xref: /illumos-gate/usr/src/lib/iconv_modules/ko/common/uhang_to_utf_main.c (revision 16d8656330ae5622ec32e5007f62145ebafdc50f)
1 /*
2  * CDDL HEADER START
3  *
4  * The contents of this file are subject to the terms of the
5  * Common Development and Distribution License (the "License").
6  * You may not use this file except in compliance with the License.
7  *
8  * You can obtain a copy of the license at src/OPENSOLARIS.LICENSE
9  * or http://www.opensolaris.org/os/licensing.
10  * See the License for the specific language governing permissions
11  * and limitations under the License.
12  *
13  * When distributing Covered Code, include this CDDL HEADER in each
14  * file and include the License file at src/OPENSOLARIS.LICENSE.
15  * If applicable, add the following below this CDDL HEADER, with the
16  * fields enclosed by brackets "[]" replaced with your own identifying
17  * information: Portions Copyright [yyyy] [name of copyright owner]
18  *
19  * CDDL HEADER END
20  */
21 /*
22  * Copyright (c) 1999 by Sun Microsystems, Inc.
23  */
24 
25 
26 #include <errno.h>
27 #include <widec.h>
28 #include "common_def.h"
29 #include "common_han.h"
30 #include "uhang_utf_api.h"
31 #include "euc_utf_api.h"
32 
33 /****  _ I C V _ O P E N  ****/
34 
_icv_open()35 void* _icv_open()
36 {
37         return((void*)MAGIC_NUMBER);
38 }  /* end of int _icv_open(). */
39 
40 
41 /****  _ I C V _ C L O S E  ****/
42 
_icv_close(int * cd)43 void _icv_close(int* cd)
44 {
45         if (!cd || cd != (int*)MAGIC_NUMBER)
46                 errno = EBADF;
47 }  /* end of void _icv_close(int*). */
48 
49 
50 /****  _ I C V _ I C O N V  ****/
51 
_icv_iconv(int * cd,char ** inbuf,size_t * inbufleft,char ** outbuf,size_t * outbufleft)52 size_t _icv_iconv(int* cd, char** inbuf, size_t* inbufleft,
53 			char** outbuf, size_t* outbufleft)
54 {
55 	size_t		ret_val = 0;
56 	unsigned char*	ib;
57 	unsigned char*	ob;
58 	unsigned char*	ibtail;
59 	unsigned char*	obtail;
60 
61 	if (!cd || cd != (int*)MAGIC_NUMBER)
62 	{
63 		errno = EBADF;
64 		return((size_t)-1);
65 	}
66 
67 	if (!inbuf || !(*inbuf))
68 		return((size_t)0);
69 
70 	ib = (unsigned char*)*inbuf;
71 	ob = (unsigned char*)*outbuf;
72 	ibtail = ib + *inbufleft;
73 	obtail = ob + *outbufleft;
74 
75 	while (ib < ibtail)
76 	{
77 		if (*ib & 0x80)  /* Korean EUC doesn't have CS2 or CS3. */
78 		{
79 			hcode_type euc_code, utf_code;
80 			int flag;
81 
82 			flag = 0;
83 
84 			if ((ibtail - ib) < 2)
85 			{
86 				errno = EINVAL;
87 				ret_val = (size_t)-1;
88 				break;
89 			}
90 
91 
92 			if(*ib<0xA1)
93 			{
94 				if((*(ib+1)>0x40 && *(ib+1)<0x5B) || (*(ib+1)>0x60 && *(ib+1)<0x7B) || (*(ib+1)>0x80 && *(ib+1)<0xFF))
95 					flag = 0;
96 				else
97 					flag = 1;
98 
99 			}
100 			else
101 			{
102 				if(*ib<0xC7)
103 				{
104 					if((*(ib+1)>0x40 && *(ib+1)<0x5B) || (*(ib+1)>0x60 && *(ib+1)<0x7B) || (*(ib+1)>0x80 && *(ib+1)<0xFF))
105 						flag = 0;
106 					else
107 						flag = 1;
108 				}
109 				else
110 				{
111 					if(*(ib+1)>0xA0 && *(ib+1)<0xFF)
112 						flag = 0;
113 					else
114 						flag = 1;
115 				}
116 
117 			}
118 
119 			if(flag)
120 			{
121 				errno = EILSEQ;
122 				ret_val = (size_t)-1;
123 				break;
124 			}
125 
126 			euc_code.code = 0;
127 			euc_code.byte.byte3 = *ib;
128 			euc_code.byte.byte4 = *(ib + 1);
129 			utf_code = _unified_hangul_to_utf8(euc_code);
130 
131 			if (utf_code.code != 0)
132 			{
133 				if ((obtail - ob) < 3)
134 				{
135 					errno = E2BIG;
136 					ret_val = (size_t)-1;
137 					break;
138 				}
139 				/* UTF8 code from 2 bytes is always 3 bytes */
140 				if(utf_code.byte.byte2)
141 					*ob++ = (char)utf_code.byte.byte2;
142 				*ob++ = (char)utf_code.byte.byte3;
143 				*ob++ = (char)utf_code.byte.byte4;
144 			}
145 			else  /* FAILED - this means input char doesn't belong to
146 			       *	  input codeset. */
147 			{
148 				errno = EILSEQ;
149 				ret_val = (size_t)-1;
150 				break;
151 			}
152 			ib += 2;
153 
154 		}
155 		else  /* CS0 */
156 		{
157 			if (ob >= obtail)
158 			{
159 				errno = E2BIG;
160 				ret_val = (size_t)-1;
161 				break;
162 			}
163 			*ob++ = *ib++;
164 		}
165 	}
166 
167 	*inbuf = (char*)ib;
168 	*inbufleft = ibtail - ib;
169 	*outbuf = (char*)ob;
170 	*outbufleft = obtail - ob;
171 
172 	return(ret_val);
173 }  /* end of size_t _icv_iconv(int*, char**, size_t*, char**, size_t*). */
174