|
sl@0
|
1 |
/*
|
|
sl@0
|
2 |
*******************************************************************************
|
|
sl@0
|
3 |
*
|
|
sl@0
|
4 |
* Copyright (C) 2003-2005, International Business Machines
|
|
sl@0
|
5 |
* Corporation and others. All Rights Reserved.
|
|
sl@0
|
6 |
*
|
|
sl@0
|
7 |
*******************************************************************************
|
|
sl@0
|
8 |
* file name: uidna.h
|
|
sl@0
|
9 |
* encoding: US-ASCII
|
|
sl@0
|
10 |
* tab size: 8 (not used)
|
|
sl@0
|
11 |
* indentation:4
|
|
sl@0
|
12 |
*
|
|
sl@0
|
13 |
* created on: 2003feb1
|
|
sl@0
|
14 |
* created by: Ram Viswanadha
|
|
sl@0
|
15 |
*/
|
|
sl@0
|
16 |
|
|
sl@0
|
17 |
#ifndef __UIDNA_H__
|
|
sl@0
|
18 |
#define __UIDNA_H__
|
|
sl@0
|
19 |
|
|
sl@0
|
20 |
#include "unicode/utypes.h"
|
|
sl@0
|
21 |
|
|
sl@0
|
22 |
#if !UCONFIG_NO_IDNA
|
|
sl@0
|
23 |
|
|
sl@0
|
24 |
#include "unicode/parseerr.h"
|
|
sl@0
|
25 |
|
|
sl@0
|
26 |
/**
|
|
sl@0
|
27 |
* \file
|
|
sl@0
|
28 |
* \brief C API: Internationalized Domain Names in Applications Tranformation
|
|
sl@0
|
29 |
*
|
|
sl@0
|
30 |
* UIDNA API implements the IDNA protocol as defined in the IDNA RFC
|
|
sl@0
|
31 |
* (http://www.ietf.org/rfc/rfc3490.txt).
|
|
sl@0
|
32 |
* The RFC defines 2 operations: ToASCII and ToUnicode. Domain labels
|
|
sl@0
|
33 |
* containing non-ASCII code points are required to be processed by
|
|
sl@0
|
34 |
* ToASCII operation before passing it to resolver libraries. Domain names
|
|
sl@0
|
35 |
* that are obtained from resolver libraries are required to be processed by
|
|
sl@0
|
36 |
* ToUnicode operation before displaying the domain name to the user.
|
|
sl@0
|
37 |
* IDNA requires that implementations process input strings with Nameprep
|
|
sl@0
|
38 |
* (http://www.ietf.org/rfc/rfc3491.txt),
|
|
sl@0
|
39 |
* which is a profile of Stringprep (http://www.ietf.org/rfc/rfc3454.txt),
|
|
sl@0
|
40 |
* and then with Punycode (http://www.ietf.org/rfc/rfc3492.txt).
|
|
sl@0
|
41 |
* Implementations of IDNA MUST fully implement Nameprep and Punycode;
|
|
sl@0
|
42 |
* neither Nameprep nor Punycode are optional.
|
|
sl@0
|
43 |
* The input and output of ToASCII and ToUnicode operations are Unicode
|
|
sl@0
|
44 |
* and are designed to be chainable, i.e., applying ToASCII or ToUnicode operations
|
|
sl@0
|
45 |
* multiple times to an input string will yield the same result as applying the operation
|
|
sl@0
|
46 |
* once.
|
|
sl@0
|
47 |
* ToUnicode(ToUnicode(ToUnicode...(ToUnicode(string)))) == ToUnicode(string)
|
|
sl@0
|
48 |
* ToASCII(ToASCII(ToASCII...(ToASCII(string))) == ToASCII(string).
|
|
sl@0
|
49 |
*
|
|
sl@0
|
50 |
*/
|
|
sl@0
|
51 |
|
|
sl@0
|
52 |
#ifndef U_HIDE_DRAFT_API
|
|
sl@0
|
53 |
|
|
sl@0
|
54 |
/**
|
|
sl@0
|
55 |
* Option to prohibit processing of unassigned codepoints in the input and
|
|
sl@0
|
56 |
* do not check if the input conforms to STD-3 ASCII rules.
|
|
sl@0
|
57 |
*
|
|
sl@0
|
58 |
* @see uidna_toASCII uidna_toUnicode
|
|
sl@0
|
59 |
* @stable ICU 2.6
|
|
sl@0
|
60 |
*/
|
|
sl@0
|
61 |
#define UIDNA_DEFAULT 0x0000
|
|
sl@0
|
62 |
/**
|
|
sl@0
|
63 |
* Option to allow processing of unassigned codepoints in the input
|
|
sl@0
|
64 |
*
|
|
sl@0
|
65 |
* @see uidna_toASCII uidna_toUnicode
|
|
sl@0
|
66 |
* @stable ICU 2.6
|
|
sl@0
|
67 |
*/
|
|
sl@0
|
68 |
#define UIDNA_ALLOW_UNASSIGNED 0x0001
|
|
sl@0
|
69 |
/**
|
|
sl@0
|
70 |
* Option to check if input conforms to STD-3 ASCII rules
|
|
sl@0
|
71 |
*
|
|
sl@0
|
72 |
* @see uidna_toASCII uidna_toUnicode
|
|
sl@0
|
73 |
* @stable ICU 2.6
|
|
sl@0
|
74 |
*/
|
|
sl@0
|
75 |
#define UIDNA_USE_STD3_RULES 0x0002
|
|
sl@0
|
76 |
|
|
sl@0
|
77 |
#endif /*U_HIDE_DRAFT_API*/
|
|
sl@0
|
78 |
|
|
sl@0
|
79 |
/**
|
|
sl@0
|
80 |
* This function implements the ToASCII operation as defined in the IDNA RFC.
|
|
sl@0
|
81 |
* This operation is done on <b>single labels</b> before sending it to something that expects
|
|
sl@0
|
82 |
* ASCII names. A label is an individual part of a domain name. Labels are usually
|
|
sl@0
|
83 |
* separated by dots; e.g." "www.example.com" is composed of 3 labels
|
|
sl@0
|
84 |
* "www","example", and "com".
|
|
sl@0
|
85 |
*
|
|
sl@0
|
86 |
*
|
|
sl@0
|
87 |
* @param src Input UChar array containing label in Unicode.
|
|
sl@0
|
88 |
* @param srcLength Number of UChars in src, or -1 if NUL-terminated.
|
|
sl@0
|
89 |
* @param dest Output UChar array with ASCII (ACE encoded) label.
|
|
sl@0
|
90 |
* @param destCapacity Size of dest.
|
|
sl@0
|
91 |
* @param options A bit set of options:
|
|
sl@0
|
92 |
*
|
|
sl@0
|
93 |
* - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points
|
|
sl@0
|
94 |
* and do not use STD3 ASCII rules
|
|
sl@0
|
95 |
* If unassigned code points are found the operation fails with
|
|
sl@0
|
96 |
* U_UNASSIGNED_ERROR error code.
|
|
sl@0
|
97 |
*
|
|
sl@0
|
98 |
* - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations
|
|
sl@0
|
99 |
* If this option is set, the unassigned code points are in the input
|
|
sl@0
|
100 |
* are treated as normal Unicode code points.
|
|
sl@0
|
101 |
*
|
|
sl@0
|
102 |
* - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions
|
|
sl@0
|
103 |
* If this option is set and the input does not satisfy STD3 rules,
|
|
sl@0
|
104 |
* the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR
|
|
sl@0
|
105 |
*
|
|
sl@0
|
106 |
* @param parseError Pointer to UParseError struct to receive information on position
|
|
sl@0
|
107 |
* of error if an error is encountered. Can be NULL.
|
|
sl@0
|
108 |
* @param status ICU in/out error code parameter.
|
|
sl@0
|
109 |
* U_INVALID_CHAR_FOUND if src contains
|
|
sl@0
|
110 |
* unmatched single surrogates.
|
|
sl@0
|
111 |
* U_INDEX_OUTOFBOUNDS_ERROR if src contains
|
|
sl@0
|
112 |
* too many code points.
|
|
sl@0
|
113 |
* U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
|
|
sl@0
|
114 |
* @return Number of ASCII characters converted.
|
|
sl@0
|
115 |
* @stable ICU 2.6
|
|
sl@0
|
116 |
*/
|
|
sl@0
|
117 |
U_STABLE int32_t U_EXPORT2
|
|
sl@0
|
118 |
uidna_toASCII(const UChar* src, int32_t srcLength,
|
|
sl@0
|
119 |
UChar* dest, int32_t destCapacity,
|
|
sl@0
|
120 |
int32_t options,
|
|
sl@0
|
121 |
UParseError* parseError,
|
|
sl@0
|
122 |
UErrorCode* status);
|
|
sl@0
|
123 |
|
|
sl@0
|
124 |
|
|
sl@0
|
125 |
/**
|
|
sl@0
|
126 |
* This function implements the ToUnicode operation as defined in the IDNA RFC.
|
|
sl@0
|
127 |
* This operation is done on <b>single labels</b> before sending it to something that expects
|
|
sl@0
|
128 |
* Unicode names. A label is an individual part of a domain name. Labels are usually
|
|
sl@0
|
129 |
* separated by dots; for e.g." "www.example.com" is composed of 3 labels
|
|
sl@0
|
130 |
* "www","example", and "com".
|
|
sl@0
|
131 |
*
|
|
sl@0
|
132 |
* @param src Input UChar array containing ASCII (ACE encoded) label.
|
|
sl@0
|
133 |
* @param srcLength Number of UChars in src, or -1 if NUL-terminated.
|
|
sl@0
|
134 |
* @param dest Output Converted UChar array containing Unicode equivalent of label.
|
|
sl@0
|
135 |
* @param destCapacity Size of dest.
|
|
sl@0
|
136 |
* @param options A bit set of options:
|
|
sl@0
|
137 |
*
|
|
sl@0
|
138 |
* - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points
|
|
sl@0
|
139 |
* and do not use STD3 ASCII rules
|
|
sl@0
|
140 |
* If unassigned code points are found the operation fails with
|
|
sl@0
|
141 |
* U_UNASSIGNED_ERROR error code.
|
|
sl@0
|
142 |
*
|
|
sl@0
|
143 |
* - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations
|
|
sl@0
|
144 |
* If this option is set, the unassigned code points are in the input
|
|
sl@0
|
145 |
* are treated as normal Unicode code points. <b> Note: </b> This option is
|
|
sl@0
|
146 |
* required on toUnicode operation because the RFC mandates
|
|
sl@0
|
147 |
* verification of decoded ACE input by applying toASCII and comparing
|
|
sl@0
|
148 |
* its output with source
|
|
sl@0
|
149 |
*
|
|
sl@0
|
150 |
*
|
|
sl@0
|
151 |
*
|
|
sl@0
|
152 |
* - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions
|
|
sl@0
|
153 |
* If this option is set and the input does not satisfy STD3 rules,
|
|
sl@0
|
154 |
* the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR
|
|
sl@0
|
155 |
*
|
|
sl@0
|
156 |
* @param parseError Pointer to UParseError struct to receive information on position
|
|
sl@0
|
157 |
* of error if an error is encountered. Can be NULL.
|
|
sl@0
|
158 |
* @param status ICU in/out error code parameter.
|
|
sl@0
|
159 |
* U_INVALID_CHAR_FOUND if src contains
|
|
sl@0
|
160 |
* unmatched single surrogates.
|
|
sl@0
|
161 |
* U_INDEX_OUTOFBOUNDS_ERROR if src contains
|
|
sl@0
|
162 |
* too many code points.
|
|
sl@0
|
163 |
* U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
|
|
sl@0
|
164 |
* @return Number of Unicode characters converted.
|
|
sl@0
|
165 |
* @stable ICU 2.6
|
|
sl@0
|
166 |
*/
|
|
sl@0
|
167 |
U_STABLE int32_t U_EXPORT2
|
|
sl@0
|
168 |
uidna_toUnicode(const UChar* src, int32_t srcLength,
|
|
sl@0
|
169 |
UChar* dest, int32_t destCapacity,
|
|
sl@0
|
170 |
int32_t options,
|
|
sl@0
|
171 |
UParseError* parseError,
|
|
sl@0
|
172 |
UErrorCode* status);
|
|
sl@0
|
173 |
|
|
sl@0
|
174 |
|
|
sl@0
|
175 |
/**
|
|
sl@0
|
176 |
* Convenience function that implements the IDNToASCII operation as defined in the IDNA RFC.
|
|
sl@0
|
177 |
* This operation is done on complete domain names, e.g: "www.example.com".
|
|
sl@0
|
178 |
* It is important to note that this operation can fail. If it fails, then the input
|
|
sl@0
|
179 |
* domain name cannot be used as an Internationalized Domain Name and the application
|
|
sl@0
|
180 |
* should have methods defined to deal with the failure.
|
|
sl@0
|
181 |
*
|
|
sl@0
|
182 |
* <b>Note:</b> IDNA RFC specifies that a conformant application should divide a domain name
|
|
sl@0
|
183 |
* into separate labels, decide whether to apply allowUnassigned and useSTD3ASCIIRules on each,
|
|
sl@0
|
184 |
* and then convert. This function does not offer that level of granularity. The options once
|
|
sl@0
|
185 |
* set will apply to all labels in the domain name
|
|
sl@0
|
186 |
*
|
|
sl@0
|
187 |
* @param src Input UChar array containing IDN in Unicode.
|
|
sl@0
|
188 |
* @param srcLength Number of UChars in src, or -1 if NUL-terminated.
|
|
sl@0
|
189 |
* @param dest Output UChar array with ASCII (ACE encoded) IDN.
|
|
sl@0
|
190 |
* @param destCapacity Size of dest.
|
|
sl@0
|
191 |
* @param options A bit set of options:
|
|
sl@0
|
192 |
*
|
|
sl@0
|
193 |
* - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points
|
|
sl@0
|
194 |
* and do not use STD3 ASCII rules
|
|
sl@0
|
195 |
* If unassigned code points are found the operation fails with
|
|
sl@0
|
196 |
* U_UNASSIGNED_CODE_POINT_FOUND error code.
|
|
sl@0
|
197 |
*
|
|
sl@0
|
198 |
* - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations
|
|
sl@0
|
199 |
* If this option is set, the unassigned code points are in the input
|
|
sl@0
|
200 |
* are treated as normal Unicode code points.
|
|
sl@0
|
201 |
*
|
|
sl@0
|
202 |
* - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions
|
|
sl@0
|
203 |
* If this option is set and the input does not satisfy STD3 rules,
|
|
sl@0
|
204 |
* the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR
|
|
sl@0
|
205 |
*
|
|
sl@0
|
206 |
* @param parseError Pointer to UParseError struct to receive information on position
|
|
sl@0
|
207 |
* of error if an error is encountered. Can be NULL.
|
|
sl@0
|
208 |
* @param status ICU in/out error code parameter.
|
|
sl@0
|
209 |
* U_INVALID_CHAR_FOUND if src contains
|
|
sl@0
|
210 |
* unmatched single surrogates.
|
|
sl@0
|
211 |
* U_INDEX_OUTOFBOUNDS_ERROR if src contains
|
|
sl@0
|
212 |
* too many code points.
|
|
sl@0
|
213 |
* U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
|
|
sl@0
|
214 |
* @return Number of ASCII characters converted.
|
|
sl@0
|
215 |
* @stable ICU 2.6
|
|
sl@0
|
216 |
*/
|
|
sl@0
|
217 |
U_STABLE int32_t U_EXPORT2
|
|
sl@0
|
218 |
uidna_IDNToASCII( const UChar* src, int32_t srcLength,
|
|
sl@0
|
219 |
UChar* dest, int32_t destCapacity,
|
|
sl@0
|
220 |
int32_t options,
|
|
sl@0
|
221 |
UParseError* parseError,
|
|
sl@0
|
222 |
UErrorCode* status);
|
|
sl@0
|
223 |
|
|
sl@0
|
224 |
/**
|
|
sl@0
|
225 |
* Convenience function that implements the IDNToUnicode operation as defined in the IDNA RFC.
|
|
sl@0
|
226 |
* This operation is done on complete domain names, e.g: "www.example.com".
|
|
sl@0
|
227 |
*
|
|
sl@0
|
228 |
* <b>Note:</b> IDNA RFC specifies that a conformant application should divide a domain name
|
|
sl@0
|
229 |
* into separate labels, decide whether to apply allowUnassigned and useSTD3ASCIIRules on each,
|
|
sl@0
|
230 |
* and then convert. This function does not offer that level of granularity. The options once
|
|
sl@0
|
231 |
* set will apply to all labels in the domain name
|
|
sl@0
|
232 |
*
|
|
sl@0
|
233 |
* @param src Input UChar array containing IDN in ASCII (ACE encoded) form.
|
|
sl@0
|
234 |
* @param srcLength Number of UChars in src, or -1 if NUL-terminated.
|
|
sl@0
|
235 |
* @param dest Output UChar array containing Unicode equivalent of source IDN.
|
|
sl@0
|
236 |
* @param destCapacity Size of dest.
|
|
sl@0
|
237 |
* @param options A bit set of options:
|
|
sl@0
|
238 |
*
|
|
sl@0
|
239 |
* - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points
|
|
sl@0
|
240 |
* and do not use STD3 ASCII rules
|
|
sl@0
|
241 |
* If unassigned code points are found the operation fails with
|
|
sl@0
|
242 |
* U_UNASSIGNED_CODE_POINT_FOUND error code.
|
|
sl@0
|
243 |
*
|
|
sl@0
|
244 |
* - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations
|
|
sl@0
|
245 |
* If this option is set, the unassigned code points are in the input
|
|
sl@0
|
246 |
* are treated as normal Unicode code points.
|
|
sl@0
|
247 |
*
|
|
sl@0
|
248 |
* - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions
|
|
sl@0
|
249 |
* If this option is set and the input does not satisfy STD3 rules,
|
|
sl@0
|
250 |
* the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR
|
|
sl@0
|
251 |
*
|
|
sl@0
|
252 |
* @param parseError Pointer to UParseError struct to receive information on position
|
|
sl@0
|
253 |
* of error if an error is encountered. Can be NULL.
|
|
sl@0
|
254 |
* @param status ICU in/out error code parameter.
|
|
sl@0
|
255 |
* U_INVALID_CHAR_FOUND if src contains
|
|
sl@0
|
256 |
* unmatched single surrogates.
|
|
sl@0
|
257 |
* U_INDEX_OUTOFBOUNDS_ERROR if src contains
|
|
sl@0
|
258 |
* too many code points.
|
|
sl@0
|
259 |
* U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
|
|
sl@0
|
260 |
* @return Number of ASCII characters converted.
|
|
sl@0
|
261 |
* @stable ICU 2.6
|
|
sl@0
|
262 |
*/
|
|
sl@0
|
263 |
U_STABLE int32_t U_EXPORT2
|
|
sl@0
|
264 |
uidna_IDNToUnicode( const UChar* src, int32_t srcLength,
|
|
sl@0
|
265 |
UChar* dest, int32_t destCapacity,
|
|
sl@0
|
266 |
int32_t options,
|
|
sl@0
|
267 |
UParseError* parseError,
|
|
sl@0
|
268 |
UErrorCode* status);
|
|
sl@0
|
269 |
|
|
sl@0
|
270 |
/**
|
|
sl@0
|
271 |
* Compare two IDN strings for equivalence.
|
|
sl@0
|
272 |
* This function splits the domain names into labels and compares them.
|
|
sl@0
|
273 |
* According to IDN RFC, whenever two labels are compared, they are
|
|
sl@0
|
274 |
* considered equal if and only if their ASCII forms (obtained by
|
|
sl@0
|
275 |
* applying toASCII) match using an case-insensitive ASCII comparison.
|
|
sl@0
|
276 |
* Two domain names are considered a match if and only if all labels
|
|
sl@0
|
277 |
* match regardless of whether label separators match.
|
|
sl@0
|
278 |
*
|
|
sl@0
|
279 |
* @param s1 First source string.
|
|
sl@0
|
280 |
* @param length1 Length of first source string, or -1 if NUL-terminated.
|
|
sl@0
|
281 |
*
|
|
sl@0
|
282 |
* @param s2 Second source string.
|
|
sl@0
|
283 |
* @param length2 Length of second source string, or -1 if NUL-terminated.
|
|
sl@0
|
284 |
* @param options A bit set of options:
|
|
sl@0
|
285 |
*
|
|
sl@0
|
286 |
* - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points
|
|
sl@0
|
287 |
* and do not use STD3 ASCII rules
|
|
sl@0
|
288 |
* If unassigned code points are found the operation fails with
|
|
sl@0
|
289 |
* U_UNASSIGNED_CODE_POINT_FOUND error code.
|
|
sl@0
|
290 |
*
|
|
sl@0
|
291 |
* - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations
|
|
sl@0
|
292 |
* If this option is set, the unassigned code points are in the input
|
|
sl@0
|
293 |
* are treated as normal Unicode code points.
|
|
sl@0
|
294 |
*
|
|
sl@0
|
295 |
* - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions
|
|
sl@0
|
296 |
* If this option is set and the input does not satisfy STD3 rules,
|
|
sl@0
|
297 |
* the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR
|
|
sl@0
|
298 |
*
|
|
sl@0
|
299 |
* @param status ICU error code in/out parameter.
|
|
sl@0
|
300 |
* Must fulfill U_SUCCESS before the function call.
|
|
sl@0
|
301 |
* @return <0 or 0 or >0 as usual for string comparisons
|
|
sl@0
|
302 |
* @stable ICU 2.6
|
|
sl@0
|
303 |
*/
|
|
sl@0
|
304 |
U_STABLE int32_t U_EXPORT2
|
|
sl@0
|
305 |
uidna_compare( const UChar *s1, int32_t length1,
|
|
sl@0
|
306 |
const UChar *s2, int32_t length2,
|
|
sl@0
|
307 |
int32_t options,
|
|
sl@0
|
308 |
UErrorCode* status);
|
|
sl@0
|
309 |
|
|
sl@0
|
310 |
#endif /* #if !UCONFIG_NO_IDNA */
|
|
sl@0
|
311 |
|
|
sl@0
|
312 |
#endif
|