[apple/icu.git] / icuSources / common / ustr_titlecase_brkiter.cpp

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
*******************************************************************************
*   Copyright (C) 2011, International Business Machines
*   Corporation and others.  All Rights Reserved.
*******************************************************************************
*   file name:  ustr_titlecase_brkiter.cpp
*   encoding:   UTF-8
*   tab size:   8 (not used)
*   indentation:4
*
*   created on: 2011may30
*   created by: Markus W. Scherer
*
*   Titlecasing functions that are based on BreakIterator
*   were moved here to break dependency cycles among parts of the common library.
*/

#include "unicode/utypes.h"

#if !UCONFIG_NO_BREAK_ITERATION

#include "unicode/brkiter.h"
#include "unicode/casemap.h"
#include "unicode/chariter.h"
#include "unicode/localpointer.h"
#include "unicode/ubrk.h"
#include "unicode/ucasemap.h"
#include "unicode/utext.h"
#include "cmemory.h"
#include "uassert.h"
#include "ucase.h"
#include "ucasemap_imp.h"

U_NAMESPACE_BEGIN

/**
 * Whole-string BreakIterator.
 * Titlecasing only calls setText(), first(), and next().
 * We implement the rest only to satisfy the abstract interface.
 */
class WholeStringBreakIterator : public BreakIterator {
public:
    WholeStringBreakIterator() : BreakIterator(), length(0) {}
    ~WholeStringBreakIterator() U_OVERRIDE;
    UBool operator==(const BreakIterator&) const U_OVERRIDE;
    WholeStringBreakIterator *clone() const U_OVERRIDE;
    static UClassID U_EXPORT2 getStaticClassID();
    UClassID getDynamicClassID() const U_OVERRIDE;
    CharacterIterator &getText() const U_OVERRIDE;
    UText *getUText(UText *fillIn, UErrorCode &errorCode) const U_OVERRIDE;
    void  setText(const UnicodeString &text) U_OVERRIDE;
    void  setText(UText *text, UErrorCode &errorCode) U_OVERRIDE;
    void  adoptText(CharacterIterator* it) U_OVERRIDE;
    int32_t first() U_OVERRIDE;
    int32_t last() U_OVERRIDE;
    int32_t previous() U_OVERRIDE;
    int32_t next() U_OVERRIDE;
    int32_t current() const U_OVERRIDE;
    int32_t following(int32_t offset) U_OVERRIDE;
    int32_t preceding(int32_t offset) U_OVERRIDE;
    UBool isBoundary(int32_t offset) U_OVERRIDE;
    int32_t next(int32_t n) U_OVERRIDE;
    WholeStringBreakIterator *createBufferClone(void *stackBuffer, int32_t &BufferSize,
                                                UErrorCode &errorCode) U_OVERRIDE;
    WholeStringBreakIterator &refreshInputText(UText *input, UErrorCode &errorCode) U_OVERRIDE;

private:
    int32_t length;
};

UOBJECT_DEFINE_RTTI_IMPLEMENTATION(WholeStringBreakIterator)

WholeStringBreakIterator::~WholeStringBreakIterator() {}
UBool WholeStringBreakIterator::operator==(const BreakIterator&) const { return FALSE; }
WholeStringBreakIterator *WholeStringBreakIterator::clone() const { return nullptr; }

CharacterIterator &WholeStringBreakIterator::getText() const {
    UPRV_UNREACHABLE;  // really should not be called
}
UText *WholeStringBreakIterator::getUText(UText * /*fillIn*/, UErrorCode &errorCode) const {
    if (U_SUCCESS(errorCode)) {
        errorCode = U_UNSUPPORTED_ERROR;
    }
    return nullptr;
}

void  WholeStringBreakIterator::setText(const UnicodeString &text) {
    length = text.length();
}
void  WholeStringBreakIterator::setText(UText *text, UErrorCode &errorCode) {
    if (U_SUCCESS(errorCode)) {
        int64_t length64 = utext_nativeLength(text);
        if (length64 <= INT32_MAX) {
            length = (int32_t)length64;
        } else {
            errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
        }
    }
}
void  WholeStringBreakIterator::adoptText(CharacterIterator*) {
    UPRV_UNREACHABLE;  // should not be called
}

int32_t WholeStringBreakIterator::first() { return 0; }
int32_t WholeStringBreakIterator::last() { return length; }
int32_t WholeStringBreakIterator::previous() { return 0; }
int32_t WholeStringBreakIterator::next() { return length; }
int32_t WholeStringBreakIterator::current() const { return 0; }
int32_t WholeStringBreakIterator::following(int32_t /*offset*/) { return length; }
int32_t WholeStringBreakIterator::preceding(int32_t /*offset*/) { return 0; }
UBool WholeStringBreakIterator::isBoundary(int32_t /*offset*/) { return FALSE; }
int32_t WholeStringBreakIterator::next(int32_t /*n*/) { return length; }

WholeStringBreakIterator *WholeStringBreakIterator::createBufferClone(
        void * /*stackBuffer*/, int32_t & /*BufferSize*/, UErrorCode &errorCode) {
    if (U_SUCCESS(errorCode)) {
        errorCode = U_UNSUPPORTED_ERROR;
    }
    return nullptr;
}
WholeStringBreakIterator &WholeStringBreakIterator::refreshInputText(
        UText * /*input*/, UErrorCode &errorCode) {
    if (U_SUCCESS(errorCode)) {
        errorCode = U_UNSUPPORTED_ERROR;
    }
    return *this;
}

U_CFUNC
BreakIterator *ustrcase_getTitleBreakIterator(
        const Locale *locale, const char *locID, uint32_t options, BreakIterator *iter,
        LocalPointer<BreakIterator> &ownedIter, UErrorCode &errorCode) {
    if (U_FAILURE(errorCode)) { return nullptr; }
    options &= U_TITLECASE_ITERATOR_MASK;
    if (options != 0 && iter != nullptr) {
        errorCode = U_ILLEGAL_ARGUMENT_ERROR;
        return nullptr;
    }
    if (iter == nullptr) {
        switch (options) {
        case 0:
            iter = BreakIterator::createWordInstance(
                locale != nullptr ? *locale : Locale(locID), errorCode);
            break;
        case U_TITLECASE_WHOLE_STRING:
            iter = new WholeStringBreakIterator();
            if (iter == nullptr) {
                errorCode = U_MEMORY_ALLOCATION_ERROR;
            }
            break;
        case U_TITLECASE_SENTENCES:
            iter = BreakIterator::createSentenceInstance(
                locale != nullptr ? *locale : Locale(locID), errorCode);
            break;
        default:
            errorCode = U_ILLEGAL_ARGUMENT_ERROR;
            break;
        }
        ownedIter.adoptInstead(iter);
    }
    return iter;
}

int32_t CaseMap::toTitle(
        const char *locale, uint32_t options, BreakIterator *iter,
        const UChar *src, int32_t srcLength,
        UChar *dest, int32_t destCapacity, Edits *edits,
        UErrorCode &errorCode) {
    LocalPointer<BreakIterator> ownedIter;
    iter = ustrcase_getTitleBreakIterator(nullptr, locale, options, iter, ownedIter, errorCode);
    if(iter==NULL) {
        return 0;
    }
    UnicodeString s(srcLength<0, src, srcLength);
    iter->setText(s);
    return ustrcase_map(
        ustrcase_getCaseLocale(locale), options, iter,
        dest, destCapacity,
        src, srcLength,
        ustrcase_internalToTitle, edits, errorCode);
}

U_NAMESPACE_END

U_NAMESPACE_USE

U_CAPI int32_t U_EXPORT2
u_strToTitle(UChar *dest, int32_t destCapacity,
             const UChar *src, int32_t srcLength,
             UBreakIterator *titleIter,
             const char *locale,
             UErrorCode *pErrorCode) {
    LocalPointer<BreakIterator> ownedIter;
    BreakIterator *iter = ustrcase_getTitleBreakIterator(
        nullptr, locale, 0, reinterpret_cast<BreakIterator *>(titleIter),
        ownedIter, *pErrorCode);
    if (iter == nullptr) {
        return 0;
    }
    UnicodeString s(srcLength<0, src, srcLength);
    iter->setText(s);
    return ustrcase_mapWithOverlap(
        ustrcase_getCaseLocale(locale), 0, iter,
        dest, destCapacity,
        src, srcLength,
        ustrcase_internalToTitle, *pErrorCode);
}

U_CAPI int32_t U_EXPORT2
ucasemap_toTitle(UCaseMap *csm,
                 UChar *dest, int32_t destCapacity,
                 const UChar *src, int32_t srcLength,
                 UErrorCode *pErrorCode) {
    if (U_FAILURE(*pErrorCode)) {
        return 0;
    }
    if (csm->iter == NULL) {
        LocalPointer<BreakIterator> ownedIter;
        BreakIterator *iter = ustrcase_getTitleBreakIterator(
            nullptr, csm->locale, csm->options, nullptr, ownedIter, *pErrorCode);
        if (iter == nullptr) {
            return 0;
        }
        csm->iter = ownedIter.orphan();
    }
    UnicodeString s(srcLength<0, src, srcLength);
    csm->iter->setText(s);
    return ustrcase_map(
        csm->caseLocale, csm->options, csm->iter,
        dest, destCapacity,
        src, srcLength,
        ustrcase_internalToTitle, NULL, *pErrorCode);
}

#endif  // !UCONFIG_NO_BREAK_ITERATION
Commit	Line	Data
f3c0d7a5 A	1	// © 2016 and later: Unicode, Inc. and others.
f3c0d7a5 A	2	// License & terms of use: http://www.unicode.org/copyright.html
4388f060 A	3	/*
	4	*******************************************************************************
	5	* Copyright (C) 2011, International Business Machines
	6	* Corporation and others. All Rights Reserved.
	7	*******************************************************************************
	8	* file name: ustr_titlecase_brkiter.cpp
f3c0d7a5	9	* encoding: UTF-8
4388f060 A	10	* tab size: 8 (not used)
	11	* indentation:4
	12	*
	13	* created on: 2011may30
	14	* created by: Markus W. Scherer
	15	*
	16	* Titlecasing functions that are based on BreakIterator
	17	* were moved here to break dependency cycles among parts of the common library.
	18	*/
	19
	20	#include "unicode/utypes.h"
	21
	22	#if !UCONFIG_NO_BREAK_ITERATION
	23
	24	#include "unicode/brkiter.h"
f3c0d7a5	25	#include "unicode/casemap.h"
0f5d89e8	26	#include "unicode/chariter.h"
f3c0d7a5	27	#include "unicode/localpointer.h"
4388f060 A	28	#include "unicode/ubrk.h"
4388f060 A	29	#include "unicode/ucasemap.h"
0f5d89e8	30	#include "unicode/utext.h"
4388f060	31	#include "cmemory.h"
0f5d89e8	32	#include "uassert.h"
4388f060	33	#include "ucase.h"
f3c0d7a5	34	#include "ucasemap_imp.h"
4388f060	35
0f5d89e8	36	U_NAMESPACE_BEGIN
4388f060	37
0f5d89e8 A	38	/**
	39	* Whole-string BreakIterator.
	40	* Titlecasing only calls setText(), first(), and next().
	41	* We implement the rest only to satisfy the abstract interface.
	42	*/
	43	class WholeStringBreakIterator : public BreakIterator {
	44	public:
	45	WholeStringBreakIterator() : BreakIterator(), length(0) {}
	46	~WholeStringBreakIterator() U_OVERRIDE;
	47	UBool operator==(const BreakIterator&) const U_OVERRIDE;
340931cb	48	WholeStringBreakIterator *clone() const U_OVERRIDE;
0f5d89e8 A	49	static UClassID U_EXPORT2 getStaticClassID();
	50	UClassID getDynamicClassID() const U_OVERRIDE;
	51	CharacterIterator &getText() const U_OVERRIDE;
	52	UText getUText(UText fillIn, UErrorCode &errorCode) const U_OVERRIDE;
	53	void setText(const UnicodeString &text) U_OVERRIDE;
	54	void setText(UText *text, UErrorCode &errorCode) U_OVERRIDE;
	55	void adoptText(CharacterIterator* it) U_OVERRIDE;
	56	int32_t first() U_OVERRIDE;
	57	int32_t last() U_OVERRIDE;
	58	int32_t previous() U_OVERRIDE;
	59	int32_t next() U_OVERRIDE;
	60	int32_t current() const U_OVERRIDE;
	61	int32_t following(int32_t offset) U_OVERRIDE;
	62	int32_t preceding(int32_t offset) U_OVERRIDE;
	63	UBool isBoundary(int32_t offset) U_OVERRIDE;
	64	int32_t next(int32_t n) U_OVERRIDE;
340931cb A	65	WholeStringBreakIterator createBufferClone(void stackBuffer, int32_t &BufferSize,
	66	UErrorCode &errorCode) U_OVERRIDE;
	67	WholeStringBreakIterator &refreshInputText(UText *input, UErrorCode &errorCode) U_OVERRIDE;
4388f060	68
0f5d89e8 A	69	private:
	70	int32_t length;
	71	};
4388f060	72
0f5d89e8 A	73	UOBJECT_DEFINE_RTTI_IMPLEMENTATION(WholeStringBreakIterator)
	74
	75	WholeStringBreakIterator::~WholeStringBreakIterator() {}
	76	UBool WholeStringBreakIterator::operator==(const BreakIterator&) const { return FALSE; }
340931cb	77	WholeStringBreakIterator *WholeStringBreakIterator::clone() const { return nullptr; }
0f5d89e8 A	78
0f5d89e8 A	79	CharacterIterator &WholeStringBreakIterator::getText() const {
3d1f044b	80	UPRV_UNREACHABLE; // really should not be called
0f5d89e8 A	81	}
	82	UText WholeStringBreakIterator::getUText(UText /fillIn/, UErrorCode &errorCode) const {
	83	if (U_SUCCESS(errorCode)) {
	84	errorCode = U_UNSUPPORTED_ERROR;
f3c0d7a5	85	}
0f5d89e8 A	86	return nullptr;
	87	}
	88
	89	void WholeStringBreakIterator::setText(const UnicodeString &text) {
	90	length = text.length();
	91	}
	92	void WholeStringBreakIterator::setText(UText *text, UErrorCode &errorCode) {
	93	if (U_SUCCESS(errorCode)) {
	94	int64_t length64 = utext_nativeLength(text);
	95	if (length64 <= INT32_MAX) {
	96	length = (int32_t)length64;
	97	} else {
	98	errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
	99	}
4388f060	100	}
0f5d89e8	101	}
3d1f044b A	102	void WholeStringBreakIterator::adoptText(CharacterIterator*) {
3d1f044b A	103	UPRV_UNREACHABLE; // should not be called
f3c0d7a5 A	104	}
f3c0d7a5 A	105
0f5d89e8 A	106	int32_t WholeStringBreakIterator::first() { return 0; }
	107	int32_t WholeStringBreakIterator::last() { return length; }
	108	int32_t WholeStringBreakIterator::previous() { return 0; }
	109	int32_t WholeStringBreakIterator::next() { return length; }
	110	int32_t WholeStringBreakIterator::current() const { return 0; }
	111	int32_t WholeStringBreakIterator::following(int32_t /offset/) { return length; }
	112	int32_t WholeStringBreakIterator::preceding(int32_t /offset/) { return 0; }
	113	UBool WholeStringBreakIterator::isBoundary(int32_t /offset/) { return FALSE; }
	114	int32_t WholeStringBreakIterator::next(int32_t /n/) { return length; }
	115
340931cb	116	WholeStringBreakIterator *WholeStringBreakIterator::createBufferClone(
0f5d89e8 A	117	void * /stackBuffer/, int32_t & /BufferSize/, UErrorCode &errorCode) {
	118	if (U_SUCCESS(errorCode)) {
	119	errorCode = U_UNSUPPORTED_ERROR;
	120	}
	121	return nullptr;
	122	}
340931cb	123	WholeStringBreakIterator &WholeStringBreakIterator::refreshInputText(
0f5d89e8 A	124	UText * /input/, UErrorCode &errorCode) {
	125	if (U_SUCCESS(errorCode)) {
	126	errorCode = U_UNSUPPORTED_ERROR;
	127	}
	128	return *this;
	129	}
	130
	131	U_CFUNC
	132	BreakIterator *ustrcase_getTitleBreakIterator(
	133	const Locale locale, const char locID, uint32_t options, BreakIterator *iter,
	134	LocalPointer<BreakIterator> &ownedIter, UErrorCode &errorCode) {
	135	if (U_FAILURE(errorCode)) { return nullptr; }
	136	options &= U_TITLECASE_ITERATOR_MASK;
	137	if (options != 0 && iter != nullptr) {
	138	errorCode = U_ILLEGAL_ARGUMENT_ERROR;
	139	return nullptr;
	140	}
	141	if (iter == nullptr) {
	142	switch (options) {
	143	case 0:
	144	iter = BreakIterator::createWordInstance(
	145	locale != nullptr ? *locale : Locale(locID), errorCode);
	146	break;
	147	case U_TITLECASE_WHOLE_STRING:
	148	iter = new WholeStringBreakIterator();
	149	if (iter == nullptr) {
	150	errorCode = U_MEMORY_ALLOCATION_ERROR;
	151	}
	152	break;
	153	case U_TITLECASE_SENTENCES:
	154	iter = BreakIterator::createSentenceInstance(
	155	locale != nullptr ? *locale : Locale(locID), errorCode);
	156	break;
	157	default:
	158	errorCode = U_ILLEGAL_ARGUMENT_ERROR;
	159	break;
	160	}
	161	ownedIter.adoptInstead(iter);
	162	}
	163	return iter;
	164	}
f3c0d7a5 A	165
	166	int32_t CaseMap::toTitle(
	167	const char locale, uint32_t options, BreakIterator iter,
	168	const UChar *src, int32_t srcLength,
	169	UChar dest, int32_t destCapacity, Edits edits,
	170	UErrorCode &errorCode) {
	171	LocalPointer<BreakIterator> ownedIter;
0f5d89e8	172	iter = ustrcase_getTitleBreakIterator(nullptr, locale, options, iter, ownedIter, errorCode);
f3c0d7a5	173	if(iter==NULL) {
f3c0d7a5	174	return 0;
4388f060	175	}
f3c0d7a5 A	176	UnicodeString s(srcLength<0, src, srcLength);
	177	iter->setText(s);
	178	return ustrcase_map(
	179	ustrcase_getCaseLocale(locale), options, iter,
	180	dest, destCapacity,
	181	src, srcLength,
	182	ustrcase_internalToTitle, edits, errorCode);
4388f060 A	183	}
4388f060 A	184
f3c0d7a5 A	185	U_NAMESPACE_END
f3c0d7a5 A	186
0f5d89e8 A	187	U_NAMESPACE_USE
	188
	189	U_CAPI int32_t U_EXPORT2
	190	u_strToTitle(UChar *dest, int32_t destCapacity,
	191	const UChar *src, int32_t srcLength,
	192	UBreakIterator *titleIter,
	193	const char *locale,
	194	UErrorCode *pErrorCode) {
	195	LocalPointer<BreakIterator> ownedIter;
	196	BreakIterator *iter = ustrcase_getTitleBreakIterator(
	197	nullptr, locale, 0, reinterpret_cast<BreakIterator *>(titleIter),
	198	ownedIter, *pErrorCode);
	199	if (iter == nullptr) {
	200	return 0;
	201	}
	202	UnicodeString s(srcLength<0, src, srcLength);
	203	iter->setText(s);
	204	return ustrcase_mapWithOverlap(
	205	ustrcase_getCaseLocale(locale), 0, iter,
	206	dest, destCapacity,
	207	src, srcLength,
	208	ustrcase_internalToTitle, *pErrorCode);
	209	}
	210
4388f060 A	211	U_CAPI int32_t U_EXPORT2
	212	ucasemap_toTitle(UCaseMap *csm,
	213	UChar *dest, int32_t destCapacity,
	214	const UChar *src, int32_t srcLength,
	215	UErrorCode *pErrorCode) {
f3c0d7a5 A	216	if (U_FAILURE(*pErrorCode)) {
	217	return 0;
	218	}
	219	if (csm->iter == NULL) {
0f5d89e8 A	220	LocalPointer<BreakIterator> ownedIter;
	221	BreakIterator *iter = ustrcase_getTitleBreakIterator(
	222	nullptr, csm->locale, csm->options, nullptr, ownedIter, *pErrorCode);
	223	if (iter == nullptr) {
	224	return 0;
	225	}
	226	csm->iter = ownedIter.orphan();
4388f060	227	}
f3c0d7a5 A	228	UnicodeString s(srcLength<0, src, srcLength);
f3c0d7a5 A	229	csm->iter->setText(s);
4388f060	230	return ustrcase_map(
f3c0d7a5	231	csm->caseLocale, csm->options, csm->iter,
4388f060 A	232	dest, destCapacity,
4388f060 A	233	src, srcLength,
f3c0d7a5	234	ustrcase_internalToTitle, NULL, *pErrorCode);
4388f060 A	235	}
	236
	237	#endif // !UCONFIG_NO_BREAK_ITERATION