1*880d7978SAlexander Pyhalov /*
2*880d7978SAlexander Pyhalov * CDDL HEADER START
3*880d7978SAlexander Pyhalov *
4*880d7978SAlexander Pyhalov * The contents of this file are subject to the terms of the
5*880d7978SAlexander Pyhalov * Common Development and Distribution License (the "License").
6*880d7978SAlexander Pyhalov * You may not use this file except in compliance with the License.
7*880d7978SAlexander Pyhalov *
8*880d7978SAlexander Pyhalov * You can obtain a copy of the license at src/OPENSOLARIS.LICENSE
9*880d7978SAlexander Pyhalov * or http://www.opensolaris.org/os/licensing.
10*880d7978SAlexander Pyhalov * See the License for the specific language governing permissions
11*880d7978SAlexander Pyhalov * and limitations under the License.
12*880d7978SAlexander Pyhalov *
13*880d7978SAlexander Pyhalov * When distributing Covered Code, include this CDDL HEADER in each
14*880d7978SAlexander Pyhalov * file and include the License file at src/OPENSOLARIS.LICENSE.
15*880d7978SAlexander Pyhalov * If applicable, add the following below this CDDL HEADER, with the
16*880d7978SAlexander Pyhalov * fields enclosed by brackets "[]" replaced with your own identifying
17*880d7978SAlexander Pyhalov * information: Portions Copyright [yyyy] [name of copyright owner]
18*880d7978SAlexander Pyhalov *
19*880d7978SAlexander Pyhalov * CDDL HEADER END
20*880d7978SAlexander Pyhalov */
21*880d7978SAlexander Pyhalov /*
22*880d7978SAlexander Pyhalov * Copyright (c) 2008, by Sun Microsystems, Inc.
23*880d7978SAlexander Pyhalov * All rights reserved.
24*880d7978SAlexander Pyhalov */
25*880d7978SAlexander Pyhalov
26*880d7978SAlexander Pyhalov #include <stdio.h>
27*880d7978SAlexander Pyhalov #include <errno.h>
28*880d7978SAlexander Pyhalov #include <stdlib.h>
29*880d7978SAlexander Pyhalov #include <sys/types.h>
30*880d7978SAlexander Pyhalov #define __NEED_TCVN_2_UNI__
31*880d7978SAlexander Pyhalov #include <unicode_tcvn.h> /* Unicode to tcvn mapping table */
32*880d7978SAlexander Pyhalov #include <vi_combine.h>
33*880d7978SAlexander Pyhalov #include "common_defs.h"
34*880d7978SAlexander Pyhalov
35*880d7978SAlexander Pyhalov
36*880d7978SAlexander Pyhalov typedef struct _icv_state {
37*880d7978SAlexander Pyhalov int _errno; /* internal errno */
38*880d7978SAlexander Pyhalov unsigned long last;
39*880d7978SAlexander Pyhalov } _iconv_st;
40*880d7978SAlexander Pyhalov
41*880d7978SAlexander Pyhalov #if defined(UCS_2LE)
42*880d7978SAlexander Pyhalov #define SET_UCS(UNI) *(*outbuf)++ = (unsigned char)((UNI)&0xff); \
43*880d7978SAlexander Pyhalov *(*outbuf)++ = (unsigned char)(((UNI)>>8)&0xff);
44*880d7978SAlexander Pyhalov #else
45*880d7978SAlexander Pyhalov #define SET_UCS(UNI) *(*outbuf)++ = (unsigned char)(((UNI)>>8)&0xff); \
46*880d7978SAlexander Pyhalov *(*outbuf)++ = (unsigned char)((UNI)&0xff);
47*880d7978SAlexander Pyhalov #endif
48*880d7978SAlexander Pyhalov
49*880d7978SAlexander Pyhalov static int binsearch(unsigned long x, Combine_map v[], int n);
50*880d7978SAlexander Pyhalov
51*880d7978SAlexander Pyhalov /*
52*880d7978SAlexander Pyhalov * Open; called from iconv_open()
53*880d7978SAlexander Pyhalov */
54*880d7978SAlexander Pyhalov void *
_icv_open()55*880d7978SAlexander Pyhalov _icv_open()
56*880d7978SAlexander Pyhalov {
57*880d7978SAlexander Pyhalov _iconv_st *st;
58*880d7978SAlexander Pyhalov
59*880d7978SAlexander Pyhalov if ((st = (_iconv_st *)malloc(sizeof(_iconv_st))) == NULL) {
60*880d7978SAlexander Pyhalov errno = ENOMEM;
61*880d7978SAlexander Pyhalov return ((void *) -1);
62*880d7978SAlexander Pyhalov }
63*880d7978SAlexander Pyhalov
64*880d7978SAlexander Pyhalov st->_errno = 0;
65*880d7978SAlexander Pyhalov st->last = 0;
66*880d7978SAlexander Pyhalov return ((void *) st);
67*880d7978SAlexander Pyhalov }
68*880d7978SAlexander Pyhalov
69*880d7978SAlexander Pyhalov
70*880d7978SAlexander Pyhalov /*
71*880d7978SAlexander Pyhalov * Close; called from iconv_close()
72*880d7978SAlexander Pyhalov */
73*880d7978SAlexander Pyhalov void
_icv_close(_iconv_st * st)74*880d7978SAlexander Pyhalov _icv_close(_iconv_st *st)
75*880d7978SAlexander Pyhalov {
76*880d7978SAlexander Pyhalov if (!st)
77*880d7978SAlexander Pyhalov errno = EBADF;
78*880d7978SAlexander Pyhalov else
79*880d7978SAlexander Pyhalov free(st);
80*880d7978SAlexander Pyhalov }
81*880d7978SAlexander Pyhalov
82*880d7978SAlexander Pyhalov
83*880d7978SAlexander Pyhalov /*
84*880d7978SAlexander Pyhalov * Actual conversion; called from iconv()
85*880d7978SAlexander Pyhalov */
86*880d7978SAlexander Pyhalov size_t
_icv_iconv(_iconv_st * st,char ** inbuf,size_t * inbytesleft,char ** outbuf,size_t * outbytesleft)87*880d7978SAlexander Pyhalov _icv_iconv(_iconv_st *st, char **inbuf, size_t *inbytesleft,
88*880d7978SAlexander Pyhalov char **outbuf, size_t *outbytesleft)
89*880d7978SAlexander Pyhalov {
90*880d7978SAlexander Pyhalov int unidx = -1;
91*880d7978SAlexander Pyhalov #ifdef DEBUG
92*880d7978SAlexander Pyhalov fprintf(stderr, "========== iconv(): TCVN5712 -->UCS-2 ==========\n");
93*880d7978SAlexander Pyhalov #endif
94*880d7978SAlexander Pyhalov if (st == NULL) {
95*880d7978SAlexander Pyhalov errno = EBADF;
96*880d7978SAlexander Pyhalov return ((size_t) -1);
97*880d7978SAlexander Pyhalov }
98*880d7978SAlexander Pyhalov
99*880d7978SAlexander Pyhalov if (inbuf == NULL || *inbuf == NULL) { /* Reset request. */
100*880d7978SAlexander Pyhalov st->_errno = 0;
101*880d7978SAlexander Pyhalov return ((size_t) 0);
102*880d7978SAlexander Pyhalov }
103*880d7978SAlexander Pyhalov
104*880d7978SAlexander Pyhalov st->_errno = 0; /* Reset internal errno */
105*880d7978SAlexander Pyhalov errno = 0; /* Reset external errno */
106*880d7978SAlexander Pyhalov
107*880d7978SAlexander Pyhalov /* Convert tcvn encoding to UCS-2 */
108*880d7978SAlexander Pyhalov while (*inbytesleft > 0 && *outbytesleft > 1) {
109*880d7978SAlexander Pyhalov unsigned long uni = 0;
110*880d7978SAlexander Pyhalov
111*880d7978SAlexander Pyhalov tcvn_2_uni((unsigned char*)*inbuf, &uni);
112*880d7978SAlexander Pyhalov if (st->last != 0) {
113*880d7978SAlexander Pyhalov if (ISCOMB_UNI(uni)) {
114*880d7978SAlexander Pyhalov /*
115*880d7978SAlexander Pyhalov * Composed characters with combine character
116*880d7978SAlexander Pyhalov */
117*880d7978SAlexander Pyhalov unsigned int k = 0;
118*880d7978SAlexander Pyhalov switch (uni) {
119*880d7978SAlexander Pyhalov case 0x0300: k = 0; break;
120*880d7978SAlexander Pyhalov case 0x0301: k = 1; break;
121*880d7978SAlexander Pyhalov case 0x0303: k = 2; break;
122*880d7978SAlexander Pyhalov case 0x0309: k = 3; break;
123*880d7978SAlexander Pyhalov case 0x0323: k = 4; break;
124*880d7978SAlexander Pyhalov default:
125*880d7978SAlexander Pyhalov break;
126*880d7978SAlexander Pyhalov }
127*880d7978SAlexander Pyhalov unidx = binsearch(st->last, vi_comb_data, VOWEL_NUM);
128*880d7978SAlexander Pyhalov if (unidx >= 0) {
129*880d7978SAlexander Pyhalov uni = vi_comb_data[unidx].composed[k];
130*880d7978SAlexander Pyhalov } else {
131*880d7978SAlexander Pyhalov errno = EBADF;
132*880d7978SAlexander Pyhalov return ((size_t)-1);
133*880d7978SAlexander Pyhalov }
134*880d7978SAlexander Pyhalov st->last = 0;
135*880d7978SAlexander Pyhalov
136*880d7978SAlexander Pyhalov } else {
137*880d7978SAlexander Pyhalov SET_UCS(st->last);
138*880d7978SAlexander Pyhalov (*outbytesleft) -= 2;
139*880d7978SAlexander Pyhalov }
140*880d7978SAlexander Pyhalov st->last = 0;
141*880d7978SAlexander Pyhalov } else {
142*880d7978SAlexander Pyhalov if (uni >= 0x0041 && uni <= 0x01b0
143*880d7978SAlexander Pyhalov && ((tcvn_comp_bases_mask[(uni-0x0040) >> 5] >> (uni & 0x1f)) & 1)) {
144*880d7978SAlexander Pyhalov /*
145*880d7978SAlexander Pyhalov * uni is vowel, it's a possible match with combine character.
146*880d7978SAlexander Pyhalov * Buffer it.
147*880d7978SAlexander Pyhalov * */
148*880d7978SAlexander Pyhalov st->last = uni;
149*880d7978SAlexander Pyhalov (*inbuf)++;
150*880d7978SAlexander Pyhalov (*inbytesleft)--;
151*880d7978SAlexander Pyhalov continue;
152*880d7978SAlexander Pyhalov }
153*880d7978SAlexander Pyhalov }
154*880d7978SAlexander Pyhalov SET_UCS(uni);
155*880d7978SAlexander Pyhalov (*outbytesleft) -= 2;
156*880d7978SAlexander Pyhalov (*inbuf)++;
157*880d7978SAlexander Pyhalov (*inbytesleft)--;
158*880d7978SAlexander Pyhalov
159*880d7978SAlexander Pyhalov }
160*880d7978SAlexander Pyhalov
161*880d7978SAlexander Pyhalov if ( *inbytesleft > 0 && *outbytesleft <= 1 ) {
162*880d7978SAlexander Pyhalov errno = E2BIG;
163*880d7978SAlexander Pyhalov st->last = 0;
164*880d7978SAlexander Pyhalov return ((size_t)-1);
165*880d7978SAlexander Pyhalov }
166*880d7978SAlexander Pyhalov if (st->last!=0 ) {
167*880d7978SAlexander Pyhalov SET_UCS(st->last);
168*880d7978SAlexander Pyhalov st->last = 0;
169*880d7978SAlexander Pyhalov (*outbytesleft) -= 2;
170*880d7978SAlexander Pyhalov }
171*880d7978SAlexander Pyhalov
172*880d7978SAlexander Pyhalov return ((size_t)(*inbytesleft));
173*880d7978SAlexander Pyhalov }
174*880d7978SAlexander Pyhalov
175*880d7978SAlexander Pyhalov /* binsearch: find x in v[0] <= v[1] <= ... <= v[n-1] */
binsearch(unsigned long x,Combine_map v[],int n)176*880d7978SAlexander Pyhalov static int binsearch(unsigned long x, Combine_map v[], int n)
177*880d7978SAlexander Pyhalov {
178*880d7978SAlexander Pyhalov int low = 0;
179*880d7978SAlexander Pyhalov int mid = 0;
180*880d7978SAlexander Pyhalov int high = n - 1;
181*880d7978SAlexander Pyhalov
182*880d7978SAlexander Pyhalov low = 0;
183*880d7978SAlexander Pyhalov while (low <= high) {
184*880d7978SAlexander Pyhalov mid = ((high - low)>>1) + low;
185*880d7978SAlexander Pyhalov if (x < v[mid].base)
186*880d7978SAlexander Pyhalov high = mid - 1;
187*880d7978SAlexander Pyhalov else if (x > v[mid].base)
188*880d7978SAlexander Pyhalov low = mid + 1;
189*880d7978SAlexander Pyhalov else
190*880d7978SAlexander Pyhalov /* found match */
191*880d7978SAlexander Pyhalov return mid;
192*880d7978SAlexander Pyhalov }
193*880d7978SAlexander Pyhalov
194*880d7978SAlexander Pyhalov /* no match */
195*880d7978SAlexander Pyhalov return (-1);
196*880d7978SAlexander Pyhalov }
197