xref: /titanic_50/usr/src/lib/iconv_modules/common/common_utf8.c (revision 880d797826457b77414b37d531cc3e1aa166ecbe)
1*880d7978SAlexander Pyhalov /*
2*880d7978SAlexander Pyhalov  * CDDL HEADER START
3*880d7978SAlexander Pyhalov  *
4*880d7978SAlexander Pyhalov  * The contents of this file are subject to the terms of the
5*880d7978SAlexander Pyhalov  * Common Development and Distribution License (the "License").
6*880d7978SAlexander Pyhalov  * You may not use this file except in compliance with the License.
7*880d7978SAlexander Pyhalov  *
8*880d7978SAlexander Pyhalov  * You can obtain a copy of the license at src/OPENSOLARIS.LICENSE
9*880d7978SAlexander Pyhalov  * or http://www.opensolaris.org/os/licensing.
10*880d7978SAlexander Pyhalov  * See the License for the specific language governing permissions
11*880d7978SAlexander Pyhalov  * and limitations under the License.
12*880d7978SAlexander Pyhalov  *
13*880d7978SAlexander Pyhalov  * When distributing Covered Code, include this CDDL HEADER in each
14*880d7978SAlexander Pyhalov  * file and include the License file at src/OPENSOLARIS.LICENSE.
15*880d7978SAlexander Pyhalov  * If applicable, add the following below this CDDL HEADER, with the
16*880d7978SAlexander Pyhalov  * fields enclosed by brackets "[]" replaced with your own identifying
17*880d7978SAlexander Pyhalov  * information: Portions Copyright [yyyy] [name of copyright owner]
18*880d7978SAlexander Pyhalov  *
19*880d7978SAlexander Pyhalov  * CDDL HEADER END
20*880d7978SAlexander Pyhalov  */
21*880d7978SAlexander Pyhalov 
22*880d7978SAlexander Pyhalov /*
23*880d7978SAlexander Pyhalov  * Copyright 2002 Sun Microsystems, Inc.  All rights reserved.
24*880d7978SAlexander Pyhalov  * Use is subject to license terms.
25*880d7978SAlexander Pyhalov  */
26*880d7978SAlexander Pyhalov 
27*880d7978SAlexander Pyhalov #include <sys/types.h>
28*880d7978SAlexander Pyhalov #include "../inc/common_defs.h"
29*880d7978SAlexander Pyhalov 
30*880d7978SAlexander Pyhalov /*
31*880d7978SAlexander Pyhalov  * convert utf8 string to unicode
32*880d7978SAlexander Pyhalov  * return value: 0 - fail
33*880d7978SAlexander Pyhalov  *               1 - success
34*880d7978SAlexander Pyhalov  */
35*880d7978SAlexander Pyhalov int
convert_utf8_to_ucs4(uchar_t * ib,int utf8_len,uint_t * unicode)36*880d7978SAlexander Pyhalov convert_utf8_to_ucs4(uchar_t *ib, int utf8_len, uint_t *unicode)
37*880d7978SAlexander Pyhalov {
38*880d7978SAlexander Pyhalov    uchar_t first_byte = *ib;
39*880d7978SAlexander Pyhalov    uint_t u4;
40*880d7978SAlexander Pyhalov 
41*880d7978SAlexander Pyhalov    if ( number_of_bytes_in_utf8_char[first_byte] != utf8_len) return 0;
42*880d7978SAlexander Pyhalov 
43*880d7978SAlexander Pyhalov    u4 = (uint_t)(*ib++ & masks_tbl[utf8_len]);
44*880d7978SAlexander Pyhalov    for (; utf8_len > 1; utf8_len--)
45*880d7978SAlexander Pyhalov      {
46*880d7978SAlexander Pyhalov 	u4 = (u4 << ICV_UTF8_BIT_SHIFT) | (((uint_t) *ib) & ICV_UTF8_BIT_MASK);
47*880d7978SAlexander Pyhalov 	++ib;
48*880d7978SAlexander Pyhalov      }
49*880d7978SAlexander Pyhalov 
50*880d7978SAlexander Pyhalov    *unicode = u4;
51*880d7978SAlexander Pyhalov 
52*880d7978SAlexander Pyhalov    return 1;
53*880d7978SAlexander Pyhalov }
54*880d7978SAlexander Pyhalov 
55*880d7978SAlexander Pyhalov /*
56*880d7978SAlexander Pyhalov  * check whether the input 'str' is valid UTF-8 byte sequence or not,
57*880d7978SAlexander Pyhalov  * which lenght is specified by 'utf8_len'
58*880d7978SAlexander Pyhalov  *
59*880d7978SAlexander Pyhalov  * return: 0 - invalid byte sequence
60*880d7978SAlexander Pyhalov  * 	   1 - valid byte sequence
61*880d7978SAlexander Pyhalov  */
62*880d7978SAlexander Pyhalov int
is_valid_utf8_string(uchar_t * str,int utf8_len)63*880d7978SAlexander Pyhalov is_valid_utf8_string(uchar_t *str, int utf8_len)
64*880d7978SAlexander Pyhalov {
65*880d7978SAlexander Pyhalov    uint_t unicode = 0;
66*880d7978SAlexander Pyhalov    uchar_t *ib = str;
67*880d7978SAlexander Pyhalov    uchar_t first_byte;
68*880d7978SAlexander Pyhalov    int is_second_byte = 0, len=utf8_len;
69*880d7978SAlexander Pyhalov 
70*880d7978SAlexander Pyhalov    if (number_of_bytes_in_utf8_char[*ib] == ICV_TYPE_ILLEGAL_CHAR ||
71*880d7978SAlexander Pyhalov        number_of_bytes_in_utf8_char[*ib] != utf8_len ) return 0;
72*880d7978SAlexander Pyhalov 
73*880d7978SAlexander Pyhalov    first_byte = *ib;
74*880d7978SAlexander Pyhalov    --utf8_len;
75*880d7978SAlexander Pyhalov    ++ib;
76*880d7978SAlexander Pyhalov    is_second_byte = 1;
77*880d7978SAlexander Pyhalov 
78*880d7978SAlexander Pyhalov    while (utf8_len != 0)
79*880d7978SAlexander Pyhalov      {
80*880d7978SAlexander Pyhalov 	if (is_second_byte)
81*880d7978SAlexander Pyhalov 	  {
82*880d7978SAlexander Pyhalov 	     if ( *ib < valid_min_2nd_byte[first_byte] || *ib > valid_max_2nd_byte[first_byte] )
83*880d7978SAlexander Pyhalov 	        return 0;
84*880d7978SAlexander Pyhalov 	     is_second_byte = 0;
85*880d7978SAlexander Pyhalov 	  }
86*880d7978SAlexander Pyhalov 	else if ((*ib & 0xc0) != 0x80) /* 0x80 -- 0xbf */
87*880d7978SAlexander Pyhalov 		return 0;
88*880d7978SAlexander Pyhalov 
89*880d7978SAlexander Pyhalov         --utf8_len;
90*880d7978SAlexander Pyhalov 	++ib;
91*880d7978SAlexander Pyhalov      }
92*880d7978SAlexander Pyhalov 
93*880d7978SAlexander Pyhalov    convert_utf8_to_ucs4(str, len, &unicode);
94*880d7978SAlexander Pyhalov    if (unicode == 0xFFFE || unicode == 0xFFFF) return 0;
95*880d7978SAlexander Pyhalov 
96*880d7978SAlexander Pyhalov    return 1;
97*880d7978SAlexander Pyhalov }
98