icuSources/i18n/unicode/plurrule.h

   1 /*
   2 *******************************************************************************
   3 * Copyright (C) 2008-2015, 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 FixedDecimal;
  42 class RuleChain;
  43 class PluralRuleParser;
  44 class PluralKeywordEnumeration;
  45 class AndConstraint;
  46 class SharedPluralRules;
  47
  48 /**
  49  * Defines rules for mapping non-negative numeric values onto a small set of
  50  * keywords. Rules are constructed from a text description, consisting
  51  * of a series of keywords and conditions.  The {@link #select} method
  52  * examines each condition in order and returns the keyword for the
  53  * first condition that matches the number.  If none match,
  54  * default rule(other) is returned.
  55  *
  56  * For more information, details, and tips for writing rules, see the
  57  * LDML spec, C.11 Language Plural Rules:
  58  * http://www.unicode.org/draft/reports/tr35/tr35.html#Language_Plural_Rules
  59  *
  60  * Examples:<pre>
  61  *   "one: n is 1; few: n in 2..4"</pre>
  62  *  This defines two rules, for 'one' and 'few'.  The condition for
  63  *  'one' is "n is 1" which means that the number must be equal to
  64  *  1 for this condition to pass.  The condition for 'few' is
  65  *  "n in 2..4" which means that the number must be between 2 and
  66  *  4 inclusive for this condition to pass.  All other numbers
  67  *  are assigned the keyword "other" by the default rule.
  68  *  </p><pre>
  69  *    "zero: n is 0; one: n is 1; zero: n mod 100 in 1..19"</pre>
  70  *  This illustrates that the same keyword can be defined multiple times.
  71  *  Each rule is examined in order, and the first keyword whose condition
  72  *  passes is the one returned.  Also notes that a modulus is applied
  73  *  to n in the last rule.  Thus its condition holds for 119, 219, 319...
  74  *  </p><pre>
  75  *    "one: n is 1; few: n mod 10 in 2..4 and n mod 100 not in 12..14"</pre>
  76  *  This illustrates conjunction and negation.  The condition for 'few'
  77  *  has two parts, both of which must be met: "n mod 10 in 2..4" and
  78  *  "n mod 100 not in 12..14".  The first part applies a modulus to n
  79  *  before the test as in the previous example.  The second part applies
  80  *  a different modulus and also uses negation, thus it matches all
  81  *  numbers _not_ in 12, 13, 14, 112, 113, 114, 212, 213, 214...
  82  *  </p>
  83  *  <p>
  84  * Syntax:<pre>
  85  * \code
  86  * rules         = rule (';' rule)*
  87  * rule          = keyword ':' condition
  88  * keyword       = <identifier>
  89  * condition     = and_condition ('or' and_condition)*
  90  * and_condition = relation ('and' relation)*
  91  * relation      = is_relation | in_relation | within_relation | 'n' <EOL>
  92  * is_relation   = expr 'is' ('not')? value
  93  * in_relation   = expr ('not')? 'in' range_list
  94  * within_relation = expr ('not')? 'within' range
  95  * expr          = ('n' | 'i' | 'f' | 'v' | 'j') ('mod' value)?
  96  * range_list    = (range | value) (',' range_list)*
  97  * value         = digit+  ('.' digit+)?
  98  * digit         = 0|1|2|3|4|5|6|7|8|9
  99  * range         = value'..'value
 100  * \endcode
 101  * </pre></p>
 102  * <p>
 103  * <p>
 104  * The i, f, and v values are defined as follows:
 105  * </p>
 106  * <ul>
 107  * <li>i to be the integer digits.</li>
 108  * <li>f to be the visible fractional digits, as an integer.</li>
 109  * <li>v to be the number of visible fraction digits.</li>
 110  * <li>j is defined to only match integers. That is j is 3 fails if v != 0 (eg for 3.1 or 3.0).</li>
 111  * </ul>
 112  * <p>
 113  * Examples are in the following table:
 114  * </p>
 115  * <table border='1' style="border-collapse:collapse">
 116  * <tbody>
 117  * <tr>
 118  * <th>n</th>
 119  * <th>i</th>
 120  * <th>f</th>
 121  * <th>v</th>
 122  * </tr>
 123  * <tr>
 124  * <td>1.0</td>
 125  * <td>1</td>
 126  * <td align="right">0</td>
 127  * <td>1</td>
 128  * </tr>
 129  * <tr>
 130  * <td>1.00</td>
 131  * <td>1</td>
 132  * <td align="right">0</td>
 133  * <td>2</td>
 134  * </tr>
 135  * <tr>
 136  * <td>1.3</td>
 137  * <td>1</td>
 138  * <td align="right">3</td>
 139  * <td>1</td>
 140  * </tr>
 141  * <tr>
 142  * <td>1.03</td>
 143  * <td>1</td>
 144  * <td align="right">3</td>
 145  * <td>2</td>
 146  * </tr>
 147  * <tr>
 148  * <td>1.23</td>
 149  * <td>1</td>
 150  * <td align="right">23</td>
 151  * <td>2</td>
 152  * </tr>
 153  * </tbody>
 154  * </table>
 155  * <p>
 156  * The difference between 'in' and 'within' is that 'in' only includes integers in the specified range, while 'within'
 157  * includes all values. Using 'within' with a range_list consisting entirely of values is the same as using 'in' (it's
 158  * not an error).
 159  * </p>
 160
 161  * An "identifier" is a sequence of characters that do not have the
 162  * Unicode Pattern_Syntax or Pattern_White_Space properties.
 163  * <p>
 164  * The difference between 'in' and 'within' is that 'in' only includes
 165  * integers in the specified range, while 'within' includes all values.
 166  * Using 'within' with a range_list consisting entirely of values is the
 167  * same as using 'in' (it's not an error).
 168  *</p>
 169  * <p>
 170  * Keywords
 171  * could be defined by users or from ICU locale data. There are 6
 172  * predefined values in ICU - 'zero', 'one', 'two', 'few', 'many' and
 173  * 'other'. Callers need to check the value of keyword returned by
 174  * {@link #select} method.
 175  * </p>
 176  *
 177  * Examples:<pre>
 178  * UnicodeString keyword = pl->select(number);
 179  * if (keyword== UnicodeString("one") {
 180  *     ...
 181  * }
 182  * else if ( ... )
 183  * </pre>
 184  * <strong>Note:</strong><br>
 185  *  <p>
 186  *   ICU defines plural rules for many locales based on CLDR <i>Language Plural Rules</i>.
 187  *   For these predefined rules, see CLDR page at
 188  *    http://unicode.org/repos/cldr-tmp/trunk/diff/supplemental/language_plural_rules.html
 189  * </p>
 190  */
 191 class U_I18N_API PluralRules : public UObject {
 192 public:
 193
 194     /**
 195      * Constructor.
 196      * @param status  Output param set to success/failure code on exit, which
 197      *                must not indicate a failure before the function call.
 198      *
 199      * @stable ICU 4.0
 200      */
 201     PluralRules(UErrorCode& status);
 202
 203     /**
 204      * Copy constructor.
 205      * @stable ICU 4.0
 206      */
 207     PluralRules(const PluralRules& other);
 208
 209     /**
 210      * Destructor.
 211      * @stable ICU 4.0
 212      */
 213     virtual ~PluralRules();
 214
 215     /**
 216      * Clone
 217      * @stable ICU 4.0
 218      */
 219     PluralRules* clone() const;
 220
 221     /**
 222       * Assignment operator.
 223       * @stable ICU 4.0
 224       */
 225     PluralRules& operator=(const PluralRules&);
 226
 227     /**
 228      * Creates a PluralRules from a description if it is parsable, otherwise
 229      * returns NULL.
 230      *
 231      * @param description rule description
 232      * @param status      Output param set to success/failure code on exit, which
 233      *                    must not indicate a failure before the function call.
 234      * @return            new PluralRules pointer. NULL if there is an error.
 235      * @stable ICU 4.0
 236      */
 237     static PluralRules* U_EXPORT2 createRules(const UnicodeString& description,
 238                                               UErrorCode& status);
 239
 240     /**
 241      * The default rules that accept any number.
 242      *
 243      * @param status  Output param set to success/failure code on exit, which
 244      *                must not indicate a failure before the function call.
 245      * @return        new PluralRules pointer. NULL if there is an error.
 246      * @stable ICU 4.0
 247      */
 248     static PluralRules* U_EXPORT2 createDefaultRules(UErrorCode& status);
 249
 250     /**
 251      * Provides access to the predefined cardinal-number <code>PluralRules</code> for a given
 252      * locale.
 253      * Same as forLocale(locale, UPLURAL_TYPE_CARDINAL, status).
 254      *
 255      * @param locale  The locale for which a <code>PluralRules</code> object is
 256      *                returned.
 257      * @param status  Output param set to success/failure code on exit, which
 258      *                must not indicate a failure before the function call.
 259      * @return        The predefined <code>PluralRules</code> object pointer for
 260      *                this locale. If there's no predefined rules for this locale,
 261      *                the rules for the closest parent in the locale hierarchy
 262      *                that has one will  be returned.  The final fallback always
 263      *                returns the default 'other' rules.
 264      * @stable ICU 4.0
 265      */
 266     static PluralRules* U_EXPORT2 forLocale(const Locale& locale, UErrorCode& status);
 267
 268     /**
 269      * Provides access to the predefined <code>PluralRules</code> for a given
 270      * locale and the plural type.
 271      *
 272      * @param locale  The locale for which a <code>PluralRules</code> object is
 273      *                returned.
 274      * @param type    The plural type (e.g., cardinal or ordinal).
 275      * @param status  Output param set to success/failure code on exit, which
 276      *                must not indicate a failure before the function call.
 277      * @return        The predefined <code>PluralRules</code> object pointer for
 278      *                this locale. If there's no predefined rules for this locale,
 279      *                the rules for the closest parent in the locale hierarchy
 280      *                that has one will  be returned.  The final fallback always
 281      *                returns the default 'other' rules.
 282      * @stable ICU 50
 283      */
 284     static PluralRules* U_EXPORT2 forLocale(const Locale& locale, UPluralType type, UErrorCode& status);
 285
 286 #ifndef U_HIDE_INTERNAL_API
 287     /**
 288      * Return a StringEnumeration over the locales for which there is plurals data.
 289      * @return a StringEnumeration over the locales available.
 290      * @internal
 291      */
 292     static StringEnumeration* U_EXPORT2 getAvailableLocales(UErrorCode &status);
 293
 294     /**
 295      * Returns whether or not there are overrides.
 296      * @param locale       the locale to check.
 297      * @return
 298      * @internal
 299      */
 300     static UBool hasOverride(const Locale &locale);
 301
 302     /**
 303      * For ICU use only.
 304      * creates a  SharedPluralRules object
 305      * @internal
 306      */
 307     static PluralRules* U_EXPORT2 internalForLocale(const Locale& locale, UPluralType type, UErrorCode& status);
 308
 309     /**
 310      * For ICU use only.
 311      * Returns handle to the shared, cached PluralRules instance.
 312      * Caller must call removeRef() on returned value once it is done with
 313      * the shared instance.
 314      * @internal
 315      */
 316     static const SharedPluralRules* U_EXPORT2 createSharedInstance(
 317             const Locale& locale, UPluralType type, UErrorCode& status);
 318
 319
 320 #endif  /* U_HIDE_INTERNAL_API */
 321
 322     /**
 323      * Given a number, returns the keyword of the first rule that applies to
 324      * the number.  This function can be used with isKeyword* functions to
 325      * determine the keyword for default plural rules.
 326      *
 327      * @param number  The number for which the rule has to be determined.
 328      * @return        The keyword of the selected rule.
 329      * @stable ICU 4.0
 330      */
 331     UnicodeString select(int32_t number) const;
 332
 333     /**
 334      * Given a number, returns the keyword of the first rule that applies to
 335      * the number.  This function can be used with isKeyword* functions to
 336      * determine the keyword for default plural rules.
 337      *
 338      * @param number  The number for which the rule has to be determined.
 339      * @return        The keyword of the selected rule.
 340      * @stable ICU 4.0
 341      */
 342     UnicodeString select(double number) const;
 343
 344 #ifndef U_HIDE_INTERNAL_API
 345     /**
 346       * @internal
 347       */
 348     UnicodeString select(const FixedDecimal &number) const;
 349 #endif  /* U_HIDE_INTERNAL_API */
 350
 351     /**
 352      * Returns a list of all rule keywords used in this <code>PluralRules</code>
 353      * object.  The rule 'other' is always present by default.
 354      *
 355      * @param status Output param set to success/failure code on exit, which
 356      *               must not indicate a failure before the function call.
 357      * @return       StringEnumeration with the keywords.
 358      *               The caller must delete the object.
 359      * @stable ICU 4.0
 360      */
 361     StringEnumeration* getKeywords(UErrorCode& status) const;
 362
 363 #ifndef U_HIDE_DEPRECATED_API
 364     /**
 365      * Deprecated Function, does not return useful results.
 366      *
 367      * Originally intended to return a unique value for this keyword if it exists,
 368      * else the constant UPLRULES_NO_UNIQUE_VALUE.
 369      *
 370      * @param keyword The keyword.
 371      * @return        Stub deprecated function returns UPLRULES_NO_UNIQUE_VALUE always.
 372      * @deprecated ICU 55
 373      */
 374     double getUniqueKeywordValue(const UnicodeString& keyword);
 375
 376     /**
 377      * Deprecated Function, does not produce useful results.
 378      *
 379      * Orginally intended to return all the values for which select() would return the keyword.
 380      * If the keyword is unknown, returns no values, but this is not an error.  If
 381      * the number of values is unlimited, returns no values and -1 as the
 382      * count.
 383      *
 384      * The number of returned values is typically small.
 385      *
 386      * @param keyword      The keyword.
 387      * @param dest         Array into which to put the returned values.  May
 388      *                     be NULL if destCapacity is 0.
 389      * @param destCapacity The capacity of the array, must be at least 0.
 390      * @param status       The error code. Deprecated function, always sets U_UNSUPPORTED_ERROR.
 391      * @return             The count of values available, or -1.  This count
 392      *                     can be larger than destCapacity, but no more than
 393      *                     destCapacity values will be written.
 394      * @deprecated ICU 55
 395      */
 396     int32_t getAllKeywordValues(const UnicodeString &keyword,
 397                                 double *dest, int32_t destCapacity,
 398                                 UErrorCode& status);
 399 #endif  /* U_HIDE_DEPRECATED_API */
 400
 401     /**
 402      * Returns sample values for which select() would return the keyword.  If
 403      * the keyword is unknown, returns no values, but this is not an error.
 404      *
 405      * The number of returned values is typically small.
 406      *
 407      * @param keyword      The keyword.
 408      * @param dest         Array into which to put the returned values.  May
 409      *                     be NULL if destCapacity is 0.
 410      * @param destCapacity The capacity of the array, must be at least 0.
 411      * @param status       The error code.
 412      * @return             The count of values written.
 413      *                     If more than destCapacity samples are available, then
 414      *                     only destCapacity are written, and destCapacity is returned as the count,
 415      *                     rather than setting a U_BUFFER_OVERFLOW_ERROR.
 416      *                     (The actual number of keyword values could be unlimited.)
 417      * @stable ICU 4.8
 418      */
 419     int32_t getSamples(const UnicodeString &keyword,
 420                        double *dest, int32_t destCapacity,
 421                        UErrorCode& status);
 422
 423     /**
 424      * Returns TRUE if the given keyword is defined in this
 425      * <code>PluralRules</code> object.
 426      *
 427      * @param keyword  the input keyword.
 428      * @return         TRUE if the input keyword is defined.
 429      *                 Otherwise, return FALSE.
 430      * @stable ICU 4.0
 431      */
 432     UBool isKeyword(const UnicodeString& keyword) const;
 433
 434
 435     /**
 436      * Returns keyword for default plural form.
 437      *
 438      * @return         keyword for default plural form.
 439      * @stable ICU 4.0
 440      */
 441     UnicodeString getKeywordOther() const;
 442
 443 #ifndef U_HIDE_INTERNAL_API
 444     /**
 445      *
 446      * @internal
 447      */
 448      UnicodeString getRules() const;
 449 #endif  /* U_HIDE_INTERNAL_API */
 450
 451     /**
 452      * Compares the equality of two PluralRules objects.
 453      *
 454      * @param other The other PluralRules object to be compared with.
 455      * @return      True if the given PluralRules is the same as this
 456      *              PluralRules; false otherwise.
 457      * @stable ICU 4.0
 458      */
 459     virtual UBool operator==(const PluralRules& other) const;
 460
 461     /**
 462      * Compares the inequality of two PluralRules objects.
 463      *
 464      * @param other The PluralRules object to be compared with.
 465      * @return      True if the given PluralRules is not the same as this
 466      *              PluralRules; false otherwise.
 467      * @stable ICU 4.0
 468      */
 469     UBool operator!=(const PluralRules& other) const  {return !operator==(other);}
 470
 471
 472     /**
 473      * ICU "poor man's RTTI", returns a UClassID for this class.
 474      *
 475      * @stable ICU 4.0
 476      *
 477     */
 478     static UClassID U_EXPORT2 getStaticClassID(void);
 479
 480     /**
 481      * ICU "poor man's RTTI", returns a UClassID for the actual class.
 482      *
 483      * @stable ICU 4.0
 484      */
 485     virtual UClassID getDynamicClassID() const;
 486
 487
 488 private:
 489     RuleChain  *mRules;
 490
 491     PluralRules();   // default constructor not implemented
 492     void            parseDescription(const UnicodeString& ruleData, UErrorCode &status);
 493     int32_t         getNumberValue(const UnicodeString& token) const;
 494     UnicodeString   getRuleFromResource(const Locale& locale, UPluralType type, UErrorCode& status);
 495     RuleChain      *rulesForKeyword(const UnicodeString &keyword) const;
 496
 497     friend class PluralRuleParser;
 498 };
 499
 500 U_NAMESPACE_END
 501
 502 #endif /* #if !UCONFIG_NO_FORMATTING */
 503
 504 #endif // _PLURRULE
 505 //eof