nsz Git - musl/blob - src/locale/iconv.c

   1 #include <iconv.h>
   2 #include <errno.h>
   3 #include <wchar.h>
   4 #include <string.h>
   5 #include <stdlib.h>
   6 #include <limits.h>
   7 #include <stdint.h>
   8
   9 #define UTF_32BE    0300
  10 #define UTF_16LE    0301
  11 #define UTF_16BE    0302
  12 #define UTF_32LE    0303
  13 #define UCS2BE      0304
  14 #define UCS2LE      0305
  15 #define US_ASCII    0306
  16 #define WCHAR_T     0307
  17 #define UTF_8       0310
  18 #define EUC_JP      0320
  19 #define SHIFT_JIS   0321
  20
  21 /* FIXME: these are not implemented yet
  22  * EUC:   A1-FE A1-FE
  23  * GBK:   81-FE 40-7E,80-FE
  24  * Big5:  A1-FE 40-7E,A1-FE
  25  */
  26
  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). */
  31
  32 static const unsigned char charmaps[] =
  33 "utf8\0\0\310"
  34 "wchart\0\0\307"
  35 "ucs2\0ucs2be\0\0\304"
  36 "ucs2le\0\0\305"
  37 "utf16\0utf16be\0\0\302"
  38 "utf16le\0\0\301"
  39 "ucs4\0ucs4be\0utf32\0utf32be\0\0\300"
  40 "ucs4le\0utf32le\0\0\303"
  41 "ascii\0usascii\0iso646\0iso646us\0\0\306"
  42 "eucjp\0\0\320"
  43 "shiftjis\0sjis\0\0\321"
  44 #include "codepages.h"
  45 ;
  46
  47 static const unsigned short legacy_chars[] = {
  48 #include "legacychars.h"
  49 };
  50
  51 static const unsigned short jis0208[84][94] = {
  52 #include "jis0208.h"
  53 };
  54
  55 static int fuzzycmp(const unsigned char *a, const unsigned char *b)
  56 {
  57         for (; *a && *b; a++, b++) {
  58                 while (*a && (*a|32U)-'a'>26 && *a-'0'>10U) a++;
  59                 if ((*a|32U) != *b) return 1;
  60         }
  61         return *a != *b;
  62 }
  63
  64 static size_t find_charmap(const void *name)
  65 {
  66         const unsigned char *s;
  67         for (s=charmaps; *s; ) {
  68                 if (!fuzzycmp(name, s)) {
  69                         for (; *s; s+=strlen((void *)s)+1);
  70                         return s+1-charmaps;
  71                 }
  72                 s += strlen((void *)s)+1;
  73                 if (!*s) {
  74                         if (s[1] > 0200) s+=2;
  75                         else s+=2+(128U-s[1])/4*5;
  76                 }
  77         }
  78         return -1;
  79 }
  80
  81 iconv_t iconv_open(const char *to, const char *from)
  82 {
  83         size_t f, t;
  84
  85         if ((t = find_charmap(to))==-1 || (f = find_charmap(from))==-1) {
  86                 errno = EINVAL;
  87                 return (iconv_t)-1;
  88         }
  89
  90         return (void *)(f<<16 | t);
  91 }
  92
  93 int iconv_close(iconv_t cd)
  94 {
  95         return 0;
  96 }
  97
  98 static unsigned get_16(const unsigned char *s, int e)
  99 {
 100         e &= 1;
 101         return s[e]<<8 | s[1-e];
 102 }
 103
 104 static void put_16(unsigned char *s, unsigned c, int e)
 105 {
 106         e &= 1;
 107         s[e] = c>>8;
 108         s[1-e] = c;
 109 }
 110
 111 static unsigned get_32(const unsigned char *s, int e)
 112 {
 113         e &= 3;
 114         return s[e]+0U<<24 | s[e^1]<<16 | s[e^2]<<8 | s[e^3];
 115 }
 116
 117 static void put_32(unsigned char *s, unsigned c, int e)
 118 {
 119         e &= 3;
 120         s[e^0] = c>>24;
 121         s[e^1] = c>>16;
 122         s[e^2] = c>>8;
 123         s[e^3] = c;
 124 }
 125
 126 /* Adapt as needed */
 127 #define mbrtowc_utf8 mbrtowc
 128 #define wctomb_utf8 wctomb
 129
 130 #include <stdio.h>
 131 size_t iconv(iconv_t cd0, char **in, size_t *inb, char **out, size_t *outb)
 132 {
 133         size_t x=0;
 134         unsigned long cd = (unsigned long)cd0;
 135         unsigned to = cd & 0xffff;
 136         unsigned from = cd >> 16;
 137         const unsigned char *map = charmaps+from+1;
 138         const unsigned char *tomap = charmaps+to+1;
 139         mbstate_t st = {0};
 140         wchar_t wc;
 141         unsigned c, d;
 142         size_t k, l;
 143         int err;
 144         unsigned char type = map[-1];
 145         unsigned char totype = tomap[-1];
 146
 147         if (!in || !*in || !*inb) return 0;
 148
 149         for (; *inb; *in+=l, *inb-=l) {
 150                 c = *(unsigned char *)*in;
 151                 l = 1;
 152
 153                 if (c >= 128) switch (type) {
 154                 case UTF_8:
 155                         l = mbrtowc_utf8(&wc, *in, *inb, &st);
 156                         if (!l) l++;
 157                         else if (l == (size_t)-1) goto ilseq;
 158                         else if (l == (size_t)-2) goto starved;
 159                         c = wc;
 160                         break;
 161                 case US_ASCII:
 162                         goto ilseq;
 163                 case WCHAR_T:
 164                         l = sizeof(wchar_t);
 165                         if (*inb < l) goto starved;
 166                         c = *(wchar_t *)*in;
 167                         if (0) {
 168                 case UTF_32BE:
 169                 case UTF_32LE:
 170                         l = 4;
 171                         if (*inb < 4) goto starved;
 172                         c = get_32((void *)*in, type);
 173                         }
 174                         if (c-0xd800u < 0x800u || c >= 0x110000u) goto ilseq;
 175                         break;
 176                 case UCS2BE:
 177                 case UCS2LE:
 178                 case UTF_16BE:
 179                 case UTF_16LE:
 180                         l = 2;
 181                         if (*inb < 2) goto starved;
 182                         c = get_16((void *)*in, type);
 183                         if ((unsigned)(c-0xdc00) < 0x400) goto ilseq;
 184                         if ((unsigned)(c-0xd800) < 0x400) {
 185                                 if (type-UCS2BE < 2U) goto ilseq;
 186                                 l = 4;
 187                                 if (*inb < 4) goto starved;
 188                                 d = get_16((void *)(*in + 2), from);
 189                                 if ((unsigned)(c-0xdc00) >= 0x400) goto ilseq;
 190                                 c = ((c-0xd800)<<10) | (d-0xdc00);
 191                         }
 192                         break;
 193                 case SHIFT_JIS:
 194                         if (c-0xa1 <= 0xdf-0xa1) {
 195                                 c += 0xff61-0xa1;
 196                                 break;
 197                         }
 198                         l = 2;
 199                         if (*inb < 2) goto starved;
 200                         d = *((unsigned char *)*in + 1);
 201                         if (c-129 <= 159-129) c -= 129;
 202                         else if (c-224 <= 239-224) c -= 193;
 203                         else goto ilseq;
 204                         c *= 2;
 205                         if (d-64 <= 158-64) {
 206                                 if (d==127) goto ilseq;
 207                                 if (d>127) d--;
 208                                 d -= 64;
 209                         } else if (d-159 <= 252-159) {
 210                                 c++;
 211                                 d -= 159;
 212                         }
 213                         c = jis0208[c][d];
 214                         if (!c) goto ilseq;
 215                         break;
 216                 case EUC_JP:
 217                         l = 2;
 218                         if (*inb < 2) goto starved;
 219                         d = *((unsigned char *)*in + 1);
 220                         if (c==0x8e) {
 221                                 c = d;
 222                                 if (c-0xa1 > 0xdf-0xa1) goto ilseq;
 223                                 c += 0xff61 - 0xa1;
 224                                 break;
 225                         }
 226                         c -= 0xa1;
 227                         d -= 0xa1;
 228                         if (c >= 84 || d >= 94) goto ilseq;
 229                         c = jis0208[c][d];
 230                         if (!c) goto ilseq;
 231                         break;
 232                 default:
 233                         if (c < 128+type) break;
 234                         c -= 128+type;
 235                         c = legacy_chars[ map[c*5/4]>>2*c%8 |
 236                                 map[c*5/4+1]<<8-2*c%8 & 1023 ];
 237                         if (!c) c = *(unsigned char *)*in;
 238                         if (c==1) goto ilseq;
 239                 }
 240
 241                 switch (totype) {
 242                 case WCHAR_T:
 243                         if (*outb < sizeof(wchar_t)) goto toobig;
 244                         *(wchar_t *)*out = c;
 245                         *out += sizeof(wchar_t);
 246                         *outb -= sizeof(wchar_t);
 247                         break;
 248                 case UTF_8:
 249                         if (*outb < 4) {
 250                                 char tmp[4];
 251                                 k = wctomb_utf8(tmp, c);
 252                                 if (*outb < k) goto toobig;
 253                                 memcpy(*out, tmp, k);
 254                         } else k = wctomb_utf8(*out, c);
 255                         *out += k;
 256                         *outb -= k;
 257                         break;
 258                 case US_ASCII:
 259                         if (c > 0x7f) subst: x++, c='*';
 260                 default:
 261                         if (*outb < 1) goto toobig;
 262                         if (c < 128+totype) {
 263                         revout:
 264                                 *(*out)++ = c;
 265                                 *outb -= 1;
 266                                 break;
 267                         }
 268                         d = c;
 269                         for (c=0; c<128-totype; c++) {
 270                                 if (d == legacy_chars[ map[c*5/4]>>2*c%8 |
 271                                         map[c*5/4+1]<<8-2*c%8 & 1023 ]) {
 272                                         c += 128;
 273                                         goto revout;
 274                                 }
 275                         }
 276                         goto subst;
 277                 case UCS2BE:
 278                 case UCS2LE:
 279                 case UTF_16BE:
 280                 case UTF_16LE:
 281                         if (c < 0x10000) {
 282                                 if (*outb < 2) goto toobig;
 283                                 put_16((void *)*out, c, totype);
 284                                 *out += 2;
 285                                 *outb -= 2;
 286                                 break;
 287                         }
 288                         if (type-UCS2BE < 2U) goto ilseq;
 289                         if (*outb < 4) goto toobig;
 290                         put_16((void *)*out, (c>>10)|0xd800, totype);
 291                         put_16((void *)(*out + 2), (c&0x3ff)|0xdc00, totype);
 292                         *out += 4;
 293                         *outb -= 4;
 294                         break;
 295                 case UTF_32BE:
 296                 case UTF_32LE:
 297                         if (*outb < 4) goto toobig;
 298                         put_32((void *)*out, c, totype);
 299                         *out += 4;
 300                         *outb -= 4;
 301                         break;
 302                 }
 303         }
 304         return x;
 305 ilseq:
 306         err = EILSEQ;
 307         x = -1;
 308         goto end;
 309 toobig:
 310         err = E2BIG;
 311         x = -1;
 312         goto end;
 313 starved:
 314         err = EINVAL;
 315         x = -1;
 316 end:
 317         errno = err;
 318         return x;
 319 }