1 /* 2 ******************************************************************************* 3 * Copyright (C) 2013-2015, International Business Machines 4 * Corporation and others. All Rights Reserved. 5 ******************************************************************************* 6 * collationfastlatinbuilder.h 7 * 8 * created on: 2013aug09 9 * created by: Markus W. Scherer 10 */ 11 12 #ifndef __COLLATIONFASTLATINBUILDER_H__ 13 #define __COLLATIONFASTLATINBUILDER_H__ 14 15 #include "unicode/utypes.h" 16 17 #if !UCONFIG_NO_COLLATION 18 19 #include "unicode/unistr.h" 20 #include "unicode/uobject.h" 21 #include "collation.h" 22 #include "collationfastlatin.h" 23 #include "uvectr64.h" 24 25 U_NAMESPACE_BEGIN 26 27 struct CollationData; 28 29 class U_I18N_API CollationFastLatinBuilder : public UObject { 30 public: 31 CollationFastLatinBuilder(UErrorCode &errorCode); 32 ~CollationFastLatinBuilder(); 33 34 UBool forData(const CollationData &data, UErrorCode &errorCode); 35 getTable()36 const uint16_t *getTable() const { 37 return reinterpret_cast<const uint16_t *>(result.getBuffer()); 38 } lengthOfTable()39 int32_t lengthOfTable() const { return result.length(); } 40 41 private: 42 // space, punct, symbol, currency (not digit) 43 enum { NUM_SPECIAL_GROUPS = UCOL_REORDER_CODE_CURRENCY - UCOL_REORDER_CODE_FIRST + 1 }; 44 45 UBool loadGroups(const CollationData &data, UErrorCode &errorCode); 46 UBool inSameGroup(uint32_t p, uint32_t q) const; 47 48 void resetCEs(); 49 void getCEs(const CollationData &data, UErrorCode &errorCode); 50 UBool getCEsFromCE32(const CollationData &data, UChar32 c, uint32_t ce32, 51 UErrorCode &errorCode); 52 UBool getCEsFromContractionCE32(const CollationData &data, uint32_t ce32, 53 UErrorCode &errorCode); 54 void addContractionEntry(int32_t x, int64_t cce0, int64_t cce1, UErrorCode &errorCode); 55 void addUniqueCE(int64_t ce, UErrorCode &errorCode); 56 uint32_t getMiniCE(int64_t ce) const; 57 UBool encodeUniqueCEs(UErrorCode &errorCode); 58 UBool encodeCharCEs(UErrorCode &errorCode); 59 UBool encodeContractions(UErrorCode &errorCode); 60 uint32_t encodeTwoCEs(int64_t first, int64_t second) const; 61 isContractionCharCE(int64_t ce)62 static UBool isContractionCharCE(int64_t ce) { 63 return (uint32_t)(ce >> 32) == Collation::NO_CE_PRIMARY && ce != Collation::NO_CE; 64 } 65 66 static const uint32_t CONTRACTION_FLAG = 0x80000000; 67 68 // temporary "buffer" 69 int64_t ce0, ce1; 70 71 int64_t charCEs[CollationFastLatin::NUM_FAST_CHARS][2]; 72 73 UVector64 contractionCEs; 74 UVector64 uniqueCEs; 75 76 /** One 16-bit mini CE per unique CE. */ 77 uint16_t *miniCEs; 78 79 // These are constant for a given root collator. 80 uint32_t lastSpecialPrimaries[NUM_SPECIAL_GROUPS]; 81 uint32_t firstDigitPrimary; 82 uint32_t firstLatinPrimary; 83 uint32_t lastLatinPrimary; 84 // This determines the first normal primary weight which is mapped to 85 // a short mini primary. It must be >=firstDigitPrimary. 86 uint32_t firstShortPrimary; 87 88 UBool shortPrimaryOverflow; 89 90 UnicodeString result; 91 int32_t headerLength; 92 }; 93 94 U_NAMESPACE_END 95 96 #endif // !UCONFIG_NO_COLLATION 97 #endif // __COLLATIONFASTLATINBUILDER_H__ 98