U_CAPI int U_EXPORT2
writePackageDatFile(const char *outFilename, const char *outComment, const char *sourcePath, const char *addList, Package *pkg, char outType) {
LocalPointer<Package> ownedPkg;
LocalPointer<Package> addListPkg;
if (pkg == NULL) {
ownedPkg.adoptInstead(new Package);
if(ownedPkg.isNull()) {
fprintf(stderr, "icupkg: not enough memory\n");
return U_MEMORY_ALLOCATION_ERROR;
}
pkg = ownedPkg.getAlias();
addListPkg.adoptInstead(readList(sourcePath, addList, TRUE, NULL));
if(addListPkg.isValid()) {
pkg->addItems(*addListPkg);
} else {
return U_ILLEGAL_ARGUMENT_ERROR;
}
}
pkg->writePackage(outFilename, outType, outComment);
return 0;
}
/* Process a file */
void
processFile(const char *filename, const char *cp,
const char *inputDir, const char *outputDir, const char *packageName,
SRBRoot *newPoolBundle,
UBool omitBinaryCollation, UErrorCode &status) {
LocalPointer<SRBRoot> data;
UCHARBUF *ucbuf = NULL;
char *rbname = NULL;
char *openFileName = NULL;
char *inputDirBuf = NULL;
char outputFileName[256];
int32_t dirlen = 0;
int32_t filelen = 0;
if (U_FAILURE(status)) {
return;
}
if(filename==NULL){
status=U_ILLEGAL_ARGUMENT_ERROR;
return;
}else{
filelen = (int32_t)uprv_strlen(filename);
}
if(inputDir == NULL) {
const char *filenameBegin = uprv_strrchr(filename, U_FILE_SEP_CHAR);
openFileName = (char *) uprv_malloc(dirlen + filelen + 2);
openFileName[0] = '\0';
if (filenameBegin != NULL) {
/*
* When a filename ../../../data/root.txt is specified,
* we presume that the input directory is ../../../data
* This is very important when the resource file includes
* another file, like UCARules.txt or thaidict.brk.
*/
int32_t filenameSize = (int32_t)(filenameBegin - filename + 1);
inputDirBuf = uprv_strncpy((char *)uprv_malloc(filenameSize), filename, filenameSize);
/* test for NULL */
if(inputDirBuf == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
goto finish;
}
inputDirBuf[filenameSize - 1] = 0;
inputDir = inputDirBuf;
dirlen = (int32_t)uprv_strlen(inputDir);
}
}else{
dirlen = (int32_t)uprv_strlen(inputDir);
if(inputDir[dirlen-1] != U_FILE_SEP_CHAR) {
openFileName = (char *) uprv_malloc(dirlen + filelen + 2);
/* test for NULL */
if(openFileName == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
goto finish;
}
openFileName[0] = '\0';
/*
* append the input dir to openFileName if the first char in
* filename is not file seperation char and the last char input directory is not '.'.
* This is to support :
* genrb -s. /home/icu/data
* genrb -s. icu/data
* The user cannot mix notations like
* genrb -s. /icu/data --- the absolute path specified. -s redundant
* user should use
* genrb -s. icu/data --- start from CWD and look in icu/data dir
*/
if( (filename[0] != U_FILE_SEP_CHAR) && (inputDir[dirlen-1] !='.')){
uprv_strcpy(openFileName, inputDir);
openFileName[dirlen] = U_FILE_SEP_CHAR;
}
openFileName[dirlen + 1] = '\0';
} else {
openFileName = (char *) uprv_malloc(dirlen + filelen + 1);
/* test for NULL */
if(openFileName == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
goto finish;
}
uprv_strcpy(openFileName, inputDir);
}
}
uprv_strcat(openFileName, filename);
ucbuf = ucbuf_open(openFileName, &cp,getShowWarning(),TRUE, &status);
if(status == U_FILE_ACCESS_ERROR) {
fprintf(stderr, "couldn't open file %s\n", openFileName == NULL ? filename : openFileName);
goto finish;
}
if (ucbuf == NULL || U_FAILURE(status)) {
fprintf(stderr, "An error occured processing file %s. Error: %s\n",
openFileName == NULL ? filename : openFileName, u_errorName(status));
goto finish;
}
/* auto detected popular encodings? */
if (cp!=NULL && isVerbose()) {
printf("autodetected encoding %s\n", cp);
}
/* Parse the data into an SRBRoot */
data.adoptInstead(parse(ucbuf, inputDir, outputDir, filename,
!omitBinaryCollation, options[NO_COLLATION_RULES].doesOccur, &status));
if (data.isNull() || U_FAILURE(status)) {
fprintf(stderr, "couldn't parse the file %s. Error:%s\n", filename, u_errorName(status));
goto finish;
}
if(options[WRITE_POOL_BUNDLE].doesOccur) {
data->fWritePoolBundle = newPoolBundle;
data->compactKeys(status);
int32_t newKeysLength;
const char *newKeys = data->getKeyBytes(&newKeysLength);
newPoolBundle->addKeyBytes(newKeys, newKeysLength, status);
if(U_FAILURE(status)) {
fprintf(stderr, "bundle_compactKeys(%s) or bundle_getKeyBytes() failed: %s\n",
filename, u_errorName(status));
goto finish;
}
/* count the number of just-added key strings */
for(const char *newKeysLimit = newKeys + newKeysLength; newKeys < newKeysLimit; ++newKeys) {
if(*newKeys == 0) {
++newPoolBundle->fKeysCount;
}
}
}
if(options[USE_POOL_BUNDLE].doesOccur) {
data->fUsePoolBundle = &poolBundle;
}
/* Determine the target rb filename */
rbname = make_res_filename(filename, outputDir, packageName, status);
if(U_FAILURE(status)) {
fprintf(stderr, "couldn't make the res fileName for bundle %s. Error:%s\n",
filename, u_errorName(status));
goto finish;
}
if(write_java== TRUE){
bundle_write_java(data.getAlias(), outputDir, outputEnc,
outputFileName, sizeof(outputFileName),
options[JAVA_PACKAGE].value, options[BUNDLE_NAME].value, &status);
}else if(write_xliff ==TRUE){
bundle_write_xml(data.getAlias(), outputDir, outputEnc,
filename, outputFileName, sizeof(outputFileName),
language, xliffOutputFileName, &status);
}else{
/* Write the data to the file */
data->write(outputDir, packageName, outputFileName, sizeof(outputFileName), status);
}
if (U_FAILURE(status)) {
fprintf(stderr, "couldn't write bundle %s. Error:%s\n", outputFileName, u_errorName(status));
}
finish:
if (inputDirBuf != NULL) {
uprv_free(inputDirBuf);
}
if (openFileName != NULL) {
uprv_free(openFileName);
}
if(ucbuf) {
ucbuf_close(ucbuf);
}
if (rbname) {
uprv_free(rbname);
}
}
NumberFormat*
NumberFormat::makeInstance(const Locale& desiredLocale,
UNumberFormatStyle style,
UBool mustBeDecimalFormat,
UErrorCode& status) {
if (U_FAILURE(status)) return NULL;
if (style < 0 || style >= UNUM_FORMAT_STYLE_COUNT) {
status = U_ILLEGAL_ARGUMENT_ERROR;
return NULL;
}
// Some styles are not supported. This is a result of merging
// the @draft ICU 4.2 NumberFormat::EStyles into the long-existing UNumberFormatStyle.
// Ticket #8503 is for reviewing/fixing/merging the two relevant implementations:
// this one and unum_open().
// The UNUM_PATTERN_ styles are not supported here
// because this method does not take a pattern string.
if (!isStyleSupported(style)) {
status = U_UNSUPPORTED_ERROR;
return NULL;
}
#if U_PLATFORM_USES_ONLY_WIN32_API
if (!mustBeDecimalFormat) {
char buffer[8];
int32_t count = desiredLocale.getKeywordValue("compat", buffer, sizeof(buffer), status);
// if the locale has "@compat=host", create a host-specific NumberFormat
if (U_SUCCESS(status) && count > 0 && uprv_strcmp(buffer, "host") == 0) {
Win32NumberFormat *f = NULL;
UBool curr = TRUE;
switch (style) {
case UNUM_DECIMAL:
curr = FALSE;
// fall-through
case UNUM_CURRENCY:
case UNUM_CURRENCY_ISO: // do not support plural formatting here
case UNUM_CURRENCY_PLURAL:
f = new Win32NumberFormat(desiredLocale, curr, status);
if (U_SUCCESS(status)) {
return f;
}
delete f;
break;
default:
break;
}
}
}
#endif
// Use numbering system cache hashtable
umtx_initOnce(gNSCacheInitOnce, &nscacheInit);
// Get cached numbering system
LocalPointer<NumberingSystem> ownedNs;
NumberingSystem *ns = NULL;
if (NumberingSystem_cache != NULL) {
// TODO: Bad hash key usage, see ticket #8504.
int32_t hashKey = desiredLocale.hashCode();
Mutex lock(&nscacheMutex);
ns = (NumberingSystem *)uhash_iget(NumberingSystem_cache, hashKey);
if (ns == NULL) {
ns = NumberingSystem::createInstance(desiredLocale,status);
uhash_iput(NumberingSystem_cache, hashKey, (void*)ns, &status);
}
} else {
ownedNs.adoptInstead(NumberingSystem::createInstance(desiredLocale,status));
ns = ownedNs.getAlias();
}
// check results of getting a numbering system
if (U_FAILURE(status)) {
return NULL;
}
if (mustBeDecimalFormat && ns->isAlgorithmic()) {
status = U_UNSUPPORTED_ERROR;
return NULL;
}
LocalPointer<DecimalFormatSymbols> symbolsToAdopt;
UnicodeString pattern;
LocalUResourceBundlePointer ownedResource(ures_open(NULL, desiredLocale.getName(), &status));
if (U_FAILURE(status)) {
// We don't appear to have resource data available -- use the last-resort data
status = U_USING_FALLBACK_WARNING;
// When the data is unavailable, and locale isn't passed in, last resort data is used.
symbolsToAdopt.adoptInstead(new DecimalFormatSymbols(status));
if (symbolsToAdopt.isNull()) {
status = U_MEMORY_ALLOCATION_ERROR;
return NULL;
}
// Creates a DecimalFormat instance with the last resort number patterns.
pattern.setTo(TRUE, gLastResortNumberPatterns[style], -1);
}
else {
// Loads the decimal symbols of the desired locale.
symbolsToAdopt.adoptInstead(new DecimalFormatSymbols(desiredLocale, status));
if (symbolsToAdopt.isNull()) {
status = U_MEMORY_ALLOCATION_ERROR;
return NULL;
}
UResourceBundle *resource = ownedResource.orphan();
UResourceBundle *numElements = ures_getByKeyWithFallback(resource, gNumberElements, NULL, &status);
resource = ures_getByKeyWithFallback(numElements, ns->getName(), resource, &status);
resource = ures_getByKeyWithFallback(resource, gPatterns, resource, &status);
ownedResource.adoptInstead(resource);
int32_t patLen = 0;
const UChar *patResStr = ures_getStringByKeyWithFallback(resource, gFormatKeys[style], &patLen, &status);
// Didn't find a pattern specific to the numbering system, so fall back to "latn"
if ( status == U_MISSING_RESOURCE_ERROR && uprv_strcmp(gLatn,ns->getName())) {
status = U_ZERO_ERROR;
resource = ures_getByKeyWithFallback(numElements, gLatn, resource, &status);
resource = ures_getByKeyWithFallback(resource, gPatterns, resource, &status);
patResStr = ures_getStringByKeyWithFallback(resource, gFormatKeys[style], &patLen, &status);
}
ures_close(numElements);
// Creates the specified decimal format style of the desired locale.
pattern.setTo(TRUE, patResStr, patLen);
}
if (U_FAILURE(status)) {
return NULL;
}
if(style==UNUM_CURRENCY || style == UNUM_CURRENCY_ISO){
const UChar* currPattern = symbolsToAdopt->getCurrencyPattern();
if(currPattern!=NULL){
pattern.setTo(currPattern, u_strlen(currPattern));
}
}
NumberFormat *f;
if (ns->isAlgorithmic()) {
UnicodeString nsDesc;
UnicodeString nsRuleSetGroup;
UnicodeString nsRuleSetName;
Locale nsLoc;
URBNFRuleSetTag desiredRulesType = URBNF_NUMBERING_SYSTEM;
nsDesc.setTo(ns->getDescription());
int32_t firstSlash = nsDesc.indexOf(gSlash);
int32_t lastSlash = nsDesc.lastIndexOf(gSlash);
if ( lastSlash > firstSlash ) {
CharString nsLocID;
nsLocID.appendInvariantChars(nsDesc.tempSubString(0, firstSlash), status);
nsRuleSetGroup.setTo(nsDesc,firstSlash+1,lastSlash-firstSlash-1);
nsRuleSetName.setTo(nsDesc,lastSlash+1);
nsLoc = Locale::createFromName(nsLocID.data());
UnicodeString SpelloutRules = UNICODE_STRING_SIMPLE("SpelloutRules");
if ( nsRuleSetGroup.compare(SpelloutRules) == 0 ) {
desiredRulesType = URBNF_SPELLOUT;
}
} else {
nsLoc = desiredLocale;
nsRuleSetName.setTo(nsDesc);
}
RuleBasedNumberFormat *r = new RuleBasedNumberFormat(desiredRulesType,nsLoc,status);
if (r == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
return NULL;
}
r->setDefaultRuleSet(nsRuleSetName,status);
f = r;
} else {
// replace single currency sign in the pattern with double currency sign
// if the style is UNUM_CURRENCY_ISO
if (style == UNUM_CURRENCY_ISO) {
pattern.findAndReplace(UnicodeString(TRUE, gSingleCurrencySign, 1),
UnicodeString(TRUE, gDoubleCurrencySign, 2));
}
// "new DecimalFormat()" does not adopt the symbols if its memory allocation fails.
DecimalFormatSymbols *syms = symbolsToAdopt.orphan();
f = new DecimalFormat(pattern, syms, style, status);
if (f == NULL) {
delete syms;
status = U_MEMORY_ALLOCATION_ERROR;
return NULL;
}
}
f->setLocaleIDs(ures_getLocaleByType(ownedResource.getAlias(), ULOC_VALID_LOCALE, &status),
ures_getLocaleByType(ownedResource.getAlias(), ULOC_ACTUAL_LOCALE, &status));
if (U_FAILURE(status)) {
delete f;
return NULL;
}
return f;
}
int32_t ULISentenceBreakIterator::next() {
int32_t n = fDelegate->next();
if(n == UBRK_DONE || // at end or
fBackwardsTrie.isNull()) { // .. no backwards table loaded == no exceptions
return n;
}
// OK, do we need to break here?
UErrorCode status = U_ZERO_ERROR;
// refresh text
fText.adoptInstead(fDelegate->getUText(fText.orphan(), status));
//if(debug2) u_printf("str, native len=%d\n", utext_nativeLength(fText.getAlias()));
do { // outer loop runs once per underlying break (from fDelegate).
// loops while 'n' points to an exception.
utext_setNativeIndex(fText.getAlias(), n); // from n..
fBackwardsTrie->reset();
UChar32 uch;
//if(debug2) u_printf(" n@ %d\n", n);
// Assume a space is following the '.' (so we handle the case: "Mr. /Brown")
if((uch=utext_previous32(fText.getAlias()))==(UChar32)0x0020) { // TODO: skip a class of chars here??
// TODO only do this the 1st time?
//if(debug2) u_printf("skipping prev: |%C| \n", (UChar)uch);
} else {
//if(debug2) u_printf("not skipping prev: |%C| \n", (UChar)uch);
uch = utext_next32(fText.getAlias());
//if(debug2) u_printf(" -> : |%C| \n", (UChar)uch);
}
UStringTrieResult r = USTRINGTRIE_INTERMEDIATE_VALUE;
int32_t bestPosn = -1;
int32_t bestValue = -1;
while((uch=utext_previous32(fText.getAlias()))!=U_SENTINEL && // more to consume backwards and..
USTRINGTRIE_HAS_NEXT(r=fBackwardsTrie->nextForCodePoint(uch))) {// more in the trie
if(USTRINGTRIE_HAS_VALUE(r)) { // remember the best match so far
bestPosn = utext_getNativeIndex(fText.getAlias());
bestValue = fBackwardsTrie->getValue();
}
//if(debug2) u_printf("rev< /%C/ cont?%d @%d\n", (UChar)uch, r, utext_getNativeIndex(fText.getAlias()));
}
if(USTRINGTRIE_MATCHES(r)) { // exact match?
//if(debug2) u_printf("rev<?/%C/?end of seq.. r=%d, bestPosn=%d, bestValue=%d\n", (UChar)uch, r, bestPosn, bestValue);
bestValue = fBackwardsTrie->getValue();
bestPosn = utext_getNativeIndex(fText.getAlias());
//if(debug2) u_printf("rev<+/%C/+end of seq.. r=%d, bestPosn=%d, bestValue=%d\n", (UChar)uch, r, bestPosn, bestValue);
}
if(bestPosn>=0) {
//if(debug2) u_printf("rev< /%C/ end of seq.. r=%d, bestPosn=%d, bestValue=%d\n", (UChar)uch, r, bestPosn, bestValue);
//if(USTRINGTRIE_MATCHES(r)) { // matched - so, now what?
//int32_t bestValue = fBackwardsTrie->getValue();
////if(debug2) u_printf("rev< /%C/ matched, skip..%d bestValue=%d\n", (UChar)uch, r, bestValue);
if(bestValue == kMATCH) { // exact match!
//if(debug2) u_printf(" exact backward match\n");
n = fDelegate->next(); // skip this one. Find the next lowerlevel break.
if(n==UBRK_DONE) return n;
continue; // See if the next is another exception.
} else if(bestValue == kPARTIAL
&& fForwardsPartialTrie.isValid()) { // make sure there's a forward trie
//if(debug2) u_printf(" partial backward match\n");
// We matched the "Ph." in "Ph.D." - now we need to run everything through the forwards trie
// to see if it matches something going forward.
fForwardsPartialTrie->reset();
UStringTrieResult rfwd = USTRINGTRIE_INTERMEDIATE_VALUE;
utext_setNativeIndex(fText.getAlias(), bestPosn); // hope that's close ..
//if(debug2) u_printf("Retrying at %d\n", bestPosn);
while((uch=utext_next32(fText.getAlias()))!=U_SENTINEL &&
USTRINGTRIE_HAS_NEXT(rfwd=fForwardsPartialTrie->nextForCodePoint(uch))) {
//if(debug2) u_printf("fwd> /%C/ cont?%d @%d\n", (UChar)uch, rfwd, utext_getNativeIndex(fText.getAlias()));
}
if(USTRINGTRIE_MATCHES(rfwd)) {
//if(debug2) u_printf("fwd> /%C/ == forward match!\n", (UChar)uch);
// only full matches here, nothing to check
// skip the next:
n = fDelegate->next();
if(n==UBRK_DONE) return n;
continue;
} else {
//if(debug2) u_printf("fwd> /%C/ no match.\n", (UChar)uch);
// no match (no exception) -return the 'underlying' break
return n;
}
} else {
return n; // internal error and/or no forwards trie
}
} else {
//if(debug2) u_printf("rev< /%C/ .. no match..%d\n", (UChar)uch, r); // no best match
return n; // No match - so exit. Not an exception.
}
} while(n != UBRK_DONE);
return n;
}