libiconv: add all ASCII aliases
[openwrt/svn-archive/archive.git] / package / libs / libiconv / src / iconv.c
1 #include <iconv.h>
2 #include <errno.h>
3 #include <wchar.h>
4 #include <string.h>
5 #include <strings.h>
6 #include <stdlib.h>
7 #include <limits.h>
8
9 #include <dirent.h>
10 #include <fcntl.h>
11 #include <sys/mman.h>
12 #include <sys/stat.h>
13 #include <unistd.h>
14 #include <stdint.h>
15
16 /* builtin charmaps */
17 #include "charmaps.h"
18
19 /* only 0-7 are valid as dest charset */
20 #define UTF_16BE 000
21 #define UTF_16LE 001
22 #define UTF_32BE 002
23 #define UTF_32LE 003
24 #define WCHAR_T 004
25 #define UTF_8 005
26 #define US_ASCII 006
27 #define LATIN_1 007
28
29 /* additional charsets with algorithmic conversion */
30 #define LATIN_9 010
31 #define TIS_620 011
32 #define JIS_0201 012
33
34 /* some programs like php need this */
35 int _libiconv_version = _LIBICONV_VERSION;
36
37 /* these must match the constants above */
38 static const unsigned char charsets[] =
39 "\005" "UTF-8" "\0"
40 "\004" "WCHAR_T" "\0"
41 "\000" "UTF-16BE" "\0"
42 "\001" "UTF-16LE" "\0"
43 "\002" "UTF-32BE" "\0"
44 "\003" "UTF-32LE" "\0"
45 "\006" "ASCII" "\0"
46 "\006" "US-ASCII" "\0"
47 "\006" "ISO646-US" "\0"
48 "\006" "ISO_646.IRV:1991" "\0"
49 "\006" "ISO-IR-6" "\0"
50 "\006" "ANSI_X3.4-1968" "\0"
51 "\006" "ANSI_X3.4-1986" "\0"
52 "\006" "CP367" "\0"
53 "\006" "IBM367" "\0"
54 "\006" "US" "\0"
55 "\006" "CSASCII" "\0"
56 "\007" "ISO-8859-1" "\0"
57 "\007" "LATIN1" "\0"
58 "\010" "ISO-8859-15""\0"
59 "\010" "LATIN9" "\0"
60 "\011" "ISO-8859-11""\0"
61 "\011" "TIS-620" "\0"
62 "\012" "JIS-0201" "\0"
63 "\377";
64
65 /* separate identifiers for sbcs/dbcs/etc map type */
66 #define UCS2_8BIT 000
67 #define UCS3_8BIT 001
68 #define EUC 002
69 #define EUC_TW 003
70 #define SHIFT_JIS 004
71 #define BIG5 005
72 #define GBK 006
73
74 /* FIXME: these are not implemented yet
75 // EUC: A1-FE A1-FE
76 // GBK: 81-FE 40-7E,80-FE
77 // Big5: A1-FE 40-7E,A1-FE
78 */
79
80 static const unsigned short maplen[] = {
81 [UCS2_8BIT] = 4+ 2* 128,
82 [UCS3_8BIT] = 4+ 3* 128,
83 [EUC] = 4+ 2* 94*94,
84 [SHIFT_JIS] = 4+ 2* 94*94,
85 [BIG5] = 4+ 2* 94*157,
86 [GBK] = 4+ 2* 126*190,
87 [EUC_TW] = 4+ 2* 2*94*94,
88 };
89
90 static int find_charmap(const char *name)
91 {
92 int i;
93 for (i = 0; i < (sizeof(charmaps) / sizeof(charmaps[0])); i++)
94 if (!strcasecmp(charmaps[i].name, name))
95 return i;
96 return -1;
97 }
98
99 static int find_charset(const char *name)
100 {
101 const unsigned char *s;
102 for (s=charsets; *s<0xff && strcasecmp(s+1, name); s+=strlen(s)+1);
103 return *s;
104 }
105
106 iconv_t iconv_open(const char *to, const char *from)
107 {
108 unsigned f, t;
109 int m;
110
111 if ((t = find_charset(to)) > 8)
112 return -1;
113
114 if ((f = find_charset(from)) < 255)
115 return 0 | (t<<1) | (f<<8);
116
117 if ((m = find_charmap(from)) > -1)
118 return 1 | (t<<1) | (m<<8);
119
120 return -1;
121 }
122
123 int iconv_close(iconv_t cd)
124 {
125 return 0;
126 }
127
128 static inline wchar_t get_16(const unsigned char *s, int endian)
129 {
130 endian &= 1;
131 return s[endian]<<8 | s[endian^1];
132 }
133
134 static inline void put_16(unsigned char *s, wchar_t c, int endian)
135 {
136 endian &= 1;
137 s[endian] = c>>8;
138 s[endian^1] = c;
139 }
140
141 static inline int utf8enc_wchar(char *outb, wchar_t c)
142 {
143 if (c <= 0x7F) {
144 *outb = c;
145 return 1;
146 }
147 else if (c <= 0x7FF) {
148 *outb++ = ((c >> 6) & 0x1F) | 0xC0;
149 *outb++ = ( c & 0x3F) | 0x80;
150 return 2;
151 }
152 else if (c <= 0xFFFF) {
153 *outb++ = ((c >> 12) & 0x0F) | 0xE0;
154 *outb++ = ((c >> 6) & 0x3F) | 0x80;
155 *outb++ = ( c & 0x3F) | 0x80;
156 return 3;
157 }
158 else if (c <= 0x10FFFF) {
159 *outb++ = ((c >> 18) & 0x07) | 0xF0;
160 *outb++ = ((c >> 12) & 0x3F) | 0x80;
161 *outb++ = ((c >> 6) & 0x3F) | 0x80;
162 *outb++ = ( c & 0x3F) | 0x80;
163 return 4;
164 }
165 else {
166 *outb++ = '?';
167 return 1;
168 }
169 }
170
171 static inline int utf8seq_is_overlong(char *s, int n)
172 {
173 switch (n)
174 {
175 case 2:
176 /* 1100000x (10xxxxxx) */
177 return (((*s >> 1) == 0x60) &&
178 ((*(s+1) >> 6) == 0x02));
179
180 case 3:
181 /* 11100000 100xxxxx (10xxxxxx) */
182 return ((*s == 0xE0) &&
183 ((*(s+1) >> 5) == 0x04) &&
184 ((*(s+2) >> 6) == 0x02));
185
186 case 4:
187 /* 11110000 1000xxxx (10xxxxxx 10xxxxxx) */
188 return ((*s == 0xF0) &&
189 ((*(s+1) >> 4) == 0x08) &&
190 ((*(s+2) >> 6) == 0x02) &&
191 ((*(s+3) >> 6) == 0x02));
192 }
193
194 return 0;
195 }
196
197 static inline int utf8seq_is_surrogate(char *s, int n)
198 {
199 return ((n == 3) && (*s == 0xED) && (*(s+1) >= 0xA0) && (*(s+1) <= 0xBF));
200 }
201
202 static inline int utf8seq_is_illegal(char *s, int n)
203 {
204 return ((n == 3) && (*s == 0xEF) && (*(s+1) == 0xBF) &&
205 (*(s+2) >= 0xBE) && (*(s+2) <= 0xBF));
206 }
207
208 static inline int utf8dec_wchar(wchar_t *c, unsigned char *in, size_t inb)
209 {
210 int i;
211 int n = -1;
212
213 /* trivial char */
214 if (*in <= 0x7F) {
215 *c = *in;
216 return 1;
217 }
218
219 /* find utf8 sequence length */
220 if ((*in & 0xE0) == 0xC0) n = 2;
221 else if ((*in & 0xF0) == 0xE0) n = 3;
222 else if ((*in & 0xF8) == 0xF0) n = 4;
223 else if ((*in & 0xFC) == 0xF8) n = 5;
224 else if ((*in & 0xFE) == 0xFC) n = 6;
225
226 /* starved? */
227 if (n > inb)
228 return -2;
229
230 /* decode ... */
231 if (n > 1 && n < 5) {
232 /* reject invalid sequences */
233 if (utf8seq_is_overlong(in, n) ||
234 utf8seq_is_surrogate(in, n) ||
235 utf8seq_is_illegal(in, n))
236 return -1;
237
238 /* decode ... */
239 *c = (char)(*in++ & (0x7F >> n));
240
241 for (i = 1; i < n; i++) {
242 /* illegal continuation byte */
243 if (*in < 0x80 || *in > 0xBF)
244 return -1;
245
246 *c = (*c << 6) | (*in++ & 0x3F);
247 }
248
249 return n;
250 }
251
252 /* unmapped sequence (> 4) */
253 return -1;
254 }
255
256 static inline wchar_t latin9_translit(wchar_t c)
257 {
258 /* a number of trivial iso-8859-15 <> utf-8 transliterations */
259 switch (c) {
260 case 0x20AC: return 0xA4; /* Euro */
261 case 0x0160: return 0xA6; /* S caron */
262 case 0x0161: return 0xA8; /* s caron */
263 case 0x017D: return 0xB4; /* Z caron */
264 case 0x017E: return 0xB8; /* z caron */
265 case 0x0152: return 0xBC; /* OE */
266 case 0x0153: return 0xBD; /* oe */
267 case 0x0178: return 0xBE; /* Y diaeresis */
268 default: return 0xFFFD; /* cannot translate */
269 }
270 }
271
272 size_t iconv(iconv_t cd, char **in, size_t *inb, char **out, size_t *outb)
273 {
274 size_t x=0;
275 unsigned char to = (cd>>1)&127;
276 unsigned char from = 255;
277 const unsigned char *map = 0;
278 char tmp[MB_LEN_MAX];
279 wchar_t c, d;
280 size_t k, l;
281 int err;
282
283 if (!in || !*in || !*inb) return 0;
284
285 if (cd & 1)
286 map = charmaps[cd>>8].map;
287 else
288 from = cd>>8;
289
290 for (; *inb; *in+=l, *inb-=l) {
291 c = *(unsigned char *)*in;
292 l = 1;
293 if (from >= UTF_8 && c < 0x80) goto charok;
294 switch (from) {
295 case WCHAR_T:
296 l = sizeof(wchar_t);
297 if (*inb < l) goto starved;
298 c = *(wchar_t *)*in;
299 break;
300 case UTF_8:
301 l = utf8dec_wchar(&c, *in, *inb);
302 if (!l) l++;
303 else if (l == (size_t)-1) goto ilseq;
304 else if (l == (size_t)-2) goto starved;
305 break;
306 case US_ASCII:
307 goto ilseq;
308 case LATIN_9:
309 if ((unsigned)c - 0xa4 <= 0xbe - 0xa4) {
310 static const unsigned char map[] = {
311 0, 0x60, 0, 0x61, 0, 0, 0, 0, 0, 0, 0,
312 0, 0, 0, 0, 0x7d, 0, 0, 0, 0x7e, 0, 0, 0,
313 0x52, 0x53, 0x78
314 };
315 if (c == 0xa4) c = 0x20ac;
316 else if (map[c-0xa5]) c = 0x100 | map[c-0xa5];
317 }
318 case LATIN_1:
319 goto charok;
320 case TIS_620:
321 if (c >= 0xa1) c += 0x0e01-0xa1;
322 goto charok;
323 case JIS_0201:
324 if (c >= 0xa1) {
325 if (c <= 0xdf) c += 0xff61-0xa1;
326 else goto ilseq;
327 }
328 goto charok;
329 case UTF_16BE:
330 case UTF_16LE:
331 l = 2;
332 if (*inb < 2) goto starved;
333 c = get_16(*in, from);
334 if ((unsigned)(c-0xdc00) < 0x400) goto ilseq;
335 if ((unsigned)(c-0xd800) < 0x400) {
336 l = 4;
337 if (*inb < 4) goto starved;
338 d = get_16(*in + 2, from);
339 if ((unsigned)(c-0xdc00) >= 0x400) goto ilseq;
340 c = ((c-0xd800)<<10) | (d-0xdc00);
341 }
342 break;
343 case UTF_32BE:
344 case UTF_32LE:
345 l = 4;
346 if (*inb < 4) goto starved;
347 // FIXME
348 // c = get_32(*in, from);
349 break;
350 default:
351 /* only support ascii supersets */
352 if (c < 0x80) break;
353 switch (map[0]) {
354 case UCS2_8BIT:
355 c -= 0x80;
356 break;
357 case EUC:
358 if ((unsigned)c - 0xa1 >= 94) goto ilseq;
359 if ((unsigned)in[0][1] - 0xa1 >= 94) goto ilseq;
360 c = (c-0xa1)*94 + (in[0][1]-0xa1);
361 l = 2;
362 break;
363 case SHIFT_JIS:
364 if ((unsigned)c - 0xa1 <= 0xdf-0xa1) {
365 c += 0xff61-0xa1;
366 goto charok;
367 }
368 // FIXME...
369 l = 2;
370 break;
371 default:
372 goto badf;
373 }
374 c = get_16(map + 4 + 2*c, 0);
375 if (c == 0xffff) goto ilseq;
376 goto charok;
377 }
378
379 if ((unsigned)c - 0xd800 < 0x800 || (unsigned)c >= 0x110000)
380 goto ilseq;
381 charok:
382 switch (to) {
383 case WCHAR_T:
384 if (*outb < sizeof(wchar_t)) goto toobig;
385 *(wchar_t *)*out = c;
386 *out += sizeof(wchar_t);
387 *outb -= sizeof(wchar_t);
388 break;
389 case UTF_8:
390 if (*outb < 4) {
391 k = utf8enc_wchar(tmp, c);
392 if (*outb < k) goto toobig;
393 memcpy(*out, tmp, k);
394 } else k = utf8enc_wchar(*out, c);
395 *out += k;
396 *outb -= k;
397 break;
398 case US_ASCII:
399 if (c > 0x7f) c = 0xfffd;
400 /* fall thru and count replacement in latin1 case */
401 case LATIN_9:
402 if (c >= 0x100 && c != 0xfffd)
403 c = latin9_translit(c);
404 /* fall through */
405 case LATIN_1:
406 if (c > 0xff) goto ilseq;
407 if (!*outb) goto toobig;
408 **out = c;
409 ++*out;
410 --*outb;
411 break;
412 case UTF_16BE:
413 case UTF_16LE:
414 if (c < 0x10000) {
415 if (*outb < 2) goto toobig;
416 put_16(*out, c, to);
417 *out += 2;
418 *outb -= 2;
419 break;
420 }
421 if (*outb < 4) goto toobig;
422 put_16(*out, (c>>10)|0xd800, to);
423 put_16(*out + 2, (c&0x3ff)|0xdc00, to);
424 *out += 4;
425 *outb -= 4;
426 break;
427 default:
428 goto badf;
429 }
430 }
431 return x;
432 ilseq:
433 err = EILSEQ;
434 x = -1;
435 goto end;
436 badf:
437 err = EBADF;
438 x = -1;
439 goto end;
440 toobig:
441 err = E2BIG;
442 x = -1;
443 goto end;
444 starved:
445 err = EINVAL;
446 end:
447 errno = err;
448 return x;
449 }