16 /* builtin charmaps */
19 /* only 0-7 are valid as dest charset */
29 /* additional charsets with algorithmic conversion */
34 /* some programs like php need this */
35 int _libiconv_version
= _LIBICONV_VERSION
;
37 /* these must match the constants above */
38 static const unsigned char charsets
[] =
41 "\000" "UTF-16BE" "\0"
42 "\001" "UTF-16LE" "\0"
43 "\002" "UTF-32BE" "\0"
44 "\003" "UTF-32LE" "\0"
46 "\006" "US-ASCII" "\0"
47 "\007" "ISO-8859-1" "\0"
49 "\010" "ISO-8859-15""\0"
51 "\011" "ISO-8859-11""\0"
53 "\012" "JIS-0201" "\0"
56 /* separate identifiers for sbcs/dbcs/etc map type */
65 /* FIXME: these are not implemented yet
67 // GBK: 81-FE 40-7E,80-FE
68 // Big5: A1-FE 40-7E,A1-FE
71 static const unsigned short maplen
[] = {
72 [UCS2_8BIT
] = 4+ 2* 128,
73 [UCS3_8BIT
] = 4+ 3* 128,
75 [SHIFT_JIS
] = 4+ 2* 94*94,
76 [BIG5
] = 4+ 2* 94*157,
77 [GBK
] = 4+ 2* 126*190,
78 [EUC_TW
] = 4+ 2* 2*94*94,
81 static int find_charmap(const char *name
)
84 for (i
= 0; i
< (sizeof(charmaps
) / sizeof(charmaps
[0])); i
++)
85 if (!strcasecmp(charmaps
[i
].name
, name
))
90 static int find_charset(const char *name
)
92 const unsigned char *s
;
93 for (s
=charsets
; *s
<0xff && strcasecmp(s
+1, name
); s
+=strlen(s
)+1);
97 iconv_t
iconv_open(const char *to
, const char *from
)
102 if ((t
= find_charset(to
)) > 8)
105 if ((f
= find_charset(from
)) < 255)
106 return 0 | (t
<<1) | (f
<<8);
108 if ((m
= find_charmap(from
)) > -1)
109 return 1 | (t
<<1) | (m
<<8);
114 int iconv_close(iconv_t cd
)
119 static inline wchar_t get_16(const unsigned char *s
, int endian
)
122 return s
[endian
]<<8 | s
[endian
^1];
125 static inline void put_16(unsigned char *s
, wchar_t c
, int endian
)
132 static inline int utf8enc_wchar(char *outb
, wchar_t c
)
138 else if (c
<= 0x7FF) {
139 *outb
++ = ((c
>> 6) & 0x1F) | 0xC0;
140 *outb
++ = ( c
& 0x3F) | 0x80;
143 else if (c
<= 0xFFFF) {
144 *outb
++ = ((c
>> 12) & 0x0F) | 0xE0;
145 *outb
++ = ((c
>> 6) & 0x3F) | 0x80;
146 *outb
++ = ( c
& 0x3F) | 0x80;
149 else if (c
<= 0x10FFFF) {
150 *outb
++ = ((c
>> 18) & 0x07) | 0xF0;
151 *outb
++ = ((c
>> 12) & 0x3F) | 0x80;
152 *outb
++ = ((c
>> 6) & 0x3F) | 0x80;
153 *outb
++ = ( c
& 0x3F) | 0x80;
162 static inline int utf8seq_is_overlong(char *s
, int n
)
167 /* 1100000x (10xxxxxx) */
168 return (((*s
>> 1) == 0x60) &&
169 ((*(s
+1) >> 6) == 0x02));
172 /* 11100000 100xxxxx (10xxxxxx) */
173 return ((*s
== 0xE0) &&
174 ((*(s
+1) >> 5) == 0x04) &&
175 ((*(s
+2) >> 6) == 0x02));
178 /* 11110000 1000xxxx (10xxxxxx 10xxxxxx) */
179 return ((*s
== 0xF0) &&
180 ((*(s
+1) >> 4) == 0x08) &&
181 ((*(s
+2) >> 6) == 0x02) &&
182 ((*(s
+3) >> 6) == 0x02));
188 static inline int utf8seq_is_surrogate(char *s
, int n
)
190 return ((n
== 3) && (*s
== 0xED) && (*(s
+1) >= 0xA0) && (*(s
+1) <= 0xBF));
193 static inline int utf8seq_is_illegal(char *s
, int n
)
195 return ((n
== 3) && (*s
== 0xEF) && (*(s
+1) == 0xBF) &&
196 (*(s
+2) >= 0xBE) && (*(s
+2) <= 0xBF));
199 static inline int utf8dec_wchar(wchar_t *c
, unsigned char *in
, size_t inb
)
210 /* find utf8 sequence length */
211 if ((*in
& 0xE0) == 0xC0) n
= 2;
212 else if ((*in
& 0xF0) == 0xE0) n
= 3;
213 else if ((*in
& 0xF8) == 0xF0) n
= 4;
214 else if ((*in
& 0xFC) == 0xF8) n
= 5;
215 else if ((*in
& 0xFE) == 0xFC) n
= 6;
222 if (n
> 1 && n
< 5) {
223 /* reject invalid sequences */
224 if (utf8seq_is_overlong(in
, n
) ||
225 utf8seq_is_surrogate(in
, n
) ||
226 utf8seq_is_illegal(in
, n
))
230 *c
= (char)(*in
++ & (0x7F >> n
));
232 for (i
= 1; i
< n
; i
++) {
233 /* illegal continuation byte */
234 if (*in
< 0x80 || *in
> 0xBF)
237 *c
= (*c
<< 6) | (*in
++ & 0x3F);
243 /* unmapped sequence (> 4) */
247 static inline char latin9_translit(wchar_t c
)
249 /* a number of trivial iso-8859-15 <> utf-8 transliterations */
251 case 0x20AC: return 0xA4; /* Euro */
252 case 0x0160: return 0xA6; /* S caron */
253 case 0x0161: return 0xA8; /* s caron */
254 case 0x017D: return 0xB4; /* Z caron */
255 case 0x017E: return 0xB8; /* z caron */
256 case 0x0152: return 0xBC; /* OE */
257 case 0x0153: return 0xBD; /* oe */
258 case 0x0178: return 0xBE; /* Y diaeresis */
263 size_t iconv(iconv_t cd
, char **in
, size_t *inb
, char **out
, size_t *outb
)
266 unsigned char to
= (cd
>>1)&127;
267 unsigned char from
= 255;
268 const unsigned char *map
= 0;
269 char tmp
[MB_LEN_MAX
];
274 if (!in
|| !*in
|| !*inb
) return 0;
277 map
= charmaps
[cd
>>8].map
;
281 for (; *inb
; *in
+=l
, *inb
-=l
) {
282 c
= *(unsigned char *)*in
;
284 if (from
>= UTF_8
&& c
< 0x80) goto charok
;
288 if (*inb
< l
) goto starved
;
292 l
= utf8dec_wchar(&c
, *in
, *inb
);
294 else if (l
== (size_t)-1) goto ilseq
;
295 else if (l
== (size_t)-2) goto starved
;
300 if ((unsigned)c
- 0xa4 <= 0xbe - 0xa4) {
301 static const unsigned char map
[] = {
302 0, 0x60, 0, 0x61, 0, 0, 0, 0, 0, 0, 0,
303 0, 0, 0, 0, 0x7d, 0, 0, 0, 0x7e, 0, 0, 0,
306 if (c
== 0xa4) c
= 0x20ac;
307 else if (map
[c
-0xa5]) c
= 0x100 | map
[c
-0xa5];
312 if (c
>= 0xa1) c
+= 0x0e01-0xa1;
316 if (c
<= 0xdf) c
+= 0xff61-0xa1;
323 if (*inb
< 2) goto starved
;
324 c
= get_16(*in
, from
);
325 if ((unsigned)(c
-0xdc00) < 0x400) goto ilseq
;
326 if ((unsigned)(c
-0xd800) < 0x400) {
328 if (*inb
< 4) goto starved
;
329 d
= get_16(*in
+ 2, from
);
330 if ((unsigned)(c
-0xdc00) >= 0x400) goto ilseq
;
331 c
= ((c
-0xd800)<<10) | (d
-0xdc00);
337 if (*inb
< 4) goto starved
;
339 // c = get_32(*in, from);
342 /* only support ascii supersets */
349 if ((unsigned)c
- 0xa1 >= 94) goto ilseq
;
350 if ((unsigned)in
[0][1] - 0xa1 >= 94) goto ilseq
;
351 c
= (c
-0xa1)*94 + (in
[0][1]-0xa1);
355 if ((unsigned)c
- 0xa1 <= 0xdf-0xa1) {
365 c
= get_16(map
+ 4 + 2*c
, 0);
366 if (c
== 0xffff) goto ilseq
;
370 if ((unsigned)c
- 0xd800 < 0x800 || (unsigned)c
>= 0x110000)
375 if (*outb
< sizeof(wchar_t)) goto toobig
;
376 *(wchar_t *)*out
= c
;
377 *out
+= sizeof(wchar_t);
378 *outb
-= sizeof(wchar_t);
382 k
= utf8enc_wchar(tmp
, c
);
383 if (*outb
< k
) goto toobig
;
384 memcpy(*out
, tmp
, k
);
385 } else k
= utf8enc_wchar(*out
, c
);
390 if (c
> 0x7f) c
= 0xfffd;
391 /* fall thru and count replacement in latin1 case */
393 if (c
>= 0x100 && c
!= 0xfffd)
394 c
= latin9_translit(c
);
397 if (!*outb
) goto toobig
;
398 if (c
< 0x100) **out
= c
;
399 else x
++, **out
= '*'; //FIXME: translit?
406 if (*outb
< 2) goto toobig
;
412 if (*outb
< 4) goto toobig
;
413 put_16(*out
, (c
>>10)|0xd800, to
);
414 put_16(*out
+ 2, (c
&0x3ff)|0xdc00, to
);