3 * Copyright 1998-2002 The OpenLDAP Foundation, Redwood City, California, USA
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted only as authorized by the OpenLDAP
8 * Public License. A copy of this license is available at
9 * http://www.OpenLDAP.org/license.html or in file LICENSE in the
10 * top-level directory of the distribution.
13 * ldap_pvt_uc.h - Header for Unicode functions.
14 * These are meant to be used by the OpenLDAP distribution only.
15 * These should be named ldap_pvt_....()
18 #ifndef _LDAP_PVT_UC_H
19 #define _LDAP_PVT_UC_H 1
21 #include <lber.h> /* get ber_slen_t */
24 #include "../libraries/liblunicode/ucdata/ucdata.h"
26 #include "../libraries/liblunicode/ucdata.h"
35 typedef short ldap_ucs2_t;
37 /* UCDATA uses UCS-2 passed in an unsigned long */
38 typedef unsigned long ldap_unicode_t;
39 #define ldap_utf8_to_unicode( p ) ldap_utf8_to_ucs4((p))
40 #define ldap_unicode_to_utf8( c, buf ) ldap_ucs4_to_ucs4((c),(buf))
42 /* returns the number of bytes in the UTF-8 string */
43 LDAP_F (ber_len_t) ldap_utf8_bytes( const char * );
44 /* returns the number of UTF-8 characters in the string */
45 LDAP_F (ber_len_t) ldap_utf8_chars( const char * );
46 /* returns the length (in bytes) of the UTF-8 character */
47 LDAP_F (int) ldap_utf8_offset( const char * );
48 /* returns the length (in bytes) indicated by the UTF-8 character */
49 LDAP_F (int) ldap_utf8_charlen( const char * );
51 /* returns the length (in bytes) indicated by the UTF-8 character
52 * also checks that shortest possible encoding was used
54 LDAP_F (int) ldap_utf8_charlen2( const char * );
56 /* copies a UTF-8 character and returning number of bytes copied */
57 LDAP_F (int) ldap_utf8_copy( char *, const char *);
59 /* returns pointer of next UTF-8 character in string */
60 LDAP_F (char*) ldap_utf8_next( const char * );
61 /* returns pointer of previous UTF-8 character in string */
62 LDAP_F (char*) ldap_utf8_prev( const char * );
64 /* primitive ctype routines -- not aware of non-ascii characters */
65 LDAP_F (int) ldap_utf8_isascii( const char * );
66 LDAP_F (int) ldap_utf8_isalpha( const char * );
67 LDAP_F (int) ldap_utf8_isalnum( const char * );
68 LDAP_F (int) ldap_utf8_isdigit( const char * );
69 LDAP_F (int) ldap_utf8_isxdigit( const char * );
70 LDAP_F (int) ldap_utf8_isspace( const char * );
72 /* span characters not in set, return bytes spanned */
73 LDAP_F (ber_len_t) ldap_utf8_strcspn( const char* str, const char *set);
74 /* span characters in set, return bytes spanned */
75 LDAP_F (ber_len_t) ldap_utf8_strspn( const char* str, const char *set);
76 /* return first occurance of character in string */
77 LDAP_F (char *) ldap_utf8_strchr( const char* str, const char *chr);
78 /* return first character of set in string */
79 LDAP_F (char *) ldap_utf8_strpbrk( const char* str, const char *set);
80 /* reentrant tokenizer */
81 LDAP_F (char*) ldap_utf8_strtok( char* sp, const char* sep, char **last);
84 LDAP_V (const char) ldap_utf8_lentab[128];
85 LDAP_V (const char) ldap_utf8_mintab[32];
87 #define LDAP_UTF8_ISASCII(p) ( !(*(unsigned char *)(p) & 0x80 ) )
88 #define LDAP_UTF8_CHARLEN(p) ( LDAP_UTF8_ISASCII(p) \
89 ? 1 : ldap_utf8_lentab[*(unsigned char *)(p) ^ 0x80] )
91 /* This is like CHARLEN but additionally validates to make sure
92 * the char used the shortest possible encoding.
93 * 'l' is used to temporarily hold the result of CHARLEN.
95 #define LDAP_UTF8_CHARLEN2(p, l) ( ( ( l = LDAP_UTF8_CHARLEN( p )) < 3 || \
96 ( ldap_utf8_mintab[*(unsigned char *)(p) & 0x1f] & (p)[1] ) ) ? \
99 #define LDAP_UTF8_OFFSET(p) ( LDAP_UTF8_ISASCII(p) \
100 ? 1 : ldap_utf8_offset((p)) )
102 #define LDAP_UTF8_COPY(d,s) ( LDAP_UTF8_ISASCII(s) \
103 ? (*(d) = *(s), 1) : ldap_utf8_copy((d),(s)) )
105 #define LDAP_UTF8_NEXT(p) ( LDAP_UTF8_ISASCII(p) \
106 ? (char *)(p)+1 : ldap_utf8_next((p)) )
108 #define LDAP_UTF8_INCR(p) ((p) = LDAP_UTF8_NEXT(p))
111 #define LDAP_UTF8_PREV(p) (ldap_utf8_prev((p)))
112 #define LDAP_UTF8_DECR(p) ((p)=LDAP_UTF8_PREV((p)))
115 /* these probably should be renamed */
116 LDAP_LUNICODE_F(int) ucstrncmp(
117 const ldap_unicode_t *,
118 const ldap_unicode_t *,
121 LDAP_LUNICODE_F(int) ucstrncasecmp(
122 const ldap_unicode_t *,
123 const ldap_unicode_t *,
126 LDAP_LUNICODE_F(ldap_unicode_t *) ucstrnchr(
127 const ldap_unicode_t *,
131 LDAP_LUNICODE_F(ldap_unicode_t *) ucstrncasechr(
132 const ldap_unicode_t *,
136 LDAP_LUNICODE_F(void) ucstr2upper(
140 #define LDAP_UTF8_NOCASEFOLD 0x0U
141 #define LDAP_UTF8_CASEFOLD 0x1U
142 #define LDAP_UTF8_ARG1NFC 0x2U
143 #define LDAP_UTF8_ARG2NFC 0x4U
144 #define LDAP_UTF8_APPROX 0x8U
146 LDAP_LUNICODE_F(struct berval *) UTF8bvnormalize(
151 LDAP_LUNICODE_F(int) UTF8bvnormcmp(