1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/i18n/stsearch.cpp Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,509 @@
1.4 +/*
1.5 +**********************************************************************
1.6 +* Copyright (C) 2001-2008 IBM and others. All rights reserved.
1.7 +**********************************************************************
1.8 +* Date Name Description
1.9 +* 03/22/2000 helena Creation.
1.10 +**********************************************************************
1.11 +*/
1.12 +
1.13 +#include "unicode/utypes.h"
1.14 +
1.15 +#if !UCONFIG_NO_COLLATION && !UCONFIG_NO_BREAK_ITERATION
1.16 +
1.17 +#include "unicode/stsearch.h"
1.18 +#include "usrchimp.h"
1.19 +#include "cmemory.h"
1.20 +
1.21 +U_NAMESPACE_BEGIN
1.22 +
1.23 +UOBJECT_DEFINE_RTTI_IMPLEMENTATION(StringSearch)
1.24 +
1.25 +// public constructors and destructors -----------------------------------
1.26 +
1.27 +StringSearch::StringSearch(const UnicodeString &pattern,
1.28 + const UnicodeString &text,
1.29 + const Locale &locale,
1.30 + BreakIterator *breakiter,
1.31 + UErrorCode &status) :
1.32 + SearchIterator(text, breakiter),
1.33 + m_collator_(),
1.34 + m_pattern_(pattern)
1.35 +{
1.36 + if (U_FAILURE(status)) {
1.37 + m_strsrch_ = NULL;
1.38 + return;
1.39 + }
1.40 +
1.41 + m_strsrch_ = usearch_open(m_pattern_.getBuffer(), m_pattern_.length(),
1.42 + m_text_.getBuffer(), m_text_.length(),
1.43 + locale.getName(), (UBreakIterator *)breakiter,
1.44 + &status);
1.45 + uprv_free(m_search_);
1.46 + m_search_ = NULL;
1.47 +
1.48 + // !!! dlf m_collator_ is an odd beast. basically it is an aliasing
1.49 + // wrapper around the internal collator and rules, which (here) are
1.50 + // owned by this stringsearch object. this means 1) it's destructor
1.51 + // _should not_ delete the ucollator or rules, and 2) changes made
1.52 + // to the exposed collator (setStrength etc) _should_ modify the
1.53 + // ucollator. thus the collator is not a copy-on-write alias, and it
1.54 + // needs to distinguish itself not merely from 'stand alone' colators
1.55 + // but also from copy-on-write ones. it needs additional state, which
1.56 + // setUCollator should set.
1.57 +
1.58 + if (U_SUCCESS(status)) {
1.59 + // Alias the collator
1.60 + m_collator_.setUCollator((UCollator *)m_strsrch_->collator);
1.61 + // m_search_ has been created by the base SearchIterator class
1.62 + m_search_ = m_strsrch_->search;
1.63 + }
1.64 +}
1.65 +
1.66 +StringSearch::StringSearch(const UnicodeString &pattern,
1.67 + const UnicodeString &text,
1.68 + RuleBasedCollator *coll,
1.69 + BreakIterator *breakiter,
1.70 + UErrorCode &status) :
1.71 + SearchIterator(text, breakiter),
1.72 + m_collator_(),
1.73 + m_pattern_(pattern)
1.74 +{
1.75 + if (U_FAILURE(status)) {
1.76 + m_strsrch_ = NULL;
1.77 + return;
1.78 + }
1.79 + if (coll == NULL) {
1.80 + status = U_ILLEGAL_ARGUMENT_ERROR;
1.81 + m_strsrch_ = NULL;
1.82 + return;
1.83 + }
1.84 + m_strsrch_ = usearch_openFromCollator(m_pattern_.getBuffer(),
1.85 + m_pattern_.length(),
1.86 + m_text_.getBuffer(),
1.87 + m_text_.length(), coll->ucollator,
1.88 + (UBreakIterator *)breakiter,
1.89 + &status);
1.90 + uprv_free(m_search_);
1.91 + m_search_ = NULL;
1.92 +
1.93 + if (U_SUCCESS(status)) {
1.94 + // Alias the collator
1.95 + m_collator_.setUCollator((UCollator *)m_strsrch_->collator);
1.96 + // m_search_ has been created by the base SearchIterator class
1.97 + m_search_ = m_strsrch_->search;
1.98 + }
1.99 +}
1.100 +
1.101 +StringSearch::StringSearch(const UnicodeString &pattern,
1.102 + CharacterIterator &text,
1.103 + const Locale &locale,
1.104 + BreakIterator *breakiter,
1.105 + UErrorCode &status) :
1.106 + SearchIterator(text, breakiter),
1.107 + m_collator_(),
1.108 + m_pattern_(pattern)
1.109 +{
1.110 + if (U_FAILURE(status)) {
1.111 + m_strsrch_ = NULL;
1.112 + return;
1.113 + }
1.114 + m_strsrch_ = usearch_open(m_pattern_.getBuffer(), m_pattern_.length(),
1.115 + m_text_.getBuffer(), m_text_.length(),
1.116 + locale.getName(), (UBreakIterator *)breakiter,
1.117 + &status);
1.118 + uprv_free(m_search_);
1.119 + m_search_ = NULL;
1.120 +
1.121 + if (U_SUCCESS(status)) {
1.122 + // Alias the collator
1.123 + m_collator_.setUCollator((UCollator *)m_strsrch_->collator);
1.124 + // m_search_ has been created by the base SearchIterator class
1.125 + m_search_ = m_strsrch_->search;
1.126 + }
1.127 +}
1.128 +
1.129 +StringSearch::StringSearch(const UnicodeString &pattern,
1.130 + CharacterIterator &text,
1.131 + RuleBasedCollator *coll,
1.132 + BreakIterator *breakiter,
1.133 + UErrorCode &status) :
1.134 + SearchIterator(text, breakiter),
1.135 + m_collator_(),
1.136 + m_pattern_(pattern)
1.137 +{
1.138 + if (U_FAILURE(status)) {
1.139 + m_strsrch_ = NULL;
1.140 + return;
1.141 + }
1.142 + if (coll == NULL) {
1.143 + status = U_ILLEGAL_ARGUMENT_ERROR;
1.144 + m_strsrch_ = NULL;
1.145 + return;
1.146 + }
1.147 + m_strsrch_ = usearch_openFromCollator(m_pattern_.getBuffer(),
1.148 + m_pattern_.length(),
1.149 + m_text_.getBuffer(),
1.150 + m_text_.length(), coll->ucollator,
1.151 + (UBreakIterator *)breakiter,
1.152 + &status);
1.153 + uprv_free(m_search_);
1.154 + m_search_ = NULL;
1.155 +
1.156 + if (U_SUCCESS(status)) {
1.157 + // Alias the collator
1.158 + m_collator_.setUCollator((UCollator *)m_strsrch_->collator);
1.159 + // m_search_ has been created by the base SearchIterator class
1.160 + m_search_ = m_strsrch_->search;
1.161 + }
1.162 +}
1.163 +
1.164 +StringSearch::StringSearch(const StringSearch &that) :
1.165 + SearchIterator(that.m_text_, that.m_breakiterator_),
1.166 + m_collator_(),
1.167 + m_pattern_(that.m_pattern_)
1.168 +{
1.169 + UErrorCode status = U_ZERO_ERROR;
1.170 +
1.171 + // Free m_search_ from the superclass
1.172 + uprv_free(m_search_);
1.173 + m_search_ = NULL;
1.174 +
1.175 + if (that.m_strsrch_ == NULL) {
1.176 + // This was not a good copy
1.177 + m_strsrch_ = NULL;
1.178 + }
1.179 + else {
1.180 + // Make a deep copy
1.181 + m_strsrch_ = usearch_openFromCollator(m_pattern_.getBuffer(),
1.182 + m_pattern_.length(),
1.183 + m_text_.getBuffer(),
1.184 + m_text_.length(),
1.185 + that.m_strsrch_->collator,
1.186 + (UBreakIterator *)that.m_breakiterator_,
1.187 + &status);
1.188 + if (U_SUCCESS(status)) {
1.189 + // Alias the collator
1.190 + m_collator_.setUCollator((UCollator *)m_strsrch_->collator);
1.191 + // m_search_ has been created by the base SearchIterator class
1.192 + m_search_ = m_strsrch_->search;
1.193 + }
1.194 + }
1.195 +}
1.196 +
1.197 +StringSearch::~StringSearch()
1.198 +{
1.199 + if (m_strsrch_ != NULL) {
1.200 + usearch_close(m_strsrch_);
1.201 + m_search_ = NULL;
1.202 + }
1.203 +}
1.204 +
1.205 +StringSearch *
1.206 +StringSearch::clone() const {
1.207 + return new StringSearch(*this);
1.208 +}
1.209 +
1.210 +// operator overloading ---------------------------------------------
1.211 +StringSearch & StringSearch::operator=(const StringSearch &that)
1.212 +{
1.213 + if ((*this) != that) {
1.214 + UErrorCode status = U_ZERO_ERROR;
1.215 + m_text_ = that.m_text_;
1.216 + m_breakiterator_ = that.m_breakiterator_;
1.217 + m_pattern_ = that.m_pattern_;
1.218 + // all m_search_ in the parent class is linked up with m_strsrch_
1.219 + usearch_close(m_strsrch_);
1.220 + m_strsrch_ = usearch_openFromCollator(m_pattern_.getBuffer(),
1.221 + m_pattern_.length(),
1.222 + m_text_.getBuffer(),
1.223 + m_text_.length(),
1.224 + that.m_strsrch_->collator,
1.225 + NULL, &status);
1.226 + // Check null pointer
1.227 + if (m_strsrch_ != NULL) {
1.228 + // Alias the collator
1.229 + m_collator_.setUCollator((UCollator *)m_strsrch_->collator);
1.230 + m_search_ = m_strsrch_->search;
1.231 + }
1.232 + }
1.233 + return *this;
1.234 +}
1.235 +
1.236 +UBool StringSearch::operator==(const SearchIterator &that) const
1.237 +{
1.238 + if (this == &that) {
1.239 + return TRUE;
1.240 + }
1.241 + if (SearchIterator::operator ==(that)) {
1.242 + StringSearch &thatsrch = (StringSearch &)that;
1.243 + return (this->m_pattern_ == thatsrch.m_pattern_ &&
1.244 + this->m_strsrch_->collator == thatsrch.m_strsrch_->collator);
1.245 + }
1.246 + return FALSE;
1.247 +}
1.248 +
1.249 +// public get and set methods ----------------------------------------
1.250 +
1.251 +void StringSearch::setOffset(int32_t position, UErrorCode &status)
1.252 +{
1.253 + // status checked in usearch_setOffset
1.254 + usearch_setOffset(m_strsrch_, position, &status);
1.255 +}
1.256 +
1.257 +int32_t StringSearch::getOffset(void) const
1.258 +{
1.259 + return usearch_getOffset(m_strsrch_);
1.260 +}
1.261 +
1.262 +void StringSearch::setText(const UnicodeString &text, UErrorCode &status)
1.263 +{
1.264 + if (U_SUCCESS(status)) {
1.265 + m_text_ = text;
1.266 + usearch_setText(m_strsrch_, text.getBuffer(), text.length(), &status);
1.267 + }
1.268 +}
1.269 +
1.270 +void StringSearch::setText(CharacterIterator &text, UErrorCode &status)
1.271 +{
1.272 + if (U_SUCCESS(status)) {
1.273 + text.getText(m_text_);
1.274 + usearch_setText(m_strsrch_, m_text_.getBuffer(), m_text_.length(), &status);
1.275 + }
1.276 +}
1.277 +
1.278 +RuleBasedCollator * StringSearch::getCollator() const
1.279 +{
1.280 + return (RuleBasedCollator *)&m_collator_;
1.281 +}
1.282 +
1.283 +void StringSearch::setCollator(RuleBasedCollator *coll, UErrorCode &status)
1.284 +{
1.285 + if (U_SUCCESS(status)) {
1.286 + usearch_setCollator(m_strsrch_, coll->getUCollator(), &status);
1.287 + // Alias the collator
1.288 + m_collator_.setUCollator((UCollator *)m_strsrch_->collator);
1.289 + }
1.290 +}
1.291 +
1.292 +void StringSearch::setPattern(const UnicodeString &pattern,
1.293 + UErrorCode &status)
1.294 +{
1.295 + if (U_SUCCESS(status)) {
1.296 + m_pattern_ = pattern;
1.297 + usearch_setPattern(m_strsrch_, m_pattern_.getBuffer(), m_pattern_.length(),
1.298 + &status);
1.299 + }
1.300 +}
1.301 +
1.302 +const UnicodeString & StringSearch::getPattern() const
1.303 +{
1.304 + return m_pattern_;
1.305 +}
1.306 +
1.307 +// public methods ----------------------------------------------------
1.308 +
1.309 +void StringSearch::reset()
1.310 +{
1.311 + usearch_reset(m_strsrch_);
1.312 +}
1.313 +
1.314 +SearchIterator * StringSearch::safeClone(void) const
1.315 +{
1.316 + UErrorCode status = U_ZERO_ERROR;
1.317 + StringSearch *result = new StringSearch(m_pattern_, m_text_,
1.318 + (RuleBasedCollator *)&m_collator_,
1.319 + m_breakiterator_,
1.320 + status);
1.321 + /* test for NULL */
1.322 + if (result == 0) {
1.323 + status = U_MEMORY_ALLOCATION_ERROR;
1.324 + return 0;
1.325 + }
1.326 + result->setOffset(getOffset(), status);
1.327 + result->setMatchStart(m_strsrch_->search->matchedIndex);
1.328 + result->setMatchLength(m_strsrch_->search->matchedLength);
1.329 + if (U_FAILURE(status)) {
1.330 + return NULL;
1.331 + }
1.332 + return result;
1.333 +}
1.334 +
1.335 +// protected method -------------------------------------------------
1.336 +
1.337 +int32_t StringSearch::handleNext(int32_t position, UErrorCode &status)
1.338 +{
1.339 + // values passed here are already in the pre-shift position
1.340 + if (U_SUCCESS(status)) {
1.341 + if (m_strsrch_->pattern.CELength == 0) {
1.342 + m_search_->matchedIndex =
1.343 + m_search_->matchedIndex == USEARCH_DONE ?
1.344 + getOffset() : m_search_->matchedIndex + 1;
1.345 + m_search_->matchedLength = 0;
1.346 + ucol_setOffset(m_strsrch_->textIter, m_search_->matchedIndex,
1.347 + &status);
1.348 + if (m_search_->matchedIndex == m_search_->textLength) {
1.349 + m_search_->matchedIndex = USEARCH_DONE;
1.350 + }
1.351 + }
1.352 + else {
1.353 + // looking at usearch.cpp, this part is shifted out to
1.354 + // StringSearch instead of SearchIterator because m_strsrch_ is
1.355 + // not accessible in SearchIterator
1.356 +#if 0
1.357 + if (position + m_strsrch_->pattern.defaultShiftSize
1.358 + > m_search_->textLength) {
1.359 + setMatchNotFound();
1.360 + return USEARCH_DONE;
1.361 + }
1.362 +#endif
1.363 + if (m_search_->matchedLength <= 0) {
1.364 + // the flipping direction issue has already been handled
1.365 + // in next()
1.366 + // for boundary check purposes. this will ensure that the
1.367 + // next match will not preceed the current offset
1.368 + // note search->matchedIndex will always be set to something
1.369 + // in the code
1.370 + m_search_->matchedIndex = position - 1;
1.371 + }
1.372 +
1.373 + ucol_setOffset(m_strsrch_->textIter, position, &status);
1.374 +
1.375 +#if 0
1.376 + for (;;) {
1.377 + if (m_search_->isCanonicalMatch) {
1.378 + // can't use exact here since extra accents are allowed.
1.379 + usearch_handleNextCanonical(m_strsrch_, &status);
1.380 + }
1.381 + else {
1.382 + usearch_handleNextExact(m_strsrch_, &status);
1.383 + }
1.384 + if (U_FAILURE(status)) {
1.385 + return USEARCH_DONE;
1.386 + }
1.387 + if (m_breakiterator_ == NULL
1.388 +#if !UCONFIG_NO_BREAK_ITERATION
1.389 + ||
1.390 + m_search_->matchedIndex == USEARCH_DONE ||
1.391 + (m_breakiterator_->isBoundary(m_search_->matchedIndex) &&
1.392 + m_breakiterator_->isBoundary(m_search_->matchedIndex +
1.393 + m_search_->matchedLength))
1.394 +#endif
1.395 + ) {
1.396 + if (m_search_->matchedIndex == USEARCH_DONE) {
1.397 + ucol_setOffset(m_strsrch_->textIter,
1.398 + m_search_->textLength, &status);
1.399 + }
1.400 + else {
1.401 + ucol_setOffset(m_strsrch_->textIter,
1.402 + m_search_->matchedIndex, &status);
1.403 + }
1.404 + return m_search_->matchedIndex;
1.405 + }
1.406 + }
1.407 +#else
1.408 + // if m_strsrch_->breakIter is always the same as m_breakiterator_
1.409 + // then we don't need to check the match boundaries here because
1.410 + // usearch_handleNextXXX will already have done it.
1.411 + if (m_search_->isCanonicalMatch) {
1.412 + // *could* actually use exact here 'cause no extra accents allowed...
1.413 + usearch_handleNextCanonical(m_strsrch_, &status);
1.414 + } else {
1.415 + usearch_handleNextExact(m_strsrch_, &status);
1.416 + }
1.417 +
1.418 + if (U_FAILURE(status)) {
1.419 + return USEARCH_DONE;
1.420 + }
1.421 +
1.422 + if (m_search_->matchedIndex == USEARCH_DONE) {
1.423 + ucol_setOffset(m_strsrch_->textIter, m_search_->textLength, &status);
1.424 + } else {
1.425 + ucol_setOffset(m_strsrch_->textIter, m_search_->matchedIndex, &status);
1.426 + }
1.427 +
1.428 + return m_search_->matchedIndex;
1.429 +#endif
1.430 + }
1.431 + }
1.432 + return USEARCH_DONE;
1.433 +}
1.434 +
1.435 +int32_t StringSearch::handlePrev(int32_t position, UErrorCode &status)
1.436 +{
1.437 + // values passed here are already in the pre-shift position
1.438 + if (U_SUCCESS(status)) {
1.439 + if (m_strsrch_->pattern.CELength == 0) {
1.440 + m_search_->matchedIndex =
1.441 + (m_search_->matchedIndex == USEARCH_DONE ? getOffset() :
1.442 + m_search_->matchedIndex);
1.443 + if (m_search_->matchedIndex == 0) {
1.444 + setMatchNotFound();
1.445 + }
1.446 + else {
1.447 + m_search_->matchedIndex --;
1.448 + ucol_setOffset(m_strsrch_->textIter, m_search_->matchedIndex,
1.449 + &status);
1.450 + m_search_->matchedLength = 0;
1.451 + }
1.452 + }
1.453 + else {
1.454 + // looking at usearch.cpp, this part is shifted out to
1.455 + // StringSearch instead of SearchIterator because m_strsrch_ is
1.456 + // not accessible in SearchIterator
1.457 +#if 0
1.458 + if (!m_search_->isOverlap &&
1.459 + position - m_strsrch_->pattern.defaultShiftSize < 0) {
1.460 + setMatchNotFound();
1.461 + return USEARCH_DONE;
1.462 + }
1.463 +
1.464 + for (;;) {
1.465 + if (m_search_->isCanonicalMatch) {
1.466 + // can't use exact here since extra accents are allowed.
1.467 + usearch_handlePreviousCanonical(m_strsrch_, &status);
1.468 + }
1.469 + else {
1.470 + usearch_handlePreviousExact(m_strsrch_, &status);
1.471 + }
1.472 + if (U_FAILURE(status)) {
1.473 + return USEARCH_DONE;
1.474 + }
1.475 + if (m_breakiterator_ == NULL
1.476 +#if !UCONFIG_NO_BREAK_ITERATION
1.477 + ||
1.478 + m_search_->matchedIndex == USEARCH_DONE ||
1.479 + (m_breakiterator_->isBoundary(m_search_->matchedIndex) &&
1.480 + m_breakiterator_->isBoundary(m_search_->matchedIndex +
1.481 + m_search_->matchedLength))
1.482 +#endif
1.483 + ) {
1.484 + return m_search_->matchedIndex;
1.485 + }
1.486 + }
1.487 +#else
1.488 + ucol_setOffset(m_strsrch_->textIter, position, &status);
1.489 +
1.490 + if (m_search_->isCanonicalMatch) {
1.491 + // *could* use exact match here since extra accents *not* allowed!
1.492 + usearch_handlePreviousCanonical(m_strsrch_, &status);
1.493 + } else {
1.494 + usearch_handlePreviousExact(m_strsrch_, &status);
1.495 + }
1.496 +
1.497 + if (U_FAILURE(status)) {
1.498 + return USEARCH_DONE;
1.499 + }
1.500 +
1.501 + return m_search_->matchedIndex;
1.502 +#endif
1.503 + }
1.504 +
1.505 + return m_search_->matchedIndex;
1.506 + }
1.507 + return USEARCH_DONE;
1.508 +}
1.509 +
1.510 +U_NAMESPACE_END
1.511 +
1.512 +#endif /* #if !UCONFIG_NO_COLLATION */
