1*16d86563SAlexander Pyhalov /*
2*16d86563SAlexander Pyhalov  * CDDL HEADER START
3*16d86563SAlexander Pyhalov  *
4*16d86563SAlexander Pyhalov  * The contents of this file are subject to the terms of the
5*16d86563SAlexander Pyhalov  * Common Development and Distribution License (the "License").
6*16d86563SAlexander Pyhalov  * You may not use this file except in compliance with the License.
7*16d86563SAlexander Pyhalov  *
8*16d86563SAlexander Pyhalov  * You can obtain a copy of the license at src/OPENSOLARIS.LICENSE
9*16d86563SAlexander Pyhalov  * or http://www.opensolaris.org/os/licensing.
10*16d86563SAlexander Pyhalov  * See the License for the specific language governing permissions
11*16d86563SAlexander Pyhalov  * and limitations under the License.
12*16d86563SAlexander Pyhalov  *
13*16d86563SAlexander Pyhalov  * When distributing Covered Code, include this CDDL HEADER in each
14*16d86563SAlexander Pyhalov  * file and include the License file at src/OPENSOLARIS.LICENSE.
15*16d86563SAlexander Pyhalov  * If applicable, add the following below this CDDL HEADER, with the
16*16d86563SAlexander Pyhalov  * fields enclosed by brackets "[]" replaced with your own identifying
17*16d86563SAlexander Pyhalov  * information: Portions Copyright [yyyy] [name of copyright owner]
18*16d86563SAlexander Pyhalov  *
19*16d86563SAlexander Pyhalov  * CDDL HEADER END
20*16d86563SAlexander Pyhalov  */
21*16d86563SAlexander Pyhalov /*
22*16d86563SAlexander Pyhalov  * Copyright 2004 Sun Microsystems, Inc.  All rights reserved.
23*16d86563SAlexander Pyhalov  * Use is subject to license terms.
24*16d86563SAlexander Pyhalov  *
25*16d86563SAlexander Pyhalov  * This particular file is to cover conversions from UCS-4, UCS-4BE, and
26*16d86563SAlexander Pyhalov  * UCS-4LE to UTF-32, UTF-32BE, and UTF-32LE.
27*16d86563SAlexander Pyhalov  */
28*16d86563SAlexander Pyhalov 
29*16d86563SAlexander Pyhalov 
30*16d86563SAlexander Pyhalov #include <stdlib.h>
31*16d86563SAlexander Pyhalov #include <errno.h>
32*16d86563SAlexander Pyhalov #include <sys/types.h>
33*16d86563SAlexander Pyhalov #include <sys/isa_defs.h>
34*16d86563SAlexander Pyhalov 
35*16d86563SAlexander Pyhalov /* We include the ucs4_to_ucs.h at the moment. */
36*16d86563SAlexander Pyhalov #include "ucs4_to_ucs.h"
37*16d86563SAlexander Pyhalov 
38*16d86563SAlexander Pyhalov 
39*16d86563SAlexander Pyhalov void *
_icv_open()40*16d86563SAlexander Pyhalov _icv_open()
41*16d86563SAlexander Pyhalov {
42*16d86563SAlexander Pyhalov 	ucs_ucs_state_t *cd;
43*16d86563SAlexander Pyhalov 
44*16d86563SAlexander Pyhalov 	cd = (ucs_ucs_state_t *)calloc(1, sizeof(ucs_ucs_state_t));
45*16d86563SAlexander Pyhalov 	if (cd == (ucs_ucs_state_t *)NULL) {
46*16d86563SAlexander Pyhalov 		errno = ENOMEM;
47*16d86563SAlexander Pyhalov 		return((void *)-1);
48*16d86563SAlexander Pyhalov 	}
49*16d86563SAlexander Pyhalov 
50*16d86563SAlexander Pyhalov #if defined(UCS_4BE)
51*16d86563SAlexander Pyhalov 	cd->input.little_endian = false;
52*16d86563SAlexander Pyhalov 	cd->input.bom_written = true;
53*16d86563SAlexander Pyhalov #elif defined(UCS_4LE)
54*16d86563SAlexander Pyhalov 	cd->input.little_endian = true;
55*16d86563SAlexander Pyhalov 	cd->input.bom_written = true;
56*16d86563SAlexander Pyhalov #elif defined(_LITTLE_ENDIAN)
57*16d86563SAlexander Pyhalov 	cd->input.little_endian = true;
58*16d86563SAlexander Pyhalov #endif
59*16d86563SAlexander Pyhalov 
60*16d86563SAlexander Pyhalov #if defined(UTF_32BE)
61*16d86563SAlexander Pyhalov 	cd->output.little_endian = false;
62*16d86563SAlexander Pyhalov 	cd->output.bom_written = true;
63*16d86563SAlexander Pyhalov #elif defined(UTF_32LE)
64*16d86563SAlexander Pyhalov 	cd->output.little_endian = true;
65*16d86563SAlexander Pyhalov 	cd->output.bom_written = true;
66*16d86563SAlexander Pyhalov #elif defined(_LITTLE_ENDIAN)
67*16d86563SAlexander Pyhalov 	cd->output.little_endian = true;
68*16d86563SAlexander Pyhalov #endif
69*16d86563SAlexander Pyhalov 
70*16d86563SAlexander Pyhalov 	return((void *)cd);
71*16d86563SAlexander Pyhalov }
72*16d86563SAlexander Pyhalov 
73*16d86563SAlexander Pyhalov 
74*16d86563SAlexander Pyhalov void
_icv_close(ucs_ucs_state_t * cd)75*16d86563SAlexander Pyhalov _icv_close(ucs_ucs_state_t *cd)
76*16d86563SAlexander Pyhalov {
77*16d86563SAlexander Pyhalov 	if (! cd)
78*16d86563SAlexander Pyhalov 		errno = EBADF;
79*16d86563SAlexander Pyhalov 	else
80*16d86563SAlexander Pyhalov 		free((void *)cd);
81*16d86563SAlexander Pyhalov }
82*16d86563SAlexander Pyhalov 
83*16d86563SAlexander Pyhalov 
84*16d86563SAlexander Pyhalov size_t
_icv_iconv(ucs_ucs_state_t * cd,char ** inbuf,size_t * inbufleft,char ** outbuf,size_t * outbufleft)85*16d86563SAlexander Pyhalov _icv_iconv(ucs_ucs_state_t *cd, char **inbuf, size_t *inbufleft, char **outbuf,
86*16d86563SAlexander Pyhalov                 size_t *outbufleft)
87*16d86563SAlexander Pyhalov {
88*16d86563SAlexander Pyhalov 	size_t ret_val = 0;
89*16d86563SAlexander Pyhalov 	uchar_t *ib;
90*16d86563SAlexander Pyhalov 	uchar_t *ob;
91*16d86563SAlexander Pyhalov 	uchar_t *ibtail;
92*16d86563SAlexander Pyhalov 	uchar_t *obtail;
93*16d86563SAlexander Pyhalov 	uint_t u4;
94*16d86563SAlexander Pyhalov 	signed char obsz;
95*16d86563SAlexander Pyhalov 	int i;
96*16d86563SAlexander Pyhalov 
97*16d86563SAlexander Pyhalov 
98*16d86563SAlexander Pyhalov 	if (! cd) {
99*16d86563SAlexander Pyhalov 		errno = EBADF;
100*16d86563SAlexander Pyhalov 		return((size_t)-1);
101*16d86563SAlexander Pyhalov 	}
102*16d86563SAlexander Pyhalov 
103*16d86563SAlexander Pyhalov 	if (!inbuf || !(*inbuf)) {
104*16d86563SAlexander Pyhalov #if defined(UCS_4)
105*16d86563SAlexander Pyhalov 		cd->input.bom_written = false;
106*16d86563SAlexander Pyhalov #endif
107*16d86563SAlexander Pyhalov #if defined(UTF_32)
108*16d86563SAlexander Pyhalov 		cd->output.bom_written = false;
109*16d86563SAlexander Pyhalov #endif
110*16d86563SAlexander Pyhalov 		return((size_t)0);
111*16d86563SAlexander Pyhalov 	}
112*16d86563SAlexander Pyhalov 
113*16d86563SAlexander Pyhalov 	ib = (uchar_t *)*inbuf;
114*16d86563SAlexander Pyhalov 	ob = (uchar_t *)*outbuf;
115*16d86563SAlexander Pyhalov 	ibtail = ib + *inbufleft;
116*16d86563SAlexander Pyhalov 	obtail = ob + *outbufleft;
117*16d86563SAlexander Pyhalov 
118*16d86563SAlexander Pyhalov #if defined(UCS_4)
119*16d86563SAlexander Pyhalov 	if (! cd->input.bom_written) {
120*16d86563SAlexander Pyhalov 		if ((ibtail - ib) < ICV_FETCH_UCS4_SIZE) {
121*16d86563SAlexander Pyhalov 			errno = EINVAL;
122*16d86563SAlexander Pyhalov 			ret_val = (size_t)-1;
123*16d86563SAlexander Pyhalov 			goto need_more_input_err;
124*16d86563SAlexander Pyhalov 		}
125*16d86563SAlexander Pyhalov 
126*16d86563SAlexander Pyhalov 		for (u4 = 0, i = 0; i < ICV_FETCH_UCS4_SIZE; i++)
127*16d86563SAlexander Pyhalov 			u4 = (u4 << 8) | ((uint_t)(*(ib + i)));
128*16d86563SAlexander Pyhalov 
129*16d86563SAlexander Pyhalov 		if (u4 == ICV_BOM_IN_BIG_ENDIAN) {
130*16d86563SAlexander Pyhalov 			ib += ICV_FETCH_UCS4_SIZE;
131*16d86563SAlexander Pyhalov 			cd->input.little_endian = false;
132*16d86563SAlexander Pyhalov 		} else if (u4 == ICV_BOM_IN_LITTLE_ENDIAN_UCS4) {
133*16d86563SAlexander Pyhalov 			ib += ICV_FETCH_UCS4_SIZE;
134*16d86563SAlexander Pyhalov 			cd->input.little_endian = true;
135*16d86563SAlexander Pyhalov 		}
136*16d86563SAlexander Pyhalov 	}
137*16d86563SAlexander Pyhalov 	cd->input.bom_written = true;
138*16d86563SAlexander Pyhalov #endif
139*16d86563SAlexander Pyhalov 
140*16d86563SAlexander Pyhalov 
141*16d86563SAlexander Pyhalov 	while (ib < ibtail) {
142*16d86563SAlexander Pyhalov 		if ((ibtail - ib) < ICV_FETCH_UCS4_SIZE) {
143*16d86563SAlexander Pyhalov 			errno = EINVAL;
144*16d86563SAlexander Pyhalov 			ret_val = (size_t)-1;
145*16d86563SAlexander Pyhalov 			break;
146*16d86563SAlexander Pyhalov 		}
147*16d86563SAlexander Pyhalov 
148*16d86563SAlexander Pyhalov 		u4 = 0;
149*16d86563SAlexander Pyhalov 		if (cd->input.little_endian) {
150*16d86563SAlexander Pyhalov 			for (i = ICV_FETCH_UCS4_SIZE - 1; i >= 0; i--)
151*16d86563SAlexander Pyhalov 				u4 = (u4 << 8) | ((uint_t)(*(ib + i)));
152*16d86563SAlexander Pyhalov 		} else {
153*16d86563SAlexander Pyhalov 			for (i = 0; i < ICV_FETCH_UCS4_SIZE; i++)
154*16d86563SAlexander Pyhalov 				u4 = (u4 << 8) | ((uint_t)(*(ib + i)));
155*16d86563SAlexander Pyhalov 		}
156*16d86563SAlexander Pyhalov 
157*16d86563SAlexander Pyhalov 		if (u4 == 0x00fffe || u4 == 0x00ffff || u4 > 0x7fffffff ||
158*16d86563SAlexander Pyhalov 		    (u4 >= 0x00d800 && u4 <= 0x00dfff)) {
159*16d86563SAlexander Pyhalov 			errno = EILSEQ;
160*16d86563SAlexander Pyhalov 			ret_val = (size_t)-1;
161*16d86563SAlexander Pyhalov 			goto illegal_char_err;
162*16d86563SAlexander Pyhalov 		}
163*16d86563SAlexander Pyhalov 
164*16d86563SAlexander Pyhalov 		if (u4 > 0x10ffff) {
165*16d86563SAlexander Pyhalov 			u4 = ICV_CHAR_UCS2_REPLACEMENT;
166*16d86563SAlexander Pyhalov 			ret_val++;
167*16d86563SAlexander Pyhalov 		}
168*16d86563SAlexander Pyhalov 
169*16d86563SAlexander Pyhalov 		obsz = (cd->output.bom_written) ? 2 : 4;
170*16d86563SAlexander Pyhalov 		if ((obtail - ob) < obsz) {
171*16d86563SAlexander Pyhalov 			errno = E2BIG;
172*16d86563SAlexander Pyhalov 			ret_val = (size_t)-1;
173*16d86563SAlexander Pyhalov 			break;
174*16d86563SAlexander Pyhalov 		}
175*16d86563SAlexander Pyhalov 
176*16d86563SAlexander Pyhalov 		if (cd->output.little_endian) {
177*16d86563SAlexander Pyhalov 			if (! cd->output.bom_written) {
178*16d86563SAlexander Pyhalov 				*ob++ = (uchar_t)0xff;
179*16d86563SAlexander Pyhalov 				*ob++ = (uchar_t)0xfe;
180*16d86563SAlexander Pyhalov 				*(ushort_t *)ob = (ushort_t)0;
181*16d86563SAlexander Pyhalov 				ob += 2;
182*16d86563SAlexander Pyhalov 				cd->output.bom_written = true;
183*16d86563SAlexander Pyhalov 			}
184*16d86563SAlexander Pyhalov 			*ob++ = (uchar_t)(u4 & 0xff);
185*16d86563SAlexander Pyhalov 			*ob++ = (uchar_t)((u4 >> 8) & 0xff);
186*16d86563SAlexander Pyhalov 			*ob++ = (uchar_t)((u4 >> 16) & 0xff);
187*16d86563SAlexander Pyhalov 			*ob++ = (uchar_t)((u4 >> 24) & 0xff);
188*16d86563SAlexander Pyhalov 		} else {
189*16d86563SAlexander Pyhalov 			if (! cd->output.bom_written) {
190*16d86563SAlexander Pyhalov 				*(ushort_t *)ob = (ushort_t)0;
191*16d86563SAlexander Pyhalov 				ob += 2;
192*16d86563SAlexander Pyhalov 				*ob++ = (uchar_t)0xfe;
193*16d86563SAlexander Pyhalov 				*ob++ = (uchar_t)0xff;
194*16d86563SAlexander Pyhalov 				cd->output.bom_written = true;
195*16d86563SAlexander Pyhalov 			}
196*16d86563SAlexander Pyhalov 			*ob++ = (uchar_t)((u4 >> 24) & 0xff);
197*16d86563SAlexander Pyhalov 			*ob++ = (uchar_t)((u4 >> 16) & 0xff);
198*16d86563SAlexander Pyhalov 			*ob++ = (uchar_t)((u4 >> 8) & 0xff);
199*16d86563SAlexander Pyhalov 			*ob++ = (uchar_t)(u4 & 0xff);
200*16d86563SAlexander Pyhalov 		}
201*16d86563SAlexander Pyhalov 		ib += ICV_FETCH_UCS4_SIZE;
202*16d86563SAlexander Pyhalov 	}
203*16d86563SAlexander Pyhalov 
204*16d86563SAlexander Pyhalov #if defined(UCS_4)
205*16d86563SAlexander Pyhalov need_more_input_err:
206*16d86563SAlexander Pyhalov #endif
207*16d86563SAlexander Pyhalov illegal_char_err:
208*16d86563SAlexander Pyhalov 	*inbuf = (char *)ib;
209*16d86563SAlexander Pyhalov 	*inbufleft = ibtail - ib;
210*16d86563SAlexander Pyhalov 	*outbuf = (char *)ob;
211*16d86563SAlexander Pyhalov 	*outbufleft = obtail - ob;
212*16d86563SAlexander Pyhalov 
213*16d86563SAlexander Pyhalov 	return(ret_val);
214*16d86563SAlexander Pyhalov }
215