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 * In this program, we assume that each table entry provided will contain
26*16d86563SAlexander Pyhalov * a valid UCS character, an illegal character, or, a replacement character.
27*16d86563SAlexander Pyhalov * In other words, it is table provider's responsibility to provide
28*16d86563SAlexander Pyhalov * an appropriate mapping for each single byte character in the table since
29*16d86563SAlexander Pyhalov * the program in this file will not do any special checking on the table
30*16d86563SAlexander Pyhalov * component values.
31*16d86563SAlexander Pyhalov *
32*16d86563SAlexander Pyhalov * This particular file is to cover conversions from various single byte
33*16d86563SAlexander Pyhalov * codesets to UCS-2, UCS-2BE, UCS-2LE, UCS-4, UCS-4BE, UCS-4LE, UTF-16,
34*16d86563SAlexander Pyhalov * UTF-16BE, UTF-16LE, UTF-32, UTF-32BE, and UTF-32LE.
35*16d86563SAlexander Pyhalov */
36*16d86563SAlexander Pyhalov
37*16d86563SAlexander Pyhalov
38*16d86563SAlexander Pyhalov #include <stdlib.h>
39*16d86563SAlexander Pyhalov #include <errno.h>
40*16d86563SAlexander Pyhalov #include <sys/types.h>
41*16d86563SAlexander Pyhalov #include <sys/isa_defs.h>
42*16d86563SAlexander Pyhalov #include "sb_to_ucs.h"
43*16d86563SAlexander Pyhalov
44*16d86563SAlexander Pyhalov
45*16d86563SAlexander Pyhalov void *
_icv_open()46*16d86563SAlexander Pyhalov _icv_open()
47*16d86563SAlexander Pyhalov {
48*16d86563SAlexander Pyhalov ucs_state_t *cd = (ucs_state_t *)calloc(1, sizeof(ucs_state_t));
49*16d86563SAlexander Pyhalov
50*16d86563SAlexander Pyhalov if (cd == (ucs_state_t *)NULL) {
51*16d86563SAlexander Pyhalov errno = ENOMEM;
52*16d86563SAlexander Pyhalov return((void *)-1);
53*16d86563SAlexander Pyhalov }
54*16d86563SAlexander Pyhalov
55*16d86563SAlexander Pyhalov #if defined(UTF_16BE) || defined(UCS_2BE) || defined(UCS_4BE) || \
56*16d86563SAlexander Pyhalov defined(UTF_32BE)
57*16d86563SAlexander Pyhalov cd->little_endian = false;
58*16d86563SAlexander Pyhalov cd->bom_written = true;
59*16d86563SAlexander Pyhalov #elif defined(UTF_16LE) || defined(UCS_2LE) || defined(UCS_4LE) || \
60*16d86563SAlexander Pyhalov defined(UTF_32LE)
61*16d86563SAlexander Pyhalov cd->little_endian = true;
62*16d86563SAlexander Pyhalov cd->bom_written = true;
63*16d86563SAlexander Pyhalov #elif defined(_LITTLE_ENDIAN)
64*16d86563SAlexander Pyhalov cd->little_endian = true;
65*16d86563SAlexander Pyhalov #endif
66*16d86563SAlexander Pyhalov
67*16d86563SAlexander Pyhalov return((void *)cd);
68*16d86563SAlexander Pyhalov }
69*16d86563SAlexander Pyhalov
70*16d86563SAlexander Pyhalov
71*16d86563SAlexander Pyhalov void
_icv_close(ucs_state_t * cd)72*16d86563SAlexander Pyhalov _icv_close(ucs_state_t *cd)
73*16d86563SAlexander Pyhalov {
74*16d86563SAlexander Pyhalov if (! cd)
75*16d86563SAlexander Pyhalov errno = EBADF;
76*16d86563SAlexander Pyhalov else
77*16d86563SAlexander Pyhalov free((void *)cd);
78*16d86563SAlexander Pyhalov }
79*16d86563SAlexander Pyhalov
80*16d86563SAlexander Pyhalov
81*16d86563SAlexander Pyhalov size_t
_icv_iconv(ucs_state_t * cd,char ** inbuf,size_t * inbufleft,char ** outbuf,size_t * outbufleft)82*16d86563SAlexander Pyhalov _icv_iconv(ucs_state_t *cd, char **inbuf, size_t *inbufleft, char **outbuf,
83*16d86563SAlexander Pyhalov size_t *outbufleft)
84*16d86563SAlexander Pyhalov {
85*16d86563SAlexander Pyhalov size_t ret_val = 0;
86*16d86563SAlexander Pyhalov unsigned char *ib;
87*16d86563SAlexander Pyhalov unsigned char *ob;
88*16d86563SAlexander Pyhalov unsigned char *ibtail;
89*16d86563SAlexander Pyhalov unsigned char *obtail;
90*16d86563SAlexander Pyhalov unsigned int u4;
91*16d86563SAlexander Pyhalov #if defined(UTF_16) || defined(UTF_16BE) || defined(UTF_16LE)
92*16d86563SAlexander Pyhalov unsigned int u4_2;
93*16d86563SAlexander Pyhalov #endif
94*16d86563SAlexander Pyhalov signed char obsz;
95*16d86563SAlexander Pyhalov
96*16d86563SAlexander Pyhalov
97*16d86563SAlexander Pyhalov if (! cd) {
98*16d86563SAlexander Pyhalov errno = EBADF;
99*16d86563SAlexander Pyhalov return((size_t)-1);
100*16d86563SAlexander Pyhalov }
101*16d86563SAlexander Pyhalov
102*16d86563SAlexander Pyhalov if (!inbuf || !(*inbuf)) {
103*16d86563SAlexander Pyhalov #if defined(UCS_2) || defined(UCS_4) || defined(UTF_16) || defined(UTF_32)
104*16d86563SAlexander Pyhalov cd->bom_written = false;
105*16d86563SAlexander Pyhalov #endif
106*16d86563SAlexander Pyhalov return((size_t)0);
107*16d86563SAlexander Pyhalov }
108*16d86563SAlexander Pyhalov
109*16d86563SAlexander Pyhalov ib = (unsigned char *)*inbuf;
110*16d86563SAlexander Pyhalov ob = (unsigned char *)*outbuf;
111*16d86563SAlexander Pyhalov ibtail = ib + *inbufleft;
112*16d86563SAlexander Pyhalov obtail = ob + *outbufleft;
113*16d86563SAlexander Pyhalov
114*16d86563SAlexander Pyhalov while (ib < ibtail) {
115*16d86563SAlexander Pyhalov u4 = sb_u4_tbl[*ib].u8;
116*16d86563SAlexander Pyhalov #if defined(UTF_16) || defined(UTF_16BE) || defined(UTF_16LE)
117*16d86563SAlexander Pyhalov u4_2 = 0;
118*16d86563SAlexander Pyhalov #endif
119*16d86563SAlexander Pyhalov
120*16d86563SAlexander Pyhalov if (sb_u4_tbl[*ib].size == ICV_TYPE_ILLEGAL_CHAR) {
121*16d86563SAlexander Pyhalov errno = EILSEQ;
122*16d86563SAlexander Pyhalov ret_val = (size_t)-1;
123*16d86563SAlexander Pyhalov break;
124*16d86563SAlexander Pyhalov }
125*16d86563SAlexander Pyhalov
126*16d86563SAlexander Pyhalov obsz = (cd->bom_written) ? ICV_FETCH_UCS_SIZE :
127*16d86563SAlexander Pyhalov ICV_FETCH_UCS_SIZE_TWO;
128*16d86563SAlexander Pyhalov #if defined(UCS_2) || defined(UCS_2BE) || defined(UCS_2LE)
129*16d86563SAlexander Pyhalov if (u4 > 0x00ffff) {
130*16d86563SAlexander Pyhalov u4 = ICV_CHAR_UCS2_REPLACEMENT;
131*16d86563SAlexander Pyhalov ret_val++;
132*16d86563SAlexander Pyhalov }
133*16d86563SAlexander Pyhalov #elif defined(UTF_16) || defined(UTF_16BE) || defined(UTF_16LE)
134*16d86563SAlexander Pyhalov if (u4 > 0x00ffff && u4 < 0x110000) {
135*16d86563SAlexander Pyhalov u4_2 = ((u4 - 0x010000) % 0x400) + 0x00dc00;
136*16d86563SAlexander Pyhalov u4 = ((u4 - 0x010000) / 0x400) + 0x00d800;
137*16d86563SAlexander Pyhalov obsz += 2;
138*16d86563SAlexander Pyhalov } else if (u4 > 0x10ffff) {
139*16d86563SAlexander Pyhalov u4 = ICV_CHAR_UCS2_REPLACEMENT;
140*16d86563SAlexander Pyhalov ret_val++;
141*16d86563SAlexander Pyhalov }
142*16d86563SAlexander Pyhalov #elif defined(UTF_32) || defined(UTF_32BE) || defined(UTF_32LE)
143*16d86563SAlexander Pyhalov if (u4 > 0x10ffff) {
144*16d86563SAlexander Pyhalov u4 = ICV_CHAR_UCS2_REPLACEMENT;
145*16d86563SAlexander Pyhalov ret_val++;
146*16d86563SAlexander Pyhalov }
147*16d86563SAlexander Pyhalov #elif defined(UCS_4) || defined(UCS_4BE) || defined(UCS_4LE)
148*16d86563SAlexander Pyhalov /* do nothing */
149*16d86563SAlexander Pyhalov #else
150*16d86563SAlexander Pyhalov #error "Fatal: one of the UCS macros need to be defined."
151*16d86563SAlexander Pyhalov #endif
152*16d86563SAlexander Pyhalov
153*16d86563SAlexander Pyhalov /*
154*16d86563SAlexander Pyhalov * The target values in the conversion tables are in UCS-4
155*16d86563SAlexander Pyhalov * without BOM and so the max target value possible would be
156*16d86563SAlexander Pyhalov * U+7FFFFFFF.
157*16d86563SAlexander Pyhalov */
158*16d86563SAlexander Pyhalov if (u4 == 0x00fffe || u4 == 0x00ffff || u4 > 0x7fffffff ||
159*16d86563SAlexander Pyhalov (u4 >= 0x00d800 && u4 <= 0x00dfff)) {
160*16d86563SAlexander Pyhalov /*
161*16d86563SAlexander Pyhalov * if conversion table is right, this should not
162*16d86563SAlexander Pyhalov * happen.
163*16d86563SAlexander Pyhalov */
164*16d86563SAlexander Pyhalov errno = EILSEQ;
165*16d86563SAlexander Pyhalov ret_val = (size_t)-1;
166*16d86563SAlexander Pyhalov break;
167*16d86563SAlexander Pyhalov }
168*16d86563SAlexander Pyhalov
169*16d86563SAlexander Pyhalov if ((obtail - ob) < obsz) {
170*16d86563SAlexander Pyhalov errno = E2BIG;
171*16d86563SAlexander Pyhalov ret_val = (size_t)-1;
172*16d86563SAlexander Pyhalov break;
173*16d86563SAlexander Pyhalov }
174*16d86563SAlexander Pyhalov
175*16d86563SAlexander Pyhalov if (cd->little_endian) {
176*16d86563SAlexander Pyhalov if (! cd->bom_written) {
177*16d86563SAlexander Pyhalov *ob++ = (uchar_t)0xff;
178*16d86563SAlexander Pyhalov *ob++ = (uchar_t)0xfe;
179*16d86563SAlexander Pyhalov #if defined(UCS_4) || defined(UCS_4BE) || defined(UCS_4LE) || \
180*16d86563SAlexander Pyhalov defined(UTF_32) || defined(UTF_32BE) || defined(UTF_32LE)
181*16d86563SAlexander Pyhalov *(ushort_t *)ob = (ushort_t)0;
182*16d86563SAlexander Pyhalov ob += 2;
183*16d86563SAlexander Pyhalov #endif
184*16d86563SAlexander Pyhalov cd->bom_written = true;
185*16d86563SAlexander Pyhalov }
186*16d86563SAlexander Pyhalov *ob++ = (uchar_t)(u4 & 0xff);
187*16d86563SAlexander Pyhalov *ob++ = (uchar_t)((u4 >> 8) & 0xff);
188*16d86563SAlexander Pyhalov #if defined(UCS_4) || defined(UCS_4BE) || defined(UCS_4LE) || \
189*16d86563SAlexander Pyhalov defined(UTF_32) || defined(UTF_32BE) || defined(UTF_32LE)
190*16d86563SAlexander Pyhalov *ob++ = (uchar_t)((u4 >> 16) & 0xff);
191*16d86563SAlexander Pyhalov *ob++ = (uchar_t)((u4 >> 24) & 0xff);
192*16d86563SAlexander Pyhalov #elif defined(UTF_16) || defined(UTF_16BE) || defined(UTF_16LE)
193*16d86563SAlexander Pyhalov if (u4_2) {
194*16d86563SAlexander Pyhalov *ob++ = (uchar_t)(u4_2 & 0xff);
195*16d86563SAlexander Pyhalov *ob++ = (uchar_t)((u4_2 >> 8) & 0xff);
196*16d86563SAlexander Pyhalov }
197*16d86563SAlexander Pyhalov #endif
198*16d86563SAlexander Pyhalov } else {
199*16d86563SAlexander Pyhalov if (! cd->bom_written) {
200*16d86563SAlexander Pyhalov #if defined(UCS_4) || defined(UCS_4BE) || defined(UCS_4LE) || \
201*16d86563SAlexander Pyhalov defined(UTF_32) || defined(UTF_32BE) || defined(UTF_32LE)
202*16d86563SAlexander Pyhalov *(ushort_t *)ob = (ushort_t)0;
203*16d86563SAlexander Pyhalov ob += 2;
204*16d86563SAlexander Pyhalov #endif
205*16d86563SAlexander Pyhalov *ob++ = (uchar_t)0xfe;
206*16d86563SAlexander Pyhalov *ob++ = (uchar_t)0xff;
207*16d86563SAlexander Pyhalov cd->bom_written = true;
208*16d86563SAlexander Pyhalov }
209*16d86563SAlexander Pyhalov #if defined(UCS_4) || defined(UCS_4BE) || defined(UCS_4LE) || \
210*16d86563SAlexander Pyhalov defined(UTF_32) || defined(UTF_32BE) || defined(UTF_32LE)
211*16d86563SAlexander Pyhalov *ob++ = (uchar_t)((u4 >> 24) & 0xff);
212*16d86563SAlexander Pyhalov *ob++ = (uchar_t)((u4 >> 16) & 0xff);
213*16d86563SAlexander Pyhalov #endif
214*16d86563SAlexander Pyhalov *ob++ = (uchar_t)((u4 >> 8) & 0xff);
215*16d86563SAlexander Pyhalov *ob++ = (uchar_t)(u4 & 0xff);
216*16d86563SAlexander Pyhalov #if defined(UTF_16) || defined(UTF_16BE) || defined(UTF_16LE)
217*16d86563SAlexander Pyhalov if (u4_2) {
218*16d86563SAlexander Pyhalov *ob++ = (uchar_t)((u4_2 >> 8) & 0xff);
219*16d86563SAlexander Pyhalov *ob++ = (uchar_t)(u4_2 & 0xff);
220*16d86563SAlexander Pyhalov }
221*16d86563SAlexander Pyhalov #endif
222*16d86563SAlexander Pyhalov }
223*16d86563SAlexander Pyhalov ib++;
224*16d86563SAlexander Pyhalov }
225*16d86563SAlexander Pyhalov
226*16d86563SAlexander Pyhalov *inbuf = (char *)ib;
227*16d86563SAlexander Pyhalov *inbufleft = ibtail - ib;
228*16d86563SAlexander Pyhalov *outbuf = (char *)ob;
229*16d86563SAlexander Pyhalov *outbufleft = obtail - ob;
230*16d86563SAlexander Pyhalov
231*16d86563SAlexander Pyhalov return(ret_val);
232*16d86563SAlexander Pyhalov }
233