Home | History | Annotate | Download | only in unicode
      1 /*
      2 ******************************************************************************
      3 *   Copyright (C) 1996-2014, International Business Machines
      4 *   Corporation and others.  All Rights Reserved.
      5 ******************************************************************************
      6 */
      7 
      8 /**
      9  * \file
     10  * \brief C++ API: Collation Service.
     11  */
     12 
     13 /**
     14 * File coll.h
     15 *
     16 * Created by: Helena Shih
     17 *
     18 * Modification History:
     19 *
     20 *  Date        Name        Description
     21 * 02/5/97      aliu        Modified createDefault to load collation data from
     22 *                          binary files when possible.  Added related methods
     23 *                          createCollationFromFile, chopLocale, createPathName.
     24 * 02/11/97     aliu        Added members addToCache, findInCache, and fgCache.
     25 * 02/12/97     aliu        Modified to create objects from RuleBasedCollator cache.
     26 *                          Moved cache out of Collation class.
     27 * 02/13/97     aliu        Moved several methods out of this class and into
     28 *                          RuleBasedCollator, with modifications.  Modified
     29 *                          createDefault() to call new RuleBasedCollator(Locale&)
     30 *                          constructor.  General clean up and documentation.
     31 * 02/20/97     helena      Added clone, operator==, operator!=, operator=, copy
     32 *                          constructor and getDynamicClassID.
     33 * 03/25/97     helena      Updated with platform independent data types.
     34 * 05/06/97     helena      Added memory allocation error detection.
     35 * 06/20/97     helena      Java class name change.
     36 * 09/03/97     helena      Added createCollationKeyValues().
     37 * 02/10/98     damiba      Added compare() with length as parameter.
     38 * 04/23/99     stephen     Removed EDecompositionMode, merged with
     39 *                          Normalizer::EMode.
     40 * 11/02/99     helena      Collator performance enhancements.  Eliminates the
     41 *                          UnicodeString construction and special case for NO_OP.
     42 * 11/23/99     srl         More performance enhancements. Inlining of
     43 *                          critical accessors.
     44 * 05/15/00     helena      Added version information API.
     45 * 01/29/01     synwee      Modified into a C++ wrapper which calls C apis
     46 *                          (ucol.h).
     47 * 2012-2014    markus      Rewritten in C++ again.
     48 */
     49 
     50 #ifndef COLL_H
     51 #define COLL_H
     52 
     53 #include "unicode/utypes.h"
     54 
     55 #if !UCONFIG_NO_COLLATION
     56 
     57 #include "unicode/uobject.h"
     58 #include "unicode/ucol.h"
     59 #include "unicode/normlzr.h"
     60 #include "unicode/locid.h"
     61 #include "unicode/uniset.h"
     62 #include "unicode/umisc.h"
     63 #include "unicode/uiter.h"
     64 #include "unicode/stringpiece.h"
     65 
     66 U_NAMESPACE_BEGIN
     67 
     68 class StringEnumeration;
     69 
     70 #if !UCONFIG_NO_SERVICE
     71 /**
     72  * @stable ICU 2.6
     73  */
     74 class CollatorFactory;
     75 #endif
     76 
     77 /**
     78 * @stable ICU 2.0
     79 */
     80 class CollationKey;
     81 
     82 /**
     83 * The <code>Collator</code> class performs locale-sensitive string
     84 * comparison.<br>
     85 * You use this class to build searching and sorting routines for natural
     86 * language text.
     87 * <p>
     88 * <code>Collator</code> is an abstract base class. Subclasses implement
     89 * specific collation strategies. One subclass,
     90 * <code>RuleBasedCollator</code>, is currently provided and is applicable
     91 * to a wide set of languages. Other subclasses may be created to handle more
     92 * specialized needs.
     93 * <p>
     94 * Like other locale-sensitive classes, you can use the static factory method,
     95 * <code>createInstance</code>, to obtain the appropriate
     96 * <code>Collator</code> object for a given locale. You will only need to
     97 * look at the subclasses of <code>Collator</code> if you need to
     98 * understand the details of a particular collation strategy or if you need to
     99 * modify that strategy.
    100 * <p>
    101 * The following example shows how to compare two strings using the
    102 * <code>Collator</code> for the default locale.
    103 * \htmlonly<blockquote>\endhtmlonly
    104 * <pre>
    105 * \code
    106 * // Compare two strings in the default locale
    107 * UErrorCode success = U_ZERO_ERROR;
    108 * Collator* myCollator = Collator::createInstance(success);
    109 * if (myCollator->compare("abc", "ABC") < 0)
    110 *   cout << "abc is less than ABC" << endl;
    111 * else
    112 *   cout << "abc is greater than or equal to ABC" << endl;
    113 * \endcode
    114 * </pre>
    115 * \htmlonly</blockquote>\endhtmlonly
    116 * <p>
    117 * You can set a <code>Collator</code>'s <em>strength</em> attribute to
    118 * determine the level of difference considered significant in comparisons.
    119 * Five strengths are provided: <code>PRIMARY</code>, <code>SECONDARY</code>,
    120 * <code>TERTIARY</code>, <code>QUATERNARY</code> and <code>IDENTICAL</code>.
    121 * The exact assignment of strengths to language features is locale dependent.
    122 * For example, in Czech, "e" and "f" are considered primary differences,
    123 * while "e" and "\u00EA" are secondary differences, "e" and "E" are tertiary
    124 * differences and "e" and "e" are identical. The following shows how both case
    125 * and accents could be ignored for US English.
    126 * \htmlonly<blockquote>\endhtmlonly
    127 * <pre>
    128 * \code
    129 * //Get the Collator for US English and set its strength to PRIMARY
    130 * UErrorCode success = U_ZERO_ERROR;
    131 * Collator* usCollator = Collator::createInstance(Locale::getUS(), success);
    132 * usCollator->setStrength(Collator::PRIMARY);
    133 * if (usCollator->compare("abc", "ABC") == 0)
    134 *     cout << "'abc' and 'ABC' strings are equivalent with strength PRIMARY" << endl;
    135 * \endcode
    136 * </pre>
    137 * \htmlonly</blockquote>\endhtmlonly
    138 * <p>
    139 * For comparing strings exactly once, the <code>compare</code> method
    140 * provides the best performance. When sorting a list of strings however, it
    141 * is generally necessary to compare each string multiple times. In this case,
    142 * sort keys provide better performance. The <code>getSortKey</code> methods
    143 * convert a string to a series of bytes that can be compared bitwise against
    144 * other sort keys using <code>strcmp()</code>. Sort keys are written as
    145 * zero-terminated byte strings. They consist of several substrings, one for
    146 * each collation strength level, that are delimited by 0x01 bytes.
    147 * If the string code points are appended for UCOL_IDENTICAL, then they are
    148 * processed for correct code point order comparison and may contain 0x01
    149 * bytes but not zero bytes.
    150 * </p>
    151 * <p>
    152 * Another set of APIs returns a <code>CollationKey</code> object that wraps
    153 * the sort key bytes instead of returning the bytes themselves.
    154 * </p>
    155 * <p>
    156 * <strong>Note:</strong> <code>Collator</code>s with different Locale,
    157 * and CollationStrength settings will return different sort
    158 * orders for the same set of strings. Locales have specific collation rules,
    159 * and the way in which secondary and tertiary differences are taken into
    160 * account, for example, will result in a different sorting order for same
    161 * strings.
    162 * </p>
    163 * @see         RuleBasedCollator
    164 * @see         CollationKey
    165 * @see         CollationElementIterator
    166 * @see         Locale
    167 * @see         Normalizer
    168 * @version     2.0 11/15/01
    169 */
    170 
    171 class U_I18N_API Collator : public UObject {
    172 public:
    173 
    174     // Collator public enums -----------------------------------------------
    175 
    176     /**
    177      * Base letter represents a primary difference. Set comparison level to
    178      * PRIMARY to ignore secondary and tertiary differences.<br>
    179      * Use this to set the strength of a Collator object.<br>
    180      * Example of primary difference, "abc" &lt; "abd"
    181      *
    182      * Diacritical differences on the same base letter represent a secondary
    183      * difference. Set comparison level to SECONDARY to ignore tertiary
    184      * differences. Use this to set the strength of a Collator object.<br>
    185      * Example of secondary difference, "&auml;" >> "a".
    186      *
    187      * Uppercase and lowercase versions of the same character represents a
    188      * tertiary difference.  Set comparison level to TERTIARY to include all
    189      * comparison differences. Use this to set the strength of a Collator
    190      * object.<br>
    191      * Example of tertiary difference, "abc" &lt;&lt;&lt; "ABC".
    192      *
    193      * Two characters are considered "identical" when they have the same unicode
    194      * spellings.<br>
    195      * For example, "&auml;" == "&auml;".
    196      *
    197      * UCollationStrength is also used to determine the strength of sort keys
    198      * generated from Collator objects.
    199      * @stable ICU 2.0
    200      */
    201     enum ECollationStrength
    202     {
    203         PRIMARY    = UCOL_PRIMARY,  // 0
    204         SECONDARY  = UCOL_SECONDARY,  // 1
    205         TERTIARY   = UCOL_TERTIARY,  // 2
    206         QUATERNARY = UCOL_QUATERNARY,  // 3
    207         IDENTICAL  = UCOL_IDENTICAL  // 15
    208     };
    209 
    210     /**
    211      * LESS is returned if source string is compared to be less than target
    212      * string in the compare() method.
    213      * EQUAL is returned if source string is compared to be equal to target
    214      * string in the compare() method.
    215      * GREATER is returned if source string is compared to be greater than
    216      * target string in the compare() method.
    217      * @see Collator#compare
    218      * @deprecated ICU 2.6. Use C enum UCollationResult defined in ucol.h
    219      */
    220     enum EComparisonResult
    221     {
    222         LESS = UCOL_LESS,  // -1
    223         EQUAL = UCOL_EQUAL,  // 0
    224         GREATER = UCOL_GREATER  // 1
    225     };
    226 
    227     // Collator public destructor -----------------------------------------
    228 
    229     /**
    230      * Destructor
    231      * @stable ICU 2.0
    232      */
    233     virtual ~Collator();
    234 
    235     // Collator public methods --------------------------------------------
    236 
    237     /**
    238      * Returns TRUE if "other" is the same as "this".
    239      *
    240      * The base class implementation returns TRUE if "other" has the same type/class as "this":
    241      * <code>typeid(*this) == typeid(other)</code>.
    242      *
    243      * Subclass implementations should do something like the following:
    244      * <pre>
    245      *   if (this == &other) { return TRUE; }
    246      *   if (!Collator::operator==(other)) { return FALSE; }  // not the same class
    247      *
    248      *   const MyCollator &o = (const MyCollator&)other;
    249      *   (compare this vs. o's subclass fields)
    250      * </pre>
    251      * @param other Collator object to be compared
    252      * @return TRUE if other is the same as this.
    253      * @stable ICU 2.0
    254      */
    255     virtual UBool operator==(const Collator& other) const;
    256 
    257     /**
    258      * Returns true if "other" is not the same as "this".
    259      * Calls ! operator==(const Collator&) const which works for all subclasses.
    260      * @param other Collator object to be compared
    261      * @return TRUE if other is not the same as this.
    262      * @stable ICU 2.0
    263      */
    264     virtual UBool operator!=(const Collator& other) const;
    265 
    266     /**
    267      * Makes a copy of this object.
    268      * @return a copy of this object, owned by the caller
    269      * @stable ICU 2.0
    270      */
    271     virtual Collator* clone(void) const = 0;
    272 
    273     /**
    274      * Creates the Collator object for the current default locale.
    275      * The default locale is determined by Locale::getDefault.
    276      * The UErrorCode& err parameter is used to return status information to the user.
    277      * To check whether the construction succeeded or not, you should check the
    278      * value of U_SUCCESS(err).  If you wish more detailed information, you can
    279      * check for informational error results which still indicate success.
    280      * U_USING_FALLBACK_ERROR indicates that a fall back locale was used. For
    281      * example, 'de_CH' was requested, but nothing was found there, so 'de' was
    282      * used. U_USING_DEFAULT_ERROR indicates that the default locale data was
    283      * used; neither the requested locale nor any of its fall back locales
    284      * could be found.
    285      * The caller owns the returned object and is responsible for deleting it.
    286      *
    287      * @param err    the error code status.
    288      * @return       the collation object of the default locale.(for example, en_US)
    289      * @see Locale#getDefault
    290      * @stable ICU 2.0
    291      */
    292     static Collator* U_EXPORT2 createInstance(UErrorCode&  err);
    293 
    294     /**
    295      * Gets the collation object for the desired locale. The
    296      * resource of the desired locale will be loaded.
    297      *
    298      * Locale::getRoot() is the base collation table and all other languages are
    299      * built on top of it with additional language-specific modifications.
    300      *
    301      * For some languages, multiple collation types are available;
    302      * for example, "de@collation=phonebook".
    303      * Starting with ICU 54, collation attributes can be specified via locale keywords as well,
    304      * in the old locale extension syntax ("el@colCaseFirst=upper")
    305      * or in language tag syntax ("el-u-kf-upper").
    306      * See <a href="http://userguide.icu-project.org/collation/api">User Guide: Collation API</a>.
    307      *
    308      * The UErrorCode& err parameter is used to return status information to the user.
    309      * To check whether the construction succeeded or not, you should check
    310      * the value of U_SUCCESS(err).  If you wish more detailed information, you
    311      * can check for informational error results which still indicate success.
    312      * U_USING_FALLBACK_ERROR indicates that a fall back locale was used.  For
    313      * example, 'de_CH' was requested, but nothing was found there, so 'de' was
    314      * used.  U_USING_DEFAULT_ERROR indicates that the default locale data was
    315      * used; neither the requested locale nor any of its fall back locales
    316      * could be found.
    317      *
    318      * The caller owns the returned object and is responsible for deleting it.
    319      * @param loc    The locale ID for which to open a collator.
    320      * @param err    the error code status.
    321      * @return       the created table-based collation object based on the desired
    322      *               locale.
    323      * @see Locale
    324      * @see ResourceLoader
    325      * @stable ICU 2.2
    326      */
    327     static Collator* U_EXPORT2 createInstance(const Locale& loc, UErrorCode& err);
    328 
    329     /**
    330      * The comparison function compares the character data stored in two
    331      * different strings. Returns information about whether a string is less
    332      * than, greater than or equal to another string.
    333      * @param source the source string to be compared with.
    334      * @param target the string that is to be compared with the source string.
    335      * @return Returns a byte value. GREATER if source is greater
    336      * than target; EQUAL if source is equal to target; LESS if source is less
    337      * than target
    338      * @deprecated ICU 2.6 use the overload with UErrorCode &
    339      */
    340     virtual EComparisonResult compare(const UnicodeString& source,
    341                                       const UnicodeString& target) const;
    342 
    343     /**
    344      * The comparison function compares the character data stored in two
    345      * different strings. Returns information about whether a string is less
    346      * than, greater than or equal to another string.
    347      * @param source the source string to be compared with.
    348      * @param target the string that is to be compared with the source string.
    349      * @param status possible error code
    350      * @return Returns an enum value. UCOL_GREATER if source is greater
    351      * than target; UCOL_EQUAL if source is equal to target; UCOL_LESS if source is less
    352      * than target
    353      * @stable ICU 2.6
    354      */
    355     virtual UCollationResult compare(const UnicodeString& source,
    356                                       const UnicodeString& target,
    357                                       UErrorCode &status) const = 0;
    358 
    359     /**
    360      * Does the same thing as compare but limits the comparison to a specified
    361      * length
    362      * @param source the source string to be compared with.
    363      * @param target the string that is to be compared with the source string.
    364      * @param length the length the comparison is limited to
    365      * @return Returns a byte value. GREATER if source (up to the specified
    366      *         length) is greater than target; EQUAL if source (up to specified
    367      *         length) is equal to target; LESS if source (up to the specified
    368      *         length) is less  than target.
    369      * @deprecated ICU 2.6 use the overload with UErrorCode &
    370      */
    371     virtual EComparisonResult compare(const UnicodeString& source,
    372                                       const UnicodeString& target,
    373                                       int32_t length) const;
    374 
    375     /**
    376      * Does the same thing as compare but limits the comparison to a specified
    377      * length
    378      * @param source the source string to be compared with.
    379      * @param target the string that is to be compared with the source string.
    380      * @param length the length the comparison is limited to
    381      * @param status possible error code
    382      * @return Returns an enum value. UCOL_GREATER if source (up to the specified
    383      *         length) is greater than target; UCOL_EQUAL if source (up to specified
    384      *         length) is equal to target; UCOL_LESS if source (up to the specified
    385      *         length) is less  than target.
    386      * @stable ICU 2.6
    387      */
    388     virtual UCollationResult compare(const UnicodeString& source,
    389                                       const UnicodeString& target,
    390                                       int32_t length,
    391                                       UErrorCode &status) const = 0;
    392 
    393     /**
    394      * The comparison function compares the character data stored in two
    395      * different string arrays. Returns information about whether a string array
    396      * is less than, greater than or equal to another string array.
    397      * <p>Example of use:
    398      * <pre>
    399      * .       UChar ABC[] = {0x41, 0x42, 0x43, 0};  // = "ABC"
    400      * .       UChar abc[] = {0x61, 0x62, 0x63, 0};  // = "abc"
    401      * .       UErrorCode status = U_ZERO_ERROR;
    402      * .       Collator *myCollation =
    403      * .                         Collator::createInstance(Locale::getUS(), status);
    404      * .       if (U_FAILURE(status)) return;
    405      * .       myCollation->setStrength(Collator::PRIMARY);
    406      * .       // result would be Collator::EQUAL ("abc" == "ABC")
    407      * .       // (no primary difference between "abc" and "ABC")
    408      * .       Collator::EComparisonResult result =
    409      * .                             myCollation->compare(abc, 3, ABC, 3);
    410      * .       myCollation->setStrength(Collator::TERTIARY);
    411      * .       // result would be Collator::LESS ("abc" &lt;&lt;&lt; "ABC")
    412      * .       // (with tertiary difference between "abc" and "ABC")
    413      * .       result = myCollation->compare(abc, 3, ABC, 3);
    414      * </pre>
    415      * @param source the source string array to be compared with.
    416      * @param sourceLength the length of the source string array.  If this value
    417      *        is equal to -1, the string array is null-terminated.
    418      * @param target the string that is to be compared with the source string.
    419      * @param targetLength the length of the target string array.  If this value
    420      *        is equal to -1, the string array is null-terminated.
    421      * @return Returns a byte value. GREATER if source is greater than target;
    422      *         EQUAL if source is equal to target; LESS if source is less than
    423      *         target
    424      * @deprecated ICU 2.6 use the overload with UErrorCode &
    425      */
    426     virtual EComparisonResult compare(const UChar* source, int32_t sourceLength,
    427                                       const UChar* target, int32_t targetLength)
    428                                       const;
    429 
    430     /**
    431      * The comparison function compares the character data stored in two
    432      * different string arrays. Returns information about whether a string array
    433      * is less than, greater than or equal to another string array.
    434      * @param source the source string array to be compared with.
    435      * @param sourceLength the length of the source string array.  If this value
    436      *        is equal to -1, the string array is null-terminated.
    437      * @param target the string that is to be compared with the source string.
    438      * @param targetLength the length of the target string array.  If this value
    439      *        is equal to -1, the string array is null-terminated.
    440      * @param status possible error code
    441      * @return Returns an enum value. UCOL_GREATER if source is greater
    442      * than target; UCOL_EQUAL if source is equal to target; UCOL_LESS if source is less
    443      * than target
    444      * @stable ICU 2.6
    445      */
    446     virtual UCollationResult compare(const UChar* source, int32_t sourceLength,
    447                                       const UChar* target, int32_t targetLength,
    448                                       UErrorCode &status) const = 0;
    449 
    450     /**
    451      * Compares two strings using the Collator.
    452      * Returns whether the first one compares less than/equal to/greater than
    453      * the second one.
    454      * This version takes UCharIterator input.
    455      * @param sIter the first ("source") string iterator
    456      * @param tIter the second ("target") string iterator
    457      * @param status ICU status
    458      * @return UCOL_LESS, UCOL_EQUAL or UCOL_GREATER
    459      * @stable ICU 4.2
    460      */
    461     virtual UCollationResult compare(UCharIterator &sIter,
    462                                      UCharIterator &tIter,
    463                                      UErrorCode &status) const;
    464 
    465     /**
    466      * Compares two UTF-8 strings using the Collator.
    467      * Returns whether the first one compares less than/equal to/greater than
    468      * the second one.
    469      * This version takes UTF-8 input.
    470      * Note that a StringPiece can be implicitly constructed
    471      * from a std::string or a NUL-terminated const char * string.
    472      * @param source the first UTF-8 string
    473      * @param target the second UTF-8 string
    474      * @param status ICU status
    475      * @return UCOL_LESS, UCOL_EQUAL or UCOL_GREATER
    476      * @stable ICU 4.2
    477      */
    478     virtual UCollationResult compareUTF8(const StringPiece &source,
    479                                          const StringPiece &target,
    480                                          UErrorCode &status) const;
    481 
    482     /**
    483      * Transforms the string into a series of characters that can be compared
    484      * with CollationKey::compareTo. It is not possible to restore the original
    485      * string from the chars in the sort key.  The generated sort key handles
    486      * only a limited number of ignorable characters.
    487      * <p>Use CollationKey::equals or CollationKey::compare to compare the
    488      * generated sort keys.
    489      * If the source string is null, a null collation key will be returned.
    490      * @param source the source string to be transformed into a sort key.
    491      * @param key the collation key to be filled in
    492      * @param status the error code status.
    493      * @return the collation key of the string based on the collation rules.
    494      * @see CollationKey#compare
    495      * @stable ICU 2.0
    496      */
    497     virtual CollationKey& getCollationKey(const UnicodeString&  source,
    498                                           CollationKey& key,
    499                                           UErrorCode& status) const = 0;
    500 
    501     /**
    502      * Transforms the string into a series of characters that can be compared
    503      * with CollationKey::compareTo. It is not possible to restore the original
    504      * string from the chars in the sort key.  The generated sort key handles
    505      * only a limited number of ignorable characters.
    506      * <p>Use CollationKey::equals or CollationKey::compare to compare the
    507      * generated sort keys.
    508      * <p>If the source string is null, a null collation key will be returned.
    509      * @param source the source string to be transformed into a sort key.
    510      * @param sourceLength length of the collation key
    511      * @param key the collation key to be filled in
    512      * @param status the error code status.
    513      * @return the collation key of the string based on the collation rules.
    514      * @see CollationKey#compare
    515      * @stable ICU 2.0
    516      */
    517     virtual CollationKey& getCollationKey(const UChar*source,
    518                                           int32_t sourceLength,
    519                                           CollationKey& key,
    520                                           UErrorCode& status) const = 0;
    521     /**
    522      * Generates the hash code for the collation object
    523      * @stable ICU 2.0
    524      */
    525     virtual int32_t hashCode(void) const = 0;
    526 
    527     /**
    528      * Gets the locale of the Collator
    529      *
    530      * @param type can be either requested, valid or actual locale. For more
    531      *             information see the definition of ULocDataLocaleType in
    532      *             uloc.h
    533      * @param status the error code status.
    534      * @return locale where the collation data lives. If the collator
    535      *         was instantiated from rules, locale is empty.
    536      * @deprecated ICU 2.8 This API is under consideration for revision
    537      * in ICU 3.0.
    538      */
    539     virtual Locale getLocale(ULocDataLocaleType type, UErrorCode& status) const = 0;
    540 
    541     /**
    542      * Convenience method for comparing two strings based on the collation rules.
    543      * @param source the source string to be compared with.
    544      * @param target the target string to be compared with.
    545      * @return true if the first string is greater than the second one,
    546      *         according to the collation rules. false, otherwise.
    547      * @see Collator#compare
    548      * @stable ICU 2.0
    549      */
    550     UBool greater(const UnicodeString& source, const UnicodeString& target)
    551                   const;
    552 
    553     /**
    554      * Convenience method for comparing two strings based on the collation rules.
    555      * @param source the source string to be compared with.
    556      * @param target the target string to be compared with.
    557      * @return true if the first string is greater than or equal to the second
    558      *         one, according to the collation rules. false, otherwise.
    559      * @see Collator#compare
    560      * @stable ICU 2.0
    561      */
    562     UBool greaterOrEqual(const UnicodeString& source,
    563                          const UnicodeString& target) const;
    564 
    565     /**
    566      * Convenience method for comparing two strings based on the collation rules.
    567      * @param source the source string to be compared with.
    568      * @param target the target string to be compared with.
    569      * @return true if the strings are equal according to the collation rules.
    570      *         false, otherwise.
    571      * @see Collator#compare
    572      * @stable ICU 2.0
    573      */
    574     UBool equals(const UnicodeString& source, const UnicodeString& target) const;
    575 
    576     /**
    577      * Determines the minimum strength that will be used in comparison or
    578      * transformation.
    579      * <p>E.g. with strength == SECONDARY, the tertiary difference is ignored
    580      * <p>E.g. with strength == PRIMARY, the secondary and tertiary difference
    581      * are ignored.
    582      * @return the current comparison level.
    583      * @see Collator#setStrength
    584      * @deprecated ICU 2.6 Use getAttribute(UCOL_STRENGTH...) instead
    585      */
    586     virtual ECollationStrength getStrength(void) const;
    587 
    588     /**
    589      * Sets the minimum strength to be used in comparison or transformation.
    590      * <p>Example of use:
    591      * <pre>
    592      *  \code
    593      *  UErrorCode status = U_ZERO_ERROR;
    594      *  Collator*myCollation = Collator::createInstance(Locale::getUS(), status);
    595      *  if (U_FAILURE(status)) return;
    596      *  myCollation->setStrength(Collator::PRIMARY);
    597      *  // result will be "abc" == "ABC"
    598      *  // tertiary differences will be ignored
    599      *  Collator::ComparisonResult result = myCollation->compare("abc", "ABC");
    600      * \endcode
    601      * </pre>
    602      * @see Collator#getStrength
    603      * @param newStrength the new comparison level.
    604      * @deprecated ICU 2.6 Use setAttribute(UCOL_STRENGTH...) instead
    605      */
    606     virtual void setStrength(ECollationStrength newStrength);
    607 
    608     /**
    609      * Retrieves the reordering codes for this collator.
    610      * @param dest The array to fill with the script ordering.
    611      * @param destCapacity The length of dest. If it is 0, then dest may be NULL and the function
    612      *  will only return the length of the result without writing any of the result string (pre-flighting).
    613      * @param status A reference to an error code value, which must not indicate
    614      * a failure before the function call.
    615      * @return The length of the script ordering array.
    616      * @see ucol_setReorderCodes
    617      * @see Collator#getEquivalentReorderCodes
    618      * @see Collator#setReorderCodes
    619      * @see UScriptCode
    620      * @see UColReorderCode
    621      * @stable ICU 4.8
    622      */
    623      virtual int32_t getReorderCodes(int32_t *dest,
    624                                      int32_t destCapacity,
    625                                      UErrorCode& status) const;
    626 
    627     /**
    628      * Sets the ordering of scripts for this collator.
    629      *
    630      * <p>The reordering codes are a combination of script codes and reorder codes.
    631      * @param reorderCodes An array of script codes in the new order. This can be NULL if the
    632      * length is also set to 0. An empty array will clear any reordering codes on the collator.
    633      * @param reorderCodesLength The length of reorderCodes.
    634      * @param status error code
    635      * @see Collator#getReorderCodes
    636      * @see Collator#getEquivalentReorderCodes
    637      * @see UScriptCode
    638      * @see UColReorderCode
    639      * @stable ICU 4.8
    640      */
    641      virtual void setReorderCodes(const int32_t* reorderCodes,
    642                                   int32_t reorderCodesLength,
    643                                   UErrorCode& status) ;
    644 
    645     /**
    646      * Retrieves the reorder codes that are grouped with the given reorder code. Some reorder
    647      * codes will be grouped and must reorder together.
    648      * @param reorderCode The reorder code to determine equivalence for.
    649      * @param dest The array to fill with the script equivalence reordering codes.
    650      * @param destCapacity The length of dest. If it is 0, then dest may be NULL and the
    651      * function will only return the length of the result without writing any of the result
    652      * string (pre-flighting).
    653      * @param status A reference to an error code value, which must not indicate
    654      * a failure before the function call.
    655      * @return The length of the of the reordering code equivalence array.
    656      * @see ucol_setReorderCodes
    657      * @see Collator#getReorderCodes
    658      * @see Collator#setReorderCodes
    659      * @see UScriptCode
    660      * @see UColReorderCode
    661      * @stable ICU 4.8
    662      */
    663     static int32_t U_EXPORT2 getEquivalentReorderCodes(int32_t reorderCode,
    664                                 int32_t* dest,
    665                                 int32_t destCapacity,
    666                                 UErrorCode& status);
    667 
    668     /**
    669      * Get name of the object for the desired Locale, in the desired langauge
    670      * @param objectLocale must be from getAvailableLocales
    671      * @param displayLocale specifies the desired locale for output
    672      * @param name the fill-in parameter of the return value
    673      * @return display-able name of the object for the object locale in the
    674      *         desired language
    675      * @stable ICU 2.0
    676      */
    677     static UnicodeString& U_EXPORT2 getDisplayName(const Locale& objectLocale,
    678                                          const Locale& displayLocale,
    679                                          UnicodeString& name);
    680 
    681     /**
    682     * Get name of the object for the desired Locale, in the langauge of the
    683     * default locale.
    684     * @param objectLocale must be from getAvailableLocales
    685     * @param name the fill-in parameter of the return value
    686     * @return name of the object for the desired locale in the default language
    687     * @stable ICU 2.0
    688     */
    689     static UnicodeString& U_EXPORT2 getDisplayName(const Locale& objectLocale,
    690                                          UnicodeString& name);
    691 
    692     /**
    693      * Get the set of Locales for which Collations are installed.
    694      *
    695      * <p>Note this does not include locales supported by registered collators.
    696      * If collators might have been registered, use the overload of getAvailableLocales
    697      * that returns a StringEnumeration.</p>
    698      *
    699      * @param count the output parameter of number of elements in the locale list
    700      * @return the list of available locales for which collations are installed
    701      * @stable ICU 2.0
    702      */
    703     static const Locale* U_EXPORT2 getAvailableLocales(int32_t& count);
    704 
    705     /**
    706      * Return a StringEnumeration over the locales available at the time of the call,
    707      * including registered locales.  If a severe error occurs (such as out of memory
    708      * condition) this will return null. If there is no locale data, an empty enumeration
    709      * will be returned.
    710      * @return a StringEnumeration over the locales available at the time of the call
    711      * @stable ICU 2.6
    712      */
    713     static StringEnumeration* U_EXPORT2 getAvailableLocales(void);
    714 
    715     /**
    716      * Create a string enumerator of all possible keywords that are relevant to
    717      * collation. At this point, the only recognized keyword for this
    718      * service is "collation".
    719      * @param status input-output error code
    720      * @return a string enumeration over locale strings. The caller is
    721      * responsible for closing the result.
    722      * @stable ICU 3.0
    723      */
    724     static StringEnumeration* U_EXPORT2 getKeywords(UErrorCode& status);
    725 
    726     /**
    727      * Given a keyword, create a string enumeration of all values
    728      * for that keyword that are currently in use.
    729      * @param keyword a particular keyword as enumerated by
    730      * ucol_getKeywords. If any other keyword is passed in, status is set
    731      * to U_ILLEGAL_ARGUMENT_ERROR.
    732      * @param status input-output error code
    733      * @return a string enumeration over collation keyword values, or NULL
    734      * upon error. The caller is responsible for deleting the result.
    735      * @stable ICU 3.0
    736      */
    737     static StringEnumeration* U_EXPORT2 getKeywordValues(const char *keyword, UErrorCode& status);
    738 
    739     /**
    740      * Given a key and a locale, returns an array of string values in a preferred
    741      * order that would make a difference. These are all and only those values where
    742      * the open (creation) of the service with the locale formed from the input locale
    743      * plus input keyword and that value has different behavior than creation with the
    744      * input locale alone.
    745      * @param keyword        one of the keys supported by this service.  For now, only
    746      *                      "collation" is supported.
    747      * @param locale        the locale
    748      * @param commonlyUsed  if set to true it will return only commonly used values
    749      *                      with the given locale in preferred order.  Otherwise,
    750      *                      it will return all the available values for the locale.
    751      * @param status ICU status
    752      * @return a string enumeration over keyword values for the given key and the locale.
    753      * @stable ICU 4.2
    754      */
    755     static StringEnumeration* U_EXPORT2 getKeywordValuesForLocale(const char* keyword, const Locale& locale,
    756                                                                     UBool commonlyUsed, UErrorCode& status);
    757 
    758     /**
    759      * Return the functionally equivalent locale for the given
    760      * requested locale, with respect to given keyword, for the
    761      * collation service.  If two locales return the same result, then
    762      * collators instantiated for these locales will behave
    763      * equivalently.  The converse is not always true; two collators
    764      * may in fact be equivalent, but return different results, due to
    765      * internal details.  The return result has no other meaning than
    766      * that stated above, and implies nothing as to the relationship
    767      * between the two locales.  This is intended for use by
    768      * applications who wish to cache collators, or otherwise reuse
    769      * collators when possible.  The functional equivalent may change
    770      * over time.  For more information, please see the <a
    771      * href="http://userguide.icu-project.org/locale#TOC-Locales-and-Services">
    772      * Locales and Services</a> section of the ICU User Guide.
    773      * @param keyword a particular keyword as enumerated by
    774      * ucol_getKeywords.
    775      * @param locale the requested locale
    776      * @param isAvailable reference to a fillin parameter that
    777      * indicates whether the requested locale was 'available' to the
    778      * collation service. A locale is defined as 'available' if it
    779      * physically exists within the collation locale data.
    780      * @param status reference to input-output error code
    781      * @return the functionally equivalent collation locale, or the root
    782      * locale upon error.
    783      * @stable ICU 3.0
    784      */
    785     static Locale U_EXPORT2 getFunctionalEquivalent(const char* keyword, const Locale& locale,
    786                                           UBool& isAvailable, UErrorCode& status);
    787 
    788 #if !UCONFIG_NO_SERVICE
    789     /**
    790      * Register a new Collator.  The collator will be adopted.
    791      * Because ICU may choose to cache collators internally, this must be
    792      * called at application startup, prior to any calls to
    793      * Collator::createInstance to avoid undefined behavior.
    794      * @param toAdopt the Collator instance to be adopted
    795      * @param locale the locale with which the collator will be associated
    796      * @param status the in/out status code, no special meanings are assigned
    797      * @return a registry key that can be used to unregister this collator
    798      * @stable ICU 2.6
    799      */
    800     static URegistryKey U_EXPORT2 registerInstance(Collator* toAdopt, const Locale& locale, UErrorCode& status);
    801 
    802     /**
    803      * Register a new CollatorFactory.  The factory will be adopted.
    804      * Because ICU may choose to cache collators internally, this must be
    805      * called at application startup, prior to any calls to
    806      * Collator::createInstance to avoid undefined behavior.
    807      * @param toAdopt the CollatorFactory instance to be adopted
    808      * @param status the in/out status code, no special meanings are assigned
    809      * @return a registry key that can be used to unregister this collator
    810      * @stable ICU 2.6
    811      */
    812     static URegistryKey U_EXPORT2 registerFactory(CollatorFactory* toAdopt, UErrorCode& status);
    813 
    814     /**
    815      * Unregister a previously-registered Collator or CollatorFactory
    816      * using the key returned from the register call.  Key becomes
    817      * invalid after a successful call and should not be used again.
    818      * The object corresponding to the key will be deleted.
    819      * Because ICU may choose to cache collators internally, this should
    820      * be called during application shutdown, after all calls to
    821      * Collator::createInstance to avoid undefined behavior.
    822      * @param key the registry key returned by a previous call to registerInstance
    823      * @param status the in/out status code, no special meanings are assigned
    824      * @return TRUE if the collator for the key was successfully unregistered
    825      * @stable ICU 2.6
    826      */
    827     static UBool U_EXPORT2 unregister(URegistryKey key, UErrorCode& status);
    828 #endif /* UCONFIG_NO_SERVICE */
    829 
    830     /**
    831      * Gets the version information for a Collator.
    832      * @param info the version # information, the result will be filled in
    833      * @stable ICU 2.0
    834      */
    835     virtual void getVersion(UVersionInfo info) const = 0;
    836 
    837     /**
    838      * Returns a unique class ID POLYMORPHICALLY. Pure virtual method.
    839      * This method is to implement a simple version of RTTI, since not all C++
    840      * compilers support genuine RTTI. Polymorphic operator==() and clone()
    841      * methods call this method.
    842      * @return The class ID for this object. All objects of a given class have
    843      *         the same class ID.  Objects of other classes have different class
    844      *         IDs.
    845      * @stable ICU 2.0
    846      */
    847     virtual UClassID getDynamicClassID(void) const = 0;
    848 
    849     /**
    850      * Universal attribute setter
    851      * @param attr attribute type
    852      * @param value attribute value
    853      * @param status to indicate whether the operation went on smoothly or
    854      *        there were errors
    855      * @stable ICU 2.2
    856      */
    857     virtual void setAttribute(UColAttribute attr, UColAttributeValue value,
    858                               UErrorCode &status) = 0;
    859 
    860     /**
    861      * Universal attribute getter
    862      * @param attr attribute type
    863      * @param status to indicate whether the operation went on smoothly or
    864      *        there were errors
    865      * @return attribute value
    866      * @stable ICU 2.2
    867      */
    868     virtual UColAttributeValue getAttribute(UColAttribute attr,
    869                                             UErrorCode &status) const = 0;
    870 
    871     /* Cannot use #ifndef U_HIDE_DRAFT_API for the following draft methods since they are virtual */
    872     /**
    873      * Sets the variable top to the top of the specified reordering group.
    874      * The variable top determines the highest-sorting character
    875      * which is affected by UCOL_ALTERNATE_HANDLING.
    876      * If that attribute is set to UCOL_NON_IGNORABLE, then the variable top has no effect.
    877      *
    878      * The base class implementation sets U_UNSUPPORTED_ERROR.
    879      * @param group one of UCOL_REORDER_CODE_SPACE, UCOL_REORDER_CODE_PUNCTUATION,
    880      *              UCOL_REORDER_CODE_SYMBOL, UCOL_REORDER_CODE_CURRENCY;
    881      *              or UCOL_REORDER_CODE_DEFAULT to restore the default max variable group
    882      * @param errorCode Standard ICU error code. Its input value must
    883      *                  pass the U_SUCCESS() test, or else the function returns
    884      *                  immediately. Check for U_FAILURE() on output or use with
    885      *                  function chaining. (See User Guide for details.)
    886      * @return *this
    887      * @see getMaxVariable
    888      * @draft ICU 53
    889      */
    890     virtual Collator &setMaxVariable(UColReorderCode group, UErrorCode &errorCode);
    891 
    892     /**
    893      * Returns the maximum reordering group whose characters are affected by UCOL_ALTERNATE_HANDLING.
    894      *
    895      * The base class implementation returns UCOL_REORDER_CODE_PUNCTUATION.
    896      * @return the maximum variable reordering group.
    897      * @see setMaxVariable
    898      * @draft ICU 53
    899      */
    900     virtual UColReorderCode getMaxVariable() const;
    901 
    902     /**
    903      * Sets the variable top to the primary weight of the specified string.
    904      *
    905      * Beginning with ICU 53, the variable top is pinned to
    906      * the top of one of the supported reordering groups,
    907      * and it must not be beyond the last of those groups.
    908      * See setMaxVariable().
    909      * @param varTop one or more (if contraction) UChars to which the variable top should be set
    910      * @param len length of variable top string. If -1 it is considered to be zero terminated.
    911      * @param status error code. If error code is set, the return value is undefined. Errors set by this function are: <br>
    912      *    U_CE_NOT_FOUND_ERROR if more than one character was passed and there is no such contraction<br>
    913      *    U_ILLEGAL_ARGUMENT_ERROR if the variable top is beyond
    914      *    the last reordering group supported by setMaxVariable()
    915      * @return variable top primary weight
    916      * @deprecated ICU 53 Call setMaxVariable() instead.
    917      */
    918     virtual uint32_t setVariableTop(const UChar *varTop, int32_t len, UErrorCode &status) = 0;
    919 
    920     /**
    921      * Sets the variable top to the primary weight of the specified string.
    922      *
    923      * Beginning with ICU 53, the variable top is pinned to
    924      * the top of one of the supported reordering groups,
    925      * and it must not be beyond the last of those groups.
    926      * See setMaxVariable().
    927      * @param varTop a UnicodeString size 1 or more (if contraction) of UChars to which the variable top should be set
    928      * @param status error code. If error code is set, the return value is undefined. Errors set by this function are: <br>
    929      *    U_CE_NOT_FOUND_ERROR if more than one character was passed and there is no such contraction<br>
    930      *    U_ILLEGAL_ARGUMENT_ERROR if the variable top is beyond
    931      *    the last reordering group supported by setMaxVariable()
    932      * @return variable top primary weight
    933      * @deprecated ICU 53 Call setMaxVariable() instead.
    934      */
    935     virtual uint32_t setVariableTop(const UnicodeString &varTop, UErrorCode &status) = 0;
    936 
    937     /**
    938      * Sets the variable top to the specified primary weight.
    939      *
    940      * Beginning with ICU 53, the variable top is pinned to
    941      * the top of one of the supported reordering groups,
    942      * and it must not be beyond the last of those groups.
    943      * See setMaxVariable().
    944      * @param varTop primary weight, as returned by setVariableTop or ucol_getVariableTop
    945      * @param status error code
    946      * @deprecated ICU 53 Call setMaxVariable() instead.
    947      */
    948     virtual void setVariableTop(uint32_t varTop, UErrorCode &status) = 0;
    949 
    950     /**
    951      * Gets the variable top value of a Collator.
    952      * @param status error code (not changed by function). If error code is set, the return value is undefined.
    953      * @return the variable top primary weight
    954      * @see getMaxVariable
    955      * @stable ICU 2.0
    956      */
    957     virtual uint32_t getVariableTop(UErrorCode &status) const = 0;
    958 
    959     /**
    960      * Get a UnicodeSet that contains all the characters and sequences
    961      * tailored in this collator.
    962      * @param status      error code of the operation
    963      * @return a pointer to a UnicodeSet object containing all the
    964      *         code points and sequences that may sort differently than
    965      *         in the root collator. The object must be disposed of by using delete
    966      * @stable ICU 2.4
    967      */
    968     virtual UnicodeSet *getTailoredSet(UErrorCode &status) const;
    969 
    970     /**
    971      * Same as clone().
    972      * The base class implementation simply calls clone().
    973      * @return a copy of this object, owned by the caller
    974      * @see clone()
    975      * @deprecated ICU 50 no need to have two methods for cloning
    976      */
    977     virtual Collator* safeClone(void) const;
    978 
    979     /**
    980      * Get the sort key as an array of bytes from a UnicodeString.
    981      * Sort key byte arrays are zero-terminated and can be compared using
    982      * strcmp().
    983      * @param source string to be processed.
    984      * @param result buffer to store result in. If NULL, number of bytes needed
    985      *        will be returned.
    986      * @param resultLength length of the result buffer. If if not enough the
    987      *        buffer will be filled to capacity.
    988      * @return Number of bytes needed for storing the sort key
    989      * @stable ICU 2.2
    990      */
    991     virtual int32_t getSortKey(const UnicodeString& source,
    992                               uint8_t* result,
    993                               int32_t resultLength) const = 0;
    994 
    995     /**
    996      * Get the sort key as an array of bytes from a UChar buffer.
    997      * Sort key byte arrays are zero-terminated and can be compared using
    998      * strcmp().
    999      * @param source string to be processed.
   1000      * @param sourceLength length of string to be processed.
   1001      *        If -1, the string is 0 terminated and length will be decided by the
   1002      *        function.
   1003      * @param result buffer to store result in. If NULL, number of bytes needed
   1004      *        will be returned.
   1005      * @param resultLength length of the result buffer. If if not enough the
   1006      *        buffer will be filled to capacity.
   1007      * @return Number of bytes needed for storing the sort key
   1008      * @stable ICU 2.2
   1009      */
   1010     virtual int32_t getSortKey(const UChar*source, int32_t sourceLength,
   1011                                uint8_t*result, int32_t resultLength) const = 0;
   1012 
   1013     /**
   1014      * Produce a bound for a given sortkey and a number of levels.
   1015      * Return value is always the number of bytes needed, regardless of
   1016      * whether the result buffer was big enough or even valid.<br>
   1017      * Resulting bounds can be used to produce a range of strings that are
   1018      * between upper and lower bounds. For example, if bounds are produced
   1019      * for a sortkey of string "smith", strings between upper and lower
   1020      * bounds with one level would include "Smith", "SMITH", "sMiTh".<br>
   1021      * There are two upper bounds that can be produced. If UCOL_BOUND_UPPER
   1022      * is produced, strings matched would be as above. However, if bound
   1023      * produced using UCOL_BOUND_UPPER_LONG is used, the above example will
   1024      * also match "Smithsonian" and similar.<br>
   1025      * For more on usage, see example in cintltst/capitst.c in procedure
   1026      * TestBounds.
   1027      * Sort keys may be compared using <TT>strcmp</TT>.
   1028      * @param source The source sortkey.
   1029      * @param sourceLength The length of source, or -1 if null-terminated.
   1030      *                     (If an unmodified sortkey is passed, it is always null
   1031      *                      terminated).
   1032      * @param boundType Type of bound required. It can be UCOL_BOUND_LOWER, which
   1033      *                  produces a lower inclusive bound, UCOL_BOUND_UPPER, that
   1034      *                  produces upper bound that matches strings of the same length
   1035      *                  or UCOL_BOUND_UPPER_LONG that matches strings that have the
   1036      *                  same starting substring as the source string.
   1037      * @param noOfLevels  Number of levels required in the resulting bound (for most
   1038      *                    uses, the recommended value is 1). See users guide for
   1039      *                    explanation on number of levels a sortkey can have.
   1040      * @param result A pointer to a buffer to receive the resulting sortkey.
   1041      * @param resultLength The maximum size of result.
   1042      * @param status Used for returning error code if something went wrong. If the
   1043      *               number of levels requested is higher than the number of levels
   1044      *               in the source key, a warning (U_SORT_KEY_TOO_SHORT_WARNING) is
   1045      *               issued.
   1046      * @return The size needed to fully store the bound.
   1047      * @see ucol_keyHashCode
   1048      * @stable ICU 2.1
   1049      */
   1050     static int32_t U_EXPORT2 getBound(const uint8_t       *source,
   1051             int32_t             sourceLength,
   1052             UColBoundMode       boundType,
   1053             uint32_t            noOfLevels,
   1054             uint8_t             *result,
   1055             int32_t             resultLength,
   1056             UErrorCode          &status);
   1057 
   1058 
   1059 protected:
   1060 
   1061     // Collator protected constructors -------------------------------------
   1062 
   1063     /**
   1064     * Default constructor.
   1065     * Constructor is different from the old default Collator constructor.
   1066     * The task for determing the default collation strength and normalization
   1067     * mode is left to the child class.
   1068     * @stable ICU 2.0
   1069     */
   1070     Collator();
   1071 
   1072 #ifndef U_HIDE_DEPRECATED_API
   1073     /**
   1074     * Constructor.
   1075     * Empty constructor, does not handle the arguments.
   1076     * This constructor is done for backward compatibility with 1.7 and 1.8.
   1077     * The task for handling the argument collation strength and normalization
   1078     * mode is left to the child class.
   1079     * @param collationStrength collation strength
   1080     * @param decompositionMode
   1081     * @deprecated ICU 2.4. Subclasses should use the default constructor
   1082     * instead and handle the strength and normalization mode themselves.
   1083     */
   1084     Collator(UCollationStrength collationStrength,
   1085              UNormalizationMode decompositionMode);
   1086 #endif  /* U_HIDE_DEPRECATED_API */
   1087 
   1088     /**
   1089     * Copy constructor.
   1090     * @param other Collator object to be copied from
   1091     * @stable ICU 2.0
   1092     */
   1093     Collator(const Collator& other);
   1094 
   1095 public:
   1096    /**
   1097     * Used internally by registration to define the requested and valid locales.
   1098     * @param requestedLocale the requested locale
   1099     * @param validLocale the valid locale
   1100     * @param actualLocale the actual locale
   1101     * @internal
   1102     */
   1103     virtual void setLocales(const Locale& requestedLocale, const Locale& validLocale, const Locale& actualLocale);
   1104 
   1105     /** Get the short definition string for a collator. This internal API harvests the collator's
   1106      *  locale and the attribute set and produces a string that can be used for opening
   1107      *  a collator with the same attributes using the ucol_openFromShortString API.
   1108      *  This string will be normalized.
   1109      *  The structure and the syntax of the string is defined in the "Naming collators"
   1110      *  section of the users guide:
   1111      *  http://userguide.icu-project.org/collation/concepts#TOC-Collator-naming-scheme
   1112      *  This function supports preflighting.
   1113      *
   1114      *  This is internal, and intended to be used with delegate converters.
   1115      *
   1116      *  @param locale a locale that will appear as a collators locale in the resulting
   1117      *                short string definition. If NULL, the locale will be harvested
   1118      *                from the collator.
   1119      *  @param buffer space to hold the resulting string
   1120      *  @param capacity capacity of the buffer
   1121      *  @param status for returning errors. All the preflighting errors are featured
   1122      *  @return length of the resulting string
   1123      *  @see ucol_openFromShortString
   1124      *  @see ucol_normalizeShortDefinitionString
   1125      *  @see ucol_getShortDefinitionString
   1126      *  @internal
   1127      */
   1128     virtual int32_t internalGetShortDefinitionString(const char *locale,
   1129                                                      char *buffer,
   1130                                                      int32_t capacity,
   1131                                                      UErrorCode &status) const;
   1132 
   1133     /**
   1134      * Implements ucol_strcollUTF8().
   1135      * @internal
   1136      */
   1137     virtual UCollationResult internalCompareUTF8(
   1138             const char *left, int32_t leftLength,
   1139             const char *right, int32_t rightLength,
   1140             UErrorCode &errorCode) const;
   1141 
   1142     /**
   1143      * Implements ucol_nextSortKeyPart().
   1144      * @internal
   1145      */
   1146     virtual int32_t
   1147     internalNextSortKeyPart(
   1148             UCharIterator *iter, uint32_t state[2],
   1149             uint8_t *dest, int32_t count, UErrorCode &errorCode) const;
   1150 
   1151 #ifndef U_HIDE_INTERNAL_API
   1152     /** @internal */
   1153     static inline Collator *fromUCollator(UCollator *uc) {
   1154         return reinterpret_cast<Collator *>(uc);
   1155     }
   1156     /** @internal */
   1157     static inline const Collator *fromUCollator(const UCollator *uc) {
   1158         return reinterpret_cast<const Collator *>(uc);
   1159     }
   1160     /** @internal */
   1161     inline UCollator *toUCollator() {
   1162         return reinterpret_cast<UCollator *>(this);
   1163     }
   1164     /** @internal */
   1165     inline const UCollator *toUCollator() const {
   1166         return reinterpret_cast<const UCollator *>(this);
   1167     }
   1168 #endif  // U_HIDE_INTERNAL_API
   1169 
   1170 private:
   1171     /**
   1172      * Assignment operator. Private for now.
   1173      */
   1174     Collator& operator=(const Collator& other);
   1175 
   1176     friend class CFactory;
   1177     friend class SimpleCFactory;
   1178     friend class ICUCollatorFactory;
   1179     friend class ICUCollatorService;
   1180     static Collator* makeInstance(const Locale& desiredLocale,
   1181                                   UErrorCode& status);
   1182 };
   1183 
   1184 #if !UCONFIG_NO_SERVICE
   1185 /**
   1186  * A factory, used with registerFactory, the creates multiple collators and provides
   1187  * display names for them.  A factory supports some number of locales-- these are the
   1188  * locales for which it can create collators.  The factory can be visible, in which
   1189  * case the supported locales will be enumerated by getAvailableLocales, or invisible,
   1190  * in which they are not.  Invisible locales are still supported, they are just not
   1191  * listed by getAvailableLocales.
   1192  * <p>
   1193  * If standard locale display names are sufficient, Collator instances can
   1194  * be registered using registerInstance instead.</p>
   1195  * <p>
   1196  * Note: if the collators are to be used from C APIs, they must be instances
   1197  * of RuleBasedCollator.</p>
   1198  *
   1199  * @stable ICU 2.6
   1200  */
   1201 class U_I18N_API CollatorFactory : public UObject {
   1202 public:
   1203 
   1204     /**
   1205      * Destructor
   1206      * @stable ICU 3.0
   1207      */
   1208     virtual ~CollatorFactory();
   1209 
   1210     /**
   1211      * Return true if this factory is visible.  Default is true.
   1212      * If not visible, the locales supported by this factory will not
   1213      * be listed by getAvailableLocales.
   1214      * @return true if the factory is visible.
   1215      * @stable ICU 2.6
   1216      */
   1217     virtual UBool visible(void) const;
   1218 
   1219     /**
   1220      * Return a collator for the provided locale.  If the locale
   1221      * is not supported, return NULL.
   1222      * @param loc the locale identifying the collator to be created.
   1223      * @return a new collator if the locale is supported, otherwise NULL.
   1224      * @stable ICU 2.6
   1225      */
   1226     virtual Collator* createCollator(const Locale& loc) = 0;
   1227 
   1228     /**
   1229      * Return the name of the collator for the objectLocale, localized for the displayLocale.
   1230      * If objectLocale is not supported, or the factory is not visible, set the result string
   1231      * to bogus.
   1232      * @param objectLocale the locale identifying the collator
   1233      * @param displayLocale the locale for which the display name of the collator should be localized
   1234      * @param result an output parameter for the display name, set to bogus if not supported.
   1235      * @return the display name
   1236      * @stable ICU 2.6
   1237      */
   1238     virtual  UnicodeString& getDisplayName(const Locale& objectLocale,
   1239                                            const Locale& displayLocale,
   1240                                            UnicodeString& result);
   1241 
   1242     /**
   1243      * Return an array of all the locale names directly supported by this factory.
   1244      * The number of names is returned in count.  This array is owned by the factory.
   1245      * Its contents must never change.
   1246      * @param count output parameter for the number of locales supported by the factory
   1247      * @param status the in/out error code
   1248      * @return a pointer to an array of count UnicodeStrings.
   1249      * @stable ICU 2.6
   1250      */
   1251     virtual const UnicodeString * getSupportedIDs(int32_t &count, UErrorCode& status) = 0;
   1252 };
   1253 #endif /* UCONFIG_NO_SERVICE */
   1254 
   1255 // Collator inline methods -----------------------------------------------
   1256 
   1257 U_NAMESPACE_END
   1258 
   1259 #endif /* #if !UCONFIG_NO_COLLATION */
   1260 
   1261 #endif
   1262