void ScriptSet::setScriptExtensions(UChar32 codePoint, UErrorCode& status) {
if (U_FAILURE(status)) { return; }
static const int32_t FIRST_GUESS_SCRIPT_CAPACITY = 5;
MaybeStackArray<UScriptCode,FIRST_GUESS_SCRIPT_CAPACITY> scripts;
UErrorCode internalStatus = U_ZERO_ERROR;
int32_t script_count = -1;
while (TRUE) {
script_count = uscript_getScriptExtensions(
codePoint, scripts.getAlias(), FIRST_GUESS_SCRIPT_CAPACITY, &internalStatus);
if (internalStatus == U_BUFFER_OVERFLOW_ERROR) {
// Need to allocate more space
if (scripts.resize(script_count) == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
return;
}
internalStatus = U_ZERO_ERROR;
} else {
break;
}
}
// Check if we failed for some reason other than buffer overflow
if (U_FAILURE(internalStatus)) {
status = internalStatus;
return;
}
// Load the scripts into the ScriptSet and return
for (int32_t i = 0; i < script_count; i++) {
this->set(scripts[i], status);
if (U_FAILURE(status)) { return; }
}
}
IdentifierInfo &IdentifierInfo::setIdentifier(const UnicodeString &identifier, UErrorCode &status) {
if (U_FAILURE(status)) {
return *this;
}
*fIdentifier = identifier;
clear();
ScriptSet scriptsForCP;
UChar32 cp;
for (int32_t i = 0; i < identifier.length(); i += U16_LENGTH(cp)) {
cp = identifier.char32At(i);
// Store a representative character for each kind of decimal digit
if (u_charType(cp) == U_DECIMAL_DIGIT_NUMBER) {
// Just store the zero character as a representative for comparison. Unicode guarantees it is cp - value
fNumerics->add(cp - (UChar32)u_getNumericValue(cp));
}
UScriptCode extensions[500];
int32_t extensionsCount = uscript_getScriptExtensions(cp, extensions, UPRV_LENGTHOF(extensions), &status);
if (U_FAILURE(status)) {
return *this;
}
scriptsForCP.resetAll();
for (int32_t j=0; j<extensionsCount; j++) {
scriptsForCP.set(extensions[j], status);
}
scriptsForCP.reset(USCRIPT_COMMON, status);
scriptsForCP.reset(USCRIPT_INHERITED, status);
switch (scriptsForCP.countMembers()) {
case 0: break;
case 1:
// Single script, record it.
fRequiredScripts->Union(scriptsForCP);
break;
default:
if (!fRequiredScripts->intersects(scriptsForCP)
&& !uhash_geti(fScriptSetSet, &scriptsForCP)) {
// If the set hasn't been added already, add it
// (Add a copy, fScriptSetSet takes ownership of the copy.)
uhash_puti(fScriptSetSet, new ScriptSet(scriptsForCP), 1, &status);
}
break;
}
}
// Now make a final pass through ScriptSetSet to remove alternates that came before singles.
// [Kana], [Kana Hira] => [Kana]
// This is relatively infrequent, so doesn't have to be optimized.
// We also compute any commonalities among the alternates.
if (uhash_count(fScriptSetSet) > 0) {
fCommonAmongAlternates->setAll();
for (int32_t it = UHASH_FIRST;;) {
const UHashElement *nextHashEl = uhash_nextElement(fScriptSetSet, &it);
if (nextHashEl == NULL) {
break;
}
ScriptSet *next = static_cast<ScriptSet *>(nextHashEl->key.pointer);
// [Kana], [Kana Hira] => [Kana]
if (fRequiredScripts->intersects(*next)) {
uhash_removeElement(fScriptSetSet, nextHashEl);
} else {
fCommonAmongAlternates->intersect(*next);
// [[Arab Syrc Thaa]; [Arab Syrc]] => [[Arab Syrc]]
for (int32_t otherIt = UHASH_FIRST;;) {
const UHashElement *otherHashEl = uhash_nextElement(fScriptSetSet, &otherIt);
if (otherHashEl == NULL) {
break;
}
ScriptSet *other = static_cast<ScriptSet *>(otherHashEl->key.pointer);
if (next != other && next->contains(*other)) {
uhash_removeElement(fScriptSetSet, nextHashEl);
break;
}
}
}
}
}
if (uhash_count(fScriptSetSet) == 0) {
fCommonAmongAlternates->resetAll();
}
return *this;
}
void TestGetScriptExtensions() {
UScriptCode scripts[20];
int32_t length;
UErrorCode errorCode;
/* errors and overflows */
errorCode=U_PARSE_ERROR;
length=uscript_getScriptExtensions(0x0640, scripts, UPRV_LENGTHOF(scripts), &errorCode);
if(errorCode!=U_PARSE_ERROR) {
log_err("uscript_getScriptExtensions(U+0640, U_PARSE_ERROR) did not preserve the UErrorCode - %s\n",
u_errorName(errorCode));
}
errorCode=U_ZERO_ERROR;
length=uscript_getScriptExtensions(0x0640, NULL, UPRV_LENGTHOF(scripts), &errorCode);
if(errorCode!=U_ILLEGAL_ARGUMENT_ERROR) {
log_err("uscript_getScriptExtensions(U+0640, NULL) did not set U_ILLEGAL_ARGUMENT_ERROR - %s\n",
u_errorName(errorCode));
}
errorCode=U_ZERO_ERROR;
length=uscript_getScriptExtensions(0x0640, scripts, -1, &errorCode);
if(errorCode!=U_ILLEGAL_ARGUMENT_ERROR) {
log_err("uscript_getScriptExtensions(U+0640, capacity<0) did not set U_ILLEGAL_ARGUMENT_ERROR - %s\n",
u_errorName(errorCode));
}
errorCode=U_ZERO_ERROR;
length=uscript_getScriptExtensions(0x0640, scripts, 0, &errorCode);
if(errorCode!=U_BUFFER_OVERFLOW_ERROR || length<3) {
log_err("uscript_getScriptExtensions(U+0640, capacity=0: pure preflighting)=%d < 3 - %s\n",
(int)length, u_errorName(errorCode));
}
errorCode=U_ZERO_ERROR;
length=uscript_getScriptExtensions(0x0640, scripts, 1, &errorCode);
if(errorCode!=U_BUFFER_OVERFLOW_ERROR || length<3) {
log_err("uscript_getScriptExtensions(U+0640, capacity=1: preflighting)=%d < 3 - %s\n",
(int)length, u_errorName(errorCode));
}
/* U+063F has only a Script code, no Script_Extensions. */
errorCode=U_ZERO_ERROR;
length=uscript_getScriptExtensions(0x063f, scripts, 0, &errorCode);
if(errorCode!=U_BUFFER_OVERFLOW_ERROR || length!=1) {
log_err("uscript_getScriptExtensions(U+063F, capacity=0)=%d != 1 - %s\n",
(int)length, u_errorName(errorCode));
}
/* invalid code points */
errorCode=U_ZERO_ERROR;
length=uscript_getScriptExtensions(-1, scripts, UPRV_LENGTHOF(scripts), &errorCode);
if(U_FAILURE(errorCode) || length!=1 || scripts[0]!=USCRIPT_UNKNOWN) {
log_err("uscript_getScriptExtensions(-1)=%d does not return {UNKNOWN} - %s\n",
(int)length, u_errorName(errorCode));
}
errorCode=U_ZERO_ERROR;
length=uscript_getScriptExtensions(0x110000, scripts, UPRV_LENGTHOF(scripts), &errorCode);
if(U_FAILURE(errorCode) || length!=1 || scripts[0]!=USCRIPT_UNKNOWN) {
log_err("uscript_getScriptExtensions(0x110000)=%d does not return {UNKNOWN} - %s\n",
(int)length, u_errorName(errorCode));
}
/* normal usage */
errorCode=U_ZERO_ERROR;
length=uscript_getScriptExtensions(0x063f, scripts, 1, &errorCode);
if(U_FAILURE(errorCode) || length!=1 || scripts[0]!=USCRIPT_ARABIC) {
log_err("uscript_getScriptExtensions(U+063F, capacity=1)=%d does not return {ARABIC} - %s\n",
(int)length, u_errorName(errorCode));
}
errorCode=U_ZERO_ERROR;
length=uscript_getScriptExtensions(0x0640, scripts, UPRV_LENGTHOF(scripts), &errorCode);
if(U_FAILURE(errorCode) || length<3 ||
!scriptsContain(scripts, length, USCRIPT_ARABIC) ||
!scriptsContain(scripts, length, USCRIPT_SYRIAC) ||
!scriptsContain(scripts, length, USCRIPT_MANDAIC)) {
log_err("uscript_getScriptExtensions(U+0640)=%d failed - %s\n",
(int)length, u_errorName(errorCode));
}
errorCode=U_ZERO_ERROR;
length=uscript_getScriptExtensions(0xfdf2, scripts, UPRV_LENGTHOF(scripts), &errorCode);
if(U_FAILURE(errorCode) || length!=2 || scripts[0]!=USCRIPT_ARABIC || scripts[1]!=USCRIPT_THAANA) {
log_err("uscript_getScriptExtensions(U+FDF2)=%d failed - %s\n",
(int)length, u_errorName(errorCode));
}
errorCode=U_ZERO_ERROR;
length=uscript_getScriptExtensions(0xff65, scripts, UPRV_LENGTHOF(scripts), &errorCode);
if(U_FAILURE(errorCode) || length!=6 || scripts[0]!=USCRIPT_BOPOMOFO || scripts[5]!=USCRIPT_YI) {
log_err("uscript_getScriptExtensions(U+FF65)=%d failed - %s\n",
(int)length, u_errorName(errorCode));
}
}
