bool SmallCapsIterator::consume(unsigned *capsLimit, SmallCapsBehavior* smallCapsBehavior)
{
if (m_atEnd)
return false;
while (m_utf16Iterator->consume(m_nextUChar32)) {
m_previousSmallCapsBehavior = m_currentSmallCapsBehavior;
// Skipping over combining marks, as these combine with the small-caps
// uppercased text as well and we do not need to split by their
// individual case-ness.
if (!u_getCombiningClass(m_nextUChar32)) {
m_currentSmallCapsBehavior = u_hasBinaryProperty(m_nextUChar32, UCHAR_CHANGES_WHEN_UPPERCASED)
? SmallCapsUppercaseNeeded
: SmallCapsSameCase;
}
if (m_previousSmallCapsBehavior != m_currentSmallCapsBehavior
&& m_previousSmallCapsBehavior != SmallCapsInvalid) {
*capsLimit = m_utf16Iterator->offset();
*smallCapsBehavior = m_previousSmallCapsBehavior;
return true;
}
m_utf16Iterator->advance();
}
*capsLimit = m_bufferSize;
*smallCapsBehavior = m_currentSmallCapsBehavior;
m_atEnd = true;
return true;
}
int32_t SpoofImpl::findHiddenOverlay(const UnicodeString& input, UErrorCode&) const {
bool sawLeadCharacter = false;
for (int32_t i=0; i<input.length();) {
UChar32 cp = input.char32At(i);
if (sawLeadCharacter && cp == 0x0307) {
return i;
}
uint8_t combiningClass = u_getCombiningClass(cp);
// Skip over characters except for those with combining class 0 (non-combining characters) or with
// combining class 230 (same class as U+0307)
U_ASSERT(u_getCombiningClass(0x0307) == 230);
if (combiningClass == 0 || combiningClass == 230) {
sawLeadCharacter = isIllegalCombiningDotLeadCharacter(cp);
}
i += U16_LENGTH(cp);
}
return -1;
}
void SimpleFontInstance::getGlyphAdvance(LEGlyphID glyph, LEPoint &advance) const
{
#if 0
if (u_getCombiningClass((UChar32) glyph) == 0) {
advance.fX = xUnitsToPoints(2048);
} else {
advance.fX = 0;
}
#else
advance.fX = xUnitsToPoints(2048);
#endif
advance.fY = 0;
}
void SimpleFontInstance::getGlyphAdvance(LEGlyphID glyph, LEPoint &advance) const
{
#if 0
if (u_getCombiningClass((UChar32) glyph) == 0) {
advance.fX = xUnitsToPoints(2048);
} else {
advance.fX = 0;
}
#else
(void)glyph; // Suppress unused parameter compiler warning.
advance.fX = xUnitsToPoints(2048);
#endif
advance.fY = 0;
}
UChar32 SurrogatePairAwareTextIterator::normalizeVoicingMarks()
{
// According to http://www.unicode.org/Public/UNIDATA/UCD.html#Canonical_Combining_Class_Values
static const uint8_t hiraganaKatakanaVoicingMarksCombiningClass = 8;
if (m_currentCharacter + 1 >= m_endCharacter)
return 0;
if (u_getCombiningClass(m_characters[1]) == hiraganaKatakanaVoicingMarksCombiningClass) {
// Normalize into composed form using 3.2 rules.
UChar normalizedCharacters[2] = { 0, 0 };
UErrorCode uStatus = U_ZERO_ERROR;
int32_t resultLength = unorm_normalize(m_characters, 2, UNORM_NFC, UNORM_UNICODE_3_2, &normalizedCharacters[0], 2, &uStatus);
if (resultLength == 1 && !uStatus)
return normalizedCharacters[0];
}
return 0;
}
U_CAPI int32_t U_EXPORT2
uprv_cnttab_constructTable(CntTable *table, uint32_t mainOffset, UErrorCode *status) {
int32_t i = 0, j = 0;
if(U_FAILURE(*status) || table->size == 0) {
return 0;
}
table->position = 0;
if(table->offsets != NULL) {
uprv_free(table->offsets);
}
table->offsets = (int32_t *)uprv_malloc(table->size*sizeof(int32_t));
if(table->offsets == NULL) {
*status = U_MEMORY_ALLOCATION_ERROR;
return 0;
}
/* See how much memory we need */
for(i = 0; i<table->size; i++) {
table->offsets[i] = table->position+mainOffset;
table->position += table->elements[i]->position;
}
/* Allocate it */
if(table->CEs != NULL) {
uprv_free(table->CEs);
}
table->CEs = (uint32_t *)uprv_malloc(table->position*sizeof(uint32_t));
if(table->CEs == NULL) {
*status = U_MEMORY_ALLOCATION_ERROR;
uprv_free(table->offsets);
table->offsets = NULL;
return 0;
}
uprv_memset(table->CEs, '?', table->position*sizeof(uint32_t));
if(table->codePoints != NULL) {
uprv_free(table->codePoints);
}
table->codePoints = (UChar *)uprv_malloc(table->position*sizeof(UChar));
if(table->codePoints == NULL) {
*status = U_MEMORY_ALLOCATION_ERROR;
uprv_free(table->offsets);
table->offsets = NULL;
uprv_free(table->CEs);
table->CEs = NULL;
return 0;
}
uprv_memset(table->codePoints, '?', table->position*sizeof(UChar));
/* Now stuff the things in*/
UChar *cpPointer = table->codePoints;
uint32_t *CEPointer = table->CEs;
for(i = 0; i<table->size; i++) {
int32_t size = table->elements[i]->position;
uint8_t ccMax = 0, ccMin = 255, cc = 0;
for(j = 1; j<size; j++) {
cc = u_getCombiningClass(table->elements[i]->codePoints[j]);
if(cc>ccMax) {
ccMax = cc;
}
if(cc<ccMin) {
ccMin = cc;
}
*(cpPointer+j) = table->elements[i]->codePoints[j];
}
*cpPointer = ((ccMin==ccMax)?1:0 << 8) | ccMax;
uprv_memcpy(CEPointer, table->elements[i]->CEs, size*sizeof(uint32_t));
for(j = 0; j<size; j++) {
if(isCntTableElement(*(CEPointer+j))) {
*(CEPointer+j) = constructContractCE(getCETag(*(CEPointer+j)), table->offsets[getContractOffset(*(CEPointer+j))]);
}
}
cpPointer += size;
CEPointer += size;
}
// TODO: this one apparently updates the contraction CEs to point to a real address (relative to the
// start of the flat file). However, what is done below is just wrong and it affects building of
// tailorings that have constructions in a bad way. At least, one should enumerate the trie. Also,
// keeping a list of code points that are contractions might be smart, although I'm not sure if it's
// feasible.
uint32_t CE;
for(i = 0; i<=0x10FFFF; i++) {
/*CE = ucmpe32_get(table->mapping, i);*/
CE = utrie_get32(table->mapping, i, NULL);
if(isCntTableElement(CE)) {
CE = constructContractCE(getCETag(CE), table->offsets[getContractOffset(CE)]);
/*ucmpe32_set(table->mapping, i, CE);*/
utrie_set32(table->mapping, i, CE);
}
}
return table->position;
}
/**
* Dumb recursive implementation of permutation.
* TODO: optimize
* @param source the string to find permutations for
* @return the results in a set.
*/
void U_EXPORT2 CanonicalIterator::permute(UnicodeString &source, UBool skipZeros, Hashtable *result, UErrorCode &status) {
if(U_FAILURE(status)) {
return;
}
//if (PROGRESS) printf("Permute: %s\n", UToS(Tr(source)));
int32_t i = 0;
// optimization:
// if zero or one character, just return a set with it
// we check for length < 2 to keep from counting code points all the time
if (source.length() <= 2 && source.countChar32() <= 1) {
UnicodeString *toPut = new UnicodeString(source);
/* test for NULL */
if (toPut == 0) {
status = U_MEMORY_ALLOCATION_ERROR;
return;
}
result->put(source, toPut, status);
return;
}
// otherwise iterate through the string, and recursively permute all the other characters
UChar32 cp;
Hashtable subpermute(status);
if(U_FAILURE(status)) {
return;
}
subpermute.setValueDeleter(uprv_deleteUObject);
for (i = 0; i < source.length(); i += U16_LENGTH(cp)) {
cp = source.char32At(i);
const UHashElement *ne = NULL;
int32_t el = UHASH_FIRST;
UnicodeString subPermuteString = source;
// optimization:
// if the character is canonical combining class zero,
// don't permute it
if (skipZeros && i != 0 && u_getCombiningClass(cp) == 0) {
//System.out.println("Skipping " + Utility.hex(UTF16.valueOf(source, i)));
continue;
}
subpermute.removeAll();
// see what the permutations of the characters before and after this one are
//Hashtable *subpermute = permute(source.substring(0,i) + source.substring(i + UTF16.getCharCount(cp)));
permute(subPermuteString.replace(i, U16_LENGTH(cp), NULL, 0), skipZeros, &subpermute, status);
/* Test for buffer overflows */
if(U_FAILURE(status)) {
return;
}
// The upper replace is destructive. The question is do we have to make a copy, or we don't care about the contents
// of source at this point.
// prefix this character to all of them
ne = subpermute.nextElement(el);
while (ne != NULL) {
UnicodeString *permRes = (UnicodeString *)(ne->value.pointer);
UnicodeString *chStr = new UnicodeString(cp);
//test for NULL
if (chStr == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
return;
}
chStr->append(*permRes); //*((UnicodeString *)(ne->value.pointer));
//if (PROGRESS) printf(" Piece: %s\n", UToS(*chStr));
result->put(*chStr, chStr, status);
ne = subpermute.nextElement(el);
}
}
//return result;
}
static unsigned int hb_icu_get_combining_class (hb_codepoint_t unicode) { return u_getCombiningClass (unicode); }
U_CAPI int32_t U_EXPORT2
u_getIntPropertyValue(UChar32 c, UProperty which) {
UErrorCode errorCode;
if(which<UCHAR_BINARY_START) {
return 0; /* undefined */
} else if(which<UCHAR_BINARY_LIMIT) {
return (int32_t)u_hasBinaryProperty(c, which);
} else if(which<UCHAR_INT_START) {
return 0; /* undefined */
} else if(which<UCHAR_INT_LIMIT) {
switch(which) {
case UCHAR_BIDI_CLASS:
return (int32_t)u_charDirection(c);
case UCHAR_BLOCK:
return (int32_t)ublock_getCode(c);
#if !UCONFIG_NO_NORMALIZATION
case UCHAR_CANONICAL_COMBINING_CLASS:
return u_getCombiningClass(c);
#endif
case UCHAR_DECOMPOSITION_TYPE:
return (int32_t)(u_getUnicodeProperties(c, 2)&UPROPS_DT_MASK);
case UCHAR_EAST_ASIAN_WIDTH:
return (int32_t)(u_getUnicodeProperties(c, 0)&UPROPS_EA_MASK)>>UPROPS_EA_SHIFT;
case UCHAR_GENERAL_CATEGORY:
return (int32_t)u_charType(c);
case UCHAR_JOINING_GROUP:
return ubidi_getJoiningGroup(GET_BIDI_PROPS(), c);
case UCHAR_JOINING_TYPE:
return ubidi_getJoiningType(GET_BIDI_PROPS(), c);
case UCHAR_LINE_BREAK:
return (int32_t)(u_getUnicodeProperties(c, UPROPS_LB_VWORD)&UPROPS_LB_MASK)>>UPROPS_LB_SHIFT;
case UCHAR_NUMERIC_TYPE: {
int32_t ntv=(int32_t)GET_NUMERIC_TYPE_VALUE(u_getUnicodeProperties(c, -1));
return UPROPS_NTV_GET_TYPE(ntv);
}
case UCHAR_SCRIPT:
errorCode=U_ZERO_ERROR;
return (int32_t)uscript_getScript(c, &errorCode);
case UCHAR_HANGUL_SYLLABLE_TYPE: {
/* see comments on gcbToHst[] above */
int32_t gcb=(int32_t)(u_getUnicodeProperties(c, 2)&UPROPS_GCB_MASK)>>UPROPS_GCB_SHIFT;
if(gcb<LENGTHOF(gcbToHst)) {
return gcbToHst[gcb];
} else {
return U_HST_NOT_APPLICABLE;
}
}
#if !UCONFIG_NO_NORMALIZATION
case UCHAR_NFD_QUICK_CHECK:
case UCHAR_NFKD_QUICK_CHECK:
case UCHAR_NFC_QUICK_CHECK:
case UCHAR_NFKC_QUICK_CHECK:
return (int32_t)unorm_getQuickCheck(c, (UNormalizationMode)(which-UCHAR_NFD_QUICK_CHECK+UNORM_NFD));
case UCHAR_LEAD_CANONICAL_COMBINING_CLASS:
return getFCD16(c)>>8;
case UCHAR_TRAIL_CANONICAL_COMBINING_CLASS:
return getFCD16(c)&0xff;
#endif
case UCHAR_GRAPHEME_CLUSTER_BREAK:
return (int32_t)(u_getUnicodeProperties(c, 2)&UPROPS_GCB_MASK)>>UPROPS_GCB_SHIFT;
case UCHAR_SENTENCE_BREAK:
return (int32_t)(u_getUnicodeProperties(c, 2)&UPROPS_SB_MASK)>>UPROPS_SB_SHIFT;
case UCHAR_WORD_BREAK:
return (int32_t)(u_getUnicodeProperties(c, 2)&UPROPS_WB_MASK)>>UPROPS_WB_SHIFT;
default:
return 0; /* undefined */
}
} else if(which==UCHAR_GENERAL_CATEGORY_MASK) {
uint8_t __hs_u_getCombiningClass(UChar32 c)
{
return u_getCombiningClass(c);
}
