icuSources/i18n/unicode/ucol.h

   1 /*
   2 *******************************************************************************
   3 * Copyright (c) 1996-2006, 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/parseerr.h"
  17 #include "unicode/uloc.h"
  18 #include "unicode/uset.h"
  19
  20 /**
  21  * \file
  22  * \brief C API: Collator
  23  *
  24  * <h2> Collator C API </h2>
  25  *
  26  * The C API for Collator performs locale-sensitive
  27  * string comparison. You use this service to build
  28  * searching and sorting routines for natural language text.
  29  * <em>Important: </em>The ICU collation service has been reimplemented
  30  * in order to achieve better performance and UCA compliance.
  31  * For details, see the
  32  * <a href="http://dev.icu-project.org/cgi-bin/viewcvs.cgi/~checkout~/icuhtml/design/collation/ICU_collation_design.htm">
  33  * collation design document</a>.
  34  * <p>
  35  * For more information about the collation service see
  36  * <a href="http://icu.sourceforge.net/userguide/Collate_Intro.html">the users guide</a>.
  37  * <p>
  38  * Collation service provides correct sorting orders for most locales supported in ICU.
  39  * If specific data for a locale is not available, the orders eventually falls back
  40  * to the <a href="http://www.unicode.org/unicode/reports/tr10/">UCA sort order</a>.
  41  * <p>
  42  * Sort ordering may be customized by providing your own set of rules. For more on
  43  * this subject see the
  44  * <a href="http://icu.sourceforge.net/userguide/Collate_Customization.html">
  45  * Collation customization</a> section of the users guide.
  46  * <p>
  47  * @see         UCollationResult
  48  * @see         UNormalizationMode
  49  * @see         UCollationStrength
  50  * @see         UCollationElements
  51  */
  52
  53 /** A collator.
  54 *  For usage in C programs.
  55 */
  56 struct UCollator;
  57 /** structure representing a collator object instance
  58  * @stable ICU 2.0
  59  */
  60 typedef struct UCollator UCollator;
  61
  62
  63 /**
  64  * UCOL_LESS is returned if source string is compared to be less than target
  65  * string in the u_strcoll() method.
  66  * UCOL_EQUAL is returned if source string is compared to be equal to target
  67  * string in the u_strcoll() method.
  68  * UCOL_GREATER is returned if source string is compared to be greater than
  69  * target string in the u_strcoll() method.
  70  * @see u_strcoll()
  71  * <p>
  72  * Possible values for a comparison result
  73  * @stable ICU 2.0
  74  */
  75 typedef enum {
  76   /** string a == string b */
  77   UCOL_EQUAL    = 0,
  78   /** string a > string b */
  79   UCOL_GREATER    = 1,
  80   /** string a < string b */
  81   UCOL_LESS    = -1
  82 } UCollationResult ;
  83
  84
  85 /** Enum containing attribute values for controling collation behavior.
  86  * Here are all the allowable values. Not every attribute can take every value. The only
  87  * universal value is UCOL_DEFAULT, which resets the attribute value to the predefined
  88  * value for that locale
  89  * @stable ICU 2.0
  90  */
  91 typedef enum {
  92   /** accepted by most attributes */
  93   UCOL_DEFAULT = -1,
  94
  95   /** Primary collation strength */
  96   UCOL_PRIMARY = 0,
  97   /** Secondary collation strength */
  98   UCOL_SECONDARY = 1,
  99   /** Tertiary collation strength */
 100   UCOL_TERTIARY = 2,
 101   /** Default collation strength */
 102   UCOL_DEFAULT_STRENGTH = UCOL_TERTIARY,
 103   UCOL_CE_STRENGTH_LIMIT,
 104   /** Quaternary collation strength */
 105   UCOL_QUATERNARY=3,
 106   /** Identical collation strength */
 107   UCOL_IDENTICAL=15,
 108   UCOL_STRENGTH_LIMIT,
 109
 110   /** Turn the feature off - works for UCOL_FRENCH_COLLATION,
 111       UCOL_CASE_LEVEL, UCOL_HIRAGANA_QUATERNARY_MODE
 112       & UCOL_DECOMPOSITION_MODE*/
 113   UCOL_OFF = 16,
 114   /** Turn the feature on - works for UCOL_FRENCH_COLLATION,
 115       UCOL_CASE_LEVEL, UCOL_HIRAGANA_QUATERNARY_MODE
 116       & UCOL_DECOMPOSITION_MODE*/
 117   UCOL_ON = 17,
 118
 119   /** Valid for UCOL_ALTERNATE_HANDLING. Alternate handling will be shifted */
 120   UCOL_SHIFTED = 20,
 121   /** Valid for UCOL_ALTERNATE_HANDLING. Alternate handling will be non ignorable */
 122   UCOL_NON_IGNORABLE = 21,
 123
 124   /** Valid for UCOL_CASE_FIRST -
 125       lower case sorts before upper case */
 126   UCOL_LOWER_FIRST = 24,
 127   /** upper case sorts before lower case */
 128   UCOL_UPPER_FIRST = 25,
 129
 130   UCOL_ATTRIBUTE_VALUE_COUNT
 131
 132 } UColAttributeValue;
 133
 134 /**
 135  * Base letter represents a primary difference.  Set comparison
 136  * level to UCOL_PRIMARY to ignore secondary and tertiary differences.
 137  * Use this to set the strength of a Collator object.
 138  * Example of primary difference, "abc" &lt; "abd"
 139  *
 140  * Diacritical differences on the same base letter represent a secondary
 141  * difference.  Set comparison level to UCOL_SECONDARY to ignore tertiary
 142  * differences. Use this to set the strength of a Collator object.
 143  * Example of secondary difference, "&auml;" >> "a".
 144  *
 145  * Uppercase and lowercase versions of the same character represents a
 146  * tertiary difference.  Set comparison level to UCOL_TERTIARY to include
 147  * all comparison differences. Use this to set the strength of a Collator
 148  * object.
 149  * Example of tertiary difference, "abc" &lt;&lt;&lt; "ABC".
 150  *
 151  * Two characters are considered "identical" when they have the same
 152  * unicode spellings.  UCOL_IDENTICAL.
 153  * For example, "&auml;" == "&auml;".
 154  *
 155  * UCollationStrength is also used to determine the strength of sort keys
 156  * generated from UCollator objects
 157  * These values can be now found in the UColAttributeValue enum.
 158  * @stable ICU 2.0
 159  **/
 160 typedef UColAttributeValue UCollationStrength;
 161
 162 /** Attributes that collation service understands. All the attributes can take UCOL_DEFAULT
 163  * value, as well as the values specific to each one.
 164  * @stable ICU 2.0
 165  */
 166 typedef enum {
 167      /** Attribute for direction of secondary weights - used in French.
 168       * Acceptable values are UCOL_ON, which results in secondary weights
 169       * being considered backwards and UCOL_OFF which treats secondary
 170       * weights in the order they appear.*/
 171      UCOL_FRENCH_COLLATION,
 172      /** Attribute for handling variable elements.
 173       * Acceptable values are UCOL_NON_IGNORABLE (default)
 174       * which treats all the codepoints with non-ignorable
 175       * primary weights in the same way,
 176       * and UCOL_SHIFTED which causes codepoints with primary
 177       * weights that are equal or below the variable top value
 178       * to be ignored on primary level and moved to the quaternary
 179       * level.*/
 180      UCOL_ALTERNATE_HANDLING,
 181      /** Controls the ordering of upper and lower case letters.
 182       * Acceptable values are UCOL_OFF (default), which orders
 183       * upper and lower case letters in accordance to their tertiary
 184       * weights, UCOL_UPPER_FIRST which forces upper case letters to
 185       * sort before lower case letters, and UCOL_LOWER_FIRST which does
 186       * the opposite. */
 187      UCOL_CASE_FIRST,
 188      /** Controls whether an extra case level (positioned before the third
 189       * level) is generated or not. Acceptable values are UCOL_OFF (default),
 190       * when case level is not generated, and UCOL_ON which causes the case
 191       * level to be generated. Contents of the case level are affected by
 192       * the value of UCOL_CASE_FIRST attribute. A simple way to ignore
 193       * accent differences in a string is to set the strength to UCOL_PRIMARY
 194       * and enable case level. */
 195      UCOL_CASE_LEVEL,
 196      /** Controls whether the normalization check and necessary normalizations
 197       * are performed. When set to UCOL_OFF (default) no normalization check
 198       * is performed. The correctness of the result is guaranteed only if the
 199       * input data is in so-called FCD form (see users manual for more info).
 200       * When set to UCOL_ON, an incremental check is performed to see whether
 201       * the input data is in the FCD form. If the data is not in the FCD form,
 202       * incremental NFD normalization is performed. */
 203      UCOL_NORMALIZATION_MODE,
 204      /** An alias for UCOL_NORMALIZATION_MODE attribute */
 205      UCOL_DECOMPOSITION_MODE = UCOL_NORMALIZATION_MODE,
 206      /** The strength attribute. Can be either UCOL_PRIMARY, UCOL_SECONDARY,
 207       * UCOL_TERTIARY, UCOL_QUATERNARY or UCOL_IDENTICAL. The usual strength
 208       * for most locales (except Japanese) is tertiary. Quaternary strength
 209       * is useful when combined with shifted setting for alternate handling
 210       * attribute and for JIS x 4061 collation, when it is used to distinguish
 211       * between Katakana  and Hiragana (this is achieved by setting the
 212       * UCOL_HIRAGANA_QUATERNARY mode to on. Otherwise, quaternary level
 213       * is affected only by the number of non ignorable code points in
 214       * the string. Identical strength is rarely useful, as it amounts
 215       * to codepoints of the NFD form of the string. */
 216      UCOL_STRENGTH,
 217      /** When turned on, this attribute positions Hiragana before all
 218       * non-ignorables on quaternary level This is a sneaky way to produce JIS
 219       * sort order */
 220      UCOL_HIRAGANA_QUATERNARY_MODE,
 221      /** When turned on, this attribute generates a collation key
 222       * for the numeric value of substrings of digits.
 223       * This is a way to get '100' to sort AFTER '2'. */
 224      UCOL_NUMERIC_COLLATION,
 225      UCOL_ATTRIBUTE_COUNT
 226 } UColAttribute;
 227
 228 /** Options for retrieving the rule string
 229  *  @stable ICU 2.0
 230  */
 231 typedef enum {
 232   /** Retrieve tailoring only */
 233   UCOL_TAILORING_ONLY,
 234   /** Retrieve UCA rules and tailoring */
 235   UCOL_FULL_RULES
 236 } UColRuleOption ;
 237
 238 /**
 239  * Open a UCollator for comparing strings.
 240  * The UCollator pointer is used in all the calls to the Collation
 241  * service. After finished, collator must be disposed of by calling
 242  * {@link #ucol_close }.
 243  * @param loc The locale containing the required collation rules.
 244  *            Special values for locales can be passed in -
 245  *            if NULL is passed for the locale, the default locale
 246  *            collation rules will be used. If empty string ("") or
 247  *            "root" are passed, UCA rules will be used.
 248  * @param status A pointer to an UErrorCode to receive any errors
 249  * @return A pointer to a UCollator, or 0 if an error occurred.
 250  * @see ucol_openRules
 251  * @see ucol_safeClone
 252  * @see ucol_close
 253  * @stable ICU 2.0
 254  */
 255 U_STABLE UCollator* U_EXPORT2
 256 ucol_open(const char *loc, UErrorCode *status);
 257
 258 /**
 259  * Produce an UCollator instance according to the rules supplied.
 260  * The rules are used to change the default ordering, defined in the
 261  * UCA in a process called tailoring. The resulting UCollator pointer
 262  * can be used in the same way as the one obtained by {@link #ucol_strcoll }.
 263  * @param rules A string describing the collation rules. For the syntax
 264  *              of the rules please see users guide.
 265  * @param rulesLength The length of rules, or -1 if null-terminated.
 266  * @param normalizationMode The normalization mode: One of
 267  *             UCOL_OFF     (expect the text to not need normalization),
 268  *             UCOL_ON      (normalize), or
 269  *             UCOL_DEFAULT (set the mode according to the rules)
 270  * @param strength The default collation strength; one of UCOL_PRIMARY, UCOL_SECONDARY,
 271  * UCOL_TERTIARY, UCOL_IDENTICAL,UCOL_DEFAULT_STRENGTH - can be also set in the rules.
 272  * @param parseError  A pointer to UParseError to recieve information about errors
 273  *                    occurred during parsing. This argument can currently be set
 274  *                    to NULL, but at users own risk. Please provide a real structure.
 275  * @param status A pointer to an UErrorCode to receive any errors
 276  * @return A pointer to a UCollator. It is not guaranteed that NULL be returned in case
 277  *         of error - please use status argument to check for errors.
 278  * @see ucol_open
 279  * @see ucol_safeClone
 280  * @see ucol_close
 281  * @stable ICU 2.0
 282  */
 283 U_STABLE UCollator* U_EXPORT2
 284 ucol_openRules( const UChar        *rules,
 285                 int32_t            rulesLength,
 286                 UColAttributeValue normalizationMode,
 287                 UCollationStrength strength,
 288                 UParseError        *parseError,
 289                 UErrorCode         *status);
 290
 291 /**
 292  * Open a collator defined by a short form string.
 293  * The structure and the syntax of the string is defined in the "Naming collators"
 294  * section of the users guide:
 295  * http://icu.sourceforge.net/userguide/Collate_Concepts.html#Naming_Collators
 296  * Attributes are overriden by the subsequent attributes. So, for "S2_S3", final
 297  * strength will be 3. 3066bis locale overrides individual locale parts.
 298  * The call to this function is equivalent to a call to ucol_open, followed by a
 299  * series of calls to ucol_setAttribute and ucol_setVariableTop.
 300  * @param definition A short string containing a locale and a set of attributes.
 301  *                   Attributes not explicitly mentioned are left at the default
 302  *                   state for a locale.
 303  * @param parseError if not NULL, structure that will get filled with error's pre
 304  *                   and post context in case of error.
 305  * @param forceDefaults if FALSE, the settings that are the same as the collator
 306  *                   default settings will not be applied (for example, setting
 307  *                   French secondary on a French collator would not be executed).
 308  *                   If TRUE, all the settings will be applied regardless of the
 309  *                   collator default value. If the definition
 310  *                   strings are to be cached, should be set to FALSE.
 311  * @param status     Error code. Apart from regular error conditions connected to
 312  *                   instantiating collators (like out of memory or similar), this
 313  *                   API will return an error if an invalid attribute or attribute/value
 314  *                   combination is specified.
 315  * @return           A pointer to a UCollator or 0 if an error occured (including an
 316  *                   invalid attribute).
 317  * @see ucol_open
 318  * @see ucol_setAttribute
 319  * @see ucol_setVariableTop
 320  * @see ucol_getShortDefinitionString
 321  * @see ucol_normalizeShortDefinitionString
 322  * @stable ICU 3.0
 323  *
 324  */
 325 U_STABLE UCollator* U_EXPORT2
 326 ucol_openFromShortString( const char *definition,
 327                           UBool forceDefaults,
 328                           UParseError *parseError,
 329                           UErrorCode *status);
 330
 331 /**
 332  * Get a set containing the contractions defined by the collator. The set includes
 333  * both the UCA contractions and the contractions defined by the collator. This set
 334  * will contain only strings. If a tailoring explicitly suppresses contractions from
 335  * the UCA (like Russian), removed contractions will not be in the resulting set.
 336  * @param coll collator
 337  * @param conts the set to hold the result. It gets emptied before
 338  *              contractions are added.
 339  * @param status to hold the error code
 340  * @return the size of the contraction set
 341  *
 342  * @deprecated ICU 3.4, use ucol_getContractionsAndExpansions instead
 343  */
 344 U_DEPRECATED int32_t U_EXPORT2
 345 ucol_getContractions( const UCollator *coll,
 346                   USet *conts,
 347                   UErrorCode *status);
 348
 349 /**
 350  * Get a set containing the expansions defined by the collator. The set includes
 351  * both the UCA expansions and the expansions defined by the tailoring
 352  * @param coll collator
 353  * @param contractions if not NULL, the set to hold the contractions
 354  * @param expansions if not NULL, the set to hold the expansions
 355  * @param addPrefixes add the prefix contextual elements to contractions
 356  * @param status to hold the error code
 357  *
 358  * @draft ICU 3.4
 359  */
 360 U_DRAFT void U_EXPORT2
 361 ucol_getContractionsAndExpansions( const UCollator *coll,
 362                   USet *contractions, USet *expansions,
 363                   UBool addPrefixes, UErrorCode *status);
 364
 365 /**
 366  * Close a UCollator.
 367  * Once closed, a UCollator should not be used. Every open collator should
 368  * be closed. Otherwise, a memory leak will result.
 369  * @param coll The UCollator to close.
 370  * @see ucol_open
 371  * @see ucol_openRules
 372  * @see ucol_safeClone
 373  * @stable ICU 2.0
 374  */
 375 U_STABLE void U_EXPORT2
 376 ucol_close(UCollator *coll);
 377
 378 /**
 379  * Compare two strings.
 380  * The strings will be compared using the options already specified.
 381  * @param coll The UCollator containing the comparison rules.
 382  * @param source The source string.
 383  * @param sourceLength The length of source, or -1 if null-terminated.
 384  * @param target The target string.
 385  * @param targetLength The length of target, or -1 if null-terminated.
 386  * @return The result of comparing the strings; one of UCOL_EQUAL,
 387  * UCOL_GREATER, UCOL_LESS
 388  * @see ucol_greater
 389  * @see ucol_greaterOrEqual
 390  * @see ucol_equal
 391  * @stable ICU 2.0
 392  */
 393 U_STABLE UCollationResult U_EXPORT2
 394 ucol_strcoll(    const    UCollator    *coll,
 395         const    UChar        *source,
 396         int32_t            sourceLength,
 397         const    UChar        *target,
 398         int32_t            targetLength);
 399
 400 /**
 401  * Determine if one string is greater than another.
 402  * This function is equivalent to {@link #ucol_strcoll } == UCOL_GREATER
 403  * @param coll The UCollator containing the comparison rules.
 404  * @param source The source string.
 405  * @param sourceLength The length of source, or -1 if null-terminated.
 406  * @param target The target string.
 407  * @param targetLength The length of target, or -1 if null-terminated.
 408  * @return TRUE if source is greater than target, FALSE otherwise.
 409  * @see ucol_strcoll
 410  * @see ucol_greaterOrEqual
 411  * @see ucol_equal
 412  * @stable ICU 2.0
 413  */
 414 U_STABLE UBool U_EXPORT2
 415 ucol_greater(const UCollator *coll,
 416              const UChar     *source, int32_t sourceLength,
 417              const UChar     *target, int32_t targetLength);
 418
 419 /**
 420  * Determine if one string is greater than or equal to another.
 421  * This function is equivalent to {@link #ucol_strcoll } != UCOL_LESS
 422  * @param coll The UCollator containing the comparison rules.
 423  * @param source The source string.
 424  * @param sourceLength The length of source, or -1 if null-terminated.
 425  * @param target The target string.
 426  * @param targetLength The length of target, or -1 if null-terminated.
 427  * @return TRUE if source is greater than or equal to target, FALSE otherwise.
 428  * @see ucol_strcoll
 429  * @see ucol_greater
 430  * @see ucol_equal
 431  * @stable ICU 2.0
 432  */
 433 U_STABLE UBool U_EXPORT2
 434 ucol_greaterOrEqual(const UCollator *coll,
 435                     const UChar     *source, int32_t sourceLength,
 436                     const UChar     *target, int32_t targetLength);
 437
 438 /**
 439  * Compare two strings for equality.
 440  * This function is equivalent to {@link #ucol_strcoll } == UCOL_EQUAL
 441  * @param coll The UCollator containing the comparison rules.
 442  * @param source The source string.
 443  * @param sourceLength The length of source, or -1 if null-terminated.
 444  * @param target The target string.
 445  * @param targetLength The length of target, or -1 if null-terminated.
 446  * @return TRUE if source is equal to target, FALSE otherwise
 447  * @see ucol_strcoll
 448  * @see ucol_greater
 449  * @see ucol_greaterOrEqual
 450  * @stable ICU 2.0
 451  */
 452 U_STABLE UBool U_EXPORT2
 453 ucol_equal(const UCollator *coll,
 454            const UChar     *source, int32_t sourceLength,
 455            const UChar     *target, int32_t targetLength);
 456
 457 /**
 458  * Compare two UTF-8 encoded trings.
 459  * The strings will be compared using the options already specified.
 460  * @param coll The UCollator containing the comparison rules.
 461  * @param sIter The source string iterator.
 462  * @param tIter The target string iterator.
 463  * @return The result of comparing the strings; one of UCOL_EQUAL,
 464  * UCOL_GREATER, UCOL_LESS
 465  * @param status A pointer to an UErrorCode to receive any errors
 466  * @see ucol_strcoll
 467  * @stable ICU 2.6
 468  */
 469 U_STABLE UCollationResult U_EXPORT2
 470 ucol_strcollIter(  const    UCollator    *coll,
 471                   UCharIterator *sIter,
 472                   UCharIterator *tIter,
 473                   UErrorCode *status);
 474
 475 /**
 476  * Get the collation strength used in a UCollator.
 477  * The strength influences how strings are compared.
 478  * @param coll The UCollator to query.
 479  * @return The collation strength; one of UCOL_PRIMARY, UCOL_SECONDARY,
 480  * UCOL_TERTIARY, UCOL_QUATERNARY, UCOL_IDENTICAL
 481  * @see ucol_setStrength
 482  * @stable ICU 2.0
 483  */
 484 U_STABLE UCollationStrength U_EXPORT2
 485 ucol_getStrength(const UCollator *coll);
 486
 487 /**
 488  * Set the collation strength used in a UCollator.
 489  * The strength influences how strings are compared.
 490  * @param coll The UCollator to set.
 491  * @param strength The desired collation strength; one of UCOL_PRIMARY,
 492  * UCOL_SECONDARY, UCOL_TERTIARY, UCOL_QUATERNARY, UCOL_IDENTICAL, UCOL_DEFAULT
 493  * @see ucol_getStrength
 494  * @stable ICU 2.0
 495  */
 496 U_STABLE void U_EXPORT2
 497 ucol_setStrength(UCollator *coll,
 498                  UCollationStrength strength);
 499
 500 /**
 501  * Get the display name for a UCollator.
 502  * The display name is suitable for presentation to a user.
 503  * @param objLoc The locale of the collator in question.
 504  * @param dispLoc The locale for display.
 505  * @param result A pointer to a buffer to receive the attribute.
 506  * @param resultLength The maximum size of result.
 507  * @param status A pointer to an UErrorCode to receive any errors
 508  * @return The total buffer size needed; if greater than resultLength,
 509  * the output was truncated.
 510  * @stable ICU 2.0
 511  */
 512 U_STABLE int32_t U_EXPORT2
 513 ucol_getDisplayName(    const    char        *objLoc,
 514             const    char        *dispLoc,
 515             UChar             *result,
 516             int32_t         resultLength,
 517             UErrorCode        *status);
 518
 519 /**
 520  * Get a locale for which collation rules are available.
 521  * A UCollator in a locale returned by this function will perform the correct
 522  * collation for the locale.
 523  * @param index The index of the desired locale.
 524  * @return A locale for which collation rules are available, or 0 if none.
 525  * @see ucol_countAvailable
 526  * @stable ICU 2.0
 527  */
 528 U_STABLE const char* U_EXPORT2
 529 ucol_getAvailable(int32_t index);
 530
 531 /**
 532  * Determine how many locales have collation rules available.
 533  * This function is most useful as determining the loop ending condition for
 534  * calls to {@link #ucol_getAvailable }.
 535  * @return The number of locales for which collation rules are available.
 536  * @see ucol_getAvailable
 537  * @stable ICU 2.0
 538  */
 539 U_STABLE int32_t U_EXPORT2
 540 ucol_countAvailable(void);
 541
 542 #if !UCONFIG_NO_SERVICE
 543 /**
 544  * Create a string enumerator of all locales for which a valid
 545  * collator may be opened.
 546  * @param status input-output error code
 547  * @return a string enumeration over locale strings. The caller is
 548  * responsible for closing the result.
 549  * @stable ICU 3.0
 550  */
 551 U_STABLE UEnumeration* U_EXPORT2
 552 ucol_openAvailableLocales(UErrorCode *status);
 553 #endif
 554
 555 /**
 556  * Create a string enumerator of all possible keywords that are relevant to
 557  * collation. At this point, the only recognized keyword for this
 558  * service is "collation".
 559  * @param status input-output error code
 560  * @return a string enumeration over locale strings. The caller is
 561  * responsible for closing the result.
 562  * @stable ICU 3.0
 563  */
 564 U_STABLE UEnumeration* U_EXPORT2
 565 ucol_getKeywords(UErrorCode *status);
 566
 567 /**
 568  * Given a keyword, create a string enumeration of all values
 569  * for that keyword that are currently in use.
 570  * @param keyword a particular keyword as enumerated by
 571  * ucol_getKeywords. If any other keyword is passed in, *status is set
 572  * to U_ILLEGAL_ARGUMENT_ERROR.
 573  * @param status input-output error code
 574  * @return a string enumeration over collation keyword values, or NULL
 575  * upon error. The caller is responsible for closing the result.
 576  * @stable ICU 3.0
 577  */
 578 U_STABLE UEnumeration* U_EXPORT2
 579 ucol_getKeywordValues(const char *keyword, UErrorCode *status);
 580
 581 /**
 582  * Return the functionally equivalent locale for the given
 583  * requested locale, with respect to given keyword, for the
 584  * collation service.  If two locales return the same result, then
 585  * collators instantiated for these locales will behave
 586  * equivalently.  The converse is not always true; two collators
 587  * may in fact be equivalent, but return different results, due to
 588  * internal details.  The return result has no other meaning than
 589  * that stated above, and implies nothing as to the relationship
 590  * between the two locales.  This is intended for use by
 591  * applications who wish to cache collators, or otherwise reuse
 592  * collators when possible.  The functional equivalent may change
 593  * over time.  For more information, please see the <a
 594  * href="http://icu.sourceforge.net/userguide/locale.html#services">
 595  * Locales and Services</a> section of the ICU User Guide.
 596  * @param result fillin for the functionally equivalent locale
 597  * @param resultCapacity capacity of the fillin buffer
 598  * @param keyword a particular keyword as enumerated by
 599  * ucol_getKeywords.
 600  * @param locale the requested locale
 601  * @param isAvailable if non-NULL, pointer to a fillin parameter that
 602  * indicates whether the requested locale was 'available' to the
 603  * collation service. A locale is defined as 'available' if it
 604  * physically exists within the collation locale data.
 605  * @param status pointer to input-output error code
 606  * @return the actual buffer size needed for the locale.  If greater
 607  * than resultCapacity, the returned full name will be truncated and
 608  * an error code will be returned.
 609  * @stable ICU 3.0
 610  */
 611 U_STABLE int32_t U_EXPORT2
 612 ucol_getFunctionalEquivalent(char* result, int32_t resultCapacity,
 613                              const char* keyword, const char* locale,
 614                              UBool* isAvailable, UErrorCode* status);
 615
 616 /**
 617  * Get the collation rules from a UCollator.
 618  * The rules will follow the rule syntax.
 619  * @param coll The UCollator to query.
 620  * @param length
 621  * @return The collation rules.
 622  * @stable ICU 2.0
 623  */
 624 U_STABLE const UChar* U_EXPORT2
 625 ucol_getRules(    const    UCollator    *coll,
 626         int32_t            *length);
 627
 628 /** Get the short definition string for a collator. This API harvests the collator's
 629  *  locale and the attribute set and produces a string that can be used for opening
 630  *  a collator with the same properties using the ucol_openFromShortString API.
 631  *  This string will be normalized.
 632  *  The structure and the syntax of the string is defined in the "Naming collators"
 633  *  section of the users guide:
 634  *  http://icu.sourceforge.net/userguide/Collate_Concepts.html#Naming_Collators
 635  *  This API supports preflighting.
 636  *  @param coll a collator
 637  *  @param locale a locale that will appear as a collators locale in the resulting
 638  *                short string definition. If NULL, the locale will be harvested
 639  *                from the collator.
 640  *  @param buffer space to hold the resulting string
 641  *  @param capacity capacity of the buffer
 642  *  @param status for returning errors. All the preflighting errors are featured
 643  *  @return length of the resulting string
 644  *  @see ucol_openFromShortString
 645  *  @see ucol_normalizeShortDefinitionString
 646  *  @stable ICU 3.0
 647  */
 648 U_STABLE int32_t U_EXPORT2
 649 ucol_getShortDefinitionString(const UCollator *coll,
 650                               const char *locale,
 651                               char *buffer,
 652                               int32_t capacity,
 653                               UErrorCode *status);
 654
 655 /** Verifies and normalizes short definition string.
 656  *  Normalized short definition string has all the option sorted by the argument name,
 657  *  so that equivalent definition strings are the same.
 658  *  This API supports preflighting.
 659  *  @param source definition string
 660  *  @param destination space to hold the resulting string
 661  *  @param capacity capacity of the buffer
 662  *  @param parseError if not NULL, structure that will get filled with error's pre
 663  *                   and post context in case of error.
 664  *  @param status     Error code. This API will return an error if an invalid attribute
 665  *                    or attribute/value combination is specified. All the preflighting
 666  *                    errors are also featured
 667  *  @return length of the resulting normalized string.
 668  *
 669  *  @see ucol_openFromShortString
 670  *  @see ucol_getShortDefinitionString
 671  *
 672  *  @stable ICU 3.0
 673  */
 674
 675 U_STABLE int32_t U_EXPORT2
 676 ucol_normalizeShortDefinitionString(const char *source,
 677                                     char *destination,
 678                                     int32_t capacity,
 679                                     UParseError *parseError,
 680                                     UErrorCode *status);
 681
 682
 683 /**
 684  * Get a sort key for a string from a UCollator.
 685  * Sort keys may be compared using <TT>strcmp</TT>.
 686  * @param coll The UCollator containing the collation rules.
 687  * @param source The string to transform.
 688  * @param sourceLength The length of source, or -1 if null-terminated.
 689  * @param result A pointer to a buffer to receive the attribute.
 690  * @param resultLength The maximum size of result.
 691  * @return The size needed to fully store the sort key..
 692  * @see ucol_keyHashCode
 693  * @stable ICU 2.0
 694  */
 695 U_STABLE int32_t U_EXPORT2
 696 ucol_getSortKey(const    UCollator    *coll,
 697         const    UChar        *source,
 698         int32_t        sourceLength,
 699         uint8_t        *result,
 700         int32_t        resultLength);
 701
 702
 703 /** Gets the next count bytes of a sort key. Caller needs
 704  *  to preserve state array between calls and to provide
 705  *  the same type of UCharIterator set with the same string.
 706  *  The destination buffer provided must be big enough to store
 707  *  the number of requested bytes. Generated sortkey is not
 708  *  compatible with sortkeys generated using ucol_getSortKey
 709  *  API, since we don't do any compression. If uncompressed
 710  *  sortkeys are required, this API can be used.
 711  *  @param coll The UCollator containing the collation rules.
 712  *  @param iter UCharIterator containing the string we need
 713  *              the sort key to be calculated for.
 714  *  @param state Opaque state of sortkey iteration.
 715  *  @param dest Buffer to hold the resulting sortkey part
 716  *  @param count number of sort key bytes required.
 717  *  @param status error code indicator.
 718  *  @return the actual number of bytes of a sortkey. It can be
 719  *          smaller than count if we have reached the end of
 720  *          the sort key.
 721  *  @stable ICU 2.6
 722  */
 723 U_STABLE int32_t U_EXPORT2
 724 ucol_nextSortKeyPart(const UCollator *coll,
 725                      UCharIterator *iter,
 726                      uint32_t state[2],
 727                      uint8_t *dest, int32_t count,
 728                      UErrorCode *status);
 729
 730 /** enum that is taken by ucol_getBound API
 731  * See below for explanation
 732  * do not change the values assigned to the
 733  * members of this enum. Underlying code
 734  * depends on them having these numbers
 735  * @stable ICU 2.0
 736  */
 737 typedef enum {
 738   /** lower bound */
 739   UCOL_BOUND_LOWER = 0,
 740   /** upper bound that will match strings of exact size */
 741   UCOL_BOUND_UPPER = 1,
 742   /** upper bound that will match all the strings that have the same initial substring as the given string */
 743   UCOL_BOUND_UPPER_LONG = 2,
 744   UCOL_BOUND_VALUE_COUNT
 745 } UColBoundMode;
 746
 747 /**
 748  * Produce a bound for a given sortkey and a number of levels.
 749  * Return value is always the number of bytes needed, regardless of
 750  * whether the result buffer was big enough or even valid.<br>
 751  * Resulting bounds can be used to produce a range of strings that are
 752  * between upper and lower bounds. For example, if bounds are produced
 753  * for a sortkey of string "smith", strings between upper and lower
 754  * bounds with one level would include "Smith", "SMITH", "sMiTh".<br>
 755  * There are two upper bounds that can be produced. If UCOL_BOUND_UPPER
 756  * is produced, strings matched would be as above. However, if bound
 757  * produced using UCOL_BOUND_UPPER_LONG is used, the above example will
 758  * also match "Smithsonian" and similar.<br>
 759  * For more on usage, see example in cintltst/capitst.c in procedure
 760  * TestBounds.
 761  * Sort keys may be compared using <TT>strcmp</TT>.
 762  * @param source The source sortkey.
 763  * @param sourceLength The length of source, or -1 if null-terminated.
 764  *                     (If an unmodified sortkey is passed, it is always null
 765  *                      terminated).
 766  * @param boundType Type of bound required. It can be UCOL_BOUND_LOWER, which
 767  *                  produces a lower inclusive bound, UCOL_BOUND_UPPER, that
 768  *                  produces upper bound that matches strings of the same length
 769  *                  or UCOL_BOUND_UPPER_LONG that matches strings that have the
 770  *                  same starting substring as the source string.
 771  * @param noOfLevels  Number of levels required in the resulting bound (for most
 772  *                    uses, the recommended value is 1). See users guide for
 773  *                    explanation on number of levels a sortkey can have.
 774  * @param result A pointer to a buffer to receive the resulting sortkey.
 775  * @param resultLength The maximum size of result.
 776  * @param status Used for returning error code if something went wrong. If the
 777  *               number of levels requested is higher than the number of levels
 778  *               in the source key, a warning (U_SORT_KEY_TOO_SHORT_WARNING) is
 779  *               issued.
 780  * @return The size needed to fully store the bound.
 781  * @see ucol_keyHashCode
 782  * @stable ICU 2.1
 783  */
 784 U_STABLE int32_t U_EXPORT2
 785 ucol_getBound(const uint8_t       *source,
 786         int32_t             sourceLength,
 787         UColBoundMode       boundType,
 788         uint32_t            noOfLevels,
 789         uint8_t             *result,
 790         int32_t             resultLength,
 791         UErrorCode          *status);
 792
 793 /**
 794  * Gets the version information for a Collator. Version is currently
 795  * an opaque 32-bit number which depends, among other things, on major
 796  * versions of the collator tailoring and UCA.
 797  * @param coll The UCollator to query.
 798  * @param info the version # information, the result will be filled in
 799  * @stable ICU 2.0
 800  */
 801 U_STABLE void U_EXPORT2
 802 ucol_getVersion(const UCollator* coll, UVersionInfo info);
 803
 804 /**
 805  * Gets the UCA version information for a Collator. Version is the
 806  * UCA version number (3.1.1, 4.0).
 807  * @param coll The UCollator to query.
 808  * @param info the version # information, the result will be filled in
 809  * @stable ICU 2.8
 810  */
 811 U_STABLE void U_EXPORT2
 812 ucol_getUCAVersion(const UCollator* coll, UVersionInfo info);
 813
 814 /**
 815  * Merge two sort keys. The levels are merged with their corresponding counterparts
 816  * (primaries with primaries, secondaries with secondaries etc.). Between the values
 817  * from the same level a separator is inserted.
 818  * example (uncompressed):
 819  * 191B1D 01 050505 01 910505 00 and 1F2123 01 050505 01 910505 00
 820  * will be merged as
 821  * 191B1D 02 1F212301 050505 02 050505 01 910505 02 910505 00
 822  * This allows for concatenating of first and last names for sorting, among other things.
 823  * If the destination buffer is not big enough, the results are undefined.
 824  * If any of source lengths are zero or any of source pointers are NULL/undefined,
 825  * result is of size zero.
 826  * @param src1 pointer to the first sortkey
 827  * @param src1Length length of the first sortkey
 828  * @param src2 pointer to the second sortkey
 829  * @param src2Length length of the second sortkey
 830  * @param dest buffer to hold the result
 831  * @param destCapacity size of the buffer for the result
 832  * @return size of the result. If the buffer is big enough size is always
 833  *         src1Length+src2Length-1
 834  * @stable ICU 2.0
 835  */
 836 U_STABLE int32_t U_EXPORT2
 837 ucol_mergeSortkeys(const uint8_t *src1, int32_t src1Length,
 838                    const uint8_t *src2, int32_t src2Length,
 839                    uint8_t *dest, int32_t destCapacity);
 840
 841 /**
 842  * Universal attribute setter
 843  * @param coll collator which attributes are to be changed
 844  * @param attr attribute type
 845  * @param value attribute value
 846  * @param status to indicate whether the operation went on smoothly or there were errors
 847  * @see UColAttribute
 848  * @see UColAttributeValue
 849  * @see ucol_getAttribute
 850  * @stable ICU 2.0
 851  */
 852 U_STABLE void U_EXPORT2
 853 ucol_setAttribute(UCollator *coll, UColAttribute attr, UColAttributeValue value, UErrorCode *status);
 854
 855 /**
 856  * Universal attribute getter
 857  * @param coll collator which attributes are to be changed
 858  * @param attr attribute type
 859  * @return attribute value
 860  * @param status to indicate whether the operation went on smoothly or there were errors
 861  * @see UColAttribute
 862  * @see UColAttributeValue
 863  * @see ucol_setAttribute
 864  * @stable ICU 2.0
 865  */
 866 U_STABLE UColAttributeValue  U_EXPORT2
 867 ucol_getAttribute(const UCollator *coll, UColAttribute attr, UErrorCode *status);
 868
 869 /** Variable top
 870  * is a two byte primary value which causes all the codepoints with primary values that
 871  * are less or equal than the variable top to be shifted when alternate handling is set
 872  * to UCOL_SHIFTED.
 873  * Sets the variable top to a collation element value of a string supplied.
 874  * @param coll collator which variable top needs to be changed
 875  * @param varTop one or more (if contraction) UChars to which the variable top should be set
 876  * @param len length of variable top string. If -1 it is considered to be zero terminated.
 877  * @param status error code. If error code is set, the return value is undefined.
 878  *               Errors set by this function are: <br>
 879  *    U_CE_NOT_FOUND_ERROR if more than one character was passed and there is no such
 880  *    a contraction<br>
 881  *    U_PRIMARY_TOO_LONG_ERROR if the primary for the variable top has more than two bytes
 882  * @return a 32 bit value containing the value of the variable top in upper 16 bits.
 883  *         Lower 16 bits are undefined
 884  * @see ucol_getVariableTop
 885  * @see ucol_restoreVariableTop
 886  * @stable ICU 2.0
 887  */
 888 U_STABLE uint32_t U_EXPORT2
 889 ucol_setVariableTop(UCollator *coll,
 890                     const UChar *varTop, int32_t len,
 891                     UErrorCode *status);
 892
 893 /**
 894  * Gets the variable top value of a Collator.
 895  * Lower 16 bits are undefined and should be ignored.
 896  * @param coll collator which variable top needs to be retrieved
 897  * @param status error code (not changed by function). If error code is set,
 898  *               the return value is undefined.
 899  * @return the variable top value of a Collator.
 900  * @see ucol_setVariableTop
 901  * @see ucol_restoreVariableTop
 902  * @stable ICU 2.0
 903  */
 904 U_STABLE uint32_t U_EXPORT2 ucol_getVariableTop(const UCollator *coll, UErrorCode *status);
 905
 906 /**
 907  * Sets the variable top to a collation element value supplied. Variable top is
 908  * set to the upper 16 bits.
 909  * Lower 16 bits are ignored.
 910  * @param coll collator which variable top needs to be changed
 911  * @param varTop CE value, as returned by ucol_setVariableTop or ucol)getVariableTop
 912  * @param status error code (not changed by function)
 913  * @see ucol_getVariableTop
 914  * @see ucol_setVariableTop
 915  * @stable ICU 2.0
 916  */
 917 U_STABLE void U_EXPORT2
 918 ucol_restoreVariableTop(UCollator *coll, const uint32_t varTop, UErrorCode *status);
 919
 920 /**
 921  * Thread safe cloning operation. The result is a clone of a given collator.
 922  * @param coll collator to be cloned
 923  * @param stackBuffer user allocated space for the new clone.
 924  * If NULL new memory will be allocated.
 925  *  If buffer is not large enough, new memory will be allocated.
 926  *  Clients can use the U_COL_SAFECLONE_BUFFERSIZE.
 927  *  This will probably be enough to avoid memory allocations.
 928  * @param pBufferSize pointer to size of allocated space.
 929  *  If *pBufferSize == 0, a sufficient size for use in cloning will
 930  *  be returned ('pre-flighting')
 931  *  If *pBufferSize is not enough for a stack-based safe clone,
 932  *  new memory will be allocated.
 933  * @param status to indicate whether the operation went on smoothly or there were errors
 934  *    An informational status value, U_SAFECLONE_ALLOCATED_ERROR, is used if any
 935  * allocations were necessary.
 936  * @return pointer to the new clone
 937  * @see ucol_open
 938  * @see ucol_openRules
 939  * @see ucol_close
 940  * @stable ICU 2.0
 941  */
 942 U_STABLE UCollator* U_EXPORT2
 943 ucol_safeClone(const UCollator *coll,
 944                void            *stackBuffer,
 945                int32_t         *pBufferSize,
 946                UErrorCode      *status);
 947
 948 /** default memory size for the new clone. It needs to be this large for os/400 large pointers
 949  * @stable ICU 2.0
 950  */
 951 #define U_COL_SAFECLONE_BUFFERSIZE 512
 952
 953 /**
 954  * Returns current rules. Delta defines whether full rules are returned or just the tailoring.
 955  * Returns number of UChars needed to store rules. If buffer is NULL or bufferLen is not enough
 956  * to store rules, will store up to available space.
 957  * @param coll collator to get the rules from
 958  * @param delta one of UCOL_TAILORING_ONLY, UCOL_FULL_RULES.
 959  * @param buffer buffer to store the result in. If NULL, you'll get no rules.
 960  * @param bufferLen lenght of buffer to store rules in. If less then needed you'll get only the part that fits in.
 961  * @return current rules
 962  * @stable ICU 2.0
 963  */
 964 U_STABLE int32_t U_EXPORT2
 965 ucol_getRulesEx(const UCollator *coll, UColRuleOption delta, UChar *buffer, int32_t bufferLen);
 966
 967 /**
 968  * gets the locale name of the collator. If the collator
 969  * is instantiated from the rules, then this function returns
 970  * NULL.
 971  * @param coll The UCollator for which the locale is needed
 972  * @param type You can choose between requested, valid and actual
 973  *             locale. For description see the definition of
 974  *             ULocDataLocaleType in uloc.h
 975  * @param status error code of the operation
 976  * @return real locale name from which the collation data comes.
 977  *         If the collator was instantiated from rules, returns
 978  *         NULL.
 979  * @deprecated ICU 2.8 Use ucol_getLocaleByType instead
 980  */
 981 U_DEPRECATED const char * U_EXPORT2
 982 ucol_getLocale(const UCollator *coll, ULocDataLocaleType type, UErrorCode *status);
 983
 984
 985 /**
 986  * gets the locale name of the collator. If the collator
 987  * is instantiated from the rules, then this function returns
 988  * NULL.
 989  * @param coll The UCollator for which the locale is needed
 990  * @param type You can choose between requested, valid and actual
 991  *             locale. For description see the definition of
 992  *             ULocDataLocaleType in uloc.h
 993  * @param status error code of the operation
 994  * @return real locale name from which the collation data comes.
 995  *         If the collator was instantiated from rules, returns
 996  *         NULL.
 997  * @stable ICU 2.8
 998  */
 999 U_STABLE const char * U_EXPORT2
1000 ucol_getLocaleByType(const UCollator *coll, ULocDataLocaleType type, UErrorCode *status);
1001
1002 /**
1003  * Get an Unicode set that contains all the characters and sequences tailored in
1004  * this collator. The result must be disposed of by using uset_close.
1005  * @param coll        The UCollator for which we want to get tailored chars
1006  * @param status      error code of the operation
1007  * @return a pointer to newly created USet. Must be be disposed by using uset_close
1008  * @see ucol_openRules
1009  * @see uset_close
1010  * @stable ICU 2.4
1011  */
1012 U_STABLE USet * U_EXPORT2
1013 ucol_getTailoredSet(const UCollator *coll, UErrorCode *status);
1014
1015 #ifndef U_HIDE_INTERNAL_API
1016 /**
1017  * Returned by ucol_collatorToIdentifier to signify that collator is
1018  * not encodable as an identifier.
1019  * @internal ICU 3.0
1020  */
1021 #define UCOL_SIT_COLLATOR_NOT_ENCODABLE 0x80000000
1022 #endif /* U_HIDE_INTERNAL_API */
1023
1024 /**
1025  * Get a 31-bit identifier given a collator.
1026  * @param coll UCollator
1027  *  @param locale a locale that will appear as a collators locale in the resulting
1028  *                short string definition. If NULL, the locale will be harvested
1029  *                from the collator.
1030  * @param status holds error messages
1031  * @return 31-bit identifier. MSB is used if the collator cannot be encoded. In that
1032  *         case UCOL_SIT_COLLATOR_NOT_ENCODABLE is returned
1033  * @see ucol_openFromIdentifier
1034  * @see ucol_identifierToShortString
1035  * @internal ICU 3.0
1036  */
1037 U_INTERNAL uint32_t U_EXPORT2
1038 ucol_collatorToIdentifier(const UCollator *coll,
1039                           const char *locale,
1040                           UErrorCode *status);
1041
1042 /**
1043  * Open a collator given a 31-bit identifier
1044  * @param identifier 31-bit identifier, encoded by calling ucol_collatorToIdentifier
1045  * @param forceDefaults if FALSE, the settings that are the same as the collator
1046  *                   default settings will not be applied (for example, setting
1047  *                   French secondary on a French collator would not be executed).
1048  *                   If TRUE, all the settings will be applied regardless of the
1049  *                   collator default value. If the definition
1050  *                   strings that can be produced from a collator instantiated by
1051  *                   calling this API are to be cached, should be set to FALSE.
1052  * @param status for returning errors
1053  * @return UCollator object
1054  * @see ucol_collatorToIdentifier
1055  * @see ucol_identifierToShortString
1056  * @internal ICU 3.0
1057  */
1058 U_INTERNAL UCollator* U_EXPORT2
1059 ucol_openFromIdentifier(uint32_t identifier,
1060                         UBool forceDefaults,
1061                         UErrorCode *status);
1062
1063
1064 /**
1065  * Calculate the short definition string given an identifier. Supports preflighting.
1066  * @param identifier 31-bit identifier, encoded by calling ucol_collatorToIdentifier
1067  * @param buffer buffer to store the result
1068  * @param capacity buffer capacity
1069  * @param forceDefaults whether the settings that are the same as the default setting
1070  *                      should be forced anyway. Setting this argument to FALSE reduces
1071  *                      the number of different configurations, but decreases performace
1072  *                      as a collator has to be instantiated.
1073  * @param status for returning errors
1074  * @return length of the short definition string
1075  * @see ucol_collatorToIdentifier
1076  * @see ucol_openFromIdentifier
1077  * @see ucol_shortStringToIdentifier
1078  * @internal ICU 3.0
1079  */
1080 U_INTERNAL int32_t U_EXPORT2
1081 ucol_identifierToShortString(uint32_t identifier,
1082                              char *buffer,
1083                              int32_t capacity,
1084                              UBool forceDefaults,
1085                              UErrorCode *status);
1086
1087 /**
1088  * Calculate the identifier given a short definition string. Supports preflighting.
1089  * @param definition short string definition
1090  * @param forceDefaults whether the settings that are the same as the default setting
1091  *                      should be forced anyway. Setting this argument to FALSE reduces
1092  *                      the number of different configurations, but decreases performace
1093  *                      as a collator has to be instantiated.
1094  * @param status for returning errors
1095  * @return identifier
1096  * @see ucol_collatorToIdentifier
1097  * @see ucol_openFromIdentifier
1098  * @see ucol_identifierToShortString
1099  * @internal ICU 3.0
1100  */
1101 U_INTERNAL uint32_t U_EXPORT2
1102 ucol_shortStringToIdentifier(const char *definition,
1103                              UBool forceDefaults,
1104                              UErrorCode *status);
1105
1106
1107
1108 /**
1109  * Universal attribute getter that returns UCOL_DEFAULT if the value is default
1110  * @param coll collator which attributes are to be changed
1111  * @param attr attribute type
1112  * @return attribute value or UCOL_DEFAULT if the value is default
1113  * @param status to indicate whether the operation went on smoothly or there were errors
1114  * @see UColAttribute
1115  * @see UColAttributeValue
1116  * @see ucol_setAttribute
1117  * @internal ICU 3.0
1118  */
1119 U_INTERNAL UColAttributeValue  U_EXPORT2
1120 ucol_getAttributeOrDefault(const UCollator *coll, UColAttribute attr, UErrorCode *status);
1121
1122 /** Check whether two collators are equal. Collators are considered equal if they
1123  *  will sort strings the same. This means that both the current attributes and the
1124  *  rules must be equivalent. Currently used for RuleBasedCollator::operator==.
1125  *  @param source first collator
1126  *  @param target second collator
1127  *  @return TRUE or FALSE
1128  *  @internal ICU 3.0
1129  */
1130 U_INTERNAL UBool U_EXPORT2
1131 ucol_equals(const UCollator *source, const UCollator *target);
1132
1133 /** Calculates the set of unsafe code points, given a collator.
1134  *   A character is unsafe if you could append any character and cause the ordering to alter significantly.
1135  *   Collation sorts in normalized order, so anything that rearranges in normalization can cause this.
1136  *   Thus if you have a character like a_umlaut, and you add a lower_dot to it,
1137  *   then it normalizes to a_lower_dot + umlaut, and sorts differently.
1138  *  @param coll Collator
1139  *  @param unsafe a fill-in set to receive the unsafe points
1140  *  @param status for catching errors
1141  *  @return number of elements in the set
1142  *  @internal ICU 3.0
1143  */
1144 U_INTERNAL int32_t U_EXPORT2
1145 ucol_getUnsafeSet( const UCollator *coll,
1146                   USet *unsafe,
1147                   UErrorCode *status);
1148
1149 /** Reset UCA's static pointers. You don't want to use this, unless your static memory can go away.
1150  * @internal ICU 3.2.1
1151  */
1152 U_INTERNAL void U_EXPORT2
1153 ucol_forgetUCA(void);
1154
1155 /** Touches all resources needed for instantiating a collator from a short string definition,
1156  *  thus filling up the cache.
1157  * @param definition A short string containing a locale and a set of attributes.
1158  *                   Attributes not explicitly mentioned are left at the default
1159  *                   state for a locale.
1160  * @param parseError if not NULL, structure that will get filled with error's pre
1161  *                   and post context in case of error.
1162  * @param forceDefaults if FALSE, the settings that are the same as the collator
1163  *                   default settings will not be applied (for example, setting
1164  *                   French secondary on a French collator would not be executed).
1165  *                   If TRUE, all the settings will be applied regardless of the
1166  *                   collator default value. If the definition
1167  *                   strings are to be cached, should be set to FALSE.
1168  * @param status     Error code. Apart from regular error conditions connected to
1169  *                   instantiating collators (like out of memory or similar), this
1170  *                   API will return an error if an invalid attribute or attribute/value
1171  *                   combination is specified.
1172  * @see ucol_openFromShortString
1173  * @internal ICU 3.2.1
1174  */
1175 U_INTERNAL void U_EXPORT2
1176 ucol_prepareShortStringOpen( const char *definition,
1177                           UBool forceDefaults,
1178                           UParseError *parseError,
1179                           UErrorCode *status);
1180
1181 /** Creates a binary image of a collator. This binary image can be stored and
1182  *  later used to instantiate a collator using ucol_openBinary.
1183  *  This API supports preflighting.
1184  *  @param coll Collator
1185  *  @param buffer a fill-in buffer to receive the binary image
1186  *  @param capacity capacity of the destination buffer
1187  *  @param status for catching errors
1188  *  @return size of the image
1189  *  @see ucol_openBinary
1190  *  @stable ICU 3.2
1191  */
1192 U_STABLE int32_t U_EXPORT2
1193 ucol_cloneBinary(const UCollator *coll,
1194                  uint8_t *buffer, int32_t capacity,
1195                  UErrorCode *status);
1196
1197 /** Opens a collator from a collator binary image created using
1198  *  ucol_cloneBinary. Binary image used in instantiation of the
1199  *  collator remains owned by the user and should stay around for
1200  *  the lifetime of the collator. The API also takes a base collator
1201  *  which usualy should be UCA.
1202  *  @param bin binary image owned by the user and required through the
1203  *             lifetime of the collator
1204  *  @param length size of the image. If negative, the API will try to
1205  *                figure out the length of the image
1206  *  @param base fallback collator, usually UCA. Base is required to be
1207  *              present through the lifetime of the collator. Currently
1208  *              it cannot be NULL.
1209  *  @param status for catching errors
1210  *  @return newly created collator
1211  *  @see ucol_cloneBinary
1212  *  @stable ICU 3.2
1213  */
1214 U_STABLE UCollator* U_EXPORT2
1215 ucol_openBinary(const uint8_t *bin, int32_t length,
1216                 const UCollator *base,
1217                 UErrorCode *status);
1218
1219
1220 #endif /* #if !UCONFIG_NO_COLLATION */
1221
1222 #endif
1223