decompiler

#!/usr/bin/perl
#
#  This is a sleazy program which decompiles compiled scripts.
#
#  It is possibly unreliable.
#
# Steve
# --
#

use strict;
use warnings;

use Getopt::Long;


#
#  Default to not showing the address of code.
#
#  Although you need the address for the jumps/calls to make sense
# adding them means you can't pipe the output back to the compiler.
#
#
my %CONFIG = ( show_address => 0 );

#
#  Parse options
#
exit if ( !GetOptions( "show-address", \$CONFIG{ 'show_address' } ) );



#
#  Process each file o nthe command-line.
#
foreach my $file (@ARGV)
{
    decompile($file);
}


#
# All done.
#
exit(0);



=begin doc

Open the named file, and attempt to decompile the binary opcodes from
within it.

Once again we're complicated by the fact that we have variable-length
instructions:

* A NOP/EXIT instruction is a single-byte.
* A JUMP/CALL instruction is three bytes.
* An inline-string might be 50+ bytes.

To cope with this we read the program entirely into a string, which
should be safe because the max-size of a binary is 64k and iterate over
in one-byte amounts.  We'll update the offset we're working with as we
go to skip over arguments & etc.

=end doc

=cut

sub decompile
{
    my ($file) = (@_);

    #
    #  OPen the file.
    #
    open( my $handle, "<", $file ) or
      die "Failed to open $file - $!";
    binmode($handle);


    #
    #  Read it entirely.
    #
    my $buff = '';
    my $rc;
    do {$rc = sysread( $handle, $buff, 5000000, length($buff) );} while ($rc);
    close($handle);

    #
    #  Now we have a buffer and we'll split that into an array.
    #
    my @data = split( //, $buff );

    #
    #  Parse the program from 0x0000 -> Exit.
    #
    #  NOTE:  The code to decompile will update the index - as it skips
    # over registers, data, & etc.
    #
    for ( my $i = 0 ; $i <= $#data ; $i++ )
    {

        #
        #  Show the address we're processing, if we should.
        #
        print sprintf( "%04X", $i ) if ( $CONFIG{ 'show_address' } );

        #
        #  Get the opcode
        #
        my $opcode = ord( $data[$i] );


        #
        #  Now work out what the hell it is..
        #


        if ( $opcode == 0x00 )
        {
            print "\texit\n";
        }
        elsif ( $opcode == 0x01 )
        {
            my $reg = ord( $data[$i + 1] );
            my $v1  = ord( $data[$i + 2] );
            my $v2  = ord( $data[$i + 3] );

            my $val = $v1 + ( 256 * $v2 );
            $val = sprintf( "0x%04X", $val );
            print "\tstore #$reg, $val\n";

            $i += 3;
        }
        elsif ( $opcode == 0x02 )
        {
            my $reg = ord( $data[$i + 1] );
            print "\tprint_int #$reg\n";
            $i += 1;
        }
        elsif ( $opcode == 0x03 )
        {
            my $reg = ord( $data[$i + 1] );
            print "\tint2string #$reg\n";
            $i += 1;
        }
        elsif ( $opcode == 0x04 )
        {
            my $reg = ord( $data[$i + 1] );
            print "\trandom #$reg\n";
            $i += 1;
        }
        elsif ( $opcode == 0x10 )
        {
            my $v1 = ord( $data[$i + 1] );
            my $v2 = ord( $data[$i + 2] );

            my $val = $v1 + ( 256 * $v2 );
            $val = sprintf( "0x%04X", $val );

            print "\tjmp $val\n";
            $i += 2;
        }
        elsif ( $opcode == 0x11 )
        {
            my $v1 = ord( $data[$i + 1] );
            my $v2 = ord( $data[$i + 2] );

            my $val = $v1 + ( 256 * $v2 );
            $val = sprintf( "0x%04X", $val );

            print "\tjmpz $val\n";

            $i += 2;
        }
        elsif ( $opcode == 0x12 )
        {
            my $v1 = ord( $data[$i + 1] );
            my $v2 = ord( $data[$i + 2] );

            my $val = $v1 + ( 256 * $v2 );
            $val = sprintf( "0x%04X", $val );
            print "\tjmpnz $val\n";

            $i += 2;
        }
        elsif ( $opcode == 0x20 )
        {
            my $r1  = ord( $data[$i + 1] );
            my $in1 = ord( $data[$i + 2] );
            my $in2 = ord( $data[$i + 3] );

            print "\txor #$r1, #$in1, #$in2\n";
            $i += 3;
        }
        elsif ( $opcode == 0x21 )
        {
            my $r1  = ord( $data[$i + 1] );
            my $in1 = ord( $data[$i + 2] );
            my $in2 = ord( $data[$i + 3] );

            print "\tadd #$r1, #$in1, #$in2\n";
            $i += 3;
        }
        elsif ( $opcode == 0x22 )
        {
            my $r1  = ord( $data[$i + 1] );
            my $in1 = ord( $data[$i + 2] );
            my $in2 = ord( $data[$i + 3] );

            print "\tsub #$r1, #$in1, #$in2\n";
            $i += 3;
        }
        elsif ( $opcode == 0x23 )
        {
            my $r1  = ord( $data[$i + 1] );
            my $in1 = ord( $data[$i + 2] );
            my $in2 = ord( $data[$i + 3] );

            print "\tmul #$r1, #$in1, #$in2\n";
            $i += 3;
        }
        elsif ( $opcode == 0x24 )
        {
            my $r1  = ord( $data[$i + 1] );
            my $in1 = ord( $data[$i + 2] );
            my $in2 = ord( $data[$i + 3] );

            print "\tdiv #$r1, #$in1, #$in2\n";
            $i += 3;
        }
        elsif ( $opcode == 0x25 )
        {
            my $reg = ord( $data[$i + 1] );
            print "\tinc #$reg\n";
            $i += 1;
        }
        elsif ( $opcode == 0x26 )
        {
            my $reg = ord( $data[$i + 1] );
            print "\tdec #$reg\n";
            $i += 1;
        }
        elsif ( $opcode == 0x27 )
        {
            my $r1  = ord( $data[$i + 1] );
            my $in1 = ord( $data[$i + 2] );
            my $in2 = ord( $data[$i + 3] );

            print "\tand #$r1, #$in1, #$in2\n";
            $i += 3;
        }
        elsif ( $opcode == 0x28 )
        {
            my $r1  = ord( $data[$i + 1] );
            my $in1 = ord( $data[$i + 2] );
            my $in2 = ord( $data[$i + 3] );

            print "\tor #$r1, #$in1, #$in2\n";
            $i += 3;
        }
        elsif ( $opcode == 0x30 )
        {
            my $reg  = ord( $data[$i + 1] );
            my $len1 = ord( $data[$i + 2] );
            my $len2 = ord( $data[$i + 3] );
            my $len  = $len1 + 256 * $len2;

            my $str = "";
            for ( my $d = 0 ; $d < $len ; $d++ )
            {
                my $c = $data[$i + 4 + $d];
                if ( $c eq "\n" )
                {
                    $c = "\\n";
                }
                if ( $c eq "\t" )
                {
                    $c = "\\t";
                }
                $str .= $c;
            }
            print "\tstore #$reg, \"$str\"\n";
            $i += 3;
            $i += $len;

        }
        elsif ( $opcode == 0x31 )
        {
            my $reg = ord( $data[$i + 1] );
            print "\tprint_str #$reg\n";
            $i += 1;
        }
        elsif ( $opcode == 0x32 )
        {
            my $reg = ord( $data[$i + 1] );
            my $in1 = ord( $data[$i + 2] );
            my $in2 = ord( $data[$i + 3] );
            print "\tconcat #$reg, #$in1, #$in2\n";
            $i += 3;
        }
        elsif ( $opcode == 0x33 )
        {
            my $reg = ord( $data[$i + 1] );
            print "\tsystem #$reg\n";
            $i += 1;
        }
        elsif ( $opcode == 0x34 )
        {
            my $reg = ord( $data[$i + 1] );
            print "\tstring2int #$reg\n";
            $i += 1;
        }
        elsif ( $opcode == 0x40 )
        {
            my $reg1 = ord( $data[$i + 1] );
            my $reg2 = ord( $data[$i + 2] );
            print "\tcmp #$reg1,#$reg2\n";
            $i += 2;
        }
        elsif ( $opcode == 0x41 )
        {
            my $reg = ord( $data[$i + 1] );
            my $v1  = ord( $data[$i + 2] );
            my $v2  = ord( $data[$i + 3] );

            my $val = $v1 + ( 256 * $v2 );
            $val = sprintf( "0x%04X", $val );

            print "\tcmp #$reg,$val\n";
            $i += 3;
        }
        elsif ( $opcode == 0x42 )
        {

            # cmp string
            my $reg  = ord( $data[$i + 1] );
            my $len1 = ord( $data[$i + 2] );
            my $len2 = ord( $data[$i + 3] );
            my $len  = $len1 + 256 * $len2;

            my $str = "";
            for ( my $d = 0 ; $d < $len ; $d++ )
            {
                my $c = $data[$i + 4 + $d];
                if ( $c eq "\n" )
                {
                    $c = "\\n";
                }
                if ( $c eq "\t" )
                {
                    $c = "\\t";
                }
                $str .= $c;
            }
            print "\tcmp #$reg, \"$str\"\n";
            $i += 3;
            $i += $len;
        }
        elsif ( $opcode == 0x43 )
        {
            my $reg = ord( $data[$i + 1] );
            print "\tis_string #$reg\n";

            $i += 1;
        }
        elsif ( $opcode == 0x44 )
        {
            my $reg = ord( $data[$i + 1] );
            print "\tis_integer #$reg\n";
            $i += 1;
        }
        elsif ( $opcode == 0x50 )
        {
            print "\tnop\n";
        }
        elsif ( $opcode == 0x51 )
        {

            # register store
            my $dst = ord( $data[$i + 1] );
            my $src = ord( $data[$i + 2] );


            $i += 2;
            print "\tstore #$dst,#$src\n";
        }
        elsif ( $opcode == 0x60 )
        {
            my $reg1 = ord( $data[$i + 1] );
            my $reg2 = ord( $data[$i + 2] );
            print "\tpeek #$reg1, #$reg2\n";
            $i += 2;
        }
        elsif ( $opcode == 0x61 )
        {
            my $reg1 = ord( $data[$i + 1] );
            my $reg2 = ord( $data[$i + 2] );
            print "\tpoke #$reg1, #$reg2\n";
            $i += 2;
        }
        elsif ( $opcode == 0x62 )
        {
            my $reg1 = ord( $data[$i + 1] );
            my $reg2 = ord( $data[$i + 2] );
            my $reg3 = ord( $data[$i + 3] );
            print "\tmemcpy #$reg1, #$reg2, #$reg3\n";
            $i += 3;
        }
        elsif ( $opcode == 0x70 )
        {
            my $reg = ord( $data[$i + 1] );
            print "\tpush #$reg\n";
            $i += 1;
        }
        elsif ( $opcode == 0x71 )
        {
            my $reg = ord( $data[$i + 1] );
            print "\tpop #$reg\n";
            $i += 1;
        }
        elsif ( $opcode == 0x72 )
        {
            print "\tret\n";
        }
        elsif ( $opcode == 0x73 )
        {
            my $v1 = ord( $data[$i + 1] );
            my $v2 = ord( $data[$i + 2] );

            my $val = $v1 + ( 256 * $v2 );
            $val = sprintf( "0x%04X", $val );

            print "\tcall $val\n";
            $i += 2;
        }
        else
        {
            print "\tDATA " . $opcode . "\n";

        }
    }
}