static const char* getICUCanonicalName(const char* name) {
UErrorCode error = U_ZERO_ERROR;
const char* canonicalName = NULL;
if ((canonicalName = ucnv_getCanonicalName(name, "MIME", &error)) != NULL) {
return canonicalName;
} else if ((canonicalName = ucnv_getCanonicalName(name, "IANA", &error)) != NULL) {
return canonicalName;
} else if ((canonicalName = ucnv_getCanonicalName(name, "", &error)) != NULL) {
return canonicalName;
} else if ((canonicalName = ucnv_getAlias(name, 0, &error)) != NULL) {
// We have some aliases in the form x-blah .. match those first.
return canonicalName;
} else if (strstr(name, "x-") == name) {
// Check if the converter can be opened with the name given.
error = U_ZERO_ERROR;
icu::LocalUConverterPointer cnv(ucnv_open(name + 2, &error));
if (U_SUCCESS(error)) {
return name + 2;
}
}
return NULL;
}
void TextCodecICU::registerCodecs(TextCodecRegistrar registrar)
{
// See comment above in registerEncodingNames.
UErrorCode error = U_ZERO_ERROR;
const char* canonicalConverterName = ucnv_getCanonicalName("ISO-8859-8-I", "IANA", &error);
ASSERT(U_SUCCESS(error));
registrar("ISO-8859-8-I", create, canonicalConverterName);
int32_t numConverters = ucnv_countAvailable();
for (int32_t i = 0; i < numConverters; ++i) {
canonicalConverterName = ucnv_getAvailableName(i);
error = U_ZERO_ERROR;
const char* webStandardName = ucnv_getStandardName(canonicalConverterName, "MIME", &error);
if (!U_SUCCESS(error) || !webStandardName) {
error = U_ZERO_ERROR;
webStandardName = ucnv_getStandardName(canonicalConverterName, "IANA", &error);
if (!U_SUCCESS(error) || !webStandardName)
continue;
}
// Don't register codecs for overridden encodings.
if (strcmp(webStandardName, "GB2312") == 0 || strcmp(webStandardName, "GB_2312-80") == 0
|| strcmp(webStandardName, "KSC_5601") == 0 || strcmp(webStandardName, "EUC-KR") == 0
|| strcmp(webStandardName, "cp1363") == 0
|| strcasecmp(webStandardName, "iso-8859-9") == 0
|| strcmp(webStandardName, "TIS-620") == 0)
continue;
registrar(webStandardName, create, fastStrDup(canonicalConverterName));
}
// These encodings currently don't have standard names, so we need to register encoders manually.
// FIXME: Is there a good way to determine the most up to date variant programmatically?
registrar("windows-874", create, "windows-874-2000");
registrar("windows-949", create, "windows-949-2000");
}
