1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/common/unicode/usprep.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,269 @@
1.4 +/*
1.5 + *******************************************************************************
1.6 + *
1.7 + * Copyright (C) 2003-2010, International Business Machines
1.8 + * Corporation and others. All Rights Reserved.
1.9 + *
1.10 + *******************************************************************************
1.11 + * file name: usprep.h
1.12 + * encoding: US-ASCII
1.13 + * tab size: 8 (not used)
1.14 + * indentation:4
1.15 + *
1.16 + * created on: 2003jul2
1.17 + * created by: Ram Viswanadha
1.18 + */
1.19 +
1.20 +#ifndef __USPREP_H__
1.21 +#define __USPREP_H__
1.22 +
1.23 +/**
1.24 + * \file
1.25 + * \brief C API: Implements the StringPrep algorithm.
1.26 + */
1.27 +
1.28 +#include "unicode/utypes.h"
1.29 +#include "unicode/localpointer.h"
1.30 +
1.31 +/**
1.32 + *
1.33 + * StringPrep API implements the StingPrep framework as described by RFC 3454.
1.34 + * StringPrep prepares Unicode strings for use in network protocols.
1.35 + * Profiles of StingPrep are set of rules and data according to with the
1.36 + * Unicode Strings are prepared. Each profiles contains tables which describe
1.37 + * how a code point should be treated. The tables are broadly classied into
1.38 + * <ul>
1.39 + * <li> Unassinged Table: Contains code points that are unassigned
1.40 + * in the Unicode Version supported by StringPrep. Currently
1.41 + * RFC 3454 supports Unicode 3.2. </li>
1.42 + * <li> Prohibited Table: Contains code points that are prohibted from
1.43 + * the output of the StringPrep processing function. </li>
1.44 + * <li> Mapping Table: Contains code ponts that are deleted from the output or case mapped. </li>
1.45 + * </ul>
1.46 + *
1.47 + * The procedure for preparing Unicode strings:
1.48 + * <ol>
1.49 + * <li> Map: For each character in the input, check if it has a mapping
1.50 + * and, if so, replace it with its mapping. </li>
1.51 + * <li> Normalize: Possibly normalize the result of step 1 using Unicode
1.52 + * normalization. </li>
1.53 + * <li> Prohibit: Check for any characters that are not allowed in the
1.54 + * output. If any are found, return an error.</li>
1.55 + * <li> Check bidi: Possibly check for right-to-left characters, and if
1.56 + * any are found, make sure that the whole string satisfies the
1.57 + * requirements for bidirectional strings. If the string does not
1.58 + * satisfy the requirements for bidirectional strings, return an
1.59 + * error. </li>
1.60 + * </ol>
1.61 + * @author Ram Viswanadha
1.62 + */
1.63 +#if !UCONFIG_NO_IDNA
1.64 +
1.65 +#include "unicode/parseerr.h"
1.66 +
1.67 +/**
1.68 + * The StringPrep profile
1.69 + * @stable ICU 2.8
1.70 + */
1.71 +typedef struct UStringPrepProfile UStringPrepProfile;
1.72 +
1.73 +
1.74 +/**
1.75 + * Option to prohibit processing of unassigned code points in the input
1.76 + *
1.77 + * @see usprep_prepare
1.78 + * @stable ICU 2.8
1.79 + */
1.80 +#define USPREP_DEFAULT 0x0000
1.81 +
1.82 +/**
1.83 + * Option to allow processing of unassigned code points in the input
1.84 + *
1.85 + * @see usprep_prepare
1.86 + * @stable ICU 2.8
1.87 + */
1.88 +#define USPREP_ALLOW_UNASSIGNED 0x0001
1.89 +
1.90 +/**
1.91 + * enums for the standard stringprep profile types
1.92 + * supported by usprep_openByType.
1.93 + * @see usprep_openByType
1.94 + * @stable ICU 4.2
1.95 + */
1.96 +typedef enum UStringPrepProfileType {
1.97 + /**
1.98 + * RFC3491 Nameprep
1.99 + * @stable ICU 4.2
1.100 + */
1.101 + USPREP_RFC3491_NAMEPREP,
1.102 + /**
1.103 + * RFC3530 nfs4_cs_prep
1.104 + * @stable ICU 4.2
1.105 + */
1.106 + USPREP_RFC3530_NFS4_CS_PREP,
1.107 + /**
1.108 + * RFC3530 nfs4_cs_prep with case insensitive option
1.109 + * @stable ICU 4.2
1.110 + */
1.111 + USPREP_RFC3530_NFS4_CS_PREP_CI,
1.112 + /**
1.113 + * RFC3530 nfs4_cis_prep
1.114 + * @stable ICU 4.2
1.115 + */
1.116 + USPREP_RFC3530_NFS4_CIS_PREP,
1.117 + /**
1.118 + * RFC3530 nfs4_mixed_prep for prefix
1.119 + * @stable ICU 4.2
1.120 + */
1.121 + USPREP_RFC3530_NFS4_MIXED_PREP_PREFIX,
1.122 + /**
1.123 + * RFC3530 nfs4_mixed_prep for suffix
1.124 + * @stable ICU 4.2
1.125 + */
1.126 + USPREP_RFC3530_NFS4_MIXED_PREP_SUFFIX,
1.127 + /**
1.128 + * RFC3722 iSCSI
1.129 + * @stable ICU 4.2
1.130 + */
1.131 + USPREP_RFC3722_ISCSI,
1.132 + /**
1.133 + * RFC3920 XMPP Nodeprep
1.134 + * @stable ICU 4.2
1.135 + */
1.136 + USPREP_RFC3920_NODEPREP,
1.137 + /**
1.138 + * RFC3920 XMPP Resourceprep
1.139 + * @stable ICU 4.2
1.140 + */
1.141 + USPREP_RFC3920_RESOURCEPREP,
1.142 + /**
1.143 + * RFC4011 Policy MIB Stringprep
1.144 + * @stable ICU 4.2
1.145 + */
1.146 + USPREP_RFC4011_MIB,
1.147 + /**
1.148 + * RFC4013 SASLprep
1.149 + * @stable ICU 4.2
1.150 + */
1.151 + USPREP_RFC4013_SASLPREP,
1.152 + /**
1.153 + * RFC4505 trace
1.154 + * @stable ICU 4.2
1.155 + */
1.156 + USPREP_RFC4505_TRACE,
1.157 + /**
1.158 + * RFC4518 LDAP
1.159 + * @stable ICU 4.2
1.160 + */
1.161 + USPREP_RFC4518_LDAP,
1.162 + /**
1.163 + * RFC4518 LDAP for case ignore, numeric and stored prefix
1.164 + * matching rules
1.165 + * @stable ICU 4.2
1.166 + */
1.167 + USPREP_RFC4518_LDAP_CI
1.168 +} UStringPrepProfileType;
1.169 +
1.170 +/**
1.171 + * Creates a StringPrep profile from the data file.
1.172 + *
1.173 + * @param path string containing the full path pointing to the directory
1.174 + * where the profile reside followed by the package name
1.175 + * e.g. "/usr/resource/my_app/profiles/mydata" on a Unix system.
1.176 + * if NULL, ICU default data files will be used.
1.177 + * @param fileName name of the profile file to be opened
1.178 + * @param status ICU error code in/out parameter. Must not be NULL.
1.179 + * Must fulfill U_SUCCESS before the function call.
1.180 + * @return Pointer to UStringPrepProfile that is opened. Should be closed by
1.181 + * calling usprep_close()
1.182 + * @see usprep_close()
1.183 + * @stable ICU 2.8
1.184 + */
1.185 +U_STABLE UStringPrepProfile* U_EXPORT2
1.186 +usprep_open(const char* path,
1.187 + const char* fileName,
1.188 + UErrorCode* status);
1.189 +
1.190 +/**
1.191 + * Creates a StringPrep profile for the specified profile type.
1.192 + *
1.193 + * @param type The profile type
1.194 + * @param status ICU error code in/out parameter. Must not be NULL.
1.195 + * Must fulfill U_SUCCESS before the function call.
1.196 + * @return Pointer to UStringPrepProfile that is opened. Should be closed by
1.197 + * calling usprep_close()
1.198 + * @see usprep_close()
1.199 + * @stable ICU 4.2
1.200 + */
1.201 +U_STABLE UStringPrepProfile* U_EXPORT2
1.202 +usprep_openByType(UStringPrepProfileType type,
1.203 + UErrorCode* status);
1.204 +
1.205 +/**
1.206 + * Closes the profile
1.207 + * @param profile The profile to close
1.208 + * @stable ICU 2.8
1.209 + */
1.210 +U_STABLE void U_EXPORT2
1.211 +usprep_close(UStringPrepProfile* profile);
1.212 +
1.213 +#if U_SHOW_CPLUSPLUS_API
1.214 +
1.215 +U_NAMESPACE_BEGIN
1.216 +
1.217 +/**
1.218 + * \class LocalUStringPrepProfilePointer
1.219 + * "Smart pointer" class, closes a UStringPrepProfile via usprep_close().
1.220 + * For most methods see the LocalPointerBase base class.
1.221 + *
1.222 + * @see LocalPointerBase
1.223 + * @see LocalPointer
1.224 + * @stable ICU 4.4
1.225 + */
1.226 +U_DEFINE_LOCAL_OPEN_POINTER(LocalUStringPrepProfilePointer, UStringPrepProfile, usprep_close);
1.227 +
1.228 +U_NAMESPACE_END
1.229 +
1.230 +#endif
1.231 +
1.232 +/**
1.233 + * Prepare the input buffer for use in applications with the given profile. This operation maps, normalizes(NFKC),
1.234 + * checks for prohited and BiDi characters in the order defined by RFC 3454
1.235 + * depending on the options specified in the profile.
1.236 + *
1.237 + * @param prep The profile to use
1.238 + * @param src Pointer to UChar buffer containing the string to prepare
1.239 + * @param srcLength Number of characters in the source string
1.240 + * @param dest Pointer to the destination buffer to receive the output
1.241 + * @param destCapacity The capacity of destination array
1.242 + * @param options A bit set of options:
1.243 + *
1.244 + * - USPREP_NONE Prohibit processing of unassigned code points in the input
1.245 + *
1.246 + * - USPREP_ALLOW_UNASSIGNED Treat the unassigned code points are in the input
1.247 + * as normal Unicode code points.
1.248 + *
1.249 + * @param parseError Pointer to UParseError struct to receive information on position
1.250 + * of error if an error is encountered. Can be NULL.
1.251 + * @param status ICU in/out error code parameter.
1.252 + * U_INVALID_CHAR_FOUND if src contains
1.253 + * unmatched single surrogates.
1.254 + * U_INDEX_OUTOFBOUNDS_ERROR if src contains
1.255 + * too many code points.
1.256 + * U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
1.257 + * @return The number of UChars in the destination buffer
1.258 + * @stable ICU 2.8
1.259 + */
1.260 +
1.261 +U_STABLE int32_t U_EXPORT2
1.262 +usprep_prepare( const UStringPrepProfile* prep,
1.263 + const UChar* src, int32_t srcLength,
1.264 + UChar* dest, int32_t destCapacity,
1.265 + int32_t options,
1.266 + UParseError* parseError,
1.267 + UErrorCode* status );
1.268 +
1.269 +
1.270 +#endif /* #if !UCONFIG_NO_IDNA */
1.271 +
1.272 +#endif
