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 (c) 1994 by Sun Microsystems, Inc.
23*16d86563SAlexander Pyhalov */
24*16d86563SAlexander Pyhalov
25*16d86563SAlexander Pyhalov
26*16d86563SAlexander Pyhalov #include <stdlib.h>
27*16d86563SAlexander Pyhalov #include <errno.h>
28*16d86563SAlexander Pyhalov #include "ktable.h"
29*16d86563SAlexander Pyhalov #include "utf_iso2022.h"
30*16d86563SAlexander Pyhalov
31*16d86563SAlexander Pyhalov
32*16d86563SAlexander Pyhalov /**** _ I C V _ O P E N ****/
33*16d86563SAlexander Pyhalov
_icv_open()34*16d86563SAlexander Pyhalov void* _icv_open()
35*16d86563SAlexander Pyhalov {
36*16d86563SAlexander Pyhalov _conv_desc* cd = (_conv_desc*)malloc(sizeof(_conv_desc));
37*16d86563SAlexander Pyhalov
38*16d86563SAlexander Pyhalov if (cd == (_conv_desc*)NULL)
39*16d86563SAlexander Pyhalov {
40*16d86563SAlexander Pyhalov errno = ENOMEM;
41*16d86563SAlexander Pyhalov return((void*)-1);
42*16d86563SAlexander Pyhalov }
43*16d86563SAlexander Pyhalov
44*16d86563SAlexander Pyhalov RESET_CONV_DESC();
45*16d86563SAlexander Pyhalov cd->designated = 0;
46*16d86563SAlexander Pyhalov cd->state = ASCII;
47*16d86563SAlexander Pyhalov
48*16d86563SAlexander Pyhalov return((void*)cd);
49*16d86563SAlexander Pyhalov } /* end of int _icv_open(). */
50*16d86563SAlexander Pyhalov
51*16d86563SAlexander Pyhalov
52*16d86563SAlexander Pyhalov /**** _ I C V _ C L O S E ****/
53*16d86563SAlexander Pyhalov
_icv_close(_conv_desc * cd)54*16d86563SAlexander Pyhalov void _icv_close(_conv_desc* cd)
55*16d86563SAlexander Pyhalov {
56*16d86563SAlexander Pyhalov if (!cd)
57*16d86563SAlexander Pyhalov errno = EBADF;
58*16d86563SAlexander Pyhalov else
59*16d86563SAlexander Pyhalov free((void*)cd);
60*16d86563SAlexander Pyhalov } /* end of void _icv_close(_conv_desc*). */
61*16d86563SAlexander Pyhalov
62*16d86563SAlexander Pyhalov
63*16d86563SAlexander Pyhalov /**** _ I C V _ I C O N V ****/
64*16d86563SAlexander Pyhalov
_icv_iconv(_conv_desc * cd,char ** inbuf,size_t * inbufleft,char ** outbuf,size_t * outbufleft)65*16d86563SAlexander Pyhalov size_t _icv_iconv(_conv_desc* cd, char** inbuf, size_t* inbufleft,
66*16d86563SAlexander Pyhalov char** outbuf, size_t* outbufleft)
67*16d86563SAlexander Pyhalov {
68*16d86563SAlexander Pyhalov extern unsigned short _johap_to_wansung(unsigned short, unsigned short,
69*16d86563SAlexander Pyhalov unsigned short);
70*16d86563SAlexander Pyhalov size_t ret_val = 0;
71*16d86563SAlexander Pyhalov unsigned char* ib;
72*16d86563SAlexander Pyhalov unsigned char* ob;
73*16d86563SAlexander Pyhalov unsigned char* ibtail;
74*16d86563SAlexander Pyhalov unsigned char* obtail;
75*16d86563SAlexander Pyhalov
76*16d86563SAlexander Pyhalov if (!cd)
77*16d86563SAlexander Pyhalov {
78*16d86563SAlexander Pyhalov errno = EBADF;
79*16d86563SAlexander Pyhalov return((size_t)-1);
80*16d86563SAlexander Pyhalov }
81*16d86563SAlexander Pyhalov
82*16d86563SAlexander Pyhalov /* To keep up with libc.so's iconv(), inbufleft checkings are added. */
83*16d86563SAlexander Pyhalov if (!inbuf || !(*inbuf) || !inbufleft || !(*inbufleft))
84*16d86563SAlexander Pyhalov {
85*16d86563SAlexander Pyhalov if (cd->state == WANSUNG && outbufleft && *outbufleft >= 1 &&
86*16d86563SAlexander Pyhalov outbuf && *outbuf)
87*16d86563SAlexander Pyhalov {
88*16d86563SAlexander Pyhalov **outbuf = SI;
89*16d86563SAlexander Pyhalov (*outbuf)++;
90*16d86563SAlexander Pyhalov (*outbufleft)--;
91*16d86563SAlexander Pyhalov }
92*16d86563SAlexander Pyhalov else
93*16d86563SAlexander Pyhalov {
94*16d86563SAlexander Pyhalov errno = E2BIG;
95*16d86563SAlexander Pyhalov return((size_t)-1);
96*16d86563SAlexander Pyhalov }
97*16d86563SAlexander Pyhalov
98*16d86563SAlexander Pyhalov RESET_CONV_DESC();
99*16d86563SAlexander Pyhalov cd->designated = 0;
100*16d86563SAlexander Pyhalov cd->state = ASCII;
101*16d86563SAlexander Pyhalov return((size_t)0);
102*16d86563SAlexander Pyhalov }
103*16d86563SAlexander Pyhalov
104*16d86563SAlexander Pyhalov ib = (unsigned char*)*inbuf;
105*16d86563SAlexander Pyhalov ob = (unsigned char*)*outbuf;
106*16d86563SAlexander Pyhalov ibtail = ib + *inbufleft;
107*16d86563SAlexander Pyhalov obtail = ob + *outbufleft;
108*16d86563SAlexander Pyhalov
109*16d86563SAlexander Pyhalov while (ib < ibtail)
110*16d86563SAlexander Pyhalov {
111*16d86563SAlexander Pyhalov if (!(*ib & 0x80)) /* 7 bits */
112*16d86563SAlexander Pyhalov {
113*16d86563SAlexander Pyhalov PROCESS_PRIOR_CVC();
114*16d86563SAlexander Pyhalov
115*16d86563SAlexander Pyhalov if ((obtail - ob) < (cd->state == WANSUNG ? 2 : 1))
116*16d86563SAlexander Pyhalov {
117*16d86563SAlexander Pyhalov errno = E2BIG;
118*16d86563SAlexander Pyhalov ret_val = (size_t)-1;
119*16d86563SAlexander Pyhalov break;
120*16d86563SAlexander Pyhalov }
121*16d86563SAlexander Pyhalov if (cd->state == WANSUNG)
122*16d86563SAlexander Pyhalov {
123*16d86563SAlexander Pyhalov *ob++ = SI;
124*16d86563SAlexander Pyhalov cd->state = ASCII;
125*16d86563SAlexander Pyhalov }
126*16d86563SAlexander Pyhalov *ob++ = *ib++;
127*16d86563SAlexander Pyhalov }
128*16d86563SAlexander Pyhalov else if ((*ib & 0xF0) == 0xE0) /* 16 bits */
129*16d86563SAlexander Pyhalov {
130*16d86563SAlexander Pyhalov unsigned long u;
131*16d86563SAlexander Pyhalov
132*16d86563SAlexander Pyhalov if ((ibtail - ib) < 3)
133*16d86563SAlexander Pyhalov {
134*16d86563SAlexander Pyhalov errno = EINVAL;
135*16d86563SAlexander Pyhalov ret_val = (size_t)-1;
136*16d86563SAlexander Pyhalov break;
137*16d86563SAlexander Pyhalov }
138*16d86563SAlexander Pyhalov
139*16d86563SAlexander Pyhalov u = ((unsigned long)(*ib) << 16) |
140*16d86563SAlexander Pyhalov ((unsigned long)(*(ib + 1)) << 8) |
141*16d86563SAlexander Pyhalov (unsigned long)(*(ib + 2));
142*16d86563SAlexander Pyhalov if (u == 0xE1859F ||
143*16d86563SAlexander Pyhalov (u >= 0xE18480 && u <= 0xE18492)) /* Ci */
144*16d86563SAlexander Pyhalov {
145*16d86563SAlexander Pyhalov PROCESS_PRIOR_CVC();
146*16d86563SAlexander Pyhalov
147*16d86563SAlexander Pyhalov cd->ci = (u == 0xE1859F) ? CVC_FILL
148*16d86563SAlexander Pyhalov : u - 0xE18480;
149*16d86563SAlexander Pyhalov cd->prev_state = CI;
150*16d86563SAlexander Pyhalov }
151*16d86563SAlexander Pyhalov else if (u == 0xE185A0 ||
152*16d86563SAlexander Pyhalov (u >= 0xE185A1 && u <= 0xE185B5)) /* V */
153*16d86563SAlexander Pyhalov {
154*16d86563SAlexander Pyhalov if (cd->prev_state != E && cd->prev_state != CI)
155*16d86563SAlexander Pyhalov PROCESS_PRIOR_CVC();
156*16d86563SAlexander Pyhalov
157*16d86563SAlexander Pyhalov cd->v = (u == 0xE185A0) ? CVC_FILL
158*16d86563SAlexander Pyhalov : u - 0xE185A1;
159*16d86563SAlexander Pyhalov cd->prev_state = V;
160*16d86563SAlexander Pyhalov }
161*16d86563SAlexander Pyhalov else if ((u >= 0xE186A8 && u <= 0xE186BF) ||
162*16d86563SAlexander Pyhalov (u >= 0xE18780 && u <= 0xE18782)) /* Cf */
163*16d86563SAlexander Pyhalov {
164*16d86563SAlexander Pyhalov if (cd->prev_state != E && cd->prev_state != V)
165*16d86563SAlexander Pyhalov PROCESS_PRIOR_CVC();
166*16d86563SAlexander Pyhalov
167*16d86563SAlexander Pyhalov cd->cf = u - ((u >= 0xE18780) ? 0xE18766
168*16d86563SAlexander Pyhalov : 0xE186A6);
169*16d86563SAlexander Pyhalov cd->prev_state = CF;
170*16d86563SAlexander Pyhalov
171*16d86563SAlexander Pyhalov PROCESS_PRIOR_CVC();
172*16d86563SAlexander Pyhalov }
173*16d86563SAlexander Pyhalov else if (u >= 0xE082A1 && u <= 0xEFBFA6)
174*16d86563SAlexander Pyhalov {
175*16d86563SAlexander Pyhalov register int h, i, j, l;
176*16d86563SAlexander Pyhalov short num;
177*16d86563SAlexander Pyhalov
178*16d86563SAlexander Pyhalov PROCESS_PRIOR_CVC();
179*16d86563SAlexander Pyhalov
180*16d86563SAlexander Pyhalov num = (cd->designated) ?
181*16d86563SAlexander Pyhalov (cd->state == ASCII ? 3 : 2) :
182*16d86563SAlexander Pyhalov (cd->state == ASCII ? 7 : 6);
183*16d86563SAlexander Pyhalov if ((obtail - ob) < num)
184*16d86563SAlexander Pyhalov {
185*16d86563SAlexander Pyhalov errno = E2BIG;
186*16d86563SAlexander Pyhalov ret_val = (size_t)-1;
187*16d86563SAlexander Pyhalov break;
188*16d86563SAlexander Pyhalov }
189*16d86563SAlexander Pyhalov
190*16d86563SAlexander Pyhalov for (l = 0, h = MAX_U2ISO2022_NUM; l < h; )
191*16d86563SAlexander Pyhalov {
192*16d86563SAlexander Pyhalov i = (l + h) / 2;
193*16d86563SAlexander Pyhalov if (utf82iso2022_tbl[i].utf8 == u)
194*16d86563SAlexander Pyhalov break;
195*16d86563SAlexander Pyhalov else if (utf82iso2022_tbl[l].utf8 == u)
196*16d86563SAlexander Pyhalov {
197*16d86563SAlexander Pyhalov i = l;
198*16d86563SAlexander Pyhalov break;
199*16d86563SAlexander Pyhalov }
200*16d86563SAlexander Pyhalov else if (utf82iso2022_tbl[h].utf8 == u)
201*16d86563SAlexander Pyhalov {
202*16d86563SAlexander Pyhalov i = h;
203*16d86563SAlexander Pyhalov break;
204*16d86563SAlexander Pyhalov }
205*16d86563SAlexander Pyhalov else if (utf82iso2022_tbl[i].utf8 < u)
206*16d86563SAlexander Pyhalov l = i + 1;
207*16d86563SAlexander Pyhalov else
208*16d86563SAlexander Pyhalov h = i - 1;
209*16d86563SAlexander Pyhalov }
210*16d86563SAlexander Pyhalov
211*16d86563SAlexander Pyhalov if (utf82iso2022_tbl[i].utf8 != u)
212*16d86563SAlexander Pyhalov {
213*16d86563SAlexander Pyhalov /* Let's assume the code is non-
214*16d86563SAlexander Pyhalov * identical. */
215*16d86563SAlexander Pyhalov if (cd->designated &&
216*16d86563SAlexander Pyhalov cd->state == WANSUNG)
217*16d86563SAlexander Pyhalov {
218*16d86563SAlexander Pyhalov if (ob >= obtail)
219*16d86563SAlexander Pyhalov {
220*16d86563SAlexander Pyhalov errno = E2BIG;
221*16d86563SAlexander Pyhalov ret_val = (size_t)-1;
222*16d86563SAlexander Pyhalov break;
223*16d86563SAlexander Pyhalov }
224*16d86563SAlexander Pyhalov *ob++ = SI;
225*16d86563SAlexander Pyhalov cd->state = ASCII;
226*16d86563SAlexander Pyhalov }
227*16d86563SAlexander Pyhalov *ob++ = NON_IDENTICAL;
228*16d86563SAlexander Pyhalov *ob++ = NON_IDENTICAL;
229*16d86563SAlexander Pyhalov ret_val += 2;
230*16d86563SAlexander Pyhalov }
231*16d86563SAlexander Pyhalov else
232*16d86563SAlexander Pyhalov {
233*16d86563SAlexander Pyhalov if (!cd->designated)
234*16d86563SAlexander Pyhalov {
235*16d86563SAlexander Pyhalov *ob++ = ESC;
236*16d86563SAlexander Pyhalov *ob++ = '$';
237*16d86563SAlexander Pyhalov *ob++ = ')';
238*16d86563SAlexander Pyhalov *ob++ = 'C';
239*16d86563SAlexander Pyhalov cd->designated = 1;
240*16d86563SAlexander Pyhalov }
241*16d86563SAlexander Pyhalov if (cd->state == ASCII)
242*16d86563SAlexander Pyhalov {
243*16d86563SAlexander Pyhalov *ob++ = SO;
244*16d86563SAlexander Pyhalov cd->state = WANSUNG;
245*16d86563SAlexander Pyhalov }
246*16d86563SAlexander Pyhalov *ob++ = (char)
247*16d86563SAlexander Pyhalov ((utf82iso2022_tbl[i].code >> 8)
248*16d86563SAlexander Pyhalov & 0x7F);
249*16d86563SAlexander Pyhalov *ob++ = (char)(utf82iso2022_tbl[i].code
250*16d86563SAlexander Pyhalov & 0x7F);
251*16d86563SAlexander Pyhalov }
252*16d86563SAlexander Pyhalov }
253*16d86563SAlexander Pyhalov else
254*16d86563SAlexander Pyhalov {
255*16d86563SAlexander Pyhalov PROCESS_PRIOR_CVC();
256*16d86563SAlexander Pyhalov
257*16d86563SAlexander Pyhalov /* Let's assume the code is non-identical. */
258*16d86563SAlexander Pyhalov if ((obtail - ob) < 2)
259*16d86563SAlexander Pyhalov {
260*16d86563SAlexander Pyhalov errno = E2BIG;
261*16d86563SAlexander Pyhalov ret_val = (size_t)-1;
262*16d86563SAlexander Pyhalov break;
263*16d86563SAlexander Pyhalov }
264*16d86563SAlexander Pyhalov *ob++ = NON_IDENTICAL;
265*16d86563SAlexander Pyhalov *ob++ = NON_IDENTICAL;
266*16d86563SAlexander Pyhalov ret_val += 2;
267*16d86563SAlexander Pyhalov }
268*16d86563SAlexander Pyhalov ib += 3;
269*16d86563SAlexander Pyhalov }
270*16d86563SAlexander Pyhalov else /* 11, 21, 26 & 31 bits codes won't be able to convert. */
271*16d86563SAlexander Pyhalov {
272*16d86563SAlexander Pyhalov short int offset;
273*16d86563SAlexander Pyhalov
274*16d86563SAlexander Pyhalov PROCESS_PRIOR_CVC();
275*16d86563SAlexander Pyhalov
276*16d86563SAlexander Pyhalov if ((*ib & 0xE0) == 0xC0) /* 11 */
277*16d86563SAlexander Pyhalov offset = 2;
278*16d86563SAlexander Pyhalov else if ((*ib & 0xF0) == 0xE0) /* 16 */
279*16d86563SAlexander Pyhalov offset = 3;
280*16d86563SAlexander Pyhalov else if ((*ib & 0xF8) == 0xF0) /* 21 */
281*16d86563SAlexander Pyhalov offset = 4;
282*16d86563SAlexander Pyhalov else if ((*ib & 0xFC) == 0xF8) /* 26 */
283*16d86563SAlexander Pyhalov offset = 5;
284*16d86563SAlexander Pyhalov else if ((*ib & 0xFE) == 0xFC) /* 31 */
285*16d86563SAlexander Pyhalov offset = 6;
286*16d86563SAlexander Pyhalov else /* Illegal sequence. */
287*16d86563SAlexander Pyhalov offset = 1;
288*16d86563SAlexander Pyhalov
289*16d86563SAlexander Pyhalov if ((ibtail - ib) < offset)
290*16d86563SAlexander Pyhalov {
291*16d86563SAlexander Pyhalov errno = EINVAL;
292*16d86563SAlexander Pyhalov ret_val = (size_t)-1;
293*16d86563SAlexander Pyhalov break;
294*16d86563SAlexander Pyhalov }
295*16d86563SAlexander Pyhalov ib += offset;
296*16d86563SAlexander Pyhalov
297*16d86563SAlexander Pyhalov /* Let's assume the code is non-identical. */
298*16d86563SAlexander Pyhalov offset = (offset > 2) ? 2 : 1;
299*16d86563SAlexander Pyhalov if ((obtail - ob) < offset)
300*16d86563SAlexander Pyhalov {
301*16d86563SAlexander Pyhalov errno = E2BIG;
302*16d86563SAlexander Pyhalov ret_val = (size_t)-1;
303*16d86563SAlexander Pyhalov break;
304*16d86563SAlexander Pyhalov }
305*16d86563SAlexander Pyhalov *ob++ = NON_IDENTICAL;
306*16d86563SAlexander Pyhalov if (offset > 1)
307*16d86563SAlexander Pyhalov *ob++ = NON_IDENTICAL;
308*16d86563SAlexander Pyhalov ret_val += offset;
309*16d86563SAlexander Pyhalov }
310*16d86563SAlexander Pyhalov }
311*16d86563SAlexander Pyhalov
312*16d86563SAlexander Pyhalov *inbuf = (char*)ib;
313*16d86563SAlexander Pyhalov *inbufleft = ibtail - ib;
314*16d86563SAlexander Pyhalov *outbuf = (char*)ob;
315*16d86563SAlexander Pyhalov *outbufleft = obtail - ob;
316*16d86563SAlexander Pyhalov
317*16d86563SAlexander Pyhalov return(ret_val);
318*16d86563SAlexander Pyhalov } /* end of size_t _icv_iconv(_conv_desc*, char**, size_t*, char**, size_t*).*/
319