Variant HHVM_STATIC_METHOD(IntlChar, charFromName,
const String& name, int64_t choice) {
UErrorCode error = U_ZERO_ERROR;
auto ret = u_charFromName((UCharNameChoice)choice, name.c_str(), &error);
if (U_FAILURE(error)) {
s_intl_error->setError(error);
return init_null();
}
return ret;
}
int
getCharacterByName (wchar_t *character, const char *name) {
#ifdef HAVE_ICU
UErrorCode error = U_ZERO_ERROR;
UChar uc = u_charFromName(U_EXTENDED_CHAR_NAME, name, &error);
if (U_SUCCESS(error)) {
*character = uc;
return 1;
}
#endif /* HAVE_ICU */
return 0;
}
UnicodeSet&
UnicodeSet::applyPropertyAlias(const UnicodeString& prop,
const UnicodeString& value,
UErrorCode& ec) {
if (U_FAILURE(ec) || isFrozen()) return *this;
// prop and value used to be converted to char * using the default
// converter instead of the invariant conversion.
// This should not be necessary because all Unicode property and value
// names use only invariant characters.
// If there are any variant characters, then we won't find them anyway.
// Checking first avoids assertion failures in the conversion.
if( !uprv_isInvariantUString(prop.getBuffer(), prop.length()) ||
!uprv_isInvariantUString(value.getBuffer(), value.length())
) {
FAIL(ec);
}
CharString pname, vname;
pname.appendInvariantChars(prop, ec);
vname.appendInvariantChars(value, ec);
if (U_FAILURE(ec)) return *this;
UProperty p;
int32_t v;
UBool mustNotBeEmpty = FALSE, invert = FALSE;
if (value.length() > 0) {
p = u_getPropertyEnum(pname.data());
if (p == UCHAR_INVALID_CODE) FAIL(ec);
// Treat gc as gcm
if (p == UCHAR_GENERAL_CATEGORY) {
p = UCHAR_GENERAL_CATEGORY_MASK;
}
if ((p >= UCHAR_BINARY_START && p < UCHAR_BINARY_LIMIT) ||
(p >= UCHAR_INT_START && p < UCHAR_INT_LIMIT) ||
(p >= UCHAR_MASK_START && p < UCHAR_MASK_LIMIT)) {
v = u_getPropertyValueEnum(p, vname.data());
if (v == UCHAR_INVALID_CODE) {
// Handle numeric CCC
if (p == UCHAR_CANONICAL_COMBINING_CLASS ||
p == UCHAR_TRAIL_CANONICAL_COMBINING_CLASS ||
p == UCHAR_LEAD_CANONICAL_COMBINING_CLASS) {
char* end;
double value = uprv_strtod(vname.data(), &end);
v = (int32_t) value;
if (v != value || v < 0 || *end != 0) {
// non-integral or negative value, or trailing junk
FAIL(ec);
}
// If the resultant set is empty then the numeric value
// was invalid.
mustNotBeEmpty = TRUE;
} else {
FAIL(ec);
}
}
}
else {
switch (p) {
case UCHAR_NUMERIC_VALUE:
{
char* end;
double value = uprv_strtod(vname.data(), &end);
if (*end != 0) {
FAIL(ec);
}
applyFilter(numericValueFilter, &value, UPROPS_SRC_CHAR, ec);
return *this;
}
case UCHAR_NAME:
{
// Must munge name, since u_charFromName() does not do
// 'loose' matching.
char buf[128]; // it suffices that this be > uprv_getMaxCharNameLength
if (!mungeCharName(buf, vname.data(), sizeof(buf))) FAIL(ec);
UChar32 ch = u_charFromName(U_EXTENDED_CHAR_NAME, buf, &ec);
if (U_SUCCESS(ec)) {
clear();
add(ch);
return *this;
} else {
FAIL(ec);
}
}
case UCHAR_UNICODE_1_NAME:
// ICU 49 deprecates the Unicode_1_Name property APIs.
FAIL(ec);
case UCHAR_AGE:
{
// Must munge name, since u_versionFromString() does not do
// 'loose' matching.
char buf[128];
if (!mungeCharName(buf, vname.data(), sizeof(buf))) FAIL(ec);
UVersionInfo version;
u_versionFromString(version, buf);
applyFilter(versionFilter, &version, UPROPS_SRC_PROPSVEC, ec);
return *this;
}
case UCHAR_SCRIPT_EXTENSIONS:
v = u_getPropertyValueEnum(UCHAR_SCRIPT, vname.data());
if (v == UCHAR_INVALID_CODE) {
FAIL(ec);
}
// fall through to calling applyIntPropertyValue()
break;
default:
// p is a non-binary, non-enumerated property that we
// don't support (yet).
FAIL(ec);
}
}
}
else {
// value is empty. Interpret as General Category, Script, or
// Binary property.
p = UCHAR_GENERAL_CATEGORY_MASK;
v = u_getPropertyValueEnum(p, pname.data());
if (v == UCHAR_INVALID_CODE) {
p = UCHAR_SCRIPT;
v = u_getPropertyValueEnum(p, pname.data());
if (v == UCHAR_INVALID_CODE) {
p = u_getPropertyEnum(pname.data());
if (p >= UCHAR_BINARY_START && p < UCHAR_BINARY_LIMIT) {
v = 1;
} else if (0 == uprv_comparePropertyNames(ANY, pname.data())) {
set(MIN_VALUE, MAX_VALUE);
return *this;
} else if (0 == uprv_comparePropertyNames(ASCII, pname.data())) {
set(0, 0x7F);
return *this;
} else if (0 == uprv_comparePropertyNames(ASSIGNED, pname.data())) {
// [:Assigned:]=[:^Cn:]
p = UCHAR_GENERAL_CATEGORY_MASK;
v = U_GC_CN_MASK;
invert = TRUE;
} else {
FAIL(ec);
}
}
}
}
applyIntPropertyValue(p, v, ec);
if(invert) {
complement();
}
if (U_SUCCESS(ec) && (mustNotBeEmpty && isEmpty())) {
// mustNotBeEmpty is set to true if an empty set indicates
// invalid input.
ec = U_ILLEGAL_ARGUMENT_ERROR;
}
if (isBogus() && U_SUCCESS(ec)) {
// We likely ran out of memory. AHHH!
ec = U_MEMORY_ALLOCATION_ERROR;
}
return *this;
}
/**
* Implements {@link Transliterator#handleTransliterate}.
*/
void NameUnicodeTransliterator::handleTransliterate(Replaceable& text, UTransPosition& offsets,
UBool isIncremental) const {
// The failure mode, here and below, is to behave like Any-Null,
// if either there is no name data (max len == 0) or there is no
// memory (malloc() => NULL).
int32_t maxLen = uprv_getMaxCharNameLength();
if (maxLen == 0) {
offsets.start = offsets.limit;
return;
}
// Accomodate the longest possible name
++maxLen; // allow for temporary trailing space
char* cbuf = (char*) uprv_malloc(maxLen);
if (cbuf == NULL) {
offsets.start = offsets.limit;
return;
}
UnicodeString openPat(TRUE, OPEN, -1);
UnicodeString str, name;
int32_t cursor = offsets.start;
int32_t limit = offsets.limit;
// Modes:
// 0 - looking for open delimiter
// 1 - after open delimiter
int32_t mode = 0;
int32_t openPos = -1; // open delim candidate pos
UChar32 c;
while (cursor < limit) {
c = text.char32At(cursor);
switch (mode) {
case 0: // looking for open delimiter
if (c == OPEN_DELIM) { // quick check first
openPos = cursor;
int32_t i =
ICU_Utility::parsePattern(openPat, text, cursor, limit);
if (i >= 0 && i < limit) {
mode = 1;
name.truncate(0);
cursor = i;
continue; // *** reprocess char32At(cursor)
}
}
break;
case 1: // after open delimiter
// Look for legal chars. If \s+ is found, convert it
// to a single space. If closeDelimiter is found, exit
// the loop. If any other character is found, exit the
// loop. If the limit is reached, exit the loop.
// Convert \s+ => SPACE. This assumes there are no
// runs of >1 space characters in names.
if (PatternProps::isWhiteSpace(c)) {
// Ignore leading whitespace
if (name.length() > 0 &&
name.charAt(name.length()-1) != SPACE) {
name.append(SPACE);
// If we are too long then abort. maxLen includes
// temporary trailing space, so use '>'.
if (name.length() > maxLen) {
mode = 0;
}
}
break;
}
if (c == CLOSE_DELIM) {
int32_t len = name.length();
// Delete trailing space, if any
if (len > 0 &&
name.charAt(len-1) == SPACE) {
--len;
}
if (uprv_isInvariantUString(name.getBuffer(), len)) {
name.extract(0, len, cbuf, maxLen, US_INV);
UErrorCode status = U_ZERO_ERROR;
c = u_charFromName(U_EXTENDED_CHAR_NAME, cbuf, &status);
if (U_SUCCESS(status)) {
// Lookup succeeded
// assert(U16_LENGTH(CLOSE_DELIM) == 1);
cursor++; // advance over CLOSE_DELIM
str.truncate(0);
str.append(c);
text.handleReplaceBetween(openPos, cursor, str);
// Adjust indices for the change in the length of
// the string. Do not assume that str.length() ==
// 1, in case of surrogates.
int32_t delta = cursor - openPos - str.length();
cursor -= delta;
limit -= delta;
// assert(cursor == openPos + str.length());
}
}
// If the lookup failed, we leave things as-is and
// still switch to mode 0 and continue.
mode = 0;
openPos = -1; // close off candidate
continue; // *** reprocess char32At(cursor)
}
// Check if c is a legal char. We assume here that
// legal.contains(OPEN_DELIM) is FALSE, so when we abort a
// name, we don't have to go back to openPos+1.
if (legal.contains(c)) {
name.append(c);
// If we go past the longest possible name then abort.
// maxLen includes temporary trailing space, so use '>='.
if (name.length() >= maxLen) {
mode = 0;
}
}
// Invalid character
else {
--cursor; // Backup and reprocess this character
mode = 0;
}
break;
}
cursor += U16_LENGTH(c);
}
offsets.contextLimit += limit - offsets.limit;
offsets.limit = limit;
// In incremental mode, only advance the cursor up to the last
// open delimiter candidate.
offsets.start = (isIncremental && openPos >= 0) ? openPos : cursor;
uprv_free(cbuf);
}
UChar32 __hs_u_charFromName(UCharNameChoice nameChoice, const char *name,
UErrorCode *pErrorCode)
{
return u_charFromName(nameChoice, name, pErrorCode);
}
