icuSources/i18n/unicode/plurrule.h

   1 /*
   2 *******************************************************************************
   3 * Copyright (C) 2008-2013, International Business Machines Corporation and
   4 * others. All Rights Reserved.
   5 *******************************************************************************
   6 *
   7 *
   8 * File PLURRULE.H
   9 *
  10 * Modification History:*
  11 *   Date        Name        Description
  12 *
  13 ********************************************************************************
  14 */
  15
  16 #ifndef PLURRULE
  17 #define PLURRULE
  18
  19 #include "unicode/utypes.h"
  20
  21 /**
  22  * \file
  23  * \brief C++ API: PluralRules object
  24  */
  25
  26 #if !UCONFIG_NO_FORMATTING
  27
  28 #include "unicode/format.h"
  29 #include "unicode/upluralrules.h"
  30
  31 /**
  32  * Value returned by PluralRules::getUniqueKeywordValue() when there is no
  33  * unique value to return.
  34  * @stable ICU 4.8
  35  */
  36 #define UPLRULES_NO_UNIQUE_VALUE ((double)-0.00123456777)
  37
  38 U_NAMESPACE_BEGIN
  39
  40 class Hashtable;
  41 class RuleChain;
  42 class RuleParser;
  43 class PluralKeywordEnumeration;
  44
  45 /**
  46  * Defines rules for mapping non-negative numeric values onto a small set of
  47  * keywords. Rules are constructed from a text description, consisting
  48  * of a series of keywords and conditions.  The {@link #select} method
  49  * examines each condition in order and returns the keyword for the
  50  * first condition that matches the number.  If none match,
  51  * default rule(other) is returned.
  52  *
  53  * For more information, details, and tips for writing rules, see the
  54  * LDML spec, C.11 Language Plural Rules:
  55  * http://www.unicode.org/draft/reports/tr35/tr35.html#Language_Plural_Rules
  56  *
  57  * Examples:<pre>
  58  *   "one: n is 1; few: n in 2..4"</pre>
  59  *  This defines two rules, for 'one' and 'few'.  The condition for
  60  *  'one' is "n is 1" which means that the number must be equal to
  61  *  1 for this condition to pass.  The condition for 'few' is
  62  *  "n in 2..4" which means that the number must be between 2 and
  63  *  4 inclusive for this condition to pass.  All other numbers
  64  *  are assigned the keyword "other" by the default rule.
  65  *  </p><pre>
  66  *    "zero: n is 0; one: n is 1; zero: n mod 100 in 1..19"</pre>
  67  *  This illustrates that the same keyword can be defined multiple times.
  68  *  Each rule is examined in order, and the first keyword whose condition
  69  *  passes is the one returned.  Also notes that a modulus is applied
  70  *  to n in the last rule.  Thus its condition holds for 119, 219, 319...
  71  *  </p><pre>
  72  *    "one: n is 1; few: n mod 10 in 2..4 and n mod 100 not in 12..14"</pre>
  73  *  This illustrates conjunction and negation.  The condition for 'few'
  74  *  has two parts, both of which must be met: "n mod 10 in 2..4" and
  75  *  "n mod 100 not in 12..14".  The first part applies a modulus to n
  76  *  before the test as in the previous example.  The second part applies
  77  *  a different modulus and also uses negation, thus it matches all
  78  *  numbers _not_ in 12, 13, 14, 112, 113, 114, 212, 213, 214...
  79  *  </p>
  80  *  <p>
  81  * Syntax:<pre>
  82  * \code
  83  * rules         = rule (';' rule)*
  84  * rule          = keyword ':' condition
  85  * keyword       = <identifier>
  86  * condition     = and_condition ('or' and_condition)*
  87  * and_condition = relation ('and' relation)*
  88  * relation      = is_relation | in_relation | within_relation | 'n' <EOL>
  89  * is_relation   = expr 'is' ('not')? value
  90  * in_relation   = expr ('not')? 'in' range_list
  91  * within_relation = expr ('not')? 'within' range
  92  * expr          = 'n' ('mod' value)?
  93  * range_list    = (range | value) (',' range_list)*
  94  * value         = digit+
  95  * digit         = 0|1|2|3|4|5|6|7|8|9
  96  * range         = value'..'value
  97  * \endcode
  98  * </pre></p>
  99  * <p>
 100  * An "identifier" is a sequence of characters that do not have the
 101  * Unicode Pattern_Syntax or Pattern_White_Space properties.
 102  * <p>
 103  * The difference between 'in' and 'within' is that 'in' only includes
 104  * integers in the specified range, while 'within' includes all values.</p>
 105  * <p>
 106  * Keywords
 107  * could be defined by users or from ICU locale data. There are 6
 108  * predefined values in ICU - 'zero', 'one', 'two', 'few', 'many' and
 109  * 'other'. Callers need to check the value of keyword returned by
 110  * {@link #select} method.
 111  * </p>
 112  *
 113  * Examples:<pre>
 114  * UnicodeString keyword = pl->select(number);
 115  * if (keyword== UnicodeString("one") {
 116  *     ...
 117  * }
 118  * else if ( ... )
 119  * </pre>
 120  * <strong>Note:</strong><br>
 121  *  <p>
 122  *   ICU defines plural rules for many locales based on CLDR <i>Language Plural Rules</i>.
 123  *   For these predefined rules, see CLDR page at
 124  *    http://unicode.org/repos/cldr-tmp/trunk/diff/supplemental/language_plural_rules.html
 125  * </p>
 126  */
 127 class U_I18N_API PluralRules : public UObject {
 128 public:
 129
 130     /**
 131      * Constructor.
 132      * @param status  Output param set to success/failure code on exit, which
 133      *                must not indicate a failure before the function call.
 134      *
 135      * @stable ICU 4.0
 136      */
 137     PluralRules(UErrorCode& status);
 138
 139     /**
 140      * Copy constructor.
 141      * @stable ICU 4.0
 142      */
 143     PluralRules(const PluralRules& other);
 144
 145     /**
 146      * Destructor.
 147      * @stable ICU 4.0
 148      */
 149     virtual ~PluralRules();
 150
 151     /**
 152      * Clone
 153      * @stable ICU 4.0
 154      */
 155     PluralRules* clone() const;
 156
 157     /**
 158       * Assignment operator.
 159       * @stable ICU 4.0
 160       */
 161     PluralRules& operator=(const PluralRules&);
 162
 163     /**
 164      * Creates a PluralRules from a description if it is parsable, otherwise
 165      * returns NULL.
 166      *
 167      * @param description rule description
 168      * @param status      Output param set to success/failure code on exit, which
 169      *                    must not indicate a failure before the function call.
 170      * @return            new PluralRules pointer. NULL if there is an error.
 171      * @stable ICU 4.0
 172      */
 173     static PluralRules* U_EXPORT2 createRules(const UnicodeString& description,
 174                                               UErrorCode& status);
 175
 176     /**
 177      * The default rules that accept any number.
 178      *
 179      * @param status  Output param set to success/failure code on exit, which
 180      *                must not indicate a failure before the function call.
 181      * @return        new PluralRules pointer. NULL if there is an error.
 182      * @stable ICU 4.0
 183      */
 184     static PluralRules* U_EXPORT2 createDefaultRules(UErrorCode& status);
 185
 186     /**
 187      * Provides access to the predefined cardinal-number <code>PluralRules</code> for a given
 188      * locale.
 189      * Same as forLocale(locale, UPLURAL_TYPE_CARDINAL, status).
 190      *
 191      * @param locale  The locale for which a <code>PluralRules</code> object is
 192      *                returned.
 193      * @param status  Output param set to success/failure code on exit, which
 194      *                must not indicate a failure before the function call.
 195      * @return        The predefined <code>PluralRules</code> object pointer for
 196      *                this locale. If there's no predefined rules for this locale,
 197      *                the rules for the closest parent in the locale hierarchy
 198      *                that has one will  be returned.  The final fallback always
 199      *                returns the default 'other' rules.
 200      * @stable ICU 4.0
 201      */
 202     static PluralRules* U_EXPORT2 forLocale(const Locale& locale, UErrorCode& status);
 203
 204 #ifndef U_HIDE_DRAFT_API
 205     /**
 206      * Provides access to the predefined <code>PluralRules</code> for a given
 207      * locale and the plural type.
 208      *
 209      * @param locale  The locale for which a <code>PluralRules</code> object is
 210      *                returned.
 211      * @param type    The plural type (e.g., cardinal or ordinal).
 212      * @param status  Output param set to success/failure code on exit, which
 213      *                must not indicate a failure before the function call.
 214      * @return        The predefined <code>PluralRules</code> object pointer for
 215      *                this locale. If there's no predefined rules for this locale,
 216      *                the rules for the closest parent in the locale hierarchy
 217      *                that has one will  be returned.  The final fallback always
 218      *                returns the default 'other' rules.
 219      * @draft ICU 50
 220      */
 221     static PluralRules* U_EXPORT2 forLocale(const Locale& locale, UPluralType type, UErrorCode& status);
 222 #endif /* U_HIDE_DRAFT_API */
 223
 224     /**
 225      * Given a number, returns the keyword of the first rule that applies to
 226      * the number.  This function can be used with isKeyword* functions to
 227      * determine the keyword for default plural rules.
 228      *
 229      * @param number  The number for which the rule has to be determined.
 230      * @return        The keyword of the selected rule.
 231      * @stable ICU 4.0
 232      */
 233     UnicodeString select(int32_t number) const;
 234
 235     /**
 236      * Given a number, returns the keyword of the first rule that applies to
 237      * the number.  This function can be used with isKeyword* functions to
 238      * determine the keyword for default plural rules.
 239      *
 240      * @param number  The number for which the rule has to be determined.
 241      * @return        The keyword of the selected rule.
 242      * @stable ICU 4.0
 243      */
 244     UnicodeString select(double number) const;
 245
 246     /**
 247      * Returns a list of all rule keywords used in this <code>PluralRules</code>
 248      * object.  The rule 'other' is always present by default.
 249      *
 250      * @param status Output param set to success/failure code on exit, which
 251      *               must not indicate a failure before the function call.
 252      * @return       StringEnumeration with the keywords.
 253      *               The caller must delete the object.
 254      * @stable ICU 4.0
 255      */
 256     StringEnumeration* getKeywords(UErrorCode& status) const;
 257
 258     /**
 259      * Returns a unique value for this keyword if it exists, else the constant
 260      * UPLRULES_NO_UNIQUE_VALUE.
 261      *
 262      * @param keyword The keyword.
 263      * @return        The unique value that generates the keyword, or
 264      *                UPLRULES_NO_UNIQUE_VALUE if the keyword is undefined or there is no
 265      *                unique value that generates this keyword.
 266      * @stable ICU 4.8
 267      */
 268     double getUniqueKeywordValue(const UnicodeString& keyword);
 269
 270     /**
 271      * Returns all the values for which select() would return the keyword.  If
 272      * the keyword is unknown, returns no values, but this is not an error.  If
 273      * the number of values is unlimited, returns no values and -1 as the
 274      * count.
 275      *
 276      * The number of returned values is typically small.
 277      *
 278      * @param keyword      The keyword.
 279      * @param dest         Array into which to put the returned values.  May
 280      *                     be NULL if destCapacity is 0.
 281      * @param destCapacity The capacity of the array, must be at least 0.
 282      * @param status       The error code.
 283      * @return             The count of values available, or -1.  This count
 284      *                     can be larger than destCapacity, but no more than
 285      *                     destCapacity values will be written.
 286      * @stable ICU 4.8
 287      */
 288     int32_t getAllKeywordValues(const UnicodeString &keyword,
 289                                 double *dest, int32_t destCapacity,
 290                                 UErrorCode& status);
 291
 292     /**
 293      * Returns sample values for which select() would return the keyword.  If
 294      * the keyword is unknown, returns no values, but this is not an error.
 295      *
 296      * The number of returned values is typically small.
 297      *
 298      * @param keyword      The keyword.
 299      * @param dest         Array into which to put the returned values.  May
 300      *                     be NULL if destCapacity is 0.
 301      * @param destCapacity The capacity of the array, must be at least 0.
 302      * @param status       The error code.
 303      * @return             The count of values written.
 304      *                     If more than destCapacity samples are available, then
 305      *                     only destCapacity are written, and destCapacity is returned as the count,
 306      *                     rather than setting a U_BUFFER_OVERFLOW_ERROR.
 307      *                     (The actual number of keyword values could be unlimited.)
 308      * @stable ICU 4.8
 309      */
 310     int32_t getSamples(const UnicodeString &keyword,
 311                        double *dest, int32_t destCapacity,
 312                        UErrorCode& status);
 313
 314     /**
 315      * Returns TRUE if the given keyword is defined in this
 316      * <code>PluralRules</code> object.
 317      *
 318      * @param keyword  the input keyword.
 319      * @return         TRUE if the input keyword is defined.
 320      *                 Otherwise, return FALSE.
 321      * @stable ICU 4.0
 322      */
 323     UBool isKeyword(const UnicodeString& keyword) const;
 324
 325
 326     /**
 327      * Returns keyword for default plural form.
 328      *
 329      * @return         keyword for default plural form.
 330      * @stable ICU 4.0
 331      */
 332     UnicodeString getKeywordOther() const;
 333
 334     /**
 335      * Compares the equality of two PluralRules objects.
 336      *
 337      * @param other The other PluralRules object to be compared with.
 338      * @return      True if the given PluralRules is the same as this
 339      *              PluralRules; false otherwise.
 340      * @stable ICU 4.0
 341      */
 342     virtual UBool operator==(const PluralRules& other) const;
 343
 344     /**
 345      * Compares the inequality of two PluralRules objects.
 346      *
 347      * @param other The PluralRules object to be compared with.
 348      * @return      True if the given PluralRules is not the same as this
 349      *              PluralRules; false otherwise.
 350      * @stable ICU 4.0
 351      */
 352     UBool operator!=(const PluralRules& other) const  {return !operator==(other);}
 353
 354
 355     /**
 356      * ICU "poor man's RTTI", returns a UClassID for this class.
 357      *
 358      * @stable ICU 4.0
 359      *
 360     */
 361     static UClassID U_EXPORT2 getStaticClassID(void);
 362
 363     /**
 364      * ICU "poor man's RTTI", returns a UClassID for the actual class.
 365      *
 366      * @stable ICU 4.0
 367      */
 368     virtual UClassID getDynamicClassID() const;
 369
 370
 371 private:
 372     RuleChain  *mRules;
 373     RuleParser *mParser;
 374     double     *mSamples;
 375     int32_t    *mSampleInfo;
 376     int32_t    mSampleInfoCount;
 377
 378     PluralRules();   // default constructor not implemented
 379     int32_t getRepeatLimit() const;
 380     void parseDescription(UnicodeString& ruleData, RuleChain& rules, UErrorCode &status);
 381     void getNextLocale(const UnicodeString& localeData, int32_t* curIndex, UnicodeString& localeName);
 382     void addRules(RuleChain& rules);
 383     int32_t getNumberValue(const UnicodeString& token) const;
 384     UnicodeString getRuleFromResource(const Locale& locale, UPluralType type, UErrorCode& status);
 385
 386     static const int32_t MAX_SAMPLES = 3;
 387
 388     int32_t getSamplesInternal(const UnicodeString &keyword, double *dest,
 389                                int32_t destCapacity, UBool includeUnlimited,
 390                                UErrorCode& status);
 391     int32_t getKeywordIndex(const UnicodeString& keyword,
 392                             UErrorCode& status) const;
 393     void initSamples(UErrorCode& status);
 394
 395 };
 396
 397 U_NAMESPACE_END
 398
 399 #endif /* #if !UCONFIG_NO_FORMATTING */
 400
 401 #endif // _PLURRULE
 402 //eof