Add support for polyphonic greek

Merge patch c0006e05f32ecd6f16825b799d2bce345c166433 from
harfbuzz and add autotests to QTextScriptEngine.

Support for polyphonic greek is implemented by mapping decomposed
greek character sequences to their composed characters in the
greek extended unicode range (U+1f00 - U+1fff).

Task-number: QTBUG-391
Reviewed-By: Simon Hausmann

5 files changed, 110 insertions, 24 deletions

diff --git a/src/3rdparty/harfbuzz/src/Makefile.am b/src/3rdparty/harfbuzz/src/Makefile.am
index 2b0fb1d..51d0652 100644
--- a/src/3rdparty/harfbuzz/src/Makefile.am
+++ b/src/3rdparty/harfbuzz/src/Makefile.am
@@ -12,7 +12,8 @@ MAINSOURCES =  \
 	harfbuzz-impl.c \
 	harfbuzz-open.c \
 	harfbuzz-shaper.cpp \
-	harfbuzz-tibetan.c \
+        harfbuzz-greek.c \
+        harfbuzz-tibetan.c \
 	harfbuzz-khmer.c \
 	harfbuzz-indic.cpp \
 	harfbuzz-hebrew.c \
diff --git a/src/3rdparty/harfbuzz/src/harfbuzz-shaper-all.cpp b/src/3rdparty/harfbuzz/src/harfbuzz-shaper-all.cpp
index d2f902f..2dae501 100644
--- a/src/3rdparty/harfbuzz/src/harfbuzz-shaper-all.cpp
+++ b/src/3rdparty/harfbuzz/src/harfbuzz-shaper-all.cpp
@@ -25,6 +25,7 @@
 #include "harfbuzz-shaper.cpp"
 #include "harfbuzz-indic.cpp"
 extern "C" {
+#include "harfbuzz-greek.c"
 #include "harfbuzz-tibetan.c"
 #include "harfbuzz-khmer.c"
 #include "harfbuzz-hebrew.c"
diff --git a/src/3rdparty/harfbuzz/src/harfbuzz-shaper-private.h b/src/3rdparty/harfbuzz/src/harfbuzz-shaper-private.h
index 80bccf8..11ed753 100644
--- a/src/3rdparty/harfbuzz/src/harfbuzz-shaper-private.h
+++ b/src/3rdparty/harfbuzz/src/harfbuzz-shaper-private.h
@@ -100,6 +100,7 @@ typedef struct {
 extern const HB_ScriptEngine hb_scriptEngines[];
 
 extern HB_Bool HB_BasicShape(HB_ShaperItem *shaper_item);
+extern HB_Bool HB_GreekShape(HB_ShaperItem *shaper_item);
 extern HB_Bool HB_TibetanShape(HB_ShaperItem *shaper_item);
 extern HB_Bool HB_HebrewShape(HB_ShaperItem *shaper_item);
 extern HB_Bool HB_ArabicShape(HB_ShaperItem *shaper_item);
diff --git a/src/3rdparty/harfbuzz/src/harfbuzz-shaper.cpp b/src/3rdparty/harfbuzz/src/harfbuzz-shaper.cpp
index bfc7bd4..4bc53c8 100644
--- a/src/3rdparty/harfbuzz/src/harfbuzz-shaper.cpp
+++ b/src/3rdparty/harfbuzz/src/harfbuzz-shaper.cpp
@@ -587,7 +587,7 @@ const HB_ScriptEngine HB_ScriptEngines[] = {
     // Common
     { HB_BasicShape, 0},
     // Greek
-    { HB_BasicShape, 0},
+    { HB_GreekShape, 0},
     // Cyrillic
     { HB_BasicShape, 0},
     // Armenian
diff --git a/src/3rdparty/harfbuzz/tests/shaping/main.cpp b/src/3rdparty/harfbuzz/tests/shaping/main.cpp
index 827ac30..b48b0a9 100644
--- a/src/3rdparty/harfbuzz/tests/shaping/main.cpp
+++ b/src/3rdparty/harfbuzz/tests/shaping/main.cpp
@@ -136,13 +136,13 @@ HB_Error hb_getPointInOutline(HB_Font font, HB_Glyph glyph, int flags, hb_uint32
     return HB_Err_Ok;
 }
 
-void hb_getGlyphMetrics(HB_Font font, HB_Glyph glyph, HB_GlyphMetrics *metrics)
+void hb_getGlyphMetrics(HB_Font, HB_Glyph, HB_GlyphMetrics *metrics)
 {
     // ###
     metrics->x = metrics->y = metrics->width = metrics->height = metrics->xOffset = metrics->yOffset = 0;
 }
 
-HB_Fixed hb_getFontMetric(HB_Font font, HB_FontMetric metric)
+HB_Fixed hb_getFontMetric(HB_Font, HB_FontMetric )
 {
     return 0; // ####
 }
@@ -169,6 +169,8 @@ public slots:
     void initTestCase();
     void cleanupTestCase();
 private slots:
+    void greek();
+
     void devanagari();
     void bengali();
     void gurmukhi();
@@ -203,18 +205,25 @@ void tst_QScriptEngine::cleanupTestCase()
     FT_Done_FreeType(freetype);
 }
 
-struct ShapeTable {
-    unsigned short unicode[16];
-    unsigned short glyphs[16];
+class Shaper
+{
+public:
+    Shaper(FT_Face face, HB_Script script, const QString &str);
+
+    HB_FontRec hbFont;
+    HB_ShaperItem shaper_item;
+    QVarLengthArray<HB_Glyph> hb_glyphs;
+    QVarLengthArray<HB_GlyphAttributes> hb_attributes;
+    QVarLengthArray<HB_Fixed> hb_advances;
+    QVarLengthArray<HB_FixedPoint> hb_offsets;
+    QVarLengthArray<unsigned short> hb_logClusters;
+
 };
 
-static bool shaping(FT_Face face, const ShapeTable *s, HB_Script script)
+Shaper::Shaper(FT_Face face, HB_Script script, const QString &str)
 {
-    QString str = QString::fromUtf16( s->unicode );
-
     HB_Face hbFace = HB_NewFace(face, hb_getSFntTable);
 
-    HB_FontRec hbFont;
     hbFont.klass = &hb_fontClass;
     hbFont.userData = face;
     hbFont.x_ppem  = face->size->metrics.x_ppem;
@@ -222,7 +231,6 @@ static bool shaping(FT_Face face, const ShapeTable *s, HB_Script script)
     hbFont.x_scale = face->size->metrics.x_scale;
     hbFont.y_scale = face->size->metrics.y_scale;
 
-    HB_ShaperItem shaper_item;
     shaper_item.kerning_applied = false;
     shaper_item.string = reinterpret_cast<const HB_UChar16 *>(str.constData());
     shaper_item.stringLength = str.length();
@@ -237,11 +245,6 @@ static bool shaping(FT_Face face, const ShapeTable *s, HB_Script script)
     shaper_item.glyphIndicesPresent = false;
     shaper_item.initialGlyphCount = 0;
 
-    QVarLengthArray<HB_Glyph> hb_glyphs(shaper_item.num_glyphs);
-    QVarLengthArray<HB_GlyphAttributes> hb_attributes(shaper_item.num_glyphs);
-    QVarLengthArray<HB_Fixed> hb_advances(shaper_item.num_glyphs);
-    QVarLengthArray<HB_FixedPoint> hb_offsets(shaper_item.num_glyphs);
-    QVarLengthArray<unsigned short> hb_logClusters(shaper_item.num_glyphs);
 
     while (1) {
         hb_glyphs.resize(shaper_item.num_glyphs);
@@ -263,10 +266,68 @@ static bool shaping(FT_Face face, const ShapeTable *s, HB_Script script)
 
         if (HB_ShapeItem(&shaper_item))
             break;
-
     }
 
     HB_FreeFace(hbFace);
+}
+
+
+#if defined(Q_WS_X11)
+static bool decomposedShaping(FT_Face face, HB_Script script, const QChar &ch)
+{
+    QString uc = QString().append(ch);
+    Shaper shaper(face, script, uc);
+
+    uc = uc.normalized(QString::NormalizationForm_D);
+    Shaper decomposed(face, script, uc);
+
+    if( shaper.shaper_item.num_glyphs != decomposed.shaper_item.num_glyphs )
+        goto error;
+
+    for (unsigned int i = 0; i < shaper.shaper_item.num_glyphs; ++i) {
+        if ((shaper.shaper_item.glyphs[i]&0xffffff) != (decomposed.shaper_item.glyphs[i]&0xffffff))
+            goto error;
+    }
+    return true;
+ error:
+    QString str = "";
+    int i = 0;
+    while (i < uc.length()) {
+        str += QString("%1 ").arg(uc[i].unicode(), 4, 16);
+        ++i;
+    }
+    qDebug("%s: decomposedShaping of char %4x failed\n    decomposedString: %s\n   nglyphs=%d, decomposed nglyphs %d",
+           face->family_name,
+           ch.unicode(), str.toLatin1().data(),
+           shaper.shaper_item.num_glyphs,
+           decomposed.shaper_item.num_glyphs);
+
+    str = "";
+    i = 0;
+    while (i < shaper.shaper_item.num_glyphs) {
+        str += QString("%1 ").arg(shaper.shaper_item.glyphs[i], 4, 16);
+        ++i;
+    }
+    qDebug("    composed glyph result   = %s", str.toLatin1().constData());
+    str = "";
+    i = 0;
+    while (i < decomposed.shaper_item.num_glyphs) {
+        str += QString("%1 ").arg(decomposed.shaper_item.glyphs[i], 4, 16);
+        ++i;
+    }
+    qDebug("    decomposed glyph result = %s", str.toLatin1().constData());
+    return false;
+}
+#endif
+
+struct ShapeTable {
+    unsigned short unicode[16];
+    unsigned short glyphs[16];
+};
+
+static bool shaping(FT_Face face, const ShapeTable *s, HB_Script script)
+{
+    Shaper shaper(face, script, QString::fromUtf16( s->unicode ));
 
     hb_uint32 nglyphs = 0;
     const unsigned short *g = s->glyphs;
@@ -275,16 +336,16 @@ static bool shaping(FT_Face face, const ShapeTable *s, HB_Script script)
 	g++;
     }
 
-    if( nglyphs != shaper_item.num_glyphs )
+    if( nglyphs != shaper.shaper_item.num_glyphs )
 	goto error;
 
     for (hb_uint32 i = 0; i < nglyphs; ++i) {
-	if ((shaper_item.glyphs[i]&0xffffff) != s->glyphs[i])
+        if ((shaper.shaper_item.glyphs[i]&0xffffff) != s->glyphs[i])
 	    goto error;
     }
     return true;
  error:
-    str = "";
+    QString str = "";
     const unsigned short *uc = s->unicode;
     while (*uc) {
 	str += QString("%1 ").arg(*uc, 4, 16);
@@ -293,18 +354,40 @@ static bool shaping(FT_Face face, const ShapeTable *s, HB_Script script)
     qDebug("%s: shaping of string %s failed, nglyphs=%d, expected %d",
            face->family_name,
            str.toLatin1().constData(),
-           shaper_item.num_glyphs, nglyphs);
+           shaper.shaper_item.num_glyphs, nglyphs);
 
     str = "";
     hb_uint32 i = 0;
-    while (i < shaper_item.num_glyphs) {
-	str += QString("%1 ").arg(shaper_item.glyphs[i], 4, 16);
+    while (i < shaper.shaper_item.num_glyphs) {
+        str += QString("%1 ").arg(shaper.shaper_item.glyphs[i], 4, 16);
 	++i;
     }
     qDebug("    glyph result = %s", str.toLatin1().constData());
     return false;
 }
 
+
+void tst_QScriptEngine::greek()
+{
+    FT_Face face = loadFace("DejaVuSans.ttf");
+    if (face) {
+        for (int uc = 0x1f00; uc <= 0x1fff; ++uc) {
+            QString str;
+            str.append(uc);
+            if (str.normalized(QString::NormalizationForm_D).normalized(QString::NormalizationForm_C) != str) {
+                //qDebug() << "skipping" << hex << uc;
+                continue;
+            }
+            if (uc == 0x1fc1 || uc == 0x1fed)
+                continue;
+            QVERIFY( decomposedShaping(face, HB_Script_Greek, QChar(uc)) );
+        }
+    } else {
+        QSKIP("couln't find DejaVu Sans", SkipAll);
+    }
+}
+
+
 void tst_QScriptEngine::devanagari()
 {
     {