+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
/*
**********************************************************************
-* Copyright (C) 2001-2006 IBM and others. All rights reserved.
+* Copyright (C) 2001-2014 IBM and others. All rights reserved.
**********************************************************************
* Date Name Description
* 03/22/2000 helena Creation.
#include "unicode/utypes.h"
-#if !UCONFIG_NO_COLLATION
+#if !UCONFIG_NO_COLLATION && !UCONFIG_NO_BREAK_ITERATION
#include "unicode/stsearch.h"
#include "usrchimp.h"
BreakIterator *breakiter,
UErrorCode &status) :
SearchIterator(text, breakiter),
- m_collator_(),
m_pattern_(pattern)
{
if (U_FAILURE(status)) {
uprv_free(m_search_);
m_search_ = NULL;
- // !!! dlf m_collator_ is an odd beast. basically it is an aliasing
- // wrapper around the internal collator and rules, which (here) are
- // owned by this stringsearch object. this means 1) it's destructor
- // _should not_ delete the ucollator or rules, and 2) changes made
- // to the exposed collator (setStrength etc) _should_ modify the
- // ucollator. thus the collator is not a copy-on-write alias, and it
- // needs to distinguish itself not merely from 'stand alone' colators
- // but also from copy-on-write ones. it needs additional state, which
- // setUCollator should set.
-
if (U_SUCCESS(status)) {
- // Alias the collator
- m_collator_.setUCollator((UCollator *)m_strsrch_->collator);
// m_search_ has been created by the base SearchIterator class
m_search_ = m_strsrch_->search;
}
BreakIterator *breakiter,
UErrorCode &status) :
SearchIterator(text, breakiter),
- m_collator_(),
m_pattern_(pattern)
{
if (U_FAILURE(status)) {
m_strsrch_ = usearch_openFromCollator(m_pattern_.getBuffer(),
m_pattern_.length(),
m_text_.getBuffer(),
- m_text_.length(), coll->ucollator,
+ m_text_.length(), coll->toUCollator(),
(UBreakIterator *)breakiter,
&status);
uprv_free(m_search_);
m_search_ = NULL;
if (U_SUCCESS(status)) {
- // Alias the collator
- m_collator_.setUCollator((UCollator *)m_strsrch_->collator);
// m_search_ has been created by the base SearchIterator class
m_search_ = m_strsrch_->search;
}
BreakIterator *breakiter,
UErrorCode &status) :
SearchIterator(text, breakiter),
- m_collator_(),
m_pattern_(pattern)
{
if (U_FAILURE(status)) {
m_search_ = NULL;
if (U_SUCCESS(status)) {
- // Alias the collator
- m_collator_.setUCollator((UCollator *)m_strsrch_->collator);
// m_search_ has been created by the base SearchIterator class
m_search_ = m_strsrch_->search;
}
BreakIterator *breakiter,
UErrorCode &status) :
SearchIterator(text, breakiter),
- m_collator_(),
m_pattern_(pattern)
{
if (U_FAILURE(status)) {
m_strsrch_ = usearch_openFromCollator(m_pattern_.getBuffer(),
m_pattern_.length(),
m_text_.getBuffer(),
- m_text_.length(), coll->ucollator,
+ m_text_.length(), coll->toUCollator(),
(UBreakIterator *)breakiter,
&status);
uprv_free(m_search_);
m_search_ = NULL;
if (U_SUCCESS(status)) {
- // Alias the collator
- m_collator_.setUCollator((UCollator *)m_strsrch_->collator);
// m_search_ has been created by the base SearchIterator class
m_search_ = m_strsrch_->search;
}
StringSearch::StringSearch(const StringSearch &that) :
SearchIterator(that.m_text_, that.m_breakiterator_),
- m_collator_(),
m_pattern_(that.m_pattern_)
{
UErrorCode status = U_ZERO_ERROR;
(UBreakIterator *)that.m_breakiterator_,
&status);
if (U_SUCCESS(status)) {
- // Alias the collator
- m_collator_.setUCollator((UCollator *)m_strsrch_->collator);
// m_search_ has been created by the base SearchIterator class
m_search_ = m_strsrch_->search;
}
m_text_.length(),
that.m_strsrch_->collator,
NULL, &status);
- // Alias the collator
- m_collator_.setUCollator((UCollator *)m_strsrch_->collator);
- m_search_ = m_strsrch_->search;
+ // Check null pointer
+ if (m_strsrch_ != NULL) {
+ m_search_ = m_strsrch_->search;
+ }
}
return *this;
}
RuleBasedCollator * StringSearch::getCollator() const
{
- return (RuleBasedCollator *)&m_collator_;
+ // Note the const_cast. It would be cleaner if this const method returned a const collator.
+ return RuleBasedCollator::rbcFromUCollator(const_cast<UCollator *>(m_strsrch_->collator));
}
void StringSearch::setCollator(RuleBasedCollator *coll, UErrorCode &status)
{
if (U_SUCCESS(status)) {
- usearch_setCollator(m_strsrch_, coll->getUCollator(), &status);
- // Alias the collator
- m_collator_.setUCollator((UCollator *)m_strsrch_->collator);
+ usearch_setCollator(m_strsrch_, coll->toUCollator(), &status);
}
}
{
UErrorCode status = U_ZERO_ERROR;
StringSearch *result = new StringSearch(m_pattern_, m_text_,
- (RuleBasedCollator *)&m_collator_,
+ getCollator(),
m_breakiterator_,
status);
/* test for NULL */
{
// values passed here are already in the pre-shift position
if (U_SUCCESS(status)) {
- if (m_strsrch_->pattern.CELength == 0) {
+ if (m_strsrch_->pattern.cesLength == 0) {
m_search_->matchedIndex =
m_search_->matchedIndex == USEARCH_DONE ?
getOffset() : m_search_->matchedIndex + 1;
// looking at usearch.cpp, this part is shifted out to
// StringSearch instead of SearchIterator because m_strsrch_ is
// not accessible in SearchIterator
+#if 0
if (position + m_strsrch_->pattern.defaultShiftSize
> m_search_->textLength) {
setMatchNotFound();
return USEARCH_DONE;
}
+#endif
if (m_search_->matchedLength <= 0) {
// the flipping direction issue has already been handled
// in next()
}
ucol_setOffset(m_strsrch_->textIter, position, &status);
- while (TRUE) {
+
+#if 0
+ for (;;) {
if (m_search_->isCanonicalMatch) {
// can't use exact here since extra accents are allowed.
usearch_handleNextCanonical(m_strsrch_, &status);
return m_search_->matchedIndex;
}
}
+#else
+ // if m_strsrch_->breakIter is always the same as m_breakiterator_
+ // then we don't need to check the match boundaries here because
+ // usearch_handleNextXXX will already have done it.
+ if (m_search_->isCanonicalMatch) {
+ // *could* actually use exact here 'cause no extra accents allowed...
+ usearch_handleNextCanonical(m_strsrch_, &status);
+ } else {
+ usearch_handleNextExact(m_strsrch_, &status);
+ }
+
+ if (U_FAILURE(status)) {
+ return USEARCH_DONE;
+ }
+
+ if (m_search_->matchedIndex == USEARCH_DONE) {
+ ucol_setOffset(m_strsrch_->textIter, m_search_->textLength, &status);
+ } else {
+ ucol_setOffset(m_strsrch_->textIter, m_search_->matchedIndex, &status);
+ }
+
+ return m_search_->matchedIndex;
+#endif
}
}
return USEARCH_DONE;
{
// values passed here are already in the pre-shift position
if (U_SUCCESS(status)) {
- if (m_strsrch_->pattern.CELength == 0) {
+ if (m_strsrch_->pattern.cesLength == 0) {
m_search_->matchedIndex =
(m_search_->matchedIndex == USEARCH_DONE ? getOffset() :
m_search_->matchedIndex);
// looking at usearch.cpp, this part is shifted out to
// StringSearch instead of SearchIterator because m_strsrch_ is
// not accessible in SearchIterator
+#if 0
if (!m_search_->isOverlap &&
position - m_strsrch_->pattern.defaultShiftSize < 0) {
setMatchNotFound();
return USEARCH_DONE;
}
- while (TRUE) {
+
+ for (;;) {
if (m_search_->isCanonicalMatch) {
// can't use exact here since extra accents are allowed.
usearch_handlePreviousCanonical(m_strsrch_, &status);
return m_search_->matchedIndex;
}
}
+#else
+ ucol_setOffset(m_strsrch_->textIter, position, &status);
+
+ if (m_search_->isCanonicalMatch) {
+ // *could* use exact match here since extra accents *not* allowed!
+ usearch_handlePreviousCanonical(m_strsrch_, &status);
+ } else {
+ usearch_handlePreviousExact(m_strsrch_, &status);
+ }
+
+ if (U_FAILURE(status)) {
+ return USEARCH_DONE;
+ }
+
+ return m_search_->matchedIndex;
+#endif
}
return m_search_->matchedIndex;
projects / apple / icu.git / blobdiff