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 various UCS formats,
26*16d86563SAlexander Pyhalov * especially, UCS-2, UCS-2BE, UCS-2LE, UTF-16, UTF-16BE, and, UTF-16LE to
27*16d86563SAlexander Pyhalov * another various UCS formats, UCS-4, UCS-4BE, UCS-4LE, UTF-32, UTF-32BE,
28*16d86563SAlexander Pyhalov * and, UTF-32LE.
29*16d86563SAlexander Pyhalov */
30*16d86563SAlexander Pyhalov
31*16d86563SAlexander Pyhalov
32*16d86563SAlexander Pyhalov #include <stdlib.h>
33*16d86563SAlexander Pyhalov #include <errno.h>
34*16d86563SAlexander Pyhalov #include <sys/types.h>
35*16d86563SAlexander Pyhalov #include <sys/isa_defs.h>
36*16d86563SAlexander Pyhalov #include "ucs_to_ucs4.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(UTF_16BE) || defined(UCS_2BE)
51*16d86563SAlexander Pyhalov cd->input.little_endian = false;
52*16d86563SAlexander Pyhalov cd->input.bom_written = true;
53*16d86563SAlexander Pyhalov #elif defined(UTF_16LE) || defined(UCS_2LE)
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(UCS_4BE) || defined(UTF_32BE)
61*16d86563SAlexander Pyhalov cd->output.little_endian = false;
62*16d86563SAlexander Pyhalov cd->output.bom_written = true;
63*16d86563SAlexander Pyhalov #elif defined(UCS_4LE) || 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 uint_t u4_2;
95*16d86563SAlexander Pyhalov register int i;
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(UTF_16)
104*16d86563SAlexander Pyhalov cd->input.bom_written = false;
105*16d86563SAlexander Pyhalov #endif
106*16d86563SAlexander Pyhalov #if defined(UCS_4) || defined(UTF_32)
107*16d86563SAlexander Pyhalov cd->output.bom_written = false;
108*16d86563SAlexander Pyhalov #endif
109*16d86563SAlexander Pyhalov return((size_t)0);
110*16d86563SAlexander Pyhalov }
111*16d86563SAlexander Pyhalov
112*16d86563SAlexander Pyhalov ib = (uchar_t *)*inbuf;
113*16d86563SAlexander Pyhalov ob = (uchar_t *)*outbuf;
114*16d86563SAlexander Pyhalov ibtail = ib + *inbufleft;
115*16d86563SAlexander Pyhalov obtail = ob + *outbufleft;
116*16d86563SAlexander Pyhalov
117*16d86563SAlexander Pyhalov #if defined(UCS_2) || defined(UTF_16)
118*16d86563SAlexander Pyhalov if (! cd->input.bom_written) {
119*16d86563SAlexander Pyhalov if ((ibtail - ib) < ICV_FETCH_UCS_SIZE) {
120*16d86563SAlexander Pyhalov errno = EINVAL;
121*16d86563SAlexander Pyhalov ret_val = (size_t)-1;
122*16d86563SAlexander Pyhalov goto need_more_input_err;
123*16d86563SAlexander Pyhalov }
124*16d86563SAlexander Pyhalov
125*16d86563SAlexander Pyhalov for (u4 = 0, i = 0; i < ICV_FETCH_UCS_SIZE; i++)
126*16d86563SAlexander Pyhalov u4 = (u4 << 8) | ((uint_t)(*(ib + i)));
127*16d86563SAlexander Pyhalov
128*16d86563SAlexander Pyhalov if (u4 == ICV_BOM_IN_BIG_ENDIAN) {
129*16d86563SAlexander Pyhalov ib += ICV_FETCH_UCS_SIZE;
130*16d86563SAlexander Pyhalov cd->input.little_endian = false;
131*16d86563SAlexander Pyhalov } else if (u4 == ICV_BOM_IN_LITTLE_ENDIAN) {
132*16d86563SAlexander Pyhalov ib += ICV_FETCH_UCS_SIZE;
133*16d86563SAlexander Pyhalov cd->input.little_endian = true;
134*16d86563SAlexander Pyhalov }
135*16d86563SAlexander Pyhalov }
136*16d86563SAlexander Pyhalov cd->input.bom_written = true;
137*16d86563SAlexander Pyhalov #endif
138*16d86563SAlexander Pyhalov
139*16d86563SAlexander Pyhalov while (ib < ibtail) {
140*16d86563SAlexander Pyhalov if ((ibtail - ib) < ICV_FETCH_UCS_SIZE) {
141*16d86563SAlexander Pyhalov errno = EINVAL;
142*16d86563SAlexander Pyhalov ret_val = (size_t)-1;
143*16d86563SAlexander Pyhalov break;
144*16d86563SAlexander Pyhalov }
145*16d86563SAlexander Pyhalov
146*16d86563SAlexander Pyhalov u4 = u4_2 = 0;
147*16d86563SAlexander Pyhalov if (cd->input.little_endian) {
148*16d86563SAlexander Pyhalov for (i = ICV_FETCH_UCS_SIZE - 1; i >= 0; i--)
149*16d86563SAlexander Pyhalov u4 = (u4 << 8) | ((uint_t)(*(ib + i)));
150*16d86563SAlexander Pyhalov } else {
151*16d86563SAlexander Pyhalov for (i = 0; i < ICV_FETCH_UCS_SIZE; i++)
152*16d86563SAlexander Pyhalov u4 = (u4 << 8) | ((uint_t)(*(ib + i)));
153*16d86563SAlexander Pyhalov }
154*16d86563SAlexander Pyhalov
155*16d86563SAlexander Pyhalov #if defined(UTF_16) || defined(UTF_16BE) || defined(UTF_16LE)
156*16d86563SAlexander Pyhalov if ((u4 >= 0x00dc00 && u4 <= 0x00dfff) || u4 >= 0x00fffe) {
157*16d86563SAlexander Pyhalov errno = EILSEQ;
158*16d86563SAlexander Pyhalov ret_val = (size_t)-1;
159*16d86563SAlexander Pyhalov break;
160*16d86563SAlexander Pyhalov }
161*16d86563SAlexander Pyhalov
162*16d86563SAlexander Pyhalov if (u4 >= 0x00d800 && u4 <= 0x00dbff) {
163*16d86563SAlexander Pyhalov if ((ibtail - ib) < ICV_FETCH_UCS_SIZE_TWO) {
164*16d86563SAlexander Pyhalov errno = EINVAL;
165*16d86563SAlexander Pyhalov ret_val = (size_t)-1;
166*16d86563SAlexander Pyhalov break;
167*16d86563SAlexander Pyhalov }
168*16d86563SAlexander Pyhalov
169*16d86563SAlexander Pyhalov if (cd->input.little_endian) {
170*16d86563SAlexander Pyhalov for (i = ICV_FETCH_UCS_SIZE_TWO - 1;
171*16d86563SAlexander Pyhalov i >= ICV_FETCH_UCS_SIZE;
172*16d86563SAlexander Pyhalov i--)
173*16d86563SAlexander Pyhalov u4_2 = (u4_2<<8)|((uint_t)(*(ib + i)));
174*16d86563SAlexander Pyhalov } else {
175*16d86563SAlexander Pyhalov for (i = ICV_FETCH_UCS_SIZE;
176*16d86563SAlexander Pyhalov i < ICV_FETCH_UCS_SIZE_TWO;
177*16d86563SAlexander Pyhalov i++)
178*16d86563SAlexander Pyhalov u4_2 = (u4_2<<8)|((uint_t)(*(ib + i)));
179*16d86563SAlexander Pyhalov }
180*16d86563SAlexander Pyhalov
181*16d86563SAlexander Pyhalov if (u4_2 < 0x00dc00 || u4_2 > 0x00dfff) {
182*16d86563SAlexander Pyhalov errno = EILSEQ;
183*16d86563SAlexander Pyhalov ret_val = (size_t)-1;
184*16d86563SAlexander Pyhalov break;
185*16d86563SAlexander Pyhalov }
186*16d86563SAlexander Pyhalov
187*16d86563SAlexander Pyhalov u4 = ((((u4 - 0x00d800) * 0x400) +
188*16d86563SAlexander Pyhalov (u4_2 - 0x00dc00)) & 0x0fffff) + 0x010000;
189*16d86563SAlexander Pyhalov }
190*16d86563SAlexander Pyhalov #elif defined(UCS_2) || defined(UCS_2BE) || defined(UCS_2LE)
191*16d86563SAlexander Pyhalov if (u4 >= 0x00fffe || (u4 >= 0x00d800 && u4 <= 0x00dfff)) {
192*16d86563SAlexander Pyhalov errno = EILSEQ;
193*16d86563SAlexander Pyhalov ret_val = (size_t)-1;
194*16d86563SAlexander Pyhalov break;
195*16d86563SAlexander Pyhalov }
196*16d86563SAlexander Pyhalov #elif defined(UCS_4) || defined(UCS_4BE) || defined(UCS_4LE) || \
197*16d86563SAlexander Pyhalov defined(UTF_32) || defined(UTF_32BE) || defined(UTF_32LE)
198*16d86563SAlexander Pyhalov /*
199*16d86563SAlexander Pyhalov * We do nothing here since these if expressions are
200*16d86563SAlexander Pyhalov * only for input characters particularly of
201*16d86563SAlexander Pyhalov * UCS-2, UCS-2BE, UCS-2LE, UTF-16, UTF-16BE, and
202*16d86563SAlexander Pyhalov * UTF-16LE.
203*16d86563SAlexander Pyhalov */
204*16d86563SAlexander Pyhalov #else
205*16d86563SAlexander Pyhalov #error "Fatal: one of the UCS macros need to be defined."
206*16d86563SAlexander Pyhalov #endif
207*16d86563SAlexander Pyhalov
208*16d86563SAlexander Pyhalov if ((obtail - ob) < ((cd->output.bom_written) ? 4 : 8)) {
209*16d86563SAlexander Pyhalov errno = E2BIG;
210*16d86563SAlexander Pyhalov ret_val = (size_t)-1;
211*16d86563SAlexander Pyhalov break;
212*16d86563SAlexander Pyhalov }
213*16d86563SAlexander Pyhalov
214*16d86563SAlexander Pyhalov if (cd->output.little_endian) {
215*16d86563SAlexander Pyhalov if (! cd->output.bom_written) {
216*16d86563SAlexander Pyhalov *ob++ = (uchar_t)0xff;
217*16d86563SAlexander Pyhalov *ob++ = (uchar_t)0xfe;
218*16d86563SAlexander Pyhalov *(ushort_t *)ob = (ushort_t)0;
219*16d86563SAlexander Pyhalov ob += 2;
220*16d86563SAlexander Pyhalov cd->output.bom_written = true;
221*16d86563SAlexander Pyhalov }
222*16d86563SAlexander Pyhalov *ob++ = (uchar_t)(u4 & 0xff);
223*16d86563SAlexander Pyhalov *ob++ = (uchar_t)((u4 >> 8) & 0xff);
224*16d86563SAlexander Pyhalov *ob++ = (uchar_t)((u4 >> 16) & 0xff);
225*16d86563SAlexander Pyhalov *ob++ = (uchar_t)((u4 >> 24) & 0xff);
226*16d86563SAlexander Pyhalov } else {
227*16d86563SAlexander Pyhalov if (! cd->output.bom_written) {
228*16d86563SAlexander Pyhalov *(ushort_t *)ob = (ushort_t)0;
229*16d86563SAlexander Pyhalov ob += 2;
230*16d86563SAlexander Pyhalov *ob++ = (uchar_t)0xfe;
231*16d86563SAlexander Pyhalov *ob++ = (uchar_t)0xff;
232*16d86563SAlexander Pyhalov cd->output.bom_written = true;
233*16d86563SAlexander Pyhalov }
234*16d86563SAlexander Pyhalov *ob++ = (uchar_t)((u4 >> 24) & 0xff);
235*16d86563SAlexander Pyhalov *ob++ = (uchar_t)((u4 >> 16) & 0xff);
236*16d86563SAlexander Pyhalov *ob++ = (uchar_t)((u4 >> 8) & 0xff);
237*16d86563SAlexander Pyhalov *ob++ = (uchar_t)(u4 & 0xff);
238*16d86563SAlexander Pyhalov }
239*16d86563SAlexander Pyhalov ib += ((u4_2) ? ICV_FETCH_UCS_SIZE_TWO : ICV_FETCH_UCS_SIZE);
240*16d86563SAlexander Pyhalov }
241*16d86563SAlexander Pyhalov
242*16d86563SAlexander Pyhalov #if defined(UCS_2) || defined(UTF_16)
243*16d86563SAlexander Pyhalov need_more_input_err:
244*16d86563SAlexander Pyhalov #endif
245*16d86563SAlexander Pyhalov *inbuf = (char *)ib;
246*16d86563SAlexander Pyhalov *inbufleft = ibtail - ib;
247*16d86563SAlexander Pyhalov *outbuf = (char *)ob;
248*16d86563SAlexander Pyhalov *outbufleft = obtail - ob;
249*16d86563SAlexander Pyhalov
250*16d86563SAlexander Pyhalov return(ret_val);
251*16d86563SAlexander Pyhalov }
252