1b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* 2b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru******************************************************************************* 3b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* 4f9878a236aa0d9662d8e40cafdaf2e04cd615835ccornelius* Copyright (C) 2004-2014, International Business Machines 5b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* Corporation and others. All Rights Reserved. 6b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* 7b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru******************************************************************************* 883a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius* file name: ucase.cpp 9b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* encoding: US-ASCII 10b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* tab size: 8 (not used) 11b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* indentation:4 12b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* 13b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* created on: 2004aug30 14b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* created by: Markus W. Scherer 15b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* 16b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* Low-level Unicode character/string case mapping code. 17b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* Much code moved here (and modified) from uchar.c. 18b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru*/ 19b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 20b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "unicode/utypes.h" 2183a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius#include "unicode/unistr.h" 22b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "unicode/uset.h" 23b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "unicode/udata.h" /* UDataInfo */ 2483a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius#include "unicode/utf16.h" 25b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "ucmndata.h" /* DataHeader */ 26b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "udatamem.h" 27b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "umutex.h" 28b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "uassert.h" 29b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "cmemory.h" 30b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru#include "utrie2.h" 31b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "ucase.h" 32b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 33b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustruct UCaseProps { 34b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UDataMemory *mem; 35b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const int32_t *indexes; 36b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const uint16_t *exceptions; 3783a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius const uint16_t *unfold; 38b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 39b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru UTrie2 trie; 40b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uint8_t formatVersion[4]; 41b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru}; 42b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 4383a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius/* ucase_props_data.h is machine-generated by gencase --csource */ 4483a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius#define INCLUDED_FROM_UCASE_CPP 4583a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius#include "ucase_props_data.h" 46b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 47b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* UCaseProps singleton ----------------------------------------------------- */ 48b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 49b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI const UCaseProps * U_EXPORT2 5027f654740f2a26ad62a5c155af9199af9e69b889clairehoucase_getSingleton() { 51b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return &ucase_props_singleton; 52b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 53b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 54b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* set of property starts for UnicodeSet ------------------------------------ */ 55b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 56b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic UBool U_CALLCONV 5783a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius_enumPropertyStartsRange(const void *context, UChar32 start, UChar32 /*end*/, uint32_t /*value*/) { 58b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* add the start code point to the USet */ 59b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const USetAdder *sa=(const USetAdder *)context; 60b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru sa->add(sa->set, start); 61b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return TRUE; 62b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 63b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 64b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CFUNC void U_EXPORT2 65b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruucase_addPropertyStarts(const UCaseProps *csp, const USetAdder *sa, UErrorCode *pErrorCode) { 66b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(U_FAILURE(*pErrorCode)) { 67b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 68b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 69b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 70b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* add the start code point of each same-value range of the trie */ 71b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru utrie2_enum(&csp->trie, NULL, _enumPropertyStartsRange, sa); 72b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 73b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* add code points with hardcoded properties, plus the ones following them */ 74b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 75b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* (none right now, see comment below) */ 76b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 77b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 78b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Omit code points with hardcoded specialcasing properties 79b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * because we do not build property UnicodeSets for them right now. 80b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 81b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 82b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 83b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* data access primitives --------------------------------------------------- */ 84b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 85b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define GET_EXCEPTIONS(csp, props) ((csp)->exceptions+((props)>>UCASE_EXC_SHIFT)) 86b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 87b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define PROPS_HAS_EXCEPTION(props) ((props)&UCASE_EXCEPTION) 88b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 89b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* number of bits in an 8-bit integer value */ 90b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic const uint8_t flagsOffset[256]={ 91b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4, 92b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, 93b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, 94b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 95b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, 96b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 97b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 98b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, 99b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, 100b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 101b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 102b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, 103b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 104b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, 105b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, 106b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 4, 5, 5, 6, 5, 6, 6, 7, 5, 6, 6, 7, 6, 7, 7, 8 107b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru}; 108b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 109b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru#define HAS_SLOT(flags, idx) ((flags)&(1<<(idx))) 110b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru#define SLOT_OFFSET(flags, idx) flagsOffset[(flags)&((1<<(idx))-1)] 111b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 112b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* 113b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * Get the value of an optional-value slot where HAS_SLOT(excWord, idx). 114b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 115b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param excWord (in) initial exceptions word 116b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * @param idx (in) desired slot index 117b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param pExc16 (in/out) const uint16_t * after excWord=*pExc16++; 118b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * moved to the last uint16_t of the value, use +1 for beginning of next slot 119b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param value (out) int32_t or uint32_t output if hasSlot, otherwise not modified 120b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 121b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru#define GET_SLOT_VALUE(excWord, idx, pExc16, value) \ 122b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(((excWord)&UCASE_EXC_DOUBLE_SLOTS)==0) { \ 123b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru (pExc16)+=SLOT_OFFSET(excWord, idx); \ 124b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru (value)=*pExc16; \ 125b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { \ 126b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru (pExc16)+=2*SLOT_OFFSET(excWord, idx); \ 127b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru (value)=*pExc16++; \ 128b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru (value)=((value)<<16)|*pExc16; \ 129b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 130b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 131b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* simple case mappings ----------------------------------------------------- */ 132b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 133b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI UChar32 U_EXPORT2 134b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruucase_tolower(const UCaseProps *csp, UChar32 c) { 135b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru uint16_t props=UTRIE2_GET16(&csp->trie, c); 136b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(!PROPS_HAS_EXCEPTION(props)) { 137b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(UCASE_GET_TYPE(props)>=UCASE_UPPER) { 138b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c+=UCASE_GET_DELTA(props); 139b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 140b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 141b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const uint16_t *pe=GET_EXCEPTIONS(csp, props); 142b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uint16_t excWord=*pe++; 143b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(HAS_SLOT(excWord, UCASE_EXC_LOWER)) { 144b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru GET_SLOT_VALUE(excWord, UCASE_EXC_LOWER, pe, c); 145b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 146b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 147b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return c; 148b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 149b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 150b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI UChar32 U_EXPORT2 151b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruucase_toupper(const UCaseProps *csp, UChar32 c) { 152b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru uint16_t props=UTRIE2_GET16(&csp->trie, c); 153b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(!PROPS_HAS_EXCEPTION(props)) { 154b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(UCASE_GET_TYPE(props)==UCASE_LOWER) { 155b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c+=UCASE_GET_DELTA(props); 156b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 157b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 158b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const uint16_t *pe=GET_EXCEPTIONS(csp, props); 159b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uint16_t excWord=*pe++; 160b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(HAS_SLOT(excWord, UCASE_EXC_UPPER)) { 161b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru GET_SLOT_VALUE(excWord, UCASE_EXC_UPPER, pe, c); 162b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 163b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 164b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return c; 165b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 166b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 167b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI UChar32 U_EXPORT2 168b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruucase_totitle(const UCaseProps *csp, UChar32 c) { 169b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru uint16_t props=UTRIE2_GET16(&csp->trie, c); 170b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(!PROPS_HAS_EXCEPTION(props)) { 171b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(UCASE_GET_TYPE(props)==UCASE_LOWER) { 172b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c+=UCASE_GET_DELTA(props); 173b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 174b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 175b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const uint16_t *pe=GET_EXCEPTIONS(csp, props); 176b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uint16_t excWord=*pe++; 177b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru int32_t idx; 178b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(HAS_SLOT(excWord, UCASE_EXC_TITLE)) { 179b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru idx=UCASE_EXC_TITLE; 180b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if(HAS_SLOT(excWord, UCASE_EXC_UPPER)) { 181b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru idx=UCASE_EXC_UPPER; 182b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 183b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return c; 184b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 185b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru GET_SLOT_VALUE(excWord, idx, pe, c); 186b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 187b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return c; 188b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 189b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 190c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Querustatic const UChar iDot[2] = { 0x69, 0x307 }; 191c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Querustatic const UChar jDot[2] = { 0x6a, 0x307 }; 192c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Querustatic const UChar iOgonekDot[3] = { 0x12f, 0x307 }; 193c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Querustatic const UChar iDotGrave[3] = { 0x69, 0x307, 0x300 }; 194c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Querustatic const UChar iDotAcute[3] = { 0x69, 0x307, 0x301 }; 195c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Querustatic const UChar iDotTilde[3] = { 0x69, 0x307, 0x303 }; 196c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 197c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 198b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CFUNC void U_EXPORT2 199b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruucase_addCaseClosure(const UCaseProps *csp, UChar32 c, const USetAdder *sa) { 200b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uint16_t props; 201b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 202b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 203b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Hardcode the case closure of i and its relatives and ignore the 204b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * data file data for these characters. 205b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * The Turkic dotless i and dotted I with their case mapping conditions 206b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * and case folding option make the related characters behave specially. 207b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * This code matches their closure behavior to their case folding behavior. 208b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 209b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 210b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru switch(c) { 211b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 0x49: 212b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* regular i and I are in one equivalence class */ 213b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru sa->add(sa->set, 0x69); 214b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 215b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 0x69: 216b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru sa->add(sa->set, 0x49); 217b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 218b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 0x130: 219b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* dotted I is in a class with <0069 0307> (for canonical equivalence with <0049 0307>) */ 220b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru sa->addString(sa->set, iDot, 2); 221b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 222b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 0x131: 223b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* dotless i is in a class by itself */ 224b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 225b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru default: 226b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* otherwise use the data file data */ 227b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 228b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 229b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 230b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru props=UTRIE2_GET16(&csp->trie, c); 231b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(!PROPS_HAS_EXCEPTION(props)) { 232b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(UCASE_GET_TYPE(props)!=UCASE_NONE) { 233b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* add the one simple case mapping, no matter what type it is */ 234b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t delta=UCASE_GET_DELTA(props); 235b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(delta!=0) { 236b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru sa->add(sa->set, c+delta); 237b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 238b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 239b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 240b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 241b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * c has exceptions, so there may be multiple simple and/or 242b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * full case mappings. Add them all. 243b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 244b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const uint16_t *pe0, *pe=GET_EXCEPTIONS(csp, props); 245b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const UChar *closure; 246b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uint16_t excWord=*pe++; 247b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru int32_t idx, closureLength, fullLength, length; 248b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 249b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pe0=pe; 250b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 251b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* add all simple case mappings */ 252b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru for(idx=UCASE_EXC_LOWER; idx<=UCASE_EXC_TITLE; ++idx) { 253b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru if(HAS_SLOT(excWord, idx)) { 254b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pe=pe0; 255b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru GET_SLOT_VALUE(excWord, idx, pe, c); 256b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru sa->add(sa->set, c); 257b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 258b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 259b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 260b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* get the closure string pointer & length */ 261b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(HAS_SLOT(excWord, UCASE_EXC_CLOSURE)) { 262b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pe=pe0; 263b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru GET_SLOT_VALUE(excWord, UCASE_EXC_CLOSURE, pe, closureLength); 264b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru closureLength&=UCASE_CLOSURE_MAX_LENGTH; /* higher bits are reserved */ 265b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru closure=(const UChar *)pe+1; /* behind this slot, unless there are full case mappings */ 266b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 267b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru closureLength=0; 268b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru closure=NULL; 269b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 270b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 271b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* add the full case folding */ 272b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(HAS_SLOT(excWord, UCASE_EXC_FULL_MAPPINGS)) { 273b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pe=pe0; 274b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru GET_SLOT_VALUE(excWord, UCASE_EXC_FULL_MAPPINGS, pe, fullLength); 275b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 276b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* start of full case mapping strings */ 277b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ++pe; 278b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 279b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru fullLength&=0xffff; /* bits 16 and higher are reserved */ 280b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 281b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* skip the lowercase result string */ 282b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pe+=fullLength&UCASE_FULL_LOWER; 283b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru fullLength>>=4; 284b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 285b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* add the full case folding string */ 286b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru length=fullLength&0xf; 287b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(length!=0) { 288b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru sa->addString(sa->set, (const UChar *)pe, length); 289b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pe+=length; 290b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 291b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 292b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* skip the uppercase and titlecase strings */ 293b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru fullLength>>=4; 294b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pe+=fullLength&0xf; 295b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru fullLength>>=4; 296b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pe+=fullLength; 297b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 298b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru closure=(const UChar *)pe; /* behind full case mappings */ 299b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 300b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 301b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* add each code point in the closure string */ 302b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru for(idx=0; idx<closureLength;) { 303b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru U16_NEXT_UNSAFE(closure, idx, c); 304b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru sa->add(sa->set, c); 305b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 306b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 307b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 308b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 309b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* 310b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * compare s, which has a length, with t, which has a maximum length or is NUL-terminated 311b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * must be length>0 and max>0 and length<=max 312b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 31383a171d1a62abf406f7f44ae671823d5ec20db7dCraig Corneliusstatic inline int32_t 314b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QuerustrcmpMax(const UChar *s, int32_t length, const UChar *t, int32_t max) { 315b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t c1, c2; 316b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 317b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru max-=length; /* we require length<=max, so no need to decrement max in the loop */ 318b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru do { 319b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c1=*s++; 320b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c2=*t++; 321b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(c2==0) { 322b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 1; /* reached the end of t but not of s */ 323b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 324b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c1-=c2; 325b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(c1!=0) { 326b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return c1; /* return difference result */ 327b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 328b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } while(--length>0); 329b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* ends with length==0 */ 330b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 331b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(max==0 || *t==0) { 332b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 0; /* equal to length of both strings */ 333b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 334b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return -max; /* return lengh difference */ 335b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 336b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 337b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 338b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CFUNC UBool U_EXPORT2 339b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruucase_addStringCaseClosure(const UCaseProps *csp, const UChar *s, int32_t length, const USetAdder *sa) { 340b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t i, start, limit, result, unfoldRows, unfoldRowWidth, unfoldStringWidth; 341b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 342b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(csp->unfold==NULL || s==NULL) { 343b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; /* no reverse case folding data, or no string */ 344b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 345b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(length<=1) { 346b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* the string is too short to find any match */ 347b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 348b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * more precise would be: 349b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * if(!u_strHasMoreChar32Than(s, length, 1)) 350b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * but this does not make much practical difference because 351b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * a single supplementary code point would just not be found 352b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 353b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; 354b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 355b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 35683a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius const uint16_t *unfold=csp->unfold; 357b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru unfoldRows=unfold[UCASE_UNFOLD_ROWS]; 358b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru unfoldRowWidth=unfold[UCASE_UNFOLD_ROW_WIDTH]; 359b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru unfoldStringWidth=unfold[UCASE_UNFOLD_STRING_WIDTH]; 360b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru unfold+=unfoldRowWidth; 361b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 362b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(length>unfoldStringWidth) { 363b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* the string is too long to find any match */ 364b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; 365b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 366b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 367b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* do a binary search for the string */ 368b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru start=0; 369b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru limit=unfoldRows; 370b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru while(start<limit) { 371b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru i=(start+limit)/2; 37283a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius const UChar *p=reinterpret_cast<const UChar *>(unfold+(i*unfoldRowWidth)); 373b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru result=strcmpMax(s, length, p, unfoldStringWidth); 374b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 375b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(result==0) { 376b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* found the string: add each code point, and its case closure */ 377b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar32 c; 378b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 379b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru for(i=unfoldStringWidth; i<unfoldRowWidth && p[i]!=0;) { 380b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru U16_NEXT_UNSAFE(p, i, c); 381b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru sa->add(sa->set, c); 382b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ucase_addCaseClosure(csp, c, sa); 383b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 384b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return TRUE; 385b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if(result<0) { 386b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru limit=i; 387b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else /* result>0 */ { 388b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru start=i+1; 389b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 390b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 391b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 392b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; /* string not found */ 393b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 394b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 39583a171d1a62abf406f7f44ae671823d5ec20db7dCraig CorneliusU_NAMESPACE_BEGIN 39683a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius 39783a171d1a62abf406f7f44ae671823d5ec20db7dCraig CorneliusFullCaseFoldingIterator::FullCaseFoldingIterator() 39883a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius : unfold(reinterpret_cast<const UChar *>(ucase_props_singleton.unfold)), 39983a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius unfoldRows(unfold[UCASE_UNFOLD_ROWS]), 40083a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius unfoldRowWidth(unfold[UCASE_UNFOLD_ROW_WIDTH]), 40183a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius unfoldStringWidth(unfold[UCASE_UNFOLD_STRING_WIDTH]), 40283a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius currentRow(0), 40383a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius rowCpIndex(unfoldStringWidth) { 40483a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius unfold+=unfoldRowWidth; 40583a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius} 40683a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius 40783a171d1a62abf406f7f44ae671823d5ec20db7dCraig CorneliusUChar32 40883a171d1a62abf406f7f44ae671823d5ec20db7dCraig CorneliusFullCaseFoldingIterator::next(UnicodeString &full) { 40983a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius // Advance past the last-delivered code point. 41083a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius const UChar *p=unfold+(currentRow*unfoldRowWidth); 41183a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius if(rowCpIndex>=unfoldRowWidth || p[rowCpIndex]==0) { 41283a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius ++currentRow; 41383a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius p+=unfoldRowWidth; 41483a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius rowCpIndex=unfoldStringWidth; 41583a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius } 41683a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius if(currentRow>=unfoldRows) { return U_SENTINEL; } 41783a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius // Set "full" to the NUL-terminated string in the first unfold column. 41883a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius int32_t length=unfoldStringWidth; 41983a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius while(length>0 && p[length-1]==0) { --length; } 42083a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius full.setTo(FALSE, p, length); 42183a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius // Return the code point. 42283a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius UChar32 c; 42383a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius U16_NEXT_UNSAFE(p, rowCpIndex, c); 42483a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius return c; 42583a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius} 42683a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius 42783a171d1a62abf406f7f44ae671823d5ec20db7dCraig CorneliusU_NAMESPACE_END 42883a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius 429b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** @return UCASE_NONE, UCASE_LOWER, UCASE_UPPER, UCASE_TITLE */ 430b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI int32_t U_EXPORT2 431b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruucase_getType(const UCaseProps *csp, UChar32 c) { 432b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru uint16_t props=UTRIE2_GET16(&csp->trie, c); 433b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return UCASE_GET_TYPE(props); 434b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 435b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 43650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho/** @return same as ucase_getType() and set bit 2 if c is case-ignorable */ 437b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI int32_t U_EXPORT2 438b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruucase_getTypeOrIgnorable(const UCaseProps *csp, UChar32 c) { 439b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru uint16_t props=UTRIE2_GET16(&csp->trie, c); 44083a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius return UCASE_GET_TYPE_AND_IGNORABLE(props); 441b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 442b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 443b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** @return UCASE_NO_DOT, UCASE_SOFT_DOTTED, UCASE_ABOVE, UCASE_OTHER_ACCENT */ 44483a171d1a62abf406f7f44ae671823d5ec20db7dCraig Corneliusstatic inline int32_t 445b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QuerugetDotType(const UCaseProps *csp, UChar32 c) { 446b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru uint16_t props=UTRIE2_GET16(&csp->trie, c); 447b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(!PROPS_HAS_EXCEPTION(props)) { 448b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return props&UCASE_DOT_MASK; 449b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 450b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const uint16_t *pe=GET_EXCEPTIONS(csp, props); 451b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return (*pe>>UCASE_EXC_DOT_SHIFT)&UCASE_DOT_MASK; 452b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 453b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 454b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 455b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI UBool U_EXPORT2 456b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruucase_isSoftDotted(const UCaseProps *csp, UChar32 c) { 457b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return (UBool)(getDotType(csp, c)==UCASE_SOFT_DOTTED); 458b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 459b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 460b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI UBool U_EXPORT2 461b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruucase_isCaseSensitive(const UCaseProps *csp, UChar32 c) { 462b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru uint16_t props=UTRIE2_GET16(&csp->trie, c); 463b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return (UBool)((props&UCASE_SENSITIVE)!=0); 464b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 465b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 466b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* string casing ------------------------------------------------------------ */ 467b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 468b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* 469b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * These internal functions form the core of string case mappings. 470b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * They map single code points to result code points or strings and take 471b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * all necessary conditions (context, locale ID, options) into account. 472b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 473b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * They do not iterate over the source or write to the destination 474b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * so that the same functions are useful for non-standard string storage, 475b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * such as in a Replaceable (for Transliterator) or UTF-8/32 strings etc. 476b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * For the same reason, the "surrounding text" context is passed in as a 477b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * UCaseContextIterator which does not make any assumptions about 478b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * the underlying storage. 479b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 480b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * This section contains helper functions that check for conditions 481b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * in the input text surrounding the current code point 482b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * according to SpecialCasing.txt. 483b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 484b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Each helper function gets the index 485b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * - after the current code point if it looks at following text 486b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * - before the current code point if it looks at preceding text 487b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 488b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Unicode 3.2 UAX 21 "Case Mappings" defines the conditions as follows: 489b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 490b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Final_Sigma 491b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * C is preceded by a sequence consisting of 492b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * a cased letter and a case-ignorable sequence, 493b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * and C is not followed by a sequence consisting of 494b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * an ignorable sequence and then a cased letter. 495b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 496b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * More_Above 497b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * C is followed by one or more characters of combining class 230 (ABOVE) 498b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * in the combining character sequence. 499b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 500b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * After_Soft_Dotted 501b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * The last preceding character with combining class of zero before C 502b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * was Soft_Dotted, 503b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * and there is no intervening combining character class 230 (ABOVE). 504b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 505b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Before_Dot 506b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * C is followed by combining dot above (U+0307). 507b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Any sequence of characters with a combining class that is neither 0 nor 230 508b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * may intervene between the current character and the combining dot above. 509b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 510b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * The erratum from 2002-10-31 adds the condition 511b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 512b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * After_I 513b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * The last preceding base character was an uppercase I, and there is no 514b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * intervening combining character class 230 (ABOVE). 515b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 516b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * (See Jitterbug 2344 and the comments on After_I below.) 517b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 518b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Helper definitions in Unicode 3.2 UAX 21: 519b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 520b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * D1. A character C is defined to be cased 521b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * if it meets any of the following criteria: 522b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 523b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * - The general category of C is Titlecase Letter (Lt) 524b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * - In [CoreProps], C has one of the properties Uppercase, or Lowercase 525b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * - Given D = NFD(C), then it is not the case that: 526b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * D = UCD_lower(D) = UCD_upper(D) = UCD_title(D) 527b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * (This third criterium does not add any characters to the list 528b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * for Unicode 3.2. Ignored.) 529b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 530b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * D2. A character C is defined to be case-ignorable 531b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * if it meets either of the following criteria: 532b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 533b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * - The general category of C is 534b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Nonspacing Mark (Mn), or Enclosing Mark (Me), or Format Control (Cf), or 535b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Letter Modifier (Lm), or Symbol Modifier (Sk) 536b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * - C is one of the following characters 537b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * U+0027 APOSTROPHE 538b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * U+00AD SOFT HYPHEN (SHY) 539b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * U+2019 RIGHT SINGLE QUOTATION MARK 540b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * (the preferred character for apostrophe) 541b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 542b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * D3. A case-ignorable sequence is a sequence of 543b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * zero or more case-ignorable characters. 544b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 545b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 546b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define is_a(c) ((c)=='a' || (c)=='A') 547c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru#define is_d(c) ((c)=='d' || (c)=='D') 548b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define is_e(c) ((c)=='e' || (c)=='E') 549b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define is_i(c) ((c)=='i' || (c)=='I') 550b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define is_l(c) ((c)=='l' || (c)=='L') 551c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru#define is_n(c) ((c)=='n' || (c)=='N') 552b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define is_r(c) ((c)=='r' || (c)=='R') 553b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define is_t(c) ((c)=='t' || (c)=='T') 554b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define is_u(c) ((c)=='u' || (c)=='U') 555b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define is_z(c) ((c)=='z' || (c)=='Z') 556b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 557b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* separator? */ 558b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define is_sep(c) ((c)=='_' || (c)=='-' || (c)==0) 559b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 560b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 561b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Requires non-NULL locale ID but otherwise does the equivalent of 562b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * checking for language codes as if uloc_getLanguage() were called: 563b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Accepts both 2- and 3-letter codes and accepts case variants. 564b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 565b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CFUNC int32_t 566b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruucase_getCaseLocale(const char *locale, int32_t *locCache) { 567b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t result; 568b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru char c; 569b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 570c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru if(locCache!=NULL && (result=*locCache)!=UCASE_LOC_UNKNOWN) { 571b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return result; 572b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 573b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 574c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru result=UCASE_LOC_ROOT; 575b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 576b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 577b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * This function used to use uloc_getLanguage(), but the current code 578b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * removes the dependency of this low-level code on uloc implementation code 579b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * and is faster because not the whole locale ID has to be 580b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * examined and copied/transformed. 581b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 582b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Because this code does not want to depend on uloc, the caller must 583b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * pass in a non-NULL locale, i.e., may need to call uloc_getDefault(). 584b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 585b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c=*locale++; 586b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(is_t(c)) { 587b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* tr or tur? */ 588b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c=*locale++; 589b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(is_u(c)) { 590b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c=*locale++; 591b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 592b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(is_r(c)) { 593b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c=*locale; 594b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(is_sep(c)) { 595c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru result=UCASE_LOC_TURKISH; 596b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 597b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 598b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if(is_a(c)) { 599b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* az or aze? */ 600b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c=*locale++; 601b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(is_z(c)) { 602b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c=*locale++; 603b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(is_e(c)) { 604b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c=*locale; 605b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 606b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(is_sep(c)) { 607c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru result=UCASE_LOC_TURKISH; 608b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 609b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 610b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if(is_l(c)) { 611b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* lt or lit? */ 612b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c=*locale++; 613b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(is_i(c)) { 614b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c=*locale++; 615b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 616b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(is_t(c)) { 617b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c=*locale; 618b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(is_sep(c)) { 619c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru result=UCASE_LOC_LITHUANIAN; 620c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru } 621c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru } 622c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru } else if(is_n(c)) { 623c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru /* nl or nld? */ 624c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru c=*locale++; 625c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru if(is_l(c)) { 626c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru c=*locale++; 627c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru if(is_d(c)) { 628c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru c=*locale; 629c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru } 630c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru if(is_sep(c)) { 631c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru result=UCASE_LOC_DUTCH; 632b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 633b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 634b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 635b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 636b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(locCache!=NULL) { 637b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru *locCache=result; 638b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 639b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return result; 640b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 641b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 64250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho/* 64350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * Is followed by 64450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * {case-ignorable}* cased 64550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * ? 64650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * (dir determines looking forward/backward) 64750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * If a character is case-ignorable, it is skipped regardless of whether 64850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * it is also cased or not. 64950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho */ 650b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic UBool 651b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruisFollowedByCasedLetter(const UCaseProps *csp, UCaseContextIterator *iter, void *context, int8_t dir) { 652b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar32 c; 653b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 654b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(iter==NULL) { 655b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; 656b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 657b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 658b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru for(/* dir!=0 sets direction */; (c=iter(context, dir))>=0; dir=0) { 65950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho int32_t type=ucase_getTypeOrIgnorable(csp, c); 66050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho if(type&4) { 661b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* case-ignorable, continue with the loop */ 66250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho } else if(type!=UCASE_NONE) { 66350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho return TRUE; /* followed by cased letter */ 664b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 66550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho return FALSE; /* uncased and not case-ignorable */ 666b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 667b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 668b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 669b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; /* not followed by cased letter */ 670b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 671b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 672b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* Is preceded by Soft_Dotted character with no intervening cc=230 ? */ 673b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic UBool 674b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruisPrecededBySoftDotted(const UCaseProps *csp, UCaseContextIterator *iter, void *context) { 675b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar32 c; 676b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t dotType; 677b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int8_t dir; 678b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 679b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(iter==NULL) { 680b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; 681b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 682b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 683b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru for(dir=-1; (c=iter(context, dir))>=0; dir=0) { 684b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru dotType=getDotType(csp, c); 685b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(dotType==UCASE_SOFT_DOTTED) { 686b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return TRUE; /* preceded by TYPE_i */ 687b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if(dotType!=UCASE_OTHER_ACCENT) { 688b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; /* preceded by different base character (not TYPE_i), or intervening cc==230 */ 689b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 690b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 691b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 692b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; /* not preceded by TYPE_i */ 693b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 694b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 695b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* 696b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * See Jitterbug 2344: 697b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * The condition After_I for Turkic-lowercasing of U+0307 combining dot above 698b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * is checked in ICU 2.0, 2.1, 2.6 but was not in 2.2 & 2.4 because 699b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * we made those releases compatible with Unicode 3.2 which had not fixed 700b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * a related bug in SpecialCasing.txt. 701b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 702b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * From the Jitterbug 2344 text: 703b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * ... this bug is listed as a Unicode erratum 704b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * from 2002-10-31 at http://www.unicode.org/uni2errata/UnicodeErrata.html 705b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * <quote> 706b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * There are two errors in SpecialCasing.txt. 707b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 1. Missing semicolons on two lines. ... [irrelevant for ICU] 708b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 2. An incorrect context definition. Correct as follows: 709b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * < 0307; ; 0307; 0307; tr After_Soft_Dotted; # COMBINING DOT ABOVE 710b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * < 0307; ; 0307; 0307; az After_Soft_Dotted; # COMBINING DOT ABOVE 711b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * --- 712b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * > 0307; ; 0307; 0307; tr After_I; # COMBINING DOT ABOVE 713b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * > 0307; ; 0307; 0307; az After_I; # COMBINING DOT ABOVE 714b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * where the context After_I is defined as: 715b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * The last preceding base character was an uppercase I, and there is no 716b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * intervening combining character class 230 (ABOVE). 717b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * </quote> 718b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 719b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Note that SpecialCasing.txt even in Unicode 3.2 described the condition as: 720b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 721b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * # When lowercasing, remove dot_above in the sequence I + dot_above, which will turn into i. 722b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * # This matches the behavior of the canonically equivalent I-dot_above 723b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 724b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * See also the description in this place in older versions of uchar.c (revision 1.100). 725b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 726b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Markus W. Scherer 2003-feb-15 727b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 728b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 729b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* Is preceded by base character 'I' with no intervening cc=230 ? */ 730b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic UBool 731b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruisPrecededBy_I(const UCaseProps *csp, UCaseContextIterator *iter, void *context) { 732b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar32 c; 733b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t dotType; 734b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int8_t dir; 735b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 736b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(iter==NULL) { 737b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; 738b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 739b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 740b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru for(dir=-1; (c=iter(context, dir))>=0; dir=0) { 741b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(c==0x49) { 742b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return TRUE; /* preceded by I */ 743b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 744b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru dotType=getDotType(csp, c); 745b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(dotType!=UCASE_OTHER_ACCENT) { 746b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; /* preceded by different base character (not I), or intervening cc==230 */ 747b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 748b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 749b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 750b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; /* not preceded by I */ 751b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 752b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 753b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* Is followed by one or more cc==230 ? */ 754b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic UBool 755b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruisFollowedByMoreAbove(const UCaseProps *csp, UCaseContextIterator *iter, void *context) { 756b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar32 c; 757b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t dotType; 758b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int8_t dir; 759b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 760b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(iter==NULL) { 761b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; 762b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 763b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 764b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru for(dir=1; (c=iter(context, dir))>=0; dir=0) { 765b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru dotType=getDotType(csp, c); 766b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(dotType==UCASE_ABOVE) { 767b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return TRUE; /* at least one cc==230 following */ 768b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if(dotType!=UCASE_OTHER_ACCENT) { 769b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; /* next base character, no more cc==230 following */ 770b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 771b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 772b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 773b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; /* no more cc==230 following */ 774b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 775b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 776b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* Is followed by a dot above (without cc==230 in between) ? */ 777b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic UBool 778b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruisFollowedByDotAbove(const UCaseProps *csp, UCaseContextIterator *iter, void *context) { 779b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar32 c; 780b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t dotType; 781b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int8_t dir; 782b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 783b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(iter==NULL) { 784b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; 785b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 786b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 787b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru for(dir=1; (c=iter(context, dir))>=0; dir=0) { 788b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(c==0x307) { 789b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return TRUE; 790b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 791b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru dotType=getDotType(csp, c); 792b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(dotType!=UCASE_OTHER_ACCENT) { 793b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; /* next base character or cc==230 in between */ 794b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 795b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 796b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 797b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; /* no dot above following */ 798b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 799b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 800b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI int32_t U_EXPORT2 801b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruucase_toFullLower(const UCaseProps *csp, UChar32 c, 802b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UCaseContextIterator *iter, void *context, 803b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const UChar **pString, 804c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru const char *locale, int32_t *locCache) 805c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru{ 806b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru UChar32 result=c; 807b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru uint16_t props=UTRIE2_GET16(&csp->trie, c); 808b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(!PROPS_HAS_EXCEPTION(props)) { 809b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(UCASE_GET_TYPE(props)>=UCASE_UPPER) { 810b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru result=c+UCASE_GET_DELTA(props); 811b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 812b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 813b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const uint16_t *pe=GET_EXCEPTIONS(csp, props), *pe2; 814b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uint16_t excWord=*pe++; 815b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t full; 816b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 817b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pe2=pe; 818b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 819b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(excWord&UCASE_EXC_CONDITIONAL_SPECIAL) { 820b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* use hardcoded conditions and mappings */ 821b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t loc=ucase_getCaseLocale(locale, locCache); 822b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 823b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 824b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Test for conditional mappings first 825b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * (otherwise the unconditional default mappings are always taken), 826b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * then test for characters that have unconditional mappings in SpecialCasing.txt, 827b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * then get the UnicodeData.txt mappings. 828b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 829c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru if( loc==UCASE_LOC_LITHUANIAN && 830b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* base characters, find accents above */ 831b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru (((c==0x49 || c==0x4a || c==0x12e) && 832b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru isFollowedByMoreAbove(csp, iter, context)) || 833b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* precomposed with accent above, no need to find one */ 834b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru (c==0xcc || c==0xcd || c==0x128)) 835b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ) { 836b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 837b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # Lithuanian 838b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 839b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # Lithuanian retains the dot in a lowercase i when followed by accents. 840b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 841b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # Introduce an explicit dot above when lowercasing capital I's and J's 842b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # whenever there are more accents above. 843b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # (of the accents used in Lithuanian: grave, acute, tilde above, and ogonek) 844b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 845b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 0049; 0069 0307; 0049; 0049; lt More_Above; # LATIN CAPITAL LETTER I 846b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 004A; 006A 0307; 004A; 004A; lt More_Above; # LATIN CAPITAL LETTER J 847b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 012E; 012F 0307; 012E; 012E; lt More_Above; # LATIN CAPITAL LETTER I WITH OGONEK 848b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 00CC; 0069 0307 0300; 00CC; 00CC; lt; # LATIN CAPITAL LETTER I WITH GRAVE 849b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 00CD; 0069 0307 0301; 00CD; 00CD; lt; # LATIN CAPITAL LETTER I WITH ACUTE 850b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 0128; 0069 0307 0303; 0128; 0128; lt; # LATIN CAPITAL LETTER I WITH TILDE 851b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 852b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru switch(c) { 853b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 0x49: /* LATIN CAPITAL LETTER I */ 854b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru *pString=iDot; 855b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 2; 856b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 0x4a: /* LATIN CAPITAL LETTER J */ 857b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru *pString=jDot; 858b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 2; 859b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 0x12e: /* LATIN CAPITAL LETTER I WITH OGONEK */ 860b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru *pString=iOgonekDot; 861b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 2; 862b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 0xcc: /* LATIN CAPITAL LETTER I WITH GRAVE */ 863b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru *pString=iDotGrave; 864b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 3; 865b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 0xcd: /* LATIN CAPITAL LETTER I WITH ACUTE */ 866b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru *pString=iDotAcute; 867b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 3; 868b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 0x128: /* LATIN CAPITAL LETTER I WITH TILDE */ 869b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru *pString=iDotTilde; 870b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 3; 871b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru default: 872b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 0; /* will not occur */ 873b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 874b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* # Turkish and Azeri */ 875c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru } else if(loc==UCASE_LOC_TURKISH && c==0x130) { 876b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 877b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # I and i-dotless; I-dot and i are case pairs in Turkish and Azeri 878b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # The following rules handle those cases. 879b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 880b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 0130; 0069; 0130; 0130; tr # LATIN CAPITAL LETTER I WITH DOT ABOVE 881b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 0130; 0069; 0130; 0130; az # LATIN CAPITAL LETTER I WITH DOT ABOVE 882b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 883b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 0x69; 884c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru } else if(loc==UCASE_LOC_TURKISH && c==0x307 && isPrecededBy_I(csp, iter, context)) { 885b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 886b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # When lowercasing, remove dot_above in the sequence I + dot_above, which will turn into i. 887b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # This matches the behavior of the canonically equivalent I-dot_above 888b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 889b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 0307; ; 0307; 0307; tr After_I; # COMBINING DOT ABOVE 890b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 0307; ; 0307; 0307; az After_I; # COMBINING DOT ABOVE 891b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 892b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 0; /* remove the dot (continue without output) */ 893c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru } else if(loc==UCASE_LOC_TURKISH && c==0x49 && !isFollowedByDotAbove(csp, iter, context)) { 894b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 895b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # When lowercasing, unless an I is before a dot_above, it turns into a dotless i. 896b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 897b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 0049; 0131; 0049; 0049; tr Not_Before_Dot; # LATIN CAPITAL LETTER I 898b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 0049; 0131; 0049; 0049; az Not_Before_Dot; # LATIN CAPITAL LETTER I 899b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 900b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 0x131; 901b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if(c==0x130) { 902b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 903b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # Preserve canonical equivalence for I with dot. Turkic is handled below. 904b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 905b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 0130; 0069 0307; 0130; 0130; # LATIN CAPITAL LETTER I WITH DOT ABOVE 906b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 907b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru *pString=iDot; 908b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 2; 909b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if( c==0x3a3 && 910b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru !isFollowedByCasedLetter(csp, iter, context, 1) && 911b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru isFollowedByCasedLetter(csp, iter, context, -1) /* -1=preceded */ 912b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ) { 913b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* greek capital sigma maps depending on surrounding cased letters (see SpecialCasing.txt) */ 914b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 915b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # Special case for final form of sigma 916b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 917b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 03A3; 03C2; 03A3; 03A3; Final_Sigma; # GREEK CAPITAL LETTER SIGMA 918b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 919b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 0x3c2; /* greek small final sigma */ 920b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 921b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* no known conditional special case mapping, use a normal mapping */ 922b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 923b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if(HAS_SLOT(excWord, UCASE_EXC_FULL_MAPPINGS)) { 924b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru GET_SLOT_VALUE(excWord, UCASE_EXC_FULL_MAPPINGS, pe, full); 925b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru full&=UCASE_FULL_LOWER; 926b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(full!=0) { 927b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* set the output pointer to the lowercase mapping */ 92883a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius *pString=reinterpret_cast<const UChar *>(pe+1); 929b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 930b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* return the string length */ 931b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return full; 932b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 933b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 934b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 935b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(HAS_SLOT(excWord, UCASE_EXC_LOWER)) { 936b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru GET_SLOT_VALUE(excWord, UCASE_EXC_LOWER, pe2, result); 937b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 938b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 939b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 940b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return (result==c) ? ~result : result; 941b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 942b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 943b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* internal */ 944b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic int32_t 945b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QuerutoUpperOrTitle(const UCaseProps *csp, UChar32 c, 946b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UCaseContextIterator *iter, void *context, 947b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const UChar **pString, 948b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const char *locale, int32_t *locCache, 949b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UBool upperNotTitle) { 950b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru UChar32 result=c; 951b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru uint16_t props=UTRIE2_GET16(&csp->trie, c); 952b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(!PROPS_HAS_EXCEPTION(props)) { 953b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(UCASE_GET_TYPE(props)==UCASE_LOWER) { 954b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru result=c+UCASE_GET_DELTA(props); 955b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 956b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 957b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const uint16_t *pe=GET_EXCEPTIONS(csp, props), *pe2; 958b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uint16_t excWord=*pe++; 959b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru int32_t full, idx; 960b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 961b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pe2=pe; 962b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 963b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(excWord&UCASE_EXC_CONDITIONAL_SPECIAL) { 964b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* use hardcoded conditions and mappings */ 965b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t loc=ucase_getCaseLocale(locale, locCache); 966b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 967c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru if(loc==UCASE_LOC_TURKISH && c==0x69) { 968b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 969b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # Turkish and Azeri 970b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 971b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # I and i-dotless; I-dot and i are case pairs in Turkish and Azeri 972b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # The following rules handle those cases. 973b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 974b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # When uppercasing, i turns into a dotted capital I 975b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 976b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 0069; 0069; 0130; 0130; tr; # LATIN SMALL LETTER I 977b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 0069; 0069; 0130; 0130; az; # LATIN SMALL LETTER I 978b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 979b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 0x130; 980c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru } else if(loc==UCASE_LOC_LITHUANIAN && c==0x307 && isPrecededBySoftDotted(csp, iter, context)) { 981b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 982b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # Lithuanian 983b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 984b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # Lithuanian retains the dot in a lowercase i when followed by accents. 985b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 986b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru # Remove DOT ABOVE after "i" with upper or titlecase 987b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 988b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 0307; 0307; ; ; lt After_Soft_Dotted; # COMBINING DOT ABOVE 989b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 990b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 0; /* remove the dot (continue without output) */ 991b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 992b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* no known conditional special case mapping, use a normal mapping */ 993b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 994b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if(HAS_SLOT(excWord, UCASE_EXC_FULL_MAPPINGS)) { 995b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru GET_SLOT_VALUE(excWord, UCASE_EXC_FULL_MAPPINGS, pe, full); 996b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 997b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* start of full case mapping strings */ 998b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ++pe; 999b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1000b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* skip the lowercase and case-folding result strings */ 1001b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pe+=full&UCASE_FULL_LOWER; 1002b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru full>>=4; 1003b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pe+=full&0xf; 1004b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru full>>=4; 1005b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1006b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(upperNotTitle) { 1007b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru full&=0xf; 1008b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 1009b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* skip the uppercase result string */ 1010b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pe+=full&0xf; 1011b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru full=(full>>4)&0xf; 1012b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1013b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1014b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(full!=0) { 1015b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* set the output pointer to the result string */ 101683a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius *pString=reinterpret_cast<const UChar *>(pe); 1017b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1018b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* return the string length */ 1019b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return full; 1020b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1021b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1022b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1023b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(!upperNotTitle && HAS_SLOT(excWord, UCASE_EXC_TITLE)) { 1024b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru idx=UCASE_EXC_TITLE; 1025b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if(HAS_SLOT(excWord, UCASE_EXC_UPPER)) { 1026b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* here, titlecase is same as uppercase */ 1027b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru idx=UCASE_EXC_UPPER; 1028b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 1029b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return ~c; 1030b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1031b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru GET_SLOT_VALUE(excWord, idx, pe2, result); 1032b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1033b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1034b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return (result==c) ? ~result : result; 1035b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1036b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1037b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI int32_t U_EXPORT2 1038b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruucase_toFullUpper(const UCaseProps *csp, UChar32 c, 1039b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UCaseContextIterator *iter, void *context, 1040b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const UChar **pString, 1041b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const char *locale, int32_t *locCache) { 1042b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return toUpperOrTitle(csp, c, iter, context, pString, locale, locCache, TRUE); 1043b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1044b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1045b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI int32_t U_EXPORT2 1046b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruucase_toFullTitle(const UCaseProps *csp, UChar32 c, 1047b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UCaseContextIterator *iter, void *context, 1048b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const UChar **pString, 1049b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const char *locale, int32_t *locCache) { 1050b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return toUpperOrTitle(csp, c, iter, context, pString, locale, locCache, FALSE); 1051b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1052b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1053b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* case folding ------------------------------------------------------------- */ 1054b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1055b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* 1056b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Case folding is similar to lowercasing. 1057b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * The result may be a simple mapping, i.e., a single code point, or 1058b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * a full mapping, i.e., a string. 1059b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * If the case folding for a code point is the same as its simple (1:1) lowercase mapping, 1060b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * then only the lowercase mapping is stored. 1061b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 1062b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Some special cases are hardcoded because their conditions cannot be 1063b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * parsed and processed from CaseFolding.txt. 1064b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 1065b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Unicode 3.2 CaseFolding.txt specifies for its status field: 1066b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1067b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru# C: common case folding, common mappings shared by both simple and full mappings. 1068b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru# F: full case folding, mappings that cause strings to grow in length. Multiple characters are separated by spaces. 1069b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru# S: simple case folding, mappings to single characters where different from F. 1070b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru# T: special case for uppercase I and dotted uppercase I 1071b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru# - For non-Turkic languages, this mapping is normally not used. 1072b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru# - For Turkic languages (tr, az), this mapping can be used instead of the normal mapping for these characters. 1073b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru# 1074b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru# Usage: 1075b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru# A. To do a simple case folding, use the mappings with status C + S. 1076b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru# B. To do a full case folding, use the mappings with status C + F. 1077b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru# 1078b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru# The mappings with status T can be used or omitted depending on the desired case-folding 1079b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru# behavior. (The default option is to exclude them.) 1080b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1081b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Unicode 3.2 has 'T' mappings as follows: 1082b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1083b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru0049; T; 0131; # LATIN CAPITAL LETTER I 1084b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru0130; T; 0069; # LATIN CAPITAL LETTER I WITH DOT ABOVE 1085b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1086b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * while the default mappings for these code points are: 1087b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1088b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru0049; C; 0069; # LATIN CAPITAL LETTER I 1089b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru0130; F; 0069 0307; # LATIN CAPITAL LETTER I WITH DOT ABOVE 1090b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1091b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * U+0130 has no simple case folding (simple-case-folds to itself). 1092b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 1093b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1094b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* return the simple case folding mapping for c */ 1095b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI UChar32 U_EXPORT2 1096b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruucase_fold(const UCaseProps *csp, UChar32 c, uint32_t options) { 1097b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru uint16_t props=UTRIE2_GET16(&csp->trie, c); 1098b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(!PROPS_HAS_EXCEPTION(props)) { 1099b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(UCASE_GET_TYPE(props)>=UCASE_UPPER) { 1100b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c+=UCASE_GET_DELTA(props); 1101b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1102b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 1103b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const uint16_t *pe=GET_EXCEPTIONS(csp, props); 1104b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uint16_t excWord=*pe++; 1105b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru int32_t idx; 1106b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(excWord&UCASE_EXC_CONDITIONAL_FOLD) { 1107b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* special case folding mappings, hardcoded */ 1108b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if((options&_FOLD_CASE_OPTIONS_MASK)==U_FOLD_CASE_DEFAULT) { 1109b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* default mappings */ 1110b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(c==0x49) { 1111b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 0049; C; 0069; # LATIN CAPITAL LETTER I */ 1112b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 0x69; 1113b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if(c==0x130) { 1114b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* no simple case folding for U+0130 */ 1115b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return c; 1116b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1117b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 1118b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* Turkic mappings */ 1119b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(c==0x49) { 1120b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 0049; T; 0131; # LATIN CAPITAL LETTER I */ 1121b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 0x131; 1122b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if(c==0x130) { 1123b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 0130; T; 0069; # LATIN CAPITAL LETTER I WITH DOT ABOVE */ 1124b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 0x69; 1125b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1126b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1127b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1128b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(HAS_SLOT(excWord, UCASE_EXC_FOLD)) { 1129b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru idx=UCASE_EXC_FOLD; 1130b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if(HAS_SLOT(excWord, UCASE_EXC_LOWER)) { 1131b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru idx=UCASE_EXC_LOWER; 1132b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 1133b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return c; 1134b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1135b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru GET_SLOT_VALUE(excWord, idx, pe, c); 1136b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1137b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return c; 1138b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1139b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1140b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* 1141b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Issue for canonical caseless match (UAX #21): 1142b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Turkic casefolding (using "T" mappings in CaseFolding.txt) does not preserve 1143b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * canonical equivalence, unlike default-option casefolding. 1144b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * For example, I-grave and I + grave fold to strings that are not canonically 1145b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * equivalent. 1146b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * For more details, see the comment in unorm_compare() in unorm.cpp 1147b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * and the intermediate prototype changes for Jitterbug 2021. 1148b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * (For example, revision 1.104 of uchar.c and 1.4 of CaseFolding.txt.) 1149b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * 1150b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * This did not get fixed because it appears that it is not possible to fix 1151b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * it for uppercase and lowercase characters (I-grave vs. i-grave) 1152b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * together in a way that they still fold to common result strings. 1153b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 1154b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1155b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI int32_t U_EXPORT2 1156b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruucase_toFullFolding(const UCaseProps *csp, UChar32 c, 1157b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const UChar **pString, 1158c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru uint32_t options) 1159c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru{ 1160b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru UChar32 result=c; 1161b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru uint16_t props=UTRIE2_GET16(&csp->trie, c); 1162b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(!PROPS_HAS_EXCEPTION(props)) { 1163b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(UCASE_GET_TYPE(props)>=UCASE_UPPER) { 1164b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru result=c+UCASE_GET_DELTA(props); 1165b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1166b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 1167b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const uint16_t *pe=GET_EXCEPTIONS(csp, props), *pe2; 1168b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uint16_t excWord=*pe++; 1169b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru int32_t full, idx; 1170b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1171b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pe2=pe; 1172b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1173b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(excWord&UCASE_EXC_CONDITIONAL_FOLD) { 1174b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* use hardcoded conditions and mappings */ 1175b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if((options&_FOLD_CASE_OPTIONS_MASK)==U_FOLD_CASE_DEFAULT) { 1176b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* default mappings */ 1177b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(c==0x49) { 1178b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 0049; C; 0069; # LATIN CAPITAL LETTER I */ 1179b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 0x69; 1180b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if(c==0x130) { 1181b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 0130; F; 0069 0307; # LATIN CAPITAL LETTER I WITH DOT ABOVE */ 1182b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru *pString=iDot; 1183b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 2; 1184b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1185b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 1186b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* Turkic mappings */ 1187b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(c==0x49) { 1188b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 0049; T; 0131; # LATIN CAPITAL LETTER I */ 1189b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 0x131; 1190b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if(c==0x130) { 1191b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* 0130; T; 0069; # LATIN CAPITAL LETTER I WITH DOT ABOVE */ 1192b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return 0x69; 1193b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1194b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1195b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if(HAS_SLOT(excWord, UCASE_EXC_FULL_MAPPINGS)) { 1196b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru GET_SLOT_VALUE(excWord, UCASE_EXC_FULL_MAPPINGS, pe, full); 1197b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1198b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* start of full case mapping strings */ 1199b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ++pe; 1200b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1201b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* skip the lowercase result string */ 1202b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pe+=full&UCASE_FULL_LOWER; 1203b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru full=(full>>4)&0xf; 1204b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1205b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(full!=0) { 1206b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* set the output pointer to the result string */ 120783a171d1a62abf406f7f44ae671823d5ec20db7dCraig Cornelius *pString=reinterpret_cast<const UChar *>(pe); 1208b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1209b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* return the string length */ 1210b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return full; 1211b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1212b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1213b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1214b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(HAS_SLOT(excWord, UCASE_EXC_FOLD)) { 1215b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru idx=UCASE_EXC_FOLD; 1216b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if(HAS_SLOT(excWord, UCASE_EXC_LOWER)) { 1217b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru idx=UCASE_EXC_LOWER; 1218b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 1219b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return ~c; 1220b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1221b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru GET_SLOT_VALUE(excWord, idx, pe2, result); 1222b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1223b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1224b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return (result==c) ? ~result : result; 1225b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1226b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1227b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* case mapping properties API ---------------------------------------------- */ 1228b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1229c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru#define GET_CASE_PROPS() &ucase_props_singleton 1230b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1231b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* public API (see uchar.h) */ 1232b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1233b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI UBool U_EXPORT2 1234b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruu_isULowercase(UChar32 c) { 1235b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return (UBool)(UCASE_LOWER==ucase_getType(GET_CASE_PROPS(), c)); 1236b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1237b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1238b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI UBool U_EXPORT2 1239b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruu_isUUppercase(UChar32 c) { 1240b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return (UBool)(UCASE_UPPER==ucase_getType(GET_CASE_PROPS(), c)); 1241b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1242b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1243b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* Transforms the Unicode character to its lower case equivalent.*/ 1244b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI UChar32 U_EXPORT2 1245b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruu_tolower(UChar32 c) { 1246b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return ucase_tolower(GET_CASE_PROPS(), c); 1247b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1248b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1249b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* Transforms the Unicode character to its upper case equivalent.*/ 1250b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI UChar32 U_EXPORT2 1251b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruu_toupper(UChar32 c) { 1252b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return ucase_toupper(GET_CASE_PROPS(), c); 1253b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1254b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1255b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* Transforms the Unicode character to its title case equivalent.*/ 1256b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI UChar32 U_EXPORT2 1257b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruu_totitle(UChar32 c) { 1258b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return ucase_totitle(GET_CASE_PROPS(), c); 1259b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1260b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1261b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* return the simple case folding mapping for c */ 1262b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CAPI UChar32 U_EXPORT2 1263b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruu_foldCase(UChar32 c, uint32_t options) { 1264b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return ucase_fold(GET_CASE_PROPS(), c, options); 1265b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1266b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1267b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CFUNC int32_t U_EXPORT2 1268b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruucase_hasBinaryProperty(UChar32 c, UProperty which) { 1269b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* case mapping properties */ 127050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho const UChar *resultString; 127150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho int32_t locCache; 1272b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const UCaseProps *csp=GET_CASE_PROPS(); 1273b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(csp==NULL) { 1274b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; 1275b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1276b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru switch(which) { 1277b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case UCHAR_LOWERCASE: 1278b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return (UBool)(UCASE_LOWER==ucase_getType(csp, c)); 1279b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case UCHAR_UPPERCASE: 1280b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return (UBool)(UCASE_UPPER==ucase_getType(csp, c)); 1281b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case UCHAR_SOFT_DOTTED: 1282b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return ucase_isSoftDotted(csp, c); 1283b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case UCHAR_CASE_SENSITIVE: 1284b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return ucase_isCaseSensitive(csp, c); 128550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho case UCHAR_CASED: 128650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho return (UBool)(UCASE_NONE!=ucase_getType(csp, c)); 128750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho case UCHAR_CASE_IGNORABLE: 128850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho return (UBool)(ucase_getTypeOrIgnorable(csp, c)>>2); 128950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho /* 129050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * Note: The following Changes_When_Xyz are defined as testing whether 129150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * the NFD form of the input changes when Xyz-case-mapped. 129250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * However, this simpler implementation of these properties, 129350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * ignoring NFD, passes the tests. 129450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * The implementation needs to be changed if the tests start failing. 129550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * When that happens, optimizations should be used to work with the 129650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * per-single-code point ucase_toFullXyz() functions unless 129750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * the NFD form has more than one code point, 129850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * and the property starts set needs to be the union of the 129950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho * start sets for normalization and case mappings. 130050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho */ 130150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho case UCHAR_CHANGES_WHEN_LOWERCASED: 130250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho locCache=UCASE_LOC_ROOT; 130350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho return (UBool)(ucase_toFullLower(csp, c, NULL, NULL, &resultString, "", &locCache)>=0); 130450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho case UCHAR_CHANGES_WHEN_UPPERCASED: 130550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho locCache=UCASE_LOC_ROOT; 130650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho return (UBool)(ucase_toFullUpper(csp, c, NULL, NULL, &resultString, "", &locCache)>=0); 130750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho case UCHAR_CHANGES_WHEN_TITLECASED: 130850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho locCache=UCASE_LOC_ROOT; 130950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho return (UBool)(ucase_toFullTitle(csp, c, NULL, NULL, &resultString, "", &locCache)>=0); 131050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho /* case UCHAR_CHANGES_WHEN_CASEFOLDED: -- in uprops.c */ 131150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho case UCHAR_CHANGES_WHEN_CASEMAPPED: 131250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho locCache=UCASE_LOC_ROOT; 131350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho return (UBool)( 131450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho ucase_toFullLower(csp, c, NULL, NULL, &resultString, "", &locCache)>=0 || 131550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho ucase_toFullUpper(csp, c, NULL, NULL, &resultString, "", &locCache)>=0 || 131650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho ucase_toFullTitle(csp, c, NULL, NULL, &resultString, "", &locCache)>=0); 1317b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru default: 1318b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; 1319b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1320b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1321