xref: /freebsd/lib/libiconv_modules/UTF1632/citrus_utf1632.c (revision ad30f8e79bd1007cc2476e491bd21b4f5e389e0a)
1*ad30f8e7SGabor Kovesdan /* $FreeBSD$ */
2*ad30f8e7SGabor Kovesdan /*	$NetBSD: citrus_utf1632.c,v 1.9 2008/06/14 16:01:08 tnozaki Exp $	*/
3*ad30f8e7SGabor Kovesdan 
4*ad30f8e7SGabor Kovesdan /*-
5*ad30f8e7SGabor Kovesdan  * Copyright (c)2003 Citrus Project,
6*ad30f8e7SGabor Kovesdan  * All rights reserved.
7*ad30f8e7SGabor Kovesdan  *
8*ad30f8e7SGabor Kovesdan  * Redistribution and use in source and binary forms, with or without
9*ad30f8e7SGabor Kovesdan  * modification, are permitted provided that the following conditions
10*ad30f8e7SGabor Kovesdan  * are met:
11*ad30f8e7SGabor Kovesdan  * 1. Redistributions of source code must retain the above copyright
12*ad30f8e7SGabor Kovesdan  *    notice, this list of conditions and the following disclaimer.
13*ad30f8e7SGabor Kovesdan  * 2. Redistributions in binary form must reproduce the above copyright
14*ad30f8e7SGabor Kovesdan  *    notice, this list of conditions and the following disclaimer in the
15*ad30f8e7SGabor Kovesdan  *    documentation and/or other materials provided with the distribution.
16*ad30f8e7SGabor Kovesdan  *
17*ad30f8e7SGabor Kovesdan  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
18*ad30f8e7SGabor Kovesdan  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19*ad30f8e7SGabor Kovesdan  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20*ad30f8e7SGabor Kovesdan  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
21*ad30f8e7SGabor Kovesdan  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22*ad30f8e7SGabor Kovesdan  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23*ad30f8e7SGabor Kovesdan  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24*ad30f8e7SGabor Kovesdan  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25*ad30f8e7SGabor Kovesdan  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26*ad30f8e7SGabor Kovesdan  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27*ad30f8e7SGabor Kovesdan  * SUCH DAMAGE.
28*ad30f8e7SGabor Kovesdan  */
29*ad30f8e7SGabor Kovesdan 
30*ad30f8e7SGabor Kovesdan #include <sys/cdefs.h>
31*ad30f8e7SGabor Kovesdan #include <sys/endian.h>
32*ad30f8e7SGabor Kovesdan #include <sys/types.h>
33*ad30f8e7SGabor Kovesdan 
34*ad30f8e7SGabor Kovesdan #include <assert.h>
35*ad30f8e7SGabor Kovesdan #include <errno.h>
36*ad30f8e7SGabor Kovesdan #include <limits.h>
37*ad30f8e7SGabor Kovesdan #include <stddef.h>
38*ad30f8e7SGabor Kovesdan #include <stdio.h>
39*ad30f8e7SGabor Kovesdan #include <stdlib.h>
40*ad30f8e7SGabor Kovesdan #include <string.h>
41*ad30f8e7SGabor Kovesdan #include <wchar.h>
42*ad30f8e7SGabor Kovesdan 
43*ad30f8e7SGabor Kovesdan #include "citrus_namespace.h"
44*ad30f8e7SGabor Kovesdan #include "citrus_types.h"
45*ad30f8e7SGabor Kovesdan #include "citrus_module.h"
46*ad30f8e7SGabor Kovesdan #include "citrus_stdenc.h"
47*ad30f8e7SGabor Kovesdan #include "citrus_bcs.h"
48*ad30f8e7SGabor Kovesdan 
49*ad30f8e7SGabor Kovesdan #include "citrus_utf1632.h"
50*ad30f8e7SGabor Kovesdan 
51*ad30f8e7SGabor Kovesdan 
52*ad30f8e7SGabor Kovesdan /* ----------------------------------------------------------------------
53*ad30f8e7SGabor Kovesdan  * private stuffs used by templates
54*ad30f8e7SGabor Kovesdan  */
55*ad30f8e7SGabor Kovesdan 
56*ad30f8e7SGabor Kovesdan typedef struct {
57*ad30f8e7SGabor Kovesdan 	int		 chlen;
58*ad30f8e7SGabor Kovesdan 	int		 current_endian;
59*ad30f8e7SGabor Kovesdan 	uint8_t		 ch[4];
60*ad30f8e7SGabor Kovesdan } _UTF1632State;
61*ad30f8e7SGabor Kovesdan 
62*ad30f8e7SGabor Kovesdan #define _ENDIAN_UNKNOWN		0
63*ad30f8e7SGabor Kovesdan #define _ENDIAN_BIG		1
64*ad30f8e7SGabor Kovesdan #define _ENDIAN_LITTLE		2
65*ad30f8e7SGabor Kovesdan #if BYTE_ORDER == BIG_ENDIAN
66*ad30f8e7SGabor Kovesdan #define _ENDIAN_INTERNAL	_ENDIAN_BIG
67*ad30f8e7SGabor Kovesdan #define _ENDIAN_SWAPPED		_ENDIAN_LITTLE
68*ad30f8e7SGabor Kovesdan #else
69*ad30f8e7SGabor Kovesdan #define _ENDIAN_INTERNAL	_ENDIAN_LITTLE
70*ad30f8e7SGabor Kovesdan #define _ENDIAN_SWAPPED	_ENDIAN_BIG
71*ad30f8e7SGabor Kovesdan #endif
72*ad30f8e7SGabor Kovesdan #define _MODE_UTF32		0x00000001U
73*ad30f8e7SGabor Kovesdan #define _MODE_FORCE_ENDIAN	0x00000002U
74*ad30f8e7SGabor Kovesdan 
75*ad30f8e7SGabor Kovesdan typedef struct {
76*ad30f8e7SGabor Kovesdan 	int		 preffered_endian;
77*ad30f8e7SGabor Kovesdan 	unsigned int	 cur_max;
78*ad30f8e7SGabor Kovesdan 	uint32_t	 mode;
79*ad30f8e7SGabor Kovesdan } _UTF1632EncodingInfo;
80*ad30f8e7SGabor Kovesdan 
81*ad30f8e7SGabor Kovesdan #define _FUNCNAME(m)			_citrus_UTF1632_##m
82*ad30f8e7SGabor Kovesdan #define _ENCODING_INFO			_UTF1632EncodingInfo
83*ad30f8e7SGabor Kovesdan #define _ENCODING_STATE			_UTF1632State
84*ad30f8e7SGabor Kovesdan #define _ENCODING_MB_CUR_MAX(_ei_)	((_ei_)->cur_max)
85*ad30f8e7SGabor Kovesdan #define _ENCODING_IS_STATE_DEPENDENT	0
86*ad30f8e7SGabor Kovesdan #define _STATE_NEEDS_EXPLICIT_INIT(_ps_)	0
87*ad30f8e7SGabor Kovesdan 
88*ad30f8e7SGabor Kovesdan 
89*ad30f8e7SGabor Kovesdan static __inline void
90*ad30f8e7SGabor Kovesdan /*ARGSUSED*/
91*ad30f8e7SGabor Kovesdan _citrus_UTF1632_init_state(_UTF1632EncodingInfo *ei __unused,
92*ad30f8e7SGabor Kovesdan     _UTF1632State *s)
93*ad30f8e7SGabor Kovesdan {
94*ad30f8e7SGabor Kovesdan 
95*ad30f8e7SGabor Kovesdan 	memset(s, 0, sizeof(*s));
96*ad30f8e7SGabor Kovesdan }
97*ad30f8e7SGabor Kovesdan 
98*ad30f8e7SGabor Kovesdan static int
99*ad30f8e7SGabor Kovesdan _citrus_UTF1632_mbrtowc_priv(_UTF1632EncodingInfo *ei, wchar_t *pwc,
100*ad30f8e7SGabor Kovesdan     char **s, size_t n, _UTF1632State *psenc, size_t *nresult)
101*ad30f8e7SGabor Kovesdan {
102*ad30f8e7SGabor Kovesdan 	char *s0;
103*ad30f8e7SGabor Kovesdan 	size_t result;
104*ad30f8e7SGabor Kovesdan 	wchar_t wc = L'\0';
105*ad30f8e7SGabor Kovesdan 	int chlenbak, endian, needlen;
106*ad30f8e7SGabor Kovesdan 
107*ad30f8e7SGabor Kovesdan 	s0 = *s;
108*ad30f8e7SGabor Kovesdan 
109*ad30f8e7SGabor Kovesdan 	if (s0 == NULL) {
110*ad30f8e7SGabor Kovesdan 		_citrus_UTF1632_init_state(ei, psenc);
111*ad30f8e7SGabor Kovesdan 		*nresult = 0; /* state independent */
112*ad30f8e7SGabor Kovesdan 		return (0);
113*ad30f8e7SGabor Kovesdan 	}
114*ad30f8e7SGabor Kovesdan 
115*ad30f8e7SGabor Kovesdan 	result = 0;
116*ad30f8e7SGabor Kovesdan 	chlenbak = psenc->chlen;
117*ad30f8e7SGabor Kovesdan 
118*ad30f8e7SGabor Kovesdan refetch:
119*ad30f8e7SGabor Kovesdan 	needlen = ((ei->mode & _MODE_UTF32) != 0 || chlenbak >= 2) ? 4 : 2;
120*ad30f8e7SGabor Kovesdan 
121*ad30f8e7SGabor Kovesdan 	while (chlenbak < needlen) {
122*ad30f8e7SGabor Kovesdan 		if (n == 0)
123*ad30f8e7SGabor Kovesdan 			goto restart;
124*ad30f8e7SGabor Kovesdan 		psenc->ch[chlenbak++] = *s0++;
125*ad30f8e7SGabor Kovesdan 		n--;
126*ad30f8e7SGabor Kovesdan 		result++;
127*ad30f8e7SGabor Kovesdan 	}
128*ad30f8e7SGabor Kovesdan 
129*ad30f8e7SGabor Kovesdan 	/* judge endian marker */
130*ad30f8e7SGabor Kovesdan 	if ((ei->mode & _MODE_UTF32) == 0) {
131*ad30f8e7SGabor Kovesdan 		/* UTF16 */
132*ad30f8e7SGabor Kovesdan 		if (psenc->ch[0] == 0xFE && psenc->ch[1] == 0xFF) {
133*ad30f8e7SGabor Kovesdan 			psenc->current_endian = _ENDIAN_BIG;
134*ad30f8e7SGabor Kovesdan 			chlenbak = 0;
135*ad30f8e7SGabor Kovesdan 			goto refetch;
136*ad30f8e7SGabor Kovesdan 		} else if (psenc->ch[0] == 0xFF && psenc->ch[1] == 0xFE) {
137*ad30f8e7SGabor Kovesdan 			psenc->current_endian = _ENDIAN_LITTLE;
138*ad30f8e7SGabor Kovesdan 			chlenbak = 0;
139*ad30f8e7SGabor Kovesdan 			goto refetch;
140*ad30f8e7SGabor Kovesdan 		}
141*ad30f8e7SGabor Kovesdan 	} else {
142*ad30f8e7SGabor Kovesdan 		/* UTF32 */
143*ad30f8e7SGabor Kovesdan 		if (psenc->ch[0] == 0x00 && psenc->ch[1] == 0x00 &&
144*ad30f8e7SGabor Kovesdan 		    psenc->ch[2] == 0xFE && psenc->ch[3] == 0xFF) {
145*ad30f8e7SGabor Kovesdan 			psenc->current_endian = _ENDIAN_BIG;
146*ad30f8e7SGabor Kovesdan 			chlenbak = 0;
147*ad30f8e7SGabor Kovesdan 			goto refetch;
148*ad30f8e7SGabor Kovesdan 		} else if (psenc->ch[0] == 0xFF && psenc->ch[1] == 0xFE &&
149*ad30f8e7SGabor Kovesdan 			   psenc->ch[2] == 0x00 && psenc->ch[3] == 0x00) {
150*ad30f8e7SGabor Kovesdan 			psenc->current_endian = _ENDIAN_LITTLE;
151*ad30f8e7SGabor Kovesdan 			chlenbak = 0;
152*ad30f8e7SGabor Kovesdan 			goto refetch;
153*ad30f8e7SGabor Kovesdan 		}
154*ad30f8e7SGabor Kovesdan 	}
155*ad30f8e7SGabor Kovesdan 	endian = ((ei->mode & _MODE_FORCE_ENDIAN) != 0 ||
156*ad30f8e7SGabor Kovesdan 	    psenc->current_endian == _ENDIAN_UNKNOWN) ? ei->preffered_endian :
157*ad30f8e7SGabor Kovesdan 	    psenc->current_endian;
158*ad30f8e7SGabor Kovesdan 
159*ad30f8e7SGabor Kovesdan 	/* get wc */
160*ad30f8e7SGabor Kovesdan 	if ((ei->mode & _MODE_UTF32) == 0) {
161*ad30f8e7SGabor Kovesdan 		/* UTF16 */
162*ad30f8e7SGabor Kovesdan 		if (needlen == 2) {
163*ad30f8e7SGabor Kovesdan 			switch (endian) {
164*ad30f8e7SGabor Kovesdan 			case _ENDIAN_LITTLE:
165*ad30f8e7SGabor Kovesdan 				wc = (psenc->ch[0] |
166*ad30f8e7SGabor Kovesdan 				    ((wchar_t)psenc->ch[1] << 8));
167*ad30f8e7SGabor Kovesdan 				break;
168*ad30f8e7SGabor Kovesdan 			case _ENDIAN_BIG:
169*ad30f8e7SGabor Kovesdan 				wc = (psenc->ch[1] |
170*ad30f8e7SGabor Kovesdan 				    ((wchar_t)psenc->ch[0] << 8));
171*ad30f8e7SGabor Kovesdan 				break;
172*ad30f8e7SGabor Kovesdan 			default:
173*ad30f8e7SGabor Kovesdan 				goto ilseq;
174*ad30f8e7SGabor Kovesdan 			}
175*ad30f8e7SGabor Kovesdan 			if (wc >= 0xD800 && wc <= 0xDBFF) {
176*ad30f8e7SGabor Kovesdan 				/* surrogate high */
177*ad30f8e7SGabor Kovesdan 				needlen = 4;
178*ad30f8e7SGabor Kovesdan 				goto refetch;
179*ad30f8e7SGabor Kovesdan 			}
180*ad30f8e7SGabor Kovesdan 		} else {
181*ad30f8e7SGabor Kovesdan 			/* surrogate low */
182*ad30f8e7SGabor Kovesdan 			wc -= 0xD800; /* wc : surrogate high (see above) */
183*ad30f8e7SGabor Kovesdan 			wc <<= 10;
184*ad30f8e7SGabor Kovesdan 			switch (endian) {
185*ad30f8e7SGabor Kovesdan 			case _ENDIAN_LITTLE:
186*ad30f8e7SGabor Kovesdan 				if (psenc->ch[3] < 0xDC || psenc->ch[3] > 0xDF)
187*ad30f8e7SGabor Kovesdan 					goto ilseq;
188*ad30f8e7SGabor Kovesdan 				wc |= psenc->ch[2];
189*ad30f8e7SGabor Kovesdan 				wc |= (wchar_t)(psenc->ch[3] & 3) << 8;
190*ad30f8e7SGabor Kovesdan 				break;
191*ad30f8e7SGabor Kovesdan 			case _ENDIAN_BIG:
192*ad30f8e7SGabor Kovesdan 				if (psenc->ch[2]<0xDC || psenc->ch[2]>0xDF)
193*ad30f8e7SGabor Kovesdan 					goto ilseq;
194*ad30f8e7SGabor Kovesdan 				wc |= psenc->ch[3];
195*ad30f8e7SGabor Kovesdan 				wc |= (wchar_t)(psenc->ch[2] & 3) << 8;
196*ad30f8e7SGabor Kovesdan 				break;
197*ad30f8e7SGabor Kovesdan 			default:
198*ad30f8e7SGabor Kovesdan 				goto ilseq;
199*ad30f8e7SGabor Kovesdan 			}
200*ad30f8e7SGabor Kovesdan 			wc += 0x10000;
201*ad30f8e7SGabor Kovesdan 		}
202*ad30f8e7SGabor Kovesdan 	} else {
203*ad30f8e7SGabor Kovesdan 		/* UTF32 */
204*ad30f8e7SGabor Kovesdan 		switch (endian) {
205*ad30f8e7SGabor Kovesdan 		case _ENDIAN_LITTLE:
206*ad30f8e7SGabor Kovesdan 			wc = (psenc->ch[0] |
207*ad30f8e7SGabor Kovesdan 			    ((wchar_t)psenc->ch[1] << 8) |
208*ad30f8e7SGabor Kovesdan 			    ((wchar_t)psenc->ch[2] << 16) |
209*ad30f8e7SGabor Kovesdan 			    ((wchar_t)psenc->ch[3] << 24));
210*ad30f8e7SGabor Kovesdan 			break;
211*ad30f8e7SGabor Kovesdan 		case _ENDIAN_BIG:
212*ad30f8e7SGabor Kovesdan 			wc = (psenc->ch[3] |
213*ad30f8e7SGabor Kovesdan 			    ((wchar_t)psenc->ch[2] << 8) |
214*ad30f8e7SGabor Kovesdan 			    ((wchar_t)psenc->ch[1] << 16) |
215*ad30f8e7SGabor Kovesdan 			    ((wchar_t)psenc->ch[0] << 24));
216*ad30f8e7SGabor Kovesdan 			break;
217*ad30f8e7SGabor Kovesdan 		default:
218*ad30f8e7SGabor Kovesdan 			goto ilseq;
219*ad30f8e7SGabor Kovesdan 		}
220*ad30f8e7SGabor Kovesdan 		if (wc >= 0xD800 && wc <= 0xDFFF)
221*ad30f8e7SGabor Kovesdan 			goto ilseq;
222*ad30f8e7SGabor Kovesdan 	}
223*ad30f8e7SGabor Kovesdan 
224*ad30f8e7SGabor Kovesdan 
225*ad30f8e7SGabor Kovesdan 	*pwc = wc;
226*ad30f8e7SGabor Kovesdan 	psenc->chlen = 0;
227*ad30f8e7SGabor Kovesdan 	*nresult = result;
228*ad30f8e7SGabor Kovesdan 	*s = s0;
229*ad30f8e7SGabor Kovesdan 
230*ad30f8e7SGabor Kovesdan 	return (0);
231*ad30f8e7SGabor Kovesdan 
232*ad30f8e7SGabor Kovesdan ilseq:
233*ad30f8e7SGabor Kovesdan 	*nresult = (size_t)-1;
234*ad30f8e7SGabor Kovesdan 	psenc->chlen = 0;
235*ad30f8e7SGabor Kovesdan 	return (EILSEQ);
236*ad30f8e7SGabor Kovesdan 
237*ad30f8e7SGabor Kovesdan restart:
238*ad30f8e7SGabor Kovesdan 	*nresult = (size_t)-2;
239*ad30f8e7SGabor Kovesdan 	psenc->chlen = chlenbak;
240*ad30f8e7SGabor Kovesdan 	*s = s0;
241*ad30f8e7SGabor Kovesdan 	return (0);
242*ad30f8e7SGabor Kovesdan }
243*ad30f8e7SGabor Kovesdan 
244*ad30f8e7SGabor Kovesdan static int
245*ad30f8e7SGabor Kovesdan _citrus_UTF1632_wcrtomb_priv(_UTF1632EncodingInfo *ei, char *s, size_t n,
246*ad30f8e7SGabor Kovesdan     wchar_t wc, _UTF1632State *psenc, size_t *nresult)
247*ad30f8e7SGabor Kovesdan {
248*ad30f8e7SGabor Kovesdan 	wchar_t wc2;
249*ad30f8e7SGabor Kovesdan 	static const char _bom[4] = {
250*ad30f8e7SGabor Kovesdan 	    0x00, 0x00, 0xFE, 0xFF,
251*ad30f8e7SGabor Kovesdan 	};
252*ad30f8e7SGabor Kovesdan 	const char *bom = &_bom[0];
253*ad30f8e7SGabor Kovesdan 	size_t cnt;
254*ad30f8e7SGabor Kovesdan 
255*ad30f8e7SGabor Kovesdan 	cnt = (size_t)0;
256*ad30f8e7SGabor Kovesdan 	if (psenc->current_endian == _ENDIAN_UNKNOWN) {
257*ad30f8e7SGabor Kovesdan 		if ((ei->mode & _MODE_FORCE_ENDIAN) == 0) {
258*ad30f8e7SGabor Kovesdan 			if (ei->mode & _MODE_UTF32)
259*ad30f8e7SGabor Kovesdan 				cnt = 4;
260*ad30f8e7SGabor Kovesdan 			else {
261*ad30f8e7SGabor Kovesdan 				cnt = 2;
262*ad30f8e7SGabor Kovesdan 				bom += 2;
263*ad30f8e7SGabor Kovesdan 			}
264*ad30f8e7SGabor Kovesdan 			if (n < cnt)
265*ad30f8e7SGabor Kovesdan 				goto e2big;
266*ad30f8e7SGabor Kovesdan 			memcpy(s, bom, cnt);
267*ad30f8e7SGabor Kovesdan 			s += cnt, n -= cnt;
268*ad30f8e7SGabor Kovesdan 		}
269*ad30f8e7SGabor Kovesdan 		psenc->current_endian = ei->preffered_endian;
270*ad30f8e7SGabor Kovesdan 	}
271*ad30f8e7SGabor Kovesdan 
272*ad30f8e7SGabor Kovesdan 	wc2 = 0;
273*ad30f8e7SGabor Kovesdan 	if ((ei->mode & _MODE_UTF32)==0) {
274*ad30f8e7SGabor Kovesdan 		/* UTF16 */
275*ad30f8e7SGabor Kovesdan 		if (wc > 0xFFFF) {
276*ad30f8e7SGabor Kovesdan 			/* surrogate */
277*ad30f8e7SGabor Kovesdan 			if (wc > 0x10FFFF)
278*ad30f8e7SGabor Kovesdan 				goto ilseq;
279*ad30f8e7SGabor Kovesdan 			if (n < 4)
280*ad30f8e7SGabor Kovesdan 				goto e2big;
281*ad30f8e7SGabor Kovesdan 			cnt += 4;
282*ad30f8e7SGabor Kovesdan 			wc -= 0x10000;
283*ad30f8e7SGabor Kovesdan 			wc2 = (wc & 0x3FF) | 0xDC00;
284*ad30f8e7SGabor Kovesdan 			wc = (wc>>10) | 0xD800;
285*ad30f8e7SGabor Kovesdan 		} else {
286*ad30f8e7SGabor Kovesdan 			if (n < 2)
287*ad30f8e7SGabor Kovesdan 				goto e2big;
288*ad30f8e7SGabor Kovesdan 			cnt += 2;
289*ad30f8e7SGabor Kovesdan 		}
290*ad30f8e7SGabor Kovesdan 
291*ad30f8e7SGabor Kovesdan surrogate:
292*ad30f8e7SGabor Kovesdan 		switch (psenc->current_endian) {
293*ad30f8e7SGabor Kovesdan 		case _ENDIAN_BIG:
294*ad30f8e7SGabor Kovesdan 			s[1] = wc;
295*ad30f8e7SGabor Kovesdan 			s[0] = (wc >>= 8);
296*ad30f8e7SGabor Kovesdan 			break;
297*ad30f8e7SGabor Kovesdan 		case _ENDIAN_LITTLE:
298*ad30f8e7SGabor Kovesdan 			s[0] = wc;
299*ad30f8e7SGabor Kovesdan 			s[1] = (wc >>= 8);
300*ad30f8e7SGabor Kovesdan 			break;
301*ad30f8e7SGabor Kovesdan 		}
302*ad30f8e7SGabor Kovesdan 		if (wc2 != 0) {
303*ad30f8e7SGabor Kovesdan 			wc = wc2;
304*ad30f8e7SGabor Kovesdan 			wc2 = 0;
305*ad30f8e7SGabor Kovesdan 			s += 2;
306*ad30f8e7SGabor Kovesdan 			goto surrogate;
307*ad30f8e7SGabor Kovesdan 		}
308*ad30f8e7SGabor Kovesdan 	} else {
309*ad30f8e7SGabor Kovesdan 		/* UTF32 */
310*ad30f8e7SGabor Kovesdan 		if (wc >= 0xD800 && wc <= 0xDFFF)
311*ad30f8e7SGabor Kovesdan 			goto ilseq;
312*ad30f8e7SGabor Kovesdan 		if (n < 4)
313*ad30f8e7SGabor Kovesdan 			goto e2big;
314*ad30f8e7SGabor Kovesdan 		cnt += 4;
315*ad30f8e7SGabor Kovesdan 		switch (psenc->current_endian) {
316*ad30f8e7SGabor Kovesdan 		case _ENDIAN_BIG:
317*ad30f8e7SGabor Kovesdan 			s[3] = wc;
318*ad30f8e7SGabor Kovesdan 			s[2] = (wc >>= 8);
319*ad30f8e7SGabor Kovesdan 			s[1] = (wc >>= 8);
320*ad30f8e7SGabor Kovesdan 			s[0] = (wc >>= 8);
321*ad30f8e7SGabor Kovesdan 			break;
322*ad30f8e7SGabor Kovesdan 		case _ENDIAN_LITTLE:
323*ad30f8e7SGabor Kovesdan 			s[0] = wc;
324*ad30f8e7SGabor Kovesdan 			s[1] = (wc >>= 8);
325*ad30f8e7SGabor Kovesdan 			s[2] = (wc >>= 8);
326*ad30f8e7SGabor Kovesdan 			s[3] = (wc >>= 8);
327*ad30f8e7SGabor Kovesdan 			break;
328*ad30f8e7SGabor Kovesdan 		}
329*ad30f8e7SGabor Kovesdan 	}
330*ad30f8e7SGabor Kovesdan 	*nresult = cnt;
331*ad30f8e7SGabor Kovesdan 
332*ad30f8e7SGabor Kovesdan 	return (0);
333*ad30f8e7SGabor Kovesdan 
334*ad30f8e7SGabor Kovesdan ilseq:
335*ad30f8e7SGabor Kovesdan 	*nresult = (size_t)-1;
336*ad30f8e7SGabor Kovesdan 	return (EILSEQ);
337*ad30f8e7SGabor Kovesdan e2big:
338*ad30f8e7SGabor Kovesdan 	*nresult = (size_t)-1;
339*ad30f8e7SGabor Kovesdan 	return (E2BIG);
340*ad30f8e7SGabor Kovesdan }
341*ad30f8e7SGabor Kovesdan 
342*ad30f8e7SGabor Kovesdan static void
343*ad30f8e7SGabor Kovesdan parse_variable(_UTF1632EncodingInfo * __restrict ei,
344*ad30f8e7SGabor Kovesdan     const void * __restrict var, size_t lenvar)
345*ad30f8e7SGabor Kovesdan {
346*ad30f8e7SGabor Kovesdan 	const char *p;
347*ad30f8e7SGabor Kovesdan 
348*ad30f8e7SGabor Kovesdan 	p = var;
349*ad30f8e7SGabor Kovesdan 	while (lenvar > 0) {
350*ad30f8e7SGabor Kovesdan 		switch (*p) {
351*ad30f8e7SGabor Kovesdan 		case 'B':
352*ad30f8e7SGabor Kovesdan 		case 'b':
353*ad30f8e7SGabor Kovesdan 			MATCH(big, ei->preffered_endian = _ENDIAN_BIG);
354*ad30f8e7SGabor Kovesdan 			break;
355*ad30f8e7SGabor Kovesdan 		case 'L':
356*ad30f8e7SGabor Kovesdan 		case 'l':
357*ad30f8e7SGabor Kovesdan 			MATCH(little, ei->preffered_endian = _ENDIAN_LITTLE);
358*ad30f8e7SGabor Kovesdan 			break;
359*ad30f8e7SGabor Kovesdan 		case 'i':
360*ad30f8e7SGabor Kovesdan 		case 'I':
361*ad30f8e7SGabor Kovesdan 			MATCH(internal, ei->preffered_endian = _ENDIAN_INTERNAL);
362*ad30f8e7SGabor Kovesdan 			break;
363*ad30f8e7SGabor Kovesdan 		case 's':
364*ad30f8e7SGabor Kovesdan 		case 'S':
365*ad30f8e7SGabor Kovesdan 			MATCH(swapped, ei->preffered_endian = _ENDIAN_SWAPPED);
366*ad30f8e7SGabor Kovesdan 			break;
367*ad30f8e7SGabor Kovesdan 		case 'F':
368*ad30f8e7SGabor Kovesdan 		case 'f':
369*ad30f8e7SGabor Kovesdan 			MATCH(force, ei->mode |= _MODE_FORCE_ENDIAN);
370*ad30f8e7SGabor Kovesdan 			break;
371*ad30f8e7SGabor Kovesdan 		case 'U':
372*ad30f8e7SGabor Kovesdan 		case 'u':
373*ad30f8e7SGabor Kovesdan 			MATCH(utf32, ei->mode |= _MODE_UTF32);
374*ad30f8e7SGabor Kovesdan 			break;
375*ad30f8e7SGabor Kovesdan 		}
376*ad30f8e7SGabor Kovesdan 		p++;
377*ad30f8e7SGabor Kovesdan 		lenvar--;
378*ad30f8e7SGabor Kovesdan 	}
379*ad30f8e7SGabor Kovesdan }
380*ad30f8e7SGabor Kovesdan 
381*ad30f8e7SGabor Kovesdan static int
382*ad30f8e7SGabor Kovesdan /*ARGSUSED*/
383*ad30f8e7SGabor Kovesdan _citrus_UTF1632_encoding_module_init(_UTF1632EncodingInfo * __restrict ei,
384*ad30f8e7SGabor Kovesdan     const void * __restrict var, size_t lenvar)
385*ad30f8e7SGabor Kovesdan {
386*ad30f8e7SGabor Kovesdan 
387*ad30f8e7SGabor Kovesdan 	memset((void *)ei, 0, sizeof(*ei));
388*ad30f8e7SGabor Kovesdan 
389*ad30f8e7SGabor Kovesdan 	parse_variable(ei, var, lenvar);
390*ad30f8e7SGabor Kovesdan 
391*ad30f8e7SGabor Kovesdan 	ei->cur_max = ((ei->mode&_MODE_UTF32) == 0) ? 6 : 8;
392*ad30f8e7SGabor Kovesdan 	/* 6: endian + surrogate */
393*ad30f8e7SGabor Kovesdan 	/* 8: endian + normal */
394*ad30f8e7SGabor Kovesdan 
395*ad30f8e7SGabor Kovesdan 	if (ei->preffered_endian == _ENDIAN_UNKNOWN) {
396*ad30f8e7SGabor Kovesdan 		ei->preffered_endian = _ENDIAN_BIG;
397*ad30f8e7SGabor Kovesdan 	}
398*ad30f8e7SGabor Kovesdan 
399*ad30f8e7SGabor Kovesdan 	return (0);
400*ad30f8e7SGabor Kovesdan }
401*ad30f8e7SGabor Kovesdan 
402*ad30f8e7SGabor Kovesdan static void
403*ad30f8e7SGabor Kovesdan /*ARGSUSED*/
404*ad30f8e7SGabor Kovesdan _citrus_UTF1632_encoding_module_uninit(_UTF1632EncodingInfo *ei __unused)
405*ad30f8e7SGabor Kovesdan {
406*ad30f8e7SGabor Kovesdan 
407*ad30f8e7SGabor Kovesdan }
408*ad30f8e7SGabor Kovesdan 
409*ad30f8e7SGabor Kovesdan static __inline int
410*ad30f8e7SGabor Kovesdan /*ARGSUSED*/
411*ad30f8e7SGabor Kovesdan _citrus_UTF1632_stdenc_wctocs(_UTF1632EncodingInfo * __restrict ei __unused,
412*ad30f8e7SGabor Kovesdan      _csid_t * __restrict csid, _index_t * __restrict idx, _wc_t wc)
413*ad30f8e7SGabor Kovesdan {
414*ad30f8e7SGabor Kovesdan 
415*ad30f8e7SGabor Kovesdan 	*csid = 0;
416*ad30f8e7SGabor Kovesdan 	*idx = (_index_t)wc;
417*ad30f8e7SGabor Kovesdan 
418*ad30f8e7SGabor Kovesdan 	return (0);
419*ad30f8e7SGabor Kovesdan }
420*ad30f8e7SGabor Kovesdan 
421*ad30f8e7SGabor Kovesdan static __inline int
422*ad30f8e7SGabor Kovesdan /*ARGSUSED*/
423*ad30f8e7SGabor Kovesdan _citrus_UTF1632_stdenc_cstowc(_UTF1632EncodingInfo * __restrict ei __unused,
424*ad30f8e7SGabor Kovesdan     _wc_t * __restrict wc, _csid_t csid, _index_t idx)
425*ad30f8e7SGabor Kovesdan {
426*ad30f8e7SGabor Kovesdan 
427*ad30f8e7SGabor Kovesdan 	if (csid != 0)
428*ad30f8e7SGabor Kovesdan 		return (EILSEQ);
429*ad30f8e7SGabor Kovesdan 
430*ad30f8e7SGabor Kovesdan 	*wc = (_wc_t)idx;
431*ad30f8e7SGabor Kovesdan 
432*ad30f8e7SGabor Kovesdan 	return (0);
433*ad30f8e7SGabor Kovesdan }
434*ad30f8e7SGabor Kovesdan 
435*ad30f8e7SGabor Kovesdan static __inline int
436*ad30f8e7SGabor Kovesdan /*ARGSUSED*/
437*ad30f8e7SGabor Kovesdan _citrus_UTF1632_stdenc_get_state_desc_generic(_UTF1632EncodingInfo * __restrict ei __unused,
438*ad30f8e7SGabor Kovesdan     _UTF1632State * __restrict psenc, int * __restrict rstate)
439*ad30f8e7SGabor Kovesdan {
440*ad30f8e7SGabor Kovesdan 
441*ad30f8e7SGabor Kovesdan 	*rstate = (psenc->chlen == 0) ? _STDENC_SDGEN_INITIAL :
442*ad30f8e7SGabor Kovesdan 	    _STDENC_SDGEN_INCOMPLETE_CHAR;
443*ad30f8e7SGabor Kovesdan 	return (0);
444*ad30f8e7SGabor Kovesdan }
445*ad30f8e7SGabor Kovesdan 
446*ad30f8e7SGabor Kovesdan /* ----------------------------------------------------------------------
447*ad30f8e7SGabor Kovesdan  * public interface for stdenc
448*ad30f8e7SGabor Kovesdan  */
449*ad30f8e7SGabor Kovesdan 
450*ad30f8e7SGabor Kovesdan _CITRUS_STDENC_DECLS(UTF1632);
451*ad30f8e7SGabor Kovesdan _CITRUS_STDENC_DEF_OPS(UTF1632);
452*ad30f8e7SGabor Kovesdan 
453*ad30f8e7SGabor Kovesdan #include "citrus_stdenc_template.h"
454