ICU 66.0.1  66.0.1
coll.h
Go to the documentation of this file.
1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 ******************************************************************************
5 * Copyright (C) 1996-2016, International Business Machines
6 * Corporation and others. All Rights Reserved.
7 ******************************************************************************
8 */
9 
52 #ifndef COLL_H
53 #define COLL_H
54 
55 #include "unicode/utypes.h"
56 
57 #if U_SHOW_CPLUSPLUS_API
58 
59 #if !UCONFIG_NO_COLLATION
60 
61 #include "unicode/uobject.h"
62 #include "unicode/ucol.h"
63 #include "unicode/unorm.h"
64 #include "unicode/locid.h"
65 #include "unicode/uniset.h"
66 #include "unicode/umisc.h"
67 #include "unicode/uiter.h"
68 #include "unicode/stringpiece.h"
69 
70 U_NAMESPACE_BEGIN
71 
72 class StringEnumeration;
73 
74 #if !UCONFIG_NO_SERVICE
75 
78 class CollatorFactory;
79 #endif
80 
84 class CollationKey;
85 
167 class U_I18N_API Collator : public UObject {
168 public:
169 
170  // Collator public enums -----------------------------------------------
171 
198  {
199  PRIMARY = UCOL_PRIMARY, // 0
200  SECONDARY = UCOL_SECONDARY, // 1
201  TERTIARY = UCOL_TERTIARY, // 2
202  QUATERNARY = UCOL_QUATERNARY, // 3
203  IDENTICAL = UCOL_IDENTICAL // 15
204  };
205 
206 
207  // Cannot use #ifndef U_HIDE_DEPRECATED_API for the following, it is
208  // used by virtual methods that cannot have that conditional.
209 #ifndef U_FORCE_HIDE_DEPRECATED_API
210 
221  {
222  LESS = UCOL_LESS, // -1
223  EQUAL = UCOL_EQUAL, // 0
224  GREATER = UCOL_GREATER // 1
225  };
226 #endif // U_FORCE_HIDE_DEPRECATED_API
227 
228  // Collator public destructor -----------------------------------------
229 
234  virtual ~Collator();
235 
236  // Collator public methods --------------------------------------------
237 
256  virtual UBool operator==(const Collator& other) const;
257 
265  virtual UBool operator!=(const Collator& other) const;
266 
272  virtual Collator* clone() const = 0;
273 
293  static Collator* U_EXPORT2 createInstance(UErrorCode& err);
294 
328  static Collator* U_EXPORT2 createInstance(const Locale& loc, UErrorCode& err);
329 
330 #ifndef U_FORCE_HIDE_DEPRECATED_API
331 
342  virtual EComparisonResult compare(const UnicodeString& source,
343  const UnicodeString& target) const;
344 #endif // U_FORCE_HIDE_DEPRECATED_API
345 
358  virtual UCollationResult compare(const UnicodeString& source,
359  const UnicodeString& target,
360  UErrorCode &status) const = 0;
361 
362 #ifndef U_FORCE_HIDE_DEPRECATED_API
363 
375  virtual EComparisonResult compare(const UnicodeString& source,
376  const UnicodeString& target,
377  int32_t length) const;
378 #endif // U_FORCE_HIDE_DEPRECATED_API
379 
393  virtual UCollationResult compare(const UnicodeString& source,
394  const UnicodeString& target,
395  int32_t length,
396  UErrorCode &status) const = 0;
397 
398 #ifndef U_FORCE_HIDE_DEPRECATED_API
399 
432  virtual EComparisonResult compare(const char16_t* source, int32_t sourceLength,
433  const char16_t* target, int32_t targetLength)
434  const;
435 #endif // U_FORCE_HIDE_DEPRECATED_API
436 
453  virtual UCollationResult compare(const char16_t* source, int32_t sourceLength,
454  const char16_t* target, int32_t targetLength,
455  UErrorCode &status) const = 0;
456 
468  virtual UCollationResult compare(UCharIterator &sIter,
469  UCharIterator &tIter,
470  UErrorCode &status) const;
471 
485  virtual UCollationResult compareUTF8(const StringPiece &source,
486  const StringPiece &target,
487  UErrorCode &status) const;
488 
507  virtual CollationKey& getCollationKey(const UnicodeString& source,
508  CollationKey& key,
509  UErrorCode& status) const = 0;
510 
530  virtual CollationKey& getCollationKey(const char16_t*source,
531  int32_t sourceLength,
532  CollationKey& key,
533  UErrorCode& status) const = 0;
538  virtual int32_t hashCode(void) const = 0;
539 
540 #ifndef U_FORCE_HIDE_DEPRECATED_API
541 
553  virtual Locale getLocale(ULocDataLocaleType type, UErrorCode& status) const = 0;
554 #endif // U_FORCE_HIDE_DEPRECATED_API
555 
565  UBool greater(const UnicodeString& source, const UnicodeString& target)
566  const;
567 
577  UBool greaterOrEqual(const UnicodeString& source,
578  const UnicodeString& target) const;
579 
589  UBool equals(const UnicodeString& source, const UnicodeString& target) const;
590 
591 #ifndef U_FORCE_HIDE_DEPRECATED_API
592 
602  virtual ECollationStrength getStrength(void) const;
603 
622  virtual void setStrength(ECollationStrength newStrength);
623 #endif // U_FORCE_HIDE_DEPRECATED_API
624 
640  virtual int32_t getReorderCodes(int32_t *dest,
641  int32_t destCapacity,
642  UErrorCode& status) const;
643 
659  virtual void setReorderCodes(const int32_t* reorderCodes,
660  int32_t reorderCodesLength,
661  UErrorCode& status) ;
662 
683  static int32_t U_EXPORT2 getEquivalentReorderCodes(int32_t reorderCode,
684  int32_t* dest,
685  int32_t destCapacity,
686  UErrorCode& status);
687 
697  static UnicodeString& U_EXPORT2 getDisplayName(const Locale& objectLocale,
698  const Locale& displayLocale,
699  UnicodeString& name);
700 
709  static UnicodeString& U_EXPORT2 getDisplayName(const Locale& objectLocale,
710  UnicodeString& name);
711 
723  static const Locale* U_EXPORT2 getAvailableLocales(int32_t& count);
724 
733  static StringEnumeration* U_EXPORT2 getAvailableLocales(void);
734 
744  static StringEnumeration* U_EXPORT2 getKeywords(UErrorCode& status);
745 
757  static StringEnumeration* U_EXPORT2 getKeywordValues(const char *keyword, UErrorCode& status);
758 
775  static StringEnumeration* U_EXPORT2 getKeywordValuesForLocale(const char* keyword, const Locale& locale,
776  UBool commonlyUsed, UErrorCode& status);
777 
805  static Locale U_EXPORT2 getFunctionalEquivalent(const char* keyword, const Locale& locale,
806  UBool& isAvailable, UErrorCode& status);
807 
808 #if !UCONFIG_NO_SERVICE
809 
820  static URegistryKey U_EXPORT2 registerInstance(Collator* toAdopt, const Locale& locale, UErrorCode& status);
821 
832  static URegistryKey U_EXPORT2 registerFactory(CollatorFactory* toAdopt, UErrorCode& status);
833 
847  static UBool U_EXPORT2 unregister(URegistryKey key, UErrorCode& status);
848 #endif /* UCONFIG_NO_SERVICE */
849 
855  virtual void getVersion(UVersionInfo info) const = 0;
856 
867  virtual UClassID getDynamicClassID(void) const = 0;
868 
877  virtual void setAttribute(UColAttribute attr, UColAttributeValue value,
878  UErrorCode &status) = 0;
879 
888  virtual UColAttributeValue getAttribute(UColAttribute attr,
889  UErrorCode &status) const = 0;
890 
909  virtual Collator &setMaxVariable(UColReorderCode group, UErrorCode &errorCode);
910 
919  virtual UColReorderCode getMaxVariable() const;
920 
921 #ifndef U_FORCE_HIDE_DEPRECATED_API
922 
938  virtual uint32_t setVariableTop(const char16_t *varTop, int32_t len, UErrorCode &status) = 0;
939 
955  virtual uint32_t setVariableTop(const UnicodeString &varTop, UErrorCode &status) = 0;
956 
968  virtual void setVariableTop(uint32_t varTop, UErrorCode &status) = 0;
969 #endif // U_FORCE_HIDE_DEPRECATED_API
970 
978  virtual uint32_t getVariableTop(UErrorCode &status) const = 0;
979 
989  virtual UnicodeSet *getTailoredSet(UErrorCode &status) const;
990 
991 #ifndef U_FORCE_HIDE_DEPRECATED_API
992 
999  virtual Collator* safeClone() const;
1000 #endif // U_FORCE_HIDE_DEPRECATED_API
1001 
1018  virtual int32_t getSortKey(const UnicodeString& source,
1019  uint8_t* result,
1020  int32_t resultLength) const = 0;
1021 
1041  virtual int32_t getSortKey(const char16_t*source, int32_t sourceLength,
1042  uint8_t*result, int32_t resultLength) const = 0;
1043 
1081  static int32_t U_EXPORT2 getBound(const uint8_t *source,
1082  int32_t sourceLength,
1083  UColBoundMode boundType,
1084  uint32_t noOfLevels,
1085  uint8_t *result,
1086  int32_t resultLength,
1087  UErrorCode &status);
1088 
1089 
1090 protected:
1091 
1092  // Collator protected constructors -------------------------------------
1093 
1101  Collator();
1102 
1103 #ifndef U_HIDE_DEPRECATED_API
1104 
1115  Collator(UCollationStrength collationStrength,
1116  UNormalizationMode decompositionMode);
1117 #endif /* U_HIDE_DEPRECATED_API */
1118 
1124  Collator(const Collator& other);
1125 
1126 public:
1134  virtual void setLocales(const Locale& requestedLocale, const Locale& validLocale, const Locale& actualLocale);
1135 
1159  virtual int32_t internalGetShortDefinitionString(const char *locale,
1160  char *buffer,
1161  int32_t capacity,
1162  UErrorCode &status) const;
1163 
1168  virtual UCollationResult internalCompareUTF8(
1169  const char *left, int32_t leftLength,
1170  const char *right, int32_t rightLength,
1171  UErrorCode &errorCode) const;
1172 
1177  virtual int32_t
1178  internalNextSortKeyPart(
1179  UCharIterator *iter, uint32_t state[2],
1180  uint8_t *dest, int32_t count, UErrorCode &errorCode) const;
1181 
1182 #ifndef U_HIDE_INTERNAL_API
1183 
1184  static inline Collator *fromUCollator(UCollator *uc) {
1185  return reinterpret_cast<Collator *>(uc);
1186  }
1188  static inline const Collator *fromUCollator(const UCollator *uc) {
1189  return reinterpret_cast<const Collator *>(uc);
1190  }
1193  return reinterpret_cast<UCollator *>(this);
1194  }
1196  inline const UCollator *toUCollator() const {
1197  return reinterpret_cast<const UCollator *>(this);
1198  }
1199 #endif // U_HIDE_INTERNAL_API
1200 
1201 private:
1205  Collator& operator=(const Collator& other);
1206 
1207  friend class CFactory;
1208  friend class SimpleCFactory;
1209  friend class ICUCollatorFactory;
1210  friend class ICUCollatorService;
1211  static Collator* makeInstance(const Locale& desiredLocale,
1212  UErrorCode& status);
1213 };
1214 
1215 #if !UCONFIG_NO_SERVICE
1216 
1233 public:
1234 
1239  virtual ~CollatorFactory();
1240 
1248  virtual UBool visible(void) const;
1249 
1257  virtual Collator* createCollator(const Locale& loc) = 0;
1258 
1269  virtual UnicodeString& getDisplayName(const Locale& objectLocale,
1270  const Locale& displayLocale,
1271  UnicodeString& result);
1272 
1282  virtual const UnicodeString * getSupportedIDs(int32_t &count, UErrorCode& status) = 0;
1283 };
1284 #endif /* UCONFIG_NO_SERVICE */
1285 
1286 // Collator inline methods -----------------------------------------------
1287 
1288 U_NAMESPACE_END
1289 
1290 #endif /* #if !UCONFIG_NO_COLLATION */
1291 
1292 #endif /* U_SHOW_CPLUSPLUS_API */
1293 
1294 #endif
uint8_t UVersionInfo[U_MAX_VERSION_LENGTH]
The binary form of a version on ICU APIs is an array of 4 uint8_t.
Definition: uversion.h:59
string a < string b
Definition: ucol.h:79
The Collator class performs locale-sensitive string comparison.
Definition: coll.h:167
const UCollator * toUCollator() const
Definition: coll.h:1196
UCollationResult
UCOL_LESS is returned if source string is compared to be less than target string in the ucol_strcoll(...
Definition: ucol.h:73
UNormalizationMode
Constants for normalization modes.
Definition: unorm.h:140
U_EXPORT UBool operator==(const StringPiece &x, const StringPiece &y)
Global operator == for StringPiece.
void * UClassID
UClassID is used to identify classes without using the compiler&#39;s RTTI.
Definition: uobject.h:96
A factory, used with registerFactory, the creates multiple collators and provides display names for t...
Definition: coll.h:1232
C API:misc definitions.
C API for code unit iteration.
Definition: uiter.h:341
Base class for &#39;pure&#39; C++ implementations of uenum api.
Definition: strenum.h:61
string a == string b
Definition: ucol.h:75
static const Collator * fromUCollator(const UCollator *uc)
Definition: coll.h:1188
ECollationStrength
Base letter represents a primary difference.
Definition: coll.h:197
UColReorderCode
Enum containing the codes for reordering segments of the collation table that are not script codes...
Definition: ucol.h:146
UColAttribute
Attributes that collation service understands.
Definition: ucol.h:242
#define U_I18N_API
Set to export library symbols from inside the i18n library, and to import them from outside...
Definition: utypes.h:301
UCollator * toUCollator()
Definition: coll.h:1192
C++ API: StringPiece: Read-only byte string wrapper class.
Secondary collation strength.
Definition: ucol.h:96
UBool operator!=(const StringPiece &x, const StringPiece &y)
Global operator != for StringPiece.
Definition: stringpiece.h:251
C API: Unicode Normalization.
virtual UClassID getDynamicClassID() const
ICU4C "poor man&#39;s RTTI", returns a UClassID for the actual ICU class.
EComparisonResult
LESS is returned if source string is compared to be less than target string in the compare() method...
Definition: coll.h:220
C API: Collator.
Collation keys are generated by the Collator class.
Definition: sortkey.h:101
Tertiary collation strength.
Definition: ucol.h:98
A mutable set of Unicode characters and multicharacter strings.
Definition: uniset.h:281
C API: Unicode Character Iteration.
C++ API: Common ICU base class UObject.
const void * URegistryKey
Opaque type returned by registerInstance, registerFactory and unregister for service registration...
Definition: umisc.h:57
string a > string b
Definition: ucol.h:77
UErrorCode
Standard ICU4C error code type, a substitute for exceptions.
Definition: utypes.h:415
ULocDataLocaleType
Constants for *_getLocale() Allow user to select whether she wants information on requested...
Definition: uloc.h:338
UColBoundMode
enum that is taken by ucol_getBound API See below for explanation do not change the values assigned t...
Definition: ucol.h:1059
C++ API: Locale ID object.
struct UCollator UCollator
structure representing a collator object instance
Definition: ucol.h:58
Basic definitions for ICU, for both C and C++ APIs.
static Collator * fromUCollator(UCollator *uc)
Definition: coll.h:1184
Identical collation strength.
Definition: ucol.h:105
Quaternary collation strength.
Definition: ucol.h:103
UnicodeString is a string class that stores Unicode characters directly and provides similar function...
Definition: unistr.h:294
UColAttributeValue
Enum containing attribute values for controling collation behavior.
Definition: ucol.h:89
A string-like object that points to a sized piece of memory.
Definition: stringpiece.h:60
UObject is the common ICU "boilerplate" class.
Definition: uobject.h:223
Primary collation strength.
Definition: ucol.h:94
int8_t UBool
The ICU boolean type.
Definition: umachine.h:261
A Locale object represents a specific geographical, political, or cultural region.
Definition: locid.h:195
C++ API: Unicode Set.