1 // © 2016 and later: Unicode, Inc. and others. 2 // License & terms of use: http://www.unicode.org/copyright.html 3 /* 4 ******************************************************************************* 5 * 6 * Copyright (C) 1999-2014, International Business Machines 7 * Corporation and others. All Rights Reserved. 8 * 9 ******************************************************************************* 10 * file name: collationweights.h 11 * encoding: UTF-8 12 * tab size: 8 (not used) 13 * indentation:4 14 * 15 * created on: 2001mar08 as ucol_wgt.h 16 * created by: Markus W. Scherer 17 */ 18 19 #ifndef __COLLATIONWEIGHTS_H__ 20 #define __COLLATIONWEIGHTS_H__ 21 22 #include "unicode/utypes.h" 23 24 #if !UCONFIG_NO_COLLATION 25 26 #include "unicode/uobject.h" 27 28 U_NAMESPACE_BEGIN 29 30 /** 31 * Allocates n collation element weights between two exclusive limits. 32 * Used only internally by the collation tailoring builder. 33 */ 34 class U_I18N_API CollationWeights : public UMemory { 35 public: 36 CollationWeights(); 37 lengthOfWeight(uint32_t weight)38 static inline int32_t lengthOfWeight(uint32_t weight) { 39 if((weight&0xffffff)==0) { 40 return 1; 41 } else if((weight&0xffff)==0) { 42 return 2; 43 } else if((weight&0xff)==0) { 44 return 3; 45 } else { 46 return 4; 47 } 48 } 49 50 void initForPrimary(UBool compressible); 51 void initForSecondary(); 52 void initForTertiary(); 53 54 /** 55 * Determine heuristically 56 * what ranges to use for a given number of weights between (excluding) 57 * two limits. 58 * 59 * @param lowerLimit A collation element weight; the ranges will be filled to cover 60 * weights greater than this one. 61 * @param upperLimit A collation element weight; the ranges will be filled to cover 62 * weights less than this one. 63 * @param n The number of collation element weights w necessary such that 64 * lowerLimit<w<upperLimit in lexical order. 65 * @return TRUE if it is possible to fit n elements between the limits 66 */ 67 UBool allocWeights(uint32_t lowerLimit, uint32_t upperLimit, int32_t n); 68 69 /** 70 * Given a set of ranges calculated by allocWeights(), 71 * iterate through the weights. 72 * The ranges are modified to keep the current iteration state. 73 * 74 * @return The next weight in the ranges, or 0xffffffff if there is none left. 75 */ 76 uint32_t nextWeight(); 77 78 /** @internal */ 79 struct WeightRange { 80 uint32_t start, end; 81 int32_t length, count; 82 }; 83 84 private: 85 /** @return number of usable byte values for byte idx */ countBytes(int32_t idx)86 inline int32_t countBytes(int32_t idx) const { 87 return (int32_t)(maxBytes[idx] - minBytes[idx] + 1); 88 } 89 90 uint32_t incWeight(uint32_t weight, int32_t length) const; 91 uint32_t incWeightByOffset(uint32_t weight, int32_t length, int32_t offset) const; 92 void lengthenRange(WeightRange &range) const; 93 /** 94 * Takes two CE weights and calculates the 95 * possible ranges of weights between the two limits, excluding them. 96 * For weights with up to 4 bytes there are up to 2*4-1=7 ranges. 97 */ 98 UBool getWeightRanges(uint32_t lowerLimit, uint32_t upperLimit); 99 UBool allocWeightsInShortRanges(int32_t n, int32_t minLength); 100 UBool allocWeightsInMinLengthRanges(int32_t n, int32_t minLength); 101 102 int32_t middleLength; 103 uint32_t minBytes[5]; // for byte 1, 2, 3, 4 104 uint32_t maxBytes[5]; 105 WeightRange ranges[7]; 106 int32_t rangeIndex; 107 int32_t rangeCount; 108 }; 109 110 U_NAMESPACE_END 111 112 #endif // !UCONFIG_NO_COLLATION 113 #endif // __COLLATIONWEIGHTS_H__ 114