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 IFixedDecimal;
  48 class RuleChain;
  49 class PluralRuleParser;
  50 class PluralKeywordEnumeration;
  51 class AndConstraint;
  52 class SharedPluralRules;
  53
  54 namespace number {
  55 class FormattedNumber;
  56 }
  57
  58 /**
  59  * Defines rules for mapping non-negative numeric values onto a small set of
  60  * keywords. Rules are constructed from a text description, consisting
  61  * of a series of keywords and conditions.  The {@link #select} method
  62  * examines each condition in order and returns the keyword for the
  63  * first condition that matches the number.  If none match,
  64  * default rule(other) is returned.
  65  *
  66  * For more information, details, and tips for writing rules, see the
  67  * LDML spec, C.11 Language Plural Rules:
  68  * http://www.unicode.org/draft/reports/tr35/tr35.html#Language_Plural_Rules
  69  *
  70  * Examples:<pre>
  71  *   "one: n is 1; few: n in 2..4"</pre>
  72  *  This defines two rules, for 'one' and 'few'.  The condition for
  73  *  'one' is "n is 1" which means that the number must be equal to
  74  *  1 for this condition to pass.  The condition for 'few' is
  75  *  "n in 2..4" which means that the number must be between 2 and
  76  *  4 inclusive for this condition to pass.  All other numbers
  77  *  are assigned the keyword "other" by the default rule.
  78  *  </p><pre>
  79  *    "zero: n is 0; one: n is 1; zero: n mod 100 in 1..19"</pre>
  80  *  This illustrates that the same keyword can be defined multiple times.
  81  *  Each rule is examined in order, and the first keyword whose condition
  82  *  passes is the one returned.  Also notes that a modulus is applied
  83  *  to n in the last rule.  Thus its condition holds for 119, 219, 319...
  84  *  </p><pre>
  85  *    "one: n is 1; few: n mod 10 in 2..4 and n mod 100 not in 12..14"</pre>
  86  *  This illustrates conjunction and negation.  The condition for 'few'
  87  *  has two parts, both of which must be met: "n mod 10 in 2..4" and
  88  *  "n mod 100 not in 12..14".  The first part applies a modulus to n
  89  *  before the test as in the previous example.  The second part applies
  90  *  a different modulus and also uses negation, thus it matches all
  91  *  numbers _not_ in 12, 13, 14, 112, 113, 114, 212, 213, 214...
  92  *  </p>
  93  *  <p>
  94  * Syntax:<pre>
  95  * \code
  96  * rules         = rule (';' rule)*
  97  * rule          = keyword ':' condition
  98  * keyword       = <identifier>
  99  * condition     = and_condition ('or' and_condition)*
 100  * and_condition = relation ('and' relation)*
 101  * relation      = is_relation | in_relation | within_relation | 'n' <EOL>
 102  * is_relation   = expr 'is' ('not')? value
 103  * in_relation   = expr ('not')? 'in' range_list
 104  * within_relation = expr ('not')? 'within' range
 105  * expr          = ('n' | 'i' | 'f' | 'v' | 'j') ('mod' value)?
 106  * range_list    = (range | value) (',' range_list)*
 107  * value         = digit+  ('.' digit+)?
 108  * digit         = 0|1|2|3|4|5|6|7|8|9
 109  * range         = value'..'value
 110  * \endcode
 111  * </pre></p>
 112  * <p>
 113  * <p>
 114  * The i, f, and v values are defined as follows:
 115  * </p>
 116  * <ul>
 117  * <li>i to be the integer digits.</li>
 118  * <li>f to be the visible fractional digits, as an integer.</li>
 119  * <li>v to be the number of visible fraction digits.</li>
 120  * <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>
 121  * </ul>
 122  * <p>
 123  * Examples are in the following table:
 124  * </p>
 125  * <table border='1' style="border-collapse:collapse">
 126  * <tr>
 127  * <th>n</th>
 128  * <th>i</th>
 129  * <th>f</th>
 130  * <th>v</th>
 131  * </tr>
 132  * <tr>
 133  * <td>1.0</td>
 134  * <td>1</td>
 135  * <td align="right">0</td>
 136  * <td>1</td>
 137  * </tr>
 138  * <tr>
 139  * <td>1.00</td>
 140  * <td>1</td>
 141  * <td align="right">0</td>
 142  * <td>2</td>
 143  * </tr>
 144  * <tr>
 145  * <td>1.3</td>
 146  * <td>1</td>
 147  * <td align="right">3</td>
 148  * <td>1</td>
 149  * </tr>
 150  * <tr>
 151  * <td>1.03</td>
 152  * <td>1</td>
 153  * <td align="right">3</td>
 154  * <td>2</td>
 155  * </tr>
 156  * <tr>
 157  * <td>1.23</td>
 158  * <td>1</td>
 159  * <td align="right">23</td>
 160  * <td>2</td>
 161  * </tr>
 162  * </table>
 163  * <p>
 164  * The difference between 'in' and 'within' is that 'in' only includes integers in the specified range, while 'within'
 165  * includes all values. Using 'within' with a range_list consisting entirely of values is the same as using 'in' (it's
 166  * not an error).
 167  * </p>
 168
 169  * An "identifier" is a sequence of characters that do not have the
 170  * Unicode Pattern_Syntax or Pattern_White_Space properties.
 171  * <p>
 172  * The difference between 'in' and 'within' is that 'in' only includes
 173  * integers in the specified range, while 'within' includes all values.
 174  * Using 'within' with a range_list consisting entirely of values is the
 175  * same as using 'in' (it's not an error).
 176  *</p>
 177  * <p>
 178  * Keywords
 179  * could be defined by users or from ICU locale data. There are 6
 180  * predefined values in ICU - 'zero', 'one', 'two', 'few', 'many' and
 181  * 'other'. Callers need to check the value of keyword returned by
 182  * {@link #select} method.
 183  * </p>
 184  *
 185  * Examples:<pre>
 186  * UnicodeString keyword = pl->select(number);
 187  * if (keyword== UnicodeString("one") {
 188  *     ...
 189  * }
 190  * else if ( ... )
 191  * </pre>
 192  * <strong>Note:</strong><br>
 193  *  <p>
 194  *   ICU defines plural rules for many locales based on CLDR <i>Language Plural Rules</i>.
 195  *   For these predefined rules, see CLDR page at
 196  *    http://unicode.org/repos/cldr-tmp/trunk/diff/supplemental/language_plural_rules.html
 197  * </p>
 198  */
 199 class U_I18N_API PluralRules : public UObject {
 200 public:
 201
 202     /**
 203      * Constructor.
 204      * @param status  Output param set to success/failure code on exit, which
 205      *                must not indicate a failure before the function call.
 206      *
 207      * @stable ICU 4.0
 208      */
 209     PluralRules(UErrorCode& status);
 210
 211     /**
 212      * Copy constructor.
 213      * @stable ICU 4.0
 214      */
 215     PluralRules(const PluralRules& other);
 216
 217     /**
 218      * Destructor.
 219      * @stable ICU 4.0
 220      */
 221     virtual ~PluralRules();
 222
 223     /**
 224      * Clone
 225      * @stable ICU 4.0
 226      */
 227     PluralRules* clone() const;
 228
 229     /**
 230       * Assignment operator.
 231       * @stable ICU 4.0
 232       */
 233     PluralRules& operator=(const PluralRules&);
 234
 235     /**
 236      * Creates a PluralRules from a description if it is parsable, otherwise
 237      * returns NULL.
 238      *
 239      * @param description rule description
 240      * @param status      Output param set to success/failure code on exit, which
 241      *                    must not indicate a failure before the function call.
 242      * @return            new PluralRules pointer. NULL if there is an error.
 243      * @stable ICU 4.0
 244      */
 245     static PluralRules* U_EXPORT2 createRules(const UnicodeString& description,
 246                                               UErrorCode& status);
 247
 248     /**
 249      * The default rules that accept any number.
 250      *
 251      * @param status  Output param set to success/failure code on exit, which
 252      *                must not indicate a failure before the function call.
 253      * @return        new PluralRules pointer. NULL if there is an error.
 254      * @stable ICU 4.0
 255      */
 256     static PluralRules* U_EXPORT2 createDefaultRules(UErrorCode& status);
 257
 258     /**
 259      * Provides access to the predefined cardinal-number <code>PluralRules</code> for a given
 260      * locale.
 261      * Same as forLocale(locale, UPLURAL_TYPE_CARDINAL, status).
 262      *
 263      * @param locale  The locale for which a <code>PluralRules</code> object is
 264      *                returned.
 265      * @param status  Output param set to success/failure code on exit, which
 266      *                must not indicate a failure before the function call.
 267      * @return        The predefined <code>PluralRules</code> object pointer for
 268      *                this locale. If there's no predefined rules for this locale,
 269      *                the rules for the closest parent in the locale hierarchy
 270      *                that has one will  be returned.  The final fallback always
 271      *                returns the default 'other' rules.
 272      * @stable ICU 4.0
 273      */
 274     static PluralRules* U_EXPORT2 forLocale(const Locale& locale, UErrorCode& status);
 275
 276     /**
 277      * Provides access to the predefined <code>PluralRules</code> for a given
 278      * locale and the plural type.
 279      *
 280      * @param locale  The locale for which a <code>PluralRules</code> object is
 281      *                returned.
 282      * @param type    The plural type (e.g., cardinal or ordinal).
 283      * @param status  Output param set to success/failure code on exit, which
 284      *                must not indicate a failure before the function call.
 285      * @return        The predefined <code>PluralRules</code> object pointer for
 286      *                this locale. If there's no predefined rules for this locale,
 287      *                the rules for the closest parent in the locale hierarchy
 288      *                that has one will  be returned.  The final fallback always
 289      *                returns the default 'other' rules.
 290      * @stable ICU 50
 291      */
 292     static PluralRules* U_EXPORT2 forLocale(const Locale& locale, UPluralType type, UErrorCode& status);
 293
 294 #ifndef U_HIDE_INTERNAL_API
 295     /**
 296      * Return a StringEnumeration over the locales for which there is plurals data.
 297      * @return a StringEnumeration over the locales available.
 298      * @internal
 299      */
 300     static StringEnumeration* U_EXPORT2 getAvailableLocales(UErrorCode &status);
 301
 302     /**
 303      * Returns whether or not there are overrides.
 304      * @param locale       the locale to check.
 305      * @return
 306      * @internal
 307      */
 308     static UBool hasOverride(const Locale &locale);
 309
 310     /**
 311      * For ICU use only.
 312      * creates a  SharedPluralRules object
 313      * @internal
 314      */
 315     static PluralRules* U_EXPORT2 internalForLocale(const Locale& locale, UPluralType type, UErrorCode& status);
 316
 317     /**
 318      * For ICU use only.
 319      * Returns handle to the shared, cached PluralRules instance.
 320      * Caller must call removeRef() on returned value once it is done with
 321      * the shared instance.
 322      * @internal
 323      */
 324     static const SharedPluralRules* U_EXPORT2 createSharedInstance(
 325             const Locale& locale, UPluralType type, UErrorCode& status);
 326
 327
 328 #endif  /* U_HIDE_INTERNAL_API */
 329
 330     /**
 331      * Given an integer, returns the keyword of the first rule
 332      * that applies to  the number.  This function can be used with
 333      * isKeyword* functions to determine the keyword for default plural rules.
 334      *
 335      * @param number  The number for which the rule has to be determined.
 336      * @return        The keyword of the selected rule.
 337      * @stable ICU 4.0
 338      */
 339     UnicodeString select(int32_t number) const;
 340
 341     /**
 342      * Given a floating-point number, returns the keyword of the first rule
 343      * that applies to  the number.  This function can be used with
 344      * isKeyword* functions to determine the keyword for default plural rules.
 345      *
 346      * @param number  The number for which the rule has to be determined.
 347      * @return        The keyword of the selected rule.
 348      * @stable ICU 4.0
 349      */
 350     UnicodeString select(double number) const;
 351
 352 #ifndef U_HIDE_DRAFT_API
 353     /**
 354      * Given a formatted number, returns the keyword of the first rule
 355      * that applies to  the number.  This function can be used with
 356      * isKeyword* functions to determine the keyword for default plural rules.
 357      *
 358      * A FormattedNumber allows you to specify an exponent or trailing zeros,
 359      * which can affect the plural category. To get a FormattedNumber, see
 360      * NumberFormatter.
 361      *
 362      * @param number  The number for which the rule has to be determined.
 363      * @param status  Set if an error occurs while selecting plural keyword.
 364      *                This could happen if the FormattedNumber is invalid.
 365      * @return        The keyword of the selected rule.
 366      * @draft ICU 64
 367      */
 368     UnicodeString select(const number::FormattedNumber& number, UErrorCode& status) const;
 369 #endif  /* U_HIDE_DRAFT_API */
 370
 371 #ifndef U_HIDE_INTERNAL_API
 372     /**
 373       * @internal
 374       */
 375     UnicodeString select(const IFixedDecimal &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      * Originally 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     /**
 525     * An internal status variable used to indicate that the object is in an 'invalid' state.
 526     * Used by copy constructor, the assignment operator and the clone method.
 527     */
 528     UErrorCode mInternalStatus;
 529
 530     friend class PluralRuleParser;
 531 };
 532
 533 U_NAMESPACE_END
 534 #endif // U_SHOW_CPLUSPLUS_API
 535
 536 #endif /* #if !UCONFIG_NO_FORMATTING */
 537
 538 #endif // _PLURRULE
 539 //eof