1fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius/* 2fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius******************************************************************************* 3fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius* 4fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius* Copyright (C) 1999-2014, International Business Machines 5fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius* Corporation and others. All Rights Reserved. 6fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius* 7fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius******************************************************************************* 8fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius* file name: collationweights.h 9fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius* encoding: US-ASCII 10fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius* tab size: 8 (not used) 11fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius* indentation:4 12fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius* 13fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius* created on: 2001mar08 as ucol_wgt.h 14fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius* created by: Markus W. Scherer 15fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius*/ 16fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 17fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius#ifndef __COLLATIONWEIGHTS_H__ 18fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius#define __COLLATIONWEIGHTS_H__ 19fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 20fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius#include "unicode/utypes.h" 21fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 22fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius#if !UCONFIG_NO_COLLATION 23fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 24fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius#include "unicode/uobject.h" 25fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 26fceb39872958b9fa2505e63f8b8699a9e0f882f4ccorneliusU_NAMESPACE_BEGIN 27fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 28fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius/** 29fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * Allocates n collation element weights between two exclusive limits. 30fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * Used only internally by the collation tailoring builder. 31fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius */ 32fceb39872958b9fa2505e63f8b8699a9e0f882f4ccorneliusclass U_I18N_API CollationWeights : public UMemory { 33fceb39872958b9fa2505e63f8b8699a9e0f882f4ccorneliuspublic: 34fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius CollationWeights(); 35fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 36fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius static inline int32_t lengthOfWeight(uint32_t weight) { 37fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius if((weight&0xffffff)==0) { 38fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius return 1; 39fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } else if((weight&0xffff)==0) { 40fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius return 2; 41fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } else if((weight&0xff)==0) { 42fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius return 3; 43fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } else { 44fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius return 4; 45fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } 46fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } 47fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 48fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius void initForPrimary(UBool compressible); 49fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius void initForSecondary(); 50fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius void initForTertiary(); 51fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 52fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius /** 53fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * Determine heuristically 54fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * what ranges to use for a given number of weights between (excluding) 55fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * two limits. 56fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * 57fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * @param lowerLimit A collation element weight; the ranges will be filled to cover 58fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * weights greater than this one. 59fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * @param upperLimit A collation element weight; the ranges will be filled to cover 60fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * weights less than this one. 61fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * @param n The number of collation element weights w necessary such that 62fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * lowerLimit<w<upperLimit in lexical order. 63fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * @return TRUE if it is possible to fit n elements between the limits 64fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius */ 65fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius UBool allocWeights(uint32_t lowerLimit, uint32_t upperLimit, int32_t n); 66fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 67fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius /** 68fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * Given a set of ranges calculated by allocWeights(), 69fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * iterate through the weights. 70fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * The ranges are modified to keep the current iteration state. 71fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * 72fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * @return The next weight in the ranges, or 0xffffffff if there is none left. 73fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius */ 74fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius uint32_t nextWeight(); 75fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 76fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius /** @internal */ 77fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius struct WeightRange { 78fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius uint32_t start, end; 79fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius int32_t length, count; 80fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius }; 81fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 82fceb39872958b9fa2505e63f8b8699a9e0f882f4ccorneliusprivate: 83fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius /** @return number of usable byte values for byte idx */ 84fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius inline int32_t countBytes(int32_t idx) const { 85fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius return (int32_t)(maxBytes[idx] - minBytes[idx] + 1); 86fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius } 87fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 88fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius uint32_t incWeight(uint32_t weight, int32_t length) const; 89fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius uint32_t incWeightByOffset(uint32_t weight, int32_t length, int32_t offset) const; 90fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius void lengthenRange(WeightRange &range) const; 91fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius /** 92fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * Takes two CE weights and calculates the 93fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * possible ranges of weights between the two limits, excluding them. 94fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius * For weights with up to 4 bytes there are up to 2*4-1=7 ranges. 95fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius */ 96fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius UBool getWeightRanges(uint32_t lowerLimit, uint32_t upperLimit); 97fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius UBool allocWeightsInShortRanges(int32_t n, int32_t minLength); 98fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius UBool allocWeightsInMinLengthRanges(int32_t n, int32_t minLength); 99fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 100fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius int32_t middleLength; 101fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius uint32_t minBytes[5]; // for byte 1, 2, 3, 4 102fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius uint32_t maxBytes[5]; 103fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius WeightRange ranges[7]; 104fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius int32_t rangeIndex; 105fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius int32_t rangeCount; 106fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius}; 107fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 108fceb39872958b9fa2505e63f8b8699a9e0f882f4ccorneliusU_NAMESPACE_END 109fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius 110fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius#endif // !UCONFIG_NO_COLLATION 111fceb39872958b9fa2505e63f8b8699a9e0f882f4ccornelius#endif // __COLLATIONWEIGHTS_H__ 112