citrus_mskanji.c revision 1.14 1 1.14 joerg /* $NetBSD: citrus_mskanji.c,v 1.14 2013/05/28 16:57:56 joerg 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 1.14 joerg __RCSID("$NetBSD: citrus_mskanji.c,v 1.14 2013/05/28 16:57:56 joerg 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 <wchar.h>
75 1.1 tshiozak #include <sys/types.h>
76 1.1 tshiozak #include <limits.h>
77 1.6 tshiozak
78 1.6 tshiozak #include "citrus_namespace.h"
79 1.6 tshiozak #include "citrus_types.h"
80 1.12 tnozaki #include "citrus_bcs.h"
81 1.1 tshiozak #include "citrus_module.h"
82 1.1 tshiozak #include "citrus_ctype.h"
83 1.6 tshiozak #include "citrus_stdenc.h"
84 1.1 tshiozak #include "citrus_mskanji.h"
85 1.1 tshiozak
86 1.1 tshiozak
87 1.1 tshiozak /* ----------------------------------------------------------------------
88 1.1 tshiozak * private stuffs used by templates
89 1.1 tshiozak */
90 1.1 tshiozak
91 1.1 tshiozak typedef struct _MSKanjiState {
92 1.1 tshiozak char ch[2];
93 1.1 tshiozak int chlen;
94 1.1 tshiozak } _MSKanjiState;
95 1.1 tshiozak
96 1.1 tshiozak typedef struct {
97 1.12 tnozaki int mode;
98 1.12 tnozaki #define MODE_JIS2004 1
99 1.1 tshiozak } _MSKanjiEncodingInfo;
100 1.1 tshiozak
101 1.1 tshiozak typedef struct {
102 1.1 tshiozak _MSKanjiEncodingInfo ei;
103 1.1 tshiozak struct {
104 1.1 tshiozak /* for future multi-locale facility */
105 1.1 tshiozak _MSKanjiState s_mblen;
106 1.1 tshiozak _MSKanjiState s_mbrlen;
107 1.1 tshiozak _MSKanjiState s_mbrtowc;
108 1.1 tshiozak _MSKanjiState s_mbtowc;
109 1.1 tshiozak _MSKanjiState s_mbsrtowcs;
110 1.14 joerg _MSKanjiState s_mbsnrtowcs;
111 1.1 tshiozak _MSKanjiState s_wcrtomb;
112 1.1 tshiozak _MSKanjiState s_wcsrtombs;
113 1.14 joerg _MSKanjiState s_wcsnrtombs;
114 1.1 tshiozak _MSKanjiState s_wctomb;
115 1.1 tshiozak } states;
116 1.1 tshiozak } _MSKanjiCTypeInfo;
117 1.1 tshiozak
118 1.1 tshiozak #define _CEI_TO_EI(_cei_) (&(_cei_)->ei)
119 1.1 tshiozak #define _CEI_TO_STATE(_cei_, _func_) (_cei_)->states.s_##_func_
120 1.1 tshiozak
121 1.1 tshiozak #define _FUNCNAME(m) _citrus_MSKanji_##m
122 1.1 tshiozak #define _ENCODING_INFO _MSKanjiEncodingInfo
123 1.1 tshiozak #define _CTYPE_INFO _MSKanjiCTypeInfo
124 1.1 tshiozak #define _ENCODING_STATE _MSKanjiState
125 1.2 yamt #define _ENCODING_MB_CUR_MAX(_ei_) 2
126 1.1 tshiozak #define _ENCODING_IS_STATE_DEPENDENT 0
127 1.4 yamt #define _STATE_NEEDS_EXPLICIT_INIT(_ps_) 0
128 1.1 tshiozak
129 1.1 tshiozak
130 1.1 tshiozak static int
131 1.1 tshiozak _mskanji1(int c)
132 1.1 tshiozak {
133 1.1 tshiozak
134 1.10 tshiozak if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xfc))
135 1.1 tshiozak return 1;
136 1.1 tshiozak else
137 1.1 tshiozak return 0;
138 1.1 tshiozak }
139 1.1 tshiozak
140 1.1 tshiozak static int
141 1.1 tshiozak _mskanji2(int c)
142 1.1 tshiozak {
143 1.1 tshiozak
144 1.1 tshiozak if ((c >= 0x40 && c <= 0x7e) || (c >= 0x80 && c <= 0xfc))
145 1.1 tshiozak return 1;
146 1.1 tshiozak else
147 1.1 tshiozak return 0;
148 1.1 tshiozak }
149 1.1 tshiozak
150 1.1 tshiozak static __inline void
151 1.1 tshiozak /*ARGSUSED*/
152 1.1 tshiozak _citrus_MSKanji_init_state(_MSKanjiEncodingInfo * __restrict ei,
153 1.1 tshiozak _MSKanjiState * __restrict s)
154 1.1 tshiozak {
155 1.12 tnozaki s->chlen = 0;
156 1.1 tshiozak }
157 1.1 tshiozak
158 1.1 tshiozak static __inline void
159 1.1 tshiozak /*ARGSUSED*/
160 1.1 tshiozak _citrus_MSKanji_pack_state(_MSKanjiEncodingInfo * __restrict ei,
161 1.1 tshiozak void * __restrict pspriv,
162 1.1 tshiozak const _MSKanjiState * __restrict s)
163 1.1 tshiozak {
164 1.1 tshiozak memcpy(pspriv, (const void *)s, sizeof(*s));
165 1.1 tshiozak }
166 1.1 tshiozak
167 1.1 tshiozak static __inline void
168 1.1 tshiozak /*ARGSUSED*/
169 1.1 tshiozak _citrus_MSKanji_unpack_state(_MSKanjiEncodingInfo * __restrict ei,
170 1.1 tshiozak _MSKanjiState * __restrict s,
171 1.1 tshiozak const void * __restrict pspriv)
172 1.1 tshiozak {
173 1.1 tshiozak memcpy((void *)s, pspriv, sizeof(*s));
174 1.1 tshiozak }
175 1.1 tshiozak
176 1.1 tshiozak static int
177 1.1 tshiozak /*ARGSUSED*/
178 1.1 tshiozak _citrus_MSKanji_mbrtowc_priv(_MSKanjiEncodingInfo * __restrict ei,
179 1.1 tshiozak wchar_t * __restrict pwc,
180 1.1 tshiozak const char ** __restrict s, size_t n,
181 1.1 tshiozak _MSKanjiState * __restrict psenc,
182 1.1 tshiozak size_t * __restrict nresult)
183 1.1 tshiozak {
184 1.1 tshiozak wchar_t wchar;
185 1.1 tshiozak int len;
186 1.1 tshiozak int chlenbak;
187 1.1 tshiozak const char *s0;
188 1.1 tshiozak
189 1.1 tshiozak _DIAGASSERT(nresult != 0);
190 1.1 tshiozak _DIAGASSERT(ei != NULL);
191 1.1 tshiozak _DIAGASSERT(s != NULL);
192 1.1 tshiozak _DIAGASSERT(psenc != NULL);
193 1.1 tshiozak
194 1.1 tshiozak s0 = *s;
195 1.1 tshiozak
196 1.1 tshiozak if (s0 == NULL) {
197 1.1 tshiozak _citrus_MSKanji_init_state(ei, psenc);
198 1.1 tshiozak *nresult = 0; /* state independent */
199 1.1 tshiozak return (0);
200 1.1 tshiozak }
201 1.1 tshiozak
202 1.1 tshiozak chlenbak = psenc->chlen;
203 1.1 tshiozak
204 1.1 tshiozak /* make sure we have the first byte in the buffer */
205 1.1 tshiozak switch (psenc->chlen) {
206 1.1 tshiozak case 0:
207 1.1 tshiozak if (n < 1)
208 1.1 tshiozak goto restart;
209 1.1 tshiozak psenc->ch[0] = *s0++;
210 1.1 tshiozak psenc->chlen = 1;
211 1.1 tshiozak n--;
212 1.1 tshiozak break;
213 1.1 tshiozak case 1:
214 1.1 tshiozak break;
215 1.1 tshiozak default:
216 1.1 tshiozak /* illegal state */
217 1.1 tshiozak goto encoding_error;
218 1.1 tshiozak }
219 1.1 tshiozak
220 1.1 tshiozak len = _mskanji1(psenc->ch[0] & 0xff) ? 2 : 1;
221 1.1 tshiozak while (psenc->chlen < len) {
222 1.1 tshiozak if (n < 1)
223 1.1 tshiozak goto restart;
224 1.1 tshiozak psenc->ch[psenc->chlen] = *s0++;
225 1.1 tshiozak psenc->chlen++;
226 1.1 tshiozak n--;
227 1.1 tshiozak }
228 1.1 tshiozak
229 1.1 tshiozak *s = s0;
230 1.1 tshiozak
231 1.1 tshiozak switch (len) {
232 1.1 tshiozak case 1:
233 1.1 tshiozak wchar = psenc->ch[0] & 0xff;
234 1.1 tshiozak break;
235 1.1 tshiozak case 2:
236 1.1 tshiozak if (!_mskanji2(psenc->ch[1] & 0xff))
237 1.1 tshiozak goto encoding_error;
238 1.1 tshiozak wchar = ((psenc->ch[0] & 0xff) << 8) | (psenc->ch[1] & 0xff);
239 1.1 tshiozak break;
240 1.1 tshiozak default:
241 1.1 tshiozak /* illegal state */
242 1.1 tshiozak goto encoding_error;
243 1.1 tshiozak }
244 1.1 tshiozak
245 1.1 tshiozak psenc->chlen = 0;
246 1.1 tshiozak
247 1.1 tshiozak if (pwc)
248 1.1 tshiozak *pwc = wchar;
249 1.1 tshiozak
250 1.1 tshiozak if (!wchar)
251 1.1 tshiozak *nresult = 0;
252 1.1 tshiozak else
253 1.1 tshiozak *nresult = len - chlenbak;
254 1.1 tshiozak
255 1.1 tshiozak return (0);
256 1.1 tshiozak
257 1.1 tshiozak encoding_error:
258 1.1 tshiozak psenc->chlen = 0;
259 1.1 tshiozak *nresult = (size_t)-1;
260 1.1 tshiozak return (EILSEQ);
261 1.1 tshiozak
262 1.1 tshiozak restart:
263 1.1 tshiozak *nresult = (size_t)-2;
264 1.1 tshiozak *s = s0;
265 1.1 tshiozak return (0);
266 1.1 tshiozak }
267 1.1 tshiozak
268 1.1 tshiozak
269 1.1 tshiozak static int
270 1.1 tshiozak _citrus_MSKanji_wcrtomb_priv(_MSKanjiEncodingInfo * __restrict ei,
271 1.1 tshiozak char * __restrict s, size_t n, wchar_t wc,
272 1.1 tshiozak _MSKanjiState * __restrict psenc,
273 1.1 tshiozak size_t * __restrict nresult)
274 1.1 tshiozak {
275 1.6 tshiozak int ret;
276 1.1 tshiozak
277 1.1 tshiozak _DIAGASSERT(ei != NULL);
278 1.1 tshiozak _DIAGASSERT(psenc != NULL);
279 1.1 tshiozak _DIAGASSERT(s != NULL);
280 1.6 tshiozak
281 1.1 tshiozak /* check invalid sequence */
282 1.6 tshiozak if (wc & ~0xffff) {
283 1.6 tshiozak ret = EILSEQ;
284 1.6 tshiozak goto err;
285 1.6 tshiozak }
286 1.1 tshiozak
287 1.1 tshiozak if (wc & 0xff00) {
288 1.6 tshiozak if (n < 2) {
289 1.6 tshiozak ret = E2BIG;
290 1.6 tshiozak goto err;
291 1.6 tshiozak }
292 1.1 tshiozak
293 1.1 tshiozak s[0] = (wc >> 8) & 0xff;
294 1.1 tshiozak s[1] = wc & 0xff;
295 1.6 tshiozak if (!_mskanji1(s[0] & 0xff) || !_mskanji2(s[1] & 0xff)) {
296 1.6 tshiozak ret = EILSEQ;
297 1.6 tshiozak goto err;
298 1.6 tshiozak }
299 1.1 tshiozak
300 1.1 tshiozak *nresult = 2;
301 1.6 tshiozak return 0;
302 1.1 tshiozak } else {
303 1.8 yamt if (n < 1) {
304 1.8 yamt ret = E2BIG;
305 1.8 yamt goto err;
306 1.8 yamt }
307 1.8 yamt
308 1.1 tshiozak s[0] = wc & 0xff;
309 1.6 tshiozak if (_mskanji1(s[0] & 0xff)) {
310 1.6 tshiozak ret = EILSEQ;
311 1.6 tshiozak goto err;
312 1.6 tshiozak }
313 1.1 tshiozak
314 1.1 tshiozak *nresult = 1;
315 1.6 tshiozak return 0;
316 1.1 tshiozak }
317 1.1 tshiozak
318 1.6 tshiozak err:
319 1.1 tshiozak *nresult = (size_t)-1;
320 1.6 tshiozak return ret;
321 1.1 tshiozak }
322 1.1 tshiozak
323 1.1 tshiozak
324 1.6 tshiozak static __inline int
325 1.6 tshiozak /*ARGSUSED*/
326 1.6 tshiozak _citrus_MSKanji_stdenc_wctocs(_MSKanjiEncodingInfo * __restrict ei,
327 1.6 tshiozak _csid_t * __restrict csid,
328 1.6 tshiozak _index_t * __restrict idx, wchar_t wc)
329 1.6 tshiozak {
330 1.6 tshiozak _index_t row, col;
331 1.12 tnozaki int offset;
332 1.6 tshiozak
333 1.6 tshiozak _DIAGASSERT(csid != NULL && idx != NULL);
334 1.6 tshiozak
335 1.6 tshiozak if ((_wc_t)wc < 0x80) {
336 1.6 tshiozak /* ISO-646 */
337 1.6 tshiozak *csid = 0;
338 1.6 tshiozak *idx = (_index_t)wc;
339 1.6 tshiozak } else if ((_wc_t)wc < 0x100) {
340 1.6 tshiozak /* KANA */
341 1.6 tshiozak *csid = 1;
342 1.6 tshiozak *idx = (_index_t)wc & 0x7F;
343 1.12 tnozaki } else {
344 1.6 tshiozak /* Kanji (containing Gaiji zone) */
345 1.6 tshiozak /*
346 1.6 tshiozak * 94^2 zone (contains a part of Gaiji (0xED40 - 0xEEFC)):
347 1.6 tshiozak * 0x8140 - 0x817E -> 0x2121 - 0x215F
348 1.6 tshiozak * 0x8180 - 0x819E -> 0x2160 - 0x217E
349 1.6 tshiozak * 0x819F - 0x81FC -> 0x2221 - 0x227E
350 1.6 tshiozak *
351 1.6 tshiozak * 0x8240 - 0x827E -> 0x2321 - 0x235F
352 1.6 tshiozak * ...
353 1.6 tshiozak * 0x9F9F - 0x9FFc -> 0x5E21 - 0x5E7E
354 1.6 tshiozak *
355 1.6 tshiozak * 0xE040 - 0xE07E -> 0x5F21 - 0x5F5F
356 1.6 tshiozak * ...
357 1.6 tshiozak * 0xEF9F - 0xEFFC -> 0x7E21 - 0x7E7E
358 1.6 tshiozak *
359 1.6 tshiozak * extended Gaiji zone:
360 1.6 tshiozak * 0xF040 - 0xFCFC
361 1.12 tnozaki *
362 1.12 tnozaki * JIS X0213-plane2:
363 1.12 tnozaki * 0xF040 - 0xF09E -> 0x2121 - 0x217E
364 1.12 tnozaki * 0xF140 - 0xF19E -> 0x2321 - 0x237E
365 1.12 tnozaki * ...
366 1.12 tnozaki * 0xF240 - 0xF29E -> 0x2521 - 0x257E
367 1.12 tnozaki *
368 1.12 tnozaki * 0xF09F - 0xF0FC -> 0x2821 - 0x287E
369 1.12 tnozaki * 0xF29F - 0xF2FC -> 0x2C21 - 0x2C7E
370 1.12 tnozaki * ...
371 1.12 tnozaki * 0xF44F - 0xF49E -> 0x2F21 - 0x2F7E
372 1.12 tnozaki *
373 1.12 tnozaki * 0xF49F - 0xF4FC -> 0x6E21 - 0x6E7E
374 1.12 tnozaki * ...
375 1.12 tnozaki * 0xFC9F - 0xFCFC -> 0x7E21 - 0x7E7E
376 1.6 tshiozak */
377 1.12 tnozaki row = ((_wc_t)wc >> 8) & 0xFF;
378 1.12 tnozaki col = (_wc_t)wc & 0xFF;
379 1.12 tnozaki if (!_mskanji1(row) || !_mskanji2(col))
380 1.12 tnozaki return EILSEQ;
381 1.12 tnozaki if ((ei->mode & MODE_JIS2004) == 0 || row < 0xF0) {
382 1.12 tnozaki *csid = 2;
383 1.12 tnozaki offset = 0x81;
384 1.12 tnozaki } else {
385 1.12 tnozaki *csid = 3;
386 1.12 tnozaki if ((_wc_t)wc <= 0xF49E) {
387 1.12 tnozaki offset = (_wc_t)wc >= 0xF29F ||
388 1.12 tnozaki ((_wc_t)wc >= 0xF09F && (_wc_t)wc <= 0xF0FC)
389 1.12 tnozaki ? 0xED : 0xF0;
390 1.12 tnozaki } else
391 1.12 tnozaki offset = 0xCE;
392 1.12 tnozaki }
393 1.12 tnozaki row -= offset;
394 1.6 tshiozak if (row >= 0x5F)
395 1.6 tshiozak row -= 0x40;
396 1.6 tshiozak row = row * 2 + 0x21;
397 1.12 tnozaki col -= 0x1F;
398 1.6 tshiozak if (col >= 0x61)
399 1.6 tshiozak col -= 1;
400 1.6 tshiozak if (col > 0x7E) {
401 1.6 tshiozak row += 1;
402 1.6 tshiozak col -= 0x5E;
403 1.6 tshiozak }
404 1.6 tshiozak *idx = ((_index_t)row << 8) | col;
405 1.12 tnozaki }
406 1.6 tshiozak
407 1.6 tshiozak return 0;
408 1.6 tshiozak }
409 1.6 tshiozak
410 1.6 tshiozak static __inline int
411 1.6 tshiozak /*ARGSUSED*/
412 1.6 tshiozak _citrus_MSKanji_stdenc_cstowc(_MSKanjiEncodingInfo * __restrict ei,
413 1.6 tshiozak wchar_t * __restrict wc,
414 1.6 tshiozak _csid_t csid, _index_t idx)
415 1.6 tshiozak {
416 1.6 tshiozak u_int32_t row, col;
417 1.12 tnozaki int offset;
418 1.6 tshiozak
419 1.6 tshiozak _DIAGASSERT(wc != NULL);
420 1.6 tshiozak
421 1.6 tshiozak switch (csid) {
422 1.6 tshiozak case 0:
423 1.6 tshiozak /* ISO-646 */
424 1.6 tshiozak if (idx >= 0x80)
425 1.6 tshiozak return EILSEQ;
426 1.6 tshiozak *wc = (wchar_t)idx;
427 1.6 tshiozak break;
428 1.6 tshiozak case 1:
429 1.6 tshiozak /* kana */
430 1.6 tshiozak if (idx >= 0x80)
431 1.6 tshiozak return EILSEQ;
432 1.6 tshiozak *wc = (wchar_t)idx + 0x80;
433 1.6 tshiozak break;
434 1.12 tnozaki case 3:
435 1.12 tnozaki if ((ei->mode & MODE_JIS2004) == 0)
436 1.12 tnozaki return EILSEQ;
437 1.12 tnozaki /*FALLTHROUGH*/
438 1.6 tshiozak case 2:
439 1.6 tshiozak /* kanji */
440 1.6 tshiozak row = (idx >> 8);
441 1.12 tnozaki if (row < 0x21)
442 1.12 tnozaki return EILSEQ;
443 1.12 tnozaki if (csid == 3) {
444 1.12 tnozaki if (row <= 0x2F)
445 1.12 tnozaki offset = (row == 0x22 || row >= 0x26)
446 1.12 tnozaki ? 0xED : 0xF0;
447 1.12 tnozaki else if (row >= 0x4D && row <= 0x7E)
448 1.12 tnozaki offset = 0xCE;
449 1.12 tnozaki else
450 1.12 tnozaki return EILSEQ;
451 1.12 tnozaki } else {
452 1.12 tnozaki if (row > 0x97)
453 1.12 tnozaki return EILSEQ;
454 1.12 tnozaki offset = (row < 0x5F) ? 0x81 : 0xC1;
455 1.12 tnozaki }
456 1.12 tnozaki col = idx & 0xFF;
457 1.12 tnozaki if (col < 0x21 || col > 0x7E)
458 1.6 tshiozak return EILSEQ;
459 1.6 tshiozak row -= 0x21; col -= 0x21;
460 1.12 tnozaki if ((row & 1) == 0) {
461 1.6 tshiozak col += 0x40;
462 1.12 tnozaki if (col >= 0x7F)
463 1.6 tshiozak col += 1;
464 1.6 tshiozak } else
465 1.6 tshiozak col += 0x9F;
466 1.12 tnozaki row = row / 2 + offset;
467 1.6 tshiozak *wc = ((wchar_t)row << 8) | col;
468 1.6 tshiozak break;
469 1.6 tshiozak default:
470 1.6 tshiozak return EILSEQ;
471 1.6 tshiozak }
472 1.6 tshiozak
473 1.6 tshiozak return 0;
474 1.6 tshiozak }
475 1.6 tshiozak
476 1.11 tshiozak static __inline int
477 1.11 tshiozak /*ARGSUSED*/
478 1.11 tshiozak _citrus_MSKanji_stdenc_get_state_desc_generic(_MSKanjiEncodingInfo * __restrict ei,
479 1.11 tshiozak _MSKanjiState * __restrict psenc,
480 1.11 tshiozak int * __restrict rstate)
481 1.11 tshiozak {
482 1.11 tshiozak
483 1.11 tshiozak if (psenc->chlen == 0)
484 1.11 tshiozak *rstate = _STDENC_SDGEN_INITIAL;
485 1.11 tshiozak else
486 1.11 tshiozak *rstate = _STDENC_SDGEN_INCOMPLETE_CHAR;
487 1.11 tshiozak
488 1.11 tshiozak return 0;
489 1.11 tshiozak }
490 1.11 tshiozak
491 1.1 tshiozak static int
492 1.1 tshiozak /*ARGSUSED*/
493 1.6 tshiozak _citrus_MSKanji_encoding_module_init(_MSKanjiEncodingInfo * __restrict ei,
494 1.6 tshiozak const void * __restrict var,
495 1.6 tshiozak size_t lenvar)
496 1.1 tshiozak {
497 1.12 tnozaki const char *p;
498 1.1 tshiozak
499 1.9 tnozaki _DIAGASSERT(ei != NULL);
500 1.1 tshiozak
501 1.12 tnozaki p = var;
502 1.12 tnozaki #define MATCH(x, act) \
503 1.12 tnozaki do { \
504 1.12 tnozaki if (lenvar >= (sizeof(#x)-1) && \
505 1.12 tnozaki _bcs_strncasecmp(p, #x, sizeof(#x)-1) == 0) { \
506 1.12 tnozaki act; \
507 1.12 tnozaki lenvar -= sizeof(#x)-1; \
508 1.12 tnozaki p += sizeof(#x)-1; \
509 1.12 tnozaki } \
510 1.12 tnozaki } while (/*CONSTCOND*/0)
511 1.12 tnozaki memset((void *)ei, 0, sizeof(*ei));
512 1.12 tnozaki while (lenvar > 0) {
513 1.12 tnozaki switch (_bcs_toupper(*p)) {
514 1.12 tnozaki case 'J':
515 1.12 tnozaki MATCH(JIS2004, ei->mode |= MODE_JIS2004);
516 1.12 tnozaki break;
517 1.12 tnozaki }
518 1.12 tnozaki ++p;
519 1.12 tnozaki --lenvar;
520 1.12 tnozaki }
521 1.12 tnozaki
522 1.12 tnozaki return 0;
523 1.1 tshiozak }
524 1.1 tshiozak
525 1.1 tshiozak static void
526 1.6 tshiozak _citrus_MSKanji_encoding_module_uninit(_MSKanjiEncodingInfo *ei)
527 1.1 tshiozak {
528 1.1 tshiozak }
529 1.1 tshiozak
530 1.1 tshiozak /* ----------------------------------------------------------------------
531 1.1 tshiozak * public interface for ctype
532 1.1 tshiozak */
533 1.1 tshiozak
534 1.1 tshiozak _CITRUS_CTYPE_DECLS(MSKanji);
535 1.1 tshiozak _CITRUS_CTYPE_DEF_OPS(MSKanji);
536 1.1 tshiozak
537 1.1 tshiozak #include "citrus_ctype_template.h"
538 1.6 tshiozak
539 1.6 tshiozak /* ----------------------------------------------------------------------
540 1.6 tshiozak * public interface for stdenc
541 1.6 tshiozak */
542 1.6 tshiozak
543 1.6 tshiozak _CITRUS_STDENC_DECLS(MSKanji);
544 1.6 tshiozak _CITRUS_STDENC_DEF_OPS(MSKanji);
545 1.6 tshiozak
546 1.6 tshiozak #include "citrus_stdenc_template.h"
547