StreamingTextDecoderICU::~StreamingTextDecoderICU()
{
releaseICUConverter();
}
TextCodecICU::~TextCodecICU()
{
releaseICUConverter();
}
DeprecatedString StreamingTextDecoderICU::toUnicode(const char* chs, int len, bool flush)
{
ASSERT_ARG(len, len >= 0);
if (!chs)
return DeprecatedString();
if (len <= 0 && !flush)
return "";
// Handle normal case.
if (!m_atStart)
return convert(chs, len, flush);
// Check to see if we found a BOM.
int numBufferedBytes = m_numBufferedBytes;
int buf1Len = numBufferedBytes;
int buf2Len = len;
const unsigned char* buf1 = m_bufferedBytes;
const unsigned char* buf2 = reinterpret_cast<const unsigned char*>(chs);
unsigned char c1 = buf1Len ? (--buf1Len, *buf1++) : buf2Len ? (--buf2Len, *buf2++) : 0;
unsigned char c2 = buf1Len ? (--buf1Len, *buf1++) : buf2Len ? (--buf2Len, *buf2++) : 0;
unsigned char c3 = buf1Len ? (--buf1Len, *buf1++) : buf2Len ? (--buf2Len, *buf2++) : 0;
int BOMLength = 0;
if (c1 == 0xFF && c2 == 0xFE) {
if (m_encoding != TextEncoding(UTF16Encoding, LittleEndian)) {
releaseICUConverter();
m_encoding = TextEncoding(UTF16Encoding, LittleEndian);
m_littleEndian = true;
}
BOMLength = 2;
} else if (c1 == 0xFE && c2 == 0xFF) {
if (m_encoding != TextEncoding(UTF16Encoding, BigEndian)) {
releaseICUConverter();
m_encoding = TextEncoding(UTF16Encoding, BigEndian);
m_littleEndian = false;
}
BOMLength = 2;
} else if (c1 == 0xEF && c2 == 0xBB && c3 == 0xBF) {
if (m_encoding != TextEncoding(UTF8Encoding)) {
releaseICUConverter();
m_encoding = TextEncoding(UTF8Encoding);
}
BOMLength = 3;
}
// Handle case where we found a BOM.
if (BOMLength != 0) {
ASSERT(numBufferedBytes + len >= BOMLength);
int skip = BOMLength - numBufferedBytes;
m_numBufferedBytes = 0;
m_atStart = false;
return len == skip ? DeprecatedString("") : convert(chs + skip, len - skip, flush);
}
// Handle case where we know there is no BOM coming.
const int bufferSize = sizeof(m_bufferedBytes);
if (numBufferedBytes + len > bufferSize || flush) {
m_atStart = false;
if (numBufferedBytes == 0) {
return convert(chs, len, flush);
}
unsigned char bufferedBytes[sizeof(m_bufferedBytes)];
memcpy(bufferedBytes, m_bufferedBytes, numBufferedBytes);
m_numBufferedBytes = 0;
return convert(bufferedBytes, numBufferedBytes, false) + convert(chs, len, flush);
}
// Continue to look for the BOM.
memcpy(&m_bufferedBytes[numBufferedBytes], chs, len);
m_numBufferedBytes += len;
return "";
}
