') {
|
| 648 |
|
|
*z = '_';
|
| 649 |
|
|
}
|
| 650 |
|
|
}
|
| 651 |
|
|
fprintf(out, "#define __CTYPE_HAS_CODESET_%s\n", buf);
|
| 652 |
|
|
}
|
| 653 |
|
|
#ifdef DO_WIDE_CHAR
|
| 654 |
|
|
fprintf(out, "#define __CTYPE_HAS_CODESET_UTF_8\n");
|
| 655 |
|
|
#endif /* DO_WIDE_CHAR */
|
| 656 |
|
|
|
| 657 |
|
|
#if 0
|
| 658 |
|
|
fprintf(out, "\n#endif /* __CTYPE_HAS_8_BIT_LOCALES */\n\n");
|
| 659 |
|
|
#endif
|
| 660 |
|
|
|
| 661 |
|
|
fclose(out);
|
| 662 |
|
|
|
| 663 |
|
|
total_size = 0;
|
| 664 |
|
|
#ifdef DO_WIDE_CHAR
|
| 665 |
|
|
printf("tt_num = %d ti_num = %d\n", tt_num, ti_num);
|
| 666 |
|
|
printf("max_wchar = %#lx\n", max_wchar);
|
| 667 |
|
|
|
| 668 |
|
|
printf("size is %d * %d + %d * %d + %d * %d = %d\n",
|
| 669 |
|
|
tt_num, 1 << TT_SHIFT, ti_num, 1 << TI_SHIFT,
|
| 670 |
|
|
((MAX_WCHAR >> (TT_SHIFT + TI_SHIFT)) + 1), numsets,
|
| 671 |
|
|
j = tt_num * (1 << TT_SHIFT) + ti_num * (1 << TI_SHIFT)
|
| 672 |
|
|
+ ((MAX_WCHAR >> (TT_SHIFT + TI_SHIFT)) + 1) * numsets);
|
| 673 |
|
|
total_size += j;
|
| 674 |
|
|
#endif /* DO_WIDE_CHAR */
|
| 675 |
|
|
|
| 676 |
|
|
#ifdef CTYPE_PACKED
|
| 677 |
|
|
i = 2;
|
| 678 |
|
|
#else
|
| 679 |
|
|
i = 1;
|
| 680 |
|
|
#endif
|
| 681 |
|
|
|
| 682 |
|
|
printf("ctype - CTYPE_IDX_SHIFT = %d -- %d * %d + %d * %d = %d\n",
|
| 683 |
|
|
CTYPE_IDX_SHIFT, numsets, CTYPE_IDX_LEN, n_ctype_rows, CTYPE_ROW_LEN / i,
|
| 684 |
|
|
j = numsets * CTYPE_IDX_LEN + n_ctype_rows * CTYPE_ROW_LEN / i);
|
| 685 |
|
|
total_size += j;
|
| 686 |
|
|
|
| 687 |
|
|
printf("uplow - UPLOW_IDX_SHIFT = %d -- %d * %d + %d * %d = %d\n",
|
| 688 |
|
|
UPLOW_IDX_SHIFT, numsets, UPLOW_IDX_LEN, n_uplow_rows, UPLOW_ROW_LEN,
|
| 689 |
|
|
j = numsets * UPLOW_IDX_LEN + n_uplow_rows * UPLOW_ROW_LEN);
|
| 690 |
|
|
total_size += j;
|
| 691 |
|
|
|
| 692 |
|
|
#ifdef DO_WIDE_CHAR
|
| 693 |
|
|
|
| 694 |
|
|
printf("c2wc - C2WC_IDX_SHIFT = %d -- %d * %d + 2 * %d * %d = %d\n",
|
| 695 |
|
|
C2WC_IDX_SHIFT, numsets, C2WC_IDX_LEN, n_c2wc_rows, C2WC_ROW_LEN,
|
| 696 |
|
|
j = numsets * C2WC_IDX_LEN + 2 * n_c2wc_rows * C2WC_ROW_LEN);
|
| 697 |
|
|
total_size += j;
|
| 698 |
|
|
|
| 699 |
|
|
#endif /* DO_WIDE_CHAR */
|
| 700 |
|
|
|
| 701 |
|
|
printf("total size = %d\n", total_size);
|
| 702 |
|
|
|
| 703 |
|
|
/* for (i=0 ; i < numsets ; i++) { */
|
| 704 |
|
|
/* printf("codeset_index[i] = %d codeset_list[ci[i]] = \"%s\"\n", */
|
| 705 |
|
|
/* (unsigned char) codeset_index[i], */
|
| 706 |
|
|
/* codeset_list + ((unsigned char)codeset_index[i])); */
|
| 707 |
|
|
/* } */
|
| 708 |
|
|
|
| 709 |
|
|
return EXIT_SUCCESS;
|
| 710 |
|
|
}
|
