1 /*

     2  *******************************************************************************

     3  *

     4  *   Copyright (C) 2003-2005, International Business Machines

     5  *   Corporation and others.  All Rights Reserved.

     6  *

     7  *******************************************************************************

     8  *   file name:  uidna.h

     9  *   encoding:   US-ASCII

    10  *   tab size:   8 (not used)

    11  *   indentation:4

    12  *

    13  *   created on: 2003feb1

    14  *   created by: Ram Viswanadha

    15  */

    16 

    17 #ifndef __UIDNA_H__

    18 #define __UIDNA_H__

    19 

    20 #include "unicode/utypes.h"

    21 

    22 #if !UCONFIG_NO_IDNA

    23 

    24 #include "unicode/parseerr.h"

    25   

    26 /**

    27  * \file

    28  * \brief C API: Internationalized Domain Names in Applications Tranformation

    29  *

    30  * UIDNA API implements the IDNA protocol as defined in the IDNA RFC 

    31  * (http://www.ietf.org/rfc/rfc3490.txt).

    32  * The RFC defines 2 operations: ToASCII and ToUnicode. Domain labels 

    33  * containing non-ASCII code points are required to be processed by

    34  * ToASCII operation before passing it to resolver libraries. Domain names

    35  * that are obtained from resolver libraries are required to be processed by

    36  * ToUnicode operation before displaying the domain name to the user.

    37  * IDNA requires that implementations process input strings with Nameprep

    38  * (http://www.ietf.org/rfc/rfc3491.txt), 

    39  * which is a profile of Stringprep (http://www.ietf.org/rfc/rfc3454.txt), 

    40  * and then with Punycode (http://www.ietf.org/rfc/rfc3492.txt). 

    41  * Implementations of IDNA MUST fully implement Nameprep and Punycode; 

    42  * neither Nameprep nor Punycode are optional.

    43  * The input and output of ToASCII and ToUnicode operations are Unicode 

    44  * and are designed to be chainable, i.e., applying ToASCII or ToUnicode operations

    45  * multiple times to an input string will yield the same result as applying the operation

    46  * once.

    47  * ToUnicode(ToUnicode(ToUnicode...(ToUnicode(string)))) == ToUnicode(string) 

    48  * ToASCII(ToASCII(ToASCII...(ToASCII(string))) == ToASCII(string).

    49  *

    50  */

    51 

    52 #ifndef U_HIDE_DRAFT_API

    53 

    54 /** 

    55  * Option to prohibit processing of unassigned codepoints in the input and

    56  * do not check if the input conforms to STD-3 ASCII rules.

    57  * 

    58  * @see  uidna_toASCII uidna_toUnicode

    59  * @stable ICU 2.6

    60  */

    61 #define UIDNA_DEFAULT          0x0000

    62 /** 

    63  * Option to allow processing of unassigned codepoints in the input

    64  * 

    65  * @see  uidna_toASCII uidna_toUnicode

    66  * @stable ICU 2.6

    67  */

    68 #define UIDNA_ALLOW_UNASSIGNED 0x0001

    69 /** 

    70  * Option to check if input conforms to STD-3 ASCII rules

    71  * 

    72  * @see  uidna_toASCII uidna_toUnicode

    73  * @stable ICU 2.6

    74  */

    75 #define UIDNA_USE_STD3_RULES   0x0002

    76 

    77 #endif /*U_HIDE_DRAFT_API*/

    78     

    79 /**

    80  * This function implements the ToASCII operation as defined in the IDNA RFC.

    81  * This operation is done on <b>single labels</b> before sending it to something that expects

    82  * ASCII names. A label is an individual part of a domain name. Labels are usually

    83  * separated by dots; e.g." "www.example.com" is composed of 3 labels 

    84  * "www","example", and "com".

    85  *

    86  *

    87  * @param src               Input UChar array containing label in Unicode.

    88  * @param srcLength         Number of UChars in src, or -1 if NUL-terminated.

    89  * @param dest              Output UChar array with ASCII (ACE encoded) label.

    90  * @param destCapacity      Size of dest.

    91  * @param options           A bit set of options:

    92  *

    93  *  - UIDNA_DEFAULT             Use default options, i.e., do not process unassigned code points

    94  *                              and do not use STD3 ASCII rules

    95  *                              If unassigned code points are found the operation fails with 

    96  *                              U_UNASSIGNED_ERROR error code.

    97  *

    98  *  - UIDNA_ALLOW_UNASSIGNED    Unassigned values can be converted to ASCII for query operations

    99  *                              If this option is set, the unassigned code points are in the input 

   100  *                              are treated as normal Unicode code points.

   101  *                          

   102  *  - UIDNA_USE_STD3_RULES      Use STD3 ASCII rules for host name syntax restrictions

   103  *                              If this option is set and the input does not satisfy STD3 rules,  

   104  *                              the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR

   105  *

   106  * @param parseError        Pointer to UParseError struct to receive information on position 

   107  *                          of error if an error is encountered. Can be NULL.

   108  * @param status            ICU in/out error code parameter.

   109  *                          U_INVALID_CHAR_FOUND if src contains

   110  *                          unmatched single surrogates.

   111  *                          U_INDEX_OUTOFBOUNDS_ERROR if src contains

   112  *                          too many code points.

   113  *                          U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough

   114  * @return                  Number of ASCII characters converted.

   115  * @stable ICU 2.6

   116  */

   117 U_STABLE int32_t U_EXPORT2

   118 uidna_toASCII(const UChar* src, int32_t srcLength, 

   119               UChar* dest, int32_t destCapacity,

   120               int32_t options,

   121               UParseError* parseError,

   122               UErrorCode* status);

   123 

   124 

   125 /**

   126  * This function implements the ToUnicode operation as defined in the IDNA RFC.

   127  * This operation is done on <b>single labels</b> before sending it to something that expects

   128  * Unicode names. A label is an individual part of a domain name. Labels are usually

   129  * separated by dots; for e.g." "www.example.com" is composed of 3 labels 

   130  * "www","example", and "com".

   131  *

   132  * @param src               Input UChar array containing ASCII (ACE encoded) label.

   133  * @param srcLength         Number of UChars in src, or -1 if NUL-terminated.

   134  * @param dest Output       Converted UChar array containing Unicode equivalent of label.

   135  * @param destCapacity      Size of dest.

   136  * @param options           A bit set of options:

   137  *  

   138  *  - UIDNA_DEFAULT             Use default options, i.e., do not process unassigned code points

   139  *                              and do not use STD3 ASCII rules

   140  *                              If unassigned code points are found the operation fails with 

   141  *                              U_UNASSIGNED_ERROR error code.

   142  *

   143  *  - UIDNA_ALLOW_UNASSIGNED      Unassigned values can be converted to ASCII for query operations

   144  *                              If this option is set, the unassigned code points are in the input 

   145  *                              are treated as normal Unicode code points. <b> Note: </b> This option is 

   146  *                              required on toUnicode operation because the RFC mandates 

   147  *                              verification of decoded ACE input by applying toASCII and comparing

   148  *                              its output with source

   149  *

   150  *                          

   151  *                          

   152  *  - UIDNA_USE_STD3_RULES      Use STD3 ASCII rules for host name syntax restrictions

   153  *                              If this option is set and the input does not satisfy STD3 rules,  

   154  *                              the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR

   155  *

   156  * @param parseError        Pointer to UParseError struct to receive information on position 

   157  *                          of error if an error is encountered. Can be NULL.

   158  * @param status            ICU in/out error code parameter.

   159  *                          U_INVALID_CHAR_FOUND if src contains

   160  *                          unmatched single surrogates.

   161  *                          U_INDEX_OUTOFBOUNDS_ERROR if src contains

   162  *                          too many code points.

   163  *                          U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough

   164  * @return                  Number of Unicode characters converted.

   165  * @stable ICU 2.6

   166  */

   167 U_STABLE int32_t U_EXPORT2

   168 uidna_toUnicode(const UChar* src, int32_t srcLength,

   169                 UChar* dest, int32_t destCapacity,

   170                 int32_t options,

   171                 UParseError* parseError,

   172                 UErrorCode* status);

   173 

   174 

   175 /**

   176  * Convenience function that implements the IDNToASCII operation as defined in the IDNA RFC.

   177  * This operation is done on complete domain names, e.g: "www.example.com". 

   178  * It is important to note that this operation can fail. If it fails, then the input 

   179  * domain name cannot be used as an Internationalized Domain Name and the application

   180  * should have methods defined to deal with the failure.

   181  * 

   182  * <b>Note:</b> IDNA RFC specifies that a conformant application should divide a domain name

   183  * into separate labels, decide whether to apply allowUnassigned and useSTD3ASCIIRules on each, 

   184  * and then convert. This function does not offer that level of granularity. The options once  

   185  * set will apply to all labels in the domain name

   186  *

   187  * @param src               Input UChar array containing IDN in Unicode.

   188  * @param srcLength         Number of UChars in src, or -1 if NUL-terminated.

   189  * @param dest              Output UChar array with ASCII (ACE encoded) IDN.

   190  * @param destCapacity      Size of dest.

   191  * @param options           A bit set of options:

   192  *  

   193  *  - UIDNA_DEFAULT             Use default options, i.e., do not process unassigned code points

   194  *                              and do not use STD3 ASCII rules

   195  *                              If unassigned code points are found the operation fails with 

   196  *                              U_UNASSIGNED_CODE_POINT_FOUND error code.

   197  *

   198  *  - UIDNA_ALLOW_UNASSIGNED    Unassigned values can be converted to ASCII for query operations

   199  *                              If this option is set, the unassigned code points are in the input 

   200  *                              are treated as normal Unicode code points.

   201  *                          

   202  *  - UIDNA_USE_STD3_RULES      Use STD3 ASCII rules for host name syntax restrictions

   203  *                              If this option is set and the input does not satisfy STD3 rules,  

   204  *                              the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR

   205  * 

   206  * @param parseError        Pointer to UParseError struct to receive information on position 

   207  *                          of error if an error is encountered. Can be NULL.

   208  * @param status            ICU in/out error code parameter.

   209  *                          U_INVALID_CHAR_FOUND if src contains

   210  *                          unmatched single surrogates.

   211  *                          U_INDEX_OUTOFBOUNDS_ERROR if src contains

   212  *                          too many code points.

   213  *                          U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough

   214  * @return                  Number of ASCII characters converted.

   215  * @stable ICU 2.6

   216  */

   217 U_STABLE int32_t U_EXPORT2

   218 uidna_IDNToASCII(  const UChar* src, int32_t srcLength,

   219                    UChar* dest, int32_t destCapacity,

   220                    int32_t options,

   221                    UParseError* parseError,

   222                    UErrorCode* status);

   223 

   224 /**

   225  * Convenience function that implements the IDNToUnicode operation as defined in the IDNA RFC.

   226  * This operation is done on complete domain names, e.g: "www.example.com". 

   227  *

   228  * <b>Note:</b> IDNA RFC specifies that a conformant application should divide a domain name

   229  * into separate labels, decide whether to apply allowUnassigned and useSTD3ASCIIRules on each, 

   230  * and then convert. This function does not offer that level of granularity. The options once  

   231  * set will apply to all labels in the domain name

   232  *

   233  * @param src               Input UChar array containing IDN in ASCII (ACE encoded) form.

   234  * @param srcLength         Number of UChars in src, or -1 if NUL-terminated.

   235  * @param dest Output       UChar array containing Unicode equivalent of source IDN.

   236  * @param destCapacity      Size of dest.

   237  * @param options           A bit set of options:

   238  *  

   239  *  - UIDNA_DEFAULT             Use default options, i.e., do not process unassigned code points

   240  *                              and do not use STD3 ASCII rules

   241  *                              If unassigned code points are found the operation fails with 

   242  *                              U_UNASSIGNED_CODE_POINT_FOUND error code.

   243  *

   244  *  - UIDNA_ALLOW_UNASSIGNED    Unassigned values can be converted to ASCII for query operations

   245  *                              If this option is set, the unassigned code points are in the input 

   246  *                              are treated as normal Unicode code points.

   247  *                          

   248  *  - UIDNA_USE_STD3_RULES      Use STD3 ASCII rules for host name syntax restrictions

   249  *                              If this option is set and the input does not satisfy STD3 rules,  

   250  *                              the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR

   251  *

   252  * @param parseError        Pointer to UParseError struct to receive information on position 

   253  *                          of error if an error is encountered. Can be NULL.

   254  * @param status            ICU in/out error code parameter.

   255  *                          U_INVALID_CHAR_FOUND if src contains

   256  *                          unmatched single surrogates.

   257  *                          U_INDEX_OUTOFBOUNDS_ERROR if src contains

   258  *                          too many code points.

   259  *                          U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough

   260  * @return                  Number of ASCII characters converted.

   261  * @stable ICU 2.6

   262  */

   263 U_STABLE int32_t U_EXPORT2

   264 uidna_IDNToUnicode(  const UChar* src, int32_t srcLength,

   265                      UChar* dest, int32_t destCapacity,

   266                      int32_t options,

   267                      UParseError* parseError,

   268                      UErrorCode* status);

   269 

   270 /**

   271  * Compare two IDN strings for equivalence.

   272  * This function splits the domain names into labels and compares them.

   273  * According to IDN RFC, whenever two labels are compared, they are 

   274  * considered equal if and only if their ASCII forms (obtained by 

   275  * applying toASCII) match using an case-insensitive ASCII comparison.

   276  * Two domain names are considered a match if and only if all labels 

   277  * match regardless of whether label separators match.

   278  *

   279  * @param s1                First source string.

   280  * @param length1           Length of first source string, or -1 if NUL-terminated.

   281  *

   282  * @param s2                Second source string.

   283  * @param length2           Length of second source string, or -1 if NUL-terminated.

   284  * @param options           A bit set of options:

   285  *  

   286  *  - UIDNA_DEFAULT             Use default options, i.e., do not process unassigned code points

   287  *                              and do not use STD3 ASCII rules

   288  *                              If unassigned code points are found the operation fails with 

   289  *                              U_UNASSIGNED_CODE_POINT_FOUND error code.

   290  *

   291  *  - UIDNA_ALLOW_UNASSIGNED    Unassigned values can be converted to ASCII for query operations

   292  *                              If this option is set, the unassigned code points are in the input 

   293  *                              are treated as normal Unicode code points.

   294  *                          

   295  *  - UIDNA_USE_STD3_RULES      Use STD3 ASCII rules for host name syntax restrictions

   296  *                              If this option is set and the input does not satisfy STD3 rules,  

   297  *                              the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR

   298  *

   299  * @param status            ICU error code in/out parameter.

   300  *                          Must fulfill U_SUCCESS before the function call.

   301  * @return <0 or 0 or >0 as usual for string comparisons

   302  * @stable ICU 2.6

   303  */

   304 U_STABLE int32_t U_EXPORT2

   305 uidna_compare(  const UChar *s1, int32_t length1,

   306                 const UChar *s2, int32_t length2,

   307                 int32_t options,

   308                 UErrorCode* status);

   309 

   310 #endif /* #if !UCONFIG_NO_IDNA */

   311 

   312 #endif