qgb18030codec.cpp File Reference

#include "qgb18030codec.h"

Go to the source code of this file.

Classes
struct	indexTbl_t

Macros
#define	InRange(c, lower, upper)   (((c) >= (lower)) && ((c) <= (upper)))
#define	Is1stByte(c)   (InRange((c), 0x81, 0xFE))
#define	Is2ndByte(c)   (Is2ndByteIn2Bytes(c) || Is2ndByteIn4Bytes(c))
#define	Is2ndByteIn2Bytes(c)   (InRange((c), 0x40, 0xFE) && (c) != 0x7F)
#define	Is2ndByteIn4Bytes(c)   (InRange((c), 0x30, 0x39))
#define	Is3rdByte(c)   (InRange((c), 0x81, 0xFE))
#define	Is4thByte(c)   (InRange((c), 0x30, 0x39))
#define	IsByteInGb2312(c)   (InRange((c), 0xA1, 0xFE))
#define	IsLatin(c)   ((c) <= 0x7F)
#define	IsUDA1(a, b)   (InRange((a), 0xAA, 0xAF) && InRange((b), 0xA1, 0xFE))
#define	IsUDA2(a, b)   (InRange((a), 0xF8, 0xFE) && InRange((b), 0xA1, 0xFE))
#define	IsUDA3(a, b)   (InRange((a), 0xA1, 0xA7) && InRange((b), 0x40, 0xA0) && ((b) != 0x7F))
#define	qValidChar(u)   ((u) ? (u) : static_cast<ushort>(QChar::ReplacementCharacter))

Functions
static uint	gb4lin_to_gb (uint gb4lin)
static uint	qt_Gb18030ToUnicode (const uchar *gbstr, int &len)
static int	qt_UnicodeToGb18030 (uint unicode, uchar *gbchar)
int	qt_UnicodeToGbk (uint unicode, uchar *gbchar)

Variables
static quint16 const	gb18030_2byte_to_ucs [22046]
static quint16 const	gb18030_4byte_to_ucs [6793]
static const indexTbl_t	gb18030_to_ucs_index [154]
static quint16 const	ucs_to_gb18030 [28839]
static const indexTbl_t	ucs_to_gb18030_index [256]

Macro Definition Documentation

InRange

#define InRange	(		c,
			lower,
			upper
	)		(((c) >= (lower)) && ((c) <= (upper)))

Definition at line 53 of file qgb18030codec.cpp.

Referenced by qt_Gb18030ToUnicode(), qt_UnicodeToGb18030(), and qt_UnicodeToGbk().

Is1stByte

#define Is1stByte

(

c

)

(InRange((c), 0x81, 0xFE))

Definition at line 56 of file qgb18030codec.cpp.

Referenced by QGb18030Codec::convertToUnicode(), QGbkCodec::convertToUnicode(), and qt_Gb18030ToUnicode().

Is2ndByte

#define Is2ndByte

(

c

)

(Is2ndByteIn2Bytes(c) || Is2ndByteIn4Bytes(c))

Definition at line 59 of file qgb18030codec.cpp.

Is2ndByteIn2Bytes

#define Is2ndByteIn2Bytes

(

c

)

(InRange((c), 0x40, 0xFE) && (c) != 0x7F)

Definition at line 57 of file qgb18030codec.cpp.

Referenced by QGb18030Codec::convertToUnicode(), QGbkCodec::convertToUnicode(), and qt_Gb18030ToUnicode().

Is2ndByteIn4Bytes

#define Is2ndByteIn4Bytes

(

c

)

(InRange((c), 0x30, 0x39))

Definition at line 58 of file qgb18030codec.cpp.

Referenced by QGb18030Codec::convertToUnicode(), and qt_Gb18030ToUnicode().

Is3rdByte

#define Is3rdByte

(

c

)

(InRange((c), 0x81, 0xFE))

Definition at line 60 of file qgb18030codec.cpp.

Referenced by QGb18030Codec::convertToUnicode(), and qt_Gb18030ToUnicode().

Is4thByte

#define Is4thByte

(

c

)

(InRange((c), 0x30, 0x39))

Definition at line 61 of file qgb18030codec.cpp.

Referenced by QGb18030Codec::convertToUnicode(), and qt_Gb18030ToUnicode().

IsByteInGb2312

#define IsByteInGb2312

(

c

)

(InRange((c), 0xA1, 0xFE))

Definition at line 55 of file qgb18030codec.cpp.

Referenced by QGb2312Codec::convertToUnicode().

IsLatin

#define IsLatin

(

c

)

((c) <= 0x7F)

Definition at line 54 of file qgb18030codec.cpp.

Referenced by QGb18030Codec::convertFromUnicode(), QGb18030Codec::convertToUnicode(), QGbkCodec::convertToUnicode(), QGb2312Codec::convertToUnicode(), qt_Gb18030ToUnicode(), qt_UnicodeToGb18030(), and qt_UnicodeToGbk().

IsUDA1

#define IsUDA1	(		a,
			b
	)		(InRange((a), 0xAA, 0xAF) && InRange((b), 0xA1, 0xFE))

Definition at line 68 of file qgb18030codec.cpp.

Referenced by qt_Gb18030ToUnicode().

IsUDA2

#define IsUDA2	(		a,
			b
	)		(InRange((a), 0xF8, 0xFE) && InRange((b), 0xA1, 0xFE))

Definition at line 69 of file qgb18030codec.cpp.

Referenced by qt_Gb18030ToUnicode().

IsUDA3

#define IsUDA3	(		a,
			b
	)		(InRange((a), 0xA1, 0xA7) && InRange((b), 0x40, 0xA0) && ((b) != 0x7F))

Definition at line 70 of file qgb18030codec.cpp.

Referenced by qt_Gb18030ToUnicode().

qValidChar

#define qValidChar

(

u

)

((u) ? (u) : static_cast<ushort>(QChar::ReplacementCharacter))

Definition at line 63 of file qgb18030codec.cpp.

Referenced by QGb18030Codec::convertToUnicode(), QGbkCodec::convertToUnicode(), and QGb2312Codec::convertToUnicode().

Function Documentation

gb4lin_to_gb()

static uint gb4lin_to_gb ( uint  gb4lin )

inlinestatic

Definition at line 9013 of file qgb18030codec.cpp.

Referenced by qt_UnicodeToGb18030().

                                             {
    uchar   a, b, c, d;
    a = 0x81 + gb4lin / 12600;
    b = 0x30 + (gb4lin / 1260) % 10;
    c = 0x81 + (gb4lin / 10) % 126;
    d = 0x30 + gb4lin % 10;
    return ((a << 24) | (b << 16) | (c << 8) | d);
}

qt_Gb18030ToUnicode()

static uint qt_Gb18030ToUnicode ( const uchar *  gbstr, int &  len  )

static

Definition at line 9022 of file qgb18030codec.cpp.

Referenced by QGb18030Codec::convertToUnicode(), QGbkCodec::convertToUnicode(), and QGb2312Codec::convertToUnicode().

                                                              {
    /* Returns Unicode. */
    uint    uni;
    uchar   first = *gbstr;

    if (IsLatin(first)) {
        len = 1;
        uni = (uint)first;
    }
    else if (Is1stByte(first) && len >= 2) {
        uchar        second = gbstr[1];

        if (Is2ndByteIn2Bytes(second)) {
            len = 2;

            if (IsUDA1(first, second))
                uni = 0xE000 + (first - 0xAA) * 94 + (second - 0xA1);
            else if (IsUDA2(first, second))
                uni = 0xE234 + (first - 0xF8) * 94 + (second - 0xA1);
            else if (IsUDA3(first, second))
                uni = 0xE4C6 + (first - 0xA1) * 96 + (second - 0x40)
                             - ((second >= 0x80) ? 1 : 0);
            else {
                // Use the mapping table
                uint i;

                i = (first - 0x81) * 190 + (second - 0x40)
                                         - ((second >= 0x80) ? 1 : 0);

                if (InRange(first, 0xA1, 0xA7))
                    i -= (first - 0xA0) * 96;
                if (first > 0xA7)
                    i -= 672;
                if (InRange(first, 0xAA, 0xAF))
                    i -= (first - 0xAA) * 94;
                if (first > 0xAF)
                    i -= 564;
                if (first >= 0xF8)
                    i -= (first - 0xF8) * 94;

                uni = (uint)gb18030_2byte_to_ucs[i];
            }
        }
        else if (Is2ndByteIn4Bytes(second) && len >= 4) {
            uchar   third  = gbstr[2],
                    fourth = gbstr[3];

            if (Is3rdByte(third) && Is4thByte(fourth)) {
                // Valid 4-byte GB18030, whether defined or not
                uint            gb4lin;
                indexTbl_t  g2u;

                gb4lin = (first - 0x81) * 12600 + (second - 0x30) * 1260
                          + (third - 0x81) * 10 + (fourth - 0x30);

                len = 4;
                if (gb4lin <= 0x99FB) {
                    /* GB+81308130 - GB+8431A439 */
                    g2u = gb18030_to_ucs_index[gb4lin >> 8];

                    if ((quint8)(gb4lin & 0xFF) >= g2u.tblBegin &&
                        (quint8)(gb4lin & 0xFF) <= g2u.tblEnd) {

                        uni = (uint)gb18030_4byte_to_ucs[gb4lin - g2u.tblOffset];
                    }
                    else {
                        uni = g2u.algOffset + (gb4lin & 0xFF);
                    }
                } else if (InRange(gb4lin, 0x2E248, 0x12E247)) {
                    /* GB+90308130 - GB+E3329A35 */
                    uni = gb4lin - 0xE248;
                } else {
                    /* undefined or reserved area */
                    len = 1;
                    uni = QChar::ReplacementCharacter;
                }
            }
            else {
                len = 1;
                uni = QChar::ReplacementCharacter;
            }
        }
        else {
            len = 1;
            uni = QChar::ReplacementCharacter;
        }
    }
    else {
        len = 1;
        uni = QChar::ReplacementCharacter;
    }
    return uni;
}

qt_UnicodeToGb18030()

int qt_UnicodeToGb18030 ( uint  unicode, uchar *  gbchar  )

static

Definition at line 9117 of file qgb18030codec.cpp.

Referenced by QGb18030Codec::convertFromUnicode().

                                                 {
    /* Returns the bytesize of the GB18030 character. */
    uint        gb, gb4lin;
    indexTbl_t        u2g;

    if (IsLatin(uni)) {
        *gbchar = (uchar)uni;
        return 1;
    }
    else if (uni <= 0xD7FF || InRange(uni, 0xE766, 0xFFFF)) {
        u2g = ucs_to_gb18030_index[uni >> 8];

        if ((quint8)(uni & 0xFF) >= u2g.tblBegin && (quint8)(uni & 0xFF) <= u2g.tblEnd) {
            // Use mapping table (2-byte or 4-byte GB18030)
            uint tblEntry;

            tblEntry = ucs_to_gb18030[uni - u2g.tblOffset];

            if (tblEntry > 0x8000) {
                // 2-byte GB18030
                gb = tblEntry;
            }
            else {
                // 4-byte GB18030 stored in a special compact format
                uchar        a, b;
                a = 0x81;
                b = 0x30 + (tblEntry >> 11);
                if (tblEntry >= 0x7000) {
                    a += 3;
                    b -= 14;
                } else if (tblEntry >= 0x6000) {
                    a += 2;
                    b -= 6;
                } else if (tblEntry >= 0x3000) {
                    a += 1;
                    b -= 6;
                } else if (b >= 0x31) {
                    b += 5;
                }
                gbchar[0] = a;
                gbchar[1] = b;
                gbchar[2] = 0x81 + ((tblEntry >> 4) & 0x7F);
                gbchar[3] = 0x30 + (tblEntry & 0xF);
                return 4;
            }
        }
        else {
            // 4-byte GB18030 calculated algorithmically
            gb4lin = u2g.algOffset + (uni & 0xFF);
            // Yikes, my index table could not cover all the bases...
            if (InRange(uni, 0x49B8, 0x49FF))
                gb4lin -= 11;
            gb = gb4lin_to_gb(gb4lin);
        }
    }
    else if (InRange(uni, 0xE000, 0xE765)) {
        // User-defined areas in GB18030 (2-byte)
        if (uni <= 0xE233)
            gb = 0xAAA1 + (((uni - 0xE000) / 94) << 8) + (uni - 0xE000) % 94;
        else if (uni <= 0xE4C5)
            gb = 0xF8A1 + (((uni - 0xE234) / 94) << 8) + (uni - 0xE234) % 94;
        else {
            gb = 0xA140 + (((uni - 0xE4C6) / 96) << 8) + (uni - 0xE4C6) % 96;
            // Skip the gap at 0x7F
            if ((gb & 0xFF) >= 0x7F)
                gb++;
        }
    }
    else if (InRange(uni, 0x10000, 0x10FFFF)) {
        // Qt 3.x does not support beyond BMP yet, but what the heck...
        // (U+10000 = GB+90308130) to (U+10FFFF = GB+E3329A35)
        gb = gb4lin_to_gb(0x1E248 + uni);
    }
    else {
        // Surrogate area and other undefined/reserved areas (discard)
        *gbchar = 0;
        return 0;
    }

    if (gb <= 0xFFFF) {
        gbchar[0] = (uchar)((gb >> 8) & 0xFF);
        gbchar[1] = (uchar)(gb & 0xFF);
        return 2;
    } else {
        gbchar[0] = (uchar)((gb >> 24) & 0xFF);
        gbchar[1] = (uchar)((gb >> 16) & 0xFF);
        gbchar[2] = (uchar)((gb >> 8) & 0xFF);
        gbchar[3] = (uchar)(gb & 0xFF);
        return 4;
    }
}

qt_UnicodeToGbk()

int qt_UnicodeToGbk	(	uint	unicode,
		uchar *	gbchar
	)

Definition at line 9210 of file qgb18030codec.cpp.

Referenced by QGbkCodec::convertFromUnicode(), QGb2312Codec::convertFromUnicode(), QFontGb2312Codec::convertFromUnicode(), and QFontGbkCodec::convertFromUnicode().

                                             {
    /* Returns the bytesize of the GBK character. */
    /* Intended for improving performance of GB2312 and GBK functions. */
    uint        gb;
    indexTbl_t        u2g;

    if (IsLatin(uni)) {
        *gbchar = (uchar)uni;
        return 1;
    }
    else if (uni <= 0xD7FF || InRange(uni, 0xE766, 0xFFFF)) {
        u2g = ucs_to_gb18030_index[uni >> 8];

        if ((quint8)(uni & 0xFF) >= u2g.tblBegin && (quint8)(uni & 0xFF) <= u2g.tblEnd) {
            // Use mapping table (2-byte GBK or 4-byte GB18030)
            uint tblEntry;

            tblEntry = ucs_to_gb18030[uni - u2g.tblOffset];

            if (tblEntry > 0x8000) {
                // GBK
                gb = tblEntry;
            }
            else {
                // 4-byte GB18030 stored in a special compact format (discard)
                *gbchar = 0;
                return 0;
            }
        }
        else {
            // 4-byte GB18030 calculated algorithmically (discard)
            *gbchar = 0;
            return 0;
        }
    }
    else if (InRange(uni, 0xE000, 0xE765)) {
        // User-defined areas in GB18030 (2-byte)
        if (uni <= 0xE233)
            gb = 0xAAA1 + (((uni - 0xE000) / 94) << 8) + (uni - 0xE000) % 94;
        else if (uni <= 0xE4C5)
            gb = 0xF8A1 + (((uni - 0xE234) / 94) << 8) + (uni - 0xE234) % 94;
        else {
            gb = 0xA140 + (((uni - 0xE4C6) / 96) << 8) + (uni - 0xE4C6) % 96;
            // Skip the gap at 0x7F
            if ((gb & 0xFF) >= 0x7F)
                gb++;
        }
    }
    else {
        // Surrogate area and other undefined/reserved areas (discard)
        *gbchar = 0;
        return 0;
    }

    gbchar[0] = (uchar)((gb >> 8) & 0xFF);
    gbchar[1] = (uchar)(gb & 0xFF);
    return 2;
}

Variable Documentation

gb18030_2byte_to_ucs

quint16 const gb18030_2byte_to_ucs[22046]

static

Definition at line 4577 of file qgb18030codec.cpp.

Referenced by qt_Gb18030ToUnicode().

gb18030_4byte_to_ucs

quint16 const gb18030_4byte_to_ucs[6793]

static

Definition at line 7629 of file qgb18030codec.cpp.

Referenced by qt_Gb18030ToUnicode().

gb18030_to_ucs_index

const indexTbl_t gb18030_to_ucs_index[154]

static

Definition at line 736 of file qgb18030codec.cpp.

ucs_to_gb18030

quint16 const ucs_to_gb18030[28839]

static

Definition at line 947 of file qgb18030codec.cpp.

Referenced by qt_UnicodeToGb18030(), and qt_UnicodeToGbk().

ucs_to_gb18030_index

const indexTbl_t ucs_to_gb18030_index[256]

static

Definition at line 816 of file qgb18030codec.cpp.