xref: /illumos-gate/usr/src/lib/libslp/clib/slp_utf8.c (revision 86d949f9497332fe19be6b5d711d265eb957439f)
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, Version 1.0 only
6  * (the "License").  You may not use this file except in compliance
7  * with the License.
8  *
9  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
10  * or http://www.opensolaris.org/os/licensing.
11  * See the License for the specific language governing permissions
12  * and limitations under the License.
13  *
14  * When distributing Covered Code, include this CDDL HEADER in each
15  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
16  * If applicable, add the following below this CDDL HEADER, with the
17  * fields enclosed by brackets "[]" replaced with your own identifying
18  * information: Portions Copyright [yyyy] [name of copyright owner]
19  *
20  * CDDL HEADER END
21  */
22 /*
23  * Copyright (c) 1999 by Sun Microsystems, Inc.
24  * All rights reserved.
25  */
26 
27 /*
28  * UTF-8 encoded Unicode parsing routines. For efficiency, we convert
29  * to wide chars only when absolutely needed. The following interfaces
30  * are exported to libslp:
31  *
32  * slp_utf_strchr:	same semantics as strchr, but handles UTF-8 strings
33  * slp_fold_space:	folds white space around and in between works;
34  *				handles UTF-8 strings
35  * slp_strcasecmp:	same semantics as strcasecmp, but also folds white
36  *				space and attempts locale-specific
37  *				case-insensitive comparisons.
38  */
39 
40 #include <stdio.h>
41 #include <string.h>
42 #include <widec.h>
43 #include <stdlib.h>
44 #include <syslog.h>
45 #include <slp-internal.h>
46 
47 /*
48  * Same semantics as strchr.
49  * Assumes that we start on a char boundry, and that c is a 7-bit
50  * ASCII char.
51  */
52 char *slp_utf_strchr(const char *s, char c) {
53 	int len;
54 	char *p;
55 
56 	for (p = (char *)s; *p; p += len) {
57 		len = mblen(p, MB_CUR_MAX);
58 		if (len == 1 && *p == c)
59 			return (p);
60 	}
61 	return (NULL);
62 }
63 
64 /*
65  * folds white space around and in between words.
66  * " aa    bb   " becomes "aa bb".
67  * returns NULL if it couldn't allocate memory. The caller must free
68  * the result when done.
69  */
70 static char *slp_fold_space(const char *s) {
71 	int len;
72 	char *folded, *f;
73 
74 	if (!(folded = malloc(strlen(s) + 1))) {
75 		slp_err(LOG_CRIT, 0, "slp_fold_space", "out of memory");
76 		return (NULL);
77 	}
78 
79 	f = folded;
80 	for (;;) {
81 		/* step 1: skip white space */
82 		for (; *s; s++) {
83 			len = mblen(s, MB_CUR_MAX);
84 			if (len != 1)
85 				break;
86 			if (!isspace(*s))
87 				break;
88 		}
89 
90 		if (!*s) {
91 			/* end of string */
92 			*f = 0;
93 			return (folded);
94 		}
95 		/* if we are in between words, keep one space */
96 		if (f != folded)
97 			*f++ = ' ';
98 
99 		/* step 2: copy into folded until we hit more white space */
100 		while (*s) {
101 			int i;
102 			len = mblen(s, MB_CUR_MAX);
103 			if (len == 1 && isspace(*s))
104 				break;
105 
106 			for (i = 0; i < len; i++)
107 				*f++ = *s++;
108 		}
109 		*f = *s;
110 		if (!*s++)
111 			return (folded);
112 	}
113 }
114 
115 /*
116  * performs like strcasecmp, but also folds white space before comparing,
117  * and will handle UTF-8 comparisons (including case). Note that the
118  * application's locale must have been set to a UTF-8 locale for this
119  * to work properly.
120  */
121 int slp_strcasecmp(const char *s1, const char *s2) {
122 	int diff = -1;
123 	char *p1, *p2;
124 	size_t wcslen1, wcslen2;
125 	wchar_t *wcs1, *wcs2;
126 
127 	p1 = p2 = NULL; wcs1 = wcs2 = NULL;
128 
129 	/* optimization: try simple case first */
130 	if (strcasecmp(s1, s2) == 0)
131 		return (0);
132 
133 	/* fold white space, and try again */
134 	p1 = slp_fold_space(s1);
135 	p2 = slp_fold_space(s2);
136 	if (!p1 || !p2)
137 		goto cleanup;
138 
139 	if ((diff = strcasecmp(p1, p2)) == 0)
140 		goto cleanup;
141 
142 	/*
143 	 * try converting to wide char -- we must be in a locale which
144 	 * supports the UTF8 codeset for this to work.
145 	 */
146 	if ((wcslen1 = mbstowcs(NULL, p1, 0)) == (size_t)-1)
147 		goto cleanup;
148 
149 	if (!(wcs1 = malloc(sizeof (*wcs1) * (wcslen1 + 1)))) {
150 		slp_err(LOG_CRIT, 0, "slp_strcasecmp", "out of memory");
151 		goto cleanup;
152 	}
153 
154 	if ((wcslen2 = mbstowcs(NULL, p2, 0)) == (size_t)-1)
155 		goto cleanup;
156 
157 	if (!(wcs2 = malloc(sizeof (*wcs2) * (wcslen2 + 1)))) {
158 		slp_err(LOG_CRIT, 0, "slp_strcasecmp", "out of memory");
159 		goto cleanup;
160 	}
161 	if (mbstowcs(wcs1, p1, wcslen1 + 1) == (size_t)-1)
162 		goto cleanup;
163 	if (mbstowcs(wcs2, p2, wcslen2 + 1) == (size_t)-1)
164 		goto cleanup;
165 
166 	diff = wscasecmp(wcs1, wcs2);
167 
168 cleanup:
169 	if (p1) free(p1);
170 	if (p2) free(p2);
171 	if (wcs1) free(wcs1);
172 	if (wcs2) free(wcs2);
173 	return (diff);
174 }
175