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