qgb18030codec.cpp

qt-x11-opensource-src-4.1.4.tar.gz源码

            }
            nbuf = 0;
            break;
        }
    }
    if (state) {
        state->remainingChars = nbuf;
        state->state_data[0] = (buf[0] << 24) + (buf[1] << 16) + (buf[2] << 8) + buf[3];
        state->invalidChars += invalid;
    }
    return result;
}


/*! \class QGbkCodec
    \reentrant
    \internal

  \brief The QGbkCodec class provides conversion to and from the Chinese
  GBK encoding.

  GBK, formally the Chinese Internal Code Specification, is a commonly
  used extension of GB 2312-80. Microsoft Windows uses it under the
  name code page 936.

  The GBK encoding has been superseded by the GB18030 encoding and
  GB18030 is backward compatible to GBK.  For this reason the QGbkCodec class
  is implemented in terms of the GB18030 codec and uses its 1-byte and
  2-byte portion for conversion from and to Unicode.

  The QGbkCodec is kept mainly for compatibility reasons with older software.
*/


/*!
    Constructs a QGbkCodec object.
*/
QGbkCodec::QGbkCodec()
    : QGb18030Codec()
{
}

int QGbkCodec::_mibEnum()
{
    return 113;
}

QByteArray QGbkCodec::_name()
{
    return "GBK";
}

QList<QByteArray> QGbkCodec::_aliases()
{
    QList<QByteArray> list;
    list << "CP936"
         << "MS936"
         << "windows-936";
    return list;
}

QString QGbkCodec::convertToUnicode(const char* chars, int len, ConverterState *state) const
{
    uchar buf[2];
    int nbuf = 0;
    QChar replacement = QChar::ReplacementCharacter;
    if (state) {
        if (state->flags & ConvertInvalidToNull)
            replacement = QChar::Null;
        nbuf = state->remainingChars;
        buf[0] = state->state_data[0];
        buf[1] = state->state_data[1];
    }
    int invalid = 0;

    QString result;

    //qDebug("QGbkDecoder::toUnicode(const char* chars = \"%s\", int len = %d)", chars, len);
    for (int i=0; i<len; i++) {
        uchar ch = chars[i];
        switch (nbuf) {
        case 0:
            if (ch < 0x80) {
                // ASCII
                result += QLatin1Char(ch);
            } else if (Is1stByte(ch)) {
                // GBK 1st byte?
                buf[0] = ch;
                nbuf = 1;
            } else {
                // Invalid
                result += replacement;
                ++invalid;
            }
            break;
        case 1:
            // GBK 2nd byte
            if (Is2ndByteIn2Bytes(ch)) {
                buf[1] = ch;
                int clen = 2;
                uint u = qt_Gb18030ToUnicode(buf, clen);
                if (clen == 2) {
                    result += QValidChar(u);
                } else {
                    result += replacement;
                    ++invalid;
                }
                nbuf = 0;
            } else {
                // Error
                result += replacement;
                ++invalid;
                nbuf = 0;
            }
            break;
        }
    }

    if (state) {
        state->remainingChars = nbuf;
        state->state_data[0] = buf[0];
        state->state_data[1] = buf[1];
        state->invalidChars += invalid;
    }
    return result;
}

QByteArray QGbkCodec::convertFromUnicode(const QChar *uc, int len, ConverterState *state) const
{
    char replacement = '?';
    if (state) {
        if (state->flags & ConvertInvalidToNull)
            replacement = 0;
    }
    int invalid = 0;

    int rlen = 2*len + 1;
    QByteArray rstr;
    rstr.resize(rlen);
    uchar* cursor = (uchar*)rstr.data();

    //qDebug("QGbkCodec::fromUnicode(const QString& uc, int& lenInOut = %d)", lenInOut);
    for (int i = 0; i < len; i++) {
        QChar ch = uc[i];
        uchar buf[2];

        if (ch.row() == 0x00 && ch.cell() < 0x80) {
            // ASCII
            *cursor++ = ch.cell();
        } else if (qt_UnicodeToGbk(ch.unicode(), buf) == 2) {
            *cursor++ = buf[0];
            *cursor++ = buf[1];
        } else {
            // Error
            *cursor += replacement;
            ++invalid;
        }
    }
    rstr.resize(cursor - (uchar*)rstr.constData());

    if (state) {
        state->invalidChars += invalid;
    }
    return rstr;
}


/*! \class QGb2312Codec
    \internal

  \brief The QGb2312Codec class provides conversion to and from the Chinese
  GB2312 encoding.

  The GB2312 encoding has been superseded by the GB18030 encoding and
  GB18030 is backward compatible to GB2312.  For this reason the QGb2312Codec
  class is implemented in terms of the GB18030 codec and uses its
  0xA1A1-0xFEFE subset for conversion from and to Unicode.

  The QGb2312Codec is kept mainly for compatibility reasons with older software.
*/


/*!
    Constructs a QGb2312Codec object.
*/
QGb2312Codec::QGb2312Codec()
    : QGb18030Codec()
{
}

int QGb2312Codec::_mibEnum()
{
    return 2025;
}

QByteArray QGb2312Codec::_name()
{
    return "GB2312";
}


QString QGb2312Codec::convertToUnicode(const char* chars, int len, ConverterState *state) const
{
    uchar buf[2];
    int nbuf = 0;
    QChar replacement = QChar::ReplacementCharacter;
    if (state) {
        if (state->flags & ConvertInvalidToNull)
            replacement = QChar::Null;
        nbuf = state->remainingChars;
        buf[0] = state->state_data[0];
        buf[1] = state->state_data[1];
    }
    int invalid = 0;

    QString result;
    //qDebug("QGb2312Decoder::toUnicode(const char* chars, int len = %d)", len);
    for (int i=0; i<len; i++) {
        uchar ch = chars[i];
        switch (nbuf) {
        case 0: