19 #define SHIFT_JIS 0321
26 /* FIXME: these are not implemented yet
28 * GBK: 81-FE 40-7E,80-FE
29 * Big5: A1-FE 40-7E,A1-FE
32 /* Definitions of charmaps. Each charmap consists of:
33 * 1. Empty-string-terminated list of null-terminated aliases.
34 * 2. Special type code or number of elided entries.
35 * 3. Character table (size determined by field 2). */
37 static const unsigned char charmaps[] =
40 "ucs2\0ucs2be\0\0\304"
42 "utf16\0utf16be\0\0\302"
44 "ucs4\0ucs4be\0utf32\0utf32be\0\0\300"
45 "ucs4le\0utf32le\0\0\303"
46 "ascii\0usascii\0iso646\0iso646us\0\0\307"
48 "shiftjis\0sjis\0\0\321"
52 "big5\0bigfive\0cp950\0big5hkscs\0\0\340"
53 "euckr\0ksc5601\0ksx1001\0cp949\0\0\350"
54 #include "codepages.h"
57 static const unsigned short legacy_chars[] = {
58 #include "legacychars.h"
61 static const unsigned short jis0208[84][94] = {
65 static const unsigned short gb18030[126][190] = {
69 static const unsigned short big5[89][157] = {
73 static const unsigned short hkscs[] = {
77 static const unsigned short ksc[93][94] = {
81 static int fuzzycmp(const unsigned char *a, const unsigned char *b)
83 for (; *a && *b; a++, b++) {
84 while (*a && (*a|32U)-'a'>26 && *a-'0'>10U) a++;
85 if ((*a|32U) != *b) return 1;
90 static size_t find_charmap(const void *name)
92 const unsigned char *s;
93 for (s=charmaps; *s; ) {
94 if (!fuzzycmp(name, s)) {
95 for (; *s; s+=strlen((void *)s)+1);
98 s += strlen((void *)s)+1;
100 if (s[1] > 0200) s+=2;
101 else s+=2+(128U-s[1])/4*5;
107 iconv_t iconv_open(const char *to, const char *from)
111 if ((t = find_charmap(to))==-1
112 || (f = find_charmap(from))==-1
113 || (charmaps[t] >= 0320)) {
118 return (void *)(f<<16 | t);
121 int iconv_close(iconv_t cd)
126 static unsigned get_16(const unsigned char *s, int e)
129 return s[e]<<8 | s[1-e];
132 static void put_16(unsigned char *s, unsigned c, int e)
139 static unsigned get_32(const unsigned char *s, int e)
142 return s[e]+0U<<24 | s[e^1]<<16 | s[e^2]<<8 | s[e^3];
145 static void put_32(unsigned char *s, unsigned c, int e)
154 /* Adapt as needed */
155 #define mbrtowc_utf8 mbrtowc
156 #define wctomb_utf8 wctomb
158 size_t iconv(iconv_t cd0, char **restrict in, size_t *restrict inb, char **restrict out, size_t *restrict outb)
161 unsigned long cd = (unsigned long)cd0;
162 unsigned to = cd & 0xffff;
163 unsigned from = cd >> 16;
164 const unsigned char *map = charmaps+from+1;
165 const unsigned char *tomap = charmaps+to+1;
171 unsigned char type = map[-1];
172 unsigned char totype = tomap[-1];
174 if (!in || !*in || !*inb) return 0;
176 for (; *inb; *in+=l, *inb-=l) {
177 c = *(unsigned char *)*in;
180 if (c >= 128 || type-UTF_32BE < 7U) switch (type) {
182 l = mbrtowc_utf8(&wc, *in, *inb, &st);
184 else if (l == (size_t)-1) goto ilseq;
185 else if (l == (size_t)-2) goto starved;
192 if (*inb < l) goto starved;
198 if (*inb < 4) goto starved;
199 c = get_32((void *)*in, type);
201 if (c-0xd800u < 0x800u || c >= 0x110000u) goto ilseq;
208 if (*inb < 2) goto starved;
209 c = get_16((void *)*in, type);
210 if ((unsigned)(c-0xdc00) < 0x400) goto ilseq;
211 if ((unsigned)(c-0xd800) < 0x400) {
212 if (type-UCS2BE < 2U) goto ilseq;
214 if (*inb < 4) goto starved;
215 d = get_16((void *)(*in + 2), type);
216 if ((unsigned)(d-0xdc00) >= 0x400) goto ilseq;
217 c = ((c-0xd7c0)<<10) + (d-0xdc00);
221 if (c-0xa1 <= 0xdf-0xa1) {
226 if (*inb < 2) goto starved;
227 d = *((unsigned char *)*in + 1);
228 if (c-129 <= 159-129) c -= 129;
229 else if (c-224 <= 239-224) c -= 193;
232 if (d-64 <= 158-64) {
233 if (d==127) goto ilseq;
236 } else if (d-159 <= 252-159) {
245 if (*inb < 2) goto starved;
246 d = *((unsigned char *)*in + 1);
249 if (c-0xa1 > 0xdf-0xa1) goto ilseq;
255 if (c >= 84 || d >= 94) goto ilseq;
260 if (c < 0xa1) goto ilseq;
264 if (c >= 126) goto ilseq;
266 if (*inb < 2) goto starved;
267 d = *((unsigned char *)*in + 1);
268 if (d < 0xa1 && type == GB2312) goto ilseq;
269 if (d-0x40>=191 || d==127) {
270 if (d-'0'>9 || type != GB18030)
273 if (*inb < 4) goto starved;
274 c = (10*c + d-'0') * 1260;
275 d = *((unsigned char *)*in + 2);
276 if (d-0x81>126) goto ilseq;
278 d = *((unsigned char *)*in + 3);
279 if (d-'0'>9) goto ilseq;
284 for (int i=0; i<126; i++)
285 for (int j=0; j<190; j++)
286 if (gb18030[i][j]-d <= c-d)
299 if (*inb < 2) goto starved;
300 d = *((unsigned char *)*in + 1);
301 if (d-0x40>=0xff-0x40 || d-0x7f<0xa1-0x7f) goto ilseq;
303 if (d > 0x3e) d -= 0x22;
304 if (c-0xa1>=0xfa-0xa1) {
305 if (c-0x87>=0xff-0x87) goto ilseq;
306 if (c < 0xa1) c -= 0x87;
307 else c -= 0x87 + (0xfa-0xa1);
308 c = (hkscs[4867+(c*157+d)/16]>>(c*157+d)%16)%2<<17
310 /* A few HKSCS characters map to pairs of UCS
311 * characters. These are mapped to surrogate
312 * range in the hkscs table then hard-coded
313 * here. Ugly, yes. */
315 if (totype-0300U > 8) k = 2;
316 else k = "\10\4\4\10\4\4\10\2\4"[totype-0300];
317 if (k > *outb) goto toobig;
318 x += iconv((iconv_t)(uintptr_t)to,
319 &(char *){"\303\212\314\204"
323 +c%256}, &(size_t){4},
331 c = big5[c][d]|(c==0x27&&(d==0x3a||d==0x3c||d==0x42))<<17;
336 if (*inb < 2) goto starved;
337 d = *((unsigned char *)*in + 1);
340 if (c >= 93 || d >= 94) {
343 if (c >= 93 || c>=0xc6-0x81 && d>0x52)
345 if (d-'A'<26) d = d-'A';
346 else if (d-'a'<26) d = d-'a'+26;
347 else if (d-0x81<0xff-0x81) d = d-0x81+52;
349 if (c < 0x20) c = 178*c + d;
350 else c = 178*0x20 + 84*(c-0x20) + d;
352 for (d=0xac00; d<=c; ) {
354 for (int i=0; i<93; i++)
355 for (int j=0; j<94; j++)
356 if (ksc[i][j]-d <= c-d)
367 if (c < 128+type) break;
369 c = legacy_chars[ map[c*5/4]>>2*c%8 |
370 map[c*5/4+1]<<8-2*c%8 & 1023 ];
371 if (!c) c = *(unsigned char *)*in;
372 if (c==1) goto ilseq;
377 if (*outb < sizeof(wchar_t)) goto toobig;
378 *(wchar_t *)*out = c;
379 *out += sizeof(wchar_t);
380 *outb -= sizeof(wchar_t);
385 k = wctomb_utf8(tmp, c);
386 if (*outb < k) goto toobig;
387 memcpy(*out, tmp, k);
388 } else k = wctomb_utf8(*out, c);
393 if (c > 0x7f) subst: x++, c='*';
395 if (*outb < 1) goto toobig;
396 if (c < 128+totype) {
403 for (c=0; c<128-totype; c++) {
404 if (d == legacy_chars[ tomap[c*5/4]>>2*c%8 |
405 tomap[c*5/4+1]<<8-2*c%8 & 1023 ]) {
415 if (c < 0x10000 || type-UCS2BE < 2U) {
416 if (c >= 0x10000) c = 0xFFFD;
417 if (*outb < 2) goto toobig;
418 put_16((void *)*out, c, totype);
423 if (*outb < 4) goto toobig;
425 put_16((void *)*out, (c>>10)|0xd800, totype);
426 put_16((void *)(*out + 2), (c&0x3ff)|0xdc00, totype);
432 if (*outb < 4) goto toobig;
433 put_32((void *)*out, c, totype);