plurrule.h (21128B)
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 FixedDecimal; 50 class RuleChain; 51 class PluralRuleParser; 52 class PluralKeywordEnumeration; 53 class AndConstraint; 54 class SharedPluralRules; 55 class StandardPluralRanges; 56 57 namespace number { 58 class FormattedNumber; 59 class FormattedNumberRange; 60 namespace impl { 61 class UFormattedNumberRangeData; 62 class DecimalQuantity; 63 class DecNum; 64 } 65 } 66 67 #ifndef U_HIDE_INTERNAL_API 68 using icu::number::impl::DecimalQuantity; 69 #endif /* U_HIDE_INTERNAL_API */ 70 71 /** 72 * Defines rules for mapping non-negative numeric values onto a small set of 73 * keywords. Rules are constructed from a text description, consisting 74 * of a series of keywords and conditions. The {@link #select} method 75 * examines each condition in order and returns the keyword for the 76 * first condition that matches the number. If none match, 77 * default rule(other) is returned. 78 * 79 * For more information, details, and tips for writing rules, see the 80 * LDML spec, Part 3.5 Language Plural Rules: 81 * https://www.unicode.org/reports/tr35/tr35-numbers.html#Language_Plural_Rules 82 * 83 * Examples:<pre> 84 * "one: n is 1; few: n in 2..4"</pre> 85 * This defines two rules, for 'one' and 'few'. The condition for 86 * 'one' is "n is 1" which means that the number must be equal to 87 * 1 for this condition to pass. The condition for 'few' is 88 * "n in 2..4" which means that the number must be between 2 and 89 * 4 inclusive for this condition to pass. All other numbers 90 * are assigned the keyword "other" by the default rule. 91 * </p><pre> 92 * "zero: n is 0; one: n is 1; zero: n mod 100 in 1..19"</pre> 93 * This illustrates that the same keyword can be defined multiple times. 94 * Each rule is examined in order, and the first keyword whose condition 95 * passes is the one returned. Also notes that a modulus is applied 96 * to n in the last rule. Thus its condition holds for 119, 219, 319... 97 * </p><pre> 98 * "one: n is 1; few: n mod 10 in 2..4 and n mod 100 not in 12..14"</pre> 99 * This illustrates conjunction and negation. The condition for 'few' 100 * has two parts, both of which must be met: "n mod 10 in 2..4" and 101 * "n mod 100 not in 12..14". The first part applies a modulus to n 102 * before the test as in the previous example. The second part applies 103 * a different modulus and also uses negation, thus it matches all 104 * numbers _not_ in 12, 13, 14, 112, 113, 114, 212, 213, 214... 105 * </p> 106 * <p> 107 * Syntax:<pre> 108 * \code 109 * rules = rule (';' rule)* 110 * rule = keyword ':' condition 111 * keyword = <identifier> 112 * condition = and_condition ('or' and_condition)* 113 * and_condition = relation ('and' relation)* 114 * relation = is_relation | in_relation | within_relation | 'n' <EOL> 115 * is_relation = expr 'is' ('not')? value 116 * in_relation = expr ('not')? 'in' range_list 117 * within_relation = expr ('not')? 'within' range 118 * expr = ('n' | 'i' | 'f' | 'v' | 'j') ('mod' value)? 119 * range_list = (range | value) (',' range_list)* 120 * value = digit+ ('.' digit+)? 121 * digit = 0|1|2|3|4|5|6|7|8|9 122 * range = value'..'value 123 * \endcode 124 * </pre></p> 125 * <p> 126 * <p> 127 * The i, f, and v values are defined as follows: 128 * </p> 129 * <ul> 130 * <li>i to be the integer digits.</li> 131 * <li>f to be the visible fractional digits, as an integer.</li> 132 * <li>v to be the number of visible fraction digits.</li> 133 * <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> 134 * </ul> 135 * <p> 136 * Examples are in the following table: 137 * </p> 138 * <table border='1' style="border-collapse:collapse"> 139 * <tr> 140 * <th>n</th> 141 * <th>i</th> 142 * <th>f</th> 143 * <th>v</th> 144 * </tr> 145 * <tr> 146 * <td>1.0</td> 147 * <td>1</td> 148 * <td align="right">0</td> 149 * <td>1</td> 150 * </tr> 151 * <tr> 152 * <td>1.00</td> 153 * <td>1</td> 154 * <td align="right">0</td> 155 * <td>2</td> 156 * </tr> 157 * <tr> 158 * <td>1.3</td> 159 * <td>1</td> 160 * <td align="right">3</td> 161 * <td>1</td> 162 * </tr> 163 * <tr> 164 * <td>1.03</td> 165 * <td>1</td> 166 * <td align="right">3</td> 167 * <td>2</td> 168 * </tr> 169 * <tr> 170 * <td>1.23</td> 171 * <td>1</td> 172 * <td align="right">23</td> 173 * <td>2</td> 174 * </tr> 175 * </table> 176 * <p> 177 * The difference between 'in' and 'within' is that 'in' only includes integers in the specified range, while 'within' 178 * includes all values. Using 'within' with a range_list consisting entirely of values is the same as using 'in' (it's 179 * not an error). 180 * </p> 181 182 * An "identifier" is a sequence of characters that do not have the 183 * Unicode Pattern_Syntax or Pattern_White_Space properties. 184 * <p> 185 * The difference between 'in' and 'within' is that 'in' only includes 186 * integers in the specified range, while 'within' includes all values. 187 * Using 'within' with a range_list consisting entirely of values is the 188 * same as using 'in' (it's not an error). 189 *</p> 190 * <p> 191 * Keywords 192 * could be defined by users or from ICU locale data. There are 6 193 * predefined values in ICU - 'zero', 'one', 'two', 'few', 'many' and 194 * 'other'. Callers need to check the value of keyword returned by 195 * {@link #select} method. 196 * </p> 197 * 198 * Examples:<pre> 199 * UnicodeString keyword = pl->select(number); 200 * if (keyword== UnicodeString("one") { 201 * ... 202 * } 203 * else if ( ... ) 204 * </pre> 205 * <strong>Note:</strong><br> 206 * <p> 207 * ICU defines plural rules for many locales based on CLDR <i>Language Plural Rules</i>. 208 * For these predefined rules, see CLDR page at 209 * https://unicode-org.github.io/cldr-staging/charts/latest/supplemental/language_plural_rules.html 210 * </p> 211 */ 212 class U_I18N_API PluralRules : public UObject { 213 public: 214 215 /** 216 * Constructor. 217 * @param status Output param set to success/failure code on exit, which 218 * must not indicate a failure before the function call. 219 * 220 * @stable ICU 4.0 221 */ 222 PluralRules(UErrorCode& status); 223 224 /** 225 * Copy constructor. 226 * @stable ICU 4.0 227 */ 228 PluralRules(const PluralRules& other); 229 230 /** 231 * Destructor. 232 * @stable ICU 4.0 233 */ 234 virtual ~PluralRules(); 235 236 /** 237 * Clone 238 * @stable ICU 4.0 239 */ 240 PluralRules* clone() const; 241 242 /** 243 * Assignment operator. 244 * @stable ICU 4.0 245 */ 246 PluralRules& operator=(const PluralRules&); 247 248 /** 249 * Creates a PluralRules from a description if it is parsable, otherwise 250 * returns nullptr. 251 * 252 * @param description rule description 253 * @param status Output param set to success/failure code on exit, which 254 * must not indicate a failure before the function call. 255 * @return new PluralRules pointer. nullptr if there is an error. 256 * @stable ICU 4.0 257 */ 258 static PluralRules* U_EXPORT2 createRules(const UnicodeString& description, 259 UErrorCode& status); 260 261 /** 262 * The default rules that accept any number. 263 * 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 new PluralRules pointer. nullptr if there is an error. 267 * @stable ICU 4.0 268 */ 269 static PluralRules* U_EXPORT2 createDefaultRules(UErrorCode& status); 270 271 /** 272 * Provides access to the predefined cardinal-number <code>PluralRules</code> for a given 273 * locale. 274 * Same as forLocale(locale, UPLURAL_TYPE_CARDINAL, status). 275 * 276 * @param locale The locale for which a <code>PluralRules</code> object is 277 * returned. 278 * @param status Output param set to success/failure code on exit, which 279 * must not indicate a failure before the function call. 280 * @return The predefined <code>PluralRules</code> object pointer for 281 * this locale. If there's no predefined rules for this locale, 282 * the rules for the closest parent in the locale hierarchy 283 * that has one will be returned. The final fallback always 284 * returns the default 'other' rules. 285 * @stable ICU 4.0 286 */ 287 static PluralRules* U_EXPORT2 forLocale(const Locale& locale, UErrorCode& status); 288 289 /** 290 * Provides access to the predefined <code>PluralRules</code> for a given 291 * locale and the plural type. 292 * 293 * @param locale The locale for which a <code>PluralRules</code> object is 294 * returned. 295 * @param type The plural type (e.g., cardinal or ordinal). 296 * @param status Output param set to success/failure code on exit, which 297 * must not indicate a failure before the function call. 298 * @return The predefined <code>PluralRules</code> object pointer for 299 * this locale. If there's no predefined rules for this locale, 300 * the rules for the closest parent in the locale hierarchy 301 * that has one will be returned. The final fallback always 302 * returns the default 'other' rules. 303 * @stable ICU 50 304 */ 305 static PluralRules* U_EXPORT2 forLocale(const Locale& locale, UPluralType type, UErrorCode& status); 306 307 #ifndef U_HIDE_INTERNAL_API 308 /** 309 * Return a StringEnumeration over the locales for which there is plurals data. 310 * @return a StringEnumeration over the locales available. 311 * @internal 312 */ 313 static StringEnumeration* U_EXPORT2 getAvailableLocales(UErrorCode &status); 314 315 /** 316 * For ICU use only. 317 * creates a SharedPluralRules object 318 * @internal 319 */ 320 static PluralRules* U_EXPORT2 internalForLocale(const Locale& locale, UPluralType type, UErrorCode& status); 321 322 /** 323 * For ICU use only. 324 * Returns handle to the shared, cached PluralRules instance. 325 * Caller must call removeRef() on returned value once it is done with 326 * the shared instance. 327 * @internal 328 */ 329 static const SharedPluralRules* U_EXPORT2 createSharedInstance( 330 const Locale& locale, UPluralType type, UErrorCode& status); 331 332 333 #endif /* U_HIDE_INTERNAL_API */ 334 335 /** 336 * Given an integer, returns the keyword of the first rule 337 * that applies to the number. This function can be used with 338 * isKeyword* functions to determine the keyword for default plural rules. 339 * 340 * @param number The number for which the rule has to be determined. 341 * @return The keyword of the selected rule. 342 * @stable ICU 4.0 343 */ 344 UnicodeString select(int32_t number) const; 345 346 /** 347 * Given a floating-point number, returns the keyword of the first rule 348 * that applies to the number. This function can be used with 349 * isKeyword* functions to determine the keyword for default plural rules. 350 * 351 * @param number The number for which the rule has to be determined. 352 * @return The keyword of the selected rule. 353 * @stable ICU 4.0 354 */ 355 UnicodeString select(double number) const; 356 357 /** 358 * Given a formatted number, returns the keyword of the first rule 359 * that applies to the number. This function can be used with 360 * isKeyword* functions to determine the keyword for default plural rules. 361 * 362 * A FormattedNumber allows you to specify an exponent or trailing zeros, 363 * which can affect the plural category. To get a FormattedNumber, see 364 * NumberFormatter. 365 * 366 * @param number The number for which the rule has to be determined. 367 * @param status Set if an error occurs while selecting plural keyword. 368 * This could happen if the FormattedNumber is invalid. 369 * @return The keyword of the selected rule. 370 * @stable ICU 64 371 */ 372 UnicodeString select(const number::FormattedNumber& number, UErrorCode& status) const; 373 374 /** 375 * Given a formatted number range, returns the overall plural form of the 376 * range. For example, "3-5" returns "other" in English. 377 * 378 * To get a FormattedNumberRange, see NumberRangeFormatter. 379 * 380 * This method only works if PluralRules was created with a locale. If it was created 381 * from PluralRules::createRules(), this method sets status code U_UNSUPPORTED_ERROR. 382 * 383 * @param range The number range onto which the rules will be applied. 384 * @param status Set if an error occurs while selecting plural keyword. 385 * This could happen if the FormattedNumberRange is invalid, 386 * or if plural ranges data is unavailable. 387 * @return The keyword of the selected rule. 388 * @stable ICU 68 389 */ 390 UnicodeString select(const number::FormattedNumberRange& range, UErrorCode& status) const; 391 392 #ifndef U_HIDE_INTERNAL_API 393 /** 394 * @internal 395 */ 396 UnicodeString select(const IFixedDecimal &number) const; 397 /** 398 * @internal 399 */ 400 UnicodeString select(const number::impl::UFormattedNumberRangeData* urange, UErrorCode& status) const; 401 #endif /* U_HIDE_INTERNAL_API */ 402 403 /** 404 * Returns a list of all rule keywords used in this <code>PluralRules</code> 405 * object. The rule 'other' is always present by default. 406 * 407 * @param status Output param set to success/failure code on exit, which 408 * must not indicate a failure before the function call. 409 * @return StringEnumeration with the keywords. 410 * The caller must delete the object. 411 * @stable ICU 4.0 412 */ 413 StringEnumeration* getKeywords(UErrorCode& status) const; 414 415 #ifndef U_HIDE_DEPRECATED_API 416 /** 417 * Deprecated Function, does not return useful results. 418 * 419 * Originally intended to return a unique value for this keyword if it exists, 420 * else the constant UPLRULES_NO_UNIQUE_VALUE. 421 * 422 * @param keyword The keyword. 423 * @return Stub deprecated function returns UPLRULES_NO_UNIQUE_VALUE always. 424 * @deprecated ICU 55 425 */ 426 double getUniqueKeywordValue(const UnicodeString& keyword); 427 428 /** 429 * Deprecated Function, does not produce useful results. 430 * 431 * Originally intended to return all the values for which select() would return the keyword. 432 * If the keyword is unknown, returns no values, but this is not an error. If 433 * the number of values is unlimited, returns no values and -1 as the 434 * count. 435 * 436 * The number of returned values is typically small. 437 * 438 * @param keyword The keyword. 439 * @param dest Array into which to put the returned values. May 440 * be nullptr if destCapacity is 0. 441 * @param destCapacity The capacity of the array, must be at least 0. 442 * @param status The error code. Deprecated function, always sets U_UNSUPPORTED_ERROR. 443 * @return The count of values available, or -1. This count 444 * can be larger than destCapacity, but no more than 445 * destCapacity values will be written. 446 * @deprecated ICU 55 447 */ 448 int32_t getAllKeywordValues(const UnicodeString &keyword, 449 double *dest, int32_t destCapacity, 450 UErrorCode& status); 451 #endif /* U_HIDE_DEPRECATED_API */ 452 453 /** 454 * Returns sample values for which select() would return the keyword. If 455 * the keyword is unknown, returns no values, but this is not an error. 456 * 457 * The number of returned values is typically small. 458 * 459 * @param keyword The keyword. 460 * @param dest Array into which to put the returned values. May 461 * be nullptr if destCapacity is 0. 462 * @param destCapacity The capacity of the array, must be at least 0. 463 * @param status The error code. 464 * @return The count of values written. 465 * If more than destCapacity samples are available, then 466 * only destCapacity are written, and destCapacity is returned as the count, 467 * rather than setting a U_BUFFER_OVERFLOW_ERROR. 468 * (The actual number of keyword values could be unlimited.) 469 * @stable ICU 4.8 470 */ 471 int32_t getSamples(const UnicodeString &keyword, 472 double *dest, int32_t destCapacity, 473 UErrorCode& status); 474 475 #ifndef U_HIDE_INTERNAL_API 476 /** 477 * Internal-only function that returns DecimalQuantitys instead of doubles. 478 * 479 * Returns sample values for which select() would return the keyword. If 480 * the keyword is unknown, returns no values, but this is not an error. 481 * 482 * The number of returned values is typically small. 483 * 484 * @param keyword The keyword. 485 * @param dest Array into which to put the returned values. May 486 * be nullptr if destCapacity is 0. 487 * @param destCapacity The capacity of the array, must be at least 0. 488 * @param status The error code. 489 * @return The count of values written. 490 * If more than destCapacity samples are available, then 491 * only destCapacity are written, and destCapacity is returned as the count, 492 * rather than setting a U_BUFFER_OVERFLOW_ERROR. 493 * (The actual number of keyword values could be unlimited.) 494 * @internal 495 */ 496 int32_t getSamples(const UnicodeString &keyword, 497 DecimalQuantity *dest, int32_t destCapacity, 498 UErrorCode& status); 499 #endif /* U_HIDE_INTERNAL_API */ 500 501 /** 502 * Returns true if the given keyword is defined in this 503 * <code>PluralRules</code> object. 504 * 505 * @param keyword the input keyword. 506 * @return true if the input keyword is defined. 507 * Otherwise, return false. 508 * @stable ICU 4.0 509 */ 510 UBool isKeyword(const UnicodeString& keyword) const; 511 512 513 /** 514 * Returns keyword for default plural form. 515 * 516 * @return keyword for default plural form. 517 * @stable ICU 4.0 518 */ 519 UnicodeString getKeywordOther() const; 520 521 #ifndef U_HIDE_INTERNAL_API 522 /** 523 * 524 * @internal 525 */ 526 UnicodeString getRules() const; 527 #endif /* U_HIDE_INTERNAL_API */ 528 529 /** 530 * Compares the equality of two PluralRules objects. 531 * 532 * @param other The other PluralRules object to be compared with. 533 * @return true if the given PluralRules is the same as this 534 * PluralRules; false otherwise. 535 * @stable ICU 4.0 536 */ 537 virtual bool operator==(const PluralRules& other) const; 538 539 /** 540 * Compares the inequality of two PluralRules objects. 541 * 542 * @param other The PluralRules object to be compared with. 543 * @return true if the given PluralRules is not the same as this 544 * PluralRules; false otherwise. 545 * @stable ICU 4.0 546 */ 547 bool operator!=(const PluralRules& other) const {return !operator==(other);} 548 549 550 /** 551 * ICU "poor man's RTTI", returns a UClassID for this class. 552 * 553 * @stable ICU 4.0 554 * 555 */ 556 static UClassID U_EXPORT2 getStaticClassID(); 557 558 /** 559 * ICU "poor man's RTTI", returns a UClassID for the actual class. 560 * 561 * @stable ICU 4.0 562 */ 563 virtual UClassID getDynamicClassID() const override; 564 565 566 private: 567 RuleChain *mRules; 568 StandardPluralRanges *mStandardPluralRanges; 569 570 PluralRules() = delete; // default constructor not implemented 571 UnicodeString getRuleFromResource(const Locale& locale, UPluralType type, UErrorCode& status); 572 RuleChain *rulesForKeyword(const UnicodeString &keyword) const; 573 PluralRules *clone(UErrorCode& status) const; 574 575 /** 576 * An internal status variable used to indicate that the object is in an 'invalid' state. 577 * Used by copy constructor, the assignment operator and the clone method. 578 */ 579 UErrorCode mInternalStatus; 580 581 friend class PluralRuleParser; 582 }; 583 584 U_NAMESPACE_END 585 586 #endif /* #if !UCONFIG_NO_FORMATTING */ 587 588 #endif /* U_SHOW_CPLUSPLUS_API */ 589 590 #endif // _PLURRULE 591 //eof