1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/common/uniset_closure.cpp Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,280 @@
1.4 +/*
1.5 +*******************************************************************************
1.6 +*
1.7 +* Copyright (C) 2011, International Business Machines
1.8 +* Corporation and others. All Rights Reserved.
1.9 +*
1.10 +*******************************************************************************
1.11 +* file name: uniset_closure.cpp
1.12 +* encoding: US-ASCII
1.13 +* tab size: 8 (not used)
1.14 +* indentation:4
1.15 +*
1.16 +* created on: 2011may30
1.17 +* created by: Markus W. Scherer
1.18 +*
1.19 +* UnicodeSet::closeOver() and related methods moved here from uniset_props.cpp
1.20 +* to simplify dependencies.
1.21 +* In particular, this depends on the BreakIterator, but the BreakIterator
1.22 +* code also builds UnicodeSets from patterns and needs uniset_props.
1.23 +*/
1.24 +
1.25 +#include "unicode/brkiter.h"
1.26 +#include "unicode/locid.h"
1.27 +#include "unicode/parsepos.h"
1.28 +#include "unicode/uniset.h"
1.29 +#include "cmemory.h"
1.30 +#include "ruleiter.h"
1.31 +#include "ucase.h"
1.32 +#include "util.h"
1.33 +#include "uvector.h"
1.34 +
1.35 +// initial storage. Must be >= 0
1.36 +// *** same as in uniset.cpp ! ***
1.37 +#define START_EXTRA 16
1.38 +
1.39 +U_NAMESPACE_BEGIN
1.40 +
1.41 +// TODO memory debugging provided inside uniset.cpp
1.42 +// could be made available here but probably obsolete with use of modern
1.43 +// memory leak checker tools
1.44 +#define _dbgct(me)
1.45 +
1.46 +//----------------------------------------------------------------
1.47 +// Constructors &c
1.48 +//----------------------------------------------------------------
1.49 +
1.50 +UnicodeSet::UnicodeSet(const UnicodeString& pattern,
1.51 + uint32_t options,
1.52 + const SymbolTable* symbols,
1.53 + UErrorCode& status) :
1.54 + len(0), capacity(START_EXTRA), list(0), bmpSet(0), buffer(0),
1.55 + bufferCapacity(0), patLen(0), pat(NULL), strings(NULL), stringSpan(NULL),
1.56 + fFlags(0)
1.57 +{
1.58 + if(U_SUCCESS(status)){
1.59 + list = (UChar32*) uprv_malloc(sizeof(UChar32) * capacity);
1.60 + /* test for NULL */
1.61 + if(list == NULL) {
1.62 + status = U_MEMORY_ALLOCATION_ERROR;
1.63 + }else{
1.64 + allocateStrings(status);
1.65 + applyPattern(pattern, options, symbols, status);
1.66 + }
1.67 + }
1.68 + _dbgct(this);
1.69 +}
1.70 +
1.71 +UnicodeSet::UnicodeSet(const UnicodeString& pattern, ParsePosition& pos,
1.72 + uint32_t options,
1.73 + const SymbolTable* symbols,
1.74 + UErrorCode& status) :
1.75 + len(0), capacity(START_EXTRA), list(0), bmpSet(0), buffer(0),
1.76 + bufferCapacity(0), patLen(0), pat(NULL), strings(NULL), stringSpan(NULL),
1.77 + fFlags(0)
1.78 +{
1.79 + if(U_SUCCESS(status)){
1.80 + list = (UChar32*) uprv_malloc(sizeof(UChar32) * capacity);
1.81 + /* test for NULL */
1.82 + if(list == NULL) {
1.83 + status = U_MEMORY_ALLOCATION_ERROR;
1.84 + }else{
1.85 + allocateStrings(status);
1.86 + applyPattern(pattern, pos, options, symbols, status);
1.87 + }
1.88 + }
1.89 + _dbgct(this);
1.90 +}
1.91 +
1.92 +//----------------------------------------------------------------
1.93 +// Public API
1.94 +//----------------------------------------------------------------
1.95 +
1.96 +UnicodeSet& UnicodeSet::applyPattern(const UnicodeString& pattern,
1.97 + uint32_t options,
1.98 + const SymbolTable* symbols,
1.99 + UErrorCode& status) {
1.100 + ParsePosition pos(0);
1.101 + applyPattern(pattern, pos, options, symbols, status);
1.102 + if (U_FAILURE(status)) return *this;
1.103 +
1.104 + int32_t i = pos.getIndex();
1.105 +
1.106 + if (options & USET_IGNORE_SPACE) {
1.107 + // Skip over trailing whitespace
1.108 + ICU_Utility::skipWhitespace(pattern, i, TRUE);
1.109 + }
1.110 +
1.111 + if (i != pattern.length()) {
1.112 + status = U_ILLEGAL_ARGUMENT_ERROR;
1.113 + }
1.114 + return *this;
1.115 +}
1.116 +
1.117 +UnicodeSet& UnicodeSet::applyPattern(const UnicodeString& pattern,
1.118 + ParsePosition& pos,
1.119 + uint32_t options,
1.120 + const SymbolTable* symbols,
1.121 + UErrorCode& status) {
1.122 + if (U_FAILURE(status)) {
1.123 + return *this;
1.124 + }
1.125 + if (isFrozen()) {
1.126 + status = U_NO_WRITE_PERMISSION;
1.127 + return *this;
1.128 + }
1.129 + // Need to build the pattern in a temporary string because
1.130 + // _applyPattern calls add() etc., which set pat to empty.
1.131 + UnicodeString rebuiltPat;
1.132 + RuleCharacterIterator chars(pattern, symbols, pos);
1.133 + applyPattern(chars, symbols, rebuiltPat, options, &UnicodeSet::closeOver, status);
1.134 + if (U_FAILURE(status)) return *this;
1.135 + if (chars.inVariable()) {
1.136 + // syntaxError(chars, "Extra chars in variable value");
1.137 + status = U_MALFORMED_SET;
1.138 + return *this;
1.139 + }
1.140 + setPattern(rebuiltPat);
1.141 + return *this;
1.142 +}
1.143 +
1.144 +// USetAdder implementation
1.145 +// Does not use uset.h to reduce code dependencies
1.146 +static void U_CALLCONV
1.147 +_set_add(USet *set, UChar32 c) {
1.148 + ((UnicodeSet *)set)->add(c);
1.149 +}
1.150 +
1.151 +static void U_CALLCONV
1.152 +_set_addRange(USet *set, UChar32 start, UChar32 end) {
1.153 + ((UnicodeSet *)set)->add(start, end);
1.154 +}
1.155 +
1.156 +static void U_CALLCONV
1.157 +_set_addString(USet *set, const UChar *str, int32_t length) {
1.158 + ((UnicodeSet *)set)->add(UnicodeString((UBool)(length<0), str, length));
1.159 +}
1.160 +
1.161 +//----------------------------------------------------------------
1.162 +// Case folding API
1.163 +//----------------------------------------------------------------
1.164 +
1.165 +// add the result of a full case mapping to the set
1.166 +// use str as a temporary string to avoid constructing one
1.167 +static inline void
1.168 +addCaseMapping(UnicodeSet &set, int32_t result, const UChar *full, UnicodeString &str) {
1.169 + if(result >= 0) {
1.170 + if(result > UCASE_MAX_STRING_LENGTH) {
1.171 + // add a single-code point case mapping
1.172 + set.add(result);
1.173 + } else {
1.174 + // add a string case mapping from full with length result
1.175 + str.setTo((UBool)FALSE, full, result);
1.176 + set.add(str);
1.177 + }
1.178 + }
1.179 + // result < 0: the code point mapped to itself, no need to add it
1.180 + // see ucase.h
1.181 +}
1.182 +
1.183 +UnicodeSet& UnicodeSet::closeOver(int32_t attribute) {
1.184 + if (isFrozen() || isBogus()) {
1.185 + return *this;
1.186 + }
1.187 + if (attribute & (USET_CASE_INSENSITIVE | USET_ADD_CASE_MAPPINGS)) {
1.188 + const UCaseProps *csp = ucase_getSingleton();
1.189 + {
1.190 + UnicodeSet foldSet(*this);
1.191 + UnicodeString str;
1.192 + USetAdder sa = {
1.193 + foldSet.toUSet(),
1.194 + _set_add,
1.195 + _set_addRange,
1.196 + _set_addString,
1.197 + NULL, // don't need remove()
1.198 + NULL // don't need removeRange()
1.199 + };
1.200 +
1.201 + // start with input set to guarantee inclusion
1.202 + // USET_CASE: remove strings because the strings will actually be reduced (folded);
1.203 + // therefore, start with no strings and add only those needed
1.204 + if (attribute & USET_CASE_INSENSITIVE) {
1.205 + foldSet.strings->removeAllElements();
1.206 + }
1.207 +
1.208 + int32_t n = getRangeCount();
1.209 + UChar32 result;
1.210 + const UChar *full;
1.211 + int32_t locCache = 0;
1.212 +
1.213 + for (int32_t i=0; i<n; ++i) {
1.214 + UChar32 start = getRangeStart(i);
1.215 + UChar32 end = getRangeEnd(i);
1.216 +
1.217 + if (attribute & USET_CASE_INSENSITIVE) {
1.218 + // full case closure
1.219 + for (UChar32 cp=start; cp<=end; ++cp) {
1.220 + ucase_addCaseClosure(csp, cp, &sa);
1.221 + }
1.222 + } else {
1.223 + // add case mappings
1.224 + // (does not add long s for regular s, or Kelvin for k, for example)
1.225 + for (UChar32 cp=start; cp<=end; ++cp) {
1.226 + result = ucase_toFullLower(csp, cp, NULL, NULL, &full, "", &locCache);
1.227 + addCaseMapping(foldSet, result, full, str);
1.228 +
1.229 + result = ucase_toFullTitle(csp, cp, NULL, NULL, &full, "", &locCache);
1.230 + addCaseMapping(foldSet, result, full, str);
1.231 +
1.232 + result = ucase_toFullUpper(csp, cp, NULL, NULL, &full, "", &locCache);
1.233 + addCaseMapping(foldSet, result, full, str);
1.234 +
1.235 + result = ucase_toFullFolding(csp, cp, &full, 0);
1.236 + addCaseMapping(foldSet, result, full, str);
1.237 + }
1.238 + }
1.239 + }
1.240 + if (strings != NULL && strings->size() > 0) {
1.241 + if (attribute & USET_CASE_INSENSITIVE) {
1.242 + for (int32_t j=0; j<strings->size(); ++j) {
1.243 + str = *(const UnicodeString *) strings->elementAt(j);
1.244 + str.foldCase();
1.245 + if(!ucase_addStringCaseClosure(csp, str.getBuffer(), str.length(), &sa)) {
1.246 + foldSet.add(str); // does not map to code points: add the folded string itself
1.247 + }
1.248 + }
1.249 + } else {
1.250 + Locale root("");
1.251 +#if !UCONFIG_NO_BREAK_ITERATION
1.252 + UErrorCode status = U_ZERO_ERROR;
1.253 + BreakIterator *bi = BreakIterator::createWordInstance(root, status);
1.254 + if (U_SUCCESS(status)) {
1.255 +#endif
1.256 + const UnicodeString *pStr;
1.257 +
1.258 + for (int32_t j=0; j<strings->size(); ++j) {
1.259 + pStr = (const UnicodeString *) strings->elementAt(j);
1.260 + (str = *pStr).toLower(root);
1.261 + foldSet.add(str);
1.262 +#if !UCONFIG_NO_BREAK_ITERATION
1.263 + (str = *pStr).toTitle(bi, root);
1.264 + foldSet.add(str);
1.265 +#endif
1.266 + (str = *pStr).toUpper(root);
1.267 + foldSet.add(str);
1.268 + (str = *pStr).foldCase();
1.269 + foldSet.add(str);
1.270 + }
1.271 +#if !UCONFIG_NO_BREAK_ITERATION
1.272 + }
1.273 + delete bi;
1.274 +#endif
1.275 + }
1.276 + }
1.277 + *this = foldSet;
1.278 + }
1.279 + }
1.280 + return *this;
1.281 +}
1.282 +
1.283 +U_NAMESPACE_END
