Mercurial > repos > rliterman > csp2

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

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
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) 1996-2015, International Business Machines Corporation and others.
jpayne@69 6 * All Rights Reserved.
jpayne@69 7 *******************************************************************************
jpayne@69 8 */
jpayne@69 9
jpayne@69 10 #ifndef UCOL_H
jpayne@69 11 #define UCOL_H
jpayne@69 12
jpayne@69 13 #include "unicode/utypes.h"
jpayne@69 14
jpayne@69 15 #if !UCONFIG_NO_COLLATION
jpayne@69 16
jpayne@69 17 #include "unicode/unorm.h"
jpayne@69 18 #include "unicode/localpointer.h"
jpayne@69 19 #include "unicode/parseerr.h"
jpayne@69 20 #include "unicode/uloc.h"
jpayne@69 21 #include "unicode/uset.h"
jpayne@69 22 #include "unicode/uscript.h"
jpayne@69 23
jpayne@69 24 /**
jpayne@69 25 * \file
jpayne@69 26 * \brief C API: Collator
jpayne@69 27 *
jpayne@69 28 * <h2> Collator C API </h2>
jpayne@69 29 *
jpayne@69 30 * The C API for Collator performs locale-sensitive
jpayne@69 31 * string comparison. You use this service to build
jpayne@69 32 * searching and sorting routines for natural language text.
jpayne@69 33 * <p>
jpayne@69 34 * For more information about the collation service see
jpayne@69 35 * <a href="http://userguide.icu-project.org/collation">the User Guide</a>.
jpayne@69 36 * <p>
jpayne@69 37 * Collation service provides correct sorting orders for most locales supported in ICU.
jpayne@69 38 * If specific data for a locale is not available, the orders eventually falls back
jpayne@69 39 * to the <a href="http://www.unicode.org/reports/tr35/tr35-collation.html#Root_Collation">CLDR root sort order</a>.
jpayne@69 40 * <p>
jpayne@69 41 * Sort ordering may be customized by providing your own set of rules. For more on
jpayne@69 42 * this subject see the <a href="http://userguide.icu-project.org/collation/customization">
jpayne@69 43 * Collation Customization</a> section of the User Guide.
jpayne@69 44 * <p>
jpayne@69 45 * @see UCollationResult
jpayne@69 46 * @see UNormalizationMode
jpayne@69 47 * @see UCollationStrength
jpayne@69 48 * @see UCollationElements
jpayne@69 49 */
jpayne@69 50
jpayne@69 51 /** A collator.
jpayne@69 52 * For usage in C programs.
jpayne@69 53 */
jpayne@69 54 struct UCollator;
jpayne@69 55 /** structure representing a collator object instance
jpayne@69 56 * @stable ICU 2.0
jpayne@69 57 */
jpayne@69 58 typedef struct UCollator UCollator;
jpayne@69 59
jpayne@69 60
jpayne@69 61 /**
jpayne@69 62 * UCOL_LESS is returned if source string is compared to be less than target
jpayne@69 63 * string in the ucol_strcoll() method.
jpayne@69 64 * UCOL_EQUAL is returned if source string is compared to be equal to target
jpayne@69 65 * string in the ucol_strcoll() method.
jpayne@69 66 * UCOL_GREATER is returned if source string is compared to be greater than
jpayne@69 67 * target string in the ucol_strcoll() method.
jpayne@69 68 * @see ucol_strcoll()
jpayne@69 69 * <p>
jpayne@69 70 * Possible values for a comparison result
jpayne@69 71 * @stable ICU 2.0
jpayne@69 72 */
jpayne@69 73 typedef enum {
jpayne@69 74 /** string a == string b */
jpayne@69 75 UCOL_EQUAL = 0,
jpayne@69 76 /** string a > string b */
jpayne@69 77 UCOL_GREATER = 1,
jpayne@69 78 /** string a < string b */
jpayne@69 79 UCOL_LESS = -1
jpayne@69 80 } UCollationResult ;
jpayne@69 81
jpayne@69 82
jpayne@69 83 /** Enum containing attribute values for controling collation behavior.
jpayne@69 84 * Here are all the allowable values. Not every attribute can take every value. The only
jpayne@69 85 * universal value is UCOL_DEFAULT, which resets the attribute value to the predefined
jpayne@69 86 * value for that locale
jpayne@69 87 * @stable ICU 2.0
jpayne@69 88 */
jpayne@69 89 typedef enum {
jpayne@69 90 /** accepted by most attributes */
jpayne@69 91 UCOL_DEFAULT = -1,
jpayne@69 92
jpayne@69 93 /** Primary collation strength */
jpayne@69 94 UCOL_PRIMARY = 0,
jpayne@69 95 /** Secondary collation strength */
jpayne@69 96 UCOL_SECONDARY = 1,
jpayne@69 97 /** Tertiary collation strength */
jpayne@69 98 UCOL_TERTIARY = 2,
jpayne@69 99 /** Default collation strength */
jpayne@69 100 UCOL_DEFAULT_STRENGTH = UCOL_TERTIARY,
jpayne@69 101 UCOL_CE_STRENGTH_LIMIT,
jpayne@69 102 /** Quaternary collation strength */
jpayne@69 103 UCOL_QUATERNARY=3,
jpayne@69 104 /** Identical collation strength */
jpayne@69 105 UCOL_IDENTICAL=15,
jpayne@69 106 UCOL_STRENGTH_LIMIT,
jpayne@69 107
jpayne@69 108 /** Turn the feature off - works for UCOL_FRENCH_COLLATION,
jpayne@69 109 UCOL_CASE_LEVEL, UCOL_HIRAGANA_QUATERNARY_MODE
jpayne@69 110 & UCOL_DECOMPOSITION_MODE*/
jpayne@69 111 UCOL_OFF = 16,
jpayne@69 112 /** Turn the feature on - works for UCOL_FRENCH_COLLATION,
jpayne@69 113 UCOL_CASE_LEVEL, UCOL_HIRAGANA_QUATERNARY_MODE
jpayne@69 114 & UCOL_DECOMPOSITION_MODE*/
jpayne@69 115 UCOL_ON = 17,
jpayne@69 116
jpayne@69 117 /** Valid for UCOL_ALTERNATE_HANDLING. Alternate handling will be shifted */
jpayne@69 118 UCOL_SHIFTED = 20,
jpayne@69 119 /** Valid for UCOL_ALTERNATE_HANDLING. Alternate handling will be non ignorable */
jpayne@69 120 UCOL_NON_IGNORABLE = 21,
jpayne@69 121
jpayne@69 122 /** Valid for UCOL_CASE_FIRST -
jpayne@69 123 lower case sorts before upper case */
jpayne@69 124 UCOL_LOWER_FIRST = 24,
jpayne@69 125 /** upper case sorts before lower case */
jpayne@69 126 UCOL_UPPER_FIRST = 25,
jpayne@69 127
jpayne@69 128 #ifndef U_HIDE_DEPRECATED_API
jpayne@69 129 /**
jpayne@69 130 * One more than the highest normal UColAttributeValue value.
jpayne@69 131 * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
jpayne@69 132 */
jpayne@69 133 UCOL_ATTRIBUTE_VALUE_COUNT
jpayne@69 134 #endif /* U_HIDE_DEPRECATED_API */
jpayne@69 135 } UColAttributeValue;
jpayne@69 136
jpayne@69 137 /**
jpayne@69 138 * Enum containing the codes for reordering segments of the collation table that are not script
jpayne@69 139 * codes. These reordering codes are to be used in conjunction with the script codes.
jpayne@69 140 * @see ucol_getReorderCodes
jpayne@69 141 * @see ucol_setReorderCodes
jpayne@69 142 * @see ucol_getEquivalentReorderCodes
jpayne@69 143 * @see UScriptCode
jpayne@69 144 * @stable ICU 4.8
jpayne@69 145 */
jpayne@69 146 typedef enum {
jpayne@69 147 /**
jpayne@69 148 * A special reordering code that is used to specify the default
jpayne@69 149 * reordering codes for a locale.
jpayne@69 150 * @stable ICU 4.8
jpayne@69 151 */
jpayne@69 152 UCOL_REORDER_CODE_DEFAULT = -1,
jpayne@69 153 /**
jpayne@69 154 * A special reordering code that is used to specify no reordering codes.
jpayne@69 155 * @stable ICU 4.8
jpayne@69 156 */
jpayne@69 157 UCOL_REORDER_CODE_NONE = USCRIPT_UNKNOWN,
jpayne@69 158 /**
jpayne@69 159 * A special reordering code that is used to specify all other codes used for
jpayne@69 160 * reordering except for the codes lised as UColReorderCode values and those
jpayne@69 161 * listed explicitly in a reordering.
jpayne@69 162 * @stable ICU 4.8
jpayne@69 163 */
jpayne@69 164 UCOL_REORDER_CODE_OTHERS = USCRIPT_UNKNOWN,
jpayne@69 165 /**
jpayne@69 166 * Characters with the space property.
jpayne@69 167 * This is equivalent to the rule value "space".
jpayne@69 168 * @stable ICU 4.8
jpayne@69 169 */
jpayne@69 170 UCOL_REORDER_CODE_SPACE = 0x1000,
jpayne@69 171 /**
jpayne@69 172 * The first entry in the enumeration of reordering groups. This is intended for use in
jpayne@69 173 * range checking and enumeration of the reorder codes.
jpayne@69 174 * @stable ICU 4.8
jpayne@69 175 */
jpayne@69 176 UCOL_REORDER_CODE_FIRST = UCOL_REORDER_CODE_SPACE,
jpayne@69 177 /**
jpayne@69 178 * Characters with the punctuation property.
jpayne@69 179 * This is equivalent to the rule value "punct".
jpayne@69 180 * @stable ICU 4.8
jpayne@69 181 */
jpayne@69 182 UCOL_REORDER_CODE_PUNCTUATION = 0x1001,
jpayne@69 183 /**
jpayne@69 184 * Characters with the symbol property.
jpayne@69 185 * This is equivalent to the rule value "symbol".
jpayne@69 186 * @stable ICU 4.8
jpayne@69 187 */
jpayne@69 188 UCOL_REORDER_CODE_SYMBOL = 0x1002,
jpayne@69 189 /**
jpayne@69 190 * Characters with the currency property.
jpayne@69 191 * This is equivalent to the rule value "currency".
jpayne@69 192 * @stable ICU 4.8
jpayne@69 193 */
jpayne@69 194 UCOL_REORDER_CODE_CURRENCY = 0x1003,
jpayne@69 195 /**
jpayne@69 196 * Characters with the digit property.
jpayne@69 197 * This is equivalent to the rule value "digit".
jpayne@69 198 * @stable ICU 4.8
jpayne@69 199 */
jpayne@69 200 UCOL_REORDER_CODE_DIGIT = 0x1004,
jpayne@69 201 #ifndef U_HIDE_DEPRECATED_API
jpayne@69 202 /**
jpayne@69 203 * One more than the highest normal UColReorderCode value.
jpayne@69 204 * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
jpayne@69 205 */
jpayne@69 206 UCOL_REORDER_CODE_LIMIT = 0x1005
jpayne@69 207 #endif /* U_HIDE_DEPRECATED_API */
jpayne@69 208 } UColReorderCode;
jpayne@69 209
jpayne@69 210 /**
jpayne@69 211 * Base letter represents a primary difference. Set comparison
jpayne@69 212 * level to UCOL_PRIMARY to ignore secondary and tertiary differences.
jpayne@69 213 * Use this to set the strength of a Collator object.
jpayne@69 214 * Example of primary difference, "abc" &lt; "abd"
jpayne@69 215 *
jpayne@69 216 * Diacritical differences on the same base letter represent a secondary
jpayne@69 217 * difference. Set comparison level to UCOL_SECONDARY to ignore tertiary
jpayne@69 218 * differences. Use this to set the strength of a Collator object.
jpayne@69 219 * Example of secondary difference, "&auml;" >> "a".
jpayne@69 220 *
jpayne@69 221 * Uppercase and lowercase versions of the same character represents a
jpayne@69 222 * tertiary difference. Set comparison level to UCOL_TERTIARY to include
jpayne@69 223 * all comparison differences. Use this to set the strength of a Collator
jpayne@69 224 * object.
jpayne@69 225 * Example of tertiary difference, "abc" &lt;&lt;&lt; "ABC".
jpayne@69 226 *
jpayne@69 227 * Two characters are considered "identical" when they have the same
jpayne@69 228 * unicode spellings. UCOL_IDENTICAL.
jpayne@69 229 * For example, "&auml;" == "&auml;".
jpayne@69 230 *
jpayne@69 231 * UCollationStrength is also used to determine the strength of sort keys
jpayne@69 232 * generated from UCollator objects
jpayne@69 233 * These values can be now found in the UColAttributeValue enum.
jpayne@69 234 * @stable ICU 2.0
jpayne@69 235 **/
jpayne@69 236 typedef UColAttributeValue UCollationStrength;
jpayne@69 237
jpayne@69 238 /** Attributes that collation service understands. All the attributes can take UCOL_DEFAULT
jpayne@69 239 * value, as well as the values specific to each one.
jpayne@69 240 * @stable ICU 2.0
jpayne@69 241 */
jpayne@69 242 typedef enum {
jpayne@69 243 /** Attribute for direction of secondary weights - used in Canadian French.
jpayne@69 244 * Acceptable values are UCOL_ON, which results in secondary weights
jpayne@69 245 * being considered backwards and UCOL_OFF which treats secondary
jpayne@69 246 * weights in the order they appear.
jpayne@69 247 * @stable ICU 2.0
jpayne@69 248 */
jpayne@69 249 UCOL_FRENCH_COLLATION,
jpayne@69 250 /** Attribute for handling variable elements.
jpayne@69 251 * Acceptable values are UCOL_NON_IGNORABLE (default)
jpayne@69 252 * which treats all the codepoints with non-ignorable
jpayne@69 253 * primary weights in the same way,
jpayne@69 254 * and UCOL_SHIFTED which causes codepoints with primary
jpayne@69 255 * weights that are equal or below the variable top value
jpayne@69 256 * to be ignored on primary level and moved to the quaternary
jpayne@69 257 * level.
jpayne@69 258 * @stable ICU 2.0
jpayne@69 259 */
jpayne@69 260 UCOL_ALTERNATE_HANDLING,
jpayne@69 261 /** Controls the ordering of upper and lower case letters.
jpayne@69 262 * Acceptable values are UCOL_OFF (default), which orders
jpayne@69 263 * upper and lower case letters in accordance to their tertiary
jpayne@69 264 * weights, UCOL_UPPER_FIRST which forces upper case letters to
jpayne@69 265 * sort before lower case letters, and UCOL_LOWER_FIRST which does
jpayne@69 266 * the opposite.
jpayne@69 267 * @stable ICU 2.0
jpayne@69 268 */
jpayne@69 269 UCOL_CASE_FIRST,
jpayne@69 270 /** Controls whether an extra case level (positioned before the third
jpayne@69 271 * level) is generated or not. Acceptable values are UCOL_OFF (default),
jpayne@69 272 * when case level is not generated, and UCOL_ON which causes the case
jpayne@69 273 * level to be generated. Contents of the case level are affected by
jpayne@69 274 * the value of UCOL_CASE_FIRST attribute. A simple way to ignore
jpayne@69 275 * accent differences in a string is to set the strength to UCOL_PRIMARY
jpayne@69 276 * and enable case level.
jpayne@69 277 * @stable ICU 2.0
jpayne@69 278 */
jpayne@69 279 UCOL_CASE_LEVEL,
jpayne@69 280 /** Controls whether the normalization check and necessary normalizations
jpayne@69 281 * are performed. When set to UCOL_OFF (default) no normalization check
jpayne@69 282 * is performed. The correctness of the result is guaranteed only if the
jpayne@69 283 * input data is in so-called FCD form (see users manual for more info).
jpayne@69 284 * When set to UCOL_ON, an incremental check is performed to see whether
jpayne@69 285 * the input data is in the FCD form. If the data is not in the FCD form,
jpayne@69 286 * incremental NFD normalization is performed.
jpayne@69 287 * @stable ICU 2.0
jpayne@69 288 */
jpayne@69 289 UCOL_NORMALIZATION_MODE,
jpayne@69 290 /** An alias for UCOL_NORMALIZATION_MODE attribute.
jpayne@69 291 * @stable ICU 2.0
jpayne@69 292 */
jpayne@69 293 UCOL_DECOMPOSITION_MODE = UCOL_NORMALIZATION_MODE,
jpayne@69 294 /** The strength attribute. Can be either UCOL_PRIMARY, UCOL_SECONDARY,
jpayne@69 295 * UCOL_TERTIARY, UCOL_QUATERNARY or UCOL_IDENTICAL. The usual strength
jpayne@69 296 * for most locales (except Japanese) is tertiary.
jpayne@69 297 *
jpayne@69 298 * Quaternary strength
jpayne@69 299 * is useful when combined with shifted setting for alternate handling
jpayne@69 300 * attribute and for JIS X 4061 collation, when it is used to distinguish
jpayne@69 301 * between Katakana and Hiragana.
jpayne@69 302 * Otherwise, quaternary level
jpayne@69 303 * is affected only by the number of non-ignorable code points in
jpayne@69 304 * the string.
jpayne@69 305 *
jpayne@69 306 * Identical strength is rarely useful, as it amounts
jpayne@69 307 * to codepoints of the NFD form of the string.
jpayne@69 308 * @stable ICU 2.0
jpayne@69 309 */
jpayne@69 310 UCOL_STRENGTH,
jpayne@69 311 #ifndef U_HIDE_DEPRECATED_API
jpayne@69 312 /** When turned on, this attribute positions Hiragana before all
jpayne@69 313 * non-ignorables on quaternary level This is a sneaky way to produce JIS
jpayne@69 314 * sort order.
jpayne@69 315 *
jpayne@69 316 * This attribute was an implementation detail of the CLDR Japanese tailoring.
jpayne@69 317 * Since ICU 50, this attribute is not settable any more via API functions.
jpayne@69 318 * Since CLDR 25/ICU 53, explicit quaternary relations are used
jpayne@69 319 * to achieve the same Japanese sort order.
jpayne@69 320 *
jpayne@69 321 * @deprecated ICU 50 Implementation detail, cannot be set via API, was removed from implementation.
jpayne@69 322 */
jpayne@69 323 UCOL_HIRAGANA_QUATERNARY_MODE = UCOL_STRENGTH + 1,
jpayne@69 324 #endif /* U_HIDE_DEPRECATED_API */
jpayne@69 325 /**
jpayne@69 326 * When turned on, this attribute makes
jpayne@69 327 * substrings of digits sort according to their numeric values.
jpayne@69 328 *
jpayne@69 329 * This is a way to get '100' to sort AFTER '2'. Note that the longest
jpayne@69 330 * digit substring that can be treated as a single unit is
jpayne@69 331 * 254 digits (not counting leading zeros). If a digit substring is
jpayne@69 332 * longer than that, the digits beyond the limit will be treated as a
jpayne@69 333 * separate digit substring.
jpayne@69 334 *
jpayne@69 335 * A "digit" in this sense is a code point with General_Category=Nd,
jpayne@69 336 * which does not include circled numbers, roman numerals, etc.
jpayne@69 337 * Only a contiguous digit substring is considered, that is,
jpayne@69 338 * non-negative integers without separators.
jpayne@69 339 * There is no support for plus/minus signs, decimals, exponents, etc.
jpayne@69 340 *
jpayne@69 341 * @stable ICU 2.8
jpayne@69 342 */
jpayne@69 343 UCOL_NUMERIC_COLLATION = UCOL_STRENGTH + 2,
jpayne@69 344
jpayne@69 345 /* Do not conditionalize the following with #ifndef U_HIDE_DEPRECATED_API,
jpayne@69 346 * it is needed for layout of RuleBasedCollator object. */
jpayne@69 347 #ifndef U_FORCE_HIDE_DEPRECATED_API
jpayne@69 348 /**
jpayne@69 349 * One more than the highest normal UColAttribute value.
jpayne@69 350 * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
jpayne@69 351 */
jpayne@69 352 UCOL_ATTRIBUTE_COUNT
jpayne@69 353 #endif // U_FORCE_HIDE_DEPRECATED_API
jpayne@69 354 } UColAttribute;
jpayne@69 355
jpayne@69 356 /** Options for retrieving the rule string
jpayne@69 357 * @stable ICU 2.0
jpayne@69 358 */
jpayne@69 359 typedef enum {
jpayne@69 360 /**
jpayne@69 361 * Retrieves the tailoring rules only.
jpayne@69 362 * Same as calling the version of getRules() without UColRuleOption.
jpayne@69 363 * @stable ICU 2.0
jpayne@69 364 */
jpayne@69 365 UCOL_TAILORING_ONLY,
jpayne@69 366 /**
jpayne@69 367 * Retrieves the "UCA rules" concatenated with the tailoring rules.
jpayne@69 368 * The "UCA rules" are an <i>approximation</i> of the root collator's sort order.
jpayne@69 369 * They are almost never used or useful at runtime and can be removed from the data.
jpayne@69 370 * See http://userguide.icu-project.org/collation/customization#TOC-Building-on-Existing-Locales
jpayne@69 371 * @stable ICU 2.0
jpayne@69 372 */
jpayne@69 373 UCOL_FULL_RULES
jpayne@69 374 } UColRuleOption ;
jpayne@69 375
jpayne@69 376 /**
jpayne@69 377 * Open a UCollator for comparing strings.
jpayne@69 378 *
jpayne@69 379 * For some languages, multiple collation types are available;
jpayne@69 380 * for example, "de@collation=phonebook".
jpayne@69 381 * Starting with ICU 54, collation attributes can be specified via locale keywords as well,
jpayne@69 382 * in the old locale extension syntax ("el@colCaseFirst=upper")
jpayne@69 383 * or in language tag syntax ("el-u-kf-upper").
jpayne@69 384 * See <a href="http://userguide.icu-project.org/collation/api">User Guide: Collation API</a>.
jpayne@69 385 *
jpayne@69 386 * The UCollator pointer is used in all the calls to the Collation
jpayne@69 387 * service. After finished, collator must be disposed of by calling
jpayne@69 388 * {@link #ucol_close }.
jpayne@69 389 * @param loc The locale containing the required collation rules.
jpayne@69 390 * Special values for locales can be passed in -
jpayne@69 391 * if NULL is passed for the locale, the default locale
jpayne@69 392 * collation rules will be used. If empty string ("") or
jpayne@69 393 * "root" are passed, the root collator will be returned.
jpayne@69 394 * @param status A pointer to a UErrorCode to receive any errors
jpayne@69 395 * @return A pointer to a UCollator, or 0 if an error occurred.
jpayne@69 396 * @see ucol_openRules
jpayne@69 397 * @see ucol_safeClone
jpayne@69 398 * @see ucol_close
jpayne@69 399 * @stable ICU 2.0
jpayne@69 400 */
jpayne@69 401 U_STABLE UCollator* U_EXPORT2
jpayne@69 402 ucol_open(const char *loc, UErrorCode *status);
jpayne@69 403
jpayne@69 404 /**
jpayne@69 405 * Produce a UCollator instance according to the rules supplied.
jpayne@69 406 * The rules are used to change the default ordering, defined in the
jpayne@69 407 * UCA in a process called tailoring. The resulting UCollator pointer
jpayne@69 408 * can be used in the same way as the one obtained by {@link #ucol_strcoll }.
jpayne@69 409 * @param rules A string describing the collation rules. For the syntax
jpayne@69 410 * of the rules please see users guide.
jpayne@69 411 * @param rulesLength The length of rules, or -1 if null-terminated.
jpayne@69 412 * @param normalizationMode The normalization mode: One of
jpayne@69 413 * UCOL_OFF (expect the text to not need normalization),
jpayne@69 414 * UCOL_ON (normalize), or
jpayne@69 415 * UCOL_DEFAULT (set the mode according to the rules)
jpayne@69 416 * @param strength The default collation strength; one of UCOL_PRIMARY, UCOL_SECONDARY,
jpayne@69 417 * UCOL_TERTIARY, UCOL_IDENTICAL,UCOL_DEFAULT_STRENGTH - can be also set in the rules.
jpayne@69 418 * @param parseError A pointer to UParseError to recieve information about errors
jpayne@69 419 * occurred during parsing. This argument can currently be set
jpayne@69 420 * to NULL, but at users own risk. Please provide a real structure.
jpayne@69 421 * @param status A pointer to a UErrorCode to receive any errors
jpayne@69 422 * @return A pointer to a UCollator. It is not guaranteed that NULL be returned in case
jpayne@69 423 * of error - please use status argument to check for errors.
jpayne@69 424 * @see ucol_open
jpayne@69 425 * @see ucol_safeClone
jpayne@69 426 * @see ucol_close
jpayne@69 427 * @stable ICU 2.0
jpayne@69 428 */
jpayne@69 429 U_STABLE UCollator* U_EXPORT2
jpayne@69 430 ucol_openRules( const UChar *rules,
jpayne@69 431 int32_t rulesLength,
jpayne@69 432 UColAttributeValue normalizationMode,
jpayne@69 433 UCollationStrength strength,
jpayne@69 434 UParseError *parseError,
jpayne@69 435 UErrorCode *status);
jpayne@69 436
jpayne@69 437 #ifndef U_HIDE_DEPRECATED_API
jpayne@69 438 /**
jpayne@69 439 * Open a collator defined by a short form string.
jpayne@69 440 * The structure and the syntax of the string is defined in the "Naming collators"
jpayne@69 441 * section of the users guide:
jpayne@69 442 * http://userguide.icu-project.org/collation/concepts#TOC-Collator-naming-scheme
jpayne@69 443 * Attributes are overriden by the subsequent attributes. So, for "S2_S3", final
jpayne@69 444 * strength will be 3. 3066bis locale overrides individual locale parts.
jpayne@69 445 * The call to this function is equivalent to a call to ucol_open, followed by a
jpayne@69 446 * series of calls to ucol_setAttribute and ucol_setVariableTop.
jpayne@69 447 * @param definition A short string containing a locale and a set of attributes.
jpayne@69 448 * Attributes not explicitly mentioned are left at the default
jpayne@69 449 * state for a locale.
jpayne@69 450 * @param parseError if not NULL, structure that will get filled with error's pre
jpayne@69 451 * and post context in case of error.
jpayne@69 452 * @param forceDefaults if FALSE, the settings that are the same as the collator
jpayne@69 453 * default settings will not be applied (for example, setting
jpayne@69 454 * French secondary on a French collator would not be executed).
jpayne@69 455 * If TRUE, all the settings will be applied regardless of the
jpayne@69 456 * collator default value. If the definition
jpayne@69 457 * strings are to be cached, should be set to FALSE.
jpayne@69 458 * @param status Error code. Apart from regular error conditions connected to
jpayne@69 459 * instantiating collators (like out of memory or similar), this
jpayne@69 460 * API will return an error if an invalid attribute or attribute/value
jpayne@69 461 * combination is specified.
jpayne@69 462 * @return A pointer to a UCollator or 0 if an error occured (including an
jpayne@69 463 * invalid attribute).
jpayne@69 464 * @see ucol_open
jpayne@69 465 * @see ucol_setAttribute
jpayne@69 466 * @see ucol_setVariableTop
jpayne@69 467 * @see ucol_getShortDefinitionString
jpayne@69 468 * @see ucol_normalizeShortDefinitionString
jpayne@69 469 * @deprecated ICU 54 Use ucol_open() with language tag collation keywords instead.
jpayne@69 470 */
jpayne@69 471 U_DEPRECATED UCollator* U_EXPORT2
jpayne@69 472 ucol_openFromShortString( const char *definition,
jpayne@69 473 UBool forceDefaults,
jpayne@69 474 UParseError *parseError,
jpayne@69 475 UErrorCode *status);
jpayne@69 476 #endif /* U_HIDE_DEPRECATED_API */
jpayne@69 477
jpayne@69 478 #ifndef U_HIDE_DEPRECATED_API
jpayne@69 479 /**
jpayne@69 480 * Get a set containing the contractions defined by the collator. The set includes
jpayne@69 481 * both the root collator's contractions and the contractions defined by the collator. This set
jpayne@69 482 * will contain only strings. If a tailoring explicitly suppresses contractions from
jpayne@69 483 * the root collator (like Russian), removed contractions will not be in the resulting set.
jpayne@69 484 * @param coll collator
jpayne@69 485 * @param conts the set to hold the result. It gets emptied before
jpayne@69 486 * contractions are added.
jpayne@69 487 * @param status to hold the error code
jpayne@69 488 * @return the size of the contraction set
jpayne@69 489 *
jpayne@69 490 * @deprecated ICU 3.4, use ucol_getContractionsAndExpansions instead
jpayne@69 491 */
jpayne@69 492 U_DEPRECATED int32_t U_EXPORT2
jpayne@69 493 ucol_getContractions( const UCollator *coll,
jpayne@69 494 USet *conts,
jpayne@69 495 UErrorCode *status);
jpayne@69 496 #endif /* U_HIDE_DEPRECATED_API */
jpayne@69 497
jpayne@69 498 /**
jpayne@69 499 * Get a set containing the expansions defined by the collator. The set includes
jpayne@69 500 * both the root collator's expansions and the expansions defined by the tailoring
jpayne@69 501 * @param coll collator
jpayne@69 502 * @param contractions if not NULL, the set to hold the contractions
jpayne@69 503 * @param expansions if not NULL, the set to hold the expansions
jpayne@69 504 * @param addPrefixes add the prefix contextual elements to contractions
jpayne@69 505 * @param status to hold the error code
jpayne@69 506 *
jpayne@69 507 * @stable ICU 3.4
jpayne@69 508 */
jpayne@69 509 U_STABLE void U_EXPORT2
jpayne@69 510 ucol_getContractionsAndExpansions( const UCollator *coll,
jpayne@69 511 USet *contractions, USet *expansions,
jpayne@69 512 UBool addPrefixes, UErrorCode *status);
jpayne@69 513
jpayne@69 514 /**
jpayne@69 515 * Close a UCollator.
jpayne@69 516 * Once closed, a UCollator should not be used. Every open collator should
jpayne@69 517 * be closed. Otherwise, a memory leak will result.
jpayne@69 518 * @param coll The UCollator to close.
jpayne@69 519 * @see ucol_open
jpayne@69 520 * @see ucol_openRules
jpayne@69 521 * @see ucol_safeClone
jpayne@69 522 * @stable ICU 2.0
jpayne@69 523 */
jpayne@69 524 U_STABLE void U_EXPORT2
jpayne@69 525 ucol_close(UCollator *coll);
jpayne@69 526
jpayne@69 527 #if U_SHOW_CPLUSPLUS_API
jpayne@69 528
jpayne@69 529 U_NAMESPACE_BEGIN
jpayne@69 530
jpayne@69 531 /**
jpayne@69 532 * \class LocalUCollatorPointer
jpayne@69 533 * "Smart pointer" class, closes a UCollator via ucol_close().
jpayne@69 534 * For most methods see the LocalPointerBase base class.
jpayne@69 535 *
jpayne@69 536 * @see LocalPointerBase
jpayne@69 537 * @see LocalPointer
jpayne@69 538 * @stable ICU 4.4
jpayne@69 539 */
jpayne@69 540 U_DEFINE_LOCAL_OPEN_POINTER(LocalUCollatorPointer, UCollator, ucol_close);
jpayne@69 541
jpayne@69 542 U_NAMESPACE_END
jpayne@69 543
jpayne@69 544 #endif
jpayne@69 545
jpayne@69 546 /**
jpayne@69 547 * Compare two strings.
jpayne@69 548 * The strings will be compared using the options already specified.
jpayne@69 549 * @param coll The UCollator containing the comparison rules.
jpayne@69 550 * @param source The source string.
jpayne@69 551 * @param sourceLength The length of source, or -1 if null-terminated.
jpayne@69 552 * @param target The target string.
jpayne@69 553 * @param targetLength The length of target, or -1 if null-terminated.
jpayne@69 554 * @return The result of comparing the strings; one of UCOL_EQUAL,
jpayne@69 555 * UCOL_GREATER, UCOL_LESS
jpayne@69 556 * @see ucol_greater
jpayne@69 557 * @see ucol_greaterOrEqual
jpayne@69 558 * @see ucol_equal
jpayne@69 559 * @stable ICU 2.0
jpayne@69 560 */
jpayne@69 561 U_STABLE UCollationResult U_EXPORT2
jpayne@69 562 ucol_strcoll( const UCollator *coll,
jpayne@69 563 const UChar *source,
jpayne@69 564 int32_t sourceLength,
jpayne@69 565 const UChar *target,
jpayne@69 566 int32_t targetLength);
jpayne@69 567
jpayne@69 568 /**
jpayne@69 569 * Compare two strings in UTF-8.
jpayne@69 570 * The strings will be compared using the options already specified.
jpayne@69 571 * Note: When input string contains malformed a UTF-8 byte sequence,
jpayne@69 572 * this function treats these bytes as REPLACEMENT CHARACTER (U+FFFD).
jpayne@69 573 * @param coll The UCollator containing the comparison rules.
jpayne@69 574 * @param source The source UTF-8 string.
jpayne@69 575 * @param sourceLength The length of source, or -1 if null-terminated.
jpayne@69 576 * @param target The target UTF-8 string.
jpayne@69 577 * @param targetLength The length of target, or -1 if null-terminated.
jpayne@69 578 * @param status A pointer to a UErrorCode to receive any errors
jpayne@69 579 * @return The result of comparing the strings; one of UCOL_EQUAL,
jpayne@69 580 * UCOL_GREATER, UCOL_LESS
jpayne@69 581 * @see ucol_greater
jpayne@69 582 * @see ucol_greaterOrEqual
jpayne@69 583 * @see ucol_equal
jpayne@69 584 * @stable ICU 50
jpayne@69 585 */
jpayne@69 586 U_STABLE UCollationResult U_EXPORT2
jpayne@69 587 ucol_strcollUTF8(
jpayne@69 588 const UCollator *coll,
jpayne@69 589 const char *source,
jpayne@69 590 int32_t sourceLength,
jpayne@69 591 const char *target,
jpayne@69 592 int32_t targetLength,
jpayne@69 593 UErrorCode *status);
jpayne@69 594
jpayne@69 595 /**
jpayne@69 596 * Determine if one string is greater than another.
jpayne@69 597 * This function is equivalent to {@link #ucol_strcoll } == UCOL_GREATER
jpayne@69 598 * @param coll The UCollator containing the comparison rules.
jpayne@69 599 * @param source The source string.
jpayne@69 600 * @param sourceLength The length of source, or -1 if null-terminated.
jpayne@69 601 * @param target The target string.
jpayne@69 602 * @param targetLength The length of target, or -1 if null-terminated.
jpayne@69 603 * @return TRUE if source is greater than target, FALSE otherwise.
jpayne@69 604 * @see ucol_strcoll
jpayne@69 605 * @see ucol_greaterOrEqual
jpayne@69 606 * @see ucol_equal
jpayne@69 607 * @stable ICU 2.0
jpayne@69 608 */
jpayne@69 609 U_STABLE UBool U_EXPORT2
jpayne@69 610 ucol_greater(const UCollator *coll,
jpayne@69 611 const UChar *source, int32_t sourceLength,
jpayne@69 612 const UChar *target, int32_t targetLength);
jpayne@69 613
jpayne@69 614 /**
jpayne@69 615 * Determine if one string is greater than or equal to another.
jpayne@69 616 * This function is equivalent to {@link #ucol_strcoll } != UCOL_LESS
jpayne@69 617 * @param coll The UCollator containing the comparison rules.
jpayne@69 618 * @param source The source string.
jpayne@69 619 * @param sourceLength The length of source, or -1 if null-terminated.
jpayne@69 620 * @param target The target string.
jpayne@69 621 * @param targetLength The length of target, or -1 if null-terminated.
jpayne@69 622 * @return TRUE if source is greater than or equal to target, FALSE otherwise.
jpayne@69 623 * @see ucol_strcoll
jpayne@69 624 * @see ucol_greater
jpayne@69 625 * @see ucol_equal
jpayne@69 626 * @stable ICU 2.0
jpayne@69 627 */
jpayne@69 628 U_STABLE UBool U_EXPORT2
jpayne@69 629 ucol_greaterOrEqual(const UCollator *coll,
jpayne@69 630 const UChar *source, int32_t sourceLength,
jpayne@69 631 const UChar *target, int32_t targetLength);
jpayne@69 632
jpayne@69 633 /**
jpayne@69 634 * Compare two strings for equality.
jpayne@69 635 * This function is equivalent to {@link #ucol_strcoll } == UCOL_EQUAL
jpayne@69 636 * @param coll The UCollator containing the comparison rules.
jpayne@69 637 * @param source The source string.
jpayne@69 638 * @param sourceLength The length of source, or -1 if null-terminated.
jpayne@69 639 * @param target The target string.
jpayne@69 640 * @param targetLength The length of target, or -1 if null-terminated.
jpayne@69 641 * @return TRUE if source is equal to target, FALSE otherwise
jpayne@69 642 * @see ucol_strcoll
jpayne@69 643 * @see ucol_greater
jpayne@69 644 * @see ucol_greaterOrEqual
jpayne@69 645 * @stable ICU 2.0
jpayne@69 646 */
jpayne@69 647 U_STABLE UBool U_EXPORT2
jpayne@69 648 ucol_equal(const UCollator *coll,
jpayne@69 649 const UChar *source, int32_t sourceLength,
jpayne@69 650 const UChar *target, int32_t targetLength);
jpayne@69 651
jpayne@69 652 /**
jpayne@69 653 * Compare two UTF-8 encoded trings.
jpayne@69 654 * The strings will be compared using the options already specified.
jpayne@69 655 * @param coll The UCollator containing the comparison rules.
jpayne@69 656 * @param sIter The source string iterator.
jpayne@69 657 * @param tIter The target string iterator.
jpayne@69 658 * @return The result of comparing the strings; one of UCOL_EQUAL,
jpayne@69 659 * UCOL_GREATER, UCOL_LESS
jpayne@69 660 * @param status A pointer to a UErrorCode to receive any errors
jpayne@69 661 * @see ucol_strcoll
jpayne@69 662 * @stable ICU 2.6
jpayne@69 663 */
jpayne@69 664 U_STABLE UCollationResult U_EXPORT2
jpayne@69 665 ucol_strcollIter( const UCollator *coll,
jpayne@69 666 UCharIterator *sIter,
jpayne@69 667 UCharIterator *tIter,
jpayne@69 668 UErrorCode *status);
jpayne@69 669
jpayne@69 670 /**
jpayne@69 671 * Get the collation strength used in a UCollator.
jpayne@69 672 * The strength influences how strings are compared.
jpayne@69 673 * @param coll The UCollator to query.
jpayne@69 674 * @return The collation strength; one of UCOL_PRIMARY, UCOL_SECONDARY,
jpayne@69 675 * UCOL_TERTIARY, UCOL_QUATERNARY, UCOL_IDENTICAL
jpayne@69 676 * @see ucol_setStrength
jpayne@69 677 * @stable ICU 2.0
jpayne@69 678 */
jpayne@69 679 U_STABLE UCollationStrength U_EXPORT2
jpayne@69 680 ucol_getStrength(const UCollator *coll);
jpayne@69 681
jpayne@69 682 /**
jpayne@69 683 * Set the collation strength used in a UCollator.
jpayne@69 684 * The strength influences how strings are compared.
jpayne@69 685 * @param coll The UCollator to set.
jpayne@69 686 * @param strength The desired collation strength; one of UCOL_PRIMARY,
jpayne@69 687 * UCOL_SECONDARY, UCOL_TERTIARY, UCOL_QUATERNARY, UCOL_IDENTICAL, UCOL_DEFAULT
jpayne@69 688 * @see ucol_getStrength
jpayne@69 689 * @stable ICU 2.0
jpayne@69 690 */
jpayne@69 691 U_STABLE void U_EXPORT2
jpayne@69 692 ucol_setStrength(UCollator *coll,
jpayne@69 693 UCollationStrength strength);
jpayne@69 694
jpayne@69 695 /**
jpayne@69 696 * Retrieves the reordering codes for this collator.
jpayne@69 697 * These reordering codes are a combination of UScript codes and UColReorderCode entries.
jpayne@69 698 * @param coll The UCollator to query.
jpayne@69 699 * @param dest The array to fill with the script ordering.
jpayne@69 700 * @param destCapacity The length of dest. If it is 0, then dest may be NULL and the function
jpayne@69 701 * will only return the length of the result without writing any codes (pre-flighting).
jpayne@69 702 * @param pErrorCode Must be a valid pointer to an error code value, which must not indicate a
jpayne@69 703 * failure before the function call.
jpayne@69 704 * @return The number of reordering codes written to the dest array.
jpayne@69 705 * @see ucol_setReorderCodes
jpayne@69 706 * @see ucol_getEquivalentReorderCodes
jpayne@69 707 * @see UScriptCode
jpayne@69 708 * @see UColReorderCode
jpayne@69 709 * @stable ICU 4.8
jpayne@69 710 */
jpayne@69 711 U_STABLE int32_t U_EXPORT2
jpayne@69 712 ucol_getReorderCodes(const UCollator* coll,
jpayne@69 713 int32_t* dest,
jpayne@69 714 int32_t destCapacity,
jpayne@69 715 UErrorCode *pErrorCode);
jpayne@69 716 /**
jpayne@69 717 * Sets the reordering codes for this collator.
jpayne@69 718 * Collation reordering allows scripts and some other groups of characters
jpayne@69 719 * to be moved relative to each other. This reordering is done on top of
jpayne@69 720 * the DUCET/CLDR standard collation order. Reordering can specify groups to be placed
jpayne@69 721 * at the start and/or the end of the collation order. These groups are specified using
jpayne@69 722 * UScript codes and UColReorderCode entries.
jpayne@69 723 *
jpayne@69 724 * <p>By default, reordering codes specified for the start of the order are placed in the
jpayne@69 725 * order given after several special non-script blocks. These special groups of characters
jpayne@69 726 * are space, punctuation, symbol, currency, and digit. These special groups are represented with
jpayne@69 727 * UColReorderCode entries. Script groups can be intermingled with
jpayne@69 728 * these special non-script groups if those special groups are explicitly specified in the reordering.
jpayne@69 729 *
jpayne@69 730 * <p>The special code OTHERS stands for any script that is not explicitly
jpayne@69 731 * mentioned in the list of reordering codes given. Anything that is after OTHERS
jpayne@69 732 * will go at the very end of the reordering in the order given.
jpayne@69 733 *
jpayne@69 734 * <p>The special reorder code DEFAULT will reset the reordering for this collator
jpayne@69 735 * to the default for this collator. The default reordering may be the DUCET/CLDR order or may be a reordering that
jpayne@69 736 * was specified when this collator was created from resource data or from rules. The
jpayne@69 737 * DEFAULT code <b>must</b> be the sole code supplied when it is used.
jpayne@69 738 * If not, then U_ILLEGAL_ARGUMENT_ERROR will be set.
jpayne@69 739 *
jpayne@69 740 * <p>The special reorder code NONE will remove any reordering for this collator.
jpayne@69 741 * The result of setting no reordering will be to have the DUCET/CLDR ordering used. The
jpayne@69 742 * NONE code <b>must</b> be the sole code supplied when it is used.
jpayne@69 743 *
jpayne@69 744 * @param coll The UCollator to set.
jpayne@69 745 * @param reorderCodes An array of script codes in the new order. This can be NULL if the
jpayne@69 746 * length is also set to 0. An empty array will clear any reordering codes on the collator.
jpayne@69 747 * @param reorderCodesLength The length of reorderCodes.
jpayne@69 748 * @param pErrorCode Must be a valid pointer to an error code value, which must not indicate a
jpayne@69 749 * failure before the function call.
jpayne@69 750 * @see ucol_getReorderCodes
jpayne@69 751 * @see ucol_getEquivalentReorderCodes
jpayne@69 752 * @see UScriptCode
jpayne@69 753 * @see UColReorderCode
jpayne@69 754 * @stable ICU 4.8
jpayne@69 755 */
jpayne@69 756 U_STABLE void U_EXPORT2
jpayne@69 757 ucol_setReorderCodes(UCollator* coll,
jpayne@69 758 const int32_t* reorderCodes,
jpayne@69 759 int32_t reorderCodesLength,
jpayne@69 760 UErrorCode *pErrorCode);
jpayne@69 761
jpayne@69 762 /**
jpayne@69 763 * Retrieves the reorder codes that are grouped with the given reorder code. Some reorder
jpayne@69 764 * codes will be grouped and must reorder together.
jpayne@69 765 * Beginning with ICU 55, scripts only reorder together if they are primary-equal,
jpayne@69 766 * for example Hiragana and Katakana.
jpayne@69 767 *
jpayne@69 768 * @param reorderCode The reorder code to determine equivalence for.
jpayne@69 769 * @param dest The array to fill with the script ordering.
jpayne@69 770 * @param destCapacity The length of dest. If it is 0, then dest may be NULL and the function
jpayne@69 771 * will only return the length of the result without writing any codes (pre-flighting).
jpayne@69 772 * @param pErrorCode Must be a valid pointer to an error code value, which must not indicate
jpayne@69 773 * a failure before the function call.
jpayne@69 774 * @return The number of reordering codes written to the dest array.
jpayne@69 775 * @see ucol_setReorderCodes
jpayne@69 776 * @see ucol_getReorderCodes
jpayne@69 777 * @see UScriptCode
jpayne@69 778 * @see UColReorderCode
jpayne@69 779 * @stable ICU 4.8
jpayne@69 780 */
jpayne@69 781 U_STABLE int32_t U_EXPORT2
jpayne@69 782 ucol_getEquivalentReorderCodes(int32_t reorderCode,
jpayne@69 783 int32_t* dest,
jpayne@69 784 int32_t destCapacity,
jpayne@69 785 UErrorCode *pErrorCode);
jpayne@69 786
jpayne@69 787 /**
jpayne@69 788 * Get the display name for a UCollator.
jpayne@69 789 * The display name is suitable for presentation to a user.
jpayne@69 790 * @param objLoc The locale of the collator in question.
jpayne@69 791 * @param dispLoc The locale for display.
jpayne@69 792 * @param result A pointer to a buffer to receive the attribute.
jpayne@69 793 * @param resultLength The maximum size of result.
jpayne@69 794 * @param status A pointer to a UErrorCode to receive any errors
jpayne@69 795 * @return The total buffer size needed; if greater than resultLength,
jpayne@69 796 * the output was truncated.
jpayne@69 797 * @stable ICU 2.0
jpayne@69 798 */
jpayne@69 799 U_STABLE int32_t U_EXPORT2
jpayne@69 800 ucol_getDisplayName( const char *objLoc,
jpayne@69 801 const char *dispLoc,
jpayne@69 802 UChar *result,
jpayne@69 803 int32_t resultLength,
jpayne@69 804 UErrorCode *status);
jpayne@69 805
jpayne@69 806 /**
jpayne@69 807 * Get a locale for which collation rules are available.
jpayne@69 808 * A UCollator in a locale returned by this function will perform the correct
jpayne@69 809 * collation for the locale.
jpayne@69 810 * @param localeIndex The index of the desired locale.
jpayne@69 811 * @return A locale for which collation rules are available, or 0 if none.
jpayne@69 812 * @see ucol_countAvailable
jpayne@69 813 * @stable ICU 2.0
jpayne@69 814 */
jpayne@69 815 U_STABLE const char* U_EXPORT2
jpayne@69 816 ucol_getAvailable(int32_t localeIndex);
jpayne@69 817
jpayne@69 818 /**
jpayne@69 819 * Determine how many locales have collation rules available.
jpayne@69 820 * This function is most useful as determining the loop ending condition for
jpayne@69 821 * calls to {@link #ucol_getAvailable }.
jpayne@69 822 * @return The number of locales for which collation rules are available.
jpayne@69 823 * @see ucol_getAvailable
jpayne@69 824 * @stable ICU 2.0
jpayne@69 825 */
jpayne@69 826 U_STABLE int32_t U_EXPORT2
jpayne@69 827 ucol_countAvailable(void);
jpayne@69 828
jpayne@69 829 #if !UCONFIG_NO_SERVICE
jpayne@69 830 /**
jpayne@69 831 * Create a string enumerator of all locales for which a valid
jpayne@69 832 * collator may be opened.
jpayne@69 833 * @param status input-output error code
jpayne@69 834 * @return a string enumeration over locale strings. The caller is
jpayne@69 835 * responsible for closing the result.
jpayne@69 836 * @stable ICU 3.0
jpayne@69 837 */
jpayne@69 838 U_STABLE UEnumeration* U_EXPORT2
jpayne@69 839 ucol_openAvailableLocales(UErrorCode *status);
jpayne@69 840 #endif
jpayne@69 841
jpayne@69 842 /**
jpayne@69 843 * Create a string enumerator of all possible keywords that are relevant to
jpayne@69 844 * collation. At this point, the only recognized keyword for this
jpayne@69 845 * service is "collation".
jpayne@69 846 * @param status input-output error code
jpayne@69 847 * @return a string enumeration over locale strings. The caller is
jpayne@69 848 * responsible for closing the result.
jpayne@69 849 * @stable ICU 3.0
jpayne@69 850 */
jpayne@69 851 U_STABLE UEnumeration* U_EXPORT2
jpayne@69 852 ucol_getKeywords(UErrorCode *status);
jpayne@69 853
jpayne@69 854 /**
jpayne@69 855 * Given a keyword, create a string enumeration of all values
jpayne@69 856 * for that keyword that are currently in use.
jpayne@69 857 * @param keyword a particular keyword as enumerated by
jpayne@69 858 * ucol_getKeywords. If any other keyword is passed in, *status is set
jpayne@69 859 * to U_ILLEGAL_ARGUMENT_ERROR.
jpayne@69 860 * @param status input-output error code
jpayne@69 861 * @return a string enumeration over collation keyword values, or NULL
jpayne@69 862 * upon error. The caller is responsible for closing the result.
jpayne@69 863 * @stable ICU 3.0
jpayne@69 864 */
jpayne@69 865 U_STABLE UEnumeration* U_EXPORT2
jpayne@69 866 ucol_getKeywordValues(const char *keyword, UErrorCode *status);
jpayne@69 867
jpayne@69 868 /**
jpayne@69 869 * Given a key and a locale, returns an array of string values in a preferred
jpayne@69 870 * order that would make a difference. These are all and only those values where
jpayne@69 871 * the open (creation) of the service with the locale formed from the input locale
jpayne@69 872 * plus input keyword and that value has different behavior than creation with the
jpayne@69 873 * input locale alone.
jpayne@69 874 * @param key one of the keys supported by this service. For now, only
jpayne@69 875 * "collation" is supported.
jpayne@69 876 * @param locale the locale
jpayne@69 877 * @param commonlyUsed if set to true it will return only commonly used values
jpayne@69 878 * with the given locale in preferred order. Otherwise,
jpayne@69 879 * it will return all the available values for the locale.
jpayne@69 880 * @param status error status
jpayne@69 881 * @return a string enumeration over keyword values for the given key and the locale.
jpayne@69 882 * @stable ICU 4.2
jpayne@69 883 */
jpayne@69 884 U_STABLE UEnumeration* U_EXPORT2
jpayne@69 885 ucol_getKeywordValuesForLocale(const char* key,
jpayne@69 886 const char* locale,
jpayne@69 887 UBool commonlyUsed,
jpayne@69 888 UErrorCode* status);
jpayne@69 889
jpayne@69 890 /**
jpayne@69 891 * Return the functionally equivalent locale for the specified
jpayne@69 892 * input locale, with respect to given keyword, for the
jpayne@69 893 * collation service. If two different input locale + keyword
jpayne@69 894 * combinations produce the same result locale, then collators
jpayne@69 895 * instantiated for these two different input locales will behave
jpayne@69 896 * equivalently. The converse is not always true; two collators
jpayne@69 897 * may in fact be equivalent, but return different results, due to
jpayne@69 898 * internal details. The return result has no other meaning than
jpayne@69 899 * that stated above, and implies nothing as to the relationship
jpayne@69 900 * between the two locales. This is intended for use by
jpayne@69 901 * applications who wish to cache collators, or otherwise reuse
jpayne@69 902 * collators when possible. The functional equivalent may change
jpayne@69 903 * over time. For more information, please see the <a
jpayne@69 904 * href="http://userguide.icu-project.org/locale#TOC-Locales-and-Services">
jpayne@69 905 * Locales and Services</a> section of the ICU User Guide.
jpayne@69 906 * @param result fillin for the functionally equivalent result locale
jpayne@69 907 * @param resultCapacity capacity of the fillin buffer
jpayne@69 908 * @param keyword a particular keyword as enumerated by
jpayne@69 909 * ucol_getKeywords.
jpayne@69 910 * @param locale the specified input locale
jpayne@69 911 * @param isAvailable if non-NULL, pointer to a fillin parameter that
jpayne@69 912 * on return indicates whether the specified input locale was 'available'
jpayne@69 913 * to the collation service. A locale is defined as 'available' if it
jpayne@69 914 * physically exists within the collation locale data.
jpayne@69 915 * @param status pointer to input-output error code
jpayne@69 916 * @return the actual buffer size needed for the locale. If greater
jpayne@69 917 * than resultCapacity, the returned full name will be truncated and
jpayne@69 918 * an error code will be returned.
jpayne@69 919 * @stable ICU 3.0
jpayne@69 920 */
jpayne@69 921 U_STABLE int32_t U_EXPORT2
jpayne@69 922 ucol_getFunctionalEquivalent(char* result, int32_t resultCapacity,
jpayne@69 923 const char* keyword, const char* locale,
jpayne@69 924 UBool* isAvailable, UErrorCode* status);
jpayne@69 925
jpayne@69 926 /**
jpayne@69 927 * Get the collation tailoring rules from a UCollator.
jpayne@69 928 * The rules will follow the rule syntax.
jpayne@69 929 * @param coll The UCollator to query.
jpayne@69 930 * @param length
jpayne@69 931 * @return The collation tailoring rules.
jpayne@69 932 * @stable ICU 2.0
jpayne@69 933 */
jpayne@69 934 U_STABLE const UChar* U_EXPORT2
jpayne@69 935 ucol_getRules( const UCollator *coll,
jpayne@69 936 int32_t *length);
jpayne@69 937
jpayne@69 938 #ifndef U_HIDE_DEPRECATED_API
jpayne@69 939 /** Get the short definition string for a collator. This API harvests the collator's
jpayne@69 940 * locale and the attribute set and produces a string that can be used for opening
jpayne@69 941 * a collator with the same attributes using the ucol_openFromShortString API.
jpayne@69 942 * This string will be normalized.
jpayne@69 943 * The structure and the syntax of the string is defined in the "Naming collators"
jpayne@69 944 * section of the users guide:
jpayne@69 945 * http://userguide.icu-project.org/collation/concepts#TOC-Collator-naming-scheme
jpayne@69 946 * This API supports preflighting.
jpayne@69 947 * @param coll a collator
jpayne@69 948 * @param locale a locale that will appear as a collators locale in the resulting
jpayne@69 949 * short string definition. If NULL, the locale will be harvested
jpayne@69 950 * from the collator.
jpayne@69 951 * @param buffer space to hold the resulting string
jpayne@69 952 * @param capacity capacity of the buffer
jpayne@69 953 * @param status for returning errors. All the preflighting errors are featured
jpayne@69 954 * @return length of the resulting string
jpayne@69 955 * @see ucol_openFromShortString
jpayne@69 956 * @see ucol_normalizeShortDefinitionString
jpayne@69 957 * @deprecated ICU 54
jpayne@69 958 */
jpayne@69 959 U_DEPRECATED int32_t U_EXPORT2
jpayne@69 960 ucol_getShortDefinitionString(const UCollator *coll,
jpayne@69 961 const char *locale,
jpayne@69 962 char *buffer,
jpayne@69 963 int32_t capacity,
jpayne@69 964 UErrorCode *status);
jpayne@69 965
jpayne@69 966 /** Verifies and normalizes short definition string.
jpayne@69 967 * Normalized short definition string has all the option sorted by the argument name,
jpayne@69 968 * so that equivalent definition strings are the same.
jpayne@69 969 * This API supports preflighting.
jpayne@69 970 * @param source definition string
jpayne@69 971 * @param destination space to hold the resulting string
jpayne@69 972 * @param capacity capacity of the buffer
jpayne@69 973 * @param parseError if not NULL, structure that will get filled with error's pre
jpayne@69 974 * and post context in case of error.
jpayne@69 975 * @param status Error code. This API will return an error if an invalid attribute
jpayne@69 976 * or attribute/value combination is specified. All the preflighting
jpayne@69 977 * errors are also featured
jpayne@69 978 * @return length of the resulting normalized string.
jpayne@69 979 *
jpayne@69 980 * @see ucol_openFromShortString
jpayne@69 981 * @see ucol_getShortDefinitionString
jpayne@69 982 *
jpayne@69 983 * @deprecated ICU 54
jpayne@69 984 */
jpayne@69 985
jpayne@69 986 U_DEPRECATED int32_t U_EXPORT2
jpayne@69 987 ucol_normalizeShortDefinitionString(const char *source,
jpayne@69 988 char *destination,
jpayne@69 989 int32_t capacity,
jpayne@69 990 UParseError *parseError,
jpayne@69 991 UErrorCode *status);
jpayne@69 992 #endif /* U_HIDE_DEPRECATED_API */
jpayne@69 993
jpayne@69 994
jpayne@69 995 /**
jpayne@69 996 * Get a sort key for a string from a UCollator.
jpayne@69 997 * Sort keys may be compared using <TT>strcmp</TT>.
jpayne@69 998 *
jpayne@69 999 * Note that sort keys are often less efficient than simply doing comparison.
jpayne@69 1000 * For more details, see the ICU User Guide.
jpayne@69 1001 *
jpayne@69 1002 * Like ICU functions that write to an output buffer, the buffer contents
jpayne@69 1003 * is undefined if the buffer capacity (resultLength parameter) is too small.
jpayne@69 1004 * Unlike ICU functions that write a string to an output buffer,
jpayne@69 1005 * the terminating zero byte is counted in the sort key length.
jpayne@69 1006 * @param coll The UCollator containing the collation rules.
jpayne@69 1007 * @param source The string to transform.
jpayne@69 1008 * @param sourceLength The length of source, or -1 if null-terminated.
jpayne@69 1009 * @param result A pointer to a buffer to receive the attribute.
jpayne@69 1010 * @param resultLength The maximum size of result.
jpayne@69 1011 * @return The size needed to fully store the sort key.
jpayne@69 1012 * If there was an internal error generating the sort key,
jpayne@69 1013 * a zero value is returned.
jpayne@69 1014 * @see ucol_keyHashCode
jpayne@69 1015 * @stable ICU 2.0
jpayne@69 1016 */
jpayne@69 1017 U_STABLE int32_t U_EXPORT2
jpayne@69 1018 ucol_getSortKey(const UCollator *coll,
jpayne@69 1019 const UChar *source,
jpayne@69 1020 int32_t sourceLength,
jpayne@69 1021 uint8_t *result,
jpayne@69 1022 int32_t resultLength);
jpayne@69 1023
jpayne@69 1024
jpayne@69 1025 /** Gets the next count bytes of a sort key. Caller needs
jpayne@69 1026 * to preserve state array between calls and to provide
jpayne@69 1027 * the same type of UCharIterator set with the same string.
jpayne@69 1028 * The destination buffer provided must be big enough to store
jpayne@69 1029 * the number of requested bytes.
jpayne@69 1030 *
jpayne@69 1031 * The generated sort key may or may not be compatible with
jpayne@69 1032 * sort keys generated using ucol_getSortKey().
jpayne@69 1033 * @param coll The UCollator containing the collation rules.
jpayne@69 1034 * @param iter UCharIterator containing the string we need
jpayne@69 1035 * the sort key to be calculated for.
jpayne@69 1036 * @param state Opaque state of sortkey iteration.
jpayne@69 1037 * @param dest Buffer to hold the resulting sortkey part
jpayne@69 1038 * @param count number of sort key bytes required.
jpayne@69 1039 * @param status error code indicator.
jpayne@69 1040 * @return the actual number of bytes of a sortkey. It can be
jpayne@69 1041 * smaller than count if we have reached the end of
jpayne@69 1042 * the sort key.
jpayne@69 1043 * @stable ICU 2.6
jpayne@69 1044 */
jpayne@69 1045 U_STABLE int32_t U_EXPORT2
jpayne@69 1046 ucol_nextSortKeyPart(const UCollator *coll,
jpayne@69 1047 UCharIterator *iter,
jpayne@69 1048 uint32_t state[2],
jpayne@69 1049 uint8_t *dest, int32_t count,
jpayne@69 1050 UErrorCode *status);
jpayne@69 1051
jpayne@69 1052 /** enum that is taken by ucol_getBound API
jpayne@69 1053 * See below for explanation
jpayne@69 1054 * do not change the values assigned to the
jpayne@69 1055 * members of this enum. Underlying code
jpayne@69 1056 * depends on them having these numbers
jpayne@69 1057 * @stable ICU 2.0
jpayne@69 1058 */
jpayne@69 1059 typedef enum {
jpayne@69 1060 /** lower bound */
jpayne@69 1061 UCOL_BOUND_LOWER = 0,
jpayne@69 1062 /** upper bound that will match strings of exact size */
jpayne@69 1063 UCOL_BOUND_UPPER = 1,
jpayne@69 1064 /** upper bound that will match all the strings that have the same initial substring as the given string */
jpayne@69 1065 UCOL_BOUND_UPPER_LONG = 2,
jpayne@69 1066 #ifndef U_HIDE_DEPRECATED_API
jpayne@69 1067 /**
jpayne@69 1068 * One more than the highest normal UColBoundMode value.
jpayne@69 1069 * @deprecated ICU 58 The numeric value may change over time, see ICU ticket #12420.
jpayne@69 1070 */
jpayne@69 1071 UCOL_BOUND_VALUE_COUNT
jpayne@69 1072 #endif /* U_HIDE_DEPRECATED_API */
jpayne@69 1073 } UColBoundMode;
jpayne@69 1074
jpayne@69 1075 /**
jpayne@69 1076 * Produce a bound for a given sortkey and a number of levels.
jpayne@69 1077 * Return value is always the number of bytes needed, regardless of
jpayne@69 1078 * whether the result buffer was big enough or even valid.<br>
jpayne@69 1079 * Resulting bounds can be used to produce a range of strings that are
jpayne@69 1080 * between upper and lower bounds. For example, if bounds are produced
jpayne@69 1081 * for a sortkey of string "smith", strings between upper and lower
jpayne@69 1082 * bounds with one level would include "Smith", "SMITH", "sMiTh".<br>
jpayne@69 1083 * There are two upper bounds that can be produced. If UCOL_BOUND_UPPER
jpayne@69 1084 * is produced, strings matched would be as above. However, if bound
jpayne@69 1085 * produced using UCOL_BOUND_UPPER_LONG is used, the above example will
jpayne@69 1086 * also match "Smithsonian" and similar.<br>
jpayne@69 1087 * For more on usage, see example in cintltst/capitst.c in procedure
jpayne@69 1088 * TestBounds.
jpayne@69 1089 * Sort keys may be compared using <TT>strcmp</TT>.
jpayne@69 1090 * @param source The source sortkey.
jpayne@69 1091 * @param sourceLength The length of source, or -1 if null-terminated.
jpayne@69 1092 * (If an unmodified sortkey is passed, it is always null
jpayne@69 1093 * terminated).
jpayne@69 1094 * @param boundType Type of bound required. It can be UCOL_BOUND_LOWER, which
jpayne@69 1095 * produces a lower inclusive bound, UCOL_BOUND_UPPER, that
jpayne@69 1096 * produces upper bound that matches strings of the same length
jpayne@69 1097 * or UCOL_BOUND_UPPER_LONG that matches strings that have the
jpayne@69 1098 * same starting substring as the source string.
jpayne@69 1099 * @param noOfLevels Number of levels required in the resulting bound (for most
jpayne@69 1100 * uses, the recommended value is 1). See users guide for
jpayne@69 1101 * explanation on number of levels a sortkey can have.
jpayne@69 1102 * @param result A pointer to a buffer to receive the resulting sortkey.
jpayne@69 1103 * @param resultLength The maximum size of result.
jpayne@69 1104 * @param status Used for returning error code if something went wrong. If the
jpayne@69 1105 * number of levels requested is higher than the number of levels
jpayne@69 1106 * in the source key, a warning (U_SORT_KEY_TOO_SHORT_WARNING) is
jpayne@69 1107 * issued.
jpayne@69 1108 * @return The size needed to fully store the bound.
jpayne@69 1109 * @see ucol_keyHashCode
jpayne@69 1110 * @stable ICU 2.1
jpayne@69 1111 */
jpayne@69 1112 U_STABLE int32_t U_EXPORT2
jpayne@69 1113 ucol_getBound(const uint8_t *source,
jpayne@69 1114 int32_t sourceLength,
jpayne@69 1115 UColBoundMode boundType,
jpayne@69 1116 uint32_t noOfLevels,
jpayne@69 1117 uint8_t *result,
jpayne@69 1118 int32_t resultLength,
jpayne@69 1119 UErrorCode *status);
jpayne@69 1120
jpayne@69 1121 /**
jpayne@69 1122 * Gets the version information for a Collator. Version is currently
jpayne@69 1123 * an opaque 32-bit number which depends, among other things, on major
jpayne@69 1124 * versions of the collator tailoring and UCA.
jpayne@69 1125 * @param coll The UCollator to query.
jpayne@69 1126 * @param info the version # information, the result will be filled in
jpayne@69 1127 * @stable ICU 2.0
jpayne@69 1128 */
jpayne@69 1129 U_STABLE void U_EXPORT2
jpayne@69 1130 ucol_getVersion(const UCollator* coll, UVersionInfo info);
jpayne@69 1131
jpayne@69 1132 /**
jpayne@69 1133 * Gets the UCA version information for a Collator. Version is the
jpayne@69 1134 * UCA version number (3.1.1, 4.0).
jpayne@69 1135 * @param coll The UCollator to query.
jpayne@69 1136 * @param info the version # information, the result will be filled in
jpayne@69 1137 * @stable ICU 2.8
jpayne@69 1138 */
jpayne@69 1139 U_STABLE void U_EXPORT2
jpayne@69 1140 ucol_getUCAVersion(const UCollator* coll, UVersionInfo info);
jpayne@69 1141
jpayne@69 1142 /**
jpayne@69 1143 * Merges two sort keys. The levels are merged with their corresponding counterparts
jpayne@69 1144 * (primaries with primaries, secondaries with secondaries etc.). Between the values
jpayne@69 1145 * from the same level a separator is inserted.
jpayne@69 1146 *
jpayne@69 1147 * This is useful, for example, for combining sort keys from first and last names
jpayne@69 1148 * to sort such pairs.
jpayne@69 1149 * See http://www.unicode.org/reports/tr10/#Merging_Sort_Keys
jpayne@69 1150 *
jpayne@69 1151 * The recommended way to achieve "merged" sorting is by
jpayne@69 1152 * concatenating strings with U+FFFE between them.
jpayne@69 1153 * The concatenation has the same sort order as the merged sort keys,
jpayne@69 1154 * but merge(getSortKey(str1), getSortKey(str2)) may differ from getSortKey(str1 + '\\uFFFE' + str2).
jpayne@69 1155 * Using strings with U+FFFE may yield shorter sort keys.
jpayne@69 1156 *
jpayne@69 1157 * For details about Sort Key Features see
jpayne@69 1158 * http://userguide.icu-project.org/collation/api#TOC-Sort-Key-Features
jpayne@69 1159 *
jpayne@69 1160 * It is possible to merge multiple sort keys by consecutively merging
jpayne@69 1161 * another one with the intermediate result.
jpayne@69 1162 *
jpayne@69 1163 * The length of the merge result is the sum of the lengths of the input sort keys.
jpayne@69 1164 *
jpayne@69 1165 * Example (uncompressed):
jpayne@69 1166 * <pre>191B1D 01 050505 01 910505 00
jpayne@69 1167 * 1F2123 01 050505 01 910505 00</pre>
jpayne@69 1168 * will be merged as
jpayne@69 1169 * <pre>191B1D 02 1F2123 01 050505 02 050505 01 910505 02 910505 00</pre>
jpayne@69 1170 *
jpayne@69 1171 * If the destination buffer is not big enough, then its contents are undefined.
jpayne@69 1172 * If any of source lengths are zero or any of the source pointers are NULL/undefined,
jpayne@69 1173 * the result is of size zero.
jpayne@69 1174 *
jpayne@69 1175 * @param src1 the first sort key
jpayne@69 1176 * @param src1Length the length of the first sort key, including the zero byte at the end;
jpayne@69 1177 * can be -1 if the function is to find the length
jpayne@69 1178 * @param src2 the second sort key
jpayne@69 1179 * @param src2Length the length of the second sort key, including the zero byte at the end;
jpayne@69 1180 * can be -1 if the function is to find the length
jpayne@69 1181 * @param dest the buffer where the merged sort key is written,
jpayne@69 1182 * can be NULL if destCapacity==0
jpayne@69 1183 * @param destCapacity the number of bytes in the dest buffer
jpayne@69 1184 * @return the length of the merged sort key, src1Length+src2Length;
jpayne@69 1185 * can be larger than destCapacity, or 0 if an error occurs (only for illegal arguments),
jpayne@69 1186 * in which cases the contents of dest is undefined
jpayne@69 1187 * @stable ICU 2.0
jpayne@69 1188 */
jpayne@69 1189 U_STABLE int32_t U_EXPORT2
jpayne@69 1190 ucol_mergeSortkeys(const uint8_t *src1, int32_t src1Length,
jpayne@69 1191 const uint8_t *src2, int32_t src2Length,
jpayne@69 1192 uint8_t *dest, int32_t destCapacity);
jpayne@69 1193
jpayne@69 1194 /**
jpayne@69 1195 * Universal attribute setter
jpayne@69 1196 * @param coll collator which attributes are to be changed
jpayne@69 1197 * @param attr attribute type
jpayne@69 1198 * @param value attribute value
jpayne@69 1199 * @param status to indicate whether the operation went on smoothly or there were errors
jpayne@69 1200 * @see UColAttribute
jpayne@69 1201 * @see UColAttributeValue
jpayne@69 1202 * @see ucol_getAttribute
jpayne@69 1203 * @stable ICU 2.0
jpayne@69 1204 */
jpayne@69 1205 U_STABLE void U_EXPORT2
jpayne@69 1206 ucol_setAttribute(UCollator *coll, UColAttribute attr, UColAttributeValue value, UErrorCode *status);
jpayne@69 1207
jpayne@69 1208 /**
jpayne@69 1209 * Universal attribute getter
jpayne@69 1210 * @param coll collator which attributes are to be changed
jpayne@69 1211 * @param attr attribute type
jpayne@69 1212 * @return attribute value
jpayne@69 1213 * @param status to indicate whether the operation went on smoothly or there were errors
jpayne@69 1214 * @see UColAttribute
jpayne@69 1215 * @see UColAttributeValue
jpayne@69 1216 * @see ucol_setAttribute
jpayne@69 1217 * @stable ICU 2.0
jpayne@69 1218 */
jpayne@69 1219 U_STABLE UColAttributeValue U_EXPORT2
jpayne@69 1220 ucol_getAttribute(const UCollator *coll, UColAttribute attr, UErrorCode *status);
jpayne@69 1221
jpayne@69 1222 /**
jpayne@69 1223 * Sets the variable top to the top of the specified reordering group.
jpayne@69 1224 * The variable top determines the highest-sorting character
jpayne@69 1225 * which is affected by UCOL_ALTERNATE_HANDLING.
jpayne@69 1226 * If that attribute is set to UCOL_NON_IGNORABLE, then the variable top has no effect.
jpayne@69 1227 * @param coll the collator
jpayne@69 1228 * @param group one of UCOL_REORDER_CODE_SPACE, UCOL_REORDER_CODE_PUNCTUATION,
jpayne@69 1229 * UCOL_REORDER_CODE_SYMBOL, UCOL_REORDER_CODE_CURRENCY;
jpayne@69 1230 * or UCOL_REORDER_CODE_DEFAULT to restore the default max variable group
jpayne@69 1231 * @param pErrorCode Standard ICU error code. Its input value must
jpayne@69 1232 * pass the U_SUCCESS() test, or else the function returns
jpayne@69 1233 * immediately. Check for U_FAILURE() on output or use with
jpayne@69 1234 * function chaining. (See User Guide for details.)
jpayne@69 1235 * @see ucol_getMaxVariable
jpayne@69 1236 * @stable ICU 53
jpayne@69 1237 */
jpayne@69 1238 U_STABLE void U_EXPORT2
jpayne@69 1239 ucol_setMaxVariable(UCollator *coll, UColReorderCode group, UErrorCode *pErrorCode);
jpayne@69 1240
jpayne@69 1241 /**
jpayne@69 1242 * Returns the maximum reordering group whose characters are affected by UCOL_ALTERNATE_HANDLING.
jpayne@69 1243 * @param coll the collator
jpayne@69 1244 * @return the maximum variable reordering group.
jpayne@69 1245 * @see ucol_setMaxVariable
jpayne@69 1246 * @stable ICU 53
jpayne@69 1247 */
jpayne@69 1248 U_STABLE UColReorderCode U_EXPORT2
jpayne@69 1249 ucol_getMaxVariable(const UCollator *coll);
jpayne@69 1250
jpayne@69 1251 #ifndef U_HIDE_DEPRECATED_API
jpayne@69 1252 /**
jpayne@69 1253 * Sets the variable top to the primary weight of the specified string.
jpayne@69 1254 *
jpayne@69 1255 * Beginning with ICU 53, the variable top is pinned to
jpayne@69 1256 * the top of one of the supported reordering groups,
jpayne@69 1257 * and it must not be beyond the last of those groups.
jpayne@69 1258 * See ucol_setMaxVariable().
jpayne@69 1259 * @param coll the collator
jpayne@69 1260 * @param varTop one or more (if contraction) UChars to which the variable top should be set
jpayne@69 1261 * @param len length of variable top string. If -1 it is considered to be zero terminated.
jpayne@69 1262 * @param status error code. If error code is set, the return value is undefined.
jpayne@69 1263 * Errors set by this function are:<br>
jpayne@69 1264 * U_CE_NOT_FOUND_ERROR if more than one character was passed and there is no such contraction<br>
jpayne@69 1265 * U_ILLEGAL_ARGUMENT_ERROR if the variable top is beyond
jpayne@69 1266 * the last reordering group supported by ucol_setMaxVariable()
jpayne@69 1267 * @return variable top primary weight
jpayne@69 1268 * @see ucol_getVariableTop
jpayne@69 1269 * @see ucol_restoreVariableTop
jpayne@69 1270 * @deprecated ICU 53 Call ucol_setMaxVariable() instead.
jpayne@69 1271 */
jpayne@69 1272 U_DEPRECATED uint32_t U_EXPORT2
jpayne@69 1273 ucol_setVariableTop(UCollator *coll,
jpayne@69 1274 const UChar *varTop, int32_t len,
jpayne@69 1275 UErrorCode *status);
jpayne@69 1276 #endif /* U_HIDE_DEPRECATED_API */
jpayne@69 1277
jpayne@69 1278 /**
jpayne@69 1279 * Gets the variable top value of a Collator.
jpayne@69 1280 * @param coll collator which variable top needs to be retrieved
jpayne@69 1281 * @param status error code (not changed by function). If error code is set,
jpayne@69 1282 * the return value is undefined.
jpayne@69 1283 * @return the variable top primary weight
jpayne@69 1284 * @see ucol_getMaxVariable
jpayne@69 1285 * @see ucol_setVariableTop
jpayne@69 1286 * @see ucol_restoreVariableTop
jpayne@69 1287 * @stable ICU 2.0
jpayne@69 1288 */
jpayne@69 1289 U_STABLE uint32_t U_EXPORT2 ucol_getVariableTop(const UCollator *coll, UErrorCode *status);
jpayne@69 1290
jpayne@69 1291 #ifndef U_HIDE_DEPRECATED_API
jpayne@69 1292 /**
jpayne@69 1293 * Sets the variable top to the specified primary weight.
jpayne@69 1294 *
jpayne@69 1295 * Beginning with ICU 53, the variable top is pinned to
jpayne@69 1296 * the top of one of the supported reordering groups,
jpayne@69 1297 * and it must not be beyond the last of those groups.
jpayne@69 1298 * See ucol_setMaxVariable().
jpayne@69 1299 * @param coll collator to be set
jpayne@69 1300 * @param varTop primary weight, as returned by ucol_setVariableTop or ucol_getVariableTop
jpayne@69 1301 * @param status error code
jpayne@69 1302 * @see ucol_getVariableTop
jpayne@69 1303 * @see ucol_setVariableTop
jpayne@69 1304 * @deprecated ICU 53 Call ucol_setMaxVariable() instead.
jpayne@69 1305 */
jpayne@69 1306 U_DEPRECATED void U_EXPORT2
jpayne@69 1307 ucol_restoreVariableTop(UCollator *coll, const uint32_t varTop, UErrorCode *status);
jpayne@69 1308 #endif /* U_HIDE_DEPRECATED_API */
jpayne@69 1309
jpayne@69 1310 /**
jpayne@69 1311 * Thread safe cloning operation. The result is a clone of a given collator.
jpayne@69 1312 * @param coll collator to be cloned
jpayne@69 1313 * @param stackBuffer <em>Deprecated functionality as of ICU 52, use NULL.</em><br>
jpayne@69 1314 * user allocated space for the new clone.
jpayne@69 1315 * If NULL new memory will be allocated.
jpayne@69 1316 * If buffer is not large enough, new memory will be allocated.
jpayne@69 1317 * Clients can use the U_COL_SAFECLONE_BUFFERSIZE.
jpayne@69 1318 * @param pBufferSize <em>Deprecated functionality as of ICU 52, use NULL or 1.</em><br>
jpayne@69 1319 * pointer to size of allocated space.
jpayne@69 1320 * If *pBufferSize == 0, a sufficient size for use in cloning will
jpayne@69 1321 * be returned ('pre-flighting')
jpayne@69 1322 * If *pBufferSize is not enough for a stack-based safe clone,
jpayne@69 1323 * new memory will be allocated.
jpayne@69 1324 * @param status to indicate whether the operation went on smoothly or there were errors
jpayne@69 1325 * An informational status value, U_SAFECLONE_ALLOCATED_ERROR, is used if any
jpayne@69 1326 * allocations were necessary.
jpayne@69 1327 * @return pointer to the new clone
jpayne@69 1328 * @see ucol_open
jpayne@69 1329 * @see ucol_openRules
jpayne@69 1330 * @see ucol_close
jpayne@69 1331 * @stable ICU 2.0
jpayne@69 1332 */
jpayne@69 1333 U_STABLE UCollator* U_EXPORT2
jpayne@69 1334 ucol_safeClone(const UCollator *coll,
jpayne@69 1335 void *stackBuffer,
jpayne@69 1336 int32_t *pBufferSize,
jpayne@69 1337 UErrorCode *status);
jpayne@69 1338
jpayne@69 1339 #ifndef U_HIDE_DEPRECATED_API
jpayne@69 1340
jpayne@69 1341 /** default memory size for the new clone.
jpayne@69 1342 * @deprecated ICU 52. Do not rely on ucol_safeClone() cloning into any provided buffer.
jpayne@69 1343 */
jpayne@69 1344 #define U_COL_SAFECLONE_BUFFERSIZE 1
jpayne@69 1345
jpayne@69 1346 #endif /* U_HIDE_DEPRECATED_API */
jpayne@69 1347
jpayne@69 1348 /**
jpayne@69 1349 * Returns current rules. Delta defines whether full rules are returned or just the tailoring.
jpayne@69 1350 * Returns number of UChars needed to store rules. If buffer is NULL or bufferLen is not enough
jpayne@69 1351 * to store rules, will store up to available space.
jpayne@69 1352 *
jpayne@69 1353 * ucol_getRules() should normally be used instead.
jpayne@69 1354 * See http://userguide.icu-project.org/collation/customization#TOC-Building-on-Existing-Locales
jpayne@69 1355 * @param coll collator to get the rules from
jpayne@69 1356 * @param delta one of UCOL_TAILORING_ONLY, UCOL_FULL_RULES.
jpayne@69 1357 * @param buffer buffer to store the result in. If NULL, you'll get no rules.
jpayne@69 1358 * @param bufferLen length of buffer to store rules in. If less than needed you'll get only the part that fits in.
jpayne@69 1359 * @return current rules
jpayne@69 1360 * @stable ICU 2.0
jpayne@69 1361 * @see UCOL_FULL_RULES
jpayne@69 1362 */
jpayne@69 1363 U_STABLE int32_t U_EXPORT2
jpayne@69 1364 ucol_getRulesEx(const UCollator *coll, UColRuleOption delta, UChar *buffer, int32_t bufferLen);
jpayne@69 1365
jpayne@69 1366 #ifndef U_HIDE_DEPRECATED_API
jpayne@69 1367 /**
jpayne@69 1368 * gets the locale name of the collator. If the collator
jpayne@69 1369 * is instantiated from the rules, then this function returns
jpayne@69 1370 * NULL.
jpayne@69 1371 * @param coll The UCollator for which the locale is needed
jpayne@69 1372 * @param type You can choose between requested, valid and actual
jpayne@69 1373 * locale. For description see the definition of
jpayne@69 1374 * ULocDataLocaleType in uloc.h
jpayne@69 1375 * @param status error code of the operation
jpayne@69 1376 * @return real locale name from which the collation data comes.
jpayne@69 1377 * If the collator was instantiated from rules, returns
jpayne@69 1378 * NULL.
jpayne@69 1379 * @deprecated ICU 2.8 Use ucol_getLocaleByType instead
jpayne@69 1380 */
jpayne@69 1381 U_DEPRECATED const char * U_EXPORT2
jpayne@69 1382 ucol_getLocale(const UCollator *coll, ULocDataLocaleType type, UErrorCode *status);
jpayne@69 1383 #endif /* U_HIDE_DEPRECATED_API */
jpayne@69 1384
jpayne@69 1385 /**
jpayne@69 1386 * gets the locale name of the collator. If the collator
jpayne@69 1387 * is instantiated from the rules, then this function returns
jpayne@69 1388 * NULL.
jpayne@69 1389 * @param coll The UCollator for which the locale is needed
jpayne@69 1390 * @param type You can choose between requested, valid and actual
jpayne@69 1391 * locale. For description see the definition of
jpayne@69 1392 * ULocDataLocaleType in uloc.h
jpayne@69 1393 * @param status error code of the operation
jpayne@69 1394 * @return real locale name from which the collation data comes.
jpayne@69 1395 * If the collator was instantiated from rules, returns
jpayne@69 1396 * NULL.
jpayne@69 1397 * @stable ICU 2.8
jpayne@69 1398 */
jpayne@69 1399 U_STABLE const char * U_EXPORT2
jpayne@69 1400 ucol_getLocaleByType(const UCollator *coll, ULocDataLocaleType type, UErrorCode *status);
jpayne@69 1401
jpayne@69 1402 /**
jpayne@69 1403 * Get a Unicode set that contains all the characters and sequences tailored in
jpayne@69 1404 * this collator. The result must be disposed of by using uset_close.
jpayne@69 1405 * @param coll The UCollator for which we want to get tailored chars
jpayne@69 1406 * @param status error code of the operation
jpayne@69 1407 * @return a pointer to newly created USet. Must be be disposed by using uset_close
jpayne@69 1408 * @see ucol_openRules
jpayne@69 1409 * @see uset_close
jpayne@69 1410 * @stable ICU 2.4
jpayne@69 1411 */
jpayne@69 1412 U_STABLE USet * U_EXPORT2
jpayne@69 1413 ucol_getTailoredSet(const UCollator *coll, UErrorCode *status);
jpayne@69 1414
jpayne@69 1415 #ifndef U_HIDE_INTERNAL_API
jpayne@69 1416 /** Calculates the set of unsafe code points, given a collator.
jpayne@69 1417 * A character is unsafe if you could append any character and cause the ordering to alter significantly.
jpayne@69 1418 * Collation sorts in normalized order, so anything that rearranges in normalization can cause this.
jpayne@69 1419 * Thus if you have a character like a_umlaut, and you add a lower_dot to it,
jpayne@69 1420 * then it normalizes to a_lower_dot + umlaut, and sorts differently.
jpayne@69 1421 * @param coll Collator
jpayne@69 1422 * @param unsafe a fill-in set to receive the unsafe points
jpayne@69 1423 * @param status for catching errors
jpayne@69 1424 * @return number of elements in the set
jpayne@69 1425 * @internal ICU 3.0
jpayne@69 1426 */
jpayne@69 1427 U_INTERNAL int32_t U_EXPORT2
jpayne@69 1428 ucol_getUnsafeSet( const UCollator *coll,
jpayne@69 1429 USet *unsafe,
jpayne@69 1430 UErrorCode *status);
jpayne@69 1431
jpayne@69 1432 /** Touches all resources needed for instantiating a collator from a short string definition,
jpayne@69 1433 * thus filling up the cache.
jpayne@69 1434 * @param definition A short string containing a locale and a set of attributes.
jpayne@69 1435 * Attributes not explicitly mentioned are left at the default
jpayne@69 1436 * state for a locale.
jpayne@69 1437 * @param parseError if not NULL, structure that will get filled with error's pre
jpayne@69 1438 * and post context in case of error.
jpayne@69 1439 * @param forceDefaults if FALSE, the settings that are the same as the collator
jpayne@69 1440 * default settings will not be applied (for example, setting
jpayne@69 1441 * French secondary on a French collator would not be executed).
jpayne@69 1442 * If TRUE, all the settings will be applied regardless of the
jpayne@69 1443 * collator default value. If the definition
jpayne@69 1444 * strings are to be cached, should be set to FALSE.
jpayne@69 1445 * @param status Error code. Apart from regular error conditions connected to
jpayne@69 1446 * instantiating collators (like out of memory or similar), this
jpayne@69 1447 * API will return an error if an invalid attribute or attribute/value
jpayne@69 1448 * combination is specified.
jpayne@69 1449 * @see ucol_openFromShortString
jpayne@69 1450 * @internal ICU 3.2.1
jpayne@69 1451 */
jpayne@69 1452 U_INTERNAL void U_EXPORT2
jpayne@69 1453 ucol_prepareShortStringOpen( const char *definition,
jpayne@69 1454 UBool forceDefaults,
jpayne@69 1455 UParseError *parseError,
jpayne@69 1456 UErrorCode *status);
jpayne@69 1457 #endif /* U_HIDE_INTERNAL_API */
jpayne@69 1458
jpayne@69 1459 /** Creates a binary image of a collator. This binary image can be stored and
jpayne@69 1460 * later used to instantiate a collator using ucol_openBinary.
jpayne@69 1461 * This API supports preflighting.
jpayne@69 1462 * @param coll Collator
jpayne@69 1463 * @param buffer a fill-in buffer to receive the binary image
jpayne@69 1464 * @param capacity capacity of the destination buffer
jpayne@69 1465 * @param status for catching errors
jpayne@69 1466 * @return size of the image
jpayne@69 1467 * @see ucol_openBinary
jpayne@69 1468 * @stable ICU 3.2
jpayne@69 1469 */
jpayne@69 1470 U_STABLE int32_t U_EXPORT2
jpayne@69 1471 ucol_cloneBinary(const UCollator *coll,
jpayne@69 1472 uint8_t *buffer, int32_t capacity,
jpayne@69 1473 UErrorCode *status);
jpayne@69 1474
jpayne@69 1475 /** Opens a collator from a collator binary image created using
jpayne@69 1476 * ucol_cloneBinary. Binary image used in instantiation of the
jpayne@69 1477 * collator remains owned by the user and should stay around for
jpayne@69 1478 * the lifetime of the collator. The API also takes a base collator
jpayne@69 1479 * which must be the root collator.
jpayne@69 1480 * @param bin binary image owned by the user and required through the
jpayne@69 1481 * lifetime of the collator
jpayne@69 1482 * @param length size of the image. If negative, the API will try to
jpayne@69 1483 * figure out the length of the image
jpayne@69 1484 * @param base Base collator, for lookup of untailored characters.
jpayne@69 1485 * Must be the root collator, must not be NULL.
jpayne@69 1486 * The base is required to be present through the lifetime of the collator.
jpayne@69 1487 * @param status for catching errors
jpayne@69 1488 * @return newly created collator
jpayne@69 1489 * @see ucol_cloneBinary
jpayne@69 1490 * @stable ICU 3.2
jpayne@69 1491 */
jpayne@69 1492 U_STABLE UCollator* U_EXPORT2
jpayne@69 1493 ucol_openBinary(const uint8_t *bin, int32_t length,
jpayne@69 1494 const UCollator *base,
jpayne@69 1495 UErrorCode *status);
jpayne@69 1496
jpayne@69 1497
jpayne@69 1498 #endif /* #if !UCONFIG_NO_COLLATION */
jpayne@69 1499
jpayne@69 1500 #endif