icuSources/test/intltest/idnaref.h

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4  *******************************************************************************
   5  *
   6  *   Copyright (C) 2003-2011, International Business Machines
   7  *   Corporation and others.  All Rights Reserved.
   8  *
   9  *******************************************************************************
  10  *   file name:  idnaref.h
  11  *   encoding:   UTF-8
  12  *   tab size:   8 (not used)
  13  *   indentation:4
  14  *
  15  *   created on: 2003feb1
  16  *   created by: Ram Viswanadha
  17  */
  18
  19 #ifndef __IDNAREF_H__
  20 #define __IDNAREF_H__
  21
  22 #include "unicode/utypes.h"
  23
  24 #if !UCONFIG_NO_IDNA
  25
  26 #include "unicode/parseerr.h"
  27
  28 #define IDNAREF_DEFAULT          0x0000
  29 #define IDNAREF_ALLOW_UNASSIGNED 0x0001
  30 #define IDNAREF_USE_STD3_RULES   0x0002
  31
  32 /**
  33  * This function implements the ToASCII operation as defined in the IDNA draft.
  34  * This operation is done on <b>single labels</b> before sending it to something that expects
  35  * ASCII names. A label is an individual part of a domain name. Labels are usually
  36  * separated by dots; for e.g." "www.example.com" is composed of 3 labels
  37  * "www","example", and "com".
  38  *
  39  *
  40  * @param src               Input Unicode label.
  41  * @param srcLength         Number of UChars in src, or -1 if NUL-terminated.
  42  * @param dest              Output Unicode array with ACE encoded ASCII label.
  43  * @param destCapacity      Size of dest.
  44  * @param options           A bit set of options:
  45  *
  46  *  - idnaref_UNASSIGNED        Unassigned values can be converted to ASCII for query operations
  47  *                          If TRUE unassigned values are treated as normal Unicode code points.
  48  *                          If FALSE the operation fails with U_UNASSIGNED_CODE_POINT_FOUND error code.
  49  *  - idnaref_USE_STD3_RULES    Use STD3 ASCII rules for host name syntax restrictions
  50  *                          If TRUE and the input does not statisfy STD3 rules, the operation
  51  *                          will fail with U_IDNA_STD3_ASCII_RULES_ERROR
  52  *
  53  * @param parseError        Pointer to UParseError struct to recieve information on position
  54  *                          of error if an error is encountered. Can be NULL.
  55  * @param status            ICU in/out error code parameter.
  56  *                          U_INVALID_CHAR_FOUND if src contains
  57  *                          unmatched single surrogates.
  58  *                          U_INDEX_OUTOFBOUNDS_ERROR if src contains
  59  *                          too many code points.
  60  *                          U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
  61  * @return                  Number of ASCII characters converted.
  62  */
  63 U_CFUNC int32_t U_EXPORT2
  64 idnaref_toASCII(const UChar* src, int32_t srcLength,
  65               UChar* dest, int32_t destCapacity,
  66               int32_t options,
  67               UParseError* parseError,
  68               UErrorCode* status);
  69
  70
  71 /**
  72  * This function implements the ToUnicode operation as defined in the IDNA draft.
  73  * This operation is done on <b>single labels</b> before sending it to something that expects
  74  * ASCII names. A label is an individual part of a domain name. Labels are usually
  75  * separated by dots; for e.g." "www.example.com" is composed of 3 labels
  76  * "www","example", and "com".
  77  *
  78  * @param src               Input ASCII (ACE encoded) label.
  79  * @param srcLength         Number of UChars in src, or -1 if NUL-terminated.
  80  * @param dest Output       Converted Unicode array.
  81  * @param destCapacity      Size of dest.
  82  * @param options           A bit set of options:
  83  *
  84  *  - idnaref_UNASSIGNED        Unassigned values can be converted to ASCII for query operations
  85  *                          If TRUE unassigned values are treated as normal Unicode code points.
  86  *                          If FALSE the operation fails with U_UNASSIGNED_CODE_POINT_FOUND error code.
  87  *  - idnaref_USE_STD3_RULES    Use STD3 ASCII rules for host name syntax restrictions
  88  *                          If TRUE and the input does not statisfy STD3 rules, the operation
  89  *                          will fail with U_IDNA_STD3_ASCII_RULES_ERROR
  90  *
  91  * @param parseError        Pointer to UParseError struct to recieve information on position
  92  *                          of error if an error is encountered. Can be NULL.
  93  * @param status            ICU in/out error code parameter.
  94  *                          U_INVALID_CHAR_FOUND if src contains
  95  *                          unmatched single surrogates.
  96  *                          U_INDEX_OUTOFBOUNDS_ERROR if src contains
  97  *                          too many code points.
  98  *                          U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
  99  * @return                  Number of Unicode characters converted.
 100  */
 101 U_CFUNC int32_t U_EXPORT2
 102 idnaref_toUnicode(const UChar* src, int32_t srcLength,
 103                 UChar* dest, int32_t destCapacity,
 104                 int32_t options,
 105                 UParseError* parseError,
 106                 UErrorCode* status);
 107
 108
 109 /**
 110  * Convenience function that implements the IDNToASCII operation as defined in the IDNA draft.
 111  * This operation is done on complete domain names, e.g: "www.example.com".
 112  * It is important to note that this operation can fail. If it fails, then the input
 113  * domain name cannot be used as an Internationalized Domain Name and the application
 114  * should have methods defined to deal with the failure.
 115  *
 116  * <b>Note:</b> IDNA draft specifies that a conformant application should divide a domain name
 117  * into separate labels, decide whether to apply allowUnassigned and useSTD3ASCIIRules on each,
 118  * and then convert. This function does not offer that level of granularity. The options once
 119  * set will apply to all labels in the domain name
 120  *
 121  * @param src               Input ASCII IDN.
 122  * @param srcLength         Number of UChars in src, or -1 if NUL-terminated.
 123  * @param dest Output       Unicode array.
 124  * @param destCapacity      Size of dest.
 125  * @param options           A bit set of options:
 126  *
 127  *  - idnaref_UNASSIGNED        Unassigned values can be converted to ASCII for query operations
 128  *                          If TRUE unassigned values are treated as normal Unicode code points.
 129  *                          If FALSE the operation fails with U_UNASSIGNED_CODE_POINT_FOUND error code.
 130  *  - idnaref_USE_STD3_RULES    Use STD3 ASCII rules for host name syntax restrictions
 131  *                          If TRUE and the input does not statisfy STD3 rules, the operation
 132  *                          will fail with U_IDNA_STD3_ASCII_RULES_ERROR
 133  *
 134  * @param parseError        Pointer to UParseError struct to recieve information on position
 135  *                          of error if an error is encountered. Can be NULL.
 136  * @param status            ICU in/out error code parameter.
 137  *                          U_INVALID_CHAR_FOUND if src contains
 138  *                          unmatched single surrogates.
 139  *                          U_INDEX_OUTOFBOUNDS_ERROR if src contains
 140  *                          too many code points.
 141  *                          U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
 142  * @return                  Number of ASCII characters converted.
 143  */
 144 U_CFUNC int32_t U_EXPORT2
 145 idnaref_IDNToASCII(  const UChar* src, int32_t srcLength,
 146                    UChar* dest, int32_t destCapacity,
 147                    int32_t options,
 148                    UParseError* parseError,
 149                    UErrorCode* status);
 150
 151 /**
 152  * Convenience function that implements the IDNToUnicode operation as defined in the IDNA draft.
 153  * This operation is done on complete domain names, e.g: "www.example.com".
 154  *
 155  * <b>Note:</b> IDNA draft specifies that a conformant application should divide a domain name
 156  * into separate labels, decide whether to apply allowUnassigned and useSTD3ASCIIRules on each,
 157  * and then convert. This function does not offer that level of granularity. The options once
 158  * set will apply to all labels in the domain name
 159  *
 160  * @param src               Input Unicode IDN.
 161  * @param srcLength         Number of UChars in src, or -1 if NUL-terminated.
 162  * @param dest Output       ASCII array.
 163  * @param destCapacity      Size of dest.
 164  * @param options           A bit set of options:
 165  *
 166  *  - idnaref_UNASSIGNED        Unassigned values can be converted to ASCII for query operations
 167  *                          If TRUE unassigned values are treated as normal Unicode code points.
 168  *                          If FALSE the operation fails with U_UNASSIGNED_CODE_POINT_FOUND error code.
 169  *  - idnaref_USE_STD3_RULES    Use STD3 ASCII rules for host name syntax restrictions
 170  *                          If TRUE and the input does not statisfy STD3 rules, the operation
 171  *                          will fail with U_IDNA_STD3_ASCII_RULES_ERROR
 172  *
 173  * @param parseError        Pointer to UParseError struct to recieve information on position
 174  *                          of error if an error is encountered. Can be NULL.
 175  * @param status            ICU in/out error code parameter.
 176  *                          U_INVALID_CHAR_FOUND if src contains
 177  *                          unmatched single surrogates.
 178  *                          U_INDEX_OUTOFBOUNDS_ERROR if src contains
 179  *                          too many code points.
 180  *                          U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
 181  * @return                  Number of ASCII characters converted.
 182  */
 183 U_CFUNC int32_t U_EXPORT2
 184 idnaref_IDNToUnicode(  const UChar* src, int32_t srcLength,
 185                      UChar* dest, int32_t destCapacity,
 186                      int32_t options,
 187                      UParseError* parseError,
 188                      UErrorCode* status);
 189
 190 /**
 191  * Compare two strings for IDNs for equivalence.
 192  * This function splits the domain names into labels and compares them.
 193  * According to IDN draft, whenever two labels are compared, they are
 194  * considered equal if and only if their ASCII forms (obtained by
 195  * applying toASCII) match using an case-insensitive ASCII comparison.
 196  * Two domain names are considered a match if and only if all labels
 197  * match regardless of whether label separators match.
 198  *
 199  * @param s1                First source string.
 200  * @param length1           Length of first source string, or -1 if NUL-terminated.
 201  *
 202  * @param s2                Second source string.
 203  * @param length2           Length of second source string, or -1 if NUL-terminated.
 204  * @param options           A bit set of options:
 205  *
 206  *  - idnaref_UNASSIGNED        Unassigned values can be converted to ASCII for query operations
 207  *                          If TRUE unassigned values are treated as normal Unicode code points.
 208  *                          If FALSE the operation fails with U_UNASSIGNED_CODE_POINT_FOUND error code.
 209  *  - idnaref_USE_STD3_RULES    Use STD3 ASCII rules for host name syntax restrictions
 210  *                          If TRUE and the input does not statisfy STD3 rules, the operation
 211  *                          will fail with U_IDNA_STD3_ASCII_RULES_ERROR
 212  *
 213  * @param status            ICU error code in/out parameter.
 214  *                          Must fulfill U_SUCCESS before the function call.
 215  * @return <0 or 0 or >0 as usual for string comparisons
 216  */
 217 U_CFUNC int32_t U_EXPORT2
 218 idnaref_compare(  const UChar *s1, int32_t length1,
 219                 const UChar *s2, int32_t length2,
 220                 int32_t options,
 221                 UErrorCode* status);
 222
 223 #endif /* #if !UCONFIG_NO_IDNA */
 224
 225 #endif