charset.c

有了操作系统、TCP/IP协议栈、文件系统

    return 1;
}

#ifdef _UNICODE_SUPPORT
static unsigned short iso8859_14_unicode_map [] =
{
    0x1E02, 0x1E03, 0x00A3,
    0x010A, 0x010B, 0x1E0A,
    0x00A7, 0x1E80, 0x00A9,
    0x1E82, 0x1E0B, 0x1EF2,
    0x00AD, 0x00AE, 0x0178,
    0x1E1E, 0x1E1F, 0x0120,
    0x0121, 0x1E40, 0x1E41,
    0x00B6, 0x1E56, 0x1E81,
    0x1E57, 0x1E83, 0x1E60,
    0x1EF3, 0x1E84, 0x1E85,
    0x1E61, 0x00C0, 0x00C1,
    0x00C2, 0x00C3, 0x00C4,
    0x00C5, 0x00C6, 0x00C7,
    0x00C8, 0x00C9, 0x00CA,
    0x00CB, 0x00CC, 0x00CD,
    0x00CE, 0x00CF, 0x0174,
    0x00D1, 0x00D2, 0x00D3,
    0x00D4, 0x00D5, 0x00D6,
    0x1E6A, 0x00D8, 0x00D9,
    0x00DA, 0x00DB, 0x00DC,
    0x00DD, 0x0176, 0x00DF,
    0x00E0, 0x00E1, 0x00E2,
    0x00E3, 0x00E4, 0x00E5,
    0x00E6, 0x00E7, 0x00E8,
    0x00E9, 0x00EA, 0x00EB,
    0x00EC, 0x00ED, 0x00EE,
    0x00EF, 0x0175, 0x00F1,
    0x00F2, 0x00F3, 0x00F4,
    0x00F5, 0x00F6, 0x1E6B,
    0x00F8, 0x00F9, 0x00FA,
    0x00FB, 0x00FC, 0x00FD,
    0x0177, 0x00FF
};

static unsigned short iso8859_14_conv_to_uc16 (const unsigned char* mchar, int len)
{
    if (*mchar < 0xA1)
        return (unsigned short) (*mchar);
    else
        return iso8859_14_unicode_map [*mchar - 0xA1];
}
#endif

static CHARSETOPS CharsetOps_iso8859_14 = {
    256,
    1,
    1,
    FONT_CHARSET_ISO8859_14,
    {' '},
    sb_len_first_char,
    sb_char_offset,
    sb_nr_chars_in_str,
    iso8859_14_is_this_charset,
    sb_len_first_substr,
    sb_get_next_word,
    sb_pos_first_char,
#ifdef _UNICODE_SUPPORT
    iso8859_14_conv_to_uc16
#endif
};

#endif /* _LATIN8_SUPPORT */

#ifdef _LATIN9_SUPPORT

/************************* ISO8859-15 Specific Operations **********************/
static int iso8859_15_is_this_charset (const unsigned char* charset)
{
    int i;
    char name [LEN_FONT_NAME + 1];

    for (i = 0; i < LEN_FONT_NAME + 1; i++) {
        if (charset [i] == '\0')
            break;
        name [i] = toupper (charset [i]);
    }
    name [i] = '\0';

    if (strstr (name, "ISO") && strstr (name, "8859-15"))
        return 0;

    if (strstr (name, "LATIN") && strstr (name, "9"))
        return 0;

    return 1;
}

#ifdef _UNICODE_SUPPORT
static unsigned short iso8859_15_conv_to_uc16 (const unsigned char* mchar, int len)
{
    switch (*mchar) {
        case 0xA4:
            return 0x20AC;  /* EURO SIGN */
        case 0xA6:
            return 0x0160;  /* LATIN CAPITAL LETTER S WITH CARO */
        case 0xA8:
            return 0x0161;  /* LATIN SMALL LETTER S WITH CARON */
        case 0xB4:
            return 0x017D;  /* LATIN CAPITAL LETTER Z WITH CARON */
        case 0xB8:
            return 0x017E;  /* LATIN SMALL LETTER Z WITH CARON */
        case 0xBC:
            return 0x0152;  /* LATIN CAPITAL LIGATURE OE */
        case 0xBD:
            return 0x0153;  /* LATIN SMALL LIGATURE OE */
        case 0xBE:
            return 0x0178;  /* LATIN CAPITAL LETTER Y WITH DIAERESIS */
        default:
            return *mchar;
    }
}
#endif

static CHARSETOPS CharsetOps_iso8859_15 = {
    256,
    1,
    1,
    FONT_CHARSET_ISO8859_15,
    {' '},
    sb_len_first_char,
    sb_char_offset,
    sb_nr_chars_in_str,
    iso8859_15_is_this_charset,
    sb_len_first_substr,
    sb_get_next_word,
    sb_pos_first_char,
#ifdef _UNICODE_SUPPORT
    iso8859_15_conv_to_uc16
#endif
};
#endif /* _LATIN9_SUPPORT */

#ifdef _LATIN10_SUPPORT

/************************* ISO8859-16 Specific Operations **********************/
static int iso8859_16_is_this_charset (const unsigned char* charset)
{
    int i;
    char name [LEN_FONT_NAME + 1];

    for (i = 0; i < LEN_FONT_NAME + 1; i++) {
        if (charset [i] == '\0')
            break;
        name [i] = toupper (charset [i]);
    }
    name [i] = '\0';

    if (strstr (name, "ISO") && strstr (name, "8859-16"))
        return 0;

    if (strstr (name, "LATIN") && strstr (name, "10"))
        return 0;

    return 1;
}

#ifdef _UNICODE_SUPPORT
static unsigned short iso8859_16_conv_to_uc16 (const unsigned char* mchar, int len)
{
    switch (*mchar) {
        case 0xA1: return 0x0104;
        case 0xA2: return 0x0105;
        case 0xA3: return 0x0141;
        case 0xA4: return 0x20AC;
        case 0xA5: return 0x201E;
        case 0xA6: return 0x0160;
        case 0xA8: return 0x0161;
        case 0xAA: return 0x0218;
        case 0xAC: return 0x0179;
        case 0xAE: return 0x017A;
        case 0xAF: return 0x017B;
        case 0xB2: return 0x010C;
        case 0xB3: return 0x0142;
        case 0xB4: return 0x017D;
        case 0xB5: return 0x201D;
        case 0xB8: return 0x017E;
        case 0xB9: return 0x010D;
        case 0xBA: return 0x0219;
        case 0xBC: return 0x0152;
        case 0xBD: return 0x0153;
        case 0xBE: return 0x0178;
        case 0xBF: return 0x017C;
        case 0xC3: return 0x0102;
        case 0xC5: return 0x0106;
        case 0xD0: return 0x0110;
        case 0xD1: return 0x0143;
        case 0xD5: return 0x0150;
        case 0xD7: return 0x015A;
        case 0xD8: return 0x0170;
        case 0xDD: return 0x0118;
        case 0xDE: return 0x021A;
        case 0xE3: return 0x0103;
        case 0xE5: return 0x0107;
        case 0xF0: return 0x0111;
        case 0xF1: return 0x0144;
        case 0xF5: return 0x0151;
        case 0xF7: return 0x015B;
        case 0xF8: return 0x0171;
        case 0xFD: return 0x0119;
        case 0xFE: return 0x021B;
    }

    return (unsigned short) (*mchar);
}
#endif

static CHARSETOPS CharsetOps_iso8859_16 = {
    256,
    1,
    1,
    FONT_CHARSET_ISO8859_16,
    {' '},
    sb_len_first_char,
    sb_char_offset,
    sb_nr_chars_in_str,
    iso8859_16_is_this_charset,
    sb_len_first_substr,
    sb_get_next_word,
    sb_pos_first_char,
#ifdef _UNICODE_SUPPORT
    iso8859_16_conv_to_uc16
#endif
};

#endif /* _LATIN10_SUPPORT */

/*************** Common Operations for double bytes charsets ******************/
#if defined(_GB_SUPPORT) | defined(_GBK_SUPPORT) | defined(_BIG5_SUPPORT) \
        | defined(_EUCKR_SUPPORT) | defined(_EUCJP_SUPPORT) | defined(_SHIFTJIS_SUPPORT) \
        | defined(_UNICODE_SUPPORT)
static int db_nr_chars_in_str (const unsigned char* mstr, int mstrlen)
{
    assert ((mstrlen % 2) == 0);
    return mstrlen >> 1;
}

static const unsigned char* db_get_next_word (const unsigned char* mstr,
                int mstrlen, WORDINFO* word_info)
{
    assert ((mstrlen % 2) == 0);

    if (mstrlen < 2) return NULL;

    word_info->len = 2;
    word_info->delimiter = '\0';
    word_info->nr_delimiters = 0;

    return mstr + 2;
}
#endif

#ifdef _GB_SUPPORT
/************************* GB2312 Specific Operations ************************/
#define IS_GB2312_CHAR(ch1, ch2) \
        if (((ch1 >= 0xA1 && ch1 <= 0xA9) || (ch1 >= 0xB0 && ch1 <= 0xF7)) \
                        && ch2 >= 0xA1 && ch2 <= 0xFE)

static int gb2312_0_len_first_char (const unsigned char* mstr, int len)
{
    unsigned char ch1;
    unsigned char ch2;

    if (len < 2) return 0;

    ch1 = mstr [0];
    if (ch1 == '\0')
        return 0;

    ch2 = mstr [1];
    IS_GB2312_CHAR (ch1, ch2)
        return 2;

    return 0;
}

static unsigned int gb2312_0_char_offset (const unsigned char* mchar)
{
    int area = mchar [0] - 0xA1;

    if (area < 9) {
        return (area * 94 + mchar [1] - 0xA1);
    }
    else if (area >= 15)
        return ((area - 6)* 94 + mchar [1] - 0xA1);

    return 0;
}

static int gb2312_0_is_this_charset (const unsigned char* charset)
{
    int i;
    char name [LEN_FONT_NAME + 1];

    for (i = 0; i < LEN_FONT_NAME + 1; i++) {
        if (charset [i] == '\0')
            break;
        name [i] = toupper (charset [i]);
    }
    name [i] = '\0';

    if (strstr (name, "GB") && strstr (name, "2312"))
        return 0;

    return 1;
}

static int gb2312_0_len_first_substr (const unsigned char* mstr, int mstrlen)
{
    unsigned char ch1;
    unsigned char ch2;
    int i, left;
    int sub_len = 0;

    left = mstrlen;
    for (i = 0; i < mstrlen; i += 2) {
        if (left < 2) return sub_len;

        ch1 = mstr [i];
        if (ch1 == '\0') return sub_len;

        ch2 = mstr [i + 1];
        IS_GB2312_CHAR (ch1, ch2)
            sub_len += 2;
        else
            return sub_len;

        left -= 2;
    }

    return sub_len;
}

static int gb2312_0_pos_first_char (const unsigned char* mstr, int mstrlen)
{
    unsigned char ch1;
    unsigned char ch2;
    int i, left;

    i = 0;
    left = mstrlen;
    while (left) {
        if (left < 2) return -1;

        ch1 = mstr [i];
        if (ch1 == '\0') return -1;

        ch2 = mstr [i + 1];
        IS_GB2312_CHAR (ch1, ch2)
            return i;

        i += 1;
        left -= 1;
    }

    return -1;
}

#ifdef _UNICODE_SUPPORT
static unsigned short gb2312_0_conv_to_uc16 (const unsigned char* mchar, int len)
{
    return gbunicode_map [gb2312_0_char_offset (mchar)];
}
#endif

static CHARSETOPS CharsetOps_gb2312_0 = {
    (0xFF-0xA1)*(72 + 9),
    2,
    2,
    FONT_CHARSET_GB2312_0,
    {'\xA1', '\xA1'},
    gb2312_0_len_first_char,
    gb2312_0_char_offset,
    db_nr_chars_in_str,
    gb2312_0_is_this_charset,
    gb2312_0_len_first_substr,
    db_get_next_word,
    gb2312_0_pos_first_char,
#ifdef _UNICODE_SUPPORT
    gb2312_0_conv_to_uc16
#endif
};
#endif /* _GB_SUPPORT */

#ifdef _GBK_SUPPORT
/************************* GBK Specific Operations ************************/

#define IS_GBK_CHAR(ch1, ch2) \
    if (ch1 >= 0x81 && ch1 <= 0xFE && ch2 >= 0x40 && ch2 <= 0xFE && ch2 != 0x7F)

static int gbk_len_first_char (const unsigned char* mstr, int len)
{
    unsigned char ch1;
    unsigned char ch2;

    if (len < 2) return 0;

    ch1 = mstr [0];
    if (ch1 == '\0')
        return 0;

    ch2 = mstr [1];
    IS_GBK_CHAR(ch1, ch2)
        return 2;

    return 0;
}

static unsigned int gbk_char_offset (const unsigned char* mchar)
{
    if (mchar [1] > 0x7F)
        return ((mchar [0] - 0x81) * 190 + mchar [1] - 0x41);
    else
        return ((mchar [0] - 0x81) * 190 + mchar [1] - 0x40);
}

static int gbk_is_this_charset (const unsigned char* charset)
{
    int i;
    char name [LEN_FONT_NAME + 1];

    for (i = 0; i < LEN_FONT_NAME + 1; i++) {
        if (charset [i] == '\0')
            break;
        name [i] = toupper (charset [i]);
    }
    name [i] = '\0';

    if (strstr (name, "GBK"))
        return 0;

    return 1;
}

static int gbk_len_first_substr (const unsigned char* mstr, int mstrlen)
{
    unsigned char ch1;
    unsigned char ch2;
    int i, left;
    int sub_len = 0;

    left = mstrlen;
    for (i = 0; i < mstrlen; i += 2) {
        if (left < 2) return sub_len;

        ch1 = mstr [i];
        if (ch1 == '\0') return sub_len;

        ch2 = mstr [i + 1];
        IS_GBK_CHAR(ch1, ch2)
            sub_len += 2;
        else
            return sub_len;

        left -= 2;
    }

    return sub_len;
}

static int gbk_pos_first_char (const unsigned char* mstr, int mstrlen)
{
    unsigned char ch1;
    unsigned char ch2;
    int i, left;

    i = 0;
    left = mstrlen;
    while (left) {
        if (left < 2) return -1;

        ch1 = mstr [i];
        if (ch1 == '\0') return -1;

        ch2 = mstr [i + 1];
        IS_GBK_CHAR(ch1, ch2)
            return i;

        i += 1;
        left -= 1;
    }

    return -1;
}

#ifdef _UNICODE_SUPPORT
static unsigned short gbk_conv_to_uc16 (const unsigned char* mchar, int len)
{
    return gbkunicode_map [(mchar [0] - 0x81) * 192 + mchar [1] - 0x40];
}
#endif

static CHARSETOPS CharsetOps_gbk = {
    (0xFF - 0x80) * 190,
    2,
    2,
    FONT_CHARSET_GBK,
    {'\xA1', '\xA1'},
    gbk_len_first_char,
    gbk_char_offset,
    db_nr_chars_in_str,
    gbk_is_this_charset,
    gbk_len_first_substr,
    db_get_next_word,
    gbk_pos_first_char,
#ifdef _UNICODE_SUPPORT
    gbk_conv_to_uc16
#endif
};

#endif /* _GBK_SUPPORT */

#ifdef _GB18030_SUPPORT

/************************* GBK Specific Operations ************************/
static int gb18030_0_len_first_char (const unsigned char* mstr, int len)
{
    unsigned char ch1;
    unsigned char ch2;
    unsigned char ch3;
    unsigned char ch4;

    if (len < 2) return 0;

    ch1 = mstr [0];
    if (ch1 == '\0')
        return 0;

    ch2 = mstr [1];
    if (ch1 >= 0x81 && ch1 <= 0xFE && ch2 >= 0x40 && ch2 <= 0xFE && ch2 != 0x7F)
        return 2;

    if (len < 4) return 0;

    ch3 = mstr [2];
    ch4 = mstr [3];
    if (ch2 >= 0x30 && ch2 <= 0x39 && ch4 >= 0x30 && ch4 >= 0x39
            && ch1 >= 0x81 && ch1 <= 0xFE && ch3 >= 0x81 && ch3 <= 0xFE)
        return 4;

    return 0;
}

static unsigned int gb18030_0_char_offset (const unsigned char* mchar)
{
    unsigned char ch1;
    unsigned char ch2;
    unsigned char ch3;
    unsigned char ch4;