icuSources/common/unicode/uidna.h

   1 /*
   2  *******************************************************************************
   3  *
   4  *   Copyright (C) 2003-2004, 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  * UIDNA API implements the IDNA protocol as defined in the IDNA RFC
  29  * (http://www.ietf.org/rfc/rfc3490.txt).
  30  * The RFC defines 2 operations: ToASCII and ToUnicode. Domain labels
  31  * containing non-ASCII code points are required to be processed by
  32  * ToASCII operation before passing it to resolver libraries. Domain names
  33  * that are obtained from resolver libraries are required to be processed by
  34  * ToUnicode operation before displaying the domain name to the user.
  35  * IDNA requires that implementations process input strings with Nameprep
  36  * (http://www.ietf.org/rfc/rfc3491.txt),
  37  * which is a profile of Stringprep (http://www.ietf.org/rfc/rfc3454.txt),
  38  * and then with Punycode (http://www.ietf.org/rfc/rfc3492.txt).
  39  * Implementations of IDNA MUST fully implement Nameprep and Punycode;
  40  * neither Nameprep nor Punycode are optional.
  41  * The input and output of ToASCII and ToUnicode operations are Unicode
  42  * and are designed to be chainable, i.e., applying ToASCII or ToUnicode operations
  43  * multiple times to an input string will yield the same result as applying the operation
  44  * once.
  45  * ToUnicode(ToUnicode(ToUnicode...(ToUnicode(string)))) == ToUnicode(string)
  46  * ToASCII(ToASCII(ToASCII...(ToASCII(string))) == ToASCII(string).
  47  *
  48  */
  49
  50 #ifndef U_HIDE_DRAFT_API
  51
  52 /**
  53  * Option to prohibit processing of unassigned codepoints in the input and
  54  * do not check if the input conforms to STD-3 ASCII rules.
  55  *
  56  * @see  uidna_toASCII uidna_toUnicode
  57  * @stable ICU 2.6
  58  */
  59 #define UIDNA_DEFAULT          0x0000
  60 /**
  61  * Option to allow processing of unassigned codepoints in the input
  62  *
  63  * @see  uidna_toASCII uidna_toUnicode
  64  * @stable ICU 2.6
  65  */
  66 #define UIDNA_ALLOW_UNASSIGNED 0x0001
  67 /**
  68  * Option to check if input conforms to STD-3 ASCII rules
  69  *
  70  * @see  uidna_toASCII uidna_toUnicode
  71  * @stable ICU 2.6
  72  */
  73 #define UIDNA_USE_STD3_RULES   0x0002
  74
  75 #endif /*U_HIDE_DRAFT_API*/
  76
  77 /**
  78  * This function implements the ToASCII operation as defined in the IDNA RFC.
  79  * This operation is done on <b>single labels</b> before sending it to something that expects
  80  * ASCII names. A label is an individual part of a domain name. Labels are usually
  81  * separated by dots; e.g." "www.example.com" is composed of 3 labels
  82  * "www","example", and "com".
  83  *
  84  *
  85  * @param src               Input UChar array containing label in Unicode.
  86  * @param srcLength         Number of UChars in src, or -1 if NUL-terminated.
  87  * @param dest              Output UChar array with ASCII (ACE encoded) label.
  88  * @param destCapacity      Size of dest.
  89  * @param options           A bit set of options:
  90  *
  91  *  - UIDNA_DEFAULT             Use default options, i.e., do not process unassigned code points
  92  *                              and do not use STD3 ASCII rules
  93  *                              If unassigned code points are found the operation fails with
  94  *                              U_UNASSIGNED_ERROR error code.
  95  *
  96  *  - UIDNA_ALLOW_UNASSIGNED    Unassigned values can be converted to ASCII for query operations
  97  *                              If this option is set, the unassigned code points are in the input
  98  *                              are treated as normal Unicode code points.
  99  *
 100  *  - UIDNA_USE_STD3_RULES      Use STD3 ASCII rules for host name syntax restrictions
 101  *                              If this option is set and the input does not satisfy STD3 rules,
 102  *                              the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR
 103  *
 104  * @param parseError        Pointer to UParseError struct to receive information on position
 105  *                          of error if an error is encountered. Can be NULL.
 106  * @param status            ICU in/out error code parameter.
 107  *                          U_INVALID_CHAR_FOUND if src contains
 108  *                          unmatched single surrogates.
 109  *                          U_INDEX_OUTOFBOUNDS_ERROR if src contains
 110  *                          too many code points.
 111  *                          U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
 112  * @return                  Number of ASCII characters converted.
 113  * @stable ICU 2.6
 114  */
 115 U_STABLE int32_t U_EXPORT2
 116 uidna_toASCII(const UChar* src, int32_t srcLength,
 117               UChar* dest, int32_t destCapacity,
 118               int32_t options,
 119               UParseError* parseError,
 120               UErrorCode* status);
 121
 122
 123 /**
 124  * This function implements the ToUnicode operation as defined in the IDNA RFC.
 125  * This operation is done on <b>single labels</b> before sending it to something that expects
 126  * Unicode names. A label is an individual part of a domain name. Labels are usually
 127  * separated by dots; for e.g." "www.example.com" is composed of 3 labels
 128  * "www","example", and "com".
 129  *
 130  * @param src               Input UChar array containing ASCII (ACE encoded) label.
 131  * @param srcLength         Number of UChars in src, or -1 if NUL-terminated.
 132  * @param dest Output       Converted UChar array containing Unicode equivalent of label.
 133  * @param destCapacity      Size of dest.
 134  * @param options           A bit set of options:
 135  *
 136  *  - UIDNA_DEFAULT             Use default options, i.e., do not process unassigned code points
 137  *                              and do not use STD3 ASCII rules
 138  *                              If unassigned code points are found the operation fails with
 139  *                              U_UNASSIGNED_ERROR error code.
 140  *
 141  *  - UIDNA_ALLOW_UNASSIGNED      Unassigned values can be converted to ASCII for query operations
 142  *                              If this option is set, the unassigned code points are in the input
 143  *                              are treated as normal Unicode code points. <b> Note: </b> This option is
 144  *                              required on toUnicode operation because the RFC mandates
 145  *                              verification of decoded ACE input by applying toASCII and comparing
 146  *                              its output with source
 147  *
 148  *
 149  *
 150  *  - UIDNA_USE_STD3_RULES      Use STD3 ASCII rules for host name syntax restrictions
 151  *                              If this option is set and the input does not satisfy STD3 rules,
 152  *                              the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR
 153  *
 154  * @param parseError        Pointer to UParseError struct to receive information on position
 155  *                          of error if an error is encountered. Can be NULL.
 156  * @param status            ICU in/out error code parameter.
 157  *                          U_INVALID_CHAR_FOUND if src contains
 158  *                          unmatched single surrogates.
 159  *                          U_INDEX_OUTOFBOUNDS_ERROR if src contains
 160  *                          too many code points.
 161  *                          U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
 162  * @return                  Number of Unicode characters converted.
 163  * @stable ICU 2.6
 164  */
 165 U_STABLE int32_t U_EXPORT2
 166 uidna_toUnicode(const UChar* src, int32_t srcLength,
 167                 UChar* dest, int32_t destCapacity,
 168                 int32_t options,
 169                 UParseError* parseError,
 170                 UErrorCode* status);
 171
 172
 173 /**
 174  * Convenience function that implements the IDNToASCII operation as defined in the IDNA RFC.
 175  * This operation is done on complete domain names, e.g: "www.example.com".
 176  * It is important to note that this operation can fail. If it fails, then the input
 177  * domain name cannot be used as an Internationalized Domain Name and the application
 178  * should have methods defined to deal with the failure.
 179  *
 180  * <b>Note:</b> IDNA RFC specifies that a conformant application should divide a domain name
 181  * into separate labels, decide whether to apply allowUnassigned and useSTD3ASCIIRules on each,
 182  * and then convert. This function does not offer that level of granularity. The options once
 183  * set will apply to all labels in the domain name
 184  *
 185  * @param src               Input UChar array containing IDN in Unicode.
 186  * @param srcLength         Number of UChars in src, or -1 if NUL-terminated.
 187  * @param dest              Output UChar array with ASCII (ACE encoded) IDN.
 188  * @param destCapacity      Size of dest.
 189  * @param options           A bit set of options:
 190  *
 191  *  - UIDNA_DEFAULT             Use default options, i.e., do not process unassigned code points
 192  *                              and do not use STD3 ASCII rules
 193  *                              If unassigned code points are found the operation fails with
 194  *                              U_UNASSIGNED_CODE_POINT_FOUND error code.
 195  *
 196  *  - UIDNA_ALLOW_UNASSIGNED    Unassigned values can be converted to ASCII for query operations
 197  *                              If this option is set, the unassigned code points are in the input
 198  *                              are treated as normal Unicode code points.
 199  *
 200  *  - UIDNA_USE_STD3_RULES      Use STD3 ASCII rules for host name syntax restrictions
 201  *                              If this option is set and the input does not satisfy STD3 rules,
 202  *                              the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR
 203  *
 204  * @param parseError        Pointer to UParseError struct to receive information on position
 205  *                          of error if an error is encountered. Can be NULL.
 206  * @param status            ICU in/out error code parameter.
 207  *                          U_INVALID_CHAR_FOUND if src contains
 208  *                          unmatched single surrogates.
 209  *                          U_INDEX_OUTOFBOUNDS_ERROR if src contains
 210  *                          too many code points.
 211  *                          U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
 212  * @return                  Number of ASCII characters converted.
 213  * @stable ICU 2.6
 214  */
 215 U_STABLE int32_t U_EXPORT2
 216 uidna_IDNToASCII(  const UChar* src, int32_t srcLength,
 217                    UChar* dest, int32_t destCapacity,
 218                    int32_t options,
 219                    UParseError* parseError,
 220                    UErrorCode* status);
 221
 222 /**
 223  * Convenience function that implements the IDNToUnicode operation as defined in the IDNA RFC.
 224  * This operation is done on complete domain names, e.g: "www.example.com".
 225  *
 226  * <b>Note:</b> IDNA RFC specifies that a conformant application should divide a domain name
 227  * into separate labels, decide whether to apply allowUnassigned and useSTD3ASCIIRules on each,
 228  * and then convert. This function does not offer that level of granularity. The options once
 229  * set will apply to all labels in the domain name
 230  *
 231  * @param src               Input UChar array containing IDN in ASCII (ACE encoded) form.
 232  * @param srcLength         Number of UChars in src, or -1 if NUL-terminated.
 233  * @param dest Output       UChar array containing Unicode equivalent of source IDN.
 234  * @param destCapacity      Size of dest.
 235  * @param options           A bit set of options:
 236  *
 237  *  - UIDNA_DEFAULT             Use default options, i.e., do not process unassigned code points
 238  *                              and do not use STD3 ASCII rules
 239  *                              If unassigned code points are found the operation fails with
 240  *                              U_UNASSIGNED_CODE_POINT_FOUND error code.
 241  *
 242  *  - UIDNA_ALLOW_UNASSIGNED    Unassigned values can be converted to ASCII for query operations
 243  *                              If this option is set, the unassigned code points are in the input
 244  *                              are treated as normal Unicode code points.
 245  *
 246  *  - UIDNA_USE_STD3_RULES      Use STD3 ASCII rules for host name syntax restrictions
 247  *                              If this option is set and the input does not satisfy STD3 rules,
 248  *                              the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR
 249  *
 250  * @param parseError        Pointer to UParseError struct to receive information on position
 251  *                          of error if an error is encountered. Can be NULL.
 252  * @param status            ICU in/out error code parameter.
 253  *                          U_INVALID_CHAR_FOUND if src contains
 254  *                          unmatched single surrogates.
 255  *                          U_INDEX_OUTOFBOUNDS_ERROR if src contains
 256  *                          too many code points.
 257  *                          U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
 258  * @return                  Number of ASCII characters converted.
 259  * @stable ICU 2.6
 260  */
 261 U_STABLE int32_t U_EXPORT2
 262 uidna_IDNToUnicode(  const UChar* src, int32_t srcLength,
 263                      UChar* dest, int32_t destCapacity,
 264                      int32_t options,
 265                      UParseError* parseError,
 266                      UErrorCode* status);
 267
 268 /**
 269  * Compare two IDN strings for equivalence.
 270  * This function splits the domain names into labels and compares them.
 271  * According to IDN RFC, whenever two labels are compared, they are
 272  * considered equal if and only if their ASCII forms (obtained by
 273  * applying toASCII) match using an case-insensitive ASCII comparison.
 274  * Two domain names are considered a match if and only if all labels
 275  * match regardless of whether label separators match.
 276  *
 277  * @param s1                First source string.
 278  * @param length1           Length of first source string, or -1 if NUL-terminated.
 279  *
 280  * @param s2                Second source string.
 281  * @param length2           Length of second source string, or -1 if NUL-terminated.
 282  * @param options           A bit set of options:
 283  *
 284  *  - UIDNA_DEFAULT             Use default options, i.e., do not process unassigned code points
 285  *                              and do not use STD3 ASCII rules
 286  *                              If unassigned code points are found the operation fails with
 287  *                              U_UNASSIGNED_CODE_POINT_FOUND error code.
 288  *
 289  *  - UIDNA_ALLOW_UNASSIGNED    Unassigned values can be converted to ASCII for query operations
 290  *                              If this option is set, the unassigned code points are in the input
 291  *                              are treated as normal Unicode code points.
 292  *
 293  *  - UIDNA_USE_STD3_RULES      Use STD3 ASCII rules for host name syntax restrictions
 294  *                              If this option is set and the input does not satisfy STD3 rules,
 295  *                              the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR
 296  *
 297  * @param status            ICU error code in/out parameter.
 298  *                          Must fulfill U_SUCCESS before the function call.
 299  * @return <0 or 0 or >0 as usual for string comparisons
 300  * @stable ICU 2.6
 301  */
 302 U_STABLE int32_t U_EXPORT2
 303 uidna_compare(  const UChar *s1, int32_t length1,
 304                 const UChar *s2, int32_t length2,
 305                 int32_t options,
 306                 UErrorCode* status);
 307
 308 #endif /* #if !UCONFIG_NO_IDNA */
 309
 310 #endif