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 2007 by Sun Microsystems, Inc.  All rights reserved.
23*16d86563SAlexander Pyhalov  * Use is subject to license terms.
24*16d86563SAlexander Pyhalov  */
25*16d86563SAlexander Pyhalov 
26*16d86563SAlexander Pyhalov 
27*16d86563SAlexander Pyhalov #include <stdlib.h>
28*16d86563SAlexander Pyhalov #include <errno.h>
29*16d86563SAlexander Pyhalov #include <sys/types.h>
30*16d86563SAlexander Pyhalov #include <sys/isa_defs.h>
31*16d86563SAlexander Pyhalov #include "utf8_to_utf_ebcdic.h"
32*16d86563SAlexander Pyhalov 
33*16d86563SAlexander Pyhalov void *
_icv_open()34*16d86563SAlexander Pyhalov _icv_open()
35*16d86563SAlexander Pyhalov {
36*16d86563SAlexander Pyhalov 	return((void *)MAGIC_NUMBER);
37*16d86563SAlexander Pyhalov }
38*16d86563SAlexander Pyhalov 
39*16d86563SAlexander Pyhalov 
40*16d86563SAlexander Pyhalov void
_icv_close(int * cd)41*16d86563SAlexander Pyhalov _icv_close(int *cd)
42*16d86563SAlexander Pyhalov {
43*16d86563SAlexander Pyhalov 	if (! cd || cd != (int *)MAGIC_NUMBER)
44*16d86563SAlexander Pyhalov 		errno = EBADF;
45*16d86563SAlexander Pyhalov }
46*16d86563SAlexander Pyhalov 
47*16d86563SAlexander Pyhalov 
48*16d86563SAlexander Pyhalov #define	OUTBUF_SIZE_CHECK(sz) \
49*16d86563SAlexander Pyhalov 	if ((obtail - ob) < (sz)) { \
50*16d86563SAlexander Pyhalov 		ib = ib_org; \
51*16d86563SAlexander Pyhalov 		errno = E2BIG; \
52*16d86563SAlexander Pyhalov 		ret_val = (size_t)-1; \
53*16d86563SAlexander Pyhalov 		break; \
54*16d86563SAlexander Pyhalov 	}
55*16d86563SAlexander Pyhalov 
56*16d86563SAlexander Pyhalov #define I8_UTFEBICDIC(i8) i8_to_utf_ebcdic[(i8)]
57*16d86563SAlexander Pyhalov 
58*16d86563SAlexander Pyhalov size_t
_icv_iconv(int * cd,char ** inbuf,size_t * inbufleft,char ** outbuf,size_t * outbufleft)59*16d86563SAlexander Pyhalov _icv_iconv(int *cd, char **inbuf, size_t *inbufleft, char **outbuf,
60*16d86563SAlexander Pyhalov                 size_t *outbufleft)
61*16d86563SAlexander Pyhalov {
62*16d86563SAlexander Pyhalov 	size_t ret_val = 0;
63*16d86563SAlexander Pyhalov 	uchar_t *ib;
64*16d86563SAlexander Pyhalov 	uchar_t *ob;
65*16d86563SAlexander Pyhalov 	uchar_t *ibtail;
66*16d86563SAlexander Pyhalov 	uchar_t *obtail;
67*16d86563SAlexander Pyhalov 
68*16d86563SAlexander Pyhalov 	if (cd != (int *)MAGIC_NUMBER) {
69*16d86563SAlexander Pyhalov 		errno = EBADF;
70*16d86563SAlexander Pyhalov 		return((size_t)-1);
71*16d86563SAlexander Pyhalov 	}
72*16d86563SAlexander Pyhalov 
73*16d86563SAlexander Pyhalov 	if (!inbuf || !(*inbuf))
74*16d86563SAlexander Pyhalov 		return((size_t)0);
75*16d86563SAlexander Pyhalov 
76*16d86563SAlexander Pyhalov 	ib = (uchar_t *)*inbuf;
77*16d86563SAlexander Pyhalov 	ob = (uchar_t *)*outbuf;
78*16d86563SAlexander Pyhalov 	ibtail = ib + *inbufleft;
79*16d86563SAlexander Pyhalov 	obtail = ob + *outbufleft;
80*16d86563SAlexander Pyhalov 
81*16d86563SAlexander Pyhalov 	while (ib < ibtail) {
82*16d86563SAlexander Pyhalov 		uchar_t *ib_org;
83*16d86563SAlexander Pyhalov 		uint_t u4;
84*16d86563SAlexander Pyhalov 		uint_t first_byte;
85*16d86563SAlexander Pyhalov 		signed char sz;
86*16d86563SAlexander Pyhalov 
87*16d86563SAlexander Pyhalov 		sz = number_of_bytes_in_utf8_char[*ib];
88*16d86563SAlexander Pyhalov 		if (sz == ICV_TYPE_ILLEGAL_CHAR) {
89*16d86563SAlexander Pyhalov 			errno = EILSEQ;
90*16d86563SAlexander Pyhalov 			ret_val = (size_t)-1;
91*16d86563SAlexander Pyhalov 			break;
92*16d86563SAlexander Pyhalov 		}
93*16d86563SAlexander Pyhalov 
94*16d86563SAlexander Pyhalov 		if ((ibtail - ib) < sz) {
95*16d86563SAlexander Pyhalov 			errno = EINVAL;
96*16d86563SAlexander Pyhalov 			ret_val = (size_t)-1;
97*16d86563SAlexander Pyhalov 			break;
98*16d86563SAlexander Pyhalov 		}
99*16d86563SAlexander Pyhalov 
100*16d86563SAlexander Pyhalov 		ib_org = ib;
101*16d86563SAlexander Pyhalov 		first_byte = *ib;
102*16d86563SAlexander Pyhalov 		u4 = (uint_t)(*ib++ & masks_tbl[sz]);
103*16d86563SAlexander Pyhalov 		for (; sz > 1; sz--) {
104*16d86563SAlexander Pyhalov 			if (first_byte) {
105*16d86563SAlexander Pyhalov 				if (((uchar_t)*ib) <
106*16d86563SAlexander Pyhalov 					valid_min_2nd_byte[first_byte] ||
107*16d86563SAlexander Pyhalov 				    ((uchar_t)*ib) >
108*16d86563SAlexander Pyhalov 					valid_max_2nd_byte[first_byte]) {
109*16d86563SAlexander Pyhalov 					ib = ib_org;
110*16d86563SAlexander Pyhalov 					errno = EILSEQ;
111*16d86563SAlexander Pyhalov 					ret_val = (size_t)-1;
112*16d86563SAlexander Pyhalov 					goto ILLEGAL_CHAR_ERR;
113*16d86563SAlexander Pyhalov 				}
114*16d86563SAlexander Pyhalov 				first_byte = 0;
115*16d86563SAlexander Pyhalov 			} else if (((uint_t)*ib) < 0x80 ||
116*16d86563SAlexander Pyhalov 				   ((uint_t)*ib) > 0xbf) {
117*16d86563SAlexander Pyhalov 				ib = ib_org;
118*16d86563SAlexander Pyhalov 				errno = EILSEQ;
119*16d86563SAlexander Pyhalov 				ret_val = (size_t)-1;
120*16d86563SAlexander Pyhalov 				goto ILLEGAL_CHAR_ERR;
121*16d86563SAlexander Pyhalov 			}
122*16d86563SAlexander Pyhalov 			u4 = (u4 << ICV_UTF8_BIT_SHIFT) |
123*16d86563SAlexander Pyhalov 				(((uint_t)*ib) & ICV_UTF8_BIT_MASK);
124*16d86563SAlexander Pyhalov 			ib++;
125*16d86563SAlexander Pyhalov 		}
126*16d86563SAlexander Pyhalov 
127*16d86563SAlexander Pyhalov 		/* Check against known non-characters. */
128*16d86563SAlexander Pyhalov 		if ((u4 & ICV_UTF32_NONCHAR_mask) == ICV_UTF32_NONCHAR_fffe ||
129*16d86563SAlexander Pyhalov 		    (u4 & ICV_UTF32_NONCHAR_mask) == ICV_UTF32_NONCHAR_ffff ||
130*16d86563SAlexander Pyhalov 		    u4 > ICV_UTF32_LAST_VALID_CHAR ||
131*16d86563SAlexander Pyhalov 		    (u4 >= ICV_UTF32_SURROGATE_START_d800 &&
132*16d86563SAlexander Pyhalov 		    u4 <= ICV_UTF32_SURROGATE_END_dfff) ||
133*16d86563SAlexander Pyhalov 		    (u4 >= ICV_UTF32_ARABIC_NONCHAR_START_fdd0 &&
134*16d86563SAlexander Pyhalov 		    u4 <= ICV_UTF32_ARABIC_NONCHAR_END_fdef)) {
135*16d86563SAlexander Pyhalov 			ib = ib_org;
136*16d86563SAlexander Pyhalov 			errno = EILSEQ;
137*16d86563SAlexander Pyhalov 			ret_val = (size_t)-1;
138*16d86563SAlexander Pyhalov 			goto ILLEGAL_CHAR_ERR;
139*16d86563SAlexander Pyhalov 		}
140*16d86563SAlexander Pyhalov 
141*16d86563SAlexander Pyhalov 		if (u4 <= 0x7f) {
142*16d86563SAlexander Pyhalov 			OUTBUF_SIZE_CHECK(1);
143*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(u4);
144*16d86563SAlexander Pyhalov 		} else if (u4 <= 0x9f) {
145*16d86563SAlexander Pyhalov 			OUTBUF_SIZE_CHECK(1);
146*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(u4);
147*16d86563SAlexander Pyhalov 		} else if (u4 <= 0x3ff) {
148*16d86563SAlexander Pyhalov 			OUTBUF_SIZE_CHECK(2);
149*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xc0 | ((u4 & 0x03e0) >> 5));
150*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 |  (u4 & 0x001f));
151*16d86563SAlexander Pyhalov 		} else if (u4 <= 0x3fff) {
152*16d86563SAlexander Pyhalov 			OUTBUF_SIZE_CHECK(3);
153*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xe0 | ((u4 & 0x3c00) >> 10));
154*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 | ((u4 & 0x03e0) >> 5));
155*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 |  (u4 & 0x001f));
156*16d86563SAlexander Pyhalov 		} else if (u4 <= 0x3ffff) {
157*16d86563SAlexander Pyhalov 			OUTBUF_SIZE_CHECK(4);
158*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xf0 | ((u4 & 0x38000) >> 15));
159*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 | ((u4 & 0x07c00) >> 10));
160*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 | ((u4 & 0x003e0) >> 5));
161*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 |  (u4 & 0x0001f));
162*16d86563SAlexander Pyhalov 		} else if (u4 <= 0x3fffff) {
163*16d86563SAlexander Pyhalov 			OUTBUF_SIZE_CHECK(5);
164*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xf8 | ((u4 & 0x300000) >> 20));
165*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 | ((u4 & 0x0f8000) >> 15));
166*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 | ((u4 & 0x007c00) >> 10));
167*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 | ((u4 & 0x0003e0) >> 5));
168*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 |  (u4 & 0x00001f));
169*16d86563SAlexander Pyhalov 		} else if (u4 <= 0x3ffffff) {
170*16d86563SAlexander Pyhalov 			OUTBUF_SIZE_CHECK(6);
171*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xfc | ((u4 & 0x2000000) >> 25));
172*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 | ((u4 & 0x1f00000) >> 20));
173*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 | ((u4 & 0x00f8000) >> 15));
174*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 | ((u4 & 0x0007c00) >> 10));
175*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 | ((u4 & 0x00003e0) >> 5));
176*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 |  (u4 & 0x000001f));
177*16d86563SAlexander Pyhalov 		} else if (u4 <= 0x7fffffff) {
178*16d86563SAlexander Pyhalov 			OUTBUF_SIZE_CHECK(7);
179*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xfe | ((u4 & 0x40000000) >> 30));
180*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 | ((u4 & 0x3e000000) >> 25));
181*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 | ((u4 & 0x01f00000) >> 20));
182*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 | ((u4 & 0x000f8000) >> 15));
183*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 | ((u4 & 0x00007c00) >> 10));
184*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 | ((u4 & 0x000003e0) >> 5));
185*16d86563SAlexander Pyhalov 			*ob++ = I8_UTFEBICDIC(0xa0 |  (u4 & 0x0000001f));
186*16d86563SAlexander Pyhalov 		} else {
187*16d86563SAlexander Pyhalov 			ib = ib_org;
188*16d86563SAlexander Pyhalov 			errno = EILSEQ;
189*16d86563SAlexander Pyhalov 			ret_val = (size_t)-1;
190*16d86563SAlexander Pyhalov 			break;
191*16d86563SAlexander Pyhalov 		}
192*16d86563SAlexander Pyhalov 	}
193*16d86563SAlexander Pyhalov 
194*16d86563SAlexander Pyhalov ILLEGAL_CHAR_ERR:
195*16d86563SAlexander Pyhalov 	*inbuf = (char *)ib;
196*16d86563SAlexander Pyhalov 	*inbufleft = ibtail - ib;
197*16d86563SAlexander Pyhalov 	*outbuf = (char *)ob;
198*16d86563SAlexander Pyhalov 	*outbufleft = obtail - ob;
199*16d86563SAlexander Pyhalov 
200*16d86563SAlexander Pyhalov 	return(ret_val);
201*16d86563SAlexander Pyhalov }
202