uidna.h (34936B)
1 // © 2016 and later: Unicode, Inc. and others. 2 // License & terms of use: http://www.unicode.org/copyright.html 3 /* 4 ******************************************************************************* 5 * 6 * Copyright (C) 2003-2014, International Business Machines 7 * Corporation and others. All Rights Reserved. 8 * 9 ******************************************************************************* 10 * file name: uidna.h 11 * encoding: UTF-8 12 * tab size: 8 (not used) 13 * indentation:4 14 * 15 * created on: 2003feb1 16 * created by: Ram Viswanadha 17 */ 18 19 #ifndef __UIDNA_H__ 20 #define __UIDNA_H__ 21 22 #include "unicode/utypes.h" 23 24 #if !UCONFIG_NO_IDNA 25 26 #include <stdbool.h> 27 #include "unicode/parseerr.h" 28 29 #if U_SHOW_CPLUSPLUS_API 30 #include "unicode/localpointer.h" 31 #endif // U_SHOW_CPLUSPLUS_API 32 33 /** 34 * \file 35 * \brief C API: Internationalizing Domain Names in Applications (IDNA) 36 * 37 * IDNA2008 is implemented according to UTS #46, see the IDNA C++ class in idna.h. 38 * 39 * The C API functions which do take a UIDNA * service object pointer 40 * implement UTS #46 and IDNA2008. 41 * 42 * IDNA2003 is obsolete. 43 * The C API functions which do not take a service object pointer 44 * implement IDNA2003. They are all deprecated. 45 */ 46 47 /* 48 * IDNA option bit set values. 49 */ 50 enum { 51 /** 52 * Default options value: UTS #46 nontransitional processing. 53 * For use in static worker and factory methods. 54 * 55 * Since ICU 76, this is the same as 56 * UIDNA_NONTRANSITIONAL_TO_ASCII | UIDNA_NONTRANSITIONAL_TO_UNICODE, 57 * corresponding to Unicode 15.1 UTS #46 deprecating transitional processing. 58 * (These options are ignored by the IDNA2003 implementation.) 59 * 60 * Before ICU 76, this constant did not set any of the options. 61 * 62 * @stable ICU 2.6 63 */ 64 UIDNA_DEFAULT=0x30, 65 #ifndef U_HIDE_DEPRECATED_API 66 /** 67 * Option to allow unassigned code points in domain names and labels. 68 * For use in static worker and factory methods. 69 * <p>This option is ignored by the UTS46 implementation. 70 * (UTS #46 disallows unassigned code points.) 71 * @deprecated ICU 55 Use UTS #46 instead via uidna_openUTS46() or class IDNA. 72 */ 73 UIDNA_ALLOW_UNASSIGNED=1, 74 #endif /* U_HIDE_DEPRECATED_API */ 75 /** 76 * Option to check whether the input conforms to the STD3 ASCII rules, 77 * for example the restriction of labels to LDH characters 78 * (ASCII Letters, Digits and Hyphen-Minus). 79 * For use in static worker and factory methods. 80 * @stable ICU 2.6 81 */ 82 UIDNA_USE_STD3_RULES=2, 83 /** 84 * IDNA option to check for whether the input conforms to the BiDi rules. 85 * For use in static worker and factory methods. 86 * <p>This option is ignored by the IDNA2003 implementation. 87 * (IDNA2003 always performs a BiDi check.) 88 * @stable ICU 4.6 89 */ 90 UIDNA_CHECK_BIDI=4, 91 /** 92 * IDNA option to check for whether the input conforms to the CONTEXTJ rules. 93 * For use in static worker and factory methods. 94 * <p>This option is ignored by the IDNA2003 implementation. 95 * (The CONTEXTJ check is new in IDNA2008.) 96 * @stable ICU 4.6 97 */ 98 UIDNA_CHECK_CONTEXTJ=8, 99 /** 100 * IDNA option for nontransitional processing in ToASCII(). 101 * For use in static worker and factory methods. 102 * 103 * <p>By default, ToASCII() uses transitional processing. 104 * Unicode 15.1 UTS #46 deprecated transitional processing. 105 * 106 * <p>This option is ignored by the IDNA2003 implementation. 107 * (This is only relevant for compatibility of newer IDNA implementations with IDNA2003.) 108 * @stable ICU 4.6 109 * @see UIDNA_DEFAULT 110 */ 111 UIDNA_NONTRANSITIONAL_TO_ASCII=0x10, 112 /** 113 * IDNA option for nontransitional processing in ToUnicode(). 114 * For use in static worker and factory methods. 115 * 116 * <p>By default, ToUnicode() uses transitional processing. 117 * Unicode 15.1 UTS #46 deprecated transitional processing. 118 * 119 * <p>This option is ignored by the IDNA2003 implementation. 120 * (This is only relevant for compatibility of newer IDNA implementations with IDNA2003.) 121 * @stable ICU 4.6 122 * @see UIDNA_DEFAULT 123 */ 124 UIDNA_NONTRANSITIONAL_TO_UNICODE=0x20, 125 /** 126 * IDNA option to check for whether the input conforms to the CONTEXTO rules. 127 * For use in static worker and factory methods. 128 * <p>This option is ignored by the IDNA2003 implementation. 129 * (The CONTEXTO check is new in IDNA2008.) 130 * <p>This is for use by registries for IDNA2008 conformance. 131 * UTS #46 does not require the CONTEXTO check. 132 * @stable ICU 49 133 */ 134 UIDNA_CHECK_CONTEXTO=0x40 135 }; 136 137 /** 138 * Opaque C service object type for the new IDNA API. 139 * @stable ICU 4.6 140 */ 141 struct UIDNA; 142 typedef struct UIDNA UIDNA; /**< C typedef for struct UIDNA. @stable ICU 4.6 */ 143 144 /** 145 * Returns a UIDNA instance which implements UTS #46. 146 * Returns an unmodifiable instance, owned by the caller. 147 * Cache it for multiple operations, and uidna_close() it when done. 148 * The instance is thread-safe, that is, it can be used concurrently. 149 * 150 * For details about the UTS #46 implementation see the IDNA C++ class in idna.h. 151 * 152 * @param options Bit set to modify the processing and error checking. 153 * These should include UIDNA_DEFAULT, or 154 * UIDNA_NONTRANSITIONAL_TO_ASCII | UIDNA_NONTRANSITIONAL_TO_UNICODE. 155 * See option bit set values in uidna.h. 156 * @param pErrorCode Standard ICU error code. Its input value must 157 * pass the U_SUCCESS() test, or else the function returns 158 * immediately. Check for U_FAILURE() on output or use with 159 * function chaining. (See User Guide for details.) 160 * @return the UTS #46 UIDNA instance, if successful 161 * @stable ICU 4.6 162 */ 163 U_CAPI UIDNA * U_EXPORT2 164 uidna_openUTS46(uint32_t options, UErrorCode *pErrorCode); 165 166 /** 167 * Closes a UIDNA instance. 168 * @param idna UIDNA instance to be closed 169 * @stable ICU 4.6 170 */ 171 U_CAPI void U_EXPORT2 172 uidna_close(UIDNA *idna); 173 174 #if U_SHOW_CPLUSPLUS_API 175 176 U_NAMESPACE_BEGIN 177 178 /** 179 * \class LocalUIDNAPointer 180 * "Smart pointer" class, closes a UIDNA via uidna_close(). 181 * For most methods see the LocalPointerBase base class. 182 * 183 * @see LocalPointerBase 184 * @see LocalPointer 185 * @stable ICU 4.6 186 */ 187 U_DEFINE_LOCAL_OPEN_POINTER(LocalUIDNAPointer, UIDNA, uidna_close); 188 189 U_NAMESPACE_END 190 191 #endif 192 193 /** 194 * Output container for IDNA processing errors. 195 * Initialize with UIDNA_INFO_INITIALIZER: 196 * \code 197 * UIDNAInfo info = UIDNA_INFO_INITIALIZER; 198 * int32_t length = uidna_nameToASCII(..., &info, &errorCode); 199 * if(U_SUCCESS(errorCode) && info.errors!=0) { ... } 200 * \endcode 201 * @stable ICU 4.6 202 */ 203 typedef struct UIDNAInfo { 204 /** sizeof(UIDNAInfo) @stable ICU 4.6 */ 205 int16_t size; 206 /** 207 * Set to true if transitional and nontransitional processing produce different results. 208 * For details see C++ IDNAInfo::isTransitionalDifferent(). 209 * @stable ICU 4.6 210 */ 211 UBool isTransitionalDifferent; 212 UBool reservedB3; /**< Reserved field, do not use. @internal */ 213 /** 214 * Bit set indicating IDNA processing errors. 0 if no errors. 215 * See UIDNA_ERROR_... constants. 216 * @stable ICU 4.6 217 */ 218 uint32_t errors; 219 int32_t reservedI2; /**< Reserved field, do not use. @internal */ 220 int32_t reservedI3; /**< Reserved field, do not use. @internal */ 221 } UIDNAInfo; 222 223 /** 224 * Static initializer for a UIDNAInfo struct. 225 * @stable ICU 4.6 226 */ 227 #define UIDNA_INFO_INITIALIZER { \ 228 (int16_t)sizeof(UIDNAInfo), \ 229 false, false, \ 230 0, 0, 0 } 231 232 /** 233 * Converts a single domain name label into its ASCII form for DNS lookup. 234 * If any processing step fails, then pInfo->errors will be non-zero and 235 * the result might not be an ASCII string. 236 * The label might be modified according to the types of errors. 237 * Labels with severe errors will be left in (or turned into) their Unicode form. 238 * 239 * The UErrorCode indicates an error only in exceptional cases, 240 * such as a U_MEMORY_ALLOCATION_ERROR. 241 * 242 * @param idna UIDNA instance 243 * @param label Input domain name label 244 * @param length Label length, or -1 if NUL-terminated 245 * @param dest Destination string buffer 246 * @param capacity Destination buffer capacity 247 * @param pInfo Output container of IDNA processing details. 248 * @param pErrorCode Standard ICU error code. Its input value must 249 * pass the U_SUCCESS() test, or else the function returns 250 * immediately. Check for U_FAILURE() on output or use with 251 * function chaining. (See User Guide for details.) 252 * @return destination string length 253 * @stable ICU 4.6 254 */ 255 U_CAPI int32_t U_EXPORT2 256 uidna_labelToASCII(const UIDNA *idna, 257 const UChar *label, int32_t length, 258 UChar *dest, int32_t capacity, 259 UIDNAInfo *pInfo, UErrorCode *pErrorCode); 260 261 /** 262 * Converts a single domain name label into its Unicode form for human-readable display. 263 * If any processing step fails, then pInfo->errors will be non-zero. 264 * The label might be modified according to the types of errors. 265 * 266 * The UErrorCode indicates an error only in exceptional cases, 267 * such as a U_MEMORY_ALLOCATION_ERROR. 268 * 269 * @param idna UIDNA instance 270 * @param label Input domain name label 271 * @param length Label length, or -1 if NUL-terminated 272 * @param dest Destination string buffer 273 * @param capacity Destination buffer capacity 274 * @param pInfo Output container of IDNA processing details. 275 * @param pErrorCode Standard ICU error code. Its input value must 276 * pass the U_SUCCESS() test, or else the function returns 277 * immediately. Check for U_FAILURE() on output or use with 278 * function chaining. (See User Guide for details.) 279 * @return destination string length 280 * @stable ICU 4.6 281 */ 282 U_CAPI int32_t U_EXPORT2 283 uidna_labelToUnicode(const UIDNA *idna, 284 const UChar *label, int32_t length, 285 UChar *dest, int32_t capacity, 286 UIDNAInfo *pInfo, UErrorCode *pErrorCode); 287 288 /** 289 * Converts a whole domain name into its ASCII form for DNS lookup. 290 * If any processing step fails, then pInfo->errors will be non-zero and 291 * the result might not be an ASCII string. 292 * The domain name might be modified according to the types of errors. 293 * Labels with severe errors will be left in (or turned into) their Unicode form. 294 * 295 * The UErrorCode indicates an error only in exceptional cases, 296 * such as a U_MEMORY_ALLOCATION_ERROR. 297 * 298 * @param idna UIDNA instance 299 * @param name Input domain name 300 * @param length Domain name length, or -1 if NUL-terminated 301 * @param dest Destination string buffer 302 * @param capacity Destination buffer capacity 303 * @param pInfo Output container of IDNA processing details. 304 * @param pErrorCode Standard ICU error code. Its input value must 305 * pass the U_SUCCESS() test, or else the function returns 306 * immediately. Check for U_FAILURE() on output or use with 307 * function chaining. (See User Guide for details.) 308 * @return destination string length 309 * @stable ICU 4.6 310 */ 311 U_CAPI int32_t U_EXPORT2 312 uidna_nameToASCII(const UIDNA *idna, 313 const UChar *name, int32_t length, 314 UChar *dest, int32_t capacity, 315 UIDNAInfo *pInfo, UErrorCode *pErrorCode); 316 317 /** 318 * Converts a whole domain name into its Unicode form for human-readable display. 319 * If any processing step fails, then pInfo->errors will be non-zero. 320 * The domain name might be modified according to the types of errors. 321 * 322 * The UErrorCode indicates an error only in exceptional cases, 323 * such as a U_MEMORY_ALLOCATION_ERROR. 324 * 325 * @param idna UIDNA instance 326 * @param name Input domain name 327 * @param length Domain name length, or -1 if NUL-terminated 328 * @param dest Destination string buffer 329 * @param capacity Destination buffer capacity 330 * @param pInfo Output container of IDNA processing details. 331 * @param pErrorCode Standard ICU error code. Its input value must 332 * pass the U_SUCCESS() test, or else the function returns 333 * immediately. Check for U_FAILURE() on output or use with 334 * function chaining. (See User Guide for details.) 335 * @return destination string length 336 * @stable ICU 4.6 337 */ 338 U_CAPI int32_t U_EXPORT2 339 uidna_nameToUnicode(const UIDNA *idna, 340 const UChar *name, int32_t length, 341 UChar *dest, int32_t capacity, 342 UIDNAInfo *pInfo, UErrorCode *pErrorCode); 343 344 /* UTF-8 versions of the processing methods --------------------------------- */ 345 346 /** 347 * Converts a single domain name label into its ASCII form for DNS lookup. 348 * UTF-8 version of uidna_labelToASCII(), same behavior. 349 * 350 * @param idna UIDNA instance 351 * @param label Input domain name label 352 * @param length Label length, or -1 if NUL-terminated 353 * @param dest Destination string buffer 354 * @param capacity Destination buffer capacity 355 * @param pInfo Output container of IDNA processing details. 356 * @param pErrorCode Standard ICU error code. Its input value must 357 * pass the U_SUCCESS() test, or else the function returns 358 * immediately. Check for U_FAILURE() on output or use with 359 * function chaining. (See User Guide for details.) 360 * @return destination string length 361 * @stable ICU 4.6 362 */ 363 U_CAPI int32_t U_EXPORT2 364 uidna_labelToASCII_UTF8(const UIDNA *idna, 365 const char *label, int32_t length, 366 char *dest, int32_t capacity, 367 UIDNAInfo *pInfo, UErrorCode *pErrorCode); 368 369 /** 370 * Converts a single domain name label into its Unicode form for human-readable display. 371 * UTF-8 version of uidna_labelToUnicode(), same behavior. 372 * 373 * @param idna UIDNA instance 374 * @param label Input domain name label 375 * @param length Label length, or -1 if NUL-terminated 376 * @param dest Destination string buffer 377 * @param capacity Destination buffer capacity 378 * @param pInfo Output container of IDNA processing details. 379 * @param pErrorCode Standard ICU error code. Its input value must 380 * pass the U_SUCCESS() test, or else the function returns 381 * immediately. Check for U_FAILURE() on output or use with 382 * function chaining. (See User Guide for details.) 383 * @return destination string length 384 * @stable ICU 4.6 385 */ 386 U_CAPI int32_t U_EXPORT2 387 uidna_labelToUnicodeUTF8(const UIDNA *idna, 388 const char *label, int32_t length, 389 char *dest, int32_t capacity, 390 UIDNAInfo *pInfo, UErrorCode *pErrorCode); 391 392 /** 393 * Converts a whole domain name into its ASCII form for DNS lookup. 394 * UTF-8 version of uidna_nameToASCII(), same behavior. 395 * 396 * @param idna UIDNA instance 397 * @param name Input domain name 398 * @param length Domain name length, or -1 if NUL-terminated 399 * @param dest Destination string buffer 400 * @param capacity Destination buffer capacity 401 * @param pInfo Output container of IDNA processing details. 402 * @param pErrorCode Standard ICU error code. Its input value must 403 * pass the U_SUCCESS() test, or else the function returns 404 * immediately. Check for U_FAILURE() on output or use with 405 * function chaining. (See User Guide for details.) 406 * @return destination string length 407 * @stable ICU 4.6 408 */ 409 U_CAPI int32_t U_EXPORT2 410 uidna_nameToASCII_UTF8(const UIDNA *idna, 411 const char *name, int32_t length, 412 char *dest, int32_t capacity, 413 UIDNAInfo *pInfo, UErrorCode *pErrorCode); 414 415 /** 416 * Converts a whole domain name into its Unicode form for human-readable display. 417 * UTF-8 version of uidna_nameToUnicode(), same behavior. 418 * 419 * @param idna UIDNA instance 420 * @param name Input domain name 421 * @param length Domain name length, or -1 if NUL-terminated 422 * @param dest Destination string buffer 423 * @param capacity Destination buffer capacity 424 * @param pInfo Output container of IDNA processing details. 425 * @param pErrorCode Standard ICU error code. Its input value must 426 * pass the U_SUCCESS() test, or else the function returns 427 * immediately. Check for U_FAILURE() on output or use with 428 * function chaining. (See User Guide for details.) 429 * @return destination string length 430 * @stable ICU 4.6 431 */ 432 U_CAPI int32_t U_EXPORT2 433 uidna_nameToUnicodeUTF8(const UIDNA *idna, 434 const char *name, int32_t length, 435 char *dest, int32_t capacity, 436 UIDNAInfo *pInfo, UErrorCode *pErrorCode); 437 438 /* 439 * IDNA error bit set values. 440 * When a domain name or label fails a processing step or does not meet the 441 * validity criteria, then one or more of these error bits are set. 442 */ 443 enum { 444 /** 445 * A non-final domain name label (or the whole domain name) is empty. 446 * @stable ICU 4.6 447 */ 448 UIDNA_ERROR_EMPTY_LABEL=1, 449 /** 450 * A domain name label is longer than 63 bytes. 451 * (See STD13/RFC1034 3.1. Name space specifications and terminology.) 452 * This is only checked in ToASCII operations, and only if the output label is all-ASCII. 453 * @stable ICU 4.6 454 */ 455 UIDNA_ERROR_LABEL_TOO_LONG=2, 456 /** 457 * A domain name is longer than 255 bytes in its storage form. 458 * (See STD13/RFC1034 3.1. Name space specifications and terminology.) 459 * This is only checked in ToASCII operations, and only if the output domain name is all-ASCII. 460 * @stable ICU 4.6 461 */ 462 UIDNA_ERROR_DOMAIN_NAME_TOO_LONG=4, 463 /** 464 * A label starts with a hyphen-minus ('-'). 465 * @stable ICU 4.6 466 */ 467 UIDNA_ERROR_LEADING_HYPHEN=8, 468 /** 469 * A label ends with a hyphen-minus ('-'). 470 * @stable ICU 4.6 471 */ 472 UIDNA_ERROR_TRAILING_HYPHEN=0x10, 473 /** 474 * A label contains hyphen-minus ('-') in the third and fourth positions. 475 * @stable ICU 4.6 476 */ 477 UIDNA_ERROR_HYPHEN_3_4=0x20, 478 /** 479 * A label starts with a combining mark. 480 * @stable ICU 4.6 481 */ 482 UIDNA_ERROR_LEADING_COMBINING_MARK=0x40, 483 /** 484 * A label or domain name contains disallowed characters. 485 * @stable ICU 4.6 486 */ 487 UIDNA_ERROR_DISALLOWED=0x80, 488 /** 489 * A label starts with "xn--" but does not contain valid Punycode. 490 * That is, an xn-- label failed Punycode decoding. 491 * @stable ICU 4.6 492 */ 493 UIDNA_ERROR_PUNYCODE=0x100, 494 /** 495 * A label contains a dot=full stop. 496 * This can occur in an input string for a single-label function. 497 * @stable ICU 4.6 498 */ 499 UIDNA_ERROR_LABEL_HAS_DOT=0x200, 500 /** 501 * An ACE label does not contain a valid label string. 502 * The label was successfully ACE (Punycode) decoded but the resulting 503 * string had severe validation errors. For example, 504 * it might contain characters that are not allowed in ACE labels, 505 * or it might not be normalized. 506 * @stable ICU 4.6 507 */ 508 UIDNA_ERROR_INVALID_ACE_LABEL=0x400, 509 /** 510 * A label does not meet the IDNA BiDi requirements (for right-to-left characters). 511 * @stable ICU 4.6 512 */ 513 UIDNA_ERROR_BIDI=0x800, 514 /** 515 * A label does not meet the IDNA CONTEXTJ requirements. 516 * @stable ICU 4.6 517 */ 518 UIDNA_ERROR_CONTEXTJ=0x1000, 519 /** 520 * A label does not meet the IDNA CONTEXTO requirements for punctuation characters. 521 * Some punctuation characters "Would otherwise have been DISALLOWED" 522 * but are allowed in certain contexts. (RFC 5892) 523 * @stable ICU 49 524 */ 525 UIDNA_ERROR_CONTEXTO_PUNCTUATION=0x2000, 526 /** 527 * A label does not meet the IDNA CONTEXTO requirements for digits. 528 * Arabic-Indic Digits (U+066x) must not be mixed with Extended Arabic-Indic Digits (U+06Fx). 529 * @stable ICU 49 530 */ 531 UIDNA_ERROR_CONTEXTO_DIGITS=0x4000 532 }; 533 534 #ifndef U_HIDE_DEPRECATED_API 535 536 /* IDNA2003 API ------------------------------------------------------------- */ 537 538 /** 539 * IDNA2003: This function implements the ToASCII operation as defined in the IDNA RFC. 540 * This operation is done on <b>single labels</b> before sending it to something that expects 541 * ASCII names. A label is an individual part of a domain name. Labels are usually 542 * separated by dots; e.g. "www.example.com" is composed of 3 labels "www","example", and "com". 543 * 544 * IDNA2003 API Overview: 545 * 546 * The uidna_ API implements the IDNA protocol as defined in the IDNA RFC 547 * (http://www.ietf.org/rfc/rfc3490.txt). 548 * The RFC defines 2 operations: ToASCII and ToUnicode. Domain name labels 549 * containing non-ASCII code points are processed by the 550 * ToASCII operation before passing it to resolver libraries. Domain names 551 * that are obtained from resolver libraries are processed by the 552 * ToUnicode operation before displaying the domain name to the user. 553 * IDNA requires that implementations process input strings with Nameprep 554 * (http://www.ietf.org/rfc/rfc3491.txt), 555 * which is a profile of Stringprep (http://www.ietf.org/rfc/rfc3454.txt), 556 * and then with Punycode (http://www.ietf.org/rfc/rfc3492.txt). 557 * Implementations of IDNA MUST fully implement Nameprep and Punycode; 558 * neither Nameprep nor Punycode are optional. 559 * The input and output of ToASCII and ToUnicode operations are Unicode 560 * and are designed to be chainable, i.e., applying ToASCII or ToUnicode operations 561 * multiple times to an input string will yield the same result as applying the operation 562 * once. 563 * ToUnicode(ToUnicode(ToUnicode...(ToUnicode(string)))) == ToUnicode(string) 564 * ToASCII(ToASCII(ToASCII...(ToASCII(string))) == ToASCII(string). 565 * 566 * @param src Input UChar array containing label in Unicode. 567 * @param srcLength Number of UChars in src, or -1 if NUL-terminated. 568 * @param dest Output UChar array with ASCII (ACE encoded) label. 569 * @param destCapacity Size of dest. 570 * @param options A bit set of options: 571 * 572 * - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points 573 * and do not use STD3 ASCII rules 574 * If unassigned code points are found the operation fails with 575 * U_UNASSIGNED_ERROR error code. 576 * 577 * - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations 578 * If this option is set, the unassigned code points are in the input 579 * are treated as normal Unicode code points. 580 * 581 * - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions 582 * If this option is set and the input does not satisfy STD3 rules, 583 * the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR 584 * 585 * @param parseError Pointer to UParseError struct to receive information on position 586 * of error if an error is encountered. Can be NULL. 587 * @param status ICU in/out error code parameter. 588 * U_INVALID_CHAR_FOUND if src contains 589 * unmatched single surrogates. 590 * U_INDEX_OUTOFBOUNDS_ERROR if src contains 591 * too many code points. 592 * U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough 593 * @return The length of the result string, if successful - or in case of a buffer overflow, 594 * in which case it will be greater than destCapacity. 595 * @deprecated ICU 55 Use UTS #46 instead via uidna_openUTS46() or class IDNA. 596 */ 597 U_DEPRECATED int32_t U_EXPORT2 598 uidna_toASCII(const UChar* src, int32_t srcLength, 599 UChar* dest, int32_t destCapacity, 600 int32_t options, 601 UParseError* parseError, 602 UErrorCode* status); 603 604 605 /** 606 * IDNA2003: This function implements the ToUnicode operation as defined in the IDNA RFC. 607 * This operation is done on <b>single labels</b> before sending it to something that expects 608 * Unicode names. A label is an individual part of a domain name. Labels are usually 609 * separated by dots; for e.g. "www.example.com" is composed of 3 labels "www","example", and "com". 610 * 611 * @param src Input UChar array containing ASCII (ACE encoded) label. 612 * @param srcLength Number of UChars in src, or -1 if NUL-terminated. 613 * @param dest Output Converted UChar array containing Unicode equivalent of label. 614 * @param destCapacity Size of dest. 615 * @param options A bit set of options: 616 * 617 * - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points 618 * and do not use STD3 ASCII rules 619 * If unassigned code points are found the operation fails with 620 * U_UNASSIGNED_ERROR error code. 621 * 622 * - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations 623 * If this option is set, the unassigned code points are in the input 624 * are treated as normal Unicode code points. <b> Note: </b> This option is 625 * required on toUnicode operation because the RFC mandates 626 * verification of decoded ACE input by applying toASCII and comparing 627 * its output with source 628 * 629 * - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions 630 * If this option is set and the input does not satisfy STD3 rules, 631 * the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR 632 * 633 * @param parseError Pointer to UParseError struct to receive information on position 634 * of error if an error is encountered. Can be NULL. 635 * @param status ICU in/out error code parameter. 636 * U_INVALID_CHAR_FOUND if src contains 637 * unmatched single surrogates. 638 * U_INDEX_OUTOFBOUNDS_ERROR if src contains 639 * too many code points. 640 * U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough 641 * @return The length of the result string, if successful - or in case of a buffer overflow, 642 * in which case it will be greater than destCapacity. 643 * @deprecated ICU 55 Use UTS #46 instead via uidna_openUTS46() or class IDNA. 644 */ 645 U_DEPRECATED int32_t U_EXPORT2 646 uidna_toUnicode(const UChar* src, int32_t srcLength, 647 UChar* dest, int32_t destCapacity, 648 int32_t options, 649 UParseError* parseError, 650 UErrorCode* status); 651 652 653 /** 654 * IDNA2003: Convenience function that implements the IDNToASCII operation as defined in the IDNA RFC. 655 * This operation is done on complete domain names, e.g: "www.example.com". 656 * It is important to note that this operation can fail. If it fails, then the input 657 * domain name cannot be used as an Internationalized Domain Name and the application 658 * should have methods defined to deal with the failure. 659 * 660 * <b>Note:</b> IDNA RFC specifies that a conformant application should divide a domain name 661 * into separate labels, decide whether to apply allowUnassigned and useSTD3ASCIIRules on each, 662 * and then convert. This function does not offer that level of granularity. The options once 663 * set will apply to all labels in the domain name 664 * 665 * @param src Input UChar array containing IDN in Unicode. 666 * @param srcLength Number of UChars in src, or -1 if NUL-terminated. 667 * @param dest Output UChar array with ASCII (ACE encoded) IDN. 668 * @param destCapacity Size of dest. 669 * @param options A bit set of options: 670 * 671 * - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points 672 * and do not use STD3 ASCII rules 673 * If unassigned code points are found the operation fails with 674 * U_UNASSIGNED_CODE_POINT_FOUND error code. 675 * 676 * - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations 677 * If this option is set, the unassigned code points are in the input 678 * are treated as normal Unicode code points. 679 * 680 * - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions 681 * If this option is set and the input does not satisfy STD3 rules, 682 * the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR 683 * 684 * @param parseError Pointer to UParseError struct to receive information on position 685 * of error if an error is encountered. Can be NULL. 686 * @param status ICU in/out error code parameter. 687 * U_INVALID_CHAR_FOUND if src contains 688 * unmatched single surrogates. 689 * U_INDEX_OUTOFBOUNDS_ERROR if src contains 690 * too many code points. 691 * U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough 692 * @return The length of the result string, if successful - or in case of a buffer overflow, 693 * in which case it will be greater than destCapacity. 694 * @deprecated ICU 55 Use UTS #46 instead via uidna_openUTS46() or class IDNA. 695 */ 696 U_DEPRECATED int32_t U_EXPORT2 697 uidna_IDNToASCII( const UChar* src, int32_t srcLength, 698 UChar* dest, int32_t destCapacity, 699 int32_t options, 700 UParseError* parseError, 701 UErrorCode* status); 702 703 /** 704 * IDNA2003: Convenience function that implements the IDNToUnicode operation as defined in the IDNA RFC. 705 * This operation is done on complete domain names, e.g: "www.example.com". 706 * 707 * <b>Note:</b> IDNA RFC specifies that a conformant application should divide a domain name 708 * into separate labels, decide whether to apply allowUnassigned and useSTD3ASCIIRules on each, 709 * and then convert. This function does not offer that level of granularity. The options once 710 * set will apply to all labels in the domain name 711 * 712 * @param src Input UChar array containing IDN in ASCII (ACE encoded) form. 713 * @param srcLength Number of UChars in src, or -1 if NUL-terminated. 714 * @param dest Output UChar array containing Unicode equivalent of source IDN. 715 * @param destCapacity Size of dest. 716 * @param options A bit set of options: 717 * 718 * - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points 719 * and do not use STD3 ASCII rules 720 * If unassigned code points are found the operation fails with 721 * U_UNASSIGNED_CODE_POINT_FOUND error code. 722 * 723 * - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations 724 * If this option is set, the unassigned code points are in the input 725 * are treated as normal Unicode code points. 726 * 727 * - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions 728 * If this option is set and the input does not satisfy STD3 rules, 729 * the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR 730 * 731 * @param parseError Pointer to UParseError struct to receive information on position 732 * of error if an error is encountered. Can be NULL. 733 * @param status ICU in/out error code parameter. 734 * U_INVALID_CHAR_FOUND if src contains 735 * unmatched single surrogates. 736 * U_INDEX_OUTOFBOUNDS_ERROR if src contains 737 * too many code points. 738 * U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough 739 * @return The length of the result string, if successful - or in case of a buffer overflow, 740 * in which case it will be greater than destCapacity. 741 * @deprecated ICU 55 Use UTS #46 instead via uidna_openUTS46() or class IDNA. 742 */ 743 U_DEPRECATED int32_t U_EXPORT2 744 uidna_IDNToUnicode( const UChar* src, int32_t srcLength, 745 UChar* dest, int32_t destCapacity, 746 int32_t options, 747 UParseError* parseError, 748 UErrorCode* status); 749 750 /** 751 * IDNA2003: Compare two IDN strings for equivalence. 752 * This function splits the domain names into labels and compares them. 753 * According to IDN RFC, whenever two labels are compared, they are 754 * considered equal if and only if their ASCII forms (obtained by 755 * applying toASCII) match using an case-insensitive ASCII comparison. 756 * Two domain names are considered a match if and only if all labels 757 * match regardless of whether label separators match. 758 * 759 * @param s1 First source string. 760 * @param length1 Length of first source string, or -1 if NUL-terminated. 761 * 762 * @param s2 Second source string. 763 * @param length2 Length of second source string, or -1 if NUL-terminated. 764 * @param options A bit set of options: 765 * 766 * - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points 767 * and do not use STD3 ASCII rules 768 * If unassigned code points are found the operation fails with 769 * U_UNASSIGNED_CODE_POINT_FOUND error code. 770 * 771 * - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations 772 * If this option is set, the unassigned code points are in the input 773 * are treated as normal Unicode code points. 774 * 775 * - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions 776 * If this option is set and the input does not satisfy STD3 rules, 777 * the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR 778 * 779 * @param status ICU error code in/out parameter. 780 * Must fulfill U_SUCCESS before the function call. 781 * @return <0 or 0 or >0 as usual for string comparisons 782 * @deprecated ICU 55 Use UTS #46 instead via uidna_openUTS46() or class IDNA. 783 */ 784 U_DEPRECATED int32_t U_EXPORT2 785 uidna_compare( const UChar *s1, int32_t length1, 786 const UChar *s2, int32_t length2, 787 int32_t options, 788 UErrorCode* status); 789 790 #endif /* U_HIDE_DEPRECATED_API */ 791 792 #endif /* #if !UCONFIG_NO_IDNA */ 793 794 #endif