XMLSize_t
MacOSTranscoder::transcodeTo(const XMLCh* const srcData
, const XMLSize_t srcCount
, XMLByte* const toFill
, const XMLSize_t maxBytes
, XMLSize_t& charsEaten
, const UnRepOpts options)
{
// Reset the tec state (since we don't know that we're part of a
// larger run of text).
TECClearConverterContextInfo(mUnicodeToText);
// Do the conversion
ByteCount bytesConsumed = 0;
ByteCount bytesProduced = 0;
OSStatus status = TECConvertText(mUnicodeToText,
(ConstTextPtr) srcData,
srcCount * sizeof(XMLCh), // inputBufferLength
&bytesConsumed, // actualInputLength
(TextPtr) toFill, // outputBuffer
maxBytes, // outputBufferLength
&bytesProduced); // actualOutputLength
// Ignorable error codes
if( status == kTECUsedFallbacksStatus
|| status == kTECOutputBufferFullStatus
|| status == kTECPartialCharErr
)
status = noErr;
std::size_t charsConsumed = bytesConsumed / sizeof(XMLCh);
// Deal with errors
if (status != noErr)
{
if (status == kTECUnmappableElementErr && options == UnRep_Throw)
{
XMLCh tmpBuf[17];
XMLString::binToText(srcData[charsConsumed], tmpBuf, 16, 16);
ThrowXML2
(
TranscodingException
, XMLExcepts::Trans_Unrepresentable
, tmpBuf
, getEncodingName()
);
}
}
charsEaten = charsConsumed;
return bytesProduced;
}
OSStatus TextCodecMac::createTECConverter() const
{
bool cachedEncodingEqual = cachedConverterEncoding == m_encoding;
cachedConverterEncoding = invalidEncoding;
if (cachedEncodingEqual && cachedConverterTEC) {
m_converterTEC = cachedConverterTEC;
cachedConverterTEC = 0;
TECClearConverterContextInfo(m_converterTEC);
} else {
OSStatus status = TECCreateConverter(&m_converterTEC, m_encoding,
CreateTextEncoding(kTextEncodingUnicodeDefault, kTextEncodingDefaultVariant, kUnicode16BitFormat));
if (status)
return status;
TECSetBasicOptions(m_converterTEC, kUnicodeForceASCIIRangeMask);
}
return noErr;
}
XMLSize_t
MacOSTranscoder::transcodeFrom( const XMLByte* const srcData
, const XMLSize_t srcCount
, XMLCh* const toFill
, const XMLSize_t maxChars
, XMLSize_t& bytesEaten
, unsigned char* const charSizes)
{
// Reset the tec state (since we don't know that we're part of a
// larger run of text).
TECClearConverterContextInfo(mTextToUnicode);
// Do the conversion
ByteCount bytesConsumed = 0;
ByteCount bytesProduced = 0;
OSStatus status = TECConvertText(mTextToUnicode,
(ConstTextPtr) srcData,
srcCount, // inputBufferLength
&bytesConsumed, // actualInputLength
(TextPtr) toFill, // outputBuffer
maxChars * sizeof(XMLCh), // outputBufferLength
&bytesProduced); // actualOutputLength
// Ignorable error codes
if( status == kTECUsedFallbacksStatus
|| status == kTECOutputBufferFullStatus
|| status == kTECPartialCharErr
)
status = noErr;
if (status != noErr)
ThrowXML(TranscodingException, XMLExcepts::Trans_BadSrcSeq);
std::size_t charsProduced = bytesProduced / sizeof(XMLCh);
bytesEaten = bytesConsumed;
return charsProduced;
}
String TextCodecMac::decode(const char* bytes, size_t length, bool flush, bool stopOnError, bool& sawError)
{
// Get a converter for the passed-in encoding.
if (!m_converterTEC && createTECConverter() != noErr)
return String();
Vector<UChar> result;
const unsigned char* sourcePointer = reinterpret_cast<const unsigned char*>(bytes);
int sourceLength = length;
bool bufferWasFull = false;
UniChar buffer[ConversionBufferSize];
while ((sourceLength || bufferWasFull) && !sawError) {
int bytesRead = 0;
int bytesWritten = 0;
OSStatus status = decode(sourcePointer, sourceLength, bytesRead, buffer, sizeof(buffer), bytesWritten);
ASSERT(bytesRead <= sourceLength);
sourcePointer += bytesRead;
sourceLength -= bytesRead;
switch (status) {
case noErr:
case kTECOutputBufferFullStatus:
break;
case kTextMalformedInputErr:
case kTextUndefinedElementErr:
// FIXME: Put FFFD character into the output string in this case?
TECClearConverterContextInfo(m_converterTEC);
if (stopOnError) {
sawError = true;
break;
}
if (sourceLength) {
sourcePointer += 1;
sourceLength -= 1;
}
break;
case kTECPartialCharErr: {
// Put the partial character into the buffer.
ASSERT(m_numBufferedBytes == 0);
const int bufferSize = sizeof(m_numBufferedBytes);
if (sourceLength < bufferSize) {
memcpy(m_bufferedBytes, sourcePointer, sourceLength);
m_numBufferedBytes = sourceLength;
} else {
LOG_ERROR("TECConvertText gave a kTECPartialCharErr, but left %u bytes in the buffer", sourceLength);
}
sourceLength = 0;
break;
}
default:
sawError = true;
return String();
}
ASSERT(!(bytesWritten % sizeof(UChar)));
result.append(buffer, bytesWritten / sizeof(UChar));
bufferWasFull = status == kTECOutputBufferFullStatus;
}
if (flush) {
unsigned long bytesWritten = 0;
TECFlushText(m_converterTEC, reinterpret_cast<unsigned char*>(buffer), sizeof(buffer), &bytesWritten);
ASSERT(!(bytesWritten % sizeof(UChar)));
result.append(buffer, bytesWritten / sizeof(UChar));
}
String resultString = String::adopt(result);
// <rdar://problem/3225472>
// Simplified Chinese pages use the code A3A0 to mean "full-width space".
// But GB18030 decodes it to U+E5E5, which is correct in theory but not in practice.
// To work around, just change all occurences of U+E5E5 to U+3000 (ideographic space).
if (m_encoding == kCFStringEncodingGB_18030_2000)
resultString.replace(0xE5E5, ideographicSpace);
return resultString;
}
