citrus_utf7.c revision 1.4 1 1.4 christos /* $NetBSD: citrus_utf7.c,v 1.4 2006/03/19 01:55:48 christos Exp $ */
2 1.1 tnozaki
3 1.1 tnozaki /*-
4 1.1 tnozaki * Copyright (c)2004, 2005 Citrus Project,
5 1.1 tnozaki * All rights reserved.
6 1.1 tnozaki *
7 1.1 tnozaki * Redistribution and use in source and binary forms, with or without
8 1.1 tnozaki * modification, are permitted provided that the following conditions
9 1.1 tnozaki * are met:
10 1.1 tnozaki * 1. Redistributions of source code must retain the above copyright
11 1.1 tnozaki * notice, this list of conditions and the following disclaimer.
12 1.1 tnozaki * 2. Redistributions in binary form must reproduce the above copyright
13 1.1 tnozaki * notice, this list of conditions and the following disclaimer in the
14 1.1 tnozaki * documentation and/or other materials provided with the distribution.
15 1.1 tnozaki *
16 1.1 tnozaki * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
17 1.1 tnozaki * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18 1.1 tnozaki * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19 1.1 tnozaki * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
20 1.1 tnozaki * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21 1.1 tnozaki * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22 1.1 tnozaki * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23 1.1 tnozaki * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24 1.1 tnozaki * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25 1.1 tnozaki * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26 1.1 tnozaki * SUCH DAMAGE.
27 1.1 tnozaki *
28 1.1 tnozaki */
29 1.1 tnozaki
30 1.1 tnozaki #include <sys/cdefs.h>
31 1.1 tnozaki #if defined(LIB_SCCS) && !defined(lint)
32 1.4 christos __RCSID("$NetBSD: citrus_utf7.c,v 1.4 2006/03/19 01:55:48 christos Exp $");
33 1.1 tnozaki #endif /* LIB_SCCS and not lint */
34 1.1 tnozaki
35 1.1 tnozaki #include <assert.h>
36 1.1 tnozaki #include <errno.h>
37 1.1 tnozaki #include <string.h>
38 1.1 tnozaki #include <stdio.h>
39 1.1 tnozaki #include <stdint.h>
40 1.1 tnozaki #include <stdlib.h>
41 1.1 tnozaki #include <limits.h>
42 1.1 tnozaki #include <wchar.h>
43 1.1 tnozaki
44 1.1 tnozaki #include "citrus_namespace.h"
45 1.1 tnozaki #include "citrus_types.h"
46 1.1 tnozaki #include "citrus_module.h"
47 1.1 tnozaki #include "citrus_ctype.h"
48 1.1 tnozaki #include "citrus_stdenc.h"
49 1.1 tnozaki #include "citrus_utf7.h"
50 1.1 tnozaki
51 1.1 tnozaki /* ----------------------------------------------------------------------
52 1.1 tnozaki * private stuffs used by templates
53 1.1 tnozaki */
54 1.1 tnozaki
55 1.1 tnozaki typedef struct {
56 1.1 tnozaki uint16_t cell[0x80];
57 1.1 tnozaki #define EI_MASK UINT16_C(0xff)
58 1.1 tnozaki #define EI_DIRECT UINT16_C(0x100)
59 1.1 tnozaki #define EI_OPTION UINT16_C(0x200)
60 1.1 tnozaki #define EI_SPACE UINT16_C(0x400)
61 1.1 tnozaki } _UTF7EncodingInfo;
62 1.1 tnozaki
63 1.1 tnozaki typedef struct {
64 1.1 tnozaki unsigned int
65 1.1 tnozaki mode: 1, /* whether base64 mode */
66 1.1 tnozaki bits: 4, /* need to hold 0 - 15 */
67 1.1 tnozaki cache: 22, /* 22 = BASE64_BIT + UTF16_BIT */
68 1.2 tnozaki surrogate: 1; /* whether surrogate pair or not */
69 1.2 tnozaki int chlen;
70 1.1 tnozaki char ch[4]; /* BASE64_IN, 3 * 6 = 18, most closed to UTF16_BIT */
71 1.1 tnozaki } _UTF7State;
72 1.1 tnozaki
73 1.1 tnozaki typedef struct {
74 1.1 tnozaki _UTF7EncodingInfo ei;
75 1.1 tnozaki struct {
76 1.1 tnozaki /* for future multi-locale facility */
77 1.1 tnozaki _UTF7State s_mblen;
78 1.1 tnozaki _UTF7State s_mbrlen;
79 1.1 tnozaki _UTF7State s_mbrtowc;
80 1.1 tnozaki _UTF7State s_mbtowc;
81 1.1 tnozaki _UTF7State s_mbsrtowcs;
82 1.1 tnozaki _UTF7State s_wcrtomb;
83 1.1 tnozaki _UTF7State s_wcsrtombs;
84 1.1 tnozaki _UTF7State s_wctomb;
85 1.1 tnozaki } states;
86 1.1 tnozaki } _UTF7CTypeInfo;
87 1.1 tnozaki
88 1.1 tnozaki #define _CEI_TO_EI(_cei_) (&(_cei_)->ei)
89 1.1 tnozaki #define _CEI_TO_STATE(_cei_, _func_) (_cei_)->states.s_##_func_
90 1.1 tnozaki
91 1.1 tnozaki #define _FUNCNAME(m) _citrus_UTF7_##m
92 1.1 tnozaki #define _ENCODING_INFO _UTF7EncodingInfo
93 1.1 tnozaki #define _CTYPE_INFO _UTF7CTypeInfo
94 1.1 tnozaki #define _ENCODING_STATE _UTF7State
95 1.1 tnozaki #define _ENCODING_MB_CUR_MAX(_ei_) 4
96 1.1 tnozaki #define _ENCODING_IS_STATE_DEPENDENT 1
97 1.1 tnozaki #define _STATE_NEEDS_EXPLICIT_INIT(_ps_) 0
98 1.1 tnozaki
99 1.1 tnozaki static __inline void
100 1.1 tnozaki /*ARGSUSED*/
101 1.1 tnozaki _citrus_UTF7_init_state(_UTF7EncodingInfo * __restrict ei,
102 1.1 tnozaki _UTF7State * __restrict s)
103 1.1 tnozaki {
104 1.1 tnozaki /* ei appears to be unused */
105 1.1 tnozaki _DIAGASSERT(s != NULL);
106 1.1 tnozaki
107 1.1 tnozaki memset((void *)s, 0, sizeof(*s));
108 1.1 tnozaki }
109 1.1 tnozaki
110 1.1 tnozaki static __inline void
111 1.1 tnozaki /*ARGSUSED*/
112 1.1 tnozaki _citrus_UTF7_pack_state(_UTF7EncodingInfo * __restrict ei,
113 1.1 tnozaki void *__restrict pspriv, const _UTF7State * __restrict s)
114 1.1 tnozaki {
115 1.1 tnozaki /* ei seem to be unused */
116 1.1 tnozaki _DIAGASSERT(pspriv != NULL);
117 1.1 tnozaki _DIAGASSERT(s != NULL);
118 1.1 tnozaki
119 1.1 tnozaki memcpy(pspriv, (const void *)s, sizeof(*s));
120 1.1 tnozaki }
121 1.1 tnozaki
122 1.1 tnozaki static __inline void
123 1.1 tnozaki /*ARGSUSED*/
124 1.1 tnozaki _citrus_UTF7_unpack_state(_UTF7EncodingInfo * __restrict ei,
125 1.1 tnozaki _UTF7State * __restrict s, const void * __restrict pspriv)
126 1.1 tnozaki {
127 1.1 tnozaki /* ei seem to be unused */
128 1.1 tnozaki _DIAGASSERT(s != NULL);
129 1.1 tnozaki _DIAGASSERT(pspriv != NULL);
130 1.1 tnozaki
131 1.1 tnozaki memcpy((void *)s, pspriv, sizeof(*s));
132 1.1 tnozaki }
133 1.1 tnozaki
134 1.1 tnozaki static const char base64[] =
135 1.1 tnozaki "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
136 1.1 tnozaki "abcdefghijklmnopqrstuvwxyz"
137 1.1 tnozaki "0123456789+/";
138 1.1 tnozaki
139 1.1 tnozaki static const char direct[] =
140 1.1 tnozaki "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
141 1.1 tnozaki "abcdefghijklmnopqrstuvwxyz"
142 1.1 tnozaki "0123456789(),-./:?";
143 1.1 tnozaki
144 1.1 tnozaki static const char option[] = "!\"#$%&';<=>@[]^_`{|}";
145 1.1 tnozaki static const char spaces[] = " \t\r\n";
146 1.1 tnozaki
147 1.1 tnozaki #define BASE64_BIT 6
148 1.1 tnozaki #define UTF16_BIT 16
149 1.1 tnozaki
150 1.1 tnozaki #define BASE64_MAX 0x3f
151 1.1 tnozaki #define UTF16_MAX UINT16_C(0xffff)
152 1.1 tnozaki #define UTF32_MAX UINT32_C(0x10ffff)
153 1.1 tnozaki
154 1.1 tnozaki #define BASE64_IN '+'
155 1.1 tnozaki #define BASE64_OUT '-'
156 1.1 tnozaki
157 1.1 tnozaki #define SHIFT7BIT(c) ((c) >> 7)
158 1.1 tnozaki #define ISSPECIAL(c) ((c) == '\0' || (c) == BASE64_IN)
159 1.1 tnozaki
160 1.1 tnozaki #define FINDLEN(ei, c) \
161 1.1 tnozaki (SHIFT7BIT((c)) ? -1 : (((ei)->cell[(c)] & EI_MASK) - 1))
162 1.1 tnozaki
163 1.1 tnozaki #define ISDIRECT(ei, c) (!SHIFT7BIT((c)) && (ISSPECIAL((c)) || \
164 1.1 tnozaki ei->cell[(c)] & (EI_DIRECT | EI_OPTION | EI_SPACE)))
165 1.1 tnozaki
166 1.1 tnozaki #define ISSAFE(ei, c) (!SHIFT7BIT((c)) && (ISSPECIAL((c)) || \
167 1.4 christos (c < 0x80 && ei->cell[(c)] & (EI_DIRECT | EI_SPACE))))
168 1.1 tnozaki
169 1.1 tnozaki /* surrogate pair */
170 1.1 tnozaki #define SRG_BASE UINT32_C(0x10000)
171 1.1 tnozaki #define HISRG_MIN UINT16_C(0xd800)
172 1.1 tnozaki #define HISRG_MAX UINT16_C(0xdbff)
173 1.1 tnozaki #define LOSRG_MIN UINT16_C(0xdc00)
174 1.1 tnozaki #define LOSRG_MAX UINT16_C(0xdfff)
175 1.1 tnozaki #define CHECK_SRG(st, c, act) \
176 1.1 tnozaki do { \
177 1.1 tnozaki if (!(st)->surrogate) { \
178 1.1 tnozaki if ((c) >= HISRG_MIN && (c) <= HISRG_MAX) \
179 1.1 tnozaki (st)->surrogate = 1; \
180 1.1 tnozaki } else { \
181 1.1 tnozaki if ((c) < LOSRG_MIN || (c) > LOSRG_MAX) \
182 1.1 tnozaki act; \
183 1.1 tnozaki (st)->surrogate = 0; \
184 1.1 tnozaki } \
185 1.1 tnozaki } while (/*CONSTCOND*/0)
186 1.1 tnozaki
187 1.1 tnozaki static int
188 1.1 tnozaki _mbtoutf16(_UTF7EncodingInfo * __restrict ei,
189 1.1 tnozaki uint16_t * __restrict u16, const char ** __restrict s, size_t n,
190 1.1 tnozaki _UTF7State * __restrict psenc, size_t * __restrict nresult)
191 1.1 tnozaki {
192 1.2 tnozaki _UTF7State sv;
193 1.1 tnozaki const char *s0;
194 1.1 tnozaki int i, done, len;
195 1.1 tnozaki
196 1.1 tnozaki _DIAGASSERT(ei != NULL);
197 1.1 tnozaki _DIAGASSERT(s != NULL && *s != NULL);
198 1.1 tnozaki _DIAGASSERT(psenc != NULL);
199 1.1 tnozaki
200 1.1 tnozaki s0 = *s;
201 1.2 tnozaki sv = *psenc;
202 1.1 tnozaki
203 1.1 tnozaki for (i = 0, done = 0; done == 0; i++) {
204 1.2 tnozaki _DIAGASSERT(i <= psenc->chlen);
205 1.2 tnozaki if (i == psenc->chlen) {
206 1.1 tnozaki if (n-- < 1) {
207 1.1 tnozaki *nresult = (size_t)-2;
208 1.1 tnozaki *s = s0;
209 1.2 tnozaki sv.chlen = psenc->chlen;
210 1.2 tnozaki *psenc = sv;
211 1.1 tnozaki return (0);
212 1.1 tnozaki }
213 1.2 tnozaki psenc->ch[psenc->chlen++] = *s0++;
214 1.1 tnozaki }
215 1.1 tnozaki if (SHIFT7BIT((int)psenc->ch[i]))
216 1.1 tnozaki goto ilseq;
217 1.2 tnozaki if (!psenc->mode) {
218 1.2 tnozaki if (psenc->bits > 0 || psenc->cache > 0)
219 1.1 tnozaki return (EINVAL);
220 1.1 tnozaki if (psenc->ch[i] == BASE64_IN) {
221 1.2 tnozaki psenc->mode = 1;
222 1.1 tnozaki } else {
223 1.1 tnozaki if (!ISDIRECT(ei, (int)psenc->ch[i]))
224 1.1 tnozaki goto ilseq;
225 1.1 tnozaki *u16 = (uint16_t)psenc->ch[i];
226 1.1 tnozaki done = 1;
227 1.1 tnozaki continue;
228 1.1 tnozaki }
229 1.1 tnozaki } else {
230 1.2 tnozaki if (psenc->ch[i] == BASE64_OUT && psenc->cache == 0) {
231 1.2 tnozaki psenc->mode = 0;
232 1.1 tnozaki *u16 = (uint16_t)BASE64_IN;
233 1.1 tnozaki done = 1;
234 1.1 tnozaki continue;
235 1.1 tnozaki }
236 1.1 tnozaki len = FINDLEN(ei, (int)psenc->ch[i]);
237 1.1 tnozaki if (len < 0) {
238 1.2 tnozaki if (psenc->bits >= BASE64_BIT)
239 1.1 tnozaki return (EINVAL);
240 1.2 tnozaki psenc->mode = 0;
241 1.2 tnozaki psenc->bits = psenc->cache = 0;
242 1.1 tnozaki if (psenc->ch[i] != BASE64_OUT) {
243 1.1 tnozaki if (!ISDIRECT(ei, (int)psenc->ch[i]))
244 1.1 tnozaki goto ilseq;
245 1.1 tnozaki *u16 = (uint16_t)psenc->ch[i];
246 1.1 tnozaki done = 1;
247 1.1 tnozaki }
248 1.1 tnozaki } else {
249 1.2 tnozaki psenc->cache = (psenc->cache << BASE64_BIT) | len;
250 1.2 tnozaki switch (psenc->bits) {
251 1.1 tnozaki case 0: case 2: case 4: case 6: case 8:
252 1.2 tnozaki psenc->bits += BASE64_BIT;
253 1.1 tnozaki break;
254 1.1 tnozaki case 10: case 12: case 14:
255 1.2 tnozaki psenc->bits -= (UTF16_BIT - BASE64_BIT);
256 1.2 tnozaki *u16 = (psenc->cache >> psenc->bits)
257 1.1 tnozaki & UTF16_MAX;
258 1.2 tnozaki CHECK_SRG(psenc, *u16, goto ilseq);
259 1.1 tnozaki done = 1;
260 1.1 tnozaki break;
261 1.1 tnozaki default:
262 1.1 tnozaki return (EINVAL);
263 1.1 tnozaki }
264 1.1 tnozaki }
265 1.1 tnozaki }
266 1.1 tnozaki }
267 1.1 tnozaki
268 1.2 tnozaki if (psenc->chlen > i)
269 1.1 tnozaki return (EINVAL);
270 1.2 tnozaki psenc->chlen = 0;
271 1.1 tnozaki *nresult = (size_t)((*u16 == 0) ? 0 : s0 - *s);
272 1.1 tnozaki *s = s0;
273 1.1 tnozaki
274 1.1 tnozaki return (0);
275 1.1 tnozaki
276 1.1 tnozaki ilseq:
277 1.1 tnozaki *nresult = (size_t)-1;
278 1.1 tnozaki return (EILSEQ);
279 1.1 tnozaki }
280 1.1 tnozaki
281 1.1 tnozaki static int
282 1.1 tnozaki _citrus_UTF7_mbrtowc_priv(_UTF7EncodingInfo * __restrict ei,
283 1.1 tnozaki wchar_t * __restrict pwc, const char ** __restrict s, size_t n,
284 1.1 tnozaki _UTF7State * __restrict psenc, size_t * __restrict nresult)
285 1.1 tnozaki {
286 1.1 tnozaki uint32_t u32;
287 1.1 tnozaki uint16_t hi, lo;
288 1.1 tnozaki size_t siz;
289 1.1 tnozaki int err;
290 1.1 tnozaki
291 1.1 tnozaki _DIAGASSERT(ei != NULL);
292 1.1 tnozaki /* pwc may be null */
293 1.1 tnozaki _DIAGASSERT(s != NULL);
294 1.1 tnozaki _DIAGASSERT(psenc != NULL);
295 1.1 tnozaki
296 1.1 tnozaki if (*s == NULL) {
297 1.1 tnozaki _citrus_UTF7_init_state(ei, psenc);
298 1.1 tnozaki *nresult = (size_t)_ENCODING_IS_STATE_DEPENDENT;
299 1.1 tnozaki return (0);
300 1.1 tnozaki }
301 1.1 tnozaki
302 1.2 tnozaki if (psenc->surrogate) {
303 1.2 tnozaki hi = (psenc->cache >> 2) & UTF16_MAX;
304 1.1 tnozaki if (hi >= HISRG_MIN && hi <= HISRG_MAX)
305 1.1 tnozaki return (EINVAL);
306 1.1 tnozaki siz = 0;
307 1.1 tnozaki } else {
308 1.1 tnozaki err = _mbtoutf16(ei, &hi, s, n, psenc, nresult);
309 1.1 tnozaki if (err || *nresult == (size_t)-2)
310 1.1 tnozaki return (err);
311 1.1 tnozaki n -= *nresult;
312 1.1 tnozaki siz = *nresult;
313 1.1 tnozaki }
314 1.2 tnozaki if (!psenc->surrogate) {
315 1.1 tnozaki u32 = (uint32_t)hi;
316 1.1 tnozaki } else {
317 1.1 tnozaki err = _mbtoutf16(ei, &lo, s, n, psenc, nresult);
318 1.1 tnozaki if (err || *nresult == (size_t)-2)
319 1.1 tnozaki return (err);
320 1.1 tnozaki _DIAGASSERT(!st->surrogate &&
321 1.1 tnozaki lo >= LOSRG_MIN && lo <= LOSRG_MAX);
322 1.1 tnozaki hi -= HISRG_MIN;
323 1.1 tnozaki lo -= LOSRG_MIN;
324 1.1 tnozaki u32 = (hi << 10 | lo) + SRG_BASE;
325 1.1 tnozaki *nresult += siz;
326 1.1 tnozaki }
327 1.1 tnozaki if (pwc != NULL)
328 1.1 tnozaki *pwc = (wchar_t)u32;
329 1.1 tnozaki
330 1.1 tnozaki return (0);
331 1.1 tnozaki }
332 1.1 tnozaki
333 1.1 tnozaki static __inline int
334 1.1 tnozaki _utf16tomb(_UTF7EncodingInfo * __restrict ei,
335 1.1 tnozaki uint16_t u16, _UTF7State * __restrict psenc)
336 1.1 tnozaki {
337 1.1 tnozaki int bits, i;
338 1.1 tnozaki
339 1.1 tnozaki _DIAGASSERT(ei != NULL);
340 1.1 tnozaki _DIAGASSERT(psenc != NULL);
341 1.1 tnozaki
342 1.2 tnozaki if (psenc->chlen != 0 || psenc->bits > BASE64_BIT)
343 1.1 tnozaki return (EINVAL);
344 1.2 tnozaki CHECK_SRG(psenc, u16, return (EILSEQ));
345 1.1 tnozaki
346 1.1 tnozaki if (ISSAFE(ei, u16)) {
347 1.2 tnozaki if (psenc->mode) {
348 1.2 tnozaki if (psenc->bits > 0) {
349 1.2 tnozaki bits = BASE64_BIT - psenc->bits;
350 1.2 tnozaki i = (psenc->cache << bits) & BASE64_MAX;
351 1.2 tnozaki psenc->ch[psenc->chlen++] = base64[i];
352 1.2 tnozaki psenc->bits = psenc->cache = 0;
353 1.1 tnozaki }
354 1.1 tnozaki if (u16 == BASE64_OUT || FINDLEN(ei, u16) >= 0)
355 1.2 tnozaki psenc->ch[psenc->chlen++] = BASE64_OUT;
356 1.2 tnozaki psenc->mode = 0;
357 1.1 tnozaki }
358 1.2 tnozaki if (psenc->bits != 0)
359 1.1 tnozaki return (EINVAL);
360 1.2 tnozaki psenc->ch[psenc->chlen++] = (char)u16;
361 1.1 tnozaki if (u16 == BASE64_IN)
362 1.2 tnozaki psenc->ch[psenc->chlen++] = BASE64_OUT;
363 1.1 tnozaki } else {
364 1.2 tnozaki if (!psenc->mode) {
365 1.2 tnozaki if (psenc->bits > 0)
366 1.1 tnozaki return (EINVAL);
367 1.2 tnozaki psenc->ch[psenc->chlen++] = BASE64_IN;
368 1.2 tnozaki psenc->mode = 1;
369 1.1 tnozaki }
370 1.2 tnozaki psenc->cache = (psenc->cache << UTF16_BIT) | u16;
371 1.2 tnozaki bits = UTF16_BIT + psenc->bits;
372 1.2 tnozaki psenc->bits = bits % BASE64_BIT;
373 1.1 tnozaki while ((bits -= BASE64_BIT) >= 0) {
374 1.2 tnozaki i = (psenc->cache >> bits) & BASE64_MAX;
375 1.2 tnozaki psenc->ch[psenc->chlen++] = base64[i];
376 1.1 tnozaki }
377 1.1 tnozaki }
378 1.1 tnozaki
379 1.1 tnozaki return (0);
380 1.1 tnozaki }
381 1.1 tnozaki
382 1.1 tnozaki static int
383 1.1 tnozaki _citrus_UTF7_wcrtomb_priv(_UTF7EncodingInfo * __restrict ei,
384 1.1 tnozaki char * __restrict s, size_t n, wchar_t wchar,
385 1.1 tnozaki _UTF7State * __restrict psenc, size_t * __restrict nresult)
386 1.1 tnozaki {
387 1.2 tnozaki _UTF7State sv;
388 1.1 tnozaki uint32_t u32;
389 1.1 tnozaki uint16_t u16[2];
390 1.1 tnozaki int err, len, i;
391 1.1 tnozaki size_t nr;
392 1.1 tnozaki
393 1.1 tnozaki _DIAGASSERT(ei != NULL);
394 1.1 tnozaki _DIAGASSERT(s != NULL);
395 1.1 tnozaki _DIAGASSERT(psenc != NULL);
396 1.1 tnozaki _DIAGASSERT(*nresult != NULL);
397 1.1 tnozaki
398 1.1 tnozaki u32 = (uint32_t)wchar;
399 1.1 tnozaki if (u32 <= UTF16_MAX) {
400 1.1 tnozaki u16[0] = (uint16_t)u32;
401 1.1 tnozaki len = 1;
402 1.1 tnozaki } else if (u32 <= UTF32_MAX) {
403 1.1 tnozaki u32 -= SRG_BASE;
404 1.1 tnozaki u16[0] = (u32 >> 10) + HISRG_MIN;
405 1.1 tnozaki u16[1] = ((uint16_t)(u32 & UINT32_C(0x3ff))) + LOSRG_MIN;
406 1.1 tnozaki len = 2;
407 1.1 tnozaki } else {
408 1.1 tnozaki *nresult = (size_t)-1;
409 1.1 tnozaki return (EILSEQ);
410 1.1 tnozaki }
411 1.1 tnozaki
412 1.2 tnozaki sv = *psenc;
413 1.1 tnozaki nr = 0;
414 1.1 tnozaki for (i = 0; i < len; i++) {
415 1.1 tnozaki err = _utf16tomb(ei, u16[i], psenc);
416 1.1 tnozaki switch (err) {
417 1.1 tnozaki case 0:
418 1.2 tnozaki if (psenc->chlen <= n)
419 1.1 tnozaki break;
420 1.2 tnozaki *psenc = sv;
421 1.1 tnozaki err = (E2BIG);
422 1.1 tnozaki case EILSEQ:
423 1.1 tnozaki *nresult = (size_t)-1;
424 1.1 tnozaki /*FALLTHROUGH*/
425 1.1 tnozaki default:
426 1.1 tnozaki return (err);
427 1.1 tnozaki }
428 1.2 tnozaki n -= psenc->chlen;
429 1.2 tnozaki memcpy(s, psenc->ch, psenc->chlen);
430 1.2 tnozaki s += psenc->chlen;
431 1.2 tnozaki nr += psenc->chlen;
432 1.2 tnozaki psenc->chlen = 0;
433 1.1 tnozaki }
434 1.1 tnozaki *nresult = nr;
435 1.1 tnozaki
436 1.1 tnozaki return (0);
437 1.1 tnozaki }
438 1.1 tnozaki
439 1.1 tnozaki static int
440 1.1 tnozaki /* ARGSUSED */
441 1.1 tnozaki _citrus_UTF7_put_state_reset(_UTF7EncodingInfo * __restrict ei,
442 1.1 tnozaki char * __restrict s, size_t n, _UTF7State * __restrict psenc,
443 1.1 tnozaki size_t * __restrict nresult)
444 1.1 tnozaki {
445 1.1 tnozaki int bits, pos;
446 1.1 tnozaki
447 1.1 tnozaki _DIAGASSERT(ei != NULL);
448 1.1 tnozaki _DIAGASSERT(s != NULL);
449 1.1 tnozaki _DIAGASSERT(psenc != NULL);
450 1.1 tnozaki _DIAGASSERT(nresult != NULL);
451 1.1 tnozaki
452 1.2 tnozaki if (psenc->chlen != 0 || psenc->bits > BASE64_BIT || psenc->surrogate)
453 1.1 tnozaki return (EINVAL);
454 1.1 tnozaki
455 1.2 tnozaki if (psenc->mode) {
456 1.2 tnozaki if (psenc->bits > 0) {
457 1.1 tnozaki if (n-- < 1)
458 1.1 tnozaki return (E2BIG);
459 1.2 tnozaki bits = BASE64_BIT - psenc->bits;
460 1.2 tnozaki pos = (psenc->cache << bits) & BASE64_MAX;
461 1.2 tnozaki psenc->ch[psenc->chlen++] = base64[pos];
462 1.2 tnozaki psenc->ch[psenc->chlen++] = BASE64_OUT;
463 1.2 tnozaki psenc->bits = psenc->cache = 0;
464 1.1 tnozaki }
465 1.2 tnozaki psenc->mode = 0;
466 1.1 tnozaki }
467 1.2 tnozaki if (psenc->bits != 0)
468 1.1 tnozaki return (EINVAL);
469 1.1 tnozaki if (n-- < 1)
470 1.1 tnozaki return (E2BIG);
471 1.1 tnozaki
472 1.2 tnozaki _DIAGASSERT(n >= psenc->chlen);
473 1.2 tnozaki *nresult = (size_t)psenc->chlen;
474 1.2 tnozaki if (psenc->chlen > 0) {
475 1.2 tnozaki memcpy(s, psenc->ch, psenc->chlen);
476 1.2 tnozaki psenc->chlen = 0;
477 1.1 tnozaki }
478 1.1 tnozaki
479 1.1 tnozaki return (0);
480 1.1 tnozaki }
481 1.1 tnozaki
482 1.1 tnozaki static __inline int
483 1.1 tnozaki /*ARGSUSED*/
484 1.1 tnozaki _citrus_UTF7_stdenc_wctocs(_UTF7EncodingInfo * __restrict ei,
485 1.1 tnozaki _csid_t * __restrict csid,
486 1.1 tnozaki _index_t * __restrict idx, wchar_t wc)
487 1.1 tnozaki {
488 1.1 tnozaki /* ei seem to be unused */
489 1.1 tnozaki _DIAGASSERT(csid != NULL);
490 1.1 tnozaki _DIAGASSERT(idx != NULL);
491 1.1 tnozaki
492 1.1 tnozaki *csid = 0;
493 1.1 tnozaki *idx = (_index_t)wc;
494 1.1 tnozaki
495 1.1 tnozaki return (0);
496 1.1 tnozaki }
497 1.1 tnozaki
498 1.1 tnozaki static __inline int
499 1.1 tnozaki /*ARGSUSED*/
500 1.1 tnozaki _citrus_UTF7_stdenc_cstowc(_UTF7EncodingInfo * __restrict ei,
501 1.1 tnozaki wchar_t * __restrict wc,
502 1.1 tnozaki _csid_t csid, _index_t idx)
503 1.1 tnozaki {
504 1.1 tnozaki /* ei seem to be unused */
505 1.1 tnozaki _DIAGASSERT(wc != NULL);
506 1.1 tnozaki
507 1.1 tnozaki if (csid != 0)
508 1.1 tnozaki return (EILSEQ);
509 1.1 tnozaki *wc = (wchar_t)idx;
510 1.1 tnozaki
511 1.1 tnozaki return (0);
512 1.1 tnozaki }
513 1.1 tnozaki
514 1.3 tshiozak static __inline int
515 1.3 tshiozak /*ARGSUSED*/
516 1.3 tshiozak _citrus_UTF7_stdenc_get_state_desc_generic(_UTF7EncodingInfo * __restrict ei,
517 1.3 tshiozak _UTF7State * __restrict psenc,
518 1.3 tshiozak int * __restrict rstate)
519 1.3 tshiozak {
520 1.3 tshiozak
521 1.3 tshiozak if (psenc->chlen == 0)
522 1.3 tshiozak *rstate = _STDENC_SDGEN_INITIAL;
523 1.3 tshiozak else
524 1.3 tshiozak *rstate = _STDENC_SDGEN_INCOMPLETE_CHAR;
525 1.3 tshiozak
526 1.3 tshiozak return 0;
527 1.3 tshiozak }
528 1.3 tshiozak
529 1.1 tnozaki static void
530 1.1 tnozaki /*ARGSUSED*/
531 1.1 tnozaki _citrus_UTF7_encoding_module_uninit(_UTF7EncodingInfo *ei)
532 1.1 tnozaki {
533 1.1 tnozaki /* ei seems to be unused */
534 1.1 tnozaki }
535 1.1 tnozaki
536 1.1 tnozaki
537 1.1 tnozaki static int
538 1.1 tnozaki /*ARGSUSED*/
539 1.1 tnozaki _citrus_UTF7_encoding_module_init(_UTF7EncodingInfo * __restrict ei,
540 1.1 tnozaki const void * __restrict var, size_t lenvar)
541 1.1 tnozaki {
542 1.1 tnozaki const char *s;
543 1.1 tnozaki
544 1.3 tshiozak _DIAGASSERT(ei != NULL);
545 1.1 tnozaki /* var may be null */
546 1.1 tnozaki
547 1.1 tnozaki memset(ei, 0, sizeof(*ei));
548 1.1 tnozaki
549 1.1 tnozaki #define FILL(str, flag) \
550 1.1 tnozaki do { \
551 1.1 tnozaki for (s = str; *s != '\0'; s++) \
552 1.1 tnozaki ei->cell[*s & 0x7f] |= flag; \
553 1.1 tnozaki } while (/*CONSTCOND*/0)
554 1.1 tnozaki
555 1.1 tnozaki FILL(base64, (s - base64) + 1);
556 1.1 tnozaki FILL(direct, EI_DIRECT);
557 1.1 tnozaki FILL(option, EI_OPTION);
558 1.1 tnozaki FILL(spaces, EI_SPACE);
559 1.1 tnozaki
560 1.1 tnozaki return (0);
561 1.1 tnozaki }
562 1.1 tnozaki
563 1.1 tnozaki /* ----------------------------------------------------------------------
564 1.1 tnozaki * public interface for ctype
565 1.1 tnozaki */
566 1.1 tnozaki
567 1.1 tnozaki _CITRUS_CTYPE_DECLS(UTF7);
568 1.1 tnozaki _CITRUS_CTYPE_DEF_OPS(UTF7);
569 1.1 tnozaki
570 1.1 tnozaki #include "citrus_ctype_template.h"
571 1.1 tnozaki
572 1.1 tnozaki /* ----------------------------------------------------------------------
573 1.1 tnozaki * public interface for stdenc
574 1.1 tnozaki */
575 1.1 tnozaki
576 1.1 tnozaki _CITRUS_STDENC_DECLS(UTF7);
577 1.1 tnozaki _CITRUS_STDENC_DEF_OPS(UTF7);
578 1.1 tnozaki
579 1.1 tnozaki #include "citrus_stdenc_template.h"
580