1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/i18n/tblcoll.cpp Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,695 @@
1.4 +/*
1.5 + ******************************************************************************
1.6 + * Copyright (C) 1996-2013, International Business Machines Corporation and
1.7 + * others. All Rights Reserved.
1.8 + ******************************************************************************
1.9 + */
1.10 +
1.11 +/**
1.12 + * File tblcoll.cpp
1.13 + *
1.14 + * Created by: Helena Shih
1.15 + *
1.16 + * Modification History:
1.17 + *
1.18 + * Date Name Description
1.19 + * 2/5/97 aliu Added streamIn and streamOut methods. Added
1.20 + * constructor which reads RuleBasedCollator object from
1.21 + * a binary file. Added writeToFile method which streams
1.22 + * RuleBasedCollator out to a binary file. The streamIn
1.23 + * and streamOut methods use istream and ostream objects
1.24 + * in binary mode.
1.25 + * 2/11/97 aliu Moved declarations out of for loop initializer.
1.26 + * Added Mac compatibility #ifdef for ios::nocreate.
1.27 + * 2/12/97 aliu Modified to use TableCollationData sub-object to
1.28 + * hold invariant data.
1.29 + * 2/13/97 aliu Moved several methods into this class from Collation.
1.30 + * Added a private RuleBasedCollator(Locale&) constructor,
1.31 + * to be used by Collator::getInstance(). General
1.32 + * clean up. Made use of UErrorCode variables consistent.
1.33 + * 2/20/97 helena Added clone, operator==, operator!=, operator=, and copy
1.34 + * constructor and getDynamicClassID.
1.35 + * 3/5/97 aliu Changed compaction cycle to improve performance. We
1.36 + * use the maximum allowable value which is kBlockCount.
1.37 + * Modified getRules() to load rules dynamically. Changed
1.38 + * constructFromFile() call to accomodate this (added
1.39 + * parameter to specify whether binary loading is to
1.40 + * take place).
1.41 + * 05/06/97 helena Added memory allocation error check.
1.42 + * 6/20/97 helena Java class name change.
1.43 + * 6/23/97 helena Adding comments to make code more readable.
1.44 + * 09/03/97 helena Added createCollationKeyValues().
1.45 + * 06/26/98 erm Changes for CollationKeys using byte arrays.
1.46 + * 08/10/98 erm Synched with 1.2 version of RuleBasedCollator.java
1.47 + * 04/23/99 stephen Removed EDecompositionMode, merged with
1.48 + * Normalizer::EMode
1.49 + * 06/14/99 stephen Removed kResourceBundleSuffix
1.50 + * 06/22/99 stephen Fixed logic in constructFromFile() since .ctx
1.51 + * files are no longer used.
1.52 + * 11/02/99 helena Collator performance enhancements. Special case
1.53 + * for NO_OP situations.
1.54 + * 11/17/99 srl More performance enhancements. Inlined some internal functions.
1.55 + * 12/15/99 aliu Update to support Thai collation. Move NormalizerIterator
1.56 + * to implementation file.
1.57 + * 01/29/01 synwee Modified into a C++ wrapper calling C APIs (ucol.h)
1.58 + */
1.59 +
1.60 +#include "unicode/utypes.h"
1.61 +
1.62 +#if !UCONFIG_NO_COLLATION
1.63 +
1.64 +#include "unicode/tblcoll.h"
1.65 +#include "unicode/coleitr.h"
1.66 +#include "unicode/ures.h"
1.67 +#include "unicode/uset.h"
1.68 +#include "ucol_imp.h"
1.69 +#include "uresimp.h"
1.70 +#include "uhash.h"
1.71 +#include "cmemory.h"
1.72 +#include "cstring.h"
1.73 +#include "putilimp.h"
1.74 +#include "ustr_imp.h"
1.75 +
1.76 +/* public RuleBasedCollator constructor ---------------------------------- */
1.77 +
1.78 +U_NAMESPACE_BEGIN
1.79 +
1.80 +/**
1.81 +* Copy constructor, aliasing, not write-through
1.82 +*/
1.83 +RuleBasedCollator::RuleBasedCollator(const RuleBasedCollator& that)
1.84 +: Collator(that)
1.85 +, dataIsOwned(FALSE)
1.86 +, isWriteThroughAlias(FALSE)
1.87 +, ucollator(NULL)
1.88 +{
1.89 + RuleBasedCollator::operator=(that);
1.90 +}
1.91 +
1.92 +RuleBasedCollator::RuleBasedCollator(const UnicodeString& rules,
1.93 + UErrorCode& status) :
1.94 +dataIsOwned(FALSE)
1.95 +{
1.96 + construct(rules,
1.97 + UCOL_DEFAULT_STRENGTH,
1.98 + UCOL_DEFAULT,
1.99 + status);
1.100 +}
1.101 +
1.102 +RuleBasedCollator::RuleBasedCollator(const UnicodeString& rules,
1.103 + ECollationStrength collationStrength,
1.104 + UErrorCode& status) : dataIsOwned(FALSE)
1.105 +{
1.106 + construct(rules,
1.107 + (UColAttributeValue)collationStrength,
1.108 + UCOL_DEFAULT,
1.109 + status);
1.110 +}
1.111 +
1.112 +RuleBasedCollator::RuleBasedCollator(const UnicodeString& rules,
1.113 + UColAttributeValue decompositionMode,
1.114 + UErrorCode& status) :
1.115 +dataIsOwned(FALSE)
1.116 +{
1.117 + construct(rules,
1.118 + UCOL_DEFAULT_STRENGTH,
1.119 + decompositionMode,
1.120 + status);
1.121 +}
1.122 +
1.123 +RuleBasedCollator::RuleBasedCollator(const UnicodeString& rules,
1.124 + ECollationStrength collationStrength,
1.125 + UColAttributeValue decompositionMode,
1.126 + UErrorCode& status) : dataIsOwned(FALSE)
1.127 +{
1.128 + construct(rules,
1.129 + (UColAttributeValue)collationStrength,
1.130 + decompositionMode,
1.131 + status);
1.132 +}
1.133 +RuleBasedCollator::RuleBasedCollator(const uint8_t *bin, int32_t length,
1.134 + const RuleBasedCollator *base,
1.135 + UErrorCode &status) :
1.136 +dataIsOwned(TRUE),
1.137 +isWriteThroughAlias(FALSE)
1.138 +{
1.139 + ucollator = ucol_openBinary(bin, length, base->ucollator, &status);
1.140 +}
1.141 +
1.142 +void
1.143 +RuleBasedCollator::setRuleStringFromCollator()
1.144 +{
1.145 + int32_t length;
1.146 + const UChar *r = ucol_getRules(ucollator, &length);
1.147 +
1.148 + if (r && length > 0) {
1.149 + // alias the rules string
1.150 + urulestring.setTo(TRUE, r, length);
1.151 + }
1.152 + else {
1.153 + urulestring.truncate(0); // Clear string.
1.154 + }
1.155 +}
1.156 +
1.157 +// not aliasing, not write-through
1.158 +void
1.159 +RuleBasedCollator::construct(const UnicodeString& rules,
1.160 + UColAttributeValue collationStrength,
1.161 + UColAttributeValue decompositionMode,
1.162 + UErrorCode& status)
1.163 +{
1.164 + ucollator = ucol_openRules(rules.getBuffer(), rules.length(),
1.165 + decompositionMode, collationStrength,
1.166 + NULL, &status);
1.167 +
1.168 + dataIsOwned = TRUE; // since we own a collator now, we need to get rid of it
1.169 + isWriteThroughAlias = FALSE;
1.170 +
1.171 + if(ucollator == NULL) {
1.172 + if(U_SUCCESS(status)) {
1.173 + status = U_MEMORY_ALLOCATION_ERROR;
1.174 + }
1.175 + return; // Failure
1.176 + }
1.177 +
1.178 + setRuleStringFromCollator();
1.179 +}
1.180 +
1.181 +/* RuleBasedCollator public destructor ----------------------------------- */
1.182 +
1.183 +RuleBasedCollator::~RuleBasedCollator()
1.184 +{
1.185 + if (dataIsOwned)
1.186 + {
1.187 + ucol_close(ucollator);
1.188 + }
1.189 + ucollator = 0;
1.190 +}
1.191 +
1.192 +/* RuleBaseCollator public methods --------------------------------------- */
1.193 +
1.194 +UBool RuleBasedCollator::operator==(const Collator& that) const
1.195 +{
1.196 + /* only checks for address equals here */
1.197 + if (this == &that) {
1.198 + return TRUE;
1.199 + }
1.200 + if (!Collator::operator==(that)) {
1.201 + return FALSE; /* not the same class */
1.202 + }
1.203 +
1.204 + RuleBasedCollator& thatAlias = (RuleBasedCollator&)that;
1.205 +
1.206 + return ucol_equals(this->ucollator, thatAlias.ucollator);
1.207 +}
1.208 +
1.209 +// aliasing, not write-through
1.210 +RuleBasedCollator& RuleBasedCollator::operator=(const RuleBasedCollator& that)
1.211 +{
1.212 + if (this == &that) { return *this; }
1.213 +
1.214 + UErrorCode intStatus = U_ZERO_ERROR;
1.215 + UCollator *ucol = ucol_safeClone(that.ucollator, NULL, NULL, &intStatus);
1.216 + if (U_FAILURE(intStatus)) { return *this; }
1.217 +
1.218 + if (dataIsOwned) {
1.219 + ucol_close(ucollator);
1.220 + }
1.221 + ucollator = ucol;
1.222 + dataIsOwned = TRUE;
1.223 + isWriteThroughAlias = FALSE;
1.224 + setRuleStringFromCollator();
1.225 + return *this;
1.226 +}
1.227 +
1.228 +// aliasing, not write-through
1.229 +Collator* RuleBasedCollator::clone() const
1.230 +{
1.231 + RuleBasedCollator* coll = new RuleBasedCollator(*this);
1.232 + // There is a small chance that the internal ucol_safeClone() call fails.
1.233 + if (coll != NULL && coll->ucollator == NULL) {
1.234 + delete coll;
1.235 + return NULL;
1.236 + }
1.237 + return coll;
1.238 +}
1.239 +
1.240 +
1.241 +CollationElementIterator* RuleBasedCollator::createCollationElementIterator
1.242 + (const UnicodeString& source) const
1.243 +{
1.244 + UErrorCode status = U_ZERO_ERROR;
1.245 + CollationElementIterator *result = new CollationElementIterator(source, this,
1.246 + status);
1.247 + if (U_FAILURE(status)) {
1.248 + delete result;
1.249 + return NULL;
1.250 + }
1.251 +
1.252 + return result;
1.253 +}
1.254 +
1.255 +/**
1.256 +* Create a CollationElementIterator object that will iterate over the
1.257 +* elements in a string, using the collation rules defined in this
1.258 +* RuleBasedCollator
1.259 +*/
1.260 +CollationElementIterator* RuleBasedCollator::createCollationElementIterator
1.261 + (const CharacterIterator& source) const
1.262 +{
1.263 + UErrorCode status = U_ZERO_ERROR;
1.264 + CollationElementIterator *result = new CollationElementIterator(source, this,
1.265 + status);
1.266 +
1.267 + if (U_FAILURE(status)) {
1.268 + delete result;
1.269 + return NULL;
1.270 + }
1.271 +
1.272 + return result;
1.273 +}
1.274 +
1.275 +/**
1.276 +* Return a string representation of this collator's rules. The string can
1.277 +* later be passed to the constructor that takes a UnicodeString argument,
1.278 +* which will construct a collator that's functionally identical to this one.
1.279 +* You can also allow users to edit the string in order to change the collation
1.280 +* data, or you can print it out for inspection, or whatever.
1.281 +*/
1.282 +const UnicodeString& RuleBasedCollator::getRules() const
1.283 +{
1.284 + return urulestring;
1.285 +}
1.286 +
1.287 +void RuleBasedCollator::getRules(UColRuleOption delta, UnicodeString &buffer)
1.288 +{
1.289 + int32_t rulesize = ucol_getRulesEx(ucollator, delta, NULL, -1);
1.290 +
1.291 + if (rulesize > 0) {
1.292 + UChar *rules = (UChar*) uprv_malloc( sizeof(UChar) * (rulesize) );
1.293 + if(rules != NULL) {
1.294 + ucol_getRulesEx(ucollator, delta, rules, rulesize);
1.295 + buffer.setTo(rules, rulesize);
1.296 + uprv_free(rules);
1.297 + } else { // couldn't allocate
1.298 + buffer.remove();
1.299 + }
1.300 + }
1.301 + else {
1.302 + buffer.remove();
1.303 + }
1.304 +}
1.305 +
1.306 +UnicodeSet *
1.307 +RuleBasedCollator::getTailoredSet(UErrorCode &status) const
1.308 +{
1.309 + if(U_FAILURE(status)) {
1.310 + return NULL;
1.311 + }
1.312 + return (UnicodeSet *)ucol_getTailoredSet(this->ucollator, &status);
1.313 +}
1.314 +
1.315 +
1.316 +void RuleBasedCollator::getVersion(UVersionInfo versionInfo) const
1.317 +{
1.318 + if (versionInfo!=NULL){
1.319 + ucol_getVersion(ucollator, versionInfo);
1.320 + }
1.321 +}
1.322 +
1.323 +/**
1.324 +* Compare two strings using this collator
1.325 +*/
1.326 +UCollationResult RuleBasedCollator::compare(
1.327 + const UnicodeString& source,
1.328 + const UnicodeString& target,
1.329 + int32_t length,
1.330 + UErrorCode &status) const
1.331 +{
1.332 + return compare(source.getBuffer(), uprv_min(length,source.length()), target.getBuffer(), uprv_min(length,target.length()), status);
1.333 +}
1.334 +
1.335 +UCollationResult RuleBasedCollator::compare(const UChar* source,
1.336 + int32_t sourceLength,
1.337 + const UChar* target,
1.338 + int32_t targetLength,
1.339 + UErrorCode &status) const
1.340 +{
1.341 + if(U_SUCCESS(status)) {
1.342 + return ucol_strcoll(ucollator, source, sourceLength, target, targetLength);
1.343 + } else {
1.344 + return UCOL_EQUAL;
1.345 + }
1.346 +}
1.347 +
1.348 +UCollationResult RuleBasedCollator::compare(
1.349 + const UnicodeString& source,
1.350 + const UnicodeString& target,
1.351 + UErrorCode &status) const
1.352 +{
1.353 + if(U_SUCCESS(status)) {
1.354 + return ucol_strcoll(ucollator, source.getBuffer(), source.length(),
1.355 + target.getBuffer(), target.length());
1.356 + } else {
1.357 + return UCOL_EQUAL;
1.358 + }
1.359 +}
1.360 +
1.361 +UCollationResult RuleBasedCollator::compare(UCharIterator &sIter,
1.362 + UCharIterator &tIter,
1.363 + UErrorCode &status) const {
1.364 + if(U_SUCCESS(status)) {
1.365 + return ucol_strcollIter(ucollator, &sIter, &tIter, &status);
1.366 + } else {
1.367 + return UCOL_EQUAL;
1.368 + }
1.369 +}
1.370 +
1.371 +/**
1.372 +* Retrieve a collation key for the specified string. The key can be compared
1.373 +* with other collation keys using a bitwise comparison (e.g. memcmp) to find
1.374 +* the ordering of their respective source strings. This is handy when doing a
1.375 +* sort, where each sort key must be compared many times.
1.376 +*
1.377 +* The basic algorithm here is to find all of the collation elements for each
1.378 +* character in the source string, convert them to an ASCII representation, and
1.379 +* put them into the collation key. But it's trickier than that. Each
1.380 +* collation element in a string has three components: primary ('A' vs 'B'),
1.381 +* secondary ('u' vs '\u00FC'), and tertiary ('A' vs 'a'), and a primary difference
1.382 +* at the end of a string takes precedence over a secondary or tertiary
1.383 +* difference earlier in the string.
1.384 +*
1.385 +* To account for this, we put all of the primary orders at the beginning of
1.386 +* the string, followed by the secondary and tertiary orders. Each set of
1.387 +* orders is terminated by nulls so that a key for a string which is a initial
1.388 +* substring of another key will compare less without any special case.
1.389 +*
1.390 +* Here's a hypothetical example, with the collation element represented as a
1.391 +* three-digit number, one digit for primary, one for secondary, etc.
1.392 +*
1.393 +* String: A a B \u00C9
1.394 +* Collation Elements: 101 100 201 511
1.395 +* Collation Key: 1125<null>0001<null>1011<null>
1.396 +*
1.397 +* To make things even trickier, secondary differences (accent marks) are
1.398 +* compared starting at the *end* of the string in languages with French
1.399 +* secondary ordering. But when comparing the accent marks on a single base
1.400 +* character, they are compared from the beginning. To handle this, we reverse
1.401 +* all of the accents that belong to each base character, then we reverse the
1.402 +* entire string of secondary orderings at the end.
1.403 +*/
1.404 +CollationKey& RuleBasedCollator::getCollationKey(
1.405 + const UnicodeString& source,
1.406 + CollationKey& sortkey,
1.407 + UErrorCode& status) const
1.408 +{
1.409 + return getCollationKey(source.getBuffer(), source.length(), sortkey, status);
1.410 +}
1.411 +
1.412 +CollationKey& RuleBasedCollator::getCollationKey(const UChar* source,
1.413 + int32_t sourceLen,
1.414 + CollationKey& sortkey,
1.415 + UErrorCode& status) const
1.416 +{
1.417 + if (U_FAILURE(status)) {
1.418 + return sortkey.setToBogus();
1.419 + }
1.420 + if (sourceLen < -1 || (source == NULL && sourceLen != 0)) {
1.421 + status = U_ILLEGAL_ARGUMENT_ERROR;
1.422 + return sortkey.setToBogus();
1.423 + }
1.424 +
1.425 + if (sourceLen < 0) {
1.426 + sourceLen = u_strlen(source);
1.427 + }
1.428 + if (sourceLen == 0) {
1.429 + return sortkey.reset();
1.430 + }
1.431 +
1.432 + int32_t resultLen = ucol_getCollationKey(ucollator, source, sourceLen, sortkey, status);
1.433 +
1.434 + if (U_SUCCESS(status)) {
1.435 + sortkey.setLength(resultLen);
1.436 + } else {
1.437 + sortkey.setToBogus();
1.438 + }
1.439 + return sortkey;
1.440 +}
1.441 +
1.442 +/**
1.443 + * Return the maximum length of any expansion sequences that end with the
1.444 + * specified comparison order.
1.445 + * @param order a collation order returned by previous or next.
1.446 + * @return the maximum length of any expansion seuences ending with the
1.447 + * specified order or 1 if collation order does not occur at the end of any
1.448 + * expansion sequence.
1.449 + * @see CollationElementIterator#getMaxExpansion
1.450 + */
1.451 +int32_t RuleBasedCollator::getMaxExpansion(int32_t order) const
1.452 +{
1.453 + uint8_t result;
1.454 + UCOL_GETMAXEXPANSION(ucollator, (uint32_t)order, result);
1.455 + return result;
1.456 +}
1.457 +
1.458 +uint8_t* RuleBasedCollator::cloneRuleData(int32_t &length,
1.459 + UErrorCode &status)
1.460 +{
1.461 + if (U_FAILURE(status)) { return NULL; }
1.462 + LocalMemory<uint8_t> buffer((uint8_t *)uprv_malloc(20000));
1.463 + if (buffer.isNull()) {
1.464 + status = U_MEMORY_ALLOCATION_ERROR;
1.465 + return NULL;
1.466 + }
1.467 + length = cloneBinary(buffer.getAlias(), 20000, status);
1.468 + if (status == U_BUFFER_OVERFLOW_ERROR) {
1.469 + if (buffer.allocateInsteadAndCopy(length, 0) == NULL) {
1.470 + status = U_MEMORY_ALLOCATION_ERROR;
1.471 + return NULL;
1.472 + }
1.473 + status = U_ZERO_ERROR;
1.474 + length = cloneBinary(buffer.getAlias(), length, status);
1.475 + }
1.476 + if (U_FAILURE(status)) { return NULL; }
1.477 + return buffer.orphan();
1.478 +}
1.479 +
1.480 +
1.481 +int32_t RuleBasedCollator::cloneBinary(uint8_t *buffer, int32_t capacity, UErrorCode &status)
1.482 +{
1.483 + return ucol_cloneBinary(ucollator, buffer, capacity, &status);
1.484 +}
1.485 +
1.486 +void RuleBasedCollator::setAttribute(UColAttribute attr,
1.487 + UColAttributeValue value,
1.488 + UErrorCode &status)
1.489 +{
1.490 + if (U_FAILURE(status))
1.491 + return;
1.492 + checkOwned();
1.493 + ucol_setAttribute(ucollator, attr, value, &status);
1.494 +}
1.495 +
1.496 +UColAttributeValue RuleBasedCollator::getAttribute(UColAttribute attr,
1.497 + UErrorCode &status) const
1.498 +{
1.499 + if (U_FAILURE(status))
1.500 + return UCOL_DEFAULT;
1.501 + return ucol_getAttribute(ucollator, attr, &status);
1.502 +}
1.503 +
1.504 +uint32_t RuleBasedCollator::setVariableTop(const UChar *varTop, int32_t len, UErrorCode &status) {
1.505 + checkOwned();
1.506 + return ucol_setVariableTop(ucollator, varTop, len, &status);
1.507 +}
1.508 +
1.509 +uint32_t RuleBasedCollator::setVariableTop(const UnicodeString &varTop, UErrorCode &status) {
1.510 + checkOwned();
1.511 + return ucol_setVariableTop(ucollator, varTop.getBuffer(), varTop.length(), &status);
1.512 +}
1.513 +
1.514 +void RuleBasedCollator::setVariableTop(uint32_t varTop, UErrorCode &status) {
1.515 + checkOwned();
1.516 + ucol_restoreVariableTop(ucollator, varTop, &status);
1.517 +}
1.518 +
1.519 +uint32_t RuleBasedCollator::getVariableTop(UErrorCode &status) const {
1.520 + return ucol_getVariableTop(ucollator, &status);
1.521 +}
1.522 +
1.523 +int32_t RuleBasedCollator::getSortKey(const UnicodeString& source,
1.524 + uint8_t *result, int32_t resultLength)
1.525 + const
1.526 +{
1.527 + return ucol_getSortKey(ucollator, source.getBuffer(), source.length(), result, resultLength);
1.528 +}
1.529 +
1.530 +int32_t RuleBasedCollator::getSortKey(const UChar *source,
1.531 + int32_t sourceLength, uint8_t *result,
1.532 + int32_t resultLength) const
1.533 +{
1.534 + return ucol_getSortKey(ucollator, source, sourceLength, result, resultLength);
1.535 +}
1.536 +
1.537 +int32_t RuleBasedCollator::getReorderCodes(int32_t *dest,
1.538 + int32_t destCapacity,
1.539 + UErrorCode& status) const
1.540 +{
1.541 + return ucol_getReorderCodes(ucollator, dest, destCapacity, &status);
1.542 +}
1.543 +
1.544 +void RuleBasedCollator::setReorderCodes(const int32_t *reorderCodes,
1.545 + int32_t reorderCodesLength,
1.546 + UErrorCode& status)
1.547 +{
1.548 + checkOwned();
1.549 + ucol_setReorderCodes(ucollator, reorderCodes, reorderCodesLength, &status);
1.550 +}
1.551 +
1.552 +int32_t RuleBasedCollator::getEquivalentReorderCodes(int32_t reorderCode,
1.553 + int32_t* dest,
1.554 + int32_t destCapacity,
1.555 + UErrorCode& status)
1.556 +{
1.557 + return ucol_getEquivalentReorderCodes(reorderCode, dest, destCapacity, &status);
1.558 +}
1.559 +
1.560 +/**
1.561 +* Create a hash code for this collation. Just hash the main rule table -- that
1.562 +* should be good enough for almost any use.
1.563 +*/
1.564 +int32_t RuleBasedCollator::hashCode() const
1.565 +{
1.566 + int32_t length;
1.567 + const UChar *rules = ucol_getRules(ucollator, &length);
1.568 + return ustr_hashUCharsN(rules, length);
1.569 +}
1.570 +
1.571 +/**
1.572 +* return the locale of this collator
1.573 +*/
1.574 +Locale RuleBasedCollator::getLocale(ULocDataLocaleType type, UErrorCode &status) const {
1.575 + const char *result = ucol_getLocaleByType(ucollator, type, &status);
1.576 + if(result == NULL) {
1.577 + Locale res("");
1.578 + res.setToBogus();
1.579 + return res;
1.580 + } else {
1.581 + return Locale(result);
1.582 + }
1.583 +}
1.584 +
1.585 +void
1.586 +RuleBasedCollator::setLocales(const Locale& requestedLocale, const Locale& validLocale, const Locale& actualLocale) {
1.587 + checkOwned();
1.588 + char* rloc = uprv_strdup(requestedLocale.getName());
1.589 + if (rloc) {
1.590 + char* vloc = uprv_strdup(validLocale.getName());
1.591 + if (vloc) {
1.592 + char* aloc = uprv_strdup(actualLocale.getName());
1.593 + if (aloc) {
1.594 + ucol_setReqValidLocales(ucollator, rloc, vloc, aloc);
1.595 + return;
1.596 + }
1.597 + uprv_free(vloc);
1.598 + }
1.599 + uprv_free(rloc);
1.600 + }
1.601 +}
1.602 +
1.603 +// RuleBaseCollatorNew private constructor ----------------------------------
1.604 +
1.605 +RuleBasedCollator::RuleBasedCollator()
1.606 + : dataIsOwned(FALSE), isWriteThroughAlias(FALSE), ucollator(NULL)
1.607 +{
1.608 +}
1.609 +
1.610 +RuleBasedCollator::RuleBasedCollator(const Locale& desiredLocale,
1.611 + UErrorCode& status)
1.612 + : dataIsOwned(FALSE), isWriteThroughAlias(FALSE), ucollator(NULL)
1.613 +{
1.614 + if (U_FAILURE(status))
1.615 + return;
1.616 +
1.617 + /*
1.618 + Try to load, in order:
1.619 + 1. The desired locale's collation.
1.620 + 2. A fallback of the desired locale.
1.621 + 3. The default locale's collation.
1.622 + 4. A fallback of the default locale.
1.623 + 5. The default collation rules, which contains en_US collation rules.
1.624 +
1.625 + To reiterate, we try:
1.626 + Specific:
1.627 + language+country+variant
1.628 + language+country
1.629 + language
1.630 + Default:
1.631 + language+country+variant
1.632 + language+country
1.633 + language
1.634 + Root: (aka DEFAULTRULES)
1.635 + steps 1-5 are handled by resource bundle fallback mechanism.
1.636 + however, in a very unprobable situation that no resource bundle
1.637 + data exists, step 5 is repeated with hardcoded default rules.
1.638 + */
1.639 +
1.640 + setUCollator(desiredLocale, status);
1.641 +
1.642 + if (U_FAILURE(status))
1.643 + {
1.644 + status = U_ZERO_ERROR;
1.645 +
1.646 + setUCollator(kRootLocaleName, status);
1.647 + if (status == U_ZERO_ERROR) {
1.648 + status = U_USING_DEFAULT_WARNING;
1.649 + }
1.650 + }
1.651 +
1.652 + if (U_SUCCESS(status))
1.653 + {
1.654 + setRuleStringFromCollator();
1.655 + }
1.656 +}
1.657 +
1.658 +void
1.659 +RuleBasedCollator::setUCollator(const char *locale,
1.660 + UErrorCode &status)
1.661 +{
1.662 + if (U_FAILURE(status)) {
1.663 + return;
1.664 + }
1.665 + if (ucollator && dataIsOwned)
1.666 + ucol_close(ucollator);
1.667 + ucollator = ucol_open_internal(locale, &status);
1.668 + dataIsOwned = TRUE;
1.669 + isWriteThroughAlias = FALSE;
1.670 +}
1.671 +
1.672 +
1.673 +void
1.674 +RuleBasedCollator::checkOwned() {
1.675 + if (!(dataIsOwned || isWriteThroughAlias)) {
1.676 + UErrorCode status = U_ZERO_ERROR;
1.677 + ucollator = ucol_safeClone(ucollator, NULL, NULL, &status);
1.678 + setRuleStringFromCollator();
1.679 + dataIsOwned = TRUE;
1.680 + isWriteThroughAlias = FALSE;
1.681 + }
1.682 +}
1.683 +
1.684 +
1.685 +int32_t RuleBasedCollator::internalGetShortDefinitionString(const char *locale,
1.686 + char *buffer,
1.687 + int32_t capacity,
1.688 + UErrorCode &status) const {
1.689 + /* simply delegate */
1.690 + return ucol_getShortDefinitionString(ucollator, locale, buffer, capacity, &status);
1.691 +}
1.692 +
1.693 +
1.694 +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(RuleBasedCollator)
1.695 +
1.696 +U_NAMESPACE_END
1.697 +
1.698 +#endif /* #if !UCONFIG_NO_COLLATION */
