UChar32 CaseFoldingUCharIterator::next() {
UChar32 foldedC;
UChar32 originalC;
if (fFoldChars == NULL) {
// We are not in a string folding of an earlier character.
// Start handling the next char from the input UText.
if (fIndex >= fLimit) {
return U_SENTINEL;
}
U16_NEXT(fChars, fIndex, fLimit, originalC);
fFoldLength = ucase_toFullFolding(fcsp, originalC, &fFoldChars, U_FOLD_CASE_DEFAULT);
if (fFoldLength >= UCASE_MAX_STRING_LENGTH || fFoldLength < 0) {
// input code point folds to a single code point, possibly itself.
// See comment in ucase.h for explanation of return values from ucase_toFullFoldings.
if (fFoldLength < 0) {
fFoldLength = ~fFoldLength;
}
foldedC = (UChar32)fFoldLength;
fFoldChars = NULL;
return foldedC;
}
// String foldings fall through here.
fFoldIndex = 0;
}
U16_NEXT(fFoldChars, fFoldIndex, fFoldLength, foldedC);
if (fFoldIndex >= fFoldLength) {
fFoldChars = NULL;
}
return foldedC;
}
U_CFUNC int32_t
ustr_foldCase(const UCaseProps *csp,
UChar *dest, int32_t destCapacity,
const UChar *src, int32_t srcLength,
uint32_t options,
UErrorCode *pErrorCode) {
int32_t srcIndex, destIndex;
const UChar *s;
UChar32 c, c2 = 0;
/* case mapping loop */
srcIndex=destIndex=0;
while(srcIndex<srcLength) {
U16_NEXT(src, srcIndex, srcLength, c);
c=ucase_toFullFolding(csp, c, &s, options);
if((destIndex<destCapacity) && (c<0 ? (c2=~c)<=0xffff : UCASE_MAX_STRING_LENGTH<c && (c2=c)<=0xffff)) {
/* fast path version of appendResult() for BMP results */
dest[destIndex++]=(UChar)c2;
} else {
destIndex=appendResult(dest, destIndex, destCapacity, c, s);
}
}
if(destIndex>destCapacity) {
*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
}
return destIndex;
}
static void U_CALLCONV
ucasemap_internalUTF8Fold(int32_t /* caseLocale */, uint32_t options, UCASEMAP_BREAK_ITERATOR_UNUSED
const uint8_t *src, int32_t srcLength,
icu::ByteSink &sink, icu::Edits *edits,
UErrorCode &errorCode) {
/* case mapping loop */
int32_t srcIndex = 0;
while (U_SUCCESS(errorCode) && srcIndex < srcLength) {
int32_t cpStart = srcIndex;
UChar32 c;
U8_NEXT(src, srcIndex, srcLength, c);
if(c<0) {
// Malformed UTF-8.
ByteSinkUtil::appendUnchanged(src+cpStart, srcIndex-cpStart,
sink, options, edits, errorCode);
} else {
const UChar *s;
c = ucase_toFullFolding(c, &s, options);
appendResult(srcIndex - cpStart, c, s, sink, options, edits, errorCode);
}
}
}
static UBool changesWhenCasefolded(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
UnicodeString nfd;
UErrorCode errorCode=U_ZERO_ERROR;
const Normalizer2 *nfcNorm2=Normalizer2Factory::getNFCInstance(errorCode);
if(U_FAILURE(errorCode)) {
return FALSE;
}
if(nfcNorm2->getDecomposition(c, nfd)) {
/* c has a decomposition */
if(nfd.length()==1) {
c=nfd[0]; /* single BMP code point */
} else if(nfd.length()<=U16_MAX_LENGTH &&
nfd.length()==U16_LENGTH(c=nfd.char32At(0))
) {
/* single supplementary code point */
} else {
c=U_SENTINEL;
}
} else if(c<0) {
return FALSE; /* protect against bad input */
}
if(c>=0) {
/* single code point */
const UCaseProps *csp=ucase_getSingleton();
const UChar *resultString;
return (UBool)(ucase_toFullFolding(csp, c, &resultString, U_FOLD_CASE_DEFAULT)>=0);
} else {
/* guess some large but stack-friendly capacity */
UChar dest[2*UCASE_MAX_STRING_LENGTH];
int32_t destLength;
destLength=u_strFoldCase(dest, LENGTHOF(dest),
nfd.getBuffer(), nfd.length(),
U_FOLD_CASE_DEFAULT, &errorCode);
return (UBool)(U_SUCCESS(errorCode) &&
0!=u_strCompare(nfd.getBuffer(), nfd.length(),
dest, destLength, FALSE));
}
}
static int32_t
utf8_foldCase(const UCaseProps *csp,
uint8_t *dest, int32_t destCapacity,
const uint8_t *src, int32_t srcLength,
uint32_t options,
UErrorCode *pErrorCode) {
int32_t srcIndex, destIndex;
const UChar *s;
UChar32 c, c2;
int32_t start;
/* case mapping loop */
srcIndex=destIndex=0;
while(srcIndex<srcLength) {
start=srcIndex;
U8_NEXT(src, srcIndex, srcLength, c);
if(c<0) {
while(destIndex<destCapacity && start<srcIndex) {
dest[destIndex++]=src[start++];
}
continue;
}
c=ucase_toFullFolding(csp, c, &s, options);
if((destIndex<destCapacity) && (c<0 ? (c2=~c)<=0x7f : UCASE_MAX_STRING_LENGTH<c && (c2=c)<=0x7f)) {
/* fast path version of appendResult() for ASCII results */
dest[destIndex++]=(uint8_t)c2;
} else {
destIndex=appendResult(dest, destIndex, destCapacity, c, s);
}
}
if(destIndex>destCapacity) {
*pErrorCode=U_BUFFER_OVERFLOW_ERROR;
}
return destIndex;
}
/* internal function */
U_CFUNC int32_t
u_strcmpFold(const UChar *s1, int32_t length1,
const UChar *s2, int32_t length2,
uint32_t options,
UErrorCode *pErrorCode) {
const UCaseProps *csp;
/* current-level start/limit - s1/s2 as current */
const UChar *start1, *start2, *limit1, *limit2;
/* case folding variables */
const UChar *p;
int32_t length;
/* stacks of previous-level start/current/limit */
CmpEquivLevel stack1[2], stack2[2];
/* case folding buffers, only use current-level start/limit */
UChar fold1[UCASE_MAX_STRING_LENGTH+1], fold2[UCASE_MAX_STRING_LENGTH+1];
/* track which is the current level per string */
int32_t level1, level2;
/* current code units, and code points for lookups */
UChar32 c1, c2, cp1, cp2;
/* no argument error checking because this itself is not an API */
/*
* assume that at least the option U_COMPARE_IGNORE_CASE is set
* otherwise this function would have to behave exactly as uprv_strCompare()
*/
csp=ucase_getSingleton();
if(U_FAILURE(*pErrorCode)) {
return 0;
}
/* initialize */
start1=s1;
if(length1==-1) {
limit1=NULL;
} else {
limit1=s1+length1;
}
start2=s2;
if(length2==-1) {
limit2=NULL;
} else {
limit2=s2+length2;
}
level1=level2=0;
c1=c2=-1;
/* comparison loop */
for(;;) {
/*
* here a code unit value of -1 means "get another code unit"
* below it will mean "this source is finished"
*/
if(c1<0) {
/* get next code unit from string 1, post-increment */
for(;;) {
if(s1==limit1 || ((c1=*s1)==0 && (limit1==NULL || (options&_STRNCMP_STYLE)))) {
if(level1==0) {
c1=-1;
break;
}
} else {
++s1;
break;
}
/* reached end of level buffer, pop one level */
do {
--level1;
start1=stack1[level1].start;
} while(start1==NULL);
s1=stack1[level1].s;
limit1=stack1[level1].limit;
}
}
if(c2<0) {
/* get next code unit from string 2, post-increment */
for(;;) {
if(s2==limit2 || ((c2=*s2)==0 && (limit2==NULL || (options&_STRNCMP_STYLE)))) {
if(level2==0) {
c2=-1;
break;
}
} else {
++s2;
break;
}
/* reached end of level buffer, pop one level */
do {
--level2;
start2=stack2[level2].start;
} while(start2==NULL);
s2=stack2[level2].s;
limit2=stack2[level2].limit;
}
}
/*
* compare c1 and c2
* either variable c1, c2 is -1 only if the corresponding string is finished
*/
if(c1==c2) {
if(c1<0) {
return 0; /* c1==c2==-1 indicating end of strings */
}
c1=c2=-1; /* make us fetch new code units */
continue;
} else if(c1<0) {
return -1; /* string 1 ends before string 2 */
} else if(c2<0) {
return 1; /* string 2 ends before string 1 */
}
/* c1!=c2 && c1>=0 && c2>=0 */
/* get complete code points for c1, c2 for lookups if either is a surrogate */
cp1=c1;
if(U_IS_SURROGATE(c1)) {
UChar c;
if(U_IS_SURROGATE_LEAD(c1)) {
if(s1!=limit1 && U16_IS_TRAIL(c=*s1)) {
/* advance ++s1; only below if cp1 decomposes/case-folds */
cp1=U16_GET_SUPPLEMENTARY(c1, c);
}
} else /* isTrail(c1) */ {
if(start1<=(s1-2) && U16_IS_LEAD(c=*(s1-2))) {
cp1=U16_GET_SUPPLEMENTARY(c, c1);
}
}
}
cp2=c2;
if(U_IS_SURROGATE(c2)) {
UChar c;
if(U_IS_SURROGATE_LEAD(c2)) {
if(s2!=limit2 && U16_IS_TRAIL(c=*s2)) {
/* advance ++s2; only below if cp2 decomposes/case-folds */
cp2=U16_GET_SUPPLEMENTARY(c2, c);
}
} else /* isTrail(c2) */ {
if(start2<=(s2-2) && U16_IS_LEAD(c=*(s2-2))) {
cp2=U16_GET_SUPPLEMENTARY(c, c2);
}
}
}
/*
* go down one level for each string
* continue with the main loop as soon as there is a real change
*/
if( level1==0 &&
(length=ucase_toFullFolding(csp, (UChar32)cp1, &p, options))>=0
) {
/* cp1 case-folds to the code point "length" or to p[length] */
if(U_IS_SURROGATE(c1)) {
if(U_IS_SURROGATE_LEAD(c1)) {
/* advance beyond source surrogate pair if it case-folds */
++s1;
} else /* isTrail(c1) */ {
/*
* we got a supplementary code point when hitting its trail surrogate,
* therefore the lead surrogate must have been the same as in the other string;
* compare this decomposition with the lead surrogate in the other string
* remember that this simulates bulk text replacement:
* the decomposition would replace the entire code point
*/
--s2;
c2=*(s2-1);
}
}
/* push current level pointers */
stack1[0].start=start1;
stack1[0].s=s1;
stack1[0].limit=limit1;
++level1;
/* copy the folding result to fold1[] */
if(length<=UCASE_MAX_STRING_LENGTH) {
u_memcpy(fold1, p, length);
} else {
int32_t i=0;
U16_APPEND_UNSAFE(fold1, i, length);
length=i;
}
/* set next level pointers to case folding */
start1=s1=fold1;
limit1=fold1+length;
/* get ready to read from decomposition, continue with loop */
c1=-1;
continue;
}
if( level2==0 &&
(length=ucase_toFullFolding(csp, (UChar32)cp2, &p, options))>=0
) {
/* cp2 case-folds to the code point "length" or to p[length] */
if(U_IS_SURROGATE(c2)) {
if(U_IS_SURROGATE_LEAD(c2)) {
/* advance beyond source surrogate pair if it case-folds */
++s2;
} else /* isTrail(c2) */ {
/*
* we got a supplementary code point when hitting its trail surrogate,
* therefore the lead surrogate must have been the same as in the other string;
* compare this decomposition with the lead surrogate in the other string
* remember that this simulates bulk text replacement:
* the decomposition would replace the entire code point
*/
--s1;
c1=*(s1-1);
}
}
/* push current level pointers */
stack2[0].start=start2;
stack2[0].s=s2;
stack2[0].limit=limit2;
++level2;
/* copy the folding result to fold2[] */
if(length<=UCASE_MAX_STRING_LENGTH) {
u_memcpy(fold2, p, length);
} else {
int32_t i=0;
U16_APPEND_UNSAFE(fold2, i, length);
length=i;
}
/* set next level pointers to case folding */
start2=s2=fold2;
limit2=fold2+length;
/* get ready to read from decomposition, continue with loop */
c2=-1;
continue;
}
/*
* no decomposition/case folding, max level for both sides:
* return difference result
*
* code point order comparison must not just return cp1-cp2
* because when single surrogates are present then the surrogate pairs
* that formed cp1 and cp2 may be from different string indexes
*
* example: { d800 d800 dc01 } vs. { d800 dc00 }, compare at second code units
* c1=d800 cp1=10001 c2=dc00 cp2=10000
* cp1-cp2>0 but c1-c2<0 and in fact in UTF-32 it is { d800 10001 } < { 10000 }
*
* therefore, use same fix-up as in ustring.c/uprv_strCompare()
* except: uprv_strCompare() fetches c=*s while this functions fetches c=*s++
* so we have slightly different pointer/start/limit comparisons here
*/
if(c1>=0xd800 && c2>=0xd800 && (options&U_COMPARE_CODE_POINT_ORDER)) {
/* subtract 0x2800 from BMP code points to make them smaller than supplementary ones */
if(
(c1<=0xdbff && s1!=limit1 && U16_IS_TRAIL(*s1)) ||
(U16_IS_TRAIL(c1) && start1!=(s1-1) && U16_IS_LEAD(*(s1-2)))
) {
/* part of a surrogate pair, leave >=d800 */
} else {
/* BMP code point - may be surrogate code point - make <d800 */
c1-=0x2800;
}
if(
(c2<=0xdbff && s2!=limit2 && U16_IS_TRAIL(*s2)) ||
(U16_IS_TRAIL(c2) && start2!=(s2-1) && U16_IS_LEAD(*(s2-2)))
) {
/* part of a surrogate pair, leave >=d800 */
} else {
/* BMP code point - may be surrogate code point - make <d800 */
c2-=0x2800;
}
}
return c1-c2;
}
}
UnicodeSet& UnicodeSet::closeOver(int32_t attribute) {
if (isFrozen() || isBogus()) {
return *this;
}
if (attribute & (USET_CASE_INSENSITIVE | USET_ADD_CASE_MAPPINGS)) {
const UCaseProps *csp = ucase_getSingleton();
{
UnicodeSet foldSet(*this);
UnicodeString str;
USetAdder sa = {
foldSet.toUSet(),
_set_add,
_set_addRange,
_set_addString,
NULL, // don't need remove()
NULL // don't need removeRange()
};
// start with input set to guarantee inclusion
// USET_CASE: remove strings because the strings will actually be reduced (folded);
// therefore, start with no strings and add only those needed
if (attribute & USET_CASE_INSENSITIVE) {
foldSet.strings->removeAllElements();
}
int32_t n = getRangeCount();
UChar32 result;
const UChar *full;
int32_t locCache = 0;
for (int32_t i=0; i<n; ++i) {
UChar32 start = getRangeStart(i);
UChar32 end = getRangeEnd(i);
if (attribute & USET_CASE_INSENSITIVE) {
// full case closure
for (UChar32 cp=start; cp<=end; ++cp) {
ucase_addCaseClosure(csp, cp, &sa);
}
} else {
// add case mappings
// (does not add long s for regular s, or Kelvin for k, for example)
for (UChar32 cp=start; cp<=end; ++cp) {
result = ucase_toFullLower(csp, cp, NULL, NULL, &full, "", &locCache);
addCaseMapping(foldSet, result, full, str);
result = ucase_toFullTitle(csp, cp, NULL, NULL, &full, "", &locCache);
addCaseMapping(foldSet, result, full, str);
result = ucase_toFullUpper(csp, cp, NULL, NULL, &full, "", &locCache);
addCaseMapping(foldSet, result, full, str);
result = ucase_toFullFolding(csp, cp, &full, 0);
addCaseMapping(foldSet, result, full, str);
}
}
}
if (strings != NULL && strings->size() > 0) {
if (attribute & USET_CASE_INSENSITIVE) {
for (int32_t j=0; j<strings->size(); ++j) {
str = *(const UnicodeString *) strings->elementAt(j);
str.foldCase();
if(!ucase_addStringCaseClosure(csp, str.getBuffer(), str.length(), &sa)) {
foldSet.add(str); // does not map to code points: add the folded string itself
}
}
} else {
Locale root("");
#if !UCONFIG_NO_BREAK_ITERATION
UErrorCode status = U_ZERO_ERROR;
BreakIterator *bi = BreakIterator::createWordInstance(root, status);
if (U_SUCCESS(status)) {
#endif
const UnicodeString *pStr;
for (int32_t j=0; j<strings->size(); ++j) {
pStr = (const UnicodeString *) strings->elementAt(j);
(str = *pStr).toLower(root);
foldSet.add(str);
#if !UCONFIG_NO_BREAK_ITERATION
(str = *pStr).toTitle(bi, root);
foldSet.add(str);
#endif
(str = *pStr).toUpper(root);
foldSet.add(str);
(str = *pStr).foldCase();
foldSet.add(str);
}
#if !UCONFIG_NO_BREAK_ITERATION
}
delete bi;
#endif
}
}
*this = foldSet;
}
}
return *this;
}
U_CAPI UBool U_EXPORT2
u_hasBinaryProperty(UChar32 c, UProperty which) {
/* c is range-checked in the functions that are called from here */
if(which<UCHAR_BINARY_START || UCHAR_BINARY_LIMIT<=which) {
/* not a known binary property */
} else {
uint32_t mask=binProps[which].mask;
int32_t column=binProps[which].column;
if(mask!=0) {
/* systematic, directly stored properties */
return (u_getUnicodeProperties(c, column)&mask)!=0;
} else {
if(column==UPROPS_SRC_CASE) {
return ucase_hasBinaryProperty(c, which);
} else if(column==UPROPS_SRC_NORM) {
#if !UCONFIG_NO_NORMALIZATION
/* normalization properties from unorm.icu */
switch(which) {
case UCHAR_SEGMENT_STARTER:
return unorm_isCanonSafeStart(c);
default:
break;
}
#endif
} else if(column==UPROPS_SRC_NFC) {
#if !UCONFIG_NO_NORMALIZATION
UErrorCode errorCode=U_ZERO_ERROR;
switch(which) {
case UCHAR_FULL_COMPOSITION_EXCLUSION: {
// By definition, Full_Composition_Exclusion is the same as NFC_QC=No.
const Normalizer2Impl *impl=Normalizer2Factory::getNFCImpl(errorCode);
return U_SUCCESS(errorCode) && impl->isCompNo(impl->getNorm16(c));
break;
}
default: {
// UCHAR_NF[CD]_INERT properties
const Normalizer2 *norm2=Normalizer2Factory::getInstance(
(UNormalizationMode)(which-UCHAR_NFD_INERT+UNORM_NFD), errorCode);
return U_SUCCESS(errorCode) && norm2->isInert(c);
}
}
#endif
} else if(column==UPROPS_SRC_NFKC) {
#if !UCONFIG_NO_NORMALIZATION
// UCHAR_NFK[CD]_INERT properties
UErrorCode errorCode=U_ZERO_ERROR;
const Normalizer2 *norm2=Normalizer2Factory::getInstance(
(UNormalizationMode)(which-UCHAR_NFD_INERT+UNORM_NFD), errorCode);
return U_SUCCESS(errorCode) && norm2->isInert(c);
#endif
} else if(column==UPROPS_SRC_NFKC_CF) {
// currently only for UCHAR_CHANGES_WHEN_NFKC_CASEFOLDED
#if !UCONFIG_NO_NORMALIZATION
UErrorCode errorCode=U_ZERO_ERROR;
const Normalizer2Impl *kcf=Normalizer2Factory::getNFKC_CFImpl(errorCode);
if(U_SUCCESS(errorCode)) {
UnicodeString src(c);
UnicodeString dest;
{
// The ReorderingBuffer must be in a block because its destructor
// needs to release dest's buffer before we look at its contents.
ReorderingBuffer buffer(*kcf, dest);
// Small destCapacity for NFKC_CF(c).
if(buffer.init(5, errorCode)) {
const UChar *srcArray=src.getBuffer();
kcf->compose(srcArray, srcArray+src.length(), FALSE,
TRUE, buffer, errorCode);
}
}
return U_SUCCESS(errorCode) && dest!=src;
}
#endif
} else if(column==UPROPS_SRC_BIDI) {
/* bidi/shaping properties */
const UBiDiProps *bdp=GET_BIDI_PROPS();
if(bdp!=NULL) {
switch(which) {
case UCHAR_BIDI_MIRRORED:
return ubidi_isMirrored(bdp, c);
case UCHAR_BIDI_CONTROL:
return ubidi_isBidiControl(bdp, c);
case UCHAR_JOIN_CONTROL:
return ubidi_isJoinControl(bdp, c);
default:
break;
}
}
/* else return FALSE below */
} else if(column==UPROPS_SRC_CHAR) {
switch(which) {
case UCHAR_POSIX_BLANK:
return u_isblank(c);
case UCHAR_POSIX_GRAPH:
return u_isgraphPOSIX(c);
case UCHAR_POSIX_PRINT:
return u_isprintPOSIX(c);
case UCHAR_POSIX_XDIGIT:
return u_isxdigit(c);
default:
break;
}
} else if(column==UPROPS_SRC_CHAR_AND_PROPSVEC) {
switch(which) {
case UCHAR_POSIX_ALNUM:
return u_isalnumPOSIX(c);
default:
break;
}
} else if(column==UPROPS_SRC_CASE_AND_NORM) {
#if !UCONFIG_NO_NORMALIZATION
UChar nfdBuffer[4];
const UChar *nfd;
int32_t nfdLength;
UErrorCode errorCode=U_ZERO_ERROR;
const Normalizer2Impl *nfcImpl=Normalizer2Factory::getNFCImpl(errorCode);
if(U_FAILURE(errorCode)) {
return FALSE;
}
switch(which) {
case UCHAR_CHANGES_WHEN_CASEFOLDED:
nfd=nfcImpl->getDecomposition(c, nfdBuffer, nfdLength);
if(nfd!=NULL) {
/* c has a decomposition */
if(nfdLength==1) {
c=nfd[0]; /* single BMP code point */
} else if(nfdLength<=U16_MAX_LENGTH) {
int32_t i=0;
U16_NEXT(nfd, i, nfdLength, c);
if(i==nfdLength) {
/* single supplementary code point */
} else {
c=U_SENTINEL;
}
} else {
c=U_SENTINEL;
}
} else if(c<0) {
return FALSE; /* protect against bad input */
}
errorCode=U_ZERO_ERROR;
if(c>=0) {
/* single code point */
const UCaseProps *csp=ucase_getSingleton(&errorCode);
const UChar *resultString;
return (UBool)(ucase_toFullFolding(csp, c, &resultString, U_FOLD_CASE_DEFAULT)>=0);
} else {
/* guess some large but stack-friendly capacity */
UChar dest[2*UCASE_MAX_STRING_LENGTH];
int32_t destLength;
destLength=u_strFoldCase(dest, LENGTHOF(dest), nfd, nfdLength, U_FOLD_CASE_DEFAULT, &errorCode);
return (UBool)(U_SUCCESS(errorCode) && 0!=u_strCompare(nfd, nfdLength, dest, destLength, FALSE));
}
default:
break;
}
#endif
}
}
}
return FALSE;
}
