icuSources/i18n/unicode/plurrule.h

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 *******************************************************************************
   5 * Copyright (C) 2008-2015, International Business Machines Corporation and
   6 * others. All Rights Reserved.
   7 *******************************************************************************
   8 *
   9 *
  10 * File PLURRULE.H
  11 *
  12 * Modification History:*
  13 *   Date        Name        Description
  14 *
  15 ********************************************************************************
  16 */
  17
  18 #ifndef PLURRULE
  19 #define PLURRULE
  20
  21 #include "unicode/utypes.h"
  22
  23 /**
  24  * \file
  25  * \brief C++ API: PluralRules object
  26  */
  27
  28 #if !UCONFIG_NO_FORMATTING
  29
  30 #include "unicode/format.h"
  31 #include "unicode/upluralrules.h"
  32 #ifndef U_HIDE_INTERNAL_API
  33 #include "unicode/numfmt.h"
  34 #endif  /* U_HIDE_INTERNAL_API */
  35
  36 /**
  37  * Value returned by PluralRules::getUniqueKeywordValue() when there is no
  38  * unique value to return.
  39  * @stable ICU 4.8
  40  */
  41 #define UPLRULES_NO_UNIQUE_VALUE ((double)-0.00123456777)
  42
  43 #if U_SHOW_CPLUSPLUS_API
  44 U_NAMESPACE_BEGIN
  45
  46 class Hashtable;
  47 class FixedDecimal;
  48 class VisibleDigitsWithExponent;
  49 class RuleChain;
  50 class PluralRuleParser;
  51 class PluralKeywordEnumeration;
  52 class AndConstraint;
  53 class SharedPluralRules;
  54
  55 /**
  56  * Defines rules for mapping non-negative numeric values onto a small set of
  57  * keywords. Rules are constructed from a text description, consisting
  58  * of a series of keywords and conditions.  The {@link #select} method
  59  * examines each condition in order and returns the keyword for the
  60  * first condition that matches the number.  If none match,
  61  * default rule(other) is returned.
  62  *
  63  * For more information, details, and tips for writing rules, see the
  64  * LDML spec, C.11 Language Plural Rules:
  65  * http://www.unicode.org/draft/reports/tr35/tr35.html#Language_Plural_Rules
  66  *
  67  * Examples:<pre>
  68  *   "one: n is 1; few: n in 2..4"</pre>
  69  *  This defines two rules, for 'one' and 'few'.  The condition for
  70  *  'one' is "n is 1" which means that the number must be equal to
  71  *  1 for this condition to pass.  The condition for 'few' is
  72  *  "n in 2..4" which means that the number must be between 2 and
  73  *  4 inclusive for this condition to pass.  All other numbers
  74  *  are assigned the keyword "other" by the default rule.
  75  *  </p><pre>
  76  *    "zero: n is 0; one: n is 1; zero: n mod 100 in 1..19"</pre>
  77  *  This illustrates that the same keyword can be defined multiple times.
  78  *  Each rule is examined in order, and the first keyword whose condition
  79  *  passes is the one returned.  Also notes that a modulus is applied
  80  *  to n in the last rule.  Thus its condition holds for 119, 219, 319...
  81  *  </p><pre>
  82  *    "one: n is 1; few: n mod 10 in 2..4 and n mod 100 not in 12..14"</pre>
  83  *  This illustrates conjunction and negation.  The condition for 'few'
  84  *  has two parts, both of which must be met: "n mod 10 in 2..4" and
  85  *  "n mod 100 not in 12..14".  The first part applies a modulus to n
  86  *  before the test as in the previous example.  The second part applies
  87  *  a different modulus and also uses negation, thus it matches all
  88  *  numbers _not_ in 12, 13, 14, 112, 113, 114, 212, 213, 214...
  89  *  </p>
  90  *  <p>
  91  * Syntax:<pre>
  92  * \code
  93  * rules         = rule (';' rule)*
  94  * rule          = keyword ':' condition
  95  * keyword       = <identifier>
  96  * condition     = and_condition ('or' and_condition)*
  97  * and_condition = relation ('and' relation)*
  98  * relation      = is_relation | in_relation | within_relation | 'n' <EOL>
  99  * is_relation   = expr 'is' ('not')? value
 100  * in_relation   = expr ('not')? 'in' range_list
 101  * within_relation = expr ('not')? 'within' range
 102  * expr          = ('n' | 'i' | 'f' | 'v' | 'j') ('mod' value)?
 103  * range_list    = (range | value) (',' range_list)*
 104  * value         = digit+  ('.' digit+)?
 105  * digit         = 0|1|2|3|4|5|6|7|8|9
 106  * range         = value'..'value
 107  * \endcode
 108  * </pre></p>
 109  * <p>
 110  * <p>
 111  * The i, f, and v values are defined as follows:
 112  * </p>
 113  * <ul>
 114  * <li>i to be the integer digits.</li>
 115  * <li>f to be the visible fractional digits, as an integer.</li>
 116  * <li>v to be the number of visible fraction digits.</li>
 117  * <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>
 118  * </ul>
 119  * <p>
 120  * Examples are in the following table:
 121  * </p>
 122  * <table border='1' style="border-collapse:collapse">
 123  * <tbody>
 124  * <tr>
 125  * <th>n</th>
 126  * <th>i</th>
 127  * <th>f</th>
 128  * <th>v</th>
 129  * </tr>
 130  * <tr>
 131  * <td>1.0</td>
 132  * <td>1</td>
 133  * <td align="right">0</td>
 134  * <td>1</td>
 135  * </tr>
 136  * <tr>
 137  * <td>1.00</td>
 138  * <td>1</td>
 139  * <td align="right">0</td>
 140  * <td>2</td>
 141  * </tr>
 142  * <tr>
 143  * <td>1.3</td>
 144  * <td>1</td>
 145  * <td align="right">3</td>
 146  * <td>1</td>
 147  * </tr>
 148  * <tr>
 149  * <td>1.03</td>
 150  * <td>1</td>
 151  * <td align="right">3</td>
 152  * <td>2</td>
 153  * </tr>
 154  * <tr>
 155  * <td>1.23</td>
 156  * <td>1</td>
 157  * <td align="right">23</td>
 158  * <td>2</td>
 159  * </tr>
 160  * </tbody>
 161  * </table>
 162  * <p>
 163  * The difference between 'in' and 'within' is that 'in' only includes integers in the specified range, while 'within'
 164  * includes all values. Using 'within' with a range_list consisting entirely of values is the same as using 'in' (it's
 165  * not an error).
 166  * </p>
 167
 168  * An "identifier" is a sequence of characters that do not have the
 169  * Unicode Pattern_Syntax or Pattern_White_Space properties.
 170  * <p>
 171  * The difference between 'in' and 'within' is that 'in' only includes
 172  * integers in the specified range, while 'within' includes all values.
 173  * Using 'within' with a range_list consisting entirely of values is the
 174  * same as using 'in' (it's not an error).
 175  *</p>
 176  * <p>
 177  * Keywords
 178  * could be defined by users or from ICU locale data. There are 6
 179  * predefined values in ICU - 'zero', 'one', 'two', 'few', 'many' and
 180  * 'other'. Callers need to check the value of keyword returned by
 181  * {@link #select} method.
 182  * </p>
 183  *
 184  * Examples:<pre>
 185  * UnicodeString keyword = pl->select(number);
 186  * if (keyword== UnicodeString("one") {
 187  *     ...
 188  * }
 189  * else if ( ... )
 190  * </pre>
 191  * <strong>Note:</strong><br>
 192  *  <p>
 193  *   ICU defines plural rules for many locales based on CLDR <i>Language Plural Rules</i>.
 194  *   For these predefined rules, see CLDR page at
 195  *    http://unicode.org/repos/cldr-tmp/trunk/diff/supplemental/language_plural_rules.html
 196  * </p>
 197  */
 198 class U_I18N_API PluralRules : public UObject {
 199 public:
 200
 201     /**
 202      * Constructor.
 203      * @param status  Output param set to success/failure code on exit, which
 204      *                must not indicate a failure before the function call.
 205      *
 206      * @stable ICU 4.0
 207      */
 208     PluralRules(UErrorCode& status);
 209
 210     /**
 211      * Copy constructor.
 212      * @stable ICU 4.0
 213      */
 214     PluralRules(const PluralRules& other);
 215
 216     /**
 217      * Destructor.
 218      * @stable ICU 4.0
 219      */
 220     virtual ~PluralRules();
 221
 222     /**
 223      * Clone
 224      * @stable ICU 4.0
 225      */
 226     PluralRules* clone() const;
 227
 228     /**
 229       * Assignment operator.
 230       * @stable ICU 4.0
 231       */
 232     PluralRules& operator=(const PluralRules&);
 233
 234     /**
 235      * Creates a PluralRules from a description if it is parsable, otherwise
 236      * returns NULL.
 237      *
 238      * @param description rule description
 239      * @param status      Output param set to success/failure code on exit, which
 240      *                    must not indicate a failure before the function call.
 241      * @return            new PluralRules pointer. NULL if there is an error.
 242      * @stable ICU 4.0
 243      */
 244     static PluralRules* U_EXPORT2 createRules(const UnicodeString& description,
 245                                               UErrorCode& status);
 246
 247     /**
 248      * The default rules that accept any number.
 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        new PluralRules pointer. NULL if there is an error.
 253      * @stable ICU 4.0
 254      */
 255     static PluralRules* U_EXPORT2 createDefaultRules(UErrorCode& status);
 256
 257     /**
 258      * Provides access to the predefined cardinal-number <code>PluralRules</code> for a given
 259      * locale.
 260      * Same as forLocale(locale, UPLURAL_TYPE_CARDINAL, status).
 261      *
 262      * @param locale  The locale for which a <code>PluralRules</code> object is
 263      *                returned.
 264      * @param status  Output param set to success/failure code on exit, which
 265      *                must not indicate a failure before the function call.
 266      * @return        The predefined <code>PluralRules</code> object pointer for
 267      *                this locale. If there's no predefined rules for this locale,
 268      *                the rules for the closest parent in the locale hierarchy
 269      *                that has one will  be returned.  The final fallback always
 270      *                returns the default 'other' rules.
 271      * @stable ICU 4.0
 272      */
 273     static PluralRules* U_EXPORT2 forLocale(const Locale& locale, UErrorCode& status);
 274
 275     /**
 276      * Provides access to the predefined <code>PluralRules</code> for a given
 277      * locale and the plural type.
 278      *
 279      * @param locale  The locale for which a <code>PluralRules</code> object is
 280      *                returned.
 281      * @param type    The plural type (e.g., cardinal or ordinal).
 282      * @param status  Output param set to success/failure code on exit, which
 283      *                must not indicate a failure before the function call.
 284      * @return        The predefined <code>PluralRules</code> object pointer for
 285      *                this locale. If there's no predefined rules for this locale,
 286      *                the rules for the closest parent in the locale hierarchy
 287      *                that has one will  be returned.  The final fallback always
 288      *                returns the default 'other' rules.
 289      * @stable ICU 50
 290      */
 291     static PluralRules* U_EXPORT2 forLocale(const Locale& locale, UPluralType type, UErrorCode& status);
 292
 293 #ifndef U_HIDE_INTERNAL_API
 294     /**
 295      * Return a StringEnumeration over the locales for which there is plurals data.
 296      * @return a StringEnumeration over the locales available.
 297      * @internal
 298      */
 299     static StringEnumeration* U_EXPORT2 getAvailableLocales(UErrorCode &status);
 300
 301     /**
 302      * Returns whether or not there are overrides.
 303      * @param locale       the locale to check.
 304      * @return
 305      * @internal
 306      */
 307     static UBool hasOverride(const Locale &locale);
 308
 309     /**
 310      * For ICU use only.
 311      * creates a  SharedPluralRules object
 312      * @internal
 313      */
 314     static PluralRules* U_EXPORT2 internalForLocale(const Locale& locale, UPluralType type, UErrorCode& status);
 315
 316     /**
 317      * For ICU use only.
 318      * Returns handle to the shared, cached PluralRules instance.
 319      * Caller must call removeRef() on returned value once it is done with
 320      * the shared instance.
 321      * @internal
 322      */
 323     static const SharedPluralRules* U_EXPORT2 createSharedInstance(
 324             const Locale& locale, UPluralType type, UErrorCode& status);
 325
 326
 327 #endif  /* U_HIDE_INTERNAL_API */
 328
 329     /**
 330      * Given a number, returns the keyword of the first rule that applies to
 331      * the number.  This function can be used with isKeyword* functions to
 332      * determine the keyword for default plural rules.
 333      *
 334      * @param number  The number for which the rule has to be determined.
 335      * @return        The keyword of the selected rule.
 336      * @stable ICU 4.0
 337      */
 338     UnicodeString select(int32_t number) const;
 339
 340     /**
 341      * Given a number, returns the keyword of the first rule that applies to
 342      * the number.  This function can be used with isKeyword* functions to
 343      * determine the keyword for default plural rules.
 344      *
 345      * @param number  The number for which the rule has to be determined.
 346      * @return        The keyword of the selected rule.
 347      * @stable ICU 4.0
 348      */
 349     UnicodeString select(double number) const;
 350
 351 #ifndef U_HIDE_INTERNAL_API
 352     /**
 353      * Given a number and a format, returns the keyword of the first applicable
 354      * rule for this PluralRules object.
 355      * Note: This internal preview interface may be removed in the future if
 356      * an architecturally cleaner solution reaches stable status.
 357      * @param obj The numeric object for which the rule should be determined.
 358      * @param fmt The NumberFormat specifying how the number will be formatted
 359      *        (this can affect the plural form, e.g. "1 dollar" vs "1.0 dollars").
 360      * @param status  Input/output parameter. If at entry this indicates a
 361      *                failure status, the method returns immediately; otherwise
 362      *                this is set to indicate the outcome of the call.
 363      * @return The keyword of the selected rule. Undefined in the case of an error.
 364      * @internal ICU 59 technology preview, may be removed in the future
 365      */
 366     UnicodeString select(const Formattable& obj, const NumberFormat& fmt, UErrorCode& status) const;
 367
 368     /**
 369       * @internal
 370       */
 371     UnicodeString select(const FixedDecimal &number) const;
 372     /**
 373       * @internal
 374       */
 375     UnicodeString select(const VisibleDigitsWithExponent &number) const;
 376 #endif  /* U_HIDE_INTERNAL_API */
 377
 378     /**
 379      * Returns a list of all rule keywords used in this <code>PluralRules</code>
 380      * object.  The rule 'other' is always present by default.
 381      *
 382      * @param status Output param set to success/failure code on exit, which
 383      *               must not indicate a failure before the function call.
 384      * @return       StringEnumeration with the keywords.
 385      *               The caller must delete the object.
 386      * @stable ICU 4.0
 387      */
 388     StringEnumeration* getKeywords(UErrorCode& status) const;
 389
 390 #ifndef U_HIDE_DEPRECATED_API
 391     /**
 392      * Deprecated Function, does not return useful results.
 393      *
 394      * Originally intended to return a unique value for this keyword if it exists,
 395      * else the constant UPLRULES_NO_UNIQUE_VALUE.
 396      *
 397      * @param keyword The keyword.
 398      * @return        Stub deprecated function returns UPLRULES_NO_UNIQUE_VALUE always.
 399      * @deprecated ICU 55
 400      */
 401     double getUniqueKeywordValue(const UnicodeString& keyword);
 402
 403     /**
 404      * Deprecated Function, does not produce useful results.
 405      *
 406      * Orginally intended to return all the values for which select() would return the keyword.
 407      * If the keyword is unknown, returns no values, but this is not an error.  If
 408      * the number of values is unlimited, returns no values and -1 as the
 409      * count.
 410      *
 411      * The number of returned values is typically small.
 412      *
 413      * @param keyword      The keyword.
 414      * @param dest         Array into which to put the returned values.  May
 415      *                     be NULL if destCapacity is 0.
 416      * @param destCapacity The capacity of the array, must be at least 0.
 417      * @param status       The error code. Deprecated function, always sets U_UNSUPPORTED_ERROR.
 418      * @return             The count of values available, or -1.  This count
 419      *                     can be larger than destCapacity, but no more than
 420      *                     destCapacity values will be written.
 421      * @deprecated ICU 55
 422      */
 423     int32_t getAllKeywordValues(const UnicodeString &keyword,
 424                                 double *dest, int32_t destCapacity,
 425                                 UErrorCode& status);
 426 #endif  /* U_HIDE_DEPRECATED_API */
 427
 428     /**
 429      * Returns sample values for which select() would return the keyword.  If
 430      * the keyword is unknown, returns no values, but this is not an error.
 431      *
 432      * The number of returned values is typically small.
 433      *
 434      * @param keyword      The keyword.
 435      * @param dest         Array into which to put the returned values.  May
 436      *                     be NULL if destCapacity is 0.
 437      * @param destCapacity The capacity of the array, must be at least 0.
 438      * @param status       The error code.
 439      * @return             The count of values written.
 440      *                     If more than destCapacity samples are available, then
 441      *                     only destCapacity are written, and destCapacity is returned as the count,
 442      *                     rather than setting a U_BUFFER_OVERFLOW_ERROR.
 443      *                     (The actual number of keyword values could be unlimited.)
 444      * @stable ICU 4.8
 445      */
 446     int32_t getSamples(const UnicodeString &keyword,
 447                        double *dest, int32_t destCapacity,
 448                        UErrorCode& status);
 449
 450     /**
 451      * Returns TRUE if the given keyword is defined in this
 452      * <code>PluralRules</code> object.
 453      *
 454      * @param keyword  the input keyword.
 455      * @return         TRUE if the input keyword is defined.
 456      *                 Otherwise, return FALSE.
 457      * @stable ICU 4.0
 458      */
 459     UBool isKeyword(const UnicodeString& keyword) const;
 460
 461
 462     /**
 463      * Returns keyword for default plural form.
 464      *
 465      * @return         keyword for default plural form.
 466      * @stable ICU 4.0
 467      */
 468     UnicodeString getKeywordOther() const;
 469
 470 #ifndef U_HIDE_INTERNAL_API
 471     /**
 472      *
 473      * @internal
 474      */
 475      UnicodeString getRules() const;
 476 #endif  /* U_HIDE_INTERNAL_API */
 477
 478     /**
 479      * Compares the equality of two PluralRules objects.
 480      *
 481      * @param other The other PluralRules object to be compared with.
 482      * @return      True if the given PluralRules is the same as this
 483      *              PluralRules; false otherwise.
 484      * @stable ICU 4.0
 485      */
 486     virtual UBool operator==(const PluralRules& other) const;
 487
 488     /**
 489      * Compares the inequality of two PluralRules objects.
 490      *
 491      * @param other The PluralRules object to be compared with.
 492      * @return      True if the given PluralRules is not the same as this
 493      *              PluralRules; false otherwise.
 494      * @stable ICU 4.0
 495      */
 496     UBool operator!=(const PluralRules& other) const  {return !operator==(other);}
 497
 498
 499     /**
 500      * ICU "poor man's RTTI", returns a UClassID for this class.
 501      *
 502      * @stable ICU 4.0
 503      *
 504     */
 505     static UClassID U_EXPORT2 getStaticClassID(void);
 506
 507     /**
 508      * ICU "poor man's RTTI", returns a UClassID for the actual class.
 509      *
 510      * @stable ICU 4.0
 511      */
 512     virtual UClassID getDynamicClassID() const;
 513
 514
 515 private:
 516     RuleChain  *mRules;
 517
 518     PluralRules();   // default constructor not implemented
 519     void            parseDescription(const UnicodeString& ruleData, UErrorCode &status);
 520     int32_t         getNumberValue(const UnicodeString& token) const;
 521     UnicodeString   getRuleFromResource(const Locale& locale, UPluralType type, UErrorCode& status);
 522     RuleChain      *rulesForKeyword(const UnicodeString &keyword) const;
 523
 524     friend class PluralRuleParser;
 525 };
 526
 527 U_NAMESPACE_END
 528 #endif // U_SHOW_CPLUSPLUS_API
 529
 530 #endif /* #if !UCONFIG_NO_FORMATTING */
 531
 532 #endif // _PLURRULE
 533 //eof