xref: /freebsd/lib/libiconv_modules/UTF1632/citrus_utf1632.c (revision 5e53a4f90f82c4345f277dd87cc9292f26e04a29)
1ad30f8e7SGabor Kovesdan /* $FreeBSD$ */
2ad30f8e7SGabor Kovesdan /*	$NetBSD: citrus_utf1632.c,v 1.9 2008/06/14 16:01:08 tnozaki Exp $	*/
3ad30f8e7SGabor Kovesdan 
4ad30f8e7SGabor Kovesdan /*-
5*5e53a4f9SPedro F. Giffuni  * SPDX-License-Identifier: BSD-2-Clause
6*5e53a4f9SPedro F. Giffuni  *
7ad30f8e7SGabor Kovesdan  * Copyright (c)2003 Citrus Project,
8ad30f8e7SGabor Kovesdan  * All rights reserved.
9ad30f8e7SGabor Kovesdan  *
10ad30f8e7SGabor Kovesdan  * Redistribution and use in source and binary forms, with or without
11ad30f8e7SGabor Kovesdan  * modification, are permitted provided that the following conditions
12ad30f8e7SGabor Kovesdan  * are met:
13ad30f8e7SGabor Kovesdan  * 1. Redistributions of source code must retain the above copyright
14ad30f8e7SGabor Kovesdan  *    notice, this list of conditions and the following disclaimer.
15ad30f8e7SGabor Kovesdan  * 2. Redistributions in binary form must reproduce the above copyright
16ad30f8e7SGabor Kovesdan  *    notice, this list of conditions and the following disclaimer in the
17ad30f8e7SGabor Kovesdan  *    documentation and/or other materials provided with the distribution.
18ad30f8e7SGabor Kovesdan  *
19ad30f8e7SGabor Kovesdan  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
20ad30f8e7SGabor Kovesdan  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21ad30f8e7SGabor Kovesdan  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22ad30f8e7SGabor Kovesdan  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
23ad30f8e7SGabor Kovesdan  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24ad30f8e7SGabor Kovesdan  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25ad30f8e7SGabor Kovesdan  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26ad30f8e7SGabor Kovesdan  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27ad30f8e7SGabor Kovesdan  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28ad30f8e7SGabor Kovesdan  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29ad30f8e7SGabor Kovesdan  * SUCH DAMAGE.
30ad30f8e7SGabor Kovesdan  */
31ad30f8e7SGabor Kovesdan 
32ad30f8e7SGabor Kovesdan #include <sys/cdefs.h>
33ad30f8e7SGabor Kovesdan #include <sys/endian.h>
34ad30f8e7SGabor Kovesdan #include <sys/types.h>
35ad30f8e7SGabor Kovesdan 
36ad30f8e7SGabor Kovesdan #include <assert.h>
37ad30f8e7SGabor Kovesdan #include <errno.h>
38ad30f8e7SGabor Kovesdan #include <limits.h>
39ad30f8e7SGabor Kovesdan #include <stddef.h>
40ad30f8e7SGabor Kovesdan #include <stdio.h>
41ad30f8e7SGabor Kovesdan #include <stdlib.h>
42ad30f8e7SGabor Kovesdan #include <string.h>
43ad30f8e7SGabor Kovesdan #include <wchar.h>
44ad30f8e7SGabor Kovesdan 
45ad30f8e7SGabor Kovesdan #include "citrus_namespace.h"
46ad30f8e7SGabor Kovesdan #include "citrus_types.h"
47ad30f8e7SGabor Kovesdan #include "citrus_module.h"
48ad30f8e7SGabor Kovesdan #include "citrus_stdenc.h"
49ad30f8e7SGabor Kovesdan #include "citrus_bcs.h"
50ad30f8e7SGabor Kovesdan 
51ad30f8e7SGabor Kovesdan #include "citrus_utf1632.h"
52ad30f8e7SGabor Kovesdan 
53ad30f8e7SGabor Kovesdan 
54ad30f8e7SGabor Kovesdan /* ----------------------------------------------------------------------
55ad30f8e7SGabor Kovesdan  * private stuffs used by templates
56ad30f8e7SGabor Kovesdan  */
57ad30f8e7SGabor Kovesdan 
58ad30f8e7SGabor Kovesdan typedef struct {
59ad30f8e7SGabor Kovesdan 	int		 chlen;
60ad30f8e7SGabor Kovesdan 	int		 current_endian;
61ad30f8e7SGabor Kovesdan 	uint8_t		 ch[4];
62ad30f8e7SGabor Kovesdan } _UTF1632State;
63ad30f8e7SGabor Kovesdan 
64ad30f8e7SGabor Kovesdan #define _ENDIAN_UNKNOWN		0
65ad30f8e7SGabor Kovesdan #define _ENDIAN_BIG		1
66ad30f8e7SGabor Kovesdan #define _ENDIAN_LITTLE		2
67ad30f8e7SGabor Kovesdan #if BYTE_ORDER == BIG_ENDIAN
68ad30f8e7SGabor Kovesdan #define _ENDIAN_INTERNAL	_ENDIAN_BIG
69ad30f8e7SGabor Kovesdan #define _ENDIAN_SWAPPED		_ENDIAN_LITTLE
70ad30f8e7SGabor Kovesdan #else
71ad30f8e7SGabor Kovesdan #define _ENDIAN_INTERNAL	_ENDIAN_LITTLE
72ad30f8e7SGabor Kovesdan #define _ENDIAN_SWAPPED	_ENDIAN_BIG
73ad30f8e7SGabor Kovesdan #endif
74ad30f8e7SGabor Kovesdan #define _MODE_UTF32		0x00000001U
75ad30f8e7SGabor Kovesdan #define _MODE_FORCE_ENDIAN	0x00000002U
76ad30f8e7SGabor Kovesdan 
77ad30f8e7SGabor Kovesdan typedef struct {
78ad30f8e7SGabor Kovesdan 	int		 preffered_endian;
79ad30f8e7SGabor Kovesdan 	unsigned int	 cur_max;
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)
87ad30f8e7SGabor Kovesdan #define _ENCODING_IS_STATE_DEPENDENT	0
88ad30f8e7SGabor Kovesdan #define _STATE_NEEDS_EXPLICIT_INIT(_ps_)	0
89ad30f8e7SGabor Kovesdan 
90ad30f8e7SGabor Kovesdan 
91ad30f8e7SGabor Kovesdan static __inline void
92ad30f8e7SGabor Kovesdan /*ARGSUSED*/
93ad30f8e7SGabor Kovesdan _citrus_UTF1632_init_state(_UTF1632EncodingInfo *ei __unused,
94ad30f8e7SGabor Kovesdan     _UTF1632State *s)
95ad30f8e7SGabor Kovesdan {
96ad30f8e7SGabor Kovesdan 
97ad30f8e7SGabor Kovesdan 	memset(s, 0, sizeof(*s));
98ad30f8e7SGabor Kovesdan }
99ad30f8e7SGabor Kovesdan 
100ad30f8e7SGabor Kovesdan static int
101ad30f8e7SGabor Kovesdan _citrus_UTF1632_mbrtowc_priv(_UTF1632EncodingInfo *ei, wchar_t *pwc,
1021243a98eSTijl Coosemans     char **s, size_t n, _UTF1632State *psenc, size_t *nresult)
103ad30f8e7SGabor Kovesdan {
1041243a98eSTijl Coosemans 	char *s0;
105ad30f8e7SGabor Kovesdan 	size_t result;
106ad30f8e7SGabor Kovesdan 	wchar_t wc = L'\0';
107ad30f8e7SGabor Kovesdan 	int chlenbak, endian, needlen;
108ad30f8e7SGabor Kovesdan 
109ad30f8e7SGabor Kovesdan 	s0 = *s;
110ad30f8e7SGabor Kovesdan 
111ad30f8e7SGabor Kovesdan 	if (s0 == NULL) {
112ad30f8e7SGabor Kovesdan 		_citrus_UTF1632_init_state(ei, psenc);
113ad30f8e7SGabor Kovesdan 		*nresult = 0; /* state independent */
114ad30f8e7SGabor Kovesdan 		return (0);
115ad30f8e7SGabor Kovesdan 	}
116ad30f8e7SGabor Kovesdan 
117ad30f8e7SGabor Kovesdan 	result = 0;
118ad30f8e7SGabor Kovesdan 	chlenbak = psenc->chlen;
119ad30f8e7SGabor Kovesdan 
120ad30f8e7SGabor Kovesdan refetch:
121ad30f8e7SGabor Kovesdan 	needlen = ((ei->mode & _MODE_UTF32) != 0 || chlenbak >= 2) ? 4 : 2;
122ad30f8e7SGabor Kovesdan 
123ad30f8e7SGabor Kovesdan 	while (chlenbak < needlen) {
124ad30f8e7SGabor Kovesdan 		if (n == 0)
125ad30f8e7SGabor Kovesdan 			goto restart;
126ad30f8e7SGabor Kovesdan 		psenc->ch[chlenbak++] = *s0++;
127ad30f8e7SGabor Kovesdan 		n--;
128ad30f8e7SGabor Kovesdan 		result++;
129ad30f8e7SGabor Kovesdan 	}
130ad30f8e7SGabor Kovesdan 
131ad30f8e7SGabor Kovesdan 	/* judge endian marker */
132ad30f8e7SGabor Kovesdan 	if ((ei->mode & _MODE_UTF32) == 0) {
133ad30f8e7SGabor Kovesdan 		/* UTF16 */
134ad30f8e7SGabor Kovesdan 		if (psenc->ch[0] == 0xFE && psenc->ch[1] == 0xFF) {
135ad30f8e7SGabor Kovesdan 			psenc->current_endian = _ENDIAN_BIG;
136ad30f8e7SGabor Kovesdan 			chlenbak = 0;
137ad30f8e7SGabor Kovesdan 			goto refetch;
138ad30f8e7SGabor Kovesdan 		} else if (psenc->ch[0] == 0xFF && psenc->ch[1] == 0xFE) {
139ad30f8e7SGabor Kovesdan 			psenc->current_endian = _ENDIAN_LITTLE;
140ad30f8e7SGabor Kovesdan 			chlenbak = 0;
141ad30f8e7SGabor Kovesdan 			goto refetch;
142ad30f8e7SGabor Kovesdan 		}
143ad30f8e7SGabor Kovesdan 	} else {
144ad30f8e7SGabor Kovesdan 		/* UTF32 */
145ad30f8e7SGabor Kovesdan 		if (psenc->ch[0] == 0x00 && psenc->ch[1] == 0x00 &&
146ad30f8e7SGabor Kovesdan 		    psenc->ch[2] == 0xFE && psenc->ch[3] == 0xFF) {
147ad30f8e7SGabor Kovesdan 			psenc->current_endian = _ENDIAN_BIG;
148ad30f8e7SGabor Kovesdan 			chlenbak = 0;
149ad30f8e7SGabor Kovesdan 			goto refetch;
150ad30f8e7SGabor Kovesdan 		} else if (psenc->ch[0] == 0xFF && psenc->ch[1] == 0xFE &&
151ad30f8e7SGabor Kovesdan 			   psenc->ch[2] == 0x00 && psenc->ch[3] == 0x00) {
152ad30f8e7SGabor Kovesdan 			psenc->current_endian = _ENDIAN_LITTLE;
153ad30f8e7SGabor Kovesdan 			chlenbak = 0;
154ad30f8e7SGabor Kovesdan 			goto refetch;
155ad30f8e7SGabor Kovesdan 		}
156ad30f8e7SGabor Kovesdan 	}
157ad30f8e7SGabor Kovesdan 	endian = ((ei->mode & _MODE_FORCE_ENDIAN) != 0 ||
158ad30f8e7SGabor Kovesdan 	    psenc->current_endian == _ENDIAN_UNKNOWN) ? ei->preffered_endian :
159ad30f8e7SGabor Kovesdan 	    psenc->current_endian;
160ad30f8e7SGabor Kovesdan 
161ad30f8e7SGabor Kovesdan 	/* get wc */
162ad30f8e7SGabor Kovesdan 	if ((ei->mode & _MODE_UTF32) == 0) {
163ad30f8e7SGabor Kovesdan 		/* UTF16 */
164ad30f8e7SGabor Kovesdan 		if (needlen == 2) {
165ad30f8e7SGabor Kovesdan 			switch (endian) {
166ad30f8e7SGabor Kovesdan 			case _ENDIAN_LITTLE:
167ad30f8e7SGabor Kovesdan 				wc = (psenc->ch[0] |
168ad30f8e7SGabor Kovesdan 				    ((wchar_t)psenc->ch[1] << 8));
169ad30f8e7SGabor Kovesdan 				break;
170ad30f8e7SGabor Kovesdan 			case _ENDIAN_BIG:
171ad30f8e7SGabor Kovesdan 				wc = (psenc->ch[1] |
172ad30f8e7SGabor Kovesdan 				    ((wchar_t)psenc->ch[0] << 8));
173ad30f8e7SGabor Kovesdan 				break;
174ad30f8e7SGabor Kovesdan 			default:
175ad30f8e7SGabor Kovesdan 				goto ilseq;
176ad30f8e7SGabor Kovesdan 			}
177ad30f8e7SGabor Kovesdan 			if (wc >= 0xD800 && wc <= 0xDBFF) {
178ad30f8e7SGabor Kovesdan 				/* surrogate high */
179ad30f8e7SGabor Kovesdan 				needlen = 4;
180ad30f8e7SGabor Kovesdan 				goto refetch;
181ad30f8e7SGabor Kovesdan 			}
182ad30f8e7SGabor Kovesdan 		} else {
183ad30f8e7SGabor Kovesdan 			/* surrogate low */
184ad30f8e7SGabor Kovesdan 			wc -= 0xD800; /* wc : surrogate high (see above) */
185ad30f8e7SGabor Kovesdan 			wc <<= 10;
186ad30f8e7SGabor Kovesdan 			switch (endian) {
187ad30f8e7SGabor Kovesdan 			case _ENDIAN_LITTLE:
188ad30f8e7SGabor Kovesdan 				if (psenc->ch[3] < 0xDC || psenc->ch[3] > 0xDF)
189ad30f8e7SGabor Kovesdan 					goto ilseq;
190ad30f8e7SGabor Kovesdan 				wc |= psenc->ch[2];
191ad30f8e7SGabor Kovesdan 				wc |= (wchar_t)(psenc->ch[3] & 3) << 8;
192ad30f8e7SGabor Kovesdan 				break;
193ad30f8e7SGabor Kovesdan 			case _ENDIAN_BIG:
194ad30f8e7SGabor Kovesdan 				if (psenc->ch[2]<0xDC || psenc->ch[2]>0xDF)
195ad30f8e7SGabor Kovesdan 					goto ilseq;
196ad30f8e7SGabor Kovesdan 				wc |= psenc->ch[3];
197ad30f8e7SGabor Kovesdan 				wc |= (wchar_t)(psenc->ch[2] & 3) << 8;
198ad30f8e7SGabor Kovesdan 				break;
199ad30f8e7SGabor Kovesdan 			default:
200ad30f8e7SGabor Kovesdan 				goto ilseq;
201ad30f8e7SGabor Kovesdan 			}
202ad30f8e7SGabor Kovesdan 			wc += 0x10000;
203ad30f8e7SGabor Kovesdan 		}
204ad30f8e7SGabor Kovesdan 	} else {
205ad30f8e7SGabor Kovesdan 		/* UTF32 */
206ad30f8e7SGabor Kovesdan 		switch (endian) {
207ad30f8e7SGabor Kovesdan 		case _ENDIAN_LITTLE:
208ad30f8e7SGabor Kovesdan 			wc = (psenc->ch[0] |
209ad30f8e7SGabor Kovesdan 			    ((wchar_t)psenc->ch[1] << 8) |
210ad30f8e7SGabor Kovesdan 			    ((wchar_t)psenc->ch[2] << 16) |
211ad30f8e7SGabor Kovesdan 			    ((wchar_t)psenc->ch[3] << 24));
212ad30f8e7SGabor Kovesdan 			break;
213ad30f8e7SGabor Kovesdan 		case _ENDIAN_BIG:
214ad30f8e7SGabor Kovesdan 			wc = (psenc->ch[3] |
215ad30f8e7SGabor Kovesdan 			    ((wchar_t)psenc->ch[2] << 8) |
216ad30f8e7SGabor Kovesdan 			    ((wchar_t)psenc->ch[1] << 16) |
217ad30f8e7SGabor Kovesdan 			    ((wchar_t)psenc->ch[0] << 24));
218ad30f8e7SGabor Kovesdan 			break;
219ad30f8e7SGabor Kovesdan 		default:
220ad30f8e7SGabor Kovesdan 			goto ilseq;
221ad30f8e7SGabor Kovesdan 		}
222ad30f8e7SGabor Kovesdan 		if (wc >= 0xD800 && wc <= 0xDFFF)
223ad30f8e7SGabor Kovesdan 			goto ilseq;
224ad30f8e7SGabor Kovesdan 	}
225ad30f8e7SGabor Kovesdan 
226ad30f8e7SGabor Kovesdan 
227ad30f8e7SGabor Kovesdan 	*pwc = wc;
228ad30f8e7SGabor Kovesdan 	psenc->chlen = 0;
229ad30f8e7SGabor Kovesdan 	*nresult = result;
230ad30f8e7SGabor Kovesdan 	*s = s0;
231ad30f8e7SGabor Kovesdan 
232ad30f8e7SGabor Kovesdan 	return (0);
233ad30f8e7SGabor Kovesdan 
234ad30f8e7SGabor Kovesdan ilseq:
235ad30f8e7SGabor Kovesdan 	*nresult = (size_t)-1;
236ad30f8e7SGabor Kovesdan 	psenc->chlen = 0;
237ad30f8e7SGabor Kovesdan 	return (EILSEQ);
238ad30f8e7SGabor Kovesdan 
239ad30f8e7SGabor Kovesdan restart:
240ad30f8e7SGabor Kovesdan 	*nresult = (size_t)-2;
241ad30f8e7SGabor Kovesdan 	psenc->chlen = chlenbak;
242ad30f8e7SGabor Kovesdan 	*s = s0;
243ad30f8e7SGabor Kovesdan 	return (0);
244ad30f8e7SGabor Kovesdan }
245ad30f8e7SGabor Kovesdan 
246ad30f8e7SGabor Kovesdan static int
247ad30f8e7SGabor Kovesdan _citrus_UTF1632_wcrtomb_priv(_UTF1632EncodingInfo *ei, char *s, size_t n,
248ad30f8e7SGabor Kovesdan     wchar_t wc, _UTF1632State *psenc, size_t *nresult)
249ad30f8e7SGabor Kovesdan {
250ad30f8e7SGabor Kovesdan 	wchar_t wc2;
251ad30f8e7SGabor Kovesdan 	static const char _bom[4] = {
252ad30f8e7SGabor Kovesdan 	    0x00, 0x00, 0xFE, 0xFF,
253ad30f8e7SGabor Kovesdan 	};
254ad30f8e7SGabor Kovesdan 	const char *bom = &_bom[0];
255ad30f8e7SGabor Kovesdan 	size_t cnt;
256ad30f8e7SGabor Kovesdan 
257ad30f8e7SGabor Kovesdan 	cnt = (size_t)0;
258ad30f8e7SGabor Kovesdan 	if (psenc->current_endian == _ENDIAN_UNKNOWN) {
259ad30f8e7SGabor Kovesdan 		if ((ei->mode & _MODE_FORCE_ENDIAN) == 0) {
260ad30f8e7SGabor Kovesdan 			if (ei->mode & _MODE_UTF32)
261ad30f8e7SGabor Kovesdan 				cnt = 4;
262ad30f8e7SGabor Kovesdan 			else {
263ad30f8e7SGabor Kovesdan 				cnt = 2;
264ad30f8e7SGabor Kovesdan 				bom += 2;
265ad30f8e7SGabor Kovesdan 			}
266ad30f8e7SGabor Kovesdan 			if (n < cnt)
267ad30f8e7SGabor Kovesdan 				goto e2big;
268ad30f8e7SGabor Kovesdan 			memcpy(s, bom, cnt);
269ad30f8e7SGabor Kovesdan 			s += cnt, n -= cnt;
270ad30f8e7SGabor Kovesdan 		}
271ad30f8e7SGabor Kovesdan 		psenc->current_endian = ei->preffered_endian;
272ad30f8e7SGabor Kovesdan 	}
273ad30f8e7SGabor Kovesdan 
274ad30f8e7SGabor Kovesdan 	wc2 = 0;
275ad30f8e7SGabor Kovesdan 	if ((ei->mode & _MODE_UTF32)==0) {
276ad30f8e7SGabor Kovesdan 		/* UTF16 */
277ad30f8e7SGabor Kovesdan 		if (wc > 0xFFFF) {
278ad30f8e7SGabor Kovesdan 			/* surrogate */
279ad30f8e7SGabor Kovesdan 			if (wc > 0x10FFFF)
280ad30f8e7SGabor Kovesdan 				goto ilseq;
281ad30f8e7SGabor Kovesdan 			if (n < 4)
282ad30f8e7SGabor Kovesdan 				goto e2big;
283ad30f8e7SGabor Kovesdan 			cnt += 4;
284ad30f8e7SGabor Kovesdan 			wc -= 0x10000;
285ad30f8e7SGabor Kovesdan 			wc2 = (wc & 0x3FF) | 0xDC00;
286ad30f8e7SGabor Kovesdan 			wc = (wc>>10) | 0xD800;
287ad30f8e7SGabor Kovesdan 		} else {
288ad30f8e7SGabor Kovesdan 			if (n < 2)
289ad30f8e7SGabor Kovesdan 				goto e2big;
290ad30f8e7SGabor Kovesdan 			cnt += 2;
291ad30f8e7SGabor Kovesdan 		}
292ad30f8e7SGabor Kovesdan 
293ad30f8e7SGabor Kovesdan surrogate:
294ad30f8e7SGabor Kovesdan 		switch (psenc->current_endian) {
295ad30f8e7SGabor Kovesdan 		case _ENDIAN_BIG:
296ad30f8e7SGabor Kovesdan 			s[1] = wc;
297ad30f8e7SGabor Kovesdan 			s[0] = (wc >>= 8);
298ad30f8e7SGabor Kovesdan 			break;
299ad30f8e7SGabor Kovesdan 		case _ENDIAN_LITTLE:
300ad30f8e7SGabor Kovesdan 			s[0] = wc;
301ad30f8e7SGabor Kovesdan 			s[1] = (wc >>= 8);
302ad30f8e7SGabor Kovesdan 			break;
303ad30f8e7SGabor Kovesdan 		}
304ad30f8e7SGabor Kovesdan 		if (wc2 != 0) {
305ad30f8e7SGabor Kovesdan 			wc = wc2;
306ad30f8e7SGabor Kovesdan 			wc2 = 0;
307ad30f8e7SGabor Kovesdan 			s += 2;
308ad30f8e7SGabor Kovesdan 			goto surrogate;
309ad30f8e7SGabor Kovesdan 		}
310ad30f8e7SGabor Kovesdan 	} else {
311ad30f8e7SGabor Kovesdan 		/* UTF32 */
312ad30f8e7SGabor Kovesdan 		if (wc >= 0xD800 && wc <= 0xDFFF)
313ad30f8e7SGabor Kovesdan 			goto ilseq;
314ad30f8e7SGabor Kovesdan 		if (n < 4)
315ad30f8e7SGabor Kovesdan 			goto e2big;
316ad30f8e7SGabor Kovesdan 		cnt += 4;
317ad30f8e7SGabor Kovesdan 		switch (psenc->current_endian) {
318ad30f8e7SGabor Kovesdan 		case _ENDIAN_BIG:
319ad30f8e7SGabor Kovesdan 			s[3] = wc;
320ad30f8e7SGabor Kovesdan 			s[2] = (wc >>= 8);
321ad30f8e7SGabor Kovesdan 			s[1] = (wc >>= 8);
322ad30f8e7SGabor Kovesdan 			s[0] = (wc >>= 8);
323ad30f8e7SGabor Kovesdan 			break;
324ad30f8e7SGabor Kovesdan 		case _ENDIAN_LITTLE:
325ad30f8e7SGabor Kovesdan 			s[0] = wc;
326ad30f8e7SGabor Kovesdan 			s[1] = (wc >>= 8);
327ad30f8e7SGabor Kovesdan 			s[2] = (wc >>= 8);
328ad30f8e7SGabor Kovesdan 			s[3] = (wc >>= 8);
329ad30f8e7SGabor Kovesdan 			break;
330ad30f8e7SGabor Kovesdan 		}
331ad30f8e7SGabor Kovesdan 	}
332ad30f8e7SGabor Kovesdan 	*nresult = cnt;
333ad30f8e7SGabor Kovesdan 
334ad30f8e7SGabor Kovesdan 	return (0);
335ad30f8e7SGabor Kovesdan 
336ad30f8e7SGabor Kovesdan ilseq:
337ad30f8e7SGabor Kovesdan 	*nresult = (size_t)-1;
338ad30f8e7SGabor Kovesdan 	return (EILSEQ);
339ad30f8e7SGabor Kovesdan e2big:
340ad30f8e7SGabor Kovesdan 	*nresult = (size_t)-1;
341ad30f8e7SGabor Kovesdan 	return (E2BIG);
342ad30f8e7SGabor Kovesdan }
343ad30f8e7SGabor Kovesdan 
344ad30f8e7SGabor Kovesdan static void
345ad30f8e7SGabor Kovesdan parse_variable(_UTF1632EncodingInfo * __restrict ei,
346ad30f8e7SGabor Kovesdan     const void * __restrict var, size_t lenvar)
347ad30f8e7SGabor Kovesdan {
348ad30f8e7SGabor Kovesdan 	const char *p;
349ad30f8e7SGabor Kovesdan 
350ad30f8e7SGabor Kovesdan 	p = var;
351ad30f8e7SGabor Kovesdan 	while (lenvar > 0) {
352ad30f8e7SGabor Kovesdan 		switch (*p) {
353ad30f8e7SGabor Kovesdan 		case 'B':
354ad30f8e7SGabor Kovesdan 		case 'b':
355ad30f8e7SGabor Kovesdan 			MATCH(big, ei->preffered_endian = _ENDIAN_BIG);
356ad30f8e7SGabor Kovesdan 			break;
357ad30f8e7SGabor Kovesdan 		case 'L':
358ad30f8e7SGabor Kovesdan 		case 'l':
359ad30f8e7SGabor Kovesdan 			MATCH(little, ei->preffered_endian = _ENDIAN_LITTLE);
360ad30f8e7SGabor Kovesdan 			break;
361ad30f8e7SGabor Kovesdan 		case 'i':
362ad30f8e7SGabor Kovesdan 		case 'I':
363ad30f8e7SGabor Kovesdan 			MATCH(internal, ei->preffered_endian = _ENDIAN_INTERNAL);
364ad30f8e7SGabor Kovesdan 			break;
365ad30f8e7SGabor Kovesdan 		case 's':
366ad30f8e7SGabor Kovesdan 		case 'S':
367ad30f8e7SGabor Kovesdan 			MATCH(swapped, ei->preffered_endian = _ENDIAN_SWAPPED);
368ad30f8e7SGabor Kovesdan 			break;
369ad30f8e7SGabor Kovesdan 		case 'F':
370ad30f8e7SGabor Kovesdan 		case 'f':
371ad30f8e7SGabor Kovesdan 			MATCH(force, ei->mode |= _MODE_FORCE_ENDIAN);
372ad30f8e7SGabor Kovesdan 			break;
373ad30f8e7SGabor Kovesdan 		case 'U':
374ad30f8e7SGabor Kovesdan 		case 'u':
375ad30f8e7SGabor Kovesdan 			MATCH(utf32, ei->mode |= _MODE_UTF32);
376ad30f8e7SGabor Kovesdan 			break;
377ad30f8e7SGabor Kovesdan 		}
378ad30f8e7SGabor Kovesdan 		p++;
379ad30f8e7SGabor Kovesdan 		lenvar--;
380ad30f8e7SGabor Kovesdan 	}
381ad30f8e7SGabor Kovesdan }
382ad30f8e7SGabor Kovesdan 
383ad30f8e7SGabor Kovesdan static int
384ad30f8e7SGabor Kovesdan /*ARGSUSED*/
385ad30f8e7SGabor Kovesdan _citrus_UTF1632_encoding_module_init(_UTF1632EncodingInfo * __restrict ei,
386ad30f8e7SGabor Kovesdan     const void * __restrict var, size_t lenvar)
387ad30f8e7SGabor Kovesdan {
388ad30f8e7SGabor Kovesdan 
389ad30f8e7SGabor Kovesdan 	memset((void *)ei, 0, sizeof(*ei));
390ad30f8e7SGabor Kovesdan 
391ad30f8e7SGabor Kovesdan 	parse_variable(ei, var, lenvar);
392ad30f8e7SGabor Kovesdan 
393ad30f8e7SGabor Kovesdan 	ei->cur_max = ((ei->mode&_MODE_UTF32) == 0) ? 6 : 8;
394ad30f8e7SGabor Kovesdan 	/* 6: endian + surrogate */
395ad30f8e7SGabor Kovesdan 	/* 8: endian + normal */
396ad30f8e7SGabor Kovesdan 
397ad30f8e7SGabor Kovesdan 	if (ei->preffered_endian == _ENDIAN_UNKNOWN) {
398ad30f8e7SGabor Kovesdan 		ei->preffered_endian = _ENDIAN_BIG;
399ad30f8e7SGabor Kovesdan 	}
400ad30f8e7SGabor Kovesdan 
401ad30f8e7SGabor Kovesdan 	return (0);
402ad30f8e7SGabor Kovesdan }
403ad30f8e7SGabor Kovesdan 
404ad30f8e7SGabor Kovesdan static void
405ad30f8e7SGabor Kovesdan /*ARGSUSED*/
406ad30f8e7SGabor Kovesdan _citrus_UTF1632_encoding_module_uninit(_UTF1632EncodingInfo *ei __unused)
407ad30f8e7SGabor Kovesdan {
408ad30f8e7SGabor Kovesdan 
409ad30f8e7SGabor Kovesdan }
410ad30f8e7SGabor Kovesdan 
411ad30f8e7SGabor Kovesdan static __inline int
412ad30f8e7SGabor Kovesdan /*ARGSUSED*/
413ad30f8e7SGabor Kovesdan _citrus_UTF1632_stdenc_wctocs(_UTF1632EncodingInfo * __restrict ei __unused,
414ad30f8e7SGabor Kovesdan      _csid_t * __restrict csid, _index_t * __restrict idx, _wc_t wc)
415ad30f8e7SGabor Kovesdan {
416ad30f8e7SGabor Kovesdan 
417ad30f8e7SGabor Kovesdan 	*csid = 0;
418ad30f8e7SGabor Kovesdan 	*idx = (_index_t)wc;
419ad30f8e7SGabor Kovesdan 
420ad30f8e7SGabor Kovesdan 	return (0);
421ad30f8e7SGabor Kovesdan }
422ad30f8e7SGabor Kovesdan 
423ad30f8e7SGabor Kovesdan static __inline int
424ad30f8e7SGabor Kovesdan /*ARGSUSED*/
425ad30f8e7SGabor Kovesdan _citrus_UTF1632_stdenc_cstowc(_UTF1632EncodingInfo * __restrict ei __unused,
426ad30f8e7SGabor Kovesdan     _wc_t * __restrict wc, _csid_t csid, _index_t idx)
427ad30f8e7SGabor Kovesdan {
428ad30f8e7SGabor Kovesdan 
429ad30f8e7SGabor Kovesdan 	if (csid != 0)
430ad30f8e7SGabor Kovesdan 		return (EILSEQ);
431ad30f8e7SGabor Kovesdan 
432ad30f8e7SGabor Kovesdan 	*wc = (_wc_t)idx;
433ad30f8e7SGabor Kovesdan 
434ad30f8e7SGabor Kovesdan 	return (0);
435ad30f8e7SGabor Kovesdan }
436ad30f8e7SGabor Kovesdan 
437ad30f8e7SGabor Kovesdan static __inline int
438ad30f8e7SGabor Kovesdan /*ARGSUSED*/
439ad30f8e7SGabor Kovesdan _citrus_UTF1632_stdenc_get_state_desc_generic(_UTF1632EncodingInfo * __restrict ei __unused,
440ad30f8e7SGabor Kovesdan     _UTF1632State * __restrict psenc, int * __restrict rstate)
441ad30f8e7SGabor Kovesdan {
442ad30f8e7SGabor Kovesdan 
443ad30f8e7SGabor Kovesdan 	*rstate = (psenc->chlen == 0) ? _STDENC_SDGEN_INITIAL :
444ad30f8e7SGabor Kovesdan 	    _STDENC_SDGEN_INCOMPLETE_CHAR;
445ad30f8e7SGabor Kovesdan 	return (0);
446ad30f8e7SGabor Kovesdan }
447ad30f8e7SGabor Kovesdan 
448ad30f8e7SGabor Kovesdan /* ----------------------------------------------------------------------
449ad30f8e7SGabor Kovesdan  * public interface for stdenc
450ad30f8e7SGabor Kovesdan  */
451ad30f8e7SGabor Kovesdan 
452ad30f8e7SGabor Kovesdan _CITRUS_STDENC_DECLS(UTF1632);
453ad30f8e7SGabor Kovesdan _CITRUS_STDENC_DEF_OPS(UTF1632);
454ad30f8e7SGabor Kovesdan 
455ad30f8e7SGabor Kovesdan #include "citrus_stdenc_template.h"
456