intl/icu/source/common/unicode/idna.h

Wed, 31 Dec 2014 06:09:35 +0100

author
Michael Schloh von Bennewitz <michael@schloh.com>
date
Wed, 31 Dec 2014 06:09:35 +0100
changeset 0
6474c204b198
permissions
-rw-r--r--

Cloned upstream origin tor-browser at tor-browser-31.3.0esr-4.5-1-build1
revision ID fc1c9ff7c1b2defdbc039f12214767608f46423f for hacking purpose.

michael@0 1 /*
michael@0 2 *******************************************************************************
michael@0 3 * Copyright (C) 2010-2012, International Business Machines
michael@0 4 * Corporation and others. All Rights Reserved.
michael@0 5 *******************************************************************************
michael@0 6 * file name: idna.h
michael@0 7 * encoding: US-ASCII
michael@0 8 * tab size: 8 (not used)
michael@0 9 * indentation:4
michael@0 10 *
michael@0 11 * created on: 2010mar05
michael@0 12 * created by: Markus W. Scherer
michael@0 13 */
michael@0 14
michael@0 15 #ifndef __IDNA_H__
michael@0 16 #define __IDNA_H__
michael@0 17
michael@0 18 /**
michael@0 19 * \file
michael@0 20 * \brief C++ API: Internationalizing Domain Names in Applications (IDNA)
michael@0 21 */
michael@0 22
michael@0 23 #include "unicode/utypes.h"
michael@0 24
michael@0 25 #if !UCONFIG_NO_IDNA
michael@0 26
michael@0 27 #include "unicode/bytestream.h"
michael@0 28 #include "unicode/stringpiece.h"
michael@0 29 #include "unicode/uidna.h"
michael@0 30 #include "unicode/unistr.h"
michael@0 31
michael@0 32 U_NAMESPACE_BEGIN
michael@0 33
michael@0 34 class IDNAInfo;
michael@0 35
michael@0 36 /**
michael@0 37 * Abstract base class for IDNA processing.
michael@0 38 * See http://www.unicode.org/reports/tr46/
michael@0 39 * and http://www.ietf.org/rfc/rfc3490.txt
michael@0 40 *
michael@0 41 * The IDNA class is not intended for public subclassing.
michael@0 42 *
michael@0 43 * This C++ API currently only implements UTS #46.
michael@0 44 * The uidna.h C API implements both UTS #46 (functions using UIDNA service object)
michael@0 45 * and IDNA2003 (functions that do not use a service object).
michael@0 46 * @stable ICU 4.6
michael@0 47 */
michael@0 48 class U_COMMON_API IDNA : public UObject {
michael@0 49 public:
michael@0 50 /**
michael@0 51 * Destructor.
michael@0 52 * @stable ICU 4.6
michael@0 53 */
michael@0 54 ~IDNA();
michael@0 55
michael@0 56 /**
michael@0 57 * Returns an IDNA instance which implements UTS #46.
michael@0 58 * Returns an unmodifiable instance, owned by the caller.
michael@0 59 * Cache it for multiple operations, and delete it when done.
michael@0 60 * The instance is thread-safe, that is, it can be used concurrently.
michael@0 61 *
michael@0 62 * UTS #46 defines Unicode IDNA Compatibility Processing,
michael@0 63 * updated to the latest version of Unicode and compatible with both
michael@0 64 * IDNA2003 and IDNA2008.
michael@0 65 *
michael@0 66 * The worker functions use transitional processing, including deviation mappings,
michael@0 67 * unless UIDNA_NONTRANSITIONAL_TO_ASCII or UIDNA_NONTRANSITIONAL_TO_UNICODE
michael@0 68 * is used in which case the deviation characters are passed through without change.
michael@0 69 *
michael@0 70 * Disallowed characters are mapped to U+FFFD.
michael@0 71 *
michael@0 72 * For available options see the uidna.h header.
michael@0 73 * Operations with the UTS #46 instance do not support the
michael@0 74 * UIDNA_ALLOW_UNASSIGNED option.
michael@0 75 *
michael@0 76 * By default, the UTS #46 implementation allows all ASCII characters (as valid or mapped).
michael@0 77 * When the UIDNA_USE_STD3_RULES option is used, ASCII characters other than
michael@0 78 * letters, digits, hyphen (LDH) and dot/full stop are disallowed and mapped to U+FFFD.
michael@0 79 *
michael@0 80 * @param options Bit set to modify the processing and error checking.
michael@0 81 * See option bit set values in uidna.h.
michael@0 82 * @param errorCode Standard ICU error code. Its input value must
michael@0 83 * pass the U_SUCCESS() test, or else the function returns
michael@0 84 * immediately. Check for U_FAILURE() on output or use with
michael@0 85 * function chaining. (See User Guide for details.)
michael@0 86 * @return the UTS #46 IDNA instance, if successful
michael@0 87 * @stable ICU 4.6
michael@0 88 */
michael@0 89 static IDNA *
michael@0 90 createUTS46Instance(uint32_t options, UErrorCode &errorCode);
michael@0 91
michael@0 92 /**
michael@0 93 * Converts a single domain name label into its ASCII form for DNS lookup.
michael@0 94 * If any processing step fails, then info.hasErrors() will be TRUE and
michael@0 95 * the result might not be an ASCII string.
michael@0 96 * The label might be modified according to the types of errors.
michael@0 97 * Labels with severe errors will be left in (or turned into) their Unicode form.
michael@0 98 *
michael@0 99 * The UErrorCode indicates an error only in exceptional cases,
michael@0 100 * such as a U_MEMORY_ALLOCATION_ERROR.
michael@0 101 *
michael@0 102 * @param label Input domain name label
michael@0 103 * @param dest Destination string object
michael@0 104 * @param info Output container of IDNA processing details.
michael@0 105 * @param errorCode Standard ICU error code. Its input value must
michael@0 106 * pass the U_SUCCESS() test, or else the function returns
michael@0 107 * immediately. Check for U_FAILURE() on output or use with
michael@0 108 * function chaining. (See User Guide for details.)
michael@0 109 * @return dest
michael@0 110 * @stable ICU 4.6
michael@0 111 */
michael@0 112 virtual UnicodeString &
michael@0 113 labelToASCII(const UnicodeString &label, UnicodeString &dest,
michael@0 114 IDNAInfo &info, UErrorCode &errorCode) const = 0;
michael@0 115
michael@0 116 /**
michael@0 117 * Converts a single domain name label into its Unicode form for human-readable display.
michael@0 118 * If any processing step fails, then info.hasErrors() will be TRUE.
michael@0 119 * The label might be modified according to the types of errors.
michael@0 120 *
michael@0 121 * The UErrorCode indicates an error only in exceptional cases,
michael@0 122 * such as a U_MEMORY_ALLOCATION_ERROR.
michael@0 123 *
michael@0 124 * @param label Input domain name label
michael@0 125 * @param dest Destination string object
michael@0 126 * @param info Output container of IDNA processing details.
michael@0 127 * @param errorCode Standard ICU error code. Its input value must
michael@0 128 * pass the U_SUCCESS() test, or else the function returns
michael@0 129 * immediately. Check for U_FAILURE() on output or use with
michael@0 130 * function chaining. (See User Guide for details.)
michael@0 131 * @return dest
michael@0 132 * @stable ICU 4.6
michael@0 133 */
michael@0 134 virtual UnicodeString &
michael@0 135 labelToUnicode(const UnicodeString &label, UnicodeString &dest,
michael@0 136 IDNAInfo &info, UErrorCode &errorCode) const = 0;
michael@0 137
michael@0 138 /**
michael@0 139 * Converts a whole domain name into its ASCII form for DNS lookup.
michael@0 140 * If any processing step fails, then info.hasErrors() will be TRUE and
michael@0 141 * the result might not be an ASCII string.
michael@0 142 * The domain name might be modified according to the types of errors.
michael@0 143 * Labels with severe errors will be left in (or turned into) their Unicode form.
michael@0 144 *
michael@0 145 * The UErrorCode indicates an error only in exceptional cases,
michael@0 146 * such as a U_MEMORY_ALLOCATION_ERROR.
michael@0 147 *
michael@0 148 * @param name Input domain name
michael@0 149 * @param dest Destination string object
michael@0 150 * @param info Output container of IDNA processing details.
michael@0 151 * @param errorCode Standard ICU error code. Its input value must
michael@0 152 * pass the U_SUCCESS() test, or else the function returns
michael@0 153 * immediately. Check for U_FAILURE() on output or use with
michael@0 154 * function chaining. (See User Guide for details.)
michael@0 155 * @return dest
michael@0 156 * @stable ICU 4.6
michael@0 157 */
michael@0 158 virtual UnicodeString &
michael@0 159 nameToASCII(const UnicodeString &name, UnicodeString &dest,
michael@0 160 IDNAInfo &info, UErrorCode &errorCode) const = 0;
michael@0 161
michael@0 162 /**
michael@0 163 * Converts a whole domain name into its Unicode form for human-readable display.
michael@0 164 * If any processing step fails, then info.hasErrors() will be TRUE.
michael@0 165 * The domain name might be modified according to the types of errors.
michael@0 166 *
michael@0 167 * The UErrorCode indicates an error only in exceptional cases,
michael@0 168 * such as a U_MEMORY_ALLOCATION_ERROR.
michael@0 169 *
michael@0 170 * @param name Input domain name
michael@0 171 * @param dest Destination string object
michael@0 172 * @param info Output container of IDNA processing details.
michael@0 173 * @param errorCode Standard ICU error code. Its input value must
michael@0 174 * pass the U_SUCCESS() test, or else the function returns
michael@0 175 * immediately. Check for U_FAILURE() on output or use with
michael@0 176 * function chaining. (See User Guide for details.)
michael@0 177 * @return dest
michael@0 178 * @stable ICU 4.6
michael@0 179 */
michael@0 180 virtual UnicodeString &
michael@0 181 nameToUnicode(const UnicodeString &name, UnicodeString &dest,
michael@0 182 IDNAInfo &info, UErrorCode &errorCode) const = 0;
michael@0 183
michael@0 184 // UTF-8 versions of the processing methods ---------------------------- ***
michael@0 185
michael@0 186 /**
michael@0 187 * Converts a single domain name label into its ASCII form for DNS lookup.
michael@0 188 * UTF-8 version of labelToASCII(), same behavior.
michael@0 189 *
michael@0 190 * @param label Input domain name label
michael@0 191 * @param dest Destination byte sink; Flush()ed if successful
michael@0 192 * @param info Output container of IDNA processing details.
michael@0 193 * @param errorCode Standard ICU error code. Its input value must
michael@0 194 * pass the U_SUCCESS() test, or else the function returns
michael@0 195 * immediately. Check for U_FAILURE() on output or use with
michael@0 196 * function chaining. (See User Guide for details.)
michael@0 197 * @return dest
michael@0 198 * @stable ICU 4.6
michael@0 199 */
michael@0 200 virtual void
michael@0 201 labelToASCII_UTF8(const StringPiece &label, ByteSink &dest,
michael@0 202 IDNAInfo &info, UErrorCode &errorCode) const;
michael@0 203
michael@0 204 /**
michael@0 205 * Converts a single domain name label into its Unicode form for human-readable display.
michael@0 206 * UTF-8 version of labelToUnicode(), same behavior.
michael@0 207 *
michael@0 208 * @param label Input domain name label
michael@0 209 * @param dest Destination byte sink; Flush()ed if successful
michael@0 210 * @param info Output container of IDNA processing details.
michael@0 211 * @param errorCode Standard ICU error code. Its input value must
michael@0 212 * pass the U_SUCCESS() test, or else the function returns
michael@0 213 * immediately. Check for U_FAILURE() on output or use with
michael@0 214 * function chaining. (See User Guide for details.)
michael@0 215 * @return dest
michael@0 216 * @stable ICU 4.6
michael@0 217 */
michael@0 218 virtual void
michael@0 219 labelToUnicodeUTF8(const StringPiece &label, ByteSink &dest,
michael@0 220 IDNAInfo &info, UErrorCode &errorCode) const;
michael@0 221
michael@0 222 /**
michael@0 223 * Converts a whole domain name into its ASCII form for DNS lookup.
michael@0 224 * UTF-8 version of nameToASCII(), same behavior.
michael@0 225 *
michael@0 226 * @param name Input domain name
michael@0 227 * @param dest Destination byte sink; Flush()ed if successful
michael@0 228 * @param info Output container of IDNA processing details.
michael@0 229 * @param errorCode Standard ICU error code. Its input value must
michael@0 230 * pass the U_SUCCESS() test, or else the function returns
michael@0 231 * immediately. Check for U_FAILURE() on output or use with
michael@0 232 * function chaining. (See User Guide for details.)
michael@0 233 * @return dest
michael@0 234 * @stable ICU 4.6
michael@0 235 */
michael@0 236 virtual void
michael@0 237 nameToASCII_UTF8(const StringPiece &name, ByteSink &dest,
michael@0 238 IDNAInfo &info, UErrorCode &errorCode) const;
michael@0 239
michael@0 240 /**
michael@0 241 * Converts a whole domain name into its Unicode form for human-readable display.
michael@0 242 * UTF-8 version of nameToUnicode(), same behavior.
michael@0 243 *
michael@0 244 * @param name Input domain name
michael@0 245 * @param dest Destination byte sink; Flush()ed if successful
michael@0 246 * @param info Output container of IDNA processing details.
michael@0 247 * @param errorCode Standard ICU error code. Its input value must
michael@0 248 * pass the U_SUCCESS() test, or else the function returns
michael@0 249 * immediately. Check for U_FAILURE() on output or use with
michael@0 250 * function chaining. (See User Guide for details.)
michael@0 251 * @return dest
michael@0 252 * @stable ICU 4.6
michael@0 253 */
michael@0 254 virtual void
michael@0 255 nameToUnicodeUTF8(const StringPiece &name, ByteSink &dest,
michael@0 256 IDNAInfo &info, UErrorCode &errorCode) const;
michael@0 257 };
michael@0 258
michael@0 259 class UTS46;
michael@0 260
michael@0 261 /**
michael@0 262 * Output container for IDNA processing errors.
michael@0 263 * The IDNAInfo class is not suitable for subclassing.
michael@0 264 * @stable ICU 4.6
michael@0 265 */
michael@0 266 class U_COMMON_API IDNAInfo : public UMemory {
michael@0 267 public:
michael@0 268 /**
michael@0 269 * Constructor for stack allocation.
michael@0 270 * @stable ICU 4.6
michael@0 271 */
michael@0 272 IDNAInfo() : errors(0), labelErrors(0), isTransDiff(FALSE), isBiDi(FALSE), isOkBiDi(TRUE) {}
michael@0 273 /**
michael@0 274 * Were there IDNA processing errors?
michael@0 275 * @return TRUE if there were processing errors
michael@0 276 * @stable ICU 4.6
michael@0 277 */
michael@0 278 UBool hasErrors() const { return errors!=0; }
michael@0 279 /**
michael@0 280 * Returns a bit set indicating IDNA processing errors.
michael@0 281 * See UIDNA_ERROR_... constants in uidna.h.
michael@0 282 * @return bit set of processing errors
michael@0 283 * @stable ICU 4.6
michael@0 284 */
michael@0 285 uint32_t getErrors() const { return errors; }
michael@0 286 /**
michael@0 287 * Returns TRUE if transitional and nontransitional processing produce different results.
michael@0 288 * This is the case when the input label or domain name contains
michael@0 289 * one or more deviation characters outside a Punycode label (see UTS #46).
michael@0 290 * <ul>
michael@0 291 * <li>With nontransitional processing, such characters are
michael@0 292 * copied to the destination string.
michael@0 293 * <li>With transitional processing, such characters are
michael@0 294 * mapped (sharp s/sigma) or removed (joiner/nonjoiner).
michael@0 295 * </ul>
michael@0 296 * @return TRUE if transitional and nontransitional processing produce different results
michael@0 297 * @stable ICU 4.6
michael@0 298 */
michael@0 299 UBool isTransitionalDifferent() const { return isTransDiff; }
michael@0 300
michael@0 301 private:
michael@0 302 friend class UTS46;
michael@0 303
michael@0 304 IDNAInfo(const IDNAInfo &other); // no copying
michael@0 305 IDNAInfo &operator=(const IDNAInfo &other); // no copying
michael@0 306
michael@0 307 void reset() {
michael@0 308 errors=labelErrors=0;
michael@0 309 isTransDiff=FALSE;
michael@0 310 isBiDi=FALSE;
michael@0 311 isOkBiDi=TRUE;
michael@0 312 }
michael@0 313
michael@0 314 uint32_t errors, labelErrors;
michael@0 315 UBool isTransDiff;
michael@0 316 UBool isBiDi;
michael@0 317 UBool isOkBiDi;
michael@0 318 };
michael@0 319
michael@0 320 U_NAMESPACE_END
michael@0 321
michael@0 322 #endif // UCONFIG_NO_IDNA
michael@0 323 #endif // __IDNA_H__

mercurial