CString TextCodecMac::encode(const UChar* characters, size_t length, UnencodableHandling handling)
{
// FIXME: We should really use TEC here instead of CFString for consistency with the other direction.
// FIXME: Since there's no "force ASCII range" mode in CFString, we change the backslash into a yen sign.
// Encoding will change the yen sign back into a backslash.
String copy(characters, length);
copy.replace('\\', m_backslashAsCurrencySymbol);
CFStringRef cfs = copy.createCFString();
CFIndex startPos = 0;
CFIndex charactersLeft = CFStringGetLength(cfs);
Vector<char> result;
size_t size = 0;
UInt8 lossByte = handling == QuestionMarksForUnencodables ? '?' : 0;
while (charactersLeft > 0) {
CFRange range = CFRangeMake(startPos, charactersLeft);
CFIndex bufferLength;
CFStringGetBytes(cfs, range, m_encoding, lossByte, false, NULL, 0x7FFFFFFF, &bufferLength);
result.grow(size + bufferLength);
unsigned char* buffer = reinterpret_cast<unsigned char*>(result.data() + size);
CFIndex charactersConverted = CFStringGetBytes(cfs, range, m_encoding, lossByte, false, buffer, bufferLength, &bufferLength);
size += bufferLength;
if (charactersConverted != charactersLeft) {
unsigned badChar = CFStringGetCharacterAtIndex(cfs, startPos + charactersConverted);
++charactersConverted;
if ((badChar & 0xFC00) == 0xD800 && charactersConverted != charactersLeft) { // is high surrogate
UniChar low = CFStringGetCharacterAtIndex(cfs, startPos + charactersConverted);
if ((low & 0xFC00) == 0xDC00) { // is low surrogate
badChar <<= 10;
badChar += low;
badChar += 0x10000 - (0xD800 << 10) - 0xDC00;
++charactersConverted;
}
}
UnencodableReplacementArray entity;
int entityLength = getUnencodableReplacement(badChar, handling, entity);
result.grow(size + entityLength);
memcpy(result.data() + size, entity, entityLength);
size += entityLength;
}
startPos += charactersConverted;
charactersLeft -= charactersConverted;
}
CFRelease(cfs);
return CString(result.data(), size);
}
CString TextCodecQt::encode(const UChar* characters, size_t length, UnencodableHandling handling)
{
QTextCodec::ConverterState state;
state.flags = QTextCodec::ConversionFlags(QTextCodec::ConvertInvalidToNull | QTextCodec::IgnoreHeader);
if (!length)
return "";
QByteArray ba = m_codec->fromUnicode(reinterpret_cast<const QChar*>(characters), length, &state);
// If some <b> characters </b> are unencodable, escape them as specified by <b> handling </b>
// We append one valid encoded chunk to a QByteArray at a time. When we encounter an unencodable chunk we
// escape it with getUnencodableReplacement, append it, then move to the next chunk.
if (state.invalidChars) {
state.invalidChars = 0;
state.remainingChars = 0;
int len = 0;
ba.clear();
for (size_t pos = 0; pos < length; ++pos) {
QByteArray tba = m_codec->fromUnicode(reinterpret_cast<const QChar*>(characters), ++len, &state);
if (state.remainingChars)
continue;
if (state.invalidChars) {
UnencodableReplacementArray replacement;
getUnencodableReplacement(characters[0], handling, replacement);
tba.replace('\0', replacement);
state.invalidChars = 0;
}
ba.append(tba);
characters += len;
len = 0;
state.remainingChars = 0;
}
}
return CString(ba.constData(), ba.length());
}
