U_CAPI UTrie2 * U_EXPORT2
upvec_compactToUTrie2WithRowIndexes(UPropsVectors *pv, UErrorCode *pErrorCode) {
UPVecToUTrie2Context toUTrie2={ NULL };
upvec_compact(pv, upvec_compactToUTrie2Handler, &toUTrie2, pErrorCode);
utrie2_freeze(toUTrie2.trie, UTRIE2_16_VALUE_BITS, pErrorCode);
if(U_FAILURE(*pErrorCode)) {
utrie2_close(toUTrie2.trie);
toUTrie2.trie=NULL;
}
return toUTrie2.trie;
}
//-----------------------------------------------------------------------------------
//
// getTrieSize() Return the size that will be required to serialize the Trie.
//
//-----------------------------------------------------------------------------------
int32_t RBBISetBuilder::getTrieSize() {
if (U_FAILURE(*fStatus)) {
return 0;
}
utrie2_freeze(fTrie, UTRIE2_16_VALUE_BITS, fStatus);
fTrieSize = utrie2_serialize(fTrie,
NULL, // Buffer
0, // Capacity
fStatus);
if (*fStatus == U_BUFFER_OVERFLOW_ERROR) {
*fStatus = U_ZERO_ERROR;
}
// RBBIDebugPrintf("Trie table size is %d\n", trieSize);
return fTrieSize;
}
/* Almost the same as utrie2_cloneAsThawed() but copies a UTrie and freezes the clone. */
U_CAPI UTrie2 * U_EXPORT2
utrie2_fromUTrie(const UTrie *trie1, uint32_t errorValue, UErrorCode *pErrorCode) {
NewTrieAndStatus context;
UChar lead;
if(U_FAILURE(*pErrorCode)) {
return NULL;
}
if(trie1==NULL) {
*pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
return NULL;
}
context.trie=utrie2_open(trie1->initialValue, errorValue, pErrorCode);
if(U_FAILURE(*pErrorCode)) {
return NULL;
}
context.exclusiveLimit=TRUE;
context.errorCode=*pErrorCode;
utrie_enum(trie1, NULL, copyEnumRange, &context);
*pErrorCode=context.errorCode;
for(lead=0xd800; lead<0xdc00; ++lead) {
uint32_t value;
if(trie1->data32==NULL) {
value=UTRIE_GET16_FROM_LEAD(trie1, lead);
} else {
value=UTRIE_GET32_FROM_LEAD(trie1, lead);
}
if(value!=trie1->initialValue) {
utrie2_set32ForLeadSurrogateCodeUnit(context.trie, lead, value, pErrorCode);
}
}
if(U_SUCCESS(*pErrorCode)) {
utrie2_freeze(context.trie,
trie1->data32!=NULL ? UTRIE2_32_VALUE_BITS : UTRIE2_16_VALUE_BITS,
pErrorCode);
}
#ifdef UTRIE2_DEBUG
if(U_SUCCESS(*pErrorCode)) {
utrie_printLengths(trie1);
utrie2_printLengths(context.trie, "fromUTrie");
}
#endif
if(U_FAILURE(*pErrorCode)) {
utrie2_close(context.trie);
context.trie=NULL;
}
return context.trie;
}
void
BiDiPropsBuilder::build(UErrorCode &errorCode) {
makeMirror(errorCode);
if(U_FAILURE(errorCode)) { return; }
utrie2_freeze(pTrie, UTRIE2_16_VALUE_BITS, &errorCode);
trieSize=utrie2_serialize(pTrie, trieBlock, sizeof(trieBlock), &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(stderr, "genprops error: utrie2_freeze()+utrie2_serialize() failed: %s (length %ld)\n",
u_errorName(errorCode), (long)trieSize);
return;
}
// Finish jgArray & jgArray2.
UChar32 jgStart; // First code point with a Joining_Group, first range.
UChar32 jgLimit; // One past the last one.
// Find the end of the range first, so that if it's empty we
// get jgStart=jgLimit=MIN_JG_START.
for(jgLimit=MAX_JG_LIMIT;
MIN_JG_START<jgLimit && jgArray[jgLimit-MIN_JG_START-1]==U_JG_NO_JOINING_GROUP;
--jgLimit) {}
for(jgStart=MIN_JG_START;
jgStart<jgLimit && jgArray[jgStart-MIN_JG_START]==U_JG_NO_JOINING_GROUP;
++jgStart) {}
UChar32 jgStart2; // First code point with a Joining_Group, second range.
UChar32 jgLimit2; // One past the last one.
for(jgLimit2=MAX_JG_LIMIT2;
MIN_JG_START2<jgLimit2 && jgArray2[jgLimit2-MIN_JG_START2-1]==U_JG_NO_JOINING_GROUP;
--jgLimit2) {}
for(jgStart2=MIN_JG_START2;
jgStart2<jgLimit2 && jgArray2[jgStart2-MIN_JG_START2]==U_JG_NO_JOINING_GROUP;
++jgStart2) {}
// Pad the total Joining_Group arrays length to a multiple of 4.
// Prefer rounding down starts before rounding up limits
// so that we are guaranteed not to increase the limits beyond
// the end of the arrays' code point ranges.
int32_t jgLength=jgLimit-jgStart+jgLimit2-jgStart2;
while(jgLength&3) {
if((jgStart<jgLimit) && (jgStart&3)) {
--jgStart;
} else if((jgStart2<jgLimit2) && (jgStart2&3)) {
--jgStart2;
} else if(jgStart<jgLimit) {
++jgLimit;
} else {
++jgLimit2;
}
++jgLength;
}
indexes[UBIDI_IX_JG_START]=jgStart;
indexes[UBIDI_IX_JG_LIMIT]=jgLimit;
indexes[UBIDI_IX_JG_START2]=jgStart2;
indexes[UBIDI_IX_JG_LIMIT2]=jgLimit2;
indexes[UBIDI_IX_TRIE_SIZE]=trieSize;
indexes[UBIDI_IX_MIRROR_LENGTH]=mirrorTop;
indexes[UBIDI_IX_LENGTH]=
(int32_t)sizeof(indexes)+
trieSize+
4*mirrorTop+
jgLength;
if(!beQuiet) {
puts("* ubidi.icu stats *");
printf("trie size in bytes: %5d\n", (int)trieSize);
printf("size in bytes of mirroring table: %5d\n", (int)(4*mirrorTop));
printf("length of Joining_Group array: %5d (U+%04x..U+%04x)\n",
(int)(jgLimit-jgStart), (int)jgStart, (int)(jgLimit-1));
printf("length of Joining_Group array 2: %5d (U+%04x..U+%04x)\n",
(int)(jgLimit2-jgStart2), (int)jgStart2, (int)(jgLimit2-1));
printf("data size: %5d\n", (int)indexes[UBIDI_IX_LENGTH]);
}
indexes[UBIDI_MAX_VALUES_INDEX]=
((int32_t)U_CHAR_DIRECTION_COUNT-1)|
(((int32_t)U_JT_COUNT-1)<<UBIDI_JT_SHIFT)|
(((int32_t)U_BPT_COUNT-1)<<UBIDI_BPT_SHIFT)|
(((int32_t)U_JG_COUNT-1)<<UBIDI_MAX_JG_SHIFT);
}
extern void
generateData(const char *dataDir, UBool csource) {
static int32_t indexes[UBIDI_IX_TOP]={
UBIDI_IX_TOP
};
static uint8_t trieBlock[40000];
static uint8_t jgArray[0x300]; /* at most for U+0600..U+08FF */
const uint32_t *row;
UChar32 start, end, prev, jgStart;
int32_t i;
UNewDataMemory *pData;
UNewTrie *pTrie;
UErrorCode errorCode=U_ZERO_ERROR;
int32_t trieSize;
long dataLength;
makeMirror();
pTrie=utrie_open(NULL, NULL, 20000, 0, 0, TRUE);
if(pTrie==NULL) {
fprintf(stderr, "genbidi error: unable to create a UNewTrie\n");
exit(U_MEMORY_ALLOCATION_ERROR);
}
prev=jgStart=0;
for(i=0; (row=upvec_getRow(pv, i, &start, &end))!=NULL && start<UPVEC_FIRST_SPECIAL_CP; ++i) {
/* store most values from vector column 0 in the trie */
if(!utrie_setRange32(pTrie, start, end+1, *row, TRUE)) {
fprintf(stderr, "genbidi error: unable to set trie value (overflow)\n");
exit(U_BUFFER_OVERFLOW_ERROR);
}
/* store Joining_Group values from vector column 1 in a simple byte array */
if(row[1]!=0) {
if(start<0x600 || 0x8ff<end) {
fprintf(stderr, "genbidi error: Joining_Group for out-of-range code points U+%04lx..U+%04lx\n",
(long)start, (long)end);
exit(U_ILLEGAL_ARGUMENT_ERROR);
}
if(prev==0) {
/* first code point with any value */
prev=jgStart=start;
} else {
/* add No_Joining_Group for code points between prev and start */
while(prev<start) {
jgArray[prev++ -jgStart]=0;
}
}
/* set Joining_Group value for start..end */
while(prev<=end) {
jgArray[prev++ -jgStart]=(uint8_t)row[1];
}
}
}
/* finish jgArray, pad to multiple of 4 */
while((prev-jgStart)&3) {
jgArray[prev++ -jgStart]=0;
}
indexes[UBIDI_IX_JG_START]=jgStart;
indexes[UBIDI_IX_JG_LIMIT]=prev;
trieSize=utrie_serialize(pTrie, trieBlock, sizeof(trieBlock), NULL, TRUE, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(stderr, "genbidi error: utrie_serialize failed: %s (length %ld)\n", u_errorName(errorCode), (long)trieSize);
exit(errorCode);
}
indexes[UBIDI_IX_TRIE_SIZE]=trieSize;
indexes[UBIDI_IX_MIRROR_LENGTH]=mirrorTop;
indexes[UBIDI_IX_LENGTH]=
(int32_t)sizeof(indexes)+
trieSize+
4*mirrorTop+
(prev-jgStart);
if(beVerbose) {
printf("trie size in bytes: %5d\n", (int)trieSize);
printf("size in bytes of mirroring table: %5d\n", (int)(4*mirrorTop));
printf("length of Joining_Group array: %5d (U+%04x..U+%04x)\n", (int)(prev-jgStart), (int)jgStart, (int)(prev-1));
printf("data size: %5d\n", (int)indexes[UBIDI_IX_LENGTH]);
}
indexes[UBIDI_MAX_VALUES_INDEX]=
((int32_t)U_CHAR_DIRECTION_COUNT-1)|
(((int32_t)U_JT_COUNT-1)<<UBIDI_JT_SHIFT)|
(((int32_t)U_JG_COUNT-1)<<UBIDI_MAX_JG_SHIFT);
if(csource) {
/* write .c file for hardcoded data */
UTrie trie={ NULL };
UTrie2 *trie2;
FILE *f;
utrie_unserialize(&trie, trieBlock, trieSize, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(
stderr,
"genbidi error: failed to utrie_unserialize(ubidi.icu trie) - %s\n",
u_errorName(errorCode));
exit(errorCode);
}
/* use UTrie2 */
dataInfo.formatVersion[0]=2;
dataInfo.formatVersion[2]=0;
dataInfo.formatVersion[3]=0;
trie2=utrie2_fromUTrie(&trie, 0, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(
stderr,
"genbidi error: utrie2_fromUTrie() failed - %s\n",
u_errorName(errorCode));
exit(errorCode);
}
{
/* delete lead surrogate code unit values */
UChar lead;
trie2=utrie2_cloneAsThawed(trie2, &errorCode);
for(lead=0xd800; lead<0xdc00; ++lead) {
utrie2_set32ForLeadSurrogateCodeUnit(trie2, lead, trie2->initialValue, &errorCode);
}
utrie2_freeze(trie2, UTRIE2_16_VALUE_BITS, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(
stderr,
"genbidi error: deleting lead surrogate code unit values failed - %s\n",
u_errorName(errorCode));
exit(errorCode);
}
}
f=usrc_create(dataDir, "ubidi_props_data.c");
if(f!=NULL) {
usrc_writeArray(f,
"static const UVersionInfo ubidi_props_dataVersion={",
dataInfo.dataVersion, 8, 4,
"};\n\n");
usrc_writeArray(f,
"static const int32_t ubidi_props_indexes[UBIDI_IX_TOP]={",
indexes, 32, UBIDI_IX_TOP,
"};\n\n");
usrc_writeUTrie2Arrays(f,
"static const uint16_t ubidi_props_trieIndex[%ld]={\n", NULL,
trie2,
"\n};\n\n");
usrc_writeArray(f,
"static const uint32_t ubidi_props_mirrors[%ld]={\n",
mirrors, 32, mirrorTop,
"\n};\n\n");
usrc_writeArray(f,
"static const uint8_t ubidi_props_jgArray[%ld]={\n",
jgArray, 8, prev-jgStart,
"\n};\n\n");
fputs(
"static const UBiDiProps ubidi_props_singleton={\n"
" NULL,\n"
" ubidi_props_indexes,\n"
" ubidi_props_mirrors,\n"
" ubidi_props_jgArray,\n",
f);
usrc_writeUTrie2Struct(f,
" {\n",
trie2, "ubidi_props_trieIndex", NULL,
" },\n");
usrc_writeArray(f, " { ", dataInfo.formatVersion, 8, 4, " }\n");
fputs("};\n", f);
fclose(f);
}
utrie2_close(trie2);
} else {
/* write the data */
pData=udata_create(dataDir, UBIDI_DATA_TYPE, UBIDI_DATA_NAME, &dataInfo,
haveCopyright ? U_COPYRIGHT_STRING : NULL, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(stderr, "genbidi: unable to create data memory, %s\n", u_errorName(errorCode));
exit(errorCode);
}
udata_writeBlock(pData, indexes, sizeof(indexes));
udata_writeBlock(pData, trieBlock, trieSize);
udata_writeBlock(pData, mirrors, 4*mirrorTop);
udata_writeBlock(pData, jgArray, prev-jgStart);
/* finish up */
dataLength=udata_finish(pData, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(stderr, "genbidi: error %d writing the output file\n", errorCode);
exit(errorCode);
}
if(dataLength!=indexes[UBIDI_IX_LENGTH]) {
fprintf(stderr, "genbidi: data length %ld != calculated size %d\n",
dataLength, (int)indexes[UBIDI_IX_LENGTH]);
exit(U_INTERNAL_PROGRAM_ERROR);
}
}
utrie_close(pTrie);
upvec_close(pv);
}
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;
}
extern void
generateData(const char *dataDir, UBool csource) {
static int32_t indexes[UPROPS_INDEX_COUNT]={
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0,
0, 0, 0, 0
};
static uint8_t trieBlock[40000];
static uint8_t additionalProps[120000];
UNewDataMemory *pData;
UErrorCode errorCode=U_ZERO_ERROR;
uint32_t size = 0;
int32_t trieSize, additionalPropsSize, offset;
long dataLength;
trieSize=utrie_serialize(pTrie, trieBlock, sizeof(trieBlock), NULL, TRUE, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(stderr, "error: utrie_serialize failed: %s (length %ld)\n", u_errorName(errorCode), (long)trieSize);
exit(errorCode);
}
offset=sizeof(indexes)/4; /* uint32_t offset to the properties trie */
/* round up trie size to 4-alignment */
trieSize=(trieSize+3)&~3;
offset+=trieSize>>2;
indexes[UPROPS_PROPS32_INDEX]= /* set indexes to the same offsets for empty */
indexes[UPROPS_EXCEPTIONS_INDEX]= /* structures from the old format version 3 */
indexes[UPROPS_EXCEPTIONS_TOP_INDEX]= /* so that less runtime code has to be changed */
indexes[UPROPS_ADDITIONAL_TRIE_INDEX]=offset;
if(beVerbose) {
printf("trie size in bytes: %5u\n", (int)trieSize);
}
if(csource) {
/* write .c file for hardcoded data */
UTrie trie={ NULL };
UTrie2 *trie2;
FILE *f;
utrie_unserialize(&trie, trieBlock, trieSize, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(
stderr,
"genprops error: failed to utrie_unserialize(uprops.icu main trie) - %s\n",
u_errorName(errorCode));
exit(errorCode);
}
/* use UTrie2 */
trie2=utrie2_fromUTrie(&trie, 0, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(
stderr,
"genprops error: utrie2_fromUTrie() failed - %s\n",
u_errorName(errorCode));
exit(errorCode);
}
{
/* delete lead surrogate code unit values */
UChar lead;
trie2=utrie2_cloneAsThawed(trie2, &errorCode);
for(lead=0xd800; lead<0xdc00; ++lead) {
utrie2_set32ForLeadSurrogateCodeUnit(trie2, lead, trie2->initialValue, &errorCode);
}
utrie2_freeze(trie2, UTRIE2_16_VALUE_BITS, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(
stderr,
"genprops error: deleting lead surrogate code unit values failed - %s\n",
u_errorName(errorCode));
exit(errorCode);
}
}
f=usrc_create(dataDir, "uchar_props_data.c");
if(f!=NULL) {
/* unused
usrc_writeArray(f,
"static const UVersionInfo formatVersion={",
dataInfo.formatVersion, 8, 4,
"};\n\n");
*/
usrc_writeArray(f,
"static const UVersionInfo dataVersion={",
dataInfo.dataVersion, 8, 4,
"};\n\n");
usrc_writeUTrie2Arrays(f,
"static const uint16_t propsTrie_index[%ld]={\n", NULL,
trie2,
"\n};\n\n");
usrc_writeUTrie2Struct(f,
"static const UTrie2 propsTrie={\n",
trie2, "propsTrie_index", NULL,
"};\n\n");
additionalPropsSize=writeAdditionalData(f, additionalProps, sizeof(additionalProps), indexes);
size=4*offset+additionalPropsSize; /* total size of data */
usrc_writeArray(f,
"static const int32_t indexes[UPROPS_INDEX_COUNT]={",
indexes, 32, UPROPS_INDEX_COUNT,
"};\n\n");
fclose(f);
}
utrie2_close(trie2);
} else {
/* write the data */
pData=udata_create(dataDir, DATA_TYPE, DATA_NAME, &dataInfo,
haveCopyright ? U_COPYRIGHT_STRING : NULL, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(stderr, "genprops: unable to create data memory, %s\n", u_errorName(errorCode));
exit(errorCode);
}
additionalPropsSize=writeAdditionalData(NULL, additionalProps, sizeof(additionalProps), indexes);
size=4*offset+additionalPropsSize; /* total size of data */
udata_writeBlock(pData, indexes, sizeof(indexes));
udata_writeBlock(pData, trieBlock, trieSize);
udata_writeBlock(pData, additionalProps, additionalPropsSize);
/* finish up */
dataLength=udata_finish(pData, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(stderr, "genprops: error %d writing the output file\n", errorCode);
exit(errorCode);
}
if(dataLength!=(long)size) {
fprintf(stderr, "genprops: data length %ld != calculated size %lu\n",
dataLength, (unsigned long)size);
exit(U_INTERNAL_PROGRAM_ERROR);
}
}
if(beVerbose) {
printf("data size: %6lu\n", (unsigned long)size);
}
}
// 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;
}
U_CFUNC int32_t
writeAdditionalData(FILE *f, uint8_t *p, int32_t capacity, int32_t indexes[UPROPS_INDEX_COUNT]) {
const uint32_t *pvArray;
int32_t pvRows, pvCount;
int32_t length;
UErrorCode errorCode;
pvArray=upvec_getArray(pv, &pvRows, NULL);
pvCount=pvRows*UPROPS_VECTOR_WORDS;
errorCode=U_ZERO_ERROR;
length=utrie_serialize(newTrie, p, capacity, NULL, TRUE, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(stderr, "genprops error: unable to serialize trie for additional properties: %s\n", u_errorName(errorCode));
exit(errorCode);
}
if(p!=NULL) {
if(beVerbose) {
printf("size in bytes of additional props trie:%5u\n", (int)length);
}
if(f!=NULL) {
UTrie trie={ NULL };
UTrie2 *trie2;
utrie_unserialize(&trie, p, length, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(
stderr,
"genprops error: failed to utrie_unserialize(trie for additional properties) - %s\n",
u_errorName(errorCode));
exit(errorCode);
}
/* use UTrie2 */
trie2=utrie2_fromUTrie(&trie, trie.initialValue, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(
stderr,
"genprops error: utrie2_fromUTrie() failed - %s\n",
u_errorName(errorCode));
exit(errorCode);
}
{
/* delete lead surrogate code unit values */
UChar lead;
trie2=utrie2_cloneAsThawed(trie2, &errorCode);
for(lead=0xd800; lead<0xdc00; ++lead) {
utrie2_set32ForLeadSurrogateCodeUnit(trie2, lead, trie2->initialValue, &errorCode);
}
utrie2_freeze(trie2, UTRIE2_16_VALUE_BITS, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(
stderr,
"genbidi error: deleting lead surrogate code unit values failed - %s\n",
u_errorName(errorCode));
exit(errorCode);
}
}
usrc_writeUTrie2Arrays(f,
"static const uint16_t propsVectorsTrie_index[%ld]={\n", NULL,
trie2,
"\n};\n\n");
usrc_writeUTrie2Struct(f,
"static const UTrie2 propsVectorsTrie={\n",
trie2, "propsVectorsTrie_index", NULL,
"};\n\n");
utrie2_close(trie2);
}
p+=length;
capacity-=length;
/* set indexes */
indexes[UPROPS_ADDITIONAL_VECTORS_INDEX]=
indexes[UPROPS_ADDITIONAL_TRIE_INDEX]+length/4;
indexes[UPROPS_ADDITIONAL_VECTORS_COLUMNS_INDEX]=UPROPS_VECTOR_WORDS;
indexes[UPROPS_RESERVED_INDEX]=
indexes[UPROPS_ADDITIONAL_VECTORS_INDEX]+pvCount;
indexes[UPROPS_MAX_VALUES_INDEX]=
(((int32_t)U_EA_COUNT-1)<<UPROPS_EA_SHIFT)|
(((int32_t)UBLOCK_COUNT-1)<<UPROPS_BLOCK_SHIFT)|
(((int32_t)USCRIPT_CODE_LIMIT-1)&UPROPS_SCRIPT_MASK);
indexes[UPROPS_MAX_VALUES_2_INDEX]=
(((int32_t)U_LB_COUNT-1)<<UPROPS_LB_SHIFT)|
(((int32_t)U_SB_COUNT-1)<<UPROPS_SB_SHIFT)|
(((int32_t)U_WB_COUNT-1)<<UPROPS_WB_SHIFT)|
(((int32_t)U_GCB_COUNT-1)<<UPROPS_GCB_SHIFT)|
((int32_t)U_DT_COUNT-1);
}
if(p!=NULL && (pvCount*4)<=capacity) {
if(f!=NULL) {
usrc_writeArray(f,
"static const uint32_t propsVectors[%ld]={\n",
pvArray, 32, pvCount,
"};\n\n");
fprintf(f, "static const int32_t countPropsVectors=%ld;\n", (long)pvCount);
fprintf(f, "static const int32_t propsVectorsColumns=%ld;\n", (long)indexes[UPROPS_ADDITIONAL_VECTORS_COLUMNS_INDEX]);
} else {
uprv_memcpy(p, pvArray, pvCount*4);
}
if(beVerbose) {
printf("number of additional props vectors: %5u\n", (int)pvRows);
printf("number of 32-bit words per vector: %5u\n", UPROPS_VECTOR_WORDS);
}
}
length+=pvCount*4;
return length;
}
