icuSources/i18n/unicode/ucol.h

   1 /*
   2 *******************************************************************************
   3 * Copyright (c) 1996-2012, International Business Machines Corporation and others.
   4 * All Rights Reserved.
   5 *******************************************************************************
   6 */
   7
   8 #ifndef UCOL_H
   9 #define UCOL_H
  10
  11 #include "unicode/utypes.h"
  12
  13 #if !UCONFIG_NO_COLLATION
  14
  15 #include "unicode/unorm.h"
  16 #include "unicode/localpointer.h"
  17 #include "unicode/parseerr.h"
  18 #include "unicode/uloc.h"
  19 #include "unicode/uset.h"
  20 #include "unicode/uscript.h"
  21
  22 /**
  23  * \file
  24  * \brief C API: Collator
  25  *
  26  * <h2> Collator C API </h2>
  27  *
  28  * The C API for Collator performs locale-sensitive
  29  * string comparison. You use this service to build
  30  * searching and sorting routines for natural language text.
  31  * <em>Important: </em>The ICU collation service has been reimplemented
  32  * in order to achieve better performance and UCA compliance.
  33  * For details, see the
  34  * <a href="http://source.icu-project.org/repos/icu/icuhtml/trunk/design/collation/ICU_collation_design.htm">
  35  * collation design document</a>.
  36  * <p>
  37  * For more information about the collation service see
  38  * <a href="http://icu-project.org/userguide/Collate_Intro.html">the users guide</a>.
  39  * <p>
  40  * Collation service provides correct sorting orders for most locales supported in ICU.
  41  * If specific data for a locale is not available, the orders eventually falls back
  42  * to the <a href="http://www.unicode.org/unicode/reports/tr10/">UCA sort order</a>.
  43  * <p>
  44  * Sort ordering may be customized by providing your own set of rules. For more on
  45  * this subject see the
  46  * <a href="http://icu-project.org/userguide/Collate_Customization.html">
  47  * Collation customization</a> section of the users guide.
  48  * <p>
  49  * @see         UCollationResult
  50  * @see         UNormalizationMode
  51  * @see         UCollationStrength
  52  * @see         UCollationElements
  53  */
  54
  55 /** A collator.
  56 *  For usage in C programs.
  57 */
  58 struct UCollator;
  59 /** structure representing a collator object instance
  60  * @stable ICU 2.0
  61  */
  62 typedef struct UCollator UCollator;
  63
  64
  65 /**
  66  * UCOL_LESS is returned if source string is compared to be less than target
  67  * string in the u_strcoll() method.
  68  * UCOL_EQUAL is returned if source string is compared to be equal to target
  69  * string in the u_strcoll() method.
  70  * UCOL_GREATER is returned if source string is compared to be greater than
  71  * target string in the u_strcoll() method.
  72  * @see u_strcoll()
  73  * <p>
  74  * Possible values for a comparison result
  75  * @stable ICU 2.0
  76  */
  77 typedef enum {
  78   /** string a == string b */
  79   UCOL_EQUAL    = 0,
  80   /** string a > string b */
  81   UCOL_GREATER    = 1,
  82   /** string a < string b */
  83   UCOL_LESS    = -1
  84 } UCollationResult ;
  85
  86
  87 /** Enum containing attribute values for controling collation behavior.
  88  * Here are all the allowable values. Not every attribute can take every value. The only
  89  * universal value is UCOL_DEFAULT, which resets the attribute value to the predefined
  90  * value for that locale
  91  * @stable ICU 2.0
  92  */
  93 typedef enum {
  94   /** accepted by most attributes */
  95   UCOL_DEFAULT = -1,
  96
  97   /** Primary collation strength */
  98   UCOL_PRIMARY = 0,
  99   /** Secondary collation strength */
 100   UCOL_SECONDARY = 1,
 101   /** Tertiary collation strength */
 102   UCOL_TERTIARY = 2,
 103   /** Default collation strength */
 104   UCOL_DEFAULT_STRENGTH = UCOL_TERTIARY,
 105   UCOL_CE_STRENGTH_LIMIT,
 106   /** Quaternary collation strength */
 107   UCOL_QUATERNARY=3,
 108   /** Identical collation strength */
 109   UCOL_IDENTICAL=15,
 110   UCOL_STRENGTH_LIMIT,
 111
 112   /** Turn the feature off - works for UCOL_FRENCH_COLLATION,
 113       UCOL_CASE_LEVEL, UCOL_HIRAGANA_QUATERNARY_MODE
 114       & UCOL_DECOMPOSITION_MODE*/
 115   UCOL_OFF = 16,
 116   /** Turn the feature on - works for UCOL_FRENCH_COLLATION,
 117       UCOL_CASE_LEVEL, UCOL_HIRAGANA_QUATERNARY_MODE
 118       & UCOL_DECOMPOSITION_MODE*/
 119   UCOL_ON = 17,
 120
 121   /** Valid for UCOL_ALTERNATE_HANDLING. Alternate handling will be shifted */
 122   UCOL_SHIFTED = 20,
 123   /** Valid for UCOL_ALTERNATE_HANDLING. Alternate handling will be non ignorable */
 124   UCOL_NON_IGNORABLE = 21,
 125
 126   /** Valid for UCOL_CASE_FIRST -
 127       lower case sorts before upper case */
 128   UCOL_LOWER_FIRST = 24,
 129   /** upper case sorts before lower case */
 130   UCOL_UPPER_FIRST = 25,
 131
 132   UCOL_ATTRIBUTE_VALUE_COUNT
 133
 134 } UColAttributeValue;
 135
 136 /**
 137  * Enum containing the codes for reordering segments of the collation table that are not script
 138  * codes. These reordering codes are to be used in conjunction with the script codes.
 139  * @see ucol_getReorderCodes
 140  * @see ucol_setReorderCodes
 141  * @see ucol_getEquivalentReorderCodes
 142  * @see UScriptCode
 143  * @stable ICU 4.8
 144  */
 145  typedef enum {
 146    /**
 147     * A special reordering code that is used to specify the default
 148     * reordering codes for a locale.
 149     * @stable ICU 4.8
 150     */
 151     UCOL_REORDER_CODE_DEFAULT       = -1,
 152    /**
 153     * A special reordering code that is used to specify no reordering codes.
 154     * @stable ICU 4.8
 155     */
 156     UCOL_REORDER_CODE_NONE          = USCRIPT_UNKNOWN,
 157    /**
 158     * A special reordering code that is used to specify all other codes used for
 159     * reordering except for the codes lised as UColReorderCode values and those
 160     * listed explicitly in a reordering.
 161     * @stable ICU 4.8
 162     */
 163     UCOL_REORDER_CODE_OTHERS        = USCRIPT_UNKNOWN,
 164    /**
 165     * Characters with the space property.
 166     * This is equivalent to the rule value "space".
 167     * @stable ICU 4.8
 168     */
 169     UCOL_REORDER_CODE_SPACE         = 0x1000,
 170    /**
 171     * The first entry in the enumeration of reordering groups. This is intended for use in
 172     * range checking and enumeration of the reorder codes.
 173     * @stable ICU 4.8
 174     */
 175     UCOL_REORDER_CODE_FIRST         = UCOL_REORDER_CODE_SPACE,
 176    /**
 177     * Characters with the punctuation property.
 178     * This is equivalent to the rule value "punct".
 179     * @stable ICU 4.8
 180     */
 181     UCOL_REORDER_CODE_PUNCTUATION   = 0x1001,
 182    /**
 183     * Characters with the symbol property.
 184     * This is equivalent to the rule value "symbol".
 185     * @stable ICU 4.8
 186     */
 187     UCOL_REORDER_CODE_SYMBOL        = 0x1002,
 188    /**
 189     * Characters with the currency property.
 190     * This is equivalent to the rule value "currency".
 191     * @stable ICU 4.8
 192     */
 193     UCOL_REORDER_CODE_CURRENCY      = 0x1003,
 194    /**
 195     * Characters with the digit property.
 196     * This is equivalent to the rule value "digit".
 197     * @stable ICU 4.8
 198     */
 199     UCOL_REORDER_CODE_DIGIT         = 0x1004,
 200    /**
 201     * The limit of the reorder codes. This is intended for use in range checking
 202     * and enumeration of the reorder codes.
 203     * @stable ICU 4.8
 204     */
 205     UCOL_REORDER_CODE_LIMIT         = 0x1005
 206 } UColReorderCode;
 207
 208 /**
 209  * Base letter represents a primary difference.  Set comparison
 210  * level to UCOL_PRIMARY to ignore secondary and tertiary differences.
 211  * Use this to set the strength of a Collator object.
 212  * Example of primary difference, "abc" &lt; "abd"
 213  *
 214  * Diacritical differences on the same base letter represent a secondary
 215  * difference.  Set comparison level to UCOL_SECONDARY to ignore tertiary
 216  * differences. Use this to set the strength of a Collator object.
 217  * Example of secondary difference, "&auml;" >> "a".
 218  *
 219  * Uppercase and lowercase versions of the same character represents a
 220  * tertiary difference.  Set comparison level to UCOL_TERTIARY to include
 221  * all comparison differences. Use this to set the strength of a Collator
 222  * object.
 223  * Example of tertiary difference, "abc" &lt;&lt;&lt; "ABC".
 224  *
 225  * Two characters are considered "identical" when they have the same
 226  * unicode spellings.  UCOL_IDENTICAL.
 227  * For example, "&auml;" == "&auml;".
 228  *
 229  * UCollationStrength is also used to determine the strength of sort keys
 230  * generated from UCollator objects
 231  * These values can be now found in the UColAttributeValue enum.
 232  * @stable ICU 2.0
 233  **/
 234 typedef UColAttributeValue UCollationStrength;
 235
 236 /** Attributes that collation service understands. All the attributes can take UCOL_DEFAULT
 237  * value, as well as the values specific to each one.
 238  * @stable ICU 2.0
 239  */
 240 typedef enum {
 241      /** Attribute for direction of secondary weights - used in French.
 242       * Acceptable values are UCOL_ON, which results in secondary weights
 243       * being considered backwards and UCOL_OFF which treats secondary
 244       * weights in the order they appear.*/
 245      UCOL_FRENCH_COLLATION,
 246      /** Attribute for handling variable elements.
 247       * Acceptable values are UCOL_NON_IGNORABLE (default)
 248       * which treats all the codepoints with non-ignorable
 249       * primary weights in the same way,
 250       * and UCOL_SHIFTED which causes codepoints with primary
 251       * weights that are equal or below the variable top value
 252       * to be ignored on primary level and moved to the quaternary
 253       * level.*/
 254      UCOL_ALTERNATE_HANDLING,
 255      /** Controls the ordering of upper and lower case letters.
 256       * Acceptable values are UCOL_OFF (default), which orders
 257       * upper and lower case letters in accordance to their tertiary
 258       * weights, UCOL_UPPER_FIRST which forces upper case letters to
 259       * sort before lower case letters, and UCOL_LOWER_FIRST which does
 260       * the opposite. */
 261      UCOL_CASE_FIRST,
 262      /** Controls whether an extra case level (positioned before the third
 263       * level) is generated or not. Acceptable values are UCOL_OFF (default),
 264       * when case level is not generated, and UCOL_ON which causes the case
 265       * level to be generated. Contents of the case level are affected by
 266       * the value of UCOL_CASE_FIRST attribute. A simple way to ignore
 267       * accent differences in a string is to set the strength to UCOL_PRIMARY
 268       * and enable case level. */
 269      UCOL_CASE_LEVEL,
 270      /** Controls whether the normalization check and necessary normalizations
 271       * are performed. When set to UCOL_OFF (default) no normalization check
 272       * is performed. The correctness of the result is guaranteed only if the
 273       * input data is in so-called FCD form (see users manual for more info).
 274       * When set to UCOL_ON, an incremental check is performed to see whether
 275       * the input data is in the FCD form. If the data is not in the FCD form,
 276       * incremental NFD normalization is performed. */
 277      UCOL_NORMALIZATION_MODE,
 278      /** An alias for UCOL_NORMALIZATION_MODE attribute */
 279      UCOL_DECOMPOSITION_MODE = UCOL_NORMALIZATION_MODE,
 280      /** The strength attribute. Can be either UCOL_PRIMARY, UCOL_SECONDARY,
 281       * UCOL_TERTIARY, UCOL_QUATERNARY or UCOL_IDENTICAL. The usual strength
 282       * for most locales (except Japanese) is tertiary. Quaternary strength
 283       * is useful when combined with shifted setting for alternate handling
 284       * attribute and for JIS x 4061 collation, when it is used to distinguish
 285       * between Katakana  and Hiragana (this is achieved by setting the
 286       * UCOL_HIRAGANA_QUATERNARY mode to on. Otherwise, quaternary level
 287       * is affected only by the number of non ignorable code points in
 288       * the string. Identical strength is rarely useful, as it amounts
 289       * to codepoints of the NFD form of the string. */
 290      UCOL_STRENGTH,
 291      /** When turned on, this attribute positions Hiragana before all
 292       * non-ignorables on quaternary level This is a sneaky way to produce JIS
 293       * sort order */
 294      UCOL_HIRAGANA_QUATERNARY_MODE,
 295      /** When turned on, this attribute generates a collation key
 296       * for the numeric value of substrings of digits.
 297       * This is a way to get '100' to sort AFTER '2'. Note that the longest
 298       * digit substring that can be treated as a single collation element is
 299       * 254 digits (not counting leading zeros). If a digit substring is
 300       * longer than that, the digits beyond the limit will be treated as a
 301       * separate digit substring associated with a separate collation element. */
 302      UCOL_NUMERIC_COLLATION,
 303      UCOL_ATTRIBUTE_COUNT
 304 } UColAttribute;
 305
 306 /** Options for retrieving the rule string
 307  *  @stable ICU 2.0
 308  */
 309 typedef enum {
 310   /** Retrieve tailoring only */
 311   UCOL_TAILORING_ONLY,
 312   /** Retrieve UCA rules and tailoring */
 313   UCOL_FULL_RULES
 314 } UColRuleOption ;
 315
 316 /**
 317  * Open a UCollator for comparing strings.
 318  * The UCollator pointer is used in all the calls to the Collation
 319  * service. After finished, collator must be disposed of by calling
 320  * {@link #ucol_close }.
 321  * @param loc The locale containing the required collation rules.
 322  *            Special values for locales can be passed in -
 323  *            if NULL is passed for the locale, the default locale
 324  *            collation rules will be used. If empty string ("") or
 325  *            "root" are passed, UCA rules will be used.
 326  * @param status A pointer to an UErrorCode to receive any errors
 327  * @return A pointer to a UCollator, or 0 if an error occurred.
 328  * @see ucol_openRules
 329  * @see ucol_safeClone
 330  * @see ucol_close
 331  * @stable ICU 2.0
 332  */
 333 U_STABLE UCollator* U_EXPORT2
 334 ucol_open(const char *loc, UErrorCode *status);
 335
 336 /**
 337  * Produce an UCollator instance according to the rules supplied.
 338  * The rules are used to change the default ordering, defined in the
 339  * UCA in a process called tailoring. The resulting UCollator pointer
 340  * can be used in the same way as the one obtained by {@link #ucol_strcoll }.
 341  * @param rules A string describing the collation rules. For the syntax
 342  *              of the rules please see users guide.
 343  * @param rulesLength The length of rules, or -1 if null-terminated.
 344  * @param normalizationMode The normalization mode: One of
 345  *             UCOL_OFF     (expect the text to not need normalization),
 346  *             UCOL_ON      (normalize), or
 347  *             UCOL_DEFAULT (set the mode according to the rules)
 348  * @param strength The default collation strength; one of UCOL_PRIMARY, UCOL_SECONDARY,
 349  * UCOL_TERTIARY, UCOL_IDENTICAL,UCOL_DEFAULT_STRENGTH - can be also set in the rules.
 350  * @param parseError  A pointer to UParseError to recieve information about errors
 351  *                    occurred during parsing. This argument can currently be set
 352  *                    to NULL, but at users own risk. Please provide a real structure.
 353  * @param status A pointer to an UErrorCode to receive any errors
 354  * @return A pointer to a UCollator. It is not guaranteed that NULL be returned in case
 355  *         of error - please use status argument to check for errors.
 356  * @see ucol_open
 357  * @see ucol_safeClone
 358  * @see ucol_close
 359  * @stable ICU 2.0
 360  */
 361 U_STABLE UCollator* U_EXPORT2
 362 ucol_openRules( const UChar        *rules,
 363                 int32_t            rulesLength,
 364                 UColAttributeValue normalizationMode,
 365                 UCollationStrength strength,
 366                 UParseError        *parseError,
 367                 UErrorCode         *status);
 368
 369 /**
 370  * Open a collator defined by a short form string.
 371  * The structure and the syntax of the string is defined in the "Naming collators"
 372  * section of the users guide:
 373  * http://icu-project.org/userguide/Collate_Concepts.html#Naming_Collators
 374  * Attributes are overriden by the subsequent attributes. So, for "S2_S3", final
 375  * strength will be 3. 3066bis locale overrides individual locale parts.
 376  * The call to this function is equivalent to a call to ucol_open, followed by a
 377  * series of calls to ucol_setAttribute and ucol_setVariableTop.
 378  * @param definition A short string containing a locale and a set of attributes.
 379  *                   Attributes not explicitly mentioned are left at the default
 380  *                   state for a locale.
 381  * @param parseError if not NULL, structure that will get filled with error's pre
 382  *                   and post context in case of error.
 383  * @param forceDefaults if FALSE, the settings that are the same as the collator
 384  *                   default settings will not be applied (for example, setting
 385  *                   French secondary on a French collator would not be executed).
 386  *                   If TRUE, all the settings will be applied regardless of the
 387  *                   collator default value. If the definition
 388  *                   strings are to be cached, should be set to FALSE.
 389  * @param status     Error code. Apart from regular error conditions connected to
 390  *                   instantiating collators (like out of memory or similar), this
 391  *                   API will return an error if an invalid attribute or attribute/value
 392  *                   combination is specified.
 393  * @return           A pointer to a UCollator or 0 if an error occured (including an
 394  *                   invalid attribute).
 395  * @see ucol_open
 396  * @see ucol_setAttribute
 397  * @see ucol_setVariableTop
 398  * @see ucol_getShortDefinitionString
 399  * @see ucol_normalizeShortDefinitionString
 400  * @stable ICU 3.0
 401  *
 402  */
 403 U_STABLE UCollator* U_EXPORT2
 404 ucol_openFromShortString( const char *definition,
 405                           UBool forceDefaults,
 406                           UParseError *parseError,
 407                           UErrorCode *status);
 408
 409 #ifndef U_HIDE_DEPRECATED_API
 410 /**
 411  * Get a set containing the contractions defined by the collator. The set includes
 412  * both the UCA contractions and the contractions defined by the collator. This set
 413  * will contain only strings. If a tailoring explicitly suppresses contractions from
 414  * the UCA (like Russian), removed contractions will not be in the resulting set.
 415  * @param coll collator
 416  * @param conts the set to hold the result. It gets emptied before
 417  *              contractions are added.
 418  * @param status to hold the error code
 419  * @return the size of the contraction set
 420  *
 421  * @deprecated ICU 3.4, use ucol_getContractionsAndExpansions instead
 422  */
 423 U_DEPRECATED int32_t U_EXPORT2
 424 ucol_getContractions( const UCollator *coll,
 425                   USet *conts,
 426                   UErrorCode *status);
 427 #endif  /* U_HIDE_DEPRECATED_API */
 428
 429 /**
 430  * Get a set containing the expansions defined by the collator. The set includes
 431  * both the UCA expansions and the expansions defined by the tailoring
 432  * @param coll collator
 433  * @param contractions if not NULL, the set to hold the contractions
 434  * @param expansions if not NULL, the set to hold the expansions
 435  * @param addPrefixes add the prefix contextual elements to contractions
 436  * @param status to hold the error code
 437  *
 438  * @stable ICU 3.4
 439  */
 440 U_STABLE void U_EXPORT2
 441 ucol_getContractionsAndExpansions( const UCollator *coll,
 442                   USet *contractions, USet *expansions,
 443                   UBool addPrefixes, UErrorCode *status);
 444
 445 /**
 446  * Close a UCollator.
 447  * Once closed, a UCollator should not be used. Every open collator should
 448  * be closed. Otherwise, a memory leak will result.
 449  * @param coll The UCollator to close.
 450  * @see ucol_open
 451  * @see ucol_openRules
 452  * @see ucol_safeClone
 453  * @stable ICU 2.0
 454  */
 455 U_STABLE void U_EXPORT2
 456 ucol_close(UCollator *coll);
 457
 458 #if U_SHOW_CPLUSPLUS_API
 459
 460 U_NAMESPACE_BEGIN
 461
 462 /**
 463  * \class LocalUCollatorPointer
 464  * "Smart pointer" class, closes a UCollator via ucol_close().
 465  * For most methods see the LocalPointerBase base class.
 466  *
 467  * @see LocalPointerBase
 468  * @see LocalPointer
 469  * @stable ICU 4.4
 470  */
 471 U_DEFINE_LOCAL_OPEN_POINTER(LocalUCollatorPointer, UCollator, ucol_close);
 472
 473 U_NAMESPACE_END
 474
 475 #endif
 476
 477 /**
 478  * Compare two strings.
 479  * The strings will be compared using the options already specified.
 480  * @param coll The UCollator containing the comparison rules.
 481  * @param source The source string.
 482  * @param sourceLength The length of source, or -1 if null-terminated.
 483  * @param target The target string.
 484  * @param targetLength The length of target, or -1 if null-terminated.
 485  * @return The result of comparing the strings; one of UCOL_EQUAL,
 486  * UCOL_GREATER, UCOL_LESS
 487  * @see ucol_greater
 488  * @see ucol_greaterOrEqual
 489  * @see ucol_equal
 490  * @stable ICU 2.0
 491  */
 492 U_STABLE UCollationResult U_EXPORT2
 493 ucol_strcoll(    const    UCollator    *coll,
 494         const    UChar        *source,
 495         int32_t            sourceLength,
 496         const    UChar        *target,
 497         int32_t            targetLength);
 498
 499 /**
 500  * Determine if one string is greater than another.
 501  * This function is equivalent to {@link #ucol_strcoll } == UCOL_GREATER
 502  * @param coll The UCollator containing the comparison rules.
 503  * @param source The source string.
 504  * @param sourceLength The length of source, or -1 if null-terminated.
 505  * @param target The target string.
 506  * @param targetLength The length of target, or -1 if null-terminated.
 507  * @return TRUE if source is greater than target, FALSE otherwise.
 508  * @see ucol_strcoll
 509  * @see ucol_greaterOrEqual
 510  * @see ucol_equal
 511  * @stable ICU 2.0
 512  */
 513 U_STABLE UBool U_EXPORT2
 514 ucol_greater(const UCollator *coll,
 515              const UChar     *source, int32_t sourceLength,
 516              const UChar     *target, int32_t targetLength);
 517
 518 /**
 519  * Determine if one string is greater than or equal to another.
 520  * This function is equivalent to {@link #ucol_strcoll } != UCOL_LESS
 521  * @param coll The UCollator containing the comparison rules.
 522  * @param source The source string.
 523  * @param sourceLength The length of source, or -1 if null-terminated.
 524  * @param target The target string.
 525  * @param targetLength The length of target, or -1 if null-terminated.
 526  * @return TRUE if source is greater than or equal to target, FALSE otherwise.
 527  * @see ucol_strcoll
 528  * @see ucol_greater
 529  * @see ucol_equal
 530  * @stable ICU 2.0
 531  */
 532 U_STABLE UBool U_EXPORT2
 533 ucol_greaterOrEqual(const UCollator *coll,
 534                     const UChar     *source, int32_t sourceLength,
 535                     const UChar     *target, int32_t targetLength);
 536
 537 /**
 538  * Compare two strings for equality.
 539  * This function is equivalent to {@link #ucol_strcoll } == UCOL_EQUAL
 540  * @param coll The UCollator containing the comparison rules.
 541  * @param source The source string.
 542  * @param sourceLength The length of source, or -1 if null-terminated.
 543  * @param target The target string.
 544  * @param targetLength The length of target, or -1 if null-terminated.
 545  * @return TRUE if source is equal to target, FALSE otherwise
 546  * @see ucol_strcoll
 547  * @see ucol_greater
 548  * @see ucol_greaterOrEqual
 549  * @stable ICU 2.0
 550  */
 551 U_STABLE UBool U_EXPORT2
 552 ucol_equal(const UCollator *coll,
 553            const UChar     *source, int32_t sourceLength,
 554            const UChar     *target, int32_t targetLength);
 555
 556 /**
 557  * Compare two UTF-8 encoded trings.
 558  * The strings will be compared using the options already specified.
 559  * @param coll The UCollator containing the comparison rules.
 560  * @param sIter The source string iterator.
 561  * @param tIter The target string iterator.
 562  * @return The result of comparing the strings; one of UCOL_EQUAL,
 563  * UCOL_GREATER, UCOL_LESS
 564  * @param status A pointer to an UErrorCode to receive any errors
 565  * @see ucol_strcoll
 566  * @stable ICU 2.6
 567  */
 568 U_STABLE UCollationResult U_EXPORT2
 569 ucol_strcollIter(  const    UCollator    *coll,
 570                   UCharIterator *sIter,
 571                   UCharIterator *tIter,
 572                   UErrorCode *status);
 573
 574 /**
 575  * Get the collation strength used in a UCollator.
 576  * The strength influences how strings are compared.
 577  * @param coll The UCollator to query.
 578  * @return The collation strength; one of UCOL_PRIMARY, UCOL_SECONDARY,
 579  * UCOL_TERTIARY, UCOL_QUATERNARY, UCOL_IDENTICAL
 580  * @see ucol_setStrength
 581  * @stable ICU 2.0
 582  */
 583 U_STABLE UCollationStrength U_EXPORT2
 584 ucol_getStrength(const UCollator *coll);
 585
 586 /**
 587  * Set the collation strength used in a UCollator.
 588  * The strength influences how strings are compared.
 589  * @param coll The UCollator to set.
 590  * @param strength The desired collation strength; one of UCOL_PRIMARY,
 591  * UCOL_SECONDARY, UCOL_TERTIARY, UCOL_QUATERNARY, UCOL_IDENTICAL, UCOL_DEFAULT
 592  * @see ucol_getStrength
 593  * @stable ICU 2.0
 594  */
 595 U_STABLE void U_EXPORT2
 596 ucol_setStrength(UCollator *coll,
 597                  UCollationStrength strength);
 598
 599 /**
 600  * Retrieves the reordering codes for this collator.
 601  * These reordering codes are a combination of UScript codes and UColReorderCode entries.
 602  * @param coll The UCollator to query.
 603  * @param dest The array to fill with the script ordering.
 604  * @param destCapacity The length of dest. If it is 0, then dest may be NULL and the function
 605  * will only return the length of the result without writing any of the result string (pre-flighting).
 606  * @param pErrorCode Must be a valid pointer to an error code value, which must not indicate a
 607  * failure before the function call.
 608  * @return The number of reordering codes written to the dest array.
 609  * @see ucol_setReorderCodes
 610  * @see ucol_getEquivalentReorderCodes
 611  * @see UScriptCode
 612  * @see UColReorderCode
 613  * @stable ICU 4.8
 614  */
 615 U_DRAFT int32_t U_EXPORT2
 616 ucol_getReorderCodes(const UCollator* coll,
 617                     int32_t* dest,
 618                     int32_t destCapacity,
 619                     UErrorCode *pErrorCode);
 620 /**
 621  * Sets the reordering codes for this collator.
 622  * Collation reordering allows scripts and some other defined blocks of characters
 623  * to be moved relative to each other as a block. This reordering is done on top of
 624  * the DUCET/CLDR standard collation order. Reordering can specify groups to be placed
 625  * at the start and/or the end of the collation order. These groups are specified using
 626  * UScript codes and UColReorderCode entries.
 627  * <p>By default, reordering codes specified for the start of the order are placed in the
 628  * order given after a group of "special" non-script blocks. These special groups of characters
 629  * are space, punctuation, symbol, currency, and digit. These special groups are represented with
 630  * UColReorderCode entries. Script groups can be intermingled with
 631  * these special non-script blocks if those special blocks are explicitly specified in the reordering.
 632  * <p>The special code OTHERS stands for any script that is not explicitly
 633  * mentioned in the list of reordering codes given. Anything that is after OTHERS
 634  * will go at the very end of the reordering in the order given.
 635  * <p>The special reorder code DEFAULT will reset the reordering for this collator
 636  * to the default for this collator. The default reordering may be the DUCET/CLDR order or may be a reordering that
 637  * was specified when this collator was created from resource data or from rules. The
 638  * DEFAULT code <b>must</b> be the sole code supplied when it used. If not
 639  * that will result in an U_ILLEGAL_ARGUMENT_ERROR being set.
 640  * <p>The special reorder code NONE will remove any reordering for this collator.
 641  * The result of setting no reordering will be to have the DUCET/CLDR ordering used. The
 642  * NONE code <b>must</b> be the sole code supplied when it used.
 643  * @param coll The UCollator to set.
 644  * @param reorderCodes An array of script codes in the new order. This can be NULL if the
 645  * length is also set to 0. An empty array will clear any reordering codes on the collator.
 646  * @param reorderCodesLength The length of reorderCodes.
 647  * @param pErrorCode Must be a valid pointer to an error code value, which must not indicate a
 648  * failure before the function call.
 649  * @see ucol_getReorderCodes
 650  * @see ucol_getEquivalentReorderCodes
 651  * @see UScriptCode
 652  * @see UColReorderCode
 653  * @stable ICU 4.8
 654  */
 655 U_DRAFT void U_EXPORT2
 656 ucol_setReorderCodes(UCollator* coll,
 657                     const int32_t* reorderCodes,
 658                     int32_t reorderCodesLength,
 659                     UErrorCode *pErrorCode);
 660
 661 /**
 662  * Retrieves the reorder codes that are grouped with the given reorder code. Some reorder
 663  * codes will be grouped and must reorder together.
 664  * @param reorderCode The reorder code to determine equivalence for.
 665  * @param dest The array to fill with the script ordering.
 666  * @param destCapacity The length of dest. If it is 0, then dest may be NULL and the function
 667  * will only return the length of the result without writing any of the result string (pre-flighting).
 668  * @param pErrorCode Must be a valid pointer to an error code value, which must not indicate
 669  * a failure before the function call.
 670  * @return The number of reordering codes written to the dest array.
 671  * @see ucol_setReorderCodes
 672  * @see ucol_getReorderCodes
 673  * @see UScriptCode
 674  * @see UColReorderCode
 675  * @stable ICU 4.8
 676  */
 677 U_DRAFT int32_t U_EXPORT2
 678 ucol_getEquivalentReorderCodes(int32_t reorderCode,
 679                     int32_t* dest,
 680                     int32_t destCapacity,
 681                     UErrorCode *pErrorCode);
 682
 683 /**
 684  * Get the display name for a UCollator.
 685  * The display name is suitable for presentation to a user.
 686  * @param objLoc The locale of the collator in question.
 687  * @param dispLoc The locale for display.
 688  * @param result A pointer to a buffer to receive the attribute.
 689  * @param resultLength The maximum size of result.
 690  * @param status A pointer to an UErrorCode to receive any errors
 691  * @return The total buffer size needed; if greater than resultLength,
 692  * the output was truncated.
 693  * @stable ICU 2.0
 694  */
 695 U_STABLE int32_t U_EXPORT2
 696 ucol_getDisplayName(    const    char        *objLoc,
 697             const    char        *dispLoc,
 698             UChar             *result,
 699             int32_t         resultLength,
 700             UErrorCode        *status);
 701
 702 /**
 703  * Get a locale for which collation rules are available.
 704  * A UCollator in a locale returned by this function will perform the correct
 705  * collation for the locale.
 706  * @param localeIndex The index of the desired locale.
 707  * @return A locale for which collation rules are available, or 0 if none.
 708  * @see ucol_countAvailable
 709  * @stable ICU 2.0
 710  */
 711 U_STABLE const char* U_EXPORT2
 712 ucol_getAvailable(int32_t localeIndex);
 713
 714 /**
 715  * Determine how many locales have collation rules available.
 716  * This function is most useful as determining the loop ending condition for
 717  * calls to {@link #ucol_getAvailable }.
 718  * @return The number of locales for which collation rules are available.
 719  * @see ucol_getAvailable
 720  * @stable ICU 2.0
 721  */
 722 U_STABLE int32_t U_EXPORT2
 723 ucol_countAvailable(void);
 724
 725 #if !UCONFIG_NO_SERVICE
 726 /**
 727  * Create a string enumerator of all locales for which a valid
 728  * collator may be opened.
 729  * @param status input-output error code
 730  * @return a string enumeration over locale strings. The caller is
 731  * responsible for closing the result.
 732  * @stable ICU 3.0
 733  */
 734 U_STABLE UEnumeration* U_EXPORT2
 735 ucol_openAvailableLocales(UErrorCode *status);
 736 #endif
 737
 738 /**
 739  * Create a string enumerator of all possible keywords that are relevant to
 740  * collation. At this point, the only recognized keyword for this
 741  * service is "collation".
 742  * @param status input-output error code
 743  * @return a string enumeration over locale strings. The caller is
 744  * responsible for closing the result.
 745  * @stable ICU 3.0
 746  */
 747 U_STABLE UEnumeration* U_EXPORT2
 748 ucol_getKeywords(UErrorCode *status);
 749
 750 /**
 751  * Given a keyword, create a string enumeration of all values
 752  * for that keyword that are currently in use.
 753  * @param keyword a particular keyword as enumerated by
 754  * ucol_getKeywords. If any other keyword is passed in, *status is set
 755  * to U_ILLEGAL_ARGUMENT_ERROR.
 756  * @param status input-output error code
 757  * @return a string enumeration over collation keyword values, or NULL
 758  * upon error. The caller is responsible for closing the result.
 759  * @stable ICU 3.0
 760  */
 761 U_STABLE UEnumeration* U_EXPORT2
 762 ucol_getKeywordValues(const char *keyword, UErrorCode *status);
 763
 764 /**
 765  * Given a key and a locale, returns an array of string values in a preferred
 766  * order that would make a difference. These are all and only those values where
 767  * the open (creation) of the service with the locale formed from the input locale
 768  * plus input keyword and that value has different behavior than creation with the
 769  * input locale alone.
 770  * @param key           one of the keys supported by this service.  For now, only
 771  *                      "collation" is supported.
 772  * @param locale        the locale
 773  * @param commonlyUsed  if set to true it will return only commonly used values
 774  *                      with the given locale in preferred order.  Otherwise,
 775  *                      it will return all the available values for the locale.
 776  * @param status error status
 777  * @return a string enumeration over keyword values for the given key and the locale.
 778  * @stable ICU 4.2
 779  */
 780 U_STABLE UEnumeration* U_EXPORT2
 781 ucol_getKeywordValuesForLocale(const char* key,
 782                                const char* locale,
 783                                UBool commonlyUsed,
 784                                UErrorCode* status);
 785
 786 /**
 787  * Return the functionally equivalent locale for the given
 788  * requested locale, with respect to given keyword, for the
 789  * collation service.  If two locales return the same result, then
 790  * collators instantiated for these locales will behave
 791  * equivalently.  The converse is not always true; two collators
 792  * may in fact be equivalent, but return different results, due to
 793  * internal details.  The return result has no other meaning than
 794  * that stated above, and implies nothing as to the relationship
 795  * between the two locales.  This is intended for use by
 796  * applications who wish to cache collators, or otherwise reuse
 797  * collators when possible.  The functional equivalent may change
 798  * over time.  For more information, please see the <a
 799  * href="http://icu-project.org/userguide/locale.html#services">
 800  * Locales and Services</a> section of the ICU User Guide.
 801  * @param result fillin for the functionally equivalent locale
 802  * @param resultCapacity capacity of the fillin buffer
 803  * @param keyword a particular keyword as enumerated by
 804  * ucol_getKeywords.
 805  * @param locale the requested locale
 806  * @param isAvailable if non-NULL, pointer to a fillin parameter that
 807  * indicates whether the requested locale was 'available' to the
 808  * collation service. A locale is defined as 'available' if it
 809  * physically exists within the collation locale data.
 810  * @param status pointer to input-output error code
 811  * @return the actual buffer size needed for the locale.  If greater
 812  * than resultCapacity, the returned full name will be truncated and
 813  * an error code will be returned.
 814  * @stable ICU 3.0
 815  */
 816 U_STABLE int32_t U_EXPORT2
 817 ucol_getFunctionalEquivalent(char* result, int32_t resultCapacity,
 818                              const char* keyword, const char* locale,
 819                              UBool* isAvailable, UErrorCode* status);
 820
 821 /**
 822  * Get the collation rules from a UCollator.
 823  * The rules will follow the rule syntax.
 824  * @param coll The UCollator to query.
 825  * @param length
 826  * @return The collation rules.
 827  * @stable ICU 2.0
 828  */
 829 U_STABLE const UChar* U_EXPORT2
 830 ucol_getRules(    const    UCollator    *coll,
 831         int32_t            *length);
 832
 833 /** Get the short definition string for a collator. This API harvests the collator's
 834  *  locale and the attribute set and produces a string that can be used for opening
 835  *  a collator with the same properties using the ucol_openFromShortString API.
 836  *  This string will be normalized.
 837  *  The structure and the syntax of the string is defined in the "Naming collators"
 838  *  section of the users guide:
 839  *  http://icu-project.org/userguide/Collate_Concepts.html#Naming_Collators
 840  *  This API supports preflighting.
 841  *  @param coll a collator
 842  *  @param locale a locale that will appear as a collators locale in the resulting
 843  *                short string definition. If NULL, the locale will be harvested
 844  *                from the collator.
 845  *  @param buffer space to hold the resulting string
 846  *  @param capacity capacity of the buffer
 847  *  @param status for returning errors. All the preflighting errors are featured
 848  *  @return length of the resulting string
 849  *  @see ucol_openFromShortString
 850  *  @see ucol_normalizeShortDefinitionString
 851  *  @stable ICU 3.0
 852  */
 853 U_STABLE int32_t U_EXPORT2
 854 ucol_getShortDefinitionString(const UCollator *coll,
 855                               const char *locale,
 856                               char *buffer,
 857                               int32_t capacity,
 858                               UErrorCode *status);
 859
 860 /** Verifies and normalizes short definition string.
 861  *  Normalized short definition string has all the option sorted by the argument name,
 862  *  so that equivalent definition strings are the same.
 863  *  This API supports preflighting.
 864  *  @param source definition string
 865  *  @param destination space to hold the resulting string
 866  *  @param capacity capacity of the buffer
 867  *  @param parseError if not NULL, structure that will get filled with error's pre
 868  *                   and post context in case of error.
 869  *  @param status     Error code. This API will return an error if an invalid attribute
 870  *                    or attribute/value combination is specified. All the preflighting
 871  *                    errors are also featured
 872  *  @return length of the resulting normalized string.
 873  *
 874  *  @see ucol_openFromShortString
 875  *  @see ucol_getShortDefinitionString
 876  *
 877  *  @stable ICU 3.0
 878  */
 879
 880 U_STABLE int32_t U_EXPORT2
 881 ucol_normalizeShortDefinitionString(const char *source,
 882                                     char *destination,
 883                                     int32_t capacity,
 884                                     UParseError *parseError,
 885                                     UErrorCode *status);
 886
 887
 888 /**
 889  * Get a sort key for a string from a UCollator.
 890  * Sort keys may be compared using <TT>strcmp</TT>.
 891  *
 892  * Like ICU functions that write to an output buffer, the buffer contents
 893  * is undefined if the buffer capacity (resultLength parameter) is too small.
 894  * Unlike ICU functions that write a string to an output buffer,
 895  * the terminating zero byte is counted in the sort key length.
 896  * @param coll The UCollator containing the collation rules.
 897  * @param source The string to transform.
 898  * @param sourceLength The length of source, or -1 if null-terminated.
 899  * @param result A pointer to a buffer to receive the attribute.
 900  * @param resultLength The maximum size of result.
 901  * @return The size needed to fully store the sort key.
 902  *      If there was an internal error generating the sort key,
 903  *      a zero value is returned.
 904  * @see ucol_keyHashCode
 905  * @stable ICU 2.0
 906  */
 907 U_STABLE int32_t U_EXPORT2
 908 ucol_getSortKey(const    UCollator    *coll,
 909         const    UChar        *source,
 910         int32_t        sourceLength,
 911         uint8_t        *result,
 912         int32_t        resultLength);
 913
 914
 915 /** Gets the next count bytes of a sort key. Caller needs
 916  *  to preserve state array between calls and to provide
 917  *  the same type of UCharIterator set with the same string.
 918  *  The destination buffer provided must be big enough to store
 919  *  the number of requested bytes. Generated sortkey is not
 920  *  compatible with sortkeys generated using ucol_getSortKey
 921  *  API, since we don't do any compression. If uncompressed
 922  *  sortkeys are required, this API can be used.
 923  *  @param coll The UCollator containing the collation rules.
 924  *  @param iter UCharIterator containing the string we need
 925  *              the sort key to be calculated for.
 926  *  @param state Opaque state of sortkey iteration.
 927  *  @param dest Buffer to hold the resulting sortkey part
 928  *  @param count number of sort key bytes required.
 929  *  @param status error code indicator.
 930  *  @return the actual number of bytes of a sortkey. It can be
 931  *          smaller than count if we have reached the end of
 932  *          the sort key.
 933  *  @stable ICU 2.6
 934  */
 935 U_STABLE int32_t U_EXPORT2
 936 ucol_nextSortKeyPart(const UCollator *coll,
 937                      UCharIterator *iter,
 938                      uint32_t state[2],
 939                      uint8_t *dest, int32_t count,
 940                      UErrorCode *status);
 941
 942 /** enum that is taken by ucol_getBound API
 943  * See below for explanation
 944  * do not change the values assigned to the
 945  * members of this enum. Underlying code
 946  * depends on them having these numbers
 947  * @stable ICU 2.0
 948  */
 949 typedef enum {
 950   /** lower bound */
 951   UCOL_BOUND_LOWER = 0,
 952   /** upper bound that will match strings of exact size */
 953   UCOL_BOUND_UPPER = 1,
 954   /** upper bound that will match all the strings that have the same initial substring as the given string */
 955   UCOL_BOUND_UPPER_LONG = 2,
 956   UCOL_BOUND_VALUE_COUNT
 957 } UColBoundMode;
 958
 959 /**
 960  * Produce a bound for a given sortkey and a number of levels.
 961  * Return value is always the number of bytes needed, regardless of
 962  * whether the result buffer was big enough or even valid.<br>
 963  * Resulting bounds can be used to produce a range of strings that are
 964  * between upper and lower bounds. For example, if bounds are produced
 965  * for a sortkey of string "smith", strings between upper and lower
 966  * bounds with one level would include "Smith", "SMITH", "sMiTh".<br>
 967  * There are two upper bounds that can be produced. If UCOL_BOUND_UPPER
 968  * is produced, strings matched would be as above. However, if bound
 969  * produced using UCOL_BOUND_UPPER_LONG is used, the above example will
 970  * also match "Smithsonian" and similar.<br>
 971  * For more on usage, see example in cintltst/capitst.c in procedure
 972  * TestBounds.
 973  * Sort keys may be compared using <TT>strcmp</TT>.
 974  * @param source The source sortkey.
 975  * @param sourceLength The length of source, or -1 if null-terminated.
 976  *                     (If an unmodified sortkey is passed, it is always null
 977  *                      terminated).
 978  * @param boundType Type of bound required. It can be UCOL_BOUND_LOWER, which
 979  *                  produces a lower inclusive bound, UCOL_BOUND_UPPER, that
 980  *                  produces upper bound that matches strings of the same length
 981  *                  or UCOL_BOUND_UPPER_LONG that matches strings that have the
 982  *                  same starting substring as the source string.
 983  * @param noOfLevels  Number of levels required in the resulting bound (for most
 984  *                    uses, the recommended value is 1). See users guide for
 985  *                    explanation on number of levels a sortkey can have.
 986  * @param result A pointer to a buffer to receive the resulting sortkey.
 987  * @param resultLength The maximum size of result.
 988  * @param status Used for returning error code if something went wrong. If the
 989  *               number of levels requested is higher than the number of levels
 990  *               in the source key, a warning (U_SORT_KEY_TOO_SHORT_WARNING) is
 991  *               issued.
 992  * @return The size needed to fully store the bound.
 993  * @see ucol_keyHashCode
 994  * @stable ICU 2.1
 995  */
 996 U_STABLE int32_t U_EXPORT2
 997 ucol_getBound(const uint8_t       *source,
 998         int32_t             sourceLength,
 999         UColBoundMode       boundType,
1000         uint32_t            noOfLevels,
1001         uint8_t             *result,
1002         int32_t             resultLength,
1003         UErrorCode          *status);
1004
1005 /**
1006  * Gets the version information for a Collator. Version is currently
1007  * an opaque 32-bit number which depends, among other things, on major
1008  * versions of the collator tailoring and UCA.
1009  * @param coll The UCollator to query.
1010  * @param info the version # information, the result will be filled in
1011  * @stable ICU 2.0
1012  */
1013 U_STABLE void U_EXPORT2
1014 ucol_getVersion(const UCollator* coll, UVersionInfo info);
1015
1016 /**
1017  * Gets the UCA version information for a Collator. Version is the
1018  * UCA version number (3.1.1, 4.0).
1019  * @param coll The UCollator to query.
1020  * @param info the version # information, the result will be filled in
1021  * @stable ICU 2.8
1022  */
1023 U_STABLE void U_EXPORT2
1024 ucol_getUCAVersion(const UCollator* coll, UVersionInfo info);
1025
1026 /**
1027  * Merge two sort keys. The levels are merged with their corresponding counterparts
1028  * (primaries with primaries, secondaries with secondaries etc.). Between the values
1029  * from the same level a separator is inserted.
1030  * example (uncompressed):
1031  * 191B1D 01 050505 01 910505 00 and 1F2123 01 050505 01 910505 00
1032  * will be merged as
1033  * 191B1D 02 1F212301 050505 02 050505 01 910505 02 910505 00
1034  * This allows for concatenating of first and last names for sorting, among other things.
1035  * If the destination buffer is not big enough, the results are undefined.
1036  * If any of source lengths are zero or any of source pointers are NULL/undefined,
1037  * result is of size zero.
1038  * @param src1 pointer to the first sortkey
1039  * @param src1Length length of the first sortkey
1040  * @param src2 pointer to the second sortkey
1041  * @param src2Length length of the second sortkey
1042  * @param dest buffer to hold the result
1043  * @param destCapacity size of the buffer for the result
1044  * @return size of the result. If the buffer is big enough size is always
1045  *         src1Length+src2Length-1
1046  * @stable ICU 2.0
1047  */
1048 U_STABLE int32_t U_EXPORT2
1049 ucol_mergeSortkeys(const uint8_t *src1, int32_t src1Length,
1050                    const uint8_t *src2, int32_t src2Length,
1051                    uint8_t *dest, int32_t destCapacity);
1052
1053 /**
1054  * Universal attribute setter
1055  * @param coll collator which attributes are to be changed
1056  * @param attr attribute type
1057  * @param value attribute value
1058  * @param status to indicate whether the operation went on smoothly or there were errors
1059  * @see UColAttribute
1060  * @see UColAttributeValue
1061  * @see ucol_getAttribute
1062  * @stable ICU 2.0
1063  */
1064 U_STABLE void U_EXPORT2
1065 ucol_setAttribute(UCollator *coll, UColAttribute attr, UColAttributeValue value, UErrorCode *status);
1066
1067 /**
1068  * Universal attribute getter
1069  * @param coll collator which attributes are to be changed
1070  * @param attr attribute type
1071  * @return attribute value
1072  * @param status to indicate whether the operation went on smoothly or there were errors
1073  * @see UColAttribute
1074  * @see UColAttributeValue
1075  * @see ucol_setAttribute
1076  * @stable ICU 2.0
1077  */
1078 U_STABLE UColAttributeValue  U_EXPORT2
1079 ucol_getAttribute(const UCollator *coll, UColAttribute attr, UErrorCode *status);
1080
1081 /** Variable top
1082  * is a two byte primary value which causes all the codepoints with primary values that
1083  * are less or equal than the variable top to be shifted when alternate handling is set
1084  * to UCOL_SHIFTED.
1085  * Sets the variable top to a collation element value of a string supplied.
1086  * @param coll collator which variable top needs to be changed
1087  * @param varTop one or more (if contraction) UChars to which the variable top should be set
1088  * @param len length of variable top string. If -1 it is considered to be zero terminated.
1089  * @param status error code. If error code is set, the return value is undefined.
1090  *               Errors set by this function are: <br>
1091  *    U_CE_NOT_FOUND_ERROR if more than one character was passed and there is no such
1092  *    a contraction<br>
1093  *    U_PRIMARY_TOO_LONG_ERROR if the primary for the variable top has more than two bytes
1094  * @return a 32 bit value containing the value of the variable top in upper 16 bits.
1095  *         Lower 16 bits are undefined
1096  * @see ucol_getVariableTop
1097  * @see ucol_restoreVariableTop
1098  * @stable ICU 2.0
1099  */
1100 U_STABLE uint32_t U_EXPORT2
1101 ucol_setVariableTop(UCollator *coll,
1102                     const UChar *varTop, int32_t len,
1103                     UErrorCode *status);
1104
1105 /**
1106  * Gets the variable top value of a Collator.
1107  * Lower 16 bits are undefined and should be ignored.
1108  * @param coll collator which variable top needs to be retrieved
1109  * @param status error code (not changed by function). If error code is set,
1110  *               the return value is undefined.
1111  * @return the variable top value of a Collator.
1112  * @see ucol_setVariableTop
1113  * @see ucol_restoreVariableTop
1114  * @stable ICU 2.0
1115  */
1116 U_STABLE uint32_t U_EXPORT2 ucol_getVariableTop(const UCollator *coll, UErrorCode *status);
1117
1118 /**
1119  * Sets the variable top to a collation element value supplied. Variable top is
1120  * set to the upper 16 bits.
1121  * Lower 16 bits are ignored.
1122  * @param coll collator which variable top needs to be changed
1123  * @param varTop CE value, as returned by ucol_setVariableTop or ucol)getVariableTop
1124  * @param status error code (not changed by function)
1125  * @see ucol_getVariableTop
1126  * @see ucol_setVariableTop
1127  * @stable ICU 2.0
1128  */
1129 U_STABLE void U_EXPORT2
1130 ucol_restoreVariableTop(UCollator *coll, const uint32_t varTop, UErrorCode *status);
1131
1132 /**
1133  * Thread safe cloning operation. The result is a clone of a given collator.
1134  * @param coll collator to be cloned
1135  * @param stackBuffer user allocated space for the new clone.
1136  * If NULL new memory will be allocated.
1137  *  If buffer is not large enough, new memory will be allocated.
1138  *  Clients can use the U_COL_SAFECLONE_BUFFERSIZE.
1139  *  This will probably be enough to avoid memory allocations.
1140  * @param pBufferSize pointer to size of allocated space.
1141  *  If *pBufferSize == 0, a sufficient size for use in cloning will
1142  *  be returned ('pre-flighting')
1143  *  If *pBufferSize is not enough for a stack-based safe clone,
1144  *  new memory will be allocated.
1145  * @param status to indicate whether the operation went on smoothly or there were errors
1146  *    An informational status value, U_SAFECLONE_ALLOCATED_ERROR, is used if any
1147  * allocations were necessary.
1148  * @return pointer to the new clone
1149  * @see ucol_open
1150  * @see ucol_openRules
1151  * @see ucol_close
1152  * @stable ICU 2.0
1153  */
1154 U_STABLE UCollator* U_EXPORT2
1155 ucol_safeClone(const UCollator *coll,
1156                void            *stackBuffer,
1157                int32_t         *pBufferSize,
1158                UErrorCode      *status);
1159
1160 /** default memory size for the new clone. It needs to be this large for os/400 large pointers
1161  * @stable ICU 2.0
1162  */
1163 #define U_COL_SAFECLONE_BUFFERSIZE 528
1164
1165 /**
1166  * Returns current rules. Delta defines whether full rules are returned or just the tailoring.
1167  * Returns number of UChars needed to store rules. If buffer is NULL or bufferLen is not enough
1168  * to store rules, will store up to available space.
1169  * @param coll collator to get the rules from
1170  * @param delta one of UCOL_TAILORING_ONLY, UCOL_FULL_RULES.
1171  * @param buffer buffer to store the result in. If NULL, you'll get no rules.
1172  * @param bufferLen lenght of buffer to store rules in. If less then needed you'll get only the part that fits in.
1173  * @return current rules
1174  * @stable ICU 2.0
1175  */
1176 U_STABLE int32_t U_EXPORT2
1177 ucol_getRulesEx(const UCollator *coll, UColRuleOption delta, UChar *buffer, int32_t bufferLen);
1178
1179 #ifndef U_HIDE_DEPRECATED_API
1180 /**
1181  * gets the locale name of the collator. If the collator
1182  * is instantiated from the rules, then this function returns
1183  * NULL.
1184  * @param coll The UCollator for which the locale is needed
1185  * @param type You can choose between requested, valid and actual
1186  *             locale. For description see the definition of
1187  *             ULocDataLocaleType in uloc.h
1188  * @param status error code of the operation
1189  * @return real locale name from which the collation data comes.
1190  *         If the collator was instantiated from rules, returns
1191  *         NULL.
1192  * @deprecated ICU 2.8 Use ucol_getLocaleByType instead
1193  */
1194 U_DEPRECATED const char * U_EXPORT2
1195 ucol_getLocale(const UCollator *coll, ULocDataLocaleType type, UErrorCode *status);
1196 #endif  /* U_HIDE_DEPRECATED_API */
1197
1198 /**
1199  * gets the locale name of the collator. If the collator
1200  * is instantiated from the rules, then this function returns
1201  * NULL.
1202  * @param coll The UCollator for which the locale is needed
1203  * @param type You can choose between requested, valid and actual
1204  *             locale. For description see the definition of
1205  *             ULocDataLocaleType in uloc.h
1206  * @param status error code of the operation
1207  * @return real locale name from which the collation data comes.
1208  *         If the collator was instantiated from rules, returns
1209  *         NULL.
1210  * @stable ICU 2.8
1211  */
1212 U_STABLE const char * U_EXPORT2
1213 ucol_getLocaleByType(const UCollator *coll, ULocDataLocaleType type, UErrorCode *status);
1214
1215 /**
1216  * Get an Unicode set that contains all the characters and sequences tailored in
1217  * this collator. The result must be disposed of by using uset_close.
1218  * @param coll        The UCollator for which we want to get tailored chars
1219  * @param status      error code of the operation
1220  * @return a pointer to newly created USet. Must be be disposed by using uset_close
1221  * @see ucol_openRules
1222  * @see uset_close
1223  * @stable ICU 2.4
1224  */
1225 U_STABLE USet * U_EXPORT2
1226 ucol_getTailoredSet(const UCollator *coll, UErrorCode *status);
1227
1228 #ifndef U_HIDE_INTERNAL_API
1229 /**
1230  * Universal attribute getter that returns UCOL_DEFAULT if the value is default
1231  * @param coll collator which attributes are to be changed
1232  * @param attr attribute type
1233  * @return attribute value or UCOL_DEFAULT if the value is default
1234  * @param status to indicate whether the operation went on smoothly or there were errors
1235  * @see UColAttribute
1236  * @see UColAttributeValue
1237  * @see ucol_setAttribute
1238  * @internal ICU 3.0
1239  */
1240 U_INTERNAL UColAttributeValue  U_EXPORT2
1241 ucol_getAttributeOrDefault(const UCollator *coll, UColAttribute attr, UErrorCode *status);
1242
1243 /** Check whether two collators are equal. Collators are considered equal if they
1244  *  will sort strings the same. This means that both the current attributes and the
1245  *  rules must be equivalent. Currently used for RuleBasedCollator::operator==.
1246  *  @param source first collator
1247  *  @param target second collator
1248  *  @return TRUE or FALSE
1249  *  @internal ICU 3.0
1250  */
1251 U_INTERNAL UBool U_EXPORT2
1252 ucol_equals(const UCollator *source, const UCollator *target);
1253
1254 /** Calculates the set of unsafe code points, given a collator.
1255  *   A character is unsafe if you could append any character and cause the ordering to alter significantly.
1256  *   Collation sorts in normalized order, so anything that rearranges in normalization can cause this.
1257  *   Thus if you have a character like a_umlaut, and you add a lower_dot to it,
1258  *   then it normalizes to a_lower_dot + umlaut, and sorts differently.
1259  *  @param coll Collator
1260  *  @param unsafe a fill-in set to receive the unsafe points
1261  *  @param status for catching errors
1262  *  @return number of elements in the set
1263  *  @internal ICU 3.0
1264  */
1265 U_INTERNAL int32_t U_EXPORT2
1266 ucol_getUnsafeSet( const UCollator *coll,
1267                   USet *unsafe,
1268                   UErrorCode *status);
1269
1270 /** Reset UCA's static pointers. You don't want to use this, unless your static memory can go away.
1271  * @internal ICU 3.2.1
1272  */
1273 U_INTERNAL void U_EXPORT2
1274 ucol_forgetUCA(void);
1275
1276 /** Touches all resources needed for instantiating a collator from a short string definition,
1277  *  thus filling up the cache.
1278  * @param definition A short string containing a locale and a set of attributes.
1279  *                   Attributes not explicitly mentioned are left at the default
1280  *                   state for a locale.
1281  * @param parseError if not NULL, structure that will get filled with error's pre
1282  *                   and post context in case of error.
1283  * @param forceDefaults if FALSE, the settings that are the same as the collator
1284  *                   default settings will not be applied (for example, setting
1285  *                   French secondary on a French collator would not be executed).
1286  *                   If TRUE, all the settings will be applied regardless of the
1287  *                   collator default value. If the definition
1288  *                   strings are to be cached, should be set to FALSE.
1289  * @param status     Error code. Apart from regular error conditions connected to
1290  *                   instantiating collators (like out of memory or similar), this
1291  *                   API will return an error if an invalid attribute or attribute/value
1292  *                   combination is specified.
1293  * @see ucol_openFromShortString
1294  * @internal ICU 3.2.1
1295  */
1296 U_INTERNAL void U_EXPORT2
1297 ucol_prepareShortStringOpen( const char *definition,
1298                           UBool forceDefaults,
1299                           UParseError *parseError,
1300                           UErrorCode *status);
1301 #endif  /* U_HIDE_INTERNAL_API */
1302
1303 /** Creates a binary image of a collator. This binary image can be stored and
1304  *  later used to instantiate a collator using ucol_openBinary.
1305  *  This API supports preflighting.
1306  *  @param coll Collator
1307  *  @param buffer a fill-in buffer to receive the binary image
1308  *  @param capacity capacity of the destination buffer
1309  *  @param status for catching errors
1310  *  @return size of the image
1311  *  @see ucol_openBinary
1312  *  @stable ICU 3.2
1313  */
1314 U_STABLE int32_t U_EXPORT2
1315 ucol_cloneBinary(const UCollator *coll,
1316                  uint8_t *buffer, int32_t capacity,
1317                  UErrorCode *status);
1318
1319 /** Opens a collator from a collator binary image created using
1320  *  ucol_cloneBinary. Binary image used in instantiation of the
1321  *  collator remains owned by the user and should stay around for
1322  *  the lifetime of the collator. The API also takes a base collator
1323  *  which usualy should be UCA.
1324  *  @param bin binary image owned by the user and required through the
1325  *             lifetime of the collator
1326  *  @param length size of the image. If negative, the API will try to
1327  *                figure out the length of the image
1328  *  @param base fallback collator, usually UCA. Base is required to be
1329  *              present through the lifetime of the collator. Currently
1330  *              it cannot be NULL.
1331  *  @param status for catching errors
1332  *  @return newly created collator
1333  *  @see ucol_cloneBinary
1334  *  @stable ICU 3.2
1335  */
1336 U_STABLE UCollator* U_EXPORT2
1337 ucol_openBinary(const uint8_t *bin, int32_t length,
1338                 const UCollator *base,
1339                 UErrorCode *status);
1340
1341
1342 #endif /* #if !UCONFIG_NO_COLLATION */
1343
1344 #endif