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 UTF-8 to various single
26*16d86563SAlexander Pyhalov  * byte codesets.
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 "utf8_to_sb.h"
34*16d86563SAlexander Pyhalov 
35*16d86563SAlexander Pyhalov 
36*16d86563SAlexander Pyhalov 
37*16d86563SAlexander Pyhalov void *
_icv_open()38*16d86563SAlexander Pyhalov _icv_open()
39*16d86563SAlexander Pyhalov {
40*16d86563SAlexander Pyhalov 	ucs_state_t *cd = (ucs_state_t *)calloc(1, sizeof(ucs_state_t));
41*16d86563SAlexander Pyhalov 	if (cd == (ucs_state_t *)NULL) {
42*16d86563SAlexander Pyhalov 		errno = ENOMEM;
43*16d86563SAlexander Pyhalov 		return((void *)-1);
44*16d86563SAlexander Pyhalov 	}
45*16d86563SAlexander Pyhalov 
46*16d86563SAlexander Pyhalov 	return((void *)cd);
47*16d86563SAlexander Pyhalov }
48*16d86563SAlexander Pyhalov 
49*16d86563SAlexander Pyhalov 
50*16d86563SAlexander Pyhalov void
_icv_close(ucs_state_t * cd)51*16d86563SAlexander Pyhalov _icv_close(ucs_state_t *cd)
52*16d86563SAlexander Pyhalov {
53*16d86563SAlexander Pyhalov 	if (! cd)
54*16d86563SAlexander Pyhalov 		errno = EBADF;
55*16d86563SAlexander Pyhalov 	else
56*16d86563SAlexander Pyhalov 		free((void *)cd);
57*16d86563SAlexander Pyhalov }
58*16d86563SAlexander Pyhalov 
59*16d86563SAlexander Pyhalov 
60*16d86563SAlexander Pyhalov size_t
_icv_iconv(ucs_state_t * cd,char ** inbuf,size_t * inbufleft,char ** outbuf,size_t * outbufleft)61*16d86563SAlexander Pyhalov _icv_iconv(ucs_state_t *cd, char **inbuf, size_t *inbufleft, char **outbuf,
62*16d86563SAlexander Pyhalov                 size_t *outbufleft)
63*16d86563SAlexander Pyhalov {
64*16d86563SAlexander Pyhalov 	size_t ret_val = 0;
65*16d86563SAlexander Pyhalov 	unsigned char *ib;
66*16d86563SAlexander Pyhalov 	unsigned char *ob;
67*16d86563SAlexander Pyhalov 	unsigned char *ibtail;
68*16d86563SAlexander Pyhalov 	unsigned char *obtail;
69*16d86563SAlexander Pyhalov 	register int i, l, h;
70*16d86563SAlexander Pyhalov 	signed char sz;
71*16d86563SAlexander Pyhalov 	unsigned long u8;
72*16d86563SAlexander Pyhalov 
73*16d86563SAlexander Pyhalov 	if (! cd) {
74*16d86563SAlexander Pyhalov 		errno = EBADF;
75*16d86563SAlexander Pyhalov 		return((size_t)-1);
76*16d86563SAlexander Pyhalov 	}
77*16d86563SAlexander Pyhalov 
78*16d86563SAlexander Pyhalov 	if (!inbuf || !(*inbuf)) {
79*16d86563SAlexander Pyhalov 		cd->bom_written = false;
80*16d86563SAlexander Pyhalov 		return((size_t)0);
81*16d86563SAlexander Pyhalov 	}
82*16d86563SAlexander Pyhalov 
83*16d86563SAlexander Pyhalov 	ib = (unsigned char *)*inbuf;
84*16d86563SAlexander Pyhalov 	ob = (unsigned char *)*outbuf;
85*16d86563SAlexander Pyhalov 	ibtail = ib + *inbufleft;
86*16d86563SAlexander Pyhalov 	obtail = ob + *outbufleft;
87*16d86563SAlexander Pyhalov 
88*16d86563SAlexander Pyhalov 	/* We skip any first signiture of UTF-8 */
89*16d86563SAlexander Pyhalov 	if (!cd->bom_written &&
90*16d86563SAlexander Pyhalov 		((ibtail - ib) >= ICV_FETCH_UTF8_BOM_SIZE)) {
91*16d86563SAlexander Pyhalov 		for (u8 = 0, i = 0; i < ICV_FETCH_UTF8_BOM_SIZE; i++)
92*16d86563SAlexander Pyhalov 			u8 = (u8 << 8) | ((uint_t)(*(ib + i)));
93*16d86563SAlexander Pyhalov 		if (u8 == ICV_BOM_IN_BIG_ENDIAN)
94*16d86563SAlexander Pyhalov 			ib += ICV_FETCH_UTF8_BOM_SIZE;
95*16d86563SAlexander Pyhalov 	}
96*16d86563SAlexander Pyhalov 	cd->bom_written = true;
97*16d86563SAlexander Pyhalov 
98*16d86563SAlexander Pyhalov 	while (ib < ibtail) {
99*16d86563SAlexander Pyhalov 		sz = number_of_bytes_in_utf8_char[*ib];
100*16d86563SAlexander Pyhalov 		if (sz == ICV_TYPE_ILLEGAL_CHAR) {
101*16d86563SAlexander Pyhalov 			errno = EILSEQ;
102*16d86563SAlexander Pyhalov 			ret_val = (size_t)-1;
103*16d86563SAlexander Pyhalov 			break;
104*16d86563SAlexander Pyhalov 		}
105*16d86563SAlexander Pyhalov 
106*16d86563SAlexander Pyhalov 		if (ob >= obtail) {
107*16d86563SAlexander Pyhalov 			errno = E2BIG;
108*16d86563SAlexander Pyhalov 			ret_val = (size_t)-1;
109*16d86563SAlexander Pyhalov 			break;
110*16d86563SAlexander Pyhalov 		}
111*16d86563SAlexander Pyhalov 
112*16d86563SAlexander Pyhalov 		if (sz == 1) {
113*16d86563SAlexander Pyhalov 			*ob++ = *ib++;
114*16d86563SAlexander Pyhalov 		} else {
115*16d86563SAlexander Pyhalov 			if ((ibtail - ib) < sz) {
116*16d86563SAlexander Pyhalov 				errno = EINVAL;
117*16d86563SAlexander Pyhalov 				ret_val = (size_t)-1;
118*16d86563SAlexander Pyhalov 				break;
119*16d86563SAlexander Pyhalov 			}
120*16d86563SAlexander Pyhalov 
121*16d86563SAlexander Pyhalov 			u8 = 0;
122*16d86563SAlexander Pyhalov 			for (i = 0; i < sz; i++) {
123*16d86563SAlexander Pyhalov 				if (((unsigned int)*ib) < 0x80) {
124*16d86563SAlexander Pyhalov 					errno = EILSEQ;
125*16d86563SAlexander Pyhalov 					ret_val = (size_t)-1;
126*16d86563SAlexander Pyhalov 					goto illegal_char_err;
127*16d86563SAlexander Pyhalov 				}
128*16d86563SAlexander Pyhalov 				u8 = (u8 << 8) | ((unsigned int)*ib);
129*16d86563SAlexander Pyhalov 				ib++;
130*16d86563SAlexander Pyhalov 			}
131*16d86563SAlexander Pyhalov 			if ((u8 & ICV_UTF8_REPRESENTATION_ffff_mask) ==
132*16d86563SAlexander Pyhalov 			    ICV_UTF8_REPRESENTATION_fffe ||
133*16d86563SAlexander Pyhalov 			    (u8 & ICV_UTF8_REPRESENTATION_ffff_mask) ==
134*16d86563SAlexander Pyhalov 			    ICV_UTF8_REPRESENTATION_ffff ||
135*16d86563SAlexander Pyhalov 			    u8 > ICV_UTF8_REPRESENTATION_10fffd ||
136*16d86563SAlexander Pyhalov 			    (u8 >= ICV_UTF8_REPRESENTATION_d800 &&
137*16d86563SAlexander Pyhalov 			    u8 <= ICV_UTF8_REPRESENTATION_dfff) ||
138*16d86563SAlexander Pyhalov 			    (u8 >= ICV_UTF8_REPRESENTATION_fdd0 &&
139*16d86563SAlexander Pyhalov 			    u8 <= ICV_UTF8_REPRESENTATION_fdef)) {
140*16d86563SAlexander Pyhalov 				ib -= sz;
141*16d86563SAlexander Pyhalov 				errno = EILSEQ;
142*16d86563SAlexander Pyhalov 				ret_val = (size_t)-1;
143*16d86563SAlexander Pyhalov 				goto illegal_char_err;
144*16d86563SAlexander Pyhalov 			}
145*16d86563SAlexander Pyhalov 
146*16d86563SAlexander Pyhalov 			i = l = 0;
147*16d86563SAlexander Pyhalov 			h = (sizeof(u8_sb_tbl) /
148*16d86563SAlexander Pyhalov 			     sizeof(to_sb_table_component_t)) - 1;
149*16d86563SAlexander Pyhalov 			while (l <= h) {
150*16d86563SAlexander Pyhalov 				i = (l + h) / 2;
151*16d86563SAlexander Pyhalov 				if (u8_sb_tbl[i].u8 == u8)
152*16d86563SAlexander Pyhalov 					break;
153*16d86563SAlexander Pyhalov 				else if (u8_sb_tbl[i].u8 < u8)
154*16d86563SAlexander Pyhalov 					l = i + 1;
155*16d86563SAlexander Pyhalov 				else
156*16d86563SAlexander Pyhalov 					h = i - 1;
157*16d86563SAlexander Pyhalov 			}
158*16d86563SAlexander Pyhalov 
159*16d86563SAlexander Pyhalov 			/*
160*16d86563SAlexander Pyhalov 			 * We just assume that either we found it or it is
161*16d86563SAlexander Pyhalov 			 * a non-identical character that we need to
162*16d86563SAlexander Pyhalov 			 * provide a replacement character.
163*16d86563SAlexander Pyhalov 			 */
164*16d86563SAlexander Pyhalov 			if (u8_sb_tbl[i].u8 == u8) {
165*16d86563SAlexander Pyhalov 				*ob++ = u8_sb_tbl[i].sb;
166*16d86563SAlexander Pyhalov 			} else {
167*16d86563SAlexander Pyhalov 				*ob++ = ICV_CHAR_ASCII_REPLACEMENT;
168*16d86563SAlexander Pyhalov 				ret_val++;
169*16d86563SAlexander Pyhalov 			}
170*16d86563SAlexander Pyhalov 		}
171*16d86563SAlexander Pyhalov 	}
172*16d86563SAlexander Pyhalov 
173*16d86563SAlexander Pyhalov illegal_char_err:
174*16d86563SAlexander Pyhalov 	*inbuf = (char *)ib;
175*16d86563SAlexander Pyhalov 	*inbufleft = ibtail - ib;
176*16d86563SAlexander Pyhalov 	*outbuf = (char *)ob;
177*16d86563SAlexander Pyhalov 	*outbufleft = obtail - ob;
178*16d86563SAlexander Pyhalov 
179*16d86563SAlexander Pyhalov 	return(ret_val);
180*16d86563SAlexander Pyhalov }
181