1ad30f8e7SGabor Kovesdan /* $NetBSD: citrus_utf1632.c,v 1.9 2008/06/14 16:01:08 tnozaki Exp $ */
2ad30f8e7SGabor Kovesdan
3ad30f8e7SGabor Kovesdan /*-
45e53a4f9SPedro F. Giffuni * SPDX-License-Identifier: BSD-2-Clause
55e53a4f9SPedro F. Giffuni *
6ad30f8e7SGabor Kovesdan * Copyright (c)2003 Citrus Project,
7ad30f8e7SGabor Kovesdan * All rights reserved.
8ad30f8e7SGabor Kovesdan *
9ad30f8e7SGabor Kovesdan * Redistribution and use in source and binary forms, with or without
10ad30f8e7SGabor Kovesdan * modification, are permitted provided that the following conditions
11ad30f8e7SGabor Kovesdan * are met:
12ad30f8e7SGabor Kovesdan * 1. Redistributions of source code must retain the above copyright
13ad30f8e7SGabor Kovesdan * notice, this list of conditions and the following disclaimer.
14ad30f8e7SGabor Kovesdan * 2. Redistributions in binary form must reproduce the above copyright
15ad30f8e7SGabor Kovesdan * notice, this list of conditions and the following disclaimer in the
16ad30f8e7SGabor Kovesdan * documentation and/or other materials provided with the distribution.
17ad30f8e7SGabor Kovesdan *
18ad30f8e7SGabor Kovesdan * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
19ad30f8e7SGabor Kovesdan * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20ad30f8e7SGabor Kovesdan * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21ad30f8e7SGabor Kovesdan * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
22ad30f8e7SGabor Kovesdan * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
23ad30f8e7SGabor Kovesdan * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
24ad30f8e7SGabor Kovesdan * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
25ad30f8e7SGabor Kovesdan * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
26ad30f8e7SGabor Kovesdan * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
27ad30f8e7SGabor Kovesdan * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
28ad30f8e7SGabor Kovesdan * SUCH DAMAGE.
29ad30f8e7SGabor Kovesdan */
30ad30f8e7SGabor Kovesdan
31ad30f8e7SGabor Kovesdan #include <sys/cdefs.h>
32ad30f8e7SGabor Kovesdan #include <sys/endian.h>
33ad30f8e7SGabor Kovesdan #include <sys/types.h>
34ad30f8e7SGabor Kovesdan
35ad30f8e7SGabor Kovesdan #include <assert.h>
36ad30f8e7SGabor Kovesdan #include <errno.h>
37ad30f8e7SGabor Kovesdan #include <limits.h>
38ad30f8e7SGabor Kovesdan #include <stddef.h>
39ad30f8e7SGabor Kovesdan #include <stdio.h>
40ad30f8e7SGabor Kovesdan #include <stdlib.h>
41ad30f8e7SGabor Kovesdan #include <string.h>
42ad30f8e7SGabor Kovesdan #include <wchar.h>
43ad30f8e7SGabor Kovesdan
44ad30f8e7SGabor Kovesdan #include "citrus_namespace.h"
45ad30f8e7SGabor Kovesdan #include "citrus_types.h"
46ad30f8e7SGabor Kovesdan #include "citrus_module.h"
47ad30f8e7SGabor Kovesdan #include "citrus_stdenc.h"
48ad30f8e7SGabor Kovesdan #include "citrus_bcs.h"
49ad30f8e7SGabor Kovesdan
50ad30f8e7SGabor Kovesdan #include "citrus_utf1632.h"
51ad30f8e7SGabor Kovesdan
52ad30f8e7SGabor Kovesdan
53ad30f8e7SGabor Kovesdan /* ----------------------------------------------------------------------
54ad30f8e7SGabor Kovesdan * private stuffs used by templates
55ad30f8e7SGabor Kovesdan */
56ad30f8e7SGabor Kovesdan
57ad30f8e7SGabor Kovesdan typedef struct {
58ad30f8e7SGabor Kovesdan int chlen;
59ad30f8e7SGabor Kovesdan int current_endian;
60ad30f8e7SGabor Kovesdan uint8_t ch[4];
61ad30f8e7SGabor Kovesdan } _UTF1632State;
62ad30f8e7SGabor Kovesdan
63ad30f8e7SGabor Kovesdan #define _ENDIAN_UNKNOWN 0
64ad30f8e7SGabor Kovesdan #define _ENDIAN_BIG 1
65ad30f8e7SGabor Kovesdan #define _ENDIAN_LITTLE 2
66ad30f8e7SGabor Kovesdan #if BYTE_ORDER == BIG_ENDIAN
67ad30f8e7SGabor Kovesdan #define _ENDIAN_INTERNAL _ENDIAN_BIG
68ad30f8e7SGabor Kovesdan #define _ENDIAN_SWAPPED _ENDIAN_LITTLE
69ad30f8e7SGabor Kovesdan #else
70ad30f8e7SGabor Kovesdan #define _ENDIAN_INTERNAL _ENDIAN_LITTLE
71ad30f8e7SGabor Kovesdan #define _ENDIAN_SWAPPED _ENDIAN_BIG
72ad30f8e7SGabor Kovesdan #endif
73ad30f8e7SGabor Kovesdan #define _MODE_UTF32 0x00000001U
74ad30f8e7SGabor Kovesdan #define _MODE_FORCE_ENDIAN 0x00000002U
75ad30f8e7SGabor Kovesdan
76ad30f8e7SGabor Kovesdan typedef struct {
77ad30f8e7SGabor Kovesdan int preffered_endian;
78ad30f8e7SGabor Kovesdan unsigned int cur_max;
792300a22cSKyle Evans unsigned int cur_min;
80ad30f8e7SGabor Kovesdan uint32_t mode;
81ad30f8e7SGabor Kovesdan } _UTF1632EncodingInfo;
82ad30f8e7SGabor Kovesdan
83ad30f8e7SGabor Kovesdan #define _FUNCNAME(m) _citrus_UTF1632_##m
84ad30f8e7SGabor Kovesdan #define _ENCODING_INFO _UTF1632EncodingInfo
85ad30f8e7SGabor Kovesdan #define _ENCODING_STATE _UTF1632State
86ad30f8e7SGabor Kovesdan #define _ENCODING_MB_CUR_MAX(_ei_) ((_ei_)->cur_max)
872300a22cSKyle Evans #define _ENCODING_MB_CUR_MIN(_ei_) ((_ei_)->cur_min)
88ad30f8e7SGabor Kovesdan #define _ENCODING_IS_STATE_DEPENDENT 0
89ad30f8e7SGabor Kovesdan #define _STATE_NEEDS_EXPLICIT_INIT(_ps_) 0
90ad30f8e7SGabor Kovesdan
91ad30f8e7SGabor Kovesdan
92ad30f8e7SGabor Kovesdan static __inline void
93ad30f8e7SGabor Kovesdan /*ARGSUSED*/
_citrus_UTF1632_init_state(_UTF1632EncodingInfo * ei __unused,_UTF1632State * s)94ad30f8e7SGabor Kovesdan _citrus_UTF1632_init_state(_UTF1632EncodingInfo *ei __unused,
95ad30f8e7SGabor Kovesdan _UTF1632State *s)
96ad30f8e7SGabor Kovesdan {
97ad30f8e7SGabor Kovesdan
98ad30f8e7SGabor Kovesdan memset(s, 0, sizeof(*s));
99ad30f8e7SGabor Kovesdan }
100ad30f8e7SGabor Kovesdan
101ad30f8e7SGabor Kovesdan static int
_citrus_UTF1632_mbrtowc_priv(_UTF1632EncodingInfo * ei,wchar_t * pwc,char ** s,size_t n,_UTF1632State * psenc,size_t * nresult)102ad30f8e7SGabor Kovesdan _citrus_UTF1632_mbrtowc_priv(_UTF1632EncodingInfo *ei, wchar_t *pwc,
1031243a98eSTijl Coosemans char **s, size_t n, _UTF1632State *psenc, size_t *nresult)
104ad30f8e7SGabor Kovesdan {
1051243a98eSTijl Coosemans char *s0;
106ad30f8e7SGabor Kovesdan size_t result;
107ad30f8e7SGabor Kovesdan wchar_t wc = L'\0';
108ad30f8e7SGabor Kovesdan int chlenbak, endian, needlen;
109ad30f8e7SGabor Kovesdan
110ad30f8e7SGabor Kovesdan s0 = *s;
111ad30f8e7SGabor Kovesdan
112ad30f8e7SGabor Kovesdan if (s0 == NULL) {
113ad30f8e7SGabor Kovesdan _citrus_UTF1632_init_state(ei, psenc);
114ad30f8e7SGabor Kovesdan *nresult = 0; /* state independent */
115ad30f8e7SGabor Kovesdan return (0);
116ad30f8e7SGabor Kovesdan }
117ad30f8e7SGabor Kovesdan
118ad30f8e7SGabor Kovesdan result = 0;
119ad30f8e7SGabor Kovesdan chlenbak = psenc->chlen;
120ad30f8e7SGabor Kovesdan
121ad30f8e7SGabor Kovesdan refetch:
122ad30f8e7SGabor Kovesdan needlen = ((ei->mode & _MODE_UTF32) != 0 || chlenbak >= 2) ? 4 : 2;
123ad30f8e7SGabor Kovesdan
124ad30f8e7SGabor Kovesdan while (chlenbak < needlen) {
125ad30f8e7SGabor Kovesdan if (n == 0)
126ad30f8e7SGabor Kovesdan goto restart;
127ad30f8e7SGabor Kovesdan psenc->ch[chlenbak++] = *s0++;
128ad30f8e7SGabor Kovesdan n--;
129ad30f8e7SGabor Kovesdan result++;
130ad30f8e7SGabor Kovesdan }
131ad30f8e7SGabor Kovesdan
132ad30f8e7SGabor Kovesdan /* judge endian marker */
133ad30f8e7SGabor Kovesdan if ((ei->mode & _MODE_UTF32) == 0) {
134ad30f8e7SGabor Kovesdan /* UTF16 */
135ad30f8e7SGabor Kovesdan if (psenc->ch[0] == 0xFE && psenc->ch[1] == 0xFF) {
136ad30f8e7SGabor Kovesdan psenc->current_endian = _ENDIAN_BIG;
137ad30f8e7SGabor Kovesdan chlenbak = 0;
138ad30f8e7SGabor Kovesdan goto refetch;
139ad30f8e7SGabor Kovesdan } else if (psenc->ch[0] == 0xFF && psenc->ch[1] == 0xFE) {
140ad30f8e7SGabor Kovesdan psenc->current_endian = _ENDIAN_LITTLE;
141ad30f8e7SGabor Kovesdan chlenbak = 0;
142ad30f8e7SGabor Kovesdan goto refetch;
143ad30f8e7SGabor Kovesdan }
144ad30f8e7SGabor Kovesdan } else {
145ad30f8e7SGabor Kovesdan /* UTF32 */
146ad30f8e7SGabor Kovesdan if (psenc->ch[0] == 0x00 && psenc->ch[1] == 0x00 &&
147ad30f8e7SGabor Kovesdan psenc->ch[2] == 0xFE && psenc->ch[3] == 0xFF) {
148ad30f8e7SGabor Kovesdan psenc->current_endian = _ENDIAN_BIG;
149ad30f8e7SGabor Kovesdan chlenbak = 0;
150ad30f8e7SGabor Kovesdan goto refetch;
151ad30f8e7SGabor Kovesdan } else if (psenc->ch[0] == 0xFF && psenc->ch[1] == 0xFE &&
152ad30f8e7SGabor Kovesdan psenc->ch[2] == 0x00 && psenc->ch[3] == 0x00) {
153ad30f8e7SGabor Kovesdan psenc->current_endian = _ENDIAN_LITTLE;
154ad30f8e7SGabor Kovesdan chlenbak = 0;
155ad30f8e7SGabor Kovesdan goto refetch;
156ad30f8e7SGabor Kovesdan }
157ad30f8e7SGabor Kovesdan }
158ad30f8e7SGabor Kovesdan endian = ((ei->mode & _MODE_FORCE_ENDIAN) != 0 ||
159ad30f8e7SGabor Kovesdan psenc->current_endian == _ENDIAN_UNKNOWN) ? ei->preffered_endian :
160ad30f8e7SGabor Kovesdan psenc->current_endian;
161ad30f8e7SGabor Kovesdan
162ad30f8e7SGabor Kovesdan /* get wc */
163ad30f8e7SGabor Kovesdan if ((ei->mode & _MODE_UTF32) == 0) {
164ad30f8e7SGabor Kovesdan /* UTF16 */
165ad30f8e7SGabor Kovesdan if (needlen == 2) {
166ad30f8e7SGabor Kovesdan switch (endian) {
167ad30f8e7SGabor Kovesdan case _ENDIAN_LITTLE:
168ad30f8e7SGabor Kovesdan wc = (psenc->ch[0] |
169ad30f8e7SGabor Kovesdan ((wchar_t)psenc->ch[1] << 8));
170ad30f8e7SGabor Kovesdan break;
171ad30f8e7SGabor Kovesdan case _ENDIAN_BIG:
172ad30f8e7SGabor Kovesdan wc = (psenc->ch[1] |
173ad30f8e7SGabor Kovesdan ((wchar_t)psenc->ch[0] << 8));
174ad30f8e7SGabor Kovesdan break;
175ad30f8e7SGabor Kovesdan default:
176ad30f8e7SGabor Kovesdan goto ilseq;
177ad30f8e7SGabor Kovesdan }
178ad30f8e7SGabor Kovesdan if (wc >= 0xD800 && wc <= 0xDBFF) {
179ad30f8e7SGabor Kovesdan /* surrogate high */
180ad30f8e7SGabor Kovesdan needlen = 4;
181ad30f8e7SGabor Kovesdan goto refetch;
182ad30f8e7SGabor Kovesdan }
183ad30f8e7SGabor Kovesdan } else {
184ad30f8e7SGabor Kovesdan /* surrogate low */
185ad30f8e7SGabor Kovesdan wc -= 0xD800; /* wc : surrogate high (see above) */
186ad30f8e7SGabor Kovesdan wc <<= 10;
187ad30f8e7SGabor Kovesdan switch (endian) {
188ad30f8e7SGabor Kovesdan case _ENDIAN_LITTLE:
189ad30f8e7SGabor Kovesdan if (psenc->ch[3] < 0xDC || psenc->ch[3] > 0xDF)
190ad30f8e7SGabor Kovesdan goto ilseq;
191ad30f8e7SGabor Kovesdan wc |= psenc->ch[2];
192ad30f8e7SGabor Kovesdan wc |= (wchar_t)(psenc->ch[3] & 3) << 8;
193ad30f8e7SGabor Kovesdan break;
194ad30f8e7SGabor Kovesdan case _ENDIAN_BIG:
195ad30f8e7SGabor Kovesdan if (psenc->ch[2]<0xDC || psenc->ch[2]>0xDF)
196ad30f8e7SGabor Kovesdan goto ilseq;
197ad30f8e7SGabor Kovesdan wc |= psenc->ch[3];
198ad30f8e7SGabor Kovesdan wc |= (wchar_t)(psenc->ch[2] & 3) << 8;
199ad30f8e7SGabor Kovesdan break;
200ad30f8e7SGabor Kovesdan default:
201ad30f8e7SGabor Kovesdan goto ilseq;
202ad30f8e7SGabor Kovesdan }
203ad30f8e7SGabor Kovesdan wc += 0x10000;
204ad30f8e7SGabor Kovesdan }
205ad30f8e7SGabor Kovesdan } else {
206ad30f8e7SGabor Kovesdan /* UTF32 */
207ad30f8e7SGabor Kovesdan switch (endian) {
208ad30f8e7SGabor Kovesdan case _ENDIAN_LITTLE:
209ad30f8e7SGabor Kovesdan wc = (psenc->ch[0] |
210ad30f8e7SGabor Kovesdan ((wchar_t)psenc->ch[1] << 8) |
211ad30f8e7SGabor Kovesdan ((wchar_t)psenc->ch[2] << 16) |
212ad30f8e7SGabor Kovesdan ((wchar_t)psenc->ch[3] << 24));
213ad30f8e7SGabor Kovesdan break;
214ad30f8e7SGabor Kovesdan case _ENDIAN_BIG:
215ad30f8e7SGabor Kovesdan wc = (psenc->ch[3] |
216ad30f8e7SGabor Kovesdan ((wchar_t)psenc->ch[2] << 8) |
217ad30f8e7SGabor Kovesdan ((wchar_t)psenc->ch[1] << 16) |
218ad30f8e7SGabor Kovesdan ((wchar_t)psenc->ch[0] << 24));
219ad30f8e7SGabor Kovesdan break;
220ad30f8e7SGabor Kovesdan default:
221ad30f8e7SGabor Kovesdan goto ilseq;
222ad30f8e7SGabor Kovesdan }
223ad30f8e7SGabor Kovesdan if (wc >= 0xD800 && wc <= 0xDFFF)
224ad30f8e7SGabor Kovesdan goto ilseq;
225ad30f8e7SGabor Kovesdan }
226ad30f8e7SGabor Kovesdan
227ad30f8e7SGabor Kovesdan
228ad30f8e7SGabor Kovesdan *pwc = wc;
229ad30f8e7SGabor Kovesdan psenc->chlen = 0;
230ad30f8e7SGabor Kovesdan *nresult = result;
231ad30f8e7SGabor Kovesdan *s = s0;
232ad30f8e7SGabor Kovesdan
233ad30f8e7SGabor Kovesdan return (0);
234ad30f8e7SGabor Kovesdan
235ad30f8e7SGabor Kovesdan ilseq:
236ad30f8e7SGabor Kovesdan *nresult = (size_t)-1;
237ad30f8e7SGabor Kovesdan psenc->chlen = 0;
238ad30f8e7SGabor Kovesdan return (EILSEQ);
239ad30f8e7SGabor Kovesdan
240ad30f8e7SGabor Kovesdan restart:
241ad30f8e7SGabor Kovesdan *nresult = (size_t)-2;
242ad30f8e7SGabor Kovesdan psenc->chlen = chlenbak;
243ad30f8e7SGabor Kovesdan *s = s0;
244ad30f8e7SGabor Kovesdan return (0);
245ad30f8e7SGabor Kovesdan }
246ad30f8e7SGabor Kovesdan
247ad30f8e7SGabor Kovesdan static int
_citrus_UTF1632_wcrtomb_priv(_UTF1632EncodingInfo * ei,char * s,size_t n,wchar_t wc,_UTF1632State * psenc,size_t * nresult)248ad30f8e7SGabor Kovesdan _citrus_UTF1632_wcrtomb_priv(_UTF1632EncodingInfo *ei, char *s, size_t n,
249ad30f8e7SGabor Kovesdan wchar_t wc, _UTF1632State *psenc, size_t *nresult)
250ad30f8e7SGabor Kovesdan {
251ad30f8e7SGabor Kovesdan wchar_t wc2;
252ad30f8e7SGabor Kovesdan static const char _bom[4] = {
253ad30f8e7SGabor Kovesdan 0x00, 0x00, 0xFE, 0xFF,
254ad30f8e7SGabor Kovesdan };
255ad30f8e7SGabor Kovesdan const char *bom = &_bom[0];
256ad30f8e7SGabor Kovesdan size_t cnt;
257ad30f8e7SGabor Kovesdan
258ad30f8e7SGabor Kovesdan cnt = (size_t)0;
259ad30f8e7SGabor Kovesdan if (psenc->current_endian == _ENDIAN_UNKNOWN) {
260ad30f8e7SGabor Kovesdan if ((ei->mode & _MODE_FORCE_ENDIAN) == 0) {
261ad30f8e7SGabor Kovesdan if (ei->mode & _MODE_UTF32)
262ad30f8e7SGabor Kovesdan cnt = 4;
263ad30f8e7SGabor Kovesdan else {
264ad30f8e7SGabor Kovesdan cnt = 2;
265ad30f8e7SGabor Kovesdan bom += 2;
266ad30f8e7SGabor Kovesdan }
267ad30f8e7SGabor Kovesdan if (n < cnt)
268ad30f8e7SGabor Kovesdan goto e2big;
269ad30f8e7SGabor Kovesdan memcpy(s, bom, cnt);
270ad30f8e7SGabor Kovesdan s += cnt, n -= cnt;
271ad30f8e7SGabor Kovesdan }
272ad30f8e7SGabor Kovesdan psenc->current_endian = ei->preffered_endian;
273ad30f8e7SGabor Kovesdan }
274ad30f8e7SGabor Kovesdan
275ad30f8e7SGabor Kovesdan wc2 = 0;
276ad30f8e7SGabor Kovesdan if ((ei->mode & _MODE_UTF32)==0) {
277ad30f8e7SGabor Kovesdan /* UTF16 */
278ad30f8e7SGabor Kovesdan if (wc > 0xFFFF) {
279ad30f8e7SGabor Kovesdan /* surrogate */
280ad30f8e7SGabor Kovesdan if (wc > 0x10FFFF)
281ad30f8e7SGabor Kovesdan goto ilseq;
282ad30f8e7SGabor Kovesdan if (n < 4)
283ad30f8e7SGabor Kovesdan goto e2big;
284ad30f8e7SGabor Kovesdan cnt += 4;
285ad30f8e7SGabor Kovesdan wc -= 0x10000;
286ad30f8e7SGabor Kovesdan wc2 = (wc & 0x3FF) | 0xDC00;
287ad30f8e7SGabor Kovesdan wc = (wc>>10) | 0xD800;
288ad30f8e7SGabor Kovesdan } else {
289ad30f8e7SGabor Kovesdan if (n < 2)
290ad30f8e7SGabor Kovesdan goto e2big;
291ad30f8e7SGabor Kovesdan cnt += 2;
292ad30f8e7SGabor Kovesdan }
293ad30f8e7SGabor Kovesdan
294ad30f8e7SGabor Kovesdan surrogate:
295ad30f8e7SGabor Kovesdan switch (psenc->current_endian) {
296ad30f8e7SGabor Kovesdan case _ENDIAN_BIG:
297ad30f8e7SGabor Kovesdan s[1] = wc;
298ad30f8e7SGabor Kovesdan s[0] = (wc >>= 8);
299ad30f8e7SGabor Kovesdan break;
300ad30f8e7SGabor Kovesdan case _ENDIAN_LITTLE:
301ad30f8e7SGabor Kovesdan s[0] = wc;
302ad30f8e7SGabor Kovesdan s[1] = (wc >>= 8);
303ad30f8e7SGabor Kovesdan break;
304ad30f8e7SGabor Kovesdan }
305ad30f8e7SGabor Kovesdan if (wc2 != 0) {
306ad30f8e7SGabor Kovesdan wc = wc2;
307ad30f8e7SGabor Kovesdan wc2 = 0;
308ad30f8e7SGabor Kovesdan s += 2;
309ad30f8e7SGabor Kovesdan goto surrogate;
310ad30f8e7SGabor Kovesdan }
311ad30f8e7SGabor Kovesdan } else {
312ad30f8e7SGabor Kovesdan /* UTF32 */
313ad30f8e7SGabor Kovesdan if (wc >= 0xD800 && wc <= 0xDFFF)
314ad30f8e7SGabor Kovesdan goto ilseq;
315ad30f8e7SGabor Kovesdan if (n < 4)
316ad30f8e7SGabor Kovesdan goto e2big;
317ad30f8e7SGabor Kovesdan cnt += 4;
318ad30f8e7SGabor Kovesdan switch (psenc->current_endian) {
319ad30f8e7SGabor Kovesdan case _ENDIAN_BIG:
320ad30f8e7SGabor Kovesdan s[3] = wc;
321ad30f8e7SGabor Kovesdan s[2] = (wc >>= 8);
322ad30f8e7SGabor Kovesdan s[1] = (wc >>= 8);
323ad30f8e7SGabor Kovesdan s[0] = (wc >>= 8);
324ad30f8e7SGabor Kovesdan break;
325ad30f8e7SGabor Kovesdan case _ENDIAN_LITTLE:
326ad30f8e7SGabor Kovesdan s[0] = wc;
327ad30f8e7SGabor Kovesdan s[1] = (wc >>= 8);
328ad30f8e7SGabor Kovesdan s[2] = (wc >>= 8);
329ad30f8e7SGabor Kovesdan s[3] = (wc >>= 8);
330ad30f8e7SGabor Kovesdan break;
331ad30f8e7SGabor Kovesdan }
332ad30f8e7SGabor Kovesdan }
333ad30f8e7SGabor Kovesdan *nresult = cnt;
334ad30f8e7SGabor Kovesdan
335ad30f8e7SGabor Kovesdan return (0);
336ad30f8e7SGabor Kovesdan
337ad30f8e7SGabor Kovesdan ilseq:
338ad30f8e7SGabor Kovesdan *nresult = (size_t)-1;
339ad30f8e7SGabor Kovesdan return (EILSEQ);
340ad30f8e7SGabor Kovesdan e2big:
341ad30f8e7SGabor Kovesdan *nresult = (size_t)-1;
342ad30f8e7SGabor Kovesdan return (E2BIG);
343ad30f8e7SGabor Kovesdan }
344ad30f8e7SGabor Kovesdan
345ad30f8e7SGabor Kovesdan static void
parse_variable(_UTF1632EncodingInfo * __restrict ei,const void * __restrict var,size_t lenvar)346ad30f8e7SGabor Kovesdan parse_variable(_UTF1632EncodingInfo * __restrict ei,
347ad30f8e7SGabor Kovesdan const void * __restrict var, size_t lenvar)
348ad30f8e7SGabor Kovesdan {
349ad30f8e7SGabor Kovesdan const char *p;
350ad30f8e7SGabor Kovesdan
351ad30f8e7SGabor Kovesdan p = var;
352ad30f8e7SGabor Kovesdan while (lenvar > 0) {
353ad30f8e7SGabor Kovesdan switch (*p) {
354ad30f8e7SGabor Kovesdan case 'B':
355ad30f8e7SGabor Kovesdan case 'b':
356ad30f8e7SGabor Kovesdan MATCH(big, ei->preffered_endian = _ENDIAN_BIG);
357ad30f8e7SGabor Kovesdan break;
358ad30f8e7SGabor Kovesdan case 'L':
359ad30f8e7SGabor Kovesdan case 'l':
360ad30f8e7SGabor Kovesdan MATCH(little, ei->preffered_endian = _ENDIAN_LITTLE);
361ad30f8e7SGabor Kovesdan break;
362ad30f8e7SGabor Kovesdan case 'i':
363ad30f8e7SGabor Kovesdan case 'I':
364ad30f8e7SGabor Kovesdan MATCH(internal, ei->preffered_endian = _ENDIAN_INTERNAL);
365ad30f8e7SGabor Kovesdan break;
366ad30f8e7SGabor Kovesdan case 's':
367ad30f8e7SGabor Kovesdan case 'S':
368ad30f8e7SGabor Kovesdan MATCH(swapped, ei->preffered_endian = _ENDIAN_SWAPPED);
369ad30f8e7SGabor Kovesdan break;
370ad30f8e7SGabor Kovesdan case 'F':
371ad30f8e7SGabor Kovesdan case 'f':
372ad30f8e7SGabor Kovesdan MATCH(force, ei->mode |= _MODE_FORCE_ENDIAN);
373ad30f8e7SGabor Kovesdan break;
374ad30f8e7SGabor Kovesdan case 'U':
375ad30f8e7SGabor Kovesdan case 'u':
376ad30f8e7SGabor Kovesdan MATCH(utf32, ei->mode |= _MODE_UTF32);
377ad30f8e7SGabor Kovesdan break;
378ad30f8e7SGabor Kovesdan }
379ad30f8e7SGabor Kovesdan p++;
380ad30f8e7SGabor Kovesdan lenvar--;
381ad30f8e7SGabor Kovesdan }
382ad30f8e7SGabor Kovesdan }
383ad30f8e7SGabor Kovesdan
384ad30f8e7SGabor Kovesdan static int
385ad30f8e7SGabor Kovesdan /*ARGSUSED*/
_citrus_UTF1632_encoding_module_init(_UTF1632EncodingInfo * __restrict ei,const void * __restrict var,size_t lenvar)386ad30f8e7SGabor Kovesdan _citrus_UTF1632_encoding_module_init(_UTF1632EncodingInfo * __restrict ei,
387ad30f8e7SGabor Kovesdan const void * __restrict var, size_t lenvar)
388ad30f8e7SGabor Kovesdan {
389ad30f8e7SGabor Kovesdan
390ad30f8e7SGabor Kovesdan memset((void *)ei, 0, sizeof(*ei));
391ad30f8e7SGabor Kovesdan
392ad30f8e7SGabor Kovesdan parse_variable(ei, var, lenvar);
393ad30f8e7SGabor Kovesdan
3942300a22cSKyle Evans ei->cur_min = ((ei->mode&_MODE_UTF32) == 0) ? 2 : 4;
395ad30f8e7SGabor Kovesdan ei->cur_max = ((ei->mode&_MODE_UTF32) == 0) ? 6 : 8;
396ad30f8e7SGabor Kovesdan /* 6: endian + surrogate */
397ad30f8e7SGabor Kovesdan /* 8: endian + normal */
398ad30f8e7SGabor Kovesdan
399ad30f8e7SGabor Kovesdan if (ei->preffered_endian == _ENDIAN_UNKNOWN) {
400ad30f8e7SGabor Kovesdan ei->preffered_endian = _ENDIAN_BIG;
401ad30f8e7SGabor Kovesdan }
402ad30f8e7SGabor Kovesdan
403ad30f8e7SGabor Kovesdan return (0);
404ad30f8e7SGabor Kovesdan }
405ad30f8e7SGabor Kovesdan
406ad30f8e7SGabor Kovesdan static void
407ad30f8e7SGabor Kovesdan /*ARGSUSED*/
_citrus_UTF1632_encoding_module_uninit(_UTF1632EncodingInfo * ei __unused)408ad30f8e7SGabor Kovesdan _citrus_UTF1632_encoding_module_uninit(_UTF1632EncodingInfo *ei __unused)
409ad30f8e7SGabor Kovesdan {
410ad30f8e7SGabor Kovesdan
411ad30f8e7SGabor Kovesdan }
412ad30f8e7SGabor Kovesdan
413ad30f8e7SGabor Kovesdan static __inline int
414ad30f8e7SGabor Kovesdan /*ARGSUSED*/
_citrus_UTF1632_stdenc_wctocs(_UTF1632EncodingInfo * __restrict ei __unused,_csid_t * __restrict csid,_index_t * __restrict idx,_wc_t wc)415ad30f8e7SGabor Kovesdan _citrus_UTF1632_stdenc_wctocs(_UTF1632EncodingInfo * __restrict ei __unused,
416ad30f8e7SGabor Kovesdan _csid_t * __restrict csid, _index_t * __restrict idx, _wc_t wc)
417ad30f8e7SGabor Kovesdan {
418ad30f8e7SGabor Kovesdan
419ad30f8e7SGabor Kovesdan *csid = 0;
420ad30f8e7SGabor Kovesdan *idx = (_index_t)wc;
421ad30f8e7SGabor Kovesdan
422ad30f8e7SGabor Kovesdan return (0);
423ad30f8e7SGabor Kovesdan }
424ad30f8e7SGabor Kovesdan
425ad30f8e7SGabor Kovesdan static __inline int
426ad30f8e7SGabor Kovesdan /*ARGSUSED*/
_citrus_UTF1632_stdenc_cstowc(_UTF1632EncodingInfo * __restrict ei __unused,_wc_t * __restrict wc,_csid_t csid,_index_t idx)427ad30f8e7SGabor Kovesdan _citrus_UTF1632_stdenc_cstowc(_UTF1632EncodingInfo * __restrict ei __unused,
428ad30f8e7SGabor Kovesdan _wc_t * __restrict wc, _csid_t csid, _index_t idx)
429ad30f8e7SGabor Kovesdan {
430ad30f8e7SGabor Kovesdan
431ad30f8e7SGabor Kovesdan if (csid != 0)
432ad30f8e7SGabor Kovesdan return (EILSEQ);
433ad30f8e7SGabor Kovesdan
434ad30f8e7SGabor Kovesdan *wc = (_wc_t)idx;
435ad30f8e7SGabor Kovesdan
436ad30f8e7SGabor Kovesdan return (0);
437ad30f8e7SGabor Kovesdan }
438ad30f8e7SGabor Kovesdan
439ad30f8e7SGabor Kovesdan static __inline int
440ad30f8e7SGabor Kovesdan /*ARGSUSED*/
_citrus_UTF1632_stdenc_get_state_desc_generic(_UTF1632EncodingInfo * __restrict ei __unused,_UTF1632State * __restrict psenc,int * __restrict rstate)441ad30f8e7SGabor Kovesdan _citrus_UTF1632_stdenc_get_state_desc_generic(_UTF1632EncodingInfo * __restrict ei __unused,
442ad30f8e7SGabor Kovesdan _UTF1632State * __restrict psenc, int * __restrict rstate)
443ad30f8e7SGabor Kovesdan {
444ad30f8e7SGabor Kovesdan
445ad30f8e7SGabor Kovesdan *rstate = (psenc->chlen == 0) ? _STDENC_SDGEN_INITIAL :
446ad30f8e7SGabor Kovesdan _STDENC_SDGEN_INCOMPLETE_CHAR;
447ad30f8e7SGabor Kovesdan return (0);
448ad30f8e7SGabor Kovesdan }
449ad30f8e7SGabor Kovesdan
450ad30f8e7SGabor Kovesdan /* ----------------------------------------------------------------------
451ad30f8e7SGabor Kovesdan * public interface for stdenc
452ad30f8e7SGabor Kovesdan */
453ad30f8e7SGabor Kovesdan
454ad30f8e7SGabor Kovesdan _CITRUS_STDENC_DECLS(UTF1632);
455ad30f8e7SGabor Kovesdan _CITRUS_STDENC_DEF_OPS(UTF1632);
456ad30f8e7SGabor Kovesdan
457ad30f8e7SGabor Kovesdan #include "citrus_stdenc_template.h"
458