citrus_euctw.c revision 1.8 1 1.8 yamt /* $NetBSD: citrus_euctw.c,v 1.8 2003/10/14 12:50:03 yamt Exp $ */
2 1.1 tshiozak
3 1.1 tshiozak /*-
4 1.1 tshiozak * Copyright (c)2002 Citrus Project,
5 1.1 tshiozak * All rights reserved.
6 1.1 tshiozak *
7 1.1 tshiozak * Redistribution and use in source and binary forms, with or without
8 1.1 tshiozak * modification, are permitted provided that the following conditions
9 1.1 tshiozak * are met:
10 1.1 tshiozak * 1. Redistributions of source code must retain the above copyright
11 1.1 tshiozak * notice, this list of conditions and the following disclaimer.
12 1.1 tshiozak * 2. Redistributions in binary form must reproduce the above copyright
13 1.1 tshiozak * notice, this list of conditions and the following disclaimer in the
14 1.1 tshiozak * documentation and/or other materials provided with the distribution.
15 1.1 tshiozak *
16 1.1 tshiozak * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
17 1.1 tshiozak * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18 1.1 tshiozak * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19 1.1 tshiozak * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
20 1.1 tshiozak * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21 1.1 tshiozak * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22 1.1 tshiozak * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23 1.1 tshiozak * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24 1.1 tshiozak * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25 1.1 tshiozak * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26 1.1 tshiozak * SUCH DAMAGE.
27 1.1 tshiozak */
28 1.1 tshiozak
29 1.1 tshiozak /*-
30 1.1 tshiozak * Copyright (c)1999 Citrus Project,
31 1.1 tshiozak * All rights reserved.
32 1.1 tshiozak *
33 1.1 tshiozak * Redistribution and use in source and binary forms, with or without
34 1.1 tshiozak * modification, are permitted provided that the following conditions
35 1.1 tshiozak * are met:
36 1.1 tshiozak * 1. Redistributions of source code must retain the above copyright
37 1.1 tshiozak * notice, this list of conditions and the following disclaimer.
38 1.1 tshiozak * 2. Redistributions in binary form must reproduce the above copyright
39 1.1 tshiozak * notice, this list of conditions and the following disclaimer in the
40 1.1 tshiozak * documentation and/or other materials provided with the distribution.
41 1.1 tshiozak *
42 1.1 tshiozak * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
43 1.1 tshiozak * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
44 1.1 tshiozak * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
45 1.1 tshiozak * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
46 1.1 tshiozak * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
47 1.1 tshiozak * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
48 1.1 tshiozak * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
49 1.1 tshiozak * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
50 1.1 tshiozak * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
51 1.1 tshiozak * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
52 1.1 tshiozak * SUCH DAMAGE.
53 1.1 tshiozak *
54 1.1 tshiozak * $Citrus: xpg4dl/FreeBSD/lib/libc/locale/euctw.c,v 1.13 2001/06/21 01:51:44 yamt Exp $
55 1.1 tshiozak */
56 1.1 tshiozak
57 1.1 tshiozak #include <sys/cdefs.h>
58 1.1 tshiozak #if defined(LIBC_SCCS) && !defined(lint)
59 1.8 yamt __RCSID("$NetBSD: citrus_euctw.c,v 1.8 2003/10/14 12:50:03 yamt Exp $");
60 1.1 tshiozak #endif /* LIBC_SCCS and not lint */
61 1.1 tshiozak
62 1.1 tshiozak #include <assert.h>
63 1.1 tshiozak #include <errno.h>
64 1.1 tshiozak #include <string.h>
65 1.1 tshiozak #include <stdio.h>
66 1.1 tshiozak #include <stdlib.h>
67 1.1 tshiozak #include <stddef.h>
68 1.1 tshiozak #include <locale.h>
69 1.1 tshiozak #include <wchar.h>
70 1.1 tshiozak #include <sys/types.h>
71 1.1 tshiozak #include <limits.h>
72 1.6 tshiozak
73 1.6 tshiozak #include "citrus_namespace.h"
74 1.6 tshiozak #include "citrus_types.h"
75 1.1 tshiozak #include "citrus_module.h"
76 1.1 tshiozak #include "citrus_ctype.h"
77 1.6 tshiozak #include "citrus_stdenc.h"
78 1.1 tshiozak #include "citrus_euctw.h"
79 1.1 tshiozak
80 1.1 tshiozak
81 1.1 tshiozak /* ----------------------------------------------------------------------
82 1.1 tshiozak * private stuffs used by templates
83 1.1 tshiozak */
84 1.1 tshiozak
85 1.1 tshiozak typedef struct {
86 1.1 tshiozak char ch[4];
87 1.1 tshiozak int chlen;
88 1.1 tshiozak } _EUCTWState;
89 1.1 tshiozak
90 1.1 tshiozak typedef struct {
91 1.1 tshiozak int dummy;
92 1.1 tshiozak } _EUCTWEncodingInfo;
93 1.1 tshiozak typedef struct {
94 1.1 tshiozak _EUCTWEncodingInfo ei;
95 1.1 tshiozak struct {
96 1.1 tshiozak /* for future multi-locale facility */
97 1.1 tshiozak _EUCTWState s_mblen;
98 1.1 tshiozak _EUCTWState s_mbrlen;
99 1.1 tshiozak _EUCTWState s_mbrtowc;
100 1.1 tshiozak _EUCTWState s_mbtowc;
101 1.1 tshiozak _EUCTWState s_mbsrtowcs;
102 1.1 tshiozak _EUCTWState s_wcrtomb;
103 1.1 tshiozak _EUCTWState s_wcsrtombs;
104 1.1 tshiozak _EUCTWState s_wctomb;
105 1.1 tshiozak } states;
106 1.1 tshiozak } _EUCTWCTypeInfo;
107 1.1 tshiozak
108 1.1 tshiozak #define _SS2 0x008e
109 1.1 tshiozak #define _SS3 0x008f
110 1.1 tshiozak
111 1.1 tshiozak #define _CEI_TO_EI(_cei_) (&(_cei_)->ei)
112 1.1 tshiozak #define _CEI_TO_STATE(_cei_, _func_) (_cei_)->states.s_##_func_
113 1.1 tshiozak
114 1.1 tshiozak #define _FUNCNAME(m) _citrus_EUCTW_##m
115 1.1 tshiozak #define _ENCODING_INFO _EUCTWEncodingInfo
116 1.1 tshiozak #define _CTYPE_INFO _EUCTWCTypeInfo
117 1.1 tshiozak #define _ENCODING_STATE _EUCTWState
118 1.2 yamt #define _ENCODING_MB_CUR_MAX(_ei_) 4
119 1.1 tshiozak #define _ENCODING_IS_STATE_DEPENDENT 0
120 1.4 yamt #define _STATE_NEEDS_EXPLICIT_INIT(_ps_) 0
121 1.1 tshiozak
122 1.1 tshiozak static __inline int
123 1.1 tshiozak _citrus_EUCTW_cs(u_int c)
124 1.1 tshiozak {
125 1.1 tshiozak c &= 0xff;
126 1.1 tshiozak
127 1.1 tshiozak return ((c & 0x80) ? (c == _SS2 ? 2 : 1) : 0);
128 1.1 tshiozak }
129 1.1 tshiozak
130 1.1 tshiozak static __inline int
131 1.1 tshiozak _citrus_EUCTW_count(int cs)
132 1.1 tshiozak {
133 1.1 tshiozak switch (cs) {
134 1.1 tshiozak case 0:
135 1.1 tshiozak return 1;
136 1.1 tshiozak case 1:
137 1.1 tshiozak return 2;
138 1.1 tshiozak case 2:
139 1.1 tshiozak return 4;
140 1.1 tshiozak case 3:
141 1.1 tshiozak abort();
142 1.1 tshiozak /*NOTREACHED*/
143 1.1 tshiozak }
144 1.1 tshiozak return 0;
145 1.1 tshiozak }
146 1.1 tshiozak
147 1.1 tshiozak static __inline void
148 1.1 tshiozak /*ARGSUSED*/
149 1.1 tshiozak _citrus_EUCTW_init_state(_EUCTWEncodingInfo * __restrict ei,
150 1.1 tshiozak _EUCTWState * __restrict s)
151 1.1 tshiozak {
152 1.1 tshiozak memset(s, 0, sizeof(*s));
153 1.1 tshiozak }
154 1.1 tshiozak
155 1.1 tshiozak static __inline void
156 1.1 tshiozak /*ARGSUSED*/
157 1.1 tshiozak _citrus_EUCTW_pack_state(_EUCTWEncodingInfo * __restrict ei,
158 1.1 tshiozak void * __restrict pspriv,
159 1.1 tshiozak const _EUCTWState * __restrict s)
160 1.1 tshiozak {
161 1.1 tshiozak memcpy(pspriv, (const void *)s, sizeof(*s));
162 1.1 tshiozak }
163 1.1 tshiozak
164 1.1 tshiozak static __inline void
165 1.1 tshiozak /*ARGSUSED*/
166 1.1 tshiozak _citrus_EUCTW_unpack_state(_EUCTWEncodingInfo * __restrict ei,
167 1.1 tshiozak _EUCTWState * __restrict s,
168 1.1 tshiozak const void * __restrict pspriv)
169 1.1 tshiozak {
170 1.1 tshiozak memcpy((void *)s, pspriv, sizeof(*s));
171 1.1 tshiozak }
172 1.1 tshiozak
173 1.1 tshiozak static int
174 1.1 tshiozak /*ARGSUSED*/
175 1.6 tshiozak _citrus_EUCTW_encoding_module_init(_EUCTWEncodingInfo * __restrict ei,
176 1.6 tshiozak const void * __restrict var, size_t lenvar)
177 1.1 tshiozak {
178 1.1 tshiozak
179 1.1 tshiozak _DIAGASSERT(ei != NULL);
180 1.1 tshiozak
181 1.1 tshiozak memset((void *)ei, 0, sizeof(*ei));
182 1.1 tshiozak
183 1.1 tshiozak return 0;
184 1.1 tshiozak }
185 1.1 tshiozak
186 1.1 tshiozak static void
187 1.1 tshiozak /*ARGSUSED*/
188 1.6 tshiozak _citrus_EUCTW_encoding_module_uninit(_EUCTWEncodingInfo *ei)
189 1.1 tshiozak {
190 1.1 tshiozak }
191 1.1 tshiozak
192 1.1 tshiozak static int
193 1.1 tshiozak _citrus_EUCTW_mbrtowc_priv(_EUCTWEncodingInfo * __restrict ei,
194 1.1 tshiozak wchar_t * __restrict pwc,
195 1.1 tshiozak const char ** __restrict s,
196 1.1 tshiozak size_t n, _EUCTWState * __restrict psenc,
197 1.1 tshiozak size_t * __restrict nresult)
198 1.1 tshiozak {
199 1.1 tshiozak wchar_t wchar;
200 1.1 tshiozak int c, cs;
201 1.1 tshiozak int chlenbak;
202 1.1 tshiozak const char *s0;
203 1.1 tshiozak
204 1.1 tshiozak _DIAGASSERT(nresult != 0);
205 1.1 tshiozak _DIAGASSERT(ei != NULL);
206 1.1 tshiozak _DIAGASSERT(psenc != NULL);
207 1.1 tshiozak _DIAGASSERT(s != NULL);
208 1.1 tshiozak
209 1.1 tshiozak s0 = *s;
210 1.1 tshiozak
211 1.1 tshiozak if (s0 == NULL) {
212 1.1 tshiozak _citrus_EUCTW_init_state(ei, psenc);
213 1.1 tshiozak *nresult = 0; /* state independent */
214 1.1 tshiozak return (0);
215 1.1 tshiozak }
216 1.8 yamt
217 1.8 yamt chlenbak = psenc->chlen;
218 1.1 tshiozak
219 1.1 tshiozak /* make sure we have the first byte in the buffer */
220 1.1 tshiozak switch (psenc->chlen) {
221 1.1 tshiozak case 0:
222 1.1 tshiozak if (n < 1)
223 1.1 tshiozak goto restart;
224 1.1 tshiozak psenc->ch[0] = *s0++;
225 1.1 tshiozak psenc->chlen = 1;
226 1.1 tshiozak n--;
227 1.1 tshiozak break;
228 1.1 tshiozak case 1:
229 1.1 tshiozak case 2:
230 1.1 tshiozak break;
231 1.1 tshiozak default:
232 1.1 tshiozak /* illgeal state */
233 1.1 tshiozak goto ilseq;
234 1.1 tshiozak }
235 1.1 tshiozak
236 1.1 tshiozak c = _citrus_EUCTW_count(cs = _citrus_EUCTW_cs(psenc->ch[0] & 0xff));
237 1.1 tshiozak if (c == 0)
238 1.1 tshiozak goto ilseq;
239 1.1 tshiozak while (psenc->chlen < c) {
240 1.1 tshiozak if (n < 1)
241 1.1 tshiozak goto ilseq;
242 1.1 tshiozak psenc->ch[psenc->chlen] = *s0++;
243 1.1 tshiozak psenc->chlen++;
244 1.1 tshiozak n--;
245 1.1 tshiozak }
246 1.1 tshiozak
247 1.1 tshiozak wchar = 0;
248 1.1 tshiozak switch (cs) {
249 1.1 tshiozak case 0:
250 1.1 tshiozak if (psenc->ch[0] & 0x80)
251 1.1 tshiozak goto ilseq;
252 1.1 tshiozak wchar = psenc->ch[0] & 0xff;
253 1.1 tshiozak break;
254 1.1 tshiozak case 1:
255 1.1 tshiozak if (!(psenc->ch[0] & 0x80) || !(psenc->ch[1] & 0x80))
256 1.1 tshiozak goto ilseq;
257 1.1 tshiozak wchar = ((psenc->ch[0] & 0xff) << 8) | (psenc->ch[1] & 0xff);
258 1.1 tshiozak wchar |= 'G' << 24;
259 1.1 tshiozak break;
260 1.1 tshiozak case 2:
261 1.1 tshiozak if ((u_char)psenc->ch[1] < 0xa1 || 0xa7 < (u_char)psenc->ch[1])
262 1.1 tshiozak goto ilseq;
263 1.1 tshiozak if (!(psenc->ch[2] & 0x80) || !(psenc->ch[3] & 0x80))
264 1.1 tshiozak goto ilseq;
265 1.1 tshiozak wchar = ((psenc->ch[2] & 0xff) << 8) | (psenc->ch[3] & 0xff);
266 1.1 tshiozak wchar |= ('G' + psenc->ch[1] - 0xa1) << 24;
267 1.1 tshiozak break;
268 1.1 tshiozak default:
269 1.1 tshiozak goto ilseq;
270 1.1 tshiozak }
271 1.1 tshiozak
272 1.1 tshiozak *s = s0;
273 1.1 tshiozak psenc->chlen = 0;
274 1.1 tshiozak
275 1.1 tshiozak if (pwc)
276 1.1 tshiozak *pwc = wchar;
277 1.1 tshiozak
278 1.1 tshiozak if (!wchar)
279 1.1 tshiozak *nresult = 0;
280 1.1 tshiozak else
281 1.1 tshiozak *nresult = c - chlenbak;
282 1.1 tshiozak
283 1.1 tshiozak return (0);
284 1.1 tshiozak
285 1.1 tshiozak ilseq:
286 1.1 tshiozak psenc->chlen = 0;
287 1.1 tshiozak *nresult = (size_t)-1;
288 1.1 tshiozak return (EILSEQ);
289 1.1 tshiozak
290 1.1 tshiozak restart:
291 1.1 tshiozak *s = s0;
292 1.1 tshiozak *nresult = (size_t)-1;
293 1.1 tshiozak return (0);
294 1.1 tshiozak }
295 1.1 tshiozak
296 1.1 tshiozak static int
297 1.1 tshiozak _citrus_EUCTW_wcrtomb_priv(_EUCTWEncodingInfo * __restrict ei,
298 1.1 tshiozak char * __restrict s, size_t n, wchar_t wc,
299 1.1 tshiozak _EUCTWState * __restrict psenc,
300 1.1 tshiozak size_t * __restrict nresult)
301 1.1 tshiozak {
302 1.1 tshiozak wchar_t cs = wc & 0x7f000080;
303 1.1 tshiozak wchar_t v;
304 1.6 tshiozak int i, len, clen, ret;
305 1.1 tshiozak
306 1.1 tshiozak _DIAGASSERT(ei != NULL);
307 1.1 tshiozak _DIAGASSERT(nresult != 0);
308 1.1 tshiozak _DIAGASSERT(s != NULL);
309 1.6 tshiozak
310 1.1 tshiozak clen = 1;
311 1.1 tshiozak if (wc & 0x00007f00)
312 1.1 tshiozak clen = 2;
313 1.1 tshiozak if ((wc & 0x007f0000) && !(wc & 0x00800000))
314 1.1 tshiozak clen = 3;
315 1.1 tshiozak
316 1.1 tshiozak if (clen == 1 && cs == 0x00000000) {
317 1.1 tshiozak /* ASCII */
318 1.1 tshiozak len = 1;
319 1.6 tshiozak if (n < len) {
320 1.6 tshiozak ret = E2BIG;
321 1.6 tshiozak goto err;
322 1.6 tshiozak }
323 1.1 tshiozak v = wc & 0x0000007f;
324 1.1 tshiozak } else if (clen == 2 && cs == ('G' << 24)) {
325 1.1 tshiozak /* CNS-11643-1 */
326 1.1 tshiozak len = 2;
327 1.6 tshiozak if (n < len) {
328 1.6 tshiozak ret = E2BIG;
329 1.6 tshiozak goto err;
330 1.6 tshiozak }
331 1.1 tshiozak v = wc & 0x00007f7f;
332 1.1 tshiozak v |= 0x00008080;
333 1.1 tshiozak } else if (clen == 2 && 'H' <= (cs >> 24) && (cs >> 24) <= 'M') {
334 1.1 tshiozak /* CNS-11643-[2-7] */
335 1.1 tshiozak len = 4;
336 1.6 tshiozak if (n < len) {
337 1.6 tshiozak ret = E2BIG;
338 1.6 tshiozak goto err;
339 1.6 tshiozak }
340 1.1 tshiozak *s++ = _SS2;
341 1.1 tshiozak *s++ = (cs >> 24) - 'H' + 0xa2;
342 1.1 tshiozak v = wc & 0x00007f7f;
343 1.1 tshiozak v |= 0x00008080;
344 1.6 tshiozak } else {
345 1.6 tshiozak ret = EILSEQ;
346 1.6 tshiozak goto err;
347 1.6 tshiozak }
348 1.1 tshiozak
349 1.1 tshiozak i = clen;
350 1.1 tshiozak while (i-- > 0)
351 1.1 tshiozak *s++ = (v >> (i << 3)) & 0xff;
352 1.1 tshiozak
353 1.1 tshiozak *nresult = len;
354 1.6 tshiozak return 0;
355 1.6 tshiozak
356 1.6 tshiozak err:
357 1.6 tshiozak *nresult = (size_t)-1;
358 1.6 tshiozak return ret;
359 1.6 tshiozak }
360 1.6 tshiozak
361 1.6 tshiozak static __inline int
362 1.6 tshiozak /*ARGSUSED*/
363 1.6 tshiozak _citrus_EUCTW_stdenc_wctocs(_EUCTWEncodingInfo * __restrict ei,
364 1.6 tshiozak _csid_t * __restrict csid,
365 1.6 tshiozak _index_t * __restrict idx, wchar_t wc)
366 1.6 tshiozak {
367 1.6 tshiozak
368 1.6 tshiozak _DIAGASSERT(ei != NULL && csid != NULL && idx != NULL);
369 1.6 tshiozak
370 1.6 tshiozak *csid = (_csid_t)(wc >> 24) & 0xFF;
371 1.6 tshiozak *idx = (_index_t)(wc & 0x7F7F);
372 1.6 tshiozak
373 1.1 tshiozak return (0);
374 1.6 tshiozak }
375 1.6 tshiozak
376 1.6 tshiozak static __inline int
377 1.6 tshiozak /*ARGSUSED*/
378 1.6 tshiozak _citrus_EUCTW_stdenc_cstowc(_EUCTWEncodingInfo * __restrict ei,
379 1.6 tshiozak wchar_t * __restrict wc,
380 1.6 tshiozak _csid_t csid, _index_t idx)
381 1.6 tshiozak {
382 1.6 tshiozak
383 1.6 tshiozak _DIAGASSERT(ei != NULL && wc != NULL);
384 1.6 tshiozak
385 1.6 tshiozak if (csid > 7 || (idx & ~0x7F7F) != 0)
386 1.6 tshiozak return (EINVAL);
387 1.6 tshiozak
388 1.6 tshiozak if (csid==0) {
389 1.6 tshiozak if ((idx & ~0x7F) != 0)
390 1.6 tshiozak return (EINVAL);
391 1.6 tshiozak *wc = (wchar_t)idx;
392 1.6 tshiozak } else {
393 1.6 tshiozak if ((idx & ~0x7F7F) != 0)
394 1.6 tshiozak return (EINVAL);
395 1.6 tshiozak *wc = (wchar_t)idx | ((wchar_t)csid<<24);
396 1.6 tshiozak }
397 1.1 tshiozak
398 1.6 tshiozak return (0);
399 1.1 tshiozak }
400 1.1 tshiozak
401 1.1 tshiozak /* ----------------------------------------------------------------------
402 1.1 tshiozak * public interface for ctype
403 1.1 tshiozak */
404 1.1 tshiozak
405 1.1 tshiozak _CITRUS_CTYPE_DECLS(EUCTW);
406 1.1 tshiozak _CITRUS_CTYPE_DEF_OPS(EUCTW);
407 1.1 tshiozak
408 1.1 tshiozak #include "citrus_ctype_template.h"
409 1.6 tshiozak
410 1.6 tshiozak /* ----------------------------------------------------------------------
411 1.6 tshiozak * public interface for stdenc
412 1.6 tshiozak */
413 1.6 tshiozak
414 1.6 tshiozak _CITRUS_STDENC_DECLS(EUCTW);
415 1.6 tshiozak _CITRUS_STDENC_DEF_OPS(EUCTW);
416 1.6 tshiozak
417 1.6 tshiozak #include "citrus_stdenc_template.h"
418