charsetConverter::charsetConverter(const charset& source, const charset& dest)
: m_desc(NULL), m_source(source), m_dest(dest)
{
// Get an iconv descriptor
const iconv_t cd = iconv_open(dest.getName().c_str(), source.getName().c_str());
if (cd != reinterpret_cast <iconv_t>(-1))
{
iconv_t* p = new iconv_t;
*p= cd;
m_desc = p;
}
}
charsetFilteredOutputStream_icu::charsetFilteredOutputStream_icu
(const charset& source, const charset& dest, outputStream* os,
const charsetConverterOptions& opts)
: m_from(NULL), m_to(NULL), m_sourceCharset(source),
m_destCharset(dest), m_stream(*os), m_options(opts)
{
UErrorCode err = U_ZERO_ERROR;
m_from = ucnv_open(source.getName().c_str(), &err);
if (!U_SUCCESS(err))
{
throw exceptions::charset_conv_error
("Cannot initialize ICU converter for source charset '" + source.getName() + "' (error code: " + u_errorName(err) + ".");
}
m_to = ucnv_open(dest.getName().c_str(), &err);
if (!U_SUCCESS(err))
{
throw exceptions::charset_conv_error
("Cannot initialize ICU converter for destination charset '" + dest.getName() + "' (error code: " + u_errorName(err) + ".");
}
// Tell ICU what to do when encountering an illegal byte sequence
if (m_options.silentlyReplaceInvalidSequences)
{
// Set replacement chars for when converting from Unicode to codepage
icu::UnicodeString substString(m_options.invalidSequence.c_str());
ucnv_setSubstString(m_to, substString.getTerminatedBuffer(), -1, &err);
if (U_FAILURE(err))
throw exceptions::charset_conv_error("[ICU] Error when setting substitution string.");
}
else
{
// Tell ICU top stop (and return an error) on illegal byte sequences
ucnv_setToUCallBack
(m_to, UCNV_TO_U_CALLBACK_STOP, UCNV_SUB_STOP_ON_ILLEGAL, NULL, NULL, &err);
if (U_FAILURE(err))
throw exceptions::charset_conv_error("[ICU] Error when setting ToU callback.");
ucnv_setFromUCallBack
(m_to, UCNV_FROM_U_CALLBACK_STOP, UCNV_SUB_STOP_ON_ILLEGAL, NULL, NULL, &err);
if (U_FAILURE(err))
throw exceptions::charset_conv_error("[ICU] Error when setting FromU callback.");
}
}
charsetFilteredOutputStream::charsetFilteredOutputStream
(const charset& source, const charset& dest, outputStream& os)
: m_desc(NULL), m_sourceCharset(source), m_destCharset(dest),
m_stream(os), m_unconvCount(0)
{
// Get an iconv descriptor
const iconv_t cd = iconv_open(dest.getName().c_str(), source.getName().c_str());
if (cd != reinterpret_cast <iconv_t>(-1))
{
iconv_t* p = new iconv_t;
*p= cd;
m_desc = p;
}
}
charsetConverter_icu::charsetConverter_icu
(const charset& source, const charset& dest, const charsetConverterOptions& opts)
: m_from(NULL), m_to(NULL), m_source(source), m_dest(dest), m_options(opts)
{
UErrorCode err = U_ZERO_ERROR;
m_from = ucnv_open(source.getName().c_str(), &err);
if (!U_SUCCESS(err))
{
throw exceptions::charset_conv_error
("Cannot initialize ICU converter for source charset '" + source.getName() + "' (error code: " + u_errorName(err) + ".");
}
m_to = ucnv_open(dest.getName().c_str(), &err);
if (!U_SUCCESS(err))
{
throw exceptions::charset_conv_error
("Cannot initialize ICU converter for destination charset '" + dest.getName() + "' (error code: " + u_errorName(err) + ".");
}
}
