1/* $OpenLDAP$ */
2/* This work is part of OpenLDAP Software <http://www.openldap.org/>.
3 *
4 * Copyright 1998-2011 The OpenLDAP Foundation.
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted only as authorized by the OpenLDAP
9 * Public License.
10 *
11 * A copy of this license is available in file LICENSE in the
12 * top-level directory of the distribution or, alternatively, at
13 * <http://www.OpenLDAP.org/license.html>.
14 */
15
16/*
17 * ldap_pvt_uc.h - Header for Unicode functions.
18 * These are meant to be used by the OpenLDAP distribution only.
19 * These should be named ldap_pvt_....()
20 */
21
22#ifndef _LDAP_PVT_UC_H
23#define _LDAP_PVT_UC_H 1
24
25#include <lber.h>				/* get ber_slen_t */
26
27#include <ac/bytes.h>
28#include "../libraries/liblunicode/ucdata/ucdata.h"
29
30LDAP_BEGIN_DECL
31
32/*
33 * UTF-8 (in utf-8.c)
34 */
35
36/* UCDATA uses UCS-2 passed in a 4 byte unsigned int */
37typedef ac_uint4 ldap_unicode_t;
38
39/* Convert a string with csize octets per character to UTF-8 */
40LDAP_F( int ) ldap_ucs_to_utf8s LDAP_P((
41	struct berval *ucs, int csize, struct berval *utf8s ));
42
43
44/* returns the number of bytes in the UTF-8 string */
45LDAP_F (ber_len_t) ldap_utf8_bytes( const char * );
46/* returns the number of UTF-8 characters in the string */
47LDAP_F (ber_len_t) ldap_utf8_chars( const char * );
48/* returns the length (in bytes) of the UTF-8 character */
49LDAP_F (int) ldap_utf8_offset( const char * );
50/* returns the length (in bytes) indicated by the UTF-8 character */
51LDAP_F (int) ldap_utf8_charlen( const char * );
52
53/* returns the length (in bytes) indicated by the UTF-8 character
54 * also checks that shortest possible encoding was used
55 */
56LDAP_F (int) ldap_utf8_charlen2( const char * );
57
58/* copies a UTF-8 character and returning number of bytes copied */
59LDAP_F (int) ldap_utf8_copy( char *, const char *);
60
61/* returns pointer of next UTF-8 character in string */
62LDAP_F (char*) ldap_utf8_next( const char * );
63/* returns pointer of previous UTF-8 character in string */
64LDAP_F (char*) ldap_utf8_prev( const char * );
65
66/* primitive ctype routines -- not aware of non-ascii characters */
67LDAP_F (int) ldap_utf8_isascii( const char * );
68LDAP_F (int) ldap_utf8_isalpha( const char * );
69LDAP_F (int) ldap_utf8_isalnum( const char * );
70LDAP_F (int) ldap_utf8_isdigit( const char * );
71LDAP_F (int) ldap_utf8_isxdigit( const char * );
72LDAP_F (int) ldap_utf8_isspace( const char * );
73
74/* span characters not in set, return bytes spanned */
75LDAP_F (ber_len_t) ldap_utf8_strcspn( const char* str, const char *set);
76/* span characters in set, return bytes spanned */
77LDAP_F (ber_len_t) ldap_utf8_strspn( const char* str, const char *set);
78/* return first occurance of character in string */
79LDAP_F (char *) ldap_utf8_strchr( const char* str, const char *chr);
80/* return first character of set in string */
81LDAP_F (char *) ldap_utf8_strpbrk( const char* str, const char *set);
82/* reentrant tokenizer */
83LDAP_F (char*) ldap_utf8_strtok( char* sp, const char* sep, char **last);
84
85/* Optimizations */
86LDAP_V (const char) ldap_utf8_lentab[128];
87LDAP_V (const char) ldap_utf8_mintab[32];
88
89#define LDAP_UTF8_ISASCII(p) ( !(*(const unsigned char *)(p) & 0x80 ) )
90#define LDAP_UTF8_CHARLEN(p) ( LDAP_UTF8_ISASCII(p) \
91	? 1 : ldap_utf8_lentab[*(const unsigned char *)(p) ^ 0x80] )
92
93/* This is like CHARLEN but additionally validates to make sure
94 * the char used the shortest possible encoding.
95 * 'l' is used to temporarily hold the result of CHARLEN.
96 */
97#define LDAP_UTF8_CHARLEN2(p, l) ( ( ( l = LDAP_UTF8_CHARLEN( p )) < 3 || \
98	( ldap_utf8_mintab[*(const unsigned char *)(p) & 0x1f] & (p)[1] ) ) ? \
99	l : 0 )
100
101#define LDAP_UTF8_OFFSET(p) ( LDAP_UTF8_ISASCII(p) \
102	? 1 : ldap_utf8_offset((p)) )
103
104#define LDAP_UTF8_COPY(d,s) ( LDAP_UTF8_ISASCII(s) \
105	? (*(d) = *(s), 1) : ldap_utf8_copy((d),(s)) )
106
107#define LDAP_UTF8_NEXT(p) (	LDAP_UTF8_ISASCII(p) \
108	? (char *)(p)+1 : ldap_utf8_next((p)) )
109
110#define LDAP_UTF8_INCR(p) ((p) = LDAP_UTF8_NEXT(p))
111
112/* For symmetry */
113#define LDAP_UTF8_PREV(p) (ldap_utf8_prev((p)))
114#define LDAP_UTF8_DECR(p) ((p)=LDAP_UTF8_PREV((p)))
115
116
117/* these probably should be renamed */
118LDAP_LUNICODE_F(int) ucstrncmp(
119	const ldap_unicode_t *,
120	const ldap_unicode_t *,
121	ber_len_t );
122
123LDAP_LUNICODE_F(int) ucstrncasecmp(
124	const ldap_unicode_t *,
125	const ldap_unicode_t *,
126	ber_len_t );
127
128LDAP_LUNICODE_F(ldap_unicode_t *) ucstrnchr(
129	const ldap_unicode_t *,
130	ber_len_t,
131	ldap_unicode_t );
132
133LDAP_LUNICODE_F(ldap_unicode_t *) ucstrncasechr(
134	const ldap_unicode_t *,
135	ber_len_t,
136	ldap_unicode_t );
137
138LDAP_LUNICODE_F(void) ucstr2upper(
139	ldap_unicode_t *,
140	ber_len_t );
141
142#define LDAP_UTF8_NOCASEFOLD	0x0U
143#define LDAP_UTF8_CASEFOLD	0x1U
144#define LDAP_UTF8_ARG1NFC	0x2U
145#define LDAP_UTF8_ARG2NFC	0x4U
146#define LDAP_UTF8_APPROX	0x8U
147
148LDAP_LUNICODE_F(struct berval *) UTF8bvnormalize(
149	struct berval *,
150	struct berval *,
151	unsigned,
152	void *memctx );
153
154LDAP_LUNICODE_F(int) UTF8bvnormcmp(
155	struct berval *,
156	struct berval *,
157	unsigned,
158	void *memctx );
159
160LDAP_END_DECL
161
162#endif
163
164