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;
108 static iconv_t combine_to_from(size_t t, size_t f)
110 return (void *)(f<<16 | t<<1 | 1);
113 static size_t extract_from(iconv_t cd)
115 return (size_t)cd >> 16;
118 static size_t extract_to(iconv_t cd)
120 return (size_t)cd >> 1 & 0x7fff;
123 iconv_t iconv_open(const char *to, const char *from)
127 if ((t = find_charmap(to))==-1
128 || (f = find_charmap(from))==-1
129 || (charmaps[t] >= 0320)) {
133 iconv_t cd = combine_to_from(t, f);
136 struct stateful_cd *scd = malloc(sizeof *scd);
137 if (!scd) return (iconv_t)-1;
146 static unsigned get_16(const unsigned char *s, int e)
149 return s[e]<<8 | s[1-e];
152 static void put_16(unsigned char *s, unsigned c, int e)
159 static unsigned get_32(const unsigned char *s, int e)
162 return s[e]+0U<<24 | s[e^1]<<16 | s[e^2]<<8 | s[e^3];
165 static void put_32(unsigned char *s, unsigned c, int e)
174 /* Adapt as needed */
175 #define mbrtowc_utf8 mbrtowc
176 #define wctomb_utf8 wctomb
178 static unsigned legacy_map(const unsigned char *map, unsigned c)
180 unsigned x = c - 128 - map[-1];
181 x = legacy_chars[ map[x*5/4]>>2*x%8 |
182 map[x*5/4+1]<<8-2*x%8 & 1023 ];
186 size_t iconv(iconv_t cd, char **restrict in, size_t *restrict inb, char **restrict out, size_t *restrict outb)
189 struct stateful_cd *scd=0;
190 if (!((size_t)cd & 1)) {
194 unsigned to = extract_to(cd);
195 unsigned from = extract_from(cd);
196 const unsigned char *map = charmaps+from+1;
197 const unsigned char *tomap = charmaps+to+1;
203 unsigned char type = map[-1];
204 unsigned char totype = tomap[-1];
205 locale_t *ploc = &CURRENT_LOCALE, loc = *ploc;
207 if (!in || !*in || !*inb) return 0;
211 for (; *inb; *in+=l, *inb-=l) {
212 c = *(unsigned char *)*in;
218 l = mbrtowc_utf8(&wc, *in, *inb, &st);
219 if (l == (size_t)-1) goto ilseq;
220 if (l == (size_t)-2) goto starved;
224 if (c >= 128) goto ilseq;
228 if (*inb < l) goto starved;
234 if (*inb < 4) goto starved;
235 c = get_32((void *)*in, type);
237 if (c-0xd800u < 0x800u || c >= 0x110000u) goto ilseq;
244 if (*inb < 2) goto starved;
245 c = get_16((void *)*in, type);
246 if ((unsigned)(c-0xdc00) < 0x400) goto ilseq;
247 if ((unsigned)(c-0xd800) < 0x400) {
248 if (type-UCS2BE < 2U) goto ilseq;
250 if (*inb < 4) goto starved;
251 d = get_16((void *)(*in + 2), type);
252 if ((unsigned)(d-0xdc00) >= 0x400) goto ilseq;
253 c = ((c-0xd7c0)<<10) + (d-0xdc00);
258 if (c-0xa1 <= 0xdf-0xa1) {
263 if (*inb < 2) goto starved;
264 d = *((unsigned char *)*in + 1);
265 if (c-129 <= 159-129) c -= 129;
266 else if (c-224 <= 239-224) c -= 193;
269 if (d-64 <= 158-64) {
270 if (d==127) goto ilseq;
273 } else if (d-159 <= 252-159) {
283 if (*inb < 2) goto starved;
284 d = *((unsigned char *)*in + 1);
287 if (c-0xa1 > 0xdf-0xa1) goto ilseq;
293 if (c >= 84 || d >= 94) goto ilseq;
299 if (c < 0xa1) goto ilseq;
304 if (c >= 126) goto ilseq;
306 if (*inb < 2) goto starved;
307 d = *((unsigned char *)*in + 1);
308 if (d < 0xa1 && type == GB2312) goto ilseq;
309 if (d-0x40>=191 || d==127) {
310 if (d-'0'>9 || type != GB18030)
313 if (*inb < 4) goto starved;
314 c = (10*c + d-'0') * 1260;
315 d = *((unsigned char *)*in + 2);
316 if (d-0x81>126) goto ilseq;
318 d = *((unsigned char *)*in + 3);
319 if (d-'0'>9) goto ilseq;
324 for (int i=0; i<126; i++)
325 for (int j=0; j<190; j++)
326 if (gb18030[i][j]-d <= c-d)
340 if (*inb < 2) goto starved;
341 d = *((unsigned char *)*in + 1);
342 if (d-0x40>=0xff-0x40 || d-0x7f<0xa1-0x7f) goto ilseq;
344 if (d > 0x3e) d -= 0x22;
345 if (c-0xa1>=0xfa-0xa1) {
346 if (c-0x87>=0xff-0x87) goto ilseq;
347 if (c < 0xa1) c -= 0x87;
348 else c -= 0x87 + (0xfa-0xa1);
349 c = (hkscs[4867+(c*157+d)/16]>>(c*157+d)%16)%2<<17
351 /* A few HKSCS characters map to pairs of UCS
352 * characters. These are mapped to surrogate
353 * range in the hkscs table then hard-coded
354 * here. Ugly, yes. */
356 if (totype-0300U > 8) k = 2;
357 else k = "\10\4\4\10\4\4\10\2\4"[totype-0300];
358 if (k > *outb) goto toobig;
359 x += iconv(combine_to_from(to, 0),
360 &(char *){"\303\212\314\204"
364 +c%256}, &(size_t){4},
372 c = big5[c][d]|(c==0x27&&(d==0x3a||d==0x3c||d==0x42))<<17;
378 if (*inb < 2) goto starved;
379 d = *((unsigned char *)*in + 1);
382 if (c >= 93 || d >= 94) {
385 if (c >= 93 || c>=0xc6-0x81 && d>0x52)
387 if (d-'A'<26) d = d-'A';
388 else if (d-'a'<26) d = d-'a'+26;
389 else if (d-0x81<0xff-0x81) d = d-0x81+52;
391 if (c < 0x20) c = 178*c + d;
392 else c = 178*0x20 + 84*(c-0x20) + d;
394 for (d=0xac00; d<=c; ) {
396 for (int i=0; i<93; i++)
397 for (int j=0; j<94; j++)
398 if (ksc[i][j]-d <= c-d)
409 if (c < 128+type) break;
410 c = legacy_map(map, c);
411 if (c==1) goto ilseq;
416 if (*outb < sizeof(wchar_t)) goto toobig;
417 *(wchar_t *)*out = c;
418 *out += sizeof(wchar_t);
419 *outb -= sizeof(wchar_t);
424 k = wctomb_utf8(tmp, c);
425 if (*outb < k) goto toobig;
426 memcpy(*out, tmp, k);
427 } else k = wctomb_utf8(*out, c);
432 if (c > 0x7f) subst: x++, c='*';
434 if (*outb < 1) goto toobig;
435 if (c < 128+totype || (c<256 && c==legacy_map(tomap, c))) {
442 for (c=128+totype; c<256; c++) {
443 if (d == legacy_map(tomap, c)) {
452 if (c < 0x10000 || type-UCS2BE < 2U) {
453 if (c >= 0x10000) c = 0xFFFD;
454 if (*outb < 2) goto toobig;
455 put_16((void *)*out, c, totype);
460 if (*outb < 4) goto toobig;
462 put_16((void *)*out, (c>>10)|0xd800, totype);
463 put_16((void *)(*out + 2), (c&0x3ff)|0xdc00, totype);
469 if (*outb < 4) goto toobig;
470 put_32((void *)*out, c, totype);