michael@0: #!/usr/bin/perl
michael@0: # parse-mozilla-encoding-table.pl, version 0.6
michael@0: #
michael@0: # Script to deassemble existing Mozilla *.uf or *.ut files
michael@0: # back to source conversion tables.
michael@0: # by Anthony Fok <anthony@thizlinux.com>, ThizLinux Laboratory Ltd., 2002/11/27
michael@0: # License: GNU General Public License, version 2 or newer
michael@0: # 
michael@0: # Used for verifying HKSCS-1999 hkscs.uf and hkscs.ut so that I can make
michael@0: # new ones for HKSCS-2001.  This script is quick-and-dirty and not very
michael@0: # robust, so if the debug output of fromu/tou ever changes, this script
michael@0: # will need to be modified too.  :-)
michael@0: 
michael@0: my %data = ();
michael@0: my $mappingPos = 0;
michael@0: my $filename = shift;
michael@0: my $mode;
michael@0: if ($filename =~ /\.(ut|uf)$/) {
michael@0:   print $filename, "\n";
michael@0:   $mode = $1;
michael@0: } else {
michael@0:   die;
michael@0: }
michael@0: 
michael@0: open(INFILE, "<$filename") or die;
michael@0: 
michael@0: # Quick-and-dirty routine to populate %data
michael@0: while (<INFILE>) {
michael@0:   if (/^Begin of Item ([[:xdigit:]]+)/) {
michael@0:     die if defined($itemId) and hex($itemId) + 1 != hex($1);
michael@0:     $itemId = $1;
michael@0:     <INFILE> =~ /Format ([012])/ or die;
michael@0:     $format = $1;
michael@0:     <INFILE> =~ /srcBegin = ([[:xdigit:]]+)/ or die;
michael@0:     $srcBegin = $1;
michael@0:     
michael@0:     if ($format == 0) {		# Range
michael@0:       <INFILE> =~ /srcEnd = ([[:xdigit:]]+)/ or die;
michael@0:       $srcEnd = $1;
michael@0:       <INFILE> =~ /destBegin = ([[:xdigit:]]+)/ or die;
michael@0:       $destBegin = $1;
michael@0: 
michael@0:       for ($i = hex($srcBegin); $i <= hex($srcEnd); $i++) {
michael@0:         $data{sprintf("%04X",$i)} = sprintf("%04X",
michael@0: 	    hex($destBegin) + $i - hex($srcBegin));
michael@0:       }
michael@0: 
michael@0:       <INFILE> =~ /^End of Item $itemId\s*$/ or die;
michael@0:     }
michael@0:     elsif ($format == 1) {	# Mapping
michael@0:       <INFILE> =~ /srcEnd = ([[:xdigit:]]+)/ or die;
michael@0:       $srcEnd = $1;
michael@0:       <INFILE> =~ /mappingOffset = ([[:xdigit:]]+)/ or die;
michael@0:       $mappingOffset = hex($1);
michael@0:       die unless $mappingOffset == $mappingPos;
michael@0:       <INFILE> =~ /Mapping  =\s*$/ or die;
michael@0:       until ($_ = <INFILE>, /^End of Item/) {
michael@0:         chop;
michael@0:         for $i (split ' ') {
michael@0: 	  $key = sprintf("%04X", hex($srcBegin) - $mappingOffset + $mappingPos++);
michael@0: 	  next if $i eq "FFFD";
michael@0: 	  if (defined($data{$key})) {
michael@0: 	    print "Error: doubly defined. $key was $data{$key}, and now $i.\n";
michael@0: 	  } else {
michael@0: 	    $data{$key} = $i;
michael@0: 	  }
michael@0: 	}
michael@0:       }
michael@0:       die unless $mappingPos - $mappingOffset == hex($srcEnd) - hex($srcBegin) + 1;
michael@0:       /^End of Item $itemId\s*$/ or die;
michael@0:     }
michael@0:     else {			# Single ($format == 2)
michael@0:       <INFILE> =~ /destBegin = ([[:xdigit:]]+)/ or die;
michael@0:       $destBegin = $1;
michael@0:       $data{$srcBegin} = $destBegin;
michael@0:       <INFILE> =~ /^End of Item $itemId\s*$/ or die;
michael@0:     }
michael@0:   }
michael@0: }
michael@0: 
michael@0: # Generate conversion table
michael@0: for $key (sort keys %data) {
michael@0:   if ($mode eq "ut") {
michael@0:     print "0x$key\t0x$data{$key}\n";
michael@0:   } elsif ($mode eq "uf") {
michael@0:     print "0x$data{$key}\t0x$key\n";
michael@0:   } else {
michael@0:     die;
michael@0:   }
michael@0: }
michael@0: 
michael@0: close INFILE;