qscriptengine.cpp

奇趣公司比较新的qt/emd版本

        return (ArabicGroup) arabic_group[uc-0x600];
    else if (uc == 0x200d)
        return Center;
    else if (QChar::category(uc) == QChar::Separator_Space)
        return ArabicSpace;
    else
        return ArabicNone;
}


/*
   Arabic shaping obeys a number of rules according to the joining classes (see Unicode book, section on
   arabic).

   Each unicode char has a joining class (right, dual (left&right), center (joincausing) or transparent).
   transparent joining is not encoded in QChar::joining(), but applies to all combining marks and format marks.

   Right join-causing: dual + center
   Left join-causing: dual + right + center

   Rules are as follows (for a string already in visual order, as we have it here):

   R1 Transparent characters do not affect joining behaviour.
   R2 A right joining character, that has a right join-causing char on the right will get form XRight
   (R3 A left joining character, that has a left join-causing char on the left will get form XLeft)
   Note: the above rule is meaningless, as there are no pure left joining characters defined in Unicode
   R4 A dual joining character, that has a left join-causing char on the left and a right join-causing char on
             the right will get form XMedial
   R5  A dual joining character, that has a right join causing char on the right, and no left join causing char on the left
         will get form XRight
   R6 A dual joining character, that has a  left join causing char on the left, and no right join causing char on the right
         will get form XLeft
   R7 Otherwise the character will get form XIsolated

   Additionally we have to do the minimal ligature support for lam-alef ligatures:

   L1 Transparent characters do not affect ligature behaviour.
   L2 Any sequence of Alef(XRight) + Lam(XMedial) will form the ligature Alef.Lam(XLeft)
   L3 Any sequence of Alef(XRight) + Lam(XLeft) will form the ligature Alef.Lam(XIsolated)

   The state table below handles rules R1-R7.
*/

enum Joining {
    JNone,
    JCausing,
    JDual,
    JRight,
    JTransparent
};

static const Joining joining_for_group[ArabicGroupsEnd] = {
    // NonJoining
    JNone, // ArabicNone
    JNone, // ArabicSpace
    // Transparent
    JTransparent, // Transparent
    // Causing
    JCausing, // Center
    JCausing, // Kashida
    // Dual
    JDual, // Beh
    JDual, // Noon
    JDual, // Yeh
    JDual, // Hah
    JDual, // Seen
    JDual, // Tah
    JDual, // Ain
    // Right
    JRight, // Alef
    JRight, // Waw
    JRight, // Dal
    JRight, // Reh
    JRight  // HamzaOnHehGoal
};


struct JoiningPair {
    QArabicShape form1;
    QArabicShape form2;
};

static const JoiningPair joining_table[5][4] =
// None, Causing, Dual, Right
{
    { { XIsolated, XIsolated }, { XIsolated, XCausing }, { XIsolated, XInitial }, { XIsolated, XIsolated } }, // XIsolated
    { { XFinal, XIsolated }, { XFinal, XCausing }, { XFinal, XInitial }, { XFinal, XIsolated } }, // XFinal
    { { XIsolated, XIsolated }, { XInitial, XCausing }, { XInitial, XMedial }, { XInitial, XFinal } }, // XInitial
    { { XFinal, XIsolated }, { XMedial, XCausing }, { XMedial, XMedial }, { XMedial, XFinal } }, // XMedial
    { { XIsolated, XIsolated }, { XIsolated, XCausing }, { XIsolated, XMedial }, { XIsolated, XFinal } }, // XCausing
};


/*
According to http://www.microsoft.com/middleeast/Arabicdev/IE6/KBase.asp

1. Find the priority of the connecting opportunities in each word
2. Add expansion at the highest priority connection opportunity
3. If more than one connection opportunity have the same highest value,
   use the opportunity closest to the end of the word.

Following is a chart that provides the priority for connection
opportunities and where expansion occurs. The character group names
are those in table 6.6 of the UNICODE 2.0 book.


PrioritY        Glyph                   Condition                                       Kashida Location

Arabic_Kashida        User inserted Kashida   The user entered a Kashida in a position.       After the user
                (Shift+j or Shift+[E with hat])    Thus, it is the highest priority to insert an   inserted kashida
                                        automatic kashida.

Arabic_Seen        Seen, Sad               Connecting to the next character.               After the character.
                                        (Initial or medial form).

Arabic_HaaDal        Teh Marbutah, Haa, Dal  Connecting to previous character.               Before the final form
                                                                                        of these characters.

Arabic_Alef     Alef, Tah, Lam,         Connecting to previous character.               Before the final form
                Kaf and Gaf                                                             of these characters.

Arabic_BaRa     Reh, Yeh                Connected to medial Beh                         Before preceding medial Baa

Arabic_Waw        Waw, Ain, Qaf, Feh      Connecting to previous character.               Before the final form of
                                                                                        these characters.

Arabic_Normal   Other connecting        Connecting to previous character.               Before the final form
                characters                                                              of these characters.



This seems to imply that we have at most one kashida point per arabic word.

*/

void qt_getArabicProperties(const unsigned short *chars, int len, QArabicProperties *properties)
{
//     qDebug("arabicSyriacOpenTypeShape: properties:");
    int lastPos = 0;
    int lastGroup = ArabicNone;

    ArabicGroup group = arabicGroup(chars[0]);
    Joining j = joining_for_group[group];
    QArabicShape shape = joining_table[XIsolated][j].form2;
    properties[0].justification = QGlyphLayout::NoJustification;

    for (int i = 1; i < len; ++i) {
        // #### fix handling for spaces and punktuation
        properties[i].justification = QGlyphLayout::NoJustification;

        group = arabicGroup(chars[i]);
        j = joining_for_group[group];

        if (j == JTransparent) {
            properties[i].shape = XIsolated;
            continue;
        }

        properties[lastPos].shape = joining_table[shape][j].form1;
        shape = joining_table[shape][j].form2;

        switch(lastGroup) {
        case Seen:
            if (properties[lastPos].shape == XInitial || properties[lastPos].shape == XMedial)
                properties[i-1].justification = QGlyphLayout::Arabic_Seen;
            break;
        case Hah:
            if (properties[lastPos].shape == XFinal)
                properties[lastPos-1].justification = QGlyphLayout::Arabic_HaaDal;
            break;
        case Alef:
            if (properties[lastPos].shape == XFinal)
                properties[lastPos-1].justification = QGlyphLayout::Arabic_Alef;
            break;
        case Ain:
            if (properties[lastPos].shape == XFinal)
                properties[lastPos-1].justification = QGlyphLayout::Arabic_Waw;
            break;
        case Noon:
            if (properties[lastPos].shape == XFinal)
                properties[lastPos-1].justification = QGlyphLayout::Arabic_Normal;
            break;
        case ArabicNone:
            break;

        default:
            Q_ASSERT(false);
        }

        lastGroup = ArabicNone;

        switch(group) {
        case ArabicNone:
        case Transparent:
        // ### Center should probably be treated as transparent when it comes to justification.
        case Center:
            break;
        case ArabicSpace:
            properties[i].justification = QGlyphLayout::Arabic_Space;
            break;
        case Kashida:
            properties[i].justification = QGlyphLayout::Arabic_Kashida;
            break;
        case Seen:
            lastGroup = Seen;
            break;

        case Hah:
        case Dal:
            lastGroup = Hah;
            break;

        case Alef:
        case Tah:
            lastGroup = Alef;
            break;

        case Yeh:
        case Reh:
            if (properties[lastPos].shape == XMedial && arabicGroup(chars[lastPos]) == Beh)
                properties[lastPos-1].justification = QGlyphLayout::Arabic_BaRa;
            break;

        case Ain:
        case Waw:
            lastGroup = Ain;
            break;

        case Noon:
        case Beh:
        case HamzaOnHehGoal:
            lastGroup = Noon;
            break;
        case ArabicGroupsEnd:
            Q_ASSERT(false);
        }

        lastPos = i;
    }
    properties[lastPos].shape = joining_table[shape][JNone].form1;


//     for (int i = 0; i < len; ++i)
//         qDebug("arabic properties(%d): uc=%x shape=%d, justification=%d", i, chars[i], properties[i].shape, properties[i].justification);
}






// The unicode to unicode shaping codec.
// does only presentation forms B at the moment, but that should be enough for
// simple display
static const ushort arabicUnicodeMapping[256][2] = {
    // base of shaped forms, and number-1 of them (0 for non shaping,
    // 1 for right binding and 3 for dual binding

    // These are just the glyphs available in Unicode,
    // some characters are in R class, but have no glyphs in Unicode.

    { 0x0600, 0 }, // 0x0600
    { 0x0601, 0 }, // 0x0601
    { 0x0602, 0 }, // 0x0602
    { 0x0603, 0 }, // 0x0603
    { 0x0604, 0 }, // 0x0604
    { 0x0605, 0 }, // 0x0605
    { 0x0606, 0 }, // 0x0606
    { 0x0607, 0 }, // 0x0607
    { 0x0608, 0 }, // 0x0608
    { 0x0609, 0 }, // 0x0609
    { 0x060A, 0 }, // 0x060A
    { 0x060B, 0 }, // 0x060B
    { 0x060C, 0 }, // 0x060C
    { 0x060D, 0 }, // 0x060D
    { 0x060E, 0 }, // 0x060E
    { 0x060F, 0 }, // 0x060F

    { 0x0610, 0 }, // 0x0610
    { 0x0611, 0 }, // 0x0611
    { 0x0612, 0 }, // 0x0612
    { 0x0613, 0 }, // 0x0613
    { 0x0614, 0 }, // 0x0614
    { 0x0615, 0 }, // 0x0615
    { 0x0616, 0 }, // 0x0616
    { 0x0617, 0 }, // 0x0617
    { 0x0618, 0 }, // 0x0618
    { 0x0619, 0 }, // 0x0619
    { 0x061A, 0 }, // 0x061A
    { 0x061B, 0 }, // 0x061B
    { 0x061C, 0 }, // 0x061C
    { 0x061D, 0 }, // 0x061D
    { 0x061E, 0 }, // 0x061E
    { 0x061F, 0 }, // 0x061F

    { 0x0620, 0 }, // 0x0620
    { 0xFE80, 0 }, // 0x0621            HAMZA
    { 0xFE81, 1 }, // 0x0622    R       ALEF WITH MADDA ABOVE
    { 0xFE83, 1 }, // 0x0623    R       ALEF WITH HAMZA ABOVE
    { 0xFE85, 1 }, // 0x0624    R       WAW WITH HAMZA ABOVE
    { 0xFE87, 1 }, // 0x0625    R       ALEF WITH HAMZA BELOW
    { 0xFE89, 3 }, // 0x0626    D       YEH WITH HAMZA ABOVE
    { 0xFE8D, 1 }, // 0x0627    R       ALEF
    { 0xFE8F, 3 }, // 0x0628    D       BEH
    { 0xFE93, 1 }, // 0x0629    R       TEH MARBUTA
    { 0xFE95, 3 }, // 0x062A    D       TEH
    { 0xFE99, 3 }, // 0x062B    D       THEH
    { 0xFE9D, 3 }, // 0x062C    D       JEEM
    { 0xFEA1, 3 }, // 0x062D    D       HAH
    { 0xFEA5, 3 }, // 0x062E    D       KHAH
    { 0xFEA9, 1 }, // 0x062F    R       DAL

    { 0xFEAB, 1 }, // 0x0630    R       THAL
    { 0xFEAD, 1 }, // 0x0631    R       REH
    { 0xFEAF, 1 }, // 0x0632    R       ZAIN
    { 0xFEB1, 3 }, // 0x0633    D       SEEN
    { 0xFEB5, 3 }, // 0x0634    D       SHEEN
    { 0xFEB9, 3 }, // 0x0635    D       SAD
    { 0xFEBD, 3 }, // 0x0636    D       DAD
    { 0xFEC1, 3 }, // 0x0637    D       TAH
    { 0xFEC5, 3 }, // 0x0638    D       ZAH
    { 0xFEC9, 3 }, // 0x0639    D       AIN
    { 0xFECD, 3 }, // 0x063A    D       GHAIN
    { 0x063B, 0 }, // 0x063B
    { 0x063C, 0 }, // 0x063C
    { 0x063D, 0 }, // 0x063D
    { 0x063E, 0 }, // 0x063E
    { 0x063F, 0 }, // 0x063F

    { 0x0640, 0 }, // 0x0640    C       TATWEEL // ### Join Causing, only one glyph
    { 0xFED1, 3 }, // 0x0641    D       FEH
    { 0xFED5, 3 }, // 0x0642    D       QAF
    { 0xFED9, 3 }, // 0x0643    D       KAF
    { 0xFEDD, 3 }, // 0x0644    D       LAM
    { 0xFEE1, 3 }, // 0x0645    D       MEEM
    { 0xFEE5, 3 }, // 0x0646    D       NOON
    { 0xFEE9, 3 }, // 0x0647    D       HEH
    { 0xFEED, 1 }, // 0x0648    R       WAW
    { 0x0649, 3 }, // 0x0649            ALEF MAKSURA // ### Dual, glyphs not consecutive, handle in code.
    { 0xFEF1, 3 }, // 0x064A    D       YEH
    { 0x064B, 0 }, // 0x064B
    { 0x064C, 0 }, // 0x064C
    { 0x064D, 0 }, // 0x064D
    { 0x064E, 0 }, // 0x064E
    { 0x064F, 0 }, // 0x064F

    { 0x0650, 0 }, // 0x0650
    { 0x0651, 0 }, // 0x0651
    { 0x0652, 0 }, // 0x0652
    { 0x0653, 0 }, // 0x0653
    { 0x0654, 0 }, // 0x0654
    { 0x0655, 0 }, // 0x0655
    { 0x0656, 0 }, // 0x0656
    { 0x0657, 0 }, // 0x0657
    { 0x0658, 0 }, // 0x0658
    { 0x0659, 0 }, // 0x0659
    { 0x065A, 0 }, // 0x065A
    { 0x065B, 0 }, // 0x065B
    { 0x065C, 0 }, // 0x065C
    { 0x065D, 0 }, // 0x065D
    { 0x065E, 0 }, // 0x065E
    { 0x065F, 0 }, // 0x065F

    { 0x0660, 0 }, // 0x0660
    { 0x0661, 0 }, // 0x0661
    { 0x0662, 0 }, // 0x0662
    { 0x0663, 0 }, // 0x0663
    { 0x0664, 0 }, // 0x0664
    { 0x0665, 0 }, // 0x0665
    { 0x0666, 0 }, // 0x0666
    { 0x0667, 0 }, // 0x0667
    { 0x0668, 0 }, // 0x0668
    { 0x0669, 0 }, // 0x0669
    { 0x066A, 0 }, // 0x066A
    { 0x066B, 0 }, // 0x066B
    { 0x066C, 0 }, // 0x066C
    { 0x066D, 0 }, // 0x066D
    { 0x066E, 0 }, // 0x066E
    { 0x066F, 0 }, // 0x066F

    { 0x0670, 0 }, // 0x0670