1*91e1e26aSAlexander Pyhalov /*
2*91e1e26aSAlexander Pyhalov * CDDL HEADER START
3*91e1e26aSAlexander Pyhalov *
4*91e1e26aSAlexander Pyhalov * The contents of this file are subject to the terms of the
5*91e1e26aSAlexander Pyhalov * Common Development and Distribution License (the "License").
6*91e1e26aSAlexander Pyhalov * You may not use this file except in compliance with the License.
7*91e1e26aSAlexander Pyhalov *
8*91e1e26aSAlexander Pyhalov * You can obtain a copy of the license at src/OPENSOLARIS.LICENSE
9*91e1e26aSAlexander Pyhalov * or http://www.opensolaris.org/os/licensing.
10*91e1e26aSAlexander Pyhalov * See the License for the specific language governing permissions
11*91e1e26aSAlexander Pyhalov * and limitations under the License.
12*91e1e26aSAlexander Pyhalov *
13*91e1e26aSAlexander Pyhalov * When distributing Covered Code, include this CDDL HEADER in each
14*91e1e26aSAlexander Pyhalov * file and include the License file at src/OPENSOLARIS.LICENSE.
15*91e1e26aSAlexander Pyhalov * If applicable, add the following below this CDDL HEADER, with the
16*91e1e26aSAlexander Pyhalov * fields enclosed by brackets "[]" replaced with your own identifying
17*91e1e26aSAlexander Pyhalov * information: Portions Copyright [yyyy] [name of copyright owner]
18*91e1e26aSAlexander Pyhalov *
19*91e1e26aSAlexander Pyhalov * CDDL HEADER END
20*91e1e26aSAlexander Pyhalov */
21*91e1e26aSAlexander Pyhalov /*
22*91e1e26aSAlexander Pyhalov * Copyright (c) 2008, by Sun Microsystems, Inc.
23*91e1e26aSAlexander Pyhalov * All rights reserved.
24*91e1e26aSAlexander Pyhalov */
25*91e1e26aSAlexander Pyhalov
26*91e1e26aSAlexander Pyhalov #include <stdio.h>
27*91e1e26aSAlexander Pyhalov #include <errno.h>
28*91e1e26aSAlexander Pyhalov #include <stdlib.h>
29*91e1e26aSAlexander Pyhalov #include <sys/types.h>
30*91e1e26aSAlexander Pyhalov #define __NEED_TCVN_2_UNI__
31*91e1e26aSAlexander Pyhalov #include <unicode_tcvn.h> /* Unicode to tcvn mapping table */
32*91e1e26aSAlexander Pyhalov #include <vi_combine.h>
33*91e1e26aSAlexander Pyhalov #include "common_defs.h"
34*91e1e26aSAlexander Pyhalov
35*91e1e26aSAlexander Pyhalov
36*91e1e26aSAlexander Pyhalov typedef struct _icv_state {
37*91e1e26aSAlexander Pyhalov int _errno; /* internal errno */
38*91e1e26aSAlexander Pyhalov unsigned long last;
39*91e1e26aSAlexander Pyhalov } _iconv_st;
40*91e1e26aSAlexander Pyhalov
41*91e1e26aSAlexander Pyhalov #if defined(UCS_2LE)
42*91e1e26aSAlexander Pyhalov #define SET_UCS(UNI) *(*outbuf)++ = (unsigned char)((UNI)&0xff); \
43*91e1e26aSAlexander Pyhalov *(*outbuf)++ = (unsigned char)(((UNI)>>8)&0xff);
44*91e1e26aSAlexander Pyhalov #else
45*91e1e26aSAlexander Pyhalov #define SET_UCS(UNI) *(*outbuf)++ = (unsigned char)(((UNI)>>8)&0xff); \
46*91e1e26aSAlexander Pyhalov *(*outbuf)++ = (unsigned char)((UNI)&0xff);
47*91e1e26aSAlexander Pyhalov #endif
48*91e1e26aSAlexander Pyhalov
49*91e1e26aSAlexander Pyhalov static int binsearch(unsigned long x, Combine_map v[], int n);
50*91e1e26aSAlexander Pyhalov
51*91e1e26aSAlexander Pyhalov /*
52*91e1e26aSAlexander Pyhalov * Open; called from iconv_open()
53*91e1e26aSAlexander Pyhalov */
54*91e1e26aSAlexander Pyhalov void *
_icv_open()55*91e1e26aSAlexander Pyhalov _icv_open()
56*91e1e26aSAlexander Pyhalov {
57*91e1e26aSAlexander Pyhalov _iconv_st *st;
58*91e1e26aSAlexander Pyhalov
59*91e1e26aSAlexander Pyhalov if ((st = (_iconv_st *)malloc(sizeof(_iconv_st))) == NULL) {
60*91e1e26aSAlexander Pyhalov errno = ENOMEM;
61*91e1e26aSAlexander Pyhalov return ((void *) -1);
62*91e1e26aSAlexander Pyhalov }
63*91e1e26aSAlexander Pyhalov
64*91e1e26aSAlexander Pyhalov st->_errno = 0;
65*91e1e26aSAlexander Pyhalov st->last = 0;
66*91e1e26aSAlexander Pyhalov return ((void *) st);
67*91e1e26aSAlexander Pyhalov }
68*91e1e26aSAlexander Pyhalov
69*91e1e26aSAlexander Pyhalov
70*91e1e26aSAlexander Pyhalov /*
71*91e1e26aSAlexander Pyhalov * Close; called from iconv_close()
72*91e1e26aSAlexander Pyhalov */
73*91e1e26aSAlexander Pyhalov void
_icv_close(_iconv_st * st)74*91e1e26aSAlexander Pyhalov _icv_close(_iconv_st *st)
75*91e1e26aSAlexander Pyhalov {
76*91e1e26aSAlexander Pyhalov if (!st)
77*91e1e26aSAlexander Pyhalov errno = EBADF;
78*91e1e26aSAlexander Pyhalov else
79*91e1e26aSAlexander Pyhalov free(st);
80*91e1e26aSAlexander Pyhalov }
81*91e1e26aSAlexander Pyhalov
82*91e1e26aSAlexander Pyhalov
83*91e1e26aSAlexander Pyhalov /*
84*91e1e26aSAlexander Pyhalov * Actual conversion; called from iconv()
85*91e1e26aSAlexander Pyhalov */
86*91e1e26aSAlexander Pyhalov size_t
_icv_iconv(_iconv_st * st,char ** inbuf,size_t * inbytesleft,char ** outbuf,size_t * outbytesleft)87*91e1e26aSAlexander Pyhalov _icv_iconv(_iconv_st *st, char **inbuf, size_t *inbytesleft,
88*91e1e26aSAlexander Pyhalov char **outbuf, size_t *outbytesleft)
89*91e1e26aSAlexander Pyhalov {
90*91e1e26aSAlexander Pyhalov int unidx = -1;
91*91e1e26aSAlexander Pyhalov #ifdef DEBUG
92*91e1e26aSAlexander Pyhalov fprintf(stderr, "========== iconv(): TCVN5712 -->UCS-2 ==========\n");
93*91e1e26aSAlexander Pyhalov #endif
94*91e1e26aSAlexander Pyhalov if (st == NULL) {
95*91e1e26aSAlexander Pyhalov errno = EBADF;
96*91e1e26aSAlexander Pyhalov return ((size_t) -1);
97*91e1e26aSAlexander Pyhalov }
98*91e1e26aSAlexander Pyhalov
99*91e1e26aSAlexander Pyhalov if (inbuf == NULL || *inbuf == NULL) { /* Reset request. */
100*91e1e26aSAlexander Pyhalov st->_errno = 0;
101*91e1e26aSAlexander Pyhalov return ((size_t) 0);
102*91e1e26aSAlexander Pyhalov }
103*91e1e26aSAlexander Pyhalov
104*91e1e26aSAlexander Pyhalov st->_errno = 0; /* Reset internal errno */
105*91e1e26aSAlexander Pyhalov errno = 0; /* Reset external errno */
106*91e1e26aSAlexander Pyhalov
107*91e1e26aSAlexander Pyhalov /* Convert tcvn encoding to UCS-2 */
108*91e1e26aSAlexander Pyhalov while (*inbytesleft > 0 && *outbytesleft > 1) {
109*91e1e26aSAlexander Pyhalov unsigned long uni = 0;
110*91e1e26aSAlexander Pyhalov
111*91e1e26aSAlexander Pyhalov tcvn_2_uni((unsigned char*)*inbuf, &uni);
112*91e1e26aSAlexander Pyhalov if (st->last != 0) {
113*91e1e26aSAlexander Pyhalov if (ISCOMB_UNI(uni)) {
114*91e1e26aSAlexander Pyhalov /*
115*91e1e26aSAlexander Pyhalov * Composed characters with combine character
116*91e1e26aSAlexander Pyhalov */
117*91e1e26aSAlexander Pyhalov unsigned int k = 0;
118*91e1e26aSAlexander Pyhalov switch (uni) {
119*91e1e26aSAlexander Pyhalov case 0x0300: k = 0; break;
120*91e1e26aSAlexander Pyhalov case 0x0301: k = 1; break;
121*91e1e26aSAlexander Pyhalov case 0x0303: k = 2; break;
122*91e1e26aSAlexander Pyhalov case 0x0309: k = 3; break;
123*91e1e26aSAlexander Pyhalov case 0x0323: k = 4; break;
124*91e1e26aSAlexander Pyhalov default:
125*91e1e26aSAlexander Pyhalov break;
126*91e1e26aSAlexander Pyhalov }
127*91e1e26aSAlexander Pyhalov unidx = binsearch(st->last, vi_comb_data, VOWEL_NUM);
128*91e1e26aSAlexander Pyhalov if (unidx >= 0) {
129*91e1e26aSAlexander Pyhalov uni = vi_comb_data[unidx].composed[k];
130*91e1e26aSAlexander Pyhalov } else {
131*91e1e26aSAlexander Pyhalov errno = EBADF;
132*91e1e26aSAlexander Pyhalov return ((size_t)-1);
133*91e1e26aSAlexander Pyhalov }
134*91e1e26aSAlexander Pyhalov st->last = 0;
135*91e1e26aSAlexander Pyhalov
136*91e1e26aSAlexander Pyhalov } else {
137*91e1e26aSAlexander Pyhalov SET_UCS(st->last);
138*91e1e26aSAlexander Pyhalov (*outbytesleft) -= 2;
139*91e1e26aSAlexander Pyhalov }
140*91e1e26aSAlexander Pyhalov st->last = 0;
141*91e1e26aSAlexander Pyhalov } else {
142*91e1e26aSAlexander Pyhalov if (uni >= 0x0041 && uni <= 0x01b0
143*91e1e26aSAlexander Pyhalov && ((tcvn_comp_bases_mask[(uni-0x0040) >> 5] >> (uni & 0x1f)) & 1)) {
144*91e1e26aSAlexander Pyhalov /*
145*91e1e26aSAlexander Pyhalov * uni is vowel, it's a possible match with combine character.
146*91e1e26aSAlexander Pyhalov * Buffer it.
147*91e1e26aSAlexander Pyhalov * */
148*91e1e26aSAlexander Pyhalov st->last = uni;
149*91e1e26aSAlexander Pyhalov (*inbuf)++;
150*91e1e26aSAlexander Pyhalov (*inbytesleft)--;
151*91e1e26aSAlexander Pyhalov continue;
152*91e1e26aSAlexander Pyhalov }
153*91e1e26aSAlexander Pyhalov }
154*91e1e26aSAlexander Pyhalov SET_UCS(uni);
155*91e1e26aSAlexander Pyhalov (*outbytesleft) -= 2;
156*91e1e26aSAlexander Pyhalov (*inbuf)++;
157*91e1e26aSAlexander Pyhalov (*inbytesleft)--;
158*91e1e26aSAlexander Pyhalov
159*91e1e26aSAlexander Pyhalov }
160*91e1e26aSAlexander Pyhalov
161*91e1e26aSAlexander Pyhalov if ( *inbytesleft > 0 && *outbytesleft <= 1 ) {
162*91e1e26aSAlexander Pyhalov errno = E2BIG;
163*91e1e26aSAlexander Pyhalov st->last = 0;
164*91e1e26aSAlexander Pyhalov return ((size_t)-1);
165*91e1e26aSAlexander Pyhalov }
166*91e1e26aSAlexander Pyhalov if (st->last!=0 ) {
167*91e1e26aSAlexander Pyhalov SET_UCS(st->last);
168*91e1e26aSAlexander Pyhalov st->last = 0;
169*91e1e26aSAlexander Pyhalov (*outbytesleft) -= 2;
170*91e1e26aSAlexander Pyhalov }
171*91e1e26aSAlexander Pyhalov
172*91e1e26aSAlexander Pyhalov return ((size_t)(*inbytesleft));
173*91e1e26aSAlexander Pyhalov }
174*91e1e26aSAlexander Pyhalov
175*91e1e26aSAlexander Pyhalov /* binsearch: find x in v[0] <= v[1] <= ... <= v[n-1] */
binsearch(unsigned long x,Combine_map v[],int n)176*91e1e26aSAlexander Pyhalov static int binsearch(unsigned long x, Combine_map v[], int n)
177*91e1e26aSAlexander Pyhalov {
178*91e1e26aSAlexander Pyhalov int low = 0;
179*91e1e26aSAlexander Pyhalov int mid = 0;
180*91e1e26aSAlexander Pyhalov int high = n - 1;
181*91e1e26aSAlexander Pyhalov
182*91e1e26aSAlexander Pyhalov low = 0;
183*91e1e26aSAlexander Pyhalov while (low <= high) {
184*91e1e26aSAlexander Pyhalov mid = ((high - low)>>1) + low;
185*91e1e26aSAlexander Pyhalov if (x < v[mid].base)
186*91e1e26aSAlexander Pyhalov high = mid - 1;
187*91e1e26aSAlexander Pyhalov else if (x > v[mid].base)
188*91e1e26aSAlexander Pyhalov low = mid + 1;
189*91e1e26aSAlexander Pyhalov else
190*91e1e26aSAlexander Pyhalov /* found match */
191*91e1e26aSAlexander Pyhalov return mid;
192*91e1e26aSAlexander Pyhalov }
193*91e1e26aSAlexander Pyhalov
194*91e1e26aSAlexander Pyhalov /* no match */
195*91e1e26aSAlexander Pyhalov return (-1);
196*91e1e26aSAlexander Pyhalov }
197