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
  19 /* FIXME: these are not implemented yet
  20  * EUC:   A1-FE A1-FE
  21  * GBK:   81-FE 40-7E,80-FE
  22  * Big5:  A1-FE 40-7E,A1-FE
  23  */
  24
  25 /* Definitions of charmaps. Each charmap consists of:
  26  * 1. Empty-string-terminated list of null-terminated aliases.
  27  * 2. Special type code or number of elided entries.
  28  * 3. Character table (size determined by field 2). */
  29
  30 static const unsigned char charmaps[] =
  31 "utf8\0\0\310"
  32 "wchart\0\0\307"
  33 "ucs2\0ucs2be\0\0\304"
  34 "ucs2le\0\0\305"
  35 "utf16\0utf16be\0\0\302"
  36 "utf16le\0\0\301"
  37 "ucs4\0ucs4be\0utf32\0utf32be\0\0\300"
  38 "ucs4le\0utf32le\0\0\303"
  39 "ascii\0usascii\0iso646\0iso646us\0\0\306"
  40 #include "codepages.h"
  41 ;
  42
  43 static const unsigned short legacy_chars[] = {
  44 #include "legacychars.h"
  45 };
  46
  47 static int fuzzycmp(const unsigned char *a, const unsigned char *b)
  48 {
  49         for (; *a && *b; a++, b++) {
  50                 while (*a && (*a|32U)-'a'>26 && *a-'0'>10U) a++;
  51                 if ((*a|32U) != *b) return 1;
  52         }
  53         return *a != *b;
  54 }
  55
  56 static size_t find_charmap(const void *name)
  57 {
  58         const unsigned char *s;
  59         for (s=charmaps; *s; ) {
  60                 if (!fuzzycmp(name, s)) {
  61                         for (; *s; s+=strlen((void *)s)+1);
  62                         return s+1-charmaps;
  63                 }
  64                 s += strlen((void *)s)+1;
  65                 if (!*s) {
  66                         if (s[1] > 0200) s+=2;
  67                         else s+=2+(128U-s[1])/4*5;
  68                 }
  69         }
  70         return -1;
  71 }
  72
  73 iconv_t iconv_open(const char *to, const char *from)
  74 {
  75         size_t f, t;
  76
  77         if ((t = find_charmap(to))==-1 || (f = find_charmap(from))==-1) {
  78                 errno = EINVAL;
  79                 return (iconv_t)-1;
  80         }
  81
  82         return (void *)(f<<16 | t);
  83 }
  84
  85 int iconv_close(iconv_t cd)
  86 {
  87         return 0;
  88 }
  89
  90 static unsigned get_16(const unsigned char *s, int e)
  91 {
  92         e &= 1;
  93         return s[e]<<8 | s[1-e];
  94 }
  95
  96 static void put_16(unsigned char *s, unsigned c, int e)
  97 {
  98         e &= 1;
  99         s[e] = c>>8;
 100         s[1-e] = c;
 101 }
 102
 103 static unsigned get_32(const unsigned char *s, int e)
 104 {
 105         e &= 3;
 106         return s[e]+0U<<24 | s[e^1]<<16 | s[e^2]<<8 | s[e^3];
 107 }
 108
 109 static void put_32(unsigned char *s, unsigned c, int e)
 110 {
 111         e &= 3;
 112         s[e^0] = c>>24;
 113         s[e^1] = c>>16;
 114         s[e^2] = c>>8;
 115         s[e^3] = c;
 116 }
 117
 118 /* Adapt as needed */
 119 #define mbrtowc_utf8 mbrtowc
 120 #define wctomb_utf8 wctomb
 121
 122 #include <stdio.h>
 123 size_t iconv(iconv_t cd0, char **in, size_t *inb, char **out, size_t *outb)
 124 {
 125         size_t x=0;
 126         unsigned long cd = (unsigned long)cd0;
 127         unsigned to = cd & 0xffff;
 128         unsigned from = cd >> 16;
 129         const unsigned char *map = charmaps+from+1;
 130         const unsigned char *tomap = charmaps+to+1;
 131         mbstate_t st = {0};
 132         wchar_t wc;
 133         unsigned c, d;
 134         size_t k, l;
 135         int err;
 136         unsigned char type = map[-1];
 137         unsigned char totype = tomap[-1];
 138
 139         if (!in || !*in || !*inb) return 0;
 140
 141         for (; *inb; *in+=l, *inb-=l) {
 142                 c = *(unsigned char *)*in;
 143                 l = 1;
 144
 145                 if (c >= 128) switch (type) {
 146                 case UTF_8:
 147                         l = mbrtowc_utf8(&wc, *in, *inb, &st);
 148                         if (!l) l++;
 149                         else if (l == (size_t)-1) goto ilseq;
 150                         else if (l == (size_t)-2) goto starved;
 151                         c = wc;
 152                         break;
 153                 case US_ASCII:
 154                         goto ilseq;
 155                 case WCHAR_T:
 156                         l = sizeof(wchar_t);
 157                         if (*inb < l) goto starved;
 158                         c = *(wchar_t *)*in;
 159                         if (0) {
 160                 case UTF_32BE:
 161                 case UTF_32LE:
 162                         l = 4;
 163                         if (*inb < 4) goto starved;
 164                         c = get_32((void *)*in, type);
 165                         }
 166                         if (c-0xd800u < 0x800u || c >= 0x110000u) goto ilseq;
 167                         break;
 168                 case UCS2BE:
 169                 case UCS2LE:
 170                 case UTF_16BE:
 171                 case UTF_16LE:
 172                         l = 2;
 173                         if (*inb < 2) goto starved;
 174                         c = get_16((void *)*in, type);
 175                         if ((unsigned)(c-0xdc00) < 0x400) goto ilseq;
 176                         if ((unsigned)(c-0xd800) < 0x400) {
 177                                 if (type-UCS2BE < 2U) goto ilseq;
 178                                 l = 4;
 179                                 if (*inb < 4) goto starved;
 180                                 d = get_16((void *)(*in + 2), from);
 181                                 if ((unsigned)(c-0xdc00) >= 0x400) goto ilseq;
 182                                 c = ((c-0xd800)<<10) | (d-0xdc00);
 183                         }
 184                         break;
 185                 default:
 186                         if (c < 128+type) break;
 187                         c -= 128+type;
 188                         c = legacy_chars[ map[c*5/4]>>2*c%8 |
 189                                 map[c*5/4+1]<<8-2*c%8 & 1023 ];
 190                         if (!c) c = *(unsigned char *)*in;
 191                         if (c==1) goto ilseq;
 192                 }
 193
 194                 switch (totype) {
 195                 case WCHAR_T:
 196                         if (*outb < sizeof(wchar_t)) goto toobig;
 197                         *(wchar_t *)*out = c;
 198                         *out += sizeof(wchar_t);
 199                         *outb -= sizeof(wchar_t);
 200                         break;
 201                 case UTF_8:
 202                         if (*outb < 4) {
 203                                 char tmp[4];
 204                                 k = wctomb_utf8(tmp, c);
 205                                 if (*outb < k) goto toobig;
 206                                 memcpy(*out, tmp, k);
 207                         } else k = wctomb_utf8(*out, c);
 208                         *out += k;
 209                         *outb -= k;
 210                         break;
 211                 case US_ASCII:
 212                         if (c > 0x7f) subst: x++, c='*';
 213                 default:
 214                         if (*outb < 1) goto toobig;
 215                         if (c < 128+totype) {
 216                         revout:
 217                                 *(*out)++ = c;
 218                                 *outb -= 1;
 219                                 break;
 220                         }
 221                         d = c;
 222                         for (c=0; c<128-totype; c++) {
 223                                 if (d == legacy_chars[ map[c*5/4]>>2*c%8 |
 224                                         map[c*5/4+1]<<8-2*c%8 & 1023 ]) {
 225                                         c += 128;
 226                                         goto revout;
 227                                 }
 228                         }
 229                         goto subst;
 230                 case UCS2BE:
 231                 case UCS2LE:
 232                 case UTF_16BE:
 233                 case UTF_16LE:
 234                         if (c < 0x10000) {
 235                                 if (*outb < 2) goto toobig;
 236                                 put_16((void *)*out, c, totype);
 237                                 *out += 2;
 238                                 *outb -= 2;
 239                                 break;
 240                         }
 241                         if (type-UCS2BE < 2U) goto ilseq;
 242                         if (*outb < 4) goto toobig;
 243                         put_16((void *)*out, (c>>10)|0xd800, totype);
 244                         put_16((void *)(*out + 2), (c&0x3ff)|0xdc00, totype);
 245                         *out += 4;
 246                         *outb -= 4;
 247                         break;
 248                 case UTF_32BE:
 249                 case UTF_32LE:
 250                         if (*outb < 4) goto toobig;
 251                         put_32((void *)*out, c, totype);
 252                         *out += 4;
 253                         *outb -= 4;
 254                         break;
 255                 }
 256         }
 257         return x;
 258 ilseq:
 259         err = EILSEQ;
 260         x = -1;
 261         goto end;
 262 toobig:
 263         err = E2BIG;
 264         x = -1;
 265         goto end;
 266 starved:
 267         err = EINVAL;
 268         x = -1;
 269 end:
 270         errno = err;
 271         return x;
 272 }