[apple/icu.git] / icuSources / common / unicode / uidna.h

/*
 *******************************************************************************
 *
 *   Copyright (C) 2003-2007, International Business Machines
 *   Corporation and others.  All Rights Reserved.
 *
 *******************************************************************************
 *   file name:  uidna.h
 *   encoding:   US-ASCII
 *   tab size:   8 (not used)
 *   indentation:4
 *
 *   created on: 2003feb1
 *   created by: Ram Viswanadha
 */

#ifndef __UIDNA_H__
#define __UIDNA_H__

#include "unicode/utypes.h"

#if !UCONFIG_NO_IDNA

#include "unicode/parseerr.h"
  
/**
 * \file
 * \brief C API: Internationalized Domain Names in Applications Tranformation
 *
 * UIDNA API implements the IDNA protocol as defined in the IDNA RFC 
 * (http://www.ietf.org/rfc/rfc3490.txt).
 * The RFC defines 2 operations: ToASCII and ToUnicode. Domain labels 
 * containing non-ASCII code points are required to be processed by
 * ToASCII operation before passing it to resolver libraries. Domain names
 * that are obtained from resolver libraries are required to be processed by
 * ToUnicode operation before displaying the domain name to the user.
 * IDNA requires that implementations process input strings with Nameprep
 * (http://www.ietf.org/rfc/rfc3491.txt), 
 * which is a profile of Stringprep (http://www.ietf.org/rfc/rfc3454.txt), 
 * and then with Punycode (http://www.ietf.org/rfc/rfc3492.txt). 
 * Implementations of IDNA MUST fully implement Nameprep and Punycode; 
 * neither Nameprep nor Punycode are optional.
 * The input and output of ToASCII and ToUnicode operations are Unicode 
 * and are designed to be chainable, i.e., applying ToASCII or ToUnicode operations
 * multiple times to an input string will yield the same result as applying the operation
 * once.
 * ToUnicode(ToUnicode(ToUnicode...(ToUnicode(string)))) == ToUnicode(string) 
 * ToASCII(ToASCII(ToASCII...(ToASCII(string))) == ToASCII(string).
 *
 */

/** 
 * Option to prohibit processing of unassigned codepoints in the input and
 * do not check if the input conforms to STD-3 ASCII rules.
 * 
 * @see  uidna_toASCII uidna_toUnicode
 * @stable ICU 2.6
 */
#define UIDNA_DEFAULT          0x0000
/** 
 * Option to allow processing of unassigned codepoints in the input
 * 
 * @see  uidna_toASCII uidna_toUnicode
 * @stable ICU 2.6
 */
#define UIDNA_ALLOW_UNASSIGNED 0x0001
/** 
 * Option to check if input conforms to STD-3 ASCII rules
 * 
 * @see  uidna_toASCII uidna_toUnicode
 * @stable ICU 2.6
 */
#define UIDNA_USE_STD3_RULES   0x0002

/**
 * This function implements the ToASCII operation as defined in the IDNA RFC.
 * This operation is done on <b>single labels</b> before sending it to something that expects
 * ASCII names. A label is an individual part of a domain name. Labels are usually
 * separated by dots; e.g." "www.example.com" is composed of 3 labels 
 * "www","example", and "com".
 *
 *
 * @param src               Input UChar array containing label in Unicode.
 * @param srcLength         Number of UChars in src, or -1 if NUL-terminated.
 * @param dest              Output UChar array with ASCII (ACE encoded) label.
 * @param destCapacity      Size of dest.
 * @param options           A bit set of options:
 *
 *  - UIDNA_DEFAULT             Use default options, i.e., do not process unassigned code points
 *                              and do not use STD3 ASCII rules
 *                              If unassigned code points are found the operation fails with 
 *                              U_UNASSIGNED_ERROR error code.
 *
 *  - UIDNA_ALLOW_UNASSIGNED    Unassigned values can be converted to ASCII for query operations
 *                              If this option is set, the unassigned code points are in the input 
 *                              are treated as normal Unicode code points.
 *                          
 *  - UIDNA_USE_STD3_RULES      Use STD3 ASCII rules for host name syntax restrictions
 *                              If this option is set and the input does not satisfy STD3 rules,  
 *                              the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR
 *
 * @param parseError        Pointer to UParseError struct to receive information on position 
 *                          of error if an error is encountered. Can be NULL.
 * @param status            ICU in/out error code parameter.
 *                          U_INVALID_CHAR_FOUND if src contains
 *                          unmatched single surrogates.
 *                          U_INDEX_OUTOFBOUNDS_ERROR if src contains
 *                          too many code points.
 *                          U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
 * @return The length of the result string, if successful - or in case of a buffer overflow,
 *         in which case it will be greater than destCapacity.
 * @stable ICU 2.6
 */
U_STABLE int32_t U_EXPORT2
uidna_toASCII(const UChar* src, int32_t srcLength, 
              UChar* dest, int32_t destCapacity,
              int32_t options,
              UParseError* parseError,
              UErrorCode* status);


/**
 * This function implements the ToUnicode operation as defined in the IDNA RFC.
 * This operation is done on <b>single labels</b> before sending it to something that expects
 * Unicode names. A label is an individual part of a domain name. Labels are usually
 * separated by dots; for e.g." "www.example.com" is composed of 3 labels 
 * "www","example", and "com".
 *
 * @param src               Input UChar array containing ASCII (ACE encoded) label.
 * @param srcLength         Number of UChars in src, or -1 if NUL-terminated.
 * @param dest Output       Converted UChar array containing Unicode equivalent of label.
 * @param destCapacity      Size of dest.
 * @param options           A bit set of options:
 *  
 *  - UIDNA_DEFAULT             Use default options, i.e., do not process unassigned code points
 *                              and do not use STD3 ASCII rules
 *                              If unassigned code points are found the operation fails with 
 *                              U_UNASSIGNED_ERROR error code.
 *
 *  - UIDNA_ALLOW_UNASSIGNED      Unassigned values can be converted to ASCII for query operations
 *                              If this option is set, the unassigned code points are in the input 
 *                              are treated as normal Unicode code points. <b> Note: </b> This option is 
 *                              required on toUnicode operation because the RFC mandates 
 *                              verification of decoded ACE input by applying toASCII and comparing
 *                              its output with source
 *
 *                          
 *                          
 *  - UIDNA_USE_STD3_RULES      Use STD3 ASCII rules for host name syntax restrictions
 *                              If this option is set and the input does not satisfy STD3 rules,  
 *                              the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR
 *
 * @param parseError        Pointer to UParseError struct to receive information on position 
 *                          of error if an error is encountered. Can be NULL.
 * @param status            ICU in/out error code parameter.
 *                          U_INVALID_CHAR_FOUND if src contains
 *                          unmatched single surrogates.
 *                          U_INDEX_OUTOFBOUNDS_ERROR if src contains
 *                          too many code points.
 *                          U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
 * @return The length of the result string, if successful - or in case of a buffer overflow,
 *         in which case it will be greater than destCapacity.
 * @stable ICU 2.6
 */
U_STABLE int32_t U_EXPORT2
uidna_toUnicode(const UChar* src, int32_t srcLength,
                UChar* dest, int32_t destCapacity,
                int32_t options,
                UParseError* parseError,
                UErrorCode* status);


/**
 * Convenience function that implements the IDNToASCII operation as defined in the IDNA RFC.
 * This operation is done on complete domain names, e.g: "www.example.com". 
 * It is important to note that this operation can fail. If it fails, then the input 
 * domain name cannot be used as an Internationalized Domain Name and the application
 * should have methods defined to deal with the failure.
 * 
 * <b>Note:</b> IDNA RFC specifies that a conformant application should divide a domain name
 * into separate labels, decide whether to apply allowUnassigned and useSTD3ASCIIRules on each, 
 * and then convert. This function does not offer that level of granularity. The options once  
 * set will apply to all labels in the domain name
 *
 * @param src               Input UChar array containing IDN in Unicode.
 * @param srcLength         Number of UChars in src, or -1 if NUL-terminated.
 * @param dest              Output UChar array with ASCII (ACE encoded) IDN.
 * @param destCapacity      Size of dest.
 * @param options           A bit set of options:
 *  
 *  - UIDNA_DEFAULT             Use default options, i.e., do not process unassigned code points
 *                              and do not use STD3 ASCII rules
 *                              If unassigned code points are found the operation fails with 
 *                              U_UNASSIGNED_CODE_POINT_FOUND error code.
 *
 *  - UIDNA_ALLOW_UNASSIGNED    Unassigned values can be converted to ASCII for query operations
 *                              If this option is set, the unassigned code points are in the input 
 *                              are treated as normal Unicode code points.
 *                          
 *  - UIDNA_USE_STD3_RULES      Use STD3 ASCII rules for host name syntax restrictions
 *                              If this option is set and the input does not satisfy STD3 rules,  
 *                              the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR
 * 
 * @param parseError        Pointer to UParseError struct to receive information on position 
 *                          of error if an error is encountered. Can be NULL.
 * @param status            ICU in/out error code parameter.
 *                          U_INVALID_CHAR_FOUND if src contains
 *                          unmatched single surrogates.
 *                          U_INDEX_OUTOFBOUNDS_ERROR if src contains
 *                          too many code points.
 *                          U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
 * @return The length of the result string, if successful - or in case of a buffer overflow,
 *         in which case it will be greater than destCapacity.
 * @stable ICU 2.6
 */
U_STABLE int32_t U_EXPORT2
uidna_IDNToASCII(  const UChar* src, int32_t srcLength,
                   UChar* dest, int32_t destCapacity,
                   int32_t options,
                   UParseError* parseError,
                   UErrorCode* status);

/**
 * Convenience function that implements the IDNToUnicode operation as defined in the IDNA RFC.
 * This operation is done on complete domain names, e.g: "www.example.com". 
 *
 * <b>Note:</b> IDNA RFC specifies that a conformant application should divide a domain name
 * into separate labels, decide whether to apply allowUnassigned and useSTD3ASCIIRules on each, 
 * and then convert. This function does not offer that level of granularity. The options once  
 * set will apply to all labels in the domain name
 *
 * @param src               Input UChar array containing IDN in ASCII (ACE encoded) form.
 * @param srcLength         Number of UChars in src, or -1 if NUL-terminated.
 * @param dest Output       UChar array containing Unicode equivalent of source IDN.
 * @param destCapacity      Size of dest.
 * @param options           A bit set of options:
 *  
 *  - UIDNA_DEFAULT             Use default options, i.e., do not process unassigned code points
 *                              and do not use STD3 ASCII rules
 *                              If unassigned code points are found the operation fails with 
 *                              U_UNASSIGNED_CODE_POINT_FOUND error code.
 *
 *  - UIDNA_ALLOW_UNASSIGNED    Unassigned values can be converted to ASCII for query operations
 *                              If this option is set, the unassigned code points are in the input 
 *                              are treated as normal Unicode code points.
 *                          
 *  - UIDNA_USE_STD3_RULES      Use STD3 ASCII rules for host name syntax restrictions
 *                              If this option is set and the input does not satisfy STD3 rules,  
 *                              the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR
 *
 * @param parseError        Pointer to UParseError struct to receive information on position 
 *                          of error if an error is encountered. Can be NULL.
 * @param status            ICU in/out error code parameter.
 *                          U_INVALID_CHAR_FOUND if src contains
 *                          unmatched single surrogates.
 *                          U_INDEX_OUTOFBOUNDS_ERROR if src contains
 *                          too many code points.
 *                          U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
 * @return The length of the result string, if successful - or in case of a buffer overflow,
 *         in which case it will be greater than destCapacity.
 * @stable ICU 2.6
 */
U_STABLE int32_t U_EXPORT2
uidna_IDNToUnicode(  const UChar* src, int32_t srcLength,
                     UChar* dest, int32_t destCapacity,
                     int32_t options,
                     UParseError* parseError,
                     UErrorCode* status);

/**
 * Compare two IDN strings for equivalence.
 * This function splits the domain names into labels and compares them.
 * According to IDN RFC, whenever two labels are compared, they are 
 * considered equal if and only if their ASCII forms (obtained by 
 * applying toASCII) match using an case-insensitive ASCII comparison.
 * Two domain names are considered a match if and only if all labels 
 * match regardless of whether label separators match.
 *
 * @param s1                First source string.
 * @param length1           Length of first source string, or -1 if NUL-terminated.
 *
 * @param s2                Second source string.
 * @param length2           Length of second source string, or -1 if NUL-terminated.
 * @param options           A bit set of options:
 *  
 *  - UIDNA_DEFAULT             Use default options, i.e., do not process unassigned code points
 *                              and do not use STD3 ASCII rules
 *                              If unassigned code points are found the operation fails with 
 *                              U_UNASSIGNED_CODE_POINT_FOUND error code.
 *
 *  - UIDNA_ALLOW_UNASSIGNED    Unassigned values can be converted to ASCII for query operations
 *                              If this option is set, the unassigned code points are in the input 
 *                              are treated as normal Unicode code points.
 *                          
 *  - UIDNA_USE_STD3_RULES      Use STD3 ASCII rules for host name syntax restrictions
 *                              If this option is set and the input does not satisfy STD3 rules,  
 *                              the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR
 *
 * @param status            ICU error code in/out parameter.
 *                          Must fulfill U_SUCCESS before the function call.
 * @return <0 or 0 or >0 as usual for string comparisons
 * @stable ICU 2.6
 */
U_STABLE int32_t U_EXPORT2
uidna_compare(  const UChar *s1, int32_t length1,
                const UChar *s2, int32_t length2,
                int32_t options,
                UErrorCode* status);

#endif /* #if !UCONFIG_NO_IDNA */

#endif
Commit	Line	Data
b75a7d8f A	1	/*
	2	*******************************************************************************
	3	*
46f4442e	4	* Copyright (C) 2003-2007, International Business Machines
b75a7d8f A	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	/**
73c04bcf A	27	* \file
	28	* \brief C API: Internationalized Domain Names in Applications Tranformation
	29	*
b75a7d8f A	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).
374ca955	49	*
b75a7d8f A	50	*/
	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
374ca955	57	* @stable ICU 2.6
b75a7d8f A	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
374ca955	64	* @stable ICU 2.6
b75a7d8f A	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
374ca955	71	* @stable ICU 2.6
b75a7d8f A	72	*/
b75a7d8f A	73	#define UIDNA_USE_STD3_RULES 0x0002
374ca955	74
b75a7d8f A	75	/**
	76	* This function implements the ToASCII operation as defined in the IDNA RFC.
	77	* This operation is done on <b>single labels</b> before sending it to something that expects
	78	* ASCII names. A label is an individual part of a domain name. Labels are usually
	79	* separated by dots; e.g." "www.example.com" is composed of 3 labels
	80	* "www","example", and "com".
	81	*
	82	*
	83	* @param src Input UChar array containing label in Unicode.
	84	* @param srcLength Number of UChars in src, or -1 if NUL-terminated.
	85	* @param dest Output UChar array with ASCII (ACE encoded) label.
	86	* @param destCapacity Size of dest.
	87	* @param options A bit set of options:
	88	*
374ca955 A	89	* - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points
	90	* and do not use STD3 ASCII rules
	91	* If unassigned code points are found the operation fails with
	92	* U_UNASSIGNED_ERROR error code.
b75a7d8f	93	*
374ca955 A	94	* - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations
	95	* If this option is set, the unassigned code points are in the input
	96	* are treated as normal Unicode code points.
b75a7d8f	97	*
374ca955 A	98	* - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions
	99	* If this option is set and the input does not satisfy STD3 rules,
	100	* the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR
b75a7d8f A	101	*
	102	* @param parseError Pointer to UParseError struct to receive information on position
	103	* of error if an error is encountered. Can be NULL.
	104	* @param status ICU in/out error code parameter.
	105	* U_INVALID_CHAR_FOUND if src contains
	106	* unmatched single surrogates.
	107	* U_INDEX_OUTOFBOUNDS_ERROR if src contains
	108	* too many code points.
	109	* U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
46f4442e A	110	* @return The length of the result string, if successful - or in case of a buffer overflow,
46f4442e A	111	* in which case it will be greater than destCapacity.
374ca955	112	* @stable ICU 2.6
b75a7d8f	113	*/
374ca955	114	U_STABLE int32_t U_EXPORT2
b75a7d8f A	115	uidna_toASCII(const UChar* src, int32_t srcLength,
	116	UChar* dest, int32_t destCapacity,
	117	int32_t options,
	118	UParseError* parseError,
	119	UErrorCode* status);
	120
	121
	122	/**
	123	* This function implements the ToUnicode operation as defined in the IDNA RFC.
	124	* This operation is done on <b>single labels</b> before sending it to something that expects
	125	* Unicode names. A label is an individual part of a domain name. Labels are usually
	126	* separated by dots; for e.g." "www.example.com" is composed of 3 labels
	127	* "www","example", and "com".
	128	*
	129	* @param src Input UChar array containing ASCII (ACE encoded) label.
	130	* @param srcLength Number of UChars in src, or -1 if NUL-terminated.
	131	* @param dest Output Converted UChar array containing Unicode equivalent of label.
	132	* @param destCapacity Size of dest.
	133	* @param options A bit set of options:
	134	*
374ca955 A	135	* - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points
	136	* and do not use STD3 ASCII rules
	137	* If unassigned code points are found the operation fails with
	138	* U_UNASSIGNED_ERROR error code.
b75a7d8f	139	*
374ca955 A	140	* - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations
	141	* If this option is set, the unassigned code points are in the input
	142	* are treated as normal Unicode code points. <b> Note: </b> This option is
	143	* required on toUnicode operation because the RFC mandates
	144	* verification of decoded ACE input by applying toASCII and comparing
	145	* its output with source
b75a7d8f A	146	*
	147	*
	148	*
374ca955 A	149	* - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions
	150	* If this option is set and the input does not satisfy STD3 rules,
	151	* the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR
b75a7d8f A	152	*
	153	* @param parseError Pointer to UParseError struct to receive information on position
	154	* of error if an error is encountered. Can be NULL.
	155	* @param status ICU in/out error code parameter.
	156	* U_INVALID_CHAR_FOUND if src contains
	157	* unmatched single surrogates.
	158	* U_INDEX_OUTOFBOUNDS_ERROR if src contains
	159	* too many code points.
	160	* U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
46f4442e A	161	* @return The length of the result string, if successful - or in case of a buffer overflow,
46f4442e A	162	* in which case it will be greater than destCapacity.
374ca955	163	* @stable ICU 2.6
b75a7d8f	164	*/
374ca955	165	U_STABLE int32_t U_EXPORT2
b75a7d8f A	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	*
374ca955 A	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.
b75a7d8f	195	*
374ca955 A	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.
b75a7d8f	199	*
374ca955 A	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
b75a7d8f A	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
46f4442e A	212	* @return The length of the result string, if successful - or in case of a buffer overflow,
46f4442e A	213	* in which case it will be greater than destCapacity.
374ca955	214	* @stable ICU 2.6
b75a7d8f	215	*/
374ca955	216	U_STABLE int32_t U_EXPORT2
b75a7d8f A	217	uidna_IDNToASCII( const UChar* src, int32_t srcLength,
	218	UChar* dest, int32_t destCapacity,
	219	int32_t options,
	220	UParseError* parseError,
	221	UErrorCode* status);
	222
	223	/**
	224	* Convenience function that implements the IDNToUnicode operation as defined in the IDNA RFC.
	225	* This operation is done on complete domain names, e.g: "www.example.com".
	226	*
	227	* <b>Note:</b> IDNA RFC specifies that a conformant application should divide a domain name
	228	* into separate labels, decide whether to apply allowUnassigned and useSTD3ASCIIRules on each,
	229	* and then convert. This function does not offer that level of granularity. The options once
	230	* set will apply to all labels in the domain name
	231	*
	232	* @param src Input UChar array containing IDN in ASCII (ACE encoded) form.
	233	* @param srcLength Number of UChars in src, or -1 if NUL-terminated.
	234	* @param dest Output UChar array containing Unicode equivalent of source IDN.
	235	* @param destCapacity Size of dest.
	236	* @param options A bit set of options:
	237	*
374ca955 A	238	* - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points
	239	* and do not use STD3 ASCII rules
	240	* If unassigned code points are found the operation fails with
	241	* U_UNASSIGNED_CODE_POINT_FOUND error code.
b75a7d8f	242	*
374ca955 A	243	* - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations
	244	* If this option is set, the unassigned code points are in the input
	245	* are treated as normal Unicode code points.
b75a7d8f	246	*
374ca955 A	247	* - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions
	248	* If this option is set and the input does not satisfy STD3 rules,
	249	* the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR
b75a7d8f A	250	*
	251	* @param parseError Pointer to UParseError struct to receive information on position
	252	* of error if an error is encountered. Can be NULL.
	253	* @param status ICU in/out error code parameter.
	254	* U_INVALID_CHAR_FOUND if src contains
	255	* unmatched single surrogates.
	256	* U_INDEX_OUTOFBOUNDS_ERROR if src contains
	257	* too many code points.
	258	* U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough
46f4442e A	259	* @return The length of the result string, if successful - or in case of a buffer overflow,
46f4442e A	260	* in which case it will be greater than destCapacity.
374ca955	261	* @stable ICU 2.6
b75a7d8f	262	*/
374ca955	263	U_STABLE int32_t U_EXPORT2
b75a7d8f A	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	/**
374ca955	271	* Compare two IDN strings for equivalence.
b75a7d8f A	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	*
374ca955 A	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.
b75a7d8f	290	*
374ca955 A	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.
b75a7d8f	294	*
374ca955 A	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
b75a7d8f A	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
374ca955	302	* @stable ICU 2.6
b75a7d8f	303	*/
374ca955	304	U_STABLE int32_t U_EXPORT2
b75a7d8f A	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