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 #if U_SHOW_CPLUSPLUS_API
  24
  25 /**
  26  * \file
  27  * \brief C++ API: PluralRules object
  28  */
  29
  30 #if !UCONFIG_NO_FORMATTING
  31
  32 #include "unicode/format.h"
  33 #include "unicode/upluralrules.h"
  34 #ifndef U_HIDE_INTERNAL_API
  35 #include "unicode/numfmt.h"
  36 #endif  /* U_HIDE_INTERNAL_API */
  37
  38 /**
  39  * Value returned by PluralRules::getUniqueKeywordValue() when there is no
  40  * unique value to return.
  41  * @stable ICU 4.8
  42  */
  43 #define UPLRULES_NO_UNIQUE_VALUE ((double)-0.00123456777)
  44
  45 U_NAMESPACE_BEGIN
  46
  47 class Hashtable;
  48 class IFixedDecimal;
  49 class RuleChain;
  50 class PluralRuleParser;
  51 class PluralKeywordEnumeration;
  52 class AndConstraint;
  53 class SharedPluralRules;
  54
  55 namespace number {
  56 class FormattedNumber;
  57 }
  58
  59 /**
  60  * Defines rules for mapping non-negative numeric values onto a small set of
  61  * keywords. Rules are constructed from a text description, consisting
  62  * of a series of keywords and conditions.  The {@link #select} method
  63  * examines each condition in order and returns the keyword for the
  64  * first condition that matches the number.  If none match,
  65  * default rule(other) is returned.
  66  *
  67  * For more information, details, and tips for writing rules, see the
  68  * LDML spec, C.11 Language Plural Rules:
  69  * http://www.unicode.org/draft/reports/tr35/tr35.html#Language_Plural_Rules
  70  *
  71  * Examples:<pre>
  72  *   "one: n is 1; few: n in 2..4"</pre>
  73  *  This defines two rules, for 'one' and 'few'.  The condition for
  74  *  'one' is "n is 1" which means that the number must be equal to
  75  *  1 for this condition to pass.  The condition for 'few' is
  76  *  "n in 2..4" which means that the number must be between 2 and
  77  *  4 inclusive for this condition to pass.  All other numbers
  78  *  are assigned the keyword "other" by the default rule.
  79  *  </p><pre>
  80  *    "zero: n is 0; one: n is 1; zero: n mod 100 in 1..19"</pre>
  81  *  This illustrates that the same keyword can be defined multiple times.
  82  *  Each rule is examined in order, and the first keyword whose condition
  83  *  passes is the one returned.  Also notes that a modulus is applied
  84  *  to n in the last rule.  Thus its condition holds for 119, 219, 319...
  85  *  </p><pre>
  86  *    "one: n is 1; few: n mod 10 in 2..4 and n mod 100 not in 12..14"</pre>
  87  *  This illustrates conjunction and negation.  The condition for 'few'
  88  *  has two parts, both of which must be met: "n mod 10 in 2..4" and
  89  *  "n mod 100 not in 12..14".  The first part applies a modulus to n
  90  *  before the test as in the previous example.  The second part applies
  91  *  a different modulus and also uses negation, thus it matches all
  92  *  numbers _not_ in 12, 13, 14, 112, 113, 114, 212, 213, 214...
  93  *  </p>
  94  *  <p>
  95  * Syntax:<pre>
  96  * \code
  97  * rules         = rule (';' rule)*
  98  * rule          = keyword ':' condition
  99  * keyword       = <identifier>
 100  * condition     = and_condition ('or' and_condition)*
 101  * and_condition = relation ('and' relation)*
 102  * relation      = is_relation | in_relation | within_relation | 'n' <EOL>
 103  * is_relation   = expr 'is' ('not')? value
 104  * in_relation   = expr ('not')? 'in' range_list
 105  * within_relation = expr ('not')? 'within' range
 106  * expr          = ('n' | 'i' | 'f' | 'v' | 'j') ('mod' value)?
 107  * range_list    = (range | value) (',' range_list)*
 108  * value         = digit+  ('.' digit+)?
 109  * digit         = 0|1|2|3|4|5|6|7|8|9
 110  * range         = value'..'value
 111  * \endcode
 112  * </pre></p>
 113  * <p>
 114  * <p>
 115  * The i, f, and v values are defined as follows:
 116  * </p>
 117  * <ul>
 118  * <li>i to be the integer digits.</li>
 119  * <li>f to be the visible fractional digits, as an integer.</li>
 120  * <li>v to be the number of visible fraction digits.</li>
 121  * <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>
 122  * </ul>
 123  * <p>
 124  * Examples are in the following table:
 125  * </p>
 126  * <table border='1' style="border-collapse:collapse">
 127  * <tr>
 128  * <th>n</th>
 129  * <th>i</th>
 130  * <th>f</th>
 131  * <th>v</th>
 132  * </tr>
 133  * <tr>
 134  * <td>1.0</td>
 135  * <td>1</td>
 136  * <td align="right">0</td>
 137  * <td>1</td>
 138  * </tr>
 139  * <tr>
 140  * <td>1.00</td>
 141  * <td>1</td>
 142  * <td align="right">0</td>
 143  * <td>2</td>
 144  * </tr>
 145  * <tr>
 146  * <td>1.3</td>
 147  * <td>1</td>
 148  * <td align="right">3</td>
 149  * <td>1</td>
 150  * </tr>
 151  * <tr>
 152  * <td>1.03</td>
 153  * <td>1</td>
 154  * <td align="right">3</td>
 155  * <td>2</td>
 156  * </tr>
 157  * <tr>
 158  * <td>1.23</td>
 159  * <td>1</td>
 160  * <td align="right">23</td>
 161  * <td>2</td>
 162  * </tr>
 163  * </table>
 164  * <p>
 165  * The difference between 'in' and 'within' is that 'in' only includes integers in the specified range, while 'within'
 166  * includes all values. Using 'within' with a range_list consisting entirely of values is the same as using 'in' (it's
 167  * not an error).
 168  * </p>
 169
 170  * An "identifier" is a sequence of characters that do not have the
 171  * Unicode Pattern_Syntax or Pattern_White_Space properties.
 172  * <p>
 173  * The difference between 'in' and 'within' is that 'in' only includes
 174  * integers in the specified range, while 'within' includes all values.
 175  * Using 'within' with a range_list consisting entirely of values is the
 176  * same as using 'in' (it's not an error).
 177  *</p>
 178  * <p>
 179  * Keywords
 180  * could be defined by users or from ICU locale data. There are 6
 181  * predefined values in ICU - 'zero', 'one', 'two', 'few', 'many' and
 182  * 'other'. Callers need to check the value of keyword returned by
 183  * {@link #select} method.
 184  * </p>
 185  *
 186  * Examples:<pre>
 187  * UnicodeString keyword = pl->select(number);
 188  * if (keyword== UnicodeString("one") {
 189  *     ...
 190  * }
 191  * else if ( ... )
 192  * </pre>
 193  * <strong>Note:</strong><br>
 194  *  <p>
 195  *   ICU defines plural rules for many locales based on CLDR <i>Language Plural Rules</i>.
 196  *   For these predefined rules, see CLDR page at
 197  *    http://unicode.org/repos/cldr-tmp/trunk/diff/supplemental/language_plural_rules.html
 198  * </p>
 199  */
 200 class U_I18N_API PluralRules : public UObject {
 201 public:
 202
 203     /**
 204      * Constructor.
 205      * @param status  Output param set to success/failure code on exit, which
 206      *                must not indicate a failure before the function call.
 207      *
 208      * @stable ICU 4.0
 209      */
 210     PluralRules(UErrorCode& status);
 211
 212     /**
 213      * Copy constructor.
 214      * @stable ICU 4.0
 215      */
 216     PluralRules(const PluralRules& other);
 217
 218     /**
 219      * Destructor.
 220      * @stable ICU 4.0
 221      */
 222     virtual ~PluralRules();
 223
 224     /**
 225      * Clone
 226      * @stable ICU 4.0
 227      */
 228     PluralRules* clone() const;
 229
 230     /**
 231       * Assignment operator.
 232       * @stable ICU 4.0
 233       */
 234     PluralRules& operator=(const PluralRules&);
 235
 236     /**
 237      * Creates a PluralRules from a description if it is parsable, otherwise
 238      * returns NULL.
 239      *
 240      * @param description rule description
 241      * @param status      Output param set to success/failure code on exit, which
 242      *                    must not indicate a failure before the function call.
 243      * @return            new PluralRules pointer. NULL if there is an error.
 244      * @stable ICU 4.0
 245      */
 246     static PluralRules* U_EXPORT2 createRules(const UnicodeString& description,
 247                                               UErrorCode& status);
 248
 249     /**
 250      * The default rules that accept any number.
 251      *
 252      * @param status  Output param set to success/failure code on exit, which
 253      *                must not indicate a failure before the function call.
 254      * @return        new PluralRules pointer. NULL if there is an error.
 255      * @stable ICU 4.0
 256      */
 257     static PluralRules* U_EXPORT2 createDefaultRules(UErrorCode& status);
 258
 259     /**
 260      * Provides access to the predefined cardinal-number <code>PluralRules</code> for a given
 261      * locale.
 262      * Same as forLocale(locale, UPLURAL_TYPE_CARDINAL, status).
 263      *
 264      * @param locale  The locale for which a <code>PluralRules</code> object is
 265      *                returned.
 266      * @param status  Output param set to success/failure code on exit, which
 267      *                must not indicate a failure before the function call.
 268      * @return        The predefined <code>PluralRules</code> object pointer for
 269      *                this locale. If there's no predefined rules for this locale,
 270      *                the rules for the closest parent in the locale hierarchy
 271      *                that has one will  be returned.  The final fallback always
 272      *                returns the default 'other' rules.
 273      * @stable ICU 4.0
 274      */
 275     static PluralRules* U_EXPORT2 forLocale(const Locale& locale, UErrorCode& status);
 276
 277     /**
 278      * Provides access to the predefined <code>PluralRules</code> for a given
 279      * locale and the plural type.
 280      *
 281      * @param locale  The locale for which a <code>PluralRules</code> object is
 282      *                returned.
 283      * @param type    The plural type (e.g., cardinal or ordinal).
 284      * @param status  Output param set to success/failure code on exit, which
 285      *                must not indicate a failure before the function call.
 286      * @return        The predefined <code>PluralRules</code> object pointer for
 287      *                this locale. If there's no predefined rules for this locale,
 288      *                the rules for the closest parent in the locale hierarchy
 289      *                that has one will  be returned.  The final fallback always
 290      *                returns the default 'other' rules.
 291      * @stable ICU 50
 292      */
 293     static PluralRules* U_EXPORT2 forLocale(const Locale& locale, UPluralType type, UErrorCode& status);
 294
 295 #ifndef U_HIDE_INTERNAL_API
 296     /**
 297      * Return a StringEnumeration over the locales for which there is plurals data.
 298      * @return a StringEnumeration over the locales available.
 299      * @internal
 300      */
 301     static StringEnumeration* U_EXPORT2 getAvailableLocales(UErrorCode &status);
 302
 303     /**
 304      * Returns whether or not there are overrides.
 305      * @param locale       the locale to check.
 306      * @return
 307      * @internal
 308      */
 309     static UBool hasOverride(const Locale &locale);
 310
 311     /**
 312      * For ICU use only.
 313      * creates a  SharedPluralRules object
 314      * @internal
 315      */
 316     static PluralRules* U_EXPORT2 internalForLocale(const Locale& locale, UPluralType type, UErrorCode& status);
 317
 318     /**
 319      * For ICU use only.
 320      * Returns handle to the shared, cached PluralRules instance.
 321      * Caller must call removeRef() on returned value once it is done with
 322      * the shared instance.
 323      * @internal
 324      */
 325     static const SharedPluralRules* U_EXPORT2 createSharedInstance(
 326             const Locale& locale, UPluralType type, UErrorCode& status);
 327
 328
 329 #endif  /* U_HIDE_INTERNAL_API */
 330
 331     /**
 332      * Given an integer, returns the keyword of the first rule
 333      * that applies to  the number.  This function can be used with
 334      * isKeyword* functions to determine the keyword for default plural rules.
 335      *
 336      * @param number  The number for which the rule has to be determined.
 337      * @return        The keyword of the selected rule.
 338      * @stable ICU 4.0
 339      */
 340     UnicodeString select(int32_t number) const;
 341
 342     /**
 343      * Given a floating-point number, returns the keyword of the first rule
 344      * that applies to  the number.  This function can be used with
 345      * isKeyword* functions to determine the keyword for default plural rules.
 346      *
 347      * @param number  The number for which the rule has to be determined.
 348      * @return        The keyword of the selected rule.
 349      * @stable ICU 4.0
 350      */
 351     UnicodeString select(double number) const;
 352
 353 #ifndef U_HIDE_DRAFT_API
 354     /**
 355      * Given a formatted number, returns the keyword of the first rule
 356      * that applies to  the number.  This function can be used with
 357      * isKeyword* functions to determine the keyword for default plural rules.
 358      *
 359      * A FormattedNumber allows you to specify an exponent or trailing zeros,
 360      * which can affect the plural category. To get a FormattedNumber, see
 361      * NumberFormatter.
 362      *
 363      * @param number  The number for which the rule has to be determined.
 364      * @param status  Set if an error occurs while selecting plural keyword.
 365      *                This could happen if the FormattedNumber is invalid.
 366      * @return        The keyword of the selected rule.
 367      * @draft ICU 64
 368      */
 369     UnicodeString select(const number::FormattedNumber& number, UErrorCode& status) const;
 370 #endif  /* U_HIDE_DRAFT_API */
 371
 372 #ifndef U_HIDE_INTERNAL_API
 373     /**
 374       * @internal
 375       */
 376     UnicodeString select(const IFixedDecimal &number) const;
 377 #endif  /* U_HIDE_INTERNAL_API */
 378
 379     /**
 380      * Returns a list of all rule keywords used in this <code>PluralRules</code>
 381      * object.  The rule 'other' is always present by default.
 382      *
 383      * @param status Output param set to success/failure code on exit, which
 384      *               must not indicate a failure before the function call.
 385      * @return       StringEnumeration with the keywords.
 386      *               The caller must delete the object.
 387      * @stable ICU 4.0
 388      */
 389     StringEnumeration* getKeywords(UErrorCode& status) const;
 390
 391 #ifndef U_HIDE_DEPRECATED_API
 392     /**
 393      * Deprecated Function, does not return useful results.
 394      *
 395      * Originally intended to return a unique value for this keyword if it exists,
 396      * else the constant UPLRULES_NO_UNIQUE_VALUE.
 397      *
 398      * @param keyword The keyword.
 399      * @return        Stub deprecated function returns UPLRULES_NO_UNIQUE_VALUE always.
 400      * @deprecated ICU 55
 401      */
 402     double getUniqueKeywordValue(const UnicodeString& keyword);
 403
 404     /**
 405      * Deprecated Function, does not produce useful results.
 406      *
 407      * Originally intended to return all the values for which select() would return the keyword.
 408      * If the keyword is unknown, returns no values, but this is not an error.  If
 409      * the number of values is unlimited, returns no values and -1 as the
 410      * count.
 411      *
 412      * The number of returned values is typically small.
 413      *
 414      * @param keyword      The keyword.
 415      * @param dest         Array into which to put the returned values.  May
 416      *                     be NULL if destCapacity is 0.
 417      * @param destCapacity The capacity of the array, must be at least 0.
 418      * @param status       The error code. Deprecated function, always sets U_UNSUPPORTED_ERROR.
 419      * @return             The count of values available, or -1.  This count
 420      *                     can be larger than destCapacity, but no more than
 421      *                     destCapacity values will be written.
 422      * @deprecated ICU 55
 423      */
 424     int32_t getAllKeywordValues(const UnicodeString &keyword,
 425                                 double *dest, int32_t destCapacity,
 426                                 UErrorCode& status);
 427 #endif  /* U_HIDE_DEPRECATED_API */
 428
 429     /**
 430      * Returns sample values for which select() would return the keyword.  If
 431      * the keyword is unknown, returns no values, but this is not an error.
 432      *
 433      * The number of returned values is typically small.
 434      *
 435      * @param keyword      The keyword.
 436      * @param dest         Array into which to put the returned values.  May
 437      *                     be NULL if destCapacity is 0.
 438      * @param destCapacity The capacity of the array, must be at least 0.
 439      * @param status       The error code.
 440      * @return             The count of values written.
 441      *                     If more than destCapacity samples are available, then
 442      *                     only destCapacity are written, and destCapacity is returned as the count,
 443      *                     rather than setting a U_BUFFER_OVERFLOW_ERROR.
 444      *                     (The actual number of keyword values could be unlimited.)
 445      * @stable ICU 4.8
 446      */
 447     int32_t getSamples(const UnicodeString &keyword,
 448                        double *dest, int32_t destCapacity,
 449                        UErrorCode& status);
 450
 451     /**
 452      * Returns TRUE if the given keyword is defined in this
 453      * <code>PluralRules</code> object.
 454      *
 455      * @param keyword  the input keyword.
 456      * @return         TRUE if the input keyword is defined.
 457      *                 Otherwise, return FALSE.
 458      * @stable ICU 4.0
 459      */
 460     UBool isKeyword(const UnicodeString& keyword) const;
 461
 462
 463     /**
 464      * Returns keyword for default plural form.
 465      *
 466      * @return         keyword for default plural form.
 467      * @stable ICU 4.0
 468      */
 469     UnicodeString getKeywordOther() const;
 470
 471 #ifndef U_HIDE_INTERNAL_API
 472     /**
 473      *
 474      * @internal
 475      */
 476      UnicodeString getRules() const;
 477 #endif  /* U_HIDE_INTERNAL_API */
 478
 479     /**
 480      * Compares the equality of two PluralRules objects.
 481      *
 482      * @param other The other PluralRules object to be compared with.
 483      * @return      True if the given PluralRules is the same as this
 484      *              PluralRules; false otherwise.
 485      * @stable ICU 4.0
 486      */
 487     virtual UBool operator==(const PluralRules& other) const;
 488
 489     /**
 490      * Compares the inequality of two PluralRules objects.
 491      *
 492      * @param other The PluralRules object to be compared with.
 493      * @return      True if the given PluralRules is not the same as this
 494      *              PluralRules; false otherwise.
 495      * @stable ICU 4.0
 496      */
 497     UBool operator!=(const PluralRules& other) const  {return !operator==(other);}
 498
 499
 500     /**
 501      * ICU "poor man's RTTI", returns a UClassID for this class.
 502      *
 503      * @stable ICU 4.0
 504      *
 505     */
 506     static UClassID U_EXPORT2 getStaticClassID(void);
 507
 508     /**
 509      * ICU "poor man's RTTI", returns a UClassID for the actual class.
 510      *
 511      * @stable ICU 4.0
 512      */
 513     virtual UClassID getDynamicClassID() const;
 514
 515
 516 private:
 517     RuleChain  *mRules;
 518
 519     PluralRules();   // default constructor not implemented
 520     void            parseDescription(const UnicodeString& ruleData, UErrorCode &status);
 521     int32_t         getNumberValue(const UnicodeString& token) const;
 522     UnicodeString   getRuleFromResource(const Locale& locale, UPluralType type, UErrorCode& status);
 523     RuleChain      *rulesForKeyword(const UnicodeString &keyword) const;
 524
 525     /**
 526     * An internal status variable used to indicate that the object is in an 'invalid' state.
 527     * Used by copy constructor, the assignment operator and the clone method.
 528     */
 529     UErrorCode mInternalStatus;
 530
 531     friend class PluralRuleParser;
 532 };
 533
 534 U_NAMESPACE_END
 535
 536 #endif /* #if !UCONFIG_NO_FORMATTING */
 537
 538 #endif /* U_SHOW_CPLUSPLUS_API */
 539
 540 #endif // _PLURRULE
 541 //eof