qtextengine_mac.cpp

Go to the documentation of this file.

/****************************************************************************
**
** Copyright (C) 2014 Digia Plc and/or its subsidiary(-ies).
** Contact: http://www.qt-project.org/legal
**
** This file is part of the QtGui module of the Qt Toolkit.
**
** $QT_BEGIN_LICENSE:LGPL$
** Commercial License Usage
** Licensees holding valid commercial Qt licenses may use this file in
** accordance with the commercial license agreement provided with the
** Software or, alternatively, in accordance with the terms contained in
** a written agreement between you and Digia.  For licensing terms and
** conditions see http://qt.digia.com/licensing.  For further information
** use the contact form at http://qt.digia.com/contact-us.
**
** GNU Lesser General Public License Usage
** Alternatively, this file may be used under the terms of the GNU Lesser
** General Public License version 2.1 as published by the Free Software
** Foundation and appearing in the file LICENSE.LGPL included in the
** packaging of this file.  Please review the following information to
** ensure the GNU Lesser General Public License version 2.1 requirements
** will be met: http://www.gnu.org/licenses/old-licenses/lgpl-2.1.html.
**
** In addition, as a special exception, Digia gives you certain additional
** rights.  These rights are described in the Digia Qt LGPL Exception
** version 1.1, included in the file LGPL_EXCEPTION.txt in this package.
**
** GNU General Public License Usage
** Alternatively, this file may be used under the terms of the GNU
** General Public License version 3.0 as published by the Free Software
** Foundation and appearing in the file LICENSE.GPL included in the
** packaging of this file.  Please review the following information to
** ensure the GNU General Public License version 3.0 requirements will be
** met: http://www.gnu.org/copyleft/gpl.html.
**
**
** $QT_END_LICENSE$
**
****************************************************************************/

#include "qtextengine_p.h"

#include <private/qfontengine_coretext_p.h>
#include <private/qfontengine_mac_p.h>

QT_BEGIN_NAMESPACE

// set the glyph attributes heuristically. Assumes a 1 to 1 relationship between chars and glyphs
// and no reordering.
// also computes logClusters heuristically
static void heuristicSetGlyphAttributes(const QChar *uc, int length, QGlyphLayout *glyphs, unsigned short *logClusters, int num_glyphs)
{
    // ### zeroWidth and justification are missing here!!!!!

    Q_UNUSED(num_glyphs);

//     qDebug("QScriptEngine::heuristicSetGlyphAttributes, num_glyphs=%d", item->num_glyphs);

    const bool symbolFont = false; // ####
    glyphs->attributes[0].mark = false;
    glyphs->attributes[0].clusterStart = true;
    glyphs->attributes[0].dontPrint = (!symbolFont && uc[0].unicode() == 0x00ad) || qIsControlChar(uc[0].unicode());

    int pos = 0;
    int lastCat = QChar::category(uc[0].unicode());
    for (int i = 1; i < length; ++i) {
        if (logClusters[i] == pos)
            // same glyph
            continue;
        ++pos;
        while (pos < logClusters[i]) {
            ++pos;
        }
        // hide soft-hyphens by default
        if ((!symbolFont && uc[i].unicode() == 0x00ad) || qIsControlChar(uc[i].unicode()))
            glyphs->attributes[pos].dontPrint = true;
        const QUnicodeTables::Properties *prop = QUnicodeTables::properties(uc[i].unicode());
        int cat = prop->category;

        // one gets an inter character justification point if the current char is not a non spacing mark.
        // as then the current char belongs to the last one and one gets a space justification point
        // after the space char.
        if (lastCat == QChar::Separator_Space)
            glyphs->attributes[pos-1].justification = HB_Space;
        else if (cat != QChar::Mark_NonSpacing)
            glyphs->attributes[pos-1].justification = HB_Character;
        else
            glyphs->attributes[pos-1].justification = HB_NoJustification;

        lastCat = cat;
    }
    pos = logClusters[length-1];
    if (lastCat == QChar::Separator_Space)
        glyphs->attributes[pos].justification = HB_Space;
    else
        glyphs->attributes[pos].justification = HB_Character;
}

struct QArabicProperties {
    unsigned char shape;
    unsigned char justification;
};
Q_DECLARE_TYPEINFO(QArabicProperties, Q_PRIMITIVE_TYPE);

enum QArabicShape {
    XIsolated,
    XFinal,
    XInitial,
    XMedial,
    // intermediate state
    XCausing
};


// these groups correspond to the groups defined in the Unicode standard.
// Some of these groups are equal with regards to both joining and line breaking behaviour,
// and thus have the same enum value
//
// I'm not sure the mapping of syriac to arabic enums is correct with regards to justification, but as
// I couldn't find any better document I'll hope for the best.
enum ArabicGroup {
    // NonJoining
    ArabicNone,
    ArabicSpace,
    // Transparent
    Transparent,
    // Causing
    Center,
    Kashida,

    // Arabic
    // Dual
    Beh,
    Noon,
    Meem = Noon,
    Heh = Noon,
    KnottedHeh = Noon,
    HehGoal = Noon,
    SwashKaf = Noon,
    Yeh,
    Hah,
    Seen,
    Sad = Seen,
    Tah,
    Kaf = Tah,
    Gaf = Tah,
    Lam = Tah,
    Ain,
    Feh = Ain,
    Qaf = Ain,
    // Right
    Alef,
    Waw,
    Dal,
    TehMarbuta = Dal,
    Reh,
    HamzaOnHehGoal,
    YehWithTail = HamzaOnHehGoal,
    YehBarre = HamzaOnHehGoal,

    // Syriac
    // Dual
    Beth = Beh,
    Gamal = Ain,
    Heth = Noon,
    Teth = Hah,
    Yudh = Noon,
    Kaph = Noon,
    Lamadh = Lam,
    Mim = Noon,
    Nun = Noon,
    Semakh = Noon,
    FinalSemakh = Noon,
    SyriacE = Ain,
    Pe = Ain,
    ReversedPe = Hah,
    Qaph = Noon,
    Shin = Noon,
    Fe = Ain,

    // Right
    Alaph = Alef,
    Dalath = Dal,
    He = Dal,
    SyriacWaw = Waw,
    Zain = Alef,
    YudhHe = Waw,
    Sadhe = HamzaOnHehGoal,
    Taw = Dal,

    // Compiler bug? Otherwise ArabicGroupsEnd would be equal to Dal + 1.
    Dummy = HamzaOnHehGoal,
    ArabicGroupsEnd
};

static const unsigned char arabic_group[0x150] = {
    ArabicNone, ArabicNone, ArabicNone, ArabicNone,
    ArabicNone, ArabicNone, ArabicNone, ArabicNone,
    ArabicNone, ArabicNone, ArabicNone, ArabicNone,
    ArabicNone, ArabicNone, ArabicNone, ArabicNone,

    Transparent, Transparent, Transparent, Transparent,
    Transparent, Transparent, ArabicNone, ArabicNone,
    ArabicNone, ArabicNone, ArabicNone, ArabicNone,
    ArabicNone, ArabicNone, ArabicNone, ArabicNone,

    ArabicNone, ArabicNone, Alef, Alef,
    Waw, Alef, Yeh, Alef,
    Beh, TehMarbuta, Beh, Beh,
    Hah, Hah, Hah, Dal,

    Dal, Reh, Reh, Seen,
    Seen, Sad, Sad, Tah,
    Tah, Ain, Ain, ArabicNone,
    ArabicNone, ArabicNone, ArabicNone, ArabicNone,

    // 0x640
    Kashida, Feh, Qaf, Kaf,
    Lam, Meem, Noon, Heh,
    Waw, Yeh, Yeh, Transparent,
    Transparent, Transparent, Transparent, Transparent,

    Transparent, Transparent, Transparent, Transparent,
    Transparent, Transparent, Transparent, Transparent,
    Transparent, ArabicNone, ArabicNone, ArabicNone,
    ArabicNone, ArabicNone, ArabicNone, ArabicNone,

    ArabicNone, ArabicNone, ArabicNone, ArabicNone,
    ArabicNone, ArabicNone, ArabicNone, ArabicNone,
    ArabicNone, ArabicNone, ArabicNone, ArabicNone,
    ArabicNone, ArabicNone, Beh, Qaf,

    Transparent, Alef, Alef, Alef,
    ArabicNone, Alef, Waw, Waw,
    Yeh, Beh, Beh, Beh,
    Beh, Beh, Beh, Beh,

    // 0x680
    Beh, Hah, Hah, Hah,
    Hah, Hah, Hah, Hah,
    Dal, Dal, Dal, Dal,
    Dal, Dal, Dal, Dal,

    Dal, Reh, Reh, Reh,
    Reh, Reh, Reh, Reh,
    Reh, Reh, Seen, Seen,
    Seen, Sad, Sad, Tah,

    Ain, Feh, Feh, Feh,
    Feh, Feh, Feh, Qaf,
    Qaf, Gaf, SwashKaf, Gaf,
    Kaf, Kaf, Kaf, Gaf,

    Gaf, Gaf, Gaf, Gaf,
    Gaf, Lam, Lam, Lam,
    Lam, Noon, Noon, Noon,
    Noon, Noon, KnottedHeh, Hah,

    // 0x6c0
    TehMarbuta, HehGoal, HamzaOnHehGoal, HamzaOnHehGoal,
    Waw, Waw, Waw, Waw,
    Waw, Waw, Waw, Waw,
    Yeh, YehWithTail, Yeh, Waw,

    Yeh, Yeh, YehBarre, YehBarre,
    ArabicNone, TehMarbuta, Transparent, Transparent,
    Transparent, Transparent, Transparent, Transparent,
    Transparent, ArabicNone, ArabicNone, Transparent,

    Transparent, Transparent, Transparent, Transparent,
    Transparent, ArabicNone, ArabicNone, Transparent,
    Transparent, ArabicNone, Transparent, Transparent,
    Transparent, Transparent, Dal, Reh,

    ArabicNone, ArabicNone, ArabicNone, ArabicNone,
    ArabicNone, ArabicNone, ArabicNone, ArabicNone,
    ArabicNone, ArabicNone, Seen, Sad,
    Ain, ArabicNone, ArabicNone, KnottedHeh,

    // 0x700
    ArabicNone, ArabicNone, ArabicNone, ArabicNone,
    ArabicNone, ArabicNone, ArabicNone, ArabicNone,
    ArabicNone, ArabicNone, ArabicNone, ArabicNone,
    ArabicNone, ArabicNone, ArabicNone, ArabicNone,

    Alaph, Transparent, Beth, Gamal,
    Gamal, Dalath, Dalath, He,
    SyriacWaw, Zain, Heth, Teth,
    Teth, Yudh, YudhHe, Kaph,

    Lamadh, Mim, Nun, Semakh,
    FinalSemakh, SyriacE, Pe, ReversedPe,
    Sadhe, Qaph, Dalath, Shin,
    Taw, Beth, Gamal, Dalath,

    Transparent, Transparent, Transparent, Transparent,
    Transparent, Transparent, Transparent, Transparent,
    Transparent, Transparent, Transparent, Transparent,
    Transparent, Transparent, Transparent, Transparent,

    Transparent, Transparent, Transparent, Transparent,
    Transparent, Transparent, Transparent, Transparent,
    Transparent, Transparent, Transparent, ArabicNone,
    ArabicNone, Zain, Kaph, Fe,
};

static inline ArabicGroup arabicGroup(unsigned short uc)
{
    if (uc >= 0x0600 && uc < 0x750)
        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 = HB_NoJustification;

    for (int i = 1; i < len; ++i) {
        // #### fix handling for spaces and punktuation
        properties[i].justification = HB_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 = HB_Arabic_Seen;
            break;
        case Hah:
            if (properties[lastPos].shape == XFinal)
                properties[lastPos-1].justification = HB_Arabic_HaaDal;
            break;
        case Alef:
            if (properties[lastPos].shape == XFinal)
                properties[lastPos-1].justification = HB_Arabic_Alef;
            break;
        case Ain:
            if (properties[lastPos].shape == XFinal)
                properties[lastPos-1].justification = HB_Arabic_Waw;
            break;
        case Noon:
            if (properties[lastPos].shape == XFinal)
                properties[lastPos-1].justification = HB_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 = HB_Arabic_Space;
            break;
        case Kashida:
            properties[i].justification = HB_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 = HB_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);
}

void QTextEngine::shapeTextMac(int item) const
{
    QScriptItem &si = layoutData->items[item];

    si.glyph_data_offset = layoutData->used;

    QFontEngine *font = fontEngine(si, &si.ascent, &si.descent, &si.leading);
    if (font->type() != QFontEngine::Multi) {
        shapeTextWithHarfbuzz(item);
        return;
    }
    
#ifndef QT_MAC_USE_COCOA
    QFontEngineMacMulti *fe = static_cast<QFontEngineMacMulti *>(font);
#else
    QCoreTextFontEngineMulti *fe = static_cast<QCoreTextFontEngineMulti *>(font);
#endif
    QTextEngine::ShaperFlags flags;
    if (si.analysis.bidiLevel % 2)
        flags |= RightToLeft;
    if (option.useDesignMetrics())
    flags |= DesignMetrics;

    attributes(); // pre-initialize char attributes

    const int len = length(item);
    int num_glyphs = length(item);
    const QChar *str = layoutData->string.unicode() + si.position;
    ushort upperCased[256];
    if (si.analysis.flags == QScriptAnalysis::SmallCaps || si.analysis.flags == QScriptAnalysis::Uppercase
            || si.analysis.flags == QScriptAnalysis::Lowercase) {
        ushort *uc = upperCased;
        if (len > 256)
            uc = new ushort[len];
        for (int i = 0; i < len; ++i) {
            if(si.analysis.flags == QScriptAnalysis::Lowercase)
                uc[i] = str[i].toLower().unicode();
            else
                uc[i] = str[i].toUpper().unicode();
        }
        str = reinterpret_cast<const QChar *>(uc);
    }

    ensureSpace(num_glyphs);
    num_glyphs = layoutData->glyphLayout.numGlyphs - layoutData->used;

    QGlyphLayout g = availableGlyphs(&si);
    g.numGlyphs = num_glyphs;
    unsigned short *log_clusters = logClusters(&si);

    bool stringToCMapFailed = false;
    // Skip shaping of line or paragraph separators since we are not
    // going to draw them anyway
    if (si.analysis.flags == QScriptAnalysis::LineOrParagraphSeparator
        && !(option.flags() & QTextOption::ShowLineAndParagraphSeparators)) {
        memset(log_clusters, 0, len * sizeof(unsigned short));
        goto cleanUp;
    }

    if (!fe->stringToCMap(str, len, &g, &num_glyphs, flags, log_clusters, attributes(), &si)) {
        ensureSpace(num_glyphs);
        g = availableGlyphs(&si);
        stringToCMapFailed = !fe->stringToCMap(str, len, &g, &num_glyphs, flags, log_clusters,
                                               attributes(), &si);
    }

    if (!stringToCMapFailed) {
        heuristicSetGlyphAttributes(str, len, &g, log_clusters, num_glyphs);

        si.num_glyphs = num_glyphs;

        layoutData->used += si.num_glyphs;

        QGlyphLayout g = shapedGlyphs(&si);

        if (si.analysis.script == QUnicodeTables::Arabic) {
            QVarLengthArray<QArabicProperties> props(len + 2);
            QArabicProperties *properties = props.data();
            int f = si.position;
            int l = len;
            if (f > 0) {
                --f;
                ++l;
                ++properties;
            }
            if (f + l < layoutData->string.length()) {
                ++l;
            }
            qt_getArabicProperties((const unsigned short *)(layoutData->string.unicode()+f), l, props.data());

            unsigned short *log_clusters = logClusters(&si);

            for (int i = 0; i < len; ++i) {
                int gpos = log_clusters[i];
                g.attributes[gpos].justification = properties[i].justification;
            }
        }
    }

cleanUp:
    const ushort *uc = reinterpret_cast<const ushort *>(str);

    if ((si.analysis.flags == QScriptAnalysis::SmallCaps || si.analysis.flags == QScriptAnalysis::Uppercase
         || si.analysis.flags == QScriptAnalysis::Lowercase)
        && uc != upperCased)
        delete [] uc;
}

QT_END_NAMESPACE