#define WCHAR_T 0306
#define US_ASCII 0307
#define UTF_8 0310
+#define UTF_16 0312
+#define UTF_32 0313
+#define UCS2 0314
#define EUC_JP 0320
#define SHIFT_JIS 0321
#define ISO2022_JP 0322
/* Definitions of charmaps. Each charmap consists of:
* 1. Empty-string-terminated list of null-terminated aliases.
- * 2. Special type code or number of elided entries.
- * 3. Character table (size determined by field 2). */
+ * 2. Special type code or number of elided quads of entries.
+ * 3. Character table (size determined by field 2), consisting
+ * of 5 bytes for every 4 characters, interpreted as 10-bit
+ * indices into the legacy_chars table. */
static const unsigned char charmaps[] =
"utf8\0char\0\0\310"
"wchart\0\0\306"
-"ucs2\0ucs2be\0\0\304"
+"ucs2be\0\0\304"
"ucs2le\0\0\305"
-"utf16\0utf16be\0\0\302"
+"utf16be\0\0\302"
"utf16le\0\0\301"
-"ucs4\0ucs4be\0utf32\0utf32be\0\0\300"
+"ucs4be\0utf32be\0\0\300"
"ucs4le\0utf32le\0\0\303"
"ascii\0usascii\0iso646\0iso646us\0\0\307"
+"utf16\0\0\312"
+"ucs4\0utf32\0\0\313"
+"ucs2\0\0\314"
"eucjp\0\0\320"
"shiftjis\0sjis\0\0\321"
"iso2022jp\0\0\322"
#include "codepages.h"
;
+/* Table of characters that appear in legacy 8-bit codepages,
+ * limited to 1024 slots (10 bit indices). The first 256 entries
+ * are elided since those characters are obviously all included. */
static const unsigned short legacy_chars[] = {
#include "legacychars.h"
};
#include "ksc.h"
};
+static const unsigned short rev_jis[] = {
+#include "revjis.h"
+};
+
static int fuzzycmp(const unsigned char *a, const unsigned char *b)
{
for (; *a && *b; a++, b++) {
s += strlen((void *)s)+1;
if (!*s) {
if (s[1] > 0200) s+=2;
- else s+=2+(128U-s[1])/4*5;
+ else s+=2+(64U-s[1])*5;
}
}
return -1;
if ((t = find_charmap(to))==-1
|| (f = find_charmap(from))==-1
- || (charmaps[t] >= 0320)) {
+ || (charmaps[t] >= 0330)) {
errno = EINVAL;
return (iconv_t)-1;
}
iconv_t cd = combine_to_from(t, f);
switch (charmaps[f]) {
+ case UTF_16:
+ case UTF_32:
+ case UCS2:
case ISO2022_JP:
scd = malloc(sizeof *scd);
if (!scd) return (iconv_t)-1;
static unsigned legacy_map(const unsigned char *map, unsigned c)
{
- unsigned x = c - 128 - map[-1];
- x = legacy_chars[ map[x*5/4]>>2*x%8 |
- map[x*5/4+1]<<8-2*x%8 & 1023 ];
- return x ? x : c;
+ if (c < 4*map[-1]) return c;
+ unsigned x = c - 4*map[-1];
+ x = map[x*5/4]>>2*x%8 | map[x*5/4+1]<<8-2*x%8 & 1023;
+ return x < 256 ? x : legacy_chars[x-256];
+}
+
+static unsigned uni_to_jis(unsigned c)
+{
+ unsigned nel = sizeof rev_jis / sizeof *rev_jis;
+ unsigned d, j, i, b = 0;
+ for (;;) {
+ i = nel/2;
+ j = rev_jis[b+i];
+ d = jis0208[j/256][j%256];
+ if (d==c) return j + 0x2121;
+ else if (nel == 1) return 0;
+ else if (c < d)
+ nel /= 2;
+ else {
+ b += i;
+ nel -= nel/2;
+ }
+ }
}
size_t iconv(iconv_t cd, char **restrict in, size_t *restrict inb, char **restrict out, size_t *restrict outb)
c = ((c-0xd7c0)<<10) + (d-0xdc00);
}
break;
+ case UCS2:
+ case UTF_16:
+ l = 0;
+ if (!scd->state) {
+ if (*inb < 2) goto starved;
+ c = get_16((void *)*in, 0);
+ scd->state = type==UCS2
+ ? c==0xfffe ? UCS2LE : UCS2BE
+ : c==0xfffe ? UTF_16LE : UTF_16BE;
+ if (c == 0xfffe || c == 0xfeff)
+ l = 2;
+ }
+ type = scd->state;
+ continue;
+ case UTF_32:
+ l = 0;
+ if (!scd->state) {
+ if (*inb < 4) goto starved;
+ c = get_32((void *)*in, 0);
+ scd->state = c==0xfffe0000 ? UTF_32LE : UTF_32BE;
+ if (c == 0xfffe0000 || c == 0xfeff)
+ l = 4;
+ }
+ type = scd->state;
+ continue;
case SHIFT_JIS:
if (c < 128) break;
if (c-0xa1 <= 0xdf-0xa1) {
* range in the hkscs table then hard-coded
* here. Ugly, yes. */
if (c/256 == 0xdc) {
- if (totype-0300U > 8) k = 2;
- else k = "\10\4\4\10\4\4\10\2\4"[totype-0300];
- if (k > *outb) goto toobig;
- x += iconv(combine_to_from(to, 0),
+ union {
+ char c[8];
+ wchar_t wc[2];
+ } tmp;
+ char *ptmp = tmp.c;
+ size_t tmpx = iconv(combine_to_from(to, find_charmap("utf8")),
&(char *){"\303\212\314\204"
"\303\212\314\214"
"\303\252\314\204"
"\303\252\314\214"
+c%256}, &(size_t){4},
- out, outb);
+ &ptmp, &(size_t){sizeof tmp});
+ size_t tmplen = ptmp - tmp.c;
+ if (tmplen > *outb) goto toobig;
+ if (tmpx) x++;
+ memcpy(*out, &tmp, tmplen);
+ *out += tmplen;
+ *outb -= tmplen;
continue;
}
if (!c) goto ilseq;
if (!c) goto ilseq;
break;
default:
- if (c < 128+type) break;
+ if (!c) break;
c = legacy_map(map, c);
- if (c==1) goto ilseq;
+ if (!c) goto ilseq;
}
switch (totype) {
if (c > 0x7f) subst: x++, c='*';
default:
if (*outb < 1) goto toobig;
- if (c < 128+totype || (c<256 && c==legacy_map(tomap, c))) {
+ if (c<256 && c==legacy_map(tomap, c)) {
revout:
+ if (*outb < 1) goto toobig;
*(*out)++ = c;
*outb -= 1;
break;
}
d = c;
- for (c=128+totype; c<256; c++) {
+ for (c=4*totype; c<256; c++) {
if (d == legacy_map(tomap, c)) {
goto revout;
}
}
goto subst;
+ case SHIFT_JIS:
+ if (c < 128) goto revout;
+ if (c == 0xa5) {
+ x++;
+ c = '\\';
+ goto revout;
+ }
+ if (c == 0x203e) {
+ x++;
+ c = '~';
+ goto revout;
+ }
+ if (c-0xff61 <= 0xdf-0xa1) {
+ c += 0xa1 - 0xff61;
+ goto revout;
+ }
+ c = uni_to_jis(c);
+ if (!c) goto subst;
+ if (*outb < 2) goto toobig;
+ d = c%256;
+ c = c/256;
+ *(*out)++ = (c+1)/2 + (c<95 ? 112 : 176);
+ *(*out)++ = c%2 ? d + 31 + d/96 : d + 126;
+ *outb -= 2;
+ break;
+ case EUC_JP:
+ if (c < 128) goto revout;
+ if (c-0xff61 <= 0xdf-0xa1) {
+ c += 0x0e00 + 0x21 - 0xff61;
+ } else {
+ c = uni_to_jis(c);
+ }
+ if (!c) goto subst;
+ if (*outb < 2) goto toobig;
+ *(*out)++ = c/256 + 0x80;
+ *(*out)++ = c%256 + 0x80;
+ *outb -= 2;
+ break;
+ case ISO2022_JP:
+ if (c < 128) goto revout;
+ if (c-0xff61 <= 0xdf-0xa1 || c==0xa5 || c==0x203e) {
+ if (*outb < 7) goto toobig;
+ *(*out)++ = '\033';
+ *(*out)++ = '(';
+ if (c==0xa5) {
+ *(*out)++ = 'J';
+ *(*out)++ = '\\';
+ } else if (c==0x203e) {
+ *(*out)++ = 'J';
+ *(*out)++ = '~';
+ } else {
+ *(*out)++ = 'I';
+ *(*out)++ = c-0xff61+0x21;
+ }
+ *(*out)++ = '\033';
+ *(*out)++ = '(';
+ *(*out)++ = 'B';
+ *outb -= 7;
+ break;
+ }
+ c = uni_to_jis(c);
+ if (!c) goto subst;
+ if (*outb < 8) goto toobig;
+ *(*out)++ = '\033';
+ *(*out)++ = '$';
+ *(*out)++ = 'B';
+ *(*out)++ = c/256;
+ *(*out)++ = c%256;
+ *(*out)++ = '\033';
+ *(*out)++ = '(';
+ *(*out)++ = 'B';
+ *outb -= 8;
+ break;
+ case UCS2:
+ totype = UCS2BE;
case UCS2BE:
case UCS2LE:
+ case UTF_16:
case UTF_16BE:
case UTF_16LE:
- if (c < 0x10000 || type-UCS2BE < 2U) {
+ if (c < 0x10000 || totype-UCS2BE < 2U) {
if (c >= 0x10000) c = 0xFFFD;
if (*outb < 2) goto toobig;
put_16((void *)*out, c, totype);
*out += 4;
*outb -= 4;
break;
+ case UTF_32:
+ totype = UTF_32BE;
case UTF_32BE:
case UTF_32LE:
if (*outb < 4) goto toobig;