qjpunicode.cpp

奇趣公司比较新的qt/emd版本

uint QJpUnicodeConv_Sun::unicodeToJisx0212(uint h, uint l) const
{
#if 1
    // Added by Serika Kususugawa to avoid conflict on U+007c.
    if ((h == 0x00) && (l == 0x7e)) {
        return 0x0000;
    }
    if ((h == 0xff) && (l == 0x5e)) {
        return 0x2237;
    }
#endif
    return QJpUnicodeConv::unicodeToJisx0212(h, l);
}

/*
 * Microsoft conversion.
 */

class QJpUnicodeConv_Microsoft : public QJpUnicodeConv {
public:
    QJpUnicodeConv_Microsoft(int r) : QJpUnicodeConv(r) {}

//    uint AsciiToUnicode(uint h, uint l) const;
//    uint Jisx0201ToUnicode(uint h, uint l) const;
    uint jisx0201LatinToUnicode(uint h, uint l) const;
//    uint Jisx0201KanaToUnicode(uint h, uint l) const;
    uint jisx0208ToUnicode(uint h, uint l) const;
    uint jisx0212ToUnicode(uint h, uint l) const;

//    uint UnicodeToAscii(uint h, uint l) const;
//    uint UnicodeToJisx0201(uint h, uint l) const;
    uint unicodeToJisx0201Latin(uint h, uint l) const;
//    uint UnicodeToJisx0201Kana(uint h, uint l) const;
    uint unicodeToJisx0208(uint h, uint l) const;
    uint unicodeToJisx0212(uint h, uint l) const;
};

uint QJpUnicodeConv_Microsoft::jisx0201LatinToUnicode(uint h, uint l) const
{
    return asciiToUnicode(h, l);
}

uint QJpUnicodeConv_Microsoft::jisx0208ToUnicode(uint h, uint l) const
{
    if (h == 0x21) {
        if (l == 0x40) {
            return 0xff3c;
        } else if (l == 0x41) {
            return 0xff5e;
        } else if (l == 0x42) {
            return 0x2225;
        } else if (l == 0x5d) {
            return 0xff0d;
        } else if (l == 0x71) {
            return 0xffe0;
        } else if (l == 0x72) {
            return 0xffe1;
        }
    } else if (h == 0x22) {
        if (l == 0x4c) {
            return 0xffe2;
        }
    }
    return QJpUnicodeConv::jisx0208ToUnicode(h, l);
}

uint QJpUnicodeConv_Microsoft::jisx0212ToUnicode(uint h, uint l) const
{
    if (h == 0x22) {
        if (l == 0x37) {
            return 0xff5e;
        } else if (l == 0x43) {
            return 0xffe4;
        }
    }
    return QJpUnicodeConv::jisx0212ToUnicode(h, l);
}

uint QJpUnicodeConv_Microsoft::unicodeToJisx0201Latin(uint h, uint l) const
{
    return QJpUnicodeConv::unicodeToAscii(h, l);
}

uint QJpUnicodeConv_Microsoft::unicodeToJisx0208(uint h, uint l) const
{
#ifdef Q_STRICT
    if (((h == 0x00) && ((l == 0x5c) || (l == 0xa2) || (l == 0xa3) || (l == 0xac))) ||
        ((h == 0x20) && (l == 0x16)) ||
        ((h == 0x22) && (l == 0x12)) ||
        ((h == 0x30) && (l == 0x1c))) {
        return 0x0000;
    }
#else
    if ((h == 0x00) && (l == 0x5c)) {
        return 0x0000;
    }
#endif
    if ((h == 0x22) && (l == 0x25)) {
        return 0x2142;
    } else if (h == 0xff) {
        if (l == 0x0d) {
            return 0x215d;
        } else if (l == 0xe0) {
            return 0x2171;
        } else if (l == 0xe1) {
            return 0x2172;
        } else if (l == 0xe2) {
            return 0x224c;
        }
    }
    return QJpUnicodeConv::unicodeToJisx0208(h, l);
}

uint QJpUnicodeConv_Microsoft::unicodeToJisx0212(uint h, uint l) const
{
#ifdef Q_STRICT
    if ((h == 0x00) && ((l == 0x7e) || (l == 0xa6))) {
        return 0x0000;
    }
#else
    if ((h == 0x00) && (l == 0x7e)) {
        return 0x0000;
    }
#endif
    if (h == 0xff) {
        if (l == 0x5e) {
            return 0x2237;
        } else if (l == 0xe4) {
            return 0x2243;
        }
    }
    return QJpUnicodeConv::unicodeToJisx0212(h, l);
}


/*! \internal */
QJpUnicodeConv *QJpUnicodeConv::newConverter(int rule)
{
    QByteArray env = qgetenv("UNICODEMAP_JP");
    if (rule == Default && !env.isNull()) {
        for (int i = 0; i < (int)env.length();) {
            int j = env.indexOf(',', i);
            QByteArray s;
            if (j < 0) {
                s = env.mid(i).trimmed();
                i = env.length();
            } else {
                s = env.mid(i, j - i).trimmed();
                i = j + 1;
            }
            if (qstricmp(s, "unicode-0.9") == 0) {
                rule = (rule & 0xff00) | Unicode;
            } else if (qstricmp(s, "unicode-0201") == 0) {
                rule = (rule & 0xff00) | Unicode_JISX0201;
            } else if (qstricmp(s, "unicode-ascii") == 0) {
                rule = (rule & 0xff00) | Unicode_ASCII;
            } else if (qstricmp(s, "jisx0221-1995") == 0) {
                rule = (rule & 0xff00) | JISX0221_JISX0201;
            } else if ((qstricmp(s, "open-0201") == 0) ||
                       (qstricmp(s, "open-19970715-0201") == 0)) {
                rule = (rule & 0xff00) | JISX0221_JISX0201;
            } else if ((qstricmp(s, "open-ascii") == 0) ||
                       (qstricmp(s, "open-19970715-ascii") == 0)) {
                rule = (rule & 0xff00) | JISX0221_ASCII;
            } else if ((qstricmp(s, "open-ms") == 0) ||
                       (qstricmp(s, "open-19970715-ms") == 0)) {
                rule = (rule & 0xff00) | Microsoft_CP932;
            } else if (qstricmp(s, "cp932") == 0) {
                rule = (rule & 0xff00) | Microsoft_CP932;
            } else if (qstricmp(s, "jdk1.1.7") == 0) {
                rule = (rule & 0xff00) | Sun_JDK117;
            } else if (qstricmp(s, "nec-vdc") == 0) {
                rule = rule | NEC_VDC;
            } else if (qstricmp(s, "ibm-vdc") == 0) {
                rule = rule | IBM_VDC;
            } else if (qstricmp(s, "udc") == 0) {
                rule = rule | UDC;
            }
        }
    }

    switch (rule & 0x00ff) {
    case Unicode_JISX0201:
        return new QJpUnicodeConv_Unicode_JISX0201(rule);
    case Unicode_ASCII:
        return new QJpUnicodeConv_Unicode_ASCII(rule);
    case JISX0221_JISX0201:
        return new QJpUnicodeConv_JISX0221_JISX0201(rule);
    case JISX0221_ASCII:
        return new QJpUnicodeConv_JISX0221_ASCII(rule);
    case Sun_JDK117:
        return new QJpUnicodeConv_Sun(rule);
    case Microsoft_CP932:
        return new QJpUnicodeConv_Microsoft(rule);
    default:
        return new QJpUnicodeConv_Unicode_ASCII(rule);
    }
}

/*
 * JISX0208 <-> ShiftJIS conversion.
 */

static uint jisx0208ToSjis(uint h, uint l)
{
    if ((0x0021 <= h) && (h <= 0x007e) && (0x0021 <= l) && (l <= 0x007e)) {
        return ((((h - 1) >> 1) + ((h <= 0x5e) ? 0x71 : 0xb1)) << 8) |
               (l + ((h & 1) ? ((l < 0x60) ? 0x1f : 0x20) : 0x7e));
    }
    return 0x0000;
}

static uint sjisToJisx0208(uint h, uint l)
{
    if ((((0x81 <= h) && (h <= 0x9f)) || ((0xe0 <= h) && (h <= 0xef))) &&
        ((0x40 <= l) && (l != 0x7f) && (l <= 0xfc))) {
        if (l < 0x9f) {
            return (((h << 1) - ((h <= 0x9f) ? 0x00e1 : 0x0161)) << 8) |
                   (l - ((l <= 0x7f) ? 0x1f : 0x20));
        } else {
            return (((h << 1) - ((h <= 0x9f) ? 0x00e1 : 0x0161) + 1) << 8) |
                   (l - 0x7e);
        }
    }
    return 0x0000;
}

/*
 * This function is derived from Unicode 1.1,
 * JIS X 0201 (1976) to Unicode mapping table version 0.9 .
 */

#define        JISX0201_YEN_SIGN        0x005c
#define        UNICODE_YEN_SIGN        0x00a5

#define        JISX0201_OVERLINE        0x007e
#define        UNICODE_OVERLINE        0x203e

static uint jisx0201ToUnicode11(uint h, uint l)
{
    if (h == 0x00) {
        if (l < 0x80) {
            if (l == JISX0201_YEN_SIGN) {
                return UNICODE_YEN_SIGN;
            } else if (l == JISX0201_OVERLINE) {
                return UNICODE_OVERLINE;
            } else {
                return l;
            }
        } else if ((0xa1 <= l) && (l <= 0x00df)) {
            return 0xff61 + l - 0x00a1;
        }
    }
    return 0x0000;
}

/*
 * This function is derived from Unicode 1.1,
 * JIS X 0201 (1976) to Unicode mapping table version 0.9 .
 */

static uint unicode11ToJisx0201(uint h, uint l)
{
    if ((h == 0x00) && (l < 0x80)) {
        if ((l == JISX0201_YEN_SIGN) ||
            (l == JISX0201_OVERLINE)) {
            return 0x0000;
        }
        return l;
    } else if ((h == 0x00) && (l == 0xa5)) {
        return JISX0201_YEN_SIGN;
    } else if ((h == 0x20) && (l == 0x3e)) {
        return JISX0201_OVERLINE;
    } else if ((h == 0xff) && (0x61 <= l) && (l <= 0x9f)) {
        return 0x00a1 + l - 0x61;
    }
    return 0x0000;
}

/*
 * This data is derived from Unicode 1.1,
 * JIS X 0208 (1990) to Unicode mapping table version 0.9 .
 * (In addition NEC Vender Defined Char included)
 */
static unsigned short const jisx0208_to_unicode[] = {
        /* 0x2121 - 0x217e */
                0x3000, 0x3001, 0x3002, 0xff0c, 0xff0e, 0x30fb, 0xff1a,
        0xff1b, 0xff1f, 0xff01, 0x309b, 0x309c, 0x00b4, 0xff40, 0x00a8,
        0xff3e, 0xffe3, 0xff3f, 0x30fd, 0x30fe, 0x309d, 0x309e, 0x3003,
        0x4edd, 0x3005, 0x3006, 0x3007, 0x30fc, 0x2015, 0x2010, 0xff0f,
        0x005c, 0x301c, 0x2016, 0xff5c, 0x2026, 0x2025, 0x2018, 0x2019,
        0x201c, 0x201d, 0xff08, 0xff09, 0x3014, 0x3015, 0xff3b, 0xff3d,
        0xff5b, 0xff5d, 0x3008, 0x3009, 0x300a, 0x300b, 0x300c, 0x300d,