U_CAPI void U_CALLCONV
upvec_compactToUTrie2Handler(void *context,
UChar32 start, UChar32 end,
int32_t rowIndex, uint32_t *row, int32_t columns,
UErrorCode *pErrorCode) {
UPVecToUTrie2Context *toUTrie2=(UPVecToUTrie2Context *)context;
if(start<UPVEC_FIRST_SPECIAL_CP) {
utrie2_setRange32(toUTrie2->trie, start, end, (uint32_t)rowIndex, TRUE, pErrorCode);
} else {
switch(start) {
case UPVEC_INITIAL_VALUE_CP:
toUTrie2->initialValue=rowIndex;
break;
case UPVEC_ERROR_VALUE_CP:
toUTrie2->errorValue=rowIndex;
break;
case UPVEC_START_REAL_VALUES_CP:
toUTrie2->maxValue=rowIndex;
if(rowIndex>0xffff) {
/* too many rows for a 16-bit trie */
*pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
} else {
toUTrie2->trie=utrie2_open(toUTrie2->initialValue,
toUTrie2->errorValue, pErrorCode);
}
break;
default:
break;
}
}
}
void
BiDiPropsBuilder::setProps(const UniProps &props, const UnicodeSet &newValues,
UErrorCode &errorCode) {
if(U_FAILURE(errorCode) || newValues.containsNone(relevantProps)) { return; }
UChar32 start=props.start;
UChar32 end=props.end;
// The runtime code relies on this invariant for returning both bmg and bpb
// from the same data.
int32_t bpt=props.getIntProp(UCHAR_BIDI_PAIRED_BRACKET_TYPE);
if(!(bpt==0 ? props.bpb==U_SENTINEL : props.bpb==props.bmg)) {
fprintf(stderr,
"genprops error: invariant not true: "
"if(bpt==None) then bpb=<none> else bpb=bmg\n");
return;
}
int32_t delta=encodeBidiMirroringGlyph(start, end, props.bmg, errorCode);
uint32_t value=(uint32_t)delta<<UBIDI_MIRROR_DELTA_SHIFT;
if(props.binProps[UCHAR_BIDI_MIRRORED]) {
value|=U_MASK(UBIDI_IS_MIRRORED_SHIFT);
}
if(props.binProps[UCHAR_BIDI_CONTROL]) {
value|=U_MASK(UBIDI_BIDI_CONTROL_SHIFT);
}
if(props.binProps[UCHAR_JOIN_CONTROL]) {
value|=U_MASK(UBIDI_JOIN_CONTROL_SHIFT);
}
value|=(uint32_t)bpt<<UBIDI_BPT_SHIFT;
value|=(uint32_t)props.getIntProp(UCHAR_JOINING_TYPE)<<UBIDI_JT_SHIFT;
value|=(uint32_t)props.getIntProp(UCHAR_BIDI_CLASS);
utrie2_setRange32(pTrie, start, end, value, TRUE, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(stderr, "genprops error: BiDiPropsBuilder utrie2_setRange32() failed - %s\n",
u_errorName(errorCode));
return;
}
// Store Joining_Group values from vector column 1 in simple byte arrays.
int32_t jg=props.getIntProp(UCHAR_JOINING_GROUP);
for(UChar32 c=start; c<=end; ++c) {
int32_t jgStart;
if(MIN_JG_START<=c && c<MAX_JG_LIMIT) {
jgArray[c-MIN_JG_START]=(uint8_t)jg;
} else if(MIN_JG_START2<=c && c<MAX_JG_LIMIT2) {
jgArray2[c-MIN_JG_START2]=(uint8_t)jg;
} else if(jg!=U_JG_NO_JOINING_GROUP) {
fprintf(stderr, "genprops error: Joining_Group for out-of-range code points U+%04lx..U+%04lx\n",
(long)start, (long)end);
errorCode=U_ILLEGAL_ARGUMENT_ERROR;
return;
}
}
}
//
// Build the Trie table for mapping UChar32 values to the corresponding
// range group number.
//
void RBBISetBuilder::buildTrie() {
RangeDescriptor *rlRange;
fTrie = utrie2_open(0, // Initial value for all code points.
0, // Error value for out-of-range input.
fStatus);
for (rlRange = fRangeList; rlRange!=0 && U_SUCCESS(*fStatus); rlRange=rlRange->fNext) {
utrie2_setRange32(fTrie,
rlRange->fStartChar, // Range start
rlRange->fEndChar, // Range end (inclusive)
rlRange->fNum, // value for range
TRUE, // Overwrite previously written values
fStatus);
}
}
static UBool U_CALLCONV
copyEnumRange(const void *context, UChar32 start, UChar32 end, uint32_t value) {
NewTrieAndStatus *nt=(NewTrieAndStatus *)context;
if(value!=nt->trie->initialValue) {
if(nt->exclusiveLimit) {
--end;
}
if(start==end) {
utrie2_set32(nt->trie, start, value, &nt->errorCode);
} else {
utrie2_setRange32(nt->trie, start, end, value, TRUE, &nt->errorCode);
}
return U_SUCCESS(nt->errorCode);
} else {
return TRUE;
}
}
int main(int argc, char** argv)
{
// Create a value array of all possible code points.
const UChar32 size = kMaxCodepoint + 1;
CharacterProperty* values = new CharacterProperty[size];
memset(values, 0, sizeof(CharacterProperty) * size);
setRanges(values,
cjkIdeographRanges, ARRAY_LENGTH(cjkIdeographRanges),
CharacterProperty::isCJKIdeographOrSymbol);
setRanges(values,
cjkSymbolRanges, ARRAY_LENGTH(cjkSymbolRanges),
CharacterProperty::isCJKIdeographOrSymbol);
setValues(values,
cjkIsolatedSymbolsArray, ARRAY_LENGTH(cjkIsolatedSymbolsArray),
CharacterProperty::isCJKIdeographOrSymbol);
setRanges(values,
isUprightInMixedVerticalRanges,
ARRAY_LENGTH(isUprightInMixedVerticalRanges),
CharacterProperty::isUprightInMixedVertical);
setValues(values,
isUprightInMixedVerticalArray,
ARRAY_LENGTH(isUprightInMixedVerticalArray),
CharacterProperty::isUprightInMixedVertical);
// Create a trie from the value array.
UErrorCode error = U_ZERO_ERROR;
UTrie2* trie = utrie2_open(0, 0, &error);
assert(error == U_ZERO_ERROR);
UChar32 start = 0;
CharacterProperty value = values[0];
for (UChar32 c = 1;; c++) {
if (c < size && values[c] == value)
continue;
if (static_cast<uint32_t>(value)) {
utrie2_setRange32(trie, start, c - 1,
static_cast<uint32_t>(value), TRUE, &error);
assert(error == U_ZERO_ERROR);
}
if (c >= size)
break;
start = c;
value = values[start];
}
// Freeze and serialize the trie to a byte array.
utrie2_freeze(trie, UTrie2ValueBits::UTRIE2_16_VALUE_BITS, &error);
assert(error == U_ZERO_ERROR);
int32_t serializedSize = utrie2_serialize(trie, nullptr, 0, &error);
error = U_ZERO_ERROR;
uint8_t* serialized = new uint8_t[serializedSize];
serializedSize = utrie2_serialize(trie, serialized, serializedSize, &error);
assert(error == U_ZERO_ERROR);
// Write the serialized array to the source file.
if (argc <= 1) {
generate(stdout, serializedSize, serialized);
} else {
FILE* fp = fopen(argv[1], "wb");
generate(fp, serializedSize, serialized);
fclose(fp);
}
utrie2_close(trie);
return 0;
}
void
CasePropsBuilder::setProps(const UniProps &props, const UnicodeSet &newValues,
UErrorCode &errorCode) {
if(U_FAILURE(errorCode) || newValues.containsNone(relevantProps)) { return; }
UChar32 start=props.start;
UChar32 end=props.end;
/* default: map to self */
int32_t delta=0;
uint32_t type;
if(props.binProps[UCHAR_LOWERCASE]) {
type=UCASE_LOWER;
} else if(props.binProps[UCHAR_UPPERCASE]) {
type=UCASE_UPPER;
} else if(props.getIntProp(UCHAR_GENERAL_CATEGORY)==U_TITLECASE_LETTER) {
type=UCASE_TITLE;
} else {
type=UCASE_NONE;
}
uint32_t value=type;
UBool hasMapping=FALSE;
if(props.suc>=0) {
/* uppercase mapping as delta if the character is lowercase */
hasMapping=TRUE;
if(type==UCASE_LOWER) {
delta=props.suc-start;
} else {
value|=UCASE_EXCEPTION;
}
}
if(props.slc>=0) {
/* lowercase mapping as delta if the character is uppercase or titlecase */
hasMapping=TRUE;
if(type>=UCASE_UPPER) {
delta=props.slc-start;
} else {
value|=UCASE_EXCEPTION;
}
}
if(props.stc>=0) {
hasMapping=TRUE;
}
if(props.suc!=props.stc) {
value|=UCASE_EXCEPTION;
}
if(!props.lc.isEmpty() || !props.uc.isEmpty() || !props.tc.isEmpty() ||
newValues.contains(PPUCD_CONDITIONAL_CASE_MAPPINGS)
) {
hasMapping=TRUE;
value|=UCASE_EXCEPTION;
}
if( (props.scf>=0 && props.scf!=props.slc) ||
(!props.cf.isEmpty() && props.cf!=UnicodeString(props.scf)) ||
newValues.contains(PPUCD_TURKIC_CASE_FOLDING)
) {
hasMapping=TRUE;
value|=UCASE_EXCEPTION;
}
// Simple case folding falls back to simple lowercasing.
// If there is no case folding but there is a lowercase mapping,
// then add a case folding mapping to the code point.
// For example: Cherokee uppercase syllables since Unicode 8.
// (Full case folding falls back to simple case folding,
// not to full lowercasing, so we need not also handle it specially
// for such cases.)
UChar32 scf=props.scf;
if(scf<0 && props.slc>=0) {
scf=start;
hasMapping=TRUE;
value|=UCASE_EXCEPTION;
}
if(delta<UCASE_MIN_DELTA || UCASE_MAX_DELTA<delta) {
value|=UCASE_EXCEPTION;
}
if(props.binProps[UCHAR_SOFT_DOTTED]) {
value|=UCASE_SOFT_DOTTED;
}
int32_t cc=props.getIntProp(UCHAR_CANONICAL_COMBINING_CLASS);
if(cc!=0) {
if(props.binProps[UCHAR_SOFT_DOTTED]) {
fprintf(stderr, "genprops error: a soft-dotted character has ccc!=0\n");
errorCode=U_ILLEGAL_ARGUMENT_ERROR;
return;
}
if(cc==230) {
value|=UCASE_ABOVE;
} else {
value|=UCASE_OTHER_ACCENT;
}
}
if(props.binProps[UCHAR_CASE_IGNORABLE]) {
value|=UCASE_IGNORABLE;
}
if((hasMapping || (value&UCASE_EXCEPTION)) && start!=end) {
fprintf(stderr,
"genprops error: range %04lX..%04lX has case mappings "
"or reasons for data structure exceptions\n",
(long)start, (long)end);
errorCode=U_ILLEGAL_ARGUMENT_ERROR;
return;
}
/* handle exceptions */
if(value&UCASE_EXCEPTION) {
/* simply store exceptions for later processing and encoding */
if(excPropsCount==MAX_EXC_COUNT) {
fprintf(stderr, "genprops error: casepropsbuilder: too many exceptions\n");
errorCode=U_INDEX_OUTOFBOUNDS_ERROR;
return;
}
ExcProps *newExcProps=new ExcProps(props);
if(newExcProps==NULL) {
fprintf(stderr,
"genprops error: casepropsbuilder out of memory allocating "
"exceptions properties\n");
errorCode=U_MEMORY_ALLOCATION_ERROR;
return;
}
newExcProps->props.scf=scf;
newExcProps->hasConditionalCaseMappings=newValues.contains(PPUCD_CONDITIONAL_CASE_MAPPINGS);
newExcProps->hasTurkicCaseFolding=newValues.contains(PPUCD_TURKIC_CASE_FOLDING);
value|=(uint32_t)excPropsCount<<UGENCASE_EXC_SHIFT;
excProps[excPropsCount++]=newExcProps;
} else {
/* store the simple case mapping delta */
value|=((uint32_t)delta<<UCASE_DELTA_SHIFT)&UCASE_DELTA_MASK;
}
utrie2_setRange32(pTrie, start, end, value, TRUE, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(stderr, "genprops error: unable to set case mapping values: %s\n",
u_errorName(errorCode));
return;
}
if(hasMapping) {
/* update the case-sensitive set */
caseSensitive.add(start);
if(scf>=0) { caseSensitive.add(scf); }
if(props.slc>=0) { caseSensitive.add(props.slc); }
if(props.suc>=0) { caseSensitive.add(props.suc); }
if(props.stc>=0) { caseSensitive.add(props.stc); }
caseSensitive.addAll(props.cf);
caseSensitive.addAll(props.lc);
caseSensitive.addAll(props.uc);
caseSensitive.addAll(props.tc);
/* update maxFullLength */
if(props.cf.length()>maxFullLength) { maxFullLength=props.cf.length(); }
if(props.lc.length()>maxFullLength) { maxFullLength=props.lc.length(); }
if(props.uc.length()>maxFullLength) { maxFullLength=props.uc.length(); }
if(props.tc.length()>maxFullLength) { maxFullLength=props.tc.length(); }
}
/* add the multi-character case folding to the "unfold" data */
if(props.cf.hasMoreChar32Than(0, 0x7fffffff, 1)) {
addUnfolding(start, props.cf, errorCode);
}
}
// Build the Whole Script Confusable data
//
// TODO: Reorganize. Either get rid of the WSConfusableDataBuilder class,
// because everything is local to this one build function anyhow,
// OR
// break this function into more reasonably sized pieces, with
// state in WSConfusableDataBuilder.
//
void buildWSConfusableData(SpoofImpl *spImpl, const char * confusablesWS,
int32_t confusablesWSLen, UParseError *pe, UErrorCode &status)
{
if (U_FAILURE(status)) {
return;
}
URegularExpression *parseRegexp = NULL;
int32_t inputLen = 0;
UChar *input = NULL;
int32_t lineNum = 0;
UVector *scriptSets = NULL;
uint32_t rtScriptSetsCount = 2;
UTrie2 *anyCaseTrie = NULL;
UTrie2 *lowerCaseTrie = NULL;
anyCaseTrie = utrie2_open(0, 0, &status);
lowerCaseTrie = utrie2_open(0, 0, &status);
// The scriptSets vector provides a mapping from TRIE values to the set of scripts.
//
// Reserved TRIE values:
// 0: Code point has no whole script confusables.
// 1: Code point is of script Common or Inherited.
// These code points do not participate in whole script confusable detection.
// (This is logically equivalent to saying that they contain confusables in
// all scripts)
//
// Because Trie values are indexes into the ScriptSets vector, pre-fill
// vector positions 0 and 1 to avoid conflicts with the reserved values.
scriptSets = new UVector(status);
if (scriptSets == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
goto cleanup;
}
scriptSets->addElement((void *)NULL, status);
scriptSets->addElement((void *)NULL, status);
// Convert the user input data from UTF-8 to UChar (UTF-16)
u_strFromUTF8(NULL, 0, &inputLen, confusablesWS, confusablesWSLen, &status);
if (status != U_BUFFER_OVERFLOW_ERROR) {
goto cleanup;
}
status = U_ZERO_ERROR;
input = static_cast<UChar *>(uprv_malloc((inputLen+1) * sizeof(UChar)));
if (input == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
goto cleanup;
}
u_strFromUTF8(input, inputLen+1, NULL, confusablesWS, confusablesWSLen, &status);
parseRegexp = uregex_openC(parseExp, 0, NULL, &status);
// Zap any Byte Order Mark at the start of input. Changing it to a space is benign
// given the syntax of the input.
if (*input == 0xfeff) {
*input = 0x20;
}
// Parse the input, one line per iteration of this loop.
uregex_setText(parseRegexp, input, inputLen, &status);
while (uregex_findNext(parseRegexp, &status)) {
lineNum++;
UChar line[200];
uregex_group(parseRegexp, 0, line, 200, &status);
if (uregex_start(parseRegexp, 1, &status) >= 0) {
// this was a blank or comment line.
continue;
}
if (uregex_start(parseRegexp, 8, &status) >= 0) {
// input file syntax error.
status = U_PARSE_ERROR;
goto cleanup;
}
if (U_FAILURE(status)) {
goto cleanup;
}
// Pick up the start and optional range end code points from the parsed line.
UChar32 startCodePoint = SpoofImpl::ScanHex(
input, uregex_start(parseRegexp, 2, &status), uregex_end(parseRegexp, 2, &status), status);
UChar32 endCodePoint = startCodePoint;
if (uregex_start(parseRegexp, 3, &status) >=0) {
endCodePoint = SpoofImpl::ScanHex(
input, uregex_start(parseRegexp, 3, &status), uregex_end(parseRegexp, 3, &status), status);
}
// Extract the two script names from the source line. We need these in an 8 bit
// default encoding (will be EBCDIC on IBM mainframes) in order to pass them on
// to the ICU u_getPropertyValueEnum() function. Ugh.
char srcScriptName[20];
char targScriptName[20];
extractGroup(parseRegexp, 4, srcScriptName, sizeof(srcScriptName), status);
extractGroup(parseRegexp, 5, targScriptName, sizeof(targScriptName), status);
UScriptCode srcScript =
static_cast<UScriptCode>(u_getPropertyValueEnum(UCHAR_SCRIPT, srcScriptName));
UScriptCode targScript =
static_cast<UScriptCode>(u_getPropertyValueEnum(UCHAR_SCRIPT, targScriptName));
if (U_FAILURE(status)) {
goto cleanup;
}
if (srcScript == USCRIPT_INVALID_CODE || targScript == USCRIPT_INVALID_CODE) {
status = U_INVALID_FORMAT_ERROR;
goto cleanup;
}
// select the table - (A) any case or (L) lower case only
UTrie2 *table = anyCaseTrie;
if (uregex_start(parseRegexp, 7, &status) >= 0) {
table = lowerCaseTrie;
}
// Build the set of scripts containing confusable characters for
// the code point(s) specified in this input line.
// Sanity check that the script of the source code point is the same
// as the source script indicated in the input file. Failure of this check is
// an error in the input file.
// Include the source script in the set (needed for Mixed Script Confusable detection).
//
UChar32 cp;
for (cp=startCodePoint; cp<=endCodePoint; cp++) {
int32_t setIndex = utrie2_get32(table, cp);
BuilderScriptSet *bsset = NULL;
if (setIndex > 0) {
U_ASSERT(setIndex < scriptSets->size());
bsset = static_cast<BuilderScriptSet *>(scriptSets->elementAt(setIndex));
} else {
bsset = new BuilderScriptSet();
if (bsset == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
goto cleanup;
}
bsset->codePoint = cp;
bsset->trie = table;
bsset->sset = new ScriptSet();
setIndex = scriptSets->size();
bsset->index = setIndex;
bsset->rindex = 0;
if (bsset->sset == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
goto cleanup;
}
scriptSets->addElement(bsset, status);
utrie2_set32(table, cp, setIndex, &status);
}
bsset->sset->Union(targScript);
bsset->sset->Union(srcScript);
if (U_FAILURE(status)) {
goto cleanup;
}
UScriptCode cpScript = uscript_getScript(cp, &status);
if (cpScript != srcScript) {
status = U_INVALID_FORMAT_ERROR;
goto cleanup;
}
}
}
// Eliminate duplicate script sets. At this point we have a separate
// script set for every code point that had data in the input file.
//
// We eliminate underlying ScriptSet objects, not the BuildScriptSets that wrap them
//
// printf("Number of scriptSets: %d\n", scriptSets->size());
{
int32_t duplicateCount = 0;
rtScriptSetsCount = 2;
for (int32_t outeri=2; outeri<scriptSets->size(); outeri++) {
BuilderScriptSet *outerSet = static_cast<BuilderScriptSet *>(scriptSets->elementAt(outeri));
if (outerSet->index != static_cast<uint32_t>(outeri)) {
// This set was already identified as a duplicate.
// It will not be allocated a position in the runtime array of ScriptSets.
continue;
}
outerSet->rindex = rtScriptSetsCount++;
for (int32_t inneri=outeri+1; inneri<scriptSets->size(); inneri++) {
BuilderScriptSet *innerSet = static_cast<BuilderScriptSet *>(scriptSets->elementAt(inneri));
if (*(outerSet->sset) == *(innerSet->sset) && outerSet->sset != innerSet->sset) {
delete innerSet->sset;
innerSet->scriptSetOwned = FALSE;
innerSet->sset = outerSet->sset;
innerSet->index = outeri;
innerSet->rindex = outerSet->rindex;
duplicateCount++;
}
// But this doesn't get all. We need to fix the TRIE.
}
}
// printf("Number of distinct script sets: %d\n", rtScriptSetsCount);
}
// Update the Trie values to be reflect the run time script indexes (after duplicate merging).
// (Trie Values 0 and 1 are reserved, and the corresponding slots in scriptSets
// are unused, which is why the loop index starts at 2.)
{
for (int32_t i=2; i<scriptSets->size(); i++) {
BuilderScriptSet *bSet = static_cast<BuilderScriptSet *>(scriptSets->elementAt(i));
if (bSet->rindex != (uint32_t)i) {
utrie2_set32(bSet->trie, bSet->codePoint, bSet->rindex, &status);
}
}
}
// For code points with script==Common or script==Inherited,
// Set the reserved value of 1 into both Tries. These characters do not participate
// in Whole Script Confusable detection; this reserved value is the means
// by which they are detected.
{
UnicodeSet ignoreSet;
ignoreSet.applyIntPropertyValue(UCHAR_SCRIPT, USCRIPT_COMMON, status);
UnicodeSet inheritedSet;
inheritedSet.applyIntPropertyValue(UCHAR_SCRIPT, USCRIPT_INHERITED, status);
ignoreSet.addAll(inheritedSet);
for (int32_t rn=0; rn<ignoreSet.getRangeCount(); rn++) {
UChar32 rangeStart = ignoreSet.getRangeStart(rn);
UChar32 rangeEnd = ignoreSet.getRangeEnd(rn);
utrie2_setRange32(anyCaseTrie, rangeStart, rangeEnd, 1, TRUE, &status);
utrie2_setRange32(lowerCaseTrie, rangeStart, rangeEnd, 1, TRUE, &status);
}
}
// Serialize the data to the Spoof Detector
{
utrie2_freeze(anyCaseTrie, UTRIE2_16_VALUE_BITS, &status);
int32_t size = utrie2_serialize(anyCaseTrie, NULL, 0, &status);
// printf("Any case Trie size: %d\n", size);
if (status != U_BUFFER_OVERFLOW_ERROR) {
goto cleanup;
}
status = U_ZERO_ERROR;
spImpl->fSpoofData->fRawData->fAnyCaseTrie = spImpl->fSpoofData->fMemLimit;
spImpl->fSpoofData->fRawData->fAnyCaseTrieLength = size;
spImpl->fSpoofData->fAnyCaseTrie = anyCaseTrie;
void *where = spImpl->fSpoofData->reserveSpace(size, status);
utrie2_serialize(anyCaseTrie, where, size, &status);
utrie2_freeze(lowerCaseTrie, UTRIE2_16_VALUE_BITS, &status);
size = utrie2_serialize(lowerCaseTrie, NULL, 0, &status);
// printf("Lower case Trie size: %d\n", size);
if (status != U_BUFFER_OVERFLOW_ERROR) {
goto cleanup;
}
status = U_ZERO_ERROR;
spImpl->fSpoofData->fRawData->fLowerCaseTrie = spImpl->fSpoofData->fMemLimit;
spImpl->fSpoofData->fRawData->fLowerCaseTrieLength = size;
spImpl->fSpoofData->fLowerCaseTrie = lowerCaseTrie;
where = spImpl->fSpoofData->reserveSpace(size, status);
utrie2_serialize(lowerCaseTrie, where, size, &status);
spImpl->fSpoofData->fRawData->fScriptSets = spImpl->fSpoofData->fMemLimit;
spImpl->fSpoofData->fRawData->fScriptSetsLength = rtScriptSetsCount;
ScriptSet *rtScriptSets = static_cast<ScriptSet *>
(spImpl->fSpoofData->reserveSpace(rtScriptSetsCount * sizeof(ScriptSet), status));
uint32_t rindex = 2;
for (int32_t i=2; i<scriptSets->size(); i++) {
BuilderScriptSet *bSet = static_cast<BuilderScriptSet *>(scriptSets->elementAt(i));
if (bSet->rindex < rindex) {
// We have already copied this script set to the serialized data.
continue;
}
U_ASSERT(rindex == bSet->rindex);
rtScriptSets[rindex] = *bSet->sset; // Assignment of a ScriptSet just copies the bits.
rindex++;
}
}
// Open new utrie2s from the serialized data. We don't want to keep the ones
// we just built because we would then have two copies of the data, one internal to
// the utries that we have already constructed, and one in the serialized data area.
// An alternative would be to not pre-serialize the Trie data, but that makes the
// spoof detector data different, depending on how the detector was constructed.
// It's simpler to keep the data always the same.
spImpl->fSpoofData->fAnyCaseTrie = utrie2_openFromSerialized(
UTRIE2_16_VALUE_BITS,
(const char *)spImpl->fSpoofData->fRawData + spImpl->fSpoofData->fRawData->fAnyCaseTrie,
spImpl->fSpoofData->fRawData->fAnyCaseTrieLength,
NULL,
&status);
spImpl->fSpoofData->fLowerCaseTrie = utrie2_openFromSerialized(
UTRIE2_16_VALUE_BITS,
(const char *)spImpl->fSpoofData->fRawData + spImpl->fSpoofData->fRawData->fLowerCaseTrie,
spImpl->fSpoofData->fRawData->fAnyCaseTrieLength,
NULL,
&status);
cleanup:
if (U_FAILURE(status)) {
pe->line = lineNum;
}
uregex_close(parseRegexp);
uprv_free(input);
int32_t i;
for (i=0; i<scriptSets->size(); i++) {
BuilderScriptSet *bsset = static_cast<BuilderScriptSet *>(scriptSets->elementAt(i));
delete bsset;
}
delete scriptSets;
utrie2_close(anyCaseTrie);
utrie2_close(lowerCaseTrie);
return;
}
