icu/unicode/ucol.h

   1 /*
   2 *******************************************************************************
   3 * Copyright (c) 1996-2005, 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://icu.sourceforge.net/cvs/icu/~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/icu/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/icu/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 collation element iterator.
  54 *  For usage in C programs.
  55 */
  56 struct collIterate;
  57 /** structure representing collation element iterator instance
  58  * @stable ICU 2.0
  59  */
  60 typedef struct collIterate collIterate;
  61
  62 /** A collator.
  63 *  For usage in C programs.
  64 */
  65 struct UCollator;
  66 /** structure representing a collator object instance
  67  * @stable ICU 2.0
  68  */
  69 typedef struct UCollator UCollator;
  70
  71
  72 /**
  73  * UCOL_LESS is returned if source string is compared to be less than target
  74  * string in the u_strcoll() method.
  75  * UCOL_EQUAL is returned if source string is compared to be equal to target
  76  * string in the u_strcoll() method.
  77  * UCOL_GREATER is returned if source string is compared to be greater than
  78  * target string in the u_strcoll() method.
  79  * @see u_strcoll()
  80  * <p>
  81  * Possible values for a comparison result
  82  * @stable ICU 2.0
  83  */
  84 typedef enum {
  85   /** string a == string b */
  86   UCOL_EQUAL    = 0,
  87   /** string a > string b */
  88   UCOL_GREATER    = 1,
  89   /** string a < string b */
  90   UCOL_LESS    = -1
  91 } UCollationResult ;
  92
  93
  94 /** Enum containing attribute values for controling collation behavior.
  95  * Here are all the allowable values. Not every attribute can take every value. The only
  96  * universal value is UCOL_DEFAULT, which resets the attribute value to the predefined
  97  * value for that locale
  98  * @stable ICU 2.0
  99  */
 100 typedef enum {
 101   /** accepted by most attributes */
 102   UCOL_DEFAULT = -1,
 103
 104   /** Primary collation strength */
 105   UCOL_PRIMARY = 0,
 106   /** Secondary collation strength */
 107   UCOL_SECONDARY = 1,
 108   /** Tertiary collation strength */
 109   UCOL_TERTIARY = 2,
 110   /** Default collation strength */
 111   UCOL_DEFAULT_STRENGTH = UCOL_TERTIARY,
 112   UCOL_CE_STRENGTH_LIMIT,
 113   /** Quaternary collation strength */
 114   UCOL_QUATERNARY=3,
 115   /** Identical collation strength */
 116   UCOL_IDENTICAL=15,
 117   UCOL_STRENGTH_LIMIT,
 118
 119   /** Turn the feature off - works for UCOL_FRENCH_COLLATION,
 120       UCOL_CASE_LEVEL, UCOL_HIRAGANA_QUATERNARY_MODE
 121       & UCOL_DECOMPOSITION_MODE*/
 122   UCOL_OFF = 16,
 123   /** Turn the feature on - works for UCOL_FRENCH_COLLATION,
 124       UCOL_CASE_LEVEL, UCOL_HIRAGANA_QUATERNARY_MODE
 125       & UCOL_DECOMPOSITION_MODE*/
 126   UCOL_ON = 17,
 127
 128   /** Valid for UCOL_ALTERNATE_HANDLING. Alternate handling will be shifted */
 129   UCOL_SHIFTED = 20,
 130   /** Valid for UCOL_ALTERNATE_HANDLING. Alternate handling will be non ignorable */
 131   UCOL_NON_IGNORABLE = 21,
 132
 133   /** Valid for UCOL_CASE_FIRST -
 134       lower case sorts before upper case */
 135   UCOL_LOWER_FIRST = 24,
 136   /** upper case sorts before lower case */
 137   UCOL_UPPER_FIRST = 25,
 138
 139   UCOL_ATTRIBUTE_VALUE_COUNT
 140
 141 } UColAttributeValue;
 142
 143 /**
 144  * Base letter represents a primary difference.  Set comparison
 145  * level to UCOL_PRIMARY to ignore secondary and tertiary differences.
 146  * Use this to set the strength of a Collator object.
 147  * Example of primary difference, "abc" &lt; "abd"
 148  *
 149  * Diacritical differences on the same base letter represent a secondary
 150  * difference.  Set comparison level to UCOL_SECONDARY to ignore tertiary
 151  * differences. Use this to set the strength of a Collator object.
 152  * Example of secondary difference, "ä" >> "a".
 153  *
 154  * Uppercase and lowercase versions of the same character represents a
 155  * tertiary difference.  Set comparison level to UCOL_TERTIARY to include
 156  * all comparison differences. Use this to set the strength of a Collator
 157  * object.
 158  * Example of tertiary difference, "abc" &lt;&lt;&lt; "ABC".
 159  *
 160  * Two characters are considered "identical" when they have the same
 161  * unicode spellings.  UCOL_IDENTICAL.
 162  * For example, "ä" == "ä".
 163  *
 164  * UCollationStrength is also used to determine the strength of sort keys
 165  * generated from UCollator objects
 166  * These values can be now found in the UColAttributeValue enum.
 167  * @stable ICU 2.0
 168  **/
 169 typedef UColAttributeValue UCollationStrength;
 170
 171 /** Attributes that collation service understands. All the attributes can take UCOL_DEFAULT
 172  * value, as well as the values specific to each one.
 173  * @stable ICU 2.0
 174  */
 175 typedef enum {
 176      /** Attribute for direction of secondary weights - used in French.\
 177       * Acceptable values are UCOL_ON, which results in secondary weights
 178       * being considered backwards and UCOL_OFF which treats secondary
 179       * weights in the order they appear.*/
 180      UCOL_FRENCH_COLLATION,
 181      /** Attribute for handling variable elements.\
 182       * Acceptable values are UCOL_NON_IGNORABLE (default)
 183       * which treats all the codepoints with non-ignorable
 184       * primary weights in the same way,
 185       * and UCOL_SHIFTED which causes codepoints with primary
 186       * weights that are equal or below the variable top value
 187       * to be ignored on primary level and moved to the quaternary
 188       * level.*/
 189      UCOL_ALTERNATE_HANDLING,
 190      /** Controls the ordering of upper and lower case letters.\
 191       * Acceptable values are UCOL_OFF (default), which orders
 192       * upper and lower case letters in accordance to their tertiary
 193       * weights, UCOL_UPPER_FIRST which forces upper case letters to
 194       * sort before lower case letters, and UCOL_LOWER_FIRST which does
 195       * the opposite. */
 196      UCOL_CASE_FIRST,
 197      /** Controls whether an extra case level (positioned before the third
 198       * level) is generated or not.\ Acceptable values are UCOL_OFF (default),
 199       * when case level is not generated, and UCOL_ON which causes the case
 200       * level to be generated.\ Contents of the case level are affected by
 201       * the value of UCOL_CASE_FIRST attribute.\ A simple way to ignore
 202       * accent differences in a string is to set the strength to UCOL_PRIMARY
 203       * and enable case level. */
 204      UCOL_CASE_LEVEL,
 205      /** Controls whether the normalization check and necessary normalizations
 206       * are performed.\ When set to UCOL_OFF (default) no normalization check
 207       * is performed.\ The correctness of the result is guaranteed only if the
 208       * input data is in so-called FCD form (see users manual for more info).\
 209       * When set to UCOL_ON, an incremental check is performed to see whether the input data
 210       * is in the FCD form.\ If the data is not in the FCD form, incremental
 211       * NFD normalization is performed. */
 212      UCOL_NORMALIZATION_MODE,
 213      /** An alias for UCOL_NORMALIZATION_MODE attribute */
 214      UCOL_DECOMPOSITION_MODE = UCOL_NORMALIZATION_MODE,
 215      /** The strength attribute.\ Can be either UCOL_PRIMARY, UCOL_SECONDARY,
 216       * UCOL_TERTIARY, UCOL_QUATERNARY or UCOL_IDENTICAL.\ The usual strength
 217       * for most locales (except Japanese) is tertiary.\ Quaternary strength
 218       * is useful when combined with shifted setting for alternate handling
 219       * attribute and for JIS x 4061 collation, when it is used to distinguish
 220       * between Katakana  and Hiragana (this is achieved by setting the
 221       * UCOL_HIRAGANA_QUATERNARY mode to on.\ Otherwise, quaternary level
 222       * is affected only by the number of non ignorable code points in
 223       * the string.\ Identical strength is rarely useful, as it amounts
 224       * to codepoints of the NFD form of the string. */
 225      UCOL_STRENGTH,
 226      /** when turned on, this attribute
 227       * positions Hiragana before all
 228       * non-ignorables on quaternary level
 229       * This is a sneaky way to produce JIS
 230       * sort order */
 231      UCOL_HIRAGANA_QUATERNARY_MODE,
 232      /** when turned on, this attribute
 233       * generates a collation key
 234       * for the numeric value of substrings
 235       * of digits. This is a way to get '100'
 236       * to sort AFTER '2'.*/
 237      UCOL_NUMERIC_COLLATION,
 238      UCOL_ATTRIBUTE_COUNT
 239 } UColAttribute;
 240
 241 /** Options for retrieving the rule string
 242  *  @stable ICU 2.0
 243  */
 244 typedef enum {
 245   /** Retrieve tailoring only */
 246   UCOL_TAILORING_ONLY,
 247   /** Retrieve UCA rules and tailoring */
 248   UCOL_FULL_RULES
 249 } UColRuleOption ;
 250
 251 /**
 252  * Open a UCollator for comparing strings.
 253  * The UCollator pointer is used in all the calls to the Collation
 254  * service. After finished, collator must be disposed of by calling
 255  * {@link #ucol_close }.
 256  * @param loc The locale containing the required collation rules.
 257  *            Special values for locales can be passed in -
 258  *            if NULL is passed for the locale, the default locale
 259  *            collation rules will be used. If empty string ("") or
 260  *            "root" are passed, UCA rules will be used.
 261  * @param status A pointer to an UErrorCode to receive any errors
 262  * @return A pointer to a UCollator, or 0 if an error occurred.
 263  * @see ucol_openRules
 264  * @see ucol_safeClone
 265  * @see ucol_close
 266  * @stable ICU 2.0
 267  */
 268 U_STABLE UCollator* U_EXPORT2
 269 ucol_open(const char *loc, UErrorCode *status);
 270
 271 /**
 272  * Produce an UCollator instance according to the rules supplied.
 273  * The rules are used to change the default ordering, defined in the
 274  * UCA in a process called tailoring. The resulting UCollator pointer
 275  * can be used in the same way as the one obtained by {@link #ucol_strcoll }.
 276  * @param rules A string describing the collation rules. For the syntax
 277  *              of the rules please see users guide.
 278  * @param rulesLength The length of rules, or -1 if null-terminated.
 279  * @param normalizationMode The normalization mode: One of
 280  *             UCOL_OFF     (expect the text to not need normalization),
 281  *             UCOL_ON      (normalize), or
 282  *             UCOL_DEFAULT (set the mode according to the rules)
 283  * @param strength The default collation strength; one of UCOL_PRIMARY, UCOL_SECONDARY,
 284  * UCOL_TERTIARY, UCOL_IDENTICAL,UCOL_DEFAULT_STRENGTH - can be also set in the rules.
 285  * @param parseError  A pointer to UParseError to recieve information about errors
 286  *                    occurred during parsing. This argument can currently be set
 287  *                    to NULL, but at users own risk. Please provide a real structure.
 288  * @param status A pointer to an UErrorCode to receive any errors
 289  * @return A pointer to a UCollator.\ It is not guaranteed that NULL be returned in case
 290  *         of error - please use status argument to check for errors.
 291  * @see ucol_open
 292  * @see ucol_safeClone
 293  * @see ucol_close
 294  * @stable ICU 2.0
 295  */
 296 U_STABLE UCollator* U_EXPORT2
 297 ucol_openRules( const UChar        *rules,
 298                 int32_t            rulesLength,
 299                 UColAttributeValue normalizationMode,
 300                 UCollationStrength strength,
 301                 UParseError        *parseError,
 302                 UErrorCode         *status);
 303
 304 /**
 305  * Open a collator defined by a short form string.
 306  * The structure and the syntax of the string is defined in the "Naming collators"
 307  * section of the users guide:
 308  * http://icu.sourceforge.net/icu/userguide/Collate_Concepts.html#Naming_Collators
 309  * Attributes are overriden by the subsequent attributes. So, for "S2_S3", final
 310  * strength will be 3. 3066bis locale overrides individual locale parts.
 311  * The call to this function is equivalent to a call to ucol_open, followed by a
 312  * series of calls to ucol_setAttribute and ucol_setVariableTop.
 313  * @param definition A short string containing a locale and a set of attributes.
 314  *                   Attributes not explicitly mentioned are left at the default
 315  *                   state for a locale.
 316  * @param parseError if not NULL, structure that will get filled with error's pre
 317  *                   and post context in case of error.
 318  * @param forceDefaults if FALSE, the settings that are the same as the collator
 319  *                   default settings will not be applied (for example, setting
 320  *                   French secondary on a French collator would not be executed).
 321  *                   If TRUE, all the settings will be applied regardless of the
 322  *                   collator default value. If the definition
 323  *                   strings are to be cached, should be set to FALSE.
 324  * @param status     Error code. Apart from regular error conditions connected to
 325  *                   instantiating collators (like out of memory or similar), this
 326  *                   API will return an error if an invalid attribute or attribute/value
 327  *                   combination is specified.
 328  * @return           A pointer to a UCollator or 0 if an error occured (including an
 329  *                   invalid attribute).
 330  * @see ucol_open
 331  * @see ucol_setAttribute
 332  * @see ucol_setVariableTop
 333  * @see ucol_getShortDefinitionString
 334  * @see ucol_normalizeShortDefinitionString
 335  * @draft ICU 3.0
 336  *
 337  */
 338 U_CAPI UCollator* U_EXPORT2
 339 ucol_openFromShortString( const char *definition,
 340                           UBool forceDefaults,
 341                           UParseError *parseError,
 342                           UErrorCode *status);
 343
 344 /**
 345  * Get a set containing the contractions defined by the collator. The set includes
 346  * both the UCA contractions and the contractions defined by the collator. This set
 347  * will contain only strings. If a tailoring explicitly suppresses contractions from
 348  * the UCA (like Russian), removed contractions will not be in the resulting set.
 349  * @param coll collator
 350  * @param conts the set to hold the result. It gets emptied before
 351  *              contractions are added.
 352  * @param status to hold the error code
 353  * @return the size of the contraction set
 354  *
 355  * @draft ICU 3.0
 356  */
 357 U_CAPI int32_t U_EXPORT2
 358 ucol_getContractions( const UCollator *coll,
 359                   USet *conts,
 360                   UErrorCode *status);
 361
 362
 363 /**
 364  * Close a UCollator.
 365  * Once closed, a UCollator should not be used.\ Every open collator should
 366  * be closed.\ Otherwise, a memory leak will result.
 367  * @param coll The UCollator to close.
 368  * @see ucol_open
 369  * @see ucol_openRules
 370  * @see ucol_safeClone
 371  * @stable ICU 2.0
 372  */
 373 U_STABLE void U_EXPORT2
 374 ucol_close(UCollator *coll);
 375
 376 /**
 377  * Compare two strings.
 378  * The strings will be compared using the options already specified.
 379  * @param coll The UCollator containing the comparison rules.
 380  * @param source The source string.
 381  * @param sourceLength The length of source, or -1 if null-terminated.
 382  * @param target The target string.
 383  * @param targetLength The length of target, or -1 if null-terminated.
 384  * @return The result of comparing the strings; one of UCOL_EQUAL,
 385  * UCOL_GREATER, UCOL_LESS
 386  * @see ucol_greater
 387  * @see ucol_greaterOrEqual
 388  * @see ucol_equal
 389  * @stable ICU 2.0
 390  */
 391 U_STABLE UCollationResult U_EXPORT2
 392 ucol_strcoll(    const    UCollator    *coll,
 393         const    UChar        *source,
 394         int32_t            sourceLength,
 395         const    UChar        *target,
 396         int32_t            targetLength);
 397
 398 /**
 399  * Determine if one string is greater than another.
 400  * This function is equivalent to {@link #ucol_strcoll } == UCOL_GREATER
 401  * @param coll The UCollator containing the comparison rules.
 402  * @param source The source string.
 403  * @param sourceLength The length of source, or -1 if null-terminated.
 404  * @param target The target string.
 405  * @param targetLength The length of target, or -1 if null-terminated.
 406  * @return TRUE if source is greater than target, FALSE otherwise.
 407  * @see ucol_strcoll
 408  * @see ucol_greaterOrEqual
 409  * @see ucol_equal
 410  * @stable ICU 2.0
 411  */
 412 U_STABLE UBool U_EXPORT2
 413 ucol_greater(const UCollator *coll,
 414              const UChar     *source, int32_t sourceLength,
 415              const UChar     *target, int32_t targetLength);
 416
 417 /**
 418  * Determine if one string is greater than or equal to another.
 419  * This function is equivalent to {@link #ucol_strcoll } != UCOL_LESS
 420  * @param coll The UCollator containing the comparison rules.
 421  * @param source The source string.
 422  * @param sourceLength The length of source, or -1 if null-terminated.
 423  * @param target The target string.
 424  * @param targetLength The length of target, or -1 if null-terminated.
 425  * @return TRUE if source is greater than or equal to target, FALSE otherwise.
 426  * @see ucol_strcoll
 427  * @see ucol_greater
 428  * @see ucol_equal
 429  * @stable ICU 2.0
 430  */
 431 U_STABLE UBool U_EXPORT2
 432 ucol_greaterOrEqual(const UCollator *coll,
 433                     const UChar     *source, int32_t sourceLength,
 434                     const UChar     *target, int32_t targetLength);
 435
 436 /**
 437  * Compare two strings for equality.
 438  * This function is equivalent to {@link #ucol_strcoll } == UCOL_EQUAL
 439  * @param coll The UCollator containing the comparison rules.
 440  * @param source The source string.
 441  * @param sourceLength The length of source, or -1 if null-terminated.
 442  * @param target The target string.
 443  * @param targetLength The length of target, or -1 if null-terminated.
 444  * @return TRUE if source is equal to target, FALSE otherwise
 445  * @see ucol_strcoll
 446  * @see ucol_greater
 447  * @see ucol_greaterOrEqual
 448  * @stable ICU 2.0
 449  */
 450 U_STABLE UBool U_EXPORT2
 451 ucol_equal(const UCollator *coll,
 452            const UChar     *source, int32_t sourceLength,
 453            const UChar     *target, int32_t targetLength);
 454
 455 /**
 456  * Compare two UTF-8 encoded trings.
 457  * The strings will be compared using the options already specified.
 458  * @param coll The UCollator containing the comparison rules.
 459  * @param sIter The source string iterator.
 460  * @param tIter The target string iterator.
 461  * @return The result of comparing the strings; one of UCOL_EQUAL,
 462  * UCOL_GREATER, UCOL_LESS
 463  * @param status A pointer to an UErrorCode to receive any errors
 464  * @see ucol_strcoll
 465  * @stable ICU 2.6
 466  */
 467 U_STABLE UCollationResult U_EXPORT2
 468 ucol_strcollIter(  const    UCollator    *coll,
 469                   UCharIterator *sIter,
 470                   UCharIterator *tIter,
 471                   UErrorCode *status);
 472
 473 /**
 474  * Get the collation strength used in a UCollator.
 475  * The strength influences how strings are compared.
 476  * @param coll The UCollator to query.
 477  * @return The collation strength; one of UCOL_PRIMARY, UCOL_SECONDARY,
 478  * UCOL_TERTIARY, UCOL_QUATERNARY, UCOL_IDENTICAL
 479  * @see ucol_setStrength
 480  * @stable ICU 2.0
 481  */
 482 U_STABLE UCollationStrength U_EXPORT2
 483 ucol_getStrength(const UCollator *coll);
 484
 485 /**
 486  * Set the collation strength used in a UCollator.
 487  * The strength influences how strings are compared.
 488  * @param coll The UCollator to set.
 489  * @param strength The desired collation strength; one of UCOL_PRIMARY,
 490  * UCOL_SECONDARY, UCOL_TERTIARY, UCOL_QUATERNARY, UCOL_IDENTICAL, UCOL_DEFAULT
 491  * @see ucol_getStrength
 492  * @stable ICU 2.0
 493  */
 494 U_STABLE void U_EXPORT2
 495 ucol_setStrength(UCollator *coll,
 496                  UCollationStrength strength);
 497
 498 /**
 499  * Get the display name for a UCollator.
 500  * The display name is suitable for presentation to a user.
 501  * @param objLoc The locale of the collator in question.
 502  * @param dispLoc The locale for display.
 503  * @param result A pointer to a buffer to receive the attribute.
 504  * @param resultLength The maximum size of result.
 505  * @param status A pointer to an UErrorCode to receive any errors
 506  * @return The total buffer size needed; if greater than resultLength,
 507  * the output was truncated.
 508  * @stable ICU 2.0
 509  */
 510 U_STABLE int32_t U_EXPORT2
 511 ucol_getDisplayName(    const    char        *objLoc,
 512             const    char        *dispLoc,
 513             UChar             *result,
 514             int32_t         resultLength,
 515             UErrorCode        *status);
 516
 517 /**
 518  * Get a locale for which collation rules are available.
 519  * A UCollator in a locale returned by this function will perform the correct
 520  * collation for the locale.
 521  * @param index The index of the desired locale.
 522  * @return A locale for which collation rules are available, or 0 if none.
 523  * @see ucol_countAvailable
 524  * @stable ICU 2.0
 525  */
 526 U_STABLE const char* U_EXPORT2
 527 ucol_getAvailable(int32_t index);
 528
 529 /**
 530  * Determine how many locales have collation rules available.
 531  * This function is most useful as determining the loop ending condition for
 532  * calls to {@link #ucol_getAvailable }.
 533  * @return The number of locales for which collation rules are available.
 534  * @see ucol_getAvailable
 535  * @stable ICU 2.0
 536  */
 537 U_STABLE int32_t U_EXPORT2
 538 ucol_countAvailable(void);
 539
 540 #if !UCONFIG_NO_SERVICE
 541 /**
 542  * Create a string enumerator of all locales for which a valid
 543  * collator may be opened.
 544  * @param status input-output error code
 545  * @return a string enumeration over locale strings. The caller is
 546  * responsible for closing the result.
 547  * @draft ICU 3.0
 548  */
 549 U_DRAFT UEnumeration* U_EXPORT2
 550 ucol_openAvailableLocales(UErrorCode *status);
 551 #endif
 552
 553 /**
 554  * Create a string enumerator of all possible keywords that are relevant to
 555  * collation. At this point, the only recognized keyword for this
 556  * service is "collation".
 557  * @param status input-output error code
 558  * @return a string enumeration over locale strings. The caller is
 559  * responsible for closing the result.
 560  * @draft ICU 3.0
 561  */
 562 U_DRAFT UEnumeration* U_EXPORT2
 563 ucol_getKeywords(UErrorCode *status);
 564
 565 /**
 566  * Given a keyword, create a string enumeration of all values
 567  * for that keyword that are currently in use.
 568  * @param keyword a particular keyword as enumerated by
 569  * ucol_getKeywords. If any other keyword is passed in, *status is set
 570  * to U_ILLEGAL_ARGUMENT_ERROR.
 571  * @param status input-output error code
 572  * @return a string enumeration over collation keyword values, or NULL
 573  * upon error. The caller is responsible for closing the result.
 574  * @draft ICU 3.0
 575  */
 576 U_DRAFT UEnumeration* U_EXPORT2
 577 ucol_getKeywordValues(const char *keyword, UErrorCode *status);
 578
 579 /**
 580  * Return the functionally equivalent locale for the given
 581  * requested locale, with respect to given keyword, for the
 582  * collation service.  If two locales return the same result, then
 583  * collators instantiated for these locales will behave
 584  * equivalently.  The converse is not always true; two collators
 585  * may in fact be equivalent, but return different results, due to
 586  * internal details.  The return result has no other meaning than
 587  * that stated above, and implies nothing as to the relationship
 588  * between the two locales.  This is intended for use by
 589  * applications who wish to cache collators, or otherwise reuse
 590  * collators when possible.  The functional equivalent may change
 591  * over time.  For more information, please see the <a
 592  * href="http://icu.sourceforge.net/icu/userguide/locale.html#services">
 593  * Locales and Services</a> section of the ICU User Guide.
 594  * @param result fillin for the functionally equivalent locale
 595  * @param resultCapacity capacity of the fillin buffer
 596  * @param keyword a particular keyword as enumerated by
 597  * ucol_getKeywords.
 598  * @param locale the requested locale
 599  * @param isAvailable if non-NULL, pointer to a fillin parameter that
 600  * indicates whether the requested locale was 'available' to the
 601  * collation service. A locale is defined as 'available' if it
 602  * physically exists within the collation locale data.
 603  * @param status pointer to input-output error code
 604  * @return the actual buffer size needed for the locale.  If greater
 605  * than resultCapacity, the returned full name will be truncated and
 606  * an error code will be returned.
 607  * @draft ICU 3.0
 608  */
 609 U_DRAFT int32_t U_EXPORT2
 610 ucol_getFunctionalEquivalent(char* result, int32_t resultCapacity,
 611                              const char* keyword, const char* locale,
 612                              UBool* isAvailable, UErrorCode* status);
 613
 614 /**
 615  * Get the collation rules from a UCollator.
 616  * The rules will follow the rule syntax.
 617  * @param coll The UCollator to query.
 618  * @param length
 619  * @return The collation rules.
 620  * @stable ICU 2.0
 621  */
 622 U_STABLE const UChar* U_EXPORT2
 623 ucol_getRules(    const    UCollator    *coll,
 624         int32_t            *length);
 625
 626 /** Get the short definition string for a collator. This API harvests the collator's
 627  *  locale and the attribute set and produces a string that can be used for opening
 628  *  a collator with the same properties using the ucol_openFromShortString API.
 629  *  This string will be normalized.
 630  *  The structure and the syntax of the string is defined in the "Naming collators"
 631  *  section of the users guide:
 632  *  http://icu.sourceforge.net/icu/userguide/Collate_Concepts.html#Naming_Collators
 633  *  This API supports preflighting.
 634  *  @param coll a collator
 635  *  @param locale a locale that will appear as a collators locale in the resulting
 636  *                short string definition. If NULL, the locale will be harvested
 637  *                from the collator.
 638  *  @param buffer space to hold the resulting string
 639  *  @param capacity capacity of the buffer
 640  *  @param status for returning errors. All the preflighting errors are featured
 641  *  @return length of the resulting string
 642  *  @see ucol_openFromShortString
 643  *  @see ucol_normalizeShortDefinitionString
 644  *  @draft ICU 3.0
 645  */
 646 U_CAPI int32_t U_EXPORT2
 647 ucol_getShortDefinitionString(const UCollator *coll,
 648                               const char *locale,
 649                               char *buffer,
 650                               int32_t capacity,
 651                               UErrorCode *status);
 652
 653 /** Verifies and normalizes short definition string.
 654  *  Normalized short definition string has all the option sorted by the argument name,
 655  *  so that equivalent definition strings are the same.
 656  *  This API supports preflighting.
 657  *  @param source definition string
 658  *  @param destination space to hold the resulting string
 659  *  @param capacity capacity of the buffer
 660  *  @param parseError if not NULL, structure that will get filled with error's pre
 661  *                   and post context in case of error.
 662  *  @param status     Error code. This API will return an error if an invalid attribute
 663  *                    or attribute/value combination is specified. All the preflighting
 664  *                    errors are also featured
 665  *  @return length of the resulting normalized string.
 666  *
 667  *  @see ucol_openFromShortString
 668  *  @see ucol_getShortDefinitionString
 669  *
 670  *  @draft ICU 3.0
 671  */
 672
 673 U_CAPI int32_t U_EXPORT2
 674 ucol_normalizeShortDefinitionString(const char *source,
 675                                     char *destination,
 676                                     int32_t capacity,
 677                                     UParseError *parseError,
 678                                     UErrorCode *status);
 679
 680
 681 /**
 682  * Get a sort key for a string from a UCollator.
 683  * Sort keys may be compared using <TT>strcmp</TT>.
 684  * @param coll The UCollator containing the collation rules.
 685  * @param source The string to transform.
 686  * @param sourceLength The length of source, or -1 if null-terminated.
 687  * @param result A pointer to a buffer to receive the attribute.
 688  * @param resultLength The maximum size of result.
 689  * @return The size needed to fully store the sort key..
 690  * @see ucol_keyHashCode
 691  * @stable ICU 2.0
 692  */
 693 U_STABLE int32_t U_EXPORT2
 694 ucol_getSortKey(const    UCollator    *coll,
 695         const    UChar        *source,
 696         int32_t        sourceLength,
 697         uint8_t        *result,
 698         int32_t        resultLength);
 699
 700
 701 /** Gets the next count bytes of a sort key. Caller needs
 702  *  to preserve state array between calls and to provide
 703  *  the same type of UCharIterator set with the same string.
 704  *  The destination buffer provided must be big enough to store
 705  *  the number of requested bytes. Generated sortkey is not
 706  *  compatible with sortkeys generated using ucol_getSortKey
 707  *  API, since we don't do any compression. If uncompressed
 708  *  sortkeys are required, this API can be used.
 709  *  @param coll The UCollator containing the collation rules.
 710  *  @param iter UCharIterator containing the string we need
 711  *              the sort key to be calculated for.
 712  *  @param state Opaque state of sortkey iteration.
 713  *  @param dest Buffer to hold the resulting sortkey part
 714  *  @param count number of sort key bytes required.
 715  *  @param status error code indicator.
 716  *  @return the actual number of bytes of a sortkey. It can be
 717  *          smaller than count if we have reached the end of
 718  *          the sort key.
 719  *  @stable ICU 2.6
 720  */
 721 U_STABLE int32_t U_EXPORT2
 722 ucol_nextSortKeyPart(const UCollator *coll,
 723                      UCharIterator *iter,
 724                      uint32_t state[2],
 725                      uint8_t *dest, int32_t count,
 726                      UErrorCode *status);
 727
 728 /** enum that is taken by ucol_getBound API
 729  * See below for explanation
 730  * do not change the values assigned to the
 731  * members of this enum. Underlying code
 732  * depends on them having these numbers
 733  * @stable ICU 2.0
 734  */
 735 typedef enum {
 736   /** lower bound */
 737   UCOL_BOUND_LOWER = 0,
 738   /** upper bound that will match strings of exact size */
 739   UCOL_BOUND_UPPER = 1,
 740   /** upper bound that will match all the strings that have the same initial substring as the given string */
 741   UCOL_BOUND_UPPER_LONG = 2,
 742   UCOL_BOUND_VALUE_COUNT
 743 } UColBoundMode;
 744
 745 /**
 746  * Produce a bound for a given sortkey and a number of levels.
 747  * Return value is always the number of bytes needed, regardless of
 748  * whether the result buffer was big enough or even valid.<br>
 749  * Resulting bounds can be used to produce a range of strings that are
 750  * between upper and lower bounds. For example, if bounds are produced
 751  * for a sortkey of string "smith", strings between upper and lower
 752  * bounds with one level would include "Smith", "SMITH", "sMiTh".<br>
 753  * There are two upper bounds that can be produced. If UCOL_BOUND_UPPER
 754  * is produced, strings matched would be as above. However, if bound
 755  * produced using UCOL_BOUND_UPPER_LONG is used, the above example will
 756  * also match "Smithsonian" and similar.<br>
 757  * For more on usage, see example in cintltst/capitst.c in procedure
 758  * TestBounds.
 759  * Sort keys may be compared using <TT>strcmp</TT>.
 760  * @param source The source sortkey.
 761  * @param sourceLength The length of source, or -1 if null-terminated.
 762  *                     (If an unmodified sortkey is passed, it is always null
 763  *                      terminated).
 764  * @param boundType Type of bound required. It can be UCOL_BOUND_LOWER, which
 765  *                  produces a lower inclusive bound, UCOL_BOUND_UPPER, that
 766  *                  produces upper bound that matches strings of the same length
 767  *                  or UCOL_BOUND_UPPER_LONG that matches strings that have the
 768  *                  same starting substring as the source string.
 769  * @param noOfLevels  Number of levels required in the resulting bound (for most
 770  *                    uses, the recommended value is 1). See users guide for
 771  *                    explanation on number of levels a sortkey can have.
 772  * @param result A pointer to a buffer to receive the resulting sortkey.
 773  * @param resultLength The maximum size of result.
 774  * @param status Used for returning error code if something went wrong. If the
 775  *               number of levels requested is higher than the number of levels
 776  *               in the source key, a warning (U_SORT_KEY_TOO_SHORT_WARNING) is
 777  *               issued.
 778  * @return The size needed to fully store the bound.
 779  * @see ucol_keyHashCode
 780  * @stable ICU 2.1
 781  */
 782 U_STABLE int32_t U_EXPORT2
 783 ucol_getBound(const uint8_t       *source,
 784         int32_t             sourceLength,
 785         UColBoundMode       boundType,
 786         uint32_t            noOfLevels,
 787         uint8_t             *result,
 788         int32_t             resultLength,
 789         UErrorCode          *status);
 790
 791 /**
 792  * Gets the version information for a Collator. Version is currently
 793  * an opaque 32-bit number which depends, among other things, on major
 794  * versions of the collator tailoring and UCA.
 795  * @param coll The UCollator to query.
 796  * @param info the version # information, the result will be filled in
 797  * @stable ICU 2.0
 798  */
 799 U_STABLE void U_EXPORT2
 800 ucol_getVersion(const UCollator* coll, UVersionInfo info);
 801
 802 /**
 803  * Gets the UCA version information for a Collator. Version is the
 804  * UCA version number (3.1.1, 4.0).
 805  * @param coll The UCollator to query.
 806  * @param info the version # information, the result will be filled in
 807  * @draft ICU 2.8
 808  */
 809 U_DRAFT void U_EXPORT2
 810 ucol_getUCAVersion(const UCollator* coll, UVersionInfo info);
 811
 812 /**
 813  * Merge two sort keys. The levels are merged with their corresponding counterparts
 814  * (primaries with primaries, secondaries with secondaries etc.). Between the values
 815  * from the same level a separator is inserted.
 816  * example (uncompressed):
 817  * 191B1D 01 050505 01 910505 00 and 1F2123 01 050505 01 910505 00
 818  * will be merged as
 819  * 191B1D 02 1F212301 050505 02 050505 01 910505 02 910505 00
 820  * This allows for concatenating of first and last names for sorting, among other things.
 821  * If the destination buffer is not big enough, the results are undefined.
 822  * If any of source lengths are zero or any of source pointers are NULL/undefined,
 823  * result is of size zero.
 824  * @param src1 pointer to the first sortkey
 825  * @param src1Length length of the first sortkey
 826  * @param src2 pointer to the second sortkey
 827  * @param src2Length length of the second sortkey
 828  * @param dest buffer to hold the result
 829  * @param destCapacity size of the buffer for the result
 830  * @return size of the result. If the buffer is big enough size is always
 831  *         src1Length+src2Length-1
 832  * @stable ICU 2.0
 833  */
 834 U_STABLE int32_t U_EXPORT2
 835 ucol_mergeSortkeys(const uint8_t *src1, int32_t src1Length,
 836                    const uint8_t *src2, int32_t src2Length,
 837                    uint8_t *dest, int32_t destCapacity);
 838
 839 /**
 840  * Universal attribute setter
 841  * @param coll collator which attributes are to be changed
 842  * @param attr attribute type
 843  * @param value attribute value
 844  * @param status to indicate whether the operation went on smoothly or there were errors
 845  * @see UColAttribute
 846  * @see UColAttributeValue
 847  * @see ucol_getAttribute
 848  * @stable ICU 2.0
 849  */
 850 U_STABLE void U_EXPORT2
 851 ucol_setAttribute(UCollator *coll, UColAttribute attr, UColAttributeValue value, UErrorCode *status);
 852
 853 /**
 854  * Universal attribute getter
 855  * @param coll collator which attributes are to be changed
 856  * @param attr attribute type
 857  * @return attribute value
 858  * @param status to indicate whether the operation went on smoothly or there were errors
 859  * @see UColAttribute
 860  * @see UColAttributeValue
 861  * @see ucol_setAttribute
 862  * @stable ICU 2.0
 863  */
 864 U_STABLE UColAttributeValue  U_EXPORT2
 865 ucol_getAttribute(const UCollator *coll, UColAttribute attr, UErrorCode *status);
 866
 867 /** Variable top
 868  * is a two byte primary value which causes all the codepoints with primary values that
 869  * are less or equal than the variable top to be shifted when alternate handling is set
 870  * to UCOL_SHIFTED.
 871  * Sets the variable top to a collation element value of a string supplied.
 872  * @param coll collator which variable top needs to be changed
 873  * @param varTop one or more (if contraction) UChars to which the variable top should be set
 874  * @param len length of variable top string. If -1 it is considered to be zero terminated.
 875  * @param status error code. If error code is set, the return value is undefined.
 876  *               Errors set by this function are: <br>
 877  *    U_CE_NOT_FOUND_ERROR if more than one character was passed and there is no such
 878  *    a contraction<br>
 879  *    U_PRIMARY_TOO_LONG_ERROR if the primary for the variable top has more than two bytes
 880  * @return a 32 bit value containing the value of the variable top in upper 16 bits.
 881  *         Lower 16 bits are undefined
 882  * @see ucol_getVariableTop
 883  * @see ucol_restoreVariableTop
 884  * @stable ICU 2.0
 885  */
 886 U_STABLE uint32_t U_EXPORT2
 887 ucol_setVariableTop(UCollator *coll,
 888                     const UChar *varTop, int32_t len,
 889                     UErrorCode *status);
 890
 891 /**
 892  * Gets the variable top value of a Collator.
 893  * Lower 16 bits are undefined and should be ignored.
 894  * @param coll collator which variable top needs to be retrieved
 895  * @param status error code (not changed by function). If error code is set,
 896  *               the return value is undefined.
 897  * @return the variable top value of a Collator.
 898  * @see ucol_setVariableTop
 899  * @see ucol_restoreVariableTop
 900  * @stable ICU 2.0
 901  */
 902 U_STABLE uint32_t U_EXPORT2 ucol_getVariableTop(const UCollator *coll, UErrorCode *status);
 903
 904 /**
 905  * Sets the variable top to a collation element value supplied. Variable top is
 906  * set to the upper 16 bits.
 907  * Lower 16 bits are ignored.
 908  * @param coll collator which variable top needs to be changed
 909  * @param varTop CE value, as returned by ucol_setVariableTop or ucol)getVariableTop
 910  * @param status error code (not changed by function)
 911  * @see ucol_getVariableTop
 912  * @see ucol_setVariableTop
 913  * @stable ICU 2.0
 914  */
 915 U_STABLE void U_EXPORT2
 916 ucol_restoreVariableTop(UCollator *coll, const uint32_t varTop, UErrorCode *status);
 917
 918 /**
 919  * Thread safe cloning operation. The result is a clone of a given collator.
 920  * @param coll collator to be cloned
 921  * @param stackBuffer user allocated space for the new clone.
 922  * If NULL new memory will be allocated.
 923  *  If buffer is not large enough, new memory will be allocated.
 924  *  Clients can use the U_COL_SAFECLONE_BUFFERSIZE.
 925  *  This will probably be enough to avoid memory allocations.
 926  * @param pBufferSize pointer to size of allocated space.
 927  *  If *pBufferSize == 0, a sufficient size for use in cloning will
 928  *  be returned ('pre-flighting')
 929  *  If *pBufferSize is not enough for a stack-based safe clone,
 930  *  new memory will be allocated.
 931  * @param status to indicate whether the operation went on smoothly or there were errors
 932  *    An informational status value, U_SAFECLONE_ALLOCATED_ERROR, is used if any
 933  * allocations were necessary.
 934  * @return pointer to the new clone
 935  * @see ucol_open
 936  * @see ucol_openRules
 937  * @see ucol_close
 938  * @stable ICU 2.0
 939  */
 940 U_STABLE UCollator* U_EXPORT2
 941 ucol_safeClone(const UCollator *coll,
 942                void            *stackBuffer,
 943                int32_t         *pBufferSize,
 944                UErrorCode      *status);
 945
 946 /** default memory size for the new clone. It needs to be this large for os/400 large pointers
 947  * @stable ICU 2.0
 948  */
 949 #define U_COL_SAFECLONE_BUFFERSIZE 512
 950
 951 /**
 952  * Returns current rules. Delta defines whether full rules are returned or just the tailoring.
 953  * Returns number of UChars needed to store rules. If buffer is NULL or bufferLen is not enough
 954  * to store rules, will store up to available space.
 955  * @param coll collator to get the rules from
 956  * @param delta one of UCOL_TAILORING_ONLY, UCOL_FULL_RULES.
 957  * @param buffer buffer to store the result in. If NULL, you'll get no rules.
 958  * @param bufferLen lenght of buffer to store rules in. If less then needed you'll get only the part that fits in.
 959  * @return current rules
 960  * @stable ICU 2.0
 961  */
 962 U_STABLE int32_t U_EXPORT2
 963 ucol_getRulesEx(const UCollator *coll, UColRuleOption delta, UChar *buffer, int32_t bufferLen);
 964
 965 /**
 966  * gets the locale name of the collator. If the collator
 967  * is instantiated from the rules, then this function returns
 968  * NULL.
 969  * @param coll The UCollator for which the locale is needed
 970  * @param type You can choose between requested, valid and actual
 971  *             locale. For description see the definition of
 972  *             ULocDataLocaleType in uloc.h
 973  * @param status error code of the operation
 974  * @return real locale name from which the collation data comes.
 975  *         If the collator was instantiated from rules, returns
 976  *         NULL.
 977  * @deprecated ICU 2.8 Use ucol_getLocaleByType instead
 978  */
 979 U_DEPRECATED const char * U_EXPORT2
 980 ucol_getLocale(const UCollator *coll, ULocDataLocaleType type, UErrorCode *status);
 981
 982
 983 /**
 984  * gets the locale name of the collator. If the collator
 985  * is instantiated from the rules, then this function returns
 986  * NULL.
 987  * @param coll The UCollator for which the locale is needed
 988  * @param type You can choose between requested, valid and actual
 989  *             locale. For description see the definition of
 990  *             ULocDataLocaleType in uloc.h
 991  * @param status error code of the operation
 992  * @return real locale name from which the collation data comes.
 993  *         If the collator was instantiated from rules, returns
 994  *         NULL.
 995  * @draft ICU 2.8 likely to change in ICU 3.0, based on feedback
 996  */
 997 U_DRAFT const char * U_EXPORT2
 998 ucol_getLocaleByType(const UCollator *coll, ULocDataLocaleType type, UErrorCode *status);
 999
1000 /**
1001  * Get an Unicode set that contains all the characters and sequences tailored in
1002  * this collator. The result must be disposed of by using uset_close.
1003  * @param coll        The UCollator for which we want to get tailored chars
1004  * @param status      error code of the operation
1005  * @return a pointer to newly created USet. Must be be disposed by using uset_close
1006  * @see ucol_openRules
1007  * @see uset_close
1008  * @stable ICU 2.4
1009  */
1010 U_STABLE USet * U_EXPORT2
1011 ucol_getTailoredSet(const UCollator *coll, UErrorCode *status);
1012
1013 /**
1014  * Returned by ucol_collatorToIdentifier to signify that collator is
1015  * not encodable as an identifier.
1016  * @internal ICU 3.0
1017  */
1018 #define UCOL_SIT_COLLATOR_NOT_ENCODABLE 0x80000000
1019
1020 /**
1021  * Get a 31-bit identifier given a collator.
1022  * @param coll UCollator
1023  *  @param locale a locale that will appear as a collators locale in the resulting
1024  *                short string definition. If NULL, the locale will be harvested
1025  *                from the collator.
1026  * @param status holds error messages
1027  * @return 31-bit identifier. MSB is used if the collator cannot be encoded. In that
1028  *         case UCOL_SIT_COLLATOR_NOT_ENCODABLE is returned
1029  * @see ucol_openFromIdentifier
1030  * @see ucol_identifierToShortString
1031  * @internal ICU 3.0
1032  */
1033 U_INTERNAL uint32_t U_EXPORT2
1034 ucol_collatorToIdentifier(const UCollator *coll,
1035                           const char *locale,
1036                           UErrorCode *status);
1037
1038 /**
1039  * Open a collator given a 31-bit identifier
1040  * @param identifier 31-bit identifier, encoded by calling ucol_collatorToIdentifier
1041  * @param forceDefaults if FALSE, the settings that are the same as the collator
1042  *                   default settings will not be applied (for example, setting
1043  *                   French secondary on a French collator would not be executed).
1044  *                   If TRUE, all the settings will be applied regardless of the
1045  *                   collator default value. If the definition
1046  *                   strings that can be produced from a collator instantiated by
1047  *                   calling this API are to be cached, should be set to FALSE.
1048  * @param status for returning errors
1049  * @return UCollator object
1050  * @see ucol_collatorToIdentifier
1051  * @see ucol_identifierToShortString
1052  * @internal ICU 3.0
1053  */
1054 U_INTERNAL UCollator* U_EXPORT2
1055 ucol_openFromIdentifier(uint32_t identifier,
1056                         UBool forceDefaults,
1057                         UErrorCode *status);
1058
1059
1060 /**
1061  * Calculate the short definition string given an identifier. Supports preflighting.
1062  * @param identifier 31-bit identifier, encoded by calling ucol_collatorToIdentifier
1063  * @param buffer buffer to store the result
1064  * @param capacity buffer capacity
1065  * @param forceDefaults whether the settings that are the same as the default setting
1066  *                      should be forced anyway. Setting this argument to FALSE reduces
1067  *                      the number of different configurations, but decreases performace
1068  *                      as a collator has to be instantiated.
1069  * @param status for returning errors
1070  * @return length of the short definition string
1071  * @see ucol_collatorToIdentifier
1072  * @see ucol_openFromIdentifier
1073  * @see ucol_shortStringToIdentifier
1074  * @internal ICU 3.0
1075  */
1076 U_INTERNAL int32_t U_EXPORT2
1077 ucol_identifierToShortString(uint32_t identifier,
1078                              char *buffer,
1079                              int32_t capacity,
1080                              UBool forceDefaults,
1081                              UErrorCode *status);
1082
1083 /**
1084  * Calculate the identifier given a short definition string. Supports preflighting.
1085  * @param definition short string definition
1086  * @param forceDefaults whether the settings that are the same as the default setting
1087  *                      should be forced anyway. Setting this argument to FALSE reduces
1088  *                      the number of different configurations, but decreases performace
1089  *                      as a collator has to be instantiated.
1090  * @param status for returning errors
1091  * @return identifier
1092  * @see ucol_collatorToIdentifier
1093  * @see ucol_openFromIdentifier
1094  * @see ucol_identifierToShortString
1095  * @internal ICU 3.0
1096  */
1097 U_INTERNAL uint32_t U_EXPORT2
1098 ucol_shortStringToIdentifier(const char *definition,
1099                              UBool forceDefaults,
1100                              UErrorCode *status);
1101
1102
1103
1104 /**
1105  * Universal attribute getter that returns UCOL_DEFAULT if the value is default
1106  * @param coll collator which attributes are to be changed
1107  * @param attr attribute type
1108  * @return attribute value or UCOL_DEFAULT if the value is default
1109  * @param status to indicate whether the operation went on smoothly or there were errors
1110  * @see UColAttribute
1111  * @see UColAttributeValue
1112  * @see ucol_setAttribute
1113  * @internal ICU 3.0
1114  */
1115 U_INTERNAL UColAttributeValue  U_EXPORT2
1116 ucol_getAttributeOrDefault(const UCollator *coll, UColAttribute attr, UErrorCode *status);
1117
1118 /** Check whether two collators are equal. Collators are considered equal if they
1119  *  will sort strings the same. This means that both the current attributes and the
1120  *  rules must be equivalent. Currently used for RuleBasedCollator::operator==.
1121  *  @param source first collator
1122  *  @param target second collator
1123  *  @return TRUE or FALSE
1124  *  @internal ICU 3.0
1125  */
1126 U_INTERNAL UBool U_EXPORT2
1127 ucol_equals(const UCollator *source, const UCollator *target);
1128
1129 /** Calculates the set of unsafe code points, given a collator.
1130  *   A character is unsafe if you could append any character and cause the ordering to alter significantly.
1131  *   Collation sorts in normalized order, so anything that rearranges in normalization can cause this.
1132  *   Thus if you have a character like a_umlaut, and you add a lower_dot to it,
1133  *   then it normalizes to a_lower_dot + umlaut, and sorts differently.
1134  *  @param coll Collator
1135  *  @param unsafe a fill-in set to receive the unsafe points
1136  *  @param status for catching errors
1137  *  @return number of elements in the set
1138  *  @internal ICU 3.0
1139  */
1140 U_INTERNAL int32_t U_EXPORT2
1141 ucol_getUnsafeSet( const UCollator *coll,
1142                   USet *unsafe,
1143                   UErrorCode *status);
1144
1145 /** Reset UCA's static pointers. You don't want to use this, unless your static memory can go away.
1146  * @internal ICU 3.2.1
1147  */
1148 U_INTERNAL void U_EXPORT2
1149 ucol_forgetUCA(void);
1150
1151 /** Touches all resources needed for instantiating a collator from a short string definition,
1152  *  thus filling up the cache.
1153  * @param definition A short string containing a locale and a set of attributes.
1154  *                   Attributes not explicitly mentioned are left at the default
1155  *                   state for a locale.
1156  * @param parseError if not NULL, structure that will get filled with error's pre
1157  *                   and post context in case of error.
1158  * @param forceDefaults if FALSE, the settings that are the same as the collator
1159  *                   default settings will not be applied (for example, setting
1160  *                   French secondary on a French collator would not be executed).
1161  *                   If TRUE, all the settings will be applied regardless of the
1162  *                   collator default value. If the definition
1163  *                   strings are to be cached, should be set to FALSE.
1164  * @param status     Error code. Apart from regular error conditions connected to
1165  *                   instantiating collators (like out of memory or similar), this
1166  *                   API will return an error if an invalid attribute or attribute/value
1167  *                   combination is specified.
1168  * @see ucol_openFromShortString
1169  * @internal ICU 3.2.1
1170  */
1171 U_INTERNAL void U_EXPORT2
1172 ucol_prepareShortStringOpen( const char *definition,
1173                           UBool forceDefaults,
1174                           UParseError *parseError,
1175                           UErrorCode *status);
1176
1177 /** Creates a binary image of a collator. This binary image can be stored and
1178  *  later used to instantiate a collator using ucol_openBinary.
1179  *  This API supports preflighting.
1180  *  @param coll Collator
1181  *  @param buffer a fill-in buffer to receive the binary image
1182  *  @param capacity capacity of the destination buffer
1183  *  @param status for catching errors
1184  *  @return size of the image
1185  *  @see ucol_openBinary
1186  *  @draft ICU 3.2
1187  */
1188 U_DRAFT int32_t U_EXPORT2
1189 ucol_cloneBinary(const UCollator *coll,
1190                  uint8_t *buffer, int32_t capacity,
1191                  UErrorCode *status);
1192
1193 /** Opens a collator from a collator binary image created using
1194  *  ucol_cloneBinary. Binary image used in instantiation of the
1195  *  collator remains owned by the user and should stay around for
1196  *  the lifetime of the collator. The API also takes a base collator
1197  *  which usualy should be UCA.
1198  *  @param bin binary image owned by the user and required through the
1199  *             lifetime of the collator
1200  *  @param length size of the image. If negative, the API will try to
1201  *                figure out the length of the image
1202  *  @param base fallback collator, usually UCA. Base is required to be
1203  *              present through the lifetime of the collator. Currently
1204  *              it cannot be NULL.
1205  *  @param status for catching errors
1206  *  @return newly created collator
1207  *  @see ucol_cloneBinary
1208  *  @draft ICU 3.2
1209  */
1210 U_DRAFT UCollator* U_EXPORT2
1211 ucol_openBinary(const uint8_t *bin, int32_t length,
1212                 const UCollator *base,
1213                 UErrorCode *status);
1214
1215
1216 #endif /* #if !UCONFIG_NO_COLLATION */
1217
1218 #endif
1219