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\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 ksc[93][94] = {
77 static int fuzzycmp(const unsigned char *a, const unsigned char *b)
79 for (; *a && *b; a++, b++) {
80 while (*a && (*a|32U)-'a'>26 && *a-'0'>10U) a++;
81 if ((*a|32U) != *b) return 1;
86 static size_t find_charmap(const void *name)
88 const unsigned char *s;
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 size_t iconv(iconv_t cd0, char **restrict in, size_t *restrict inb, char **restrict out, size_t *restrict outb)
157 unsigned long cd = (unsigned long)cd0;
158 unsigned to = cd & 0xffff;
159 unsigned from = cd >> 16;
160 const unsigned char *map = charmaps+from+1;
161 const unsigned char *tomap = charmaps+to+1;
167 unsigned char type = map[-1];
168 unsigned char totype = tomap[-1];
170 if (!in || !*in || !*inb) return 0;
172 for (; *inb; *in+=l, *inb-=l) {
173 c = *(unsigned char *)*in;
176 if (c >= 128 || type-UTF_32BE < 7U) switch (type) {
178 l = mbrtowc_utf8(&wc, *in, *inb, &st);
180 else if (l == (size_t)-1) goto ilseq;
181 else if (l == (size_t)-2) goto starved;
188 if (*inb < l) goto starved;
194 if (*inb < 4) goto starved;
195 c = get_32((void *)*in, type);
197 if (c-0xd800u < 0x800u || c >= 0x110000u) goto ilseq;
204 if (*inb < 2) goto starved;
205 c = get_16((void *)*in, type);
206 if ((unsigned)(c-0xdc00) < 0x400) goto ilseq;
207 if ((unsigned)(c-0xd800) < 0x400) {
208 if (type-UCS2BE < 2U) goto ilseq;
210 if (*inb < 4) goto starved;
211 d = get_16((void *)(*in + 2), type);
212 if ((unsigned)(d-0xdc00) >= 0x400) goto ilseq;
213 c = ((c-0xd7c0)<<10) + (d-0xdc00);
217 if (c-0xa1 <= 0xdf-0xa1) {
222 if (*inb < 2) goto starved;
223 d = *((unsigned char *)*in + 1);
224 if (c-129 <= 159-129) c -= 129;
225 else if (c-224 <= 239-224) c -= 193;
228 if (d-64 <= 158-64) {
229 if (d==127) goto ilseq;
232 } else if (d-159 <= 252-159) {
241 if (*inb < 2) goto starved;
242 d = *((unsigned char *)*in + 1);
245 if (c-0xa1 > 0xdf-0xa1) goto ilseq;
251 if (c >= 84 || d >= 94) goto ilseq;
256 if (c < 0xa1) goto ilseq;
260 if (c >= 126) goto ilseq;
262 if (*inb < 2) goto starved;
263 d = *((unsigned char *)*in + 1);
264 if (d < 0xa1 && type == GB2312) goto ilseq;
265 if (d-0x40>=191 || d==127) {
266 if (d-'0'>9 || type != GB18030)
269 if (*inb < 4) goto starved;
270 c = (10*c + d-'0') * 1260;
271 d = *((unsigned char *)*in + 2);
272 if (d-0x81>126) goto ilseq;
274 d = *((unsigned char *)*in + 3);
275 if (d-'0'>9) goto ilseq;
280 for (int i=0; i<126; i++)
281 for (int j=0; j<190; j++)
282 if (gb18030[i][j]-d <= c-d)
295 if (*inb < 2) goto starved;
296 d = *((unsigned char *)*in + 1);
297 if (c-0xa1>=0xfa-0xa1) goto ilseq;
299 if (d-0x40>=0xff-0x40 || d-0x7f<0xa1-0x7f) goto ilseq;
301 if (d > 0x3e) d -= 0x22;
307 if (*inb < 2) goto starved;
308 d = *((unsigned char *)*in + 1);
311 if (c >= 93 || d >= 94) {
314 if (c >= 93 || c>=0xc6-0x81 && d>0x52)
316 if (d-'A'<26) d = d-'A';
317 else if (d-'a'<26) d = d-'a'+26;
318 else if (d-0x81<0xff-0x81) d = d-0x81+52;
320 if (c < 0x20) c = 178*c + d;
321 else c = 178*0x20 + 84*(c-0x20) + d;
323 for (d=0xac00; d<=c; ) {
325 for (int i=0; i<93; i++)
326 for (int j=0; j<94; j++)
327 if (ksc[i][j]-d <= c-d)
338 if (c < 128+type) break;
340 c = legacy_chars[ map[c*5/4]>>2*c%8 |
341 map[c*5/4+1]<<8-2*c%8 & 1023 ];
342 if (!c) c = *(unsigned char *)*in;
343 if (c==1) goto ilseq;
348 if (*outb < sizeof(wchar_t)) goto toobig;
349 *(wchar_t *)*out = c;
350 *out += sizeof(wchar_t);
351 *outb -= sizeof(wchar_t);
356 k = wctomb_utf8(tmp, c);
357 if (*outb < k) goto toobig;
358 memcpy(*out, tmp, k);
359 } else k = wctomb_utf8(*out, c);
364 if (c > 0x7f) subst: x++, c='*';
366 if (*outb < 1) goto toobig;
367 if (c < 128+totype) {
374 for (c=0; c<128-totype; c++) {
375 if (d == legacy_chars[ tomap[c*5/4]>>2*c%8 |
376 tomap[c*5/4+1]<<8-2*c%8 & 1023 ]) {
386 if (c < 0x10000 || type-UCS2BE < 2U) {
387 if (c >= 0x10000) c = 0xFFFD;
388 if (*outb < 2) goto toobig;
389 put_16((void *)*out, c, totype);
394 if (*outb < 4) goto toobig;
396 put_16((void *)*out, (c>>10)|0xd800, totype);
397 put_16((void *)(*out + 2), (c&0x3ff)|0xdc00, totype);
403 if (*outb < 4) goto toobig;
404 put_32((void *)*out, c, totype);