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) 1996 by Sun Microsystems, Inc.
23  */
24 
25 
26 #include <errno.h>
27 #include <widec.h>
28 #include "common_def.h"
29 #include "common_thai.h"
30 #include "utf_euc_api.h"
31 
32 #include "common_defs.h"
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))		/* 7 bits */
78 		{
79 			if (ob >= obtail)
80 			{
81 				errno = E2BIG;
82 				ret_val = (size_t)-1;
83 				break;
84 			}
85 			*ob++ = *ib++;
86 		}
87 		else if ((*ib & 0xF0) == 0xE0)	/* 16 bits */
88 		{
89 			hcode_type utf8_code, euc_code;
90 
91 			if ((ibtail - ib) < 3)
92 			{
93 				errno = EINVAL;
94 				ret_val = (size_t)-1;
95 				break;
96 			}
97 
98             /* UNICODE 3.2 :eliminate non-secure UTF-8 sequence  */
99 			if (*(ib+1)<valid_min_2nd_byte[*ib] || *(ib+1)>valid_max_2nd_byte[*ib])
100 			{
101 				errno = EILSEQ;
102 				ret_val = (size_t)-1;
103 				break;
104 			}
105 			else if ((*(ib+2)&0xC0)^0x80)
106 			{
107 				errno = EILSEQ;
108 			        ret_val = (size_t)-1;
109 				break;
110 			}
111 			else if (*ib==0xEF && *(ib+1)==0xBF && (*(ib+2)==0xBF || *(ib+2)==0xBE))
112 			{ /*U+FFFE , U+FFFF*/
113 				errno = EILSEQ;
114 			        ret_val = (size_t)-1;
115 				break;
116 			}
117 
118 			utf8_code.byte.byte1 = 0;
119 			utf8_code.byte.byte2 = *ib;
120 			utf8_code.byte.byte3 = *(ib + 1);
121 			utf8_code.byte.byte4 = *(ib + 2);
122 
123 			euc_code = _utf8_to_eucTH(utf8_code);
124 
125 			if (euc_code.code != 0) {
126 				/* If find something -> EUC code */
127 				*ob++ = euc_code.byte.byte4;
128 			}
129 			else
130 			{
131 				/* Let's assume the code is non-identical. */
132 				if ((obtail - ob) < 1)
133 				{
134 					errno = E2BIG;
135 					ret_val = (size_t)-1;
136 					break;
137 				}
138 				*ob++ = NON_IDENTICAL;
139 				ret_val += 1;
140 			}
141 			ib += 3;
142 		}
143 		else  /* 11, 21, 26 & 31 bits codes won't be able to convert. */
144 		{
145 			short int offset;
146 
147 			errno = 0;
148 			offset = number_of_bytes_in_utf8_char[*ib];
149 			if (offset == ICV_TYPE_ILLEGAL_CHAR )
150 			{
151 				errno = EILSEQ;
152 			}
153 			else if (ibtail-ib < offset)
154 			{
155 				errno = EINVAL;
156 			}
157 			else if (*(ib+1)<valid_min_2nd_byte[*ib] || *(ib+1)>valid_max_2nd_byte[*ib])
158 			{
159 				errno = EILSEQ;
160 			}
161 			else /* check the remaining byte [0x80, 0xBF] */
162 			{
163 				int i;
164 				for(i=2;i<offset;i++)
165 				{
166 					if ((*(ib+i) & 0xC0) ^ 0x80)
167 					{
168 						errno = EILSEQ;
169 						break;
170 					}
171 				}
172 			}
173 			if(errno)
174 			{
175 				ret_val = (size_t)-1;
176 				break;
177 			}
178 
179 			ib += offset;
180 
181 			/* Let's assume the code is non-identical. */
182 			offset = (offset > 2) ? 2 : 1;
183 			if ((obtail - ob) < offset)
184 			{
185 				errno = E2BIG;
186 				ret_val = (size_t)-1;
187 				break;
188 			}
189 			*ob++ = NON_IDENTICAL;
190 			if (offset > 1)
191 				*ob++ = NON_IDENTICAL;
192 			ret_val += offset;
193 		}
194 	}
195 
196 	*inbuf = (char*)ib;
197 	*inbufleft = ibtail - ib;
198 	*outbuf = (char*)ob;
199 	*outbufleft = obtail - ob;
200 
201 	return(ret_val);
202 }  /* end of size_t _icv_iconv(int*, char**, size_t*, char**, size_t*).*/
203