michael@0: /*
michael@0:  *******************************************************************************
michael@0:  *
michael@0:  *   Copyright (C) 2003-2010, International Business Machines
michael@0:  *   Corporation and others.  All Rights Reserved.
michael@0:  *
michael@0:  *******************************************************************************
michael@0:  *   file name:  usprep.h
michael@0:  *   encoding:   US-ASCII
michael@0:  *   tab size:   8 (not used)
michael@0:  *   indentation:4
michael@0:  *
michael@0:  *   created on: 2003jul2
michael@0:  *   created by: Ram Viswanadha
michael@0:  */
michael@0: 
michael@0: #ifndef __USPREP_H__
michael@0: #define __USPREP_H__
michael@0: 
michael@0: /**
michael@0:  * \file 
michael@0:  * \brief C API: Implements the StringPrep algorithm.
michael@0:  */
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: #include "unicode/localpointer.h"
michael@0: 
michael@0: /**
michael@0:  *
michael@0:  * StringPrep API implements the StingPrep framework as described by RFC 3454.
michael@0:  * StringPrep prepares Unicode strings for use in network protocols.
michael@0:  * Profiles of StingPrep are set of rules and data according to with the
michael@0:  * Unicode Strings are prepared. Each profiles contains tables which describe
michael@0:  * how a code point should be treated. The tables are broadly classied into
michael@0:  * <ul>
michael@0:  *     <li> Unassinged Table: Contains code points that are unassigned 
michael@0:  *          in the Unicode Version supported by StringPrep. Currently 
michael@0:  *          RFC 3454 supports Unicode 3.2. </li>
michael@0:  *     <li> Prohibited Table: Contains code points that are prohibted from
michael@0:  *          the output of the StringPrep processing function. </li>
michael@0:  *     <li> Mapping Table: Contains code ponts that are deleted from the output or case mapped. </li>
michael@0:  * </ul>
michael@0:  * 
michael@0:  * The procedure for preparing Unicode strings:
michael@0:  * <ol>
michael@0:  *      <li> Map: For each character in the input, check if it has a mapping
michael@0:  *           and, if so, replace it with its mapping. </li>
michael@0:  *      <li> Normalize: Possibly normalize the result of step 1 using Unicode
michael@0:  *           normalization. </li>
michael@0:  *      <li> Prohibit: Check for any characters that are not allowed in the
michael@0:  *        output.  If any are found, return an error.</li>
michael@0:  *      <li> Check bidi: Possibly check for right-to-left characters, and if
michael@0:  *           any are found, make sure that the whole string satisfies the
michael@0:  *           requirements for bidirectional strings.  If the string does not
michael@0:  *           satisfy the requirements for bidirectional strings, return an
michael@0:  *           error.  </li>
michael@0:  * </ol>
michael@0:  * @author Ram Viswanadha
michael@0:  */
michael@0: #if !UCONFIG_NO_IDNA
michael@0: 
michael@0: #include "unicode/parseerr.h"
michael@0: 
michael@0: /**
michael@0:  * The StringPrep profile
michael@0:  * @stable ICU 2.8
michael@0:  */
michael@0: typedef struct UStringPrepProfile UStringPrepProfile;
michael@0: 
michael@0: 
michael@0: /** 
michael@0:  * Option to prohibit processing of unassigned code points in the input
michael@0:  * 
michael@0:  * @see  usprep_prepare
michael@0:  * @stable ICU 2.8
michael@0:  */
michael@0: #define USPREP_DEFAULT 0x0000
michael@0: 
michael@0: /** 
michael@0:  * Option to allow processing of unassigned code points in the input
michael@0:  * 
michael@0:  * @see  usprep_prepare
michael@0:  * @stable ICU 2.8
michael@0:  */
michael@0: #define USPREP_ALLOW_UNASSIGNED 0x0001
michael@0: 
michael@0: /**
michael@0:  * enums for the standard stringprep profile types
michael@0:  * supported by usprep_openByType.
michael@0:  * @see usprep_openByType
michael@0:  * @stable ICU 4.2
michael@0:  */
michael@0: typedef enum UStringPrepProfileType {
michael@0:     /**
michael@0:      * RFC3491 Nameprep
michael@0:      * @stable ICU 4.2
michael@0:      */
michael@0:     USPREP_RFC3491_NAMEPREP,
michael@0:     /**
michael@0:      * RFC3530 nfs4_cs_prep
michael@0:      * @stable ICU 4.2
michael@0:      */
michael@0: 	USPREP_RFC3530_NFS4_CS_PREP,
michael@0:     /**
michael@0:      * RFC3530 nfs4_cs_prep with case insensitive option
michael@0:      * @stable ICU 4.2
michael@0:      */
michael@0: 	USPREP_RFC3530_NFS4_CS_PREP_CI,
michael@0:     /**
michael@0:      * RFC3530 nfs4_cis_prep
michael@0:      * @stable ICU 4.2
michael@0:      */
michael@0: 	USPREP_RFC3530_NFS4_CIS_PREP,
michael@0:     /**
michael@0:      * RFC3530 nfs4_mixed_prep for prefix
michael@0:      * @stable ICU 4.2
michael@0:      */
michael@0: 	USPREP_RFC3530_NFS4_MIXED_PREP_PREFIX,
michael@0:     /**
michael@0:      * RFC3530 nfs4_mixed_prep for suffix
michael@0:      * @stable ICU 4.2
michael@0:      */
michael@0: 	USPREP_RFC3530_NFS4_MIXED_PREP_SUFFIX,
michael@0:     /**
michael@0:      * RFC3722 iSCSI
michael@0:      * @stable ICU 4.2
michael@0:      */
michael@0: 	USPREP_RFC3722_ISCSI,
michael@0:     /**
michael@0:      * RFC3920 XMPP Nodeprep
michael@0:      * @stable ICU 4.2
michael@0:      */
michael@0: 	USPREP_RFC3920_NODEPREP,
michael@0:     /**
michael@0:      * RFC3920 XMPP Resourceprep
michael@0:      * @stable ICU 4.2
michael@0:      */
michael@0: 	USPREP_RFC3920_RESOURCEPREP,
michael@0:     /**
michael@0:      * RFC4011 Policy MIB Stringprep
michael@0:      * @stable ICU 4.2
michael@0:      */
michael@0: 	USPREP_RFC4011_MIB,
michael@0:     /**
michael@0:      * RFC4013 SASLprep
michael@0:      * @stable ICU 4.2
michael@0:      */
michael@0:     USPREP_RFC4013_SASLPREP,
michael@0:     /**
michael@0:      * RFC4505 trace
michael@0:      * @stable ICU 4.2
michael@0:      */
michael@0: 	USPREP_RFC4505_TRACE,
michael@0:     /**
michael@0:      * RFC4518 LDAP
michael@0:      * @stable ICU 4.2
michael@0:      */
michael@0: 	USPREP_RFC4518_LDAP,
michael@0:     /**
michael@0:      * RFC4518 LDAP for case ignore, numeric and stored prefix
michael@0:      * matching rules
michael@0:      * @stable ICU 4.2
michael@0:      */
michael@0: 	USPREP_RFC4518_LDAP_CI
michael@0: } UStringPrepProfileType;
michael@0: 
michael@0: /**
michael@0:  * Creates a StringPrep profile from the data file.
michael@0:  *
michael@0:  * @param path      string containing the full path pointing to the directory
michael@0:  *                  where the profile reside followed by the package name
michael@0:  *                  e.g. "/usr/resource/my_app/profiles/mydata" on a Unix system.
michael@0:  *                  if NULL, ICU default data files will be used.
michael@0:  * @param fileName  name of the profile file to be opened
michael@0:  * @param status    ICU error code in/out parameter. Must not be NULL.
michael@0:  *                  Must fulfill U_SUCCESS before the function call.
michael@0:  * @return Pointer to UStringPrepProfile that is opened. Should be closed by
michael@0:  * calling usprep_close()
michael@0:  * @see usprep_close()
michael@0:  * @stable ICU 2.8
michael@0:  */
michael@0: U_STABLE UStringPrepProfile* U_EXPORT2
michael@0: usprep_open(const char* path, 
michael@0:             const char* fileName,
michael@0:             UErrorCode* status);
michael@0: 
michael@0: /**
michael@0:  * Creates a StringPrep profile for the specified profile type.
michael@0:  *
michael@0:  * @param type		The profile type
michael@0:  * @param status    ICU error code in/out parameter. Must not be NULL.
michael@0:  *                  Must fulfill U_SUCCESS before the function call.
michael@0:  * @return          Pointer to UStringPrepProfile that is opened. Should be closed by
michael@0:  *                  calling usprep_close()
michael@0:  * @see usprep_close()
michael@0:  * @stable ICU 4.2
michael@0:  */
michael@0: U_STABLE UStringPrepProfile* U_EXPORT2
michael@0: usprep_openByType(UStringPrepProfileType type,
michael@0: 				  UErrorCode* status);
michael@0: 
michael@0: /**
michael@0:  * Closes the profile
michael@0:  * @param profile The profile to close
michael@0:  * @stable ICU 2.8
michael@0:  */
michael@0: U_STABLE void U_EXPORT2
michael@0: usprep_close(UStringPrepProfile* profile);
michael@0: 
michael@0: #if U_SHOW_CPLUSPLUS_API
michael@0: 
michael@0: U_NAMESPACE_BEGIN
michael@0: 
michael@0: /**
michael@0:  * \class LocalUStringPrepProfilePointer
michael@0:  * "Smart pointer" class, closes a UStringPrepProfile via usprep_close().
michael@0:  * For most methods see the LocalPointerBase base class.
michael@0:  *
michael@0:  * @see LocalPointerBase
michael@0:  * @see LocalPointer
michael@0:  * @stable ICU 4.4
michael@0:  */
michael@0: U_DEFINE_LOCAL_OPEN_POINTER(LocalUStringPrepProfilePointer, UStringPrepProfile, usprep_close);
michael@0: 
michael@0: U_NAMESPACE_END
michael@0: 
michael@0: #endif
michael@0: 
michael@0: /**
michael@0:  * Prepare the input buffer for use in applications with the given profile. This operation maps, normalizes(NFKC),
michael@0:  * checks for prohited and BiDi characters in the order defined by RFC 3454
michael@0:  * depending on the options specified in the profile.
michael@0:  *
michael@0:  * @param prep          The profile to use 
michael@0:  * @param src           Pointer to UChar buffer containing the string to prepare
michael@0:  * @param srcLength     Number of characters in the source string
michael@0:  * @param dest          Pointer to the destination buffer to receive the output
michael@0:  * @param destCapacity  The capacity of destination array
michael@0:  * @param options       A bit set of options:
michael@0:  *
michael@0:  *  - USPREP_NONE               Prohibit processing of unassigned code points in the input
michael@0:  *
michael@0:  *  - USPREP_ALLOW_UNASSIGNED   Treat the unassigned code points are in the input 
michael@0:  *                              as normal Unicode code points.
michael@0:  *
michael@0:  * @param parseError        Pointer to UParseError struct to receive information on position 
michael@0:  *                          of error if an error is encountered. Can be NULL.
michael@0:  * @param status            ICU in/out error code parameter.
michael@0:  *                          U_INVALID_CHAR_FOUND if src contains
michael@0:  *                          unmatched single surrogates.
michael@0:  *                          U_INDEX_OUTOFBOUNDS_ERROR if src contains
michael@0:  *                          too many code points.
michael@0:  *                          U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
michael@0:  * @return The number of UChars in the destination buffer
michael@0:  * @stable ICU 2.8
michael@0:  */
michael@0: 
michael@0: U_STABLE int32_t U_EXPORT2
michael@0: usprep_prepare(   const UStringPrepProfile* prep,
michael@0:                   const UChar* src, int32_t srcLength, 
michael@0:                   UChar* dest, int32_t destCapacity,
michael@0:                   int32_t options,
michael@0:                   UParseError* parseError,
michael@0:                   UErrorCode* status );
michael@0: 
michael@0: 
michael@0: #endif /* #if !UCONFIG_NO_IDNA */
michael@0: 
michael@0: #endif