8 #include "locale_impl.h"
20 #define SHIFT_JIS 0321
27 /* Definitions of charmaps. Each charmap consists of:
28 * 1. Empty-string-terminated list of null-terminated aliases.
29 * 2. Special type code or number of elided entries.
30 * 3. Character table (size determined by field 2). */
32 static const unsigned char charmaps[] =
35 "ucs2\0ucs2be\0\0\304"
37 "utf16\0utf16be\0\0\302"
39 "ucs4\0ucs4be\0utf32\0utf32be\0\0\300"
40 "ucs4le\0utf32le\0\0\303"
41 "ascii\0usascii\0iso646\0iso646us\0\0\307"
43 "shiftjis\0sjis\0\0\321"
47 "big5\0bigfive\0cp950\0big5hkscs\0\0\340"
48 "euckr\0ksc5601\0ksx1001\0cp949\0\0\350"
49 #include "codepages.h"
52 static const unsigned short legacy_chars[] = {
53 #include "legacychars.h"
56 static const unsigned short jis0208[84][94] = {
60 static const unsigned short gb18030[126][190] = {
64 static const unsigned short big5[89][157] = {
68 static const unsigned short hkscs[] = {
72 static const unsigned short ksc[93][94] = {
76 static int fuzzycmp(const unsigned char *a, const unsigned char *b)
78 for (; *a && *b; a++, b++) {
79 while (*a && (*a|32U)-'a'>26 && *a-'0'>10U) a++;
80 if ((*a|32U) != *b) return 1;
85 static size_t find_charmap(const void *name)
87 const unsigned char *s;
88 if (!*(char *)name) name=charmaps; /* "utf8" */
89 for (s=charmaps; *s; ) {
90 if (!fuzzycmp(name, s)) {
91 for (; *s; s+=strlen((void *)s)+1);
94 s += strlen((void *)s)+1;
96 if (s[1] > 0200) s+=2;
97 else s+=2+(128U-s[1])/4*5;
103 iconv_t iconv_open(const char *to, const char *from)
107 if ((t = find_charmap(to))==-1
108 || (f = find_charmap(from))==-1
109 || (charmaps[t] >= 0320)) {
114 return (void *)(f<<16 | t);
117 int iconv_close(iconv_t cd)
122 static unsigned get_16(const unsigned char *s, int e)
125 return s[e]<<8 | s[1-e];
128 static void put_16(unsigned char *s, unsigned c, int e)
135 static unsigned get_32(const unsigned char *s, int e)
138 return s[e]+0U<<24 | s[e^1]<<16 | s[e^2]<<8 | s[e^3];
141 static void put_32(unsigned char *s, unsigned c, int e)
150 /* Adapt as needed */
151 #define mbrtowc_utf8 mbrtowc
152 #define wctomb_utf8 wctomb
154 static unsigned legacy_map(const unsigned char *map, unsigned c)
156 unsigned x = c - 128 - map[-1];
157 x = legacy_chars[ map[x*5/4]>>2*x%8 |
158 map[x*5/4+1]<<8-2*x%8 & 1023 ];
162 size_t iconv(iconv_t cd0, char **restrict in, size_t *restrict inb, char **restrict out, size_t *restrict outb)
165 unsigned long cd = (unsigned long)cd0;
166 unsigned to = cd & 0xffff;
167 unsigned from = cd >> 16;
168 const unsigned char *map = charmaps+from+1;
169 const unsigned char *tomap = charmaps+to+1;
175 unsigned char type = map[-1];
176 unsigned char totype = tomap[-1];
177 locale_t *ploc = &CURRENT_LOCALE, loc = *ploc;
179 if (!in || !*in || !*inb) return 0;
183 for (; *inb; *in+=l, *inb-=l) {
184 c = *(unsigned char *)*in;
187 if (c >= 128 || type-UTF_32BE < 7U) switch (type) {
189 l = mbrtowc_utf8(&wc, *in, *inb, &st);
191 else if (l == (size_t)-1) goto ilseq;
192 else if (l == (size_t)-2) goto starved;
199 if (*inb < l) goto starved;
205 if (*inb < 4) goto starved;
206 c = get_32((void *)*in, type);
208 if (c-0xd800u < 0x800u || c >= 0x110000u) goto ilseq;
215 if (*inb < 2) goto starved;
216 c = get_16((void *)*in, type);
217 if ((unsigned)(c-0xdc00) < 0x400) goto ilseq;
218 if ((unsigned)(c-0xd800) < 0x400) {
219 if (type-UCS2BE < 2U) goto ilseq;
221 if (*inb < 4) goto starved;
222 d = get_16((void *)(*in + 2), type);
223 if ((unsigned)(d-0xdc00) >= 0x400) goto ilseq;
224 c = ((c-0xd7c0)<<10) + (d-0xdc00);
228 if (c-0xa1 <= 0xdf-0xa1) {
233 if (*inb < 2) goto starved;
234 d = *((unsigned char *)*in + 1);
235 if (c-129 <= 159-129) c -= 129;
236 else if (c-224 <= 239-224) c -= 193;
239 if (d-64 <= 158-64) {
240 if (d==127) goto ilseq;
243 } else if (d-159 <= 252-159) {
252 if (*inb < 2) goto starved;
253 d = *((unsigned char *)*in + 1);
256 if (c-0xa1 > 0xdf-0xa1) goto ilseq;
262 if (c >= 84 || d >= 94) goto ilseq;
267 if (c < 0xa1) goto ilseq;
271 if (c >= 126) goto ilseq;
273 if (*inb < 2) goto starved;
274 d = *((unsigned char *)*in + 1);
275 if (d < 0xa1 && type == GB2312) goto ilseq;
276 if (d-0x40>=191 || d==127) {
277 if (d-'0'>9 || type != GB18030)
280 if (*inb < 4) goto starved;
281 c = (10*c + d-'0') * 1260;
282 d = *((unsigned char *)*in + 2);
283 if (d-0x81>126) goto ilseq;
285 d = *((unsigned char *)*in + 3);
286 if (d-'0'>9) goto ilseq;
291 for (int i=0; i<126; i++)
292 for (int j=0; j<190; j++)
293 if (gb18030[i][j]-d <= c-d)
306 if (*inb < 2) goto starved;
307 d = *((unsigned char *)*in + 1);
308 if (d-0x40>=0xff-0x40 || d-0x7f<0xa1-0x7f) goto ilseq;
310 if (d > 0x3e) d -= 0x22;
311 if (c-0xa1>=0xfa-0xa1) {
312 if (c-0x87>=0xff-0x87) goto ilseq;
313 if (c < 0xa1) c -= 0x87;
314 else c -= 0x87 + (0xfa-0xa1);
315 c = (hkscs[4867+(c*157+d)/16]>>(c*157+d)%16)%2<<17
317 /* A few HKSCS characters map to pairs of UCS
318 * characters. These are mapped to surrogate
319 * range in the hkscs table then hard-coded
320 * here. Ugly, yes. */
322 if (totype-0300U > 8) k = 2;
323 else k = "\10\4\4\10\4\4\10\2\4"[totype-0300];
324 if (k > *outb) goto toobig;
325 x += iconv((iconv_t)(uintptr_t)to,
326 &(char *){"\303\212\314\204"
330 +c%256}, &(size_t){4},
338 c = big5[c][d]|(c==0x27&&(d==0x3a||d==0x3c||d==0x42))<<17;
343 if (*inb < 2) goto starved;
344 d = *((unsigned char *)*in + 1);
347 if (c >= 93 || d >= 94) {
350 if (c >= 93 || c>=0xc6-0x81 && d>0x52)
352 if (d-'A'<26) d = d-'A';
353 else if (d-'a'<26) d = d-'a'+26;
354 else if (d-0x81<0xff-0x81) d = d-0x81+52;
356 if (c < 0x20) c = 178*c + d;
357 else c = 178*0x20 + 84*(c-0x20) + d;
359 for (d=0xac00; d<=c; ) {
361 for (int i=0; i<93; i++)
362 for (int j=0; j<94; j++)
363 if (ksc[i][j]-d <= c-d)
374 if (c < 128+type) break;
375 c = legacy_map(map, c);
376 if (c==1) goto ilseq;
381 if (*outb < sizeof(wchar_t)) goto toobig;
382 *(wchar_t *)*out = c;
383 *out += sizeof(wchar_t);
384 *outb -= sizeof(wchar_t);
389 k = wctomb_utf8(tmp, c);
390 if (*outb < k) goto toobig;
391 memcpy(*out, tmp, k);
392 } else k = wctomb_utf8(*out, c);
397 if (c > 0x7f) subst: x++, c='*';
399 if (*outb < 1) goto toobig;
400 if (c < 128+totype || (c<256 && c==legacy_map(tomap, c))) {
407 for (c=128+totype; c<256; c++) {
408 if (d == legacy_map(tomap, c)) {
417 if (c < 0x10000 || type-UCS2BE < 2U) {
418 if (c >= 0x10000) c = 0xFFFD;
419 if (*outb < 2) goto toobig;
420 put_16((void *)*out, c, totype);
425 if (*outb < 4) goto toobig;
427 put_16((void *)*out, (c>>10)|0xd800, totype);
428 put_16((void *)(*out + 2), (c&0x3ff)|0xdc00, totype);
434 if (*outb < 4) goto toobig;
435 put_32((void *)*out, c, totype);