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) 2001, Sun Microsystems, Inc.
23*16d86563SAlexander Pyhalov * All rights reserved.
24*16d86563SAlexander Pyhalov */
25*16d86563SAlexander Pyhalov
26*16d86563SAlexander Pyhalov #include <stdio.h>
27*16d86563SAlexander Pyhalov #include <stdlib.h>
28*16d86563SAlexander Pyhalov #include <errno.h>
29*16d86563SAlexander Pyhalov #include <gb18030_big5hk.h>
30*16d86563SAlexander Pyhalov
31*16d86563SAlexander Pyhalov #define NON_ID_CHAR '_' /* non-identified character */
32*16d86563SAlexander Pyhalov #define MSB 0x80
33*16d86563SAlexander Pyhalov #define ONEBYTE 0xff
34*16d86563SAlexander Pyhalov
35*16d86563SAlexander Pyhalov #define gbk4_2nd_byte(v) ( (v) >= 0x30 && (v) <= 0x39 )
36*16d86563SAlexander Pyhalov #define gbk4_3rd_byte(v) ( (v) >= 0x81 && (v) <= 0xfe )
37*16d86563SAlexander Pyhalov #define gbk4_4th_byte(v) gbk4_2nd_byte(v)
38*16d86563SAlexander Pyhalov
39*16d86563SAlexander Pyhalov int binsearch(unsigned long x, table_t table[], int n);
40*16d86563SAlexander Pyhalov int gbk_2nd_byte(char inbuf);
41*16d86563SAlexander Pyhalov int gbk_to_hkscs(char keepc[], char *buf, size_t buflen);
42*16d86563SAlexander Pyhalov int gbk4_to_hkscs(char keepc[], char *buf, size_t buflen);
43*16d86563SAlexander Pyhalov
44*16d86563SAlexander Pyhalov typedef struct _icv_state {
45*16d86563SAlexander Pyhalov char keepc[4]; /* maximum # byte of GB chararor in two bytes area */
46*16d86563SAlexander Pyhalov short cstate;
47*16d86563SAlexander Pyhalov int _errno; /* internal errno */
48*16d86563SAlexander Pyhalov } _iconv_st;
49*16d86563SAlexander Pyhalov
50*16d86563SAlexander Pyhalov enum _CSTATE { C0, C1, C2, C3 };
51*16d86563SAlexander Pyhalov
52*16d86563SAlexander Pyhalov /*=======================================================
53*16d86563SAlexander Pyhalov *
54*16d86563SAlexander Pyhalov * State Machine for interpreting GBK code
55*16d86563SAlexander Pyhalov *
56*16d86563SAlexander Pyhalov *=======================================================
57*16d86563SAlexander Pyhalov *
58*16d86563SAlexander Pyhalov * 3rd C
59*16d86563SAlexander Pyhalov * C2--------> C3
60*16d86563SAlexander Pyhalov * ^ |
61*16d86563SAlexander Pyhalov * 2nd C | 4th C |
62*16d86563SAlexander Pyhalov * 1st C | |
63*16d86563SAlexander Pyhalov * +--------> C0 ----------> C1 |
64*16d86563SAlexander Pyhalov * | ascii | 2nd C | |
65*16d86563SAlexander Pyhalov * ^ v v V
66*16d86563SAlexander Pyhalov * +----<-----+-----<--------+-----<------+
67*16d86563SAlexander Pyhalov *
68*16d86563SAlexander Pyhalov *=======================================================*/
69*16d86563SAlexander Pyhalov /*
70*16d86563SAlexander Pyhalov * Open; called from iconv_open()
71*16d86563SAlexander Pyhalov */
_icv_open()72*16d86563SAlexander Pyhalov void * _icv_open() {
73*16d86563SAlexander Pyhalov _iconv_st * st;
74*16d86563SAlexander Pyhalov
75*16d86563SAlexander Pyhalov if ((st = (_iconv_st *) malloc(sizeof(_iconv_st))) == NULL) {
76*16d86563SAlexander Pyhalov errno = ENOMEM;
77*16d86563SAlexander Pyhalov return ((void *) -1);
78*16d86563SAlexander Pyhalov }
79*16d86563SAlexander Pyhalov
80*16d86563SAlexander Pyhalov st->cstate = C0;
81*16d86563SAlexander Pyhalov st->_errno = 0;
82*16d86563SAlexander Pyhalov
83*16d86563SAlexander Pyhalov return ((void *) st);
84*16d86563SAlexander Pyhalov }
85*16d86563SAlexander Pyhalov
86*16d86563SAlexander Pyhalov /*
87*16d86563SAlexander Pyhalov * Close; called from iconv_close()
88*16d86563SAlexander Pyhalov */
_icv_close(_iconv_st * st)89*16d86563SAlexander Pyhalov void _icv_close(_iconv_st * st) {
90*16d86563SAlexander Pyhalov if (!st)
91*16d86563SAlexander Pyhalov errno = EBADF;
92*16d86563SAlexander Pyhalov else
93*16d86563SAlexander Pyhalov free(st);
94*16d86563SAlexander Pyhalov }
95*16d86563SAlexander Pyhalov
96*16d86563SAlexander Pyhalov /*
97*16d86563SAlexander Pyhalov * Actual conversion; called from iconv()
98*16d86563SAlexander Pyhalov */
99*16d86563SAlexander Pyhalov
_icv_iconv(_iconv_st * st,char ** inbuf,size_t * inbytesleft,char ** outbuf,size_t * outbytesleft)100*16d86563SAlexander Pyhalov size_t _icv_iconv(_iconv_st * st, char **inbuf, size_t *inbytesleft,
101*16d86563SAlexander Pyhalov char ** outbuf, size_t *outbytesleft) {
102*16d86563SAlexander Pyhalov int n;
103*16d86563SAlexander Pyhalov if (st == NULL) {
104*16d86563SAlexander Pyhalov errno = EBADF;
105*16d86563SAlexander Pyhalov return ((size_t) -1);
106*16d86563SAlexander Pyhalov }
107*16d86563SAlexander Pyhalov
108*16d86563SAlexander Pyhalov if (inbuf == NULL || *inbuf == NULL) { /* Reset request. */
109*16d86563SAlexander Pyhalov st->cstate = C0;
110*16d86563SAlexander Pyhalov st->_errno = 0;
111*16d86563SAlexander Pyhalov return ((size_t) 0);
112*16d86563SAlexander Pyhalov }
113*16d86563SAlexander Pyhalov
114*16d86563SAlexander Pyhalov errno = st->_errno = 0;
115*16d86563SAlexander Pyhalov
116*16d86563SAlexander Pyhalov while (*inbytesleft > 0 && *outbytesleft > 0) {
117*16d86563SAlexander Pyhalov switch (st->cstate) {
118*16d86563SAlexander Pyhalov case C0:
119*16d86563SAlexander Pyhalov if (**inbuf & MSB) { /* gb charactor */
120*16d86563SAlexander Pyhalov st->keepc[0] = (**inbuf);
121*16d86563SAlexander Pyhalov st->cstate = C1;
122*16d86563SAlexander Pyhalov } else { /* ASCII */
123*16d86563SAlexander Pyhalov **outbuf = **inbuf;
124*16d86563SAlexander Pyhalov (*outbuf)++;
125*16d86563SAlexander Pyhalov (*outbytesleft)--;
126*16d86563SAlexander Pyhalov }
127*16d86563SAlexander Pyhalov break;
128*16d86563SAlexander Pyhalov case C1: /* GBK charactor 2nd byte */
129*16d86563SAlexander Pyhalov if (gbk_2nd_byte(**inbuf) == 0) {
130*16d86563SAlexander Pyhalov st->keepc[1] = (**inbuf);
131*16d86563SAlexander Pyhalov n = gbk_to_hkscs(st->keepc, *outbuf, *outbytesleft);
132*16d86563SAlexander Pyhalov if (n > 0) {
133*16d86563SAlexander Pyhalov (*outbuf) += n;
134*16d86563SAlexander Pyhalov (*outbytesleft) -= n;
135*16d86563SAlexander Pyhalov
136*16d86563SAlexander Pyhalov st->cstate = C0;
137*16d86563SAlexander Pyhalov } else {
138*16d86563SAlexander Pyhalov st->_errno = errno = E2BIG;
139*16d86563SAlexander Pyhalov }
140*16d86563SAlexander Pyhalov } else if ( gbk4_2nd_byte((unsigned char)**inbuf) ) {
141*16d86563SAlexander Pyhalov st->keepc[1] = (**inbuf);
142*16d86563SAlexander Pyhalov st->cstate = C2;
143*16d86563SAlexander Pyhalov } else { /* illegal input, don't reset state */
144*16d86563SAlexander Pyhalov st->_errno = errno = EILSEQ;
145*16d86563SAlexander Pyhalov }
146*16d86563SAlexander Pyhalov break;
147*16d86563SAlexander Pyhalov case C2:
148*16d86563SAlexander Pyhalov if ( gbk4_3rd_byte((unsigned char)**inbuf) ) {
149*16d86563SAlexander Pyhalov st->keepc[2] = (**inbuf);
150*16d86563SAlexander Pyhalov st->cstate = C3;
151*16d86563SAlexander Pyhalov } else {
152*16d86563SAlexander Pyhalov /* illegal input, don't reset state */
153*16d86563SAlexander Pyhalov st->_errno = errno = EILSEQ;
154*16d86563SAlexander Pyhalov }
155*16d86563SAlexander Pyhalov break;
156*16d86563SAlexander Pyhalov case C3:
157*16d86563SAlexander Pyhalov if ( gbk4_4th_byte((unsigned char)**inbuf) ) {
158*16d86563SAlexander Pyhalov st->keepc[3] = (**inbuf);
159*16d86563SAlexander Pyhalov
160*16d86563SAlexander Pyhalov n = gbk4_to_hkscs(st->keepc, *outbuf, *outbytesleft);
161*16d86563SAlexander Pyhalov if (n > 0) {
162*16d86563SAlexander Pyhalov (*outbuf) += n;
163*16d86563SAlexander Pyhalov (*outbytesleft) -= n;
164*16d86563SAlexander Pyhalov
165*16d86563SAlexander Pyhalov st->cstate = C0;
166*16d86563SAlexander Pyhalov } else {
167*16d86563SAlexander Pyhalov st->_errno = errno = E2BIG;
168*16d86563SAlexander Pyhalov }
169*16d86563SAlexander Pyhalov
170*16d86563SAlexander Pyhalov } else {
171*16d86563SAlexander Pyhalov /* illegal input, don't reset state */
172*16d86563SAlexander Pyhalov st->_errno = errno = EILSEQ;
173*16d86563SAlexander Pyhalov }
174*16d86563SAlexander Pyhalov break;
175*16d86563SAlexander Pyhalov default: /* un-reachable */
176*16d86563SAlexander Pyhalov st->_errno = errno = EILSEQ;
177*16d86563SAlexander Pyhalov st->cstate = C0;
178*16d86563SAlexander Pyhalov break;
179*16d86563SAlexander Pyhalov }
180*16d86563SAlexander Pyhalov
181*16d86563SAlexander Pyhalov if (st->_errno) break;
182*16d86563SAlexander Pyhalov
183*16d86563SAlexander Pyhalov (*inbuf)++;
184*16d86563SAlexander Pyhalov (*inbytesleft)--;
185*16d86563SAlexander Pyhalov }
186*16d86563SAlexander Pyhalov
187*16d86563SAlexander Pyhalov if (errno) return ((size_t) -1);
188*16d86563SAlexander Pyhalov
189*16d86563SAlexander Pyhalov if ( *inbytesleft == 0 && st->cstate != C0 ) {
190*16d86563SAlexander Pyhalov errno = EINVAL;
191*16d86563SAlexander Pyhalov return ((size_t) -1);
192*16d86563SAlexander Pyhalov }
193*16d86563SAlexander Pyhalov
194*16d86563SAlexander Pyhalov if ( *inbytesleft > 0 && *outbytesleft == 0 ) {
195*16d86563SAlexander Pyhalov errno = E2BIG;
196*16d86563SAlexander Pyhalov return ((size_t) -1);
197*16d86563SAlexander Pyhalov }
198*16d86563SAlexander Pyhalov
199*16d86563SAlexander Pyhalov return (size_t)(*inbytesleft);
200*16d86563SAlexander Pyhalov }
201*16d86563SAlexander Pyhalov
202*16d86563SAlexander Pyhalov /*
203*16d86563SAlexander Pyhalov * Test whether inbuf is a valid character for
204*16d86563SAlexander Pyhalov * 2nd byte of GB2312 charactor:
205*16d86563SAlexander Pyhalov * Return: 0 --- valid GBK 2nd byte
206*16d86563SAlexander Pyhalov * 1 --- invalid GBK 2nd byte
207*16d86563SAlexander Pyhalov */
gbk_2nd_byte(inbuf)208*16d86563SAlexander Pyhalov int gbk_2nd_byte(inbuf)
209*16d86563SAlexander Pyhalov char inbuf;
210*16d86563SAlexander Pyhalov {
211*16d86563SAlexander Pyhalov
212*16d86563SAlexander Pyhalov unsigned int buf = (unsigned int) (inbuf & ONEBYTE);
213*16d86563SAlexander Pyhalov
214*16d86563SAlexander Pyhalov if ((buf >= 0x40) && (buf <= 0x7e))
215*16d86563SAlexander Pyhalov return 0;
216*16d86563SAlexander Pyhalov if ((buf >= 0x80) && (buf <= 0xfe))
217*16d86563SAlexander Pyhalov return 0;
218*16d86563SAlexander Pyhalov return 1;
219*16d86563SAlexander Pyhalov }
220*16d86563SAlexander Pyhalov
221*16d86563SAlexander Pyhalov /*
222*16d86563SAlexander Pyhalov * gbk_to_hkscs: Convert gbk charactor to hkscs.
223*16d86563SAlexander Pyhalov * Return: >0 --- converted with enough space in output buffer
224*16d86563SAlexander Pyhalov * =0 --- no space in outbuf
225*16d86563SAlexander Pyhalov */
226*16d86563SAlexander Pyhalov
gbk_to_hkscs(char keepc[],char * buf,size_t buflen)227*16d86563SAlexander Pyhalov int gbk_to_hkscs(char keepc[], char *buf, size_t buflen) {
228*16d86563SAlexander Pyhalov
229*16d86563SAlexander Pyhalov unsigned long gbk_val; /* GBK value */
230*16d86563SAlexander Pyhalov int index;
231*16d86563SAlexander Pyhalov unsigned long hkscs_val; /* hkscs value */
232*16d86563SAlexander Pyhalov
233*16d86563SAlexander Pyhalov if (buflen < 2) {
234*16d86563SAlexander Pyhalov errno = E2BIG;
235*16d86563SAlexander Pyhalov return 0;
236*16d86563SAlexander Pyhalov }
237*16d86563SAlexander Pyhalov
238*16d86563SAlexander Pyhalov gbk_val = ((keepc[0] & ONEBYTE) << 8) + (keepc[1] & ONEBYTE);
239*16d86563SAlexander Pyhalov index = binsearch(gbk_val, gbk_hkscs_tab, GBKMAX);
240*16d86563SAlexander Pyhalov if (index >= 0) {
241*16d86563SAlexander Pyhalov hkscs_val = gbk_hkscs_tab[index].value;
242*16d86563SAlexander Pyhalov *buf = (hkscs_val >> 8) & ONEBYTE;
243*16d86563SAlexander Pyhalov *(buf + 1) = hkscs_val & ONEBYTE;
244*16d86563SAlexander Pyhalov } else
245*16d86563SAlexander Pyhalov *buf = *(buf + 1) = (char)NON_ID_CHAR;
246*16d86563SAlexander Pyhalov return 2;
247*16d86563SAlexander Pyhalov }
248*16d86563SAlexander Pyhalov
gbk4_to_hkscs(char keepc[],char * buf,size_t buflen)249*16d86563SAlexander Pyhalov int gbk4_to_hkscs(char keepc[], char *buf, size_t buflen) {
250*16d86563SAlexander Pyhalov
251*16d86563SAlexander Pyhalov unsigned long gbk_val; /* GBK value */
252*16d86563SAlexander Pyhalov int index;
253*16d86563SAlexander Pyhalov unsigned long hkscs_val; /* hkscs value */
254*16d86563SAlexander Pyhalov
255*16d86563SAlexander Pyhalov if (buflen < 2) {
256*16d86563SAlexander Pyhalov errno = E2BIG;
257*16d86563SAlexander Pyhalov return 0;
258*16d86563SAlexander Pyhalov }
259*16d86563SAlexander Pyhalov
260*16d86563SAlexander Pyhalov gbk_val = ((keepc[0] & ONEBYTE) << 24) + ((keepc[1] & ONEBYTE) << 16) +
261*16d86563SAlexander Pyhalov ((keepc[2] & ONEBYTE) << 8 ) + (keepc[3] & ONEBYTE);
262*16d86563SAlexander Pyhalov index = binsearch(gbk_val, gbk4_hkscs_tab, GBK4MAX);
263*16d86563SAlexander Pyhalov if (index >= 0) {
264*16d86563SAlexander Pyhalov hkscs_val = gbk4_hkscs_tab[index].value;
265*16d86563SAlexander Pyhalov *buf = (hkscs_val >> 8) & ONEBYTE;
266*16d86563SAlexander Pyhalov *(buf + 1) = hkscs_val & ONEBYTE;
267*16d86563SAlexander Pyhalov } else
268*16d86563SAlexander Pyhalov *buf = *(buf + 1) = (char)NON_ID_CHAR;
269*16d86563SAlexander Pyhalov
270*16d86563SAlexander Pyhalov return 2;
271*16d86563SAlexander Pyhalov }
272*16d86563SAlexander Pyhalov
273*16d86563SAlexander Pyhalov /*
274*16d86563SAlexander Pyhalov * binsearch()
275*16d86563SAlexander Pyhalov */
binsearch(unsigned long x,table_t table[],int n)276*16d86563SAlexander Pyhalov int binsearch(unsigned long x, table_t table[], int n) {
277*16d86563SAlexander Pyhalov int low, high, mid;
278*16d86563SAlexander Pyhalov
279*16d86563SAlexander Pyhalov low = 0;
280*16d86563SAlexander Pyhalov high = n - 1;
281*16d86563SAlexander Pyhalov while (low <= high) {
282*16d86563SAlexander Pyhalov mid = (low + high) >> 1;
283*16d86563SAlexander Pyhalov if (x < table[mid].key)
284*16d86563SAlexander Pyhalov high = mid - 1;
285*16d86563SAlexander Pyhalov else if (x > table[mid].key)
286*16d86563SAlexander Pyhalov low = mid + 1;
287*16d86563SAlexander Pyhalov else
288*16d86563SAlexander Pyhalov return mid;
289*16d86563SAlexander Pyhalov }
290*16d86563SAlexander Pyhalov return -1;
291*16d86563SAlexander Pyhalov }
292*16d86563SAlexander Pyhalov
293*16d86563SAlexander Pyhalov #ifdef DEBUG
main(int argc,char * argv[])294*16d86563SAlexander Pyhalov main(int argc, char * argv[]) {
295*16d86563SAlexander Pyhalov _iconv_st * ist;
296*16d86563SAlexander Pyhalov char * inbuf = "�������е�ÿһ�������һ���Ѱ�װ��ע����������ʾ�� ��Ʒϵ�� ��";
297*16d86563SAlexander Pyhalov char * outbuf;
298*16d86563SAlexander Pyhalov char * ib, * oub;
299*16d86563SAlexander Pyhalov int inbyteleft;
300*16d86563SAlexander Pyhalov int outbyteleft;
301*16d86563SAlexander Pyhalov
302*16d86563SAlexander Pyhalov ist = (_iconv_st *) _icv_open();
303*16d86563SAlexander Pyhalov inbyteleft = outbyteleft = 2 * strlen(inbuf);
304*16d86563SAlexander Pyhalov outbuf = (char *)malloc(outbyteleft);
305*16d86563SAlexander Pyhalov ib = inbuf;
306*16d86563SAlexander Pyhalov oub = outbuf;
307*16d86563SAlexander Pyhalov _icv_iconv(ist, &inbuf, &inbyteleft, &outbuf, &outbyteleft);
308*16d86563SAlexander Pyhalov printf("IN -- %s\n", ib);
309*16d86563SAlexander Pyhalov printf("OUT -- %s\n", oub);
310*16d86563SAlexander Pyhalov }
311*16d86563SAlexander Pyhalov #endif
312