/* open uprops.icu */
static void
_openProps(UCharProps *ucp, UErrorCode *pErrorCode) {
const uint32_t *p;
int32_t length;
ucp->propsData=udata_openChoice(NULL, DATA_TYPE, DATA_NAME, isAcceptable, NULL, pErrorCode);
if(U_FAILURE(*pErrorCode)) {
return;
}
ucp->pData32=p=(const uint32_t *)udata_getMemory(ucp->propsData);
/* unserialize the trie; it is directly after the int32_t indexes[UPROPS_INDEX_COUNT] */
length=(int32_t)p[UPROPS_PROPS32_INDEX]*4;
length=utrie_unserialize(&ucp->propsTrie, (const uint8_t *)(p+UPROPS_INDEX_COUNT), length-64, pErrorCode);
if(U_FAILURE(*pErrorCode)) {
return;
}
/* unserialize the properties vectors trie */
length=(int32_t)(p[UPROPS_ADDITIONAL_VECTORS_INDEX]-p[UPROPS_ADDITIONAL_TRIE_INDEX])*4;
if(length>0) {
length=utrie_unserialize(&ucp->propsVectorsTrie, (const uint8_t *)(p+p[UPROPS_ADDITIONAL_TRIE_INDEX]), length, pErrorCode);
}
if(length<=0 || U_FAILURE(*pErrorCode)) {
/*
* length==0:
* Allow the properties vectors trie to be missing -
* also requires propsVectorsColumns=indexes[UPROPS_ADDITIONAL_VECTORS_COLUMNS_INDEX]
* to be zero so that this trie is never accessed.
*/
uprv_memset(&ucp->propsVectorsTrie, 0, sizeof(ucp->propsVectorsTrie));
}
}
static UCaseProps *
ucase_openData(UCaseProps *cspProto,
const uint8_t *bin, int32_t length, UErrorCode *pErrorCode) {
UCaseProps *csp;
int32_t size;
cspProto->indexes=(const int32_t *)bin;
if( (length>=0 && length<16*4) ||
cspProto->indexes[UCASE_IX_INDEX_TOP]<16
) {
/* length or indexes[] too short for minimum indexes[] length of 16 */
*pErrorCode=U_INVALID_FORMAT_ERROR;
return NULL;
}
size=cspProto->indexes[UCASE_IX_INDEX_TOP]*4;
if(length>=0) {
if(length>=size && length>=cspProto->indexes[UCASE_IX_LENGTH]) {
length-=size;
} else {
/* length too short for indexes[] or for the whole data length */
*pErrorCode=U_INVALID_FORMAT_ERROR;
return NULL;
}
}
bin+=size;
/* from here on, assume that the sizes of the items fit into the total length */
/* unserialize the trie, after indexes[] */
size=cspProto->indexes[UCASE_IX_TRIE_SIZE];
utrie_unserialize(&cspProto->trie, bin, size, pErrorCode);
if(U_FAILURE(*pErrorCode)) {
return NULL;
}
bin+=size;
/* get exceptions[] */
size=2*cspProto->indexes[UCASE_IX_EXC_LENGTH];
cspProto->exceptions=(const uint16_t *)bin;
bin+=size;
/* get unfold[] */
size=2*cspProto->indexes[UCASE_IX_UNFOLD_LENGTH];
if(size!=0) {
cspProto->unfold=(const UChar *)bin;
bin+=size;
} else {
cspProto->unfold=NULL;
}
/* allocate, copy, and return the new UCaseProps */
csp=(UCaseProps *)uprv_malloc(sizeof(UCaseProps));
if(csp==NULL) {
*pErrorCode=U_MEMORY_ALLOCATION_ERROR;
return NULL;
} else {
uprv_memcpy(csp, cspProto, sizeof(UCaseProps));
return csp;
}
}
static UBiDiProps *
ubidi_openData(UBiDiProps *bdpProto,
const uint8_t *bin, int32_t length, UErrorCode *pErrorCode) {
UBiDiProps *bdp;
int32_t size;
bdpProto->indexes=(const int32_t *)bin;
if( (length>=0 && length<16*4) ||
bdpProto->indexes[UBIDI_IX_INDEX_TOP]<16
) {
/* length or indexes[] too short for minimum indexes[] length of 16 */
*pErrorCode=U_INVALID_FORMAT_ERROR;
return NULL;
}
size=bdpProto->indexes[UBIDI_IX_INDEX_TOP]*4;
if(length>=0) {
if(length>=size && length>=bdpProto->indexes[UBIDI_IX_LENGTH]) {
length-=size;
} else {
/* length too short for indexes[] or for the whole data length */
*pErrorCode=U_INVALID_FORMAT_ERROR;
return NULL;
}
}
bin+=size;
/* from here on, assume that the sizes of the items fit into the total length */
/* unserialize the trie, after indexes[] */
size=bdpProto->indexes[UBIDI_IX_TRIE_SIZE];
utrie_unserialize(&bdpProto->trie, bin, size, pErrorCode);
if(U_FAILURE(*pErrorCode)) {
return NULL;
}
bin+=size;
/* get mirrors[] */
size=4*bdpProto->indexes[UBIDI_IX_MIRROR_LENGTH];
bdpProto->mirrors=(const uint32_t *)bin;
bin+=size;
/* get jgArray[] */
size=bdpProto->indexes[UBIDI_IX_JG_LIMIT]-bdpProto->indexes[UBIDI_IX_JG_START];
bdpProto->jgArray=bin;
bin+=size;
/* allocate, copy, and return the new UBiDiProps */
bdp=(UBiDiProps *)uprv_malloc(sizeof(UBiDiProps));
if(bdp==NULL) {
*pErrorCode=U_MEMORY_ALLOCATION_ERROR;
return NULL;
} else {
uprv_memcpy(bdp, bdpProto, sizeof(UBiDiProps));
return bdp;
}
}
//-----------------------------------------------------------------------------
//
// init(). Does most of the work of construction, shared between the
// constructors.
//
//-----------------------------------------------------------------------------
void RBBIDataWrapper::init(const RBBIDataHeader *data, UErrorCode &status) {
if (U_FAILURE(status)) {
return;
}
fHeader = data;
if (fHeader->fMagic != 0xb1a0 ||
!(fHeader->fFormatVersion[0] == 3 || // ICU 3.4
*(int32_t *)fHeader->fFormatVersion == 1)) // ICU 3.2 and earlier.
{
status = U_INVALID_FORMAT_ERROR;
return;
}
fUDataMem = NULL;
fReverseTable = NULL;
fSafeFwdTable = NULL;
fSafeRevTable = NULL;
if (data->fFTableLen != 0) {
fForwardTable = (RBBIStateTable *)((char *)data + fHeader->fFTable);
}
if (data->fRTableLen != 0) {
fReverseTable = (RBBIStateTable *)((char *)data + fHeader->fRTable);
}
if (data->fSFTableLen != 0) {
fSafeFwdTable = (RBBIStateTable *)((char *)data + fHeader->fSFTable);
}
if (data->fSRTableLen != 0) {
fSafeRevTable = (RBBIStateTable *)((char *)data + fHeader->fSRTable);
}
utrie_unserialize(&fTrie,
(uint8_t *)data + fHeader->fTrie,
fHeader->fTrieLen,
&status);
if (U_FAILURE(status)) {
return;
}
fTrie.getFoldingOffset=getFoldingOffset;
fRuleSource = (UChar *)((char *)data + fHeader->fRuleSource);
fRuleString.setTo(TRUE, fRuleSource, -1);
U_ASSERT(data->fRuleSourceLen > 0);
fRuleStatusTable = (int32_t *)((char *)data + fHeader->fStatusTable);
fStatusMaxIdx = data->fStatusTableLen / sizeof(int32_t);
fRefCount = 1;
#ifdef RBBI_DEBUG
char *debugEnv = getenv("U_RBBIDEBUG");
if (debugEnv && uprv_strstr(debugEnv, "data")) {this->printData();}
#endif
}
void RBBIDataWrapper::init(const RBBIDataHeader *data, UErrorCode &status) {
if (U_FAILURE(status)) {
return;
}
fHeader = data;
if (fHeader->fMagic != 0xb1a0 || fHeader->fFormatVersion[0] != 3)
{
status = U_INVALID_FORMAT_ERROR;
return;
}
// Note: in ICU version 3.2 and earlier, there was a formatVersion 1
// that is no longer supported. At that time fFormatVersion was
// an int32_t field, rather than an array of 4 bytes.
fDontFreeData = FALSE;
if (data->fFTableLen != 0) {
fForwardTable = (RBBIStateTable *)((char *)data + fHeader->fFTable);
}
if (data->fRTableLen != 0) {
fReverseTable = (RBBIStateTable *)((char *)data + fHeader->fRTable);
}
if (data->fSFTableLen != 0) {
fSafeFwdTable = (RBBIStateTable *)((char *)data + fHeader->fSFTable);
}
if (data->fSRTableLen != 0) {
fSafeRevTable = (RBBIStateTable *)((char *)data + fHeader->fSRTable);
}
utrie_unserialize(&fTrie,
(uint8_t *)data + fHeader->fTrie,
fHeader->fTrieLen,
&status);
if (U_FAILURE(status)) {
return;
}
fTrie.getFoldingOffset=getFoldingOffset;
fRuleSource = (UChar *)((char *)data + fHeader->fRuleSource);
fRuleString.setTo(TRUE, fRuleSource, -1);
U_ASSERT(data->fRuleSourceLen > 0);
fRuleStatusTable = (int32_t *)((char *)data + fHeader->fStatusTable);
fStatusMaxIdx = data->fStatusTableLen / sizeof(int32_t);
fRefCount = 1;
#ifdef RBBI_DEBUG
char *debugEnv = getenv("U_RBBIDEBUG");
if (debugEnv && uprv_strstr(debugEnv, "data")) {this->printData();}
#endif
}
void BreakIterator::init(const RBBIDataHeader* data, UErrorCode &status)
{
if (U_FAILURE(status)) {
return;
}
fHeader = data;
if (fHeader->fMagic != 0xb1a0 ||
!(fHeader->fFormatVersion[0] == 3 || // ICU 3.4
*(int32_t *)fHeader->fFormatVersion == 1)) // ICU 3.2 and earlier.
{
status = U_INVALID_FORMAT_ERROR;
return;
}
fReverseTable = NULL;
fSafeFwdTable = NULL;
fSafeRevTable = NULL;
if (data->fFTableLen != 0) {
fForwardTable = (RBBIStateTable *)((char *)data + fHeader->fFTable);
}
if (data->fRTableLen != 0) {
fReverseTable = (RBBIStateTable *)((char *)data + fHeader->fRTable);
}
if (data->fSFTableLen != 0) {
fSafeFwdTable = (RBBIStateTable *)((char *)data + fHeader->fSFTable);
}
if (data->fSRTableLen != 0) {
fSafeRevTable = (RBBIStateTable *)((char *)data + fHeader->fSRTable);
}
utrie_unserialize(&fTrie,
(uint8_t *)data + fHeader->fTrie,
fHeader->fTrieLen,
&status);
if (U_FAILURE(status)) {
return;
}
fTrie.getFoldingOffset=getFoldingOffset;
fRuleSource = (UChar *)((char *)data + fHeader->fRuleSource);
fRuleString.setTo(TRUE, fRuleSource, -1);
U_ASSERT(data->fRuleSourceLen > 0);
fRuleStatusTable = (int32_t *)((char *)data + fHeader->fStatusTable);
fStatusMaxIdx = data->fStatusTableLen / sizeof(int32_t);
}
//-----------------------------------------------------------------------------
//
// init(). Does most of the work of construction, shared between the
// constructors.
//
//-----------------------------------------------------------------------------
void RBBIDataWrapper::init(const RBBIDataHeader *data, UErrorCode &status) {
if (U_FAILURE(status)) {
return;
}
fHeader = data;
if (fHeader->fMagic != 0xb1a0) {
status = U_BRK_INTERNAL_ERROR;
return;
}
fUDataMem = NULL;
fForwardTable = (RBBIStateTable *)((char *)data + fHeader->fFTable);
fReverseTable = NULL;
if (data->fRTableLen != 0) {
fReverseTable = (RBBIStateTable *)((char *)data + fHeader->fRTable);
}
utrie_unserialize(&fTrie,
(uint8_t *)data + fHeader->fTrie,
fHeader->fTrieLen,
&status);
if (U_FAILURE(status)) {
return;
}
fTrie.getFoldingOffset=getFoldingOffset;
fRuleSource = (UChar *)((char *)data + fHeader->fRuleSource);
fRuleString.setTo(TRUE, fRuleSource, -1);
fRefCount = 1;
#ifdef RBBI_DEBUG
char *debugEnv = getenv("U_RBBIDEBUG");
if (debugEnv && uprv_strstr(debugEnv, "data")) {this->printData();}
#endif
}
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);
}
static UBool U_CALLCONV
loadData(UStringPrepProfile* profile,
const char* path,
const char* name,
const char* type,
UErrorCode* errorCode) {
/* load Unicode SPREP data from file */
UTrie _sprepTrie={ 0,0,0,0,0,0,0 };
UDataMemory *dataMemory;
const int32_t *p=NULL;
const uint8_t *pb;
UVersionInfo normUnicodeVersion;
int32_t normUniVer, sprepUniVer, normCorrVer;
if(errorCode==NULL || U_FAILURE(*errorCode)) {
return 0;
}
/* open the data outside the mutex block */
//TODO: change the path
dataMemory=udata_openChoice(path, type, name, isSPrepAcceptable, NULL, errorCode);
if(U_FAILURE(*errorCode)) {
return FALSE;
}
p=(const int32_t *)udata_getMemory(dataMemory);
pb=(const uint8_t *)(p+_SPREP_INDEX_TOP);
utrie_unserialize(&_sprepTrie, pb, p[_SPREP_INDEX_TRIE_SIZE], errorCode);
_sprepTrie.getFoldingOffset=getSPrepFoldingOffset;
if(U_FAILURE(*errorCode)) {
udata_close(dataMemory);
return FALSE;
}
/* in the mutex block, set the data for this process */
umtx_lock(usprepMutex());
if(profile->sprepData==NULL) {
profile->sprepData=dataMemory;
dataMemory=NULL;
uprv_memcpy(&profile->indexes, p, sizeof(profile->indexes));
uprv_memcpy(&profile->sprepTrie, &_sprepTrie, sizeof(UTrie));
} else {
p=(const int32_t *)udata_getMemory(profile->sprepData);
}
umtx_unlock(usprepMutex());
/* initialize some variables */
profile->mappingData=(uint16_t *)((uint8_t *)(p+_SPREP_INDEX_TOP)+profile->indexes[_SPREP_INDEX_TRIE_SIZE]);
u_getUnicodeVersion(normUnicodeVersion);
normUniVer = (normUnicodeVersion[0] << 24) + (normUnicodeVersion[1] << 16) +
(normUnicodeVersion[2] << 8 ) + (normUnicodeVersion[3]);
sprepUniVer = (dataVersion[0] << 24) + (dataVersion[1] << 16) +
(dataVersion[2] << 8 ) + (dataVersion[3]);
normCorrVer = profile->indexes[_SPREP_NORM_CORRECTNS_LAST_UNI_VERSION];
if(U_FAILURE(*errorCode)){
udata_close(dataMemory);
return FALSE;
}
if( normUniVer < sprepUniVer && /* the Unicode version of SPREP file must be less than the Unicode Vesion of the normalization data */
normUniVer < normCorrVer && /* the Unicode version of the NormalizationCorrections.txt file should be less than the Unicode Vesion of the normalization data */
((profile->indexes[_SPREP_OPTIONS] & _SPREP_NORMALIZATION_ON) > 0) /* normalization turned on*/
){
*errorCode = U_INVALID_FORMAT_ERROR;
udata_close(dataMemory);
return FALSE;
}
profile->isDataLoaded = TRUE;
/* if a different thread set it first, then close the extra data */
if(dataMemory!=NULL) {
udata_close(dataMemory); /* NULL if it was set correctly */
}
return profile->isDataLoaded;
}
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);
}
}
U_CDECL_END
#if !UNORM_HARDCODE_DATA
static int8_t
loadNormData(UErrorCode &errorCode) {
/* load Unicode normalization data from file */
/*
* This lazy intialization with double-checked locking (without mutex protection for
* haveNormData==0) is transiently unsafe under certain circumstances.
* Check the readme and use u_init() if necessary.
*
* While u_init() initializes the main normalization data via this functions,
* it does not do so for exclusion sets (which are fully mutexed).
* This is because
* - there can be many exclusion sets
* - they are rarely used
* - they are not usually used in execution paths that are
* as performance-sensitive as others
* (e.g., IDNA takes more time than unorm_quickCheck() anyway)
*
* TODO: Remove code in support for non-hardcoded data. u_init() is now advertised
* as not being required for thread safety, and we can't reasonably
* revert to requiring it.
*/
if(haveNormData==0) {
UTrie _normTrie={ 0,0,0,0,0,0,0 }, _fcdTrie={ 0,0,0,0,0,0,0 }, _auxTrie={ 0,0,0,0,0,0,0 };
UDataMemory *data;
const int32_t *p=NULL;
const uint8_t *pb;
if(&errorCode==NULL || U_FAILURE(errorCode)) {
return 0;
}
/* open the data outside the mutex block */
data=udata_openChoice(NULL, DATA_TYPE, DATA_NAME, isAcceptable, NULL, &errorCode);
dataErrorCode=errorCode;
if(U_FAILURE(errorCode)) {
return haveNormData=-1;
}
p=(const int32_t *)udata_getMemory(data);
pb=(const uint8_t *)(p+_NORM_INDEX_TOP);
utrie_unserialize(&_normTrie, pb, p[_NORM_INDEX_TRIE_SIZE], &errorCode);
_normTrie.getFoldingOffset=getFoldingNormOffset;
pb+=p[_NORM_INDEX_TRIE_SIZE]+p[_NORM_INDEX_UCHAR_COUNT]*2+p[_NORM_INDEX_COMBINE_DATA_COUNT]*2;
if(p[_NORM_INDEX_FCD_TRIE_SIZE]!=0) {
utrie_unserialize(&_fcdTrie, pb, p[_NORM_INDEX_FCD_TRIE_SIZE], &errorCode);
}
pb+=p[_NORM_INDEX_FCD_TRIE_SIZE];
if(p[_NORM_INDEX_AUX_TRIE_SIZE]!=0) {
utrie_unserialize(&_auxTrie, pb, p[_NORM_INDEX_AUX_TRIE_SIZE], &errorCode);
_auxTrie.getFoldingOffset=getFoldingAuxOffset;
}
if(U_FAILURE(errorCode)) {
dataErrorCode=errorCode;
udata_close(data);
return haveNormData=-1;
}
/* in the mutex block, set the data for this process */
umtx_lock(NULL);
if(normData==NULL) {
normData=data;
data=NULL;
uprv_memcpy(&indexes, p, sizeof(indexes));
uprv_memcpy(&normTrie, &_normTrie, sizeof(UTrie));
uprv_memcpy(&fcdTrie, &_fcdTrie, sizeof(UTrie));
uprv_memcpy(&auxTrie, &_auxTrie, sizeof(UTrie));
} else {
p=(const int32_t *)udata_getMemory(normData);
}
/* initialize some variables */
extraData=(uint16_t *)((uint8_t *)(p+_NORM_INDEX_TOP)+indexes[_NORM_INDEX_TRIE_SIZE]);
combiningTable=extraData+indexes[_NORM_INDEX_UCHAR_COUNT];
formatVersion_2_1=formatVersion[0]>2 || (formatVersion[0]==2 && formatVersion[1]>=1);
formatVersion_2_2=formatVersion[0]>2 || (formatVersion[0]==2 && formatVersion[1]>=2);
if(formatVersion_2_1) {
canonStartSets=combiningTable+
indexes[_NORM_INDEX_COMBINE_DATA_COUNT]+
(indexes[_NORM_INDEX_FCD_TRIE_SIZE]+indexes[_NORM_INDEX_AUX_TRIE_SIZE])/2;
}
haveNormData=1;
ucln_common_registerCleanup(UCLN_COMMON_UNORM, unorm_cleanup);
umtx_unlock(NULL);
/* if a different thread set it first, then close the extra data */
if(data!=NULL) {
udata_close(data); /* NULL if it was set correctly */
}
}
return haveNormData;
}
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;
}
static void
testTrieRanges(const char *testName,
const SetRange setRanges[], int32_t countSetRanges,
const CheckRange checkRanges[], int32_t countCheckRanges,
UBool dataIs32, UBool latin1Linear) {
union{
double bogus; /* needed for aligining the storage */
uint8_t storage[32768];
} storageHolder;
UTrieGetFoldingOffset *getFoldingOffset;
UNewTrieGetFoldedValue *getFoldedValue;
const CheckRange *enumRanges;
UNewTrie *newTrie;
UTrie trie={ 0 };
uint32_t value, value2;
UChar32 start, limit;
int32_t i, length;
UErrorCode errorCode;
UBool overwrite, ok;
log_verbose("\ntesting Trie '%s'\n", testName);
newTrie=utrie_open(NULL, NULL, 2000,
checkRanges[0].value, checkRanges[0].value,
latin1Linear);
/* set values from setRanges[] */
ok=TRUE;
for(i=0; i<countSetRanges; ++i) {
start=setRanges[i].start;
limit=setRanges[i].limit;
value=setRanges[i].value;
overwrite=setRanges[i].overwrite;
if((limit-start)==1 && overwrite) {
ok&=utrie_set32(newTrie, start, value);
} else {
ok&=utrie_setRange32(newTrie, start, limit, value, overwrite);
}
}
if(!ok) {
log_err("error: setting values into a trie failed (%s)\n", testName);
return;
}
/* verify that all these values are in the new Trie */
start=0;
for(i=0; i<countCheckRanges; ++i) {
limit=checkRanges[i].limit;
value=checkRanges[i].value;
while(start<limit) {
if(value!=utrie_get32(newTrie, start, NULL)) {
log_err("error: newTrie(%s)[U+%04lx]==0x%lx instead of 0x%lx\n",
testName, start, utrie_get32(newTrie, start, NULL), value);
}
++start;
}
}
if(dataIs32) {
getFoldingOffset=_testFoldingOffset32;
getFoldedValue=_testFoldedValue32;
} else {
getFoldingOffset=_testFoldingOffset16;
getFoldedValue=_testFoldedValue16;
}
/*
* code coverage for utrie.c/defaultGetFoldedValue(),
* pick some combination of parameters for selecting the UTrie defaults
*/
if(!dataIs32 && latin1Linear) {
getFoldingOffset=NULL;
getFoldedValue=NULL;
}
errorCode=U_ZERO_ERROR;
length=utrie_serialize(newTrie, storageHolder.storage, sizeof(storageHolder.storage),
getFoldedValue,
(UBool)!dataIs32,
&errorCode);
if(U_FAILURE(errorCode)) {
log_err("error: utrie_serialize(%s) failed: %s\n", testName, u_errorName(errorCode));
utrie_close(newTrie);
return;
}
if (length >= (int32_t)sizeof(storageHolder.storage)) {
log_err("error: utrie_serialize(%s) needs more memory\n", testName);
utrie_close(newTrie);
return;
}
/* test linear Latin-1 range from utrie_getData() */
if(latin1Linear) {
uint32_t *data;
int32_t dataLength;
data=utrie_getData(newTrie, &dataLength);
start=0;
for(i=0; i<countCheckRanges && start<=0xff; ++i) {
limit=checkRanges[i].limit;
value=checkRanges[i].value;
while(start<limit && start<=0xff) {
if(value!=data[UTRIE_DATA_BLOCK_LENGTH+start]) {
log_err("error: newTrie(%s).latin1Data[U+%04lx]==0x%lx instead of 0x%lx\n",
testName, start, data[UTRIE_DATA_BLOCK_LENGTH+start], value);
}
++start;
}
}
}
utrie_close(newTrie);
errorCode=U_ZERO_ERROR;
if(!utrie_unserialize(&trie, storageHolder.storage, length, &errorCode)) {
log_err("error: utrie_unserialize() failed, %s\n", u_errorName(errorCode));
return;
}
if(getFoldingOffset!=NULL) {
trie.getFoldingOffset=getFoldingOffset;
}
if(dataIs32!=(trie.data32!=NULL)) {
log_err("error: trie serialization (%s) did not preserve 32-bitness\n", testName);
}
if(latin1Linear!=trie.isLatin1Linear) {
log_err("error: trie serialization (%s) did not preserve Latin-1-linearity\n", testName);
}
/* verify that all these values are in the unserialized Trie */
start=0;
for(i=0; i<countCheckRanges; ++i) {
limit=checkRanges[i].limit;
value=checkRanges[i].value;
if(start==0xd800) {
/* skip surrogates */
start=limit;
continue;
}
while(start<limit) {
if(start<=0xffff) {
if(dataIs32) {
value2=UTRIE_GET32_FROM_BMP(&trie, start);
} else {
value2=UTRIE_GET16_FROM_BMP(&trie, start);
}
if(value!=value2) {
log_err("error: unserialized trie(%s).fromBMP(U+%04lx)==0x%lx instead of 0x%lx\n",
testName, start, value2, value);
}
if(!U16_IS_LEAD(start)) {
if(dataIs32) {
value2=UTRIE_GET32_FROM_LEAD(&trie, start);
} else {
value2=UTRIE_GET16_FROM_LEAD(&trie, start);
}
if(value!=value2) {
log_err("error: unserialized trie(%s).fromLead(U+%04lx)==0x%lx instead of 0x%lx\n",
testName, start, value2, value);
}
}
}
if(dataIs32) {
UTRIE_GET32(&trie, start, value2);
} else {
UTRIE_GET16(&trie, start, value2);
}
if(value!=value2) {
log_err("error: unserialized trie(%s).get(U+%04lx)==0x%lx instead of 0x%lx\n",
testName, start, value2, value);
}
++start;
}
}
/* enumerate and verify all ranges */
enumRanges=checkRanges+1;
utrie_enum(&trie, _testEnumValue, _testEnumRange, &enumRanges);
/* test linear Latin-1 range */
if(trie.isLatin1Linear) {
if(trie.data32!=NULL) {
const uint32_t *latin1=UTRIE_GET32_LATIN1(&trie);
for(start=0; start<0x100; ++start) {
if(latin1[start]!=UTRIE_GET32_FROM_LEAD(&trie, start)) {
log_err("error: (%s) trie.latin1[U+%04lx]=0x%lx!=0x%lx=trie.get32(U+%04lx)\n",
testName, start, latin1[start], UTRIE_GET32_FROM_LEAD(&trie, start), start);
}
}
} else {
const uint16_t *latin1=UTRIE_GET16_LATIN1(&trie);
for(start=0; start<0x100; ++start) {
if(latin1[start]!=UTRIE_GET16_FROM_LEAD(&trie, start)) {
log_err("error: (%s) trie.latin1[U+%04lx]=0x%lx!=0x%lx=trie.get16(U+%04lx)\n",
testName, start, latin1[start], UTRIE_GET16_FROM_LEAD(&trie, start), start);
}
}
}
}
testTrieIteration(testName, &trie, checkRanges, countCheckRanges);
}
