citrus_mskanji.c revision 1.12 1 1.12 tnozaki /* $NetBSD: citrus_mskanji.c,v 1.12 2007/03/05 16:57:06 tnozaki 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 * ja_JP.SJIS locale table for BSD4.4/rune
31 1.1 tshiozak * version 1.0
32 1.1 tshiozak * (C) Sin'ichiro MIYATANI / Phase One, Inc
33 1.1 tshiozak * May 12, 1995
34 1.1 tshiozak *
35 1.1 tshiozak * Redistribution and use in source and binary forms, with or without
36 1.1 tshiozak * modification, are permitted provided that the following conditions
37 1.1 tshiozak * are met:
38 1.1 tshiozak * 1. Redistributions of source code must retain the above copyright
39 1.1 tshiozak * notice, this list of conditions and the following disclaimer.
40 1.1 tshiozak * 2. Redistributions in binary form must reproduce the above copyright
41 1.1 tshiozak * notice, this list of conditions and the following disclaimer in the
42 1.1 tshiozak * documentation and/or other materials provided with the distribution.
43 1.1 tshiozak * 3. All advertising materials mentioning features or use of this software
44 1.1 tshiozak * must display the following acknowledgement:
45 1.1 tshiozak * This product includes software developed by Phase One, Inc.
46 1.1 tshiozak * 4. The name of Phase One, Inc. may be used to endorse or promote products
47 1.1 tshiozak * derived from this software without specific prior written permission.
48 1.1 tshiozak *
49 1.1 tshiozak * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
50 1.1 tshiozak * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
51 1.1 tshiozak * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
52 1.1 tshiozak * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
53 1.1 tshiozak * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
54 1.1 tshiozak * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
55 1.1 tshiozak * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
56 1.1 tshiozak * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
57 1.1 tshiozak * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
58 1.1 tshiozak * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
59 1.1 tshiozak * SUCH DAMAGE.
60 1.1 tshiozak */
61 1.1 tshiozak
62 1.1 tshiozak
63 1.1 tshiozak #include <sys/cdefs.h>
64 1.1 tshiozak #if defined(LIBC_SCCS) && !defined(lint)
65 1.12 tnozaki __RCSID("$NetBSD: citrus_mskanji.c,v 1.12 2007/03/05 16:57:06 tnozaki Exp $");
66 1.1 tshiozak #endif /* LIBC_SCCS and not lint */
67 1.1 tshiozak
68 1.1 tshiozak #include <assert.h>
69 1.1 tshiozak #include <errno.h>
70 1.1 tshiozak #include <string.h>
71 1.1 tshiozak #include <stdio.h>
72 1.1 tshiozak #include <stdlib.h>
73 1.1 tshiozak #include <stddef.h>
74 1.1 tshiozak #include <locale.h>
75 1.1 tshiozak #include <wchar.h>
76 1.1 tshiozak #include <sys/types.h>
77 1.1 tshiozak #include <limits.h>
78 1.6 tshiozak
79 1.6 tshiozak #include "citrus_namespace.h"
80 1.6 tshiozak #include "citrus_types.h"
81 1.12 tnozaki #include "citrus_bcs.h"
82 1.1 tshiozak #include "citrus_module.h"
83 1.1 tshiozak #include "citrus_ctype.h"
84 1.6 tshiozak #include "citrus_stdenc.h"
85 1.1 tshiozak #include "citrus_mskanji.h"
86 1.1 tshiozak
87 1.1 tshiozak
88 1.1 tshiozak /* ----------------------------------------------------------------------
89 1.1 tshiozak * private stuffs used by templates
90 1.1 tshiozak */
91 1.1 tshiozak
92 1.1 tshiozak typedef struct _MSKanjiState {
93 1.1 tshiozak char ch[2];
94 1.1 tshiozak int chlen;
95 1.1 tshiozak } _MSKanjiState;
96 1.1 tshiozak
97 1.1 tshiozak typedef struct {
98 1.12 tnozaki int mode;
99 1.12 tnozaki #define MODE_JIS2004 1
100 1.1 tshiozak } _MSKanjiEncodingInfo;
101 1.1 tshiozak
102 1.1 tshiozak typedef struct {
103 1.1 tshiozak _MSKanjiEncodingInfo ei;
104 1.1 tshiozak struct {
105 1.1 tshiozak /* for future multi-locale facility */
106 1.1 tshiozak _MSKanjiState s_mblen;
107 1.1 tshiozak _MSKanjiState s_mbrlen;
108 1.1 tshiozak _MSKanjiState s_mbrtowc;
109 1.1 tshiozak _MSKanjiState s_mbtowc;
110 1.1 tshiozak _MSKanjiState s_mbsrtowcs;
111 1.1 tshiozak _MSKanjiState s_wcrtomb;
112 1.1 tshiozak _MSKanjiState s_wcsrtombs;
113 1.1 tshiozak _MSKanjiState s_wctomb;
114 1.1 tshiozak } states;
115 1.1 tshiozak } _MSKanjiCTypeInfo;
116 1.1 tshiozak
117 1.1 tshiozak #define _CEI_TO_EI(_cei_) (&(_cei_)->ei)
118 1.1 tshiozak #define _CEI_TO_STATE(_cei_, _func_) (_cei_)->states.s_##_func_
119 1.1 tshiozak
120 1.1 tshiozak #define _FUNCNAME(m) _citrus_MSKanji_##m
121 1.1 tshiozak #define _ENCODING_INFO _MSKanjiEncodingInfo
122 1.1 tshiozak #define _CTYPE_INFO _MSKanjiCTypeInfo
123 1.1 tshiozak #define _ENCODING_STATE _MSKanjiState
124 1.2 yamt #define _ENCODING_MB_CUR_MAX(_ei_) 2
125 1.1 tshiozak #define _ENCODING_IS_STATE_DEPENDENT 0
126 1.4 yamt #define _STATE_NEEDS_EXPLICIT_INIT(_ps_) 0
127 1.1 tshiozak
128 1.1 tshiozak
129 1.1 tshiozak static int
130 1.1 tshiozak _mskanji1(int c)
131 1.1 tshiozak {
132 1.1 tshiozak
133 1.10 tshiozak if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xfc))
134 1.1 tshiozak return 1;
135 1.1 tshiozak else
136 1.1 tshiozak return 0;
137 1.1 tshiozak }
138 1.1 tshiozak
139 1.1 tshiozak static int
140 1.1 tshiozak _mskanji2(int c)
141 1.1 tshiozak {
142 1.1 tshiozak
143 1.1 tshiozak if ((c >= 0x40 && c <= 0x7e) || (c >= 0x80 && c <= 0xfc))
144 1.1 tshiozak return 1;
145 1.1 tshiozak else
146 1.1 tshiozak return 0;
147 1.1 tshiozak }
148 1.1 tshiozak
149 1.1 tshiozak static __inline void
150 1.1 tshiozak /*ARGSUSED*/
151 1.1 tshiozak _citrus_MSKanji_init_state(_MSKanjiEncodingInfo * __restrict ei,
152 1.1 tshiozak _MSKanjiState * __restrict s)
153 1.1 tshiozak {
154 1.12 tnozaki s->chlen = 0;
155 1.1 tshiozak }
156 1.1 tshiozak
157 1.1 tshiozak static __inline void
158 1.1 tshiozak /*ARGSUSED*/
159 1.1 tshiozak _citrus_MSKanji_pack_state(_MSKanjiEncodingInfo * __restrict ei,
160 1.1 tshiozak void * __restrict pspriv,
161 1.1 tshiozak const _MSKanjiState * __restrict s)
162 1.1 tshiozak {
163 1.1 tshiozak memcpy(pspriv, (const void *)s, sizeof(*s));
164 1.1 tshiozak }
165 1.1 tshiozak
166 1.1 tshiozak static __inline void
167 1.1 tshiozak /*ARGSUSED*/
168 1.1 tshiozak _citrus_MSKanji_unpack_state(_MSKanjiEncodingInfo * __restrict ei,
169 1.1 tshiozak _MSKanjiState * __restrict s,
170 1.1 tshiozak const void * __restrict pspriv)
171 1.1 tshiozak {
172 1.1 tshiozak memcpy((void *)s, pspriv, sizeof(*s));
173 1.1 tshiozak }
174 1.1 tshiozak
175 1.1 tshiozak static int
176 1.1 tshiozak /*ARGSUSED*/
177 1.1 tshiozak _citrus_MSKanji_mbrtowc_priv(_MSKanjiEncodingInfo * __restrict ei,
178 1.1 tshiozak wchar_t * __restrict pwc,
179 1.1 tshiozak const char ** __restrict s, size_t n,
180 1.1 tshiozak _MSKanjiState * __restrict psenc,
181 1.1 tshiozak size_t * __restrict nresult)
182 1.1 tshiozak {
183 1.1 tshiozak wchar_t wchar;
184 1.1 tshiozak int len;
185 1.1 tshiozak int chlenbak;
186 1.1 tshiozak const char *s0;
187 1.1 tshiozak
188 1.1 tshiozak _DIAGASSERT(nresult != 0);
189 1.1 tshiozak _DIAGASSERT(ei != NULL);
190 1.1 tshiozak _DIAGASSERT(s != NULL);
191 1.1 tshiozak _DIAGASSERT(psenc != NULL);
192 1.1 tshiozak
193 1.1 tshiozak s0 = *s;
194 1.1 tshiozak
195 1.1 tshiozak if (s0 == NULL) {
196 1.1 tshiozak _citrus_MSKanji_init_state(ei, psenc);
197 1.1 tshiozak *nresult = 0; /* state independent */
198 1.1 tshiozak return (0);
199 1.1 tshiozak }
200 1.1 tshiozak
201 1.1 tshiozak chlenbak = psenc->chlen;
202 1.1 tshiozak
203 1.1 tshiozak /* make sure we have the first byte in the buffer */
204 1.1 tshiozak switch (psenc->chlen) {
205 1.1 tshiozak case 0:
206 1.1 tshiozak if (n < 1)
207 1.1 tshiozak goto restart;
208 1.1 tshiozak psenc->ch[0] = *s0++;
209 1.1 tshiozak psenc->chlen = 1;
210 1.1 tshiozak n--;
211 1.1 tshiozak break;
212 1.1 tshiozak case 1:
213 1.1 tshiozak break;
214 1.1 tshiozak default:
215 1.1 tshiozak /* illegal state */
216 1.1 tshiozak goto encoding_error;
217 1.1 tshiozak }
218 1.1 tshiozak
219 1.1 tshiozak len = _mskanji1(psenc->ch[0] & 0xff) ? 2 : 1;
220 1.1 tshiozak while (psenc->chlen < len) {
221 1.1 tshiozak if (n < 1)
222 1.1 tshiozak goto restart;
223 1.1 tshiozak psenc->ch[psenc->chlen] = *s0++;
224 1.1 tshiozak psenc->chlen++;
225 1.1 tshiozak n--;
226 1.1 tshiozak }
227 1.1 tshiozak
228 1.1 tshiozak *s = s0;
229 1.1 tshiozak
230 1.1 tshiozak switch (len) {
231 1.1 tshiozak case 1:
232 1.1 tshiozak wchar = psenc->ch[0] & 0xff;
233 1.1 tshiozak break;
234 1.1 tshiozak case 2:
235 1.1 tshiozak if (!_mskanji2(psenc->ch[1] & 0xff))
236 1.1 tshiozak goto encoding_error;
237 1.1 tshiozak wchar = ((psenc->ch[0] & 0xff) << 8) | (psenc->ch[1] & 0xff);
238 1.1 tshiozak break;
239 1.1 tshiozak default:
240 1.1 tshiozak /* illegal state */
241 1.1 tshiozak goto encoding_error;
242 1.1 tshiozak }
243 1.1 tshiozak
244 1.1 tshiozak psenc->chlen = 0;
245 1.1 tshiozak
246 1.1 tshiozak if (pwc)
247 1.1 tshiozak *pwc = wchar;
248 1.1 tshiozak
249 1.1 tshiozak if (!wchar)
250 1.1 tshiozak *nresult = 0;
251 1.1 tshiozak else
252 1.1 tshiozak *nresult = len - chlenbak;
253 1.1 tshiozak
254 1.1 tshiozak return (0);
255 1.1 tshiozak
256 1.1 tshiozak encoding_error:
257 1.1 tshiozak psenc->chlen = 0;
258 1.1 tshiozak *nresult = (size_t)-1;
259 1.1 tshiozak return (EILSEQ);
260 1.1 tshiozak
261 1.1 tshiozak restart:
262 1.1 tshiozak *nresult = (size_t)-2;
263 1.1 tshiozak *s = s0;
264 1.1 tshiozak return (0);
265 1.1 tshiozak }
266 1.1 tshiozak
267 1.1 tshiozak
268 1.1 tshiozak static int
269 1.1 tshiozak _citrus_MSKanji_wcrtomb_priv(_MSKanjiEncodingInfo * __restrict ei,
270 1.1 tshiozak char * __restrict s, size_t n, wchar_t wc,
271 1.1 tshiozak _MSKanjiState * __restrict psenc,
272 1.1 tshiozak size_t * __restrict nresult)
273 1.1 tshiozak {
274 1.6 tshiozak int ret;
275 1.1 tshiozak
276 1.1 tshiozak _DIAGASSERT(ei != NULL);
277 1.1 tshiozak _DIAGASSERT(psenc != NULL);
278 1.1 tshiozak _DIAGASSERT(s != NULL);
279 1.6 tshiozak
280 1.1 tshiozak /* check invalid sequence */
281 1.6 tshiozak if (wc & ~0xffff) {
282 1.6 tshiozak ret = EILSEQ;
283 1.6 tshiozak goto err;
284 1.6 tshiozak }
285 1.1 tshiozak
286 1.1 tshiozak if (wc & 0xff00) {
287 1.6 tshiozak if (n < 2) {
288 1.6 tshiozak ret = E2BIG;
289 1.6 tshiozak goto err;
290 1.6 tshiozak }
291 1.1 tshiozak
292 1.1 tshiozak s[0] = (wc >> 8) & 0xff;
293 1.1 tshiozak s[1] = wc & 0xff;
294 1.6 tshiozak if (!_mskanji1(s[0] & 0xff) || !_mskanji2(s[1] & 0xff)) {
295 1.6 tshiozak ret = EILSEQ;
296 1.6 tshiozak goto err;
297 1.6 tshiozak }
298 1.1 tshiozak
299 1.1 tshiozak *nresult = 2;
300 1.6 tshiozak return 0;
301 1.1 tshiozak } else {
302 1.8 yamt if (n < 1) {
303 1.8 yamt ret = E2BIG;
304 1.8 yamt goto err;
305 1.8 yamt }
306 1.8 yamt
307 1.1 tshiozak s[0] = wc & 0xff;
308 1.6 tshiozak if (_mskanji1(s[0] & 0xff)) {
309 1.6 tshiozak ret = EILSEQ;
310 1.6 tshiozak goto err;
311 1.6 tshiozak }
312 1.1 tshiozak
313 1.1 tshiozak *nresult = 1;
314 1.6 tshiozak return 0;
315 1.1 tshiozak }
316 1.1 tshiozak
317 1.6 tshiozak err:
318 1.1 tshiozak *nresult = (size_t)-1;
319 1.6 tshiozak return ret;
320 1.1 tshiozak }
321 1.1 tshiozak
322 1.1 tshiozak
323 1.6 tshiozak static __inline int
324 1.6 tshiozak /*ARGSUSED*/
325 1.6 tshiozak _citrus_MSKanji_stdenc_wctocs(_MSKanjiEncodingInfo * __restrict ei,
326 1.6 tshiozak _csid_t * __restrict csid,
327 1.6 tshiozak _index_t * __restrict idx, wchar_t wc)
328 1.6 tshiozak {
329 1.6 tshiozak _index_t row, col;
330 1.12 tnozaki int offset;
331 1.6 tshiozak
332 1.6 tshiozak _DIAGASSERT(csid != NULL && idx != NULL);
333 1.6 tshiozak
334 1.6 tshiozak if ((_wc_t)wc < 0x80) {
335 1.6 tshiozak /* ISO-646 */
336 1.6 tshiozak *csid = 0;
337 1.6 tshiozak *idx = (_index_t)wc;
338 1.6 tshiozak } else if ((_wc_t)wc < 0x100) {
339 1.6 tshiozak /* KANA */
340 1.6 tshiozak *csid = 1;
341 1.6 tshiozak *idx = (_index_t)wc & 0x7F;
342 1.12 tnozaki } else {
343 1.6 tshiozak /* Kanji (containing Gaiji zone) */
344 1.6 tshiozak /*
345 1.6 tshiozak * 94^2 zone (contains a part of Gaiji (0xED40 - 0xEEFC)):
346 1.6 tshiozak * 0x8140 - 0x817E -> 0x2121 - 0x215F
347 1.6 tshiozak * 0x8180 - 0x819E -> 0x2160 - 0x217E
348 1.6 tshiozak * 0x819F - 0x81FC -> 0x2221 - 0x227E
349 1.6 tshiozak *
350 1.6 tshiozak * 0x8240 - 0x827E -> 0x2321 - 0x235F
351 1.6 tshiozak * ...
352 1.6 tshiozak * 0x9F9F - 0x9FFc -> 0x5E21 - 0x5E7E
353 1.6 tshiozak *
354 1.6 tshiozak * 0xE040 - 0xE07E -> 0x5F21 - 0x5F5F
355 1.6 tshiozak * ...
356 1.6 tshiozak * 0xEF9F - 0xEFFC -> 0x7E21 - 0x7E7E
357 1.6 tshiozak *
358 1.6 tshiozak * extended Gaiji zone:
359 1.6 tshiozak * 0xF040 - 0xFCFC
360 1.12 tnozaki *
361 1.12 tnozaki * JIS X0213-plane2:
362 1.12 tnozaki * 0xF040 - 0xF09E -> 0x2121 - 0x217E
363 1.12 tnozaki * 0xF140 - 0xF19E -> 0x2321 - 0x237E
364 1.12 tnozaki * ...
365 1.12 tnozaki * 0xF240 - 0xF29E -> 0x2521 - 0x257E
366 1.12 tnozaki *
367 1.12 tnozaki * 0xF09F - 0xF0FC -> 0x2821 - 0x287E
368 1.12 tnozaki * 0xF29F - 0xF2FC -> 0x2C21 - 0x2C7E
369 1.12 tnozaki * ...
370 1.12 tnozaki * 0xF44F - 0xF49E -> 0x2F21 - 0x2F7E
371 1.12 tnozaki *
372 1.12 tnozaki * 0xF49F - 0xF4FC -> 0x6E21 - 0x6E7E
373 1.12 tnozaki * ...
374 1.12 tnozaki * 0xFC9F - 0xFCFC -> 0x7E21 - 0x7E7E
375 1.6 tshiozak */
376 1.12 tnozaki row = ((_wc_t)wc >> 8) & 0xFF;
377 1.12 tnozaki col = (_wc_t)wc & 0xFF;
378 1.12 tnozaki if (!_mskanji1(row) || !_mskanji2(col))
379 1.12 tnozaki return EILSEQ;
380 1.12 tnozaki if ((ei->mode & MODE_JIS2004) == 0 || row < 0xF0) {
381 1.12 tnozaki *csid = 2;
382 1.12 tnozaki offset = 0x81;
383 1.12 tnozaki } else {
384 1.12 tnozaki *csid = 3;
385 1.12 tnozaki if ((_wc_t)wc <= 0xF49E) {
386 1.12 tnozaki offset = (_wc_t)wc >= 0xF29F ||
387 1.12 tnozaki ((_wc_t)wc >= 0xF09F && (_wc_t)wc <= 0xF0FC)
388 1.12 tnozaki ? 0xED : 0xF0;
389 1.12 tnozaki } else
390 1.12 tnozaki offset = 0xCE;
391 1.12 tnozaki }
392 1.12 tnozaki row -= offset;
393 1.6 tshiozak if (row >= 0x5F)
394 1.6 tshiozak row -= 0x40;
395 1.6 tshiozak row = row * 2 + 0x21;
396 1.12 tnozaki col -= 0x1F;
397 1.6 tshiozak if (col >= 0x61)
398 1.6 tshiozak col -= 1;
399 1.6 tshiozak if (col > 0x7E) {
400 1.6 tshiozak row += 1;
401 1.6 tshiozak col -= 0x5E;
402 1.6 tshiozak }
403 1.6 tshiozak *idx = ((_index_t)row << 8) | col;
404 1.12 tnozaki }
405 1.6 tshiozak
406 1.6 tshiozak return 0;
407 1.6 tshiozak }
408 1.6 tshiozak
409 1.6 tshiozak static __inline int
410 1.6 tshiozak /*ARGSUSED*/
411 1.6 tshiozak _citrus_MSKanji_stdenc_cstowc(_MSKanjiEncodingInfo * __restrict ei,
412 1.6 tshiozak wchar_t * __restrict wc,
413 1.6 tshiozak _csid_t csid, _index_t idx)
414 1.6 tshiozak {
415 1.6 tshiozak u_int32_t row, col;
416 1.12 tnozaki int offset;
417 1.6 tshiozak
418 1.6 tshiozak _DIAGASSERT(wc != NULL);
419 1.6 tshiozak
420 1.6 tshiozak switch (csid) {
421 1.6 tshiozak case 0:
422 1.6 tshiozak /* ISO-646 */
423 1.6 tshiozak if (idx >= 0x80)
424 1.6 tshiozak return EILSEQ;
425 1.6 tshiozak *wc = (wchar_t)idx;
426 1.6 tshiozak break;
427 1.6 tshiozak case 1:
428 1.6 tshiozak /* kana */
429 1.6 tshiozak if (idx >= 0x80)
430 1.6 tshiozak return EILSEQ;
431 1.6 tshiozak *wc = (wchar_t)idx + 0x80;
432 1.6 tshiozak break;
433 1.12 tnozaki case 3:
434 1.12 tnozaki if ((ei->mode & MODE_JIS2004) == 0)
435 1.12 tnozaki return EILSEQ;
436 1.12 tnozaki /*FALLTHROUGH*/
437 1.6 tshiozak case 2:
438 1.6 tshiozak /* kanji */
439 1.6 tshiozak row = (idx >> 8);
440 1.12 tnozaki if (row < 0x21)
441 1.12 tnozaki return EILSEQ;
442 1.12 tnozaki if (csid == 3) {
443 1.12 tnozaki if (row <= 0x2F)
444 1.12 tnozaki offset = (row == 0x22 || row >= 0x26)
445 1.12 tnozaki ? 0xED : 0xF0;
446 1.12 tnozaki else if (row >= 0x4D && row <= 0x7E)
447 1.12 tnozaki offset = 0xCE;
448 1.12 tnozaki else
449 1.12 tnozaki return EILSEQ;
450 1.12 tnozaki } else {
451 1.12 tnozaki if (row > 0x97)
452 1.12 tnozaki return EILSEQ;
453 1.12 tnozaki offset = (row < 0x5F) ? 0x81 : 0xC1;
454 1.12 tnozaki }
455 1.12 tnozaki col = idx & 0xFF;
456 1.12 tnozaki if (col < 0x21 || col > 0x7E)
457 1.6 tshiozak return EILSEQ;
458 1.6 tshiozak row -= 0x21; col -= 0x21;
459 1.12 tnozaki if ((row & 1) == 0) {
460 1.6 tshiozak col += 0x40;
461 1.12 tnozaki if (col >= 0x7F)
462 1.6 tshiozak col += 1;
463 1.6 tshiozak } else
464 1.6 tshiozak col += 0x9F;
465 1.12 tnozaki row = row / 2 + offset;
466 1.6 tshiozak *wc = ((wchar_t)row << 8) | col;
467 1.6 tshiozak break;
468 1.6 tshiozak default:
469 1.6 tshiozak return EILSEQ;
470 1.6 tshiozak }
471 1.6 tshiozak
472 1.6 tshiozak return 0;
473 1.6 tshiozak }
474 1.6 tshiozak
475 1.11 tshiozak static __inline int
476 1.11 tshiozak /*ARGSUSED*/
477 1.11 tshiozak _citrus_MSKanji_stdenc_get_state_desc_generic(_MSKanjiEncodingInfo * __restrict ei,
478 1.11 tshiozak _MSKanjiState * __restrict psenc,
479 1.11 tshiozak int * __restrict rstate)
480 1.11 tshiozak {
481 1.11 tshiozak
482 1.11 tshiozak if (psenc->chlen == 0)
483 1.11 tshiozak *rstate = _STDENC_SDGEN_INITIAL;
484 1.11 tshiozak else
485 1.11 tshiozak *rstate = _STDENC_SDGEN_INCOMPLETE_CHAR;
486 1.11 tshiozak
487 1.11 tshiozak return 0;
488 1.11 tshiozak }
489 1.11 tshiozak
490 1.1 tshiozak static int
491 1.1 tshiozak /*ARGSUSED*/
492 1.6 tshiozak _citrus_MSKanji_encoding_module_init(_MSKanjiEncodingInfo * __restrict ei,
493 1.6 tshiozak const void * __restrict var,
494 1.6 tshiozak size_t lenvar)
495 1.1 tshiozak {
496 1.12 tnozaki const char *p;
497 1.1 tshiozak
498 1.9 tnozaki _DIAGASSERT(ei != NULL);
499 1.1 tshiozak
500 1.12 tnozaki p = var;
501 1.12 tnozaki #define MATCH(x, act) \
502 1.12 tnozaki do { \
503 1.12 tnozaki if (lenvar >= (sizeof(#x)-1) && \
504 1.12 tnozaki _bcs_strncasecmp(p, #x, sizeof(#x)-1) == 0) { \
505 1.12 tnozaki act; \
506 1.12 tnozaki lenvar -= sizeof(#x)-1; \
507 1.12 tnozaki p += sizeof(#x)-1; \
508 1.12 tnozaki } \
509 1.12 tnozaki } while (/*CONSTCOND*/0)
510 1.12 tnozaki memset((void *)ei, 0, sizeof(*ei));
511 1.12 tnozaki while (lenvar > 0) {
512 1.12 tnozaki switch (_bcs_toupper(*p)) {
513 1.12 tnozaki case 'J':
514 1.12 tnozaki MATCH(JIS2004, ei->mode |= MODE_JIS2004);
515 1.12 tnozaki break;
516 1.12 tnozaki }
517 1.12 tnozaki ++p;
518 1.12 tnozaki --lenvar;
519 1.12 tnozaki }
520 1.12 tnozaki
521 1.12 tnozaki return 0;
522 1.1 tshiozak }
523 1.1 tshiozak
524 1.1 tshiozak static void
525 1.6 tshiozak _citrus_MSKanji_encoding_module_uninit(_MSKanjiEncodingInfo *ei)
526 1.1 tshiozak {
527 1.1 tshiozak }
528 1.1 tshiozak
529 1.1 tshiozak /* ----------------------------------------------------------------------
530 1.1 tshiozak * public interface for ctype
531 1.1 tshiozak */
532 1.1 tshiozak
533 1.1 tshiozak _CITRUS_CTYPE_DECLS(MSKanji);
534 1.1 tshiozak _CITRUS_CTYPE_DEF_OPS(MSKanji);
535 1.1 tshiozak
536 1.1 tshiozak #include "citrus_ctype_template.h"
537 1.6 tshiozak
538 1.6 tshiozak /* ----------------------------------------------------------------------
539 1.6 tshiozak * public interface for stdenc
540 1.6 tshiozak */
541 1.6 tshiozak
542 1.6 tshiozak _CITRUS_STDENC_DECLS(MSKanji);
543 1.6 tshiozak _CITRUS_STDENC_DEF_OPS(MSKanji);
544 1.6 tshiozak
545 1.6 tshiozak #include "citrus_stdenc_template.h"
546