icuSources/i18n/unicode/ucol.h

   1 /*
   2 *******************************************************************************
   3 * Copyright (c) 1996-2003, 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://oss.software.ibm.com/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://oss.software.ibm.com/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://oss.software.ibm.com/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  * \Ref{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_CAPI 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 \Ref{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_CAPI 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  * Close a UCollator.
 306  * Once closed, a UCollator should not be used.\ Every open collator should
 307  * be closed.\ Otherwise, a memory leak will result.
 308  * @param coll The UCollator to close.
 309  * @see ucol_open
 310  * @see ucol_openRules
 311  * @see ucol_safeClone
 312  * @stable ICU 2.0
 313  */
 314 U_CAPI void U_EXPORT2
 315 ucol_close(UCollator *coll);
 316
 317 /**
 318  * Compare two strings.
 319  * The strings will be compared using the options already specified.
 320  * @param coll The UCollator containing the comparison rules.
 321  * @param source The source string.
 322  * @param sourceLength The length of source, or -1 if null-terminated.
 323  * @param target The target string.
 324  * @param targetLength The length of target, or -1 if null-terminated.
 325  * @return The result of comparing the strings; one of UCOL_EQUAL,
 326  * UCOL_GREATER, UCOL_LESS
 327  * @see ucol_greater
 328  * @see ucol_greaterOrEqual
 329  * @see ucol_equal
 330  * @stable ICU 2.0
 331  */
 332 U_CAPI UCollationResult U_EXPORT2
 333 ucol_strcoll(    const    UCollator    *coll,
 334         const    UChar        *source,
 335         int32_t            sourceLength,
 336         const    UChar        *target,
 337         int32_t            targetLength);
 338
 339 /**
 340  * Determine if one string is greater than another.
 341  * This function is equivalent to \Ref{ucol_strcoll} == UCOL_GREATER
 342  * @param coll The UCollator containing the comparison rules.
 343  * @param source The source string.
 344  * @param sourceLength The length of source, or -1 if null-terminated.
 345  * @param target The target string.
 346  * @param targetLength The length of target, or -1 if null-terminated.
 347  * @return TRUE if source is greater than target, FALSE otherwise.
 348  * @see ucol_strcoll
 349  * @see ucol_greaterOrEqual
 350  * @see ucol_equal
 351  * @stable ICU 2.0
 352  */
 353 U_CAPI UBool U_EXPORT2
 354 ucol_greater(const UCollator *coll,
 355              const UChar     *source, int32_t sourceLength,
 356              const UChar     *target, int32_t targetLength);
 357
 358 /**
 359  * Determine if one string is greater than or equal to another.
 360  * This function is equivalent to \Ref{ucol_strcoll} != UCOL_LESS
 361  * @param coll The UCollator containing the comparison rules.
 362  * @param source The source string.
 363  * @param sourceLength The length of source, or -1 if null-terminated.
 364  * @param target The target string.
 365  * @param targetLength The length of target, or -1 if null-terminated.
 366  * @return TRUE if source is greater than or equal to target, FALSE otherwise.
 367  * @see ucol_strcoll
 368  * @see ucol_greater
 369  * @see ucol_equal
 370  * @stable ICU 2.0
 371  */
 372 U_CAPI UBool U_EXPORT2
 373 ucol_greaterOrEqual(const UCollator *coll,
 374                     const UChar     *source, int32_t sourceLength,
 375                     const UChar     *target, int32_t targetLength);
 376
 377 /**
 378  * Compare two strings for equality.
 379  * This function is equivalent to \Ref{ucol_strcoll} == UCOL_EQUAL
 380  * @param coll The UCollator containing the comparison rules.
 381  * @param source The source string.
 382  * @param sourceLength The length of source, or -1 if null-terminated.
 383  * @param target The target string.
 384  * @param targetLength The length of target, or -1 if null-terminated.
 385  * @return TRUE if source is equal to target, FALSE otherwise
 386  * @see ucol_strcoll
 387  * @see ucol_greater
 388  * @see ucol_greaterOrEqual
 389  * @stable ICU 2.0
 390  */
 391 U_CAPI UBool U_EXPORT2
 392 ucol_equal(const UCollator *coll,
 393            const UChar     *source, int32_t sourceLength,
 394            const UChar     *target, int32_t targetLength);
 395
 396 /**
 397  * Compare two UTF-8 encoded trings.
 398  * The strings will be compared using the options already specified.
 399  * @param coll The UCollator containing the comparison rules.
 400  * @param sIter The source string iterator.
 401  * @param tIter The target string iterator.
 402  * @return The result of comparing the strings; one of UCOL_EQUAL,
 403  * UCOL_GREATER, UCOL_LESS
 404  * @param status A pointer to an UErrorCode to receive any errors
 405  * @see ucol_strcoll
 406  * @draft ICU 2.6
 407  */
 408 U_CAPI UCollationResult U_EXPORT2
 409 ucol_strcollIter(  const    UCollator    *coll,
 410                   UCharIterator *sIter,
 411                   UCharIterator *tIter,
 412                   UErrorCode *status);
 413
 414 /**
 415  * Get the collation strength used in a UCollator.
 416  * The strength influences how strings are compared.
 417  * @param coll The UCollator to query.
 418  * @return The collation strength; one of UCOL_PRIMARY, UCOL_SECONDARY,
 419  * UCOL_TERTIARY, UCOL_QUATERNARY, UCOL_IDENTICAL
 420  * @see ucol_setStrength
 421  * @stable ICU 2.0
 422  */
 423 U_CAPI UCollationStrength U_EXPORT2
 424 ucol_getStrength(const UCollator *coll);
 425
 426 /**
 427  * Set the collation strength used in a UCollator.
 428  * The strength influences how strings are compared.
 429  * @param coll The UCollator to set.
 430  * @param strength The desired collation strength; one of UCOL_PRIMARY,
 431  * UCOL_SECONDARY, UCOL_TERTIARY, UCOL_QUATERNARY, UCOL_IDENTICAL, UCOL_DEFAULT
 432  * @see ucol_getStrength
 433  * @stable ICU 2.0
 434  */
 435 U_CAPI void U_EXPORT2
 436 ucol_setStrength(UCollator *coll,
 437                  UCollationStrength strength);
 438
 439 /**
 440  * Get the display name for a UCollator.
 441  * The display name is suitable for presentation to a user.
 442  * @param objLoc The locale of the collator in question.
 443  * @param dispLoc The locale for display.
 444  * @param result A pointer to a buffer to receive the attribute.
 445  * @param resultLength The maximum size of result.
 446  * @param status A pointer to an UErrorCode to receive any errors
 447  * @return The total buffer size needed; if greater than resultLength,
 448  * the output was truncated.
 449  * @stable ICU 2.0
 450  */
 451 U_CAPI int32_t U_EXPORT2
 452 ucol_getDisplayName(    const    char        *objLoc,
 453             const    char        *dispLoc,
 454             UChar             *result,
 455             int32_t         resultLength,
 456             UErrorCode        *status);
 457
 458 /**
 459  * Get a locale for which collation rules are available.
 460  * A UCollator in a locale returned by this function will perform the correct
 461  * collation for the locale.
 462  * @param index The index of the desired locale.
 463  * @return A locale for which collation rules are available, or 0 if none.
 464  * @see ucol_countAvailable
 465  * @stable ICU 2.0
 466  */
 467 U_CAPI const char* U_EXPORT2
 468 ucol_getAvailable(int32_t index);
 469
 470 /**
 471  * Determine how many locales have collation rules available.
 472  * This function is most useful as determining the loop ending condition for
 473  * calls to \Ref{ucol_getAvailable}.
 474  * @return The number of locales for which collation rules are available.
 475  * @see ucol_getAvailable
 476  * @stable ICU 2.0
 477  */
 478 U_CAPI int32_t U_EXPORT2
 479 ucol_countAvailable(void);
 480
 481 /**
 482  * Get the collation rules from a UCollator.
 483  * The rules will follow the rule syntax.
 484  * @param coll The UCollator to query.
 485  * @param length
 486  * @return The collation rules.
 487  * @stable ICU 2.0
 488  */
 489 U_CAPI const UChar* U_EXPORT2
 490 ucol_getRules(    const    UCollator    *coll,
 491         int32_t            *length);
 492
 493 /**
 494  * Get a sort key for a string from a UCollator.
 495  * Sort keys may be compared using <TT>strcmp</TT>.
 496  * @param coll The UCollator containing the collation rules.
 497  * @param source The string to transform.
 498  * @param sourceLength The length of source, or -1 if null-terminated.
 499  * @param result A pointer to a buffer to receive the attribute.
 500  * @param resultLength The maximum size of result.
 501  * @return The size needed to fully store the sort key..
 502  * @see ucol_keyHashCode
 503  * @stable ICU 2.0
 504  */
 505 U_CAPI int32_t U_EXPORT2
 506 ucol_getSortKey(const    UCollator    *coll,
 507         const    UChar        *source,
 508         int32_t        sourceLength,
 509         uint8_t        *result,
 510         int32_t        resultLength);
 511
 512
 513 /** Gets the next count bytes of a sort key. Caller needs
 514  *  to preserve state array between calls and to provide
 515  *  the same type of UCharIterator set with the same string.
 516  *  The destination buffer provided must be big enough to store
 517  *  the number of requested bytes. Generated sortkey is not
 518  *  compatible with sortkeys generated using ucol_getSortKey
 519  *  API, since we don't do any compression. If uncompressed
 520  *  sortkeys are required, this API can be used.
 521  *  @param coll The UCollator containing the collation rules.
 522  *  @param iter UCharIterator containing the string we need
 523  *              the sort key to be calculated for.
 524  *  @param state Opaque state of sortkey iteration.
 525  *  @param dest Buffer to hold the resulting sortkey part
 526  *  @param count number of sort key bytes required.
 527  *  @param status error code indicator.
 528  *  @return the actual number of bytes of a sortkey. It can be
 529  *          smaller than count if we have reached the end of
 530  *          the sort key.
 531  *  @draft ICU 2.6
 532  */
 533 U_CAPI int32_t U_EXPORT2
 534 ucol_nextSortKeyPart(const UCollator *coll,
 535                      UCharIterator *iter,
 536                      uint32_t state[2],
 537                      uint8_t *dest, int32_t count,
 538                      UErrorCode *status);
 539
 540 /** enum that is taken by ucol_getBound API
 541  * See below for explanation
 542  * do not change the values assigned to the
 543  * members of this enum. Underlying code
 544  * depends on them having these numbers
 545  * @stable ICU 2.0
 546  */
 547 typedef enum {
 548   /** lower bound */
 549   UCOL_BOUND_LOWER = 0,
 550   /** upper bound that will match strings of exact size */
 551   UCOL_BOUND_UPPER = 1,
 552   /** upper bound that will match all the strings that have the same initial substring as the given string */
 553   UCOL_BOUND_UPPER_LONG = 2,
 554   UCOL_BOUND_VALUE_COUNT
 555 } UColBoundMode;
 556
 557 /**
 558  * Produce a bound for a given sortkey and a number of levels.
 559  * Return value is always the number of bytes needed, regardless of
 560  * whether the result buffer was big enough or even valid.<br>
 561  * Resulting bounds can be used to produce a range of strings that are
 562  * between upper and lower bounds. For example, if bounds are produced
 563  * for a sortkey of string "smith", strings between upper and lower
 564  * bounds with one level would include "Smith", "SMITH", "sMiTh".<br>
 565  * There are two upper bounds that can be produced. If UCOL_BOUND_UPPER
 566  * is produced, strings matched would be as above. However, if bound
 567  * produced using UCOL_BOUND_UPPER_LONG is used, the above example will
 568  * also match "Smithsonian" and similar.<br>
 569  * For more on usage, see example in cintltst/capitst.c in procedure
 570  * TestBounds.
 571  * Sort keys may be compared using <TT>strcmp</TT>.
 572  * @param source The source sortkey.
 573  * @param sourceLength The length of source, or -1 if null-terminated.
 574  *                     (If an unmodified sortkey is passed, it is always null
 575  *                      terminated).
 576  * @param boundType Type of bound required. It can be UCOL_BOUND_LOWER, which
 577  *                  produces a lower inclusive bound, UCOL_BOUND_UPPER, that
 578  *                  produces upper bound that matches strings of the same length
 579  *                  or UCOL_BOUND_UPPER_LONG that matches strings that have the
 580  *                  same starting substring as the source string.
 581  * @param noOfLevels  Number of levels required in the resulting bound (for most
 582  *                    uses, the recommended value is 1). See users guide for
 583  *                    explanation on number of levels a sortkey can have.
 584  * @param result A pointer to a buffer to receive the resulting sortkey.
 585  * @param resultLength The maximum size of result.
 586  * @param status Used for returning error code if something went wrong. If the
 587  *               number of levels requested is higher than the number of levels
 588  *               in the source key, a warning (U_SORT_KEY_TOO_SHORT_WARNING) is
 589  *               issued.
 590  * @return The size needed to fully store the bound.
 591  * @see ucol_keyHashCode
 592  * @stable ICU 2.1
 593  */
 594 U_CAPI int32_t U_EXPORT2
 595 ucol_getBound(const uint8_t       *source,
 596         int32_t             sourceLength,
 597         UColBoundMode       boundType,
 598         uint32_t            noOfLevels,
 599         uint8_t             *result,
 600         int32_t             resultLength,
 601         UErrorCode          *status);
 602
 603 /**
 604  * Gets the version information for a Collator. Version is currently
 605  * an opaque 32-bit number which depends, among other things, on major
 606  * versions of the collator tailoring and UCA.
 607  * @param coll The UCollator to query.
 608  * @param info the version # information, the result will be filled in
 609  * @stable ICU 2.0
 610  */
 611 U_CAPI void U_EXPORT2
 612 ucol_getVersion(const UCollator* coll, UVersionInfo info);
 613
 614
 615 /**
 616  * Merge two sort keys. The levels are merged with their corresponding counterparts
 617  * (primaries with primaries, secondaries with secondaries etc.). Between the values
 618  * from the same level a separator is inserted.
 619  * example (uncompressed):
 620  * 191B1D 01 050505 01 910505 00 and 1F2123 01 050505 01 910505 00
 621  * will be merged as
 622  * 191B1D 02 1F212301 050505 02 050505 01 910505 02 910505 00
 623  * This allows for concatenating of first and last names for sorting, among other things.
 624  * If the destination buffer is not big enough, the results are undefined.
 625  * If any of source lengths are zero or any of source pointers are NULL/undefined,
 626  * result is of size zero.
 627  * @param src1 pointer to the first sortkey
 628  * @param src1Length length of the first sortkey
 629  * @param src2 pointer to the second sortkey
 630  * @param src2Length length of the second sortkey
 631  * @param dest buffer to hold the result
 632  * @param destCapacity size of the buffer for the result
 633  * @return size of the result. If the buffer is big enough size is always
 634  *         src1Length+src2Length-1
 635  * @stable ICU 2.0
 636  */
 637 U_CAPI int32_t U_EXPORT2
 638 ucol_mergeSortkeys(const uint8_t *src1, int32_t src1Length,
 639                    const uint8_t *src2, int32_t src2Length,
 640                    uint8_t *dest, int32_t destCapacity);
 641
 642 /**
 643  * Universal attribute setter
 644  * @param coll collator which attributes are to be changed
 645  * @param attr attribute type
 646  * @param value attribute value
 647  * @param status to indicate whether the operation went on smoothly or there were errors
 648  * @see UColAttribute
 649  * @see UColAttributeValue
 650  * @see ucol_getAttribute
 651  * @stable ICU 2.0
 652  */
 653 U_CAPI void U_EXPORT2
 654 ucol_setAttribute(UCollator *coll, UColAttribute attr, UColAttributeValue value, UErrorCode *status);
 655
 656 /**
 657  * Universal attribute getter
 658  * @param coll collator which attributes are to be changed
 659  * @param attr attribute type
 660  * @return attribute value
 661  * @param status to indicate whether the operation went on smoothly or there were errors
 662  * @see UColAttribute
 663  * @see UColAttributeValue
 664  * @see ucol_setAttribute
 665  * @stable ICU 2.0
 666  */
 667 U_CAPI UColAttributeValue  U_EXPORT2
 668 ucol_getAttribute(const UCollator *coll, UColAttribute attr, UErrorCode *status);
 669
 670 /** Variable top
 671  * is a two byte primary value which causes all the codepoints with primary values that
 672  * are less or equal than the variable top to be shifted when alternate handling is set
 673  * to UCOL_SHIFTED.
 674  * Sets the variable top to a collation element value of a string supplied.
 675  * @param coll collator which variable top needs to be changed
 676  * @param varTop one or more (if contraction) UChars to which the variable top should be set
 677  * @param len length of variable top string. If -1 it is considered to be zero terminated.
 678  * @param status error code. If error code is set, the return value is undefined.
 679  *               Errors set by this function are: <br>
 680  *    U_CE_NOT_FOUND_ERROR if more than one character was passed and there is no such
 681  *    a contraction<br>
 682  *    U_PRIMARY_TOO_LONG_ERROR if the primary for the variable top has more than two bytes
 683  * @return a 32 bit value containing the value of the variable top in upper 16 bits.
 684  *         Lower 16 bits are undefined
 685  * @see ucol_getVariableTop
 686  * @see ucol_restoreVariableTop
 687  * @stable ICU 2.0
 688  */
 689 U_CAPI uint32_t U_EXPORT2
 690 ucol_setVariableTop(UCollator *coll,
 691                     const UChar *varTop, int32_t len,
 692                     UErrorCode *status);
 693
 694 /**
 695  * Gets the variable top value of a Collator.
 696  * Lower 16 bits are undefined and should be ignored.
 697  * @param coll collator which variable top needs to be retrieved
 698  * @param status error code (not changed by function). If error code is set,
 699  *               the return value is undefined.
 700  * @return the variable top value of a Collator.
 701  * @see ucol_setVariableTop
 702  * @see ucol_restoreVariableTop
 703  * @stable ICU 2.0
 704  */
 705 U_CAPI uint32_t U_EXPORT2 ucol_getVariableTop(const UCollator *coll, UErrorCode *status);
 706
 707 /**
 708  * Sets the variable top to a collation element value supplied. Variable top is
 709  * set to the upper 16 bits.
 710  * Lower 16 bits are ignored.
 711  * @param coll collator which variable top needs to be changed
 712  * @param varTop CE value, as returned by ucol_setVariableTop or ucol)getVariableTop
 713  * @param status error code (not changed by function)
 714  * @see ucol_getVariableTop
 715  * @see ucol_setVariableTop
 716  * @stable ICU 2.0
 717  */
 718 U_CAPI void U_EXPORT2
 719 ucol_restoreVariableTop(UCollator *coll, const uint32_t varTop, UErrorCode *status);
 720
 721 /**
 722  * Thread safe cloning operation. The result is a clone of a given collator.
 723  * @param coll collator to be cloned
 724  * @param stackBuffer user allocated space for the new clone.
 725  * If NULL new memory will be allocated.
 726  *  If buffer is not large enough, new memory will be allocated.
 727  *  Clients can use the U_COL_SAFECLONE_BUFFERSIZE.
 728  *  This will probably be enough to avoid memory allocations.
 729  * @param pBufferSize pointer to size of allocated space.
 730  *  If *pBufferSize == 0, a sufficient size for use in cloning will
 731  *  be returned ('pre-flighting')
 732  *  If *pBufferSize is not enough for a stack-based safe clone,
 733  *  new memory will be allocated.
 734  * @param status to indicate whether the operation went on smoothly or there were errors
 735  *    An informational status value, U_SAFECLONE_ALLOCATED_ERROR, is used if any
 736  * allocations were necessary.
 737  * @return pointer to the new clone
 738  * @see ucol_open
 739  * @see ucol_openRules
 740  * @see ucol_close
 741  * @stable ICU 2.0
 742  */
 743 U_CAPI UCollator* U_EXPORT2
 744 ucol_safeClone(const UCollator *coll,
 745                void            *stackBuffer,
 746                int32_t         *pBufferSize,
 747                UErrorCode      *status);
 748
 749 /** default memory size for the new clone. It needs to be this large for os/400 large pointers
 750  * @stable ICU 2.0
 751  */
 752 #define U_COL_SAFECLONE_BUFFERSIZE 512
 753
 754 /**
 755  * Returns current rules. Delta defines whether full rules are returned or just the tailoring.
 756  * Returns number of UChars needed to store rules. If buffer is NULL or bufferLen is not enough
 757  * to store rules, will store up to available space.
 758  * @param coll collator to get the rules from
 759  * @param delta one of UCOL_TAILORING_ONLY, UCOL_FULL_RULES.
 760  * @param buffer buffer to store the result in. If NULL, you'll get no rules.
 761  * @param bufferLen lenght of buffer to store rules in. If less then needed you'll get only the part that fits in.
 762  * @return current rules
 763  * @stable ICU 2.0
 764  */
 765 U_CAPI int32_t U_EXPORT2
 766 ucol_getRulesEx(const UCollator *coll, UColRuleOption delta, UChar *buffer, int32_t bufferLen);
 767
 768 /**
 769  * gets the locale name of the collator. If the collator
 770  * is instantiated from the rules, then this function returns
 771  * NULL.
 772  * @param coll The UCollator for which the locale is needed
 773  * @param type You can choose between requested, valid and actual
 774  *             locale. For description see the definition of
 775  *             ULocDataLocaleType in uloc.h
 776  * @param status error code of the operation
 777  * @return real locale name from which the collation data comes.
 778  *         If the collator was instantiated from rules, returns
 779  *         NULL.
 780  * @stable ICU 2.1
 781  */
 782 U_CAPI const char * U_EXPORT2
 783 ucol_getLocale(const UCollator *coll, ULocDataLocaleType type, UErrorCode *status);
 784
 785
 786 /**
 787  * Get an Unicode set that contains all the characters and sequences tailored in
 788  * this collator. The result must be disposed of by using uset_close.
 789  * @param coll        The UCollator for which we want to get tailored chars
 790  * @param status      error code of the operation
 791  * @return a pointer to newly created USet. Must be be disposed by using uset_close
 792  * @see ucol_openRules
 793  * @see uset_close
 794  * @draft ICU 2.4
 795  */
 796 U_CAPI USet * U_EXPORT2
 797 ucol_getTailoredSet(const UCollator *coll, UErrorCode *status);
 798
 799 #endif /* #if !UCONFIG_NO_COLLATION */
 800
 801 #endif