annotate CSP2/CSP2_env/env-d9b9114564458d9d-741b3de822f2aaca6c6caa4325c4afce/include/unicode/tblcoll.h @ 69:33d812a61356

planemo upload commit 2e9511a184a1ca667c7be0c6321a36dc4e3d116d
author jpayne
date Tue, 18 Mar 2025 17:55:14 -0400
parents
children
rev   line source
jpayne@69 1 // © 2016 and later: Unicode, Inc. and others.
jpayne@69 2 // License & terms of use: http://www.unicode.org/copyright.html
jpayne@69 3 /*
jpayne@69 4 ******************************************************************************
jpayne@69 5 * Copyright (C) 1996-2016, International Business Machines Corporation and
jpayne@69 6 * others. All Rights Reserved.
jpayne@69 7 ******************************************************************************
jpayne@69 8 */
jpayne@69 9
jpayne@69 10 /**
jpayne@69 11 * \file
jpayne@69 12 * \brief C++ API: The RuleBasedCollator class implements the Collator abstract base class.
jpayne@69 13 */
jpayne@69 14
jpayne@69 15 /**
jpayne@69 16 * File tblcoll.h
jpayne@69 17 *
jpayne@69 18 * Created by: Helena Shih
jpayne@69 19 *
jpayne@69 20 * Modification History:
jpayne@69 21 *
jpayne@69 22 * Date Name Description
jpayne@69 23 * 2/5/97 aliu Added streamIn and streamOut methods. Added
jpayne@69 24 * constructor which reads RuleBasedCollator object from
jpayne@69 25 * a binary file. Added writeToFile method which streams
jpayne@69 26 * RuleBasedCollator out to a binary file. The streamIn
jpayne@69 27 * and streamOut methods use istream and ostream objects
jpayne@69 28 * in binary mode.
jpayne@69 29 * 2/12/97 aliu Modified to use TableCollationData sub-object to
jpayne@69 30 * hold invariant data.
jpayne@69 31 * 2/13/97 aliu Moved several methods into this class from Collation.
jpayne@69 32 * Added a private RuleBasedCollator(Locale&) constructor,
jpayne@69 33 * to be used by Collator::createDefault(). General
jpayne@69 34 * clean up.
jpayne@69 35 * 2/20/97 helena Added clone, operator==, operator!=, operator=, and copy
jpayne@69 36 * constructor and getDynamicClassID.
jpayne@69 37 * 3/5/97 aliu Modified constructFromFile() to add parameter
jpayne@69 38 * specifying whether or not binary loading is to be
jpayne@69 39 * attempted. This is required for dynamic rule loading.
jpayne@69 40 * 05/07/97 helena Added memory allocation error detection.
jpayne@69 41 * 6/17/97 helena Added IDENTICAL strength for compare, changed getRules to
jpayne@69 42 * use MergeCollation::getPattern.
jpayne@69 43 * 6/20/97 helena Java class name change.
jpayne@69 44 * 8/18/97 helena Added internal API documentation.
jpayne@69 45 * 09/03/97 helena Added createCollationKeyValues().
jpayne@69 46 * 02/10/98 damiba Added compare with "length" parameter
jpayne@69 47 * 08/05/98 erm Synched with 1.2 version of RuleBasedCollator.java
jpayne@69 48 * 04/23/99 stephen Removed EDecompositionMode, merged with
jpayne@69 49 * Normalizer::EMode
jpayne@69 50 * 06/14/99 stephen Removed kResourceBundleSuffix
jpayne@69 51 * 11/02/99 helena Collator performance enhancements. Eliminates the
jpayne@69 52 * UnicodeString construction and special case for NO_OP.
jpayne@69 53 * 11/23/99 srl More performance enhancements. Updates to NormalizerIterator
jpayne@69 54 * internal state management.
jpayne@69 55 * 12/15/99 aliu Update to support Thai collation. Move NormalizerIterator
jpayne@69 56 * to implementation file.
jpayne@69 57 * 01/29/01 synwee Modified into a C++ wrapper which calls C API
jpayne@69 58 * (ucol.h)
jpayne@69 59 * 2012-2014 markus Rewritten in C++ again.
jpayne@69 60 */
jpayne@69 61
jpayne@69 62 #ifndef TBLCOLL_H
jpayne@69 63 #define TBLCOLL_H
jpayne@69 64
jpayne@69 65 #include "unicode/utypes.h"
jpayne@69 66
jpayne@69 67 #if U_SHOW_CPLUSPLUS_API
jpayne@69 68
jpayne@69 69 #if !UCONFIG_NO_COLLATION
jpayne@69 70
jpayne@69 71 #include "unicode/coll.h"
jpayne@69 72 #include "unicode/locid.h"
jpayne@69 73 #include "unicode/uiter.h"
jpayne@69 74 #include "unicode/ucol.h"
jpayne@69 75
jpayne@69 76 U_NAMESPACE_BEGIN
jpayne@69 77
jpayne@69 78 struct CollationCacheEntry;
jpayne@69 79 struct CollationData;
jpayne@69 80 struct CollationSettings;
jpayne@69 81 struct CollationTailoring;
jpayne@69 82 /**
jpayne@69 83 * @stable ICU 2.0
jpayne@69 84 */
jpayne@69 85 class StringSearch;
jpayne@69 86 /**
jpayne@69 87 * @stable ICU 2.0
jpayne@69 88 */
jpayne@69 89 class CollationElementIterator;
jpayne@69 90 class CollationKey;
jpayne@69 91 class SortKeyByteSink;
jpayne@69 92 class UnicodeSet;
jpayne@69 93 class UnicodeString;
jpayne@69 94 class UVector64;
jpayne@69 95
jpayne@69 96 /**
jpayne@69 97 * The RuleBasedCollator class provides the implementation of
jpayne@69 98 * Collator, using data-driven tables. The user can create a customized
jpayne@69 99 * table-based collation.
jpayne@69 100 * <p>
jpayne@69 101 * For more information about the collation service see
jpayne@69 102 * <a href="http://userguide.icu-project.org/collation">the User Guide</a>.
jpayne@69 103 * <p>
jpayne@69 104 * Collation service provides correct sorting orders for most locales supported in ICU.
jpayne@69 105 * If specific data for a locale is not available, the orders eventually falls back
jpayne@69 106 * to the <a href="http://www.unicode.org/reports/tr35/tr35-collation.html#Root_Collation">CLDR root sort order</a>.
jpayne@69 107 * <p>
jpayne@69 108 * Sort ordering may be customized by providing your own set of rules. For more on
jpayne@69 109 * this subject see the <a href="http://userguide.icu-project.org/collation/customization">
jpayne@69 110 * Collation Customization</a> section of the User Guide.
jpayne@69 111 * <p>
jpayne@69 112 * Note, RuleBasedCollator is not to be subclassed.
jpayne@69 113 * @see Collator
jpayne@69 114 */
jpayne@69 115 class U_I18N_API RuleBasedCollator : public Collator {
jpayne@69 116 public:
jpayne@69 117 /**
jpayne@69 118 * RuleBasedCollator constructor. This takes the table rules and builds a
jpayne@69 119 * collation table out of them. Please see RuleBasedCollator class
jpayne@69 120 * description for more details on the collation rule syntax.
jpayne@69 121 * @param rules the collation rules to build the collation table from.
jpayne@69 122 * @param status reporting a success or an error.
jpayne@69 123 * @stable ICU 2.0
jpayne@69 124 */
jpayne@69 125 RuleBasedCollator(const UnicodeString& rules, UErrorCode& status);
jpayne@69 126
jpayne@69 127 /**
jpayne@69 128 * RuleBasedCollator constructor. This takes the table rules and builds a
jpayne@69 129 * collation table out of them. Please see RuleBasedCollator class
jpayne@69 130 * description for more details on the collation rule syntax.
jpayne@69 131 * @param rules the collation rules to build the collation table from.
jpayne@69 132 * @param collationStrength strength for comparison
jpayne@69 133 * @param status reporting a success or an error.
jpayne@69 134 * @stable ICU 2.0
jpayne@69 135 */
jpayne@69 136 RuleBasedCollator(const UnicodeString& rules,
jpayne@69 137 ECollationStrength collationStrength,
jpayne@69 138 UErrorCode& status);
jpayne@69 139
jpayne@69 140 /**
jpayne@69 141 * RuleBasedCollator constructor. This takes the table rules and builds a
jpayne@69 142 * collation table out of them. Please see RuleBasedCollator class
jpayne@69 143 * description for more details on the collation rule syntax.
jpayne@69 144 * @param rules the collation rules to build the collation table from.
jpayne@69 145 * @param decompositionMode the normalisation mode
jpayne@69 146 * @param status reporting a success or an error.
jpayne@69 147 * @stable ICU 2.0
jpayne@69 148 */
jpayne@69 149 RuleBasedCollator(const UnicodeString& rules,
jpayne@69 150 UColAttributeValue decompositionMode,
jpayne@69 151 UErrorCode& status);
jpayne@69 152
jpayne@69 153 /**
jpayne@69 154 * RuleBasedCollator constructor. This takes the table rules and builds a
jpayne@69 155 * collation table out of them. Please see RuleBasedCollator class
jpayne@69 156 * description for more details on the collation rule syntax.
jpayne@69 157 * @param rules the collation rules to build the collation table from.
jpayne@69 158 * @param collationStrength strength for comparison
jpayne@69 159 * @param decompositionMode the normalisation mode
jpayne@69 160 * @param status reporting a success or an error.
jpayne@69 161 * @stable ICU 2.0
jpayne@69 162 */
jpayne@69 163 RuleBasedCollator(const UnicodeString& rules,
jpayne@69 164 ECollationStrength collationStrength,
jpayne@69 165 UColAttributeValue decompositionMode,
jpayne@69 166 UErrorCode& status);
jpayne@69 167
jpayne@69 168 #ifndef U_HIDE_INTERNAL_API
jpayne@69 169 /**
jpayne@69 170 * TODO: document & propose as public API
jpayne@69 171 * @internal
jpayne@69 172 */
jpayne@69 173 RuleBasedCollator(const UnicodeString &rules,
jpayne@69 174 UParseError &parseError, UnicodeString &reason,
jpayne@69 175 UErrorCode &errorCode);
jpayne@69 176 #endif /* U_HIDE_INTERNAL_API */
jpayne@69 177
jpayne@69 178 /**
jpayne@69 179 * Copy constructor.
jpayne@69 180 * @param other the RuleBasedCollator object to be copied
jpayne@69 181 * @stable ICU 2.0
jpayne@69 182 */
jpayne@69 183 RuleBasedCollator(const RuleBasedCollator& other);
jpayne@69 184
jpayne@69 185
jpayne@69 186 /** Opens a collator from a collator binary image created using
jpayne@69 187 * cloneBinary. Binary image used in instantiation of the
jpayne@69 188 * collator remains owned by the user and should stay around for
jpayne@69 189 * the lifetime of the collator. The API also takes a base collator
jpayne@69 190 * which must be the root collator.
jpayne@69 191 * @param bin binary image owned by the user and required through the
jpayne@69 192 * lifetime of the collator
jpayne@69 193 * @param length size of the image. If negative, the API will try to
jpayne@69 194 * figure out the length of the image
jpayne@69 195 * @param base Base collator, for lookup of untailored characters.
jpayne@69 196 * Must be the root collator, must not be NULL.
jpayne@69 197 * The base is required to be present through the lifetime of the collator.
jpayne@69 198 * @param status for catching errors
jpayne@69 199 * @return newly created collator
jpayne@69 200 * @see cloneBinary
jpayne@69 201 * @stable ICU 3.4
jpayne@69 202 */
jpayne@69 203 RuleBasedCollator(const uint8_t *bin, int32_t length,
jpayne@69 204 const RuleBasedCollator *base,
jpayne@69 205 UErrorCode &status);
jpayne@69 206
jpayne@69 207 /**
jpayne@69 208 * Destructor.
jpayne@69 209 * @stable ICU 2.0
jpayne@69 210 */
jpayne@69 211 virtual ~RuleBasedCollator();
jpayne@69 212
jpayne@69 213 /**
jpayne@69 214 * Assignment operator.
jpayne@69 215 * @param other other RuleBasedCollator object to copy from.
jpayne@69 216 * @stable ICU 2.0
jpayne@69 217 */
jpayne@69 218 RuleBasedCollator& operator=(const RuleBasedCollator& other);
jpayne@69 219
jpayne@69 220 /**
jpayne@69 221 * Returns true if argument is the same as this object.
jpayne@69 222 * @param other Collator object to be compared.
jpayne@69 223 * @return true if arguments is the same as this object.
jpayne@69 224 * @stable ICU 2.0
jpayne@69 225 */
jpayne@69 226 virtual UBool operator==(const Collator& other) const;
jpayne@69 227
jpayne@69 228 /**
jpayne@69 229 * Makes a copy of this object.
jpayne@69 230 * @return a copy of this object, owned by the caller
jpayne@69 231 * @stable ICU 2.0
jpayne@69 232 */
jpayne@69 233 virtual RuleBasedCollator* clone() const;
jpayne@69 234
jpayne@69 235 /**
jpayne@69 236 * Creates a collation element iterator for the source string. The caller of
jpayne@69 237 * this method is responsible for the memory management of the return
jpayne@69 238 * pointer.
jpayne@69 239 * @param source the string over which the CollationElementIterator will
jpayne@69 240 * iterate.
jpayne@69 241 * @return the collation element iterator of the source string using this as
jpayne@69 242 * the based Collator.
jpayne@69 243 * @stable ICU 2.2
jpayne@69 244 */
jpayne@69 245 virtual CollationElementIterator* createCollationElementIterator(
jpayne@69 246 const UnicodeString& source) const;
jpayne@69 247
jpayne@69 248 /**
jpayne@69 249 * Creates a collation element iterator for the source. The caller of this
jpayne@69 250 * method is responsible for the memory management of the returned pointer.
jpayne@69 251 * @param source the CharacterIterator which produces the characters over
jpayne@69 252 * which the CollationElementItgerator will iterate.
jpayne@69 253 * @return the collation element iterator of the source using this as the
jpayne@69 254 * based Collator.
jpayne@69 255 * @stable ICU 2.2
jpayne@69 256 */
jpayne@69 257 virtual CollationElementIterator* createCollationElementIterator(
jpayne@69 258 const CharacterIterator& source) const;
jpayne@69 259
jpayne@69 260 // Make deprecated versions of Collator::compare() visible.
jpayne@69 261 using Collator::compare;
jpayne@69 262
jpayne@69 263 /**
jpayne@69 264 * The comparison function compares the character data stored in two
jpayne@69 265 * different strings. Returns information about whether a string is less
jpayne@69 266 * than, greater than or equal to another string.
jpayne@69 267 * @param source the source string to be compared with.
jpayne@69 268 * @param target the string that is to be compared with the source string.
jpayne@69 269 * @param status possible error code
jpayne@69 270 * @return Returns an enum value. UCOL_GREATER if source is greater
jpayne@69 271 * than target; UCOL_EQUAL if source is equal to target; UCOL_LESS if source is less
jpayne@69 272 * than target
jpayne@69 273 * @stable ICU 2.6
jpayne@69 274 **/
jpayne@69 275 virtual UCollationResult compare(const UnicodeString& source,
jpayne@69 276 const UnicodeString& target,
jpayne@69 277 UErrorCode &status) const;
jpayne@69 278
jpayne@69 279 /**
jpayne@69 280 * Does the same thing as compare but limits the comparison to a specified
jpayne@69 281 * length
jpayne@69 282 * @param source the source string to be compared with.
jpayne@69 283 * @param target the string that is to be compared with the source string.
jpayne@69 284 * @param length the length the comparison is limited to
jpayne@69 285 * @param status possible error code
jpayne@69 286 * @return Returns an enum value. UCOL_GREATER if source (up to the specified
jpayne@69 287 * length) is greater than target; UCOL_EQUAL if source (up to specified
jpayne@69 288 * length) is equal to target; UCOL_LESS if source (up to the specified
jpayne@69 289 * length) is less than target.
jpayne@69 290 * @stable ICU 2.6
jpayne@69 291 */
jpayne@69 292 virtual UCollationResult compare(const UnicodeString& source,
jpayne@69 293 const UnicodeString& target,
jpayne@69 294 int32_t length,
jpayne@69 295 UErrorCode &status) const;
jpayne@69 296
jpayne@69 297 /**
jpayne@69 298 * The comparison function compares the character data stored in two
jpayne@69 299 * different string arrays. Returns information about whether a string array
jpayne@69 300 * is less than, greater than or equal to another string array.
jpayne@69 301 * @param source the source string array to be compared with.
jpayne@69 302 * @param sourceLength the length of the source string array. If this value
jpayne@69 303 * is equal to -1, the string array is null-terminated.
jpayne@69 304 * @param target the string that is to be compared with the source string.
jpayne@69 305 * @param targetLength the length of the target string array. If this value
jpayne@69 306 * is equal to -1, the string array is null-terminated.
jpayne@69 307 * @param status possible error code
jpayne@69 308 * @return Returns an enum value. UCOL_GREATER if source is greater
jpayne@69 309 * than target; UCOL_EQUAL if source is equal to target; UCOL_LESS if source is less
jpayne@69 310 * than target
jpayne@69 311 * @stable ICU 2.6
jpayne@69 312 */
jpayne@69 313 virtual UCollationResult compare(const char16_t* source, int32_t sourceLength,
jpayne@69 314 const char16_t* target, int32_t targetLength,
jpayne@69 315 UErrorCode &status) const;
jpayne@69 316
jpayne@69 317 /**
jpayne@69 318 * Compares two strings using the Collator.
jpayne@69 319 * Returns whether the first one compares less than/equal to/greater than
jpayne@69 320 * the second one.
jpayne@69 321 * This version takes UCharIterator input.
jpayne@69 322 * @param sIter the first ("source") string iterator
jpayne@69 323 * @param tIter the second ("target") string iterator
jpayne@69 324 * @param status ICU status
jpayne@69 325 * @return UCOL_LESS, UCOL_EQUAL or UCOL_GREATER
jpayne@69 326 * @stable ICU 4.2
jpayne@69 327 */
jpayne@69 328 virtual UCollationResult compare(UCharIterator &sIter,
jpayne@69 329 UCharIterator &tIter,
jpayne@69 330 UErrorCode &status) const;
jpayne@69 331
jpayne@69 332 /**
jpayne@69 333 * Compares two UTF-8 strings using the Collator.
jpayne@69 334 * Returns whether the first one compares less than/equal to/greater than
jpayne@69 335 * the second one.
jpayne@69 336 * This version takes UTF-8 input.
jpayne@69 337 * Note that a StringPiece can be implicitly constructed
jpayne@69 338 * from a std::string or a NUL-terminated const char * string.
jpayne@69 339 * @param source the first UTF-8 string
jpayne@69 340 * @param target the second UTF-8 string
jpayne@69 341 * @param status ICU status
jpayne@69 342 * @return UCOL_LESS, UCOL_EQUAL or UCOL_GREATER
jpayne@69 343 * @stable ICU 51
jpayne@69 344 */
jpayne@69 345 virtual UCollationResult compareUTF8(const StringPiece &source,
jpayne@69 346 const StringPiece &target,
jpayne@69 347 UErrorCode &status) const;
jpayne@69 348
jpayne@69 349 /**
jpayne@69 350 * Transforms the string into a series of characters
jpayne@69 351 * that can be compared with CollationKey.compare().
jpayne@69 352 *
jpayne@69 353 * Note that sort keys are often less efficient than simply doing comparison.
jpayne@69 354 * For more details, see the ICU User Guide.
jpayne@69 355 *
jpayne@69 356 * @param source the source string.
jpayne@69 357 * @param key the transformed key of the source string.
jpayne@69 358 * @param status the error code status.
jpayne@69 359 * @return the transformed key.
jpayne@69 360 * @see CollationKey
jpayne@69 361 * @stable ICU 2.0
jpayne@69 362 */
jpayne@69 363 virtual CollationKey& getCollationKey(const UnicodeString& source,
jpayne@69 364 CollationKey& key,
jpayne@69 365 UErrorCode& status) const;
jpayne@69 366
jpayne@69 367 /**
jpayne@69 368 * Transforms a specified region of the string into a series of characters
jpayne@69 369 * that can be compared with CollationKey.compare.
jpayne@69 370 *
jpayne@69 371 * Note that sort keys are often less efficient than simply doing comparison.
jpayne@69 372 * For more details, see the ICU User Guide.
jpayne@69 373 *
jpayne@69 374 * @param source the source string.
jpayne@69 375 * @param sourceLength the length of the source string.
jpayne@69 376 * @param key the transformed key of the source string.
jpayne@69 377 * @param status the error code status.
jpayne@69 378 * @return the transformed key.
jpayne@69 379 * @see CollationKey
jpayne@69 380 * @stable ICU 2.0
jpayne@69 381 */
jpayne@69 382 virtual CollationKey& getCollationKey(const char16_t *source,
jpayne@69 383 int32_t sourceLength,
jpayne@69 384 CollationKey& key,
jpayne@69 385 UErrorCode& status) const;
jpayne@69 386
jpayne@69 387 /**
jpayne@69 388 * Generates the hash code for the rule-based collation object.
jpayne@69 389 * @return the hash code.
jpayne@69 390 * @stable ICU 2.0
jpayne@69 391 */
jpayne@69 392 virtual int32_t hashCode() const;
jpayne@69 393
jpayne@69 394 #ifndef U_FORCE_HIDE_DEPRECATED_API
jpayne@69 395 /**
jpayne@69 396 * Gets the locale of the Collator
jpayne@69 397 * @param type can be either requested, valid or actual locale. For more
jpayne@69 398 * information see the definition of ULocDataLocaleType in
jpayne@69 399 * uloc.h
jpayne@69 400 * @param status the error code status.
jpayne@69 401 * @return locale where the collation data lives. If the collator
jpayne@69 402 * was instantiated from rules, locale is empty.
jpayne@69 403 * @deprecated ICU 2.8 likely to change in ICU 3.0, based on feedback
jpayne@69 404 */
jpayne@69 405 virtual Locale getLocale(ULocDataLocaleType type, UErrorCode& status) const;
jpayne@69 406 #endif // U_FORCE_HIDE_DEPRECATED_API
jpayne@69 407
jpayne@69 408 /**
jpayne@69 409 * Gets the tailoring rules for this collator.
jpayne@69 410 * @return the collation tailoring from which this collator was created
jpayne@69 411 * @stable ICU 2.0
jpayne@69 412 */
jpayne@69 413 const UnicodeString& getRules() const;
jpayne@69 414
jpayne@69 415 /**
jpayne@69 416 * Gets the version information for a Collator.
jpayne@69 417 * @param info the version # information, the result will be filled in
jpayne@69 418 * @stable ICU 2.0
jpayne@69 419 */
jpayne@69 420 virtual void getVersion(UVersionInfo info) const;
jpayne@69 421
jpayne@69 422 #ifndef U_HIDE_DEPRECATED_API
jpayne@69 423 /**
jpayne@69 424 * Returns the maximum length of any expansion sequences that end with the
jpayne@69 425 * specified comparison order.
jpayne@69 426 *
jpayne@69 427 * This is specific to the kind of collation element values and sequences
jpayne@69 428 * returned by the CollationElementIterator.
jpayne@69 429 * Call CollationElementIterator::getMaxExpansion() instead.
jpayne@69 430 *
jpayne@69 431 * @param order a collation order returned by CollationElementIterator::previous
jpayne@69 432 * or CollationElementIterator::next.
jpayne@69 433 * @return maximum size of the expansion sequences ending with the collation
jpayne@69 434 * element, or 1 if the collation element does not occur at the end of
jpayne@69 435 * any expansion sequence
jpayne@69 436 * @see CollationElementIterator#getMaxExpansion
jpayne@69 437 * @deprecated ICU 51 Use CollationElementIterator::getMaxExpansion() instead.
jpayne@69 438 */
jpayne@69 439 int32_t getMaxExpansion(int32_t order) const;
jpayne@69 440 #endif /* U_HIDE_DEPRECATED_API */
jpayne@69 441
jpayne@69 442 /**
jpayne@69 443 * Returns a unique class ID POLYMORPHICALLY. Pure virtual override. This
jpayne@69 444 * method is to implement a simple version of RTTI, since not all C++
jpayne@69 445 * compilers support genuine RTTI. Polymorphic operator==() and clone()
jpayne@69 446 * methods call this method.
jpayne@69 447 * @return The class ID for this object. All objects of a given class have
jpayne@69 448 * the same class ID. Objects of other classes have different class
jpayne@69 449 * IDs.
jpayne@69 450 * @stable ICU 2.0
jpayne@69 451 */
jpayne@69 452 virtual UClassID getDynamicClassID(void) const;
jpayne@69 453
jpayne@69 454 /**
jpayne@69 455 * Returns the class ID for this class. This is useful only for comparing to
jpayne@69 456 * a return value from getDynamicClassID(). For example:
jpayne@69 457 * <pre>
jpayne@69 458 * Base* polymorphic_pointer = createPolymorphicObject();
jpayne@69 459 * if (polymorphic_pointer->getDynamicClassID() ==
jpayne@69 460 * Derived::getStaticClassID()) ...
jpayne@69 461 * </pre>
jpayne@69 462 * @return The class ID for all objects of this class.
jpayne@69 463 * @stable ICU 2.0
jpayne@69 464 */
jpayne@69 465 static UClassID U_EXPORT2 getStaticClassID(void);
jpayne@69 466
jpayne@69 467 #ifndef U_HIDE_DEPRECATED_API
jpayne@69 468 /**
jpayne@69 469 * Do not use this method: The caller and the ICU library might use different heaps.
jpayne@69 470 * Use cloneBinary() instead which writes to caller-provided memory.
jpayne@69 471 *
jpayne@69 472 * Returns a binary format of this collator.
jpayne@69 473 * @param length Returns the length of the data, in bytes
jpayne@69 474 * @param status the error code status.
jpayne@69 475 * @return memory, owned by the caller, of size 'length' bytes.
jpayne@69 476 * @deprecated ICU 52. Use cloneBinary() instead.
jpayne@69 477 */
jpayne@69 478 uint8_t *cloneRuleData(int32_t &length, UErrorCode &status) const;
jpayne@69 479 #endif /* U_HIDE_DEPRECATED_API */
jpayne@69 480
jpayne@69 481 /** Creates a binary image of a collator. This binary image can be stored and
jpayne@69 482 * later used to instantiate a collator using ucol_openBinary.
jpayne@69 483 * This API supports preflighting.
jpayne@69 484 * @param buffer a fill-in buffer to receive the binary image
jpayne@69 485 * @param capacity capacity of the destination buffer
jpayne@69 486 * @param status for catching errors
jpayne@69 487 * @return size of the image
jpayne@69 488 * @see ucol_openBinary
jpayne@69 489 * @stable ICU 3.4
jpayne@69 490 */
jpayne@69 491 int32_t cloneBinary(uint8_t *buffer, int32_t capacity, UErrorCode &status) const;
jpayne@69 492
jpayne@69 493 /**
jpayne@69 494 * Returns current rules. Delta defines whether full rules are returned or
jpayne@69 495 * just the tailoring.
jpayne@69 496 *
jpayne@69 497 * getRules(void) should normally be used instead.
jpayne@69 498 * See http://userguide.icu-project.org/collation/customization#TOC-Building-on-Existing-Locales
jpayne@69 499 * @param delta one of UCOL_TAILORING_ONLY, UCOL_FULL_RULES.
jpayne@69 500 * @param buffer UnicodeString to store the result rules
jpayne@69 501 * @stable ICU 2.2
jpayne@69 502 * @see UCOL_FULL_RULES
jpayne@69 503 */
jpayne@69 504 void getRules(UColRuleOption delta, UnicodeString &buffer) const;
jpayne@69 505
jpayne@69 506 /**
jpayne@69 507 * Universal attribute setter
jpayne@69 508 * @param attr attribute type
jpayne@69 509 * @param value attribute value
jpayne@69 510 * @param status to indicate whether the operation went on smoothly or there were errors
jpayne@69 511 * @stable ICU 2.2
jpayne@69 512 */
jpayne@69 513 virtual void setAttribute(UColAttribute attr, UColAttributeValue value,
jpayne@69 514 UErrorCode &status);
jpayne@69 515
jpayne@69 516 /**
jpayne@69 517 * Universal attribute getter.
jpayne@69 518 * @param attr attribute type
jpayne@69 519 * @param status to indicate whether the operation went on smoothly or there were errors
jpayne@69 520 * @return attribute value
jpayne@69 521 * @stable ICU 2.2
jpayne@69 522 */
jpayne@69 523 virtual UColAttributeValue getAttribute(UColAttribute attr,
jpayne@69 524 UErrorCode &status) const;
jpayne@69 525
jpayne@69 526 /**
jpayne@69 527 * Sets the variable top to the top of the specified reordering group.
jpayne@69 528 * The variable top determines the highest-sorting character
jpayne@69 529 * which is affected by UCOL_ALTERNATE_HANDLING.
jpayne@69 530 * If that attribute is set to UCOL_NON_IGNORABLE, then the variable top has no effect.
jpayne@69 531 * @param group one of UCOL_REORDER_CODE_SPACE, UCOL_REORDER_CODE_PUNCTUATION,
jpayne@69 532 * UCOL_REORDER_CODE_SYMBOL, UCOL_REORDER_CODE_CURRENCY;
jpayne@69 533 * or UCOL_REORDER_CODE_DEFAULT to restore the default max variable group
jpayne@69 534 * @param errorCode Standard ICU error code. Its input value must
jpayne@69 535 * pass the U_SUCCESS() test, or else the function returns
jpayne@69 536 * immediately. Check for U_FAILURE() on output or use with
jpayne@69 537 * function chaining. (See User Guide for details.)
jpayne@69 538 * @return *this
jpayne@69 539 * @see getMaxVariable
jpayne@69 540 * @stable ICU 53
jpayne@69 541 */
jpayne@69 542 virtual Collator &setMaxVariable(UColReorderCode group, UErrorCode &errorCode);
jpayne@69 543
jpayne@69 544 /**
jpayne@69 545 * Returns the maximum reordering group whose characters are affected by UCOL_ALTERNATE_HANDLING.
jpayne@69 546 * @return the maximum variable reordering group.
jpayne@69 547 * @see setMaxVariable
jpayne@69 548 * @stable ICU 53
jpayne@69 549 */
jpayne@69 550 virtual UColReorderCode getMaxVariable() const;
jpayne@69 551
jpayne@69 552 #ifndef U_FORCE_HIDE_DEPRECATED_API
jpayne@69 553 /**
jpayne@69 554 * Sets the variable top to the primary weight of the specified string.
jpayne@69 555 *
jpayne@69 556 * Beginning with ICU 53, the variable top is pinned to
jpayne@69 557 * the top of one of the supported reordering groups,
jpayne@69 558 * and it must not be beyond the last of those groups.
jpayne@69 559 * See setMaxVariable().
jpayne@69 560 * @param varTop one or more (if contraction) char16_ts to which the variable top should be set
jpayne@69 561 * @param len length of variable top string. If -1 it is considered to be zero terminated.
jpayne@69 562 * @param status error code. If error code is set, the return value is undefined. Errors set by this function are: <br>
jpayne@69 563 * U_CE_NOT_FOUND_ERROR if more than one character was passed and there is no such contraction<br>
jpayne@69 564 * U_ILLEGAL_ARGUMENT_ERROR if the variable top is beyond
jpayne@69 565 * the last reordering group supported by setMaxVariable()
jpayne@69 566 * @return variable top primary weight
jpayne@69 567 * @deprecated ICU 53 Call setMaxVariable() instead.
jpayne@69 568 */
jpayne@69 569 virtual uint32_t setVariableTop(const char16_t *varTop, int32_t len, UErrorCode &status);
jpayne@69 570
jpayne@69 571 /**
jpayne@69 572 * Sets the variable top to the primary weight of the specified string.
jpayne@69 573 *
jpayne@69 574 * Beginning with ICU 53, the variable top is pinned to
jpayne@69 575 * the top of one of the supported reordering groups,
jpayne@69 576 * and it must not be beyond the last of those groups.
jpayne@69 577 * See setMaxVariable().
jpayne@69 578 * @param varTop a UnicodeString size 1 or more (if contraction) of char16_ts to which the variable top should be set
jpayne@69 579 * @param status error code. If error code is set, the return value is undefined. Errors set by this function are: <br>
jpayne@69 580 * U_CE_NOT_FOUND_ERROR if more than one character was passed and there is no such contraction<br>
jpayne@69 581 * U_ILLEGAL_ARGUMENT_ERROR if the variable top is beyond
jpayne@69 582 * the last reordering group supported by setMaxVariable()
jpayne@69 583 * @return variable top primary weight
jpayne@69 584 * @deprecated ICU 53 Call setMaxVariable() instead.
jpayne@69 585 */
jpayne@69 586 virtual uint32_t setVariableTop(const UnicodeString &varTop, UErrorCode &status);
jpayne@69 587
jpayne@69 588 /**
jpayne@69 589 * Sets the variable top to the specified primary weight.
jpayne@69 590 *
jpayne@69 591 * Beginning with ICU 53, the variable top is pinned to
jpayne@69 592 * the top of one of the supported reordering groups,
jpayne@69 593 * and it must not be beyond the last of those groups.
jpayne@69 594 * See setMaxVariable().
jpayne@69 595 * @param varTop primary weight, as returned by setVariableTop or ucol_getVariableTop
jpayne@69 596 * @param status error code
jpayne@69 597 * @deprecated ICU 53 Call setMaxVariable() instead.
jpayne@69 598 */
jpayne@69 599 virtual void setVariableTop(uint32_t varTop, UErrorCode &status);
jpayne@69 600 #endif // U_FORCE_HIDE_DEPRECATED_API
jpayne@69 601
jpayne@69 602 /**
jpayne@69 603 * Gets the variable top value of a Collator.
jpayne@69 604 * @param status error code (not changed by function). If error code is set, the return value is undefined.
jpayne@69 605 * @return the variable top primary weight
jpayne@69 606 * @see getMaxVariable
jpayne@69 607 * @stable ICU 2.0
jpayne@69 608 */
jpayne@69 609 virtual uint32_t getVariableTop(UErrorCode &status) const;
jpayne@69 610
jpayne@69 611 /**
jpayne@69 612 * Get a UnicodeSet that contains all the characters and sequences tailored in
jpayne@69 613 * this collator.
jpayne@69 614 * @param status error code of the operation
jpayne@69 615 * @return a pointer to a UnicodeSet object containing all the
jpayne@69 616 * code points and sequences that may sort differently than
jpayne@69 617 * in the root collator. The object must be disposed of by using delete
jpayne@69 618 * @stable ICU 2.4
jpayne@69 619 */
jpayne@69 620 virtual UnicodeSet *getTailoredSet(UErrorCode &status) const;
jpayne@69 621
jpayne@69 622 /**
jpayne@69 623 * Get the sort key as an array of bytes from a UnicodeString.
jpayne@69 624 *
jpayne@69 625 * Note that sort keys are often less efficient than simply doing comparison.
jpayne@69 626 * For more details, see the ICU User Guide.
jpayne@69 627 *
jpayne@69 628 * @param source string to be processed.
jpayne@69 629 * @param result buffer to store result in. If NULL, number of bytes needed
jpayne@69 630 * will be returned.
jpayne@69 631 * @param resultLength length of the result buffer. If if not enough the
jpayne@69 632 * buffer will be filled to capacity.
jpayne@69 633 * @return Number of bytes needed for storing the sort key
jpayne@69 634 * @stable ICU 2.0
jpayne@69 635 */
jpayne@69 636 virtual int32_t getSortKey(const UnicodeString& source, uint8_t *result,
jpayne@69 637 int32_t resultLength) const;
jpayne@69 638
jpayne@69 639 /**
jpayne@69 640 * Get the sort key as an array of bytes from a char16_t buffer.
jpayne@69 641 *
jpayne@69 642 * Note that sort keys are often less efficient than simply doing comparison.
jpayne@69 643 * For more details, see the ICU User Guide.
jpayne@69 644 *
jpayne@69 645 * @param source string to be processed.
jpayne@69 646 * @param sourceLength length of string to be processed. If -1, the string
jpayne@69 647 * is 0 terminated and length will be decided by the function.
jpayne@69 648 * @param result buffer to store result in. If NULL, number of bytes needed
jpayne@69 649 * will be returned.
jpayne@69 650 * @param resultLength length of the result buffer. If if not enough the
jpayne@69 651 * buffer will be filled to capacity.
jpayne@69 652 * @return Number of bytes needed for storing the sort key
jpayne@69 653 * @stable ICU 2.2
jpayne@69 654 */
jpayne@69 655 virtual int32_t getSortKey(const char16_t *source, int32_t sourceLength,
jpayne@69 656 uint8_t *result, int32_t resultLength) const;
jpayne@69 657
jpayne@69 658 /**
jpayne@69 659 * Retrieves the reordering codes for this collator.
jpayne@69 660 * @param dest The array to fill with the script ordering.
jpayne@69 661 * @param destCapacity The length of dest. If it is 0, then dest may be NULL and the function
jpayne@69 662 * will only return the length of the result without writing any codes (pre-flighting).
jpayne@69 663 * @param status A reference to an error code value, which must not indicate
jpayne@69 664 * a failure before the function call.
jpayne@69 665 * @return The length of the script ordering array.
jpayne@69 666 * @see ucol_setReorderCodes
jpayne@69 667 * @see Collator#getEquivalentReorderCodes
jpayne@69 668 * @see Collator#setReorderCodes
jpayne@69 669 * @stable ICU 4.8
jpayne@69 670 */
jpayne@69 671 virtual int32_t getReorderCodes(int32_t *dest,
jpayne@69 672 int32_t destCapacity,
jpayne@69 673 UErrorCode& status) const;
jpayne@69 674
jpayne@69 675 /**
jpayne@69 676 * Sets the ordering of scripts for this collator.
jpayne@69 677 * @param reorderCodes An array of script codes in the new order. This can be NULL if the
jpayne@69 678 * length is also set to 0. An empty array will clear any reordering codes on the collator.
jpayne@69 679 * @param reorderCodesLength The length of reorderCodes.
jpayne@69 680 * @param status error code
jpayne@69 681 * @see ucol_setReorderCodes
jpayne@69 682 * @see Collator#getReorderCodes
jpayne@69 683 * @see Collator#getEquivalentReorderCodes
jpayne@69 684 * @stable ICU 4.8
jpayne@69 685 */
jpayne@69 686 virtual void setReorderCodes(const int32_t* reorderCodes,
jpayne@69 687 int32_t reorderCodesLength,
jpayne@69 688 UErrorCode& status) ;
jpayne@69 689
jpayne@69 690 /**
jpayne@69 691 * Implements ucol_strcollUTF8().
jpayne@69 692 * @internal
jpayne@69 693 */
jpayne@69 694 virtual UCollationResult internalCompareUTF8(
jpayne@69 695 const char *left, int32_t leftLength,
jpayne@69 696 const char *right, int32_t rightLength,
jpayne@69 697 UErrorCode &errorCode) const;
jpayne@69 698
jpayne@69 699 /** Get the short definition string for a collator. This internal API harvests the collator's
jpayne@69 700 * locale and the attribute set and produces a string that can be used for opening
jpayne@69 701 * a collator with the same attributes using the ucol_openFromShortString API.
jpayne@69 702 * This string will be normalized.
jpayne@69 703 * The structure and the syntax of the string is defined in the "Naming collators"
jpayne@69 704 * section of the users guide:
jpayne@69 705 * http://userguide.icu-project.org/collation/concepts#TOC-Collator-naming-scheme
jpayne@69 706 * This function supports preflighting.
jpayne@69 707 *
jpayne@69 708 * This is internal, and intended to be used with delegate converters.
jpayne@69 709 *
jpayne@69 710 * @param locale a locale that will appear as a collators locale in the resulting
jpayne@69 711 * short string definition. If NULL, the locale will be harvested
jpayne@69 712 * from the collator.
jpayne@69 713 * @param buffer space to hold the resulting string
jpayne@69 714 * @param capacity capacity of the buffer
jpayne@69 715 * @param status for returning errors. All the preflighting errors are featured
jpayne@69 716 * @return length of the resulting string
jpayne@69 717 * @see ucol_openFromShortString
jpayne@69 718 * @see ucol_normalizeShortDefinitionString
jpayne@69 719 * @see ucol_getShortDefinitionString
jpayne@69 720 * @internal
jpayne@69 721 */
jpayne@69 722 virtual int32_t internalGetShortDefinitionString(const char *locale,
jpayne@69 723 char *buffer,
jpayne@69 724 int32_t capacity,
jpayne@69 725 UErrorCode &status) const;
jpayne@69 726
jpayne@69 727 /**
jpayne@69 728 * Implements ucol_nextSortKeyPart().
jpayne@69 729 * @internal
jpayne@69 730 */
jpayne@69 731 virtual int32_t internalNextSortKeyPart(
jpayne@69 732 UCharIterator *iter, uint32_t state[2],
jpayne@69 733 uint8_t *dest, int32_t count, UErrorCode &errorCode) const;
jpayne@69 734
jpayne@69 735 // Do not enclose the default constructor with #ifndef U_HIDE_INTERNAL_API
jpayne@69 736 /**
jpayne@69 737 * Only for use in ucol_openRules().
jpayne@69 738 * @internal
jpayne@69 739 */
jpayne@69 740 RuleBasedCollator();
jpayne@69 741
jpayne@69 742 #ifndef U_HIDE_INTERNAL_API
jpayne@69 743 /**
jpayne@69 744 * Implements ucol_getLocaleByType().
jpayne@69 745 * Needed because the lifetime of the locale ID string must match that of the collator.
jpayne@69 746 * getLocale() returns a copy of a Locale, with minimal lifetime in a C wrapper.
jpayne@69 747 * @internal
jpayne@69 748 */
jpayne@69 749 const char *internalGetLocaleID(ULocDataLocaleType type, UErrorCode &errorCode) const;
jpayne@69 750
jpayne@69 751 /**
jpayne@69 752 * Implements ucol_getContractionsAndExpansions().
jpayne@69 753 * Gets this collator's sets of contraction strings and/or
jpayne@69 754 * characters and strings that map to multiple collation elements (expansions).
jpayne@69 755 * If addPrefixes is TRUE, then contractions that are expressed as
jpayne@69 756 * prefix/pre-context rules are included.
jpayne@69 757 * @param contractions if not NULL, the set to hold the contractions
jpayne@69 758 * @param expansions if not NULL, the set to hold the expansions
jpayne@69 759 * @param addPrefixes include prefix contextual mappings
jpayne@69 760 * @param errorCode in/out ICU error code
jpayne@69 761 * @internal
jpayne@69 762 */
jpayne@69 763 void internalGetContractionsAndExpansions(
jpayne@69 764 UnicodeSet *contractions, UnicodeSet *expansions,
jpayne@69 765 UBool addPrefixes, UErrorCode &errorCode) const;
jpayne@69 766
jpayne@69 767 /**
jpayne@69 768 * Adds the contractions that start with character c to the set.
jpayne@69 769 * Ignores prefixes. Used by AlphabeticIndex.
jpayne@69 770 * @internal
jpayne@69 771 */
jpayne@69 772 void internalAddContractions(UChar32 c, UnicodeSet &set, UErrorCode &errorCode) const;
jpayne@69 773
jpayne@69 774 /**
jpayne@69 775 * Implements from-rule constructors, and ucol_openRules().
jpayne@69 776 * @internal
jpayne@69 777 */
jpayne@69 778 void internalBuildTailoring(
jpayne@69 779 const UnicodeString &rules,
jpayne@69 780 int32_t strength,
jpayne@69 781 UColAttributeValue decompositionMode,
jpayne@69 782 UParseError *outParseError, UnicodeString *outReason,
jpayne@69 783 UErrorCode &errorCode);
jpayne@69 784
jpayne@69 785 /** @internal */
jpayne@69 786 static inline RuleBasedCollator *rbcFromUCollator(UCollator *uc) {
jpayne@69 787 return dynamic_cast<RuleBasedCollator *>(fromUCollator(uc));
jpayne@69 788 }
jpayne@69 789 /** @internal */
jpayne@69 790 static inline const RuleBasedCollator *rbcFromUCollator(const UCollator *uc) {
jpayne@69 791 return dynamic_cast<const RuleBasedCollator *>(fromUCollator(uc));
jpayne@69 792 }
jpayne@69 793
jpayne@69 794 /**
jpayne@69 795 * Appends the CEs for the string to the vector.
jpayne@69 796 * @internal for tests & tools
jpayne@69 797 */
jpayne@69 798 void internalGetCEs(const UnicodeString &str, UVector64 &ces, UErrorCode &errorCode) const;
jpayne@69 799 #endif // U_HIDE_INTERNAL_API
jpayne@69 800
jpayne@69 801 protected:
jpayne@69 802 /**
jpayne@69 803 * Used internally by registration to define the requested and valid locales.
jpayne@69 804 * @param requestedLocale the requested locale
jpayne@69 805 * @param validLocale the valid locale
jpayne@69 806 * @param actualLocale the actual locale
jpayne@69 807 * @internal
jpayne@69 808 */
jpayne@69 809 virtual void setLocales(const Locale& requestedLocale, const Locale& validLocale, const Locale& actualLocale);
jpayne@69 810
jpayne@69 811 private:
jpayne@69 812 friend class CollationElementIterator;
jpayne@69 813 friend class Collator;
jpayne@69 814
jpayne@69 815 RuleBasedCollator(const CollationCacheEntry *entry);
jpayne@69 816
jpayne@69 817 /**
jpayne@69 818 * Enumeration of attributes that are relevant for short definition strings
jpayne@69 819 * (e.g., ucol_getShortDefinitionString()).
jpayne@69 820 * Effectively extends UColAttribute.
jpayne@69 821 */
jpayne@69 822 enum Attributes {
jpayne@69 823 ATTR_VARIABLE_TOP = UCOL_ATTRIBUTE_COUNT,
jpayne@69 824 ATTR_LIMIT
jpayne@69 825 };
jpayne@69 826
jpayne@69 827 void adoptTailoring(CollationTailoring *t, UErrorCode &errorCode);
jpayne@69 828
jpayne@69 829 // Both lengths must be <0 or else both must be >=0.
jpayne@69 830 UCollationResult doCompare(const char16_t *left, int32_t leftLength,
jpayne@69 831 const char16_t *right, int32_t rightLength,
jpayne@69 832 UErrorCode &errorCode) const;
jpayne@69 833 UCollationResult doCompare(const uint8_t *left, int32_t leftLength,
jpayne@69 834 const uint8_t *right, int32_t rightLength,
jpayne@69 835 UErrorCode &errorCode) const;
jpayne@69 836
jpayne@69 837 void writeSortKey(const char16_t *s, int32_t length,
jpayne@69 838 SortKeyByteSink &sink, UErrorCode &errorCode) const;
jpayne@69 839
jpayne@69 840 void writeIdenticalLevel(const char16_t *s, const char16_t *limit,
jpayne@69 841 SortKeyByteSink &sink, UErrorCode &errorCode) const;
jpayne@69 842
jpayne@69 843 const CollationSettings &getDefaultSettings() const;
jpayne@69 844
jpayne@69 845 void setAttributeDefault(int32_t attribute) {
jpayne@69 846 explicitlySetAttributes &= ~((uint32_t)1 << attribute);
jpayne@69 847 }
jpayne@69 848 void setAttributeExplicitly(int32_t attribute) {
jpayne@69 849 explicitlySetAttributes |= (uint32_t)1 << attribute;
jpayne@69 850 }
jpayne@69 851 UBool attributeHasBeenSetExplicitly(int32_t attribute) const {
jpayne@69 852 // assert(0 <= attribute < ATTR_LIMIT);
jpayne@69 853 return (UBool)((explicitlySetAttributes & ((uint32_t)1 << attribute)) != 0);
jpayne@69 854 }
jpayne@69 855
jpayne@69 856 /**
jpayne@69 857 * Tests whether a character is "unsafe" for use as a collation starting point.
jpayne@69 858 *
jpayne@69 859 * @param c code point or code unit
jpayne@69 860 * @return TRUE if c is unsafe
jpayne@69 861 * @see CollationElementIterator#setOffset(int)
jpayne@69 862 */
jpayne@69 863 UBool isUnsafe(UChar32 c) const;
jpayne@69 864
jpayne@69 865 static void U_CALLCONV computeMaxExpansions(const CollationTailoring *t, UErrorCode &errorCode);
jpayne@69 866 UBool initMaxExpansions(UErrorCode &errorCode) const;
jpayne@69 867
jpayne@69 868 void setFastLatinOptions(CollationSettings &ownedSettings) const;
jpayne@69 869
jpayne@69 870 const CollationData *data;
jpayne@69 871 const CollationSettings *settings; // reference-counted
jpayne@69 872 const CollationTailoring *tailoring; // alias of cacheEntry->tailoring
jpayne@69 873 const CollationCacheEntry *cacheEntry; // reference-counted
jpayne@69 874 Locale validLocale;
jpayne@69 875 uint32_t explicitlySetAttributes;
jpayne@69 876
jpayne@69 877 UBool actualLocaleIsSameAsValid;
jpayne@69 878 };
jpayne@69 879
jpayne@69 880 U_NAMESPACE_END
jpayne@69 881
jpayne@69 882 #endif // !UCONFIG_NO_COLLATION
jpayne@69 883
jpayne@69 884 #endif /* U_SHOW_CPLUSPLUS_API */
jpayne@69 885
jpayne@69 886 #endif // TBLCOLL_H