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 /*
23*91e1e26aSAlexander Pyhalov * Copyright 2002 Sun Microsystems, Inc. All rights reserved.
24*91e1e26aSAlexander Pyhalov * Use is subject to license terms.
25*91e1e26aSAlexander Pyhalov */
26*91e1e26aSAlexander Pyhalov
27*91e1e26aSAlexander Pyhalov #include <sys/types.h>
28*91e1e26aSAlexander Pyhalov #include "../inc/common_defs.h"
29*91e1e26aSAlexander Pyhalov
30*91e1e26aSAlexander Pyhalov /*
31*91e1e26aSAlexander Pyhalov * convert utf8 string to unicode
32*91e1e26aSAlexander Pyhalov * return value: 0 - fail
33*91e1e26aSAlexander Pyhalov * 1 - success
34*91e1e26aSAlexander Pyhalov */
35*91e1e26aSAlexander Pyhalov int
convert_utf8_to_ucs4(uchar_t * ib,int utf8_len,uint_t * unicode)36*91e1e26aSAlexander Pyhalov convert_utf8_to_ucs4(uchar_t *ib, int utf8_len, uint_t *unicode)
37*91e1e26aSAlexander Pyhalov {
38*91e1e26aSAlexander Pyhalov uchar_t first_byte = *ib;
39*91e1e26aSAlexander Pyhalov uint_t u4;
40*91e1e26aSAlexander Pyhalov
41*91e1e26aSAlexander Pyhalov if ( number_of_bytes_in_utf8_char[first_byte] != utf8_len) return 0;
42*91e1e26aSAlexander Pyhalov
43*91e1e26aSAlexander Pyhalov u4 = (uint_t)(*ib++ & masks_tbl[utf8_len]);
44*91e1e26aSAlexander Pyhalov for (; utf8_len > 1; utf8_len--)
45*91e1e26aSAlexander Pyhalov {
46*91e1e26aSAlexander Pyhalov u4 = (u4 << ICV_UTF8_BIT_SHIFT) | (((uint_t) *ib) & ICV_UTF8_BIT_MASK);
47*91e1e26aSAlexander Pyhalov ++ib;
48*91e1e26aSAlexander Pyhalov }
49*91e1e26aSAlexander Pyhalov
50*91e1e26aSAlexander Pyhalov *unicode = u4;
51*91e1e26aSAlexander Pyhalov
52*91e1e26aSAlexander Pyhalov return 1;
53*91e1e26aSAlexander Pyhalov }
54*91e1e26aSAlexander Pyhalov
55*91e1e26aSAlexander Pyhalov /*
56*91e1e26aSAlexander Pyhalov * check whether the input 'str' is valid UTF-8 byte sequence or not,
57*91e1e26aSAlexander Pyhalov * which lenght is specified by 'utf8_len'
58*91e1e26aSAlexander Pyhalov *
59*91e1e26aSAlexander Pyhalov * return: 0 - invalid byte sequence
60*91e1e26aSAlexander Pyhalov * 1 - valid byte sequence
61*91e1e26aSAlexander Pyhalov */
62*91e1e26aSAlexander Pyhalov int
is_valid_utf8_string(uchar_t * str,int utf8_len)63*91e1e26aSAlexander Pyhalov is_valid_utf8_string(uchar_t *str, int utf8_len)
64*91e1e26aSAlexander Pyhalov {
65*91e1e26aSAlexander Pyhalov uint_t unicode = 0;
66*91e1e26aSAlexander Pyhalov uchar_t *ib = str;
67*91e1e26aSAlexander Pyhalov uchar_t first_byte;
68*91e1e26aSAlexander Pyhalov int is_second_byte = 0, len=utf8_len;
69*91e1e26aSAlexander Pyhalov
70*91e1e26aSAlexander Pyhalov if (number_of_bytes_in_utf8_char[*ib] == ICV_TYPE_ILLEGAL_CHAR ||
71*91e1e26aSAlexander Pyhalov number_of_bytes_in_utf8_char[*ib] != utf8_len ) return 0;
72*91e1e26aSAlexander Pyhalov
73*91e1e26aSAlexander Pyhalov first_byte = *ib;
74*91e1e26aSAlexander Pyhalov --utf8_len;
75*91e1e26aSAlexander Pyhalov ++ib;
76*91e1e26aSAlexander Pyhalov is_second_byte = 1;
77*91e1e26aSAlexander Pyhalov
78*91e1e26aSAlexander Pyhalov while (utf8_len != 0)
79*91e1e26aSAlexander Pyhalov {
80*91e1e26aSAlexander Pyhalov if (is_second_byte)
81*91e1e26aSAlexander Pyhalov {
82*91e1e26aSAlexander Pyhalov if ( *ib < valid_min_2nd_byte[first_byte] || *ib > valid_max_2nd_byte[first_byte] )
83*91e1e26aSAlexander Pyhalov return 0;
84*91e1e26aSAlexander Pyhalov is_second_byte = 0;
85*91e1e26aSAlexander Pyhalov }
86*91e1e26aSAlexander Pyhalov else if ((*ib & 0xc0) != 0x80) /* 0x80 -- 0xbf */
87*91e1e26aSAlexander Pyhalov return 0;
88*91e1e26aSAlexander Pyhalov
89*91e1e26aSAlexander Pyhalov --utf8_len;
90*91e1e26aSAlexander Pyhalov ++ib;
91*91e1e26aSAlexander Pyhalov }
92*91e1e26aSAlexander Pyhalov
93*91e1e26aSAlexander Pyhalov convert_utf8_to_ucs4(str, len, &unicode);
94*91e1e26aSAlexander Pyhalov if (unicode == 0xFFFE || unicode == 0xFFFF) return 0;
95*91e1e26aSAlexander Pyhalov
96*91e1e26aSAlexander Pyhalov return 1;
97*91e1e26aSAlexander Pyhalov }
98