dwrite/arabic: Set per-glyph action.

dlls/dwrite/Makefile.in

@@ -18,4 +18,5 @@ C_SRCS = \
 	opentype.c \
 	scripts.c \
 	shape.c \
-	shapers/arabic.c
+	shapers/arabic.c \
+	shapers/arabic_table.c

dlls/dwrite/shapers/arabic.c

@@ -75,9 +75,91 @@ static void arabic_collect_features(struct scriptshaping_context *context,
     shape_enable_feature(features, DWRITE_MAKE_OPENTYPE_TAG('m','s','e','t'), 0);
 }
 
+enum arabic_joining_type
+{
+    JOINING_TYPE_U = 0,
+    JOINING_TYPE_L = 1,
+    JOINING_TYPE_R = 2,
+    JOINING_TYPE_D = 3,
+    JOINING_TYPE_C = JOINING_TYPE_D,
+    JOINING_GROUP_ALAPH = 4,
+    JOINING_GROUP_DALATH_RISH = 5,
+    JOINING_TYPES = 6,
+    JOINING_TYPE_T = 6,
+};
+
+static const struct arabic_state_table_entry
+{
+    unsigned char prev_action;
+    unsigned char curr_action;
+    unsigned char next_state;
+}
+arabic_state_table[][JOINING_TYPES] =
+{
+    /*     U,              L,             R,             D,           ALAPH,      DALATH_RISH  */
+    /* State 0: prev was U, not willing to join. */
+    { {NONE,NONE,0}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,6}, },
+
+    /* State 1: prev was R or ISOL/ALAPH, not willing to join. */
+    { {NONE,NONE,0}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,FIN2,5}, {NONE,ISOL,6}, },
+
+    /* State 2: prev was D/L in ISOL form, willing to join. */
+    { {NONE,NONE,0}, {NONE,ISOL,2}, {INIT,FINA,1}, {INIT,FINA,3}, {INIT,FINA,4}, {INIT,FINA,6}, },
+
+    /* State 3: prev was D in FINA form, willing to join. */
+    { {NONE,NONE,0}, {NONE,ISOL,2}, {MEDI,FINA,1}, {MEDI,FINA,3}, {MEDI,FINA,4}, {MEDI,FINA,6}, },
+
+    /* State 4: prev was FINA ALAPH, not willing to join. */
+    { {NONE,NONE,0}, {NONE,ISOL,2}, {MED2,ISOL,1}, {MED2,ISOL,2}, {MED2,FIN2,5}, {MED2,ISOL,6}, },
+
+    /* State 5: prev was FIN2/FIN3 ALAPH, not willing to join. */
+    { {NONE,NONE,0}, {NONE,ISOL,2}, {ISOL,ISOL,1}, {ISOL,ISOL,2}, {ISOL,FIN2,5}, {ISOL,ISOL,6}, },
+
+    /* State 6: prev was DALATH/RISH, not willing to join. */
+    { {NONE,NONE,0}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,FIN3,5}, {NONE,ISOL,6}, }
+};
+
+extern const unsigned short arabic_shaping_table[] DECLSPEC_HIDDEN;
+
+static unsigned short arabic_get_joining_type(WCHAR ch)
+{
+    const unsigned short *table = arabic_shaping_table;
+    return table[table[table[ch >> 8] + ((ch >> 4) & 0x0f)] + (ch & 0xf)];
+}
+
+static void arabic_set_shaping_action(struct scriptshaping_context *context,
+        unsigned int idx, enum arabic_shaping_action action)
+{
+    context->glyph_infos[idx].props &= ~(0xf << 16);
+    context->glyph_infos[idx].props |= (action & 0xf) << 16;
+}
+
 static void arabic_setup_masks(struct scriptshaping_context *context,
         const struct shaping_features *features)
 {
+    unsigned int i, prev = ~0u, state = 0;
+
+    for (i = 0; i < context->length; ++i)
+    {
+        unsigned short this_type = arabic_get_joining_type(context->text[i]);
+        const struct arabic_state_table_entry *entry;
+
+        if (this_type == JOINING_TYPE_T)
+        {
+            arabic_set_shaping_action(context, i, NONE);
+            continue;
+        }
+
+        entry = &arabic_state_table[state][this_type];
+
+        if (entry->prev_action != NONE && prev != ~0u)
+            arabic_set_shaping_action(context, prev, entry->prev_action);
+
+        arabic_set_shaping_action(context, i, entry->curr_action);
+
+        prev = i;
+        state = entry->next_state;
+    }
 }
 
 const struct shaper arabic_shaper =

tools/make_unicode

@@ -435,7 +435,7 @@ my @tolower_table = ();
 my @toupper_table = ();
 my @digitmap_table = ();
 my @category_table = ();
-my @joining_table = ();
+my @initial_joining_table = ();
 my @direction_table = ();
 my @decomp_table = ();
 my @combining_class_table = ();
@@ -645,11 +645,11 @@ sub load_data()
         $direction_table[$src] = $bidi;
         if ($cat eq "Mn" || $cat eq "Me" || $cat eq "Cf")
         {
-            $joining_table[$src] = $joining_types{"T"};
+            $initial_joining_table[$src] = $joining_types{"T"};
         }
         else
         {
-            $joining_table[$src] = $joining_types{"U"};
+            $initial_joining_table[$src] = $joining_types{"U"};
         }
 
         if ($lower ne "")
@@ -1609,6 +1609,7 @@ sub dump_bracket($)
 sub dump_shaping($)
 {
     my $filename = shift;
+    my @joining_table = @initial_joining_table;
 
     my $INPUT = open_data_file( $UNIDATA, "ArabicShaping.txt" );
     while (<$INPUT>)
@@ -1651,6 +1652,52 @@ sub dump_shaping($)
 }
 
 ################################################################
+# dump the Arabic shaping table
+sub dump_arabic_shaping($)
+{
+    my $filename = shift;
+    my @joining_table = @initial_joining_table;
+
+    my $INPUT = open_data_file( $UNIDATA, "ArabicShaping.txt" );
+    while (<$INPUT>)
+    {
+        next if /^\#/;  # skip comments
+        next if /^\s*$/;  # skip empty lines
+        next if /\x1a/;  # skip ^Z
+        if (/^\s*([0-9a-fA-F]+)\s*;.*;\s*([RLDCUT])\s*;\s*(\w+)/)
+        {
+            my $type = $2;
+            my $group = $3;
+
+            if ($group eq "ALAPH" || $group eq "DALATH RISH")
+            {
+                $joining_table[hex $1] = $joining_types{$group};
+            }
+            else
+            {
+                $joining_table[hex $1] = $joining_types{$type};
+            }
+
+            next;
+        }
+        die "malformed line $_";
+    }
+    close $INPUT;
+
+    open OUTPUT,">$filename.new" or die "Cannot create $filename";
+    print "Building $filename\n";
+    print OUTPUT "/* Unicode Arabic shaping */\n";
+    print OUTPUT "/* generated from $UNIDATA:ArabicShaping.txt */\n";
+    print OUTPUT "/* DO NOT EDIT!! */\n\n";
+    print OUTPUT "#include \"windef.h\"\n\n";
+
+    dump_two_level_mapping( "arabic_shaping_table", 0, 16, @joining_table );
+
+    close OUTPUT;
+    save_file($filename);
+}
+
+################################################################
 # dump the Vertical Orientation table
 sub dump_vertical($)
 {
@@ -2710,6 +2757,7 @@ dump_mirroring( "dlls/dwrite/mirror.c" );
 dump_bracket( "dlls/gdi32/uniscribe/bracket.c" );
 dump_bracket( "dlls/dwrite/bracket.c" );
 dump_shaping( "dlls/gdi32/uniscribe/shaping.c" );
+dump_arabic_shaping( "dlls/dwrite/shapers/arabic_table.c" );
 dump_linebreak( "dlls/gdi32/uniscribe/linebreak.c" );
 dump_linebreak( "dlls/dwrite/linebreak.c" );
 dump_scripts( "dlls/dwrite/scripts" );