1b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/* 2b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru******************************************************************************* 3b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* 4b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho* Copyright (C) 1999-2011, International Business Machines 5b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* Corporation and others. All Rights Reserved. 6b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* 7b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru******************************************************************************* 8b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* file name: uniset_props.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: 2004aug25 14b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* created by: Markus W. Scherer 15b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* 16b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru* Character property dependent functions moved here from uniset.cpp 17b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru*/ 18b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 19b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "unicode/utypes.h" 20b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "unicode/uniset.h" 21b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "unicode/parsepos.h" 22b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "unicode/uchar.h" 23b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "unicode/uscript.h" 24b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "unicode/symtable.h" 25b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "unicode/uset.h" 26b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "unicode/locid.h" 27b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "unicode/brkiter.h" 28b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "uset_imp.h" 29b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "ruleiter.h" 30b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "cmemory.h" 31b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "ucln_cmn.h" 32b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "util.h" 33b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "uvector.h" 34b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "uprops.h" 35b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "propname.h" 3650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho#include "normalizer2impl.h" 37b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "ucase.h" 38b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "ubidi_props.h" 39b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "uinvchar.h" 4050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho#include "uprops.h" 41b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "charstr.h" 42b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "cstring.h" 4350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho#include "mutex.h" 44b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "umutex.h" 45b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "uassert.h" 46b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#include "hash.h" 47b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 48b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_NAMESPACE_USE 49b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 50b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0])) 51b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 52b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// initial storage. Must be >= 0 53b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// *** same as in uniset.cpp ! *** 54b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define START_EXTRA 16 55b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 56b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// Define UChar constants using hex for EBCDIC compatibility 57b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// Used #define to reduce private static exports and memory access time. 58b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define SET_OPEN ((UChar)0x005B) /*[*/ 59b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define SET_CLOSE ((UChar)0x005D) /*]*/ 60b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define HYPHEN ((UChar)0x002D) /*-*/ 61b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define COMPLEMENT ((UChar)0x005E) /*^*/ 62b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define COLON ((UChar)0x003A) /*:*/ 63b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define BACKSLASH ((UChar)0x005C) /*\*/ 64b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define INTERSECTION ((UChar)0x0026) /*&*/ 65b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define UPPER_U ((UChar)0x0055) /*U*/ 66b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define LOWER_U ((UChar)0x0075) /*u*/ 67b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define OPEN_BRACE ((UChar)123) /*{*/ 68b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define CLOSE_BRACE ((UChar)125) /*}*/ 69b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define UPPER_P ((UChar)0x0050) /*P*/ 70b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define LOWER_P ((UChar)0x0070) /*p*/ 71b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define UPPER_N ((UChar)78) /*N*/ 72b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define EQUALS ((UChar)0x003D) /*=*/ 73b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 74b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//static const UChar POSIX_OPEN[] = { SET_OPEN,COLON,0 }; // "[:" 75b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic const UChar POSIX_CLOSE[] = { COLON,SET_CLOSE,0 }; // ":]" 76b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//static const UChar PERL_OPEN[] = { BACKSLASH,LOWER_P,0 }; // "\\p" 77b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic const UChar PERL_CLOSE[] = { CLOSE_BRACE,0 }; // "}" 78b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//static const UChar NAME_OPEN[] = { BACKSLASH,UPPER_N,0 }; // "\\N" 79b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic const UChar HYPHEN_RIGHT_BRACE[] = {HYPHEN,SET_CLOSE,0}; /*-]*/ 80b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 81b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// Special property set IDs 82b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic const char ANY[] = "ANY"; // [\u0000-\U0010FFFF] 83b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic const char ASCII[] = "ASCII"; // [\u0000-\u007F] 84b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic const char ASSIGNED[] = "Assigned"; // [:^Cn:] 85b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 86b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// Unicode name property alias 87b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define NAME_PROP "na" 88b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define NAME_PROP_LENGTH 2 89b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 90b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 91b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Delimiter string used in patterns to close a category reference: 92b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * ":]". Example: "[:Lu:]". 93b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 94b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//static const UChar CATEGORY_CLOSE[] = {COLON, SET_CLOSE, 0x0000}; /* ":]" */ 95b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 9650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho// Cached sets ------------------------------------------------------------- *** 9750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 9850294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_CDECL_BEGIN 9950294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehostatic UBool U_CALLCONV uset_cleanup(); 10050294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_CDECL_END 10150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 10250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho// Not a TriStateSingletonWrapper because we think the UnicodeSet constructor 10350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho// can only fail with an out-of-memory error 10450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho// if we have a correct pattern and the properties data is hardcoded and always available. 10550294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoclass UnicodeSetSingleton : public SimpleSingletonWrapper<UnicodeSet> { 10650294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehopublic: 10750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UnicodeSetSingleton(SimpleSingleton &s, const char *pattern) : 10850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho SimpleSingletonWrapper<UnicodeSet>(s), fPattern(pattern) {} 10950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UnicodeSet *getInstance(UErrorCode &errorCode) { 11050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho return SimpleSingletonWrapper<UnicodeSet>::getInstance(createInstance, fPattern, errorCode); 11150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho } 11250294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoprivate: 11350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho static void *createInstance(const void *context, UErrorCode &errorCode) { 11450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UnicodeString pattern((const char *)context, -1, US_INV); 11550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UnicodeSet *set=new UnicodeSet(pattern, errorCode); 11650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho if(set==NULL) { 11750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho errorCode=U_MEMORY_ALLOCATION_ERROR; 118b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho return NULL; 11950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho } 12050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho set->freeze(); 12150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho ucln_common_registerCleanup(UCLN_COMMON_USET, uset_cleanup); 12250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho return set; 12350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho } 12450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 12550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho const char *fPattern; 12650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho}; 12750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 128b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CDECL_BEGIN 129b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 130b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic UnicodeSet *INCLUSIONS[UPROPS_SRC_COUNT] = { NULL }; // cached getInclusions() 131b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 13250294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoSTATIC_SIMPLE_SINGLETON(uni32Singleton); 13350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 134b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//---------------------------------------------------------------- 135b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// Inclusions list 136b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//---------------------------------------------------------------- 137b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 138b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// USetAdder implementation 139b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// Does not use uset.h to reduce code dependencies 140b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic void U_CALLCONV 141b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru_set_add(USet *set, UChar32 c) { 142b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ((UnicodeSet *)set)->add(c); 143b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 144b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 145b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic void U_CALLCONV 146b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru_set_addRange(USet *set, UChar32 start, UChar32 end) { 147b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ((UnicodeSet *)set)->add(start, end); 148b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 149b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 150b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic void U_CALLCONV 151b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru_set_addString(USet *set, const UChar *str, int32_t length) { 152b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ((UnicodeSet *)set)->add(UnicodeString((UBool)(length<0), str, length)); 153b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 154b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 155b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 156b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Cleanup function for UnicodeSet 157b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 158b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic UBool U_CALLCONV uset_cleanup(void) { 159b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t i; 160b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 161b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru for(i = UPROPS_SRC_NONE; i < UPROPS_SRC_COUNT; ++i) { 162b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (INCLUSIONS[i] != NULL) { 163b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru delete INCLUSIONS[i]; 164b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru INCLUSIONS[i] = NULL; 165b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 166b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 16750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho UnicodeSetSingleton(uni32Singleton, NULL).deleteInstance(); 168b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return TRUE; 169b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 170b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 171b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_CDECL_END 172b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 173b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_NAMESPACE_BEGIN 174b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 175c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru/* 176c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste QueruReduce excessive reallocation, and make it easier to detect initialization 177c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queruproblems. 178c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste QueruUsually you don't see smaller sets than this for Unicode 5.0. 179c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru*/ 180c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru#define DEFAULT_INCLUSION_CAPACITY 3072 181c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru 182c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queruconst UnicodeSet* UnicodeSet::getInclusions(int32_t src, UErrorCode &status) { 183b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UBool needInit; 184b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UMTX_CHECK(NULL, (INCLUSIONS[src] == NULL), needInit); 185b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (needInit) { 186b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UnicodeSet* incl = new UnicodeSet(); 187b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru USetAdder sa = { 188b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru (USet *)incl, 189b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru _set_add, 190b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru _set_addRange, 191b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru _set_addString, 192c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru NULL, // don't need remove() 193c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru NULL // don't need removeRange() 194b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru }; 195b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (incl != NULL) { 196b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2claireho incl->ensureCapacity(DEFAULT_INCLUSION_CAPACITY, status); 197b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru switch(src) { 198b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case UPROPS_SRC_CHAR: 199b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uchar_addPropertyStarts(&sa, &status); 200b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 201b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case UPROPS_SRC_PROPSVEC: 202b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru upropsvec_addPropertyStarts(&sa, &status); 203b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 204b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case UPROPS_SRC_CHAR_AND_PROPSVEC: 205b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uchar_addPropertyStarts(&sa, &status); 206b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru upropsvec_addPropertyStarts(&sa, &status); 207b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 208b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#if !UCONFIG_NO_NORMALIZATION 20927f654740f2a26ad62a5c155af9199af9e69b889claireho case UPROPS_SRC_CASE_AND_NORM: { 21027f654740f2a26ad62a5c155af9199af9e69b889claireho const Normalizer2Impl *impl=Normalizer2Factory::getNFCImpl(status); 21127f654740f2a26ad62a5c155af9199af9e69b889claireho if(U_SUCCESS(status)) { 21227f654740f2a26ad62a5c155af9199af9e69b889claireho impl->addPropertyStarts(&sa, status); 21327f654740f2a26ad62a5c155af9199af9e69b889claireho } 21427f654740f2a26ad62a5c155af9199af9e69b889claireho ucase_addPropertyStarts(ucase_getSingleton(), &sa, &status); 21550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho break; 21627f654740f2a26ad62a5c155af9199af9e69b889claireho } 21750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho case UPROPS_SRC_NFC: { 21850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho const Normalizer2Impl *impl=Normalizer2Factory::getNFCImpl(status); 21950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho if(U_SUCCESS(status)) { 22050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho impl->addPropertyStarts(&sa, status); 22150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho } 22250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho break; 22350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho } 22450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho case UPROPS_SRC_NFKC: { 22550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho const Normalizer2Impl *impl=Normalizer2Factory::getNFKCImpl(status); 22650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho if(U_SUCCESS(status)) { 22750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho impl->addPropertyStarts(&sa, status); 22850294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho } 22950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho break; 23050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho } 23150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho case UPROPS_SRC_NFKC_CF: { 23250294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho const Normalizer2Impl *impl=Normalizer2Factory::getNFKC_CFImpl(status); 23350294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho if(U_SUCCESS(status)) { 23450294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho impl->addPropertyStarts(&sa, status); 23550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho } 23650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho break; 23750294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho } 23827f654740f2a26ad62a5c155af9199af9e69b889claireho case UPROPS_SRC_NFC_CANON_ITER: { 23927f654740f2a26ad62a5c155af9199af9e69b889claireho const Normalizer2Impl *impl=Normalizer2Factory::getNFCImpl(status); 24027f654740f2a26ad62a5c155af9199af9e69b889claireho if(U_SUCCESS(status)) { 24127f654740f2a26ad62a5c155af9199af9e69b889claireho impl->addCanonIterPropertyStarts(&sa, status); 24227f654740f2a26ad62a5c155af9199af9e69b889claireho } 24327f654740f2a26ad62a5c155af9199af9e69b889claireho break; 24427f654740f2a26ad62a5c155af9199af9e69b889claireho } 245b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#endif 246b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case UPROPS_SRC_CASE: 24727f654740f2a26ad62a5c155af9199af9e69b889claireho ucase_addPropertyStarts(ucase_getSingleton(), &sa, &status); 248b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 249b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case UPROPS_SRC_BIDI: 25027f654740f2a26ad62a5c155af9199af9e69b889claireho ubidi_addPropertyStarts(ubidi_getSingleton(), &sa, &status); 251b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 252b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru default: 253b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru status = U_INTERNAL_PROGRAM_ERROR; 254b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 255b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 256b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_SUCCESS(status)) { 257b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Compact for caching 258b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru incl->compact(); 259b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru umtx_lock(NULL); 260b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (INCLUSIONS[src] == NULL) { 261b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru INCLUSIONS[src] = incl; 262b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru incl = NULL; 263b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ucln_common_registerCleanup(UCLN_COMMON_USET, uset_cleanup); 264b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 265b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru umtx_unlock(NULL); 266b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 267b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru delete incl; 268b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 269b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru status = U_MEMORY_ALLOCATION_ERROR; 270b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 271b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 272b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return INCLUSIONS[src]; 273b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 274b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 27550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho// Cache some sets for other services -------------------------------------- *** 27650294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 27750294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_CFUNC UnicodeSet * 27850294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehouniset_getUnicode32Instance(UErrorCode &errorCode) { 27950294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho return UnicodeSetSingleton(uni32Singleton, "[:age=3.2:]").getInstance(errorCode); 28050294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho} 28150294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 282b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// helper functions for matching of pattern syntax pieces ------------------ *** 283b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// these functions are parallel to the PERL_OPEN etc. strings above 284b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 285b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// using these functions is not only faster than UnicodeString::compare() and 286b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// caseCompare(), but they also make UnicodeSet work for simple patterns when 287b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// no Unicode properties data is available - when caseCompare() fails 288b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 289b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic inline UBool 290b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruisPerlOpen(const UnicodeString &pattern, int32_t pos) { 291b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar c; 292b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return pattern.charAt(pos)==BACKSLASH && ((c=pattern.charAt(pos+1))==LOWER_P || c==UPPER_P); 293b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 294b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 295b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/*static inline UBool 296b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruisPerlClose(const UnicodeString &pattern, int32_t pos) { 297b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return pattern.charAt(pos)==CLOSE_BRACE; 298b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru}*/ 299b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 300b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic inline UBool 301b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruisNameOpen(const UnicodeString &pattern, int32_t pos) { 302b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return pattern.charAt(pos)==BACKSLASH && pattern.charAt(pos+1)==UPPER_N; 303b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 304b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 305b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic inline UBool 306b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruisPOSIXOpen(const UnicodeString &pattern, int32_t pos) { 307b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return pattern.charAt(pos)==SET_OPEN && pattern.charAt(pos+1)==COLON; 308b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 309b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 310b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/*static inline UBool 311b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruisPOSIXClose(const UnicodeString &pattern, int32_t pos) { 312b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return pattern.charAt(pos)==COLON && pattern.charAt(pos+1)==SET_CLOSE; 313b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru}*/ 314b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 315b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// TODO memory debugging provided inside uniset.cpp 316b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// could be made available here but probably obsolete with use of modern 317b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// memory leak checker tools 318b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define _dbgct(me) 319b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 320b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//---------------------------------------------------------------- 321b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// Constructors &c 322b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//---------------------------------------------------------------- 323b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 324b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 325b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Constructs a set from the given pattern, optionally ignoring 326b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * white space. See the class description for the syntax of the 327b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * pattern language. 328b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param pattern a string specifying what characters are in the set 329b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 330b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruUnicodeSet::UnicodeSet(const UnicodeString& pattern, 331b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode& status) : 332b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru len(0), capacity(START_EXTRA), list(0), bmpSet(0), buffer(0), 333c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru bufferCapacity(0), patLen(0), pat(NULL), strings(NULL), stringSpan(NULL), 334c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru fFlags(0) 335b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru{ 336b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(U_SUCCESS(status)){ 337b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru list = (UChar32*) uprv_malloc(sizeof(UChar32) * capacity); 338b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* test for NULL */ 339b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(list == NULL) { 340b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru status = U_MEMORY_ALLOCATION_ERROR; 341b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru }else{ 342b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru allocateStrings(status); 343b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru applyPattern(pattern, USET_IGNORE_SPACE, NULL, status); 344b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 345b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 346b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru _dbgct(this); 347b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 348b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 349b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 350b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Constructs a set from the given pattern, optionally ignoring 351b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * white space. See the class description for the syntax of the 352b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * pattern language. 353b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param pattern a string specifying what characters are in the set 354b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param options bitmask for options to apply to the pattern. 355b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Valid options are USET_IGNORE_SPACE and USET_CASE_INSENSITIVE. 356b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 357b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruUnicodeSet::UnicodeSet(const UnicodeString& pattern, 358b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uint32_t options, 359b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const SymbolTable* symbols, 360b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode& status) : 361b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru len(0), capacity(START_EXTRA), list(0), bmpSet(0), buffer(0), 362c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru bufferCapacity(0), patLen(0), pat(NULL), strings(NULL), stringSpan(NULL), 363c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru fFlags(0) 364b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru{ 365b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(U_SUCCESS(status)){ 366b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru list = (UChar32*) uprv_malloc(sizeof(UChar32) * capacity); 367b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* test for NULL */ 368b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(list == NULL) { 369b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru status = U_MEMORY_ALLOCATION_ERROR; 370b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru }else{ 371b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru allocateStrings(status); 372b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru applyPattern(pattern, options, symbols, status); 373b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 374b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 375b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru _dbgct(this); 376b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 377b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 378b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruUnicodeSet::UnicodeSet(const UnicodeString& pattern, ParsePosition& pos, 379b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uint32_t options, 380b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const SymbolTable* symbols, 381b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode& status) : 382b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru len(0), capacity(START_EXTRA), list(0), bmpSet(0), buffer(0), 383c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru bufferCapacity(0), patLen(0), pat(NULL), strings(NULL), stringSpan(NULL), 384c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru fFlags(0) 385b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru{ 386b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(U_SUCCESS(status)){ 387b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru list = (UChar32*) uprv_malloc(sizeof(UChar32) * capacity); 388b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* test for NULL */ 389b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(list == NULL) { 390b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru status = U_MEMORY_ALLOCATION_ERROR; 391b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru }else{ 392b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru allocateStrings(status); 393b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru applyPattern(pattern, pos, options, symbols, status); 394b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 395b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 396b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru _dbgct(this); 397b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 398b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 399b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//---------------------------------------------------------------- 400b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// Public API 401b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//---------------------------------------------------------------- 402b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 403b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruUnicodeSet& UnicodeSet::applyPattern(const UnicodeString& pattern, 404b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode& status) { 405b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return applyPattern(pattern, USET_IGNORE_SPACE, NULL, status); 406b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 407b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 408b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 409b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 410b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Modifies this set to represent the set specified by the given 411b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * pattern, optionally ignoring white space. See the class 412b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * description for the syntax of the pattern language. 413b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param pattern a string specifying what characters are in the set 414b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param options bitmask for options to apply to the pattern. 415b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Valid options are USET_IGNORE_SPACE and USET_CASE_INSENSITIVE. 416b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 417b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruUnicodeSet& UnicodeSet::applyPattern(const UnicodeString& pattern, 418b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uint32_t options, 419b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const SymbolTable* symbols, 420b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode& status) { 421b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_FAILURE(status) || isFrozen()) { 422b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return *this; 423b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 424b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 425b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ParsePosition pos(0); 426b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru applyPattern(pattern, pos, options, symbols, status); 427b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_FAILURE(status)) return *this; 428b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 429b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t i = pos.getIndex(); 430b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 431b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (options & USET_IGNORE_SPACE) { 432b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Skip over trailing whitespace 433b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ICU_Utility::skipWhitespace(pattern, i, TRUE); 434b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 435b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 436b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (i != pattern.length()) { 437b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru status = U_ILLEGAL_ARGUMENT_ERROR; 438b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 439b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return *this; 440b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 441b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 442b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruUnicodeSet& UnicodeSet::applyPattern(const UnicodeString& pattern, 443b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ParsePosition& pos, 444b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uint32_t options, 445b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const SymbolTable* symbols, 446b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode& status) { 447b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_FAILURE(status) || isFrozen()) { 448b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return *this; 449b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 450b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Need to build the pattern in a temporary string because 451b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // _applyPattern calls add() etc., which set pat to empty. 452b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UnicodeString rebuiltPat; 453b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru RuleCharacterIterator chars(pattern, symbols, pos); 454b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru applyPattern(chars, symbols, rebuiltPat, options, status); 455b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_FAILURE(status)) return *this; 456b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (chars.inVariable()) { 457b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // syntaxError(chars, "Extra chars in variable value"); 458b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru status = U_MALFORMED_SET; 459b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return *this; 460b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 461b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru setPattern(rebuiltPat); 462b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return *this; 463b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 464b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 465b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 466b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Return true if the given position, in the given pattern, appears 467b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * to be the start of a UnicodeSet pattern. 468b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 469b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruUBool UnicodeSet::resemblesPattern(const UnicodeString& pattern, int32_t pos) { 470b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return ((pos+1) < pattern.length() && 471b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pattern.charAt(pos) == (UChar)91/*[*/) || 472b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru resemblesPropertyPattern(pattern, pos); 473b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 474b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 475b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//---------------------------------------------------------------- 476b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// Implementation: Pattern parsing 477b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//---------------------------------------------------------------- 478b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 479b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 480b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * A small all-inline class to manage a UnicodeSet pointer. Add 481b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * operator->() etc. as needed. 482b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 483b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruclass UnicodeSetPointer { 484b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UnicodeSet* p; 485b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querupublic: 486b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru inline UnicodeSetPointer() : p(0) {} 487b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru inline ~UnicodeSetPointer() { delete p; } 488b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru inline UnicodeSet* pointer() { return p; } 489b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru inline UBool allocate() { 490b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (p == 0) { 491b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru p = new UnicodeSet(); 492b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 493b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return p != 0; 494b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 495b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru}; 496b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 497b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 498b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Parse the pattern from the given RuleCharacterIterator. The 499b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * iterator is advanced over the parsed pattern. 500b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param chars iterator over the pattern characters. Upon return 501b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * it will be advanced to the first character after the parsed 502b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * pattern, or the end of the iteration if all characters are 503b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * parsed. 504b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param symbols symbol table to use to parse and dereference 505b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * variables, or null if none. 506b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param rebuiltPat the pattern that was parsed, rebuilt or 507b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * copied from the input pattern, as appropriate. 508b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param options a bit mask of zero or more of the following: 509b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * IGNORE_SPACE, CASE. 510b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 511b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruvoid UnicodeSet::applyPattern(RuleCharacterIterator& chars, 512b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const SymbolTable* symbols, 513b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UnicodeString& rebuiltPat, 514b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru uint32_t options, 515b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode& ec) { 516b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_FAILURE(ec)) return; 517b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 518b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Syntax characters: [ ] ^ - & { } 519b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 520b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Recognized special forms for chars, sets: c-c s-s s&s 521b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 522b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t opts = RuleCharacterIterator::PARSE_VARIABLES | 523b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru RuleCharacterIterator::PARSE_ESCAPES; 524b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if ((options & USET_IGNORE_SPACE) != 0) { 525b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru opts |= RuleCharacterIterator::SKIP_WHITESPACE; 526b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 527b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 528b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UnicodeString patLocal, buf; 529b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UBool usePat = FALSE; 530b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UnicodeSetPointer scratch; 531b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru RuleCharacterIterator::Pos backup; 532b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 533b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // mode: 0=before [, 1=between [...], 2=after ] 534b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // lastItem: 0=none, 1=char, 2=set 535b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int8_t lastItem = 0, mode = 0; 536b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar32 lastChar = 0; 537b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar op = 0; 538b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 539b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UBool invert = FALSE; 540b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 541b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru clear(); 542b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 543b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru while (mode != 2 && !chars.atEnd()) { 544b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru U_ASSERT((lastItem == 0 && op == 0) || 545b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru (lastItem == 1 && (op == 0 || op == HYPHEN /*'-'*/)) || 546b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru (lastItem == 2 && (op == 0 || op == HYPHEN /*'-'*/ || 547b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru op == INTERSECTION /*'&'*/))); 548b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 549b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar32 c = 0; 550b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UBool literal = FALSE; 551b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UnicodeSet* nested = 0; // alias - do not delete 552b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 553b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // -------- Check for property pattern 554b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 555b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // setMode: 0=none, 1=unicodeset, 2=propertypat, 3=preparsed 556b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int8_t setMode = 0; 557b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (resemblesPropertyPattern(chars, opts)) { 558b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru setMode = 2; 559b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 560b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 561b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // -------- Parse '[' of opening delimiter OR nested set. 562b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // If there is a nested set, use `setMode' to define how 563b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // the set should be parsed. If the '[' is part of the 564b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // opening delimiter for this pattern, parse special 565b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // strings "[", "[^", "[-", and "[^-". Check for stand-in 566b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // characters representing a nested set in the symbol 567b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // table. 568b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 569b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru else { 570b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Prepare to backup if necessary 571b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru chars.getPos(backup); 572b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c = chars.next(opts, literal, ec); 573b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_FAILURE(ec)) return; 574b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 575b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (c == 0x5B /*'['*/ && !literal) { 576b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (mode == 1) { 577b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru chars.setPos(backup); // backup 578b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru setMode = 1; 579b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 580b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Handle opening '[' delimiter 581b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru mode = 1; 582b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru patLocal.append((UChar) 0x5B /*'['*/); 583b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru chars.getPos(backup); // prepare to backup 584b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c = chars.next(opts, literal, ec); 585b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_FAILURE(ec)) return; 586b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (c == 0x5E /*'^'*/ && !literal) { 587b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru invert = TRUE; 588b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru patLocal.append((UChar) 0x5E /*'^'*/); 589b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru chars.getPos(backup); // prepare to backup 590b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c = chars.next(opts, literal, ec); 591b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_FAILURE(ec)) return; 592b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 593b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Fall through to handle special leading '-'; 594b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // otherwise restart loop for nested [], \p{}, etc. 595b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (c == HYPHEN /*'-'*/) { 596b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru literal = TRUE; 597b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Fall through to handle literal '-' below 598b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 599b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru chars.setPos(backup); // backup 600b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru continue; 601b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 602b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 603b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if (symbols != 0) { 604b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const UnicodeFunctor *m = symbols->lookupMatcher(c); 605b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (m != 0) { 60627f654740f2a26ad62a5c155af9199af9e69b889claireho const UnicodeSet *ms = dynamic_cast<const UnicodeSet *>(m); 60727f654740f2a26ad62a5c155af9199af9e69b889claireho if (ms == NULL) { 608b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ec = U_MALFORMED_SET; 609b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 610b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 611b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // casting away const, but `nested' won't be modified 612b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // (important not to modify stored set) 61327f654740f2a26ad62a5c155af9199af9e69b889claireho nested = const_cast<UnicodeSet*>(ms); 614b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru setMode = 3; 615b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 616b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 617b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 618b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 619b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // -------- Handle a nested set. This either is inline in 620b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // the pattern or represented by a stand-in that has 621b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // previously been parsed and was looked up in the symbol 622b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // table. 623b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 624b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (setMode != 0) { 625b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (lastItem == 1) { 626b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (op != 0) { 627b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // syntaxError(chars, "Char expected after operator"); 628b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ec = U_MALFORMED_SET; 629b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 630b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 631b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru add(lastChar, lastChar); 632b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru _appendToPat(patLocal, lastChar, FALSE); 633b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru lastItem = 0; 634b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru op = 0; 635b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 636b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 637b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (op == HYPHEN /*'-'*/ || op == INTERSECTION /*'&'*/) { 638b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru patLocal.append(op); 639b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 640b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 641b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (nested == 0) { 642b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // lazy allocation 643b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (!scratch.allocate()) { 644b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ec = U_MEMORY_ALLOCATION_ERROR; 645b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 646b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 647b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru nested = scratch.pointer(); 648b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 649b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru switch (setMode) { 650b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 1: 651b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru nested->applyPattern(chars, symbols, patLocal, options, ec); 652b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 653b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 2: 654b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru chars.skipIgnored(opts); 655b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru nested->applyPropertyPattern(chars, patLocal, ec); 656b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_FAILURE(ec)) return; 657b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 658b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 3: // `nested' already parsed 659b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru nested->_toPattern(patLocal, FALSE); 660b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 661b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 662b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 663b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru usePat = TRUE; 664b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 665b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (mode == 0) { 666b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Entire pattern is a category; leave parse loop 667b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru *this = *nested; 668b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru mode = 2; 669b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 670b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 671b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 672b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru switch (op) { 673b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case HYPHEN: /*'-'*/ 674b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru removeAll(*nested); 675b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 676b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case INTERSECTION: /*'&'*/ 677b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru retainAll(*nested); 678b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 679b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 0: 680b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru addAll(*nested); 681b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 682b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 683b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 684b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru op = 0; 685b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru lastItem = 2; 686b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 687b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru continue; 688b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 689b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 690b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (mode == 0) { 691b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // syntaxError(chars, "Missing '['"); 692b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ec = U_MALFORMED_SET; 693b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 694b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 695b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 696b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // -------- Parse special (syntax) characters. If the 697b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // current character is not special, or if it is escaped, 698b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // then fall through and handle it below. 699b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 700b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (!literal) { 701b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru switch (c) { 702b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 0x5D /*']'*/: 703b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (lastItem == 1) { 704b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru add(lastChar, lastChar); 705b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru _appendToPat(patLocal, lastChar, FALSE); 706b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 707b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Treat final trailing '-' as a literal 708b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (op == HYPHEN /*'-'*/) { 709b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru add(op, op); 710b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru patLocal.append(op); 711b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if (op == INTERSECTION /*'&'*/) { 712b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // syntaxError(chars, "Trailing '&'"); 713b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ec = U_MALFORMED_SET; 714b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 715b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 716b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru patLocal.append((UChar) 0x5D /*']'*/); 717b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru mode = 2; 718b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru continue; 719b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case HYPHEN /*'-'*/: 720b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (op == 0) { 721b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (lastItem != 0) { 722b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru op = (UChar) c; 723b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru continue; 724b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 725b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Treat final trailing '-' as a literal 726b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru add(c, c); 727b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c = chars.next(opts, literal, ec); 728b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_FAILURE(ec)) return; 729b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (c == 0x5D /*']'*/ && !literal) { 730b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru patLocal.append(HYPHEN_RIGHT_BRACE); 731b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru mode = 2; 732b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru continue; 733b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 734b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 735b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 736b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // syntaxError(chars, "'-' not after char or set"); 737b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ec = U_MALFORMED_SET; 738b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 739b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case INTERSECTION /*'&'*/: 740b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (lastItem == 2 && op == 0) { 741b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru op = (UChar) c; 742b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru continue; 743b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 744b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // syntaxError(chars, "'&' not after set"); 745b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ec = U_MALFORMED_SET; 746b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 747b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 0x5E /*'^'*/: 748b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // syntaxError(chars, "'^' not after '['"); 749b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ec = U_MALFORMED_SET; 750b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 751b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 0x7B /*'{'*/: 752b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (op != 0) { 753b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // syntaxError(chars, "Missing operand after operator"); 754b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ec = U_MALFORMED_SET; 755b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 756b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 757b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (lastItem == 1) { 758b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru add(lastChar, lastChar); 759b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru _appendToPat(patLocal, lastChar, FALSE); 760b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 761b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru lastItem = 0; 762b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru buf.truncate(0); 763b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru { 764b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UBool ok = FALSE; 765b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru while (!chars.atEnd()) { 766b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c = chars.next(opts, literal, ec); 767b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_FAILURE(ec)) return; 768b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (c == 0x7D /*'}'*/ && !literal) { 769b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ok = TRUE; 770b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 771b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 772b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru buf.append(c); 773b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 774b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (buf.length() < 1 || !ok) { 775b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // syntaxError(chars, "Invalid multicharacter string"); 776b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ec = U_MALFORMED_SET; 777b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 778b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 779b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 780b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // We have new string. Add it to set and continue; 781b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // we don't need to drop through to the further 782b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // processing 783b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru add(buf); 784b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru patLocal.append((UChar) 0x7B /*'{'*/); 785b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru _appendToPat(patLocal, buf, FALSE); 786b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru patLocal.append((UChar) 0x7D /*'}'*/); 787b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru continue; 788b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case SymbolTable::SYMBOL_REF: 789b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // symbols nosymbols 790b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // [a-$] error error (ambiguous) 791b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // [a$] anchor anchor 792b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // [a-$x] var "x"* literal '$' 793b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // [a-$.] error literal '$' 794b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // *We won't get here in the case of var "x" 795b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru { 796b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru chars.getPos(backup); 797b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c = chars.next(opts, literal, ec); 798b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_FAILURE(ec)) return; 799b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UBool anchor = (c == 0x5D /*']'*/ && !literal); 800b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (symbols == 0 && !anchor) { 801b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru c = SymbolTable::SYMBOL_REF; 802b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru chars.setPos(backup); 803b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; // literal '$' 804b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 805b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (anchor && op == 0) { 806b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (lastItem == 1) { 807b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru add(lastChar, lastChar); 808b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru _appendToPat(patLocal, lastChar, FALSE); 809b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 810b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru add(U_ETHER); 811b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru usePat = TRUE; 812b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru patLocal.append((UChar) SymbolTable::SYMBOL_REF); 813b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru patLocal.append((UChar) 0x5D /*']'*/); 814b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru mode = 2; 815b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru continue; 816b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 817b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // syntaxError(chars, "Unquoted '$'"); 818b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ec = U_MALFORMED_SET; 819b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 820b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 821b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru default: 822b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 823b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 824b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 825b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 826b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // -------- Parse literal characters. This includes both 827b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // escaped chars ("\u4E01") and non-syntax characters 828b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // ("a"). 829b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 830b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru switch (lastItem) { 831b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 0: 832b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru lastItem = 1; 833b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru lastChar = c; 834b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 835b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 1: 836b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (op == HYPHEN /*'-'*/) { 837b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (lastChar >= c) { 838b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Don't allow redundant (a-a) or empty (b-a) ranges; 839b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // these are most likely typos. 840b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // syntaxError(chars, "Invalid range"); 841b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ec = U_MALFORMED_SET; 842b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 843b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 844b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru add(lastChar, c); 845b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru _appendToPat(patLocal, lastChar, FALSE); 846b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru patLocal.append(op); 847b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru _appendToPat(patLocal, c, FALSE); 848b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru lastItem = 0; 849b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru op = 0; 850b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 851b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru add(lastChar, lastChar); 852b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru _appendToPat(patLocal, lastChar, FALSE); 853b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru lastChar = c; 854b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 855b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 856b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case 2: 857b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (op != 0) { 858b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // syntaxError(chars, "Set expected after operator"); 859b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ec = U_MALFORMED_SET; 860b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 861b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 862b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru lastChar = c; 863b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru lastItem = 1; 864b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 865b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 866b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 867b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 868b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (mode != 2) { 869b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // syntaxError(chars, "Missing ']'"); 870b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ec = U_MALFORMED_SET; 871b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 872b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 873b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 874b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru chars.skipIgnored(opts); 875b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 876b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /** 877b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Handle global flags (invert, case insensitivity). If this 878b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * pattern should be compiled case-insensitive, then we need 879b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * to close over case BEFORE COMPLEMENTING. This makes 880b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * patterns like /[^abc]/i work. 881b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 882b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if ((options & USET_CASE_INSENSITIVE) != 0) { 883b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru closeOver(USET_CASE_INSENSITIVE); 884b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 885b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru else if ((options & USET_ADD_CASE_MAPPINGS) != 0) { 886b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru closeOver(USET_ADD_CASE_MAPPINGS); 887b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 888b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (invert) { 889b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru complement(); 890b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 891b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 892b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Use the rebuilt pattern (patLocal) only if necessary. Prefer the 893b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // generated pattern. 894b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (usePat) { 895b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru rebuiltPat.append(patLocal); 896b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 897b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru _generatePattern(rebuiltPat, FALSE); 898b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 899c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru if (isBogus() && U_SUCCESS(ec)) { 900c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru // We likely ran out of memory. AHHH! 901c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru ec = U_MEMORY_ALLOCATION_ERROR; 902c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru } 903b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 904b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 905b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//---------------------------------------------------------------- 906b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// Property set implementation 907b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//---------------------------------------------------------------- 908b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 909b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic UBool numericValueFilter(UChar32 ch, void* context) { 910b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return u_getNumericValue(ch) == *(double*)context; 911b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 912b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 913b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic UBool generalCategoryMaskFilter(UChar32 ch, void* context) { 914b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t value = *(int32_t*)context; 915b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return (U_GET_GC_MASK((UChar32) ch) & value) != 0; 916b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 917b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 918b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic UBool versionFilter(UChar32 ch, void* context) { 91927f654740f2a26ad62a5c155af9199af9e69b889claireho static const UVersionInfo none = { 0, 0, 0, 0 }; 92027f654740f2a26ad62a5c155af9199af9e69b889claireho UVersionInfo v; 921b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru u_charAge(ch, v); 92227f654740f2a26ad62a5c155af9199af9e69b889claireho UVersionInfo* version = (UVersionInfo*)context; 923b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return uprv_memcmp(&v, &none, sizeof(v)) > 0 && uprv_memcmp(&v, version, sizeof(v)) <= 0; 924b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 925b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 926b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querutypedef struct { 927b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UProperty prop; 928b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t value; 929b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} IntPropertyContext; 930b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 931b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic UBool intPropertyFilter(UChar32 ch, void* context) { 932b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru IntPropertyContext* c = (IntPropertyContext*)context; 933b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return u_getIntPropertyValue((UChar32) ch, c->prop) == c->value; 934b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 935b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 93627f654740f2a26ad62a5c155af9199af9e69b889clairehostatic UBool scriptExtensionsFilter(UChar32 ch, void* context) { 93727f654740f2a26ad62a5c155af9199af9e69b889claireho return uscript_hasScript(ch, *(UScriptCode*)context); 93827f654740f2a26ad62a5c155af9199af9e69b889claireho} 939b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 940b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 941b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Generic filter-based scanning code for UCD property UnicodeSets. 942b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 943b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruvoid UnicodeSet::applyFilter(UnicodeSet::Filter filter, 944b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru void* context, 945b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t src, 946b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode &status) { 94727f654740f2a26ad62a5c155af9199af9e69b889claireho if (U_FAILURE(status)) return; 94827f654740f2a26ad62a5c155af9199af9e69b889claireho 94927f654740f2a26ad62a5c155af9199af9e69b889claireho // Logically, walk through all Unicode characters, noting the start 950b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // and end of each range for which filter.contain(c) is 951b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // true. Add each range to a set. 952b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // 95327f654740f2a26ad62a5c155af9199af9e69b889claireho // To improve performance, use an inclusions set which 954b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // encodes information about character ranges that are known 95527f654740f2a26ad62a5c155af9199af9e69b889claireho // to have identical properties. 95627f654740f2a26ad62a5c155af9199af9e69b889claireho // getInclusions(src) contains exactly the first characters of 95727f654740f2a26ad62a5c155af9199af9e69b889claireho // same-value ranges for the given properties "source". 958b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const UnicodeSet* inclusions = getInclusions(src, status); 959b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_FAILURE(status)) { 960b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 961b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 962b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 963b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru clear(); 964b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 965b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar32 startHasProperty = -1; 966c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru int32_t limitRange = inclusions->getRangeCount(); 967b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 968b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru for (int j=0; j<limitRange; ++j) { 969b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // get current range 970b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar32 start = inclusions->getRangeStart(j); 971b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar32 end = inclusions->getRangeEnd(j); 972b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 973b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // for all the code points in the range, process 974b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru for (UChar32 ch = start; ch <= end; ++ch) { 975b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // only add to this UnicodeSet on inflection points -- 976b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // where the hasProperty value changes to false 977b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if ((*filter)(ch, context)) { 978b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (startHasProperty < 0) { 979b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru startHasProperty = ch; 980b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 981b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if (startHasProperty >= 0) { 982b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru add(startHasProperty, ch-1); 983b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru startHasProperty = -1; 984b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 985b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 986b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 987b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (startHasProperty >= 0) { 988b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru add((UChar32)startHasProperty, (UChar32)0x10FFFF); 989b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 990c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru if (isBogus() && U_SUCCESS(status)) { 991c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru // We likely ran out of memory. AHHH! 992c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru status = U_MEMORY_ALLOCATION_ERROR; 993c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru } 994b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 995b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 996b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic UBool mungeCharName(char* dst, const char* src, int32_t dstCapacity) { 997b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru /* Note: we use ' ' in compiler code page */ 998b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t j = 0; 999b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru char ch; 1000b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru --dstCapacity; /* make room for term. zero */ 1001b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru while ((ch = *src++) != 0) { 1002b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (ch == ' ' && (j==0 || (j>0 && dst[j-1]==' '))) { 1003b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru continue; 1004b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1005b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (j >= dstCapacity) return FALSE; 1006b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru dst[j++] = ch; 1007b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1008b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (j > 0 && dst[j-1] == ' ') --j; 1009b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru dst[j] = 0; 1010b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return TRUE; 1011b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1012b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1013b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//---------------------------------------------------------------- 1014b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// Property set API 1015b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//---------------------------------------------------------------- 1016b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1017b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#define FAIL(ec) {ec=U_ILLEGAL_ARGUMENT_ERROR; return *this;} 1018b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1019b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruUnicodeSet& 1020b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruUnicodeSet::applyIntPropertyValue(UProperty prop, int32_t value, UErrorCode& ec) { 1021b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_FAILURE(ec) || isFrozen()) return *this; 1022b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1023b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (prop == UCHAR_GENERAL_CATEGORY_MASK) { 1024b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru applyFilter(generalCategoryMaskFilter, &value, UPROPS_SRC_CHAR, ec); 102527f654740f2a26ad62a5c155af9199af9e69b889claireho } else if (prop == UCHAR_SCRIPT_EXTENSIONS) { 102627f654740f2a26ad62a5c155af9199af9e69b889claireho UScriptCode script = (UScriptCode)value; 102727f654740f2a26ad62a5c155af9199af9e69b889claireho applyFilter(scriptExtensionsFilter, &script, UPROPS_SRC_PROPSVEC, ec); 1028b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 1029b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru IntPropertyContext c = {prop, value}; 1030b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru applyFilter(intPropertyFilter, &c, uprops_getSource(prop), ec); 1031b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1032b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return *this; 1033b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1034b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1035b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruUnicodeSet& 1036b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruUnicodeSet::applyPropertyAlias(const UnicodeString& prop, 1037b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const UnicodeString& value, 1038b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode& ec) { 1039b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_FAILURE(ec) || isFrozen()) return *this; 1040b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1041b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // prop and value used to be converted to char * using the default 1042b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // converter instead of the invariant conversion. 1043b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // This should not be necessary because all Unicode property and value 1044b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // names use only invariant characters. 1045b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // If there are any variant characters, then we won't find them anyway. 1046b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Checking first avoids assertion failures in the conversion. 1047b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if( !uprv_isInvariantUString(prop.getBuffer(), prop.length()) || 1048b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru !uprv_isInvariantUString(value.getBuffer(), value.length()) 1049b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ) { 1050b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru FAIL(ec); 1051b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 105227f654740f2a26ad62a5c155af9199af9e69b889claireho CharString pname, vname; 105327f654740f2a26ad62a5c155af9199af9e69b889claireho pname.appendInvariantChars(prop, ec); 105427f654740f2a26ad62a5c155af9199af9e69b889claireho vname.appendInvariantChars(value, ec); 105527f654740f2a26ad62a5c155af9199af9e69b889claireho if (U_FAILURE(ec)) return *this; 1056b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1057b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UProperty p; 1058b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t v; 1059b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UBool mustNotBeEmpty = FALSE, invert = FALSE; 1060b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1061b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (value.length() > 0) { 106227f654740f2a26ad62a5c155af9199af9e69b889claireho p = u_getPropertyEnum(pname.data()); 1063b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (p == UCHAR_INVALID_CODE) FAIL(ec); 1064b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1065b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Treat gc as gcm 1066b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (p == UCHAR_GENERAL_CATEGORY) { 1067b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru p = UCHAR_GENERAL_CATEGORY_MASK; 1068b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1069b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1070b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if ((p >= UCHAR_BINARY_START && p < UCHAR_BINARY_LIMIT) || 1071b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru (p >= UCHAR_INT_START && p < UCHAR_INT_LIMIT) || 1072b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru (p >= UCHAR_MASK_START && p < UCHAR_MASK_LIMIT)) { 107327f654740f2a26ad62a5c155af9199af9e69b889claireho v = u_getPropertyValueEnum(p, vname.data()); 1074b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (v == UCHAR_INVALID_CODE) { 1075b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Handle numeric CCC 1076b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (p == UCHAR_CANONICAL_COMBINING_CLASS || 1077b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru p == UCHAR_TRAIL_CANONICAL_COMBINING_CLASS || 1078b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru p == UCHAR_LEAD_CANONICAL_COMBINING_CLASS) { 1079b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru char* end; 108027f654740f2a26ad62a5c155af9199af9e69b889claireho double value = uprv_strtod(vname.data(), &end); 1081b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru v = (int32_t) value; 1082b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (v != value || v < 0 || *end != 0) { 1083b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // non-integral or negative value, or trailing junk 1084b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru FAIL(ec); 1085b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1086b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // If the resultant set is empty then the numeric value 1087b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // was invalid. 1088b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru mustNotBeEmpty = TRUE; 1089b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 1090b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru FAIL(ec); 1091b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1092b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1093b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1094b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1095b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru else { 1096b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1097b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru switch (p) { 1098b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case UCHAR_NUMERIC_VALUE: 1099b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru { 1100b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru char* end; 110127f654740f2a26ad62a5c155af9199af9e69b889claireho double value = uprv_strtod(vname.data(), &end); 1102b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (*end != 0) { 1103b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru FAIL(ec); 1104b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1105b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru applyFilter(numericValueFilter, &value, UPROPS_SRC_CHAR, ec); 1106b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return *this; 1107b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1108b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 1109b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case UCHAR_NAME: 1110b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case UCHAR_UNICODE_1_NAME: 1111b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru { 1112b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Must munge name, since u_charFromName() does not do 1113b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // 'loose' matching. 1114b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru char buf[128]; // it suffices that this be > uprv_getMaxCharNameLength 111527f654740f2a26ad62a5c155af9199af9e69b889claireho if (!mungeCharName(buf, vname.data(), sizeof(buf))) FAIL(ec); 1116b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UCharNameChoice choice = (p == UCHAR_NAME) ? 1117b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru U_EXTENDED_CHAR_NAME : U_UNICODE_10_CHAR_NAME; 1118b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar32 ch = u_charFromName(choice, buf, &ec); 1119b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_SUCCESS(ec)) { 1120b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru clear(); 1121b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru add(ch); 1122b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return *this; 1123b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 1124b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru FAIL(ec); 1125b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1126b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1127b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 1128b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru case UCHAR_AGE: 1129b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru { 1130b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Must munge name, since u_versionFromString() does not do 1131b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // 'loose' matching. 1132b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru char buf[128]; 113327f654740f2a26ad62a5c155af9199af9e69b889claireho if (!mungeCharName(buf, vname.data(), sizeof(buf))) FAIL(ec); 1134b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UVersionInfo version; 1135b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru u_versionFromString(version, buf); 1136b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru applyFilter(versionFilter, &version, UPROPS_SRC_PROPSVEC, ec); 1137b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return *this; 1138b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1139b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru break; 114027f654740f2a26ad62a5c155af9199af9e69b889claireho case UCHAR_SCRIPT_EXTENSIONS: 114127f654740f2a26ad62a5c155af9199af9e69b889claireho v = u_getPropertyValueEnum(UCHAR_SCRIPT, vname.data()); 114227f654740f2a26ad62a5c155af9199af9e69b889claireho if (v == UCHAR_INVALID_CODE) { 114327f654740f2a26ad62a5c155af9199af9e69b889claireho FAIL(ec); 114427f654740f2a26ad62a5c155af9199af9e69b889claireho } 114527f654740f2a26ad62a5c155af9199af9e69b889claireho // fall through to calling applyIntPropertyValue() 114627f654740f2a26ad62a5c155af9199af9e69b889claireho break; 1147b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru default: 1148b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // p is a non-binary, non-enumerated property that we 1149b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // don't support (yet). 1150b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru FAIL(ec); 1151b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1152b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1153b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1154b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1155b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru else { 1156b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // value is empty. Interpret as General Category, Script, or 1157b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Binary property. 1158b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru p = UCHAR_GENERAL_CATEGORY_MASK; 115927f654740f2a26ad62a5c155af9199af9e69b889claireho v = u_getPropertyValueEnum(p, pname.data()); 1160b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (v == UCHAR_INVALID_CODE) { 1161b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru p = UCHAR_SCRIPT; 116227f654740f2a26ad62a5c155af9199af9e69b889claireho v = u_getPropertyValueEnum(p, pname.data()); 1163b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (v == UCHAR_INVALID_CODE) { 116427f654740f2a26ad62a5c155af9199af9e69b889claireho p = u_getPropertyEnum(pname.data()); 1165b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (p >= UCHAR_BINARY_START && p < UCHAR_BINARY_LIMIT) { 1166b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru v = 1; 116727f654740f2a26ad62a5c155af9199af9e69b889claireho } else if (0 == uprv_comparePropertyNames(ANY, pname.data())) { 1168b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru set(MIN_VALUE, MAX_VALUE); 1169b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return *this; 117027f654740f2a26ad62a5c155af9199af9e69b889claireho } else if (0 == uprv_comparePropertyNames(ASCII, pname.data())) { 1171b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru set(0, 0x7F); 1172b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return *this; 117327f654740f2a26ad62a5c155af9199af9e69b889claireho } else if (0 == uprv_comparePropertyNames(ASSIGNED, pname.data())) { 1174b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // [:Assigned:]=[:^Cn:] 1175b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru p = UCHAR_GENERAL_CATEGORY_MASK; 1176b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru v = U_GC_CN_MASK; 1177b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru invert = TRUE; 1178b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 1179b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru FAIL(ec); 1180b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1181b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1182b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1183b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 118427f654740f2a26ad62a5c155af9199af9e69b889claireho 1185b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru applyIntPropertyValue(p, v, ec); 1186b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(invert) { 1187b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru complement(); 1188b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1189b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1190b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_SUCCESS(ec) && (mustNotBeEmpty && isEmpty())) { 1191b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // mustNotBeEmpty is set to true if an empty set indicates 1192b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // invalid input. 1193b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ec = U_ILLEGAL_ARGUMENT_ERROR; 1194b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1195b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1196c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru if (isBogus() && U_SUCCESS(ec)) { 1197c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru // We likely ran out of memory. AHHH! 1198c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru ec = U_MEMORY_ALLOCATION_ERROR; 1199c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru } 1200b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return *this; 1201b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1202b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1203b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//---------------------------------------------------------------- 1204b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// Property set patterns 1205b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//---------------------------------------------------------------- 1206b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1207b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 1208b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Return true if the given position, in the given pattern, appears 1209b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * to be the start of a property set pattern. 1210b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 1211b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruUBool UnicodeSet::resemblesPropertyPattern(const UnicodeString& pattern, 1212b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t pos) { 1213b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Patterns are at least 5 characters long 1214b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if ((pos+5) > pattern.length()) { 1215b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return FALSE; 1216b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1217b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1218b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Look for an opening [:, [:^, \p, or \P 1219b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return isPOSIXOpen(pattern, pos) || isPerlOpen(pattern, pos) || isNameOpen(pattern, pos); 1220b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1221b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1222b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 1223b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Return true if the given iterator appears to point at a 1224b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * property pattern. Regardless of the result, return with the 1225b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * iterator unchanged. 1226b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param chars iterator over the pattern characters. Upon return 1227b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * it will be unchanged. 1228b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param iterOpts RuleCharacterIterator options 1229b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 1230b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruUBool UnicodeSet::resemblesPropertyPattern(RuleCharacterIterator& chars, 1231b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t iterOpts) { 1232b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // NOTE: literal will always be FALSE, because we don't parse escapes. 1233b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UBool result = FALSE, literal; 1234b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode ec = U_ZERO_ERROR; 1235b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru iterOpts &= ~RuleCharacterIterator::PARSE_ESCAPES; 1236b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru RuleCharacterIterator::Pos pos; 1237b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru chars.getPos(pos); 1238b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar32 c = chars.next(iterOpts, literal, ec); 1239b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (c == 0x5B /*'['*/ || c == 0x5C /*'\\'*/) { 1240b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar32 d = chars.next(iterOpts & ~RuleCharacterIterator::SKIP_WHITESPACE, 1241b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru literal, ec); 1242b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru result = (c == 0x5B /*'['*/) ? (d == 0x3A /*':'*/) : 1243b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru (d == 0x4E /*'N'*/ || d == 0x70 /*'p'*/ || d == 0x50 /*'P'*/); 1244b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1245b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru chars.setPos(pos); 1246b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return result && U_SUCCESS(ec); 1247b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1248b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1249b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 1250b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Parse the given property pattern at the given parse position. 1251b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 1252b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruUnicodeSet& UnicodeSet::applyPropertyPattern(const UnicodeString& pattern, 1253b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ParsePosition& ppos, 1254b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode &ec) { 1255b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t pos = ppos.getIndex(); 1256b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1257b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UBool posix = FALSE; // true for [:pat:], false for \p{pat} \P{pat} \N{pat} 1258b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UBool isName = FALSE; // true for \N{pat}, o/w false 1259b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UBool invert = FALSE; 1260b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1261b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_FAILURE(ec)) return *this; 1262b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1263b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Minimum length is 5 characters, e.g. \p{L} 1264b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if ((pos+5) > pattern.length()) { 1265b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru FAIL(ec); 1266b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1267b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1268b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // On entry, ppos should point to one of the following locations: 1269b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Look for an opening [:, [:^, \p, or \P 1270b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (isPOSIXOpen(pattern, pos)) { 1271b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru posix = TRUE; 1272b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pos += 2; 1273b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pos = ICU_Utility::skipWhitespace(pattern, pos); 1274b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (pos < pattern.length() && pattern.charAt(pos) == COMPLEMENT) { 1275b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ++pos; 1276b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru invert = TRUE; 1277b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1278b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else if (isPerlOpen(pattern, pos) || isNameOpen(pattern, pos)) { 1279b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar c = pattern.charAt(pos+1); 1280b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru invert = (c == UPPER_P); 1281b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru isName = (c == UPPER_N); 1282b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pos += 2; 1283b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pos = ICU_Utility::skipWhitespace(pattern, pos); 1284b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (pos == pattern.length() || pattern.charAt(pos++) != OPEN_BRACE) { 1285b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Syntax error; "\p" or "\P" not followed by "{" 1286b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru FAIL(ec); 1287b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1288b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 1289b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Open delimiter not seen 1290b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru FAIL(ec); 1291b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1292b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1293b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Look for the matching close delimiter, either :] or } 1294b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t close = pattern.indexOf(posix ? POSIX_CLOSE : PERL_CLOSE, pos); 1295b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (close < 0) { 1296b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Syntax error; close delimiter missing 1297b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru FAIL(ec); 1298b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1299b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1300b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Look for an '=' sign. If this is present, we will parse a 1301b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // medium \p{gc=Cf} or long \p{GeneralCategory=Format} 1302b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // pattern. 1303b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t equals = pattern.indexOf(EQUALS, pos); 1304b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UnicodeString propName, valueName; 1305b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (equals >= 0 && equals < close && !isName) { 1306b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Equals seen; parse medium/long pattern 1307b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pattern.extractBetween(pos, equals, propName); 1308b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pattern.extractBetween(equals+1, close, valueName); 1309b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1310b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1311b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru else { 1312b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Handle case where no '=' is seen, and \N{} 1313b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pattern.extractBetween(pos, close, propName); 1314b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1315b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Handle \N{name} 1316b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (isName) { 1317b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // This is a little inefficient since it means we have to 1318b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // parse NAME_PROP back to UCHAR_NAME even though we already 1319b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // know it's UCHAR_NAME. If we refactor the API to 1320b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // support args of (UProperty, char*) then we can remove 1321b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // NAME_PROP and make this a little more efficient. 1322b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru valueName = propName; 1323b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru propName = UnicodeString(NAME_PROP, NAME_PROP_LENGTH, US_INV); 1324b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1325b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1326b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1327b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru applyPropertyAlias(propName, valueName, ec); 1328b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1329b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_SUCCESS(ec)) { 1330b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (invert) { 1331b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru complement(); 1332b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1333b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1334b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // Move to the limit position after the close delimiter if the 1335b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // parse succeeded. 1336b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ppos.setIndex(close + (posix ? 2 : 1)); 1337b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1338b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1339b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return *this; 1340b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1341b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1342b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru/** 1343b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * Parse a property pattern. 1344b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param chars iterator over the pattern characters. Upon return 1345b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * it will be advanced to the first character after the parsed 1346b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * pattern, or the end of the iteration if all characters are 1347b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * parsed. 1348b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * @param rebuiltPat the pattern that was parsed, rebuilt or 1349b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru * copied from the input pattern, as appropriate. 1350b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru */ 1351b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queruvoid UnicodeSet::applyPropertyPattern(RuleCharacterIterator& chars, 1352b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UnicodeString& rebuiltPat, 1353b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UErrorCode& ec) { 1354b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_FAILURE(ec)) return; 1355b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UnicodeString pattern; 1356b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru chars.lookahead(pattern); 1357b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ParsePosition pos(0); 1358b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru applyPropertyPattern(pattern, pos, ec); 1359b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_FAILURE(ec)) return; 1360b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (pos.getIndex() == 0) { 1361b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // syntaxError(chars, "Invalid property pattern"); 1362b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ec = U_MALFORMED_SET; 1363b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return; 1364b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1365b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru chars.jumpahead(pos.getIndex()); 1366b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru rebuiltPat.append(pattern, 0, pos.getIndex()); 1367b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1368b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1369b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//---------------------------------------------------------------- 1370b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// Case folding API 1371b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru//---------------------------------------------------------------- 1372b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1373b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// add the result of a full case mapping to the set 1374b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru// use str as a temporary string to avoid constructing one 1375b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Querustatic inline void 1376b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruaddCaseMapping(UnicodeSet &set, int32_t result, const UChar *full, UnicodeString &str) { 1377b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(result >= 0) { 1378b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(result > UCASE_MAX_STRING_LENGTH) { 1379b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // add a single-code point case mapping 1380b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru set.add(result); 1381b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 1382b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // add a string case mapping from full with length result 1383b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru str.setTo((UBool)FALSE, full, result); 1384b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru set.add(str); 1385b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1386b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1387b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // result < 0: the code point mapped to itself, no need to add it 1388b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // see ucase.h 1389b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1390b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1391b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruUnicodeSet& UnicodeSet::closeOver(int32_t attribute) { 1392c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru if (isFrozen() || isBogus()) { 1393b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return *this; 1394b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1395b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (attribute & (USET_CASE_INSENSITIVE | USET_ADD_CASE_MAPPINGS)) { 139627f654740f2a26ad62a5c155af9199af9e69b889claireho const UCaseProps *csp = ucase_getSingleton(); 139727f654740f2a26ad62a5c155af9199af9e69b889claireho { 1398b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UnicodeSet foldSet(*this); 1399b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UnicodeString str; 1400b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru USetAdder sa = { 140127f654740f2a26ad62a5c155af9199af9e69b889claireho foldSet.toUSet(), 1402b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru _set_add, 1403b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru _set_addRange, 1404b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru _set_addString, 1405c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru NULL, // don't need remove() 1406c69afcec261fc345fda8daf46f0ea6b4351dc777Jean-Baptiste Queru NULL // don't need removeRange() 1407b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru }; 1408b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1409b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // start with input set to guarantee inclusion 1410b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // USET_CASE: remove strings because the strings will actually be reduced (folded); 1411b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // therefore, start with no strings and add only those needed 1412b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (attribute & USET_CASE_INSENSITIVE) { 1413b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru foldSet.strings->removeAllElements(); 1414b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1415b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1416b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t n = getRangeCount(); 1417b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar32 result; 1418b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const UChar *full; 1419b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru int32_t locCache = 0; 1420b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1421b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru for (int32_t i=0; i<n; ++i) { 1422b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar32 start = getRangeStart(i); 1423b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru UChar32 end = getRangeEnd(i); 1424b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1425b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (attribute & USET_CASE_INSENSITIVE) { 1426b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // full case closure 1427b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru for (UChar32 cp=start; cp<=end; ++cp) { 1428b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru ucase_addCaseClosure(csp, cp, &sa); 1429b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1430b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 1431b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // add case mappings 1432b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru // (does not add long s for regular s, or Kelvin for k, for example) 1433b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru for (UChar32 cp=start; cp<=end; ++cp) { 1434b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru result = ucase_toFullLower(csp, cp, NULL, NULL, &full, "", &locCache); 1435b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru addCaseMapping(foldSet, result, full, str); 1436b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1437b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru result = ucase_toFullTitle(csp, cp, NULL, NULL, &full, "", &locCache); 1438b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru addCaseMapping(foldSet, result, full, str); 1439b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1440b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru result = ucase_toFullUpper(csp, cp, NULL, NULL, &full, "", &locCache); 1441b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru addCaseMapping(foldSet, result, full, str); 1442b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1443b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru result = ucase_toFullFolding(csp, cp, &full, 0); 1444b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru addCaseMapping(foldSet, result, full, str); 1445b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1446b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1447b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1448b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (strings != NULL && strings->size() > 0) { 1449b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (attribute & USET_CASE_INSENSITIVE) { 1450b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru for (int32_t j=0; j<strings->size(); ++j) { 1451b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru str = *(const UnicodeString *) strings->elementAt(j); 1452b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru str.foldCase(); 1453b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if(!ucase_addStringCaseClosure(csp, str.getBuffer(), str.length(), &sa)) { 1454b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru foldSet.add(str); // does not map to code points: add the folded string itself 1455b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1456b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1457b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } else { 1458b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru Locale root(""); 1459b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#if !UCONFIG_NO_BREAK_ITERATION 146027f654740f2a26ad62a5c155af9199af9e69b889claireho UErrorCode status = U_ZERO_ERROR; 1461b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru BreakIterator *bi = BreakIterator::createWordInstance(root, status); 1462b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru if (U_SUCCESS(status)) { 146327f654740f2a26ad62a5c155af9199af9e69b889claireho#endif 1464b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru const UnicodeString *pStr; 1465b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1466b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru for (int32_t j=0; j<strings->size(); ++j) { 1467b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru pStr = (const UnicodeString *) strings->elementAt(j); 1468b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru (str = *pStr).toLower(root); 1469b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru foldSet.add(str); 1470b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#if !UCONFIG_NO_BREAK_ITERATION 1471b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru (str = *pStr).toTitle(bi, root); 1472b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru foldSet.add(str); 1473b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#endif 1474b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru (str = *pStr).toUpper(root); 1475b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru foldSet.add(str); 1476b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru (str = *pStr).foldCase(); 1477b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru foldSet.add(str); 1478b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1479b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#if !UCONFIG_NO_BREAK_ITERATION 148027f654740f2a26ad62a5c155af9199af9e69b889claireho } 1481b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru delete bi; 1482b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru#endif 1483b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1484b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1485b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru *this = foldSet; 1486b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1487b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru } 1488b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru return *this; 1489b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru} 1490b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste Queru 1491b13da9df870a61b11249bf741347908dbea0edd8Jean-Baptiste QueruU_NAMESPACE_END 1492