qscriptengine.cpp

奇趣公司比较新的qt/emd版本

    { 0xFB50, 1 }, // 0x0671    R       ALEF WASLA
    { 0x0672, 0 }, // 0x0672
    { 0x0673, 0 }, // 0x0673
    { 0x0674, 0 }, // 0x0674
    { 0x0675, 0 }, // 0x0675
    { 0x0676, 0 }, // 0x0676
    { 0x0677, 0 }, // 0x0677
    { 0x0678, 0 }, // 0x0678
    { 0xFB66, 3 }, // 0x0679    D       TTEH
    { 0xFB5E, 3 }, // 0x067A    D       TTEHEH
    { 0xFB52, 3 }, // 0x067B    D       BEEH
    { 0x067C, 0 }, // 0x067C
    { 0x067D, 0 }, // 0x067D
    { 0xFB56, 3 }, // 0x067E    D       PEH
    { 0xFB62, 3 }, // 0x067F    D       TEHEH

    { 0xFB5A, 3 }, // 0x0680    D       BEHEH
    { 0x0681, 0 }, // 0x0681
    { 0x0682, 0 }, // 0x0682
    { 0xFB76, 3 }, // 0x0683    D       NYEH
    { 0xFB72, 3 }, // 0x0684    D       DYEH
    { 0x0685, 0 }, // 0x0685
    { 0xFB7A, 3 }, // 0x0686    D       TCHEH
    { 0xFB7E, 3 }, // 0x0687    D       TCHEHEH
    { 0xFB88, 1 }, // 0x0688    R       DDAL
    { 0x0689, 0 }, // 0x0689
    { 0x068A, 0 }, // 0x068A
    { 0x068B, 0 }, // 0x068B
    { 0xFB84, 1 }, // 0x068C    R       DAHAL
    { 0xFB82, 1 }, // 0x068D    R       DDAHAL
    { 0xFB86, 1 }, // 0x068E    R       DUL
    { 0x068F, 0 }, // 0x068F

    { 0x0690, 0 }, // 0x0690
    { 0xFB8C, 1 }, // 0x0691    R       RREH
    { 0x0692, 0 }, // 0x0692
    { 0x0693, 0 }, // 0x0693
    { 0x0694, 0 }, // 0x0694
    { 0x0695, 0 }, // 0x0695
    { 0x0696, 0 }, // 0x0696
    { 0x0697, 0 }, // 0x0697
    { 0xFB8A, 1 }, // 0x0698    R       JEH
    { 0x0699, 0 }, // 0x0699
    { 0x069A, 0 }, // 0x069A
    { 0x069B, 0 }, // 0x069B
    { 0x069C, 0 }, // 0x069C
    { 0x069D, 0 }, // 0x069D
    { 0x069E, 0 }, // 0x069E
    { 0x069F, 0 }, // 0x069F

    { 0x06A0, 0 }, // 0x06A0
    { 0x06A1, 0 }, // 0x06A1
    { 0x06A2, 0 }, // 0x06A2
    { 0x06A3, 0 }, // 0x06A3
    { 0xFB6A, 3 }, // 0x06A4    D       VEH
    { 0x06A5, 0 }, // 0x06A5
    { 0xFB6E, 3 }, // 0x06A6    D       PEHEH
    { 0x06A7, 0 }, // 0x06A7
    { 0x06A8, 0 }, // 0x06A8
    { 0xFB8E, 3 }, // 0x06A9    D       KEHEH
    { 0x06AA, 0 }, // 0x06AA
    { 0x06AB, 0 }, // 0x06AB
    { 0x06AC, 0 }, // 0x06AC
    { 0xFBD3, 3 }, // 0x06AD    D       NG
    { 0x06AE, 0 }, // 0x06AE
    { 0xFB92, 3 }, // 0x06AF    D       GAF

    { 0x06B0, 0 }, // 0x06B0
    { 0xFB9A, 3 }, // 0x06B1    D       NGOEH
    { 0x06B2, 0 }, // 0x06B2
    { 0xFB96, 3 }, // 0x06B3    D       GUEH
    { 0x06B4, 0 }, // 0x06B4
    { 0x06B5, 0 }, // 0x06B5
    { 0x06B6, 0 }, // 0x06B6
    { 0x06B7, 0 }, // 0x06B7
    { 0x06B8, 0 }, // 0x06B8
    { 0x06B9, 0 }, // 0x06B9
    { 0xFB9E, 1 }, // 0x06BA    R       NOON GHUNNA
    { 0xFBA0, 3 }, // 0x06BB    D       RNOON
    { 0x06BC, 0 }, // 0x06BC
    { 0x06BD, 0 }, // 0x06BD
    { 0xFBAA, 3 }, // 0x06BE    D       HEH DOACHASHMEE
    { 0x06BF, 0 }, // 0x06BF

    { 0xFBA4, 1 }, // 0x06C0    R       HEH WITH YEH ABOVE
    { 0xFBA6, 3 }, // 0x06C1    D       HEH GOAL
    { 0x06C2, 0 }, // 0x06C2
    { 0x06C3, 0 }, // 0x06C3
    { 0x06C4, 0 }, // 0x06C4
    { 0xFBE0, 1 }, // 0x06C5    R       KIRGHIZ OE
    { 0xFBD9, 1 }, // 0x06C6    R       OE
    { 0xFBD7, 1 }, // 0x06C7    R       U
    { 0xFBDB, 1 }, // 0x06C8    R       YU
    { 0xFBE2, 1 }, // 0x06C9    R       KIRGHIZ YU
    { 0x06CA, 0 }, // 0x06CA
    { 0xFBDE, 1 }, // 0x06CB    R       VE
    { 0xFBFC, 3 }, // 0x06CC    D       FARSI YEH
    { 0x06CD, 0 }, // 0x06CD
    { 0x06CE, 0 }, // 0x06CE
    { 0x06CF, 0 }, // 0x06CF

    { 0xFBE4, 3 }, // 0x06D0    D       E
    { 0x06D1, 0 }, // 0x06D1
    { 0xFBAE, 1 }, // 0x06D2    R       YEH BARREE
    { 0xFBB0, 1 }, // 0x06D3    R       YEH BARREE WITH HAMZA ABOVE
    { 0x06D4, 0 }, // 0x06D4
    { 0x06D5, 0 }, // 0x06D5
    { 0x06D6, 0 }, // 0x06D6
    { 0x06D7, 0 }, // 0x06D7
    { 0x06D8, 0 }, // 0x06D8
    { 0x06D9, 0 }, // 0x06D9
    { 0x06DA, 0 }, // 0x06DA
    { 0x06DB, 0 }, // 0x06DB
    { 0x06DC, 0 }, // 0x06DC
    { 0x06DD, 0 }, // 0x06DD
    { 0x06DE, 0 }, // 0x06DE
    { 0x06DF, 0 }, // 0x06DF

    { 0x06E0, 0 }, // 0x06E0
    { 0x06E1, 0 }, // 0x06E1
    { 0x06E2, 0 }, // 0x06E2
    { 0x06E3, 0 }, // 0x06E3
    { 0x06E4, 0 }, // 0x06E4
    { 0x06E5, 0 }, // 0x06E5
    { 0x06E6, 0 }, // 0x06E6
    { 0x06E7, 0 }, // 0x06E7
    { 0x06E8, 0 }, // 0x06E8
    { 0x06E9, 0 }, // 0x06E9
    { 0x06EA, 0 }, // 0x06EA
    { 0x06EB, 0 }, // 0x06EB
    { 0x06EC, 0 }, // 0x06EC
    { 0x06ED, 0 }, // 0x06ED
    { 0x06EE, 0 }, // 0x06EE
    { 0x06EF, 0 }, // 0x06EF

    { 0x06F0, 0 }, // 0x06F0
    { 0x06F1, 0 }, // 0x06F1
    { 0x06F2, 0 }, // 0x06F2
    { 0x06F3, 0 }, // 0x06F3
    { 0x06F4, 0 }, // 0x06F4
    { 0x06F5, 0 }, // 0x06F5
    { 0x06F6, 0 }, // 0x06F6
    { 0x06F7, 0 }, // 0x06F7
    { 0x06F8, 0 }, // 0x06F8
    { 0x06F9, 0 }, // 0x06F9
    { 0x06FA, 0 }, // 0x06FA
    { 0x06FB, 0 }, // 0x06FB
    { 0x06FC, 0 }, // 0x06FC
    { 0x06FD, 0 }, // 0x06FD
    { 0x06FE, 0 }, // 0x06FE
    { 0x06FF, 0 }  // 0x06FF
};

// the arabicUnicodeMapping does not work for U+0649 ALEF MAKSURA, this table does
static const ushort alefMaksura[4] = {0xFEEF, 0xFEF0, 0xFBE8, 0xFBE9};

// this is a bit tricky. Alef always binds to the right, so the second parameter descibing the shape
// of the lam can be either initial of medial. So initial maps to the isolated form of the ligature,
// medial to the final form
static const ushort arabicUnicodeLamAlefMapping[6][4] = {
    { 0xfffd, 0xfffd, 0xfef5, 0xfef6 }, // 0x622        R       Alef with Madda above
    { 0xfffd, 0xfffd, 0xfef7, 0xfef8 }, // 0x623        R       Alef with Hamza above
    { 0xfffd, 0xfffd, 0xfffd, 0xfffd }, // 0x624        // Just to fill the table ;-)
    { 0xfffd, 0xfffd, 0xfef9, 0xfefa }, // 0x625        R       Alef with Hamza below
    { 0xfffd, 0xfffd, 0xfffd, 0xfffd }, // 0x626        // Just to fill the table ;-)
    { 0xfffd, 0xfffd, 0xfefb, 0xfefc }  // 0x627        R       Alef
};

static inline int getShape(uchar cell, int shape)
{
    // the arabicUnicodeMapping does not work for U+0649 ALEF MAKSURA, handle this here
    uint ch = (cell != 0x49)
              ? (shape ? arabicUnicodeMapping[cell][0] + shape : 0x600+cell)
              : alefMaksura[shape] ;
    return ch;
}


/*
  Two small helper functions for arabic shaping.
*/
static inline const QChar prevChar(const QString *str, int pos)
{
    //qDebug("leftChar: pos=%d", pos);
    pos--;
    const QChar *ch = str->unicode() + pos;
    while(pos > -1) {
        if(QChar::category(ch->unicode()) != QChar::Mark_NonSpacing)
            return *ch;
        pos--;
        ch--;
    }
    return QChar::ReplacementCharacter;
}

static inline const QChar nextChar(const QString *str, int pos)
{
    pos++;
    int len = str->length();
    const QChar *ch = str->unicode() + pos;
    while(pos < len) {
        //qDebug("rightChar: %d isLetter=%d, joining=%d", pos, ch.isLetter(), ch.joining());
        if(QChar::category(ch->unicode()) != QChar::Mark_NonSpacing)
            return *ch;
        // assume it's a transparent char, this might not be 100% correct
        pos++;
        ch++;
    }
    return QChar::ReplacementCharacter;
}

#ifndef Q_WS_MAC
static void shapedString(const QString *uc, int from, int len, QChar *shapeBuffer, int *shapedLength,
                         bool reverse, QGlyphLayout *glyphs, unsigned short *logClusters)
{
    Q_ASSERT(uc->length() >= from + len);

    if(len == 0) {
        *shapedLength = 0;
        return;
    }

    QVarLengthArray<QArabicProperties> props(len+2);
    QArabicProperties *properties = props.data();
    int f = from;
    int l = len;
    if (from > 0) {
        --f;
        ++l;
        ++properties;
    }
    if (f + l < uc->length()) {
        ++l;
    }
    qt_getArabicProperties((const unsigned short *)(uc->unicode()+f), l, props.data());

    const QChar *ch = uc->unicode() + from;
    QChar *data = shapeBuffer;
    int clusterStart = 0;

    for (int i = 0; i < len; i++) {
        uchar r = ch->row();
        int gpos = data - shapeBuffer;

        if (r != 0x06) {
            if (r == 0x20) {
                uchar c = ch->cell();
                if (c == 0x0c || c == 0x0d)
                    // remove ZWJ and ZWNJ
                    goto skip;
            }
            if (reverse)
                *data = QChar::mirroredChar(ch->unicode());
            else
                *data = *ch;
        } else {
            uchar c = ch->cell();
            int pos = i + from;
            int shape = properties[i].shape;
//            qDebug("mapping U+%x to shape %d glyph=0x%x", ch->unicode(), shape, getShape(c, shape));
            // take care of lam-alef ligatures (lam right of alef)
            ushort map;
            switch (c) {
                case 0x44: { // lam
                    const QChar pch = nextChar(uc, pos);
                    if (pch.row() == 0x06) {
                        switch (pch.cell()) {
                            case 0x22:
                            case 0x23:
                            case 0x25:
                            case 0x27:
//                                 qDebug(" lam of lam-alef ligature");
                                map = arabicUnicodeLamAlefMapping[pch.cell() - 0x22][shape];
                                goto next;
                            default:
                                break;
                        }
                    }
                    break;
                }
                case 0x22: // alef with madda
                case 0x23: // alef with hamza above
                case 0x25: // alef with hamza below
                case 0x27: // alef
                    if (prevChar(uc, pos).unicode() == 0x0644) {
                        // have a lam alef ligature
                        //qDebug(" alef of lam-alef ligature");
                        goto skip;
                    }
                default:
                    break;
            }
            map = getShape(c, shape);
        next:
            *data = map;
        }
        // ##### Fixme
        //glyphs[gpos].attributes.zeroWidth = zeroWidth;
        if (QChar::category(ch->unicode()) == QChar::Mark_NonSpacing) {
            glyphs[gpos].attributes.mark = true;
//             qDebug("glyph %d (char %d) is mark!", gpos, i);
        } else {
            glyphs[gpos].attributes.mark = false;
            clusterStart = data - shapeBuffer;
        }
        glyphs[gpos].attributes.clusterStart = !glyphs[gpos].attributes.mark;
        glyphs[gpos].attributes.combiningClass = QChar::combiningClass(ch->unicode());
        glyphs[gpos].attributes.justification = properties[i].justification;
//         qDebug("data[%d] = %x (from %x)", gpos, (uint)data->unicode(), ch->unicode());
        data++;
    skip:
        ch++;
        logClusters[i] = clusterStart;
    }
    *shapedLength = data - shapeBuffer;
}
#endif

#ifndef QT_NO_OPENTYPE

enum {
    InitProperty = 0x2,
    IsolProperty = 0x4,
    FinaProperty = 0x8,
    MediProperty = 0x10,
    RligProperty = 0x20,
    CaltProperty = 0x40,
    LigaProperty = 0x80,
    DligProperty = 0x100,
    CswhProperty = 0x200,
    MsetProperty = 0x400
};

static const QOpenType::Features arabic_features[] = {
    { FT_MAKE_TAG('c', 'c', 'm', 'p'), CcmpProperty },
    { FT_MAKE_TAG('i', 's', 'o', 'l'), IsolProperty },
    { FT_MAKE_TAG('f', 'i', 'n', 'a'), FinaProperty },
    { FT_MAKE_TAG('m', 'e', 'd', 'i'), MediProperty },
    { FT_MAKE_TAG('i', 'n', 'i', 't'), InitProperty },
    { FT_MAKE_TAG('r', 'l', 'i', 'g'), RligProperty },
    { FT_MAKE_TAG('c', 'a', 'l', 't'), CaltProperty },
    { FT_MAKE_TAG('l', 'i', 'g', 'a'), LigaProperty },
    { FT_MAKE_TAG('d', 'l', 'i', 'g'), DligProperty },
    { FT_MAKE_TAG('c', 's', 'w', 'h'), CswhProperty },
    // mset is used in old Win95 fonts that don't have a 'mark' positioning table.
    { FT_MAKE_TAG('m', 's', 'e', 't'), MsetProperty },
    {0, 0}
};

static const QOpenType::Features syriac_features[] = {
    { FT_MAKE_TAG('c', 'c', 'm', 'p'), CcmpProperty },
    { FT_MAKE_TAG('i', 's', 'o', 'l'), IsolProperty },
    { FT_MAKE_TAG('f', 'i', 'n', 'a'), FinaProperty },
    { FT_MAKE_TAG('f', 'i', 'n', '2'), FinaProperty },
    { FT_MAKE_TAG('f', 'i', 'n', '3'), FinaProperty },
    { FT_MAKE_TAG('m', 'e', 'd', 'i'), MediProperty },
    { FT_MAKE_TAG('m', 'e', 'd', '2'), MediProperty },
    { FT_MAKE_TAG('i', 'n', 'i', 't'), InitProperty },
    { FT_MAKE_TAG('r', 'l', 'i', 'g'), RligProperty },
    { FT_MAKE_TAG('c', 'a', 'l', 't'), CaltProperty },
    { FT_MAKE_TAG('l', 'i', 'g', 'a'), LigaProperty },
    { FT_MAKE_TAG('d', 'l', 'i', 'g'), DligProperty },
    {0, 0}
};

static bool arabicSyriacOpenTypeShape(QOpenType *openType, QShaperItem *item, bool *ot_ok)
{
    *ot_ok = true;

    openType->selectScript(item, item->script, item->script == QUnicodeTables::Arabic ? arabic_features : syriac_features);
    const int nglyphs = item->num_glyphs;
    if (!item->font->stringToCMap(item->string->unicode()+item->from, item->length, item->glyphs, &item->num_glyphs, QFlag(item->flags)))
        return false;
    heuristicSetGlyphAttributes(item);

    const unsigned short *uc = (const unsigned short *)item->string->unicode() + item->from;

    QVarLengthArray<QArabicProperties> props(item->length+2);
    QArabicProperties *properties = props.data();
    int f = 0;
    int l = item->length;