1 files changed, 118 insertions, 2 deletions

diff --git a/po/bin/po2txt b/po/bin/po2txt
index f10ee5b..ba29001 100755
--- a/po/bin/po2txt
+++ b/po/bin/po2txt
@@ -5,23 +5,37 @@
 # Note: en.po ist treated specially!
 
 use Getopt::Long;
+use Encode;
+
+sub arabic_is_letter;
+sub arabic_read_map;
+sub arabic_conv;
 
 sub read_texts;
 sub join_msg;
+sub fribidi;
 
-$opt_product = "SUSE Linux";
+$opt_product = "openSUSE";
 
 GetOptions(
-  'product=s' => \$opt_product
+  'product=s' => \$opt_product,
+  'verbose|v' => \$opt_arabic_verbose,
 );
 
 
+chomp ($tmp_file = `mktemp /tmp/po2txt.XXXXXXXXXX`);
+
+arabic_read_map;
+
 for $lang (@ARGV) {
   $lang = 'en' if $lang eq 'bootloader.pot';
   $lang =~ s/\.po$//;
   read_texts $lang;
 }
 
+unlink $tmp_file;
+
+
 sub read_texts
 {
   local $_;
@@ -100,6 +114,7 @@ sub read_texts
     $txt =~ s/\\"/"/g;
     $txt =~ s/\\\\/\\/g;
     $txt =~ s/\\n/\n/g;
+    $txt = fribidi $txt if $lang =~ /^(ar|he)/;
     print "$txt\x00"
   }
 
@@ -163,3 +178,104 @@ sub join_msg
 
   return $m;
 }
+
+
+sub fribidi
+{
+  local $_;
+  my $m;
+
+  $m = shift;
+
+  open F, ">$tmp_file";
+  print F arabic_conv($m);
+  close F;
+
+  open F, "cat $tmp_file | fribidi --nopad --nobreak |";
+  $m = undef;
+  while(read F, $_, 0x10000) {
+    $m .= $_;
+  }
+  close F;
+
+  return $m;
+}
+
+
+sub arabic_conv {
+  local $_;
+  my (@c, @m, $i, @attr, @attr_name);
+
+  push @c, 0;
+  push @c, unpack("V*", encode("utf32le", decode("utf8", $_[0])));
+  push @c, 0;
+
+  # isolated: 0, initial: 1, final: 2, medial: 3
+  for ($i = 1; $i < @c - 1; $i++) {
+    next if !arabic_is_letter $c[$i];
+    $attr[$i-1] += 2 if arabic_is_letter $c[$i-1];
+    $attr[$i-1] += 1 if arabic_is_letter $c[$i+1];
+  }
+
+  shift @c;
+  pop @c;
+
+  @attr_name = ( "isolated", "initial", "final", "medial" );
+
+  for ($i = 0; $i < @c; $i++) {
+    $m = $c[$i];
+    if(arabic_is_letter $c[$i]) {
+      $m = $arabic_map->{$c[$i]}{$attr_name[$attr[$i]]};
+      if(!$m && $attr[$i] == 3) {		# medial -> final
+        $attr[$i] = 2;
+        $m = $arabic_map->{$c[$i]}{$attr_name[$attr[$i]]};
+        if($m && $i < @c - 1) {
+          if($attr[$i+1] == 3) {		# next char: medial -> initial
+            $attr[$i+1] = 1;
+          }
+          elsif($attr[$i+1] == 1) {	# next char: initial ->isolated
+            $attr[$i+1] = 0;
+          }
+        }
+      }
+      $m = $c[$i] unless $m;
+    }
+    push @m, $m;
+
+    printf STDERR "%04x -> %04x (%s)\n", $c[$i], $m, $attr_name[$attr[$i]] if $opt_arabic_verbose;
+  }
+
+  return encode("utf8", decode("utf32le", pack("V*", @m)));
+}
+
+
+sub arabic_is_letter
+{
+  return $_[0] >= 0x600 && $_[0] <= 0x6ff ? 1 : 0
+}
+
+
+sub arabic_read_map
+{
+  local $_;
+  my (@i, $u, $m);
+
+  open F, "bin/arabic.txt";
+  while(<F>) {
+    @i = split /;/;
+    $u = hex $i[0];
+    if($i[5] =~ /^<(isolated|initial|final|medial)> (\S{4})$/) {
+      $m = hex $2;
+      if($arabic_map->{$m}{$1}) {
+        printf STDERR "%04x already has a '$1' form: %04x\n", $arabic_map->{$m}{$1}
+      }
+      $arabic_map->{$m}{$1} = $u;
+      # printf STDERR "%04x %s %04x\n", $u, $1, $m;
+    }
+  }
+  close F;
+
+  print STDERR "warning: no arabic support\n" unless $arabic_map
+}
+
+