1 /* 2 ******************************************************************************* 3 * 4 * Copyright (C) 2003-2013, International Business Machines 5 * Corporation and others. All Rights Reserved. 6 * 7 ******************************************************************************* 8 * file name: uidna.h 9 * encoding: US-ASCII 10 * tab size: 8 (not used) 11 * indentation:4 12 * 13 * created on: 2003feb1 14 * created by: Ram Viswanadha 15 */ 16 17 #ifndef __UIDNA_H__ 18 #define __UIDNA_H__ 19 20 #include "unicode/utypes.h" 21 22 #if !UCONFIG_NO_IDNA 23 24 #include "unicode/localpointer.h" 25 #include "unicode/parseerr.h" 26 27 /** 28 * \file 29 * \brief C API: Internationalizing Domain Names in Applications (IDNA) 30 * 31 * IDNA2008 is implemented according to UTS #46, see the IDNA C++ class in idna.h. 32 * 33 * The C API functions which do take a UIDNA * service object pointer 34 * implement UTS #46 and IDNA2008. 35 * The C API functions which do not take a service object pointer 36 * implement IDNA2003. 37 */ 38 39 /* 40 * IDNA option bit set values. 41 */ 42 enum { 43 /** 44 * Default options value: None of the other options are set. 45 * For use in static worker and factory methods. 46 * @stable ICU 2.6 47 */ 48 UIDNA_DEFAULT=0, 49 /** 50 * Option to allow unassigned code points in domain names and labels. 51 * For use in static worker and factory methods. 52 * <p>This option is ignored by the UTS46 implementation. 53 * (UTS #46 disallows unassigned code points.) 54 * @stable ICU 2.6 55 */ 56 UIDNA_ALLOW_UNASSIGNED=1, 57 /** 58 * Option to check whether the input conforms to the STD3 ASCII rules, 59 * for example the restriction of labels to LDH characters 60 * (ASCII Letters, Digits and Hyphen-Minus). 61 * For use in static worker and factory methods. 62 * @stable ICU 2.6 63 */ 64 UIDNA_USE_STD3_RULES=2, 65 /** 66 * IDNA option to check for whether the input conforms to the BiDi rules. 67 * For use in static worker and factory methods. 68 * <p>This option is ignored by the IDNA2003 implementation. 69 * (IDNA2003 always performs a BiDi check.) 70 * @stable ICU 4.6 71 */ 72 UIDNA_CHECK_BIDI=4, 73 /** 74 * IDNA option to check for whether the input conforms to the CONTEXTJ rules. 75 * For use in static worker and factory methods. 76 * <p>This option is ignored by the IDNA2003 implementation. 77 * (The CONTEXTJ check is new in IDNA2008.) 78 * @stable ICU 4.6 79 */ 80 UIDNA_CHECK_CONTEXTJ=8, 81 /** 82 * IDNA option for nontransitional processing in ToASCII(). 83 * For use in static worker and factory methods. 84 * <p>By default, ToASCII() uses transitional processing. 85 * <p>This option is ignored by the IDNA2003 implementation. 86 * (This is only relevant for compatibility of newer IDNA implementations with IDNA2003.) 87 * @stable ICU 4.6 88 */ 89 UIDNA_NONTRANSITIONAL_TO_ASCII=0x10, 90 /** 91 * IDNA option for nontransitional processing in ToUnicode(). 92 * For use in static worker and factory methods. 93 * <p>By default, ToUnicode() uses transitional processing. 94 * <p>This option is ignored by the IDNA2003 implementation. 95 * (This is only relevant for compatibility of newer IDNA implementations with IDNA2003.) 96 * @stable ICU 4.6 97 */ 98 UIDNA_NONTRANSITIONAL_TO_UNICODE=0x20, 99 /** 100 * IDNA option to check for whether the input conforms to the CONTEXTO rules. 101 * For use in static worker and factory methods. 102 * <p>This option is ignored by the IDNA2003 implementation. 103 * (The CONTEXTO check is new in IDNA2008.) 104 * <p>This is for use by registries for IDNA2008 conformance. 105 * UTS #46 does not require the CONTEXTO check. 106 * @stable ICU 49 107 */ 108 UIDNA_CHECK_CONTEXTO=0x40 109 }; 110 111 /** 112 * Opaque C service object type for the new IDNA API. 113 * @stable ICU 4.6 114 */ 115 struct UIDNA; 116 typedef struct UIDNA UIDNA; /**< C typedef for struct UIDNA. @stable ICU 4.6 */ 117 118 /** 119 * Returns a UIDNA instance which implements UTS #46. 120 * Returns an unmodifiable instance, owned by the caller. 121 * Cache it for multiple operations, and uidna_close() it when done. 122 * The instance is thread-safe, that is, it can be used concurrently. 123 * 124 * For details about the UTS #46 implementation see the IDNA C++ class in idna.h. 125 * 126 * @param options Bit set to modify the processing and error checking. 127 * See option bit set values in uidna.h. 128 * @param pErrorCode Standard ICU error code. Its input value must 129 * pass the U_SUCCESS() test, or else the function returns 130 * immediately. Check for U_FAILURE() on output or use with 131 * function chaining. (See User Guide for details.) 132 * @return the UTS #46 UIDNA instance, if successful 133 * @stable ICU 4.6 134 */ 135 U_STABLE UIDNA * U_EXPORT2 136 uidna_openUTS46(uint32_t options, UErrorCode *pErrorCode); 137 138 /** 139 * Closes a UIDNA instance. 140 * @param idna UIDNA instance to be closed 141 * @stable ICU 4.6 142 */ 143 U_STABLE void U_EXPORT2 144 uidna_close(UIDNA *idna); 145 146 #if U_SHOW_CPLUSPLUS_API 147 148 U_NAMESPACE_BEGIN 149 150 /** 151 * \class LocalUIDNAPointer 152 * "Smart pointer" class, closes a UIDNA via uidna_close(). 153 * For most methods see the LocalPointerBase base class. 154 * 155 * @see LocalPointerBase 156 * @see LocalPointer 157 * @stable ICU 4.6 158 */ 159 U_DEFINE_LOCAL_OPEN_POINTER(LocalUIDNAPointer, UIDNA, uidna_close); 160 161 U_NAMESPACE_END 162 163 #endif 164 165 /** 166 * Output container for IDNA processing errors. 167 * Initialize with UIDNA_INFO_INITIALIZER: 168 * \code 169 * UIDNAInfo info = UIDNA_INFO_INITIALIZER; 170 * int32_t length = uidna_nameToASCII(..., &info, &errorCode); 171 * if(U_SUCCESS(errorCode) && info.errors!=0) { ... } 172 * \endcode 173 * @stable ICU 4.6 174 */ 175 typedef struct UIDNAInfo { 176 /** sizeof(UIDNAInfo) @stable ICU 4.6 */ 177 int16_t size; 178 /** 179 * Set to TRUE if transitional and nontransitional processing produce different results. 180 * For details see C++ IDNAInfo::isTransitionalDifferent(). 181 * @stable ICU 4.6 182 */ 183 UBool isTransitionalDifferent; 184 UBool reservedB3; /**< Reserved field, do not use. @internal */ 185 /** 186 * Bit set indicating IDNA processing errors. 0 if no errors. 187 * See UIDNA_ERROR_... constants. 188 * @stable ICU 4.6 189 */ 190 uint32_t errors; 191 int32_t reservedI2; /**< Reserved field, do not use. @internal */ 192 int32_t reservedI3; /**< Reserved field, do not use. @internal */ 193 } UIDNAInfo; 194 195 /** 196 * Static initializer for a UIDNAInfo struct. 197 * @stable ICU 4.6 198 */ 199 #define UIDNA_INFO_INITIALIZER { \ 200 (int16_t)sizeof(UIDNAInfo), \ 201 FALSE, FALSE, \ 202 0, 0, 0 } 203 204 /** 205 * Converts a single domain name label into its ASCII form for DNS lookup. 206 * If any processing step fails, then pInfo->errors will be non-zero and 207 * the result might not be an ASCII string. 208 * The label might be modified according to the types of errors. 209 * Labels with severe errors will be left in (or turned into) their Unicode form. 210 * 211 * The UErrorCode indicates an error only in exceptional cases, 212 * such as a U_MEMORY_ALLOCATION_ERROR. 213 * 214 * @param idna UIDNA instance 215 * @param label Input domain name label 216 * @param length Label length, or -1 if NUL-terminated 217 * @param dest Destination string buffer 218 * @param capacity Destination buffer capacity 219 * @param pInfo Output container of IDNA processing details. 220 * @param pErrorCode Standard ICU error code. Its input value must 221 * pass the U_SUCCESS() test, or else the function returns 222 * immediately. Check for U_FAILURE() on output or use with 223 * function chaining. (See User Guide for details.) 224 * @return destination string length 225 * @stable ICU 4.6 226 */ 227 U_STABLE int32_t U_EXPORT2 228 uidna_labelToASCII(const UIDNA *idna, 229 const UChar *label, int32_t length, 230 UChar *dest, int32_t capacity, 231 UIDNAInfo *pInfo, UErrorCode *pErrorCode); 232 233 /** 234 * Converts a single domain name label into its Unicode form for human-readable display. 235 * If any processing step fails, then pInfo->errors will be non-zero. 236 * The label might be modified according to the types of errors. 237 * 238 * The UErrorCode indicates an error only in exceptional cases, 239 * such as a U_MEMORY_ALLOCATION_ERROR. 240 * 241 * @param idna UIDNA instance 242 * @param label Input domain name label 243 * @param length Label length, or -1 if NUL-terminated 244 * @param dest Destination string buffer 245 * @param capacity Destination buffer capacity 246 * @param pInfo Output container of IDNA processing details. 247 * @param pErrorCode Standard ICU error code. Its input value must 248 * pass the U_SUCCESS() test, or else the function returns 249 * immediately. Check for U_FAILURE() on output or use with 250 * function chaining. (See User Guide for details.) 251 * @return destination string length 252 * @stable ICU 4.6 253 */ 254 U_STABLE int32_t U_EXPORT2 255 uidna_labelToUnicode(const UIDNA *idna, 256 const UChar *label, int32_t length, 257 UChar *dest, int32_t capacity, 258 UIDNAInfo *pInfo, UErrorCode *pErrorCode); 259 260 /** 261 * Converts a whole domain name into its ASCII form for DNS lookup. 262 * If any processing step fails, then pInfo->errors will be non-zero and 263 * the result might not be an ASCII string. 264 * The domain name might be modified according to the types of errors. 265 * Labels with severe errors will be left in (or turned into) their Unicode form. 266 * 267 * The UErrorCode indicates an error only in exceptional cases, 268 * such as a U_MEMORY_ALLOCATION_ERROR. 269 * 270 * @param idna UIDNA instance 271 * @param name Input domain name 272 * @param length Domain name length, or -1 if NUL-terminated 273 * @param dest Destination string buffer 274 * @param capacity Destination buffer capacity 275 * @param pInfo Output container of IDNA processing details. 276 * @param pErrorCode Standard ICU error code. Its input value must 277 * pass the U_SUCCESS() test, or else the function returns 278 * immediately. Check for U_FAILURE() on output or use with 279 * function chaining. (See User Guide for details.) 280 * @return destination string length 281 * @stable ICU 4.6 282 */ 283 U_STABLE int32_t U_EXPORT2 284 uidna_nameToASCII(const UIDNA *idna, 285 const UChar *name, int32_t length, 286 UChar *dest, int32_t capacity, 287 UIDNAInfo *pInfo, UErrorCode *pErrorCode); 288 289 /** 290 * Converts a whole domain name into its Unicode form for human-readable display. 291 * If any processing step fails, then pInfo->errors will be non-zero. 292 * The domain name might be modified according to the types of errors. 293 * 294 * The UErrorCode indicates an error only in exceptional cases, 295 * such as a U_MEMORY_ALLOCATION_ERROR. 296 * 297 * @param idna UIDNA instance 298 * @param name Input domain name 299 * @param length Domain name length, or -1 if NUL-terminated 300 * @param dest Destination string buffer 301 * @param capacity Destination buffer capacity 302 * @param pInfo Output container of IDNA processing details. 303 * @param pErrorCode Standard ICU error code. Its input value must 304 * pass the U_SUCCESS() test, or else the function returns 305 * immediately. Check for U_FAILURE() on output or use with 306 * function chaining. (See User Guide for details.) 307 * @return destination string length 308 * @stable ICU 4.6 309 */ 310 U_STABLE int32_t U_EXPORT2 311 uidna_nameToUnicode(const UIDNA *idna, 312 const UChar *name, int32_t length, 313 UChar *dest, int32_t capacity, 314 UIDNAInfo *pInfo, UErrorCode *pErrorCode); 315 316 /* UTF-8 versions of the processing methods --------------------------------- */ 317 318 /** 319 * Converts a single domain name label into its ASCII form for DNS lookup. 320 * UTF-8 version of uidna_labelToASCII(), same behavior. 321 * 322 * @param idna UIDNA instance 323 * @param label Input domain name label 324 * @param length Label length, or -1 if NUL-terminated 325 * @param dest Destination string buffer 326 * @param capacity Destination buffer capacity 327 * @param pInfo Output container of IDNA processing details. 328 * @param pErrorCode Standard ICU error code. Its input value must 329 * pass the U_SUCCESS() test, or else the function returns 330 * immediately. Check for U_FAILURE() on output or use with 331 * function chaining. (See User Guide for details.) 332 * @return destination string length 333 * @stable ICU 4.6 334 */ 335 U_STABLE int32_t U_EXPORT2 336 uidna_labelToASCII_UTF8(const UIDNA *idna, 337 const char *label, int32_t length, 338 char *dest, int32_t capacity, 339 UIDNAInfo *pInfo, UErrorCode *pErrorCode); 340 341 /** 342 * Converts a single domain name label into its Unicode form for human-readable display. 343 * UTF-8 version of uidna_labelToUnicode(), same behavior. 344 * 345 * @param idna UIDNA instance 346 * @param label Input domain name label 347 * @param length Label length, or -1 if NUL-terminated 348 * @param dest Destination string buffer 349 * @param capacity Destination buffer capacity 350 * @param pInfo Output container of IDNA processing details. 351 * @param pErrorCode Standard ICU error code. Its input value must 352 * pass the U_SUCCESS() test, or else the function returns 353 * immediately. Check for U_FAILURE() on output or use with 354 * function chaining. (See User Guide for details.) 355 * @return destination string length 356 * @stable ICU 4.6 357 */ 358 U_STABLE int32_t U_EXPORT2 359 uidna_labelToUnicodeUTF8(const UIDNA *idna, 360 const char *label, int32_t length, 361 char *dest, int32_t capacity, 362 UIDNAInfo *pInfo, UErrorCode *pErrorCode); 363 364 /** 365 * Converts a whole domain name into its ASCII form for DNS lookup. 366 * UTF-8 version of uidna_nameToASCII(), same behavior. 367 * 368 * @param idna UIDNA instance 369 * @param name Input domain name 370 * @param length Domain name length, or -1 if NUL-terminated 371 * @param dest Destination string buffer 372 * @param capacity Destination buffer capacity 373 * @param pInfo Output container of IDNA processing details. 374 * @param pErrorCode Standard ICU error code. Its input value must 375 * pass the U_SUCCESS() test, or else the function returns 376 * immediately. Check for U_FAILURE() on output or use with 377 * function chaining. (See User Guide for details.) 378 * @return destination string length 379 * @stable ICU 4.6 380 */ 381 U_STABLE int32_t U_EXPORT2 382 uidna_nameToASCII_UTF8(const UIDNA *idna, 383 const char *name, int32_t length, 384 char *dest, int32_t capacity, 385 UIDNAInfo *pInfo, UErrorCode *pErrorCode); 386 387 /** 388 * Converts a whole domain name into its Unicode form for human-readable display. 389 * UTF-8 version of uidna_nameToUnicode(), same behavior. 390 * 391 * @param idna UIDNA instance 392 * @param name Input domain name 393 * @param length Domain name length, or -1 if NUL-terminated 394 * @param dest Destination string buffer 395 * @param capacity Destination buffer capacity 396 * @param pInfo Output container of IDNA processing details. 397 * @param pErrorCode Standard ICU error code. Its input value must 398 * pass the U_SUCCESS() test, or else the function returns 399 * immediately. Check for U_FAILURE() on output or use with 400 * function chaining. (See User Guide for details.) 401 * @return destination string length 402 * @stable ICU 4.6 403 */ 404 U_STABLE int32_t U_EXPORT2 405 uidna_nameToUnicodeUTF8(const UIDNA *idna, 406 const char *name, int32_t length, 407 char *dest, int32_t capacity, 408 UIDNAInfo *pInfo, UErrorCode *pErrorCode); 409 410 /* 411 * IDNA error bit set values. 412 * When a domain name or label fails a processing step or does not meet the 413 * validity criteria, then one or more of these error bits are set. 414 */ 415 enum { 416 /** 417 * A non-final domain name label (or the whole domain name) is empty. 418 * @stable ICU 4.6 419 */ 420 UIDNA_ERROR_EMPTY_LABEL=1, 421 /** 422 * A domain name label is longer than 63 bytes. 423 * (See STD13/RFC1034 3.1. Name space specifications and terminology.) 424 * This is only checked in ToASCII operations, and only if the output label is all-ASCII. 425 * @stable ICU 4.6 426 */ 427 UIDNA_ERROR_LABEL_TOO_LONG=2, 428 /** 429 * A domain name is longer than 255 bytes in its storage form. 430 * (See STD13/RFC1034 3.1. Name space specifications and terminology.) 431 * This is only checked in ToASCII operations, and only if the output domain name is all-ASCII. 432 * @stable ICU 4.6 433 */ 434 UIDNA_ERROR_DOMAIN_NAME_TOO_LONG=4, 435 /** 436 * A label starts with a hyphen-minus ('-'). 437 * @stable ICU 4.6 438 */ 439 UIDNA_ERROR_LEADING_HYPHEN=8, 440 /** 441 * A label ends with a hyphen-minus ('-'). 442 * @stable ICU 4.6 443 */ 444 UIDNA_ERROR_TRAILING_HYPHEN=0x10, 445 /** 446 * A label contains hyphen-minus ('-') in the third and fourth positions. 447 * @stable ICU 4.6 448 */ 449 UIDNA_ERROR_HYPHEN_3_4=0x20, 450 /** 451 * A label starts with a combining mark. 452 * @stable ICU 4.6 453 */ 454 UIDNA_ERROR_LEADING_COMBINING_MARK=0x40, 455 /** 456 * A label or domain name contains disallowed characters. 457 * @stable ICU 4.6 458 */ 459 UIDNA_ERROR_DISALLOWED=0x80, 460 /** 461 * A label starts with "xn--" but does not contain valid Punycode. 462 * That is, an xn-- label failed Punycode decoding. 463 * @stable ICU 4.6 464 */ 465 UIDNA_ERROR_PUNYCODE=0x100, 466 /** 467 * A label contains a dot=full stop. 468 * This can occur in an input string for a single-label function. 469 * @stable ICU 4.6 470 */ 471 UIDNA_ERROR_LABEL_HAS_DOT=0x200, 472 /** 473 * An ACE label does not contain a valid label string. 474 * The label was successfully ACE (Punycode) decoded but the resulting 475 * string had severe validation errors. For example, 476 * it might contain characters that are not allowed in ACE labels, 477 * or it might not be normalized. 478 * @stable ICU 4.6 479 */ 480 UIDNA_ERROR_INVALID_ACE_LABEL=0x400, 481 /** 482 * A label does not meet the IDNA BiDi requirements (for right-to-left characters). 483 * @stable ICU 4.6 484 */ 485 UIDNA_ERROR_BIDI=0x800, 486 /** 487 * A label does not meet the IDNA CONTEXTJ requirements. 488 * @stable ICU 4.6 489 */ 490 UIDNA_ERROR_CONTEXTJ=0x1000, 491 /** 492 * A label does not meet the IDNA CONTEXTO requirements for punctuation characters. 493 * Some punctuation characters "Would otherwise have been DISALLOWED" 494 * but are allowed in certain contexts. (RFC 5892) 495 * @stable ICU 49 496 */ 497 UIDNA_ERROR_CONTEXTO_PUNCTUATION=0x2000, 498 /** 499 * A label does not meet the IDNA CONTEXTO requirements for digits. 500 * Arabic-Indic Digits (U+066x) must not be mixed with Extended Arabic-Indic Digits (U+06Fx). 501 * @stable ICU 49 502 */ 503 UIDNA_ERROR_CONTEXTO_DIGITS=0x4000 504 }; 505 506 /* IDNA2003 API ------------------------------------------------------------- */ 507 508 /** 509 * IDNA2003: This function implements the ToASCII operation as defined in the IDNA RFC. 510 * This operation is done on <b>single labels</b> before sending it to something that expects 511 * ASCII names. A label is an individual part of a domain name. Labels are usually 512 * separated by dots; e.g. "www.example.com" is composed of 3 labels "www","example", and "com". 513 * 514 * IDNA2003 API Overview: 515 * 516 * The uidna_ API implements the IDNA protocol as defined in the IDNA RFC 517 * (http://www.ietf.org/rfc/rfc3490.txt). 518 * The RFC defines 2 operations: ToASCII and ToUnicode. Domain name labels 519 * containing non-ASCII code points are processed by the 520 * ToASCII operation before passing it to resolver libraries. Domain names 521 * that are obtained from resolver libraries are processed by the 522 * ToUnicode operation before displaying the domain name to the user. 523 * IDNA requires that implementations process input strings with Nameprep 524 * (http://www.ietf.org/rfc/rfc3491.txt), 525 * which is a profile of Stringprep (http://www.ietf.org/rfc/rfc3454.txt), 526 * and then with Punycode (http://www.ietf.org/rfc/rfc3492.txt). 527 * Implementations of IDNA MUST fully implement Nameprep and Punycode; 528 * neither Nameprep nor Punycode are optional. 529 * The input and output of ToASCII and ToUnicode operations are Unicode 530 * and are designed to be chainable, i.e., applying ToASCII or ToUnicode operations 531 * multiple times to an input string will yield the same result as applying the operation 532 * once. 533 * ToUnicode(ToUnicode(ToUnicode...(ToUnicode(string)))) == ToUnicode(string) 534 * ToASCII(ToASCII(ToASCII...(ToASCII(string))) == ToASCII(string). 535 * 536 * @param src Input UChar array containing label in Unicode. 537 * @param srcLength Number of UChars in src, or -1 if NUL-terminated. 538 * @param dest Output UChar array with ASCII (ACE encoded) label. 539 * @param destCapacity Size of dest. 540 * @param options A bit set of options: 541 * 542 * - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points 543 * and do not use STD3 ASCII rules 544 * If unassigned code points are found the operation fails with 545 * U_UNASSIGNED_ERROR error code. 546 * 547 * - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations 548 * If this option is set, the unassigned code points are in the input 549 * are treated as normal Unicode code points. 550 * 551 * - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions 552 * If this option is set and the input does not satisfy STD3 rules, 553 * the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR 554 * 555 * @param parseError Pointer to UParseError struct to receive information on position 556 * of error if an error is encountered. Can be NULL. 557 * @param status ICU in/out error code parameter. 558 * U_INVALID_CHAR_FOUND if src contains 559 * unmatched single surrogates. 560 * U_INDEX_OUTOFBOUNDS_ERROR if src contains 561 * too many code points. 562 * U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough 563 * @return The length of the result string, if successful - or in case of a buffer overflow, 564 * in which case it will be greater than destCapacity. 565 * @stable ICU 2.6 566 */ 567 U_STABLE int32_t U_EXPORT2 568 uidna_toASCII(const UChar* src, int32_t srcLength, 569 UChar* dest, int32_t destCapacity, 570 int32_t options, 571 UParseError* parseError, 572 UErrorCode* status); 573 574 575 /** 576 * IDNA2003: This function implements the ToUnicode operation as defined in the IDNA RFC. 577 * This operation is done on <b>single labels</b> before sending it to something that expects 578 * Unicode names. A label is an individual part of a domain name. Labels are usually 579 * separated by dots; for e.g. "www.example.com" is composed of 3 labels "www","example", and "com". 580 * 581 * @param src Input UChar array containing ASCII (ACE encoded) label. 582 * @param srcLength Number of UChars in src, or -1 if NUL-terminated. 583 * @param dest Output Converted UChar array containing Unicode equivalent of label. 584 * @param destCapacity Size of dest. 585 * @param options A bit set of options: 586 * 587 * - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points 588 * and do not use STD3 ASCII rules 589 * If unassigned code points are found the operation fails with 590 * U_UNASSIGNED_ERROR error code. 591 * 592 * - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations 593 * If this option is set, the unassigned code points are in the input 594 * are treated as normal Unicode code points. <b> Note: </b> This option is 595 * required on toUnicode operation because the RFC mandates 596 * verification of decoded ACE input by applying toASCII and comparing 597 * its output with source 598 * 599 * - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions 600 * If this option is set and the input does not satisfy STD3 rules, 601 * the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR 602 * 603 * @param parseError Pointer to UParseError struct to receive information on position 604 * of error if an error is encountered. Can be NULL. 605 * @param status ICU in/out error code parameter. 606 * U_INVALID_CHAR_FOUND if src contains 607 * unmatched single surrogates. 608 * U_INDEX_OUTOFBOUNDS_ERROR if src contains 609 * too many code points. 610 * U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough 611 * @return The length of the result string, if successful - or in case of a buffer overflow, 612 * in which case it will be greater than destCapacity. 613 * @stable ICU 2.6 614 */ 615 U_STABLE int32_t U_EXPORT2 616 uidna_toUnicode(const UChar* src, int32_t srcLength, 617 UChar* dest, int32_t destCapacity, 618 int32_t options, 619 UParseError* parseError, 620 UErrorCode* status); 621 622 623 /** 624 * IDNA2003: Convenience function that implements the IDNToASCII operation as defined in the IDNA RFC. 625 * This operation is done on complete domain names, e.g: "www.example.com". 626 * It is important to note that this operation can fail. If it fails, then the input 627 * domain name cannot be used as an Internationalized Domain Name and the application 628 * should have methods defined to deal with the failure. 629 * 630 * <b>Note:</b> IDNA RFC specifies that a conformant application should divide a domain name 631 * into separate labels, decide whether to apply allowUnassigned and useSTD3ASCIIRules on each, 632 * and then convert. This function does not offer that level of granularity. The options once 633 * set will apply to all labels in the domain name 634 * 635 * @param src Input UChar array containing IDN in Unicode. 636 * @param srcLength Number of UChars in src, or -1 if NUL-terminated. 637 * @param dest Output UChar array with ASCII (ACE encoded) IDN. 638 * @param destCapacity Size of dest. 639 * @param options A bit set of options: 640 * 641 * - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points 642 * and do not use STD3 ASCII rules 643 * If unassigned code points are found the operation fails with 644 * U_UNASSIGNED_CODE_POINT_FOUND error code. 645 * 646 * - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations 647 * If this option is set, the unassigned code points are in the input 648 * are treated as normal Unicode code points. 649 * 650 * - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions 651 * If this option is set and the input does not satisfy STD3 rules, 652 * the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR 653 * 654 * @param parseError Pointer to UParseError struct to receive information on position 655 * of error if an error is encountered. Can be NULL. 656 * @param status ICU in/out error code parameter. 657 * U_INVALID_CHAR_FOUND if src contains 658 * unmatched single surrogates. 659 * U_INDEX_OUTOFBOUNDS_ERROR if src contains 660 * too many code points. 661 * U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough 662 * @return The length of the result string, if successful - or in case of a buffer overflow, 663 * in which case it will be greater than destCapacity. 664 * @stable ICU 2.6 665 */ 666 U_STABLE int32_t U_EXPORT2 667 uidna_IDNToASCII( const UChar* src, int32_t srcLength, 668 UChar* dest, int32_t destCapacity, 669 int32_t options, 670 UParseError* parseError, 671 UErrorCode* status); 672 673 /** 674 * IDNA2003: Convenience function that implements the IDNToUnicode operation as defined in the IDNA RFC. 675 * This operation is done on complete domain names, e.g: "www.example.com". 676 * 677 * <b>Note:</b> IDNA RFC specifies that a conformant application should divide a domain name 678 * into separate labels, decide whether to apply allowUnassigned and useSTD3ASCIIRules on each, 679 * and then convert. This function does not offer that level of granularity. The options once 680 * set will apply to all labels in the domain name 681 * 682 * @param src Input UChar array containing IDN in ASCII (ACE encoded) form. 683 * @param srcLength Number of UChars in src, or -1 if NUL-terminated. 684 * @param dest Output UChar array containing Unicode equivalent of source IDN. 685 * @param destCapacity Size of dest. 686 * @param options A bit set of options: 687 * 688 * - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points 689 * and do not use STD3 ASCII rules 690 * If unassigned code points are found the operation fails with 691 * U_UNASSIGNED_CODE_POINT_FOUND error code. 692 * 693 * - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations 694 * If this option is set, the unassigned code points are in the input 695 * are treated as normal Unicode code points. 696 * 697 * - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions 698 * If this option is set and the input does not satisfy STD3 rules, 699 * the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR 700 * 701 * @param parseError Pointer to UParseError struct to receive information on position 702 * of error if an error is encountered. Can be NULL. 703 * @param status ICU in/out error code parameter. 704 * U_INVALID_CHAR_FOUND if src contains 705 * unmatched single surrogates. 706 * U_INDEX_OUTOFBOUNDS_ERROR if src contains 707 * too many code points. 708 * U_BUFFER_OVERFLOW_ERROR if destCapacity is not enough 709 * @return The length of the result string, if successful - or in case of a buffer overflow, 710 * in which case it will be greater than destCapacity. 711 * @stable ICU 2.6 712 */ 713 U_STABLE int32_t U_EXPORT2 714 uidna_IDNToUnicode( const UChar* src, int32_t srcLength, 715 UChar* dest, int32_t destCapacity, 716 int32_t options, 717 UParseError* parseError, 718 UErrorCode* status); 719 720 /** 721 * IDNA2003: Compare two IDN strings for equivalence. 722 * This function splits the domain names into labels and compares them. 723 * According to IDN RFC, whenever two labels are compared, they are 724 * considered equal if and only if their ASCII forms (obtained by 725 * applying toASCII) match using an case-insensitive ASCII comparison. 726 * Two domain names are considered a match if and only if all labels 727 * match regardless of whether label separators match. 728 * 729 * @param s1 First source string. 730 * @param length1 Length of first source string, or -1 if NUL-terminated. 731 * 732 * @param s2 Second source string. 733 * @param length2 Length of second source string, or -1 if NUL-terminated. 734 * @param options A bit set of options: 735 * 736 * - UIDNA_DEFAULT Use default options, i.e., do not process unassigned code points 737 * and do not use STD3 ASCII rules 738 * If unassigned code points are found the operation fails with 739 * U_UNASSIGNED_CODE_POINT_FOUND error code. 740 * 741 * - UIDNA_ALLOW_UNASSIGNED Unassigned values can be converted to ASCII for query operations 742 * If this option is set, the unassigned code points are in the input 743 * are treated as normal Unicode code points. 744 * 745 * - UIDNA_USE_STD3_RULES Use STD3 ASCII rules for host name syntax restrictions 746 * If this option is set and the input does not satisfy STD3 rules, 747 * the operation will fail with U_IDNA_STD3_ASCII_RULES_ERROR 748 * 749 * @param status ICU error code in/out parameter. 750 * Must fulfill U_SUCCESS before the function call. 751 * @return <0 or 0 or >0 as usual for string comparisons 752 * @stable ICU 2.6 753 */ 754 U_STABLE int32_t U_EXPORT2 755 uidna_compare( const UChar *s1, int32_t length1, 756 const UChar *s2, int32_t length2, 757 int32_t options, 758 UErrorCode* status); 759 760 #endif /* #if !UCONFIG_NO_IDNA */ 761 762 #endif 763