Add ZWS to hacky break algorithm.

Fri May  5 18:56:45 2000  Owen Taylor  <otaylor@redhat.com>

	* pango/break.c (pango_break): Add ZWS to hacky
	break algorithm.

	* modules/basic/basic.c (basic_engine_shape): Special
	case zero-width-space as a temporary hack. (What's the
	right solution?)

	* modules/basic/tables-big.i: Added support for TIS-620
	encoding.

	* configure.in pango/modules.[ch] pango/Makefile.am
	modules/**: First stab at support for linking modules
	directly into Pango. Add a --with-included-modules=
	flag that causes the specified modules to be built
	as convenience libraries and linked directly into
	libpangox.

Tue May  2 22:59:52 2000  Owen Taylor  <otaylor@redhat.com>

	* modules/basic/basic.c: Get rid of link list of masks
	in cache structure in favor of an array. (This is
	easy to do now since we already have linear indices
	for the masks from the new table format.)

	* pango/modules.c pango/pango-context.c pango/pangox.c:
	Modify _pango_find_map() to take quarks for the
	engine type and render type instead of strings.
	Get rid of the map hash table in favor of a GList
	with the most recently used map at the beginning.

	* pango/modules.[ch] pango/pango-context.c pango/pangox.c:
	Add some utility functions for getting the engine
	for a particular character in a map. Using modules.c
	knowledge of map structure, this allows us to save
	a bunch of useless strcmps.

	* pango/pango-context.c (add_engines): Remove unused
	lookup of shape mask.

	* modules/basic/tables-{small,big}.i modules/basic/basic.c
	tools/compress-table.pl: Reencode mask table to avoid binary
	searches and save a bit of space.

	* modules/basic/basic.c (find_converter): Get rid
	of gratuitous use of hash tables for looking up
	iconv converters.

	* modules/basic/tables-{small,big}.i modules/basic/basic.c:
	Use conv_ucs4 instead of conv_8bit for latin-1.

	* pango/pango-layout.c: Avoid calling pango_glyph_string_extents() -
	just add up the widths from shaping.

1 files changed, 78 insertions, 13 deletions

diff --git a/tools/compress-table.pl b/tools/compress-table.pl
index 83b268cb..a11f3d8a 100755
--- a/tools/compress-table.pl
+++ b/tools/compress-table.pl
@@ -7,6 +7,37 @@ sub convert {
     return $s;
 }
 
+$combo_index = 1;
+
+sub add {
+    my $combo = shift;
+    if (!exists($combos{$combo})) {
+	$combos{$combo} = $combo_index++;
+	printf "  $combo,\n", $combos{$combo};
+    }
+}
+
+my $col = 0;
+
+sub output {
+    my ($start,$u,$index) = @_;
+    
+    for (my $i = $start; $i < $u; $i++) {
+	print "  0,";
+	$col = ($col + 1) % 16;
+	if ($col == 0) {
+	    print "\n";
+	}
+    }
+    printf " %2d,", $index;
+    $col = ($col + 1) % 16;
+    if ($col == 0) {
+	print "\n";
+    }
+}
+
+print "const guint32 char_mask_map[] = {\n  0,\n";
+
 open TABLE, "table";
 
 $encodings = "";
@@ -16,28 +47,62 @@ while (<TABLE>) {
 	($u, $e) = ($1, $2);
 
 	$u = oct($u);
-	
-	if (!defined $start) {
-	    $start = $u;
+
+	if (!defined $old_u) {
 	    $old_u = $u;
 	    $encodings = convert($e);
-	    $end = $u;
 	} elsif ($old_u ne $u) {
-	    if (!defined $old_encodings) {
-		$old_encodings = $encodings;
-	    } elsif ($old_encodings ne $encodings || $old_u != $end + 1) {
-		     
-		printf "{ %#x, %#x, $old_encodings },\n", $start, $end;
-		$start = $old_u;
-		$old_encodings = $encodings;
-	    }
-	    $end = $old_u;
+	    add($encodings);
+	    $old_u = $u;
+	    $encodings = convert($e);
+	} else {
+	    $encodings .= "|".convert($e);
+	}
+    }
+}
+
+if (defined $old_u) {
+    add($encodings);
+} 
+
+close TABLE;
+
+print <<EOF;
+};
+
+const guchar char_masks[] = {
+EOF
+
+open TABLE, "table";
+
+$encodings = "";
+
+undef $old_u;
+$start = 0;
+while (<TABLE>) {
+    if (/^(0x[0-9a-fA-F]+)\s+([^:]*):(0x[0-9a-fA-F]+)/) {
+	($u, $e) = ($1, $2);
+
+	$u = oct($u);
+
+	if (!defined $old_u) {
+	    $old_u = $u;
 	    $encodings = convert($e);
+	} elsif ($old_u ne $u) {
+	    output($start, $old_u, $combos{$encodings});
+	    $start = $old_u + 1;
 	    $old_u = $u;
+	    $encodings = convert($e);
 	} else {
 	    $encodings .= "|".convert($e);
 	}
     }
 }
 
+if (defined $old_u) {
+    output($start, $old_u, $combos{$encodings});
+} 
+
 close TABLE;
+
+print "\n};\n";