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 /*
23*16d86563SAlexander Pyhalov * Copyright (c) 1997, by Sun Microsystems, Inc.
24*16d86563SAlexander Pyhalov * All rights reserved.
25*16d86563SAlexander Pyhalov */
26*16d86563SAlexander Pyhalov
27*16d86563SAlexander Pyhalov #include <stdio.h>
28*16d86563SAlexander Pyhalov #include <stdlib.h>
29*16d86563SAlexander Pyhalov #include <errno.h>
30*16d86563SAlexander Pyhalov #include <sys/types.h>
31*16d86563SAlexander Pyhalov
32*16d86563SAlexander Pyhalov #include "tab_lookup.h" /* table lookup data types */
33*16d86563SAlexander Pyhalov
34*16d86563SAlexander Pyhalov #define MSB 0x80 /* most significant bit */
35*16d86563SAlexander Pyhalov #define MBYTE 0x8e /* multi-byte (4 byte character) */
36*16d86563SAlexander Pyhalov #define PMASK 0xa0 /* plane number mask */
37*16d86563SAlexander Pyhalov #define ONEBYTE 0xff /* right most byte */
38*16d86563SAlexander Pyhalov
39*16d86563SAlexander Pyhalov /* non-identified character */
40*16d86563SAlexander Pyhalov #define UTF8_NON_ID_CHAR1 0xEF
41*16d86563SAlexander Pyhalov #define UTF8_NON_ID_CHAR2 0xBF
42*16d86563SAlexander Pyhalov #define UTF8_NON_ID_CHAR3 0xBD
43*16d86563SAlexander Pyhalov
44*16d86563SAlexander Pyhalov enum _USTATE { C0, C1 };
45*16d86563SAlexander Pyhalov
46*16d86563SAlexander Pyhalov
47*16d86563SAlexander Pyhalov
48*16d86563SAlexander Pyhalov
49*16d86563SAlexander Pyhalov /*
50*16d86563SAlexander Pyhalov * Actual conversion; called from iconv()
51*16d86563SAlexander Pyhalov * Input is UTF-8 data.
52*16d86563SAlexander Pyhalov * first convert to UCS2
53*16d86563SAlexander Pyhalov */
54*16d86563SAlexander Pyhalov size_t
_icv_iconv(_icv_state * st,char ** inbuf,size_t * inbytesleft,char ** outbuf,size_t * outbytesleft)55*16d86563SAlexander Pyhalov _icv_iconv(_icv_state *st, char **inbuf, size_t *inbytesleft,
56*16d86563SAlexander Pyhalov char **outbuf, size_t *outbytesleft)
57*16d86563SAlexander Pyhalov {
58*16d86563SAlexander Pyhalov /*
59*16d86563SAlexander Pyhalov * Actual conversion; called from iconv()
60*16d86563SAlexander Pyhalov */
61*16d86563SAlexander Pyhalov
62*16d86563SAlexander Pyhalov char c1, c2;
63*16d86563SAlexander Pyhalov int n, unidx;
64*16d86563SAlexander Pyhalov unsigned long ibm_code;
65*16d86563SAlexander Pyhalov
66*16d86563SAlexander Pyhalov #ifdef DEBUG
67*16d86563SAlexander Pyhalov fprintf(stderr, "========== iconv(): IBM --> UTF8 ==========\n");
68*16d86563SAlexander Pyhalov #endif
69*16d86563SAlexander Pyhalov
70*16d86563SAlexander Pyhalov if (st == NULL) {
71*16d86563SAlexander Pyhalov errno = EBADF;
72*16d86563SAlexander Pyhalov return ((size_t) -1);
73*16d86563SAlexander Pyhalov }
74*16d86563SAlexander Pyhalov
75*16d86563SAlexander Pyhalov if (inbuf == NULL || *inbuf == NULL) { /* Reset request. */
76*16d86563SAlexander Pyhalov st->ustate = C0;
77*16d86563SAlexander Pyhalov st->_errno = 0;
78*16d86563SAlexander Pyhalov return ((size_t) 0);
79*16d86563SAlexander Pyhalov }
80*16d86563SAlexander Pyhalov
81*16d86563SAlexander Pyhalov st->_errno = 0; /* reset internal errno */
82*16d86563SAlexander Pyhalov errno = 0; /* reset external errno */
83*16d86563SAlexander Pyhalov
84*16d86563SAlexander Pyhalov /* a state machine for interpreting UTF8 code */
85*16d86563SAlexander Pyhalov while (*inbytesleft > 0 && *outbytesleft > 0) {
86*16d86563SAlexander Pyhalov switch (st->ustate) {
87*16d86563SAlexander Pyhalov case C0 :
88*16d86563SAlexander Pyhalov st->keepc[0] = (**inbuf);
89*16d86563SAlexander Pyhalov st->ustate = C1;
90*16d86563SAlexander Pyhalov break;
91*16d86563SAlexander Pyhalov
92*16d86563SAlexander Pyhalov case C1 :
93*16d86563SAlexander Pyhalov st->keepc[1] = (**inbuf);
94*16d86563SAlexander Pyhalov n = ibm_to_utf8(st, *outbuf, *outbytesleft);
95*16d86563SAlexander Pyhalov if (n > 0) {
96*16d86563SAlexander Pyhalov (*outbuf) += n;
97*16d86563SAlexander Pyhalov (*outbytesleft) -= n;
98*16d86563SAlexander Pyhalov } else {
99*16d86563SAlexander Pyhalov st->_errno = errno;
100*16d86563SAlexander Pyhalov return((size_t)-1);
101*16d86563SAlexander Pyhalov }
102*16d86563SAlexander Pyhalov st->ustate = C0;
103*16d86563SAlexander Pyhalov st->_errno = 0;
104*16d86563SAlexander Pyhalov break;
105*16d86563SAlexander Pyhalov
106*16d86563SAlexander Pyhalov default: /* should never come here */
107*16d86563SAlexander Pyhalov st->_errno = errno = EILSEQ;
108*16d86563SAlexander Pyhalov st->ustate = C0; /* reset state */
109*16d86563SAlexander Pyhalov break;
110*16d86563SAlexander Pyhalov }
111*16d86563SAlexander Pyhalov
112*16d86563SAlexander Pyhalov
113*16d86563SAlexander Pyhalov (*inbuf)++;
114*16d86563SAlexander Pyhalov (*inbytesleft)--;
115*16d86563SAlexander Pyhalov
116*16d86563SAlexander Pyhalov if (st->_errno) {
117*16d86563SAlexander Pyhalov #ifdef DEBUG
118*16d86563SAlexander Pyhalov fprintf(stderr, "!!!!!\tst->_errno = %d\tst->ustate = %d\n",
119*16d86563SAlexander Pyhalov st->_errno, st->ustate);
120*16d86563SAlexander Pyhalov #endif
121*16d86563SAlexander Pyhalov break;
122*16d86563SAlexander Pyhalov }
123*16d86563SAlexander Pyhalov
124*16d86563SAlexander Pyhalov if (errno)
125*16d86563SAlexander Pyhalov return((size_t)-1);
126*16d86563SAlexander Pyhalov }
127*16d86563SAlexander Pyhalov
128*16d86563SAlexander Pyhalov if (*outbytesleft == 0) {
129*16d86563SAlexander Pyhalov errno = E2BIG;
130*16d86563SAlexander Pyhalov return((size_t)-1);
131*16d86563SAlexander Pyhalov }
132*16d86563SAlexander Pyhalov return (*inbytesleft);
133*16d86563SAlexander Pyhalov }
134*16d86563SAlexander Pyhalov
135*16d86563SAlexander Pyhalov /*
136*16d86563SAlexander Pyhalov * IBM code --> (Unicode)
137*16d86563SAlexander Pyhalov * Unicode --> UTF8 (FSS-UTF)
138*16d86563SAlexander Pyhalov * (File System Safe Universal Character Set Transformation Format)
139*16d86563SAlexander Pyhalov * Return: > 0 - converted with enough space in output buffer
140*16d86563SAlexander Pyhalov * = 0 - no space in outbuf
141*16d86563SAlexander Pyhalov */
ibm_to_utf8(st,buf,buflen)142*16d86563SAlexander Pyhalov int ibm_to_utf8(st, buf, buflen)
143*16d86563SAlexander Pyhalov _icv_state *st;
144*16d86563SAlexander Pyhalov char *buf;
145*16d86563SAlexander Pyhalov size_t buflen;
146*16d86563SAlexander Pyhalov {
147*16d86563SAlexander Pyhalov unsigned long ibm_val; /* Big-5 value */
148*16d86563SAlexander Pyhalov int unidx; /* Unicode index */
149*16d86563SAlexander Pyhalov unsigned long uni_val; /* Unicode */
150*16d86563SAlexander Pyhalov
151*16d86563SAlexander Pyhalov ibm_val = ((st->keepc[0]&ONEBYTE) << 8) + (st->keepc[1]&ONEBYTE);
152*16d86563SAlexander Pyhalov #ifdef DEBUG
153*16d86563SAlexander Pyhalov fprintf(stderr, "%x\t", ibm_val);
154*16d86563SAlexander Pyhalov #endif
155*16d86563SAlexander Pyhalov
156*16d86563SAlexander Pyhalov
157*16d86563SAlexander Pyhalov unidx = bisearch(ibm_val, st, st->table_size);
158*16d86563SAlexander Pyhalov
159*16d86563SAlexander Pyhalov if (unidx >= 0)
160*16d86563SAlexander Pyhalov {
161*16d86563SAlexander Pyhalov if ( st->left_to_right )
162*16d86563SAlexander Pyhalov uni_val = st->table[unidx].right_code;
163*16d86563SAlexander Pyhalov else
164*16d86563SAlexander Pyhalov uni_val = st->table[unidx].left_code;
165*16d86563SAlexander Pyhalov }
166*16d86563SAlexander Pyhalov
167*16d86563SAlexander Pyhalov #ifdef DEBUG
168*16d86563SAlexander Pyhalov fprintf(stderr, "unidx = %d, unicode = %x\t", unidx, uni_val);
169*16d86563SAlexander Pyhalov #endif
170*16d86563SAlexander Pyhalov
171*16d86563SAlexander Pyhalov if (unidx >= 0) { /* do Unicode to UTF8 conversion */
172*16d86563SAlexander Pyhalov if (uni_val > 0x0000 && uni_val <= 0x07ff) {
173*16d86563SAlexander Pyhalov if (buflen < 2) {
174*16d86563SAlexander Pyhalov #ifdef DEBUG
175*16d86563SAlexander Pyhalov fprintf(stderr, "outbuf overflow in ibm_to_utf8()!!\n");
176*16d86563SAlexander Pyhalov #endif
177*16d86563SAlexander Pyhalov errno = E2BIG;
178*16d86563SAlexander Pyhalov return(0);
179*16d86563SAlexander Pyhalov }
180*16d86563SAlexander Pyhalov *buf = (char)((uni_val >> 6) & 0x1f) | 0xc0;
181*16d86563SAlexander Pyhalov *(buf+1) = (char)(uni_val & 0x3f) | 0x80;
182*16d86563SAlexander Pyhalov #ifdef DEBUG
183*16d86563SAlexander Pyhalov fprintf(stderr, "%x %x\n", *buf&ONEBYTE, *(buf+1)&ONEBYTE);
184*16d86563SAlexander Pyhalov #endif
185*16d86563SAlexander Pyhalov return(2);
186*16d86563SAlexander Pyhalov }
187*16d86563SAlexander Pyhalov if (uni_val > 0x0800 && uni_val <= 0xffff) {
188*16d86563SAlexander Pyhalov if (buflen < 3) {
189*16d86563SAlexander Pyhalov #ifdef DEBUG
190*16d86563SAlexander Pyhalov fprintf(stderr, "outbuf overflow in ibm_to_utf8()!!\n");
191*16d86563SAlexander Pyhalov #endif
192*16d86563SAlexander Pyhalov errno = E2BIG;
193*16d86563SAlexander Pyhalov return(0);
194*16d86563SAlexander Pyhalov }
195*16d86563SAlexander Pyhalov *buf = (char)((uni_val >> 12) & 0xf) | 0xe0;
196*16d86563SAlexander Pyhalov *(buf+1) = (char)((uni_val >>6) & 0x3f) | 0x80;
197*16d86563SAlexander Pyhalov *(buf+2) = (char)(uni_val & 0x3f) | 0x80;
198*16d86563SAlexander Pyhalov #ifdef DEBUG
199*16d86563SAlexander Pyhalov fprintf(stderr, "%x %x %x\n", *buf&ONEBYTE, *(buf+1)&ONEBYTE, *(buf+2)&ONEBYTE);
200*16d86563SAlexander Pyhalov #endif
201*16d86563SAlexander Pyhalov return(3);
202*16d86563SAlexander Pyhalov }
203*16d86563SAlexander Pyhalov }
204*16d86563SAlexander Pyhalov
205*16d86563SAlexander Pyhalov /* can't find a match in IBM --> UTF8 table or illegal UTF8 code */
206*16d86563SAlexander Pyhalov if (buflen < 3) {
207*16d86563SAlexander Pyhalov #ifdef DEBUG
208*16d86563SAlexander Pyhalov fprintf(stderr, "outbuf overflow in ibm_to_utf8()!!\n");
209*16d86563SAlexander Pyhalov #endif
210*16d86563SAlexander Pyhalov errno = E2BIG;
211*16d86563SAlexander Pyhalov return(0);
212*16d86563SAlexander Pyhalov }
213*16d86563SAlexander Pyhalov
214*16d86563SAlexander Pyhalov *buf = (char)UTF8_NON_ID_CHAR1;
215*16d86563SAlexander Pyhalov *(buf+1) = (char)UTF8_NON_ID_CHAR2;
216*16d86563SAlexander Pyhalov *(buf+2) = (char)UTF8_NON_ID_CHAR3;
217*16d86563SAlexander Pyhalov
218*16d86563SAlexander Pyhalov #ifdef DEBUG
219*16d86563SAlexander Pyhalov fprintf(stderr, "%c %c %c\n", *buf, *(buf+1), *(buf+2));
220*16d86563SAlexander Pyhalov #endif
221*16d86563SAlexander Pyhalov return(3);
222*16d86563SAlexander Pyhalov }
223