godot/thirdparty/icu4c/common/ubrk.cpp

362 lines
8.7 KiB
C++

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
********************************************************************************
* Copyright (C) 1996-2015, International Business Machines
* Corporation and others. All Rights Reserved.
********************************************************************************
*/
#include "unicode/utypes.h"
#if !UCONFIG_NO_BREAK_ITERATION
#include "unicode/ubrk.h"
#include "unicode/brkiter.h"
#include "unicode/uloc.h"
#include "unicode/ustring.h"
#include "unicode/uchriter.h"
#include "unicode/rbbi.h"
#include "rbbirb.h"
#include "uassert.h"
#include "cmemory.h"
U_NAMESPACE_USE
//------------------------------------------------------------------------------
//
// ubrk_open Create a canned type of break iterator based on type (word, line, etc.)
// and locale.
//
//------------------------------------------------------------------------------
U_CAPI UBreakIterator* U_EXPORT2
ubrk_open(UBreakIteratorType type,
const char *locale,
const UChar *text,
int32_t textLength,
UErrorCode *status)
{
if(U_FAILURE(*status)) return 0;
BreakIterator *result = 0;
switch(type) {
case UBRK_CHARACTER:
result = BreakIterator::createCharacterInstance(Locale(locale), *status);
break;
case UBRK_WORD:
result = BreakIterator::createWordInstance(Locale(locale), *status);
break;
case UBRK_LINE:
result = BreakIterator::createLineInstance(Locale(locale), *status);
break;
case UBRK_SENTENCE:
result = BreakIterator::createSentenceInstance(Locale(locale), *status);
break;
case UBRK_TITLE:
result = BreakIterator::createTitleInstance(Locale(locale), *status);
break;
default:
*status = U_ILLEGAL_ARGUMENT_ERROR;
}
// check for allocation error
if (U_FAILURE(*status)) {
return 0;
}
if(result == 0) {
*status = U_MEMORY_ALLOCATION_ERROR;
return 0;
}
UBreakIterator *uBI = (UBreakIterator *)result;
if (text != NULL) {
ubrk_setText(uBI, text, textLength, status);
}
return uBI;
}
//------------------------------------------------------------------------------
//
// ubrk_openRules open a break iterator from a set of break rules.
// Invokes the rule builder.
//
//------------------------------------------------------------------------------
U_CAPI UBreakIterator* U_EXPORT2
ubrk_openRules( const UChar *rules,
int32_t rulesLength,
const UChar *text,
int32_t textLength,
UParseError *parseErr,
UErrorCode *status) {
if (status == NULL || U_FAILURE(*status)){
return 0;
}
BreakIterator *result = 0;
UnicodeString ruleString(rules, rulesLength);
result = RBBIRuleBuilder::createRuleBasedBreakIterator(ruleString, parseErr, *status);
if(U_FAILURE(*status)) {
return 0;
}
UBreakIterator *uBI = (UBreakIterator *)result;
if (text != NULL) {
ubrk_setText(uBI, text, textLength, status);
}
return uBI;
}
U_CAPI UBreakIterator* U_EXPORT2
ubrk_openBinaryRules(const uint8_t *binaryRules, int32_t rulesLength,
const UChar * text, int32_t textLength,
UErrorCode * status)
{
if (U_FAILURE(*status)) {
return NULL;
}
if (rulesLength < 0) {
*status = U_ILLEGAL_ARGUMENT_ERROR;
return NULL;
}
LocalPointer<RuleBasedBreakIterator> lpRBBI(new RuleBasedBreakIterator(binaryRules, rulesLength, *status), *status);
if (U_FAILURE(*status)) {
return NULL;
}
UBreakIterator *uBI = reinterpret_cast<UBreakIterator *>(lpRBBI.orphan());
if (text != NULL) {
ubrk_setText(uBI, text, textLength, status);
}
return uBI;
}
U_CAPI UBreakIterator * U_EXPORT2
ubrk_safeClone(
const UBreakIterator *bi,
void * /*stackBuffer*/,
int32_t *pBufferSize,
UErrorCode *status)
{
if (status == NULL || U_FAILURE(*status)){
return NULL;
}
if (bi == NULL) {
*status = U_ILLEGAL_ARGUMENT_ERROR;
return NULL;
}
if (pBufferSize != NULL) {
int32_t inputSize = *pBufferSize;
*pBufferSize = 1;
if (inputSize == 0) {
return NULL; // preflighting for deprecated functionality
}
}
BreakIterator *newBI = ((BreakIterator *)bi)->clone();
if (newBI == NULL) {
*status = U_MEMORY_ALLOCATION_ERROR;
} else if (pBufferSize != NULL) {
*status = U_SAFECLONE_ALLOCATED_WARNING;
}
return (UBreakIterator *)newBI;
}
U_CAPI UBreakIterator * U_EXPORT2
ubrk_clone(const UBreakIterator *bi, UErrorCode *status) {
return ubrk_safeClone(bi, nullptr, nullptr, status);
}
U_CAPI void U_EXPORT2
ubrk_close(UBreakIterator *bi)
{
delete (BreakIterator *)bi;
}
U_CAPI void U_EXPORT2
ubrk_setText(UBreakIterator* bi,
const UChar* text,
int32_t textLength,
UErrorCode* status)
{
UText ut = UTEXT_INITIALIZER;
utext_openUChars(&ut, text, textLength, status);
((BreakIterator*)bi)->setText(&ut, *status);
// A stack allocated UText wrapping a UChar * string
// can be dumped without explicitly closing it.
}
U_CAPI void U_EXPORT2
ubrk_setUText(UBreakIterator *bi,
UText *text,
UErrorCode *status)
{
((BreakIterator*)bi)->setText(text, *status);
}
U_CAPI int32_t U_EXPORT2
ubrk_current(const UBreakIterator *bi)
{
return ((BreakIterator*)bi)->current();
}
U_CAPI int32_t U_EXPORT2
ubrk_next(UBreakIterator *bi)
{
return ((BreakIterator*)bi)->next();
}
U_CAPI int32_t U_EXPORT2
ubrk_previous(UBreakIterator *bi)
{
return ((BreakIterator*)bi)->previous();
}
U_CAPI int32_t U_EXPORT2
ubrk_first(UBreakIterator *bi)
{
return ((BreakIterator*)bi)->first();
}
U_CAPI int32_t U_EXPORT2
ubrk_last(UBreakIterator *bi)
{
return ((BreakIterator*)bi)->last();
}
U_CAPI int32_t U_EXPORT2
ubrk_preceding(UBreakIterator *bi,
int32_t offset)
{
return ((BreakIterator*)bi)->preceding(offset);
}
U_CAPI int32_t U_EXPORT2
ubrk_following(UBreakIterator *bi,
int32_t offset)
{
return ((BreakIterator*)bi)->following(offset);
}
U_CAPI const char* U_EXPORT2
ubrk_getAvailable(int32_t index)
{
return uloc_getAvailable(index);
}
U_CAPI int32_t U_EXPORT2
ubrk_countAvailable()
{
return uloc_countAvailable();
}
U_CAPI UBool U_EXPORT2
ubrk_isBoundary(UBreakIterator *bi, int32_t offset)
{
return ((BreakIterator*)bi)->isBoundary(offset);
}
U_CAPI int32_t U_EXPORT2
ubrk_getRuleStatus(UBreakIterator *bi)
{
return ((BreakIterator*)bi)->getRuleStatus();
}
U_CAPI int32_t U_EXPORT2
ubrk_getRuleStatusVec(UBreakIterator *bi, int32_t *fillInVec, int32_t capacity, UErrorCode *status)
{
return ((BreakIterator*)bi)->getRuleStatusVec(fillInVec, capacity, *status);
}
U_CAPI const char* U_EXPORT2
ubrk_getLocaleByType(const UBreakIterator *bi,
ULocDataLocaleType type,
UErrorCode* status)
{
if (bi == NULL) {
if (U_SUCCESS(*status)) {
*status = U_ILLEGAL_ARGUMENT_ERROR;
}
return NULL;
}
return ((BreakIterator*)bi)->getLocaleID(type, *status);
}
U_CAPI void U_EXPORT2
ubrk_refreshUText(UBreakIterator *bi,
UText *text,
UErrorCode *status)
{
BreakIterator *bii = reinterpret_cast<BreakIterator *>(bi);
bii->refreshInputText(text, *status);
}
U_CAPI int32_t U_EXPORT2
ubrk_getBinaryRules(UBreakIterator *bi,
uint8_t * binaryRules, int32_t rulesCapacity,
UErrorCode * status)
{
if (U_FAILURE(*status)) {
return 0;
}
if ((binaryRules == NULL && rulesCapacity > 0) || rulesCapacity < 0) {
*status = U_ILLEGAL_ARGUMENT_ERROR;
return 0;
}
RuleBasedBreakIterator* rbbi;
if ((rbbi = dynamic_cast<RuleBasedBreakIterator*>(reinterpret_cast<BreakIterator*>(bi))) == NULL) {
*status = U_ILLEGAL_ARGUMENT_ERROR;
return 0;
}
uint32_t rulesLength;
const uint8_t * returnedRules = rbbi->getBinaryRules(rulesLength);
if (rulesLength > INT32_MAX) {
*status = U_INDEX_OUTOFBOUNDS_ERROR;
return 0;
}
if (binaryRules != NULL) { // if not preflighting
// Here we know rulesLength <= INT32_MAX and rulesCapacity >= 0, can cast safely
if ((int32_t)rulesLength > rulesCapacity) {
*status = U_BUFFER_OVERFLOW_ERROR;
} else {
uprv_memcpy(binaryRules, returnedRules, rulesLength);
}
}
return (int32_t)rulesLength;
}
#endif /* #if !UCONFIG_NO_BREAK_ITERATION */