Bug in Sinhala script processing

Hi there,

There is a bug in processing Sinhala script. In Sinhala, ZWJ is
needed to group characters, which is different from other Indic
languages. We have fixed this in our Sinhala GNU/Linux distribution
at http://www.linux.lk, but would be great if it is fixed in QT
proper, too :

http://cvs.linux.lk/cgi-bin/cvsweb/sinhala/qt/

Here are the screenshots of incorrect/correct rendering:

http://sinhala.linux.lk/pub/0.2/screenshots/halant-konqueror-before.png
http://sinhala.linux.lk/pub/0.2/screenshots/halant-konqueror-after.png

Please apply. It may also apply to QT4 cleanly. Please let us know
if it doesn't, so that we can create a seperate one.

        Anuradha

--

http://www.linux.lk/~anuradha/
http://www.gnu.org/philosophy/no-word-attachments.html
diff -rua qt-x11-free-3.3.3.old/src/kernel/qscriptengine_x11.cpp qt-x11-free-3.3.3/src/kernel/qscriptengine_x11.cpp

--- qt-x11-free-3.3.3.old/src/kernel/qscriptengine_x11.cpp    2004-08-05 20:42:08.000000000 +0600
+++ qt-x11-free-3.3.3/src/kernel/qscriptengine_x11.cpp    2004-09-26 12:30:55.000000000 +0600
@@ -522,7 +522,7 @@
     Invalid, Consonant, Invalid, Invalid,

     Consonant, Consonant, Consonant, Consonant,
-    Consonant, Consonant, Invalid, Invalid,
+    Consonant, Consonant, Consonant, Invalid,
     Invalid, Invalid, Halant, Invalid,
     Invalid, Invalid, Invalid, Matra,

@@ -1092,6 +1092,14 @@
     { (Form)0, None }
 };

+static const IndicOrdering sinhala_order [] = {
+    { Matra, Below },
+    { Matra, Above },
+    { Matra, Post },
+    { VowelMark, Post },
+    { (Form)0, None }
+};
+
 static const IndicOrdering * const indic_order[] = {
     devanagari_order, // Devanagari
     bengali_order, // Bengali
@@ -1102,7 +1110,7 @@
     telugu_order, // Telugu
     kannada_order, // Kannada
     malayalam_order, // Malayalam
-    devanagari_order // Sinhala // ### no OT specs available, we use devanagari
+    sinhala_order // Sinhala
 };

@@ -1725,7 +1733,8 @@
         break;
         goto finish;
     case Consonant:
-        if ( state == Halant )
+        if ( state == Halant && (( script != QFont::Sinhala )
+        || ( uc[pos-1].unicode() == 0x200d )))
         break;
         goto finish;
     case Halant: