citrus_mskanji.c revision 1.13.6.2 1 1.13.6.2 tnozaki /* $NetBSD: citrus_mskanji.c,v 1.13.6.2 2008/06/14 16:01:09 tnozaki Exp $ */
2 1.13.6.2 tnozaki
3 1.13.6.2 tnozaki /*-
4 1.13.6.2 tnozaki * Copyright (c)2002 Citrus Project,
5 1.13.6.2 tnozaki * All rights reserved.
6 1.13.6.2 tnozaki *
7 1.13.6.2 tnozaki * Redistribution and use in source and binary forms, with or without
8 1.13.6.2 tnozaki * modification, are permitted provided that the following conditions
9 1.13.6.2 tnozaki * are met:
10 1.13.6.2 tnozaki * 1. Redistributions of source code must retain the above copyright
11 1.13.6.2 tnozaki * notice, this list of conditions and the following disclaimer.
12 1.13.6.2 tnozaki * 2. Redistributions in binary form must reproduce the above copyright
13 1.13.6.2 tnozaki * notice, this list of conditions and the following disclaimer in the
14 1.13.6.2 tnozaki * documentation and/or other materials provided with the distribution.
15 1.13.6.2 tnozaki *
16 1.13.6.2 tnozaki * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
17 1.13.6.2 tnozaki * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18 1.13.6.2 tnozaki * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19 1.13.6.2 tnozaki * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
20 1.13.6.2 tnozaki * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21 1.13.6.2 tnozaki * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22 1.13.6.2 tnozaki * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23 1.13.6.2 tnozaki * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24 1.13.6.2 tnozaki * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25 1.13.6.2 tnozaki * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26 1.13.6.2 tnozaki * SUCH DAMAGE.
27 1.13.6.2 tnozaki */
28 1.13.6.2 tnozaki
29 1.13.6.2 tnozaki /*
30 1.13.6.2 tnozaki * ja_JP.SJIS locale table for BSD4.4/rune
31 1.13.6.2 tnozaki * version 1.0
32 1.13.6.2 tnozaki * (C) Sin'ichiro MIYATANI / Phase One, Inc
33 1.13.6.2 tnozaki * May 12, 1995
34 1.13.6.2 tnozaki *
35 1.13.6.2 tnozaki * Redistribution and use in source and binary forms, with or without
36 1.13.6.2 tnozaki * modification, are permitted provided that the following conditions
37 1.13.6.2 tnozaki * are met:
38 1.13.6.2 tnozaki * 1. Redistributions of source code must retain the above copyright
39 1.13.6.2 tnozaki * notice, this list of conditions and the following disclaimer.
40 1.13.6.2 tnozaki * 2. Redistributions in binary form must reproduce the above copyright
41 1.13.6.2 tnozaki * notice, this list of conditions and the following disclaimer in the
42 1.13.6.2 tnozaki * documentation and/or other materials provided with the distribution.
43 1.13.6.2 tnozaki * 3. All advertising materials mentioning features or use of this software
44 1.13.6.2 tnozaki * must display the following acknowledgement:
45 1.13.6.2 tnozaki * This product includes software developed by Phase One, Inc.
46 1.13.6.2 tnozaki * 4. The name of Phase One, Inc. may be used to endorse or promote products
47 1.13.6.2 tnozaki * derived from this software without specific prior written permission.
48 1.13.6.2 tnozaki *
49 1.13.6.2 tnozaki * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
50 1.13.6.2 tnozaki * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
51 1.13.6.2 tnozaki * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
52 1.13.6.2 tnozaki * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
53 1.13.6.2 tnozaki * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
54 1.13.6.2 tnozaki * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
55 1.13.6.2 tnozaki * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
56 1.13.6.2 tnozaki * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
57 1.13.6.2 tnozaki * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
58 1.13.6.2 tnozaki * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
59 1.13.6.2 tnozaki * SUCH DAMAGE.
60 1.13.6.2 tnozaki */
61 1.13.6.2 tnozaki
62 1.13.6.2 tnozaki
63 1.13.6.2 tnozaki #include <sys/cdefs.h>
64 1.13.6.2 tnozaki #if defined(LIBC_SCCS) && !defined(lint)
65 1.13.6.2 tnozaki __RCSID("$NetBSD: citrus_mskanji.c,v 1.13.6.2 2008/06/14 16:01:09 tnozaki Exp $");
66 1.13.6.2 tnozaki #endif /* LIBC_SCCS and not lint */
67 1.13.6.2 tnozaki
68 1.13.6.2 tnozaki #include <assert.h>
69 1.13.6.2 tnozaki #include <errno.h>
70 1.13.6.2 tnozaki #include <string.h>
71 1.13.6.2 tnozaki #include <stdio.h>
72 1.13.6.2 tnozaki #include <stdlib.h>
73 1.13.6.2 tnozaki #include <stddef.h>
74 1.13.6.2 tnozaki #include <wchar.h>
75 1.13.6.2 tnozaki #include <sys/types.h>
76 1.13.6.2 tnozaki #include <limits.h>
77 1.13.6.2 tnozaki
78 1.13.6.2 tnozaki #include "citrus_namespace.h"
79 1.13.6.2 tnozaki #include "citrus_types.h"
80 1.13.6.2 tnozaki #include "citrus_bcs.h"
81 1.13.6.2 tnozaki #include "citrus_module.h"
82 1.13.6.2 tnozaki #include "citrus_ctype.h"
83 1.13.6.2 tnozaki #include "citrus_stdenc.h"
84 1.13.6.2 tnozaki #include "citrus_mskanji.h"
85 1.13.6.2 tnozaki
86 1.13.6.2 tnozaki
87 1.13.6.2 tnozaki /* ----------------------------------------------------------------------
88 1.13.6.2 tnozaki * private stuffs used by templates
89 1.13.6.2 tnozaki */
90 1.13.6.2 tnozaki
91 1.13.6.2 tnozaki typedef struct _MSKanjiState {
92 1.13.6.2 tnozaki char ch[2];
93 1.13.6.2 tnozaki int chlen;
94 1.13.6.2 tnozaki } _MSKanjiState;
95 1.13.6.2 tnozaki
96 1.13.6.2 tnozaki typedef struct {
97 1.13.6.2 tnozaki int mode;
98 1.13.6.2 tnozaki #define MODE_JIS2004 1
99 1.13.6.2 tnozaki } _MSKanjiEncodingInfo;
100 1.13.6.2 tnozaki
101 1.13.6.2 tnozaki typedef struct {
102 1.13.6.2 tnozaki _MSKanjiEncodingInfo ei;
103 1.13.6.2 tnozaki struct {
104 1.13.6.2 tnozaki /* for future multi-locale facility */
105 1.13.6.2 tnozaki _MSKanjiState s_mblen;
106 1.13.6.2 tnozaki _MSKanjiState s_mbrlen;
107 1.13.6.2 tnozaki _MSKanjiState s_mbrtowc;
108 1.13.6.2 tnozaki _MSKanjiState s_mbtowc;
109 1.13.6.2 tnozaki _MSKanjiState s_mbsrtowcs;
110 1.13.6.2 tnozaki _MSKanjiState s_wcrtomb;
111 1.13.6.2 tnozaki _MSKanjiState s_wcsrtombs;
112 1.13.6.2 tnozaki _MSKanjiState s_wctomb;
113 1.13.6.2 tnozaki } states;
114 1.13.6.2 tnozaki } _MSKanjiCTypeInfo;
115 1.13.6.2 tnozaki
116 1.13.6.2 tnozaki #define _CEI_TO_EI(_cei_) (&(_cei_)->ei)
117 1.13.6.2 tnozaki #define _CEI_TO_STATE(_cei_, _func_) (_cei_)->states.s_##_func_
118 1.13.6.2 tnozaki
119 1.13.6.2 tnozaki #define _FUNCNAME(m) _citrus_MSKanji_##m
120 1.13.6.2 tnozaki #define _ENCODING_INFO _MSKanjiEncodingInfo
121 1.13.6.2 tnozaki #define _CTYPE_INFO _MSKanjiCTypeInfo
122 1.13.6.2 tnozaki #define _ENCODING_STATE _MSKanjiState
123 1.13.6.2 tnozaki #define _ENCODING_MB_CUR_MAX(_ei_) 2
124 1.13.6.2 tnozaki #define _ENCODING_IS_STATE_DEPENDENT 0
125 1.13.6.2 tnozaki #define _STATE_NEEDS_EXPLICIT_INIT(_ps_) 0
126 1.13.6.2 tnozaki
127 1.13.6.2 tnozaki
128 1.13.6.2 tnozaki static int
129 1.13.6.2 tnozaki _mskanji1(int c)
130 1.13.6.2 tnozaki {
131 1.13.6.2 tnozaki
132 1.13.6.2 tnozaki if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xfc))
133 1.13.6.2 tnozaki return 1;
134 1.13.6.2 tnozaki else
135 1.13.6.2 tnozaki return 0;
136 1.13.6.2 tnozaki }
137 1.13.6.2 tnozaki
138 1.13.6.2 tnozaki static int
139 1.13.6.2 tnozaki _mskanji2(int c)
140 1.13.6.2 tnozaki {
141 1.13.6.2 tnozaki
142 1.13.6.2 tnozaki if ((c >= 0x40 && c <= 0x7e) || (c >= 0x80 && c <= 0xfc))
143 1.13.6.2 tnozaki return 1;
144 1.13.6.2 tnozaki else
145 1.13.6.2 tnozaki return 0;
146 1.13.6.2 tnozaki }
147 1.13.6.2 tnozaki
148 1.13.6.2 tnozaki static __inline void
149 1.13.6.2 tnozaki /*ARGSUSED*/
150 1.13.6.2 tnozaki _citrus_MSKanji_init_state(_MSKanjiEncodingInfo * __restrict ei,
151 1.13.6.2 tnozaki _MSKanjiState * __restrict s)
152 1.13.6.2 tnozaki {
153 1.13.6.2 tnozaki s->chlen = 0;
154 1.13.6.2 tnozaki }
155 1.13.6.2 tnozaki
156 1.13.6.2 tnozaki static __inline void
157 1.13.6.2 tnozaki /*ARGSUSED*/
158 1.13.6.2 tnozaki _citrus_MSKanji_pack_state(_MSKanjiEncodingInfo * __restrict ei,
159 1.13.6.2 tnozaki void * __restrict pspriv,
160 1.13.6.2 tnozaki const _MSKanjiState * __restrict s)
161 1.13.6.2 tnozaki {
162 1.13.6.2 tnozaki memcpy(pspriv, (const void *)s, sizeof(*s));
163 1.13.6.2 tnozaki }
164 1.13.6.2 tnozaki
165 1.13.6.2 tnozaki static __inline void
166 1.13.6.2 tnozaki /*ARGSUSED*/
167 1.13.6.2 tnozaki _citrus_MSKanji_unpack_state(_MSKanjiEncodingInfo * __restrict ei,
168 1.13.6.2 tnozaki _MSKanjiState * __restrict s,
169 1.13.6.2 tnozaki const void * __restrict pspriv)
170 1.13.6.2 tnozaki {
171 1.13.6.2 tnozaki memcpy((void *)s, pspriv, sizeof(*s));
172 1.13.6.2 tnozaki }
173 1.13.6.2 tnozaki
174 1.13.6.2 tnozaki static int
175 1.13.6.2 tnozaki /*ARGSUSED*/
176 1.13.6.2 tnozaki _citrus_MSKanji_mbrtowc_priv(_MSKanjiEncodingInfo * __restrict ei,
177 1.13.6.2 tnozaki wchar_t * __restrict pwc,
178 1.13.6.2 tnozaki const char ** __restrict s, size_t n,
179 1.13.6.2 tnozaki _MSKanjiState * __restrict psenc,
180 1.13.6.2 tnozaki size_t * __restrict nresult)
181 1.13.6.2 tnozaki {
182 1.13.6.2 tnozaki wchar_t wchar;
183 1.13.6.2 tnozaki int len;
184 1.13.6.2 tnozaki int chlenbak;
185 1.13.6.2 tnozaki const char *s0;
186 1.13.6.2 tnozaki
187 1.13.6.2 tnozaki _DIAGASSERT(nresult != 0);
188 1.13.6.2 tnozaki _DIAGASSERT(ei != NULL);
189 1.13.6.2 tnozaki _DIAGASSERT(s != NULL);
190 1.13.6.2 tnozaki _DIAGASSERT(psenc != NULL);
191 1.13.6.2 tnozaki
192 1.13.6.2 tnozaki s0 = *s;
193 1.13.6.2 tnozaki
194 1.13.6.2 tnozaki if (s0 == NULL) {
195 1.13.6.2 tnozaki _citrus_MSKanji_init_state(ei, psenc);
196 1.13.6.2 tnozaki *nresult = 0; /* state independent */
197 1.13.6.2 tnozaki return (0);
198 1.13.6.2 tnozaki }
199 1.13.6.2 tnozaki
200 1.13.6.2 tnozaki chlenbak = psenc->chlen;
201 1.13.6.2 tnozaki
202 1.13.6.2 tnozaki /* make sure we have the first byte in the buffer */
203 1.13.6.2 tnozaki switch (psenc->chlen) {
204 1.13.6.2 tnozaki case 0:
205 1.13.6.2 tnozaki if (n < 1)
206 1.13.6.2 tnozaki goto restart;
207 1.13.6.2 tnozaki psenc->ch[0] = *s0++;
208 1.13.6.2 tnozaki psenc->chlen = 1;
209 1.13.6.2 tnozaki n--;
210 1.13.6.2 tnozaki break;
211 1.13.6.2 tnozaki case 1:
212 1.13.6.2 tnozaki break;
213 1.13.6.2 tnozaki default:
214 1.13.6.2 tnozaki /* illegal state */
215 1.13.6.2 tnozaki goto encoding_error;
216 1.13.6.2 tnozaki }
217 1.13.6.2 tnozaki
218 1.13.6.2 tnozaki len = _mskanji1(psenc->ch[0] & 0xff) ? 2 : 1;
219 1.13.6.2 tnozaki while (psenc->chlen < len) {
220 1.13.6.2 tnozaki if (n < 1)
221 1.13.6.2 tnozaki goto restart;
222 1.13.6.2 tnozaki psenc->ch[psenc->chlen] = *s0++;
223 1.13.6.2 tnozaki psenc->chlen++;
224 1.13.6.2 tnozaki n--;
225 1.13.6.2 tnozaki }
226 1.13.6.2 tnozaki
227 1.13.6.2 tnozaki *s = s0;
228 1.13.6.2 tnozaki
229 1.13.6.2 tnozaki switch (len) {
230 1.13.6.2 tnozaki case 1:
231 1.13.6.2 tnozaki wchar = psenc->ch[0] & 0xff;
232 1.13.6.2 tnozaki break;
233 1.13.6.2 tnozaki case 2:
234 1.13.6.2 tnozaki if (!_mskanji2(psenc->ch[1] & 0xff))
235 1.13.6.2 tnozaki goto encoding_error;
236 1.13.6.2 tnozaki wchar = ((psenc->ch[0] & 0xff) << 8) | (psenc->ch[1] & 0xff);
237 1.13.6.2 tnozaki break;
238 1.13.6.2 tnozaki default:
239 1.13.6.2 tnozaki /* illegal state */
240 1.13.6.2 tnozaki goto encoding_error;
241 1.13.6.2 tnozaki }
242 1.13.6.2 tnozaki
243 1.13.6.2 tnozaki psenc->chlen = 0;
244 1.13.6.2 tnozaki
245 1.13.6.2 tnozaki if (pwc)
246 1.13.6.2 tnozaki *pwc = wchar;
247 1.13.6.2 tnozaki
248 1.13.6.2 tnozaki if (!wchar)
249 1.13.6.2 tnozaki *nresult = 0;
250 1.13.6.2 tnozaki else
251 1.13.6.2 tnozaki *nresult = len - chlenbak;
252 1.13.6.2 tnozaki
253 1.13.6.2 tnozaki return (0);
254 1.13.6.2 tnozaki
255 1.13.6.2 tnozaki encoding_error:
256 1.13.6.2 tnozaki psenc->chlen = 0;
257 1.13.6.2 tnozaki *nresult = (size_t)-1;
258 1.13.6.2 tnozaki return (EILSEQ);
259 1.13.6.2 tnozaki
260 1.13.6.2 tnozaki restart:
261 1.13.6.2 tnozaki *nresult = (size_t)-2;
262 1.13.6.2 tnozaki *s = s0;
263 1.13.6.2 tnozaki return (0);
264 1.13.6.2 tnozaki }
265 1.13.6.2 tnozaki
266 1.13.6.2 tnozaki
267 1.13.6.2 tnozaki static int
268 1.13.6.2 tnozaki _citrus_MSKanji_wcrtomb_priv(_MSKanjiEncodingInfo * __restrict ei,
269 1.13.6.2 tnozaki char * __restrict s, size_t n, wchar_t wc,
270 1.13.6.2 tnozaki _MSKanjiState * __restrict psenc,
271 1.13.6.2 tnozaki size_t * __restrict nresult)
272 1.13.6.2 tnozaki {
273 1.13.6.2 tnozaki int ret;
274 1.13.6.2 tnozaki
275 1.13.6.2 tnozaki _DIAGASSERT(ei != NULL);
276 1.13.6.2 tnozaki _DIAGASSERT(psenc != NULL);
277 1.13.6.2 tnozaki _DIAGASSERT(s != NULL);
278 1.13.6.2 tnozaki
279 1.13.6.2 tnozaki /* check invalid sequence */
280 1.13.6.2 tnozaki if (wc & ~0xffff) {
281 1.13.6.2 tnozaki ret = EILSEQ;
282 1.13.6.2 tnozaki goto err;
283 1.13.6.2 tnozaki }
284 1.13.6.2 tnozaki
285 1.13.6.2 tnozaki if (wc & 0xff00) {
286 1.13.6.2 tnozaki if (n < 2) {
287 1.13.6.2 tnozaki ret = E2BIG;
288 1.13.6.2 tnozaki goto err;
289 1.13.6.2 tnozaki }
290 1.13.6.2 tnozaki
291 1.13.6.2 tnozaki s[0] = (wc >> 8) & 0xff;
292 1.13.6.2 tnozaki s[1] = wc & 0xff;
293 1.13.6.2 tnozaki if (!_mskanji1(s[0] & 0xff) || !_mskanji2(s[1] & 0xff)) {
294 1.13.6.2 tnozaki ret = EILSEQ;
295 1.13.6.2 tnozaki goto err;
296 1.13.6.2 tnozaki }
297 1.13.6.2 tnozaki
298 1.13.6.2 tnozaki *nresult = 2;
299 1.13.6.2 tnozaki return 0;
300 1.13.6.2 tnozaki } else {
301 1.13.6.2 tnozaki if (n < 1) {
302 1.13.6.2 tnozaki ret = E2BIG;
303 1.13.6.2 tnozaki goto err;
304 1.13.6.2 tnozaki }
305 1.13.6.2 tnozaki
306 1.13.6.2 tnozaki s[0] = wc & 0xff;
307 1.13.6.2 tnozaki if (_mskanji1(s[0] & 0xff)) {
308 1.13.6.2 tnozaki ret = EILSEQ;
309 1.13.6.2 tnozaki goto err;
310 1.13.6.2 tnozaki }
311 1.13.6.2 tnozaki
312 1.13.6.2 tnozaki *nresult = 1;
313 1.13.6.2 tnozaki return 0;
314 1.13.6.2 tnozaki }
315 1.13.6.2 tnozaki
316 1.13.6.2 tnozaki err:
317 1.13.6.2 tnozaki *nresult = (size_t)-1;
318 1.13.6.2 tnozaki return ret;
319 1.13.6.2 tnozaki }
320 1.13.6.2 tnozaki
321 1.13.6.2 tnozaki
322 1.13.6.2 tnozaki static __inline int
323 1.13.6.2 tnozaki /*ARGSUSED*/
324 1.13.6.2 tnozaki _citrus_MSKanji_stdenc_wctocs(_MSKanjiEncodingInfo * __restrict ei,
325 1.13.6.2 tnozaki _csid_t * __restrict csid,
326 1.13.6.2 tnozaki _index_t * __restrict idx, wchar_t wc)
327 1.13.6.2 tnozaki {
328 1.13.6.2 tnozaki _index_t row, col;
329 1.13.6.2 tnozaki int offset;
330 1.13.6.2 tnozaki
331 1.13.6.2 tnozaki _DIAGASSERT(csid != NULL && idx != NULL);
332 1.13.6.2 tnozaki
333 1.13.6.2 tnozaki if ((_wc_t)wc < 0x80) {
334 1.13.6.2 tnozaki /* ISO-646 */
335 1.13.6.2 tnozaki *csid = 0;
336 1.13.6.2 tnozaki *idx = (_index_t)wc;
337 1.13.6.2 tnozaki } else if ((_wc_t)wc < 0x100) {
338 1.13.6.2 tnozaki /* KANA */
339 1.13.6.2 tnozaki *csid = 1;
340 1.13.6.2 tnozaki *idx = (_index_t)wc & 0x7F;
341 1.13.6.2 tnozaki } else {
342 1.13.6.2 tnozaki /* Kanji (containing Gaiji zone) */
343 1.13.6.2 tnozaki /*
344 1.13.6.2 tnozaki * 94^2 zone (contains a part of Gaiji (0xED40 - 0xEEFC)):
345 1.13.6.2 tnozaki * 0x8140 - 0x817E -> 0x2121 - 0x215F
346 1.13.6.2 tnozaki * 0x8180 - 0x819E -> 0x2160 - 0x217E
347 1.13.6.2 tnozaki * 0x819F - 0x81FC -> 0x2221 - 0x227E
348 1.13.6.2 tnozaki *
349 1.13.6.2 tnozaki * 0x8240 - 0x827E -> 0x2321 - 0x235F
350 1.13.6.2 tnozaki * ...
351 1.13.6.2 tnozaki * 0x9F9F - 0x9FFc -> 0x5E21 - 0x5E7E
352 1.13.6.2 tnozaki *
353 1.13.6.2 tnozaki * 0xE040 - 0xE07E -> 0x5F21 - 0x5F5F
354 1.13.6.2 tnozaki * ...
355 1.13.6.2 tnozaki * 0xEF9F - 0xEFFC -> 0x7E21 - 0x7E7E
356 1.13.6.2 tnozaki *
357 1.13.6.2 tnozaki * extended Gaiji zone:
358 1.13.6.2 tnozaki * 0xF040 - 0xFCFC
359 1.13.6.2 tnozaki *
360 1.13.6.2 tnozaki * JIS X0213-plane2:
361 1.13.6.2 tnozaki * 0xF040 - 0xF09E -> 0x2121 - 0x217E
362 1.13.6.2 tnozaki * 0xF140 - 0xF19E -> 0x2321 - 0x237E
363 1.13.6.2 tnozaki * ...
364 1.13.6.2 tnozaki * 0xF240 - 0xF29E -> 0x2521 - 0x257E
365 1.13.6.2 tnozaki *
366 1.13.6.2 tnozaki * 0xF09F - 0xF0FC -> 0x2821 - 0x287E
367 1.13.6.2 tnozaki * 0xF29F - 0xF2FC -> 0x2C21 - 0x2C7E
368 1.13.6.2 tnozaki * ...
369 1.13.6.2 tnozaki * 0xF44F - 0xF49E -> 0x2F21 - 0x2F7E
370 1.13.6.2 tnozaki *
371 1.13.6.2 tnozaki * 0xF49F - 0xF4FC -> 0x6E21 - 0x6E7E
372 1.13.6.2 tnozaki * ...
373 1.13.6.2 tnozaki * 0xFC9F - 0xFCFC -> 0x7E21 - 0x7E7E
374 1.13.6.2 tnozaki */
375 1.13.6.2 tnozaki row = ((_wc_t)wc >> 8) & 0xFF;
376 1.13.6.2 tnozaki col = (_wc_t)wc & 0xFF;
377 1.13.6.2 tnozaki if (!_mskanji1(row) || !_mskanji2(col))
378 1.13.6.2 tnozaki return EILSEQ;
379 1.13.6.2 tnozaki if ((ei->mode & MODE_JIS2004) == 0 || row < 0xF0) {
380 1.13.6.2 tnozaki *csid = 2;
381 1.13.6.2 tnozaki offset = 0x81;
382 1.13.6.2 tnozaki } else {
383 1.13.6.2 tnozaki *csid = 3;
384 1.13.6.2 tnozaki if ((_wc_t)wc <= 0xF49E) {
385 1.13.6.2 tnozaki offset = (_wc_t)wc >= 0xF29F ||
386 1.13.6.2 tnozaki ((_wc_t)wc >= 0xF09F && (_wc_t)wc <= 0xF0FC)
387 1.13.6.2 tnozaki ? 0xED : 0xF0;
388 1.13.6.2 tnozaki } else
389 1.13.6.2 tnozaki offset = 0xCE;
390 1.13.6.2 tnozaki }
391 1.13.6.2 tnozaki row -= offset;
392 1.13.6.2 tnozaki if (row >= 0x5F)
393 1.13.6.2 tnozaki row -= 0x40;
394 1.13.6.2 tnozaki row = row * 2 + 0x21;
395 1.13.6.2 tnozaki col -= 0x1F;
396 1.13.6.2 tnozaki if (col >= 0x61)
397 1.13.6.2 tnozaki col -= 1;
398 1.13.6.2 tnozaki if (col > 0x7E) {
399 1.13.6.2 tnozaki row += 1;
400 1.13.6.2 tnozaki col -= 0x5E;
401 1.13.6.2 tnozaki }
402 1.13.6.2 tnozaki *idx = ((_index_t)row << 8) | col;
403 1.13.6.2 tnozaki }
404 1.13.6.2 tnozaki
405 1.13.6.2 tnozaki return 0;
406 1.13.6.2 tnozaki }
407 1.13.6.2 tnozaki
408 1.13.6.2 tnozaki static __inline int
409 1.13.6.2 tnozaki /*ARGSUSED*/
410 1.13.6.2 tnozaki _citrus_MSKanji_stdenc_cstowc(_MSKanjiEncodingInfo * __restrict ei,
411 1.13.6.2 tnozaki wchar_t * __restrict wc,
412 1.13.6.2 tnozaki _csid_t csid, _index_t idx)
413 1.13.6.2 tnozaki {
414 1.13.6.2 tnozaki u_int32_t row, col;
415 1.13.6.2 tnozaki int offset;
416 1.13.6.2 tnozaki
417 1.13.6.2 tnozaki _DIAGASSERT(wc != NULL);
418 1.13.6.2 tnozaki
419 1.13.6.2 tnozaki switch (csid) {
420 1.13.6.2 tnozaki case 0:
421 1.13.6.2 tnozaki /* ISO-646 */
422 1.13.6.2 tnozaki if (idx >= 0x80)
423 1.13.6.2 tnozaki return EILSEQ;
424 1.13.6.2 tnozaki *wc = (wchar_t)idx;
425 1.13.6.2 tnozaki break;
426 1.13.6.2 tnozaki case 1:
427 1.13.6.2 tnozaki /* kana */
428 1.13.6.2 tnozaki if (idx >= 0x80)
429 1.13.6.2 tnozaki return EILSEQ;
430 1.13.6.2 tnozaki *wc = (wchar_t)idx + 0x80;
431 1.13.6.2 tnozaki break;
432 1.13.6.2 tnozaki case 3:
433 1.13.6.2 tnozaki if ((ei->mode & MODE_JIS2004) == 0)
434 1.13.6.2 tnozaki return EILSEQ;
435 1.13.6.2 tnozaki /*FALLTHROUGH*/
436 1.13.6.2 tnozaki case 2:
437 1.13.6.2 tnozaki /* kanji */
438 1.13.6.2 tnozaki row = (idx >> 8);
439 1.13.6.2 tnozaki if (row < 0x21)
440 1.13.6.2 tnozaki return EILSEQ;
441 1.13.6.2 tnozaki if (csid == 3) {
442 1.13.6.2 tnozaki if (row <= 0x2F)
443 1.13.6.2 tnozaki offset = (row == 0x22 || row >= 0x26)
444 1.13.6.2 tnozaki ? 0xED : 0xF0;
445 1.13.6.2 tnozaki else if (row >= 0x4D && row <= 0x7E)
446 1.13.6.2 tnozaki offset = 0xCE;
447 1.13.6.2 tnozaki else
448 1.13.6.2 tnozaki return EILSEQ;
449 1.13.6.2 tnozaki } else {
450 1.13.6.2 tnozaki if (row > 0x97)
451 1.13.6.2 tnozaki return EILSEQ;
452 1.13.6.2 tnozaki offset = (row < 0x5F) ? 0x81 : 0xC1;
453 1.13.6.2 tnozaki }
454 1.13.6.2 tnozaki col = idx & 0xFF;
455 1.13.6.2 tnozaki if (col < 0x21 || col > 0x7E)
456 1.13.6.2 tnozaki return EILSEQ;
457 1.13.6.2 tnozaki row -= 0x21; col -= 0x21;
458 1.13.6.2 tnozaki if ((row & 1) == 0) {
459 1.13.6.2 tnozaki col += 0x40;
460 1.13.6.2 tnozaki if (col >= 0x7F)
461 1.13.6.2 tnozaki col += 1;
462 1.13.6.2 tnozaki } else
463 1.13.6.2 tnozaki col += 0x9F;
464 1.13.6.2 tnozaki row = row / 2 + offset;
465 1.13.6.2 tnozaki *wc = ((wchar_t)row << 8) | col;
466 1.13.6.2 tnozaki break;
467 1.13.6.2 tnozaki default:
468 1.13.6.2 tnozaki return EILSEQ;
469 1.13.6.2 tnozaki }
470 1.13.6.2 tnozaki
471 1.13.6.2 tnozaki return 0;
472 1.13.6.2 tnozaki }
473 1.13.6.2 tnozaki
474 1.13.6.2 tnozaki static __inline int
475 1.13.6.2 tnozaki /*ARGSUSED*/
476 1.13.6.2 tnozaki _citrus_MSKanji_stdenc_get_state_desc_generic(_MSKanjiEncodingInfo * __restrict ei,
477 1.13.6.2 tnozaki _MSKanjiState * __restrict psenc,
478 1.13.6.2 tnozaki int * __restrict rstate)
479 1.13.6.2 tnozaki {
480 1.13.6.2 tnozaki
481 1.13.6.2 tnozaki if (psenc->chlen == 0)
482 1.13.6.2 tnozaki *rstate = _STDENC_SDGEN_INITIAL;
483 1.13.6.2 tnozaki else
484 1.13.6.2 tnozaki *rstate = _STDENC_SDGEN_INCOMPLETE_CHAR;
485 1.13.6.2 tnozaki
486 1.13.6.2 tnozaki return 0;
487 1.13.6.2 tnozaki }
488 1.13.6.2 tnozaki
489 1.13.6.2 tnozaki static int
490 1.13.6.2 tnozaki /*ARGSUSED*/
491 1.13.6.2 tnozaki _citrus_MSKanji_encoding_module_init(_MSKanjiEncodingInfo * __restrict ei,
492 1.13.6.2 tnozaki const void * __restrict var,
493 1.13.6.2 tnozaki size_t lenvar)
494 1.13.6.2 tnozaki {
495 1.13.6.2 tnozaki const char *p;
496 1.13.6.2 tnozaki
497 1.13.6.2 tnozaki _DIAGASSERT(ei != NULL);
498 1.13.6.2 tnozaki
499 1.13.6.2 tnozaki p = var;
500 1.13.6.2 tnozaki #define MATCH(x, act) \
501 1.13.6.2 tnozaki do { \
502 1.13.6.2 tnozaki if (lenvar >= (sizeof(#x)-1) && \
503 1.13.6.2 tnozaki _bcs_strncasecmp(p, #x, sizeof(#x)-1) == 0) { \
504 1.13.6.2 tnozaki act; \
505 1.13.6.2 tnozaki lenvar -= sizeof(#x)-1; \
506 1.13.6.2 tnozaki p += sizeof(#x)-1; \
507 1.13.6.2 tnozaki } \
508 1.13.6.2 tnozaki } while (/*CONSTCOND*/0)
509 1.13.6.2 tnozaki memset((void *)ei, 0, sizeof(*ei));
510 1.13.6.2 tnozaki while (lenvar > 0) {
511 1.13.6.2 tnozaki switch (_bcs_toupper(*p)) {
512 1.13.6.2 tnozaki case 'J':
513 1.13.6.2 tnozaki MATCH(JIS2004, ei->mode |= MODE_JIS2004);
514 1.13.6.2 tnozaki break;
515 1.13.6.2 tnozaki }
516 1.13.6.2 tnozaki ++p;
517 1.13.6.2 tnozaki --lenvar;
518 1.13.6.2 tnozaki }
519 1.13.6.2 tnozaki
520 1.13.6.2 tnozaki return 0;
521 1.13.6.2 tnozaki }
522 1.13.6.2 tnozaki
523 1.13.6.2 tnozaki static void
524 1.13.6.2 tnozaki _citrus_MSKanji_encoding_module_uninit(_MSKanjiEncodingInfo *ei)
525 1.13.6.2 tnozaki {
526 1.13.6.2 tnozaki }
527 1.13.6.2 tnozaki
528 1.13.6.2 tnozaki /* ----------------------------------------------------------------------
529 1.13.6.2 tnozaki * public interface for ctype
530 1.13.6.2 tnozaki */
531 1.13.6.2 tnozaki
532 1.13.6.2 tnozaki _CITRUS_CTYPE_DECLS(MSKanji);
533 1.13.6.2 tnozaki _CITRUS_CTYPE_DEF_OPS(MSKanji);
534 1.13.6.2 tnozaki
535 1.13.6.2 tnozaki #include "citrus_ctype_template.h"
536 1.13.6.2 tnozaki
537 1.13.6.2 tnozaki /* ----------------------------------------------------------------------
538 1.13.6.2 tnozaki * public interface for stdenc
539 1.13.6.2 tnozaki */
540 1.13.6.2 tnozaki
541 1.13.6.2 tnozaki _CITRUS_STDENC_DECLS(MSKanji);
542 1.13.6.2 tnozaki _CITRUS_STDENC_DEF_OPS(MSKanji);
543 1.13.6.2 tnozaki
544 1.13.6.2 tnozaki #include "citrus_stdenc_template.h"
545