icuSources/common/unicode/usprep.h

   1 /*
   2  *******************************************************************************
   3  *
   4  *   Copyright (C) 2003-2010, International Business Machines
   5  *   Corporation and others.  All Rights Reserved.
   6  *
   7  *******************************************************************************
   8  *   file name:  usprep.h
   9  *   encoding:   US-ASCII
  10  *   tab size:   8 (not used)
  11  *   indentation:4
  12  *
  13  *   created on: 2003jul2
  14  *   created by: Ram Viswanadha
  15  */
  16
  17 #ifndef __USPREP_H__
  18 #define __USPREP_H__
  19
  20 /**
  21  * \file
  22  * \brief C API: Implements the StringPrep algorithm.
  23  */
  24
  25 #include "unicode/utypes.h"
  26 #include "unicode/localpointer.h"
  27
  28 /**
  29  *
  30  * StringPrep API implements the StingPrep framework as described by RFC 3454.
  31  * StringPrep prepares Unicode strings for use in network protocols.
  32  * Profiles of StingPrep are set of rules and data according to with the
  33  * Unicode Strings are prepared. Each profiles contains tables which describe
  34  * how a code point should be treated. The tables are broadly classied into
  35  * <ul>
  36  *     <li> Unassinged Table: Contains code points that are unassigned
  37  *          in the Unicode Version supported by StringPrep. Currently
  38  *          RFC 3454 supports Unicode 3.2. </li>
  39  *     <li> Prohibited Table: Contains code points that are prohibted from
  40  *          the output of the StringPrep processing function. </li>
  41  *     <li> Mapping Table: Contains code ponts that are deleted from the output or case mapped. </li>
  42  * </ul>
  43  *
  44  * The procedure for preparing Unicode strings:
  45  * <ol>
  46  *      <li> Map: For each character in the input, check if it has a mapping
  47  *           and, if so, replace it with its mapping. </li>
  48  *      <li> Normalize: Possibly normalize the result of step 1 using Unicode
  49  *           normalization. </li>
  50  *      <li> Prohibit: Check for any characters that are not allowed in the
  51  *        output.  If any are found, return an error.</li>
  52  *      <li> Check bidi: Possibly check for right-to-left characters, and if
  53  *           any are found, make sure that the whole string satisfies the
  54  *           requirements for bidirectional strings.  If the string does not
  55  *           satisfy the requirements for bidirectional strings, return an
  56  *           error.  </li>
  57  * </ol>
  58  * @author Ram Viswanadha
  59  */
  60 #if !UCONFIG_NO_IDNA
  61
  62 #include "unicode/parseerr.h"
  63
  64 /**
  65  * The StringPrep profile
  66  * @stable ICU 2.8
  67  */
  68 typedef struct UStringPrepProfile UStringPrepProfile;
  69
  70
  71 /**
  72  * Option to prohibit processing of unassigned code points in the input
  73  *
  74  * @see  usprep_prepare
  75  * @stable ICU 2.8
  76  */
  77 #define USPREP_DEFAULT 0x0000
  78
  79 /**
  80  * Option to allow processing of unassigned code points in the input
  81  *
  82  * @see  usprep_prepare
  83  * @stable ICU 2.8
  84  */
  85 #define USPREP_ALLOW_UNASSIGNED 0x0001
  86
  87 /**
  88  * enums for the standard stringprep profile types
  89  * supported by usprep_openByType.
  90  * @see usprep_openByType
  91  * @stable ICU 4.2
  92  */
  93 typedef enum UStringPrepProfileType {
  94     /**
  95      * RFC3491 Nameprep
  96      * @stable ICU 4.2
  97      */
  98     USPREP_RFC3491_NAMEPREP,
  99     /**
 100      * RFC3530 nfs4_cs_prep
 101      * @stable ICU 4.2
 102      */
 103         USPREP_RFC3530_NFS4_CS_PREP,
 104     /**
 105      * RFC3530 nfs4_cs_prep with case insensitive option
 106      * @stable ICU 4.2
 107      */
 108         USPREP_RFC3530_NFS4_CS_PREP_CI,
 109     /**
 110      * RFC3530 nfs4_cis_prep
 111      * @stable ICU 4.2
 112      */
 113         USPREP_RFC3530_NFS4_CIS_PREP,
 114     /**
 115      * RFC3530 nfs4_mixed_prep for prefix
 116      * @stable ICU 4.2
 117      */
 118         USPREP_RFC3530_NFS4_MIXED_PREP_PREFIX,
 119     /**
 120      * RFC3530 nfs4_mixed_prep for suffix
 121      * @stable ICU 4.2
 122      */
 123         USPREP_RFC3530_NFS4_MIXED_PREP_SUFFIX,
 124     /**
 125      * RFC3722 iSCSI
 126      * @stable ICU 4.2
 127      */
 128         USPREP_RFC3722_ISCSI,
 129     /**
 130      * RFC3920 XMPP Nodeprep
 131      * @stable ICU 4.2
 132      */
 133         USPREP_RFC3920_NODEPREP,
 134     /**
 135      * RFC3920 XMPP Resourceprep
 136      * @stable ICU 4.2
 137      */
 138         USPREP_RFC3920_RESOURCEPREP,
 139     /**
 140      * RFC4011 Policy MIB Stringprep
 141      * @stable ICU 4.2
 142      */
 143         USPREP_RFC4011_MIB,
 144     /**
 145      * RFC4013 SASLprep
 146      * @stable ICU 4.2
 147      */
 148     USPREP_RFC4013_SASLPREP,
 149     /**
 150      * RFC4505 trace
 151      * @stable ICU 4.2
 152      */
 153         USPREP_RFC4505_TRACE,
 154     /**
 155      * RFC4518 LDAP
 156      * @stable ICU 4.2
 157      */
 158         USPREP_RFC4518_LDAP,
 159     /**
 160      * RFC4518 LDAP for case ignore, numeric and stored prefix
 161      * matching rules
 162      * @stable ICU 4.2
 163      */
 164         USPREP_RFC4518_LDAP_CI
 165 } UStringPrepProfileType;
 166
 167 /**
 168  * Creates a StringPrep profile from the data file.
 169  *
 170  * @param path      string containing the full path pointing to the directory
 171  *                  where the profile reside followed by the package name
 172  *                  e.g. "/usr/resource/my_app/profiles/mydata" on a Unix system.
 173  *                  if NULL, ICU default data files will be used.
 174  * @param fileName  name of the profile file to be opened
 175  * @param status    ICU error code in/out parameter. Must not be NULL.
 176  *                  Must fulfill U_SUCCESS before the function call.
 177  * @return Pointer to UStringPrepProfile that is opened. Should be closed by
 178  * calling usprep_close()
 179  * @see usprep_close()
 180  * @stable ICU 2.8
 181  */
 182 U_STABLE UStringPrepProfile* U_EXPORT2
 183 usprep_open(const char* path,
 184             const char* fileName,
 185             UErrorCode* status);
 186
 187 /**
 188  * Creates a StringPrep profile for the specified profile type.
 189  *
 190  * @param type          The profile type
 191  * @param status    ICU error code in/out parameter. Must not be NULL.
 192  *                  Must fulfill U_SUCCESS before the function call.
 193  * @return          Pointer to UStringPrepProfile that is opened. Should be closed by
 194  *                  calling usprep_close()
 195  * @see usprep_close()
 196  * @stable ICU 4.2
 197  */
 198 U_STABLE UStringPrepProfile* U_EXPORT2
 199 usprep_openByType(UStringPrepProfileType type,
 200                                   UErrorCode* status);
 201
 202 /**
 203  * Closes the profile
 204  * @param profile The profile to close
 205  * @stable ICU 2.8
 206  */
 207 U_STABLE void U_EXPORT2
 208 usprep_close(UStringPrepProfile* profile);
 209
 210 #if U_SHOW_CPLUSPLUS_API
 211
 212 U_NAMESPACE_BEGIN
 213
 214 /**
 215  * \class LocalUStringPrepProfilePointer
 216  * "Smart pointer" class, closes a UStringPrepProfile via usprep_close().
 217  * For most methods see the LocalPointerBase base class.
 218  *
 219  * @see LocalPointerBase
 220  * @see LocalPointer
 221  * @stable ICU 4.4
 222  */
 223 U_DEFINE_LOCAL_OPEN_POINTER(LocalUStringPrepProfilePointer, UStringPrepProfile, usprep_close);
 224
 225 U_NAMESPACE_END
 226
 227 #endif
 228
 229 /**
 230  * Prepare the input buffer for use in applications with the given profile. This operation maps, normalizes(NFKC),
 231  * checks for prohited and BiDi characters in the order defined by RFC 3454
 232  * depending on the options specified in the profile.
 233  *
 234  * @param prep          The profile to use
 235  * @param src           Pointer to UChar buffer containing the string to prepare
 236  * @param srcLength     Number of characters in the source string
 237  * @param dest          Pointer to the destination buffer to receive the output
 238  * @param destCapacity  The capacity of destination array
 239  * @param options       A bit set of options:
 240  *
 241  *  - USPREP_NONE               Prohibit processing of unassigned code points in the input
 242  *
 243  *  - USPREP_ALLOW_UNASSIGNED   Treat the unassigned code points are in the input
 244  *                              as normal Unicode code points.
 245  *
 246  * @param parseError        Pointer to UParseError struct to receive information on position
 247  *                          of error if an error is encountered. Can be NULL.
 248  * @param status            ICU in/out error code parameter.
 249  *                          U_INVALID_CHAR_FOUND if src contains
 250  *                          unmatched single surrogates.
 251  *                          U_INDEX_OUTOFBOUNDS_ERROR if src contains
 252  *                          too many code points.
 253  *                          U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
 254  * @return The number of UChars in the destination buffer
 255  * @stable ICU 2.8
 256  */
 257
 258 U_STABLE int32_t U_EXPORT2
 259 usprep_prepare(   const UStringPrepProfile* prep,
 260                   const UChar* src, int32_t srcLength,
 261                   UChar* dest, int32_t destCapacity,
 262                   int32_t options,
 263                   UParseError* parseError,
 264                   UErrorCode* status );
 265
 266
 267 #endif /* #if !UCONFIG_NO_IDNA */
 268
 269 #endif